Introduction

In the brain, astrocytes play active roles in cell-to-cell signaling by releasing gliotransmitters, such as glutamate and ATP, and thus forming neuron-glia and glia-glia networks 1, 2, 3, 4, 5. Astrocytic ATP release is involved in modulating a variety of neuronal activities, affecting excitability 6, synaptic transmission 7, 8, cell death 9, growth and survival 9, 10. ATP released from astrocytes also modulates the activities of astrocytes 11, 12 and microglia 13, 14, 15 through the stimulation of purinergic receptors.

In a variety of cell types, the release of ATP has been reported to be mediated by a number of non-lytic membrane transport pathways including exocytotic vesicular transport, connexin or pannexin hemichannels, the P2X7 receptor, ABC transporters such as MDR1 and MRP, as well as anion channels such as CFTR, volume-sensitive outwardly rectifying (VSOR) anion channels and maxi-anion channels 17, 18, 19: for Reviews.

Astrocytes have been shown to release ATP in a non-lytic manner in response to osmotic swelling 11, mechanical stimulation 8, 20, 21, 22, 23, deprivation of extracellular Ca2+ 22, 23, 24, and stimulation with glutamate 7, UTP 25, 26, noradrenaline 27 or NO 28. However, the precise pathway for ATP release remains controversial. In the present study, we aimed to identify the pathway for swelling-induced ATP release from mouse astrocytes in primary culture by performing RT-PCR analyses and pharmacological studies to test candidate pathways hitherto reported.

Results

Swelling-induced ATP release from mouse astrocytes

In normal Ringer solution, the basal release of ATP from mouse astrocytes was low and did not exceed 1.00 ± 0.13 nM over a 35-min incubation time (Figure 1, open squares). In contrast, when cell swelling was induced by exposure to a hypotonic solution (210 mosmol/kg-H2O), the extracellular ATP concentration rapidly reached a maximum level of approx. 2.5-3 nM within 15-25 min (Figure 1, open circles).

Figure 1
figure 1

Time course of ATP release from mouse astrocytes in response to hypotonic stress. The concentration of ATP released from mouse astrocytes cultured in 12-well plates was measured by a luciferin-luciferase assay after application of control isotonic or hypotonic solution (290 or 210 mosmol/kg-H2O). The extracellular ATP concentration is plotted as a function of the incubation time in control basal (open squares) and hypotonic (open circles) conditions. Each symbol represents the mean ± SEM (vertical bar). *Significantly different from the isotonic control values at given times at P < 0.05.

Full size image

When the cells were incubated in solutions of different osmolality (ranging from 290 to 130 mosmol/kg-H2O) for 15 min, the release of ATP increased with decreasing medium osmolality (Figure 2). The osmolality dependence of ATP release had a sigmoidal shape with half-maximal ATP release observed at a medium osmolality of 228 ± 16 mosmol/kg-H2O.

Figure 2
figure 2

Osmolality dependence of ATP release from mouse astrocytes. The concentration of ATP released from mouse astrocytes cultured in 24-well plates was measured 15 min after application of isotonic or hypotonic solution. The extracellular ATP concentration is plotted as a function of the osmolality of the extracellular medium. Each symbol represents the mean ± SEM (vertical bar). *Significantly different from the isotonic control values at P < 0.05.

Full size image

Gene expression of molecules comprising candidate pathways for ATP release in mouse astrocytes

A number of possible ATP-releasing pathways have been suggested over the past several years. To determine which candidate pathway may be important in mouse astrocytes, we assayed the expression of candidate pathway molecule mRNAs by RT-PCR analysis. Under normotonic conditions, as shown in Figure 3A and 3B, RT-PCR yielded amplified cDNA fragments of expected size for connexins (Cx32, Cx37, Cx43), pannexin 1 (Px1), MRP1, P2X7 receptors (P2X7), MDR1a, and MDR1b. However, CFTR mRNA was not detected in mouse astrocytes but could be detected in mouse lung homogenates (MLHs), which were used as a positive control (Figure 3C). This profile of gene expression was not affected by osmotic swelling of the astrocytes induced by exposure to hypotonic solution (210 mosmol/kg-H2O) for 15 min (data not shown).

