Abstract
Na-K-Cl cotransporter isoform 1 (NKCC1) plays an important role in maintenance of intracellular Na+, K+, and Cl- levels in astrocytes. We propose that NKCC1 may contribute to perturbations of ionic homeostasis in astrocytes under ischemic conditions. After 3-8 hr of oxygen and glucose deprivation (OGD), NKCC1-mediated 86Rb influx was significantly increased in astrocytes from NKCC1 wild-type (NKCC1+/+) and heterozygous mutant (NKCC1+/-) mice. Phosphorylated NKCC1 protein was increased in NKCC1+/+ astrocytes at 2 hr of OGD. Two hours of OGD and 1 hr of reoxygenation (OGD/REOX) triggered an ∼3.6-fold increase in intracellular Na+ concentration ([Na+]i) in NKCC1+/+ astrocytes. Inhibition of NKCC1 activity by bumetanide or ablation of the NKCC1 gene significantly attenuated the rise in [Na+]i. Moreover, NKCC1+/+ astrocytes swelled by 10-30% during 20-60 min of OGD. Either genetic ablation of NKCC1 or inhibition of NKCC1 by bumetanide-attenuated OGD-mediated swelling. An NKCC1-mediated increase in [Na+]i may subsequently affect Ca2+ signaling through the Na+/Ca2+ exchanger (NCX). A rise in [Ca2+]i was detected after OGD/REOX in the presence of a sarcoplasmic-endoplasmic reticulum (ER) Ca2+-ATPase inhibitor thapsigargin. Moreover, OGD/REOX led to a significant increase in Ca2+ release from ER Ca2+ stores. Furthermore, KB-R7943 (2-[2-[4(4-nitrobenzyloxy)phenyl]ethyl]isothiourea mesylate), an inhibitor of reverse-mode operation of NCX, abolished the OGD/REOX-induced enhancement in filling of ER Ca2+ stores. OGD/REOX-mediated Ca2+ accumulation in ER Ca2+ stores was absent when NKCC1 activity was ablated or pharmacologically inhibited. These findings imply that stimulation of NKCC1 activity leads to Na+ accumulation after OGD/REOX and that subsequent reverse-mode operation of NCX contributes to increased Ca2+ accumulation by intracellular Ca2+ stores.
Introduction
Na-K-Cl cotransporter isoform 1 (NKCC1) belongs to the cation-dependent Cl- transporter family and transports Na+, K+, and Cl- into cells under physiological conditions (Russell, 2000). NKCC1 is expressed in rat cortical astrocytes (Yan et al., 2001), oligodendrocytes (Hoppe and Kettenmann, 1989; Wang et al., 2003), and Schwann cells (Alvarez-Leefmans, 2001). Functions of NKCC1 in these cells include K+ and Na+ uptake (Su et al., 2000, 2002b) and accumulation of Cl- above its electrochemical equilibrium (Hoppe and Kettenmann, 1989; Wang et al., 2003). In recent studies, we found that NKCC1 contributes to K+ uptake, swelling, and swelling-induced glutamate release in astrocytes in the presence of high extracellular K+ (Su et al., 2002a,b).
Disruption of Na+ and Ca2+ homeostasis plays an important role in ischemic cell damage (Siesjö, 1992). A steep inwardly directed Na+ gradient is essential for glial functions, such as glutamate reuptake and regulation of intracellular ion concentrations by other secondary ion transporters (Walz, 1989; Longuemare et al., 1999). Breakdown of the Na+ gradient is one of the key elements in promoting cellular damage in astrocytes during energy failure (Longuemare et al., 1999). An increase in intracellular Na+ concentration ([Na+]i) was found in rat spinal cord astrocytes (Rose et al., 1998), rat cortical astrocytes (Longuemare et al., 1999), and mouse cortical astrocytes (Silver et al., 1997) when these cells were exposed to glucose deprivation, NaN3-mediated chemical hypoxia, and simulated ischemia, respectively. Mechanisms of the rise in [Na+]i under reduced energy production conditions are not well understood. Inhibition of Na+/K+-ATPase activity via limited energy production results in Na+ accumulation (Silver et al., 1997). However, blocking voltage-gated Na+ channels had no effect on the rise of [Na+]i in rat spinal astrocytes during chemical hypoxia (Rose et al., 1998). It remains to be clarified which other Na+ entry pathways contribute to Na+ accumulation in astrocytes under energy deprivation.
Reduction in the Na+ gradient affects Ca2+ homeostasis in cells. Elevation of [Na+]i increases intracellular Ca2+ via reverse-mode operation of the Na+/Ca2+ exchanger (NCX) in neurons and causes irreversible injury during anoxia and ischemia (Li et al., 2000). An increase in intracellular Ca2+ via the reverse-mode operation of NCX occurs in rat astrocytes when [Na+]i was raised by inhibition of Na+/K+-ATPase activity or activation of AMPA channels (Goldman et al., 1994; Smith et al., 2000). However, it is unknown whether accumulation of intracellular Na+ during anoxia and ischemia can trigger Ca2+ entry in astrocytes via the reverse-mode operation of NCX.
In the present study, we examined Na+ entry mediated by NKCC1 after 2 hr of oxygen and glucose deprivation and 1 hr of reoxygenation (OGD/REOX) and the subsequent effects on Ca2+ signaling in mouse cortical astrocytes. We found that NKCC1 activity contributes to intracellular Na+ overload and Ca2+ accumulation in intracellular Ca2+ stores. The latter is in part a result of the reverse-mode operation of NCX.
Materials and Methods
Materials. Eagle's modified essential medium and HBSS were from Mediatech Cellgro (Herndon, VA). Fetal bovine serum was obtained from HyClone (Logan, UT). Collagen type I was from Collaborative Biomedical Products (Bedford, MA). Nigericin, bradykinin, thapsigargin, carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP), oligomycin, rotenone, and β-actin monoclonal antibody were purchased from Sigma (St. Louis, MO). GFAP monoclonal antibody was purchased from Sternberger Monoclonals (Lutherville, MD). Sodium binding benzofuran isophthalate AM (SBFI AM), fluorescein-methylene-iminodiacetate AM (calcein AM), 4-bromo A-23187 (calcimycin/5-(methylamino)-2[(2R,3R,6S,8S,9R,11R)-3,9,11-trimethyl-8-[(1S-1-methyl-2-oxo-2-(1H)-pryrrol-2-yl)-ethyl]-1,7-dioxaspiro[5.5]undec-2-yl]methyl-4-benzoxazolecarboxylic acid), and fura-2 AM were from Molecular Probes (Eugene, OR). Rubidium-86 was purchased from PerkinElmer Life Sciences (Emeryville, CA). KB-R7943 (2-[2-[4(4-nitrobenzyloxy) phenyl]ethyl]isothiourea mesylate) was from Tocris (Ellisville, MO). 2-Aminoethoxydiphenylborate (2-APB) was from EMC Biosciences (San Diego, CA). R5 polyclonal antibody was characterized previously (Flemmer et al., 2002) and was a kind gift from Dr. Biff Forbush (Yale University, New Haven, CT).
Animals and genotype analysis. NKCC1 homozygous mutant and wild-type mice were obtained by breeding gene-targeted NKCC1 heterozygous mutant mice, and genotypes were determined by PCR analysis of DNA from tail biopsies as described previously (Flagella et al., 1999; Su et al., 2002b).
