Stimulated Efflux of Amino Acids and Glutathione from Cultured Hippocampal Slices by Omission of Extracellular Calcium

Abstract

Omission of extracellular Ca²⁺ for 15 min from the incubation medium of cultured hippocampal slices stimulated the efflux of glutathione, phosphoethanolamine, hypotaurine, and taurine. The efflux was reduced by several blockers of gap junctions, i.e. carbenoxolone, flufenamic acid, and endothelin-1, and by the connexin43 hemichannel blocking peptide Gap26 but was unchanged by the P2X₇ receptor inhibitor oxidized ATP, a pannexin1 hemichannel blocking peptide and an inactive analogue of carbenoxolone. Pretreatment of the slices with the neurotoxin N-methyl-d -aspartate left the efflux by Ca²⁺ omission unchanged, indicating that the stimulated efflux primarily originated from glia. Elevated glutamate efflux was detected when Ca²⁺ omission was combined with the glutamate uptake blocker l-trans-pyrrolidine-2,4-dicarboxylate and when both Ca²⁺ and Mg²⁺ were omitted from the medium. Omission of Ca²⁺ for 15 min alone did not induce delayed toxicity, but in combination with blocked glutamate uptake, significant cell death was observed 24 h later. Our results indicate that omission of extracellular Ca²⁺ stimulates efflux of glutathione and specific amino acids including glutamate via opening of glial hemichannels. This type of efflux may have protective functions via glutathione efflux but can aggravate toxicity in situations when glutamate reuptake is impaired, such as following a stroke.

Chemical communication between brain cells occurs either through release and diffusion of substances, for example neurotransmitters, via the extracellular space or directly through intercellular gap junctions. It has become clear that types of efflux other than release by conventional exocytosis also contribute to the composition of the extracellular fluid. Because the major cell type in brain is astroglia, it follows that efflux from this cell type may have strong influence on the extracellular neurochemistry and thus on various aspects of brain function. During the last two decades, it has been shown that efflux from cultured glial cells can be achieved via various channels and processes: volume-regulated anion channels, exocytosis, reversal of uptake carriers, ionotropic purinergic receptors, and also half-gap junction channels named hemichannels or connexons, the latter thoroughly reviewed by Bennett et al. (1) and Spray et al. (2). The most recent player in this respect is “pannexons,” consisting of proteins termed pannexins (3).

In an earlier study we showed selective efflux of GSH, taurine, and phosphoethanolamine (PEA)4 from organotypic hippocampal slice cultures when Ca²⁺ was omitted from the incubation medium (4). The mechanisms behind the efflux were, however, not investigated. Because omission of Ca²⁺ is a conventional protocol to activate hemichannel opening from astrocyte cultures and excised optic nerve (5, 6), we have in this study further characterized efflux stimulated by omission of extracellular Ca²⁺ by using different means of selecting between various efflux possibilities. Principally, the study of hemichannels (or any other channel that is activated by omission of extracellular Ca²⁺) is complicated by the finding that other plasma membrane channels/pores, including P2X₇ receptor channels and various anion channels, may open parallel to or possibly as a result of hemichannel opening. For example the efflux of ATP and glutamate by hemichannels may activate purine and glutamate receptors. Some of these, such as the P2X₇ receptor, are ionotropic receptors that can mediate passage of molecules such as glutamate and aspartate (7), possibly by forming complexes with pannexin hemichannels (3, 8, 9). In addition, P2X₇ receptors and connexin hemichannels share the characteristics of increased opening in medium with low concentrations of divalent cations (10, 11). Overactivation of glutamate receptors can also cause indirect opening of large pore channels. This may occur via influx of inorganic ions followed by water influx and volume changes that open volume-sensitive anion channels (12). In addition, increased intracellular Ca²⁺ can activate opening of channels that mediate efflux of anions from neurons (13). Adding to the complexity in studying these channels is that some of the antagonists used in studying hemichannels, P2X₇ receptors and anion channels show cross-reactivity (14, 15). In this study we investigated the stimulated efflux of GSH and amino acids following omission of Ca²⁺. Various types of antagonists and hyperosmotic medium were used in an attempt to estimate the involvement of hemichannels, P2X₇ channels, and volume-sensitive channels, respectively. The selectivity of the efflux was investigated by analysis of changes in the concentrations of GSH and a large number of amino acids in the incubation medium from cultured hippocampal slices. Determination of the neuronal amino acid N-acetylaspartate in the incubation medium and the use of slices pretreated with high concentrations of the neurotoxin N-methyl-d -aspartate (NMDA) allowed us to evaluate to what extent neurons contributed to the efflux. In short, the results show that omission of extracellular Ca²⁺ for 15 min is nontoxic within the time period studied (24 h) and causes efflux of GSH, PEA, hypotaurine, and taurine. The efflux was stimulated further by omission of Mg²⁺ but was still nontoxic even if this regimen also caused efflux of glutamate. However, in combination with blocked glutamate uptake, omission of Ca²⁺ caused significant nerve cell death.

