Stretch‐induced Ca²⁺ independent ATP release in hippocampal astrocytes

Abstract

Key points

• Similar to neurons, astrocytes actively participate in synaptic transmission via releasing gliotransmitters.

• The Ca²⁺‐dependent release of gliotransmitters includes glutamate and ATP.

• Following an ‘on‐cell‐like’ mechanical stimulus to a single astrocyte, Ca²⁺ independent single, large, non‐quantal, ATP release occurs.

• Astrocytic ATP release is inhibited by either selective antagonist treatment or genetic knockdown of P2X7 receptor channels.

• Our work suggests that ATP can be released from astrocytes via two independent pathways in hippocampal astrocytes; in addition to the known Ca²⁺‐dependent vesicular release, larger non‐quantal ATP release depends on P2X7 channels following mechanical stretch.

Abstract

Astrocytic ATP release is essential for brain functions such as synaptic long‐term potentiation for learning and memory. However, whether and how ATP is released via exocytosis remains hotly debated. All previous studies of non‐vesicular ATP release have used indirect assays. By contrast, two recent studies report vesicular ATP release using more direct assays. In the present study, using patch clamped ‘ATP‐sniffer cells’, we re‐investigated astrocytic ATP release at single‐vesicle resolution in hippocampal astrocytes. Following an ‘on‐cell‐like’ mechanical stimulus of a single astrocyte, a Ca²⁺ independent single large non‐quantal ATP release occurred, in contrast to the Ca²⁺‐dependent multiple small quantal ATP release in a chromaffin cell. The mechanical stimulation‐induced ATP release from an astrocyte was inhibited by either exposure to a selective antagonist or genetic knockdown of P2X7 receptor channels. Functional P2X7 channels were expressed in astrocytes in hippocampal brain slices. Thus, in addition to small quantal ATP release, larger non‐quantal ATP release depends on P2X7 channels in astrocytes.

Key points

• Similar to neurons, astrocytes actively participate in synaptic transmission via releasing gliotransmitters.

• The Ca²⁺‐dependent release of gliotransmitters includes glutamate and ATP.

• Following an ‘on‐cell‐like’ mechanical stimulus to a single astrocyte, Ca²⁺ independent single, large, non‐quantal, ATP release occurs.

• Astrocytic ATP release is inhibited by either selective antagonist treatment or genetic knockdown of P2X7 receptor channels.

• Our work suggests that ATP can be released from astrocytes via two independent pathways in hippocampal astrocytes; in addition to the known Ca²⁺‐dependent vesicular release, larger non‐quantal ATP release depends on P2X7 channels following mechanical stretch.

Introduction

The astrocyte is proposed to be an active partner in the ‘tripartite synapse’, with important roles in the CNS under physiological and pathological conditions (Volterra & Meldolesi, 2005; Haydon & Carmignoto, 2006; Halassa et al. 2007). Astrocytes regulate neuronal activity by releasing gliotransmitters, which bind to their respective receptors on neurons, glia and vascular cells to modulate synaptic plasticity and oxygen supply (Fellin et al. 2004; Haydon & Carmignoto, 2006; Jourdain et al. 2007; Gordon et al. 2008). ATP is one of a few primary gliotransmitters and functions as a key extracellular messenger in the CNS (Salter & Hicks, 1994; Cotrina et al. 2000). Astrocytic ATP suppresses synaptic transmission (Zhang et al. 2003; Pascual et al. 2005), regulates breathing and sleep (Gourine et al. 2010; Blutstein & Haydon, 2013), and triggers cellular responses to trauma, ischaemia and neurodegenerative diseases in vivo (Parvathenani et al. 2003; Rossi et al. 2007).

Despite the importance of evoked ATP release from astrocytes, its mechanism remains controversial. Both non‐quantal release and quantal/vesicular release have been proposed (Stout et al. 2002; Pascual et al. 2005; Pangrsic et al. 2007; Zhang et al. 2007; Kang et al. 2008). Evidence supporting non‐quantal ATP release via open membrane pores such as connexin channels, pannexin channels, swelling‐activated anion channels and P2X7 receptor channels is mainly based on the indirect methods of dye uptake (Lucifer Yellow or YO‐PRO‐1) and the luciferase‐luciferin test (Ballerini et al. 1996; Cotrina et al. 1998; Darby et al. 2003; Anderson et al. 2004; Pelegrin & Surprenant, 2006; Suadicani et al. 2006; Kang et al. 2008). Although the molecular weights of these dyes are similar to ATP, their distinct molecular structures and charges make the results inconclusive. Evidence supporting Ca²⁺‐dependent quantal/vesicular ATP release comes from imaging experiments using cells preloaded with Mant‐ATP or quinacrine, or SNARE‐specific treatments (Coco et al. 2003; Pascual et al. 2005; Pangrsic et al. 2007; Zhang et al. 2007; Lalo et al. 2014; Lee et al. 2015; Gucek et al. 2016). In addition, the lysosome has been reported to be responsible for ATP release in astrocytes (Zhang et al. 2007). This conclusion was inferred from two independent findings: (i) astrocytic lysosomes undergo exocytosis based on TIRF imaging and (ii) ATP is stored in lysosomes, based on Mant‐ATP localization and the detection of ATP in isolated lysosomes (Zhang et al. 2007). However, ATP release via lysosome exocytosis was not directly recorded. Thus, direct ATP recording at the spatiotemporal resolution of single vesicles is necessary to determine whether ATP is released via vesicular exocytosis in astrocytes.

In the present study, we used whole‐cell patch clamped HEK293 cells, which express the ATP‐gated channel P2X4 with fast activation and slow desensitization (North, 2002), as ‘ATP sniffers’ to record ATP release at the single‐vesicle level. To determine whether lysosome exocytosis releases ATP, we used the most reliable method of stimulating astrocytes, mechanical stimulation (Mstim), and found that Mstim triggered not only lysosome exocytosis, but also ATP release. However, most ATP release by Mstim was dependent on P2X7 channels in hippocampal astrocytes.