Primary culture of mouse cortical astrocytes. Dissociated cortical astrocyte cultures were established as described previously (Su et al., 2002b). Cerebral cortices were removed from 1-d-old NKCC1+/+, NKCC1+/-, or NKCC1-/- mice. The cortices were incubated in a trypsin solution (0.25 mg/ml HBSS) for 25 min at 37°C. The dissociated cells were rinsed and resuspended in Eagle's minimum essential medium (EMEM) containing 10% fetal bovine serum. Viable cells (1 × 104 cells per well) were plated in 24-well plates or on glass coverslips in 6-well plates and coated with collagen type-1. Cultures were maintained in a 5% CO2 atmosphere at 37°C and refed every 3 d throughout the study. To obtain morphologically differentiated astrocytes, confluent cultures [10 d in culture, 10 d in vitro (DIV)] were treated with EMEM containing 0.25 mm dibutyryl cAMP (dBcAMP) to induce differentiation. dBcAMP has been widely used to mimic neuronal influences on astrocyte differentiation (Swanson et al., 1997). Experiments were routinely performed in cultures 10-21 DIV.
OGD model. NKCC1+/+, NKCC1+/-, or NKCC1-/- astrocyte cultures in 24-well plates were rinsed twice with an isotonic OGD solution, pH 7.4, containing as follows (in mm): 0 glucose, 26 NaHCO3, 120 NaCl, 5.36 KCl, 0.33 Na2HPO4, 0.44 KH2PO4, 1.27 CaCl2, and 0.81 MgSO4. Cells were incubated in 0.5 ml of the OGD solution in a hypoxic incubator (model 3130; Forma Scientific, Marietta, OH) containing 94% N2, 1% O2, and 5% CO2. During the initial 30 min of OGD, the plates were placed on an orbital shaker (model M48215; Thermolyne, Dubuque, IA) to facilitate equilibration of the hypoxic gases. The OGD incubation was 2 or 8 hr. Normoxic control cells were incubated in 5% CO2 and atmospheric air in an isotonic control solution for 2 or 8 hr. The solutions used for normoxic control and reoxygenation contained 5.5 mm glucose, and the rest of the components in the buffer were identical to the OGD solution. Cells grown on coverslips in six-well plates were treated similarly as described above, except the solution volumes were 1.0 ml.
K+ influx determination. NKCC1 activity was measured as bumetanide-sensitive K+ influx, using 86Rb as a tracer for K+ (Sun and Murali, 1999). Briefly, cultured astrocytes were equilibrated for 3 min at 37°C with an isotonic HEPES-buffered minimum essential medium (MEM; 312 mOsm). The concentrations of components in HEPES-MEM were described previously (Schomberg et al., 2001). Cells were preincubated for 5 min in HEPES-MEM containing either 10 μm bumetanide and or 1 mm ouabain. For assay of NKCC1 or Na+/K+-ATPase activity, cells were exposed to 1 μCi/ml 86Rb in HEPES-MEM for 3 min in the presence of 10 μm bumetanide and/or 1 mm ouabain. Radioactivity of cells extracted in 1% SDS was analyzed by liquid scintillation counting (1900CA; Packard, Downers Grove, IL). The 86Rb influx rate was then calculated as the slope of 86Rb uptake over time and expressed as nanomoles of 86Rb per milligram of protein per minute. 86Rb influx is linear over 7 min (Schomberg et al., 2001). Quadruplet determinations were obtained in each experiment throughout the study, and protein content was determined using the BCA method (Pierce, Rockford, IL).
Immunoprecipitation, gel electrophoresis, and immunoblotting. After normoxic or OGD treatments, cortical astrocytes were washed with ice-cold PBS, pH 7.4, containing phosphatase inhibitors (100 mm NaF, 10 mm Na4P2O7, 2 mm NaVO3, and 0.2 μm microcystin) and protease inhibitors as described previously (Sun and Murali, 1999). Immunoprecipitation was performed as described previously (Schomberg et al., 2003). Cells were lysed and centrifuged at 320 × g for 5 min. The supernatant fraction was collected and centrifuged at 45,000 rpm (SW 55 rotor; Beckman, Fullerton, CA) at 4°C for 30 min. The membrane pellet was resuspended in 0.25 ml of PBS-1% SDS for 30 min at room temperature, and the protein content of the membrane suspension was determined. Two hundred fifty micrograms of membrane protein from each sample were incubated in 0.875 ml of PBS-2% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) at 4°C for 30 min. Samples were centrifuged at 320 × g for 5 min, and the supernatant was collected. Fifteen microliters of a T4 monoclonal antibody (T4 ascites; Departmental Studies Hybridoma Bank, Iowa City, IA) (Lytle et al., 1995) were then added, and samples were rotated on a shaker overnight at 4°C. T4 monoclonal antibody recognizes NKCC1, NKCC2, and spliced isoforms of NKCC1. For immunoprecipitation, 40 μl of a 30% slurry of protein G-Sepharose was added, and the samples were rotated for 2 hr at 4°C. The samples were washed four times with PBS-2% CHAPS and one time with PBS. The samples were denatured in SDS reducing buffer (Bio-Rad, Hercules, CA) and heated at 55°C for 30 min before gel electrophoresis.
The samples and prestained molecular mass markers (Bio-Rad) were then electrophoretically separated on a 6% acrylamide gel (Laemmli, 1970), and the resolved proteins were electrophoretically transferred to a polyvinylidene difluoride membrane (Sun and Murali, 1999). The blots were incubated in 7.5% nonfat dry milk in Tris-buffered saline (TBS) for 2 hr and then incubated for 3 hr with a primary polyclonal antibody (R5) against a diphosphopeptide containing T184 and T189 of shark NKCC1 (NKCC-p; 1:2000) (Flemmer et al., 2002). The blots were rinsed with TBS and incubated with horseradish peroxidase-conjugated secondary IgG for 1 hr (1:3000). Bound secondary antibody was visualized using the enhanced chemiluminescence (ECL) assay (Amersham Biosciences, Piscataway, NJ).
To confirm that each sample contained a similar amount of nonphosphorylated NKCC1 protein before immunoprecipitation, 30 μg of membrane protein from each sample was used directly for immunoblotting with the T4 antibody (1:3000). Expression of β-actin was monitored as an internal control with a monoclonal antibody (1:200).
After the ECL reaction, NKCC1 protein bands on the film were scanned using a Hewlett-Packard ScanJet (4c/T) scanner. The intensity of each band was measured by UN-SCAN-IT gel software (Silk Scientific, Orem, UT). We determined previously that the T4 signal is proportional to protein with loads between 15 and 45 μg (Yan et al., 2001).
Intracellular Na+ measurement. [Na+]i was measured in cultured astrocytes grown on coverslips with the fluorescent dye SBFI AM as described previously (Su et al., 2002b). At the end of 2 hr of OGD, astrocytes were loaded with 10 μm SBFI AM at room temperature in HEPES-MEM containing 0.05% pluronic acid for 45 min. The coverslips were then placed in an open-bath imaging chamber (model RC24; Warner Instruments, Hamden, CT) containing HEPES-MEM at ambient temperature. Astrocytes were excited every 10 sec at 345 and 385 nm, and the emission fluorescences at 510 nm were recorded with a Nikon (Melville, NY) TE 300 inverted epifluorescence microscope and a 40× Super Fluor oil immersion objective lens. Images were collected at 60 min REOX; 340/380 ratios were recorded for 5 min; and absolute [Na+]i was determined for each cell as described previously (Su et al., 2002b).
