Side-by-side comparison of the effects of Gq- and Gi-DREADD-mediated astrocyte modulation on intracellular calcium dynamics and synaptic plasticity in the hippocampal CA1

Astrocytes are one of the most abundant resident cells of the brain, well known for providing metabolic support, ensuring homeostasis, and maintaining the blood–brain barrier. Astrocytes express a wide array of G protein-coupled receptors (GPCRs), allowing them to respond to synaptic activity. Indeed, astrocytes have emerged as key players at quad-partite synapses and were found to actively participate in bidirectional neuron-astrocyte communication via transient variations in intracellular Ca²⁺ [1, 2] and the release of gliotransmitters, such as glutamate, γ-aminobutyric acid (GABA), and/or adenosine triphosphate (ATP)/adenosine (reviewed by [3, 4]). Astrocytic Ca²⁺ signaling generally remains a complex and poorly understood concept [2] and is often dysregulated in various CNS disorders [5‐9]. Therefore, it is of utmost importance to understand how astrocytes tune neuronal activity and modulate higher brain functions, as this can aid in the discovery of potential novel therapeutic strategies for multiple brain disorders. It is well-established that increases in astrocytic Ca²⁺ transients can be elicited by activation of Gq-coupled GPCRs [3, 10‐13]. This astrocytic Gq-GPCR activation recruits a similar downstream intracellular cascade as observed in response to Gq-GPCR activation in neurons. Briefly, astrocytic Gq-GPCR activation recruits the Gαq protein, which in turn leads to 1,4,5-trisphosphate (IP₃)-mediated Ca²⁺ release from internal stores [13]. In contrast, although there is evidence of a role of Gi-GPCR signalling in astrocytes, the nature of the signalling has been less consistently reported in the literature. Canonically, Gi-GPCR activation recruits the G_αi/o protein which inhibits adenylate cyclase and thus, reduces intracellular cyclic adenosine monophosphate (cAMP), while the Gβγ subunit activates inwardly rectifying K⁺ channels and/or inhibits Ca²⁺ channels [14]. Indeed, a reduction in cAMP has been reported upon Gi-GPCR activation in astrocytes [15], suggesting that these receptors are coupled to the expected G_αi/o protein. Regarding the effect on intracellular Ca²⁺, either no effect on Ca²⁺ levels [16] or decreased baseline Ca²⁺ levels, presumed to be the resulting action of the Gβγ subunit [17], have been reported after Gi-GPCR activation in hippocampal [16] and ventral midbrain astrocytes [16, 17], respectively. However there is also ample evidence showing an increase in astrocytic Ca²⁺ transients upon Gi-GPCR activation, such as activation of GABA_BRs and dopamine D1 and D2-type receptors in the hippocampus [18‐21], GABA_BRs in the somatosensory cortex [22], GABA_BRs in the ventral tegmental area [23] and the µ-opioid receptor (MOR) and dopamine D1 receptors in the nucleus accumbens [24, 25]. Moreover, these GABA_BR and MOR-induced increases in cytosolic Ca²⁺ were suggested to be derived from intracellular stores [19, 22, 24, 26], as these Ca²⁺ transients were shown to be abolished in inositol 1,4,5-trisphosphate receptor type 2 (IP₃R2) knock-out mice [19, 22, 24] or when IP₃-gated internal Ca²⁺ stores were depleted [19, 26]. The underlying mechanisms mediating astrocytic Ca²⁺ transients via Gi-GPCR activation remain elusive. In addition, activation of either Gq- or Gi-GPCRs on astrocyte surfaces impact synaptic transmission and plasticity via either Ca²⁺-dependent [10, 19, 27‐31] or Ca²⁺-independent gliotransmitter release [16, 32]. Interestingly, the magnitude and temporal characteristics of astrocyte-mediated synaptic modulation, were found to be significantly affected by the pattern of interneuron activity, that caused GABA release and in turn stimulated hippocampal astrocytic GABA_BRs [33]. This means that astrocytes can interpret and integrate time- and frequency-related variations in stimulation and, thus suggests that differences in experimental protocols related to GPCR agonist application, such as duration and concentration, could potentially complicate correct interpretation of the obtained results.