Methods

Animals

The use and care of animals followed the guidelines of the Animal Care and Use Committee at Peking University (Beijing, China) who approved the protocols (IACUC # IMM‐ZhouZ‐11). For most of the experiments in cultured astrocytes, we used neonatal or postnatal 1‐day‐old Sprague–Dawley rats. IP3R2‐knockout mice (Petravicz et al. 2008) were also used in some of the experiments as specifically noted. For the experiments in brain slices, we used 2‐week‐old Sprague–Dawley rats and P2RX7‐EGFP transgenic mice of either sex (Gong et al. 2003).

Cell culture

Primary cultures of rat hippocampal astrocytes were prepared as described previously (Chen et al. 2005a), with minor modifications. Briefly, neonatal or 1‐day‐old Sprague–Dawley rats were anaesthetized by hypothermia and the brains were removed into ice‐cold dissection solution (pH 7.3) containing (in mm): 137 NaCl, 5.5 KCl, 0.2 Na₂HPO₄, 0.2 KH₂PO₄, 10 Hepes, 18 glucose and 22 sucrose. Hippocampi were dissociated and trypsinized for 10 min and then inactivated with Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (FBS; Gibco, Gaithersburg, MD, USA). Cells were plated in 25‐cm² culture flasks (Corning, New York, NY, USA) at 1–3 × 10⁶ mL^–1 and incubated in antibiotic‐free DMEM with 10% FBS in 5% CO₂ at 37°C. After 10–12 days, the astrocytes were purified, re‐plated on coverslips and used after 24 h. The cultured astrocytes were polygonal in shape under these conditions and expressed glial fibrillary acidic protein (GFAP) as in vivo.

HEK293 cells were cultured as described previously (Yu et al. 2004). Rat adrenal chromaffin cells were also cultured as described previously (Chen et al. 2005b).

Sniffer cell recording

HEK293A cells expressing ATP‐gated P2X4 receptors were used as sniffers. The pIRES2‐EGFP rP2X4 construct was a kind gift from Dr Stanko S. Stojilkovic (NICHD, Rockville, MD, USA). Briefly, 2 μg of pIRES2‐EGFP rP2X4 plasmid was transfected by VigoFect (Vigorous, Beijing, China) and HEK293A cells were cultured for an additional 1–2 days. Thirty minutes to 1 h before electrophysiological recording, transfected HEK293A cells were trypsinized and plated on coverslips on which astrocytes had grown. Cells were bathed in standard extracellular solution (SE) (in mm: 150 NaCl, 5 KCl, 2.5 CaCl₂, 1 MgCl₂, 10 Hepes and 10 glucose, pH adjusted to 7.4 by NaOH) at room temperature. For some experiments in Ca²⁺‐free extracellular solution, CaCl₂ was replaced with 1 mm EGTA. An Axopatch 200B amplifier and Clampex 8.0 software (Molecular Devices, Sunnyvale, CA, USA) were used to perform patch clamp experiments. The pipette solution contained (in mm): 140 KCl, 10 EGTA, 10 Hepes, pH adjusted to 7.4 with KOH. Voltage clamp in the whole‐cell configuration was used to record Mstim‐induced ATP release in astrocytes. All sniffer cells had an average capacitance of 10 pF and the membrane potential was held at −70 mV. Signals were low‐pass filtered at 1 kHz and digitized at 5 kHz. Especially for patch clamp of cultured astrocytes, cells were first trypsinized from flat and polygonal to a spherical shape. P2X4‐sniffer was confirmed by the P2X4 blocker TNP‐ATP (see Supporting information, Fig. S1). Consistent with previous studies (North & Surprenant, 2000; Bo et al. 2001), TNP‐ATP blocked the P2X4 currents by 33% at 2 μm and 94% at 100 μm, respectively.

Dopamine loading and amperometric recording

Astrocytes were loaded in a bath solution containing 70 mm dopamine (DA) for 45 min at 37°C (Zhang & Zhou, 2002; Chen et al. 2005a). Ascorbate (1.2%) was added to protect against DA oxidation. After DA loading, the astrocytes were washed five times with standard extracellular solution.

Standard polypropylene‐insulated carbon fibre electrodes (CFEs) Dagan Instruments, Minneapolis, MN, USA) were used to detect quantal release in DA‐loaded astrocytes as described previously (Zhang & Zhou, 2002; Chen et al. 2005a). The 5‐μm diameter sensor tip of a CFE was lightly positioned on the surface of an astrocyte and the electrode was held at 780 mV. The output signal amperometric current (I _amp) was low‐pass filtered at 300 Hz and digitized at 2 kHz.

Immunohistochemistry

Cells cultured on coverslips were washed with PBS and fixed in 4% paraformaldehyde for 10 min. After permeabilization with 0.1% Triton X‐100 and blocking with 10% FBS for 1 h at room temperature, cells were incubated overnight with primary antibodies (rabbit anti‐P2X7 polyclonal anti‐serum; dilution 1:200; Sigma, St Louis, MO, USA; mouse anti‐GFAP monoclonal antibody; dilution 1:1500; Sigma). After washing with 0.1 m PBS, secondary antibodies were added and kept for 2 h at room temperature (Alexa Fluor 488‐ or 594‐conjugated secondary antibodies; dilution 1:500; Life Technologies, Grand Island, NY, USA). Control experiments were performed without primary or secondary antibodies. In RNA interference (RNAi) experiments, goat anti‐P2X7 polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and Alexa Fluor 488‐conjugated secondary antibody (dilution 1:500; Life Technologies) were used. Nuclei were counterstained with 4′,6‐diamidino‐2‐phenylindole (DAPI) (Life Technologies) and mounted on slides with mounting solution (Dako, Glostrup, Denmark). Fluorescence images were acquired on an inverted confocal laser scanning microscope (LSM 710; Carl Zeiss, Oberkochen, Germany) with a 40× oil‐immersion objective. For cryosection preparation, adult mice were anaesthetized and pre‐fixed with 4% paraformaldehyde via transcardial perfusion. The brain was removed, post‐fixed in 4% paraformaldehyde for 3 h at 4°C, and then cryoprotected in 30% sucrose. Coronal sections were cut at 50 μm on a cryostat. After washing with PBS, sections were permeabilized with 0.1% Triton X‐100 and blocked with 10% FBS for 1 h at room temperature. Sections were then incubated for 24 h at 4°C with primary antibody (mouse anti‐GFAP monoclonal antibody; dilution 1:1000; SYSY Goettingen, Germany). Sections were washed again in PBS and incubated with secondary antibody for 3 h in the dark at room temperature (Alexa Fluor 594‐conjugated secondary antibodies; dilution 1:500; Life Technologies). Sections were then washed, stained with 1 μm DAPI (Life Technologies) and mounted on slides with mounting solution (Dako). Fluorescence images were acquired on an upright confocal laser scanning microscope (LSM 700; Carl Zeiss) with a 10× or 20× water‐immersion objective.

