G_i/o PROTEIN-COUPLED RECEPTORS INHIBIT NEURONS BUT ACTIVATE ASTROCYTES AND STIMULATE GLIOTRANSMISSION

Abstract

G protein-coupled receptors (GPCRs) play key roles in intercellular signaling in the brain. Their effects on cellular function have been largely studied in neurons, but their functional consequences on astrocytes are less known. Using both endogenous and chemogenetic approaches with DREADDs, we have investigated the effects of G_q and G_i/o GPCR activation on astroglial Ca²⁺-based activity, gliotransmitter release, and the functional consequences on neuronal electrical activity. We found that while G_qGPCR activation led to cellular activation in both neurons and astrocytes, G_i/oGPCR activation led to cellular inhibition in neurons and cellular activation in astrocytes. Astroglial activation by either G_q or G_i/o protein-mediated signaling stimulated gliotransmitter release, which increased neuronal excitability. Additionally, activation of G_q and G_i/o DREADDs in vivo increased astrocyte Ca²⁺ activity and modified neuronal network electrical activity. Present results reveal additional complexity of the signaling consequences of excitatory and inhibitory neurotransmitters in astroglia-neuron network operation and brain function.

INTRODUCTION

Intercellular chemical communication in the brain is based on neurotransmitters released by neurons and neurotransmitter receptors in the target cells, which upon activation can enhance or inhibit cellular activity. Two major classes of neurotransmitter receptors are responsible for transducing the intercellular chemical signaling – ligand-gated channels with ionotropic function that directly change membrane ionic permeability, and G protein-coupled receptors (GPCRs) with metabotropic function that activate intracellular signaling pathways. G proteins can be classified into four major families (i.e. G_s, G_q, G_i/o, G_12/13) according to the downstream signaling triggered by the activation of their α subunit. Briefly, some of the main signaling pathways are as follows: G_q activates phospholipase C (PLC), G_i/o decreases cyclic AMP levels through inhibition of adenylyl cyclase, G_s stimulates adenylyl cyclase, and G_12/13 regulates small GTPases (Huang & Thathiah, 2015; Stewart & Fisher, 2015). Although there are a multitude of effects downstream of G protein activation, the classical view of G_q and G_i/o protein signaling in neurons is excitation and inhibition, respectively (Huang & Thathiah, 2015). Whether these functional consequences of different G protein-mediated signaling occurs in other brain cells, like astrocytes, remains largely unknown.

Astrocytes are emerging as important cells in brain function by exchanging signaling with neurons through their participation in the tripartite synapse (Araque, Parpura, Sanzgiri, & Haydon, 1999). Rather than exhibiting electrical excitability like neurons, astrocytes display a form of excitability based on changes in intracellular Ca²⁺ concentration (Araque, Carmignoto, & Haydon, 2001; Araque et al., 2014; Di Castro et al., 2011; Haydon & Carmignoto, 2006; Mariotti et al., 2018; Navarrete & Araque, 2010; Panatier et al., 2011; Perea & Araque, 2005; Perea et al., 2016; Perea, Navarrete, & Araque, 2009; Verkhratsky & Kettenmann, 1996; Volterra, Liaudet, & Savtchouk, 2014; Volterra & Meldolesi, 2005). Upon activation, astrocytes can release their own signaling molecules, termed gliotransmitters, that regulate neuronal excitability, synaptic transmission, and plasticity (Araque et al., 2014; Bezzi et al., 2004; Halassa & Haydon, 2010; Rusakov, 2015; Volterra & Meldolesi, 2005) see however (Fiacco & McCarthy, 2018). A major route of inducing Ca²⁺ elevations is through activation of G_q GPCRs that results in the release of Ca²⁺ from internal stores through activation of PLC and IP₃ receptors (IP₃R) (Di Castro et al., 2011; Panatier et al., 2011; Pasti, Volterra, Pozzan, & Carmignoto, 1997; Perea & Araque, 2005; Porter & McCarthy, 1996). The effects of G_i/o GPCR activation in astrocytes are less known. Reports indicate that neurotransmitter receptors typically coupled to G_i/o GPCRs, such as CB1 and GABA_B receptors, can increase Ca²⁺ levels and stimulate gliotransmission (Covelo & Araque, 2018; Kang, Jiang, Goldman, & Nedergaard, 1998; Mariotti, Losi, Sessolo, Marcon, & Carmignoto, 2016; Meier, Kafitz, & Rose, 2008; Navarrete & Araque, 2008; Perea et al., 2016). While CB1 receptors have been shown to promiscuously couple to G_q proteins (Lauckner, Hille, & Mackie, 2005; Navarrete & Araque, 2008), whether G_i/o protein activation inhibits cellular activity in astrocytes, like in neurons, remains unclear.

Here, we have investigated the functional consequences of selective activation of G_q and G_i/o GPCRs on astrocytic and neuronal activity using endogenous and chemogenetic (i.e. Designer Receptors Exclusively Activated by Designer Drugs, DREADDs) approaches. Combining electrophysiological and Ca²⁺ imaging techniques in hippocampal mouse brain slices, we found that G_q GPCR activation in both neurons and astrocytes led to cellular activation. Hereafter, cellular activation is defined as calcium increases in both neurons and astrocytes and depolarization in neurons. In contrast, G_i/o GPCR activation inhibited neurons, but led to astrocyte activation. Furthermore, G_q or G_i/o GPCR activation in astrocytes stimulated the release of glutamate, which enhanced neuronal excitability. Activation of both G_q and G_i/o DREADDs in astrocytes of the primary somatosensory cortex in vivo similarly led to the enhancement of both astrocyte Ca²⁺ elevations and cortical delta rhythms. These results indicate that while activation of different GPCR pathways in neurons led to either excitation or inhibition, in astrocytes they led only to activation, suggesting that inhibitory signaling is a particular property of neurons and not astrocytes.

METHODS

Experimental model and subject details

Hippocampal coronal slices were obtained from both male and female 6–12 weeks old C57BL/6, GCaMP3 (GLAST-CreE RT2 x R26-lsl-GCaMP3; (Paukert et al., 2014)), GCaMP6f (Tg(Slc1a3-cre/ERT)1Nat/J (JAX:012586) x Ai95(RCL-GCaMP6f)-D (JAX:028865); (Agarwal et al., 2017) and IP₃R2^−/− (Li, Zima, Sheikh, Blatter, & Chen, 2005) mice. GCaMP3 mice received 8 daily doses (57mg/kg) of tamoxifen injections two weeks prior to experimentation. Mice were housed in a 14/10 light/dark cycle with ad libitum food and water. All experiments were in compliance with the Animal Care and Use Committee at the University of Minnesota.

Stereotaxic surgery

4–6-week-old mice were anesthetized with ketamine/xylazine (10 ml/kg) and underwent stereotaxic surgery of AAV5-CaMKIIα-hM3Dq or AAV5-CaMKIIα-hM4Di to assess G_q and G_i/o signaling in neurons, respectively, or AAV8-GFAP-hM3Dq or AAV8-GFAP-hM4Di to assess G_q and G_i/o signaling in astrocytes, respectively. Stereotaxic coordinates used for all hippocampal injections were (relative to Bregma in mm) −2.65 A-P, ±2.25 M-L and −1.75 D-V. Stereotaxic coordinates used for all in vivo cortical injections were (relative to Bregma in mm): −−2.00 A-P, ±2.00 M-L and −1.00, −0.80, and −0.60 D-V. For in vivo control experiments, mice underwent stereotaxic surgery of AAV8-GFAP-mCherry. Experiments were performed 2–6 weeks after surgery. All DREADDs viruses were purchased from the UNC Vector Core, while the plasmid to generate the AAV8-GFAP-mCherry was purchased from Addgene (Plasmid #58909) and was generated by the University of Minnesota Viral Vector and Cloning Core.

