Autocrine Regulation of Volume-sensitive Anion Channels in Airway Epithelial Cells by Adenosine*

The activity of volume-sensitive Cl− channels was studied in human tracheal epithelial cells (9HTEo−) by taurine efflux experiments. The efflux elicited by a hypotonic shock was partially inhibited by adenosine receptor antagonists, by α,β-methyleneadenosine 5′-diphosphate (αβMeADP), an inhibitor of the 5′-ectonucleotidase, and by adenosine deaminase. On the other hand, dipyridamole, a nucleoside transporter inhibitor, increased the swelling-induced taurine efflux. Extracellular ATP and adenosine increased taurine efflux by potentiating the effect of hypotonic shock. αβMeADP strongly inhibited the effect of extracellular ATP but not that of adenosine. These results suggest that anion channel activation involves the release of intracellular ATP, which is then degraded to adenosine by specific ectoenzymes. Adenosine then binds to purinergic receptors, causing the activation of the channels. To directly demonstrate ATP efflux, cells were loaded with [3H]AMP, and the release of radiolabeled molecules was analyzed by high performance liquid chromatography. During hypotonic shock, cell supernatants showed the presence of ATP, ADP, and adenosine. αβMeADP inhibited adenosine formation and caused the appearance of AMP. Under hypotonic conditions, elevation of intracellular Ca2+ by ionomycin caused an increase of ATP and adenosine in the extracellular solution. Our results demonstrate that volume-sensitive anion channels are regulated with an autocrine mechanism involving swelling-induced ATP release and then hydrolysis to adenosine.

Virtually all cells in a multicellular organism undergo swelling or shrinking following changes of intracellular or extracellular osmotic pressure. Although the osmolarity of body fluids, particularly in mammalians, is tightly controlled, significant variations may occur in physiological or pathological conditions. For example, a dramatic drop in extracellular osmolarity occurs in the renal medulla during the passage from antidiuresis to diuresis (1). Cell swelling may also occur in the brain during hypoxic and toxic conditions or following treatment of diabetic ketoacidosis (2)(3)(4)(5) and in the heart following myocardial ischemia (6,7).
Usually cells respond to osmotic stresses by transporting organic and inorganic osmolytes through the plasma mem-brane. The lowering of extracellular osmolarity and the consequent cell swelling activate potassium and chloride channels (8). The resulting exit of KCl drives water efflux and therefore restores the original cell volume, a mechanism termed regulatory volume decrease. A key element in this process is represented by volume-sensitive Cl Ϫ channels, called volume-sensitive organic anion channel (VSOAC), 1 which are also permeable to organic osmolytes such as taurine (9 -11). Because taurine in many cells is accumulated at high concentrations through Na ϩ -dependent transporters, the opening of VSOAC causes a significant taurine efflux, thus contributing to regulatory volume decrease. The mechanism underlying VSOAC activation is unknown. In a previous work performed on a tracheal epithelial cell line, we found that extracellular ATP is a potent modulator of swelling-induced VSOAC activation (11). This finding led us to test the hypothesis that release of endogenous ATP by the cells could be responsible for VSOAC activation during the hypotonic shock. Our experiments did not support this hypothesis because ATP was not revealed by the luciferin-luciferase method, and exogenous hexokinase did not inhibit VSOAC activity (11). In contrast with our results, other investigators have found that there is a release of ATP induced by cell swelling (12,13). To reconcile these conflicting results, we have considered the possibility that ATP efflux in our cells could be masked by the rapid catabolism caused at the extracellular side of the membrane by specific ectoenzymes (14,15). The present article deals with the hypothesis that ATP is indeed released upon cell swelling and rapidly degraded to adenosine and that this nucleoside is the modulator of volumesensitive anion channels, at least in our cell model.

EXPERIMENTAL PROCEDURES
Cell Culture-The 9HTEoϪ cell line was obtained by immortalization of human tracheal epithelial cells (16). Cells were grown at 37°C in an atmosphere of 5% CO 2 using a medium containing 45% Dulbecco's modified Eagle's medium, 45% Ham's F-12, and 10% fetal clone II serum plus 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine. Cells were plated on 35-mm Petri dishes at a density of 40,000/cm 2 and cultured for 4 days before performing taurine and ATP transport studies.