Fluorescent dyes and confocal imaging

FM2‐10, FM1‐43 and LysoTracker (Life Technologies) were used to label vesicles in cultured astrocytes (Li, 2008; Duan, 2007). Astrocytes were incubated with 5 μm FM2‐10, 5 μm FM1‐43 or 50 nm LysoTracker in SE at 37°C for 30 min, and then washed three times before the experiments. The fluorescence images of single cells were acquired on an inverted confocal microscope (LSM 710; Carl Zeiss) with a 40× oil‐immersion objective. The FM dyes were excited by a 488 nm laser and LysoTracker was excited by a 594 nm laser.

Confocal FM‐fluorescence imaging was performed as described previously (Chen et al. 2005a), with slight modifications. Briefly, astrocytes were incubated for 5 min in 50 μm FM 2–10 and 100 μm glutamate at 37°C, then washed 5–10 times. The FM dyes were excited at 488 nm on either a LSM 710 confocal or an LSM 510 two‐photon inverted microscope (Carl Zeiss). The de‐staining of FM spots was sampled at 1 Hz. Laser stimulation (790 nm, 0.5 mW, 12 × 12 pixels) was used to mimic Mstim, as described previously (Chen et al. 2005a).

TIRF imaging

We used an inverted microscope (IX‐81; Olympus, Tokyo, Japan) equipped with a 100× oil lens (Olympus; numerical aperture 1.45). Images were acquired with an iXonEM+ EMCCD (Andor Technology, Belfast, Northern Ireland). The acquisition frequency was ∼4 Hz and the exposure time was 200 ms. All experiments were performed at room temperature (22–25°C). Vesicle fusion events were defined as previously described (Bowser & Khakh, 2007; Liu et al. 2011).

RNA interference and western blot analysis

P2X7 short hairpin (sh)RNA(r) and scrambled lentiviral particles were obtained from Santa Cruz Biotechnology. Three target‐specific shRNA sequences (Table 1) were used in a mixture in accordance with the manufacturer's instructions. For western blot analysis, anti‐P2X7 polyclonal antibody (dilution 1:200; Santa Cruz Biotechnology) and anti‐β‐actin monoclonal antibody (dilution 1:5000; Sigma) were used for 2 h at room temperature. IRDye‐conjugated secondary antibodies (dilution 1:5000; Li‐Cor, Lincoln, NE, USA) were applied for 2 h at room temperature and signals were detected using an Odyssey infrared imaging system (Li‐Cor) and analysed with ImageJ, version 1.42 (NIH, Bethesda, MD, USA).

Table 1.

rP2X7 shRNA sequences

5′‐gCCAAgTACTATAAggAAATTCAAgAgATTTCCTTATAgTACTTggCTTTTT‐3′

5′‐CCAAgCCgACgTTAAAgTATTCAAgAgATACTTTAACgTCggCTTggTTTTT‐3′

5′‐ggATgTgACgTCACATgTATTCAAgAgATACATgTgACgTCACATCCTTTTT‐3′

Preparation of hippocampal slices and [Ca]_i imaging in situ

Brain slices were obtained from 2‐week‐old P2X7‐EGFP transgenic mice. The mice were killed by decapitation and immediately dissected. Brain slices (200–300 μm) were cut on a vibratome (VT1200s; Leica, Wetzlar, Germany) in ice‐cold Ca²⁺‐free artificial cerebrospinal fluid (aCSF). The standard aCSF contained (in mm): 125 NaCl, 2.5 KCl, 10 glucose,1.25 NaH₂PO₄, 2 Na pyruvate, 0.5 ascorbic acid, 26 NaHCO₃, 1 MgCl₂ and 2 CaCl₂ (pH 7.4, when oxygenated with 95% O₂ and 5% CO₂). Slices were incubated for 1 h at 37°C in standard aCSF before the experiments.

For [Ca²⁺]_i imaging, brain slices were preloaded with 20 μm Fluo4‐AM (Life Technologies) at 37°C for 20 min. P2X7‐expressing cells were identified by the fluorescence of EGFP, and astrocytes were identified by post‐staining with 0.5 μm SR101 at 34°C for 25 min. Fluorescence images were acquired on an upright confocal microscope (LSM 700; Carl Zeiss) with a 20× water‐immersion objective. Images were sampled at 1.5 Hz. [Ca²⁺]_i signals were quantitated by: ΔF/F ₀ (%) = 100 [(F ₁ – B ₁) – (F ₀ – B ₀)/(F ₀ – B ₀)], where F ₁ and F ₀ are the measured fluorescence intensity at a given time and B ₁ and B ₀ are the mean intensity before stimulation as background fluorescence signals.

Drug and stimulus application

All chemicals and reagents were obtained from Sigma, except as noted elsewhere. Drugs were freshly prepared in SE. Control solutions and drugs were puffed locally onto the cell with a multichannel microperfusion system during recording (Zhang & Zhou, 2002). For osmolysis of lysosomes as described previously (Jadot et al. 1984; Zhang et al. 2007), astrocytes were incubated for 15 min in 200 μm glycyl‐phenylalanine 2‐naphthylamide (GPN) at 37°C and washed out three times in SE. Mechanical stimulation was applied with a polished pipette as described previously (Charles et al. 1991; Newman & Zahs, 1997; Chen et al. 2005a). The jet‐flow stimulation was produced by a glass pipette (inner diameter 10 μm at the tip) filled with standard extracellular solution. The pipette tip was placed on the astrocyte at intervals of 10–20 μm. The pressure level was controlled by a 5 mL injector.