For a more severe ischemic insult, a hypoxic, acidic, ion-shifted Ringer's (HAIR) buffer was used. Astrocytes grown on coverslips were loaded with 10 μm SBFI AM in HEPES-MEM containing 0.05% pluronic acid for 60 min at 37°C. The coverslips were then placed in an imaging chamber and continuously superfused with HEPES-MEM at 37°C to collect baseline data. After 2 min of baseline perfusion, the cells were superfused with HAIR buffer for 3 min. For reoxygenation, cells were perfused with normoxic HEPES-MEM for 6 min. For the bumetanide treatment study, 5 μm bumetanide was present in loading and all of the subsequent perfusion buffers. The ionic composition of HAIR was as follows (in mm): 34 NaCl, 13 NaHCO3, 3 Na-gluconate, 65 K-gluconate, 38 NMDG-Cl, 1 NaH2PO4, 0.13 CaCl2, 1.5 MgSO4, and 0 glucose (Bondarenko and Chesler, 2001). HAIR buffer, pH 6.6, was sparged with 15% CO2 and 85% N2 for at least 1 hr before the experiment. An average O2 of 1.9 ± 0.2% was determined at the entrance of the chamber using a flow-through oxygen electrode (model 16-730; Microelectrodes, Bedford, NH).
Intracellular 36Cl measurement. Intracellular 36Cl content was measured in astrocytes grown in 24-well plates. After OGD or normoxia, cells were preincubated at 37°C for 30 min in HEPES-MEM containing 36Cl (0.4 μCi/ml), as described previously (Su et al., 2002a), in the presence or absence of 10 μm bumetanide. In our previous study, steady-state intracellular 36Cl was reached by 10 min incubation and maintained for 60 min (Su et al., 2002a). Thus, a 30 min preincubation with 36Cl was used in the current study. Intracellular 36Cl content measurement was terminated by three washes with 1 ml of ice-cold washing buffer (Su et al., 2002a). Radioactivity of the cellular extract in 1% SDS was analyzed by liquid scintillation counting (1900CA; Packard). In each experiment, specific activities (counts per micromole per minute) of 36Cl were determined for each assay condition and used to calculate intracellular Cl- content (micromoles per milligram per protein). Protein content was measured in each sample using the BCA method (Pierce).
Intracellular Ca2+ measurement. To determine changes in [Ca2+]i at 1 hr of REOX after 2 hr of OGD, astrocytes were incubated with 5 μm fura-2 AM at the end of 2 hr of OGD for 45 min (at room temperature). For experiments in determination of [Ca2+]i at the end of 2 hr of OGD, the astrocytes were incubated with 5 μm fura-2 AM during the 2 hr of OGD treatment. Our pilot study indicated that a short period of OGD (2 hr) did not significantly decrease the energy level in astrocytes and that the fluorescence dye loading during OGD was sufficient.
Astrocytes on coverslips were then placed in the open-bath imaging chamber containing HEPES-MEM at ambient temperature. Using a Nikon TE 300 inverted epifluorescence microscope and a 40× Super Fluor oil immersion objective lens, astrocytes were excited every 10 sec at 345 and 385 nm, and the emission fluorescence at 510 nm was recorded. Images were collected and analyzed with the MetaFluor image-processing software. At the end of each experiment, the cells were exposed to 1 mm MnCl2 in Ca2+-free HEPES-MEM. The Ca2+-insensitive fluorescence was subtracted from each wavelength before calculations (Haworth and Redon, 1998). The MnCl2-corrected 345/385 emission ratios were converted to concentration using the Grynkiewicz equation (Grynkiewicz et al., 1985) as follows:
where R is the experimental 345/385 ratio; Rmin is the ratio in HEPES-MEM without Ca2+ and containing 1 mm EGTA and 1 μm 4-bromo A-23187; Rmax is the ratio in HEPES-MEM with 10 mm Ca2+ and 1 μm
4-bromo A-23187 plus 4 μm rotenone and 2 μm FCCP to inhibit active Ca2+ transport by mitochondria; β is the ratio of the 385 nm fluorescence intensities determined at Rmin and Rmax; and Kd is the in vivo dissociation constant for the dye and was assumed to be 370 nm (Petr and Wurster, 1997).
Bradykinin-induced Ca2+ release measurement. Bradykinin-induced Ca2+ release from intracellular stores was assessed by exposing astrocytes to 1 μm bradykinin in Ca2+-free HEPES-MEM (1 mm EGTA) to prevent entry of extracellular Ca2+. Peak Ca2+ release was measured after application of bradykinin. Within 5 min, [Ca2+]i returned to basal levels. In some cells, spontaneous Ca2+ signaling was present before application of bradykinin, which prevented an accurate prebradykinin [Ca2+]i baseline determination. In those cases, a Ca2+-free baseline after 5 min of bradykinin exposure was considered to be the basal [Ca2+]i. After washout of bradykinin, cells were exposed to 1 μm thapsigargin to assess residual Ca2+ release from bradykinin-insensitive Ca2+ stores.
In some experiments, Ca2+ release from mitochondria was monitored after application of FCCP (10 μm) to collapse the mitochondria membrane potential and oligomycin (2.5 μg/ml) to block F1F0-ATP synthase. After a new baseline of [Ca2+]i was established after this treatment, cells were exposed to 1 μm bradykinin in the presence of FCCP-oligomycin.
Measurement of reverse-mode operation of NCX. NKCC1+/+ astrocytes grown on coverslips were incubated with 5 μm fura-2 AM, and cytosolic Ca2+ concentrations were measured as described above. The reverse-mode operation of NCX was induced by exposing cells to a Na+- and Ca2+-free buffer (1 min) followed by a Na+-free buffer that contained Ca2+ (2 min). The effect of an NCX inhibitor, KB-R7943 (0-10 μm), on reverse-mode activity of NCX was investigated. Intracellular Ca2+ levels after induction of the reverse-mode NCX activity were measured, and the percentage of inhibition at given concentrations of KB-R7943 was calculated in each cell by comparing the [Ca2+]i levels at 0 μm KB-R7943 with [Ca2+]i levels in the presence of the particular concentration of KB-R7943.
Measurement of cell swelling. Relative cell volume changes were monitored during OGD on either NKCC1+/+ or NKCC1-/- cells grown on coverslips. Coverslips were placed in a closed-bath imaging chamber (model RC20; Warner Instruments) and perfused continuously at 36°C. The chamber was connected via a manifold to two reservoirs. One contained HEPES-MEM, and the other contained HEPES-MEM lacking glucose and bubbled with 100% N2 for OGD. Oxygen levels of the perfusate were monitored at the entrance of the chamber using a flow-through oxygen electrode. After 30 min of normoxic control perfusion, the chamber was switched to OGD. Oxygen levels reached 2-5% within 2 min and were then maintained at the same level for 60 min.
Relative cell volume changes during OGD were monitored using video-enhanced differential interference contrast (DIC) microscopy, as described in our previous study (Su et al., 2002b). A single astrocyte was visualized every 10 min using a 60× Plan Fluor DIC objective lens, and the images were recorded. The mean cross-sectional area (CSA) was calculated after tracing the perimeter of the cell body with MetaMorph image-processing software.