EXPERIMENTAL PROCEDURES

Materials—Gey's balanced salt solution, Basal medium Eagle, Earl's basal salt solution, NMDA, β-mercaptoethanol, horse serum, penicillin/streptomycin, l -glutamine, d -glucose, propidium iodide (PI), and o-phthaldialdehyde (OPA) were purchased from Sigma. The culture medium included Basal medium Eagle and Earl's basal salt solution (50 and 20%, respectively), horse serum (23%), penicillin/streptomycin (25 units/ml), l -glutamine (1 mm), and d -glucose (41.6 mm). The artificial cerebrospinal fluid (ACSF) consisted of 128 mm NaCl, 3 mm KCl, 2 mm CaCl₂, 1.2 mm MgSO₄, 0.4 mm KH₂PO₄, 25 mm NaHCO₃, and 10 mm d-glucose, and Gey's balanced salt solution consisted of 119 mm NaCl, 5 mm KCl, 2 mm CaCl₂, 1 mm MgCl₂, 0.3 mm MgSO₄, 0.2 mm KH₂PO₄, 27 mm NaHCO₃, 0.67 mm Na₂HPO₄, and 5.5 mm d-glucose. The solutions were equilibrated with gas mixtures (see further below) containing 5% CO₂, which resulted in a pH of ∼7.4. All of the salts in ACSF were from Sigma or Merck, N-acetylaspartate was from Aldrich, and methanol was from Rathburn (Rathburn Chemicals Ltd., Walkersburn, UK). Carbenoxolone (CBX), flufenamic acid (FFA), glycyrrhizic acid, oxidized ATP (ox-ATP), and l -trans-pyrrolidine-2,4-dicarboxylic acid (PDC) were all bought from Sigma. d l-threo-β-Benzyloxyaspartic acid (TBOA) was obtained from Tocris Cookson (Bristol, UK), and endothelin-1 was from Neosystem (Illkirch, France). The connexin43 mimetic and blocking peptide (Gap26) (VCYDKSFPISHVR) and the pannexin1 mimetic blocking peptide (WRQAAFVDSY), ¹⁰Panx1, were synthesized by solid phase chemistry and purified by HPLC to 95% purity (Sigma-Genosys) (9, 16). FFA, endothelin-1, and ¹⁰Panx1 were dissolved in DMSO and diluted in ACSF and ACSF/0Ca²⁺ to a final DMSO concentration of 0.1%.

Organotypic Hippocampal Cultures—Organotypic hippocampal slice cultures were prepared according to Stoppini et al. (17). In brief, hippocampi of 8–9-day-old Sprague-Dawley rat pups were dissected and cut in 400-μm-thick slices using a McIlwain tissue chopper. The slices were transferred to a Petri dish containing Gey's balanced salt solution with 0.45 g/liter of d -glucose. Four slices were put on a porous membrane insert (Millicell CM, Bedford, MA) in 6-well plates with 1.3 ml of culture medium. The slices were cultured for 12–14 days at 36 °C in a humidified atmosphere containing 5% CO₂ and 95% air. 1.2 ml of culture medium was changed twice a week. Slice cultures with a low number of neurons were prepared by incubating slice cultures with 300 μm NMDA for 24 h 3–4 days prior to efflux experiments. All of the animal procedures were approved of by the local ethical committee at Göteborg University.

Efflux Experiments and Cell Death Analysis—The efflux experiments and cell death estimations were carried out as described previously (13). In brief, we used PI uptake, which correlates well with other markers of cell death (18). The slices were incubated in culture medium containing PI (2 μm) 24 h before efflux experiments, and the incubation continued for 24 h following the experiment. Photographs of the slices were taken before and 24 h after the efflux experiments with a digital camera (Olympus DP50) coupled to an inverted fluorescence microscope (Olympus IX70) equipped with a rhodamine filter. The photographs were captured using Studio Lite and View Finder Lite softwares (Pixera Corporation, Los Gatos). The efflux experiments were carried out by transferring the inserts with the slices to a 6-well plate kept in a water bath set at 36 °C (for details see (13)). The atmosphere inside the plate was kept at 60% O₂, 35% N₂, and 5% CO₂ by directing a flow of gas into a water filled container inside the plate and performing the incubation with the lid on. All of the solutions were equilibrated with a gas mixture of 60% O₂, 35% N₂, and 5% CO₂ (19). The efflux experiments were carried out by incubating the slices with ACSF (400 μl) on top of the membrane for 5 min. The fluid was then removed and filtered before immediate HPLC analysis or storage in –20 °C (maximally 2 weeks). This incubation procedure was repeated nine times (45 min in total) with Ca²⁺ omission during the 5th, 6th, and 7th incubation periods (20–35 min). All of the inhibitors were present during the 30-min preincubation period and the whole incubation period (50 min in total before Ca²⁺ removal). After the experiments, the slices were cultured in culture medium with added PI. After 24 h, the slices were photographed again.

To calculate cell death in the slices, the photographs were converted to grayscale. Then the CA1 and CA3 areas and part of the background (∼10% of total) were encircled, and the fluorescence intensity of each area was measured by Scion Image software (Scion Corporation, Frederick, MI). Before further calculation the background level in each photograph was subtracted. The fluorescence intensities obtained in slices before the efflux experiments were subtracted before calculation as described earlier (13). The fluorescence intensity measured 24 h after adding 300 μm NMDA to the culture medium was, as described earlier, used as a value of maximal nerve cell death (20). Histologic degeneration has earlier been shown to be limited to neurons 24 h after NMDA exposure and consistent with the PI staining (21). The fluorescence intensity in the incubated slices above that of controls (i.e. nonincubated slices) was expressed as the percentage of maximum fluorescence intensity.

