1. Genetics and Genomics
Download icon

Drosophila serotonin 2A receptor signaling coordinates central metabolic processes to modulate aging in response to nutrient choice

  1. Yang Lyu
  2. Kristina J Weaver
  3. Humza A Shaukat
  4. Marta L Plumoff
  5. Maria Tjilos
  6. Daniel EL Promislow
  7. Scott D Pletcher  Is a corresponding author
  1. Department of Molecular and Integrative Physiology and Geriatrics Center, Biomedical Sciences and Research Building, University of Michigan, United States
  2. College of Literature, Science, and the Arts, University of Michigan, United States
  3. Department of Lab Medicine & Pathology, University of Washington School of Medicine, United States
  4. Department of Biology, University of Washington, United States
Research Article
  • Cited 2
  • Views 2,308
  • Annotations
Cite this article as: eLife 2021;10:e59399 doi: 10.7554/eLife.59399

Abstract

It has been recognized for nearly a century that diet modulates aging. Despite early experiments suggesting that reduced caloric intake augmented lifespan, accumulating evidence indicates that other characteristics of the diet may be equally or more influential in modulating aging. We demonstrate that behavior, metabolism, and lifespan in Drosophila are affected by whether flies are provided a choice of different nutrients or a single, complete medium, largely independent of the amount of nutrients that are consumed. Meal choice elicits a rapid metabolic reprogramming that indicates a potentiation of TCA cycle and amino acid metabolism, which requires serotonin 2A receptor. Knockdown of glutamate dehydrogenase, a key TCA pathway component, abrogates the effect of dietary choice on lifespan. Our results reveal a mechanism of aging that applies in natural conditions, including our own, in which organisms continuously perceive and evaluate nutrient availability to promote fitness and well-being.

eLife digest

The foods we eat can affect our lifespan, but it is also possible that thinking about food may have effects on our health. Choosing what to eat is one of the main ways we think about food, and most animals, including the fruit fly Drosophila melanogaster, choose their foods. The effects of these choices can affect health via a chemical in the brain called serotonin. This chemical interacts with proteins called serotonin 2A receptors in the brain, which then likely primes the body to process nutrients.

To understand how serotonin affected the lifespan and health of fruit flies, Lyu et al. compared flies that were offered a single food to those that could choose between several foods. The flies that had a choice of foods lived shorter lives and produced more serotonin, but these effects were reversed when Lyu et al. limited the amount of a protein called glutamate dehydrogenase, which helps cells process nutrients.

These results suggest that choosing what we eat can impact lifespan, ageing and health. Human and fly brains share many similarities, but human brain chemistry is more complex, as is our experience of food. This work demonstrates that food choices can affect lifespan. More research into this phenomenon may shed further light onto how our thoughts and decision-making impact our health.

Introduction

For nearly all metazoans, many aspects of biology are continuously influenced by information about nutrient availability. Interest in the nutritional determinants of aging date back to the early 1900s, as experiments in fruit flies (Chippindale et al., 1993; Chapman and Partridge, 1996) and rodents (McCay et al., 1935; Weindruch et al., 1986; Weindruch and Walford, 1982) revealed that limiting total food intake or reducing diet quality, without malnutrition, increased lifespan. Dietary restriction (DR, also called caloric restriction) is now known to effectively increase the lifespan of nearly every species to which it has been applied. In mammals, increased lifespan in response to DR is accompanied by a broad-spectrum improvement in health during aging (Weindruch and Walford, 1988Longo and Finch, 2003; Barger et al., 2008; Contestabile, 2009).

Research over the past century on the effects of DR has revealed how food characteristics influence healthy aging (Fontana and Partridge, 2015; Tatar et al., 2014). For many years, it was believed that the caloric content of the food and the amount of energy consumed by the animal were the driving forces behind the effects of restriction on lifespan (Masoro, 2005). More recently, however, conceptual advances such as the geometric framework for nutrition (Lee et al., 2008; Solon-Biet et al., 2014; Skorupa et al., 2008), together with comprehensive experiments that manipulate diet in model systems (Skorupa et al., 2008; Bruce et al., 2013; Solon-Biet et al., 2014; Jensen et al., 2015), have made it clear that dietary composition and availability of specific nutrients are often more relevant than calories in shaping life histories. For example, in homogenous nutritional environments where all nutrients are mixed together in a single food, animals fed a high carbohydrate–low protein diet generally live longer than siblings fed a calorically equivalent low carbohydrate–high protein diet (Lee et al., 2008; Bruce et al., 2013). Short-term experiments in heterogeneous nutritional environments, where animals are given a choice of two foods with different macronutrient ratios, reveal a behavioral preference for a diet that is high in carbohydrate and low in protein, suggesting that animals will adjust their dietary choices to optimize evolutionary fitness (Lee et al., 2008; Jensen et al., 2015; Bunning et al., 2016). The long-term effects of a complex nutritional environment and the accompanying dietary choices on physiology and healthy aging are unknown. Macronutrient intake will almost surely not be the only factor here because sensory perception of specific nutrients (Ostojic et al., 2014; Waterson et al., 2014), and even the presence of choice itself (Avondo et al., 2013; Steenfeldt et al., 2019), may be influential.

We have recently reported that how a meal is presented, or perhaps the way in which it is eaten, modulates aging in Drosophila (Ro et al., 2016). We found that flies were longer-lived when aged in an environment in which all dietary components were mixed together (termed a fixed diet) compared to one in which the two macronutrients, yeast and sugar, were available separately (termed a choice diet). When presented with a choice diet, flies consume an equivalent amount of protein and a higher carbohydrate:protein ratio than they do when fed a nutritionally equivalent fixed diet. Contrary to the results of most geometric nutritional studies (Bruce et al., 2013; Lee et al., 2008), however, their lifespan is shorter than those fed on high carbohydrate:protein fixed diets. This led us to speculate that the effects of dietary choice were derived, at least in part, from unknown effects of the complex nutritional environment.

Here we present a series of experimental results that indicate that dietary choice elicits significant changes in the physiology and lifespan of male Drosophila melanogaster through mechanisms that are distinct, at least in part, from nutrient consumption. Coincident changes in behavioral and health metrics suggest that flies exhibit distinct neural and metabolic states when presented with a choice diet than they do when presented with one that is nutritionally homogeneous. We report that neuronal expression of serotonin receptor 2A (5-HT2A) during the adult stage is required for many of the effects caused by dietary choice, including a 5-HT2A-dependent metabolic reprogramming that is consistent with a significant upregulation of the tricarboxylic acid (TCA) cycle and amino acid metabolism when nutrients were presented separately. Transgenic knockdown of a key enzyme, glutamate dehydrogenase (GDH), which is predicted to diminish the metabolic changes induced by a choice diet, abrogated its effects on lifespan. Future research may elucidate the downstream targets through which neuronal 5-HT2A signals acts as a potent modulator of aging and physiology in an environment much like our own, where animals must constantly choose between different nutrients.

Results

As with most model organisms, D. melanogaster reared in a conventional laboratory setting is normally provided with a homogenous, complete media that meets all of its nutritional needs. For flies, this medium is comprised predominantly of a source of protein and micronutrients (e.g., brewer’s yeast) and a carbohydrate source (e.g., sucrose). In natural conditions, however, a complete food source is unlikely to be found, and animals must seek out and identify nutritionally complementary foods (Csata et al., 2020; Mayntz et al., 2005). Indeed, fruit flies are known to recognize and choose among different macronutrients based on palatability (Weiss et al., 2011; Harris et al., 2015) and physiological demand (Ribeiro and Dickson, 2010; Vargas et al., 2010; Ro et al., 2016; Steck et al., 2018). While studying the mechanisms underlying protein feeding, we observed significant modulation of survival depending on whether the animals were given a complete diet or were provided with a dietary choice by presenting the two macronutrient sources separately (Figure 1A, also see Ro et al., 2016). These effects were observed in both sexes (Figure 1A) and in different laboratory strains (Supplementary file 1). The magnitude of the choice effect was consistently greater in males (29–37% change in mean lifespan) than in females (5–15% change in mean lifespan, see Supplementary file 1). Similar effects were observed using two D. melanogaster lines that were recently collected from the wild: the lifespans of male strains were consistently and significantly reduced in response to dietary choice and there was variability in the female response (Figure 1B). The effects on lifespan are not due to some sort of acute toxicity because flies maintained on a choice diet for a short period (e.g., 2 weeks) exhibited general health metrics, such as intestinal integrity and climbing ability (Figure 1C,D), that were indistinguishable from siblings that were fed a fixed diet. Together these results indicate that the effects of dietary choice on lifespan are sex specific and are not an artifact of long-term laboratory culture.

Four lines of evidence indicate that the effect of a choice diet on lifespan is driven, at least in part, by mechanisms that are independent of increased sucrose consumption. First, increased sugar consumption alone is generally not associated with a large reduction in fly lifespan; even extremely sugar-rich diets (e.g., greater than 8× sugar intake) have only modest effects (Dobson et al., 2017), much smaller than we observed following dietary choice. New data that we obtained when we measured lifespan on fixed diets designed to approximate the nutritional composition that flies voluntarily constructed from their choice diet were consistent with this interpretation. We compared the lifespans of flies aged on a choice diet with those obtained from flies aged on one of two high-sugar, fixed diets (S30Y3 and S24Y3; i.e., 30% [w/v] sugar or 24% [w/v] sucrose mixed with 3% [w/v] yeast, respectively), among which flies consumed statistically indistinguishable amounts of both sugar and yeast (Figure 1F, represented as mass of nutrient consumed). The effect of dietary choice, relative to the standard SY10 diet, was much greater than that from homogeneous high-sugar diets (Figure 1G). Second, the effects of the choice diet on gross measures of fly physiology were distinctly different from those caused by increased sugar intake in the fixed diet. Flies exposed to a choice diet exhibited a significant reduction in triglyceride/protein ratio (Figure 1H) and triglyceride abundance (Figure 1—figure supplement 1; total weight was not affected; ), while increased sugar in the fixed diets had no such effect (Figure 1H, Figure 1—figure supplement 1). Third, we observed that simply changing the nature of the choice, by manipulating the concentration of the yeast component of the choice diet, was sufficient to modulate lifespan even though both cohorts of flies consumed the same amount of sugar and protein (Figure 1I,J, represented as mass of nutrient consumed). Moreover, the effect was opposite to that normally seen when similar manipulations were applied to a homogeneous diet (Mair et al., 2005); an increased concentration of the yeast component of choice extended lifespan (Figure 1I). Fourth, the transcription factor dFoxo is required both for transcriptional changes and for nutritional programming of lifespan caused by excess dietary sugar (Dobson et al., 2017), but it is dispensable for the lifespan effect of dietary choice (Figure 1—figure supplement 2). Together, these observations suggest that animals respond differently to macronutrients depending on how they are presented and that the effects of dietary choice on lifespan and physiology may be mediated by processes that are distinct from macronutrient intake.

Figure 1 with 3 supplements see all
The presentation of nutrients significantly affects lifespan independent of the consumption of each nutrient in Drosophila.