In addition, cell volume change in single cells was also confirmed by using calcein as a marker of intracellular water volume as described previously (Hamann et al., 2002). Briefly, astrocytes grown on coverslips were placed in a closed-bath imaging chamber and incubated with 2.5 μm calcein AM. Calcein fluorescence in cells was monitored using a 40× Super Fluor oil immersion objective and an FITC filter set [excitation (Ex), 480 nm; emission (Em), 535 nm] (Chroma Technology, Rockingham, VT) until the fluorescence plateaued (∼30 min). Images were collected every 60 sec for 10 min to determine baseline data. The chamber was then switched to OGD buffer for 60 min.
The fluorescence signals were corrected for baseline drift by fitting a single exponential curve to the 10 min control period. The relationship between baseline-corrected calcein fluorescence and external osmolality was characterized by exposing the cells to salt-constant hyperosmotic and hypo-osmotic solutions for 1-2 min, during which time no regulatory volume decrease had yet been initiated (Su et al., 2002b). The osmolality of the buffers was adjusted as described before (Su et al., 2002b). The standards were corrected for the relative background signal. This signal is the fraction of intracellular calcein fluorescence that is insensitive to changes in external osmolality (the y-intercept of a plot of relative cell volume vs the reciprocal of the relative osmotic pressure) (Hamann et al., 2002). The standard fluorescence changes were plotted as a function of the reciprocal of the relative osmotic pressure, and the resulting calibration curve was applied to each experiment.
Measurement of intracellular ATP content. Intracellular ATP content was measured using a luminescence ATP detection assay (PerkinElmer Life Sciences). After 8 hr of OGD or 2 hr of OGD plus 1 hr of REOX, culture medium was removed, and 100 μl of fresh OGD or normoxic control buffer was added to each well. Fifty microliters of lysis solution were added, and plates were gently swirled for 5 min. Aliquots of cell lysates (diluted at 1:10) were added to 50 μl of substrate buffer. Luminescence was detected with an LMax II Luminometer (Molecular Devices, Sunnyvale, CA). In each experiment, ATP levels were determined using ATP standards run concurrently with samples. Protein content in cell lysates was determined in each sample using the BCA method (Pierce). Intracellular ATP content is expressed as nanomoles per milligram of protein.
Measurement of cell death. Cell viability was assessed by propidium iodide (PI) uptake and retention of calcein using a Nikon TE 300 inverted epifluorescence microscope. Cultured astrocytes were rinsed with HEPES-MEM buffer and incubated with 1 μg/ml calcein AM and 10 μg/ml PI in the same buffer at 37°C for 25 min. For cell counting, cells were rinsed and visualized using a Nikon 20× objective lens. Calcein fluorescence was visualized using FITC filters (Ex, 488 nm; Em, 515 nm), and PI fluorescence was visualized with Texas Red filters (Ex, 536 nm; Em, 645 nm). Images were collected in a blind manner, and a total of 1000 cells per condition were counted using MetaMorph image-processing software (Universal Imaging, Downingtown, PA). Cell mortality was expressed as the ratio of PI-positive cells to the sum of calceinand PI-positive cells.
Statistics. Throughout the study, n values represented numbers of cultures used in each experiment. Routinely, parameters were measured in ∼20 cells on a coverslip from each culture. Statistical significance was determined by ANOVA (Bonferroni-Dunn) at a confidence of 95% (p < 0.05).
Results
Expression of NKCC1 protein in astrocytes
To determine relative NKCC1 protein expression in astrocytes of three genotypes, immunoblotting was performed with lysate samples of astrocytes from NKCC1+/+, NKCC1+/-, and NKCC1-/- mice (Fig. 1). Densitometric optical intensity was performed on immunoblots. NKCC1+/+ astrocytes expressed abundant NKCC1 protein, and ∼30% less NKCC1 protein was detected in NKCC1+/- astrocytes. No NKCC1 protein was found in NKCC1-/- astrocytes.
NKCC1 activity is increased after OGD
NKCC1 activity was measured using 86Rb as a tracer for K+ after 3-8 hr of OGD. Bumetanide-sensitive 86Rb influx in NKCC1+/+ astrocytes was increased from a control level of 17.3 ± 3.1 to 33.1 ± 2.8 nmol/mg of protein after 3 hr of OGD (p < 0.05). Bumetanide-sensitive 86Rb influx remained elevated after 8 hr of OGD (23.1 ± 2.8 vs 13.7 ± 2.4 nmol/mg of protein) (Fig. 2A). Similarly, in NKCC1+/- astrocytes, 8 hr of OGD also led to an increase in bumetanide-sensitive 86Rb influx (Fig. 2A). The bumetanide-sensitive 86Rb influx was negligible in NKCC1-/- astrocytes (Su et al., 2002b). Therefore, no measurements were performed in NKCC1-/- astrocytes.
To determine whether Na+/K+-ATPase-mediated K+ influx was altered after OGD, ouabain-sensitive 86Rb influx was measured. After 8 hr of OGD, total 86Rb influx was reduced from 93.7 ± 8.8 to 72.9 ± 7.2 nmol/mg of protein (Fig. 2B) (p < 0.05).
Ouabain-sensitive 86Rb influx was reduced from a control level of 50.7 ± 6.9 to 32.6 ± 4.5 nmol/mg of protein (p < 0.05). Moreover, 86Rb influx that was insensitive to both ouabain and bumetanide was reduced from 36.8 ± 10.0 to 23.2 ± 4.9 nmol/mg of protein. Thus, only NKCC1-mediated 86Rb influx in astrocytes was elevated after OGD, whereas 86Rb influx from Na+/K+-ATPase activity, other transporters, or channels was decreased.
We then investigated whether stimulation of NKCC1 in astrocytes after OGD can be detected by measuring changes in phosphorylated NKCC1. The R5 polyclonal antibody against a diphosphopeptide containing T184 and T189 of NKCC1 (Darman and Forbush, 2002) was used to probe the phosphorylation state of NKCC1. T184, T189, and T202 of the N terminus represent a conserved phosphoregulatory domain of NKCC1 (Darman and Forbush, 2002). Total NKCC1 protein was immunoprecipitated with T4 monoclonal antibody. Subsequently, immunoblots were probed with the R5 antibody. After 2 hr of OGD, phosphorylated NKCC1 protein (NKCC1-p) in NKCC1+/+ astrocytes was increased compared with controls (Fig. 3A). OGD increased levels of NKCC1-p by ∼300% (299 ± 82%; n = 3; p < 0.05). To determine whether there was an upregulation of total NKCC1 protein, the same protein samples were probed with T4 antibody without immunoprecipitation (Fig. 3B). NKCC1 expression was similar in control and OGD-treated cells. In the same blot, no significant changes were observed in expression of β-actin. Together, OGD stimulates NKCC1 phosphorylation in NKCC1+/+ astrocytes, which is consistent with elevation of NKCC1 activity measured with 86Rb influx.