HPLC Determination of Amino Acids and GSH—All chromatography was performed using a Varian 5000 or 5500 HPLC pump coupled to a UV detector (Applied Biosystems 757) for absorbance measurements at 210 nm of the amide bond of N-acetylaspartate or a fluorescence detector (Schoeffel FS 970) for detection of the fluorescence intensities of OPA-derivatives of GSH and amino acid (see below for details). The data were processed with Millennium and Maxima software (Waters Corporation, Milford, MI). All of the separations were performed at room temperature. Sample injection was made using a Waters 717 autosampler or a CMA 200/240 auto-injection system controlled by CMA software (Solna, Stockholm).

Separation of N-acetylaspartate was performed as described earlier (22) using a TSK-GEL ODS-80T_M (250 × 4.6 mm; Tosoh, Tokyo, Japan). The mobile phase consisted of 50 mm NaH₂PO₄ (pH 2.15) with a flow rate of 1 ml/min. Prior to injection the samples were mixed with HCl (0.2 m) in a 6:1 ratio (sample/HCl). Sample injection volume was 90 μl. The buffers were sonicated and degassed with N₂ before use.

GSH and amino acids were determined using OPA derivatization and fluorescence detection essentially as described earlier (23, 24). A solution of β-mercaptoethanol, Na₂-EDTA, and NaN₃ (final concentrations, 20, 1, and 5 mm, respectively) was added to the samples and standards to keep GSH in its reduced form and prevent bacterial growth. The OPA solution was prepared weekly and consisted of OPA (40 mg) dissolved in methanol (400 μl), β-mercaptoethanol (40 μl), borate buffer (2.0 ml, 0.8 m, pH 12), and H₂O (1.6 ml). Every 2 days β-mercaptoethanol (10 μl) was added to the solution. Amino acids were derivatized (25 μl of sample mixed with 25 μl OPA solution) in the autosampler before injection. The amino acid derivatives were separated on a Nucleosil C₁₈ column (200 × 4.6 mm; Macherey-Nagel, Germany) with a mobile phase consisting of NaH₂PO₄ (50 mm, pH 5.28) and methanol in a gradient from 25–95% methanol. A flow rate of 1 ml/min was used. Detection was carried out by excitation at 333 nm and emission over 418 nm.

To test for possible interactions with GSH, standard solutions with or without drugs were analyzed. No differences in the peak heights of GSH were found with any of the drugs used.

Slice Homogenization and Protein Determination—After the experiments, the slices were washed twice with 0.9% NaCl solution, both on top and under the membrane, and then immediately frozen in liquid nitrogen. The four slices were then quickly cut out from the membrane and put in a precooled 1.5-ml microcentrifuge tube and replaced in liquid nitrogen. Ice-cold HClO₄ (250 μl, 0.3 m) was added to the tubes, and the tissue was sonicated (Branson Sonifier 250; Branson, Danbury, CT). After storage at 4 °C overnight, the tubes were centrifuged at 11,000 × g, and the supernatant was removed and filtered (Acrodisc, 0.2 μm; Pall Corporation, Ann Arbor, MI). The supernatant was used to determine the slice content of amino acids, and the remaining pellet was dissolved in 250 μl of NaOH (2 m) overnight for protein determination (25).

Statistics—Statistical analysis was performed with SPSS software (SPSS Inc., Chicago, IL). One-way analysis of variance followed by Tukey's test for multiple comparisons was used for statistical analysis. A p value <0.05 was considered statistically significant. The data shown in figures are expressed as the means ± S.E.

RESULTS

Omission of Extracellular Ca²⁺ Changed the Efflux Pattern: Marked Effects for GSH—Incubation of organotypic hippocampal slices for 15 min with ACSF without Ca²⁺ (ACSF/0Ca²⁺) increased the efflux of GSH, taurine, hypotaurine, and PEA (Fig. 1). The effect in relation to base-line efflux was most pronounced for GSH, showing an elevation from 19.0 ± 2.1 pmol/min/mg protein at base-line efflux to 74.8 ± 6.4 pmol/min/mg protein at the end of the 15-min period of Ca²⁺ omission, which corresponds to a relative increase of ∼300% above control (Fig. 2). The maximal increases after 15 min of Ca²⁺ omission for other substances that showed elevated efflux were: PEA, 17.2 ± 2.0 to 36.0 ± 6.1 pmol/min/mg protein (∼110%); taurine, 120.5 ± 9.3 to 194.7 ± 21.4 pmol/min/mg protein (∼60%); and hypotaurine, 110.9 ± 11.1 to 177.0 ± 22.2 pmol/min/mg protein (∼60%). The efflux rates of glutamate and valine were unchanged (Fig. 2).

FIGURE 1.

HPLC chromatogram showing the pattern of amino acid efflux in ACSF (solid line) and in ACSF/0Ca²⁺ (dashed line). Removing Ca²⁺ induced a selective efflux of GSH, PEA, taurine (Tau) and hypotaurine (Hypotau), whereas the levels of glutamate (Glu) and valine (Val) among others remained unaffected. Imp marks a peak caused by an impurity in the reagent.

FIGURE 2.

a, time course of efflux of GSH, PEA, and Glu caused by removal of Ca²⁺ showing an elevated efflux of GSH and PEA, but not of Glu. The efflux rates reached their maxima 10 min after the introduction of ACSF/0Ca²⁺. b, time course of efflux for taurine, Val, and hypotaurine showing an increased efflux for taurine and hypotaurine, but not for Val. The data are presented as the mean efflux rates ± S.E. (n = 17 for GSH, PEA, taurine, Val, and hypotaurine and n = 6 for Glu). The symbols in a and b indicate a significant difference compared with incubation in ACSF (10 min) (p < 0.05).