Hippocampal slice preparation

The brain was removed quickly after decapitation and placed in ice-cold artificial cerebrospinal fluid (ACSF). Coronal hippocampal slices (350um thick) were made with a vibratome and incubated (>1 h) in a holding chamber at room temperature (21–24 degrees C) in ACSF containing (in mM): NaCl 124, KCl 2.69, KH2PO4 1.25, MgSO₄ 2, NaHCO₃ 26, CaCl₂ 2, ascorbic acid 0.4, and glucose 10, and continuously bubbled with carbogen (95% O2 and 5% CO₂) (pH 7.3). Slices were transferred to an immersion recording chamber and superfused at 2mL/min with gassed magnesium-free ACSF containing (in mM): NaCl 124, KCl 2.69, KH₂PO₄ 1.25, NaHCO₃ 26, CaCl₂ 4, glucose 10, and glycine 4. Tetrodotoxin (TTX, 1μM) was included in the perfusion system for all slice experiments to block action potential-mediated neurotransmission, except for experiments in Figure 7. To isolate GABAB receptor-mediated events in the GABA application experiments, picrotoxin (GABAA receptor antagonist, 50 μM) was added to the perfusion system. In experiments to block G_q-PLC signaling, the PLC inhibitor U73122 (4 μM) was included in the perfusion system. In experiments to block G_i/o signaling, slices were incubated in ACSF containing the G_i/o inhibitor pertussis toxin (PTX; 7.5 μg/ml) for 3–4 hours prior to experiments. In experiments to block muscarinic ACh receptors, atropine (50 μM) was included in the perfusion system. In experiments to block GABAB receptors, CGP54626 (1 mM) was included in the perfusion system. Cells were visualized under 40x water immersion objective using differential interface contrast (DIC) in an Olympus microscope.

Figure 7. Astrocytic G_q and G_i/o DREADD activation increases neuronal action potential firing.

(A and B) Representative trace of spontaneous action potential firing of CA1 pyramidal neuron before and after CNO application to G_qDREADD-expressing astrocytes (A) and G_i/oDREADD-expressing astrocytes (B), and the normalized (from basal) firing frequency vs. time. CNO application was at t=0 (as in the other panels). (C and D) Representative trace (top) of action potential firing induced by short depolarizing pulses before and after CNO application to G_qDREADD-expressing astrocytes (C) and G_i/oDREADD-expressing astrocytes (D), expanded traces (bottom left panels), and normalized firing frequency vs. time (bottom right panels). (E and F) Response to CNO in the presence of D-AP5. (G and H) Normalized firing frequency before and after CNO application to G_qDREADD-expressing astrocytes (G) and G_qDREADD-expressing astrocytes (H) in control and D-AP5. Data are represented as mean ± SEM. P < 0.05 (*).

Slice electrophysiology

Whole-cell electrophysiological recordings were performed in pyramidal neurons of the CA1 region of the hippocampus. Patch pipettes were pulled from thick-walled borosilicate glass (1.5 mm outer diameter) on a Sutter Instruments P-1000 puller. Pipettes (3–8 MΩ) were filled with the internal solution that contained (in mM): K-Gluconate 135, KCl 10, HEPES 10, MgCl2 1, ATP-Na2 2 (pH = 7.3 adjusted with KOH; osmolality 280–290 mOsm/L). Recordings were obtained and filtered (1KHz) by PC-ONE amplifiers (Dagan Instruments, Minneapolis, MN). Signals were fed to a Pentium-based PC through a DigiData 1440A interface board. The pCLAMP 10.2 (Axon Instruments) software was used for stimulus generation, data display, acquisition and storage. For neuronal DREADDs experiments, only mCherry-positive CA1 neurons were patched, and mCherry-negative neurons were recorded for sham conditions. For neuronal slow inward current (SIC) and action potential firing responses after astrocytic DREADDs activation, only mCherry-positive slices were used, and mCherry-negative slices were used for sham conditions. ACh- and CNO-evoked neuronal currents and SICs were recorded in voltage-clamp at a holding potential of −70mV and GABA-mediated currents were recorded at a holding potential of −40mV. To mimic excitatory inputs and observe changes in action potential firing over time (Figure 7), a continuous train of depolarizing current pulses (20ms, ~200 pA, delivered every 50ms) was applied. SICs were defined as currents with a τ_on > 5 and lasting > 40ms. Experiments were performed at room temperature (21–24 °C).

Agonist application

For all slice agonist application experiments, pipettes (3–8 MΩ) were filled with ACSF plus 1 mM ACh or 1 mM GABA or 1 mM CNO, or 2mM (S)-3,5-Dihydroxyphenylglycine (DHPG). Pipettes were positioned over the pyramidal layer to assess neuronal responses to the agonist (Figures 1, 3), or over the stratum radiatum to assess astrocytic Ca²⁺ response to agonist application, SICs, and firing modulation (Figures 2, 4-7). To illustrate the time course of the agonist effects on SIC frequency in Figure 6, the number of SICs were grouped in 10s bins. To compare the agonist effects, the mean SIC frequency recorded 1 minute before and after agonist application was calculated. Agonists were delivered using a Dagan PMI-100 Pressure Micro-injector (15 PSI, 2–5s for all experiments except for the following: 30s for chemogenetic SIC experiments and firing modulation experiments).

Figure 1. G_q signaling induces neuronal activation.

(A) Scheme of neuronal G_q GPCR activation by ACh, and TexasRed and Fluo4-filled CA1 neuron (scale bar, 20µm). (B) Representative traces showing effects of local ACh application in current clamp (CC) and voltage clamp (VC). (C) ACh-induced current amplitude in different conditions (Kruskal-Wallis One-way ANOVA, p < 0.001). (D) Pseudocolor Fluo4 fluorescence images before and after ACh application in the neuron depicted in (A) (scale bar, 20µm), and corresponding fluorescent Ca²⁺ trace. (E) Percentage of neurons that responded with a Ca²⁺ increase to ACh in different conditions. (F) Immunohistochemical images of hippocampus injected with AAV5-CaMKIIα-hM3Dq-mCherry. From left to right: mCherry, NeuN, GFAP and merge (scale bar, 25 μm; s.o., stratum oriens; s.p., stratum pyramidale; s.r., stratum radiatum). (G) Scheme of neuronal chemogenetic G_qDREADD activation and image showing mCherry-expressing CA1 neurons (scale bar, 15 µm). (H) Representative traces showing effects of local CNO application in CC and VC. (I) CNO-induced current amplitude in different conditions (Kruskal-Wallis One-way ANOVA, p < 0.001). (J) Pseudocolor Fluo4 fluorescence image before and after CNO application (scale bar, 10µm), and corresponding fluorescent Ca²⁺ trace. (K) Percentage of neurons that responded with a Ca²⁺ increase to CNO in different conditions. Data are represented as mean ± SEM. P < 0.01 (**), and P < 0.001 (***).

Figure 3. G_i/o activation in neurons is inhibitory.

(A) Scheme of neuronal G_i/o activation by GABA, and TexasRed and Fluo4-filled CA1 neuron (scale bar, 20μm). (B) Representative traces showing GABA-induced responses in current clamp (CC) and voltage clamp (VC), and GABA-induced current amplitude in different conditions (Kruskal-Wallis One-way ANOVA, p < 0.01, p < 0.05). (C) Pseudocolor Fluo4 images before and after GABA application from the neuron depicted in (A) (scale bar, 20µm), and the corresponding fluorescent Ca²⁺ trace. (D) Immunohistochemical images of AAV5-CaMKIIα-hM4Di-mCherry expression in hippocampus. From left to right: mCherry, NeuN, GFAP and merge (scale bar, 50 μm; s.o., stratum oriens; s.p., stratum pyramidale; s.r., stratum radiatum). (E) Scheme of neuronal chemogenetic G_i/oDREADD activation and fluorescent image showing mCherry-expressing neurons (scale bar, 50µm). (F) Representative traces showing CNO-evoked responses in CC and VC, and CNO-induced current amplitude in different conditions (Kruskal-Wallis One-way ANOVA, p < 0.05). (G) Pseudocolor Fluo4 images before and after CNO application (scale bar, 20µm), and corresponding fluorescent Ca²⁺ trace. Data are represented as mean ± SEM. P < 0.05 (*) and P < 0.01 (**).