(A) Separating macronutrients in diet is sufficient to modulate aging. The left panel illustrates the two-well food containers for presenting either a fixed diet (FD, 10% [w/v] sugar and 10% [w/v] yeast mixed together) or a choice diet (CD, 10% [w/v] sugar and 10% [w/v] yeast presented on different sides). The right panel shows that both Canton-S males and females fed on the choice diet have shorter lifespans relative to their siblings that fed on the fixed diet (N = 104–112 per group). (B) Dietary effects are presented in two Drosophila melanogaster strains that were recently collected from the wild. Males from both strains exhibit a choice diet effect, whereas the responses of females are variable: one strain shows no lifespan difference between the two diets and the other one lives longer on a choice diet (N = 144–152 for each group). (C–D) Males fed on a choice diet exhibit the same intestinal integrity (C, N = 246–291 per diet) and climbing ability (D, N = 14 per diet) compared to fixed diet siblings. (E) Males on a choice diet consume more sugar and the same amount of yeast compared to their siblings that are fed on a fixed diet (N = 68–89 per diet). (F) Males on a choice diet consume a similar amount of sugar compared to their siblings on high-sugar fixed diets (S30Y3 and S24Y3; i.e., 30% [w/v] sugar or 24% [w/v] sucrose mixed with 3% [w/v] yeast, respectively; N = 65–82 per diet). (G–H) Dietary effects on lifespan (G, N = 144–149 for each group) and triglycerides/protein ratio (H, N = 60 for each group) differ between a choice diet and two high-sugar fixed diets. (I–J) Increased dietary protein in a choice diet is sufficient to promote longevity (I, N = 102–154 for each group) without affecting the macronutrient intake (J, N = 52–71). p-Values were obtained for lifespan using a log-rank test and for smurf assay, climbing ability, nutrient consumption, and TAG amounts using a Mann–Whitney U test. Boxplots in (C–F, H and J) represent the medians (the middle bar) and the first and the third quartiles (box boundaries) for each experimental group. Male flies were used in the experiments unless otherwise mentioned.

Figure 1—source data 1

Effects of dietary choice on lifespan and behavior.

https://cdn.elifesciences.org/articles/59399/elife-59399-fig1-data1-v2.xlsx

Although protein intake was not different in flies fed different diets, it seemed possible that dietary choice improved protein absorption, which might be important considering the influential role of this macronutrient in reducing lifespan (Fontana and Partridge, 2015). Evidence suggests that this is not the case. Individual weight was not different among flies fed different diets (Figure 1—figure supplement 1), and total protein was lower in flies given a dietary choice than it was in their siblings fed a fixed diet, which is the opposite of what would be expected from the effect of dietary choice on lifespan. Furthermore, molecular and physiological markers that were strongly affected by protein availability, including target of rapamycin (TOR) activity (Fontana and Partridge, 2015) and female fecundity (Jensen et al., 2015), were not altered by dietary choice (Figure 1—figure supplement 3).

Previously, we hypothesized that neuronal mechanisms involved in nutrient evaluation might influence lifespan when macronutrients are presented separately (Ro et al., 2016). Our candidate screen of neurotransmitters and their receptors revealed that the presence of a P-element insertion in the promoter region of serotonin receptor 2A (5-HT2APL00052), which creates a putative loss of function mutation in this gene (Nichols, 2007), reduced protein preference in starved female flies and significantly extended lifespan under dietary choice (Ro et al., 2016). Loss of function in other serotonin receptors had no influence on protein preference (Ro et al., 2016), yet it remains to be tested whether they influence the aging processes in a choice environment. Here we focused on elucidating the mechanisms underlying the effects of the 5-HT2A receptor and expanded our analysis of its effects on dietary choice and male lifespan.

Indeed, we found that the P-element insertion in 5-HT2A fully abrogated the male lifespan differences between fixed and choice dietary environments, largely by increasing lifespan when nutrients were presented individually (see Figure 2A). Mutant flies were modestly short-lived on the standard, fixed diet, which is likely due to the pleotropic effects of this receptor (Gasque et al., 2013; Chakraborty et al., 2019; Howard et al., 2019). We also investigated the influence of RNAi-based knockdown of 5-HT2A expression using multiple RNAi lines. Because 5-HT2A is highly expressed in the central nervous system (Supplementary file 2), we used the pan-neuronal driver elav-GAL4 to achieve 5-HT2A knockdown, and we observed that this manipulation abrogated the lifespan effect of dietary choice using one of the RNAi lines (UAS-5-HT2A-RNAiTRiP) and strongly reduced the effect in the second (UAS-5-HT2A-RNAiKK; Figure 2B). We also observed that the survivorship of adult flies fed with 200 µM pirenperone, an anxiolytic known to antagonize 5-HT2A receptor signaling in mammals, was unaffected by nutrient presentation, further supporting a role for this receptor and ruling out a necessity for loss of function during development (Figure 2C, Figure 2—figure supplement 1 for the second replicate). Together, these results implicate neuronal expression of receptor 5-HT2A in the adult male fly in mediating the effect of dietary choice on lifespan.

Although we concluded that nutrient intake was unlikely to be responsible for the difference in lifespan in wild-type flies, it remained formally possible that feeding differences between control and 5-HT2A mutant flies were responsible for the dissimilar response to dietary choice. Our previous work demonstrated that 5-HT2A influenced protein preference after starvation (Ro et al., 2016), and we asked whether it influenced nutrient consumption in fed flies. Surprisingly, we found that, in the choice environment, 5-HT2A mutant flies exhibited feeding characteristics similar to control animals: each consumed significantly more sucrose than yeast when given a dietary choice (Figure 2D), despite spending the majority of their time physically located on the yeast food (Figure 2E). Unlike control animals, however, the behavioral differences in 5-HT2A mutants were not coupled to changes in lifespan.

DR, and the extended lifespan that accompanies it, are associated with increased locomotion behavior in several species (Weindruch et al., 1986; Weed et al., 1997; Chen, 2005), presumably reflecting a motivational state of hunger. By video tracking control and mutant flies in each of the dietary environments (using our DTrack system, see Materials and methods), we found that control flies were generally more active when given a dietary choice: they traveled significantly farther in a 6 hr window when compared to siblings housed on a fixed diet. 5-HT2A mutant flies responded similarly (Figure 2F, two-way ANOVA on the square root transformed values, pDiet*Genotype>0.05). Behavioral classification of the movements during this time (Corrales-Carvajal et al., 2016) revealed that diet-dependent differences were mirrored in both the time spent walking (speed > 2 mm/s), in executing micromovements (e.g., grooming or when 0.2 mm/s < movement speed ≤ 2 mm/s), and in the time spent sleeping or resting (speed ≤ 0.2 mm/s) (Figure 2G). As with total distance moved, 5-HT2A mutants exhibited similar responses as control animals in all of these measures (Figure 2G, two-way ANOVA on the square root transformed values, pDiet * Genotype>0.05). From these data we concluded that diet-dependent changes in locomotion are not causal to changes in lifespan.

Increased locomotor behavior has been linked with stress in mice (Sestakova et al., 2013) and perhaps in flies (Mohammad et al., 2016), leading us to examine whether other stress-related phenotypes were changed in flies fed a choice diet (Strekalova et al., 2005; Linford et al., 2015; Yang et al., 2015). First, we measured alcohol consumption because ethanol intake is known to be potentiated by conditions of stress or reward depletion (Devineni and Heberlein, 2009; Shohat-Ophir et al., 2012). We found that when flies were provided a dietary choice for 10 days prior to being exposed to sucrose food containing 15% (v/v) ethanol, they interacted with the ethanol food significantly more than did flies previously fed a homogenous fixed diet (Figure 2H, see Materials and methods for more details). 5-HT2A mutant flies did not exhibit a difference in ethanol food interactions based on previous diet (Figure 2H). Second, we measured triglyceride (TAG) abundance (Figures 1H and 2J), which has been postulated to be a physiological indicator of stress (Starzec et al., 1981; Starzec and Berger, 1986). We found that flies given a dietary choice had reduced TAG and lived shorter following starvation on 2% agar than did flies fed a homogenous food (Figure 2I). Both phenotypes were 5-HT2A dependent (Figure 2I ,J).

Figure 2 with 1 supplement see all
Dietary choice influences lifespan, behavior, and physiology through the serotonin receptor 2A (5-HT2A).

(A) P-element insertion in 5-HT2A abrogates the differences in male lifespan caused by dietary choice (N = 145–148 per group). (B) Pan-neuronal knockdown of 5-HT2A expression eliminates (with UAS-5-HT2A-RNAiTRiP) or reduces (with UAS-5-HT2A-RNAiKK) the choice effect compared to control flies (N = 145–150 per group). (C) Pharmacological treatment of Canton-S males with a serotonin receptor 5HT2-antagonist, pirenperone, during adulthood effectively abrogates the lifespan differences caused by dietary choice (N = 145–153 per treatment). (D) 5-HT2A males exhibit similar feeding preferences on the choice diet, but consume more nutrients than w1118 animals on both diets (N = 97–115 for each group). (E) Both w1118 and 5-HT2A males spent a majority of their time on yeast food in a choice environment (PI = (Tyeast− Tsugar)/(Tyeast + Tsugar); N = 24 for each group). (F–G) Males in a choice environment exhibit increased activity: they traveled farther (F), demonstrated more walking and micromovements, and spent less time sleeping and resting (G) when they are exposed to choice diet, and such effects are independent of 5-HT2A (N = 10 or 11 per group.) (H) A schematic illustrating the ethanol feeding assay. Males were fed on choice diet or fixed diet for 10 days before being placed in chambers where they could feed from wells containing food solutions with 15% ethanol (refer to the Materials and methods for more details). w1118 flies previously fed on choice diet show significantly more licks on ethanol-containing food than sibling flies fed on fixed diet. 5-HT2A mutants show the same trend independent of the diet they previously fed on (N = 16–21 for each treatment). (I–J) One to 2 week’s feeding on choice diet decreases starvation resistance (I, N = 72–88 per group) and TAG amounts (J, N = 48 per group), and both effects require 5-HT2A. p-Values were obtained for lifespan and starvation resistance using a log-rank test, for nutrient consumption and activity using a two-way ANOVA analysis, for positional preference using a Mann–Whitney U test, and for ethanol feeding using a one-tail Student’s t-test. Boxplots in (D, E, F, G, H, and J) represent the medians (the middle bar) and the first and the third quartiles (box boundaries) for each experimental group.

Figure 2—source data 1

Requirement of 5-HT2A in the dietary choice effects.

https://cdn.elifesciences.org/articles/59399/elife-59399-fig2-data1-v2.xlsx

Our results indicated that 5-HT2A is not involved in determining how animals interact with the choice environment, yet the requirement of this neuronal signaling in lifespan (Figure 2B) and physiological traits (Figure 2I,J) suggested that it may play a role in mediating internal nutrient homeostasis when flies are presented with a dietary choice. We therefore sought to identify changes in metabolite abundance using targeted metabolomics. To delineate an appropriate window of time to study, we completed a demographic analysis of the effect of dietary choice on age-specific mortality. Mortality rates of male flies whose nutrients were presented separately following eclosion exhibited measurable and high mortality rates by day 10, while similar levels of mortality in flies exposed to a homogenous food did not appear until day 40 (Figure 3A, compare black and orange solid lines). To determine how quickly the effects of nutrient presentation manifest, we performed a switch experiment in which flies fed a homogenous medium since eclosion were given a dietary choice beginning on day 20. We observed that the mortality rate of the switched cohort rose within 48 hr after choice began, and after roughly 5 days, it reached the level of the cohort whose nutrients had been separate since eclosion. We also performed the opposite experiment in which nutrients were combined into a fixed diet at day 20. Flies in this cohort exhibited a rapid decrease in mortality, again within 48 hr. Mortality rates in this cohort reached the level observed for flies that had been fed a homogenous food since eclosion, but this process took roughly 10 days (Figure 3B), which was longer than the reverse (Figure 3A). Cohorts that comprised 5-HT2A mutant animals exhibited similar patterns of mortality regardless of nutrient presentation, and these patterns were also unaffected by dietary switch (Figure 3C,D).

Figure 3 with 3 supplements see all
Temporal effects of dietary choice on mortality and metabolomes.

