ATP release during cell swelling activates a Ca2+-dependent Cl− current by autocrine mechanism in mouse hippocampal microglia

Microglia cells, resident immune cells of the brain, survey brain parenchyma by dynamically extending and retracting their processes. Cl− channels, activated in the cellular response to stretch/swelling, take part in several functions deeply connected with microglia physiology, including cell shape changes, proliferation, differentiation and migration. However, the molecular identity and functional properties of these Cl− channels are largely unknown. We investigated the properties of swelling-activated currents in microglial from acute hippocampal slices of Cx3cr1+/GFP mice by whole-cell patch-clamp and imaging techniques. The exposure of cells to a mild hypotonic medium, caused an outward rectifying current, developing in 5–10 minutes and reverting upon stimulus washout. This current, required for microglia ability to extend processes towards a damage signal, was carried mainly by Cl− ions and dependent on intracellular Ca2+. Moreover, it involved swelling-induced ATP release. We identified a purine-dependent mechanism, likely constituting an amplification pathway of current activation: under hypotonic conditions, ATP release triggered the Ca2+-dependent activation of anionic channels by autocrine purine receptors stimulation. Our study on native microglia describes for the first time the functional properties of stretch/swelling-activated currents, representing a key element in microglia ability to monitor the brain parenchyma.

displaying outward rectification 6,13 ; in addition current can be activated by decrease in ionic strength or intracellular stimuli 14,15 . Although extensively characterized by electrophysiology and pharmacology, the molecular identity of volume activated anionic channels is not yet fully clarified 12 . Recently, it has been proposed that LRRC8 is essential component of volume-regulated Cl − channels, while several unrelated molecules have been previously involved, including bestrophins and TMEM16 proteins 16 . In addition, membrane stretch can result in the activation of pannexin hemichannels and maxi anion channels 17,18 .
Importantly, volume regulated channels are permeable to organic anions 6 and together with pannexins and maxi anion channels, are readily gated in response to hypotonic stress, constituting a preferential path for ATP efflux upon cell swelling 18 . Due to the role for ATP as paracrine and autocrine mediator, all the mechanisms by which intracellular nucleotides are exported to extracellular compartment deserve elucidation. This is particularly relevant in microglia, given the central role of ATP in microglia biology 19 and the possibility of influencing neuronal activity through purine release. Aberrations in such functions are believed to underlie many disease states in the brain, as swell-activated anion channel can be involved in the release of glutamate after a stroke or trauma exacerbating excitotoxic damage and causing neuronal cell death 14,20 . Thus, the relationship between changes in cell structure and chloride permeability could be relevant for microglia behaviour in physiological and pathological contexts.
Volume activated Cl − current has been characterized in rat cultured microglia 7,8,14 as well as in microglia cell lines 21 . However, although largely used, these reduced preparations cannot be considered as an exhaustive model of microglia as they cannot tell much about the modifications of microglia physiological properties, arising from tissue interactions 22 . Here, we report for the first time the expression of a volume activated current in microglia cells in acute murine brain slices. In addition, using a combination of patch-clamp technique and genetically encoded sensors for the analysis of changes in intracellular concentration of Cl − and ATP in brain slices and cultured cell lines, we determined the physiological and pharmacological properties of swell-activated currents in microglia. We report that these currents are characterized by Cl − selectivity, Ca 2+ -dependency and autocrine modulation by purines, highlighting the importance of volume activated ATP release as a potential signaling pathway triggered by microglia.