Omission of both Mg²⁺ and Ca²⁺ further increased the efflux of GSH and amino acids (Fig. 3). The increased efflux rates in pmol/min/mg protein were: for GSH, 19.0 ± 2.1 to 534.2 ± 74.6 (∼2700%); for PEA, 17.2 ± 2.0 to 161.8 ± 19.7 (∼850%); for taurine, 25.6 ± 9.3 to 1069.6 ± 204.8 (∼4000%); and for hypotaurine, 155.3 ± 11.4 to 1814.3 ± 184.4 (∼1100%). In contrast to the lack of effect of Ca²⁺ removal on glutamate efflux, omission of both Mg²⁺ and Ca²⁺ caused an increased glutamate efflux rate, from 15.7 ± 4.8 to 99.4 ± 17.9 pmol/min/mg protein (∼530%).

FIGURE 3.

a, removal of both Mg²⁺ and Ca²⁺ caused an efflux of Glu not observed by Ca²⁺ omission only. The efflux of GSH and PEA was further increased by omission of both Mg²⁺ and Ca²⁺. b, efflux of taurine and hypotaurine were also increased further, whereas that of Val remained unaffected. The data are presented as the mean efflux rates ± S.E. (n = 4). The symbols in a and b indicate a significant difference compared with incubation in ACSF (10 min) (p < 0.05).

Incubating the slices in a hypertonic solution (addition of 240 mm sucrose to the ACSF and ACSF/0Ca²⁺) did not reduce the effect of Ca²⁺ removal on GSH efflux (Fig. 4) or the basal efflux levels (not shown). Incubation in ACSF where Ca²⁺ was replaced with Mg²⁺ did not cause stimulated efflux of GSH (Fig. 4). Similarly, incubation of slices in medium containing 0.1 mm Ca²⁺ induced no elevated efflux of GSH, PEA, taurine, or hypotaurine. Removal of only Mg²⁺ also failed to generate an increased efflux of GSH (Fig. 4).

FIGURE 4.

Removal of Mg²⁺ in addition to Ca²⁺ further increased the efflux of GSH, whereas replacing Ca²⁺ with Mg²⁺ abolished the effect of Ca²⁺ omission. Incubation in ACSF/0Mg²⁺ generated no efflux of GSH, and neither did incubation in ACSF containing 0.1 mm Ca²⁺. Incubation in hyperosmotic ACSF (240 mm sucrose in both normal and ACSF/0Ca²⁺) increased the efflux of GSH. The data are presented as the mean efflux rates (n = 6 ± S.E.) during treatment (at 25 min, see Fig. 2). Asterisks indicate a significant difference (p < 0.05) compared with incubation in ACSF.

Gap Junction Blockers Inhibited the Efflux of GSH Caused by Ca²⁺ Omission—Incubation of the slices with gap junction blocker (starting from the preincubation, i.e. present 50 min before Ca²⁺ omission) CBX (100 μm) or FFA (200 μm) blocked the elevated efflux induced by incubation in ACSF without Ca²⁺ (Fig. 5). The structural analogue of CBX, glycyrrhizic acid (100 μm), did not inhibit the efflux. The specificities of several gap junction blockers have been questioned (14, 15). Therefore we also used the structurally and functionally nonrelated gap junction blocker endothelin-1. Although CBX and FFA likely act directly on the connexins or the membrane protein interface (26–28), endothelin-1 most likely blocks astrocyte gap junction coupling through receptor mediated intracellular signal transduction (29, 30). Endothelin-1 (1 μm) inhibited the efflux induced by Ca²⁺ removal to the same extent as the more conventional gap junction blockers CBX and FFA (Fig. 5b). In addition to blocking the stimulated efflux caused by Ca²⁺ omission, CBX and endothelin-1 also significantly reduced the basal efflux of GSH (Fig. 5a). The connexin43 mimetic peptide Gap26 (300 μm) reduced the stimulated efflux considerably, whereas the pannexin1 mimetic peptide ¹⁰Panx1 (300 μm) was ineffective (Fig. 5, c and d). The efflux of GSH in ACSF and ACSF/0Ca²⁺ was unchanged by the addition of 0.1% DMSO. The efflux values in ACSF + 0.1% DMSO was 105% ± 12 (n = 6) compared with efflux in ACSF. Efflux in ACSF/0Ca²⁺ with added 0.1% DMSO was 114% ± 14 (n = 6) compared with efflux in ACSF/0Ca²⁺.

FIGURE 5.

a, the basal efflux of GSH in ACSF is reduced by the gap junction inhibitors CBX and endothelin-1. The data are presented as the mean efflux rates (n = 6 ± S.E.). Asterisks indicate a significant difference (p < 0.05) compared with ACSF. b, inhibitors of gap junctions significantly reduced the efflux of GSH caused by omission of Ca²⁺, whereas the inactive structural analogue of CBX, glycyrrhizic acid, was ineffective. The data are presented as the mean efflux rates (n = 6 ± S.E.). Asterisks indicate a significant difference (p < 0.05) compared with ACSF/0Ca²⁺ treatment without gap junction inhibitors. c, incubation of the slices with either connexin (Gap26) or pannexin (¹⁰Panx1) mimicking peptides did not significantly alter base-line efflux of GSH. d, Gap26 inhibited efflux of GSH in ACSF/0Ca²⁺, but ¹⁰Panx1 had no effect. The data are presented as percent of basal GSH efflux (n = 6 ± S.E.). Asterisks indicate a significant difference (p < 0.05) compared with incubation in ACSF/0Ca²⁺ without mimicking peptide.

