Maternally-transmitted microbiota affects odor emission and preference in Drosophila larva

Abstract

Experimental studies show that early sensory experience often affects subsequent sensory preference, suggesting that the heterogeneity of sensory cues in nature could induce significant inter-individual behavioral variation, potentially contributing to maintain intraspecific diversity. To test this hypothesis, we explored the behavioral effect induced by variation in the levels of a self-produced chemical, acetoin, and its link with intraspecific diversity. Acetoin is a pheromone-like substance produced by gut-associated microorganisms in Drosophila. Using wild-type Drosophila melanogaster populations producing variable acetoin levels, we (i) characterized factors involved in this variation and (ii) manipulated some of these factors to affect acetoin responses in larvae. We found that increased and decreased variations in acetoin levels were caused by microorganisms associated with the outside and inside of the egg, respectively. Wild-type larvae preferred acetoin-rich food only when they both produced and were exposed to substantial amounts of acetoin. The removal of the outside of the egg or the genetic alteration of olfaction abolished this preference. In contrast, larvae exposed to high doses of synthetic acetoin were repulsed by acetoin. The similar effects obtained with freshly caught wild-type lines suggest that this acetoin "production-preference" link underlies the diversity of acetoin-producing microorganisms among natural D. melanogaster populations.

Figure 1

Acetoin content in food patches processed by D. melanogaster larvae. We measured the quantity of acetoin in food patches (ng/patch) impregnated by larvae of different strains or conditions. Similar patches were used in parallel for behavioral tests (Fig. 3). Food either contained no larvae (Ø; a–d), or was processed either by Dijon 2000 larvae (Di2; a–d) or Canton-S larvae (Cs; a,b). In all tests, we analyzed the acetoin levels in vials 3 days and 6 days after egg-laying (+3d, +6d; light grey and dark grey shaded bars, respectively). We tested the effect of control larvae either on plain food (a) or on food mixed with 0.02% acetoin (Food+H₃B₂; b). We also tested the effect of Di2 larvae resulting of control eggs (CE), or of washed eggs (WE), or of dechorionated eggs (DE) either in plain food (c) or in “Food+H₃B₂” (d). We also tested the effect of an extract of crushed larvae (CL) or crushed adults added in plain food (c). Data are shown as box plots representing the 50% median data (second and third quartile separated with a small horizontal bar indicating the median value). For each graph, the quantitative variation of each compound was tested using a Mann-Whitney test (***p < 0.001, *p < 0.05; NS: not significant; N = 10–77).

Figure 2

Variation of acetoin content during 10 generations. We simultaneously measured acetoin levels (ng/patch) in food patches processed by larvae in five Di2-derived lines (A–E; +3d vials) selected from a larger sample (F0; Fig. S2). In each line, acetoin production was measured during 10 successive generations (F1–F10). At each generation, progenitors flies inducing the next generation were kept only two days for egg-laying on the food in order to reduce microorganism transmission (see Material and methods).

Figure 3

Acetoin preference in Di2 and mutant larvae. (a) Each individual early third instar larva was given a binary choice between two food patches either impregnated with plain food (empty circle and bars) or with acetoin-rich food (Food+H₃B₂; filled circle and bars). We tested Di2 larvae (b,d,e) and anosmic Orco ² mutant larvae (c). Each double-sided bar represents the frequency for the first chosen food in the binary choice test. Acetoin was either added at variable quantities: 0.05–10 µg (b,c) or only 2 µg dose (d,e). We also performed the control experiment associating two patches of plain food (Ø; b,c). Larvae were either raised on plain food (b–d) or on acetoin-rich food (e). We tested larvae resulting of control eggs (all tests in b, c; CE in d,e) and of manipulated eggs (washed: WE; dechorionated: DE; dechorionated + raised in isolation DEx1; d,e). For each test, significant differences for binary choice between the two food patches were detected with a Fisher exact test (***p < 0.001; **p < 0.01; *p < 0.05; N = 50 except for b-2 µg, d-DE, e-CE, e-DE experiments: N = 100). The “+” and “−” signs within bars indicate significantly increased and decreased preference for acetoin when all responses corresponding to a complete set of experiment were simultaneously tested (chi-square and post-hoc cell partitioning tests; p < 0.05).

Figure 4

Acetoin production and preference in freshly caught lines larvae. Eight lines sampled in 2016 (Dijon 2016 = Di16) were characterized during their first three generations in the lab (F1–F3). (a) The measure of acetoin content (ng/patch) revealed that 7 lines had very low levels (#1–3, #5–8: LOW) whereas one line had a high acetoin level (#4: HIGH). These differences remained very stable, at least until F3. (b) F3 larvae of the HIGH and LOW lines were tested for their food preference. Larvae resulting of control eggs (CE) diverged for their response differently to larvae resulting of dechorionated eggs (DE). The “+” and “−” signs within bars indicate significantly increased and decreased preference for acetoin, when all responses corresponding to a complete set of experiment were simultaneously tested (chi-square and post-hoc cell partitioning tests; p < 0.05; N = 50). A significant difference for binary choice between the two food patches was also detected with a Fisher exact test (*p < 0.05). (c) Differences for acetoin production. Only HIGH line individual larvae (resulting of control eggs: CE) and adults (crushed and mixed in plain food: CA) produced high acetoin levels in +3d vials. In both cases, the levels in +6d vials strongly decreased. In all other tests (performed with other conditions and vial age) showed no, or very low, acetoin production. Significant differences were detected with a Mann-Whitney test (***p < 0.001; N = 5–10). Note also that all LOW lines were pooled (b,c).