Swelling-induced chloride currents activation in hippocampal microglia. Membrane currents
were recorded from microglia cells in CA1 stratum radiatum of acute hippocampal slices (P10-P25). When these cells, usually displaying a limited array of voltage dependent currents 23 , were exposed to a mild hypotonic medium (8-10% dilution), we observed the time-dependent activation of an outward rectifying current ( Fig. 1A and B), which was absent when slices were maintained in isotonic conditions ( Supplementary Fig. S1). The voltage dependence of the swelling-activated current (I Swell ) was investigated by applying a series of voltage steps from a holding potential of −70 mV. The current, associated with a progressive increase in the leakage current at −70 mV (Fig. 1A), displayed a mild outward rectification, but not the time dependent inactivation at positive potentials, typical of swelling-induced currents 6 . The amplitude of I Swell increased over time, reaching a plateau within 10-12 minutes (Fig. 1B) and disappeared upon stimulus washout, when slices were perfused with normotonic extracellular solution (n = 6; Fig. 1A and B; p < 0.005 after 15 min wash). Swelling-activated currents were inhibited by anion channel blockers. In particular, the acute application of flufenamic acid (FFA, 200 μM; n = 6, p < 0.05), indanyloxyacetic acid 94 (IAA-94, 500 μM; n = 5, p < 0.05) or diisothiocyanatostilbene-2,2′-disolfonic acid (DIDS, 150 μM; n = 4, p < 0.05) strongly reduced the amplitude of I Swell (Fig. 1C). The reversal potential of I Swell was accurately determined in a separate set of experiments, where hypotonic medium was applied acutely after few minutes of whole cell dialysis, to allow a complete equilibrium of intracellular solution ( Fig. 1D and E). In these experiments, hypotonic stimulation activated in microglia cells the typical outward rectifying current, reverting at −0.2 ± 2.0 mV (n = 7; Fig. 1E), showing the typical time-dependent activation (not shown) and disappearing after washout to normotonic medium (Fig. 1D).
To establish the ion selectivity of the swelling-activated current, intracellular Cl − concentration was decreased to 4 mM by substitution with gluconate ( Fig. 2a and b). The use of K-gluconate based pipette solution caused a shift of the current reversal potential to more negative values (E rev = −51.0 ± 7.9 mV with K-gluconate; p < 0.01; n = 4 Fig. 2b). Recordings performed with a N-methyl-D-glucamine based intracellular solution did not show a depolarizing shift of I Swell reversal potential (E rev = −7.8 ± 0.8 mV; n = 4; p < 0.001; data not shown), as expected in case of a relevant cationic component. These observations indicate that swelling-activated current are mainly based on Cl − permeability.
To directly monitor the Cl − flux caused by the activation of swelling activated currents, a set of experiments was performed in cultured BV-2 cells transfected with a fluorescent Cl − Sensor, a protein indicator that allows a ratiometric and not invasive analysis of intracellular chloride concentration 24,25 . When exposed to a hypotonic stimulus, BV-2 cells displayed a swelling induced current, showing the typical outward rectification ( Supplementary Fig. S2), Cl − permeability (Fig. 2c) and sensitivity to FFA ( Supplementary Fig. S2). To visualize Cl − flux, Cl-Sensor expressing BV-2 cells were exposed to a hypotonic extracellular medium designed to simultaneously activate I Swell and depolarize membrane potential (Hypotonic/High K + solution), in order to change the electrochemical gradient for Cl − and favor its entry into the cells. In this condition, we observed an increase in the fluorescence ratio, indicating Cl − influx, which was reverted upon shifting to a low Cl − (6 mM) extracellular solution (Fig. 2d). These data indicate that in BV-2 microglia cells, hypotonic stimulation activates a membrane conductance carried by Cl − ions, whose flux is driven by electrochemical Cl − gradient.
Together, these results demonstrate that hypotonic stimulus induces in microglia cells the activation of membrane currents through Cl − permeable channels. Swelling-activated currents require microglia ATP release. Swelling activated anionic channels are present in a large variety of cells, showing different mechanisms of activation and control. In particular, intracellular ATP is known to be necessary for the activation of volume-regulated anionic currents 11,14 . Thus, to determine whether ATP is required for the activation of I Swell , we exposed hippocampal microglia to hypotonic stimulus  Consistently, when microglia cells were acutely treated with hypotonic medium after intracellular dialysis, I Swell was completely abolished (n = 6; Fig. 3b). Each experimental set was compared with respective internal controls (recorded in the same experimental days with Mg-ATP in the pipette solution). CTR bar represents mean I Swell amplitude of all internal controls (2 mM Mg-ATP in the pipette solution, n = 48; blue bar). Na-ATP (2 mM, orange; n = 7 vs n = 10; p > 0.05, t-test), ATPγS (2 mM, violet; n = 11 vs n = 6; p > 0.05, t-test), Na-ADP (2 mM, pink; n = 12 vs n = 15; p > 0.05, t-test), adenosine (ADO, 2 mM, white; n = 10 vs n = 18; ***, p < 0.001, t-test) or with an ATP-free solution (gray; n = 23 vs n = 17; ***, p < 0.001, t-test).