Statistical analysis

IGOR Pro, version 6.22 (WaveMetrics Inc., Portland, OR, USA) was used to analyse all the electrophysiological data. ImageJ, version 1.42 (NIH) was used to analyse imaging data. Data are expressed as the mean ± SEM. Statistical differences between two groups were evaluated with Student's t test. P < 0.05 was considered statistically significant.

Results

Evoked non‐vesicular ATP release in astrocytes

To record possible quantal ATP release from astrocytes, we used patch clamped ATP‐sniffer cells (Young & Poo, 1983; Liu et al. 2011) expressing ATP‐gated P2X4 channels that had sufficient spatial (μm) and temporal (ms) resolution to record quantal ATP release from single vesicles (Fig. 1). It would be ideal to study quantal ATP release in vivo or in brain slices; however, the sniffer method of recording quantal release requires direct contact between the sniffer cell and astrocytes, which is more practical for cells in culture vs. slices. In cultured astrocytes, to determine whether lysosomes release ATP, we needed to trigger lysosome exocytosis. Previously, it has been demonstrated that lysosome exocytosis is most reliably triggered by Mstim (Chen et al. 2005a; Haydon & Carmignoto, 2006; Jaiswal et al. 2007; Li et al. 2008; Liu et al. 2011). In the present study, our standard Mstim protocol used a micromanipulator to deliver a slight touch for 1–2 s followed by a detachment. Typically, detaching the pipette from the membrane triggered a reproducible ATP release, as measured by the ATP‐sniffer current (Fig. 1 A). The sniffer current of Mstim‐induced ATP release from an astrocyte, termed MARA, quickly reached ∼1 nA in amplitude, followed by a slow decay with a time constant of 2–4 s (Fig. 1 A). Decay of the MARA signal was around three times faster than the inherent decay of the ATP‐sniffer (∼10 s) as a result of the desensitization of P2X4 channels (data not shown). The MARA was probably not an injury artefact because it was also induced by an Mstim of half‐strength to form a ‘giga‐seal’ in a neuron.

Figure 1. Mechanical stimulation (Mstim) induced non‐quantal ATP release in astrocytes (MARA).

A, typical MARA current (I _P2X4) recorded with a P2X4‐expressing sniffer cell (ATP sniffer) on an astrocyte. The astrocyte was stimulated twice and the MARA signal was reproducible. Inset: cartoon of the experimental protocol for MARA recording. A whole‐cell patched sniffer cell (green) was positioned on top of an astrocyte to record the ATP release evoked by Mstim. B, KCl‐induced (70 mm) quantal ATP release from a rat adrenal chromaffin cell (RACC) recorded with a P2X4‐expressing sniffer cell. Inset: cartoon of the experimental protocol for ATP recording on an RACC. A whole‐cell patched sniffer cell (green) was positioned near an RACC to record the ATP release evoked by 70 mm KCl. C, comparison of evoked quantal ATP release events in an RACC (blue, enlarged from B) and a typical MARA event in an astrocyte (red). The signals are superimposed on the same scale. Inset: quantal ATP release event from an RACC on an expanded scale. D, corresponding sniffer signals in astrocytes vs. RACCs were: amplitude 1030 ± 103 vs. 150 ± 13 pA, charge 3159 ± 438 vs. 13 ± 2 pC, half‐width 2388 ± 199 vs. 55 ± 2 ms and rise‐time 421 ± 37 vs. 10 ± 1 ms. ^*** P < 0.001. [Color figure can be viewed at http://wileyonlinelibrary.com]

To determine whether the ATP‐sniffer was sufficiently sensitive to detect quantal ATP release from single vesicles, the patch clamped sniffer was also used to record the well‐known vesicular release of ATP, which is co‐released with catecholamine from single chromaffin cells (Hollins & Ikeda, 1997). After stimulation with 70 mm KCl, a burst of individual spikes of ATP release was recorded from chromaffin cells (Fig. 1 B). These were temporally coincident with the individual amperometric spikes of vesicular catecholamine release (Fig. 2) (Wightman et al. 1991; Chow et al. 1992; Zhou & Misler, 1995), indicating that each single‐vesicle release corresponded to co‐release of the transmitters (catecholamine and ATP). Thus, ATP‐sniffer cells were sufficient to record quantal ATP release events in chromaffin cells. By contrast to the burst of quantal ATP spikes from vesicles in a chromaffin cell, the MARA signal showed a single but huge current spike (Fig. 1 C), ∼300‐fold greater in charge and ∼50 times slower in kinetics (Fig. 1 D), indicating that MARA occurs through a non‐vesicular pathway.

Figure 2. Coincidence of evoked quantal catecholamine and ATP release events in single rat chromaffin cells.

A, combined recording of CFE amperometry (red, upper) and ATP sniffer (blue, lower) in a chromaffin cell. Insert: recording protocol. All the CFE spikes were coincident with ATP spikes, indicating that both types of spikes arose from the same single vesicles. It has been established that the depolarization‐induced individual amperometric spikes represent quantal release events from single chromaffin vesicles (Wightman et al. 1991; Chow et al. 1992; Zhou & Misler, 1995), indicating that the individual I _P2X4 spikes represented quantal ATP release events. B, cartoon showing the recording protocol. C, latency coincidence plot of individual ATP‐sniffer events and CFE events from (A). Nine of the total of 10 ATP spike events are plotted (one ATP event was missed by the CFE). [Color figure can be viewed at http://wileyonlinelibrary.com]

Mstim‐evoked astrocytic ATP release is not via Ca²⁺‐dependent exocytosis

The lysosome is one of the major vesicular compartments for Ca²⁺‐dependent exocytosis (Jaiswal et al. 2007; Li et al. 2008; Liu et al. 2011). Based on indirect assays, lysosomes have been proposed to be responsible for evoked vesicular ATP release in astrocytes (Zhang et al. 2007). As shown above, however, our direct recordings with ATP‐sniffer cells suggested that ATP is released via a non‐vesicular pathway (Fig. 1). Thus, we re‐investigated whether ATP is released by lysosome exocytosis following an Mstim.