Figure 3
figure 3

Expression of mRNAs of various candidates for ATP release pathways in cultured mouse astrocytes. (A) RT-PCR performed for Cx32 (lane 2), Cx37 (lane 3), Cx43 (lane 4), Px1 (lanes 5 and 6), MRP1 (lane 7), P2X7 (lane 8), GAPDH (lane 9), and β-actin (lane 10) in mouse astrocytes. (B) RT-PCR for MDR1a (lane 2), MDR1b (lane 3), β-actin (lane 4), and GAPDH (lane 5) in mouse astrocytes. (C) RT-PCR for CFTR (lanes 2 and 4), for β-actin (lanes 3 and 5) in mouse astrocytes (lanes 2 and 3) and mouse lung homogenates (MLH: lanes 4 and 5). The first and last lanes show size markers (100-bp ladders).

Full size image

Pharmacological sensitivity of swelling-induced ATP release from mouse astrocytes

We next tested the sensitivity of ATP release from swollen astrocytes to blockers of not only the candidate pathways examined in the above RT-PCR analysis but also other possible pathways including exocytotic vesicular transport as well as the VSOR anion channel and the maxi-anion channel, which have not yet been molecularly identified. When mouse astrocytes were exposed to hypotonic solution of 210 mosmol/kg-H2O for 15 min in the absence of any blockers, massive ATP release was observed as before (Figure 4A, first column). We then observed the effects of different blockers for candidate pathways by adding to the hypotonic solution at concentrations that were previously reported to be most effective. The data are summarized in Figure 4A (other columns). Swelling-induced ATP release was unaffected by the presence of glibenclamide (200 μM), a known blocker of the CFTR anion channel 29, a member of the ABC transporter superfamily. This is in accord with the RT-PCR data for CFTR (Figure 3C). Similarly, swelling-induced ATP release was not significantly affected by blockers, probenecid (1 mM) and verapamil (10 μM), which are known to block other ABC transporter proteins, MRP1 30, 31 and MDR1 32, respectively. Blockers of gap junction hemichannels also failed to affect swelling-induced ATP release; no significant effects were observed by adding 2 mM 1-octanol, a blocker of connexins 33, or 100 μM carbenoxolone, a blocker of both connexins and pannexins 34. A blocker of the P2X7 receptor 24, 35, brilliant blue G (1 μM), was also ineffective. When exocytotic vesicular transport was suppressed by adding an intracellular Ca2+ chelator, BAPTA-AM (50 μM), together with an inhibitor of vesicular transport 23, brefeldin A (BFA; 5 μM), ATP release from swollen astrocytes was not significantly affected. We then observed the effects of anion channel blockers. The ATP release was suppressed to an intermediate degree by NPPB (100 μM) and SITS (100 μM), both of which are known to moderately inhibit maxi-anion channels, but more strongly block VSOR anion channels in mouse astrocytes 36. In contrast, phloretin (100 μM), which is a blocker relatively specific to the VSOR anion channel 37, had no significant effect on swelling-induced ATP release. Also, DCPIB (10 μM), another blocker specific to the VSOR anion channel 38, failed to affect swelling-induced ATP release. We next examined the effect of Gd3+ (50 μM), which blocks, from the extracellular side, maxi-anion channels expressed in many cell types 17, 39, 40, 41 including mouse astrocytes 36. As reported previously 42, Gd3+itself suppressed the luciferin-luciferase reaction. As shown in Figure 4B, the slope of a calibration curve obtained by adding known amounts of ATP to normal Ringer solution in the presence of Gd3+ (triangles) was significantly smaller than that in the absence of Gd3+ (squares). However, the Gd3+ effect on the luciferin-luciferase reaction was completely eliminated by adding 600 μM EDTA to the luciferase solution (Figure 4B, circles). By supplementing the luciferin-luciferase reaction with EDTA, Gd3+ was found to be most effective in inhibiting hypotonicity-induced ATP release (Figure 4A, last column).

Figure 4
figure 4

Effects of blockers of various candidate pathways for ATP release on the hypotonicity-induced ATP release from mouse astrocytes (A) and effects of Gd3+ and EDTA on the calibration for ATP measurements by the luciferin-luciferase reaction (B). In (A), the concentration of ATP released from mouse astrocytes cultured in 24-well plates was measured 15 min after application of hypotonic solution (210 mosmol/kg-H2O). The concentrations of blockers are given in the text. Each symbol represents the mean ± SEM (vertical bar). *Significantly different from the control value (hypotonicity in the absence of drugs for A and Gd3+- and EDTA-free Ringer solution for B) at P < 0.05.

Full size image

As shown in Figure 5, marked, moderate and little blocking effects of Gd3+, SITS and DCPIB, respectively, were observed on swelling-induced ATP release during the whole time range (up to 35 min) of observations.

