Maintenance of neural circuit activity requires appropriate regulation of excitatory and inhibitory synaptic transmission. Recently, glia have emerged as key partners in the modulation of neuronal excitability; however, the mechanisms by which glia regulate neuronal signaling are still being elucidated. Here, we describe an analysis of how Ca2+ signals within Drosophila astrocyte-like glia regulate excitability in the nervous system. We find that Drosophila astrocytes exhibit robust Ca2+ oscillatory activity manifested by fast, recurrent microdomain Ca2+ fluctuations within processes that infiltrate the synaptic neuropil. Unlike the enhanced neuronal activity and behavioral seizures that were previously observed during manipulations that trigger Ca2+ influx into Drosophila cortex glia, we find that acute induction of astrocyte Ca2+ influx leads to a rapid onset of behavioral paralysis and a suppression of neuronal activity. We observe that Ca2+ influx triggers rapid endocytosis of the GABA transporter (GAT) from astrocyte plasma membranes, suggesting that increased synaptic GABA levels contribute to the neuronal silencing and paralysis. We identify Rab11 as a novel regulator of GAT trafficking that is required for this form of activity regulation. Suppression of Rab11 function strongly offsets the reduction of neuronal activity caused by acute astrocyte Ca2+ influx, likely by inhibiting GAT endocytosis. Our data provide new insights into astrocyte Ca2+ signaling and indicate that distinct glial subtypes in the Drosophila brain can mediate opposing effects on neuronal excitability.
Complex brain functions require precise control of neuronal activity, which is often disrupted in neurologic disorders such as epilepsy. Here, we show that Drosophila astrocytes exhibit endogenous spontaneous Ca2+ oscillatory activity. Acute elevation of astrocyte Ca2+ suppresses neuronal activity and causes rapid paralysis. This is accompanied by a rapid reduction in the levels of membrane GABA transporter (GAT), a transporter for the inhibitory neurotransmitter GABA. We identify Rab11 as a novel regulator of GAT trafficking that is required for this form of Ca2+-dependent astrocyte regulation of neuronal activity. Our study provides new insight into the acute regulation of neuronal activity by glia.
The regulation of excitatory and inhibitory balance is critically important to neuronal circuit output. Mounting evidence indicates that glial cells are key regulators of neuronal activity. In particular, astrocytes have been suggested to modulate neuronal output via glia-neuron gap junctions (Nedergaard, 1994; Alvarez-Maubecin et al., 2000) or through the release of gliotransmitters, including glutamate, adenosine/ATP, and D-serine (Parpura et al., 1994; Henneberger et al., 2010; Harada et al., 2015; Ma et al., 2016). Despite being nonexcitable cells, astrocytes exhibit robust Ca2+ dynamics (Volterra et al., 2014; Bazargani and Attwell, 2016; Shigetomi et al., 2016). Early work in the field demonstrated that astrocytes could detect released neurotransmitters and initiate slow somatic Ca2+ waves secondary to Ca2+ release from internal stores (Cornell-Bell et al., 1990; Dani et al., 1992). However, a key role of this slower astrocytic Ca2+ oscillatory activity has fallen out of favor following the finding that animals lacking inositol trisphosphate receptor in astrocytes, which lack this type of glial calcium signaling, have relatively normal brain function (Petravicz et al., 2008; Agulhon et al., 2010). More recently, a form of faster, localized microdomain astrocyte Ca2+ oscillatory activity was discovered (Shigetomi et al., 2010). Such Ca2+ fluctuations are mainly contributed through influx of extracellular Ca2+ (Shigetomi et al., 2012; Rungta et al., 2016) and may represent a more important Ca2+ signaling pathway for astrocytes to regulate neural activity (Shigetomi et al., 2013).
