Serotonergic Modulation of Aggression in Drosophila Involves GABAergic and Cholinergic Opposing Pathways

Abstract

Pathological aggression is commonly associated with psychiatric and neurological disorders and can impose a substantial burden and cost on human society. Serotonin (5HT) has long been implicated in the regulation of aggression in a wide variety of animal species. In Drosophila, a small group of serotonergic neurons selectively modulates the escalation of aggression. Here, we identified downstream targets of serotonergic input-two types of neurons with opposing roles in aggression control. The dendritic fields of both neurons converge on a single optic glomerulus LC12, suggesting a key pathway linking visual input to the aggression circuitry. The first type is an inhibitory GABAergic neuron: its activation leads to a decrease in aggression. The second neuron type is excitatory: its silencing reduces and its activation increases aggression. RNA sequencing (RNA-seq) profiling of this neuron type identified that it uses acetylcholine as a neurotransmitter and likely expresses 5HT1A, short neuropeptide F receptor (sNPFR), and the resistant to dieldrin (RDL) category of GABA receptors. Knockdown of RDL receptors in these neurons increases aggression, suggesting the possibility of a direct crosstalk between the inhibitory GABAergic and the excitatory cholinergic neurons. Our data show further that neurons utilizing serotonin, GABA, ACh, and short neuropeptide F interact in the LC12 optic glomerulus. Parallel cholinergic and GABAergic pathways descending from this sensory integration area may be key elements in fine-tuning the regulation of aggression.

Keywords: GABA; LC12 optic glomerulus; RDL; acetylcholine; fruit fly; lunge; serotonin; short neuropeptide F; wing threat.

Figure 1.. Two types of 5HT1A-expressing neurons project to the VLP and modulate aggression.

A. Schematic of serotonergic modulation of aggression by 5HT-PLP neurons in the ventrolateral protocerebrum (VLP) [28]. B. 5HT-PLP neurons (green) arborize in the VLP. Neuropil areas are visualized by anti-Bruchpilot staining (nc82, gray). Genotype: UAS>stop>CD8:GFP/417-FLP; TRH-Gal4. C. The 5HT1A-expressing neurons arborize in the VLP. Genotype: UAS-CD8:GFP; 5HT1A-Gal4. D. Schematic of intersectional approach to isolate individual 5HT1A-expressing neurons. E. An individual VLP-projecting neuron. The cell body (white arrow) is GABA-positive (magenta in the close-up panel E’, scale bars - 10 um). See also Table S1. Genotype: UAS>stop>CD8:GFP/282-FLP; 5HT1A-Gal4. F. A second VLP-projecting neuron has its cell body in the anterior protocerebrum (white arrows). The lower panels in E and F show similar polarity of both neurons: the axonal marker labels both ipsi- and contralateral VLPs (middle panels, genotype: UAS>stop>nsyb:GFP/282-FLP; 5HT1A-Gal4), while the dendritic marker labels only the ipsilateral VLP (bottom panels, genotype: UAS>stop>DsCam:GFP/282-FLP; 5HT1A-Gal4). In B,C,E and F - The full z-projections are shown. Scale bars represent 50 um. G. 5HT1A-expressing neuron subsets targeted and manipulated via dTrpA1:Myc. Induced activation of the subset #1 (VLP-GABA+) results in lower numbers of lunges. For anatomical description of other targeted subsets see Figure S1. H. Induced activation of the VLP-GABA+ neurons did not change the fight outcomes against standard opponents (Chi-Square=2.485; N=75, DF=2; p=0.289). Only the fights with winning outcomes were used for the aggression analyses. I. Flies with activated VLP-GABA+ neurons (green) showed significantly lower number of lunges (Chi-Square=6.5076; DF=1; p=0.0107) performed in 30 min. The results are presented as boxplots with a median line. **p ≥ 0.01 versus controls (gray), analyzed by a nonparametric Wilcoxon/Kruskal-Wallis Test. See also Figure S2. Genotype in G-I: UAS>stop>dTRPA1:Myc/282-FLP; 5HT1A-Gal4.

Figure 2.. The anterior VLP 5HT1A-expressing neurons are targeted by sNPFR-Gal4.

A-B. The sNPFR-Gal4 (VT033628) driver targets anterior (A) and posterior (B) groups of neurons. White arrows point to the cell bodies in the anterior protocerebrum. Genotype: UAS-CD8: GFP; sNPFR-Gal4. See also Figure S3. A’-B’. Arborization sites of the anterior (A’), but not the posterior (B’) sNPFR-Gal4-targeted neurons (green) are in close apposition with sNPF antibody immunopositive area (magenta). See also Table S1. C-C’. The aggression-modulating 5HT-PLP neuron (green) arborizes through the LC17 and LC12 optic glomeruli visualized with nc82 (gray). Anti-5HT counterstaining is shown in magenta. Images represent Z-projections through the entire VLP. Partial anterior (for LC17) and posterior (for LC12) Z-stacks are shown in C’. Genotype: UAS>stop>CD8:GFP/417-FLP; TRH-Gal4. D. A reconstituted GFP signal (green) is observed between 5HT neurons (magenta) and sNPFR-expressing neurons in the area of the VLP including the LC12 glomerulus. Genotype: LexAop-spGFP¹¹; UAS-spGFP¹⁻¹⁰,TRH-LexA/sNPFR-GAL4. A-D. Scale bars represent 50 um. E. A 5HT1A::HA signal (magenta) co-localizes with the sNPFR-driven GFP signal (green) in the cell body of the anterior VLP neuron (white arrow). Scale bars represent 10 um. See also Figure S3. Genotype: 5HT1A::HA/UAS-CD8::GFP, sNPFR-Gal4.