Figure 5
figure 5

Effects of Gd3+ (50 μM), SITS (100 μM) and DCPIB (10 μM) on the time course of swelling-induced ATP release from mouse astrocytes. Each symbol represents the mean ± SEM (vertical bar). *Significantly different from the control hypotonic values in the absence of blockers.

Full size image

From the above results, it is concluded that swelling-induced astrocytic ATP release does not involve vesicular transport, ABC transporters such as CFTR, MDR1 and MRP1, connexin or pannexin hemichannels, or swelling-activated VSOR anion channels. Instead, the evidence suggests that the Gd3+-sensitive maxi-anion channel serves as the pathway for ATP release from swollen mouse astrocytes.

Discussion

Under stimulation of receptors or exposure to stress, brain astrocytes affect neuronal brain functions by releasing a number of gliotransmitters, including glutamate and ATP. Ubiquitously expressed swelling-activated anion channels, called VSOR anion channels or volume-regulated anion channels (VRACs), are believed to serve as a pathway for swelling-induced release of the excitatory amino acids glutamate and aspartate from astrocytes 43, 44. In fact, Abdullaev et al. 45 have recently shown that swelling-induced aspartate efflux is mediated by Gd3+-insensitive VSOR anion channels in rat astrocytes. We have provided compelling evidence that not only the VSOR anion channel but also a Gd3+-sensitive maxi-anion channel represent major conductive pathways for the release of glutamate from mouse astrocytes challenged by hypotonic or ischemic stress 36. Recent pore sizing studies using a nonelectrolyte exclusion technique showed that the pore radius of the maxi-anion channel (∼1.3 nm) 46, but not of the VSOR anion channel (∼0.63 nm) 47, is large enough for the channel to be permeated not only by glutamate (cut-off radius ∼0.35 nm) but also by the anionic forms of ATP (0.58∼0.65 nm). Direct evidence has accumulated for the ATP conductivity of maxi-anion channels in a number of other cell types 39, 40, 41. It is therefore highly likely that swelling-induced ATP release from mouse astrocytes is also mediated by Gd3+-sensitive maxi-anion channels that are known to be activated by hypotonic and ischemic stresses in this cell type 36. The present study showed, in fact, that ATP release from osmotically swollen astrocytes was most effectively blocked by Gd3+.

Thus far, a large variety of molecules have been suggested to constitute the non-lytic and non-exocytotic pathways of ATP release. These include CFTR, MDR1 and MRP1 31 for rat astrocytes, the connexins 22 and P2X7 receptors 24 for mouse astrocytes, as well as the pannexins 19 for other cell types. The present RT-PCR study demonstrated molecular expression of connexins (Cx32, Cx37, Cx43), Px1, MDR1, MRP1 and the P2X7 receptor in both non-swollen and swollen mouse astrocytes. These data are in agreement with those from previous RT-PCR studies that examined MRP1 30, 48 and the P2X7 receptor 49 in mouse astrocytes in normotonic conditions, and Px1 50, MDR1 31, 45, 51 and MRP1 31, 48 in rat astrocytes, also in normotonic conditions. However, the present study failed to detect the CFTR transcript in mouse astrocytes, in contrast to a previous report 30. Previous pharmacological studies suggested that ATP release involves ABC transporters in rat cortical astrocytes 11, 31 and the P2X7 receptor in mouse spinal cord astrocytes 24. However, the present pharmacological study showed that swelling-induced ATP release from mouse cortical astrocytes was insensitive to blockers of ABC transporters (glibenclamide, probenecid, verapamil) and the P2X7 receptor (brilliant blue G). In addition, it was found that blockers of connexin and pannexin hemichannels (1-octanol, carbenoxolone) and VSOR anion channels (phloretin, DCPIB) did not significantly affect swelling-induced ATP release from mouse astrocytes.

It is evident that exocytotic release of secretory granules or vesicles is a non-lytic source of ATP in some secretory cell types 17. A recent single-vesicle imaging study has provided evidence for astrocyte exocytosis 52. On the basis of sensitivity to bafilomycin A1, tetanus neurotoxin, botulinum neurotoxin C and Ca2+ chelators, it was suggested that mechanostress- or NO-induced ATP release from rat astrocytes is mediated by exocytosis 23, 28. In the present study, however, intracellular Ca2+ chelation by pretreatment with a Ca2+ chelator, BAPTA-AM, and an inhibitor of vesicular transport, BFA, failed to inhibit swelling-induced ATP release from mouse astrocytes. Although the Ca2+ chelator and BFA may affect many processes, it appears that swelling-induced ATP release does not involve exocytosis which is an event essentially dependent on cytosolic Ca2+ and sensitive to BFA.