We recorded lysosome exocytosis using three independent assays. First, we used electrochemical CFEs to record quantal exocytosis of a false transmitter (dopamine) in astrocytes, as we introduced previously (Chen et al. 2005a). Following an Mstim, Ca²⁺‐dependent quantal spikes were recorded with CFEs in astrocytes. The quantal spikes were reduced when extracellular Ca²⁺ was changed from 2.5 mm to 0 (Fig. 3 A and B). Second, we used confocal fluorescence imaging to visualize quantal lysosome exocytosis of a false transmitter (FM2‐10) (Chen et al. 2005a), which labelled lysosomes in astrocytes (Zhang et al. 2007; Li et al. 2008; Liu et al. 2011). FM2‐10‐labelled puncta overlapped >90% with LysoTracker (Fig. 4 A), which labels acidic organelles including lysosomes, confirming that lysosomes were indeed labelled by the FM dye (Li et al. 2009). After Mstim‐like laser stimulation, a subset of FM2‐10 puncta were de‐stained, indicating that lysosomes underwent exocytosis. The FM2‐10 de‐staining signal was decreased in the absence of extracellular Ca²⁺ (Fig. 3 C) (Chen et al. 2005a). Third, we used total internal reflection fluorescence microscopy (TIRFM) to obtain real‐time images of single‐lysosome exocytosis in astrocytes. The event of a lysosome fusion to the plasma membrane was observed with EGFP‐LAMP1, which is a specific marker for lysosomes. Again, Mstim induced robust lysosome fusion events in the presence but not in the absence of 2.5 mm extracellular Ca²⁺ (Fig. 3 E and F) (Liu et al. 2011). Taken together, these three independent results demonstrated that the Mstim‐induced lysosome exocytosis is Ca²⁺‐dependent in astrocytes.

Figure 3. Calcium‐dependent lysosome exocytosis was not involved in MARA.

A, Ca²⁺‐dependent exocytosis in astrocytes recorded by amperometry. Left: cartoon of the false‐transmitter loading process and amperometric recording with a carbon fibre electrode (CFE). Right: Mstim‐induced quantal release events in 0 Ca²⁺ and 2.5 mm Ca²⁺. Inset: amplified view of quantal spikes inside the grey dashed box. B, summary of evoked exocytosis in the presence or absence of extracellular Ca²⁺. ^** P < 0.01. C, Ca²⁺‐dependence of lysosome exocytosis imaged with FM2‐10 fluorescent dye. Left: FM2‐10 puncta before and after stimulation. Scale bar = 5 μm. Right: FM2‐10 discharge in extracellular solution with 0 Ca²⁺ and 2.5 mm Ca²⁺. Laser stimulation was used to mimic Mstim in the FM‐dye imaging experiments (Chen et al. 2005a). D, summary of evoked exocytosis in the presence or absence of extracellular Ca²⁺. ^** P < 0.01. E, left upper: astrocyte transfected with EGFP‐LAMP1. Scale bar = 5 μm. Left lower: a representative event of lysosome fusion observed by EGFP‐LAMP1. The interval between each frame is 255 ms and the image size is 2 μm × 2 μm. Scale bar = 2 μm. Right: cumulative curve of releasable lysosome fusion events following Mstim observed by TIRFM. F, summary of evoked exocytosis in the presence or absence of extracellular Ca²⁺. ^*** P < 0.001. G–I, representative traces showing Mstim‐induced ATP release was not inhibited by reduction of extracellular Ca²⁺ from 2 mm to 0 (G), or by pre‐incubation of the astrocytes with 1 μm thapsigargin for 30 min (H), or by pre‐incubation of the astrocytes with 50 μm BAPTA (I). J, summary of astrocytic ATP release in 0 Ca²⁺ bath solution (n = 6), after pretreatment with 1 μm thapsigargin (TG) for 30 min (n = 8), after pretreatment with 50 μm BAPTA in 0 Ca²⁺ bath solution (n = 4) and in IP3R2‐knockout mice. [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 4. MARA was not inhibited by GPN.

A, images showing FM2‐10 and LysoTracker co‐localized in astrocytes, indicating FM2‐10‐labelled vesicles were lysosomes. Scale bar = 10 μm. B, GPN destroyed lysosomes but mitochondria were not affected. Images showing that the fluorescence of LysoTracker decreased after GPN treatment (200 μm, 15 min), indicating that lysosomes were destroyed by GPN. Scale bar = 20 μm. C, images showing astrocytes loaded with TMRM (red) and FM2‐10 (green) before and after GPN treatment. Left: mitochondria were labelled with TMRM and lysosomes were loaded with FM2‐10 before treatment with GPN. Right: the fluorescence of TMRM was not affected, whereas the fluorescence of FM2‐10 decreased after treatment with 200 μm GPN. Scale bar = 5 μm. D, the MARA signal was not inhibited by pretreatment with 200 μm GPN for 15 min. E, summary of the effects of GPN on MARA (n = 15) and FM 2–10 discharge (n = 8). ^*** P < 0.001.

Therefore, we used the Ca²⁺‐free condition to test whether Mstim still induced MARA when lysosome exocytosis was blocked. Surprisingly, MARA was not inhibited in BAPTA‐AM‐loaded astrocytes in Ca²⁺‐free extracellular solution (Fig. 3 I). Pretreatment with thapsigargin, pretreatment with BAPTA‐AM in the absence of extracellular Ca²⁺ and IP3R2‐knockout all failed to block the MARA signal (Fig. 3 H–J). To further determine whether lysosome exocytosis was involved in MARA, we pretreated astrocytes with GPN (200 μm, 15 min), a substrate of the lysosomal exopeptidase cathepsin C that selectively induces osmodialysis of lysosomes. GPN greatly decreased the number of LysoTracker‐ and FM 2‐10‐loaded puncta and consequently abolished the lysosome exocytosis, although it did not affect mitochondrial integrity (Fig. 4 B and C). However, GPN pretreatment did not inhibit the MARA signal (Fig. 4 D and E). By contrast to GPN, application of carbonylcyanide‐p‐trifluoromethoxyphenylhydrazone (FCCP), an uncoupler of oxidative phosphorylation in mitochondria, progressively blocked the signal (after the second Mstim) within 2 min (data not shown), suggesting that MARA is dependent on functional mitochondria.

Together with the three measures of Mstim‐induced Ca²⁺‐dependent lysosome exocytosis and the Ca²⁺‐independence, lysosome‐independence and mitochondrial‐dependence of MARA, these results suggested that Mstim‐induced ATP release is not through the Ca²⁺‐dependent exocytosis of lysosomes (as well as other recycling vesicles taking up DA).