Taurine and ATP Efflux-For this study, we have taken advantage of * This work was supported by Telethon-Italy Grant E.593. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed: Laboratorio di Genetica Molecolare, Istituto Giannina Gaslini, Largo Gerolamo Gaslini 5, 16148 Genova, Italy. Tel.: 39-010-5636532; Fax: ϩ39-010-3779797; Email: galietta@unige.it. 1 The abbreviations used are: VSOAC, volume-sensitive organic anion channel; HS, hypotonic solution; FE, fractional efflux; DPCPX, 8-cyclopentyl-1,3-dipropylxanthine; DMPX, 3,7-dimethyl-1-propargylxanthine; ␣␤MeADP, ␣,␤-methyleneadenosine 5Ј-diphosphate; 2MeSATP, 2-methylthioadenosine triphosphate; ADA, adenosine deaminase; NBMPR, S-(4-nitrobenzyl)-6-thioinosine; AMP-PNP, 5Јadenylylimidodiphosphate; HPLC, high performance liquid chromatography; PLC, phospholipase C. the relatively large taurine permeability of VSOAC (11) to perform efflux experiments. This approach seems more appropriate than patchclamp measurements to study the regulation of volume-sensitive channels, because taurine efflux experiments leave the cell interior undisturbed. The procedure for taurine efflux has been previously described (11). Briefly, cells were incubated for 1 h at 37°C with 10 M taurine and 0.2 Ci/ml [ 3 H]taurine. At the end of incubation, the loading medium was discarded, and the cells were washed three times with ice-cold taurine-free medium. Taurine efflux experiments started with the addition of 1 ml of standard isotonic solution prewarmed at 37°C. Every 5 min, for a total of 35 min, the efflux medium containing the taurine released by the cells was removed from the Petri dish and rapidly replaced by another ml of solution. The hypotonic shock was applied 15 min after the beginning of efflux. At the end of the experiment, cells were lysed by overnight incubation with 0.25 M NaOH. The radioactivity present in the efflux samples and in cell lysates was determined by liquid scintillation. Total incorporated taurine T 0 was obtained by adding the radioactivity of all efflux samples to that remaining in the NaOH extracts. The taurine remaining in the cells at a given time t (T res ) was determined by subtracting from T 0 the amount of radioactivity in the efflux samples up to time t. The time course of taurine efflux was expressed by plotting fractional efflux (FE) versus time. The FE at a given time was calculated according to FE ϭ T ex /T res , where T ex is the amount of taurine released in a single efflux interval, and T res is the amount of residual taurine remaining in the cells at the beginning of that interval.
For ATP release experiments, cells were loaded for 1 h with 2 Ci/ml [ 3 H]AMP. After loading, the cells were treated with the same protocol of taurine efflux experiments except for the lysis, which was accomplished by three rapid freeze-thaw cycles. The radioactivity collected in the efflux samples was normalized for the total radioactivity accumulated into the cells as explained above. Except where indicated, isotonic and hypotonic solutions during the efflux always contained 1 M dipyridamole to block the uptake of adenosine.
Ion Pairing Reverse-phase HPLC Analysis of 3 H-labeled Nucleotides and Their Derivatives-Two hundred l of cell supernatants or 100 l of cell lysates were analyzed in an LKB liquid chromatograph (Amersham Pharmacia Biotech) equipped with a Supelcosil LC-18-T column (15 cm x 4.6-mm inner diameter, 3-mm particle size; Supelco, Bellefonte, PA) and a Supelguard LC-18-T cartridge precolumn (Supelco). Buffer A was 60 mM KH 2 PO 4 , 5 mM tetrabutylammonium hydrogen sulfate, pH 6.0, containing 5% (v/v) methanol. Buffer B was 70% buffer A and 30% (v/v) methanol. The mobile phase was developed at a constant flow rate of 1.5 ml/min as follows: from 0 to 6 min, 0 -60% buffer B; from 6 to 10 min, 60% buffer B; from 10 to 12 min, 60 -0% buffer B. The system and column were reequilibrated in buffer A for an additional 8 min before the subsequent injection. The column eluate was monitored through a model Flow-one Beta radioactivity detector (Packard Instruments Co, Meriden, CT). Scintillation liquid was Ultima-Flo M (Packard) at a flow rate of 3 ml/min, which yielded an approximately 25% efficiency. Peaks were quantitated by means of the Flo-one/Beta software (Packard), and substances were identified on the basis of coelution with their unlabeled counterparts monitored at 254 nm by a LKB multiwavelength UV detector.
