Astrocytic GABA transporter controls sleep by modulating GABAergic signaling in Drosophila circadian neurons

Abstract

A precise balance between sleep and wakefulness is essential to sustain a good quality of life and optimal brain function. GABA is known to play a key and conserved role in sleep control, and GABAergic tone should, therefore, be tightly controlled in sleep circuits. Here, we examined the role of the astrocytic GABA transporter (GAT) in sleep regulation using Drosophila melanogaster. We found that a hypomorphic gat mutation (gat^33-1) increased sleep amount, decreased sleep latency, and increased sleep consolidation at night. Interestingly, sleep defects were suppressed when gat^33-1 was combined with a mutation disrupting wide-awake (wake), a gene that regulates the cell-surface levels of the GABA_A receptor resistance to dieldrin (RDL) in the wake-promoting large ventral lateral neurons (l-LNvs). Moreover, RNAi knockdown of rdl and its modulators dnlg4 and wake in these circadian neurons also suppressed gat^33-1 sleep phenotypes. Brain immunohistochemistry showed that GAT-expressing astrocytes were located near RDL-positive l-LNv cell bodies and dendritic processes. We concluded that astrocytic GAT decreases GABAergic tone and RDL activation in arousal-promoting LNvs, thus determining proper sleep amount and quality in Drosophila.

Keywords: GABA; GAT; astrocytes; large ventral lateral neurons; sleep.

Figure 1.. gat^33-1 is a hypomorphic mutant of GABA transporter (GAT):

(A) Schematic of the gat gene locus. TALEN targeting site in the first exon is indicated. (B) Nucleotide sequence of the targeted region of gat. The grey boxes indicate the target sites for TALEN-mediated mutagenesis in the 1^st exon, the start codon is indicated in green. gat^33-1 mutants harbor a 7 nucleotides deletion that removes a Bcc1 endonuclease restriction site and leads to a frameshift and early termination of the GAT protein. (C) Protein sequence of the GAT N-terminus. Amino acids encoded by exon 1 and 2 are indicated in orange and blue respectively. The beginning of the first transmembrane domain (TM) is indicated by a light blue box. The frameshift in gat^33-1 should lead to an early termination after 20 amino acids (red), however downstream methionines (Green boxes) could act as alternative start sites. (D) Schematic overview of the putative GAT membrane topology. The sites of the two putative alternative start sites in gat^33-1 mutants are indicated (Blue arrows) and they would lead to 22aa or 64aa truncations of the cytoplasmic N-terminus. (E) Western blot analysis of GAT protein of wild type and gat³³⁻¹ mutant fly heads using C- and N-terminal anti-GAT antibodies. The C-terminal anti-GAT antibody recognizes a strongly reduced and truncated GAT specific band (white arrow) in gat^33-1 mutant extracts while the N-terminal anti-GAT antibody only recognized a GAT specific band in wild type control flies (black arrow). See also Figure S1.

Figure 2.. GABA transporter (GAT) is required for normal sleep-wake behavior

(A) Sleep profiles of gat^33-1 (green, n=110) and control w¹¹¹⁸ (black, n=93) male flies in 12/12 h LD cycle showing enhanced amount of sleep in the gat hypomorphic mutant during both day and night. (B-E) Total sleep amounts (min) (B), number of sleep episodes (C), mean sleep episode’s durations (min) (D), and mean wake episode’s durations (min) (E) during daytime and during night showing increase in sleep consolidation in gat^33-1. (F) Latency to sleep (min) during the day and nighttime indicates accelerated sleep initiation in gat^33-1. (G) Sleep profile of gat^33-1 (green, n=32) and control w¹¹¹⁸ (black, n=32) male flies in constant darkness (DD). Results show that enhancement of gat^33-1 sleep is not light dependent. (H) Total amount of sleep during DD. (I) Mean activity per min while moving during DD. (J) Mean activity per min while moving during LD cycle. Error bar represents SEM. For normally distributed data, a two-tailed Student’s t-test was used. In case of non-gaussian distribution, a Mann–Whitney U-test was performed, and statistical significance is shown as ***P < 0.001; **P < 0.01; and ‘ns’ as not significant. See also figure S2–3 and Video S1–2.

Figure 3.. gat^33-1 sleep defects are rescued by expressing gat in astrocytes

(A) Sleep profile of male flies with genotype gat^33-1 (green, n=90), control w¹¹¹⁸ (black, n=45), UAS-gat;; gat^33-1 (olive green, n=64), alrm-Gal4;; gat^33-1 (sea-green, n=44), alrm-Gal4/UAS-genomic-gat;; gat^33-1 (red, n=96), and 1b12;; gat^33-1 (crimson red, n=37) in 12/12 h LD cycle. The results show that GAT functions in astrocytes. (B-D) Total sleep amounts (min) (B), mean sleep episode’s durations (min) (C), and number of sleep episodes (beam breaks) (D) during day and night. (E) Latency to sleep (min) during day and nighttime. (F) Mean activity per min while moving during day and night. Error bar represents SEM. For normally distributed data, one-way ANOVA followed by a Tukey post-hoc test was performed. For non-gaussian distribution, a Kruskal-Wallis test with Dunn’s post-hoc test was performed. Different alphabetical letters (a, b, c, and d) illustrate statistically significant differences between groups. The same statistical tests were used in Figures 5–7, as well as Supplemental Figures S4–S7. See also Figure S4.

Figure 4.. RDL and PDF expressing wake promoting clock neurons (l-LNvs) are in close proximity with GAT expressing astrocytes.