ATP is the major donor of phosphoric group during phosphorylation processes, thus the decrease of swelling-induced current in the absence of ATP might result from "run-down" processes 26 . To investigate if ATP-dependent phosphorylation processes were required for I Swell activation, membrane currents were recorded with a pipette solution in which Mg-ATP was substituted by Na-ATP, unable to participate to classical reactions of phosphate group transfer 27 . In these conditions of disfavoring phosphorylations, hypotonic stimulus induced a swell-activated current similar to control (n = 9; p = 0.6; Fig. 3c). Similar results were obtained when intracellular ATP was substituted by ATPγS (2 mM), a slowly hydrolyzable ATP analog which prevents dephosphorylation, due to the formation of irreversibly thiophosphorylated residues (n = 21; p > 0.16; Fig. 3c). Remarkably, ATP could be also replaced by ADP. Indeed, in presence of ADP, I Swell occurred showing typical amplitude (n = 13; Fig. 3c), while when the pipette solution contained adenosine (ADO, n = 10; Fig. 3c), the current failed to be activated. Together, these results show that intracellular purines are required for I Swell activation, but their role is likely not linked to phosphorylation or ATPase reactions.
It is known that brain cells can extrude ATP or ADP in different ways including pannexin hemichannels 17, 28-30 . Thus, we hypothesized that in our experimental conditions, ATP could be released by microglia cells, favoring I Swell activation by triggering a purinergic pathway.
To verify this hypothesis we, first, ascertained purine release in primary cultured microglia under hypotonic conditions. On this purpose, we collected cell culture medium after hypotonic challenge and analyzed it by ultraperformance liquid chromatography. When microglia cultures were exposed for 5 minutes to hypotonic stimulus, the culture medium contained a submicromolar concentration of adenosine, which was not observed in unstimulated cultures (Fig. 4). Concentrations of ATP, ADP and AMP were unchanged by hypotonic stimulation (Fig. 4). These results show that microglia cells are able to release purines when exposed to swelling conditions. To demonstrate that microglia are able to release ATP in our experimental conditions, we took advantage of their ability to rearrange processes toward a source of ATP 31 (see also the legend to Supplementary Fig. S4). On this purpose, we analyzed the rearrangement of GFP-positive microglia processes around a recorded microglia cell under hypotonic stimulation (Fig. 5a), by tracking the movement of single processes 23 . Reconstructed processes trajectories showed that during I Swell activation, microglia processes move in the direction of the recorded cell under hypotonic stimulation, in an ATP concentration-dependent manner. Indeed, the mean displacement of surrounding processes indicated a rearrangement towards the swelled cell when microglia cells were dialyzed with a pipette solution containing 10 mM ATP and a null net movement in presence of 2 mM ATP (p < 0.001, t-test; Fig. 5a). Conversely, in normotonic conditions, increasing intracellular ATP was ineffective in attracting microglia processes towards the recorded cell (not shown).
Altogether, these data show that ATP is necessary for current activation and is released during I Swell activation raising the possibility of an extracellular site of action through purininergic receptors.

Involvement of hemichannels in microglia ATP release.