The Efflux Caused by Ca²⁺ Omission Was Not Decreased by Inhibitors of P2X₇ Receptors—Some gap junction blockers inhibit efflux of ATP following stimulation of P2X₇ receptors (15), possibly via interaction with pannexin hemichannels or a process that stimulate pannexin hemichannel opening (3). We therefore evaluated the effect of the P2X₇ receptor antagonist ox-ATP (300 μm) in combination with 15 min of Ca²⁺ removal. Blocking P2X₇ receptors with ox-ATP did not reduce, but in fact increased, the efflux of GSH following Ca²⁺ omission (Fig. 6). Activating the P2X₇ receptors with 100 μm of the agonist BzATP generated a slight increase in the efflux of GSH that was comparable with the increase caused by application of ox-ATP.

FIGURE 6.

Both the P2X₇ receptor agonist BzATP (100 μm) and the antagonist ox-ATP (300 μm) generated a slight increase in the GSH efflux induced by a 15-min period of Ca²⁺-free ACSF. The data are presented as the mean efflux rates (n = 4 ± S.E.) during incubation in ACSF (a) and in ACSF/0Ca²⁺ (b). Asterisks indicate a significant difference (p < 0.05) compared with Ca²⁺-free ACSF treatment.

GSH Efflux Caused by Ca²⁺ Omission Was Not Changed by Neuronal Loss—The cellular origin of the efflux was addressed using cultured slices in which the neurons were degenerated by adding 300 μm NMDA to the culture medium for 24 h, 3–4 days prior to the experiment (Fig. 7). Extensive neurodegeneration was confirmed by PI uptake and HPLC-based analysis (see “Experimental Procedures” for more details) of the more or less selective neuronal amino acids γ-aminobutyric acid and N-acetylaspartate, which were reduced by 83 and 87%, respectively, after the NMDA treatment (data not shown, n = 6). Incubation of these slices for 15 min in ACSF/0 Ca²⁺ did not change the efflux of GSH compared with non-NMDA-treated slices. The efflux induced by Ca²⁺ omission in NMDA-treated slices was completely blocked by CBX (100 μm; Fig. 7).

FIGURE 7.

The efflux of GSH induced by ACSF/0Ca²⁺ was not decreased in cultured slices that were pretreated with high concentrations of NMDA before the experiment. The stimulated efflux in both NMDA-treated and nontreated slices were blocked by 100 μm CBX. The basal efflux was reduced by CBX in the nontreated slices. The data are presented as the mean efflux rates (n = 6 ± S.E.). Asterisks indicate a significant difference (p < 0.05) compared with ACSF (a) and to treatment with ACSF/0Ca²⁺ (b).

Increased Extracellular Glutamate and Delayed Toxicity by Ca²⁺ Omission in Combination with Blocked Glutamate Uptake—Omission of both Ca²⁺ and Mg²⁺ increased glutamate efflux by ∼530% (Fig. 3a), whereas omission of only Ca²⁺ had no effect (Fig. 2a). To evaluate whether glutamate was released by Ca²⁺ omission but taken up by glutamate transporters we used PDC and TBOA, where the former is an uptake inhibitor that is transported by glutamate carriers and the latter is a nontransportable inhibitor. The efflux of glutamate was raised from 20.6 ± 3.9 pmol/min/protein at base-line levels to maximally 102.8 ± 20.8 (∼400%) after Ca²⁺ removal in combination with PDC (Fig. 8a) and from 19.2 ± 2.1 to maximally 142.4 ± 15.7 (∼640%) pmol/min/mg protein when 100 μm TBOA was combined with Ca²⁺/Mg²⁺ omission (Fig. 8b). The effects of PDC and TBOA differed considerably because PDC increased glutamate in combination with Ca²⁺ omission (Fig. 8a), whereas TBOA had no effect (data not shown) unless Mg²⁺ was also withdrawn (Fig. 8b).

FIGURE 8.