Since ATP exists in anionic forms at physiological pH, there is a possibility that some anion channel type (other than CFTR) may serve as the pathway for ATP release from mouse astrocytes. An involvement of the Ca2+-dependent anion channel may be ruled out because of insensitivity of the ATP release to cytosolic Ca2+ chelation. Its insensitivity to phloretin and DCPIB may similarly rule out an involvement of the VSOR anion channel. The most effective blocking agent for the ATP release from swollen astrocytes was Gd3+. Gd3+ is known to block not only maxi-anion channels 17, 36, 39, 40, 41 but also some cation channels such as mechano-gated cation channels 53 and several members of the TRP cation channel family 54 at similar concentrations, although it is not known how Gd3+ blocks both cation and anion channels. However, involvements of cation channels in release of anionic ATP would be implausible.

There is a possibility that the data shown in Figure 1 represent underestimates of the actual amount of ATP released from mouse astrocytes, because the activity of ecto-ATPases was not inhibited in the present experiments. To test this possibility, the effects of potent ecto-ATPase blockers, such as suramin and DIDS 55, 56 as well as ARL 67156 57, could be examined. However, suramin was found to almost completely inhibit the luciferin-luciferase reaction (HT Liu and Y Okada, unpublished), and DIDS was reported to block the maxi-anion channel activity 58. Thus, we made preliminary experiments by applying ARL 67156 (100 μM) and found that this ecto-ATPase blocker markedly increased the extracellular concentration of ATP released from mouse astrocytes in response to a hypotonic challenge (210 mosmol/kg-H2O for 15 min) (HT Liu and Y Okada, unpublished). Even in the presence of ARL 67156, the hypotonicity-induced ATP release was found to be significantly suppressed by Gd3+ (HT Liu and Y Okada, unpublished). Although it is not known at present whether ARL 67156 affects any candidate pathways for ATP release, it appears that the Gd3+-sensitive maxi-anion channel serves as a major pathway for swelling-induced ATP release from mouse astrocytes even under the conditions where ecto-ATPases were inhibited.

Taken together, our results suggest that swelling-induced release of anionic ATP from mouse cortical astrocytes is mediated by Gd3+-sensitive maxi-anion channels. There is little information about the molecular nature of the maxi-anion channel. A mitochondrial porin (voltage-dependent anion channel (VDAC)) located in the plasma membrane has long been considered as the molecule underlying the maxi-anion channel activity 59, 60, 61, 62, based upon similarities in the biophysical properties of these two channels and the purported presence of VDAC protein in the plasma membrane. In our recent study 63, we have deleted all the three genes encoding the VDAC isoforms and demonstrated that the maxi-anion channel activity in VDAC-deficient mouse fibroblasts was unaltered. The lack of correlation between VDAC protein expression and maxi-anion channel activity strongly argues against the long-held hypothesis of plasmalemmal VDAC being the maxi-anion channel. Molecular identification of the maxi-anion channel awaits further investigation.

Brain astrocytes are known to exhibit swelling in response to ischemic and traumatic brain injury 44. Also, astrocytes are found to undergo swelling after stimulation with glutamate 64 as well as during hyperammonemia 65, 66 and lactacidosis 67. Therefore, it is conceivable that ATP release from swollen astrocytes plays some role in such pathological conditions.

Materials and methods

Solutions and chemicals

The normal Ringer solution contained (mM): 135 NaCl, 5 KCl, 2 CaCl2, 1 MgCl2, 5 Na-HEPES, 6 HEPES, and 5 glucose (pH 7.4, 290 mosmol/kg-H2O). Hypotonic solutions were prepared by mixing the normal Ringer solution with a HEPES-buffered solution containing (mM): 5 KCl, 2 CaCl2, 1 MgCl2, 5 Na-HEPES, 6 HEPES, and 5 glucose (pH 7.4, 38 mosmol/kg-H2O).

GdCl3 was stored as a 50-mM stock solution in water and added directly to the hypotonic solution immediately before each experiment. NPPB, glibenclamide, SITS, phloretin, BFA, carbenoxolone, 1-octanol, probenecid and brilliant blue G were purchased from Sigma-Aldrich (St. Louis, MO). Verapamil and BAPTA/AM were obtained from Nacalai Tesque (Kyoto, Japan) and DOJINDO (Kumamoto, Japan), respectively. The drugs were added to hypotonic solution immediately before use from stock solutions in DMSO. DMSO did not have any effect, when added alone at a concentration less than 0.1%. Osmolality of all solutions was measured using a freezing-point depression osmometer (OM802: Vogel, Kevelaer, Germany).

