Ancient Anxiety Pathways Influence Drosophila Defense Behaviors

Abstract

Anxiety helps us anticipate and assess potential danger in ambiguous situations [1-3]; however, the anxiety disorders are the most prevalent class of psychiatric illness [4-6]. Emotional states are shared between humans and other animals [7], as observed by behavioral manifestations [8], physiological responses [9], and gene conservation [10]. Anxiety research makes wide use of three rodent behavioral assays-elevated plus maze, open field, and light/dark box-that present a choice between sheltered and exposed regions [11]. Exposure avoidance in anxiety-related defense behaviors was confirmed to be a correlate of rodent anxiety by treatment with known anxiety-altering agents [12-14] and is now used to characterize anxiety systems. Modeling anxiety with a small neurogenetic animal would further aid the elucidation of its neuronal and molecular bases. Drosophila neurogenetics research has elucidated the mechanisms of fundamental behaviors and implicated genes that are often orthologous across species. In an enclosed arena, flies stay close to the walls during spontaneous locomotion [15, 16], a behavior proposed to be related to anxiety [17]. We tested this hypothesis with manipulations of the GABA receptor, serotonin signaling, and stress. The effects of these interventions were strikingly concordant with rodent anxiety, verifying that these behaviors report on an anxiety-like state. Application of this method was able to identify several new fly anxiety genes. The presence of conserved neurogenetic pathways in the insect brain identifies Drosophila as an attractive genetic model for the study of anxiety and anxiety-related disorders, complementing existing rodent systems.

Graphical abstract

Figure 1

Drosophila Wall Following Behavior Is Reduced by Diazepam (A) Flies in a glass-topped arena walk on all interior surfaces. (B) Tracking data from a 10-min experiment reveal that flies mainly walk in the perimeter of the arena. (C) Flies fed with diazepam had decreased WAFO compared with controls (g1 = −0.32, g2 = −8.0, g3 = −0.67, g4 = −0.83, n = 40, 40). Fly WAFO was measured as mean distance from center in millimeters. Dots indicate the mean distance from center for individual flies; horizontal line indicates the mean distance from center (mm). p values determined by Mann-Whitney U. The lower axis represents the effect size in Hedges’ g with 95% CI. Green circles and asterisk (^∗) mark a statistically significant (p < 0.05) decrease in behavior. (D) Standardized mean effect sizes of diazepam effects on rodent anxiety (−0.85 g [95 CI −0.74, −0.96]) and fly WAFO (−0.83 g [95 CI −0.42, −0.91]) have comparable magnitudes. See also Figure S1.

Figure 2

Anxiety-Concordant Effects of Serotonin Gene Lesions on Fly WAFO (A) Genetic lesions of d5-HT1A produced only minor effects in WAFO. Blue dots are untreated flies; orange dots are pre-warmed to 31°C as for GAL80^ts derepression. The lower axes show Hedges’ g; responder alleles are named in the boxes. The driver is nSyb-Gal4, Tub-Gal80^ts. (B) Genetic lesions of d5-HT1B had moderate and statistically significant effects on WAFO: knockdown caused increases (d5-HT1B^KK112342g = 0.51, p = 9 × 10⁻³; d5-HT1B^KK115609g = 0.58, p = 2 × 10⁻³), while overexpression elicited a decrease (g = −0.82, p = 7.4 × 10⁻⁵, n = 53, 54). Red and green circles indicate a statistically significant WAFO change. (C) Knockdowns of mSerT with two RNAi lines produced consistent WAFO increases (SerT^GD3824g = 0.63, p = 8.2 × 10⁻⁴, n = 60, 55; SerT^KK108310g2 = 0.48, p = 0.2 × 10⁻², n = 60, 40), and overexpression decreased WAFO (dSerT^Scer∖UAS·^cPag = −0.53, p = 1.8 × 10⁻³, n = 73, 75). Warm-treated controls for d5-HT1A, d5-HT1B, and mSerT UAS transgenes underwent modest, non-statistically significant changes. (D) A comparison of mouse anxiety gene effect sizes and fly ortholog WAFO effect sizes indicates they are concordant in direction and magnitude, except for d5-HT1A knockdowns. Diamonds indicate averaged meta-analytic values; circles indicate fly WAFO effect; lateral vertices and error lines are 95% CI. See also Figure S2.

Figure 3

Fly Defense Behavior Outcomes Are Concordant with Anxiety Outcomes (A) A strong correlation between rodent anxiety and fly WAFO data for nine comparable manipulations (R²_adj = 0.77 [95 CI 0.58, 0.83]). The horizontal axis shows rodent meta-analytic g values; the vertical axis displays fly WAFO g values. The red line is the least-squares fit; p is for the F statistic of the model. (B) Walking speed changes in the square arena are weakly correlated with rodent anxiety outcomes (R²_adj = 0.22 [95 CI 0.0, 0.30]). (C) WAFO is moderately related to locomotion in the square arena (R²_adj = 0.38 [95 CI 0.06, 0.49]). (D) Light/dark choice outcomes are strongly correlated with rodent effect sizes (R²_adj = 0.81 [95 CI 0.64, 0.86]). (E) Changes in locomotion in the light/dark arena are weakly correlated with rodent anxiety outcomes (R²_adj = 0.06 [95 CI 0.0, 0.09]). (F) Light/dark choice outcomes are poorly correlated with locomotion (R²_adj = 0.11 [95 CI 0.0, 0.14]). See also Figure S3.

Figure 4

Identification of Candidate Fly Anxiety Genes (A) RNAi knockdown with d5-HT2A^KK110704 increased WAFO (g = 0.48, p = 1 × 10⁻²), but this effect was not confirmed by a second RNAi allele (d5-HT2A^JF02157g = −0.07, p = 6.9 × 10⁻¹) or overexpression (d5-HT2A^{Scer∖UAS.cPa}g = −0.21, p = 0.28). Warm-treated controls underwent non-statistical WAFO alterations. (B) Knockdown of d5-HT2B with d5-HT2B^KK111548 produced a decrease in WAFO (g = −1.1, p = 6.8 × 10⁻⁰⁸) as did a Minos transposon insertion into the gene: d-HT2B^MB11858 (g = −0.88, p = 4.1 × 10⁻⁰⁶). (C) Orthologs of candidate mouse anxiety genes were knocked down in the adult fly and tested for WAFO changes. Four knockdowns produced statistically significant reductions in WAFO: tsr^KK108706 (g = −0.89, p = 5.0 × 10⁻⁶); tmod^KK108701 (g = −0.81, p = 1.8 × 10⁻⁵); CCKLR−17D1^KK108482 (g = −0.45, p = 3.5 × 10⁻⁴); and CCKLR−17D3^KK110484 (g = −0.40, p = 1.2 × 10⁻²). Sample sizes are indicated at the base of the bars. (D) Seventeen randomly selected orthologs’ knockdowns had trivial effects on WAFO. See also Figure S4.