To investigate the mechanisms of ATP release from microglia during hypotonic stimulation, we used transfected BV2 cells expressing a FRET based ATP sensor (Fig. 6a). In this reduced system, we monitored the dynamics of ATP levels in BV2 cells during the application of an acute hypotonic stimulus. Figure 6a shows sequential images of the ratio of CFP and YFP channels at monitoring of ATP sensor fluorescence. Hypotonic stimulus induced a significant decrease in the YFP/CFP emission ratio, indicating a decrease of intracellular [ATP] in microglia cells (Fig. 6a). To verify if the observed decrease in cell fluorescence was due to a release of ATP during hypotonic challenge, FRET experiments were performed in presence of pannexin hemichannel blockers CBX 32 (100 μM) or probenecid 33 (500 μM and 1 mM). The application of these blockers prevented the hypotonic-induced decrease of cell fluorescence, while that of FFA (200 μM mM) was ineffective, indicating that the decrement of intracellular [ATP] was associated to ATP release through hemichannels ( Fig. 6a and b). A similar decrease in fluorescence ratio was observed in primary microglia cultures trasnsfected with ATP sensor (n = 5; not shown). These results suggest that pannexin1 hemichannels may be involved in ATP release during hypotonic stimulation 34,35 .
Since ATP removal prevented I Swell activation and hemichannels are associated to microglia purine release in hypotonic conditions, we used pannexin blockers to ascertain whether the activity of these conduits could interfere with hypotonic-induced current. When experiments were performed in the presence of carbenoxolone (CBX, 100 μM), I Swell amplitude was strongly reduced (n = 7; Fig. 6c). Consistently, recordings performed in presence of the selective pannexin antagonist, probenecid (500 μM and 1 mM), showed a dose-dependent reduction in current amplitude (Fig. 6c), supporting the view of hemichannels involvement in swell-induced current activation. On the other hand, gadolinium (500 μM) a blocker of maxi anion channels was ineffective (n = 3; p = 0.4; not shown).

Microglia swelling-activated current is [Ca 2+
] i -dependent. Among the transduction pathways controlled by purine P2Y receptors activation, we hypothesized that ATP might promote the activation of I Swell by inducing intracellular Ca 2+ increase in microglia cells. Indeed Ca 2+ -activated anionic channels, present in a large variety of cells, are involved in microglia response to cell swelling 36 .
Combined recordings of membrane currents and intracellular Ca 2+ in Fura2 loaded hippocampal microglia cells in slices, indicated that intracellular Ca 2+ increased during hypotonic stimulation, as shown by the increase in fluorescence ratio F 340 /F 380 (from 0.40 ± 0.03 to 0.50 ± 0.06 after 20 minutes of hypotonic stimulus; p = 0.03; Fig. 7a). Then, to highlight the role of Ca 2+ in the mechanisms of activation of swell-induced current, we changed the Ca 2+ buffering capacity of the pipette solution. On this purpose, EGTA was substituted by BAPTA (0.5, 5 e 30 mM). With BAPTA 0.5 mM, swell-activated current was similar to that recorded in control condition; while, at increasing BAPTA concentrations, I Swell amplitude was significantly reduced (5 mM), or failed to be activated by the hypoosmotic stimulus (30 mM ; Fig. 7b). These data demonstrate that the activation of swelling current requires at least minimal increase in [Ca 2+ ] i , which is avoided by the fast chelator BAPTA in the intracellular solution. However, [Ca 2+ ] i manipulation was not sufficient for I Swell activation in the absence of hypotonic stimulation (Fig. 7c). Indeed, in microglia cells recorded in presence of high intracellular Ca 2+ (1 µM), we did not observe the time dependent activation of the typical outward rectifying current, when kept in normotonic conditions (Fig. 7c), indicating that the sole presence of high intracellular Ca 2+ is unable to elicit I Swell . Consistently, in the presence of high intracellular Ca 2+ , hypotonic challenge elicited a current response, which was indistinguishable from control (Fig. 7c). Thus, the swelling current is Ca 2+ -dependent but Ca 2+ increase is not sufficient to activate it, likely constituting an amplificatory pathway or a precondition. To determine whether the source of Ca 2+ involved in current activation was intracellular, we performed experiments with a pipette solution containing thapsigargin (1 μM), in order to deplete intracellular Ca 2+ stores of the recorded microglia cell, limiting the effects on surrounding cells. To allow stores depletion, in these experiments we applied the hypotonic stimulus acutely, after 9 minutes of intracellular dialysis. In these experimental conditions, microglia recorded with thapsigargin showed reduced I Swell amplitude in respect to controls (Fig. 7d). On the other hand, when we exposed slices to hypotonic challenge in the nominal absence of external Ca 2+ , I Swell displayed the typical activation (data not shown; n = 5; p > 0.1 vs internal controls, n = 5; two-way ANOVA).