Figure 2. G_q activation in astrocytes elevates their Ca2+ levels.

(A) Scheme of G_q GPCR activation in astrocytes by ACh locally applied over stratum radiatum astrocytes. (B) Fluorescence images of an astrocyte showing selected domains in soma and processes (top), pseudocolor Ca²⁺ images before (basal) and after local application of ACh (bottom) (scale bar, 5 μm), and Ca²⁺ traces from domains shown in top right panel. (C) Left: Raster plot showing Ca²⁺ events for each ROI (somas (s) below reference line, processes (p) above reference line). Right: Average Ca²⁺ traces from responding somas (top) and processes (bottom). T-test compares 10s before and 10s after agonist. (D and E) Ca²⁺ event probability in somas (D) and processes (E) vs. time, and maximum values in different conditions. (F) Immunohistochemical images of AAV8-GFAP-hM3Dq-mCherry expression in the hippocampus. From left to right: Expression of mCherry, GFAP, NeuN, a merge of all three, and a higher magnification merge image showing colocalization of GFAP and mCherry (25x; scale bar, 20 μm; s.o., stratum oriens; s.p., stratum pyramidale; s.r., stratum radiatum). (G-K) as (A-E) but with local application of CNO to G_qDREADD expressing astrocytes instead of ACh. Data are represented as mean ± SEM. P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***).

Figure 4. G_i/o activation in astrocytes elevates their Ca2+ levels.

(A) Scheme of endogenous astrocyte G_i/o GPCR activation by GABA. (B) Fluorescence images of an astrocyte showing selected domains in soma and processes (top row), pseudocolor Ca²⁺ images before and after local application of GABA (bottom row; scale bar, 5 μm), and Ca²⁺ traces from domains shown in top left panel (right). (C) Left: Raster plot showing Ca²⁺ events for each ROI (somas (s) below reference line, processes (p) above reference line). Right: Average Ca²⁺ trace from responding somas (top) and processes (bottom). T-test compares 10s before and 10s after agonist. (D and E) Ca²⁺ event probability in somas (D) and processes (E) vs. time, and maximum values in different conditions. (F) Immunohistochemical images of AAV8-GFAP-hM4Di-mCherry expression in the hippocampus. From left to right: mCherry, GFAP, NeuN, a merge of all three, and a higher magnification merge image showing colocalization of GFAP and mCherry (25x; scale bar, 40 μm; s.o., stratum oriens; s.p., stratum pyramidale; s.r., stratum radiatum). (G-K) as (A-E) but with local application of CNO to G_i/oDREADD-expressing astrocytes instead of GABA. Data are represented as mean ± SEM. P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***).

Figure 6. Astrocyte activation via G_q and G_i/o DREADDs induces slow inward currents in neurons.

(A) Representative excitatory postsynaptic current (EPSC) and slow inward current (SIC) recorded from a CA1 pyramidal neuron. (B) Representative traces showing SICs (asterisks) in control and in D-AP5. (C) Scheme of neuronal-astroglial synaptic elements and gliotransmission. (D) Scheme and representative traces showing SICs (asterisks) before and after ACh application. (E) Left: Mean number of SICs vs. time, binned at 10s (ACh was applied at t=0). T-test compares one minute before and one minute after agonist. Right: SIC frequency (per min) one minute before and one minute after ACh in different conditions. Statistical significance was determined by the Student’s t-test comparing mean values one minute before and after agonist application. (F-G) as in (D-E) but with CNO application to G_qDREADD-expressing astrocytes. (H-I) as in (D-E) but with GABA application instead of ACh. (J-K) as in (D-E) but with CNO application to G_i/oDREADD-expressing astrocytes. Data are represented as mean ± SEM. P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***).

Ca²⁺ imaging

Ca²⁺ levels in astrocytes were monitored by two-photon microscopy (Leica DM6000 CFS upright multiphoton microscope with TCS SP5 MP laser) using the genetically encoded Ca²⁺ indicator mice GCaMP3 and GCaMP6 expressed in GLAST-positive cells. Some (n=199/512) soma Ca²⁺ responses were monitored by an Olympus BX51W1 epifluorescent microscope using Fluo-4-AM (5 μM in 0.01% of pluronic, 45min incubation at room temperature). Astrocytic Ca²⁺ levels were recorded from the astrocyte cell body and processes, and Ca²⁺ variations were estimated as changes in the fluorescence signal over the baseline (∆F/F₀). A Ca²⁺ signal was considered a Ca²⁺ event when F/F₀ >3 standard deviations greater than baseline fluorescence. Our measurement of Ca²⁺ event probability was calculated as the number of Ca²⁺ elevations grouped in 5s bins recorded from the astrocytes in the field of view (3–6 astrocytes per analyzed region), and mean values were obtained by averaging each different experiment. References to “n” in the text correspond to the number of experiments performed for each condition. Ca²⁺ analysis was performed only on astrocyte soma and processes within the field of view that showed fluctuations in Ca²⁺ levels throughout the recording. A domain was only selected along an active process length if its activity pattern differed from neighboring regions, so as to avoid oversampling of an active process. The number of process domains per experiment was 16.6 ± 0.48. Ca²⁺ levels in neurons were monitored using bulk loading of Oregon Green BAPTA-1 (5 μM in 0.01% of pluronic, 1hr incubation at room temperature) or adding Fluo4 (50 μM) to the recording pipette. A change in fluorescence (∆F/F₀ > 3 standard deviations from baseline fluorescence) occurring within 5s of agonist application was defined as a neuronal Ca²⁺ response.

In vivo experiments and analysis

Two- to six-month-old mice were anesthetized using 1.8 mg/kg urethane injected intraperitoneally, faux tears were applied, and body temperature was maintained at 37°C. Once anesthetized, animals were placed in a stereotaxic frame and a midline incision was made along the scalp. A 3mm diameter craniotomy was performed centered over the primary somatosensory cortex (relative to Bregma in mm: −2.00 A-P, −2.00 M-L). A 0.25mm tungsten wire was placed over the exposed cortex to measure electrocorticography (ECoG), and agar was applied before a glass coverslip was cemented atop the craniotomy. A screw soldered to wire was inserted over the cerebellum to act as the reference and another screw was placed over the ipsilateral frontal plate. A 3D printed frame was mounted to the exposed skull with dental acrylic and fastened to a holder. The holder and animal were then placed underneath a Leica SP5 two-photon microscope for imaging. ECoG was recorded by connecting the lead to the exposed wire over the cortical surface, and reference and ground were attached to the wire soldered to the screw placed over the cerebellum. Prior to data acquisition, a needle was fed into the intraperitoneal cavity connected to a syringe with CNO. ECoG was sent to an AM Systems 3000 AC/DC differential amplifier sampled at 10kHz, filtered at 1Hz-3kHz and digitized using an Axon Digidata 1550 Acquisition System connected to a Dell Optiplex 7010 PC. In this arrangement, ECoG was recorded simultaneously with astrocyte Ca²⁺ monitored in cortical layers 2/3 (i.e. 100–300 µm below the cortical surface). Ca²⁺ imaging was obtained at 1 frame per second through a 25X objective with an additional 1.7X digital zoom. Following 20 minutes of baseline, 2–3 mg/kg CNO was injected intraperitoneally. Following CNO injections, measurements were obtained for 90 minutes. In a subset of mice, ECoG was measured in the absence of Ca²⁺ imaging. Raw ECoG data was lowpass filtered at 300 Hz and analyzed using a custom MATLAB program where a short-time Fourier transform was done using a hamming window to measure spectral content every minute. Normalized ECoG power spectra were created by using min-max normalization with respect to the baseline. In vivo Ca²⁺ data was analyzed using a custom MATLAB program to detect Ca²⁺ events via change in fluorescence, or the derivative of the fluorescence, when the derivative of fluorescence was > 2 standard deviations of the baseline speed over the average baseline speed. After an event was detected, the program detected the end of the event when the trace decayed back to the onset amplitude; from this the duration of the event was quantified, and the amplitude was calculated by taking the maximum value between onset and decay and subtracting it from the amplitude of the onset.