Mstim‐induced ATP release is mediated by P2X7 channels

The results reported above demonstrated that Mstim‐induced lysosome exocytosis did not release ATP, implying that MARA occurs through a membrane channel or transporter. Thus, we further investigated which channel/transporter is responsible for MARA.

Using indirect assays, several candidates for non‐vesicular pathways of ATP release have been proposed, including gap junctions, anion channels, pannexin channels and P2X7 channels. Taking advantage of the direct real‐time recording of ATP release, we assessed reagents against the possible candidates and found that MARA was sensitive to a selective antagonist of P2X7. P2X7 channels are expressed in cultured hippocampal astrocytes (Fig. 5 A) (Panenka et al. 2001), in astrocytes in hippocampal slices (Fig. 6 A) and in astrocytes under ischemic conditions in vivo (Franke et al. 2004). Mstim induced a brilliant blue G (BBG) (100 nm)‐sensitive inward current under whole‐cell recording (data not shown), indicating the activation of P2X7 channels in astrocytes. BBG, a potent selective antagonist of rat P2X7 channels (EC₅₀ 10–15 nm) (Jiang et al. 2000), does not affect most other P2X subtypes until the concentration reaches 5–10 μm (Anderson & Nedergaard, 2006). BBG (100 nm) significantly reduced the MARA signal without affecting the P2X4 sensitivity of sniffer cells (Fig. 5 B and C), indicating that P2X7 channels meditate ATP release following Mstim. The involvement of P2X7 channels was further confirmed in astrocytes expressing shRNAs against P2X7 (Fig. 5 D–G). The expression of P2X7 in knockdown cells was reduced to 39 ± 13% of that in control cells. Consistently, the MARA amplitude was reduced significantly by 83 ± 23% in P2X7 knockdown cells (Fig. 5 E), whereas both the P2X7 expression and MARA amplitude remained unchanged in scrambled shRNA‐transfected cells (Fig. 5 D–G). Pannexin has been proposed to participate in P2X7 pore formation in astrocytes (Pelegrin & Surprenant, 2006). We investigated whether the pannexin hemichannel mediates MARA in C6 cells, which are a natural pannexin‐null glioma cell line expressing P2X7 channels (Lai et al. 2007). After Mstim, MARA‐like ATP release signals occurred in C6 cells, and the kinetics of ATP release in astrocytes and C6 cells were similar (Fig. 6), implying that pannexin is not involved in generating MARA. Thus, the MARA signals are probably mediated by P2X7 receptor channels in astrocytes.

Figure 5. P2X7 channels were involved in evoked ATP release from astrocytes.

A, images of representative astrocytes double‐labelled with antibodies against GFAP (red) and P2X7 receptors (green). Scale bar = 20 μm. B, BBG (100 nm), a specific P2X7 receptor antagonist, inhibited MARA in astrocytes. Inset: BBG (100 nm) did not affect the sensitivity of ATP sniffer currents (n = 6). C, summary of the effect of BBG on evoked ATP release (n = 6). ^** P < 0.01. D, immunofluorescence staining showing partial knockdown of P2X7 receptors. Scale bar = 20 μm. E, upper, western blots showing partial knockdown of P2X7 receptors by RNAi. Lower: normalized P2X7 protein levels were 100% (control), 93 ± 9% (scrambled) and 39 ± 13% (RNAi) (n = 9). ^** P < 0.01. F, MARA was markedly lower in P2X7 knockdown astrocytes than in normal astrocytes, or scrambled RNAi‐transfected astrocytes. G, summary of the RNAi knockdown effect on MARA (n = 20). ^*** P < 0.001.

Figure 6. Mstim‐induced ATP release exists in the pannexin‐knockout glioma cell line.

A and B, RT‐PCR for mRNA of P2X7 receptors and pannexin‐1 in cultured astrocytes (A) and C6 cells (B). C, comparison of two normalized typical MARA recordings in an astrocyte (grey trace) and a pannexin‐1‐null glioma C6 cell (black trace). D, statistical comparisons of half‐widths and rise‐times of cultured astrocytes (n = 78) and C6 cells (n = 12).

Functional P2X7 channels are expressed in astrocytes in hippocampal slices

In addition to cultured astrocytes, we further investigated whether P2X7 channels were functionally expressed in astrocytes in situ. We performed immunostaining and [Ca²⁺]_i imaging in hippocampal slices from P2X7‐EGFP transgenic mice, which express EGFP under the control of the promoter of the P2X7 channel (Gong et al. 2003). By staining the slices with the antibody against GFAP, a marker for astrocytes, we found that a group of astrocytes expressed P2X7‐EGFP in situ (Fig. 7 A). The expression of P2X7 in astrocytes in situ was also detected in rat hippocampal slices immunostained with antibodies against GFAP and P2X7 (data not shown).

Figure 7. Expression and function of the P2X7 receptor channel in hippocampal astrocytes in slices.

A, expression of P2X7 channels in astrocytes in hippocampal slices. Upper: hippocampal slice from a P2X7‐EGFP transgenic mouse immunostained for GFAP (red) to identify astrocytes; the nuclei were stained with DAPI (blue). Scale bar = 250 μm. Middle: magnified image of boxed area in upper panel. Scale bar = 50 μm. Lower: magnified view of boxed area in middle panel reveals that P2X7 and GFAP co‐localized in astrocytes. Scale bar = 10 μm. B, hippocampal slice from a P2X7‐EGFP transgenic mouse loaded with Fluo4 (10 μm, white) and SR101 (0.5 μm, red). C, left: ATP‐induced Ca²⁺ elevation in a P2X7‐positive astrocyte was blocked by BBG (200 nm). Right: summary of the BBG effect in P2X7‐positive astrocytes (n = 15). D, left: ATP‐induced Ca²⁺ elevation in a P2X7‐negative cell was not blocked by BBG (200 nm). Right: summary of the BBG effect in P2X7‐negative cells (n = 6).