Astrocytes also control neural activity through their function in neurotransmitter uptake. In the “tripartite synapse” model, astrocyte terminals remove neurotransmitters and terminate their action via neurotransmitter transporters (Eulenburg and Gomeza, 2010). Astrocyte uptake of GABA, the primary inhibitory neurotransmitter, via GABA transporters (GATs) is critical for proper regulation of the activity of neuronal circuits (Minelli et al., 1996). In humans, mutations in GAT that disrupt GABA uptake are associated with severe forms of epilepsy (Carvill et al., 2015). GAT has been shown to exhibit a very high turnover rate, with a third of the protein residing in a recycling pool (Wang and Quick, 2005), implying that regulation of GAT plasma membrane expression may serve as an acute mechanism to modulate neuronal activity via changes in the rate of GABA uptake. Several factors involved in GAT trafficking have been identified (Scimemi, 2014), but the mechanisms regulating fast turnover of GAT remain elusive.
Drosophila has several classes of glial cells (Freeman, 2015). Astrocyte-like glia (hereafter referred to as astrocytes) and cortex glia are the two main subtypes found intimately associated with neurons in the CNS (Kremer et al., 2017). It has recently been shown that Drosophila cortex glia exhibit near-membrane microdomain Ca2+ oscillatory activity (Melom and Littleton, 2013). Mutations in Zydeco, a glial-specific K+-dependent Na+/Ca2+ exchanger (NCKX), eliminate microdomain Ca2+ oscillations in cortex glia and lead to higher intracellular Ca2+ levels. This increase in cortex glial Ca2+ leads to enhanced seizure susceptibility, as does acute induction of Ca2+ influx through ectopic expression of TRP channels in cortex glia (Melom and Littleton, 2013). These observations have led to a model whereby increased Ca2+ influx into cortex glia leads to neuronal hyperexcitability. However, it is unknown whether Drosophila astrocytes also exhibit Ca2+ activity and whether astrocyte Ca2+ signals have a similar role in exciting neighboring neurons. In the present study, we report that Drosophila astrocytes exhibit spontaneous, microdomain Ca2+ transients, resembling those observed in their mammalian counterparts. Surprisingly, unlike cortex glia, acute Ca2+ influx in astrocytes causes behavioral paralysis and a rapid loss of neuronal activity. We find this suppression of neuronal activity is due in part to rapid GAT endocytosis from astrocyte membranes, leading to enhanced GABA levels in the synaptic cleft. We identify Rab11 as a key regulator of GAT trafficking downstream of astrocyte Ca2+ influx and find that reduction in GAT turnover via suppression of Rab11 function ameliorates the induced suppression of neuronal activity.
Materials and Methods
All Drosophila stocks were maintained on standard media at room temperature (22°C). The following stocks were used: UAS-TrpA1 (BDSC 26263), Alrm-Gal4 (Doherty et al., 2009), UAS-Rab11-GFP (BDSC 8506), GAD1-Gal4 (Ng et al., 2002), UAS-Rab11-RNAi [#I, VDRC 22189 (Dietzl et al., 2007); #II, BDSC 27730 (Perkins et al., 2015)], UAS-Rab11N124I (Satoh et al., 2005), repo-Gal80 (Awasaki et al., 2008), and elav-Gal80 (Yang et al., 2009). To create a stable UAS-TrpA1-myc line, UAS > stop > TrpA1-myc flies (von Philipsborn et al., 2011) were crossed with βTub85D-FLP (BDSC 7196) to excise the transcriptional stop codon. For all experiments, flies of either sex were assayed.
Generation of transgenic flies
To construct the plasmid pBID-UASc-myrGCaMP6s, GCaMP6s cDNA (Addgene #40753) was PCR amplified by designing primers with restriction enzyme cutting sites added on each end. Plasmid pBID-UASc-myrGCaMP5G (Melom and Littleton, 2013) was digested with NotI and XbaI and ligated with GCaMP6s using standard procedure. Cloning was verified by sequence analysis (Genewiz). To generate transgenic Drosophila, plasmids were injected into embryos of a strain with a third chromosome attP docking site (BDSC 8622) by Bestgene.
For testing the temperature sensitive behavioral phenotype of adult flies, 2-d-old flies were used. A total of 10-12 flies were transferred to each vial and placed in a water bath held at the indicated temperature. The number of flies with motor impairment was observed and manually recorded for up to 5 min, with an interval of 15 s in the first minute and 30 s in the following 4 min. Paralysis of adult flies was defined by complete loss of movement. For RNAi screening conducted at 30°C, Alrm > TrpA1 flies do not show 100% penetration of the paralysis phenotype. In this case, the number of flies that showed severe motor impairment and were unable to climb up the vial wall were recorded at each time point. For testing the behavior of headless flies, heads of adult flies were severed by fine scissors under CO2 anesthetization. These flies were then allowed to recover at room temperature in a closed Petri dish for 1 h before being transferred to Petri dishes preheated to the indicated temperature.