Inhibition of NKCC1 activity attenuated Na+ and Cl- accumulation after OGD
We speculated that the increased activity of NKCC1 after OGD may lead to Na+ accumulation in astrocytes. Because we observed a significant increase in NKCC1 phosphorylation after 2 hr of OGD, we determined whether this increase in NKCC1 activity would cause intracellular Na+ accumulation after 2 hr of OGD. As shown in Figure 4, the baseline [Na+]i level in NKCC1+/+ astrocytes was 12.7 ± 0.6 mm after 3 hr normoxic incubation. After OGD/REOX, the [Na+]i level in NKCC1+/+ astrocytes rose to 46.4 ± 5.0 mm (p < 0.01). Additionally, in NKCC1+/+ cells, inhibition of NKCC1 activity with 5 μm bumetanide reduced the rise in [Na+]i (23.2 ± 0.9 mm; p < 0.01) (Fig. 4). Basal [Na+]i in NKCC1-/- cells was 12.6 ± 1.1 mm under control conditions, which was not different from that of NKCC1+/+ astrocytes. In contrast, in NKCC1-/- astrocytes, the OGD/REOX-mediated rise in [Na+]i was only 19.8 ± 0.6 mm (p < 0.01), significantly less than in NKCC1+/+ astrocytes (Fig. 4). Although the rise in [Na+]i was attenuated in NKCC1-/- cells after OGD/REOX, it remained significantly higher than in NKCC1-/- cells under normoxic conditions. Together, blocking of astrocyte NKCC1 activity, either by genetic ablation or pharmacological inhibition, significantly reduced the accumulation of intracellular Na+ after OGD and REOX.
We also measured intracellular 36Cl content in NKCC1+/+ astrocytes after either 3 hr of normoxia or 2 hr of OGD and 1 hr of REOX. 36Cl content was 0.27 ± 0.01 μg/mg of protein in the normoxia control group. It was increased to 0.31 ± 0.02 μg/mg of protein in the OGD/REOX-treated astrocytes (p < 0.05). Moreover, inhibition of NKCC1 activity with bumetanide (10 μm) abolished the increase in 36Cl accumulation after OGD/REOX (0.22 ± 0.01 μg/mg of protein). Bumetanide had no significant affects on the basal level of 36Cl content in astrocytes (0.26 ± 0.01 μg/mg of protein; p > 0.05). Thus, these data support the hypothesis that NKCC1 activation causes Na+ and Cl- accumulation in astrocytes after OGD and reoxygenation.
In the current OGD/REOX study, the ionic composition of the buffers is held constant. However, during in vivo ischemia, major ionic shifts occur in the extracellular space. Therefore, we investigated whether NKCC1 can contribute to Na+ accumulation when astrocytes are exposed to a hypoxic buffer whose ionic composition was modified to mimic the interstitial milieu of the hypoxic brain (Bondarenko and Chesler, 2001). As shown in Table 1, [Na+]i decreased to 70% of baseline in response to 3 min of HAIR treatment. This decrease can be attributable to either the experimental decrease in extracellular [Na+] ([Na+]o; 69 mm) or the stimulation of Na+/K+-ATPase by the elevated external K+ (62 mm). In NKCC1+/+ astrocytes treated with 5 μm bumetanide, [Na+]i decreased further to 38% of baseline during HAIR exposure. When NKCC1+/+ astrocytes were returned to HEPES-MEM, [Na+]i quickly increased and plateaued to a level 4.6 times baseline within 6 min of REOX (Table 1). However, in bumetanide-treated NKCC1+/+ astrocytes, the increase in [Na+]i at 6 min of REOX was only 2.6 times baseline (Table 1). Thus, NKCC1 plays a similar role in Na+ accumulation under the more pathophysiologically relevant HAIR model.
NKCC1-/- astrocytes exhibited less swelling after OGD
Increases in intracellular Na+ and Cl- will inevitably lead to cell swelling. We therefore investigated whether OGD would induce an increase in relative cross-sectional area (CSAr) in NKCC1+/+ and NKCC1-/- astrocytes. We monitored swelling in NKCC1+/+ and NKCC1-/- astroctyes with time. As shown in Figure 5A, after 20 min of OGD, CSAr in NKCC1+/+ astroctyes increased by 10%, and after 60 min of OGD, it increased to ∼30%. In contrast, changes in CSAr developed slowly in NKCC1-/- astroctyes. NKCC1-/- astroctyes exhibited only a 3% increase in CSAr at 20 min of OGD. After 1 hr of OGD, CSAr increased by 14% in NKCC1-/- astrocytes (Fig. 5A). These increases in CSAr are significantly less than those observed in NKCC1+/+ astroctyes after 30, 40, 50, or 60 min of OGD (Fig. 5A).
We also examined whether pharmacological inhibition of NKCC1 activity blocked OGD-mediated swelling in NKCC1+/+ astrocytes. After 1 hr of OGD, the CSAr of NKCC1+/+ astrocytes increased by 47%. In the presence of bumetanide (5 μm), the increase in CSAr was 14% after 1 hr of OGD, similar to that of NKCC1-/- astrocytes (Fig. 5B).
To validate the DIC method as an accurate reflection of cell volume changes, we used a fluorescence-based calcein method. NKCC1+/+ astrocytes were exposed to isotonic or hypotonic buffers, and cell swelling was monitored using both methods. As shown in Figure 5C, changes in relative cell volume correlated with the reciprocal of the relative osmotic pressure using either method (DIC, r2 = 0.98; calcein, r2 = 0.99). However, the slope of the DIC fit (0.64) was somewhat shallower than that of the calcein fit (0.88). This implies that the calcein method exhibits more ideal osmometric behavior and is likely to be more sensitive. The y-intercept corresponded to the fraction of the calcein that is insensitive to changes in external osmolarity.
Figure 5D shows that similar cell swelling was detected with the calcein method. During control perfusion, relative cell volume in NKCC1+/+ astrocytes did not change (Fig. 5D). OGD caused a rapid increase in relative cell volume that reached levels similar to those measured with the DIC method. In NKCC1+/+ cells treated with bumetanide, the increase in relative cell volume during OGD was greatly reduced (Fig. 5D). Together, the results with the calcein method confirm our conclusion that NKCC1 has an important role in OGD-mediated cell swelling.
OGD/REOX did not trigger significant changes in bulk [Ca2+]i in astrocytes
Accumulation of intracellular Na+ may increase [Ca2+]i via reverse-mode operation of NCX in astrocytes (Smith et al., 2000). To test this hypothesis, we first measured [Ca2+]i of NKCC1+/+ and NKCC1-/- astrocytes under normoxic control conditions, 2 hr of OGD, and OGD/REOX. As shown in Figure 6A, similar basal [Ca2+]i levels were found between NKCC1+/+ and NKCC1-/- astrocytes under control conditions. In addition, neither OGD nor OGD/REOX caused a significant increase in [Ca2+]i in NKCC1+/+ or NKCC1-/- astrocytes.
Intracellular Ca2+ stores in astrocytes are a key component in astrocyte Ca2+ signaling (Smith et al., 2003). Therefore, it is possible that Ca2+ entry after OGD was buffered by an increase in Ca2+ uptake into intracellular Ca2+ stores. To test this possibility, we examined whether inhibition of Ca2+ uptake by an irreversible inhibitor of the endoplasmic reticulum (ER) Ca2+-ATPase, thapsigargin, would reveal a rise in [Ca2+]i. As shown in Figure 6B, in the presence of 1.2 mm [Ca2+]o and thapsigargin, a small elevation of [Ca2+]i was detected in control NKCC1+/+ astrocytes. Two hours of OGD and 1 hr of REOX triggered a twofold increase in [Ca2+]i (Fig. 6B, inset). In contrast, inhibition of NKCC1 activity with bumetanide significantly attenuated the Ca2+ rise (Fig. 6B, inset).