Immunohistochemistry

Anesthetized mice were perfused through the left cardiac ventricle with 4% paraformaldehyde in 0.1M PBS. The brains were removed, postfixed with paraformaldehyde and cut into 50 µm slices using a Leica VT 1000S vibratome. Brain sections were blocked with phosphate-buffered saline (PBS) containing 10% normal goat serum (NGS) or normal donkey serum (NDS) and 0.2% Triton X-100 and incubated overnight with the primary antibody diluted in blocking solution. Appropriate fluorochrome-labeled secondary antibodies (Life Technologies, Waltham, MA) were used for detection. An antibody against glial fibrillary acidic protein (GFAP, 1:1000; Sigma-Aldrich) was used as a marker for astrocytes. An antibody against neuron-specific nuclear protein (NeuN, 1:500; Millipore) was used as a marker for neurons. Fluorescent stained sections were mounted with Vectashield mounting medium (Vector Laboratories). The stained sections were visualized with a Nikon NiE C2 or an Olympus FluoView FV1000 upright confocal microscope and analyzed for colocalization of DREADDs with either neurons or astrocytes using ImageJ software (NIH, Bethesda, MD).

Quantification and statistical analyses

Data are expressed as mean ± standard error of the mean (SEM). Data normal distribution was assessed using a Shapiro-Wilk Normality test. Ca²⁺ event probability, Ca²⁺ fluorescence, SIC frequency, and in vivo Ca²⁺ events were analyzed using a two-tailed paired Student’s t-test (α = 0.05) comparing baseline to post-stimulus. In vivo delta power was analyzed using a Wilcoxon signed-rank test (α = 0.05) comparing baseline to post-stimulus. A one-way ANOVA was used with Holm-Sidak posthoc test to compare treatment groups against the control group for neuronal electrophysiology response data. When the Shapiro-Wilk Normality test failed, a Kruskal-Wallis ANOVA with a Dunn’s posthoc was used. Statistical differences were established with P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***).

RESULTS

G_q GPCR activation in neurons

We first investigated the effects on neurons of endogenous G_q GPCR activation by acetylcholine (ACh), which activates muscarinic ACh receptors (mAChRs). Type 1, 3 and 5 mAChRs, known to be coupled to G_q proteins, are highly expressed in hippocampal neurons (Berkeley et al., 2001; Park & Spruston, 2012; Scheiderer et al., 2008). We performed whole-cell electrophysiological recordings from hippocampal CA1 pyramidal neurons, monitored neuronal Ca²⁺ levels, and locally applied ACh (Figure 1A). CA1 neurons responded to ACh with an inward current in voltage clamp conditions, a transient depolarization in current clamp, and Ca²⁺ elevations (n = 9; Figures 1B–1E). These responses were abolished by the mAChR antagonist atropine (50 μM; n = 5; Figures 1C and 1E). Furthermore, ACh-evoked responses were prevented by the PLC inhibitor U73122 (4 μM; n = 7), but present after incubation with the G_i/o inhibitor pertussis toxin (7.5 μg/ml; PTX; n = 6; Figures 1C and 1E), which is consistent with the canonical G_q signaling pathway (Huang & Thathiah, 2015; Stewart & Fisher, 2015). These results indicate that ACh-induced activation of G_q GPCR signaling evokes neuronal responses (i.e. inward currents and Ca²⁺ increases) associated with neuronal excitation.

To confirm the effects of G_q signaling in neurons, we selectively expressed G_qDREADDs in hippocampal CA1 pyramidal neurons using the viral vector AAV5-CaMKIIα-hM3Dq-mCherry, which contains the G_qDREADD and the fluorescent reporter mCherry under the CaMKIIα promoter, which is highly expressed in CA1 pyramidal neurons (Erondu & Kennedy, 1985; Schulman & Lou, 1989). The selective expression of G_qDREADDs in CA1 pyramidal neurons was confirmed by immunohistochemistry (Figure 1F) and the electrophysiological properties of the recorded neurons expressing mCherry (Gasparini & Magee, 2006; Magee & Carruth, 1999). Local application of the DREADD agonist Clozapine-N-oxide (CNO) evoked transient depolarizations, inward currents, and Ca²⁺ increases in G_qDREADD-expressing neurons (n = 8; Figures 1G–1K), but not in neurons lacking G_qDREADDs (n = 6; Figures 1I and 1K). CNO-evoked responses were diminished by U73122 (n = 5), but not by PTX (n = 6; Figures 1I and 1K). These results indicate that G_q activation by endogenous or chemogenetic means induces neuronal responses associated with cellular excitation.

G_q GPCR activation in astrocytes

Hippocampal astrocytes respond with Ca²⁺ elevations to ACh through activation of mAChRs (Araque, Martin, Perea, Arellano, & Buno, 2002; Shelton & McCarthy, 2000), but the signaling pathway activated remains undefined. Thus, we locally applied ACh and monitored Ca²⁺ levels in the soma and processes of astrocytes located in the CA1 stratum radiatum using transgenic mice that expressed the genetically-encoded Ca²⁺ indicators GCaMP3 or GCaMP6 under the astroglial GLAST promoter (Figure 2A). ACh transiently elevated Ca²⁺ in both astrocyte somas and processes (Figures 2B-E) (Araque et al., 2002; Navarrete et al., 2012; Perea & Araque, 2005; Shelton & McCarthy, 2000; Takata et al., 2011), as quantitatively shown by the increase of the Ca²⁺ event probability (36 astrocyte somas, n = 6; 122 process domains, n = 5; Figures 2D–2E). ACh-induced Ca²⁺ increases were blocked by the mAChR antagonist atropine (18 astrocyte somas, n = 6; 78 process domains, n = 6; Figures 2D and 2E) and by the PLC inhibitor U73122 (27 astrocyte somas, n = 7; 106 process domains, n = 7; Figures 2D and 2E), but were present in slices incubated with the G_i/o inhibitor PTX (54 astrocyte somas, n = 9; 83 process domains, n = 7; Figures 2D and 2E). These results indicate that mAChRs in astrocytes activate G_q proteins that increase intracellular Ca²⁺ levels through stimulation of PLC.