For immunofluorescence, 3rd instar stage larvae were dissected in HL3.1 (70 mM NaCl, 5 mM KCl, 10 mM NaHCO3, 4 mM MgCl2, 5 mM trehalose, 115 mM sucrose, and 5 mM HEPES; pH 7.2; Feng et al., 2004) and fixed for 10 min either in Bouin’s fixative or 4% paraformaldehyde. Following washes with PBST (PBS containing 0.1% Triton X-100), larval fillets were blocked for 30 min with 5% normal goat serum in PBST, then incubated with primary antibody overnight at 4°C. Fixed larvae were then washed with PBST. Secondary antibody incubation was performed at room temperature for 2 h. Samples were washed and mounted in vectashield (Vector Laboratories). Antibodies for immunostaining and their dilution were as follows: rabbit anti-GAT, 1:2000 (Stork et al., 2014); mouse anti-Repo, 1:100, (DSHB 8D12); mouse anti-Brp, 1:100 (DSHB NC82); mouse anti-GFP, 1:500 (Thermo Fisher Scientific 33-2600); DyLight 649-conjugated anti-horseradish peroxidase, 1:500 (Jackson ImmunoResearch); Alexa Fluor 488-conjugated goat anti-mouse, fluorecein isothiocyanate-conjugated goat anti-rabbit; and Alexa Fluor 546-conjugated goat anti-mouse, 1:500 (Thermo Fisher Scientific). Images were acquired on a Zeiss LSM700 Laser Scanning confocal microscope with a 40× 1.3 NA, or a 100× 1.3 NA oil-immersion objective (CarlZeiss).
Western blot analysis
Adult flies were snap frozen in liquid nitrogen and five heads per sample were collected and homogenized in 50 µl of 1× Laemmli loading buffer (60 mM Tris, pH 6.8; 10% glycerol; 2% SDS; 1% β-mercaptoethanol; and 0.01% bromophenol blue), followed by centrifugation at 16,000 × g for 10 min to clear the homogenate. Supernatant was analyzed by SDS-PAGE (Bio-Rad 456-1086) and transferred to nitrocellulose membrane. Membranes were blocked with blocking buffer (2% bovine serum albumin, 2% nonfat dry milk, 0.02% Tween 20 in TBS) and then incubated with appropriate antibodies for immunoblotting. Primary antibody incubation were performed with gentle shaking at 4°C overnight followed by four 5-min washes using TBS containing 0.01% Tween 20, secondary antibody incubation for 1 h at room temperature, and four more 5-min washes using TBS containing 0.01% Tween 20. Then the membrane was briefly rinsed with TBS, and the blot was visualized using an Odyssey infrared scanner (Li-Cor). Antibodies for Western blotting: rabbit anti-GAT, 1:5000 (gift from Marc Freeman; Stork et al., 2014); mouse anti-Tubulin, 1:60,000 (Sigma-Aldrich, clone B5-1-2). Alexa Fluor 680-conjugated anti-rabbit and Alexa Fluor 800-conjugated anti-mouse, 1:10,000 (Invitrogen). All antibodies were diluted using blocking buffer.
Extracellular recordings were obtained from dissected 3rd instar larvae using a fire-polished patch electrode. To record central pattern generator (CPG) activity, the larval brain was left intact and care was taken during dissection to avoid damaging segmental nerves. Recordings were amplified using an AxoClamp 2B amplifier and digitized via an Molecular Devices Digidata 1550 (Molecular Devices). Clampfit 6.1 software was used to record and process electrophysiological data. Larvae were constantly perfused via a gravity system with HL3.1 solution containing 1.5 mM Ca2+. Saline was heated using an in-line solution heater (Harvard Apparatus), controlled by a Dual Channel Bipolar Temperature Controller (CL-200A). Additionally, the recording dish was temperature controlled using a Peltier device regulated by a fan-based heat sink and a Dual Channel Temperature Controller (Koolance, Warner Instruments).