Next, we studied the function of P2X7 channels in situ. We loaded hippocampal slices from P2X7‐EGFP mice with the fluorescent Ca²⁺ indicator Fluo4‐AM to measure astrocytic [Ca²⁺]_i elevation, and with the astrocyte‐specific dye SR101 to identify astrocytes (Fig. 7 B). Application of 200 μm ATP induced [Ca²⁺]_i elevation in P2X7‐positive astrocytes, and this Ca²⁺ signal was inhibited by the P2X7‐specific antagonist BBG (200 nm), suggesting that the [Ca²⁺]_i elevation in astrocytes is mediated by P2X7 channels (Fig. 7 C). By contrast, BBG failed to block the ATP‐induced [Ca²⁺]_i elevation in P2X7‐negative cells (Fig. 7 D). These results confirmed the functional expression of P2X7 channels in hippocampal astrocytes in situ.

Discussion

We have provided several lines of evidence indicating that, following Mstim‐triggered lysosome exocytosis, most ATP release (MARA) was not via lysosomes but, instead, was dependent on P2X7 channels in astrocytes: (i) in contrast to the established burst of quantal spikes of ATP‐release in chromaffin cells (Figs 1 B and 2), MARA showed a single spike with dramatic kinetic characteristics (Fig. 1); (ii) lysosome exocytosis required Ca²⁺, as determined by three independent assays: TIRF imaging of EGFP‐LAMP1, confocal imaging of FM2‐10 and CFE recording of false transmitter (Fig. 3 A–F) (Chen et al. 2005a; Jaiswal et al. 2007; Li et al. 2008; Liu et al. 2011), whereas MARA did not require Ca²⁺ (Fig. 3 G–J); (iii) GPN that destroyed lysosomes had no effect on MARA (Fig. 4 B–E); (iv) FCCP, a mitochondrial uncoupler, inhibited MARA (data not shown); (v) MARA was inhibited by BBG, a selective P2X7 antagonist (Fig. 5 B and C); (vi) pannexin‐1, which also permeates ATP, was not involved in MARA (Fig. 6); (vii) P2X7 was expressed not only in cultured, but also in brain‐slice astrocytes (Figs 5 A and 7 A); and (viii) MARA was inhibited by RNAi‐based P2X7 knockdown (Fig. 5 D–G), indicating that most ATP release by Mstim is via the MARA pathway in hippocampal astrocytes.

Our experiments demonstrated that ATP was not released by exocytosis following Mstim. ATP‐sniffer cells had sufficient spatiotemporal resolution to record quantal ATP release in cultured chromaffin cells (Figs 1 and 2). In astrocytes, however, although Mstim reliably triggers lysosome exocytosis (Jaiswal et al. 2007; Li et al. 2008; Liu et al. 2011), the ATP release signal (MARA) recorded by sniffer cells was not that of multiple quantal spikes but, instead, a single huge spike. The kinetics of MARA was ∼300 times larger and ∼50 times slower than that of quantal spikes in chromaffin cells (Fig. 1). Thus, the MARA was not from lysosomes. Rather, MARA was most probably from whole astrocytes. Because astrocytic lysosome release is asynchronous (Fig. 3 A) (Chen et al. 2005a; Liu et al. 2011), similar to secretion in chromaffin cells (Figs 1 B and 2 A), quantal release of ATP from individual lysosomes would be resolved by ATP‐sniffer recording. Thus, the huge and slow ATP sniffer current of MARA could not be from the synchronous release of multiple vesicles/lysosomes. Taken together, the kinetic evidence suggests that MARA is most probably ATP release from the whole astrocyte, rather than from a vesicle.

Based on Ca²⁺‐dependence and GPN‐sensitivity, we further demonstrated that the majority of Mstim‐induced astrocytic ATP release was via a lysosome independent pathway. According to previous studies, including our own work, lysosome vesicle exocytosis is dependent on intracellular Ca²⁺ in astrocytes (Jaiswal et al. 2007; Li et al. 2008; Liu et al. 2011). Also, it has been established that Ca²⁺‐dependent transmitter release is via exocytosis. Thus, to determine whether lysosome exocytosis produces ATP release, we separately investigated the Mstim‐induced signals (lysosome exocytosis and ATP release). Regarding Mstim‐induced lysosome exocytosis, we carried out three independent experiments (Fig. 3 A–F). First, astrocytes were preloaded with the false transmitter dopamine. After Mstim, we recorded Ca²⁺‐dependent quantal exocytosis with CFEs (Fig. 3 A and B). Dopamine and FM2‐10 are loaded into the same type of vesicles (Chen et al. 2005a; Li et al. 2008), which are predominantly lysosomes (Fig. 4 A) (Zhang et al. 2007; Li et al. 2008; Liu et al. 2011). This indicates that the quantal release events illustrated in Fig. 3 A and B were from lysosome exocytosis. Second, FM2‐10 was preloaded into astrocytes to selectively label lysosomes (Fig. 4 A) (Zhang et al. 2007; Li et al. 2008; Liu et al. 2011). Lysosome exocytosis was observed as FM2‐10 discharge, and was Ca²⁺‐dependent (Fig. 3 C and D). Third, EGFP‐LAMP1 was transfected into astrocytes to label lysosomes. Again, Ca²⁺‐dependent lysosome exocytosis was observed by TIRFM (Fig. 3 E and F). Taken together, these experiments established that lysosome exocytosis is Ca²⁺‐dependent. Regarding Mstim‐induced ATP release in astrocytes (MARA), we carried out two independent types of experiments. First, when Mstim‐induced lysosome exocytosis was abolished in Ca²⁺‐free solution, the sniffer recording of MARA was not inhibited (Fig. 3 G). To ensure the absence of Ca²⁺, we measured the MARA signal under the following conditions: (i) 0 Ca²⁺ bath and BAPTA‐AM pretreatment to keep both the extracellular and intracellular environments free of Ca²⁺; (ii) thapsigargin pretreatment to deplete the endoplasmic reticulum Ca²⁺ store; and (iii) IP3R2‐knockout mice in which the IP3R2‐mediated Ca²⁺ elevation is abolished. These experiments demonstrated that Ca²⁺ was not required for MARA (Fig. 3 H–J). Second, when lysosomes were destroyed, MARA was not inhibited (Fig. 4 B–E). To destroy lysosomes, astrocytes were pretreated with GPN, which induces osmodialysis of lysosomes (Fig. 4 B and C), and lysosome exocytosis was abolished (Fig. 4 E). However, MARA was not reduced by GPN (Fig. 4 D and E). Taken together, Ca²⁺ and lysosomes are required for lysosome exocytosis but are not required for MARA. Thus, ATP release was largely independent of lysosomal exocytosis under our experimental conditions.