To selectively stimulate G_q-linked GPCRs in astrocytes, we targeted stratum radiatum hippocampal astrocytes with the viral vector AAV8-GFAP-hM3Dq-mCherry, which contains the G_qDREADD and the fluorescent reporter mCherry under the control of the astroglial GFAP promoter (Figure 2F). CNO application mimicked the ACh effects, elevating Ca²⁺ in both somas and processes of G_qDREADD-expressing astrocytes (35 astrocyte somas, n = 6; 153 process domains, n = 6; Figures 2G–2K), but not in astrocytes lacking G_qDREADD expression (18 astrocyte somas, n = 6; 100 process domains, n = 6; Figures 2J and 2K). CNO-evoked responses in both somas and processes were blocked by U73122 (20 astrocyte somas, n = 5; 83 process domains, n = 6), but were still present in slices incubated with PTX (18 astrocyte somas, n = 5; 53 process domains, n = 5; Figures 2J and 2K), indicating that G_q GPCR signaling pathways increase Ca²⁺ in astrocytes by activation of PLC. To further test this idea and to determine the source of the Ca²⁺ elevation and the intracellular signaling downstream of G_qDREADD activation, we performed the experiment in IP₃R2^−/− mice (Li et al., 2005), which lack type 2 IP₃ receptors, the main receptor subtype responsible for astrocytic Ca²⁺ mobilization from internal stores (Di Castro et al., 2011; Gomez-Gonzalo et al., 2015; Martin-Fernandez et al., 2017; Martin, Bajo-Graneras, Moratalla, Perea, & Araque, 2015; Navarrete et al., 2012; Petravicz, Fiacco, & McCarthy, 2008). CNO application in G_qDREADD-expressing slices from IP₃R2^−/− mice showed no significant increase in Ca²⁺ event probability (14 astrocyte somas, n = 5; 46 process domains, n = 5; Figures 2J and 2K), indicating that G_q-GPCR activation induces astrocyte Ca²⁺ mobilization from internal stores. Taken together, these results indicate that stimulating G_q GPCR signaling evokes astrocyte responses (i.e. Ca²⁺ increases) associated with astrocyte activation.

G_i/o GPCR activation in neurons

To investigate G_i/o signaling cascade effects on neurons, we used GABA to activate endogenous GABA_B receptors, which are known to be coupled to G_i/o proteins (Logothetis, Kurachi, Galper, Neer, & Clapham, 1987; North, 1989; Wickman et al., 1994). GABA_B-mediated responses were isolated by blocking GABA_A receptors with picrotoxin (50 μM). In CA1 pyramidal neurons, GABA application evoked a hyperpolarization in current-clamp conditions, and an outward current in voltage-clamp conditions (n = 8; Figures 3A and 3B), with no changes in Ca²⁺ levels (n = 6; see Figure 3C). The outward currents were blocked by the GABA_B receptor antagonist CGP54626 (1 mM; n = 5), as well as by the G_i/o inhibitor PTX (n = 5), but were unchanged by the PLC inhibitor U73122 (n = 5; Figure 3B). These data indicate that G_i/o intracellular signaling cascades evoke effects (i.e. outward currents and absence of Ca²⁺ increases) consistent with neuronal inhibition.

We further investigated the effects of neuronal G_i/o-linked GPCRs using the G_i/oDREADD (Armbruster, Li, Pausch, Herlitze, & Roth, 2007), which was specifically expressed in CA1 pyramidal neurons with the viral vector AAV5-CaMKIIα-hM4Di-mCherry (Figure 3D and 3E). CNO application hyperpolarized and evoked outward currents in G_i/oDREADD-expressing neurons (n = 7), but not in neurons lacking G_i/oDREADD expression (n = 5; Figure 3F). These effects were absent in PTX-incubated slices (n = 5), but were similar to control in the presence of U73122 (n = 5; Figure 3F). Like GABA, CNO did not modify Ca²⁺ levels (n = 6; see Figure 3G). These data indicate that G_i/o signaling evoked by GABA or DREADDs has inhibitory actions on neurons.

G_i/o GPCR activation in astrocytes

We then tested the effects of G_i/oGPCR stimulation on astrocyte Ca²⁺ activity. Local application of GABA increased the Ca²⁺ event probability in both somas and processes (49 astrocyte somas, n = 10; 147 process domains, n = 7; Figures 4A–4E). These effects were not observed in PTX-treated slices (57 astrocyte somas, n = 9; 69 processes domains, n = 5; Figures 4D and 4E) or in the presence of the GABA_B receptor antagonist CGP54626 (26 astrocyte somas, n = 5; 61 process domains, n = 6; Figures 4D and 4E), but were still present in U73122 conditions (38 astrocyte somas, n = 8; 161 process domains, n = 5; Figures 4D and 4E), indicating that, unlike in neurons, stimulation of G_i/oGPCR signaling in astrocytes leads to cellular activation.

To further test this hypothesis, G_i/oDREADDs were specifically expressed in astrocytes using the viral vector AAV8-GFAP-hM4Di-mCherry (Figure 4F). CNO application transiently increased the Ca²⁺ event probability in somas and processes of G_i/oDREADD-expressing astrocytes (21 astrocyte somas, n = 5; 163 process domains, n = 8), but not in non-expressing astrocytes (19 astrocyte somas, n = 5; 71 process domains, n = 5; Figures 4G–4K). These Ca²⁺ responses were present in U73122 (20 astrocyte somas, n = 5; 160 process domains, n = 6), but were absent in PTX-treated slices (20 astrocyte somas, n = 6; 144 process domains, n = 6), and in slices from IP₃R2^−/− mice (22 astrocyte somas, n = 5; 48 process domains, n = 5; Figures 4J and 4K). Taken together, these results indicate that G_i/o signaling evoked by endogenous and chemogenetic approaches resulted in astrocyte activation manifested as Ca²⁺ increases.

Simultaneous activation of G_q and G_i/o GPCR signaling in astrocytes

We further tested the hypothesis that G_q and G_i/o GPCR signaling in astrocytes operates through distinct intracellular pathways. We hypothesized that two agonists acting on similar GPCR signaling pathways would interact producing a non-linear response, whereas two agonists activating different GPCR signaling would evoke a linear response. We first tested two agonists that are known to activate G_q GPCR signaling, ACh and group I mGluR agonist (S)-3,5-Dihydroxyphenylglycine (DHPG). We either applied them separately or both simultaneously and monitored the Ca²⁺ responses. We found that the observed Ca²⁺ response elicited by the simultaneous ACh and DHPG application was lower than the response expected if there was no interaction; i.e., the linear summation of the peak Ca²⁺ elevation evoked by each agonist applied separately (n = 26; Figures 5A, 5B, and 5E) (Perea & Araque, 2005). This relative reduction of the response evoked by these agonists indicates an interaction of the signaling pathways activated and suggests that they operated through the same intracellular pathways. Next, we applied separately ACh, GABA, or both simultaneously to measure the relative contribution of G_q and G_i/o GPCR signaling, respectively, to the astrocyte Ca²⁺ response. We found that the observed amplitude of the Ca²⁺ response to simultaneous ACh and GABA application was similar to that of the expected response; i.e., the linear summation of the responses evoked independently (n = 29; Figures 5C, 5D, and 5E). Moreover, the peak amplitude of the Ca²⁺ response could be further increased by increasing the concentration of ACh (peak fluorescence evoked by 1 mM ACh: 39.4 ± 4.8%, peak fluorescence evoked by 10 mM ACh: 64.6 ± 8.3%, p = 0.040, n = 4 videos, 13 astrocytes), indicating that the receptors and signaling pathways were not saturated. The fact that the Ca²⁺ response linearly summated with simultaneous G_q and G_i/o GPCR agonist application suggests that distinct intracellular pathways were activated, indicating that the actions of G_i/o and G_q GPCRs do not occlude and that the detected Ca²⁺ signal is not saturated. Furthermore, these results indicate that two distinct intracellular signaling mechanisms are activated downstream of G_q and G_i/o GPCR signaling in astrocytes.

Figure 5. G_i/o and G_q GPCRs stimulate different signaling pathways that do not occlude.

(A and B) Scheme of experimental set-up, and astrocyte Ca²⁺ responses to ACh (left), DHPG (middle), and both simultaneously (right), showing the observed (black trace) and expected (green trace) responses. Expected response corresponds to the linear summation of the responses evoked by independent application of ACh and DHPG. (C and D) as in (A and B), but with GABA instead of DHPG. (E) Histogram of the ratio of the observed and expected Ca²⁺ responses evoked by simultaneous application of GPCR agonists in the different pharmacological conditions. Expected responses were considered the linear summation of the responses elicited by independent stimulation of the agonists. Data are represented as mean ± SEM. P < 0.001 (***).