a, incubation with the Glu uptake blocker PDC (100 μm) increased Glu efflux both during and after Ca²⁺ removal. Also GSH efflux was elevated by PDC after ACSF/0Ca²⁺ treatment in combination with PDC. The data are presented as the mean efflux rates (n = 4 ± S.E.). The symbols indicate a significant difference (p < 0.05) comparing values with and without PDC treatment. b, the nontransportable Glu uptake blocker TBOA increased Glu efflux only when combined with omission of both Mg²⁺ and Ca²⁺ and only in two fractions. The efflux of GSH was not affected by TBOA. Asterisks indicate a significant difference (p < 0.05) comparing values with and without TBOA treatment. Note the difference in scale between panels a and b. c, the combined treatment of PDC and 15 min of ACSF/0Ca²⁺ was toxic as shown by increased PI uptake in both CA1 and CA3 areas 24 h after the efflux experiment. The separate treatments with 100 μm PDC or 15 min of ACSF/0Ca²⁺, respectively, did not cause any delayed toxicity. The data are presented as the mean PI intensity expressed as percentages of cell death induced by 300 μm NMDA for 24 h (n = 4 ± S.E.). Asterisks indicate a significant difference (n = 4, p < 0.05) compared with controls (ACSF treatment). d, incubation with 100 μm TBOA during the efflux experiments only generated nearly maximal cell death per se when PI uptake was determined 24 h after the efflux experiment. Incubation in ACSF/0Ca²⁺, 0Mg²⁺ caused no cell death. The symbols mark a significant difference (n = 6, p < 0.05) compared with controls (ACSF treatment). e, fluorescence micrographs taken 24 h after the efflux experiments. Top row, from the left): a control slice incubated only in ACSF; a slice treated with 15 min of ACSF/0Ca²⁺ showing no significant PI uptake 24 h later; a slice incubated with both 100 μm PDC and 15 min of ACSF/0Ca²⁺ showing increased PI uptake in the CA1 and CA3 areas 24 h later; and a slice treated with 300 μm NMDA for 24 h representing maximum cell death in this model. Note that in the third panel from the left, the PI fluorescence is confined to neuronal-rich areas, whereas non-neuronal areas (marked by asterisks) show low fluorescence. Bottom row, from the left: a slice incubated in ACSF/0Ca²⁺, 0Mg²⁺; a slice incubated in 100 μm TBOA; and a slice incubated in ACSF/0Ca²⁺, 0Mg²⁺, and 100 μm TBOA. Note that PI uptake was measured 24 h after the efflux experiments.

When Ca²⁺ omission was combined with PDC, the efflux of GSH (Fig. 8a) was further increased in comparison with Ca²⁺ omission only. Interestingly, this effect was observed after, but not during the 15-min period of Ca²⁺ omission. The stimulating effect was specific, because the efflux of valine (data not shown, n = 4) was unaffected throughout the experiments.

Uptake of PI was evaluated 24 h after efflux experiments to investigate whether stimulated efflux was correlated with toxicity (Fig. 8, c and d). The addition of the uptake blockers to the incubation medium with normal Ca²⁺ revealed that PDC was nontoxic but that TBOA (included during the efflux experiment only) induced significant delayed toxicity when evaluated 24 h after the efflux experiment (Fig. 8, c and d). Omission of Ca²⁺ for 15 min did not cause delayed nerve cell death, but the combination of PDC and Ca²⁺ removal caused delayed neurotoxicity in both the CA3 and CA1 areas 24 h after the efflux experiments (Fig. 8c).

The combination of PDC and Ca²⁺ omission also caused an efflux of N-acetylaspartate, which could not be detected by only removing Ca²⁺ from the incubation medium (Fig. 9a). The efflux of N-acetylaspartate was also increased by complete removal of divalent cations from the medium (Fig. 9b). It is worth noting that the maximal efflux of N-acetylaspartate was delayed compared with the efflux of glutamate (compare Fig. 9, a and b, with Figs. 2, a and b, and 3, a and b).

FIGURE 9.

a, omission of Ca²⁺ in combination with blocking Glu uptake with PDC caused an efflux of the neuronal marker N-acetylaspartate. N-Acetylaspartate could not be detected in the incubation medium from slices treated with only Ca²⁺-free ACSF for a period of 15 min (data not shown). The maximum efflux rate of N-acetylaspartate was reached later (at 30 min) than that for GSH and glutamate (at 25 min) (compare with Fig. 3a). The data are presented as the mean efflux rates (n = 4 ± S.E.). Asterisks indicate a significant difference (p < 0.05) compared with efflux from slices incubated without PDC. b, omission of both Mg²⁺ and Ca²⁺ also caused an efflux of N-acetylaspartate. The data are presented as the mean efflux rates (n = 4 ± S.E.). Asterisks indicate a significant difference (p < 0.05) to efflux in ACSF-Ca²⁺.

DISCUSSION

The conventional gap junction blockers CBX, FFA as well as two unconventional blockers, endothelin-1 and the connexin43 mimetic peptide Gap26, reduced the efflux of GSH, taurine, PEA, and hypotaurine induced by Ca²⁺ omission, whereas the structural CBX analogue glycyrrhizic acid, which does not block gap junctions, was ineffective. Whatever the exact mechanism of endothelin-1 is, the blocking action is on intrinsic signaling pathways (29, 30) rather than at the connexin proteins directly, which is likely the mechanism of action for the blocking peptide Gap26 and glycyrrhizic acid derivatives such as CBX (26, 27).

The connexin mimetic peptide Gap26 has previously been used to block ATP efflux induced by Ca²⁺ omission from endothelial cells in cornea (16). Importantly it was found that the gap junction communication was not altered by Gap26, which is different from the effects of CBX, which blocks both gap junction communication and hemichannel activation. Concerning the pannexin1 mimetic peptide ¹⁰panx1, it has been shown that it reduces the ATP stimulated efflux from P2X₇ containing cells that overexpressed pannexin1, whereas no effect was observed by a connexin mimetic peptide (9). These studies agree well with our results in that the peptides appear selective, i.e. we observed no effect of the pannexin1 mimetic peptide on efflux stimulated by Ca²⁺ omission. This is in contrast to the recent study showing that both pannexin and connexin mimetic peptides have unspecific effects (31). We have no clear-cut explanation for these discrepancies, but one major difference is the model system that in the latter case was oocytes transfected with connexins and pannexin1. In the former case a variety of primary cultures and human cells such as macrophages were used. Obviously, more studies are needed to sort out these apparently contrasting findings of the mimetic peptides.

