Previous studies indicate that in Drosophila, body size is more nutritionally plastic in females than in males, a phenomenon referred to sex-specific plasticity (SSP) [19]. Nutritional plasticity is canonically regulated by the insulin/IGF-signaling and TOR signaling pathways in Drosophila and most other animals. We hypothesized, therefore, that differences in nutritional plasticity between the sexes reflects sex-specific differences in the activation and activity of these pathways. In this study, we applied a nutritional geometry framework to test this hypothesis, and explored the sex-specific effects of protein and carbohydrates on body size and the expression of genes that activate and are transcriptionally regulated by IIS/TOR. Our results indicate that, broadly, gene expression is more nutritionally sensitive in females than in males, although the effect of nutrition on the expression of IIS and TOR-signaling genes is both quantitively and qualitatively different between males and females.
The simplest hypothesis to account for sex-specific differences in the nutritional plasticity of body size in Drosophila is that females have the same response to nutrition as males, but at a higher sensitivity. Under this hypothesis, the response surfaces of body size and gene expression across a nutritional landscape of varying protein and carbohydrate concentrations should be the same shape in both sexes, but with lower gradients in males. This would be evident as lower parameter values in males versus females. This is not supported by the data, however. In particular, body size in males shows more or less the same sensitivity to changes in protein concentration as females, but has no detectable sensitivity to changes in carbohydrate concentration. This in turn suggests that an important aspect of sex-specific nutritional plasticity in Drosophila is differences in the response to variation in dietary carbohydrates.
The best understood regulator of body size with respect to nutrition are the insulin/IGF-signaling and TOR-signaling pathways [29, 30]. The IIS pathway is activated by circulating dILPs, some of which are released in a nutrition-dependent manner [27]. The TOR-signaling pathway is in part regulated by IIS, but also responds directly to circulating amino acids [31–33]. Because the SSP of body size in flies appears to primarily driven by a differential response to carbohydrates but not to protein, we might expect that the effect of diet on body size is mediated primarily by IIS rather than TOR-signaling. Our data do not support this hypothesis. We used the expression of InR and 4E-BP as a measure of IIS activity: Both are transcriptionally regulated by the forkhead transcription factor FOXO, which is activated when nutrition and IIS activity is low [26, 34]. If sex-specific plasticity is mediated by IIS then we would expect to see sex-specific differences in Inr and 4EBP expression in response to changes in carbohydrate but not to protein. In both males and females, however, InR expression responded only to protein level – and to the same extant – but did not respond to carbohydrate level. Expression of 4E-BP only responded to protein level in females, but did not respond to carbohydrate level in either sex. Collectively, variation in the expression of 4E-BP and InR were the strongest correlates with body size in females and males respectively (Table 3), consistent with the hypothesis that the IIS is the major regulator of body size with respect to nutrition in Drosophila. However, the expression of these genes do not reflect sex-specific differences in the effect of carbohydrate on body size, and so do not explain sex-specific plasticity.
In contrast to InR and 4E-BP, the expression of Ash2L and CG3071 both responded to carbohydrates in a sex-specific manner. However, the response is not consistent with a previous report on the regulation of the these genes’ expression by TOR-signaling [28]. This study reported that Ash2L is negatively regulated by TOR-signaling while CG3071 is positively regulated. If TOR-signaling increases with protein, then we would expect Ash2L expression to decrease correspondingly, which is true in females. However, Ash2L expression also decreases with increasing carbohydrate, which is difficult to reconcile with the negative effects that carbohydrate has on female growth. Even more challenging to interpret is the observation that Ash2L expression increases with carbohydrate but not protein in males, even though male do not appear to have a growth response to carbohydrate. Thus the relationship between Ash2L expression and body size is not a simple one. The expression pattern of CG3071 is similarly unclear. While CG3071 also has a carbohydrate and protein response in females, this response is qualitatively similar to the response of Ash2L, which is not expected if TOR-signaling positively regulates CG3071 expression but negatively regulates Ash2L expression. Thus the role that TOR signaling plays in regulating SSD and SSP is equivocal based on our data. Problematically, the mechanism by which TOR-signaling regulates Ash2L and CG3071 expression has not been fully explored, which makes interpreting changes in Ash2L and CG3071 expression with diet even more challenging.
Our data suggests a complex relationship between sex, diet, dILP expression and dILP retention levels in the IPCs. In females, there were stark differences in the pattern of expression among the dILPS. Of all the dILPs, dILP5 expression was most strongly correlated with body size, increasing with increasing protein level but decreasing with increasing carbohydrate level. We also found that dILP5 levels in the IPCs were lower in larvae fed on lower food concentrations. Thus, chronic reductions in nutrition reduce dILP5 expression and dILP5 IPCs level. This is consistent with dILP5 being a positive regulator of growth in response to nutrition [35]. The negative effect of low nutrition on dILP5 expression has been observed in previous studies [30, 35, 36]. However, at least one of these studies reported an increase in dILP5 retention in the IPCs of larvae that were either starved for 24 hours or reared on a low protein diet since birth [36], which is inconsistent with our data. The low protein diet in this study was generated by reducing the amount of protein from ~16g/l to ~0.8g/l, while keeping the concentration of carbohydrate at ~60g/l, thereby decreasing the protein-to-carbohydrate ratio. In our study we reduced overall food concentration from 360g/l to 45g/l while retaining a 1:2 protein-to-carbohydrate ratio. It is possible, therefore, that the retention of dILP5 in the IPCs in the earlier study is due to a low protein-to-carbohydrate ratio. Indeed, a second earlier study showed that a chronic high sugar diet with low protein-to-carbohydrate ratio also increased dILP5 levels in the IPCs [37], although this time accompanied by an increase in dILP5 expression. Thus the relationship between dILP5 expression and dILP5 levels in the IPCs may depend not only on the overall food concentration, but also on food composition, in particular the protein-to-carbohydrate level.