Figure 3.. Functional characterization of the anterior VLP/sNPFR neurons.

A. The anterior VLP/sNPFR neurons show a Ca2⁺-activating responses to bath-applied serotonin. Dotted circles in the close-up panels show the cell body ROI used for the quantification (A’). Genotype: UAS-GCamp6(s); sNPFR-Gal4. See also Figure S4. B-D. Activation of the anterior VLP/sNPFR neurons increases aggression. A Kruskal-Wallis 1-way test revealed significant effects of genotype on the numbers of aggressive encounters (B, Chi-Square=7.022; DF=2; p=0.0299), lunges (C, Chi-Square=5.443; DF=2; p=0.0658) and wing threats (D, Chi-Square=31.974; DF=2; p<0.0001) performed in 10 min. Pair-wise comparisons are shown on the graphs. *p > 0.05; ***p>0.001 versus controls (open circles). Data are presented as boxplots with a median line. Genotype: sNPFR-Gal4/UAS-dTrpA1. E-G. Inhibition of the anterior VLP/sNPFR neurons reduces aggression. A Wilcoxon/Kruskal-Wallis rank sum test revealed significant differences in the numbers of aggressive encounters (E, Chi-Square=5.2730; DF=1; p=0.0217) and lunges (F, Chi-Square=6.1342; DF=1; p=0.0133) performed in 10 min. Data are presented as boxplots with a median line. *p > 0.05; **p>0.01 versus controls (open circles). Genotype: UAS-Kir 2.1/ tub-Gal80^ts; sNPFR-Gal4.

Figure 4.. Anterior VLP/sNPFR neurons are cholinergic.

A. Heatmap of the anterior VLP/sNPFR pooled- and single-neuron expression profiles in comparison to pools of 5HT-enriched neurons. B. Anterior VLP/sNPFR neurons (green) are immunopositive for the cholinergic neurotransmission marker ChAT (magenta) in the cell body (B’, white arrow) and in the LC12 glomerulus (B”). See also Table S1. The scale bar represents 50 um. The neighboring LC17 glomerulus is also ChAT-positive (magenta). Fine anterior VLP arbors outside of LC12 (white arrowheads) appear to invade nearby optic glomeruli. The scale bars in B’ and B” represent 10 um. Genotype: UAS-CD8:GFP; sNPFR-Gal4.

Figure 5.. RNAi-mediated knockdown of RDL receptors increases aggression.

A-C. RDL (black), but not GABA-B-R2 (gray), receptor silencing in the anterior VLP/sNPFR neurons increases aggression. Kruskal-Wallis 1-way test revealed significant effects of the genotype on the numbers of lunges (B, Chi-Square=8.439; DF=3; p=0.0378), and wing threats (C, Chi-Square=13.371; DF=3; p<0.0039) performed in 10 min. Pair-wise comparisons are shown on the graphs. *p > 0.05; **p>0.01 versus controls (open circles). Data are presented as boxplots with a median line. Genotype: tub-Gal80^ts; UAS-RNAi (or genetic control)/sNPFR-Gal4. D-F. Silencing (black) or overexpressing (gray) of 5HT1A receptors in the anterior VLP/sNPFR neurons has no effects on aggression. Kruskal-Wallis 1-way test showed no significant effect of genotype on the aggressive behavioral elements scored in 10 min. Individual pair comparisons are shown on the graphs, where *p > 0.05 versus controls (open circles). Data are presented as boxplots with a median line. Genotype: tub-Gal80^ts;UAS-RNAi (or genetic control)/sNPFR-Gal4. See also Figure S4.

Figure 6.. A model of how visual input might integrate into an aggression-modulating circuitry.

An aggression-promoting cholinergic neuron (red) and an aggression-inhibiting GABAergic neuron (blue) collect synaptic input from the LC12 optic glomerulus formed by lobula visual projection neurons (gray). Both neuron types putatively express G_i/0-coupled 5HT1A receptors, which should, when activated, inhibit the activity of both neurons. Yet, the cholinergic neurons are activated by serotonin (Figure 3A). When the serotonergic 5HT-PLP neurons (green) fire, they activate 5HT1A receptors on the GABAergic neurons, leading to a reduction in GABA release in the LC12 glomerulus. This relieves an ongoing potent RDL-mediated GABAergic inhibition of the cholinergic neurons, allowing for their spontaneous or vision-evoked firing, which via further downstream circuitry enhances aggression display through increased number of lunges and wing threats.