Analysis of ATP Degradation-Cells were plated and cultured as described for efflux studies. To detect the activity of ectoenzymes, cells were incubated at 37°C with the saline solution containing nonradioactive ATP or AMP-PNP. The cell supernatants were then analyzed by HPLC as explained above.
Materials-Culture media and fetal clone II were purchased from HyClone (Cramlington, UK). [ 3 H]Taurine and [ 3 H]AMP were from Amersham Pharmacia Biotech. DMPX, DPCPX, U73122, and CGS21680 were from RBI (Natick, MA). AMP-PNP was purchased from Roche Molecular Biochemicals. All other chemicals were from Sigma.
Statistics-Data are presented as mean ϮS.E. The points shown in each figure represent the mean of at least four experiments. Statistical significance was assessed by Student's t test for unpaired data.

RESULTS
According to our hypothesis, the hypotonic shock would cause a release of endogenous ATP. ATP would then be converted to adenosine by extracellular local catabolism. Adenosine, by interacting with adenosine receptors, would be responsible for the activation of VSOAC during the hypotonic shock. To test this model, we applied the hypotonic shock (68% HS) in combination with known adenosine receptor antagonists. DPCPX, which is a specific inhibitor of A 1 receptors at low nM concentrations (17), significantly inhibited the peak of swelling-induced taurine release (Fig. 1, A and B). The effect was partial and dose-dependent with a maximal inhibition of 34% and a half-effective concentration equal to 16 nM (Fig. 1B). Similarly, DMPX, an antagonist that acts at concentrations in the low M range with a slight selectivity for A 2 versus A 1 receptors (18), was also able to inhibit taurine efflux (Fig. 1, A and B). Maximal inhibition and half-effective concentration were 44% and 560 nM, respectively. When DMPX and DPCPX were used together at maximal concentrations (10 and 1 M, respectively), the inhibition of taurine efflux was not different from that obtained by each antagonist individually. The inhibition of taurine efflux by adenosine receptor antagonists suggested that endogenous adenosine really contributes to swellingdependent VSOAC activation. The next step was to assess the source of adenosine. Adenosine could directly originate from the cells or by hydrolysis of released ATP. Accordingly, we used ␣␤MeADP, a potent blocker of 5Ј-ectonucleotidase, the ectoenzyme that hydrolyses AMP to adenosine (15,19). ␣␤MeADP inhibited in a dose-dependent fashion the taurine efflux elicited by the hypotonic shock (Fig. 2). Maximal effect was 37%, and half-effective concentration was 510 nM.
We previously showed that extracellular ATP application results in a strong potentiation of hypotonic shock effect (11). We asked whether exogenous ATP also acts through the conversion to adenosine. Therefore, we applied 100 M ATP in the 68% HS with or without ␣␤MeADP. Interestingly, ␣␤MeADP strongly reduced the taurine efflux elicited by the combined stimulation with ATP and hypotonic shock (Fig. 3). Actually, ␣␤MeADP inhibition seems to result from complete block of ATP effect in addition to the expected partial inhibition of the efflux induced by the hypotonic shock alone. Extracellular application of adenosine also potentiated swelling-induced taurine release as ATP (Fig. 3). Nevertheless, under these conditions, ␣␤MeADP was completely ineffective. Previous studies suggested that extracellular adenosine and ATP act in 9HTEoϪ and other airway cell lines by increasing intracellular Ca 2ϩ through a phospholipase C (PLC)-based mechanism (20). If the hypotonic shock induced ATP and adenosine release, a PLC inhibitor should reduce the swellinginduced taurine efflux. We used U73122, an amino steroid that is able to potently block PLC when used at concentrations in the low M range (21). U73122 (2 M) reduced the peak of taurine efflux by 40% (Fig. 4A).