Unfortunately, further investigation of reciprocal neuron-astrocyte interactions have long been hindered due to the lack of selective tools to modulate astrocytic GPCR signalling [34, 35]. Designer receptors exclusively activated by designer drugs (DREADDs) are an interesting tool to tackle these issues as they use GPCR-mediated intracellular signal transduction pathways naturally present in astrocytes. DREADDs have been increasingly used to study effects on intracellular GPCR-dependent astrocytic Ca²⁺ signalling and on synaptic transmission and/or plasticity. Similar to endogenous GPCRs, Gq- and Gi-coupled DREADDs have both been shown to induce Ca²⁺ responses in astrocytes [36‐42]. However, while Gq-DREADD-induced astrocyte modulation is consistently reported to increase astrocytic Ca²⁺ events or baseline Ca²⁺ levels [36‐42], the effect of Gi-DREADDs on astrocytic Ca²⁺ signalling remains less clear [39, 40, 43]. Differences in intracellular Ca²⁺ responses upon Gi-DREADD mediated astrocyte modulation have been reported, not only between brain regions [40], but also within the same region i.e., hippocampus [39, 40, 43]. More specifically, application of the DREADD agonist, clozapine-N-oxide (CNO), to hippocampal slices transduced with Gi-DREADD has been shown alternately to elicit transient increases in intracellular Ca²⁺ [39], to decrease intracellular baseline Ca²⁺ levels [43] or to have no discerable effect [40].

To date, only a limited amount of evidence has been gathered regarding gliotransmitter release upon Gq- or Gi-DREADD activation in hippocampal astrocytes. For instance, slow inward currents (SICs) in surrounding neurons were observed upon CNO bath application in hippocampal slices expressing either astrocytic Gq-DREADD or Gi-DREADD [39], which is typically attributed to the activation of extrasynaptic neuronal N-methyl-d-aspartate receptors (NMDARs) via astrocytic released glutamate [44]. Interestingly, Chai et al. [40] did not observe any changes in SIC amplitude or frequency, nor did they observe increased glutamate release using a genetically encoded glutamate sensor (iGluSnFR) upon CNO bath application in Gq-DREADD transduced hippocampal astrocytes. In addition, the release of d-serine has been implied upon astrocytic Gq-DREADD activation [36]. It is however accepted that activation of both Gq- and Gi-coupled DREADDs can alter hippocampal synaptic plasticity [16, 36, 39, 43]. For instance, Gq-DREADD-mediated astrocytic activation was found to induce long-term potentiation (LTP) in the Cornu Ammonis 1 (CA1) of the hippocampus without any additional, sub-threshold electrical stimulus [36], while Gi-DREADD activation was proposed to reduce LTP threshold, as LTP was induced solely upon application of an additional 40-Hz stimulus [16].

In this research paper we compare in parallel Gq- and Gi-DREADD-mediated activation of hippocampal astrocytes and evaluate their effects on intracellular Ca²⁺ dynamics in astrocytes and synaptic plasticity in the hippocampal CA1 Schaffer collateral pathway. In addition, we emphasize the importance of reporting the application duration and concentration of DREADD agonists, to allow proper comparison of the effects on astrocytic Ca²⁺ and synaptic plasticity in ex vivo slices.

In the past few years, chemogenetic approaches have been used to evaluate the crucial role of astrocytes in higher brain functions and this has drastically broadened our understanding of these star-like cells. However, proficient comprehension of the intracellular processes and subsequent effects on the surrounding neuropil are imperative to fully grasp their contribution in physiological and pathological conditions. Ongoing research on this topic unveiled astrocyte-mediated effects on synaptic plasticity [16, 36], but also, several inconsistencies on astrocytic Ca²⁺ signalling [39, 40, 43] preventing unambiguous characterization of the downstream effects of the DREADDs.

In the present study, we performed a side-by-side comparison of astrocyte activation using the two most commonly used DREADDs, hM3Dq (Gq protein-coupled GPCR) and hM4Di (Gi protein-coupled GPCR), and studied their effects on intracellular astrocytic Ca²⁺-signalling and synaptic plasticity in the hippocampal CA1 neuronal network. We showed that 35-min application of DREADD agonist CNO (10 µM), induced a robust increase in intracellular astrocytic Ca²⁺ events in Gq-DREADD, but not Gi-DREADD, expressing CA1 hippocampal astrocytes compared to control vector transduced astrocytes. In addition, we showed that both Gq- as well as Gi-mediated astrocyte modulation successfully potentiated synaptic transmission in the Schaffer collateral pathway, which was long-lasting and in the complete absence of any additional high-frequency stimulation.