Previous studies suggest that dILP2 expression is unaffected by reduced nutrition [30, 36] (although see supplementary data in [38]), but that dILP2 peptides are retained in the IPCs in larvae that have been starved 24h or are reared on a low protein diet [36, 39, 40]. A high sugar diet is also associated with increased retention of dILP2 peptide in the IPCs, but with an increase in dILP2 expression [37]. In our study, however, we found that dILP2 expression declined with an increase in both protein and carbohydrate, and that dILP2 peptide was lower in the IPCs of larvae fed on lower food concentrations. Our observed expression of dILP3 was also inconsistent with previous studies that found that acute starvation reduced expression [30], while rearing on a high protein diet increased expression [41]. In contrast, we found that, as for dILP2, dILP3 expression declined with increasing protein, although was unaffected by carbohydrate level, which is surprising given that circulating sugars promote the release of dILP3 from the IPCs [40]. Our finding that dILP2 and dILP3 expression is lowest on low protein diets that restrict growth, and that dILP2 peptides levels are higher when dILP2 expression is lower, suggests that dILP2 and dILP3 may negatively regulate their own expression, which is a common characteristic of hormone regulation.
Finally, the expression of dILP8 showed a similar pattern to dILP3, increasing with a decrease in protein concentration. dILP8 is involved in regulating growth and developmental timing through its inhibitory effects on ecdysone synthesis [42, 43]. It is released from imaginal discs in response to damage or growth perturbation, as well as showing periodic changes in expression throughout development . Ecdysone synthesis is also inhibited by low nutrition early in the third larval instar, leading to a delay in metamorphosis [44]. The observation that females on low protein diets also show elevated levels of dILP8 expression, suggests that dILP8 may play a role in regulating growth and/or development in response to nutrition, at least in females.
We did not detect an effect of diet on the expression of any of the dILPs in male larvae. This did not, however, translate into a significant interaction between diet and sex on dILP expression, and therefore does not help explain the sex-specific differences in nutritional plasticity of body size. This was because variation in dILP 2,3,5 and 8 expression among samples within diets was significantly higher for male samples than for females samples (Supplementary Table 1), reducing the statistical power to detect diet-by-sex interactions. The only exception was for dILP8, such that dILP8 expression was significantly more plastic in females than in males. Nevertheless, dILP8 expression in females did not correlate with carbohydrate level, and so is unlikely to explain the differential effect of carbohydrate on body size in females versus males. The elevated variation in dILP expression among male samples was not seen for expression of 4eBP, InR, Ash2L, or CG3072, and is therefore not likely due to sample degradation. It is also not due to lower levels of expression in males relative to females. It is possible, therefore, that dILP expression levels are more developmentally dynamic in males then females, and that we are capturing aspects of that instability by measuring gene expression at only a single point in development.
Collectively, there appears to be no simple explanatory relationship between the sex-specific nutritional geometry of body size with the sex-specific nutritional geometry of IIS/TOR-signaling gene expression. Further, many of our findings do not replicate what has been reported in previous studies. An earlier study looking at dILP expression in adults across a nutritional landscape saw similarly complex and non-intuitive relationships between diet and gene expression [23]. What is very clear from both studies is that the protein-to-carbohydrate ratio affects the transcriptional response of IIS/TOR-signaling genes to changes in total diet, and our study indicates that it does so in a sex-specific way. This has important implications for interpreting those studies that have looked at changes in the expression of IIS/TOR-signaling genes in response to changes in nutrition. First, many studies simply dilute diet to reduce overall nutrition. Problematically, while most Drosophila labs use ‘standard’ food recipes, the use of different types of yeast, cornmeal, mollasses etc means that the composition of these diets may be unique to most research groups [45]. Our data indicate that the effect of diet dilution on both body size and IIS/TOR-signaling gene expression – and potentially the activity of the IIS/TOR-signaling pathways – will depend on the P:C ratio of that specific diet. Second, these responses are sex-specific, and studies that do not consider sex may come to different conclusions than studies that do. Third, the relationship between dILP expression, dILP retention and diet is similarly complex, and may also depend on the composition of the diet being manipulated.
One important caveat with our, and almost all other studies of IIS/TOR-signaling gene expression during development, is that we measured expression at a single developmental time point, at the very beginning of larval wandering. Previous studies have shown that the activity of the IIS and TOR-signaling pathways change dynamically during development, and that different dILPs are expressed at different life stages. This may also account for differences between our and other studies of the effects of nutrition on IIS/TOR-signaling gene expression. Ideally, one would like to conduct a multidimensional study of gene expression across both time and a nutritional landscape, and to tie this with a corresponding study of dILP levels both in the IPCs and circulating in the hemolymph. Such a study would not only help elucidate the relationship between IIS/TOR-signaling and the sex-specific effects of nutrition on body size, but also help us understand better how nutrition regulates the transcription, translation, storage and release of dILPs.