All together, these results indicate that ATP is released by microglia under hypotonic conditions through pannexin hemichannel and favors the Ca 2+ -dependent mechanisms of I Swell activation acting on purinergic receptors (Fig. 8).

Discussion
We report that hypotonic stimulation on microglia in acute hippocampal slices induces an ATP-and Ca 2+ -dependent chloride current. Membrane swell is also able to cause pannexin hemichannels-mediated ATP release that could amplify current activation through purinergic receptors and an increase of intracellular Ca 2+ (Fig. 8). Swelling-released ATP may constitute a signal for microglia process recruitment. Our results show that microglia in acute hippocampal slices of Cx3cr1 +/GFP mice display an outward rectifying current activated by hypotonic stimulation, similar to the current previously observed in microglia cultures 11,14 . To activate the swelling current in microglia cells, we stimulated slices chronically or acutely with a mild hypo-osmotic stimulus, establishing a slight osmolarity delta between intracellular and extracellular sides. Our conditions, avoiding the use of very strong hypotonic stimuli 7,14 , are suitable for the study of microglia currents activated by membrane swelling in slices. Several facts support the notion that I Swell is specifically due to the induction of an osmotic delta across microglia membrane, disfavoring the possibility of chemical stimulation due to unknown substances released from other cells in the slice 37 . Indeed, (i) in the chronic protocol, all the cells in the slices are allowed to rebalance before starting the recording; (ii) I Swell slowly activates after whole cell break through due to the transmembrane osmotic delta and (iii) disappears on return to normotonic condition; (iv) it can be activated also by hypertonic intracellular solution, while (v) it fails to develop when intracellular and extracellular solutions have similar osmolarity (Supplementary Fig. S1). Finally, I Swell could be activated in several cells sequentially recorded in the same slice, which was always kept in the same medium. , expressed as percentage of the current amplitude in internal controls (recorded in the same experimental days in untreated slices). Probenecid 500 μM: red; n = 13 vs n = 5; **p < 0.01, t-test. Probenecid 1 mM: green, n = 9 vs n = 5; **p < 0.01, t-test. CBX: blue; n = 6 vs n = 4; **p < 0.01, t-test. I Swell amplitude as in 5c. Statistics were performed on raw amplitude values.
In line with prevalent Cl − selectivity, the reversal potential of the swell-activated currents was close to 0 mV in solutions with symmetrical Cl − concentrations. The positive deviation observed in chronic experiments is likely due to the progressive dialysis with Cl-based solution during whole cell recording. Indeed, in experiment performed in acute conditions, after intracellular equilibration, I Swell reversal potential was coincident with GHK equation predictions. In addition, upon equimolar substitution of Cl − with gluconate in the intracellular solution, the expected negative shift in reversal potential was observed 11 . The deviation from predictions could be explained considering a significant permeability to gluconate, as reported by Gilbertson 38 in rat taste neurons (gluconate/Cl permeability ratio 0.2). In alternative, it could be considered the possible permeability to other physiologically relevant anions, including ATP 7,8,14,39 . It should be noted that the amplitude of I Swell was reduced in Cl − substituted solutions, suggesting sensitivity to the concentration of Cl − ions 8,11 .
Cl − selectivity was confirmed by imaging experiments in BV2 cells transfected with Cl-sensor 24, 25 , highlighting Cl − fluxes dependent on [Cl − ] e . In these experiments, BV2 cells were concomitantly depolarized to increase Cl − driving force under hypotonic stimulation.