Live imaging of astrocyte Ca2+ transients
UAS-myrGCaMP6s was expressed in astrocytes using the Alrm-Gal4 driver. L2 stage larvae were mounted on their ventral side in a small drop of water on a 25 × 60 mm coverslip. A single layer of scotch tape was applied on each side of the larva as a spacer to avoid damaging the animal when an 18 × 18 mm coverslip was placed on top. An upright PerkinElmer Ultraview Vox spinning disk confocal microscope equipped with a high-speed EM CCD camera, and a 43× 1.3 NA oil objective was used for image acquisition. Velocity software was used for data analysis. When imaging, the assembly was flipped so that the ventral side of the larvae faced the objective. Time series were acquired at a speed of ∼9 Hz using a single optical plane in the neuropil layer of the larval ventral nerve cord (VNC).
Statistical analysis was performed using GraphPad Prism 6. Data are presented as mean ± SEM in all figures. Statistical analysis is summarized in Table 1. Superscript letters following p values are used to refer to the corresponding comparisons described in the table.
Drosophila astrocytes show microdomain Ca2+ transients
Expression of a genetically encoded membrane targeted myristoylated Ca2+ indicator (GECI) in cortex glia, a type of glia that associate with neuronal cell bodies in the Drosophila CNS, has revealed that these cells display robust near-membrane microdomain oscillatory activities (Melom and Littleton, 2013). To determine whether astrocytes, the other major glial cell type closely associated with neurons in the Drosophila CNS, show similar Ca2+ activity, we generated transgenic lines expressing a myristoylated form of GECI, myrGCaMP6s, under the control of an astrocyte-specific promoter, Alrm-Gal4 (Fig. 1A; Doherty et al., 2009). We found that astrocytes display robust spontaneous Ca2+ oscillatory activity in membrane microdomains of processes that invaded the synaptic neuropil, largely resembling the Ca2+ transients observed in cortex glia (Fig. 1B; Movie 1). We also monitored astrocyte Ca2+ activity using standard dissected preparations bathed in HL3.1 saline with up to 4 mM Ca2+. Even when the dissections were completed in <5 min, and extreme care was taken to avoid damage to the CNS, astrocyte microdomain Ca2+ transients were rarely observed. Because these spontaneous astrocyte Ca2+ events were very sensitive to injury, we conducted Ca2+ imaging experiments in live, undissected L2 stage larvae that were confined between cover slips (see Materials and Methods).
Astrocyte near-membrane Ca2+ transients had an average duration of 3.69 ± 0.24 s, a half decay time of 1.00 ± 0.06 s, and a mean ΔF/Favg of 34.30 ± 1.70% (n = 46 individual Ca2+ oscillatory events). These events were similar to Ca2+ microdomains previously described in cortex glia, which had an average duration of 1.35 s, and a mean ΔF/Favg of 35% (Melom and Littleton, 2013). Recurrent Ca2+ transients within astrocyte processes were frequently observed in the same area (Fig. 1D), suggesting functional or structural subdomains where Ca2+ entry and removal are sequestered. Taken together, these results suggest that Drosophila astrocytes exhibit spontaneous Ca2+ oscillations that are similar to those previously described in cortex glia.