We reasoned that if adenosine were the endogenous agonist that mediates the effect of hypotonic shock on taurine channels, adenosine deaminase (ADA), which converts adenosine to inosine, should have an inhibitory effect. ADA caused a significant decrease of the swelling-induced taurine efflux, although smaller than that of ␣␤MeADP, DPCPX, and DMPX (Fig. 4B).
Nucleosides are taken up by the cells through specific transporters, some of which are sensitive to inhibitors like nitrobenzylthioinosine (NBMPR) and dipyridamole (22,23). We applied these compounds during the hypotonic shock to assess if they had an effect. Dipyridamole significantly increased taurine efflux under hyposmotic conditions (Fig. 5A). NBMPR was instead almost ineffective (Fig. 5B).
Previous experiments revealed that a mild hypotonic shock (83% HS) causes only a slight activation of the Cl Ϫ channels. However, this effect can be strongly potentiated by Ca 2ϩ -elevating agents such as ATP and ionomycin to obtain a channel activation comparable with that achieved with a stronger hypotonic shock (11). Fig. 6 shows that adenosine also acts by synergistically increasing the volume-sensitive taurine efflux. CGS21680, an agonist selective for A 2A adenosine receptors (24), also increased taurine efflux (n ϭ 3, not shown). Interestingly, ␣␤MeADP did not inhibit adenosine effect but strongly antagonized that of ATP (Fig. 6). We also evaluated other nucleotides to clarify the mechanisms through which ATP affects channel activity. UTP and 2MeSATP were slightly active. In both cases ␣␤MeADP was ineffective. We tested AMP-PNP because it has been often used as a nonhydrolyzable ATP analog. Accordingly, this compound had to be ineffective if ATP acts through the conversion to adenosine. Unexpectedly, AMP-PNP was active, although less than ATP, and its activity was strongly reduced by ␣␤MeADP (Fig. 6).
Previous attempts to measure ATP release from 9HTEoϪ cells by the luciferin-luciferase assay were unsuccessful (11). In the present work, we have therefore utilized an HPLC-based approach. We reasoned that this method, by revealing the presence of ATP catabolites, would have allowed detection of ATP release even in the presence of a rapid extracellular hy- drolysis. Accordingly, 9HTEoϪ cells were loaded with [ 3 H]AMP for 1 h. After this period, analysis of cell lysates revealed that 53.7 Ϯ 2.3% (n ϭ 5) of the radioactivity was due to labeled ATP. ADP, AMP, adenosine, and inosine accounted for 22.1 Ϯ 1.1%, 8.6 Ϯ 0.9%, 3.8 Ϯ 0.4%, and 7.8 Ϯ 0.5% total accumulated radioactivity, respectively. It is probable that [ 3 H]AMP enter the cells as adenosine by the action of ectonucleotidase and adenosine transporters. Indeed, cell loading was strongly inhibited by dipyridamole or ␣␤MeADP (not shown). Under isotonic conditions, cells loaded with [ 3 H]AMP released small amounts of radioactivity (Fig. 7A). Upon applying the hypotonic solution (68% HS), a small but significant increase of released radioactivity was observed in the first 10 min (Fig. 7,  A and B). A milder hypotonic shock (83% HS) was instead without effect (not shown). We also wanted to assess the effect of intracellular Ca 2ϩ elevation alone or in combination with the hypotonic shock. We found that ionomycin elicited per se an increased radioactivity release from [ 3 H]AMP-loaded cells (Fig.  7B). Stimulation with ionomycin plus hypotonic shock (68% HS) induced a much higher efflux (Fig. 7B). Indeed, the effect was larger than the sum of the single responses to ionomycin and hypotonic shock.