Swelling-activated Cl − current had a dependence on intracellular ATP, as the current showed only a transient activation in the absence of ATP in the intracellular solution. Remarkably, a striking ATP dependency was observed after intracellular dialysis, when the application of an acute hypotonic stimulus was ineffective in the absence of intracellular ATP. The role of intracellular ATP in the regulation of chloride currents activated by mechanical membrane deformation has been already highlighted in microglia and other cell types and is generally described in terms of an ATP-dependent current run down 40 . However, ATP effect is apparently not dependent on ATP hydrolysis 11,14,[40][41][42] . Differently from what reported in rat carotid body cells 43 , in our conditions, intracellular Mg-ATP could be substituted by analogs, like Na-ATP, ATPɣS or ADP, impairing phosphate transfer to substrate. This indicates that ATP is not necessary as P i -donor as in phosphorylation-based regulation mechanisms 27 . Conversly, we hypothesize that ATP is released by microglia cells during swelling, contributing to I Swell activation in autocrine fashion 39,42,44 (Fig. 8), as proposed for the mechanisms leading to cell volume regulation in different cellular systems 45 and immune cells activation 46 .This conclusion is based on the absence of I Swell when slices are treated with apyrase, an enzyme that degrades ATP and ADP to AMP and free inorganic phosphate 47 .
The effects of ATP removal by apyrase support the conclusion that ATP (or ADP) binds to one or more extracellular sites, stimulating anionic channels. Consistently, the broad-spectrum purinergic receptor antagonists, suramin and PPADS, abolished I Swell , suggesting the involvement of purinergic signaling in current activation. It should be noted that these compounds, although currently used, are not selective and even potentially targeting volume-activated channels 35,48,49 . In particular, the effect of PPADS and suramin could also be ascribed to their inhibition of ectoATPases 50, 51 , potentially impairing ATP conversion to ADP and adenosine. The involvement of purinergic signaling is further supported by the inhibitory action of selective antagonists for P2Y1 receptor. Indeed, P2YRs could be activated by ATP released during cell swelling and favor I Swell activation by Ca 2+ -dependent mechanisms. Together, our data suggest that purinergic receptors are not required for I Swell activation, but participate to an amplificatory pathway (Fig. 8), as the effects exerted by P2YRs antagonists are only partial. However, it cannot be excluded that a specific extracellular purine-binding site is functionally associated to anionic channels activation.
A critical issue is the mechanism by which ATP is released by microglia cells. The release of ATP through volume regulated channels has been proposed long time ago in hepatoma 39 and endothelial cells 52 and more recently in astrocytes and Raw macrophages 44,53,54 . Our conclusion that ATP is released by microglia under hypotonic stimulation is based on severeal lines of evidence. First, we demonstrated purine release by ultraperformance liquid chromatography in cultured microglia. However, it is unlikely to observe ATP or ADP, which have been shown to be very unstable 55 and the accumulation of adenosine represents a highly likely proof of their release 50 . In addition, we directly measured ATP efflux induced by cell swelling with a FRET based cytoplasmic ATP-sensor 56 . Moreover, we obtained an independent proof of ATP release in slices, using the paradigm of ATP-induced microglia processes recruitment 23,31 . Indeed, our results show that the activation of I Swell in hypotonic conditions is associated to ATP dose dependent recruitment of microglia processes. Although indirect, this assay has the advantage to highlight the functional effect of ATP release in the same conditions used for swelling activated currents. However, the nature of the channels or transporters involved in swelling-induced ATP release is unknown. Pannexin hemichannels mediate ATP release in astrocytes 57 and neurons 17 . Moreover these channels are mechanosensitive 58 and allow ATP release upon swelling 17 . Data from BV-2 cells support the involvement of pannexin hemichannels in microglia ATP release under hypotonic stimulation. Indeed, the decrease in intracellular ATP concentration is abolished both by CBX and probenecid, supporting pannexin mediated ATP release. This is consistent with results concerning pannexin involvement in I Swell activation, which is prevented by both blockers in our conditions. On the other hand, gadolinium, a blocker of maxi anion channels, had no effect on I Swell . Based on these results, we hypothesized that in response to hypotonic stimulus, ATP or ADP, could leak from microglia cell through pannexins 59,60 (Fig. 8). It should be noted that the interpretation of CBX data is limited by the unselectivity of this compound 61 . Indeed, CBX might block directly both volume-activated currents [61][62][63] and connexin hemichannels 64 . However, the inhibition of I Swell with the selective pannexin blocker probenecid supports the involvement of pannexins in the mechanism leading to I Swell activation.