Downstream effects of astrocytic G_q and G_i/o GPCR activation on neurons

Next, we investigated whether activation of G_q and G_i/o protein signaling in astrocytes can lead to gliotransmitter release and regulation of neuronal excitability. Astrocyte Ca²⁺ elevations have been shown to stimulate the release of glutamate, which evokes slow inward currents (SICs) mediated by activation of neuronal NMDA receptors (Araque, Parpura, Sanzgiri, & Haydon, 1998; Araque et al., 1999; Araque, Sanzgiri, Parpura, & Haydon, 1998; Fellin et al., 2004; Mariotti et al., 2016; Martin et al., 2015; Navarrete & Araque, 2008; Perea & Araque, 2005; Perea, Sur, & Araque, 2014). SICs can be distinguished from synaptic currents by their different time courses and their sensitivity to the NMDA receptor antagonist D-AP5 (50 µM; average SIC amplitude = 19.7 pA ± 1.6 pA; n = 182; Figures 6A–6C). We used SICs as a biological assay to detect astrocytic glutamate release. ACh application, which increased astrocyte Ca²⁺ via G_q protein activation (see Figure 2), transiently increased the frequency of SICs in CA1 pyramidal neurons (n = 11; Figures 6D and 6E). In correspondence with the ACh-evoked effects on Ca²⁺, this effect was blocked by atropine (n = 5) and U73122 (n = 7), but was present in PTX-treated slices (n = 5; Figure 6E). Similarly, CNO application in slices with G_qDREADD-expressing astrocytes transiently increased SIC frequency (n = 5), an effect that remained in PTX-treated slices (n = 4; Figures 6F–6G). This effect was absent in the presence of U73122 (n = 5), in sham conditions with no expression of DREADDs (n = 4), and in slices from IP₃R2^−/− mice (n = 8; Figures 6G). These data indicate that G_q-linked GPCR activation in astrocytes stimulates the release of glutamate that activates neuronal NMDA receptors.

We then quantified the effects of G_i/o GPCR activation on SIC frequency. GABA transiently increased the frequency of SICs (n = 9; Figures 6H and5I). As in the case of GABA-evoked Ca²⁺ responses, SIC frequency increases were abolished by CGP54626 (n = 5) and PTX (n = 6), but were detected in slices treated with U73122 (n = 9; Figure 6I). Likewise, CNO application in slices with G_i/oDREADD-expressing astrocytes increased SIC frequency in control conditions (n = 7) and in the presence of U73122 (n = 4), but this effect was absent in PTX-incubated slices (n = 5), in sham conditions (n = 5), and in slices from IP₃R2^−/− mice (n = 8; Figures 6J and 6K). Taken together, G_q and G_i/o GPCR activation in astrocytes led to increases in Ca²⁺, which were associated with an increase in the number of SICs in nearby neurons.

SICs are proposed to enhance neuronal excitability and regulate neuronal synchronization (Angulo, Kozlov, Charpak, & Audinat, 2004; Fellin et al., 2004). We directly tested this idea by investigating the effects of G_q and G_i/o GPCR stimulation in astrocytes on action potential firing in nearby neurons (Figure 7). We activated G_q- and G_i/oDREADD-expressing astrocytes while recording hippocampal CA1 pyramidal neurons in current-clamp conditions. In a first approach, the neuronal membrane potential was slightly depolarized from resting membrane potential by injecting a continuous current to facilitate firing of spontaneous action potentials (Figures 7A and 7B). In a second approach, a continuous train of depolarizing current pulses (20ms, ~200 pA, delivered every 50ms) was applied to observe any potential changes in action potential firing (Figures 7C and 7D). In both conditions, CNO application to either G_q- or G_i/oDREADD-expressing astrocytes enhanced the frequency of action potential firing (Figures 7A–7D, and 7G-7H), which agrees with the gliotransmitter-evoked enhancement of neuronal excitability (see Figure 6). Indeed, the astrocyte G_q- and G_i/oDREADD effects on action potential firing were accompanied by small, but conspicuous, transient depolarizations (insets in Figures 7C and 7D), which appear consistent with the SICs recorded in voltage-clamp (see Figures 6F and 6J), although experimental limitations cannot provide direct evidence that SICs (recorded in voltage-clamp) are responsible for these depolarizations (recorded in current-clamp). Nevertheless, activation of both G_q- and G_i/oDREADDs in the presence of the NMDA receptor antagonist D-AP5 failed to evoke the previously observed transient depolarizations and increase in action potential firing (Figures 7E–7H), supporting the idea that these phenomena were due to neuronal NMDAR activation by astrocytic glutamate. Taken together, these results indicate that activation of both G_q and G_i/o proteins activated astrocytes in the form of a Ca²⁺ elevation. This cellular activation in astrocytes subsequently stimulated gliotransmitter release that led to the enhancement of neuronal excitability.

Effects of astrocytic G_q and G_i/o GPCR signaling in vivo

Finally, we investigated whether similar phenomena occurred in vivo. We injected the viral vectors AAV8-GFAP-hM3Dq-mCherry or AAV8-GFAP-hM4Di-mCherry into the primary somatosensory cortex to selectively express in cortical astrocytes G_q or G_i/o DREADDs, respectively (Figure 8A). Two weeks after injection, we monitored astrocyte Ca²⁺ activity and cortical local field potentials in anesthetized mice, in basal conditions and 1 hour after intraperitoneal injection of CNO (2–3 mg/kg). In mice (n = 6) with G_qDREADD-expressing astrocytes, CNO increased the frequency (from 1.30 ± 0.06 to 2.70 ± 0.28 events/min; n = 21 astrocytes), amplitude (to 218.9 ± 31.2 % from basal values between each astrocyte), and duration of the Ca²⁺ events (from 2.84 ± 0.17 to 6.00 ± 0.69 s) (Figures 8B-8G). Similarly, CNO injection to mice (n = 6) with G_i/oDREADD-expressing astrocytes also increased the frequency (from 1.21 ± 0.08 to 2.00 ± 0.23 events/min; n = 24 astrocytes), amplitude (to 229.5 ± 32.724 % from basal values of each astrocyte), and duration of Ca²⁺ events (from 4.20 ± 0.79 to 6.62 ± 0.80 s) (Figures 8B-8G). Introducing saline into mice (n = 3) with G_qDREADD-expressing astrocytes did not increase the frequency (from 1.35 ± 0.05 to 1.30 ± 0.05 events/min in 37 astrocytes), amplitude (to 98.1 ± 6.0 % from basal values between astrocytes), or duration of the Ca²⁺ events (from 2.81 ± 0.11 to 2.80 ± 0.12 s). Similarly, introducing saline in mice (n = 3) with G_i/o-DREADD-expressing astrocytes did not increase frequency (from 1.30 ± 0.04 to 1.34 ± 0.06 events/min; n = 53), amplitude (to 114.8 ± 10.3 % from basal values between astrocytes), or duration (from 2.80 ± 0.10 to 2.83 ± 0.13 s) of the Ca²⁺ events. Additionally, introducing CNO in mice (n = 3) expressing mCherry in astrocytes did not increase the frequency (from 1.32 ± 0.04 to 1.44 ± 0.06 events/min; n = 77), amplitude (to 118.7 ± 9.3 % from baseline between astrocytes), or duration of the Ca²⁺ events (from 2.78 ± 0.09 to 3.10 ± 0.16 s).

Figure 8. In vivo activation of both G_q and G_i/o DREADDs in astrocytes elevates their Ca2+ levels and regulates neuronal electrical activity.