The efflux caused by Ca²⁺ omission was enhanced by Mg²⁺ omission. This is in line with the finding that the presence of extracellular Mg²⁺ accelerated deactivation and slowed activation of connexin46 hemichannels (32). It also agrees well with the blocking effect of extracellular Mg²⁺ on efflux of glutamate from astroglial cultures stimulated by divalent cation-free solution (6).

It has recently been shown that several gap junction blockers inhibit ATP efflux following activation of P2X₇ receptors (15). This effect may be exerted via an inhibitory effect on pannexin hemichannels, which is one pathway for ATP efflux following activation of P2X₇ receptors (3, 8) rather than a direct effect on the purine receptor. However, the lack of effect of ox-ATP on the efflux at concentrations that effectively block these receptors in other systems (33, 34) suggests that P2X₇ channels are not a pathway for the efflux induced by Ca²⁺ omission. Another recent finding in line with this interpretation is that Brilliant Blue G, another blocker of P2X₇ receptor channels, does not inhibit efflux of glutathione from astroglial cultures in divalent cation-free solution (35).

The lack of effect of hyperosmotic medium (240 mm sucrose) indicates that the stimulated efflux is unrelated to opening of volume-regulated anion channels. The efflux pattern induced by hypoosmotic medium is also different from the efflux induced by Ca²⁺ omission. In the former case taurine is the predominant released compound (12), whereas in the case of Ca²⁺ omission it is GSH.

The uniform potent effect by the different types of gap junction inhibitors and the strong dependence of the efflux on divalent cations (6) together with the lack of effect of P2X₇ inhibition and hyperosmotic medium on the efflux suggests that the efflux is mediated by connexin hemichannels. Alternative candidates are hemichannels formed by pannexin proteins (3). Such hemichannels have been shown to release ATP (36) and to be a part of the pore-forming unit of the P2X₇ receptor death complex when co-expressed in Xenopus oocytes (8). Although pannexins are potently blocked by CBX (8, 37), hemichannels formed by these proteins lack the regulation by extracellular Ca²⁺ that is common to most connexins (37). Furthermore, pannexins in the hippocampus are mainly localized to postsynaptic structures, and although cultured astroglia express pannexin1, there is no evidence for functional channels (38, 39). This is in line with our observation that the pannexin1 mimetic peptide ¹⁰Panx1 was ineffective. When the effect of conventional gap junction blockers was evaluated on activation of pannexin hemichannels, it was found that pannexin hemichannels were less sensitive to inhibition by FFA compared with connexin hemichannels (37). Thus, although we lack a positive control of ¹⁰Panx1 in the slice model, our results of inhibitory effects with the connexin mimetic peptide Gap26, the stimulated efflux by Ca²⁺ omission, near 100% blockage with FFA, and persistent effect of Ca²⁺ omission when neurons are degenerated by NMDA treatment strongly indicate the involvement of connexin hemichannels rather than pannexin hemichannels.

Strangely both ox-ATP, a P2 receptor antagonist, and BzATP, a P2X₇ receptor agonist, increased the efflux stimulated by Ca²⁺ omission. The stimulating effect of BzATP agrees well with a study showing that Ca²⁺ omission potentiates BzATP-induced tonic current in CA3 neurons via stimulated efflux of glutamate from astrocytes (40). This increased effect in low calcium was probably not related to hemichannel or pannexin hemichannels because CBX did not reduce the efflux. The similar stimulating effect of the antagonist ox-ATP is very puzzling, but effects on transporter function by ox-ATP (41) and changes in intracellular levels of reducing agents such as superoxide (42) have been reported earlier and cannot be excluded as possible explanations for changed efflux.

Treatment with CBX fully blocked the effect of Ca²⁺ omission in NMDA-treated slices. However, in CBX-treated control slices Ca²⁺ omission about doubled the efflux of GSH (Fig. 7). This CBX-insensitive efflux indicates that in neurons Ca²⁺ omission may induce efflux via a pathway that is unrelated to connexin or pannexin hemichannels. This putative pathway is at present not known, but an interesting finding in this respect is that Ca²⁺ omission elevates extracellular taurine via interaction with the carrier system (43).

Efflux of glutamate was only observed when Ca²⁺ omission was combined with the glutamate uptake blocker PDC or when both Ca²⁺ and Mg²⁺ were omitted from the medium. In our model system PDC did not induce heteroexchange because no effect was observed on extracellular glutamate during control experiments with ACSF only. We also used TBOA, a nontransported glutamate uptake blocker. Unfortunately, incubation with TBOA alone during the incubation period induced delayed toxicity, which limited its use. Unexpectedly, incubation with TBOA increased extracellular glutamate during omission of both Ca²⁺ and Mg²⁺ but not during omission of Ca²⁺ alone. We have no explanation for the apparent discrepancy between PDC and TBOA, but it may be related to differences in diffusion into the slice and/or to slight differences in inhibitory profile for the different glutamate carriers. Indeed, it has been shown that TBOA and PDC have different toxic profiles, the former resulting in NMDA receptor-mediated toxicity, whereas PDC also caused gliotoxicity (44).