Another relevant point concerns the role of Ca 2+ in I Swell activation. Our data strongly support a Ca 2+ -dependent mechanism, as we demonstrated that hypotonic stimulation induces [Ca 2+ ] i increase and that I Swell is inhibited by intracellular Ca 2+ buffering. Regarding the source of calcium necessary for the activation of the current, we highlighted a role for intracellular Ca 2+ . Indeed, we observed that I Swell is not significantly affected by removal of extracellular Ca 2+ , but shows a significant reduction as a result of intracellular stores depletion by thapsigargin. Consistently with pharmacological evidences on purinergic receptors, we can argue the involvement of P2Y receptors in I Swell amplification through intracellular Ca 2+ increase. However, we found some discrepancy between the amplitude reduction observed with thapsigargin and the block caused by 30 mM intracellular BAPTA. It is possible that the full block could arise from direct effects of BAPTA on I Swell , similar to those reported in other cell types. Indeed, Cl − channels are described as highly sensitive to extracellular BAPTA 65 , although in a dose range and experimental conditions very different from those here reported; in addition, a subpopulation of Cl − channels in brown adipocytes is blocked by intracellular BAPTA 66 . On the other hand, reasoning that the action of high BAPTA is genuinely due to its Ca 2+ chelating capacity, the strict dependency of I Swell activation on Ca 2+ buffering would suggest the involvement of a high affinity Ca 2+ -binding site in the mechanism of channel activation 67 . It should be noted that, increasing free intracellular Ca 2+ was not sufficient to induce I Swell activation, in the absence of the hypotonic stimulus. Thus, Ca 2+ increase may represent a permissive factor, favoring current activation 16,43 , rather by modulating volume-sensitive anion channels 16,36,68 , than by activating calcium-sensitive ones 16,69,70 . In addition, Ca 2+ rise could amplify I Swell activation by favoring pannexin-mediated ATP release 71,72 .
Altogether, our data demonstrate that swelling-activated chloride current in native microglia cells is dependent on intracellular Ca 2+ and associated to release of purines, likely causing current amplification by an autocrine pathway (Fig. 8).
Furthermore, our study raises the possibility that membrane swell and the consequent ATP release are used by microglia cells to send signals to other cells, providing a novel mechanism by which microglia could communicate and potentially affect neuronal activity.
Voltage-clamp recordings were performed at RT using Axopatch 200 A amplifier (Molecular Devices). Currents were filtered at 2 kHz, digitized (10 kHz) and collected using Clampex 10 (Molecular Devices); off-line was performed using Clampfit 10 (Molecular Devices). Recordings were completed in 5-6 hours after slicing and performed at least 25 μm under slice surface, to avoid microglia activated during slicing.
Unless otherwise stated, I Swell was evoked exposing slices chronically to hypoosmotic ACSF (8-10% dilution of standard ACSF; in order to establish 20 mOsm Δ between pipette and extracellular solutions). Slices were placed after cutting in hypoosmotic medium and perfused with it during recordings. In specific experiments, hypoosmotic stimulus was applied acutely after 9 minutes of recording.
Resting membrane potential and membrane capacitance were monitored along the experiments. The current/ voltage (I/V) relationship were determined applying voltage steps (50 ms) from −170 to +70 mV (HP = −70 mV; ΔV = 10 mV) every 3 minutes. I Swell amplitude was measured by subtracting the current amplitude measured just after membrane rupture (chronic protocol; 12 min of recording) or that recorded just before hypotonic application (acute protocol; 18 min hypotonic stimulation). To generate I/V relationships, leak current was added to the recorded current amplitude at each time point. Although not quantified, cell swelling was observed during hypoosmotic challenge.