Acute astrocyte Ca2+ influx via ectopic TRPA1 activation leads to paralysis
It has previously been shown that mutations in zyd, a Drosophila glial-specific K+-dependent Na+/Ca2+ exchanger (NCKX), led to elevated intracellular Ca2+ in cortex glia and enhanced seizure susceptibility, suggesting a role for glial Ca2+ signaling in regulating neuroexcitability (Melom and Littleton, 2013). We were interested in whether astrocyte Ca2+ signaling also regulates neural activity similar to that observed for cortex glia. To test this model, we used transgenic flies expressing TrpA1, a temperature-sensitive Ca2+ channel, under the control of a UAS promoter (UAS-TrpA1; Hamada et al., 2008). Prior studies expressing TrpA1 in cortex glia demonstrated this manipulation resulted in acute and robust seizure activity in neurons, with epileptic behavior observed in both larval and adult flies. When expressed using the astrocyte driver Alrm-Gal4, transgenic flies did not show any abnormality at room temperature (∼22°C). Strikingly, these animals became paralyzed within ∼30 s when acute Ca2+ influx was induced by shifting the temperature to 33°C, whereas control flies did not show any obvious behavioral defects (Fig. 2A; Movie 2). Similarly, acute astrocyte Ca2+ influx led to paralysis in 3rd instar stage Alrm > TrpA1 larvae but not in control animals (Movie 3). Pan-glial expression of Gal80, a Gal4 inhibitor, suppressed the paralysis in Alrm > TrpA1 flies, whereas pan-neuronal expression of Gal80 had no effect, indicating the observed paralysis was caused by acute Ca2+ influx into astrocytes rather than leaky expression in neurons (Fig. 2B).
We next recorded CPG activity from the neuromuscular junction (NMJ) of 3rd instar stage larvae. Both control and Alrm > TrpA1 animals demonstrated normal spaced bursting activity that underlies larval locomotion at room temperature (Fig. 2C). However, when the temperature was ramped from ∼22°C to ∼34°C, neuronal activity in the Alrm > TrpA1 animals quickly diminished, whereas controls were unaffected (Fig. 2C). In contrast to cortex glia, our data indicate that acute astrocyte Ca2+ influx suppresses neuronal activity and leads to paralysis.
Acute astrocyte Ca2+ influx-induced paralysis does not require central brain function
In Drosophila, astrocytes extend fine processes and associate with synapses in the neuropil of both the central brain and the ventral ganglion (Muthukumar et al., 2014). The ventral ganglion is sufficient for conveying many motor functions. Headless flies with intact ventral ganglion survive for many hours and maintain posture, walk, and can learn to avoid electric shock (Booker and Quinn, 1981). It is possible that acute Ca2+ influx into astrocytes could suppress central brain function and cause motor impairment through descending neural circuit(s). Alternatively, the observed paralysis could be the consequence of direct interactions between astrocytes and local neurons in the ventral ganglion. To determine whether central brain function is necessary for conveying the astrocyte Ca2+ influx-induced paralysis, we severed the heads of adult flies and preformed behavioral tests on headless flies. Headless flies expressing TrpA1 in astrocytes rapidly became paralyzed when the temperature was shifted to 33°C (Movie 4). Therefore, acute Ca2+ influx in astrocytes within the ventral ganglion is sufficient to trigger paralysis; central brain function is dispensable for this process.
A candidate screen for modifiers of astrocyte Ca2+ influx-induced paralysis reveals a key role for Rab11 in GAT trafficking
To understand the mechanism of acute astrocyte Ca2+ influx-induced suppression of neuronal activity, we conducted a candidate RNAi screen to identify modifiers of the paralysis phenotype of Alrm > TrpA1 flies. Candidate UAS-RNAi constructs were expressed in astrocytes by crossing with Alrm > TrpA1 flies, and the progeny were subjected to behavioral analysis in vials placed in a heated water bath. The screen was conducted at a lower temperature of 30°C, at which ∼80% of the Alrm > TrpA1 became paralyzed, so that both suppressors and enhancers of motor impairment could be identified. We assembled a collection of ∼140 RNAi lines for proteins that are (1) involved in vesicle transport, (2) enriched in astrocytes, or (3) involved in regulation of Ca2+ signaling. Using this collection, we identified 13 RNAi lines that either enhanced or suppressed the motor impairments of Alrm > TrpA1 flies (Table 2).