HPLC analysis revealed that hypotonic shock, in the first 10 min of stimulation, caused the release of radioactive species whose elution times corresponded to those of adenosine, ADP, and ATP (Fig. 8, B and E). These peaks were absent in isotonic conditions (Fig. 8, A and C). On the other hand, supernatants of cells stimulated with ionomycin did not show a significant ATP, ADP, and adenosine release but a marked peak corresponding to inosine (Fig. 8D). When cells were treated with ionomycin plus the hypotonic shock (68% HS), supernatants had a higher content of ATP with respect to that observed with only the hypotonic shock ( Fig. 8F; p Ͻ 0.05). Adenosine was also significantly increased (p Ͻ 0.01). Interestingly, when hypotonic shock (with or without ionomycin) was applied in the presence of ␣␤MeADP, the adenosine peak was strongly reduced, whereas AMP, which was always undetectable in the other experimental conditions, was clearly visible (Fig. 8, E and F).
Experiments performed using dipyridamole and ADA (Figs. 4 and 5) showed changes in VSOAC activity that we interpreted as due to changes in extracellular adenosine concentration. To validate this interpretation we removed dipyridamole from the efflux medium or, alternatively, we added exogenous ADA (0.2 units/ml). As expected, both procedures decreased the radioactive adenosine in the extracellular medium. Removal of dipyridamole reduced the adenosine peak from 1061 Ϯ 133 to 585 Ϯ 20 cpm (p Ͻ 0.05; not shown). On the other hand, application of ADA completely abolished the adenosine peak and caused an increase of inosine from 128 Ϯ 68 to 1401 Ϯ 144 cpm (p Ͻ 0.01).
Our results suggest that 9HTEoϪ cells possess ectoenzymes able to degrade ATP. To directly demonstrate this process, cells were incubated for variable times with ATP (100 M). The supernatants were then analyzed by HPLC. After 1 h (not shown) and, more markedly, after 2 h (Fig. 9B), the peak corresponding to ATP decreased, and additional peaks appeared. These peaks corresponded to ADP, AMP, adenosine, and inosine. The process of ATP hydrolysis did not occur when ATP was incubated in the absence of cells. When ␣␤MeADP was included in the extracellular solution, the adenosine peak was abolished, whereas that of AMP was significantly increased (not shown). AMP-PNP was able to activate VSOAC in a ␣␤MeADP-sensitive way (Fig. 6). This result suggested that this compound can be converted to adenosine. Accordingly, we incubated the cells with this compound. HPLC analysis revealed the presence of a contaminant in the AMP-PNP stock solution (see Fig. 9C). According to the manufacturer instructions, this compound should be AMP-PNP after the loss of a phosphate moiety (i.e. ADP-NH 2 ). After incubation with 9HTEoϪ cells, the main and the secondary peaks were decreased, whereas a peak corresponding to adenosine was clearly evident (Fig. 9D). Incubation without the cells did not change the pattern of AMP-PNP-related peaks. Also in this case, ␣␤MeADP decreased the adenosine peak and increased the AMP signal (not shown).

