Dietary effects on cuticular hydrocarbons and sexual attractiveness in Drosophila

Abstract

Dietary composition is known to have profound effects on many aspects of animal physiology, including lifespan, general health, and reproductive potential. We have previously shown that aging and insulin signaling significantly influence the composition and sexual attractiveness of Drosophila melanogaster female cuticular hydrocarbons (CHCs), some of which are known to be sex pheromones. Because diet is intimately linked to aging and to the activity of nutrient-sensing pathways, we asked how diet affects female CHCs and attractiveness. Here we report consistent and significant effects of diet composition on female CHC profiles across ages, with dietary yeast and sugar driving CHC changes in opposite directions. Surprisingly, however, we found no evidence that these changes affect female attractiveness. Multivariate comparisons among responses of CHC profiles to diet, aging, and insulin signaling suggest that diet may alter the levels of some CHCs in a way that results in profiles that are more attractive while simultaneously altering other CHCs in a way that makes them less attractive. For example, changes in short-chain CHCs induced by a high-yeast diet phenocopy changes caused by aging and by decreased insulin signaling, both of which result in less attractive females. On the other hand, changes in long-chain CHCs in response to the same diet result in levels that are comparable to those observed in attractive young females and females with increased insulin signaling. The effects of a high-sugar diet tend in the opposite direction, as levels of short-chain CHCs resemble those in attractive females with increased insulin signaling and changes in long-chain CHCs are similar to those caused by decreased insulin signaling. Together, these data suggest that diet-dependent changes in female CHCs may be sending conflicting messages to males.

Figure 1. Total amounts of CHCs determined by GC-MS analysis for females D. melanogaster maintained on different diets.

ANOVA on log-transformed data yielded highly significant effects of diet (P<0.0001), age (P = 0.0038), and diet by age interaction (P = 0.0003) on total amount of CHC. Based on Tukey HSD post-hoc tests total amounts of CHC in S5Y5 and S20Y5 females are not different from each other and are both different from S5Y20 and S20Y20 females. This indicates that protein in the diet is the main determinant of the total CHC amount.

Figure 2. Principal component analysis of CHCs detected by GC-MS in D. melanogaster females fed four different diets.

Three PCs were retained that explained 56% total variance, with PC1, PC2, and PC3 explaining 22%, 18%, and 16% of variance after Varimax rotation. PC scores are plotted in 3D space (A), and orthogonal best fit lines for PC scores serve as axes of cones with apexes pointing in the direction of increasing fly age. The numbers designate fly ages: 1 = 7d, 2 = 23d, 3 = 49d, 4 = 65d. Colors indicate food treatments: Grey = S5Y5, Red = S5Y20, Blue = S20Y5, Green = S20Y20. The arrows on PC1-PC2 plane represent the projections of each cone’s axis. PC loadings are shown in the table (B) with shading reflecting the strength of each CHC’s load on each PC (as rendered by JMP). For statistical treatment of individual CHCs, see Fig. S2.

Figure 3. Principal component analysis of CHCs detected by LDI-MS in D. melanogaster females fed four different diets.

Three retained PCs explain 80% of the total variation in CHCs, with PC1, PC2, and PC3 explaining 36%, 24%, and 19% of variance respectively after Varimax rotation. Table (A) shows PC loadings. ANOVA analyses were performed on log-transformed component scores for rotated PCs. Three PC scores are each plotted against female age of 7, 28, 43, and 58 days (B). Colors and symbols indicate food treatments: Grey squares = S5Y5, Red circles = S5Y20, Blue triangles = S20Y5, Green inverted triangles = S20Y20. Abbreviations used: ns = not significant (at α<0.05), inter. = interaction effect. Letters on the right, where different, specify statistically significant difference (at α<0.05) between diets by Tukey HSD tests. For statistical treatment of individual CHCs, see Fig. S3.

Figure 4. Effects of dietary sugar and yeast on D. melanogaster female CHCs.

Normalized intensities of individual CHCs are shown across 4 ages (7, 23, 49, and 65 days) for females fed balanced S5Y5 diet (control) as opposed to high-sugar S20Y5 diet (A) or high-protein S5Y20 diet (B). Stars indicate significant effect based on 2-factorial ANOVA models of diet and/or diet by age interaction after sequential Bonferroni correction.

Figure 5. Comparison of diet, age, and insulin signaling effects on CHCs.

Arrows indicate an increase or decrease in the relative abundance of individual CHCs: 1) for aging - from young to old age on balanced diets , 2) for protein – based on change from S5Y5 to S5Y20, and for sugar – based on change from S5Y5 to S20Y5, 3) for insulin signaling (IIS) - change in transgenic flies with increased insulin signaling (via insulin receptor overexpression - InR^OX) or decreased insulin signaling (insulin substrate mutants chico) compared to controls . Yellow and blue background indicates statistically significant (at α = 0.05) decrease and increase, respectively. Our previous studies determined that increased insulin signaling or young age makes females more attractive to males and that decreased insulin signaling or old age makes them less attractive , .

Figure 6. Effect of diet on female attractiveness.

Females fed the two unbalanced diets (S20Y5 and S5Y20) are not different in attractiveness when tested against S5Y5 females (A,B), or against each other (C). Female attractiveness is expressed as % time a male spends in courtship proximity of a target female in 2-choice assays. P-values are shown above corresponding box plots. While few replicates show a difference from 50% (dotted line), these are not consistent in direction and not significant after Bonferroni correction.