Our results show that long-term Gq-DREADD activation in astrocytes resulted in a robust long-lasting increase in Ca²⁺ events, which is in line with the literature [36] and expectations with regard to the well-known downstream signalling upon Gq-coupled GPCRs. In addition, after switching bath perfusion from CNO-containing aCSF to CNO-free, normal aCSF, the amount of Ca²⁺ transients returned to baseline, suggesting wash-out of the DREADD agonist. However, we cannot exclude that other mechanisms, such as desensitization of the Gq-DREADD, occur, causing this decrease in Ca²⁺ transients. Interestingly, upon astrocytic Gi-DREADD activation, different effects on astrocytic Ca²⁺ have been previously described. For instance, Kol et al. [43] found that Gi-DREADD activation (10 min, 10 µM) induced a slight decrease in baseline Ca²⁺ levels of hippocampal astrocytes, while Durkee et al. [39] reported increases in intracellular Ca²⁺ upon puff application of CNO (2–5 s, 1 mM) in Gi-DREADDs expressing hippocampal astrocytes. In addition, Chai et al. [40] did not find consistent increases in intracellular Ca²⁺ hippocampal astrocytes during CNO bath application (4 min, 1 µM). Likewise, we observed no changes in the frequency of Ca²⁺ transients of Gi-DREADD transduced slices following DREADD agonism (35 min, 10 µM). Although, the mechanisms involving such variations on intracellular Ca²⁺ are hitherto still unknown, we propose that these discrepancies might be ascribed to a difference in astrocyte stimulation intensity, i.e. combination of the concentration and duration of the Gi-GPCR agonist application. Intriguingly, recently individual hippocampal astrocytes were found to be able to release both ATP/adenosine and glutamate and this in a time-dependent and activity-sensitive manner upon GABA released during interneuron activity [33]. This suggests that the type and intensity of astrocyte stimulation is crucial in determining which downstream signalling pathway is induced by astrocytic GABA_BR activation. However, whether this differentiation in downstream signalling pathway upon GABA_BR activation affects Ca²⁺ responses and if this is the case for Gi DREADDs as well, remains to be investigated. Nonetheless, astrocytic Gi-DREADD modulation has been shown to inhibit astrocytic cAMP signalling [49, 50], thus it is reasonable to assume that the hM4Di-DREADD receptor is coupled to the G_αi/o protein, but that the βγ subunits of the G protein act differently upon activation depending on the intensity of stimulation. Interestingly, Gi-DREADD-induced Ca²⁺ elevations in hippocampal astrocytes upon short-term, high concentration CNO application, were recently shown to be mediated via direct binding of the βγ subunits to IP₃R2 [39].

Next, we investigated the effects of either Gq-DREADD or Gi-DREADD-mediated astrocyte activation on synaptic plasticity in the Schaffer collateral-CA1 hippocampal network. As experimental protocols and slicing conditions can affect the characteristics of cells and networks and, thus experimental outcomes, we lay out the details of the methods provided in two papers focusing on DREADD-based LTP induction [16, 36] in comparison to our experimental procedures, in Table 1. Upon Gq-DREADD activation in hippocampal astrocytes, we observed LTP in the absence of any high frequency electrical stimulus, as previously reported [36]. Significant potentiation of fEPSP responses occurred following the wash-out phase of the DREADD agonist CNO. This is in contrast to observations made by Adamsky et al. [36], who reported a greater magnitude of potentiation that was apparent immediately during the first stimulation after CNO application. This discrepancy may be explained by differences in experimental set-up (see Table 1). For example, we prepared our slices in sucrose containing cutting solution, a manipulation that has previously been shown to reduce the magnitude of LTP [51].

The most novel finding of our study is the fact that we show for the first time that Gi-DREADD activation of astrocytes induces de novo long-lasting potentiation, as significant increases in fEPSP amplitude were already observed from the first stimulation after CNO application onwards, which lasted longer than 30 min after wash-out. Previously, Gi-DREADD activation of hippocampal astrocytes was shown to reduce the threshold of LTP induction, as LTP was elicited when 40-Hz stimulation was performed in addition to CNO bath application [16]. However, in our study we showed that low-frequency stimulation (0.033 Hz) is sufficient to evoke LTP. This might be ascribed to variations in astrocytic GPCR stimulation, i.e. duration and concentration of GPCR agonist application, as previously discussed, which can affect outcomes on synaptic plasticity [33]. In addition, other experimental differences, for instance, slice recovery conditions, such as submerging of the slice, in combination with Mg²⁺ concentrations are critical players in permitting metaplastic effects on LTP duration [52], which were not fully disclosed in [16]. Moreover, Ca²⁺ uncaging in astrocytes, has also been reported to reduce LTP threshold, and only to induce LTP in combination with an additional depolarization [31]. However, as no effect on astrocytic Ca²⁺ transients were observed upon Gi-DREADD activation, our data suggests that Gi-DREADD activation of astrocytes affects synaptic plasticity in Ca²⁺-independent manner.