We speculated that our OGD/REOX protocol causes a mild insult to astrocytes, which was reflected in the lack of change in bulk [Ca2+]i. To verify this, cell death assay was performed. Cell death in NKCC1+/+ astrocytes after 2 hr of OGD or 2 hr of OGD plus 1 hr of REOX was negligible (Fig. 6C). Eight hours of OGD led to ∼20% cell death (p = 0.10). To evaluate further the severity of the OGD insult, intracellular ATP content was measured after OGD/REOX and 8 hr of OGD. As shown in Figure 6C (inset), ATP content in control cells was 7.02 ± 0.90 nmol of ATP/mg of protein. After 2 hr of OGD, intracellular ATP was 4.71 ± 1.3 nmol of ATP/mg of protein but not significantly different from control levels. Eight hours of OGD caused a significant decrease in ATP levels (2.95 ± 0.60 nmol of ATP/mg of protein; p < 0.05).
Inhibition of NKCC1 activity reduced Ca2+ uptake mediated by intracellular Ca2+ stores
We investigated further Ca2+ release from Ca2+ stores in control and OGD/REOX-treated cells in the absence of extracellular Ca2+. As shown in Figure 7A, 1.0 μm bradykinin triggered Ca2+ release from intracellular Ca2+ stores in NKCC1+/+ astrocytes and raised [Ca2+]i from a baseline of 70 ± 6 to 402 ± 38 nm under normoxic conditions. However, after 2 hr of OGD and 1 hr of REOX, bradykinin-induced Ca2+ release increased to 714 ± 34 nm (p < 0.05) (Fig. 7A). In NKCC1-/- astrocytes, bradykinin evoked a Ca2+ release under normoxic control conditions that was similar to that of NKCC1+/+ cells (465 ± 50 nm) (Fig. 7C). In contrast, bradykinin-induced Ca2+ release in OGD-treated NKCC1-/- astrocytes (506 ± 89 nm) was not significantly different from that of control NKCC1-/- cells (Fig. 7C) but was significantly lower than in OGD-treated NKCC1+/+ astrocytes. This suggests that OGD caused an increase in Ca2+ uptake by the intracellular Ca2+ stores in NKCC1+/+ astrocytes but not in NKCC1-/- astrocytes.
As shown in Figure 7, B and D, a small fraction of Ca2+ release was insensitive to bradykinin but was triggered by thapsigargin. In contrast, when Ca2+ release was first evoked by 1.0 μm thapsigargin, bradykinin did not induce any additional Ca2+ release (data not shown). These results imply that astrocytes have both bradykinin-sensitive and -insensitive Ca2+ stores. Interestingly, NKCC1-/- astrocytes also exhibited significantly less Ca2+ release from bradykinin-insensitive Ca2+ stores than in NKCC1+/+ astrocytes (p < 0.05) (Fig. 7B,D). To confirm further that a lack of Ca2+ uptake by the intracellular Ca2+ stores in NKCC1-/- astrocytes is indeed attributable to ablation of NKCC1 activity, we performed similar experiments with bumetanide. As shown in Figure 8A, 2 hr of OGD and 1 hr of REOX led to an enhanced filling of the intracellular bradykinin-sensitive Ca2+ stores in NKCC1+/+ astrocytes (714 ± 34 nm; p < 0.05). OGD/REOX treatment also resulted in a significant increase in the second Ca2+ release by thapsigargin (from 88 ± 9 to 156 ± 20 nm; p < 0.05) (Fig. 8A). In contrast, treatment of NKCC1+/+ astrocytes with bumetanide abolished the OGD-induced increase in Ca2+ transients mediated by either bradykinin or thapsigargin (Fig. 8A,B). This is consistent with the observations in NKCC1-/- astrocytes.
Mitochondrial Ca2+ content in NKCC1+/+ astrocytes was moderately increased after OGD
Our data indicate that bradykinin-sensitive Ca2+ stores contain higher levels of Ca2+ after OGD. However, it is also possible that an increase in bradykinin-mediated Ca2+ release reflects a decrease in the efficiency of cytosolic Ca2+ clearance. Mitochondria are known to buffer increases in cytosolic Ca2+ after physiological and pathological stimuli. We therefore investigated whether Ca2+ loading in mitochondria of NKCC1+/+ astrocytes is affected during OGD. NKCC1+/+ astrocytes were exposed to 10 μm FCCP and 2.5 μg/ml oligomycin under Ca2+-free conditions, after either 3 hr of normoxia (Fig. 9A) or 2 hr of OGD and 1 hr of REOX (Fig. 9B). After a 10-20 sec delay, [Ca2+]i increased in response to FCCP/oligomycin in both normoxic and OGD/REOX-treated astrocytes (Fig. 9A,B), with a peak increase in [Ca2+]i of 166 ± 4 nm under normoxic conditions and 261 ± 15 nm under OGD/REOX (p < 0.05) (Fig. 9C). The area under the Ca2+ transient was also determined in terms of arbitrary units (AU; Ca2+ concentration × minutes) and significantly greater in OGD/REOX-treated astrocytes (177 ± 15 AU) than in normoxic controls (84 ± 13 AU; n = 4; p < 0.01). This suggests that mitochondrial buffering of Ca2+ is only moderately increased after OGD/REOX and has not yet reached its capacity (250-1000 nm) (Babcock et al., 1997). On the other hand, the increases in Ca2+ release induced by OGD/REOX were three times greater from bradykinin-sensitive stores than from mitochondria (Figs. 8A, 9C). This implies that the intracellular stores are more active in buffering Ca2+ overload after OGD/REOX than mitochondria in the current study.
We examined further bradykinin-induced Ca2+ release from intracellular stores when mitochondrial Ca2+ buffering ability is blocked by FCCP-oligomycin. However, as shown in Figure 9, A and B, bradykinin-mediated Ca2+ release was significantly reduced in the presence of FCCP-oligomycin in both normoxia- and OGD/REOX-treated NKCC1+/+ astrocytes. This is consistent with a report that inhibition of mitochondrial ATP production affects Ca2+ loading in astrocyte Ca2+ stores (Kahlert and Reiser, 2000). Thus, we were unable to characterize bradykinin-induced Ca2+ release in the absence of mitochondrial Ca2+ buffering function.
Inhibition of the reverse-mode operation of NCX abolished OGD-induced increase in Ca2+ transients
An increase in bradykinin-evoked Ca2+ release after OGD/REOX may reflect a stimulation of Ca2+ entry via the reverse-mode activity of NCX as a result of accumulation of intracellular Na+. To investigate further the role of NCX in the OGD/REOX-induced increase in Ca2+ store filling, we used KB-R7943, an inhibitor of the reverse-mode operation of NCX activity. First, we established that KB-R7943 blocked the reverse-mode operation of NCX in a dose-dependent manner (Fig. 10A-C). The reverse-mode operation of NCX was induced by exposing cells to a Na+- and Ca2+-free buffer to reduce [Na+]i and subsequently to an Na+-free buffer containing Ca2+ (Fig. 10A, a,b). The reverse-mode operation of NCX triggered an abrupt increase in [Ca2+]i. After returning to HEPES-MEM, [Ca2+]i quickly recovered to baseline values. The reverse-mode operation of NCX can be stimulated repeatedly (Fig. 10A). However, when the cell was exposed to 10 μm KB-R7943, the reverse-mode operation of NCX was inhibited (Fig. 10A).