We demonstrated that MARA was mainly dependent on P2X7 channels. This conclusion is based on five pieces of evidence. First, immunocytochemistry showed that P2X7 and GFAP co‐localized in cultured (Fig. 5 A) and hippocampal brain‐slice astrocytes (Fig. 7 A). Second, activation of P2X7 channels produced an inward membrane current in cultured cells (data not shown) and [Ca²⁺]_i elevation in hippocampal slice cells (Fig. 7 C and D). Third, MARA was inhibited by BBG, a selective P2X7 antagonist (Fig. 5 B and C). Fourth, Pannexin‐1, which is also permeable to ATP (Pelegrin & Surprenant, 2006), was not involved in MARA (Fig. 6). And fifth, RNAi‐based knockdown of P2X7 also blocked the MARA signal (Fig. 5 D–G). This evidence demonstrated that Mstim‐induced ATP release is predominantly dependent on P2X7 channels. Upon activation, P2X7 channel is known to permeate cytosolic molecules ≤900 Da including ATP (507 Da) (Yan et al. 2008; Nagasawa et al. 2009).

In previous studies, ATP release via gap junctions, swelling‐activated anion channels, pannexin channels and P2X7 channels has been proposed (Darby et al. 2003; Pelegrin & Surprenant, 2006; Suadicani et al. 2006; Kang et al. 2008). As a result of a lack of ATP recording with spatiotemporal resolution at the single‐vesicle level, it remained unclear how ATP is released. A more recent report proposed that lysosome exocytosis might release ATP in cultured astrocytes after chemical stimulation (Zhang et al. 2007). This work has received considerable attention because of three independent lines of supporting evidence: (i) the lysosomal localization of pre‐loaded MANT‐ATP, a fluorescent ATP analogue; (ii) the presence of ATP in biochemically‐purified lysosomes; and (iii) real‐time visualization of Ca²⁺‐dependent lysosomal exocytosis by TIRFM imaging. However, the ATP release from lysosome exocytosis was not directly recorded. The first two pieces of evidence, lysosomal localization of MANT‐ATP and the presence of ATP in purified lysosomes, are indirect and cannot determine whether lysosomes release ATP. Similar to other false transmitters, including dopamine and FM dyes that can be loaded into the lysosomes of astrocytes (Chen et al. 2005a; Li et al. 2008), exogenous MANT‐ATP probably localizes differently from endogenous ATP. That the lysosomal fraction contains ATP also cannot fully support the release of ATP through lysosome exocytosis because only a subset of lysosomes undergo exocytosis (Blott & Griffiths, 2002; Li et al. 2008) and subcellular fractionation cannot distinguish secretory lysosomes from conventional non‐secretory lysosomes. The third item of evidence, TIRFM imaging, only shows the existence of Ca²⁺‐dependent lysosome exocytosis without direct ATP recording; whether lysosome exocytosis releases ATP cannot be determined by this technique. By contrast, with direct and real‐time ATP recording with a P2X4‐sniffer, we demonstrated that Mstim‐induced ATP release was persistent even when Mstim‐triggered lysosome exocytosis was abolished by removing extracellular Ca²⁺ or by GPN treatment (Figs 3 and 4).

Regarding the physiological relevance of P2X7‐mediated MARA, functional P2X7 channels existed in most cultured astrocytes (Figs 1, 2, 3, 4, 5), a proportion of freshly‐isolated astrocytes (data not shown) and part of the brain‐slice astrocytes (Fig. 7). Astrocytic P2X7 expression is greatly enhanced upon brain/spinal cord inflammation/injury and is strongly associated with pathology (Franke et al. 2004; Narcisse et al. 2005; Peng et al. 2009). Because cerebrovascular vessels are surrounded by astrocytes (Haydon & Carmignoto, 2006), they might be subjected to mechanical stimuli capable of inducing ATP release via changes in the cerebrovascular blood pressure. In conclusion, by real‐time and direct ATP recordings, we have provided compelling evidence that the Mstim‐induced ATP release is predominantly dependent on P2X7 channels, which are expressed and function not only in cultured astrocytes, but also in brain slices. The question of the major pathway(s) of astrocytic ATP release in vivo is once again open for future studies.

Recently, Lalo et al. (2014) reported Ca²⁺‐dependent quantal release from small non‐lysosomal vesicles in freshly‐isolated cortical astrocytes. Although we did not record quantal ATP spikes in cultured astrocytes (but see Lee et al. 2015) similar to those proposed by Lalo et al. (2014), it is possible that non‐quantal and small‐vesicle quantal ATP release coexists in astrocytes in brain slices or in vivo. On the other hand, our present work implies that Mstim‐induced ATP release via P2X7‐dependent pathway might release more ATP in brain injury under pathological conditions.

Additional information

Competing interests

The authors declare that they have no competing interests.

Author contributions

YX, ST, LZ, SS, JL, NG, ML, LW, FZ, CW, ZR and ZZ carried out the experiments. ZZ, ZR and YX planned the work and wrote the paper. YX, ST, LZ, SS and ZZ contributed to the critical revision of the manuscript. All authors approved the final version submitted for publication.

Funding

The work was supported by grants from the National Key Research and Development Program of China (2016YFA0500401), the Natural Science Foundation of China (31761133016, 21790394, 31330024, 31171026, 31327901, 31521062, 21790390, 31670843 and 31400708) and the National Basic Research Program of China (2012CB518006).

Supporting information

Disclaimer: Supporting information has been peer‐reviewed but not copyedited.

Figure S1. TNP‐ATP blocks ATP induced P2X4R current.

Biographies

Yingfei Xiong is a medical doctor in Affiliated Hospital of Air Force Institute of Aeromedicine in Beijing, China.

Sasa Teng is a postdoctoral fellow in Columbia University in New York, USA.

Lianghong Zheng is a manager in a biomedical company in Guangzhou, China.