Cell culture and tissue preparation

The experimental protocol was approved in advance by the Ethics Review Committee for Animal Experimentation of the National Institute for Physiological Sciences. Astrocytes were obtained from 2–3-day old pups of Slc:ddy mice (Japan SLC, Inc., Hamamatsu, Japan) using the method described previously 36. The cells were cultured in 12- or 24-well plates, and, upon reaching confluency, were used for ATP measurements and RT-PCR assays. The cell number per well in 12- or 24-well plates was (2.35 ± 0.19) × 105 or (9.49 ± 1.03) × 104, respectively.

The MLHs were prepared from 100 mg tissue samples of the lung dissected from 90-day old mice after homogenizing with a glass/Teflon homogenizer (Iuchi, Osaka, Japan) in 1 ml Sepasol RNA I reagent.

Luciferin-luciferase ATP assay

The bulk extracellular ATP concentration was measured by the luciferin-luciferase assay, as described previously 36, 39, using astrocytes cultured in 12- or 24-well plates. After the culture medium was totally replaced with isotonic Ringer solution (1 000 and 425μl for 12- and 24-well plates, respectively), cells were incubated at 37°C for 60 min. As a control sample for background ATP release, an aliquot (100 μl) of the extracellular solution was collected. An osmotic challenge was then applied by gently removing most of the remaining extracellular solution (875 and 300 μl for 12- and 24-well plates, respectively), adding the hypotonic solution (1 000 and 400μl for 12- and 24-well plates, respectively), and then placing the plates in an incubator at 37 °C. At specified time points, the plate was carefully rocked again to ensure homogeneity of the extracellular solution. For the luminometric ATP assay, samples (20 and 50 μl for 12- and 24-well plates, respectively) were collected from each well at specified time points. The ATP concentration was measured, under nearly normotonic conditions, by mixing 20 or 50 μl of sample solution with 530 or 500 μl normal Ringer solution, respectively, and 50 μl of luciferin-luciferase reagent. Ionic salt sensitivity of the luciferase reaction was negligible. When required, drugs were added to the hypotonic solution to give the final concentrations as indicated. We supplemented the luciferin-luciferase assay mixture with 600 μM EDTA when the samples contained Gd3+, because Gd3+ interfered with the luciferin-luciferase reaction (see Figure 4B), as reported previously 42. Other drugs employed in the present study had no significant effect on the luciferin-luciferase reaction.

RT-PCR analysis of mRNA expression

Total RNA was isolated from cultured mouse cortical astrocytes before and after 15-min exposure to hypotonic solution (210 mosmol/kg-H2O) or from homogenates of the mouse lung using Sepasol RNA I reagent (Nacalai Tesque) according to the manufacturer's instructions. First-strand cDNA was synthesized from the isolated RNA using reverse transcriptase (Roche Diagnostics GmbH, Mannheim, Germany) and oligo-dT primers (Invitrogen Corp., Carlsbad, CA). The synthesis of cDNA was performed according to the manufacturer's protocol. The sequences of primers for P2X7, MRP1 and CFTR were the same as reported previously 30, 49. All other gene-specific primers used for PCR were designed using Primer3 software (www.genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi). The sequences of primers for RT-PCR experiments are given in supplementary information, Table S1. PCR was carried out with Blend Taq (TOYOBO, Osaka, Japan) and a Gene Amp PCR System 9600 thermal cycler (Perkin-Elmer Life Sciences, Boston, MA). Cycling conditions were 4 min at 94 °C, followed by 35 cycles of 1 min at 94 °C, 1 min at the annealing temperature given for the individual primers (supplementary information, Table S1), and 1 min at 72 °C, and, finally, 10 min at 72 °C. The integrity of the isolated RNA and the reverse transcription reaction were examined using specific primers for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and β-actin. All PCR products were analyzed on a 2% agarose gel and had the sizes expected from the known cDNA sequences.

Data analysis

Data were analyzed with OriginPro 7.0 (MicroCal Software, Northampton, MA). Pooled data are given as means ± SEM of observations (n). Statistical differences of the data were evaluated by ANOVA and the paired or unpaired Student's t test where appropriate, and considered significant at P < 0.05.

(Supplementary Information is linked to the online version of the paper on the Cell Research website.)