Astrocytes actively shape synaptic transmission via release of neuroactive substances, such as gliotransmitters (glutamate, GABA, ATP, adenosine, and d-serine), synaptogenic cues (e.g. thrombospondins, tumor necrosis factor alpha) and metabolic substrates (e.g. lactate, lipids) (reviewed by [53]). The context-specific release of gliotransmitters is regulated via activation of GPCRs, which in turn can induce enhanced Ca²⁺ transients (reviewed by [54]). Indeed, multiple endogenous astrocytic Gq-GPCRs and Gi-GPCRs have been reported to elicit increases in Ca²⁺ transients, followed by a release of active substances and/or gliotransmitters [10, 18, 19, 22, 24, 27‐29, 31, 55]. Likewise, activation of the exogenous Gq-DREADDs expressed in hippocampal astrocytes, was previously proposed to induce synaptic potentiation via Ca²⁺-dependent release of d-serine [36]. The underlying mechanisms of GPCR-induced Ca²⁺-dependent gliotransmitter release can be divided in two major categories: vesicular exocytosis, relying on soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complexes [56, 57] and non-vesicular release via gliotransmitter-permeable ion channels (such as opening of the Ca²⁺-regulated Bestrophin-1 channel [58, 59]). Besides Ca²⁺-dependent release of gliotransmitters, there is some evidence showing that activation of astrocytic GPCRs is able to trigger Ca²⁺-independent release of gliotransmitters. Particularly, several Gi-GPCRs, such as the µ-opioid receptor (MOR) [16, 60], GABA_B receptor, cannabinoid receptor 1, and adenosine 1 receptor [58], have been shown to trigger glutamate release via a Ca²⁺-independent mechanism upon activation. This Ca²⁺-independent release of glutamate was found to rely on the interaction of the dissociated Gβγ complex with the glutamate-permeable, TWIK-related K⁺ (TREK-1) channel i.e., a two-pore domain potassium channel [58]. Intriguingly, such Ca²⁺-independent mediated glutamate release has been shown to affect synaptic plasticity. More specifically, astrocytic MOR activation, co-localized with and known to release glutamate via the TREK-1 channel [60], was shown to induce LTP in the presence of an additional high frequency (40-Hz) stimulus [16]. This effect was ascribed to activation of presynaptic metabotropic glutamate receptors 1 (mGluRs1) and, notably, could be successfully mimicked by activation of Gi-DREADDs in astrocytes [16]. However, also other mediators can be released independent of increases in Ca²⁺ transients. For instance, activation of astrocytic mGluR3, a Gi-GPCRs, was recently shown to release prostaglandin E₂ (PGE₂) in the presence of a blocker of the endoplasmic reticulum Ca²⁺ pump (cyclopiazonic acid). Moreover, they showed that this astrocytic PGE₂ release led to altered synaptic plasticity, i.e. enhanced mEPSPs in surrounding neurons [61]. Noteworthy, peripheral astrocytic processes (PAPs) are critically involved in synaptic plasticity and are known to display spotty Ca²⁺ transients in their microdomains [27, 62, 63]. Although no elevations in Ca²⁺ transients were observed in our study, assessment of Ca²⁺ transients was restricted to the soma and large processes, as Ca²⁺-imaging was performed using two-photon Ca²⁺ imaging after bolus loading of a chemical dye (OGB-1 AM), which is known to have limited spatial resolution [64‐66]. Imaging of PAPs typically requires high-resolution and time-consuming imaging techniques, such as 3D two-photon Ca²⁺ imaging in combination with genetically-encoded Ca²⁺ indicator [64] or electron microscopy [66]. Therefore, we cannot fully exclude that there is no Ca²⁺-dependent process in the PAPs involved in the Gi-DREADD-mediated synaptic plasticity in our experiments. Nevertheless, taken together, these findings emphasize the possibility of Gi-GPCR-mediated Ca²⁺-independent release of active substances, which in turn can affect synaptic plasticity. Particularly, Gi-GPCR-mediated Ca²⁺-independent glutamate release via the two-pore domain potassium TREK-1 channel is the most promising mechanism for the observed Ca²⁺-independent Gi-DREADD-induced de novo potentiation. Notwithstanding, additional data should be gathered to verify if this Gi-DREADD-induced synaptic potentiation is Ca²⁺-independent.