(A) Immunohistochemical representative image of G_qDREADDs virus injected into the primary somatosensory cortex staining for neurons (NeuN; blue), astroglia (GFAP; green), the reporter mCherry (red), and colocalization of mCherry with GFAP (top) and NeuN (bottom; scale bar, 20 μm). (B and C) Left: Representative images of mCherry in astrocytes reporting expression of G_qDREADDs and G_i/oDREADDs (scale bar, 50 μm). Pseudocolor Ca²⁺images before (middle) and after (right) intraperitoneal injection (i.p.) of CNO. (D and E) Raster plots of Ca²⁺ events before and after i.p. injection of CNO in mice with astrocytes expressing G_q and G_i/oDREADDs, respectively. (F and G) Representative Ca²⁺ traces during basal (left) and after i.p. CNO injection (right) in G_qDREADD- (F) and G_i/oDREADD-injected (G) mice. Ca²⁺ oscillation frequency normalized to baseline for DREADDs-infected cortex with i.p. CNO injections (solid bars), for DREADDs-infected cortex with i.p. saline injections (hashed bars), and for mCherry control-infected cortex with i.p. CNO injections (unfilled bars) for G_qDREADDs- (F) and G_i/oDREADDs-injected (G) mice. (H and I) Representative cortical local field potential recordings prior to CNO (left) and after (right) in G_qDREADDs- (H) and G_i/oDREADDs-injected (I) mice. Lowpass filtered at 40Hz. (J and K) Normalized power spectrum of cortical local field potentials before (black) and after CNO in G_qDREADDs- (J: blue) and G_i/oDREADDs-injected (K: red) mice. Normalized power spectrum of slow-wave delta (0–4 Hz) activity for DREADDs-infected cortex with i.p. CNO injections (solid bars), DREADDs-infected cortex with i.p. saline injections (hashed bars), and mCherry control-infected cortex with i.p. CNO injections for G_qDREADDs- (J) and G_i/oDREADDs-injected (K) mice. Error bars are SEM. P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***).

Associated with the CNO-evoked astrocyte activation, CNO altered the electrical network activity manifested as an enhancement of the slow-wave delta activity (0–4 Hz) in mice with astrocytes expressing either G_q or G_i/oDREADDs (n = 6 mice for each case; Figures 8H-8K). These effects were not observed when mice with either G_q or G_i/oDREADDs-expressing astrocytes were injected with saline (n=3 mice for each case; Figures 8F, 8G, 8J, and 8K), or when CNO was delivered to mice infected with the control virus GFAP-mCherry that lacks DREADDs (n=3 mice; Figures 8F, 8G, 8J, and 8K). These results indicate that in vivo activation of either G_q or G_i/o GPCR signaling in astrocytes similarly led to the activation of astrocytes, and that these effects were associated with alterations of the neural delta network activity.

DISCUSSION

The present results show that G_q GPCR activation in neurons and astrocytes activated both cell types. In contrast, while neuronal G_i/o GPCR activation inhibited cellular activity, G_i/o GPCR stimulation in astrocytes enhanced their Ca²⁺-based cellular activity. We found that activation of endogenous G_q or G_i/o GPCRs or DREADDs in astrocytes induced Ca²⁺ increases in astrocyte somas and processes and stimulated the release of the gliotransmitter glutamate, which led to an increase in SIC frequency and action potential firing in hippocampal neurons, and impacted neuronal network activity in vivo. Hence, astrocyte G protein-mediated signaling, whether G_q or G_i/o, activated astrocytes by way of an evoked Ca²⁺ response. Therefore, neurotransmitters acting on G_i/o GPCRs proteins directly inhibit neurons but activate astrocytes, which then can feed-forward excite neurons. Consistent with this idea, we have found that astrocyte specific G_q or G_i/o GPCR activation enhanced astrocyte Ca²⁺ events and delta activity in vivo. These results add further complexity to the signaling mechanisms and effects of GPCR-mediated signaling in the brain, revealing important and unexpected functional consequences on the actions of inhibitory neurotransmitters in astroglial-neuronal networks.

Our results show that neuronal G_q GPCR activation either with ACh or G_qDREADDs increase neuronal excitability, both inducing inward currents, membrane depolarizations, and Ca²⁺ elevations through PLC-mediated intracellular signaling. These results are in agreement with previous studies that reported that electrophysiological and Ca²⁺ signal effects of ACh on neurons are mediated by PLC activation (Aiken, Lampe, Murphy, & Brown, 1995; Brown & Yu, 2000; Dasari & Gulledge, 2011; Suh & Hille, 2002; Zhang et al., 2003). G_qDREADD activation has also been shown to induce neuronal Ca²⁺ increases (Alexander et al., 2009; Armbruster et al., 2007). Present data further show that G_qDREADD activation also evokes similar electrophysiological responses as ACh, and that the Ca²⁺ elevations are mediated by the PLC signaling pathway. In contrast, we found that endogenous or chemogenetic G_i/o GPCR activation elicited neuronal outward currents and hyperpolarizations (Armbruster et al., 2007; Logothetis et al., 1987; North, 1989; Wickman et al., 1994; Zhu et al., 2014), and no Ca²⁺ changes. Regarding Ca²⁺ signaling, the βγ subunit dissociation after G_i/o activation can activate either PLC (Singer, Brown, & Sternweis, 1997) or the IP₃R directly (Zeng et al., 2003), leading to Ca²⁺mobilization from internal stores. It was interesting that we did not see this increase in neuronal Ca²⁺, suggesting either that this signaling pathway was not activated, or that we were not able to detect it. This finding contrasts with results from G_i/oDREADD signaling in astrocytes (see below), where G_i/oDREADD activation led to Ca²⁺ increases. Taken together, these data are in agreement with and expand upon previous literature showing that G_i/o activation is inhibitory in neurons.

In contrast to the differential effects of G_q and G_i/o GPCR signaling in neurons, both types of G proteins led to cellular activation in astrocytes in the form of Ca²⁺ increases. Our results show that ACh and G_qDREADD activation similarly increases Ca²⁺ in astrocyte somas and processes through PLC- and IP₃R2-mediated signaling pathways. Our data expand upon existing literature regarding astrocyte responsiveness to ACh (Araque et al., 2002; Chen et al., 2012; Navarrete et al., 2012; Perea & Araque, 2005; Shelton & McCarthy, 2000; Takata et al., 2011). Since previous studies used either a nonselective mAChR antagonist or techniques to manipulate Ca²⁺ release from internal stores, the specific GPCR pathway mediating the Ca²⁺ responses was not known. Here we used PLC and G_i/o inhibitors to block G_q- and G_i/o-mediated effects, respectively. Since G_i/o GPCR signaling has also been reported to activate PLC through the βγ subunit (Singer et al., 1997), U73122 treatment could also block any potential G_i/o-induced activation of PLC. Additionally, the βγ subunits that dissociate from G_i/o GPCR activation could directly activate the IP₃R (Zeng et al., 2003). Accordingly, there could be residual Ca²⁺ elevation with U73122 treatment alone, due to βγ activation of the IP₃R, as well as with PTX treatment alone, due to PLC-activated production of IP₃. However, we observed that the ACh-induced Ca²⁺ responses were largely mediated through G_q-PLC-IP₃ than through G_i/o-βγ-IP₃R2 signaling.

G_qDREADDs have been previously used to specifically activate astrocytes (Adamsky et al., 2018; Agulhon et al., 2013; Bonder & McCarthy, 2014; Bull et al., 2014; Chai et al., 2017; Martin-Fernandez et al., 2017; Scofield et al., 2015), but the Ca²⁺ dynamics and intracellular signaling pathways involved were not fully characterized. We show that G_qDREADD activation stimulated Ca²⁺ increases in a manner akin to endogenous Ca²⁺ dynamics evoked by an endogenous stimulus (i.e., ACh; see Figure 2), and through the canonical signaling of G_q GPCRs, i.e., PLC activation and IP₃-mediated Ca²⁺ mobilization. These results indicate that astrocyte activation via G_qDREADDs closely mimics physiological stimuli.