We have earlier shown that NMDA and kainate cause opening of neuronal channels/pores and efflux of N-acetylaspartate, taurine, PEA, and GSH (13, 45). In our present study omission of both Ca²⁺ and Mg²⁺ increased efflux of N-acetylaspartate, whereas Ca²⁺ omission was more selective with no effect on extracellular N-acetylaspartate. However, treatment with PDC in combination with Ca²⁺ omission increased the efflux of N-acetylaspartate in parallel to elevated efflux of glutamate. N-Acetylaspartate is localized more or less selectively to neurons. Thus, our interpretation is that the omission of both Ca²⁺ and Mg²⁺ or treatment with PDC in combination with Ca²⁺ omission increases the efflux of glutamate and reduces the uptake of glutamate, respectively, leading to concentrations that are high enough to overactivate neuronal glutamate receptors. This in turn opens channels that release anions, including GSH and N-acetylaspartate, from the neurons as we have reported earlier (13, 45). An alternative explanation for the delayed efflux is activation of neuronal connexin or pannexin hemichannels. However, CBX does not block the delayed efflux induced by NMDA receptor overactivation,5 which favors the former interpretation.

Why are not all small intracellular substances (<1.0 kDa) released when a pore-like structure such as the hemichannel is opened? First of all it seems that different connexins have different permeability characteristics, which may explain some of the selectivity (46, 47). Furthermore, in a recent study it was shown that phosphorylation of connexin43 by protein kinase C alters the size selectivity of the pore, excluding larger hydrophilic solutes but still allowing small inorganic ions to pass (48). In addition, a strongly negative membrane potential, i.e. negative on the inside, will favor efflux of anions because of the potential gradient. The electrochemical gradient may thus explain the relatively dramatic efflux of the anionic GSH, glutamate (see further below), and PEA when the membrane potential is negative.

Cultured hippocampal slices represent a model system that in many respects is more in vivo-like than monolayer cell cultures. Whether the glial cells in the cultured slice are similar to those in vivo concerning sensitivity to omission of Ca²⁺ remains to be investigated. The efflux of glutamate and taurine via channels that are inhibited by gap junction blockers but not by P2X₇ receptor inhibitors fits well with the study by Ransom and co-workers (6) who used primary cultures of astrocytes. They also demonstrated efflux of glutamate from acutely dissected optic nerves by omission of divalent cations, indicating that glial cells in situ may release glutamate by hemichannels.

From computational models it appears that substantial changes in extracellular Ca²⁺ can occur in physiological situations (49). Other factors, for example intracellular acidification (50, 51), metabolic status (52, 53), redox situation (5, 54), phosphorylation status (55), changes in extracellular monovalent cations (56), and the membrane potential (57), also influence gap junction and hemichannel opening. It follows, but has not been investigated as far as we know, that hemichannels may open by moderate changes in extracellular Ca²⁺ in combination with changes in these other parameters that also regulate hemichannel open probability. Efflux of glutamate from astroglia cultures occurs in medium with 0.2 mm Ca²⁺ (6), whereas we did not observe any efflux of GSH or amino acids with 0.1 mm. One major difference between these studies is the model system, i.e. cultured slices (our study) versus cultured monolayers of astroglia. The extracellular Ca²⁺ concentration in the interior of the slice when incubating in nominal Ca²⁺ free ACSF is unknown but likely much higher than that of the incubating medium that has direct contact with a monolayer. One equally important difference is that in the former study the efflux was evaluated in Mg²⁺-free medium, which changes hemichannel opening considerably (6, 32). It has earlier been suggested that a small proportion of hemichannels may be open during more physiological extracellular concentrations (6). Our results, that basal efflux of GSH in physiological concentrations of extracellular Ca²⁺ and Mg²⁺ was reduced by gap junction blockers, indeed indicate that efflux via hemichannels may contribute to the extracellular composition also during physiological conditions.

High concentrations of glutamate and aspartate as well as anoxia can cause decreases in extracellular Ca²⁺ levels down to levels that conventionally are used experimentally to study hemichannel opening (58, 59). The massive decrease in extracellular Ca²⁺ is probably due to opening of both ligand-gated and voltage-gated Ca²⁺ channels. Hemichannel opening in vivo induced by decreased extracellular Ca²⁺ might thus occur when glutamate receptors are overactivated, for example in stroke and trauma as suggested by Ransom and co-workers (2).

What would the consequences of channel opening be if it occurs in vivo? Our results clearly show that opening for at least 15 min is nontoxic to neurons as long as the glutamate release is cleared by uptake. Opening of hemichannels in combination with intact glutamate uptake could therefore possibly be considered a form of autocrine or paracrine signaling. By our studies we demonstrate that GSH and glutamate can be released also from slice cultures and add PEA and hypotaurine to the list of substances that can be released by the glial hemichannels. Export of GSH from glial cells and extracellular breakdown of GSH has been suggested to be a major route to supply neurons with cysteine and/or cysteinylglycine for de novo synthesis of GSH inside neurons (60). Decreased extracellular Ca²⁺ implies increased intracellular Ca²⁺, which in turn causes increased oxidative stress in neurons. The efflux of GSH from astrocytes may thus be a way to supply neurons with GSH precursors. However, if the efflux from glial cells is prolonged, a decreased capability to handle stress because of depletion of GSH may lead to glial cell death (35). This has also been shown to be true for depletion of mitochondrial GSH in astrocytes (61). In addition, GSH is an endogenous ligand of glutamate receptors with the capability of modulating central excitability (62–64). GSH efflux via hemichannels could thus also be coupled to its proposed role as a neuromodulator.