Among the RNAi lines that suppressed the paralysis phenotype of Alrm > TrpA1 flies, Rab11-knockdown displayed the strongest effect (Fig. 3A; Alrm > TrpA1: 87.92 ± 2.043%; Alrm > TrpA1, Rab11-RNAi: 31.12 ± 4.813%; n > 100 flies for both groups; p b < 0.0001). We verified the RNAi knockdown result using a dominant-negative mutant form of Rab11, Rab11N124I (Satoh et al., 2005). Overexpression of UAS-Rab11N124I in astrocytes resulted in a comparable suppression of the paralysis in Alrm > TrpA1 flies (Fig. 3B; Alrm > TrpA1: 80.56 ± 7.425%; Alrm > TrpA1, Rab11-DN: 3.75 ± 2.192%; n > 45 flies for both groups; p c = 0.0095). Rab11 is a member of the Rab family of small-GTPases, and it is crucial for recycling endosome function and post-Golgi transport (Welz et al., 2014). It is possible that the reduced susceptibility to paralysis after suppression of Rab11 function in astrocytes is a result of impaired TrpA1 transport to astrocyte membranes and hence reduced Ca2+ influx on temperature shifts. However, following expression of a UAS-TrpA1-myc transgene, we did not observe alteration in TrpA1-Myc localization (Fig. 3C) or intensity with rab11 knockdown (Fig. 3D; control: 1 ± 0.0903, Rab11-RNAi: 0.9323 ± 0.0764, n = 8 single optical sections for each group, p d = 0.5761).
One of the major functions of astrocytes is to uptake neurotransmitters following synaptic activity (Eulenburg and Gomeza, 2010). Drosophila astrocytes express GAT, the transporter responsible for synaptic clearance of the major inhibitory neurotransmitter GABA (Stork et al., 2014). It has been shown that membrane GAT levels are highly dynamic, and changes in GAT level can regulate neuronal excitability (Wang and Quick, 2005; Muthukumar et al., 2014). If GAT function or surface levels were suppressed by Ca2+ influx into astrocytes, one might predict an acute behavioral paralysis due to high synaptic GABA levels and a general suppression of neuronal excitability. We tested whether suppression of Rab11 alters the transport or localization of GAT. Indeed, Rab11-knockdown in astrocytes led to accumulation of large amounts of GAT-containing vesicles in the cytoplasm, although GAT is normally predominantly localized to the plasma membrane (Fig. 3E). Overexpression of Rab11N124I caused a similar defect in GAT localization (Fig. 3E). We also assayed whether suppression of Rab11 function altered the overall level of GAT. Compared with controls, rab11-knockdown animals have similar levels of overall GAT (Fig. 3F,G; control: 1, Rab11-RNAi: 0.8368 ± 0.1102, n = 6 experiments, p e = 0.3125). These data suggest that Rab11 is a novel regulator of GAT trafficking and likely ameliorates the acute astrocyte Ca2+ influx-induced paralysis of Alrm > TrpA1 flies by indirectly regulating GABA uptake.
Acute astrocyte Ca2+ influx leads to reduced membrane GAT
At room temperature, Alrm > TrpA1 larvae display normal GAT expression and localization to astrocyte plasma membrane processes (Fig. 4A,B; Alrm-Gal4: 1 ± 0.03759, Alrm > TrpA1: 1.051 ± 0.06748, n = 8-10 single optical sections for each group, p f = 0.4974). However, after subjecting animals to 33°C, Alrm > TrpA1 larvae showed a severe reduction of surface GAT expression compared with controls (Fig. 4A,B; Alrm-Gal4: 1 ± 0.06333, Alrm > TrpA1: 0.4451 ± 0.05023, n = 8 single optical sections for each group, p g < 0.0001). Large, GAT-containing membranous structures were frequently observed in the cytoplasm following acute Ca2+ influx in astrocytes, indicating Ca2+ elevation may drive GAT endocytosis (Fig. 4C).
These results suggest a Ca2+-dependent endocytotic pathway dynamically regulates membrane GAT levels. We hypothesize that excessive synaptic GABA due to the reduction in astrocyte surface GAT levels triggers the Ca2+ influx-induced paralysis. To test whether increased GABAergic signaling could lead to rapid paralysis, we expressed TrpA1 using a GABAergic neuron driver, GAD1-Gal4 (Ng et al., 2002). These flies became paralyzed within ∼30 s after being transferred to 30°C, indicating excessive GABA synaptic levels result in a strong suppression of neuronal activity and rapid behavioral paralysis in Drosophila (Fig. 5A).