The dose-dependent effect of KB-R7943 on the reverse-mode operation of NCX was determined (Fig. 10B). A plot of percentage of inhibition versus KB-R7943 concentration was fit with a hyperbolic equation that yielded an IC50 value of 0.88 μm (Fig. 10C). This is similar to the IC50 values reported previously for Na+-dependent Ca2+ influx inhibition by KB-R7943: 0.32 μm in guinea pig cardiac ventricular cells (Watano et al., 1996), 1.6 μm for cardiac NCX1-transfected CCL39 cells (Iwamoto et al., 1996), and 2.4 μm for rat cardiomyocytes (Iwamoto et al., 1996).
Subsequently, we investigated whether inhibition of NCX with KB-R7943 would reduce Ca2+ entry and subsequently Ca2+ uptake via Ca2+ stores. Three micromolar KB-R7943 (86% inhibition of NCX) (Fig. 10C) was used in these studies because cell damage occurred with prolonged incubation with 10 μm KB-R7943 in our pilot study (data not shown). Bradykinin triggered a peak release of Ca2+ in normoxic control cells (407 ± 48 nm [Ca2+]i) (Fig. 10D). Similar to data in Figure 7, after 2 hr of OGD and 1 hr of REOX, bradykinin-evoked Ca2+ release reached 750 ± 35 nm in NKCC1+/+ astrocytes (Fig. 10D). In contrast, in the presence of 3 μm KB-R7943, bradykinin-evoked Ca2+ release was reduced by 82% during 2 hr of OGD and 1 hr of REOX and by 98% in normoxic cells, in which [Ca2+]i levels were essentially the same under baseline conditions or after treatment with bradykinin. These findings suggest that the reverse-mode operation of NCX contributes to the loading of bradykinin-sensitive Ca2+ stores in astrocytes during OGD and REOX.
Differential effects of KB-R7943 and store-operated Ca2+ entry channel inhibitor 2-APB on [Ca2+]i after OGD/REOX
In our present study, we believe that the primary mechanism underlying KB-R7943-mediated reduction of Ca2+ uptake in ER (shown above) is through inhibition of NCX. This is because these experiments were performed in the absence of extracellular Ca2+; thus, other channels such as store-operated Ca2+ entry (SOCE) were inactive.
However, it was reported that 5 μm KB-R7943 reduced SOCE by ∼30% in astrocytes (Arakawa et al., 2000). To investigate further whether 3 μm KB-R7943 would cause effects on SOCE, we compared the effects of KB-R7943 and SOCE channel inhibitor 2-APB on [Ca2+]i after 2 hr of OGD and 1 hr of REOX. When NKCC1+/+ cells were exposed to 1 μm bradykinin in a Ca2+-free buffer, there was a large transient increase in [Ca2+]i (Fig. 11A). Within 2-3 min, the increase in intracellular Ca2+ was cleared from the cell. As the cells were exposed to a 1.2 mm Ca2+ buffer, store-operated Ca2+ influx was initiated, and [Ca2+]i increased by ∼2.5-fold (<40 sec), which gradually declined over the next 2 min (Fig. 11B). In the presence of 2-APB, cells showed no significant increase in [Ca2+]i at <40 sec (Fig. 11B). The [Ca2+]i remained within normal values for baseline [Ca2+]i, although it tended to slowly increase over the next 2 min. This suggests that 2-APB abolished store-operated Ca2+ influx in astrocytes. In contrast, KB-R7943 treatment resulted in a sharp reduction in the initial rise of store-operated Ca2+ influx. However, over the next 2 min, [Ca2+]i continued to increase in the KB-R7943-treated cells and eventually reached levels similar to those in untreated cells. Thus, KB-R7943-mediated effects can be differentiated from those of 2-APB, implying that 3 μm KB-R7943 does not effectively inhibit SOCE.
Discussion
Stimulation of NKCC1 activity after OGD
In the current study, we found that NKCC1-mediated 86Rb influx was significantly increased in both NKCC1+/+ and NKCC1+/- astrocytes after OGD. In addition, OGD also caused increased levels of phosphorylated NKCC1 protein in NKCC1+/+ astrocytes, recognized by the phosphospecific Thr184/Thr189 NKCC antibody. No changes in total NKCC1 protein expression were detected in NKCC1+/+ astrocytes after OGD. Phosphorylation-dependent stimulation of NKCC1 activity has been shown in numerous experimental systems using various stimuli (Lytle and Forbush, 1992; Russell, 2000). The three threonines (Thr184, Thr189, and Thr202) on the N-terminal regulatory domain of NKCC1 modulate NKCC1 activity in species ranging from shark to human (Flemmer et al., 2002). Phosphorylation of Thr184/Thr189 correlates with NKCC1 activation in epithelial cells (Darman and Forbush, 2002). Such a correlation has also been found in cultured cortical neurons when intracellular Cl- is reduced (Schomberg et al., 2003). Together, our study demonstrates that OGD induces phosphorylation of NKCC1 and stimulation of NKCC1 activity in cortical astrocytes.
The identity of the kinases responsible for the phosphorylation of NKCC1 in astrocytes after OGD is unknown. Recently, OSR1 (oxidative stress response kinase) and SPAK (STE20/SPS-1-related, proline-alanine-rich kinase), along with its rat homolog PASK (proline-alanine-rich STE20-related kinase), were found to bind to the third conserved domain in the NKCC1 N terminus and coimmunoprecipitated with NKCC1 protein (Piechotta et al., 2002; Dowd and Forbush, 2003). Overexpression of PASK in human embryonic kidney cells causes a small but significant increase in NKCC1 activity, and the phosphorylation of PASK coincides with that of NKCC1 (Dowd and Forbush, 2003).
PASK and SPAK are preferentially expressed in the brain and in cells that are active in ion transport and rich in Na+/K+-ATPase (Ushiro et al., 1998; Johnston et al., 2000; Piechotta et al., 2002). SPAK activates the p38 MAP kinase pathway and has been suggested to play a role in initiation of the cellular stress responses (Johnston et al., 2000). Although PASK and SPAK have been found in neurons, their location in astrocytes has not been reported. However, Fray, a Drosophila serine-threonine kinase that is homologous to mammalian PASK, is expressed in peripheral glia, and its function is required for normal axonal ensheathment (Leiserson et al., 2000). Additional studies will be required to determine whether the proline-alanine-rich STE20-related kinase mediates the observed phosphorylation and stimulation of NKCC1 in astrocytes after OGD.
NKCC1-mediated Na+ entry and an effect on Ca2+ homeostasis
Pathological insults such as anoxia and ischemia are characterized by changes in ion homeostasis, including intracellular Na+ accumulation. In the current study, we found that OGD led to an ∼3.6-fold increase in [Na+]i. This finding is consistent with a report on [Na+]i accumulation in spinal cord astrocytes induced by glucose deprivation, chemical hypoxia, and simulated ischemia (Rose et al., 1998). Ischemia-mediated Na+ accumulation in astrocytes has been attributed to a reduction in Na+/K+-ATPase activity and activation of Na+ channels (Rose et al., 1998; Stys, 1998). Our data show that inhibition of NKCC1 blocked ∼80% of the Na+ loading. The remaining Na+ accumulation could be the result of a decrease in Na+/K+-ATPase activity because its activity was reduced by ∼30% in astrocytes after OGD (Fig. 2). In addition, either genetic ablation or pharmacological inhibition of NKCC1 abolished OGD-mediated Cl- influx and swelling. Thus, these data provide the first line of evidence that NKCC1 plays a pivotal role in Na+ homeostasis in cortical astrocytes after OGD.