(A) Confocal analysis of w¹¹¹⁸ wild type fly brains for localization of RDL (blue) in PDF (red) positive l-LNvs. Astrocytes are expressing membrane bound GFP (green) (alarm-Gal4/UAS-CD8-GFP). (a-a3) Labeling in large ventral lateral neurons (l-LNvs). (b-b3) Labeling in medulla region of optic lobe. l-LNvs neurons also extend its projections in the medulla. (c) 3D view of image-stacks demonstrate projections of astrocytes (green) are extending towards /close to the cell body of l-LNvs and co-localized with RDL, and PDF. (B) Magnified view of l-LNvs labeled with RDL (blue) (a), astrocytes (green) (b), PDF (red) in l-LNvs (c). Merged image (d) was used for orthogonal projection (d’ and d”) where yellow arrows point to astrocytes situated very closely to RDL, and PDF expressing in l-LNvs. (C) Magnified view of the medulla region in the optic lobe labeled for RDL (blue) (a), astrocytes (green) (b), and PDF (red) (c). Merged image (d) was used for orthogonal projections (d’ and d”) where yellow arrows point to astrocytes processes approaching RDL and PDF positive l-LNvs projections innervating into medulla. (D) Confocal analysis of w¹¹¹⁸ brains for localization of GAT (blue), PDF positive circadian neurons (red), and astrocytes expressing membrane bound GFP (green) (alrm-Gal4/UAS-CD8-GFP). (a-a3) Labeling in large ventral lateral neurons (l-LNvs). (b-b3) Labeling in the medulla region of the optic lobe. (E) Magnified view of l-LNvs region labeled with GAT (blue) (a), astrocytes (green) (b), PDF (red) in l-LNvs (c). Merged image (d) was used for orthogonal projection (d’ and d”) where yellow arrows point to GAT present in astrocytes and closely placed to PDF expressing in l-LNvs. (F) Magnified view of the medulla region of the optic lobe labeled for GAT (blue) (a), astrocytes (green) (b), and PDF (red) (c). Merged image (d) was used for orthogonal projections (d’ and d”) where yellow arrows point to GAT in astrocytes processes approaching to PDF positive l-LNvs. Scale bars are represented in respective image panels. See also Video S3–7.

Figure 5.. A wide-awake (wake) mutation suppresses gat^33-1 sleep phenotypes

(A) Sleep profile of male flies with genotype gat^33-1 (green, n=34), control w¹¹¹⁸ (black, n=38), wide-awake mutant wake^d2 (orange, n=38), and double-mutant with wake^d2; gat³³⁻¹ (pink, n=31) in 12/12 h LD cycle. The results show that excessive sleeping in gat^33-1 mutants is suppressed in the double-mutant flies (B) Quantification of total sleep amounts (min) during day and during night. (C) Quantification of mean sleep-episode’s durations lengths (min). (D) Number of sleep episodes indicates fragmented sleep in wake^d2 and double-mutant flies. (E) Higher latency to sleep (min) shows difficulty in initiation of sleep in wake^d2 and double-mutant flies compared to gat^33-1 and w¹¹¹⁸ controls. (F) Mean activity per min while moving during day and night did not show any significant difference between wake^d2 flies, double-mutants, and controls. Error bar represents SEM. Statistics as in figure 3. See also Figure S5 and S7.

Figure 6.. GABA_A receptor RDL depletion in PDF neurons suppresses the gat^33-1 sleep phenotypes.

(A) Sleep profile of male flies with genotype PD2/+;; gat^33-1 (brown, n=21), UAS-rdl^RNAi/+;; gat^33-1 (navy blue, n=25), PD2/UAS-rdl^RNAi;; gat^33-1 (red, n=30), and PD2 /UAS-rdl^RNAi (lavender, n=33) in 12/12 h LD cycle. PD2 is short for pdf-GAL4 combined with UAS-dcr2 (see material and methods). The results show that excessive sleeping in gat^33-1 mutants is suppressed by rdl RNAi knockdown in PDF neurons (B) Quantification of total sleep amount (min) during day and night. (C) Quantification of mean sleep episode’s durations (min). (D) Number of sleep episodes indicate fragmented sleep in rdl RNAi knockdown flies with or without gat^33-1. (E) Latency to sleep (min) in rdl RNAi knockdown flies with and without gat^33-1 (F) Mean activity per min while moving during day and night. Error bar represents SEM. Statistics as in Figure 3. See also Figure S6

Figure 7.. DNLG4 depletion in PDF neurons suppresses the gat^33-1 sleep phenotypes.

(A) Sleep profile of male flies with genotype PD2/+;; gat^33-1 (brown, n=23), UAS-dnlg4^RNAi/+;;; gat^33-1 (teal, n=24), UAS-dnlg4^RNAi; PD2/+;; gat^33-1 (purple, n=23), and UAS-dnlg4^RNAi; PD2/+ (blue, n=24) in 12/12 h LD cycle. The results show that excessive sleeping in gat^33-1 mutants is suppressed during nighttime by dnlg4 RNAi knockdown in PDF neurons (B) Quantification of total sleep amounts (min) during day and night. Results show reduction in sleep during night in dnlg4 knockdown flies with and without gat^33-1. (C) Quantification of mean sleep-episode’s durations (min). (D) Number of sleep episodes. (E) Higher latency to sleep (min) shows difficulty in initiation of sleep in dnlg4 RNAi knockdown flies with and without gat^33-1. (F) Mean activity per min while moving during day and night. Error bar represents SEM. Statistics as in Figure 3. See also Figure S6.