DISCUSSION
Cells are able to face a hypotonic shock by activating ion channels (VSOAC) permeable to Cl Ϫ and organic osmolytes. The resulting efflux of osmolytes and water restores the original cell volume. The mechanisms responsible for VSOAC activation are unknown, although protein phosphorylation, membrane stretch, and changes in cytoskeletal organization have been proposed (25). Recently, it has been postulated that hypotonically induced ATP release represents an autocrine stimulus for VSOAC activation (12,13). Our results indicate that the hypotonic shock elicits an ATP release also in 9HTEoϪ tracheal epithelial cells. ATP is then hydrolyzed to adenosine by specific ectoenzymes. However, in our cells, adenosine and not ATP is the agonist responsible for VSOAC activation. Several evidences support this scheme. First, taurine efflux is significantly reduced by adenosine receptor antagonists such as DPCPX and DMPX. These compounds act with half-effective concentrations, consistent with an inhibition at the level of adenosine receptors (17,18). It is not clear which adenosine receptor is involved in the regulation of volume-sensitive Cl Ϫ channels in 9HTEoϪ cells. The inhibition by DMPX and the activation by CGS21680 suggests the involvement of the A 2A type (24). However, DPCPX, a selective A 1 antagonist, has an unusually low IC 50 value for an A 2A receptor (24). Future studies with a series of agonists and antagonists could elucidate this point. The 5Ј-ectonucleotidase inhibitor ␣␤MeADP, which blocks the conversion of AMP to adenosine (14,15,19), reduces the swelling-induced taurine release to an extent similar to adenosine receptor antagonists. A comparable reduction is also obtained by using the PLC inhibitor U73122. This result is consistent with the finding that exogenous adenosine stimulates PLC with consequent intracellular Ca 2ϩ increase in airway epithelial cell lines (20). Our results are also consistent with previous patch-clamp studies in which we found that stimulation of adenosine receptors results in VSOAC activation (26). Another indirect indication of adenosine involvement is the finding that nucleoside transporter inhibitors increase and that ADA, conversely, decreases taurine efflux. Such results can be explained by hypothesizing local variations of adenosine concentrations sensed by adenosine receptors. Indeed, dipyridamole would increase the availability of extracellular adenosine by blocking the reuptake. The stronger effect of dipyridamole compared with NBMPR can be explained by its ability to also inhibit NBMPR-insensitive transporters (22,23). On the other hand, ADA probably decreases taurine efflux by degrading adenosine to inosine, thus lowering the adenosine concentration, which is sensed by adenosine receptors. ADA-dependent inhibition is smaller than that elicited by other treatments (DPCPX, DMPX, ␣␤MeADP, U73122). This could be because of the difficulty encountered by a large enzyme to reach the sites on the plasma membrane where ATP release and hydrolysis occur or because the rate of adenosine production from ATP is faster than the rate of adenosine deamination to inosine.
To further verify our hypothesis, we applied extracellular ATP to assess whether it mimics the effect of the endogenous ATP. We found that ATP potentiates the effect of mild and strong hypotonic shocks (83% HS and 68% HS, respectively) and that this effect is completely blocked by ␣␤MeADP. Such a result suggests that exogenous ATP is degraded to AMP by ectoATPase and ectoADPase and then to adenosine by 5Ј-ectonucleotidase. Adenosine then activates VSOAC by interacting with adenosine receptors. Indeed, the inhibition by ␣␤MeADP can be bypassed by application of exogenous adenosine. In support of this model is the finding that extracellular UTP, which is not able to activate adenosine receptors, is a poor agonist of VSOAC. Surprisingly, AMP-PNP, which is often used as a hydrolysis-resistant ATP analog, activates taurine efflux. Furthermore, its effect can be blocked by ␣␤MeADP. Our experiments, however, demonstrate that incubation of 9HTEoϪ with AMP-PNP results in the appearance of adenosine in the cell supernatant. This process is probably due to hydrolysis because it is abolished by ␣␤MeADP. Our observation is consistent with recent findings that AMP-PNP can be actually hydrolyzed by extracellular ATPases (27).
Interestingly, other investigators have recently found that extracellular ATP may act in hypocampal slices by rapid conversion to adenosine and stimulation of adenosine receptors (27). These authors have proposed a "channeling" model to explain the process. According to this model, the enzymes responsible for ATP degradation, and adenosine receptors are closely clustered. This organization ensures that the product of each reaction goes directly to the next enzyme in the cascade. This model, termed preferential substrate delivery, has been also proposed to explain the kinetics of extracellular ATP hydrolysis in endothelial cells (15). The close organization of ectoenzymes and receptors and the consequent channeling of the compounds through the degrading cascade would account for the poor effect of large molecules like ADA. It has been reported that ectonucleotidases are clustered in caveolae, i.e. small invaginations of plasma membrane (28). This organization could also contribute to confine ATP and its catabolites so that ectoenzymes and purinergic receptors would sense local and not bulk concentrations. It is interesting to observe that our results clarify previously unresolved observations done with patch-clamp experiments on 9HTEoϪ cells (26). Actually, we found that the activation of volume-sensitive Cl Ϫ channels by extracellular ATP could be inhibited by an adenosine receptor antagonist (DMPX). We may now conclude that this surprising behavior was because of the hydrolysis of ATP to adenosine, which is the actual agonist. Such a conclusion reveals that the mechanism of ATP hydrolysis by ectoenzymes might be relevant even during patch-clamp experiments in which cells are continuously perfused with fresh solution.