The mechanisms of glial ischemic injury include Ca2+ entry and Ca2+-mediated cell death. The Ca2+ entry pathways involve L-type voltage-gated Ca2+ channels, AMPA and metabotropic glutamate receptors, and reverse operation of the NCX (Stys, 1998). In the current study, we detected an increase in bulk cytosolic [Ca2+] in astrocytes after OGD in the presence of thapsigargin. Moreover, Ca2+ release from bradykinin-sensitive Ca2+ stores was increased by ∼78% in NKCC1+/+ astrocytes after OGD. This implies that Ca2+ entry was elevated after OGD and Ca2+ stores subsequently buffered [Ca2+]i. Inhibition of reverse operation of NCX with 3 μm KB-R7943 abolished the OGD-mediated increase in releasable Ca2+ in bradykinin-sensitive Ca2+ stores. These findings suggest that reverse operation of NCX plays a role in Ca2+ entry and Ca2+ signaling in cortical astrocytes after OGD.
It is generally accepted that inhibition of the Na+/K+-ATPase and reduction of the [Na+] gradient across the plasma membrane promote Ca2+ entry via NCX in most types of cells (Blaustein and Lederer, 1999). Recently, it has been suggested that NCX and the Na+/K+-ATPase (α2 or α3 subunits) are confined to the same plasma membrane microdomains. Thus, NCX may be regulated by Na+/K+-ATPase activity via its influence on the Na+ electrochemical gradient across the plasma membrane (Golovina et al., 2003). Reduction in Na+/K+-ATPase activity elevates local [Na+]i, which in turn activates the Ca2+ entry via NCX and enhances filling of the ER Ca2+ stores (Golovina et al., 2003). In the current study, we observed for the first time that prevention of NKCC1-mediated Na+ accumulation in astrocytes can also inhibit Ca2+ entry via NCX and subsequently decreases loading of intracellular Ca2+ stores. This suggests that the functions of NKCC1 and NCX may be closely associated, as has been proposed for NCX and the Na+/K+-ATPase. Thus, both the Na+/K+-ATPase and NKCC1 may be involved in regulating intracellular Na+ homeostasis, which in turn affects NCX activity and Ca2+ signaling in astrocytes. In addition, in a recent report (Rosker et al., 2004), activation of the TRPC3 (transient receptor potential subfamily) cation channel causes Na+ loading. The Na+ entry and membrane depolarization subsequently lead to a rise in [Ca2+]i via shifting NCX to the reverse mode. This is similar to our proposed scenario that NKCC1 mediates Na+ accumulation during OGD/REOX, which triggers reverse operation of NCX. Moreover, 5 μm KB-R7943 had no effects on the typical TRPC3-mediated inward Na+ currents (Rosker et al., 2004).
Two classes of spatially resolved ER Ca2+ stores, cyclopiazonic acid (CPA)-sensitive and caffeine-ryanodine-sensitive, have been found in rat cortical astrocytes (Golovina and Blaustein, 2000). The CPA-sensitive Ca2+ stores release Ca2+ in response to an agonist and IP3-mediated Ca2+ signaling cascade, whereas caffeine-ryanodine trigger Ca2+-induced Ca2+ entry. In the current study, we observed both bradykinin-sensitive and -insensitive Ca2+ release in mouse cortical astrocytes. This may reflect the CPA- and caffeine-ryanodine-sensitive Ca2+ stores, respectively. Interestingly, OGD/REOX treatment of NKCC1+/+ astrocytes led to an increase in Ca2+ sequestration in both Ca2+ stores. Either pharmacological inhibition or genetic ablation of NKCC1 abolished the OGD/REOX-induced increase in the Ca2+ sequestration. Thus, it is possible that NKCC1-mediated Na+ accumulation during OGD indirectly affects both Ca2+ stores.
Thermodynamic analysis of NKCC1 and NCX under ischemic conditions
We calculated the net free energy (ΔG) for NKCC1 activity, as determined by the concentration gradients of the three ions (Russell, 2000), and investigated whether changes in ΔG during OGD/REOX would allow for NKCC1 to mediate Na+ accumulation. As shown in Table 2, we calculated the ΔG for NKCC1 using the known concentrations of [Na+]o, [Na+]i, [K+]o, and [Cl-]o and assumed concentrations for [K+]i and [Cl-]i of 111 and 36 mm, respectively. Under control conditions, ΔG was quite negative, indicating that net uptake of Na+ was strongly favored (Table 2). During OGD/REOX, ΔG decreased by 28% (assuming no changes in [K+]i and [Cl-]i), and the net direction of Na+ remained strongly inward (Table 2). Furthermore, if it is assumed that during OGD/REOX, [K+]i decreased by one-half and [Cl-]i doubled, there is no effect on the calculated ΔG (Table 2). Similar results were also obtained under conditions of HAIR and HAIR/REOX, in which the known concentrations of [Na+]o, [Na+]i, [K+]o, and [Cl-]o were substituted in the calculations (Table 2).
An additional important question to consider is whether the thermodynamics of NCX during OGD/REOX would make accumulation of Ca2+ possible. Ca2+ can be transported via NCX either into or out of the cell, depending on the difference between the membrane potential and the reversal potential (ENa/Ca) of NCX (Blaustein and Lederer, 1999). ENa/Ca is defined as ENa/Ca = 3 × ENa - 2 × ECa, where ENa = (RT/F) × ln([Na+]o/[Na+]i), and ECa = (RT/2F) × ln([Ca2+]o/[Ca2+]i) (Blaustein and Lederer, 1999). If RT/F is considered to be 57 at 22°C, then ENa/Ca can be calculated as approximately -65 mV using the concentrations for [Na+]o, [Na+]i, [Ca2+]o, and [Ca2+]i in NKCC1+/+ astrocytes. Under control conditions in which the plasma membrane potential (Vm) averages about -60 mV (McKhann et al., 1997), the difference between Vm and ENa/Ca is small and there is little flux through NCX. However, Vm of cultured cortical astrocytes is highly heterogeneous (-22 to -82 mV) (McKhann et al., 1997). It can be expected that when Vm is more negative in respect to ENa/Ca, Ca2+ ions would be driven out of cells.
In the presence of the ionic concentrations under OGD/REOX conditions, ENa/Ca would become more negative (approximately -160 mV), which favors the net influx of Ca2+ at constant Vm. Although we did not measure Vm under OGD/REOX conditions, we speculate that the average Vm is likely to be depolarized (Duffy and MacVicar, 1994). The estimated influx of Ca2+ via the reversal of NCX, at depolarized Vm (-20 mV) and increased [Na+]i (∼40 mm), could be ∼1000 nm. This is similar to our measured [Ca2+]i values under OGD/REOX conditions. Together, from a thermodynamic point of view, there is no reason to a priori rule out reversal of NCX during OGD/REOX.
Footnotes
This work was supported in part by National Institutes of Health (NIH) Grant R01NS38118, National Science Foundation CAREER Award IBN998 1826 (D.S.), and NIH Grant R01HL61974 (G.E.S.). We thank Dr. Robert Haworth for helpful discussion.
Correspondence should be addressed to Dr. Dandan Sun, Department of Neurological Surgery, University of Wisconsin Medical School, H4/332 Clinical Sciences Center, 600 Highland Avenue, Madison, WI 53792. E-mail: sun{at}neurosurg.wisc.edu.
Copyright © 2004 Society for Neuroscience 0270-6474/04/249585-13$15.00/0