Tracking analysis of single microglia processes. Microglia processes tracking was performed as previously described 23 . Briefly, images were processed using ImageJ software and data analyzed with ImageJ and Origin 8 (OriginLab Co.) software to obtain quantitative distributions of track parameters The ability of the processes to be attracted towards the pipette tip with different ATP-concentration was defined as ΔR = R f − R 0 , where R 0 and R f are the distances from the initial and last sample point of the track i respectively. Negative ΔR values reflect directed process movement to the pipette tip. ATP-dependent displacement was expressed as a function of radial distances of microglia processes from the pipette tip in normotonic and hypotonic medium conditions. FRET analysis of cytosolic ATP concentration. To monitor single-cell cytosolic [ATP] changes, primary microglia and BV-2 mouse microglia cells were transfected with a plasmid encoding the ATP sensor AT1.03 59 , using 2 μl Lipofectamine-2000 and 1 μg DNA per 250 μl of medium (50 μl OptiMem and 200 μl of the cell transfection medium). Incubation medium was removed leaving 200 μl per well and 50 μl of DNA/Lipofectamin mixure was added into each well. Cells were maintained in CO 2 incubator (37 °C, 2-2.5 h), then medium was discarded and cells washed twice with fresh medium. Fluorescence measurements were performed 24 hours after transfection 59 using inverted microscope Olympus IX73 (Olympus Europe) with a 40× objective. Fluorescence of AT1.03 was excited at 436 nm with a xenon lamp (Lumen 200PRO, Prior) using a filter wheel (X-light, CrestOptics); emission was monitored at 527 nm and 475 nm using an emission splitting system (DV2, Photometrics). Images were acquired by cooled CCD camera (CoolSNAP Myo, Photometrics). Imaging data were collected and analyzed using MetaFluor 6.1 (Molecular Device, USA). Cells were treated with CBX (100 μM), probenecid (500 μM and 1 mM) or FFA (200 μM) for 30 minutes before FRET experiments and during the entire time lapse.
Ultra-performance/pressure Liquid Chromatography. Samples were collected from multiwell plated primary microglia cultures (1*10 5 cells/well); after 24 hours after plating, the culture medium was removed and cells were equilibrated for 16 hours in NES-340 (NES adjusted to 340 mOsm by adding NaCl), isotonic to the culture medium. Cells were, then, exposed to hypotonic stimulus, by diluting NES-340 to 255 mOsm with water. An equal volume of NES-340 was added to control microglia cell, in order to treat cells similarly, maintaining osmolarity unaltered. From each sample, the total volume was collected and centrifuged at 14000 g for 30 minutes at 4 °C and 380 μl of supernatants were immediately stored at −80 °C. Samples were then concentrated by freeze-drying, suspended in 50 μl of ultrapure water, centrifuged at 14000 g (30 minutes, 4 °C) and 10 μl of supernatant were injected directly onto UPLC column. Chromatographic determination and quantification of ATP and adenosine nucleotides was performed on a Waters Acquity H-Class UPLC system (Waters, Milford, MA, USA), including quaternary solvent manager (QSM), sample manager with flow through needle system (FTN), and photodiode array detector (DAD). Column temperature, 30 °C; flow rate 0.3 ml/min; peaks detected at 260 nm. The method employs a reverse-phase column with the use of phosphate buffer in mobile phase to enhance retention and separation of the compounds of interest 75,76 . Statistical analysis. Paired and unpaired t-test, one-way and two-way ANOVA were used for parametrical data; for multiple comparisons (Holm-Sidak method), multiplicity-adjusted p values are indicated in figures when appropriate.
Drugs. Drugs and reagents used were purchased from Sigma Aldrich, Life Technologies, Ascent Scientific and Invitrogen.