Our present results suggest that Gq- and Gi-DREADD, based astrocyte modulation, rely on different mechanisms to induce LTP. This postulation is reinforced by the different responses observed during long-term Ca²⁺ signalling in Gq- and Gi-DREADDed astrocytes. Activation of Gq-DREADDs expressed in astrocytes was previously proposed to induce NMDAR-dependent LTP formation via the release of d-serine [36]. More specifically, slices pre-treated with 10 μM d-serine, failed to induce synaptic potentiation upon additional CNO-mediated Gq-DREADD activation, likely due to occlusion [36]. Moreover, another study showed that clamping of intra-astrocytic Ca²⁺ levels suppressed LTP at nearby synapses, which was rescued by adding 10 μM d-serine [29]. Taken together, this indicates Ca²⁺-dependent d-serine from astrocytes release is crucial for LTP induction. On the contrary, previous research showed that activation of Gi-DREADDs in astrocytes lowered the threshold for LTP induction, which was ascribed to astrocytic glutamate release and the subsequent activation of neuronal presynaptic metabotropic glutamate receptors 1 (mGluRs 1) [16]. This is in line with the proposed Ca²⁺-independent TREK-1 channel-mediated release of glutamate, as these channels were reported to induce a fast release that facilitates high extracellular peak concentrations of glutamate, necessary to activate mGluRs [58]. Agonism of group I mGluRs (mGluR1 and mGluR5) has been previously reported to facilitate LTP [67‐69], however an additional mechanism is required to induce de novo LTP in CA1 synapses [70, 71]. It is therefore likely that the observed astrocytic Gi-DREADD-induced de novo potentiation is evoked via an additional mechanism, besides mGluR agonism. Indeed, astrocyte-mediated glutamatergic signalling has been repeatedly reported to induce LTP via various mechanisms in addition to mGluR activation [3, 10]. Moreover, even though TREK-1 channel-mediated glutamate release predominantly activates mGluRs [16, 58, 60], it was also reported to activate NMDARs [58]. Therefore, it is expected that Gi-DREADD activation in astrocytes induces activation of both glutamate receptors i.e., NMDAR and mGluR, and thus triggers an NMDAR-dependent LTP. Additionally, astrocytes have also been reported to be involved in NMDAR-independent LTP [10]. Specifically, application of acetylcholine induced an NMDAR-independent, cholinergic-LTP in CA3-CA1 synapses. This effect relied on muscarinic acetylcholine receptor (mAChR)-mediated glutamate release from astrocytes, which subsequently activated neuronal mGluRs, in addition to simultaneous depolarization of the postsynaptic neuron [10]. However, in this study, mAChR-mediated glutamate release was shown to rely on increased intracellular Ca²⁺ in the astrocytic processes [10]. In addition, ATP, shown to be co-released with glutamate from individual hippocampal astrocytes [33], is able to further facilitate LTP induction upon group I mGluR and NMDAR co-activation [72]. Notably, a caveat of the current study is that we cannot fully exclude that the small, but not significant, increase in fEPSP amplitude after a break in the stimulation protocol observed in control slices (Fig. 2d), is crucial for NMDAR co-activation together with Gi-DREADD-mediated mGluR1, and thus contributes to Gi-DREADD-mediated potentiation. Nonetheless, it should be further investigated to confirm whether Gi-DREADD activation truly elicits an NMDAR-dependent LTP. Therefore, we encourage future research comparing the effects of both DREADDs on synaptic plasticity and to elucidate the underlying mechanisms. Moreover, we wish to emphasize the importance of reporting detailed descriptions of the experimental procedures to avoid discrepancies.

In summary, we showed that Gq-DREADD activation in astrocytes triggered a robust increase in Ca²⁺ events, which was not observed upon astrocytic Gi-DREADD activation. Importantly, we demonstrated that both Gq-DREADD and Gi-DREADD activation of astrocytes are sufficient to induce long-lasting potentiation of CA1 synapses, even in the absence of high frequency stimulation. In particular, we show that Gi-DREADD activation of hippocampal astrocytes induces de novo potentiation, potentially in a Ca²⁺-independent manner. Further investigation is required to fully characterize the downstream mechanisms involved in the modulation of intracellular Ca²⁺ signalling upon Gi- and Gq-DREADD mediated astrocyte activation. Comprehensive knowledge of astrocytic GPCR downstream signalling will help unveil the various roles of the excitatory and inhibitory neuro- and gliotransmitters in the hippocampal network and the resulting effects on brain function. Moreover, recent research has put astrocytic DREADD modulation forward as a potential therapeutic strategy in the treatment of CNS diseases [49, 73, 74] and understanding the relevant signalling cascades in physiological conditions will be the first step towards translation to the clinic.