Unlike in neurons, endogenous and chemogenetic G_i/o GPCR signaling led to astrocyte activation in the form of Ca²⁺ increases in a PTX-sensitive manner, demonstrating that the responses were indeed mediated by the G_i/o type of G protein. This is in line with other studies showing Ca²⁺ elevations in astrocyte somas upon G_i/o-coupled GABA_B receptor activation (Covelo & Araque, 2018; Kang et al., 1998; Meier et al., 2008; Navarrete & Araque, 2008; Perea et al., 2016; Serrano, Haddjeri, Lacaille, & Robitaille, 2006). Here, we show GABA_B activation also induces Ca²⁺ increases in astrocytic processes. It seems highly likely that other endogenous neurotransmitters coupled to G_i/o GPCRs may also activate astrocytes, but further studies are required to test this hypothesis.

Boddum et al. (2016) have recently reported that activation of astrocyte GABA transporters lead to astrocytic Na⁺ concentration elevations and consequent astrocytic Ca²⁺ increases through Na⁺/Ca²⁺ exchange (Boddum et al., 2016). Present results indicate that the GABA effects on astrocyte Ca²⁺ are mediated by GABA_B receptor activation because they are blocked by the receptor antagonist CGP. These results are consistent with two different and complementary mechanisms mediated by GABA transporters and receptors that are revealed by different experimental conditions. Indeed, Boddum et al. used long (≥ 5min) bath applications of GABA, which led to astrocytic Na⁺ concentration raise, consequent increase in astrocytic Ca²⁺ through Na⁺/Ca²⁺ exchange and downstream effects with slow time course (on a temporal scale of minutes). In our study, we used local puff application of GABA (2s) and monitored the short-term effects on astrocyte Ca²⁺ (on a temporal scale of seconds). Therefore, GABA signaling in astrocytes may occur through two complementary mechanisms with different time courses mediated by GABA transporters and receptors.

It has been reported that GABA_B receptor-induced Ca²⁺ elevations can be attributed to PLC signaling (Hirono, Yoshioka, & Konishi, 2001; New, An, Ip, & Wong, 2006) and PTX-sensitive Ca²⁺ mobilization through the IP₃R2 (Mariotti et al., 2016). G_i/o GPCR activation can result in PLC-induced Ca²⁺ mobilization through G_i/o-βγ-PLC signaling (Singer et al., 1997), however, we found no effect of U73122 in GABA- and CNO-induced Ca²⁺ signaling in astrocytes, and instead found that PTX treatment alone abolished the Ca²⁺ increase. Moreover, the lack of CNO-evoked Ca²⁺ increases in IP₃R2^−/− mice suggests that both the G_q and G_i/o protein-induced Ca²⁺ increases require IP₃R2 activation. These data suggest that G_i/oGPCR-induced Ca²⁺ elevations in astrocytes may have been mediated via βγ subunits directly binding the IP₃R2 (Zeng et al., 2003) and not through PLC activation. Future experiments, for example using βγ inhibitors, are necessary to determine the specific signaling cascade leading to Ca²⁺ elevations. Taken together, both GPCR pathways appear to converge on the IP₃R2, but differ in the exact signaling mechanisms leading to IP₃R2 activation. Lastly, we found in vivo that activation of either G_q or G_i/o DREADDs targeted to cortical astrocytes induced an increase in Ca²⁺ events following CNO injection. These data provide evidence that this chemogenetic approach to selectively activate astrocytes remains feasible in a more intact preparation.

Both G_q- and G_i/o-induced astrocyte Ca²⁺ elevations, whether elicited by endogenous receptors or DREADDs, stimulated glutamate release from astrocytes, which was detected as an increase in the frequency of SICs. SICs, which are known to be mediated by activation of neuronal NMDA receptors, have been found to enhance neuronal excitability and increase neuronal synchrony (Angulo et al., 2004; Araque, Sanzgiri, et al., 1998; Fellin et al., 2004; Perea & Araque, 2005; Shigetomi, Bowser, Sofroniew, & Khakh, 2008). While a recent study using G_qDREADDs in astrocytes failed to detect significant changes in SIC frequency upon CNO stimulation (Chai et al., 2017), we observed robust effects. Although the origin of these discrepant results is unclear, it may be due to different sensitivity in detecting SICs, as indicated by our relatively lower mean SIC amplitude.

In addition to an increase in SIC frequency in nearby neurons, selective astrocyte activation via either G_qDREADD or G_i/oDREADD signaling led to an increase in action potential firing of hippocampal neurons. While most reports of neuronal-glial interactions have shown synaptic transmission regulation by astrocytes (Araque et al., 2014), present results add to recent studies showing that astrocytes can also regulate neuronal firing and network activity (Lee et al., 2014; Poskanzer & Yuste, 2011, 2016; Shen, Nikolic, Meunier, Pfrieger, & Audinat, 2017; Tan et al., 2017). Indeed, the observed increase in astrocyte Ca²⁺ activity upon CNO injection in vivo co-occurred with an upregulated delta range of the slow-wave activity. This is consistent with previous studies showing the contribution of gliotransmission to in vivo network function (Fellin et al., 2009; Poskanzer & Yuste, 2011). Interestingly, our findings show that stimulating either the G_q or G_i/o protein pathways in astrocytes led to increases in astrocyte Ca²⁺ and slow-wave activity in the delta range in vivo, suggesting both pathways play similar regulatory roles in the living brain. The absence of effects of CNO administration in slice and in vivo in mice injected with the control virus AAV8-GFAP-mCherry indicates that the observed effects of CNO are not due to off-target effects but rather direct activation of astrocytes. It is important to note that mechanistic correlations should not be made between the results obtained in situ with those obtained in vivo, as these two regions have distinct neuronal-glial network interactions. Rather, we aimed to expand our results in vivo to test the hypothesis that activating astrocytes either via G_q or G_i/o protein signaling pathways influenced neuronal network activity. Taken together, astrocytes have multiple modulatory roles, controlling neuronal information processing via synaptic transmission as well as directly regulating neuronal output and network activity.

The excitation/inhibition balance is important for proper brain function, and its dysfunction may lead to brain disorders, such as epilepsy and autism. In addition to the importance of excitatory transmission in brain communication, inhibitory transmission controls synaptic transmission and neuronal firing frequency (Buzsaki & Chrobak, 1995; Engel et al., 2001; Salinas & Sejnowski, 2001). Beyond the direct effects on cell signaling reduction, inhibition is essential in the operation of neural networks, contributing to the generation of rhythmic activity by synchronizing the discharge of principal cell populations (Bartos, Vida, & Jonas, 2007; Kullmann, 2011; Pouille & Scanziani, 2001). Present findings indicate that inhibition is a specific property of neurons and may be fundamentally different between neurons and astrocytes. While inhibitory neurotransmitters have direct inhibitory effects on neuronal activity through direct activation of neuronal receptors, present data indicate that they may also have an indirect excitatory effect on neurons through activation of astrocytes. Hence, the same neurotransmitters that directly inhibit neurons may also activate astrocytes, which then feed-forward excite neurons (Perea et al., 2016). Therefore, the present results reveal additional complexity of the signaling consequences of excitatory and inhibitory neurotransmitters in network operation and brain function.

Main points.

G_q signaling in neurons and astrocytes leads to cellular activation, but while G_i/o signaling in neurons leads to cellular inhibition, it leads to activation in astrocytes and downstream gliotransmission that modulates neuronal activity.