In the present study, we investigated Ca2+ influx in Drosophila astrocytes and the role of astrocyte Ca2+ signals in regulating neuronal activity. We found that similar to cortex glia (Melom and Littleton, 2013), Drosophila astrocytes exhibit spontaneous Ca2+ transients. However, in contrast to cortex glia, acute astrocyte Ca2+ influx leads to suppression of neuronal activity. Elevated cytoplasmic Ca2+ in astrocytes is associated with fast endocytosis of GAT and a reduction in surface expression, which is likely to disrupt GABA uptake and induce rapid paralysis. Suppression of Rab11 led to a disruption in GAT distribution and ameliorated the paralysis caused by astrocyte Ca2+ influx, revealing a novel mechanism of Rab11-dependent control of membrane GAT and neuronal activity.
Drosophila astrocytes demonstrate spontaneous microdomain Ca2+ activity
Current data indicate that astrocytes display diverse forms of intracellular Ca2+ oscillations through distinct cellular mechanisms (Bazargani and Attwell, 2016). Changes in astrocyte Ca2+ levels could be a response to neuronal activity. In isolated rat retinas, light flashing causes an increase in Ca2+ in retinal glial cells (Newman, 2005). In the cortex of anesthetized mice, astrocytes were also found to elevate intracellular Ca2+ in response to whisker stimulation (Wang et al., 2006). However, the Ca2+ oscillatory activities that we observe in Drosophila astrocytes appear to more closely resemble another form of fast, microdomain Ca2+ fluctuation, which has not yet been demonstrated to correlate with neuronal activity. In a rodent neuron-astrocyte coculture system, Ca2+ oscillations in astrocytes appeared to be mediated by spontaneous opening of TRPA1 channels independent of neuronal spiking (Shigetomi et al., 2012). Another study using acute mouse hippocampal brain slices also showed that pharmacological blockage of neuronal activity did not affect either the frequency or the amplitude of localized Ca2+ transients at fine processes of CA1 astrocytes (Rungta et al., 2016).
It is interesting that Ca2+ oscillations in Drosophila astrocytes bear remarkable similarity with those observed in cortex glia (Melom and Littleton, 2013), despite the distinction in the morphology between these two types of glial cells. The similar patterns of Ca2+ oscillations may suggest a common cellular mechanism for their origin. Unlike astrocytes, cortex glia cells are segregated from the synaptic neuropil and, instead, enwrap neuronal cell bodies in the cortex region of the CNS with their thin processes (Ito et al., 1995). It is unclear whether cortex glial might directly sense neuronal activity as a trigger for Ca2+ oscillations due to their lack of direct contact with chemical synapses. With the caveat that we could not use pharmacological methods to directly test the origin of astrocyte Ca2+ transients due to their sensitivity to injury, these microdomain Ca2+ fluctuations are likely a result of spontaneous glial channel opening rather than being directly triggered by neuronal activity.
Distinct functions of glial Ca2+ signals on neuronal activity
We find that acute Ca2+ influx into astrocytes leads to suppression of neuronal activity, an effect opposite to the enhancement of neuronal firing following elevated cortex glia Ca2+ signaling (Melom and Littleton, 2013). Although the exact mechanism is unclear, increased cortex glial Ca2+ level may trigger excessive neuronal activity through their ability to influence the environment around neuronal cell bodies, or through gliotransmission that activates surface receptors on neuronal cell bodies. Based on our results, astrocyte Ca2+ influx appears to balance cortex glia excitation through inhibition of neuronal activity, and together they constitute a glial mechanism to fine-tune the balance between excitatory and inhibitory signals to maintain normal neuronal excitability (Fig. 5B).
Perturbations in glial Ca2+ activity have been implicated in several forms of neurologic disorders (Nedergaard et al., 2010; Shigetomi et al., 2016). For example, it was found that both the basal astrocyte Ca2+ level and the frequency of somatic Ca2+ activity increase in a mouse model of Alzheimer’s disease, which might contribute to alterations in brain function (Kuchibhotla et al., 2009). Although we did not find evidence that the transient induction of astrocyte Ca2+ influx used in our study resulted in long-lasting pathology (data not shown), it will be interesting to explore the role of chronic increases in glial Ca2+ signals in the pathogenesis of neurologic diseases using the Drosophila model.