We have demonstrated by an HPLC-based approach that ATP is indeed released during the hypotonic shock (68% HS) in 9HTEoϪ tracheal epithelial cells. Previous attempts to detect ATP by the luciferase assay were unsuccessful (11). We have no explanation for these previous negative results. It is possible that rapid nucleotide hydrolysis and the mechanism of preferential substrate delivery might reduce the amount of ATP that diffuses from the cells into the bulk solution.
Interestingly, the ATP release does not occur when the milder hypotonic shock (83% HS) is applied. This behavior explains, at least in part, the quite different channel activity observed with the two hypotonic solutions. Indeed, with 68% HS, the taurine efflux is so fast that more than 50% of intracellular taurine is released in the first 5 min of stimulation (Fig. 1). On the contrary, with 83% HS, only 3% of taurine is released (Fig. 6). In other words, the relationship between the efflux and the extracellular osmolality is not linear, but a threshold for activation exists below 250 mosmol/kg. This threshold is probably because of the activation of ATP release only with strong hypotonic shocks. Actually, when exogenous adenosine is provided under mild hypotonic conditions (Fig. 6), the taurine efflux increases to levels comparable with those attained with 68% HS. Intracellular Ca 2ϩ increase by application of ionomycin in isotonic conditions did not result in ATP efflux but caused a large release of inosine. However, combined stimulation with hypotonic solution (68% HS) and ionomycin elicited a stronger release of ATP if compared with hypotonic shock alone. In parallel, extracellular adenosine was also increased. ␣␤MeADP markedly decreased adenosine and caused the appearance of AMP in hypotonic conditions. This behavior suggests that adenosine is not directly released by the cells but probably arises from ATP hydrolysis. Inosine was not lowered by ␣␤MeADP, thus indicating that it does not derive from adenosine. Actually, the effect of ionomycin in isotonic conditions suggests that there is a separate Ca 2ϩ -dependent mechanism for inosine release.
Our results do not match exactly those of other investigators (12). First of all, we find that adenosine and not ATP is the autocrine agonist of VSOAC during the hypotonic shock. Another difference lies in the degree of channel activation, which can be accounted for by the autocrine signaling. Actually, Wang et al. (12) reported that treating hepatoma cells with ATPdegrading enzymes like apyrase completely blocks the swelling-induced chloride channel activation. In our experiments, the various treatments only partially inhibited the effect of the stronger hypotonic shock. It appears therefore that in 9HTEoϪ cells, adenosine and ATP release are not essential for swellinginduced activation of VSOAC. However, the release of these substances appears to play an important modulatory role. Our findings suggest that important tissue-specific differences might exist in the mechanism of VSOAC activation.
In conclusion, our experiments reveal that hypotonic shock induces a release of ATP from airway epithelial cells. ATP is then converted on the extracellular side of the membrane to adenosine, which binds to adenosine receptors to increase the activity of volume-sensitive chloride channels. This autocrine mechanism, which probably affects the speed and extent of regulatory volume decrease (29), is likely modulated by intracellular Ca 2ϩ . The regulation of ATP release by cell volume and Ca 2ϩ is probably important in the process of transepithelial chloride transport. Indeed, extracellular ATP represents a strong stimulus for chloride secretion in airway epithelia (30). Several agonists, including ATP, induce an intracellular Ca 2ϩ increase in airway epithelial cells (30 -32). This response could elicit intracellular ATP release, which would further stimulate the cells in an autocrine/paracrine fashion. The sensitivity of the ATP release process to cell volume could help to control the activation of chloride channels. Indeed, excessive Cl Ϫ channel activity and Cl Ϫ secretion would cause cell shrinkage. This should stop the release of ATP, which otherwise would continue to stimulate Cl Ϫ channels.