(A–D) 5-HT2A mediates mortality responses to the introduction or removal of a choice diet. Mortality rate is plotted on the natural log scale (N = 225–282). Male flies are either fed on a choice diet (CD) or a fixed diet (FD) at the onset of adulthood. The dietary switch regime, either switched from FD to CD or switched from CD to FD, was applied at 20 days of adulthood, where the mortality rate of w1118 males fed on CD is significantly higher than w1118 males fed on FD. (A) w1118 flies that previously fed on FD showed a rapid increase in mortality within 48 hr when switched to CD at 20 days of adulthood, compared to those of the same age that continually fed on FD. About 5 days after the switch, the switched cohorts were indistinguishable from those of the same-age flies maintained on a CD through adulthood. (B) Reciprocal switches from CD to FD resulted in immediate reduction in mortality rate within 48 hr. At 30 days of adulthood, the mortality rate of the switched cohorts was virtually the same as flies that fed on a CD throughout the adulthood. (C–D) 5-HT2A flies from different dietary groups did not show differences in mortality rate, suggesting that the mortality effect is mediated through serotonin signaling. (E) Dietary switch paradigm illustrated at the different time points for metabolomics samples. (F–I) Principal component analysis (PCA) isolated the metabolome component that encodes the memory of previous diet and the component that represents the response to new diet. PCA plot shows the distribution of samples for each treatment in the head metabolomes and in body metabolomes. In the head (F–H), metabolites weighted heavily in PC3 (in the top panel), PC6 (in the bottom panel), and PC2 (in both panels) distinguish the memory of previous diet, the response to new diet, and the genotype, respectively. In the body (G–I), the effect of previous diet, the effect of new diet, and the effect of genotype were distinguished by the metabolites heavily weighted in PC2, PC6, and PC3, respectively (N = 8–10 samples for each treatment, and 30–50 heads or bodies were used for each sample).

We reasoned that metabolic changes that are causally related to aging would be identified by focusing on those that were abrogated by loss of 5-HT2A. Others, perhaps those triggered by differences in feeding behaviors or food intake (and independence of the lifespan effect), would persist in the two strains. We also chose to sample flies 48 hr after a dietary switch because this period was sufficient to significantly affect mortality rates (see Figure 3A and B) and because we felt that a comparison of switched to unswitched cohorts at this time would help further isolate changes that were causal to lifespan (Figure 3E). Finally, although neural expression of 5-HT2A was required for the lifespan effects of dietary choice (Figure 2B above), it is also putatively expressed in salivary glands, in digestive tissues such as the crop, and in reproductive organs (Supplementary file 2). We therefore collected heads and bodies separately to determine whether the metabolomic effects of the two nutritional environments were enriched in the brain or whether they also manifested in peripheral tissues.

Using principal component analysis (PCA), we identified three principal components (PCs) in each of the tissue samples by which metabolome profiles effectively separated genotypic effects from those caused by nutrient presentation (additional PCA are presented in Figure 3—figure supplement 1 and Figure 3—figure supplement 2). Genotype effects on the metabolome were demonstrated by PC2 and PC3 in the head and body data, respectively (Figure 3F,G). A second PC in each data set identified diet history (PC3 and PC2 in heads and bodies, respectively; Figure 3F,G). These combinations of metabolites provided separation based on how nutrients were presented to the flies for the 20 days prior to the dietary switch, and they were largely unaffected by the switch itself. Separation was also observed in 5-HT2A mutant flies, suggesting that the metabolites that load strongly on these axes exhibit slow, persistent changes in response to differences in nutrient presentation that are not associated with changes in mortality. Finally, a single PC in both heads and bodies (represented by PC6 in both data sets) distinguished acute responses to dietary switch in both tissues that were abrogated in 5-HT2A mutant animals (Figure 3H,I). These PCs identified specific metabolites in either heads or bodies that were correlated with changes in mortality and that shared a common 5-HT2A dependence.

The PCA results suggested that nutrient presentation influenced two processes: (A) a relatively stable 5-HT2A-independent metabolic network that responded slowly to dietary environment and that accounted for a significant amount of metabolome variation (10.2–11.2%) and (B) a dynamic 5-HT2A-dependent metabolic network that accounted for a smaller component of variation (2.0–2.2%) that responded quickly to dietary choice and that was associated with mortality. Using generalized linear models, we identified individual metabolites that exhibited statistically significant changes consistent with membership in each of these two groups (a summary of the analysis of a third group that distinguished genotype effects is presented in Figure 3—figure supplement 3; statistics for each metabolite are presented in Supplementary file 3). Criteria for inclusion in Group A consisted of a statistically significant effect of diet (pDiet<0.05) and a non-significant diet × genotype interaction (pDiet*Genotype>0.05). Of the 103 and 125 metabolites that were detected in each tissue sample, respectively, we found that 45 (heads) and 46 (bodies) satisfied these criteria (Figure 4A). Among those metabolites, 13 (heads) and 21 (bodies) exhibited an increase in abundance when nutrients were presented independently, while abundances of the remaining metabolites were reduced. Although our sample of measured metabolites was small and precluded a comprehensive pathway analysis (see Materials and methods), MetaboAnalystR (Pang et al., 2020) applied to the full list of significant metabolites revealed several candidate pathways (i.e., those with more than three significantly affected metabolites in each), nearly all involved in amino acid metabolism, which may be specifically affected by dietary choice (Figure 4C, see Materials and methods). Over 40% of the metabolites (N = 19 of 45 or 46 metabolites in the head or body, respectively) were modulated in both samples (starred in Figure 4A). While 75% of these (N = 14) were modulated in the same direction, there remained nearly a quarter of them that exhibited opposite changes in abundance.

Nutrient presentation mediates metabolic changes in both head and body.

(A–B) Generalized linear model separates the metabolites that respond to dietary choice in a 5-HT2A-independent (A) or -dependent (B) manner. Up- or down-regulation of metabolites upon choice are represented by red and blue, respectively. Metabolites that were altered in both heads and bodies are indicated by asterisks. (C–D) Metabolic pathways that are influenced by choice diet are revealed by MetaboAnalystR. Pathways with a number of hits > 3 are shown.

Changes in Group B, which represented the 5-HT2A-dependent metabolome, were fewer in number but more consistent across tissue samples. We identified 18 of 103 total metabolites in the head and 19 of 125 total metabolites in the body that were present in this group (Figure 4B, ANOVA, pDiet*Genotype≤0.05). In this case, six of seven of the metabolites that were present in both groups (starred in Figure 4B) were modulated in the same direction. Furthermore, MetaboAnalystR identified related pathways, involving the synthesis of aminoacyl-tRNAs or specific amino acids, which were found either in both tissues or only in the body (Figure 4D). Among them, ‘alanine, aspartate and glutamate metabolism’ was the only category of interest in both tissues, specifically in a 5-HT2A-dependent manner. A closer look into its metabolic map (KEGG ID: map00250) revealed that candidate metabolites are either members of the TCA cycle or the direct precursors of TCA metabolites, which indicated the modulation of this metabolic process in response to dietary choice.

A detailed examination of 5-HT2A-dependent metabolites, and the relationships among them, indicated significant upregulation of the TCA cycle and amino acid metabolism when nutrients were presented separately (Figure 5A). The TCA cycle is a series of biochemical reactions that generates energy (ATP) through the oxidation of acetyl-CoA derived from different nutrients, while the same reactions also provide precursors for amino acids (Figure 5B). Specifically, we found that oxaloacetate abundance was increased in both tissue samples in a 5-HT2A-dependent manner, while α-ketoglutarate and fumarate were upregulated only in heads and bodies, respectively. Further analysis revealed consistency among the abundances of the amino acid precursors of these intermediates, which were also increased in their corresponding tissues (Figure 5A,B). Asparagine, the immediate precursor for oxaloacetate, was increased in both heads and bodies; glutamine, the major precursor for α-ketoglutarate, was increased only in heads; and aspartate, the precursor of fumarate, was increased only in bodies. Although we were unable to measure abundance of acetyl-CoA directly, we observed that the abundance of lysine, its precursor, was increased in both tissue samples.

Figure 5 with 2 supplements see all
Nutrient presentation modulates central metabolic pathways through serotonin signaling.

(A) Boxplots show the TCA intermediates and amino acids that are modulated by choice diet in a 5-HT2A-dependent manner (two-way ANOVA, p<0.05 for the interaction between diet and genotype). (B) The metabolites of the TCA cycle. Orange and purple arrows indicate the metabolites that are regulated in head and body when the flies are given choice diet conditions, respectively. (C) Ubiquitous knockdown of glutamine dehydrogenase (GDH) using the GeneSwitch system fully abrogates the lifespan differences between the two diets when RU-486 was administrated in the food (N = 144–153 per group; p-values are calculated using a log-rank test). (D) Knockdown of GDH has no effects in nutrient consumption on either diet (N = 71–94 per group, p-Values are calculated using a Mann–Whitney U test).

Figure 5—source data 1

Lifespan and feeding data from GDH knock-down.

https://cdn.elifesciences.org/articles/59399/elife-59399-fig5-data1-v2.xlsx

Inspired by concurrent changes in TCA intermediates and their precursors, we investigated the direct roles of biochemical reactions that produce the TCA metabolites in the modulation of aging in response to dietary choice. We targeted the enzyme GDH, which converts glutamate to α-ketoglutarate, for RNAi-mediated knockdown. Our metabolomic data revealed significantly increased abundance of α-ketoglutarate in the heads of animals given a dietary choice, and α-ketoglutarate has been shown to modulate longevity in Caenorhabditis elegans (Chin et al., 2014). Knocking-down GDH abundance in adult flies using the RU486-dependent GeneSwitch system (Roman et al., 2001) (tub-GS-GAL4) coupled with UAS-GDH-RNAi blocked the lifespan differences between two dietary environments (Figure 5C). This was not caused by the transcriptional inducer, RU486, because control animals (tub-GS-GAL4 coupled with UAS-GFP) responded as expected to dietary choice when RU486 was presented (Figure 5—figure supplement 1). GDH knockdown had no effect on nutrient consumption (Figure 5D) or total protein abundance in the fly (Figure 5—figure supplement 2). Which tissues/cells are critical for the diet-specific lifespan effects induced by GDH knockdown, and whether such influences are due to changes in α-ketoglutarate per se or to alterations in the metabolic network more broadly (Hoffman et al., 2017) remain to be determined.

Discussion

Here we studied the effects of meal presentation on aging and physiology in Drosophila. By separating the two macronutrients of the standard Drosophila laboratory diet (i.e., brewer’s yeast and sucrose) and allowing the animals to behaviorally structure their own diet, we found that lifespan was significantly shortened in both male and female flies. In addition, changes in activity, stress resistance, lipid metabolism, and critical metabolic processes were also elicited by dietary choice. Flies carrying a loss-of-function mutation in serotonin receptor 5A (5-HT2A), with RNAi-mediated neuronal knockdown of 5-HT2A expression, or fed the putative receptor antagonist pirenperone, exhibited no differences in lifespan between choice and fixed diets, indicating that serotonin signaling through this receptor plays an essential role in the effect of nutrient presentation on aging.

Flies voluntarily consumed more sugar than yeast when presented with a choice diet. Although it is difficult to definitively rule out differences in consumption as a cause for the phenotypes we observe, several lines of evidence indicate that is not the major factor influencing lifespan: (1) the effect of dietary choice on lifespan is significantly larger than that caused by even the most extreme carbohydrate-rich fixed diets (as shown in both Figure 1G and previous studies from Skorupa et al., 2008; Dobson et al., 2017); (2) the choice diet induced physiological effects (e.g., reduced triglyceride abundance) opposite of those associated with a high-sugar fixed diet (e.g., increased or unchanged triglyceride abundance depending on the genetic background, see Skorupa et al., 2008); (3) the insulin-responsive transcription factor, dFoxo, which is an essential effector of dietary sugar on aging (Dobson et al., 2017), is not required for the effect of dietary choice on lifespan; (4) manipulating the nature of the choice is sufficient to influence lifespan without altering the amount of nutrients eaten in a choice paradigm; and (5) flies with loss of 5-HT2A exhibited the same diet-dependent changes in feeding as did control animals, yet they experienced no diet-induced changes in lifespan (Figure 2A,D).

We found that dietary choice influenced lifespan in both male and female laboratory flies and that 5-HT2A is required for this effect in both sexes (see female data in Ro et al., 2016), suggesting the mechanisms are at least partially shared between them. However, sex-specific phenomena were also observed: lifespan was significantly shortened in male flies for all strains tested (from 11 to 37% in laboratory and wild strains), whereas the effects on lifespan were variable in recently wild-caught females (an average of 3% reduction) relative to a fixed diet. This is, perhaps, not surprising given the effects of different macronutrients among different strains (Jin et al., 2020) and sexes (Hudry et al., 2019; Camus et al., 2019; Jensen et al., 2015). Future studies that focus on dissecting the role of individual macronutrients on choice effects or on investigating the involvement of sex-specific nutrient sensing proteins (see Hudry et al., 2019), perhaps with a focus on population genetics of natural populations, will be useful for understanding the nature of sexual dimorphism on lifespan and physiology.

We found that dietary choice led to a rapid and reversible change in age-specific mortality only during a defined period of life. Dietary choice significantly increased mortality within 48 hr of exposure, and this effect was observed only up until middle age (roughly 40 days in our experiments) after which time mortality rates in cohorts provided a fixed or choice diet were indistinguishable. This was true regardless of whether the flies had been presented with a dietary choice starting at eclosion or starting at 20 days of age, and these effects required serotonin receptor 5-HT2A. The effects of dietary choice on demographic aging, therefore, do not appear to accumulate, and they are simply absent in older flies, perhaps because of an age-dependent reduction in serotonin signaling, a lack of consequences of such signaling in older animals, or the manifestation of demographic heterogeneity (Vaupel and Yashin, 1985). These mortality patterns are distinct from those observed following other types of dietary or environmental perturbations. DR in flies is typically carried out by diluting both the yeast and carbohydrate components of a complete, fixed diet (Chapman and Partridge, 1996). DR rapidly influences mortality dynamics, and it is reversible at early ages. Unlike dietary choice, however, DR influences demographic aging across the whole of adulthood (Mair et al., 2003). The effect of different temperatures on mortality is cumulative and permanent (Mair et al., 2003; Carvalho et al., 2017).

The neuronal response to dietary choice reshapes metabolic networks that are associated with energy homeostasis in peripheral tissues (see our model in Figure 6). Some of these changes are mediated by 5-HT2A, which is widely expressed in the central nervous system (Gnerer et al., 2015) and in the thoracicoabdominal ganglion (Supplementary file 2). Our metabolomic data revealed that TCA cycle components, as well as key components of amino acid metabolism, are strongly affected by food choice and that such changes require 5-HT2A. There is evidence that at least one such metabolite, α-ketoglutarate (αKG), is directly involved in the effect of dietary choice on lifespan. αKG was more abundant when flies were presented with a choice diet, and inhibition of GDH, which is involved in αKG synthesis, abrogated the difference in lifespan between choice and fixed diets (see also Talbert et al., 2015). In C. elegans, it has been reported that αKG modulates longevity by inhibiting TOR signaling (Chin et al., 2014) and that oxaloacetate increases lifespan through an AMPK/FOXO-dependent pathway (Williams et al., 2009). The effect of αKG on lifespan has recently been independently documented in flies (Su et al., 2019), and, more recently, in mice (Asadi Shahmirzadi et al., 2020), suggesting a conserved role for αKG in modulation of aging. Published data in Drosophila suggest that supplementation of αKG to a standard laboratory diet increased transcript abundance of AMPK and FOXO and reduces that of TOR (Su et al., 2019). However, our data indicate that neither FOXO (Figure 1—figure supplement 2) nor TOR (Figure 1—figure supplement 3) signaling is involved in the effects of dietary choice on lifespan, suggesting that αKG may recruit different mechanisms in this environment. Perhaps instead of energy sensing pathways, metabolic enzymes that control the branch point of crucial reactions (such as GDH) are more relevant in the context of a choice environment (Smith et al., 2019).

Our working model showing how 5-HT2A regulates energy homeostasis to influence aging and physiology in response to nutrient presentation.

We propose the idea that the effects of dietary choice may result from metabolic changes stimulated by an alteration in metabolic state induced by the need to continuously evaluate macronutrient availability against physiological demand, which would not exist in a fixed-diet environment. Flies housed in a choice environment, for example, are allowed to interact with each macronutrient individually, and this behavior would be expected to result in the sequential activation of sugar and protein sensing pathways. Flies exposed to a complete, fixed diet would only experience nutrients simultaneously. The perceived flavor of each macronutrient may be different when yeast and sugar are consumed separately compared to when they are paired together on a fixed diet (Ahn et al., 2011) and that may also influence lifespan and metabolism (Ostojic et al., 2014). Notably, when given a choice, flies consumed more sugar than yeast but spent the majority of time physically on the yeast food. Loss of 5-HT2A did not alter these behaviors, establishing that control flies are capable of recognizing the components of a choice diet and that animals lacking 5-HT2A retain this ability.

Perhaps the act of decision-making per se is costly? In nature, decisions are made continuously following evaluation of environmental resources to allow animals to adjust their behavior in a way that would maximize their fitness (Galef, 1996; Lima, 1998). In humans, an abundance of choices can be stressful, which is thought to reflect the perceived risk of making a bad choice and therefore losing out on other, more valuable resources (Schwartz, 2004; Inbar et al., 2011). Although acute stress can be beneficial (Dhabhar et al., 2012), chronic stress is often detrimental to lifespan and health (Epel et al., 2004; Glaser and Kiecolt-Glaser, 2005; Lupien et al., 2009). Drosophila exhibit stress-related behaviors that are mediated by neuronal signaling pathways that are conserved in mice and humans, including serotonin signaling (Yang et al., 2013; Sachs et al., 2015; Mohammad et al., 2016; Ries et al., 2017). It seems reasonable, therefore, to consider the possibility that 5-HT2A influences metabolic decisions in response to the perception or consumption of individual nutrients in ways that are important for aging. Biogenic amines, including serotonin and dopamine, have been shown to encode protein hunger and reward (Ribeiro and Dickson, 2010; Vargas et al., 2010; Ro et al., 2016; Liu et al., 2017), and serotonin receptor 5-HT2A has been implicated in mammalian stress (Weisstaub et al., 2006; Sachs et al., 2015) and in a putatively stressful situation in flies in which prolonged sight of dead conspecifics increases mortality and shortens lifespan (Chakraborty et al., 2019). Notably, a recent study in mice reported that unknown characteristics of each meal modulate aging independent of dietary composition or calories (Mitchell et al., 2019), suggesting that how animals recognize, interact with, or interpret their food may be a mechanism of aging that is conserved across taxa.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional information
Gene
(Drosophila melanogaster)
5-HT2AFlyBaseFLYB:FBgn0087012
Gene
(Drosophila melanogaster)
GDHFlyBaseFLYB:FBgn0001098
Genetic reagent (Drosophila melanogaster)w1118Bloomington Drosophila Stock Center
Genetic reagent (Drosophila melanogaster)Canton-SBloomington Drosophila stock Center
Genetic reagent (Drosophila melanogaster)5-HT2APL00052Bloomington Drosophila Stock CenterBDSC Cat# 19367, RRID:BDSC_19367Also see Materials and methods, section 1
Genetic reagent (Drosophila melanogaster)wDahomeyOtherL. Partridge (University College London, UK)
Genetic reagent (Drosophila melanogaster)dFoxoΔ94OtherBDSC Cat# 42220; RRID:BDSC_42220L. Partridge (University College London, UK)
Genetic reagent (Drosophila melanogaster)Tub-GS-GAL4OtherR. Davis (The Scripps Research Institute, Jupiter, FL)
Genetic reagent (Drosophila melanogaster)elav-GAL4Bloomington Drosophila Stock CenterBDSC Cat# 458
Genetic reagent (Drosophila melanogaster)UAS-5-HT2A-RNAiTRiPBloomington Drosophila Stock CenterBDSC Cat# 31882; RRID:BDSC_31882
Genetic reagent (Drosophila melanogaster)UAS-5-HT2A-RNAiKKVienna Drosophila Resource CenterVDRC Cat# 102105; RRID:FlyBase_FBst0475582
Genetic reagent (Drosophila melanogaster)UAS-GDH-RNAiBloomington Drosophila Stock CenterBDSC Cat# 51473; RRID:BDSC_51473
Genetic reagent (Drosophila melanogaster)UAS-eGFPBloomington Drosophila Stock CenterBDSC Cat# 79025; RRID:BDSC_79025
Strain, strain background (Drosophila melanogaster)18In08-17OtherP. Schmidt (University of Pennsylvania, PA)
Strain, strain background (Drosophila melanogaster)18Ln10-4OtherP. Schmidt (University of Pennsylvania, PA)
Software, algorithmDLifeOtherS. Pletcher (University of Michigan)
Software, algorithmDTrackOtherS. Pletcher (University of Michigan)
Software, algorithmDDropOtherS. Pletcher (University of Michigan)
Software, algorithmRStudioRStudio

Fly stocks

Request a detailed protocol

The standard laboratory stocks w1118, Canton‐S, and the mutant 5-HT2APL00052 were originally obtained from the Bloomington Drosophila Stock Center. As described in our previous publication (Ro et al., 2016), 5-HT2APL00052 were backcrossed 10 generations to w1118 prior to all the experiments. The wDahomey and dFoxoΔ94 lines were provided by L. Partridge (University College London, UK). dFoxoΔ94 were backcrossed into wDahomey prior to the lifespan experiments. Two isofemale lines (18In08-17 and 18Ln10-4, progeny from single wild females that were collected from two different orchards in Media, PA, in 2018) were gifts from P. Schmidt (University of Pennsylvania, PA). Tub-GS-GAL4 was obtained from R. Davis (The Scripps Research Institute, Jupiter, FL). elav-GAL4 (BDSC#458) and UAS transgenic lines, including UAS-5-HT2A-RNAiTRiP (BDSC#31882), UAS-5-HT2A-RNAiKK (VDRC#102105), UAS-GDH-RNAi (BDSC#51473), and UAS-eGFP (BDSC#79025), were purchased from either the Bloomington Drosophila Stock Center or the Vienna Drosophila Resource Center.

Husbandry

Request a detailed protocol

All fly stocks were maintained on a standard cornmeal-based larval growth medium (produced by LabScientific Inc and purchased from Fisher Scientific) and in a controlled environment (25 °C, 60% humidity) with a 12:12 hr light:dark cycle. We controlled the developmental larval density by manually aliquoting 32 µl of collected eggs into individual bottles containing 25 ml of food. Following eclosion, mixed-sex flies were kept on SY10 (10% [w/v] sucrose and 10% [w/v] yeast) medium for 2–3 days until they were used for experiments. Pioneer table sugar (purchased from Gordon Food Service, MI) and MP Biomedicals Brewer’s Yeast (purchased from Fisher Scientific) were used in our study.

Dietary choice environment

Request a detailed protocol

To study the lifespan, behavior, and physiology of flies that have a dietary choice, we created food wells that were divided in the middle (Figure 1A). This allowed us to expose the flies to two separate sources of food simultaneously. For these experiments, we either loaded different foods on each side (choice diet) or the same food on both sides (fixed diet). Unless otherwise mentioned, the choice diets contained 10% (w/v) sucrose or 10% (w/v) yeast on either side, and the fixed diets contained a mix of 10% (w/v) sucrose and 10% (w/v) yeast on each side.

Drug administration

Request a detailed protocol

We purchased all drugs from Sigma-Aldrich. In the pirenperone experiment, drug was initially dissolved in 100% dimethyl sulfoxide (DMSO) at 10 mM concentration, aliquoted and stored at −20°C. Prior to the experiment, 200 µM pirenperone and the same dilution of DMSO (control vehicle) were made. Then, 50 µl of the diluted drug or control vehicle was added to each food well. After the liquid evaporated (~2 hr), the food plates were ready for use as described in Survival assays.

In the GeneSwitch experiment, RU486 (mifepristone) was first dissolved in 80% (v/v) ethanol at 10 mM concentration, marked by blue dye (5% [w/v] FD and C Blue No. 1; Spectrum Chemical), and stored at −20°C. For experimental food, 200 µM RU486 or the same dilution of control vehicle (80% [v/v] ethanol) was made from the stock and added to the food. To ensure flies consumed the same amount of RU486 between the choice diet and the fixed diet, drug was only administrated in the yeast portion of the choice diet, since the yeast intake on a choice diet is indistinguishable from the sugar–yeast intake on the fixed diet (Figure 1E). Lifespan assay was carried out as described below.

Survival assays

Request a detailed protocol

Lifespans were measured using established protocols (Linford et al., 2013). Normally, six replicate vials (~150 experimental flies) were established for each treatment. Flies were transferred to fresh media every 2‐3 days, at which time dead flies were removed and recorded using the DLife system developed in the Pletcher Laboratory (Linford et al., 2013). Flies were kept in constant temperature (25°C) and humidity (60%) conditions with a 12:12 hr light:dark cycle.

Consumption–excretion (feeding assays)

Request a detailed protocol

We used the protocol described previously (Shell et al., 2018). Experimental flies were co-housed in vials (10 flies per vial, 8–10 replicates for each group) for 2–3 days following eclosion and then sorted into individual sex cohorts. After exposure to the choice or fixed diet for 12 days (food was changed every 2–3 days), flies were transferred to the choice or fixed diet with 1% (w/v) FD and C Blue No. 1 in either (1) only the S10 (10% [w/v] sucrose) food in the choice diet; (2) only the Y10 (10% [w/v] yeast) food in the choice diet; or (3) food in both wells in the fixed diet SY10 (10% [w/v] sucrose and 10% [w/v] yeast). Vials were discarded if one or more dead flies were observed after the 24 hr feeding period. Excreted dye (ExVial) was collected by addition of 3 ml of Milli-Q water of vials followed by vortexing. Concentration of the ExVial dye in water extracts was determined by absorbance at 630 nm, which was used to infer macronutrient consumption.

Fly Liquid-Food Interaction Counter feeding assays

Request a detailed protocol

Flies were tested on the Fly Liquid-Food Interaction Counter (FLIC) system as previously described to monitor ethanol feeding (Ro et al., 2014) after exposure to a choice or fixed diet for 10 days. We filled the liquid-food reservoir with 15% (v/v) ethanol + 5% (w/v) sucrose in 4 mg/l MgCl2. We synchronized the feeding state prior to loading flies onto the FLIC system. Flies were removed from their food environment to kimwipes with 2 ml of water 2 hr before their natural breakfast time. At the peak of their normal breakfast meal (lights on), we flipped flies back onto their respective choice or fixed diets and let them feed for 2 hr. Following breakfast feeding, flies were briefly fasted for 3 hr on a wet kimwipe with 2 ml of water. We began the FLIC recording software before aspirating flies into the system and analyzed food or ethanol interactions for 6 hr. Flies were anesthetized briefly on ice and manually aspirated into the Drosophila feeding monitors (DFMs). Each DFM was loaded with flies from at least two treatment groups to reduce technical bias from the DFM signals. FLIC data were analyzed using custom R code, which is available at wikiflic.com. Default thresholds were used for analysis except for the following: minimum feeding threshold = 20, minimum events = 1, and tasting threshold = (10, 20). Flies that had zero feeding events over the testing interval were removed from the analysis.

Video tracking and analyses

Request a detailed protocol

Experimental males were reared on SY10 food for 2–3 days shortly after eclosion. We aspirated individual flies into the two-choice food arenas (diameter = 31 mm, either a choice diet or a fixed diet, N = 12 for each group) and let them adapt to the environment for 1 day. Then, 6 hr video recording (10:00–16:00, at two frames per second with a resolution at 528 × 320) was taken and analyzed with DTrack, a video analysis package developed in the Pletcher Laboratory. Using the same speed threshold published before (Corrales-Carvajal et al., 2016), we classified the movements of flies into resting or sleeping (speed ≤ 0.2 mm/s), micromovement (0.2 mm/s < speed ≤2 mm/s), and walking (speed > 2 mm/s). For positional preference assay, we aspirated individual male flies into the two-choice food arenas (N = 24 for each genotype) after the designated exposure to the choice diet for 7 days. Three hour video recordings were taken (at two frames per second with a resolution at 528 × 320) and analyzed with DTrack. The fraction of time flies spent on each macronutrient was computed and used for statistical analyses.

Climbing activity assay

Request a detailed protocol

After the designated exposure to the choice or fixed diet for 10 days (food was changed every 2–3 days), male flies were transferred to the negative geotaxis apparatus using brief CO2 anesthesia, after which they were allowed to recover for 30 min. We used an automated process in which flies were dropped from 24’. After freefall, which effectively knocked the flies to the bottom of the chamber, a video camera was triggered and individual fly movements were tracked for 10 s using the DDrop software developed in our laboratory. From the tracking data, we were able to calculate, for each male, the total distance traveled up in 10 s.

Fly weight, TAG, and protein assays

Request a detailed protocol

After the designated exposure to the choice or fixed diet for 2 or 3 weeks (food was changed every 2–3 days), experimental flies were quickly frozen, collected into groups of five, weighed, and then homogenized in 200 μl of cold phosphate-buffered saline containing 0.1% Triton X-100 (IBI Scientific) for 30 s at 30 Hz using a QIAGEN TissueLyser. For TAG quantification, the homogenate (20 μl) was added into 200 μl of Infinity Triglyceride Reagent (Thermo Electron Corp.) and incubated at 37°C for 10 min with constant agitation. TAG concentrations were determined by the absorbance at 520 nm and estimated by a known triglyceride standard. For protein measurement, 5 μl of fly homogenate was incubated with 200 μl of (1:50) 4% (w/v) cupric sulfate/bicinchoninic acid solution (Novagen) at room temperature for 30 min. Protein concentrations were estimated by the absorbance at 562 nm through the comparison with bovine serum albumin standards. Average weight, TAG, and protein values were based on at least six independent biological replicates (of five flies each) from multiple vials.

Starvation assay

Request a detailed protocol

After the designated exposure to the choice or fixed diet for 20 days (food was changed every 2–3 days), flies were transferred to fresh vials containing 1% (w/v) agar. The number of dead flies was recorded approximately every 2–5 hr using DLife system.

Smurf assay

Request a detailed protocol

We investigated intestinal integrity using the Smurf assay (Regan et al., 2016; Rera et al., 2012). Experimental flies were grouped in vials (25 flies per vial and 12 replicates per each diet) 2–3 days following eclosion. Following exposure to a choice or fixed diet for 12 days (food was changed every 2–3 days and survivorship of w1118 males on choice diet was roughly 70%), flies were transferred to S30 (30% [w/v] sucrose) food with 2.5% (w/v) blue dye (FD and C Blue No. 1; Spectrum Chemical) for 24 hr. A fly was counted as a Smurf when blue was observed outside of the digestive tract (Regan et al., 2016; Rera et al., 2012). The proportion of Smurf flies in each replicate was calculated and used as a single observation for statistical analysis.

Egg laying assays

Request a detailed protocol

Flies were raised according to our standard husbandry protocols (see above), and following eclosion, males and females were allowed to mate on SY10 food for 48–72 hr. Eight females and eight males were then sorted into individual vials, which contained either a choice or a fixed diet. Egg counts were taken on days 1, 2, 5, and 8 after sorting. Flies were transferred to fresh SY food 24 hr prior to the counting.

Western blot

Request a detailed protocol

After exposure to the choice or fixed diet for 20 days, experimental flies were flash frozen in liquid nitrogen. Heads and bodies were separated using a metal sieve on dry ice, and 10 heads were pooled for each biological replicate. Heads were first pulverized to a fine powder using a plastic pestle on dry ice. Protein extraction was carried out on ice using RIPA buffer (Sigma–Aldrich) supplemented with protease inhibitor cocktail (Sigma), phosphatase inhibitor cocktail (Sigma), sodium orthovanadate (NEB, 1 mM), and sodium fluoride (NEB, 1 mM). Ice cold buffer (105µl) was added to heads followed by immediate homogenization with a motorized pestle for 10 s on ice. Lysates were incubated on ice for 10 min followed by 15 s of sonication and centrifugation at 16,000 × g at 4°C for 10 min. Supernatants were recovered, and protein concentration was determined using the Pierce BCA Protein Assay Kit (Fisher Scientific). Thirty micrograms of protein lysate was added to 2× protein sample buffer (1 mM Tris–HCL pH 6.8, 10% sodium dodecyl sulphate [SDS], 1% bromophenol blue, and 1M dithiothreitol) and denatured at 95°C for 10 min. Protein was separated by SDS–polyacrylamide gel electrophoresis on a 4–12% gel (Bio-Rad) at 200 V for 30 min, followed by electrophoretic transfer to PVDF membrane at 70 V for 1 hr. Blots were incubated in 5% milk in 1% TBS-T at room temperature for 1 hr, followed by overnight incubation with primary antibodies overnight at 4°C (anti-dS6K gift from Thomas Neufeld; dilution at 1:3500, anti-pS6KThr398 Cell Signaling Technologies, #9209; dilution at 1:1000, anti-GAPDH; dilution 1:6000). Membranes were washed with 1% TBS-T and incubated with HRP-conjugated secondary antibodies (Abcam) at room temperature for 4 hr. Membranes were washed again with 1% TBS-T and then incubated briefly in ECL substrate (SuperSignal West Femto, ThermoFisher) before imaging.

Metabolomics analysis

Request a detailed protocol

Male flies were exposed to the choice or fixed diet for 20 days (food was changed every 2–3 days) and then switched to the other diet (experimental groups) or the same diet (control groups). Heads were removed via vortexing and manually separated from the bodies. Heads or bodies were then homogenized for 20 s in 200 μl of a 1:4 (v:v) water:MeOH solvent mixture using the Fast Prep 24 (MP Biomedicals). Following the addition of 800 μl of methanol, the samples were incubated for 30 min on dry ice, then homogenized again. The mixture was spun at 13,000 RPM for 5 min at 4°C, and the soluble extract was collected into vials. This extract was then dried in a speedvac at 30°C for approximately 3 hr. Using a liquid chromatography triple quadrupole tandem mass spectrometry (LC–QQQ–MS) machine in the multiple reaction monitoring (MRM) mode, we targeted 205 metabolites in 25 important metabolic pathways, in both positive MS and negative MS modes.

After removing any metabolites missing from more than 8 of 76 head samples (10.5%) and 8 of 78 body samples (10.3%), we were left with 103 head metabolites and 125 body metabolites. Metabolite abundance for remaining missing values in this data set were log-transformed and imputed using the k-nearest neighbor algorithm with the impute package of R Bioconductor (http://www.bioconductor.org). We then normalized the data to the standard normal distribution (μ = 0, σ2 = 1). PCA was performed using the made4 package of R Bioconductor.

Our PCA results suggest the choice diet can elicit two processes: (1) a 5-HT2A-dependent, fast-responding metabolic network that is associated with aging despite only accounting for a small number of variation in the metabolomes (2.0–2.2%) and (2) a 5-HT2A-independent, stable metabolic component that accounts for a significant amount of metabolome variation (10.2–11.2%). To distinguish these two groups of metabolites, we used a linear model to analyze the variance attributed to diet, genotype, and interaction of the two for each metabolite:

Metabolite abundance (Y) = Diet + Genotype + Diet * Genotype.

We found 18 of 103 total metabolites in the head and 19 of 125 total metabolites in the body are modulated by choice diet in a 5-HT2A-dependent manner (Figure 4A,B, ANOVA, pDiet*Genotype≤0.05). To identify 5-HT2A independent change, we searched for metabolites that have a diet effect (pDiet≤ 0.05) but did not show a diet–genotype interaction (pDiet*Genotype>0.05). In doing so, we found 45 and 46 metabolites in the head and body, respectively. We plotted a heatmap to illustrate the expression changes of the metabolite upon nutrient presentation using the R package gplots. The metabolites were mapped to biochemical pathways using MetaboAnalystR 3.0 (Pang et al., 2020). Because the small number of detectable compounds in our targeted metabolomics panel (N = 103 in heads or 125 in bodies) limits the power of the enrichment analysis, candidate pathways were therefore identified as those with the number of hits greater than 3.

Statistics

Unless otherwise indicated, pairwise comparisons between different treatment survivorship curves (both lifespan and starvation resistance) were carried out using the statistical package R within DLife (Linford et al., 2013). p‐Values were obtained using log‐rank test. For testing the interaction between genotypes and diets, we used Cox-regression analysis to report p-value for the interaction term. Mortality rates were calculated using standard methods (Partridge et al., 2005). To test the effects of diet and genotype involved in behavior classification and metabolites, we performed two-way ANOVA (on the square root- or log-transformed values) followed by post hoc significance test. To test for a diet effect on gut integrity, climbing ability, weight, TAG/protein levels, nutrient assumption, and positional preference, we used Mann–Whitney U test. Two-tailed p-values were reported unless otherwise mentioned. We designed the layout of individual samples for lifespan, behavior, physiology, and metabolomic analyses, so non-biological factors (e.g. the location of samples) do not contribute to the differences between groups.

Data availability

Request a detailed protocol

Source data for all quantifications shown in Data Figures 15, figures supplements, and the supplementary files are provided with the paper. Metabolomic raw data, analyses, and statistics can be obtained from Supplementary files 34 and our GitHub repository (github.com/ylyu-fly/Metabolomics-FlyChoiceDietLyu, 2021; copy archived at swh:1:rev:86daef110d9a1eb27262bc5310b54cc2d315f7c1).

Data availability

Source data for all quantifications shown in Data Figures 1-5, figures supplements and the supplementary files are provided with the paper. Metabolomic raw data, analyses and statistics can be obtained from Supplementary Files 3-4 and our GitHub repository (http://github.com/ylyu-fly/Metabolomics-FlyChoiceDiet). Copy archived at https://archive.softwareheritage.org/swh:1:rev:86daef110d9a1eb27262bc5310b54cc2d315f7c1/.

The following data sets were generated
    1. Lyu Y
    2. Promislow DEL
    3. Pletcher SD
    (2016) Supplementary data
    ID FlyChoiceDiet. Head and body metabolomes in response to nutrient choice.

References

  1. Book
    1. Schwartz B
    (2004)
    The Paradox of Choice: Why Less Is More
    New York Ecco.
    1. Vaupel JW
    2. Yashin AI
    (1985)
    "Heterogeneity's ruses: some surprising effects of selection on population dynamics."
    The American Statistician 39:176–185.
  2. Book
    1. Weindruch R
    2. Walford RL
    (1988)
    Retardation of Aging and Disease by Dietary Restriction
    CC Thomas.

Decision letter

  1. Meng C Wang
    Reviewing Editor; Baylor College of Medicine, HHMI, United States
  2. Jessica K Tyler
    Senior Editor; Weill Cornell Medicine, United States
  3. David Walker
    Reviewer; UCLA, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

Your work highlights the importance of choice making in the regulation of organism lifespan and reveals the crucial role of serotonin signaling in this regulation. We believe that this finding will provide a new angle for us to understand how diet modulates aging and longevity. In addition, in order to provide a clear indication of the biological system under investigation, we would like to suggest a revision of the title, "Drosophila Serotonin 2A receptor signaling coordinates central metabolic processes to modulate aging in response to nutrient choice".

Decision letter after peer review:

Thank you for submitting your article "Serotonin 2A receptor signaling coordinates central metabolic processes to modulate aging in response to nutrient choice" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Jessica Tyler as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: David Walker (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

Summary:

Lyu et al. present interesting findings that nutritional choice affects organism physiology and lifespan. They provide compelling evidence to support that such effects are not a simple result of altered nutrient consumed. They further implicated the serotonin receptor 5-HT2A in this regulation by the dietary choice. Through metabolomic analysis, they also revealed changes in the TCA cycle and amino acid metabolism. Although reviewers agree the concept of nutritional choice regulating longevity is novel and will be of great interest to the field of aging, they suggested additional data and edits would help strengthen this claim.

Essential revisions:

1) The key claim that the observed changes are due to nutritional choice requires additional experiments.

a) When macronutrients are provided separately, there might be difference in consumption and absorption, contributing to the observed phenotypes. The authors have assayed consumption, showing flies consuming more sugar and the same amount of yeast on the choice diet. But the change in absorption has not been ruled out. The authors should determine whether consumption and/or absorption of each nutrient is altered in the choice diet when compared to mixed diet and perform this comparison in young and middle-aged files. For example, assessment of fecundity and mTOR signaling will help find possible changes in protein intake.

b) The olfactory perception of yeast can be different between the choice diet and the mixed diet. The previous finding from the lab has revealed the importance of yeast smell on fly lifespan. To rule out the smell of yeast as a confounding factor, the olfactory mutants that are smell blind should be subject to the choice diet.

2) The mechanistic follow-up of metabolomic analysis is weak. To strengthen this area, the authors should examine whether protein absorption is altered (also mentioned in the point 1) and whether GDH RNAi flies eat more under the choice diet.

3) Linking 5HT2A receptor to the regulation by the choice diet has added mechanistic insight, which needs to be strengthened by provide sufficient rationale for its selection. Are there other receptors screened? Did any additional receptors exhibit potential interactions?

4) There are many changes in the metabolomic data. What is the rationale for focusing on the TCA cycle intermediates? In addition, all metabolomic data (with statistics) should be shown without cherry-picking.

5) Statistics should be provided to Figure 2F and G.

6) Method description should be revised, including nutrient consumption (Figure 2D), Smurf assay for intestinal integrity (adding the missing reference), behavioral analysis (Figure 2F and G).

7) Additional discussion should be included to address sex specificity and genetic background difference.

Reviewer #1:

Lyu et al., present a study in which the main claim is that the choice of different nutrients itself modulates behavior, metabolism and lifespan, rather than AMOUNT of nutrients themselves. The implication of this is that caloric restriction is only one means to modulate metabolism and lifespan; food choice is another. As such, this is an interesting idea. However, several major concerns (see points below) made it unclear whether the observed effects were a true reflection of choice, or of increased sugar consumption. Further, choice of 5HT2A receptor for further study was not seen as a substantial advance, given what is already known. Finally, it was unclear from the data analysis/presentation of the metabolomic data whether TCA cycle changes were the true driver of the observed effects.

1) In Figure 1, do they have a measure of how much time the flies spend on each nutrient source in the choice diet? Without this, how do you know the flies are actually making a choice? The whole phenotype/effect could be explained by flies dwelling mostly on the sugary food (which they prefer), which the authors have shown to reduce lifespan.

2) Although the authors show in Figure 1 that intestinal integrity and climbing ability are not compromised in a 2-week period on the choice diet, it would be important to show that these measures are not more compromised than the fixed diet in older flies that have been on the choice diet for longer: 10, 20 and 30 days to assess the true effect of CD-based reduced lifespan.

3) Are the differences in Figure 2F and G significant? No statistics are provided and the data do not look convincing; correspondingly the conclusion that the diet-dependent changes in locomotion are not causal to changes in lifespan does not seem warranted.

4) The rationale for choosing 5HT2A mutants is presented as having reduced protein preference in starved female flies, and increased lifespan under dietary choice. However, it is possible that the effects they observe in the present study are an aspect of the same observation (see point 1).

5) In Figure 3, it is important to show all the PCA data in graphic format, not just those that differed amongst the various conditions.

6) Many metabolites were altered in the different conditions (Figure 4). What is the rationale for choosing to focus on the TCA cycle intermediates? It appears that there are broader changes to amino acid metabolism. Also, fold changes are not mentioned and cutoffs not given.

Reviewer #2:

This is an interesting extension of previous work from the Pletcher lab, published in eLife (Ro et al., 2016). The concept of nutritional choice impacting organismal physiology and longevity is very interesting.

The challenge, of course, is to provide irrefutable evidence that it is the “choice” that causes the observed physiological changes. On the whole, the authors have done an admirable job to exclude potential confounding factors. I have, however, some remaining concerns/suggestions for alternative explanations.

I am happy to leave it to the authors as to whether to attempt to address these concerns with additional experiments and/or to discuss in the manuscript.

Figure 1:

The entire premise of the paper assumes that the observed effects are “largely independent of nutrients consumed”. Technically, the authors have done a very good job, when focusing on nutrient consumption.

However, there could be more to this than just food consumption.

Is it possible/likely that on the choice diet the flies are absorbing more yeast/protein? Isn't the Figure 1H data consistent with this?

i.e. When the macronutrients are provided separately, are there are differences in absorption?

In the fixed diet, sugar and yeast are always consumed together.

Is consuming yeast without sugar, even for short periods, detrimental to longevity because of “elevated protein uptake”?

My (limited) understanding of human nutrition is that it really matters which ingredients are paired together in terms of health benefits, e.g. pairing tomatoes with olive oil.

Another potential issue, that given the lab's history is particularly relevant, is the issue of olfaction.

Could the smell of yeast be more pronounced when provided separately? If so, could this be a confounding factor?

Two suggestions to investigate this hypothesis:

1) Assay fecundity, which is linked to protein intake in flies in choice vs. fixed diet

2) Assay TOR signaling which is also linked to protein intake

3) Examine olfactory mutants (smell blind)

If the authors could demonstrate no difference in these parameters, it would strengthen the claims of the paper.

Figure 2:

Figure 2A-E is w1118 really an appropriate control? Precise P element revertant?

Not surprisingly, different genotypes appear to have varying degrees of reduced longevity in response to the “choice food”. I.e. Canton S in Figure 1, w1118 in Figure 2 and the RNAi strain in Figure 5. In this case (Figure 5), the effects of the choice diet appear to be modest.

Can the authors speculate as to why the 5-HT2A mutants are short-lived on the fixed diet?

The metabolic data presented (Figure 4 and 5) are interesting, but, there is little in terms of mechanistic understanding provided.

It is interesting that amino acid metabolism is upregulated in response to the choice diet. Again, as outlined above, couldn't this reflect elevated protein intake?

The GDH RNAi experiment is very interesting, but, there appears to be little follow up and/or validation. Some suggestions to validate and/or strengthen the claims:

– Validate that the RNAi line works “as advertised”.

– Do GDH RNAi flies also eat more under the choice diet? This would strengthen the claims.

– Does neuronal RNAi also alter the response?

Reviewer #3:

This is an important contribution to the field of aging, nutrition and neuroscience. It will be of great interest to the general scientific audience of eLife and I am highly supportive of its acceptance for publication.

In this manuscript, Lyu et al. report that it is how nutrients are presented, or the way that they are eaten, which modulates aging and physiology in the fruit fly, Drosophila melanogaster. The authors provide compelling evidence to demonstrate that such effects are independent of the amount of nutrients consumed and that neuronal expression of the serotonin receptor 5-HT2A is required for metabolic changes induced by dietary choice, which include upregulation of the TCA cycle and amino acid metabolism. Based on global metabolic data they identify and show that one key enzyme in these processes (Glutamate Dehydrogenase) is causal in the effect of diet choice in lifespan.

The results are intriguing and, together with previously presented data related to the role of 5-HT2A for evaluating protein demand, these discoveries demonstrate that nutrient assessment per se, and/or the feeding choices that result, have significant biological impact on aging. The effects of dietary calories and composition on aging are well-known and highly studied, and it seems reasonable to consider that specific characteristics of each meal, such as the ways in which they are presented (perhaps involving sensory perception) or the pattern in which nutrients are consumed, are important as well. These mechanisms may also be relevant to understanding the effects of different types of fasting/feeding regimes that have experienced a resurgence of interest in research (e.g., Mitchell et al., 2019) and in the general public.

The data presented in Lyu et al. are extensive, and the authors do an outstanding job of disentangling the confounding effects of behavior, feeding, and lifespan, from dietary presentation. Most importantly, the authors have provide convincing evidence that changes in lifespan in flies given a choice diet are not caused solely by a change in nutrient intake (i.e., an increase in sugar consumption). They make the extraordinary observation that at least for flies and likely many other organisms, there something of great interest occurring in making choices in the environment that is outside of the standard foci of calories or amount consumed. The support for 5-HT2A receptor in this process is absolutely solid, as is the link to specific metabolic changes through metabolic analysis and genetic manipulation. These are all major strengths of the manuscript. The authors speculate that the psychological effect of choice or the act of decision making itself may somehow be costly, and it is these assertions that while perhaps less-well supported, nonetheless follow from their work and are well appropriate for speculation given their strong data.

https://doi.org/10.7554/eLife.59399.sa1

Author response

Essential revisions:

1) The key claim that the observed changes are due to nutritional choice requires additional experiments.

a) When macronutrients are provided separately, there might be difference in consumption and absorption, contributing to the observed phenotypes. The authors have assayed consumption, showing flies consuming more sugar and the same amount of yeast on the choice diet. But the change in absorption has not been ruled out. The authors should determine whether consumption and/or absorption of each nutrient is altered in the choice diet when compared to mixed diet and perform this comparison in young and middle-aged files. For example, assessment of fecundity and mTOR signaling will help find possible changes in protein intake.

We appreciate that reviewers bring up this idea that the process of nutrient integration is complex, it happens at multiple levels and can be regulated by the characteristics of a diet. Here, and in the revised manuscript, we hope to assuage the concern about nutrient absorption by adding the fecundity and TOR data per reviewer’s requests. Indeed, neither these two measures indicate effects of dietary choice. In addition, we highlight the weight and total protein data (presented in Figure 1—figure supplement 3) to discuss the involvement of protein intake. Together, these data support the interpretation that differences in protein intake/absorption is not the cause of lifespan effects of a choice diet.

“Although protein intake was not different in flies fed different diets, it seemed possible that dietary choice improved protein absorption, which might be important considering the influential role of this macronutrient in reducing lifespan (Fontana and Partridge, 2015). […] Furthermore, molecular and physiological markers that are strongly affected by protein availability, including mTOR activity (Fontana and Partridge, 2015) and female fecundity (Jensen et al., 2015) were not altered by dietary choice (Figure 1—figure supplement 3).”

The process and effects of sugar absorption are, however, less clear than that of protein. While of interest, we believe that they are beyond the current scope of the manuscript considering the evidence that lifespan is distinct from molecular and physiological mechanisms of sugar effects, whether they be due to differences in sugar consumption or absorption.

b) The olfactory perception of yeast can be different between the choice diet and the mixed diet. The previous finding from the lab has revealed the importance of yeast smell on fly lifespan. To rule out the smell of yeast as a confounding factor, the olfactory mutants that are smell blind should be subject to the choice diet.

Our current, ongoing experiment establishes that olfactory-deficient mutants exhibit pronounced lifespan effects of a choice effects, excluding the involvement of yeast smell as a confounding factor. We will be happy to provide the data in the final version if the editors and reviewers feel this is an essential component of this study.

Author response image 1

2) The mechanistic follow-up of metabolomic analysis is weak. To strengthen this area, the authors should examine whether protein absorption is altered (also mentioned in the point 1) and whether GDH RNAi flies eat more under the choice diet.

In addition to new information about protein absorption (see above point 1a), we have provided new data on the effects of GDH knockdown. Neither feeding nor total protein in the fly was affected by GDH knock-down. We revised the text as follows:

GDH knockdown had no effect on nutrient consumption (Figure 5D) nor total protein abundance in the fly (Figure 5—figure supplement 2). Which tissues/cells are critical for the diet-specific lifespan effects that induced by GDH knockdown, and whether such influences are due to changes in α-ketoglutarate per se or to alterations in the metabolic network (Hoffman et al., 2017) more broadly, remain to be determined.”

3) Linking 5HT2A receptor to the regulation by the choice diet has added mechanistic insight, which needs to be strengthened by provide sufficient rationale for its selection. Are there other receptors screened? Did any additional receptors exhibit potential interactions?

Our choice of the 5-HT2A receptor was based on earlier studies that identified this receptor as a modulator of protein reward and originally documented diet-specific effects on lifespan (Ro et al., 2016). Our goal in the manuscript was to expand our understanding of the behavioral, physiological and aging effects of the choice diet, to determine the extent to which this effect was present in males, and to study the underlying mechanisms at play in this pathway. Loss of function in other serotonin receptors had no influence on protein preference (Ro et al., 2016), however it remains to be tested whether they might influence the aging processes in a choice environment through unknown mechanisms. We have revised text to the manuscript that clarifies this:

“Previously we hypothesized that neuronal mechanisms involved in nutrient evaluation might influence lifespan when macronutrients are presented separately (Ro et al., 2016). […] Here we focused on the elucidating the mechanisms underlying the effects of the 5-HT2A receptor and expanded our analysis of its effects on dietary choice and male lifespan.”

4) There are many changes in the metabolomic data. What is the rationale for focusing on the TCA cycle intermediates? In addition, all metabolomic data (with statistics) should be shown without cherry-picking.

We now include the statistics for all metabolites in the new Supplementary file 3. The rationale for investigating the TCA cycle has been added to the manuscript:

“Furthermore, MetaboAnalystR identified related pathways, involving the synthesis of aminoacyl-tRNAs or specific amino acids, which were found either in both tissues or only in the body (Figure 4D). […] A closer look into its metabolic map (KEGG ID: map00250) revealed that candidate metabolites are either members of the TCA cycle or are the direct precursors of TCA metabolites, which indicated the modulation of this metabolic process in response to dietary choice.”

5) Statistics should be provided to Figure 2F and G.

These statistics were provided in the original manuscript, but we realize the difficulty in identifying them. They are now highlighted in pink boxes.

6) Method description should be revised, including nutrient consumption (Figure 2D), Smurf assay for intestinal integrity (adding the missing reference), behavioral analysis (Figure 2F and G).

We have revised the Materials and methods accordingly for nutrient consumption, Smurf assay and for behavioral analysis, and the missing reference has been added now.

7) Additional discussion should be included to address sex specificity and genetic background difference.

We appreciated the general genetics context that the reviewer provided with us. A new paragraph that focuses on sexual dimorphism and genetic background has now been added in the Discussion:

“We found that dietary choice influenced lifespan in both male and female laboratory flies and that 5-HT2A is required for this effect in both sexes (see female data in Ro et al., 2016), suggesting the mechanisms are at least partially shared between them. […] Future studies that focus on dissecting the role of individual macronutrients in the choice effects or on investigating the involvement of sex-specific nutrient sensing proteins (see Hudry et al., 2019), perhaps with a focus on population genetics of natural populations, will be useful for understanding the nature of sexual dimorphism on lifespan and physiology.”

Reviewer #1:

Lyu et al., present a study in which the main claim is that the choice of different nutrients itself modulates behavior, metabolism and lifespan, rather than AMOUNT of nutrients themselves. The implication of this is that caloric restriction is only one means to modulate metabolism and lifespan; food choice is another. As such, this is an interesting idea. However, several major concerns (see points below) made it unclear whether the observed effects were a true reflection of choice, or of increased sugar consumption. Further, choice of 5HT2A receptor for further study was not seen as a substantial advance, given what is already known. Finally, it was unclear from the data analysis/presentation of the metabolomic data whether TCA cycle changes were the true driver of the observed effects.

1) In Figure 1, do they have a measure of how much time the flies spend on each nutrient source in the choice diet? Without this, how do you know the flies are actually making a choice? The whole phenotype/effect could be explained by flies dwelling mostly on the sugary food (which they prefer), which the authors have shown to reduce lifespan.

Our positional preference data were shown in Figure 2E, where we observed that flies (both controls and mutants) spend most of their time on the protein side. Furthermore, the positional preference is 5-HT2A-independent and therefore does not link mechanistically with the lifespan effects.

2) Although the authors show in Figure 1 that intestinal integrity and climbing ability are not compromised in a 2-week period on the choice diet, it would be important to show that these measures are not more compromised than the fixed diet in older flies that have been on the choice diet for longer: 10, 20 and 30 days to assess the true effect of CD-based reduced lifespan.

It is unclear to us what hypotheses the reviewer is seeking to distinguish. We specifically used these two assays (Figure 1C and D) to directly test whether the choice diet induces acute effects on general health. In our opinion, effects on long term health are more difficult to determine because inference of this sort would be directly confounded by known differences in the rates of aging between diets.

3) Are the differences in Figure 2F and G significant? No statistics are provided and the data do not look convincing; correspondingly the conclusion that the diet-dependent changes in locomotion are not causal to changes in lifespan does not seem warranted.

The relevant P-values from a two-way ANOVA were originally presented above each panel. Recognizing the complexity of this figure, however, we have now highlighted the values in pink boxes.

4) The rationale for choosing 5HT2A mutants is presented as having reduced protein preference in starved female flies, and increased lifespan under dietary choice. However, it is possible that the effects they observe in the present study are an aspect of the same observation (see point 1).

Contrary to our previously published work in which flies were purposefully starved, we find no evidence that flies fed a choice diet are starved or nutrient (protein) deprived: they spend most of their time on the yeast food (see response to point #1 above); they exhibit normal levels of egg production and TOR signaling; and they are of normal weight. It is possible that mechanisms of nutrient perception or that cell-nonautonomous effects of perceived protein reward underlie the mechanisms through which lifespan is affected, and if so, that would be a significant advance. Similarly, we document 5-HT2A-dependent metabolic changes in response to dietary choice, among which αKG seems to be causal for effects on lifespan. Whether neuronal and perception effects are common to both phenomena remain to be determined, but either way, in our opinion the data presented in this manuscript highlight new mechanisms of aging.

5) In Figure 3, it is important to show all the PCA data in graphic format, not just those that differed amongst the various conditions.

The clustering patterns of major PC combinations (PC1-PC6, which account for 85% (in head) or 87% (in body) of the total variance), as well as the proportion of variance for each PC, have been added as Figure 3—figure supplement 1 (for head data) and Figure 3—figure supplement 2 (for body data).

6) Many metabolites were altered in the different conditions (Figure 4). What is the rationale for choosing to focus on the TCA cycle intermediates? It appears that there are broader changes to amino acid metabolism. Also, fold changes are not mentioned and cutoffs not given.

We have added detailed statistics for each metabolite in Supplementary file 3. The full datasets and analyses can also be accessed from GitHub (github.com/ylyu-fly/Metabolomics-FlyChoiceDiet). As noted in the main text, we used P-values, not fold-changes as cut-offs:

“Criteria for inclusion in Group A consisted of a statistically significant effect of diet (PDiet < 0.05) and a non-significant diet x genotype interaction (PDiet * Genotype > 0.05).”

“We identified 18 out of 103 total metabolites in the head and 19 out of 125 total metabolites in the body that were present in this group (Figure 4B, ANOVA, PDiet * Genotype ≤ 0.05).”

We have also added text to the Results describing our rationale for focusing on TCA cycle.

Reviewer #2:

This is an interesting extension of previous work from the Pletcher lab, published in eLife (Ro et al., 2016). The concept of nutritional choice impacting organismal physiology and longevity is very interesting.

The challenge, of course, is to provide irrefutable evidence that it is the “choice” that causes the observed physiological changes. On the whole, the authors have done an admirable job to exclude potential confounding factors. I have, however, some remaining concerns/suggestions for alternative explanations.

I am happy to leave it to the authors as to whether to attempt to address these concerns with additional experiments and/or to discuss in the manuscript.

Figure 1:

The entire premise of the paper assumes that the observed effects are “largely independent of nutrients consumed”. Technically, the authors have done a very good job, when focusing on nutrient consumption.

However, there could be more to this than just food consumption.

Is it possible/likely that on the choice diet the flies are absorbing more yeast/protein? Isn't the Figure 1H data consistent with this?

i.e. When the macronutrients are provided separately, are there are differences in absorption?

In the fixed diet, sugar and yeast are always consumed together.

Is consuming yeast without sugar, even for short periods, detrimental to longevity because of “elevated protein uptake”?

My (limited) understanding of human nutrition is that it really matters which ingredients are paired together in terms of health benefits, e.g. pairing tomatoes with olive oil.

Another potential issue, that given the lab's history is particularly relevant, is the issue of olfaction.

Could the smell of yeast be more pronounced when provided separately? If so, could this be a confounding factor?

Two suggestions to investigate this hypothesis:

1) Assay fecundity, which is linked to protein intake in flies in choice vs. fixed diet

2) Assay TOR signaling which is also linked to protein intake

3) Examine olfactory mutants (smell blind)

If the authors could demonstrate no difference in these parameters, it would strengthen the claims of the paper.

We appreciate the insights and critiques provided by this reviewer, particularly those focused on protein sensing and absorption in the dietary choice effects, as we recognize that many uncharacterized features of diets may modulate aging. We have worked hard during difficult times in the lab to add data from new experiments that address these issues in the revised manuscript:

a) Specifically, three new experiments provide insight into the point about the possibility of diet-dependent protein absorption. Briefly, we measured fecundity and TOR activity (based on phosphorylation status of S6K) and neither measure indicated an effect of dietary choice. Furthermore, total protein in the fly was also not affected by protein choice (nor was weight), which also indicates that, given similar amounts of protein intake, protein absorption/utilization is not different between diets. The manuscript has been revised accordingly (Results).

b) This reviewer raised an interesting point that the choice effects could be the result of “pairing” different nutrients/foods. We find this (and other effects of patterning in feeding) to be very intriguing, and to our knowledge, it is largely unknown how flies adjust feeding behaviors in complex environments (but see Ahn et al., 2011). We believe that the choice diet paradigm (and modifications to it) provides an excellent opportunity to investigate such mechanisms in the future. For now, we are happy to include this possibility in the Discussion, as show below:

“Flies housed in a choice environment, for example, are allowed to interact with each macronutrient individually, and this behavior would be expected to result in the sequential activation of sugar and protein sensing pathways. […] The perceived flavor of each macronutrient may be different when yeast and sugar are consumed separately compared to when they are paired together on a fixed diet (Ahn et al., 2011), and that may also influence lifespan and metabolism (Ostojic et al., 2014).”

c) We agree that this would be a good experiment test the role of sensory perception in dietary choice effects. While the characterization of sensory requirements is complex and is well beyond the focus of a single manuscript (e.g., if you can’t smell or taste the food can you extract reward from it?), our current ongoing experiments clearly indicate that it is not simply a more odorant rich environment that is reducing lifespan upon dietary choice. This experiment is on-going, although the results are clear (Author response image 1). We will be willing to include the final data in a final version of the manuscript if the editors and reviewers feel it is an essential. Indeed, a more complete investigation of external and internal sensory mechanisms, together with patterned feeding, is a likely next step, although, we believe, outside the scope of the current manuscript.

Figure 2:

Figure 2A-E is w1118 really an appropriate control? Precise P element revertant?

We appreciate and recognize the importance of controlling for background effects. We chose in this instance to use extensive backcrossing of 5-HT2A mutants (ten generations; as described in the Materials and methods) together with background-independent RNAi. Precise excision (together with backcrossing) is also a reasonable approach, which we chose not to pursue. The primary reason for this is the potential for unintended genetic effects of excision and the time and effort required to achieve it. We felt the two manipulations described above provided strong inference.

Not surprisingly, different genotypes appear to have varying degrees of reduced longevity in response to the “choice food”. I.e. Canton-S in Figure 1, w1118 in Figure 2 and the RNAi strain in Figure 5. In this case (Figure 5), the effects of the choice diet appear to be modest.

We acknowledged the observations on the effects of genetic background, where we added a new paragraph in the Discussion on this topic.

Can the authors speculate as to why the 5-HT2A mutants are short-lived on the fixed diet?

A challenge here is that serotonin signaling has many pleotropic effects, although the extent to which some/all are 5-HT2A dependent is unclear. In Drosophila to name a few, 5-HT2A is involved in circadian rhythms (Nichols et al., 2006), walking (Howard et al., 2019), feeding (as shown in this paper, Gasque et al., 2013 and Ro et al., 2016), and sensory perception of dead conspecifics (Chakraborty et al., 2019). All of these effects could potentially contribute to its lifespan effect on a fixed diet. We have added some discussion of this to the revision.

“Mutant flies were modestly short-lived on the standard, fixed diet, which is likely due to the pleotropic effects of this receptor (Gasque et al., 2013; Chakraborty et al., 2019; Howard et al., 2019).”

The metabolic data presented (Figures 4 and 5) are interesting, but, there is little in terms of mechanistic understanding provided.

It is interesting that amino acid metabolism is upregulated in response to the choice diet. Again, as outlined above, couldn't this reflect elevated protein intake?

As related to our responses to the point #1, there is no evidence showing that flies on a choice diet have elevated protein intake. The upregulation of amino acid metabolism therefore seems to reflect differences in the way that a fixed amount of amino acids are metabolized (also see our response to reviewer #1, point #6).

The GDH RNAi experiment is very interesting, but, there appears to be little follow up and/or validation. Some suggestions to validate and/or strengthen the claims:

– Validate that the RNAi line works “as advertised”.

qPCR data reveal variable knockdown (up to 20% decrease) in GDH transcript amount, indicating that either partial knock-down is sufficient to influence lifespan or that the RNAi may function at post-transcriptional levels. We interpret this as implying that this particular pathway is subject to complex regulatory mechanisms (Smith et al., 2019). Indeed, our results simply provide one direction forward in identifying specific molecules that are responsible for the aging effect, which we believe is a significant advance. Given the difficulty of laboratory work, we would argue that detailed molecular analyses of multiple components of this pathway are beyond the scope of the current contribution.

– Do GDH RNAi flies also eat more under the choice diet? This would strengthen the claims.

We have provided new data on the food intake and protein absorption of GDH RNAi flies. In short, no, neither of these factors seems to contribute to the lifespan effects. Please see our revised text below:

GDH knockdown had no effect on nutrient consumption (Figure 5D) nor total protein abundance in the fly (Figure 5—figure supplement 2).”

– Does neuronal RNAi also alter the response?

We are interested in testing the tissue requirement of GDH knockdown, but respectively argue that such experiments are best addressed in a follow-up manuscript (see our second response to point 3, above). Nevertheless, the relevance of this point is now addressed in the revision:

“Which tissues/cells are critical for the diet-specific lifespan effects that induced by GDH knockdown, and whether such influences are due to changes in α-ketoglutarate per se or to alterations in the metabolic network (Hoffman et al., 2017) more broadly, remain to be determined.”

https://doi.org/10.7554/eLife.59399.sa2

Article and author information

Author details

  1. Yang Lyu

    Department of Molecular and Integrative Physiology and Geriatrics Center, Biomedical Sciences and Research Building, University of Michigan, Ann Arbor, United States
    Contribution
    Conceptualization, Investigation, Supervision, Formal analysis, Resources, Data curation, Software, Methodology, Project administration, Writing-original draft, Writing - review and editing, Funding acquisition
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4618-9043
  2. Kristina J Weaver

    Department of Molecular and Integrative Physiology and Geriatrics Center, Biomedical Sciences and Research Building, University of Michigan, Ann Arbor, United States
    Contribution
    Formal analysis, Investigation, Visualization, Writing - original draft, Funding acquisition
    Competing interests
    No competing interests declared
  3. Humza A Shaukat

    College of Literature, Science, and the Arts, University of Michigan, Ann Arbor, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  4. Marta L Plumoff

    College of Literature, Science, and the Arts, University of Michigan, Ann Arbor, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  5. Maria Tjilos

    College of Literature, Science, and the Arts, University of Michigan, Ann Arbor, United States
    Contribution
    Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
  6. Daniel EL Promislow

    1. Department of Lab Medicine & Pathology, University of Washington School of Medicine, Seattle, United States
    2. Department of Biology, University of Washington, Seattle, United States
    Contribution
    Resources, Software, Supervision, Funding acquisition, Methodology, Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  7. Scott D Pletcher

    Department of Molecular and Integrative Physiology and Geriatrics Center, Biomedical Sciences and Research Building, University of Michigan, Ann Arbor, United States
    Contribution
    Conceptualization, Investigation, Supervision, Formal analysis, Resources, Data curation, Software, Methodology, Project administration, Writing-original draft, Writing - review and editing, Funding acquisition
    For correspondence
    spletch@med.umich.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4812-3785

Funding

Burroughs Wellcome Fund (Collaborative Research Travel Grant BWF1017452)

  • Yang Lyu

National Science Foundation (Graduate Research Fellowship Program DGE 1256260)

  • Kristina J Weaver

National Institutes of Health (R01 AG049494)

  • Daniel EL Promislow

National Institutes of Health (R01 AG051649)

  • Scott D Pletcher

National Institutes of Health (P30 AG013280)

  • Daniel EL Promislow

National Institutes of Health (R01 AG030593)

  • Scott D Pletcher

National Institutes of Health (R01 AG063371)

  • Scott D Pletcher

Glenn Foundation for Medical Research (Paul F. Glenn Center for Aging Research)

  • Scott D Pletcher

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the editors and reviewers for insightful comments that have resulted in a considerably improved manuscript. We would like to acknowledge the members of the Pletcher laboratory for their comments on the experimental designs and analyses, members of the Promislow laboratory for their valuable opinions on the manuscript draft, the fly kitchen and undergraduate research assistants Kyrinna Wei and Justin Eagy of the Pletcher laboratory for experimental food preparation and fly husbandry, and Drs. Cheng Peng and Tong Shen for suggestions on the interpretation of the metabolomics results. This research was supported by the US National Institutes of Health, National Institute on Aging (R01 AG051649, R01 AG030593 to SDP; R01 AG063371 to SDP and DELP; and R01 AG049494 and P30 AG013280 to DELP), the Glenn Medical Foundation (to SDP), the Burroughs Wellcome Fund Collaborative Research Travel Grant (No. BWF1017452 to YL), and National Science Foundation Graduate Research Fellowship Program (No. DGE 1256260 to KJW).

Senior Editor

  1. Jessica K Tyler, Weill Cornell Medicine, United States

Reviewing Editor

  1. Meng C Wang, Baylor College of Medicine, HHMI, United States

Reviewer

  1. David Walker, UCLA, United States

Publication history

  1. Received: May 28, 2020
  2. Accepted: January 4, 2021
  3. Accepted Manuscript published: January 19, 2021 (version 1)
  4. Version of Record published: February 26, 2021 (version 2)

Copyright

© 2021, Lyu et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,308
    Page views
  • 253
    Downloads
  • 2
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

  1. Further reading

Further reading

    1. Chromosomes and Gene Expression
    2. Cell Biology
    Edited by Matt Kaeberlein et al.
    Collection

    eLife is pleased to present a Special Issue to highlight recent advances in the mechanistic understanding of aging and interventions that extend longevity.

    1. Genetics and Genomics
    2. Neuroscience
    Li Hou et al.
    Research Article Updated

    Long-term flight depends heavily on intensive energy metabolism in animals; however, the neuroendocrine mechanisms underlying efficient substrate utilization remain elusive. Here, we report that the adipokinetic hormone/corazonin-related peptide (ACP) can facilitate muscle lipid utilization in a famous long-term migratory flighting species, Locusta migratoria. By peptidomic analysis and RNAi screening, we identified brain-derived ACP as a key flight-related neuropeptide. ACP gene expression increased notably upon sustained flight. CRISPR/Cas9-mediated knockout of ACP gene and ACP receptor gene (ACPR) significantly abated prolonged flight of locusts. Transcriptomic and metabolomic analyses further revealed that genes and metabolites involved in fatty acid transport and oxidation were notably downregulated in the flight muscle of ACP mutants. Finally, we demonstrated that a fatty-acid-binding protein (FABP) mediated the effects of ACP in regulating muscle lipid metabolism during long-term flight in locusts. Our results elucidated a previously undescribed neuroendocrine mechanism underlying efficient energy utilization associated with long-term flight.