In the Drosophila CNS, astrocytes invade the entire neuropil with their meshwork of fine processes, and each astrocyte establishes unique and stereotypical territory (Muthukumar et al., 2014; Stork et al., 2014). Although much is known about the distinct origin and functions of astrocytes in different mammalian brain regions (Bayraktar et al., 2014), the functional heterogeneity and/or redundancy of astrocytes in Drosophila is still largely unexplored. Here, as a first step, we show that astrocytes in the ventral ganglion are sufficient to convey the suppression of neuronal activity caused by acute Ca2+ influx.
A mechanism regulating astrocyte membrane GAT and neuroactivity
Modulation of GABA uptake is an important glial mechanism to actively participate in the regulation of neuronal activity. It has been previously shown that global GAT levels in Drosophila astrocytes are developmentally regulated and sensitive to GABAergic neuronal output during synaptogenesis (Muthukumar et al., 2014). Suppression of metabotropic GABAB receptor signaling reduces GAT levels and suppresses seizure induction in hyperexcitable Drosophila seizure mutants, likely through dampening the rate of GABA uptake (Muthukumar et al., 2014).
Our study suggests the existence of a new mechanism for astrocytes to acutely regulate neuronal activity through fast modulation of GAT turnover. Compared with other known membrane proteins, GAT has one of the highest turnover rates (Tanner and Lienhard, 1987; Harwood and Pellarin, 1997; Menne et al., 2002; Wang and Quick, 2005). Plasma membrane GAT levels are dynamically regulated via a balance between fast endocytosis and exocytosis in primary rat cortical neuron cultures (Wang and Quick, 2005). Pharmacological activation of protein kinase C (PKC) can trigger enhanced endocytosis of GAT, leading to reduced membrane expression and reduction in GABA uptake (Wang and Quick, 2005). Here, our results suggest that astrocyte Ca2+ influx induces a rapid decrease in membrane GAT levels via endocytosis, which correlates with acute suppression of neural activity, representing a mechanism for glial cells to actively modulate neuronal function.
Our data reveal Rab11 as a novel regulator of GAT trafficking in Drosophila astrocytes. The accumulation of GAT-containing vesicles in rab11-knowdown animals likely represents a strong suppression of post-Golgi transport of GAT to the plasma membrane. However, it is important to note that suppression of Rab11 function did not totally abolish GAT localization to astrocyte membranes (Fig. 3D). Consistently, rab11-knowdown flies do not have strong locomotion defects like GAT-deficient animals (Stork et al., 2014). Indeed, it has been shown that GAT is delivered to the plasma membrane via multiple pathways (Reiterer et al., 2008). Since rab11-knockdown did not alter total GAT levels (Fig. 3E,F), constitutive membrane GAT in these animals seems sufficient to support baseline GABA uptake. Although primarily considered a regulator of recycling endosomes, Rab11 has been shown to promote endocytosis through indirect mechanisms. In the Drosophila eye, suppression of Rab11 function decreases the rate of endocytosis and eliminates the formation of endocytosis-dependent multivesicular bodies (Satoh et al., 2005). Ectopic expression of a dominant negative form of Rab11 also disrupts endocytosis in cultured Drosophila Garland cells (Satoh et al., 2005). Consistent with these observations, our data support a model in which suppression of Rab11 function blocks GAT endocytosis triggered by astrocyte Ca2+ influx, leading to sustained GABA uptake and a suppression of the paralysis and loss of neural activity normally observed following excessive astrocyte Ca2+ influx (Fig. 5C).
Taken together, our study reveals an important function of glial Ca2+ signaling in regulating neuronal activity. Acute modulation of the rate of neurotransmitter uptake through Ca2+-dependent glial signaling provides a robust mechanism for astrocytes to actively modulate the function of neural circuits.
Acknowledgements: We thank Marc Freeman, Tzumin Lee, Yuh Nung Jan, Akiko Satoh, the Developmental Studies Hybridoma Bank at the university of Iowa, the Bloomington Drosophila Stock Center at Indiana University, the Vienne Drosophila Resource Center, and the Drosophila stock center at National Institute of Genetics of Japan for providing reagents and flies strains.
The authors declare no competing financial interests.
This work was supported by the HHS/NIH Grant NS40296.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license, which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed.