Dietary Macronutrient Imbalances Lead to Compensatory Changes in Peripheral Taste via Independent Signaling Pathways

Dietary sugar-to-protein ratio sways feeding preference for sugar

The experimental strategy we adopted to investigate how changes in macronutrient balance in the diet influence taste is summarized in Figure 1A. We prepared three different diets: a “balanced” diet containing glucose and yeast extract (which contains amino acids, peptides, nucleic acids, carbohydrates including trehalose, water-soluble vitamins and trace elements) in the same proportions as in the standard cornmeal-dextrose diet, a “sugar-reduced” diet containing no added glucose and subsequently enriched in yeast extract, and a “sugar-enriched” diet containing ∼1.26× the amount of glucose and depleted of yeast extract. The composition of the diets is listed in the methods, and was designed to maintain caloric content within 20% of that of the balanced diet. Adult flies raised on standard cornmeal-dextrose media were collected at 0–2 d after eclosion, transferred to vials containing one of the three diets, and tested for cellular and/or behavioral responses to selected tastants. Male and female flies were housed together, and we assumed that all female flies would be mated at the time of testing. Sexed flies were analyzed separately, since previous studies have found sex-specific differences in feeding preference for yeast or yeast extract (Ribeiro and Dickson, 2010; Ganguly et al., 2017).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Dietary sugar-protein imbalance leads to shifts in feeding preference. A, Schematic representing the experimental strategy to evaluate diet-induced alteration of taste sensitivity and preference. B, Preference of mated females (and males, see Extended Data Fig. 1-1) fed on indicated diet for 5 mm sucrose (blue dye) tested against 1% yeast extract (pink dye) in binary choice assays; n = 6 balanced (0 d), 6 sugar-reduced (0 d), 15 balanced (1 d), 13 sugar-reduced (1 d), 13 balanced (2 d), 8 sugar-reduced (2 d), 17 balanced (3 d), 15 sugar-reduced (3 d). Feeding participation of males and females prefed on the balanced or sugar-reduced diets is shown in Extended Data Figures 1-1, 1-2. C, Ingested volume (μl) of indicated diet normalized to body weight (mg) in mated females prefed with indicated diet for 1 d and then tested for 2-h consumption of sugar-reduced (n = 35, 18), balanced (n = 35, 19), or sugar-enriched (n = 31, 10) diets. Head and body weights of females prefed on the balanced or sugar-reduced diets for 1 d are shown in Extended Data Figure 1-3. D, Preference of mated females prefed on indicated diet for 5 mm sugar (blue dye) tested against 1% yeast extract (pink dye) in binary choice assays; n = 6 sucrose, 7 fructose, 10 glucose for both balanced and sugar-reduced. E, Preference of mated females (and males, see Extended Data Fig. 1-4) prefed on indicated diet for 5 mm sucrose (blue dye) tested against 1% yeast extract in binary choice assays; n = 6 balanced, 6 sugar-enriched (0 d), n = 9 balanced, 6 sugar-enriched (1 d), n = 13 balanced, 6 sugar-enriched (2 d), n = 17 balanced, 10 sugar-enriched (3 d), n = 7 balanced, 16 sugar-enriched (4 d), n = 6 balanced, 6 sugar-enriched (5 d), n = 7 balanced, 6 sugar-enriched (6 d), n = 10 balanced, 11 sugar-enriched (7 d). Feeding participation of males and females prefed on the balanced or sugar-enriched diets are shown in Extended Data Figures 1-4, 1-5. F, Ingested volume (μl) of indicated diet normalized to body weight (mg) in mated females prefed with indicated diet for 4 d and then tested for 2-h consumption of sugar-reduced (n = 34, 10), balanced (n = 28, 21), or sugar-enriched (n = 31, 21) diets. Head and body weights of females prefed on the balanced or sugar-enriched diets for 4 d are shown in Extended Data Figure 1-6. G, Preference of mated females prefed on indicated diet for 5 mm sugar (blue dye) tested against 1% yeast extract (pink dye) in binary choice assays; n = 6 sucrose, 6 fructose, 6 glucose for both balanced and sugar-enriched. Diet-induced changes in feeding preference are compared in starved and unstarved females in Extended Data Figure 1-7. For all graphs, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data in B, C, E, F were analyzed using two-way ANOVA with Sidak's post hoc multiple comparisons test. Data in D, G were analyzed with the Mann–Whitney test.

We first determined whether feeding on the test diets altered preference in binary choice feeding assays. Flies were tested daily for a period of 3 to 7 d after transferring to sugar-reduced or sugar-enriched diets, respectively. Age-matched flies fed on the balanced diet were tested in parallel. Preference for sugar was assessed in feeding choice assays with 5 mm sucrose and 1% yeast extract as the two stimuli. We found that female flies that were fed on sugar-reduced food exhibited higher preference for sucrose as compared with those fed on the balanced diet. A significant increase in sugar preference was observed after just 1 d on the sugar-reduced diet, and the elevated preference remained with continued feeding on the same diet (Fig. 1B, w¹¹¹⁸ balanced vs sugar-reduced diet for 0 d p = 0.9997, 1 d p < 0.0001, 2 d p < 0.0001, 3 d p < 0.0001, Sidak's post hoc multiple comparisons test; see also Table 3). Although the baseline sucrose preference is higher in males to begin with, a potentiation of sugar preference was observed after feeding on the sugar-reduced diet, which was significant after 3 d (Extended Data Fig. 1-1, w¹¹¹⁸ balanced vs sugar-reduced diet for 3 d p = 0.0012, Sidak's post hoc multiple comparisons test; see also Table 3). The dietary change did not affect participation of the flies in binary choice feeding assays, except for a small drop in males at 3 d after feeding on the sugar-reduced diet (Extended Data Figs. 1-1, 1-2, w¹¹¹⁸ balanced vs sugar-reduced diet for 3 d p < 0.0001, Sidak's post hoc multiple comparisons test; see also Table 3). To rule out the possibility that the sugar-reduced diet triggers a change in amount of food intake rather than a shift in food preference, we used short-term consumption assays to compare intake of the three test diets, presented individually, after prefeeding on either balanced or sugar-reduced diets. We found that flies fed on the sugar-reduced diet for a day consumed the same amounts of each of the three diets as those fed on the balanced diet (Fig. 1C, w¹¹¹⁸ balanced vs sugar-reduced pretreatment, p = 0.6303 for sugar-reduced consumption diet, p = 0.0792 balanced consumption diet, p = 0.9368 for sugar-enriched consumption diet; Sidak's post hoc multiple comparisons test; see also Table 3). Neither head weight nor body weight were affected on feeding on the sugar-reduced diet (Extended Data Fig. 1-3, p = 0.0833 for head, p = 0.6883 for body, Sidak's post hoc multiple comparisons test; see also Table 3). Notably, flies subjected to the sugar-reduced diet also showed increased preference for fructose and glucose in binary choice assays with yeast extract, indicating that relative feeding preference for sugar was broadly altered with this dietary regimen (Fig. 1D, w¹¹¹⁸ balanced vs sugar-reduced diet, p = 0.0022 sucrose, p = 0.0006 fructose, p = 0.0014 glucose; Mann–Whitney test).

Experiments with the sugar-enriched diet yielded complementary results, however the onset of change in preference was delayed in comparison with that elicited by the sugar-reduced diet. Females displayed a reduced feeding preference for sugar in binary choice assays with yeast extract only after 4 d on the sugar-enriched diet (Fig. 1E, w¹¹¹⁸ balanced vs sugar-enriched diet for 4 d p < 0.0001, 5 d p = 0.0003, 6 d p < 0.0001, 7 d p < 0.0001; Sidak's post hoc multiple comparisons test; see also Table 3), and males exhibited a shift toward yeast extract preference after 6 d (Extended Data Fig. 1-4, w¹¹¹⁸ balanced vs sugar-enriched diet for 6 d p = 0.025, 7 d p < 0.0001; Sidak's post hoc multiple comparisons test; see also Table 3). For both sexes, the altered preference persisted for the remainder of the 7-d test period, during which flies were maintained on the same diet. In no-choice consumption assays, the test flies exhibited no differences in intake of the balanced diet or of the sugar-reduced/yeast-enriched diet (Fig. 1F, w¹¹¹⁸ balanced vs sugar-enriched pretreatment, p = 0.3549 for sugar-reduced consumption diet, p = 0.135 balanced consumption diet; Sidak's post hoc multiple comparisons test; see also Table 3), consistent with the idea that the observed shift in behavioral preference is unlikely to arise solely from changes in food intake caused by dietary protein deprivation (Steck et al., 2018). Flies fed with the sugar-enriched diet exhibited no difference in participation in binary choice feeding assays compared with those fed on balanced diet (Extended Data Figs. 1-4, 1-5; Table 3). We did, however, observe reduced intake of the sugar-enriched diet (Fig. 1F, p < 0.0001 for sugar-enriched consumption diet; Sidak's post hoc multiple comparisons test; see also Table 3), indicative of sugar-enriched diet-dependent changes in food intake over time. This was reflected in the body weight, as flies fed on the sugar-enriched diet showed a reduction in body weight, but not head weight, as compared with those fed on the balanced diet (Extended Data Fig. 1-6, w¹¹¹⁸ balanced vs sugar-enriched diet, head weight p = 0.9662, body weight p = 0.0008; Sidak's post hoc multiple comparisons test; see also Table 3). As observed for experiments with the sugar-reduced diet, sugar preference was broadly altered with the sugar-enriched dietary treatment, with reduced preference for fructose and glucose in addition to sucrose (Fig. 1G, w¹¹¹⁸ balanced vs sugar-enriched diet, p = 0.0238 sucrose, p < 0.0001 fructose, p = 0.0001 glucose; Mann–Whitney test). With both imbalanced diets, shifts in feeding preference were independent of starved or fed state (Extended Data Fig. 1-7; Table 3). Combined, these results suggest that relative feeding preference for sugar shifts in a manner that compensates for the amount of sugar in the diet.

Dietary sugar-protein variation alters sweet taste

One possible mechanism that would account for behavioral adjustments in feeding preference is a change in sensitivity of sweet taste neurons. Indeed, previous studies have reported that flies given very sweet diets, containing either a high concentration of sucrose or an artificial sweetener sucralose, experienced changes in neuronal sensitivity to sugar (Wang et al., 2016; May et al., 2019). We therefore measured responses of labellar taste hairs to sucrose using extracellular tip recordings. Individual hairs were tested with a range of sucrose concentrations to make comparisons of sucrose sensitivity between flies subjected to the different diet treatments. Sucrose responses from labellar L-type sensilla of flies fed on sugar-reduced diet for 1 d were significantly higher at all tested concentrations than those of age-matched flies fed on balanced diet (Fig. 2A,B, w¹¹¹⁸ balanced vs sugar-reduced diet, p = 0.0282 for 10 mm sucrose, p < 0.0001 for 30 mm sucrose, p = 0.0001 for 100 mm sucrose; Sidak's post hoc multiple comparisons test; see also Table 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Dietary sugar imbalance alters sweet taste. A, Representative traces of the first 500-ms period of tip recordings from sensilla stimulated with the indicated concentrations of sucrose. Before testing, flies were prefed with indicated diet for 1 d. B, Neuronal responses in the first 500-ms period on stimulation with sucrose. Flies were prefed with the indicated diet for 1 d and tested without starvation (A, B) or after a 24-h period of wet starvation (Extended Data Fig. 2-1); n = 13 (balanced), n = 20 (sugar-reduced) sensilla from three to four flies. C, Representative traces of the first 500-ms period of tip recordings from sensilla stimulated with the indicated concentrations of sucrose. Before testing, flies were prefed with the indicated diet for 4 d. D, Neuronal responses in the first 500-ms period on stimulation with sucrose. Flies were prefed with the indicated diet for 4 d and tested without starvation (C, D) or after a 24-h period of wet starvation (Extended Data Fig. 2-2); n = 21 (balanced), n = 20 (sugar-enriched) sensilla from four flies. All recordings were taken from L-type labellar sensilla in mated females. Neuronal responses in B, D were compared using two-way ANOVA with Sidak's post hoc multiple comparisons test; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

In complementary experiments, flies were fed on the sugar-enriched diet for 4 d and tested in parallel with age-matched control flies fed on the balanced diet. In this case, we observed a reduction in responses of labellar sweet taste neurons to sucrose, with significant differences observed between responses to 10 and 30 mm sucrose, but not to 100 mm sucrose (Fig. 2C,D, w¹¹¹⁸ balanced vs sugar-enriched diet, p = 0.0093 for 10 mm sucrose, p < 0.0001 for 30 mm sucrose, p = 0.0.1411 for 100 mm sucrose; Sidak's post hoc multiple comparisons test; see also Table 3), indicating that response threshold but not peak response is altered. Overall, we found that taste sensitivity to sucrose was heightened by exposure to the sugar-reduced diet and dampened by the sugar-enriched diet. Since flies are starved for 24 h before binary choice assays, we verified that the changes in taste responses were associated with the specific diet rather than with starvation by recording sucrose responses from flies that were starved for 24 h after being prefed on the sugar-reduced or sugar-enriched diets (Extended Data Fig. 2-1, w¹¹¹⁸ balanced vs sugar-reduced diet, p = 0.0212 for 10 mm sucrose, p = 0.0069 for 30 mm sucrose, p = 0.5976 for 100 mm sucrose and Extended Data Fig. 2-2, w¹¹¹⁸ balanced vs sugar-reduced diet, p = 0.001 for 10 mm sucrose, p < 0.0001 for 30 mm sucrose, p = 0.1023 for 100 mm sucrose; Sidak's post hoc multiple comparisons test; see also Table 3). Overall, the diet-induced changes in taste sensitivity aligned with the observed shifts in behavioral preference in short-term choice assays, supporting the idea that behavioral modification in response to dietary imbalances occurs, at least in part, via alteration of taste sensitivity.

Diet-induced changes in sweet taste are persistent but reversible

Since modifications in taste sensitivity to sucrose were apparent only after a day or more following transfer of flies to test diets, we wanted to assess the extent to which the observed changes persisted on return to a balanced diet. To address this, we prefed flies on the sugar-reduced or sugar-enriched diet for 1 or 4 d, respectively, which were the time frames at which increases or decreases in sugar sensitivity were observed (Fig. 1B,E). Flies were then returned to the balanced diet and behavioral preference for sugar was tested daily with batches of flies. We found that the increase in sugar preference produced by the sugar-reduced diet persisted for at least a day following reintroduction of the balanced diet, returning by 2 d to levels observed for flies fed continuously on the balanced diet (Fig. 3A, w¹¹¹⁸ balanced → balanced vs sugar-reduced → balanced; 1 d p < 0.0001, 2 d p = 0.03; Sidak's post hoc multiple comparisons test; see also Table 3). Consistent with these observations, we found that firing rates of labellar sweet taste neurons remained elevated at the higher concentrations tested (30 and 100 mm) for 1 d after resuming the balanced diet as compared with those obtained from control flies (Fig. 3B,C, w¹¹¹⁸ balanced 1 d vs sugar-reduced 1 d, p = 0.0004 for 10 mm sucrose, p < 0.0001 for 30 mm sucrose, p < 0.0001 for 100 mm sucrose; w¹¹¹⁸ balanced 2 d vs sugar-reduced 1 d + balanced 1 d, p = 0.01 for 30 mm sucrose, p = 0.0007 for 100 mm sucrose; w¹¹¹⁸ balanced 3 d vs sugar-reduced 1 d + balanced 2 d, p = 0.0264 for 100 mm sucrose; Tukey's post hoc multiple comparisons test; see also Table 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Dietary sugar-protein imbalance-induced alterations in sweet taste and behavior are reversible. A, Preference of mated females for 5 mm sucrose (blue dye) tested against 1% yeast extract (pink dye) in binary choice assays. Flies were treated to indicated dietary regimens and tested daily; n = 6 balanced, 6 sugar-reduced (0 d); n = 15 balanced, 13 sugar-reduced (1 d); n = 17 balanced, 6 sugar-reduced (2 d); n = 6 balanced, 6 sugar-reduced (3 d). B, Ratio of mean sweet neuron response in flies fed on sugar-reduced diet (test) to that from flies fed on balanced diet (control). Flies were treated to different dietary regimens as in A and tested on indicated days with different concentrations of sucrose; n = 8–25 sensilla from two to five flies. The dotted line indicates a ratio of 1, i.e., equal taste sensitivity to sucrose in test and control flies. Tip recordings were taken from L-type labellar hairs in mated females. C, Mean neuronal responses used to generate the spike ratios shown in B. D, Preference of mated females for 5 mm sucrose (blue dye) tested against 1% yeast extract (pink dye) in binary choice assays. Flies were treated to indicated dietary regimens and tested daily; n = 9 balanced, 6 sugar-enriched (1 d); n = 13 balanced, 6 sugar-enriched (2 d); n = 17 balanced, 10 sugar-enriched (3 d); n = 7 balanced, 16 sugar-enriched (4 d); n = 7 balanced, 6 sugar-enriched (5 d); n = 6 balanced, 7 sugar-enriched (6 d); n = 6 balanced, 7 sugar-enriched (7 d); n = 6 balanced, 6 sugar-enriched (8 d). E, Ratio of mean sweet neuron response in flies fed on sugar-enriched diet (test) to that from flies fed on balanced diet (control). Flies were treated to different dietary regimens as in C and tested on indicated days with different concentrations of sucrose; n = 15–29 sensilla from three to six flies. The dotted line indicates a ratio of 1, i.e., equal taste sensitivity to sucrose in test and control flies. F, Mean neuronal responses used to generate spike ratios shown in E. Data in A, D were analyzed using two-way ANOVA with Sidak's post hoc multiple comparisons test; data in C, F were analyzed using two-way ANOVA with Tukey's post hoc multiple comparisons test; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

In similar experiments with flies fed on the sugar-enriched diet, we found that feeding preference for sucrose bounced back quickly on resuming the balanced diet (Fig. 3D, w¹¹¹⁸ balanced → balanced vs sugar-enriched → balanced; 4 d p < 0.0001, 5 d p = 0.9312; Sidak's post hoc multiple comparisons test; see also Table 3). However, sucrose response in labellar sweet taste neurons remained lower than that of flies fed continuously on the balanced diet, rising to comparable levels only after 2 d of recovery on the balanced diet (Fig. 3E,F, w¹¹¹⁸ balanced 4 d vs sugar-enriched 4 d, 10 mm sucrose p = 0.0365, 30 mm sucrose p < 0.0001; w¹¹¹⁸ balanced 5 d vs sugar-enriched 4 d + balanced 1 d, p = 0.0266 for 10 mm sucrose, p < 0.0001 for 30 mm sucrose, p < 0.0001 for 100 mm sucrose; Tukey's post hoc multiple comparisons test; see also Table 3). Together, the results suggest that dietary sugar-to-protein ratio alters sweet taste in a manner that can persist for some time after recommencement of the balanced diet. This is in agreement with a previous study, which reported that the increased preference for yeast in yeast-deprived flies disappears once the flies are reintroduced to a more complete diet (Ribeiro and Dickson, 2010; Vargas et al., 2010; Ganguly et al., 2017). Moreover, in both instances of recovery we observed a “rebound” effect of overcorrected neuronal firing rate in response to sucrose, at some part of the concentration range tested. Although analogous effects were not observed in binary choice assays, it remains possible that behavioral rebounds might be observed via other assays.

Dietary sugar-to-protein ratio alters feeding preference for amino acids

Since the proportion of protein also varies in the sugar-reduced and sugar-enriched diets we tested whether these diets affect feeding preference for amino acids. Flies were fed on sugar-reduced (protein-enriched) diet for 1 d and subsequently evaluated for feeding preference for an appetitive mixture of three amino acids (serine, phenylalanine, threonine) in binary choice assays (Ganguly et al., 2017). To promote feeding, 2 mm sucrose was added to both tastant alternatives, water and amino acid mixture, respectively. We found that flies fed on the sugar-reduced/protein-enriched diet had a lower preference for amino acids in comparison with those fed on the balanced diet (Fig. 4A, w¹¹¹⁸ balanced vs sugar-reduced, 1 d, p = 0.0043, Mann–Whitney test), consistent with the idea that the sugar-protein imbalance affects taste sensitivity to sugar as well as to amino acids in opposing ways. The apparent aversion for the sucrose-amino acid mixture was somewhat surprising. Previous studies have found that certain amino acids, including phenylalanine, can elicit firing from S-type hairs that house bitter taste neurons (Park and Carlson, 2018), which may result in feeding avoidance when combined with a reduction in appetitive value.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Dietary sugar-protein imbalance alters feeding preference for amino acids. A, Preference of mated females prefed on indicated diet for 1 d for 2 mm sucrose mixed with 25 mm each of serine, phenylalanine, and threonine (pink dye, sucrose + 3AA) tested against 2 mm sucrose alone (blue dye) in binary choice assays; n = 5 (balanced), n = 6 (sugar-reduced). B, Preference of mated females fed on indicated diet for 2 mm sucrose mixed with 25 mm each of serine, phenylalanine, and threonine (pink dye, sucrose + 3AA) tested against 2 mm sucrose alone (blue dye) in binary choice assays. Flies were tested daily on start of the indicated diet; n = 6 balanced (0–3 d), 10 (4 d), 6 (5–7 d); n = 6 sugar-enriched (0–3 d), 7 (4 d), 6 (5–7 d). C, Preference of mated females for 2 mM sucrose mixed with 25 mM each of serine, phenylalanine, and threonine (pink dye, sucrose + 3AA) tested against 2 mM sucrose alone (blue dye) in binary choice assays. Flies were treated to indicated dietary regimens and tested daily; n = 6 for each condition. D, Preference of mated females of control (w¹¹¹⁸) and ΔIr76b (Ir76b¹) genotypes prefed on the indicated diet for 4 d for 1% yeast extract (pink dye) tested against 5 mm sucrose (blue dye) in binary choice assays; n = 7 balanced, 7 sugar-enriched (control), n = 10 balanced, 9 sugar-enriched (ΔIr76b). E, Representative traces of the first 500-ms period of tip recordings from sensilla stimulated with a 3AA mixture of serine (100 mm), threonine (100 mm), and phenylalanine (25 mm) or 100 mm sucrose (left). Before testing, flies were prefed with the indicated diet for 4 d. Neuronal responses in the first 500-ms period on stimulation with indicated tastant or 30 mm tricholine citrate electrolyte (TCC; right). Flies were prefed with the indicated diet for 4 d; n = 11 (balanced), n = 12 (sugar-enriched) from five to six flies. For all graphs, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Statistical comparisons were made using the Mann–Whitney test (A, D) or two-way ANOVA with Sidak's post hoc multiple comparisons test (B, C, E).

Likewise, flies fed on sugar-enriched/protein-depleted diet were tested daily in choice assays and showed an enhancement of amino acid feeding preference at 4 d after starting the test diet, they preferred a mixture of amino acids and sucrose over sucrose alone as compared with flies fed on the balanced diet (Fig. 4B, w¹¹¹⁸ balanced vs sugar-enriched, 4 d p = 0.0003, 5 d p < 0.0001, 6 d p < 0.0001, 7 d p < 0.0001; Sidak's post hoc multiple comparisons test; see also Table 3). In fact, the increased amino acid preference persisted for a day after flies were returned to a balanced diet (Fig. 4C, w¹¹¹⁸ balanced → balanced vs sugar-enriched → balanced, 4 d p = 0.0018, 5 d p = 0.0283; Sidak's post hoc multiple comparisons test; see also Table 3). Moreover, flies fed with sugar-enriched/protein-depleted diet showed an increased preference for whole yeast extract, which was dependent on the function of Ir76b (Fig. 4D, balanced vs sugar-enriched diet w¹¹¹⁸ p = 0.0017, Ir76b¹ p = 0.0851, Mann–Whitney test), a receptor required for feeding behavior to amino acids as well as other components in yeast (Croset et al., 2016; Hussain et al., 2016; Ganguly et al., 2017; Steck et al., 2018). Taken together, the results suggest that behavioral shifts to accommodate changes in protein hunger are caused, at least in part, by altered feeding preference for amino acids.

A previous study showed that proboscis extension in response to stimulation of labellar hairs with amino acids was increased with dietary protein deprivation (Toshima and Tanimura, 2012). We therefore wished to test whether labellar taste responses to amino acids, which are ordinarily weak (Dahanukar et al., 2007; Ganguly et al., 2017; Park and Carlson, 2018), are strengthened on prefeeding with the sugar-enriched/protein-depleted diet. As reported previously, we observed no response to a mixture of three amino acids (100 mm serine, 100 mm threonine, and 25 mm phenylalanine) in L-type sensilla of flies fed on the balanced diet (Fig. 4E). A small but significant increase in the response was observed when flies were fed on the sugar-enriched/protein-depleted diet (Fig. 4E, w¹¹¹⁸ balanced vs sugar-enriched 4 d, 100 mm sucrose p = 0.1164, 3AA mixture p = 0.0215; Sidak's post hoc multiple comparisons test; see also Table 3). Thus, the sugar-enriched/protein-depleted diet appears to have opposing effects on taste sensitivity of sucrose and amino acids that correlate with reduced feeding preference for the former and increased feeding preference for the latter.

Increased sweet taste sensitivity following dietary sugar reduction depends on Dop2R

To begin to understand the molecular mechanisms underlying diet-induced alterations in taste sensitivity, we first focused on candidate genes that are known to be involved in modulating feeding behavior responses on the basis of internal nutrient state. Many recent studies have identified such functions for the dopamine pathway (Okamoto and Nishimura, 2015; Post and Tatar, 2016), and a requirement for Dop2R has been uncovered in responses to starvation and protein hunger (Liu et al., 2017; Marella et al., 2012). Since the sugar-reduced/protein-enriched diet altered both sweet taste and amino acid preference, we sought to test whether one or both of these changes were mediated via Dop2R. With qRT-PCR analysis, the difference in Dop2R expression between flies fed on the balanced or test diet for 1 d was not statistically significant, but there was a trend toward lower Dop2R expression in the brains of flies prefed with the sugar-reduced diet (Extended Data Fig. 5-1, balanced vs sugar-reduced Dop2R expression p = 0.324177, independent sample t test). We therefore knocked down Dop2R pan-neuronally using the elav-GAL4 driver and two independent transgenic UAS-Dop2R RNAi lines (Extended Data Fig. 5-2). UAS-Dcr2 was also included in the genetic background. Flies were exposed to the sugar-reduced/protein-enriched diet as described above and tested in binary choice assays using sucrose and yeast extract as the two tastants. The baseline preference for sucrose was different in the two elav > Dop2R RNAi lines. elav> Dop2R RNAi1 showed a similar baseline sucrose preference as the GAL4 control, whereas elav>Dop2R RNAi2 exhibited an enhanced baseline sucrose preference (Table 3). The elav > Dop2R RNAi1 line had lost any diet-induced behavioral shift that was observed in control elav-GAL4 flies tested in parallel; this was also the case for elav > Dop2R RNAi2, but it is possible that the high baseline preference of these flies does not leave significant room for enhancement (Fig. 5A, balanced sugar reduced diet for 1 d, elav-GAL4 ctrl p = 0.0008, elav>Dop2R RNAi1 p = 0.3067, elav>Dop2R RNAi2 p = 0.7995; Sidak's post hoc multiple comparisons test; see also Table 3). We therefore selected Dop2R RNAi1 for additional experiments.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Dop2R is required for sugar-deprivation induced rise in sweet taste and preference. A, Preference of mated females prefed on indicated diet for 1 d and tested in binary choice assays with 5 mm sucrose (blue dye) against 1% yeast extract (pink dye). Genotypes were as follows: elav-Gal4 (n = 6 balanced, 7 sugar-reduced), elav-Gal4>UAS-Dop2R RNAi1 (elav>D2R RNAi1, n = 9 balanced, 9 sugar-reduced), and elav-Gal4>UAS-Dop2R RNAi2 (elav>D2R RNAi2, n = 6 balanced, 6 sugar-reduced). Levels of Dop2R expression were measured by reverse transcriptase-PCR in wild-type females fed on balanced or sugar-reduced diet for 1 d (see Extended Data Fig. 5-1) and in elav-Gal4>UAS-Dop2R RNAi1 flies (see Extended Data Fig. 5-2). B, Feeding preference and participation (see Extended Data Fig. 5-3) of mated females prefed on indicated diet for 1 d and tested in binary choice assays with 5 mm sucrose (blue dye) against 1% yeast extract (pink dye). Genotypes were as follows: w¹¹¹⁸ (control, n = 6 balanced, 6 sugar-reduced), BDSC#52025 (Dop2R¹, n = 6 balanced, 6 sugar-reduced) and BDSC#52517 (Dop2R², n = 10 balanced, 10 sugar-reduced). The observed shift in food choice differs from a starvation-induced effect (see Extended Data Figs. 5-4, 5-5). C, Preference of mated females prefed on indicated diet for 4 d and tested in binary choice assays with 5 mm sucrose (blue dye) against 1% yeast extract (pink dye). Genotypes were w¹¹¹⁸ (control, n = 5 balanced, 5 sugar-enriched), BDSC#52025 (Dop2R¹, n = 6 balanced, 6 sugar-enriched), and BDSC#52517 (Dop2R², n = 6 balanced, 6 sugar-enriched). D, Preference of mated females prefed on indicated diet for 1 d and tested in binary choice assays with 2 mm sucrose mixed with 25 mm each of serine, phenylalanine, and threonine (sucrose + 3AA, pink dye) against 2 mm sucrose alone (blue dye). Genotypes were w¹¹¹⁸ (control, n = 6 balanced, 6 sugar-reduced), BDSC#52025 (Dop2R¹, n = 9 balanced, 9 sugar-reduced), and BDSC#52517 (Dop2R², n = 6 balanced, 6 sugar-reduced). E, Representative traces obtained from Dop2R¹ mutant females (left) and ratio of mean sweet neuron response (right) of flies fed on sugar-reduced diet (test) to that from flies fed on balanced diet (control) for 1 d. Mean neuronal responses are shown in Extended Data Figure 5-6. Recordings were taken from labellar L-type sensilla of five to six flies for each genotype. Genotypes tested were w¹¹¹⁸ (control, n = 25–29) and BDSC#52025 (Dop2R¹, n = 22–23). The dotted line indicates a ratio of 1, i.e., equal taste sensitivity to sucrose in test and control flies. Data in A–D were analyzed using two-way ANOVA with Sidak's post hoc multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Expression of Dop2R RNAi in Gad1-GAL4 neurons but not in Gr64f-GAL4 or other classes of neurons phenocopied the Dop2R mutant (see Extended Data Figs. 5-7, 5-8, 5-9, 5-10, 5-11).

To validate the observations with Dop2R RNAi, we next tested two different Dop2R mutants using the same experimental treatment and assays. Both mutants exhibited baseline sucrose preference similar to that of controls (Table 3). Consistent with the results of knock-down analysis, both failed to show increases in sucrose feeding preference on dietary sugar reduction (Fig. 5B, balanced vs sugar-reduced diet, w¹¹¹⁸ p = 0.0075, Dop2R¹ p = 0.9423, Dop2R² p = 0.3635; Sidak's post hoc multiple comparisons test; see also Table 3). Loss of Dop2R did not affect participation of the flies in binary choice assays (Extended Data Fig. 5-3, balanced vs sugar-reduced diet, w¹¹¹⁸ p > 0.9999, Dop2R¹ p > 0.9999, Dop2R² p = 0.6014; Sidak's post hoc multiple comparisons test; see also Table 3). However, Dop2R mutants retained the ability to adjust feeding preference in the appropriate direction on exposure to the sugar-enriched/protein-depleted diet (Fig. 5C, balanced vs sugar-enriched, w¹¹¹⁸ p = 0.0019, Dop2R¹ p < 0.0001, Dop2R² p < 0.0001; Sidak's post hoc multiple comparisons test; see also Table 3), indicating a role for Dop2R in response to a specific dietary imbalance.

Importantly, since Dop2R has been found to be involved in response to starvation (Inagaki et al., 2012; Marella et al., 2012), we considered the possibility that flies feeding on the sugar-reduced diet were being starved. To probe this, we first determined whether flies were in fact consuming the sugar-reduced/protein-enriched food by including red dye in the diet. Flies were examined 1 d following transfer to the diet and all flies showed red dye in the abdomen (data not shown). Second, we compared starvation resistance of flies that had been subjected to 1 d of the sugar-reduced/protein-enriched diet versus water alone. No mortality was observed in the former group until ∼30–35 h of starvation. By contrast, individuals of the latter group began dying at ∼10–15 h of starvation (Extended Data Fig. 5-4, balanced vs sugar-reduced pretreatment p = 0.4661, Log-rank Mantel–Cox test; see also Table 3). Next, we compared survival rates on water (wet starvation) and the sugar-reduced/protein-enriched diet. While all the wet-starved flies succumbed in 4 d or less, the median survival time was ∼16 d for those fed on the sugar-reduced/protein-enriched diet (Extended Data Fig. 5-5). Together, these results support the idea that physiological outcomes of feeding on the sugar-reduced/protein-enriched diet are somewhat different from those of acute starvation. Thus, Dop2R-mediated changes in behavioral responses to sugar occur not only on starvation but also on exposure to a calorie-rich diet that is a poor source of sugar. Interestingly, the Dop2R mutants failed to lower preference for amino acids on feeding on the sugar-reduced/protein-enriched diet (Fig. 5D, balanced vs sugar-reduced diet, w¹¹¹⁸ p = 0.0002, Dop2R¹ p = 0.0993, Dop2R² p = 0.3417; Sidak's post hoc multiple comparisons test; see also Table 3).

We next tested whether Dop2R is required for diet-induced alteration of sweet taste. We selected the Dop2R¹ mutant for electrophysiological analysis and compared responses of labellar sweet neurons in flies fed with the balanced or sugar-reduced diet for 1 d. We found that the diet-induced enhancement of sucrose response was reduced in Dop2R mutants as compared with control flies; this effect was more pronounced for the lower concentrations that were tested (10 and 30 mm; Fig. 5E; Extended Data Fig. 5-6, w¹¹¹⁸ balanced vs sugar-reduced, 10 mm sucrose p = 0.0017, 30 mm sucrose p < 0.0001, 100 mm sucrose p < 0.0001, Dop2R¹ balanced vs Dop2R¹ sugar-reduced, 10 mm sucrose p > 0.9999, 30 mm sucrose p = 0.0244, 100 mm sucrose p = 0.0002; Sidak's post hoc multiple comparisons test; see also Table 3). Overall, these results suggest that Dop2R plays a role in increasing sweet taste sensitivity on restriction of dietary sugar.

To identify the class of neurons in which Dop2R function is necessary for altering sweet taste following dietary sugar depletion, we expressed Dop2R RNAi in different subsets of neurons and compared feeding preferences of the resulting flies fed on balanced or sugar-reduced diets. Interestingly, expression of Dop2R RNAi via Gad1-GAL4, which labels GABAergic neurons, abolished the increase in sugar feeding preference produced by the sugar-reduced diet (Extended Data Fig. 5-7, balanced vs sugar-reduced diet, Gad1-GAL4>Dop2R RNAi p = 0.7004; Sidak's post hoc multiple comparisons test; see also Table 3). Both GAL4 and UAS controls, which were treated and tested in parallel, displayed diet-induced alterations in sugar preference as expected (Extended Data Fig. 5-7; Table 3). We observed Gad1-GAL4 to be widely expressed in the fly brain with particularly strong expression in the optic lobes and also in the protocerebral bridge (Extended Data Fig. 5-8). We did not find a significant reduction in Dop2R transcript levels in Gad1 > Dop2R RNAi flies using qRT-PCR analysis (Extended Data Fig. 5-9, Gad1-Gal4 vs Gad1 > Dop2R RNAi p = 0.6521, independent sample t test). One possible explanation is that there is only partial overlap between Dop2R and Gad1-GAL4 neurons, which obscures the effects of knock-down in the selected subset of cells.

Flies in which Dop2R RNAi was expressed in primary sweet taste neurons using Gr64f-GAL4 retained the ability to increase sugar preference on dietary sugar reduction (Extended Data Fig. 5-10, balanced vs sugar-reduced, Gr64f-GAL4 p < 0.0001, UAS-D2R RNAi1 p = 0.0057, Gr64f-GAL4>D2R RNAi1 p < 0.0001; Sidak's post hoc multiple comparisons test; see also Table 3). Knock-down of Dop2R in octopaminergic (Tdc2-GAL4), dopaminergic (Ple-GAL4), serotonergic (TRH-GAL4), or glutamatergic (v-Glut-GAL4) neurons also had little appreciable consequence on this phenotype (Extended Data Fig. 5-10, balanced vs sugar-reduced, Tdc2-GAL4>Dop2R RNAi p = 0.0128, Tdc2-GAL4 p = 0.0001, Ple>Dop2R RNAi p = 0.0026, Ple-GAL4 p < 0.0001, Trh>Dop2R RNAi p = 0.0002, Trh-GAL4 p = 0.0219, v-Glut>Dop2R RNAi p < 0.0001, v-Glut-GAL4 p = 0.0276; Sidak's post hoc multiple comparisons test; see also Table 3), although we cannot fairly evaluate partial contributions with these assay conditions. Consistent with the results of behavioral experiments, the differences in sweet taste neuron response to sucrose between flies fed on balanced versus sugar-reduced diet were greatly diminished or abolished in Gad1 > Dop2R RNAi flies, as compared with UAS-Dop2R RNAi controls tested in parallel (Extended Data Fig. 5-11, UAS-D2R RNAi balanced vs sugar-reduced, 10 mm sucrose p = 0.0027, 30 mm sucrose p = 0.0164, 100 mm sucrose p = 0.0004; Gad1-GAL4>UAS-D2R RNAi balanced versus sugar-reduced, 10 mm sucrose p = 0.0993, 30 mm sucrose p = 0.2012, 100 mm sucrose p = 0.1977; Sidak's post hoc multiple comparisons test; see also Table 3). Since the Gad1-GAL4 driver labels numerous neurons in the brain, the precise subsets of neurons in which Dop2R is required remain to be identified, but our results raise the possibility that GABAergic neurons in the brain are involved.

Transcriptome analysis identifies diet-induced changes in gene expression in taste tissues and the brain

To understand the molecular changes that accompany the observed physiological and behavioral shifts, we performed RNA-seq from various tissues and compared flies fed on the sugar-enriched/protein-depleted diet to flies fed on the balanced diet for 4 d. We analyzed the brain transcriptome, as well as that of tissues housing the majority of peripheral taste neurons, the proboscis and legs. In the brain, we found just over 200 genes that were differentially expressed, 120 of which were higher in flies exposed to the sugar-enriched diet and 93 that were lower (Fig. 6A, left). In the peripheral tissues, however, we found a disproportionately larger number of genes that showed reduced expression in flies given the sugar-enriched diet. In the proboscis, only 31 genes were significantly upregulated with feeding on the sugar-enriched diet, whereas 430 were downregulated (Fig. 6A, middle). Similarly, transcript levels of 145 genes increased, while those of 664 genes were significantly lower in the legs of flies fed on the sugar-enriched diet (Fig. 6A, right). Comparison of the DEGs revealed little overlap in upregulated genes between the three tissues (Fig. 6B, left). By contrast, we found significant overlap in the downregulated genes, 18 of which were decreased in all three tissues (Fig. 6B, right, C; Table 4). Among the commonly downregulated genes is Fit, which encodes the female-specific independent of transformer (FIT) peptide. Fit expression was previously shown to increase with protein intake in female flies, and it functions as a hormone to signal protein satiety (Sun et al., 2017). Consistent with these findings, we found a decrease in Fit expression in the brain, proboscis, and legs in female flies fed on sugar-enriched/yeast-depleted diet (Table 4). Interestingly, we also found that the expression of many genes encoding yolk proteins and serum proteins was reduced across all three tissues, suggesting that this dietary imbalance has profound effects on fat body gene expression in females (Table 4). The most substantial overlap, however, occurred among genes downregulated in the proboscis and legs, suggesting that similar molecular changes occur in these peripheral taste tissues (Fig. 6B,C). Gene Ontology (GO)-enrichment analysis for the genes altered in the brains of flies fed on the sugar-enriched diet revealed an enrichment for “G-protein-coupled photoreceptor activity [GO:0008020]” (Fig. 6D, top). This group includes the seven rhodopsin family genes found in Drosophila, typically expressed in the retina at high levels and markedly reduced in the brain (Senthilan et al., 2019) We found that four rhodopsin genes were significantly increased, between 2- and 3-fold, following feeding on sugar-enriched diet. In addition, two molecular function terms, “hormone activity [GO:0005179]” and “chitin binding [GO:0008061]” were enriched in genes that were downregulated in the brains of flies fed on sugar-enriched diet (Fig. 6D, bottom). Given the importance of hormones in regulating feeding behavior, we examined the expression of these hormone genes in peripheral taste tissues. We found that neuropeptide genes amn and CCAP were altered in the proboscis and leg, whereas Nplp3 was reduced in all three tissues (Fig. 6E). Similarly, two of the eight insulin-like peptide genes, dilp3 and dilp5, were downregulated in the brain and proboscis of flies fed on sugar-enriched diet; expression in the legs was also reduced for these genes, although they did not meet our false discovery rate (FDR) cutoff (Fig. 6E). Since dilp2, dilp3, and dilp5 have been reported to be expressed only in the brain (Brogiolo et al., 2001; Ikeya et al., 2002; Okamoto and Nishimura, 2015), detection of reads in other tissues, albeit at very low levels, was a little surprising. The dilp5 transcriptome findings were supported by qRT-PCR analysis, which showed a significant reduction in expression in the brains of flies fed on the sugar-enriched diet for 4 d (Extended Data Fig. 6-1). qRT-PCR also showed a trend toward lower expression of dilp3 and dilp5 in the proboscis, although the differences were not significant (Extended Data Fig. 6-1, balanced vs sugar-enriched diet 4 d, proboscis, dilp3 p = 0.753426, dilp5 p = 0.680805, independent samples t test). Overall, our results indicate profound effects of dietary sugar-enrichment on gene expression in the brain and taste organs, with changes in transcript abundance for a variety of genes. Moreover, mechanisms of diet-induced reduction in gene expression appear to have many common targets in the three tissues that we examined.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Gene expression analyses reveal a downregulation of insulin-like peptides (dILPs) on feeding on the sugar-enriched diet. A, Plots highlighting upregulated and downregulated genes in the brain, proboscis, and legs. Red and blue dots represent upregulated genes [Log₂fold change (LFC) > 1, FDR < 0.05] and downregulated genes (LFC < −1, FDR < 0.05), respectively. B, Overlap of upregulated (left) and downregulated (right) genes in each of the three tissues. C, Significance of overlap of indicated gene sets (p values indicated in boxes; colors denote odds ratio from Fisher's exact test). D, Fold-enrichment for molecular function GO terms in genes upregulated and downregulated in the brain (p < 0.05). E, Heatmap depicting expression of brain-regulated genes associated with the “hormone activity” GO terms across all experiments. Each row represents the expression of one gene, normalized across samples (red = high expression, blue = low expression). All comparisons are relative to the balanced diet. Diet-induced changes in expression of dilp3 and dilp5 were tested by qRT-PCR (see Extended Data Fig. 6-1).

Dilp5 is required in IPCs for diet-induced dampening of sweet taste

We focused on dilp5, whose transcripts were downregulated in the brain after flies were transferred to the sugar-enriched diet (Fig. 6E; Extended Data Fig. 6-1, balanced vs sugar-enriched diet 4 d, brain, dilp3 p = 0.680683, dilp5 p = 0.009507, independent samples t test). These findings are consistent with previous studies that report changes in dilp5 expression with altered protein or amino acid concentration in food (Manière et al., 2016). Incidentally Dilp2, Dilp3, and Dilp5 are secreted from the same group of cells, namely the IPCs in the brain. Moreover, a recent study has shown that IPCs can directly sense the amino acid leucine, and secrete Dilp2 and Dilp5 in response (Manière et al., 2016).

Based on our transcriptome analysis, we hypothesized that mutant flies lacking dilp5 that are fed on the balanced diet would behave as if they were fed on the sugar-enriched diet and exhibit reduced feeding preference for sugar as compared with control flies. Indeed, this is what we observed for dilp5 mutants, but not for dilp2 or dilp3 mutants (Fig. 7A; Extended Data Fig. 7-1, balanced vs sugar-enriched diet for 4 d w¹¹¹⁸ p < 0.0001, dilp2 p < 0.0001, dilp3 p = 0.0035, dilp5 p = 0.4748; Sidak's post hoc multiple comparisons test; see also Table 3). We note that there were some apparent differences in baseline sucrose preference between the different genotypes, and in particular, dilp5 mutants had a lower mean preference for sucrose as compared with control flies or dilp2 and dilp3 mutants, whose behavior was identical to control flies. However, the differences in baseline preference across the genotypes were not statistically significant (Table 3). Moreover, dilp5 mutants participated to the same extent as controls in binary choice assays (Extended Data Fig. 7-2, participation in balanced vs sugar-enriched, w¹¹¹⁸ p > 0.9999, dilp2 p = 0.973, dilp3 p = 0.9983, dilp5 p = 0.9848; Sidak's post hoc multiple comparisons test; see also Table 3), but were not able to significantly reduce sucrose feeding preference in response to the sugar-enriched diet; dilp2 and dilp3 mutants, however, retained the ability to do so (Fig. 7A). Interestingly, we found no differences in behavioral preference for sucrose between control and dilp5 mutants following exposure to the sugar-reduced diet, indicating a requirement for dilp5 only in adjusting behavioral responses to dietary sugar enrichment but not to sugar depletion (Fig. 7B, balanced vs sugar-reduced, w¹¹¹⁸ p < 0.0001, dilp5 = 0.0008; Sidak's post hoc multiple comparisons test; see also Table 3). Consistent with these results, dilp5 mutants did not exhibit a diminished neuronal response to sugar after feeding on the sugar-enriched diet, as compared with wild-type flies that were co-tested (Fig. 7C,D; Extended Data Fig. 7-3, w¹¹¹⁸ balanced vs w¹¹¹⁸ sugar-enriched 4 d, 10 mm sucrose p = 0.03, 30 mm sucrose p < 0.0001, 100 mm sucrose p = 0.0653; dilp5 balanced vs dilp5 sugar-enriched, 10 mm sucrose p = 0.0174, 30 mm sucrose p = 0.9728, 100 mm sucrose p = 0.0006; Sidak's post hoc multiple comparisons test; see also Table 3). On the contrary, at least for some concentrations there was a slight increase in neuronal response to sucrose in the dilp5 mutants (Fig. 7D; Extended Data Fig. 7-3; Table 3). Interestingly, baseline levels of sweet neuron response and sugar preference of dilp5 mutants fed on balanced diet were lower than that of control flies (Fig. 7A,D; Extended Data Fig. 7-3; Table 3), suggesting that an optimum titer of dilp5 may be required to maintain sugar preference in flies.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

dilp5 is required for dietary sugar-enrichment-induced decline in sweet taste and preference. A, Preference of mated females prefed on indicated diet for 4 d and tested in binary choice assays with 5 mm sucrose (blue dye) against 1% yeast extract (pink dye). Genotypes were w¹¹¹⁸ (control, n = 8 balanced, 8 sugar-enriched), dilp2 (n = 7 balanced, 8 sugar-enriched), dilp3 (n = 7 balanced, 6 sugar-enriched), and dilp5 (n = 7 balanced, 9 sugar-enriched). dilp3 and dilp5 mutants were verified by genomic PCR (see Extended Data Fig. 7-1). Feeding participation of females prefed on balanced or sugar-enriched diet is shown in Extended Data Figure 7-2. B, Preference of mated females prefed on indicated diet for 1 d and tested in binary choice assays with 5 mm sucrose (blue dye) against 1% yeast extract (pink dye). Genotypes were w¹¹¹⁸ (control, n = 19 balanced, 17 sugar-reduced), dilp5 (n = 12 balanced, 12 sugar-reduced). C, Representative traces obtained from dilp5 mutant females fed on indicated diet for 4 d. D, Ratio of mean sweet neuron response of flies fed on sugar-enriched diet (test) to that from flies fed on balanced diet (control) for 4 d. Mean neuronal responses are shown in Extended Data Figure 7-3. Recordings were taken from labellar L-type sensilla in mated females. Genotypes were as follows: w¹¹¹⁸ (control) and dilp5, n = 19–30 sensilla from four to six flies. The dotted line indicates a ratio of 1, i.e., equal taste sensitivity to sucrose in test and control flies. E, Preference of mated females prefed on indicated diet for 4 d and tested in binary choice assays with 5 mm sucrose (blue dye) against 1% yeast extract (pink dye). Genotypes were as follows: dilp2-Gal4 (n = 9 balanced, 10 sugar-enriched), UAS-dilp5 RNAi (n = 6 balanced, 6 sugar-enriched), dilp2>dilp5 RNAi (n = 10 balanced, 12 sugar-enriched), UAS-TNTG (n = 9 balanced, 9 sugar-enriched), and dilp2>TNTG (n = 17 balanced, 17 sugar-enriched). Expression of dilp5 in dilp2>dilp5 RNAi flies was tested by qRT-PCR (see Extended Data Fig. 7-4). F, Preference of mated female flies prefed on indicated diet for 4 d and tested in binary choice assays with 2 mm sucrose mixed with 25 mm each of serine, phenylalanine, and threonine (sucrose + 3AA, pink dye) against 2 mm sucrose alone (blue dye). Genotypes were as follows: w¹¹¹⁸ (control, n = 6 balanced, 6 sugar-enriched), dilp5 (8 balanced, 10 sugar-enriched). G, Preference of mated females prefed on indicated diet for 4 d and tested in binary choice assays with 5 mm sucrose (blue dye) against 1% yeast extract (pink dye). Genotypes were as follows: w¹¹¹⁸ (control, n = 9 balanced, 9 sugar-enriched), InR¹ (BL11661)/InR² (BL9646; n = 6 balanced, 6 sugar-enriched). H, Preference of mated females prefed on indicated diet for 4 d and tested in binary choice assays with 5 mm sucrose (blue dye) against 1% yeast extract (pink dye). Genotypes were as follows: UAS-InR RNAi (n = 7 balanced, 6 sugar-enriched), elav-Gal4 (n = 9 balanced, 8 sugar-enriched), elav-Gal4 > UAS-InR RNAi (n = 6 balanced, 6 sugar-enriched). I, Preference of mated females prefed on indicated diet for 4 d and tested in binary choice assays with 5 mm sucrose (blue dye) against 1% yeast extract (pink dye). Genotypes were as follows: UAS-InR RNAi (control n = 6 balanced, 6 sugar-enriched), v-Glut>UAS-InR RNAi (n = 7 balanced, 7 sugar-enriched), Gad1>UAS-InR RNAi (n = 7 balanced, 7 sugar-enriched), Ple>UAS-InR RNAi (n = 6 balanced, 7 sugar-enriched), Trh>UAS-InR RNAi (n = 8 balanced, 8 sugar-enriched), Tdc2>UAS-InR RNAi (n = 7 balanced, 7 sugar-enriched), npF>UAS-InR RNAi (n = 7 balanced, 7 sugar-enriched). Expression of InR in Trh-GAL4>UAS-InR RNAi flies was measured by qRT-PCR (see Extended Data Fig. 7-5). For all except C, D, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns - not significant, two-way ANOVA with Sidak's post hoc multiple comparisons test.

We next expressed dilp5 RNAi specifically in the IPCs under the control of dilp2-GAL4. We chose dilp2-GAL4 since among the dilps, dilp2 had the most stable expression between balanced and test diet treatments. Although not statistically significant, we found a decrease in the mean relative expression level of dilp5 in the dilp2 > dilp5 RNAi flies (Extended Data Fig. 7-4, dilp5 expression in brain p = 0.360222, independent samples t test). Interestingly, the dilp2 > dilp5 RNAi flies phenocopied the dilp5 mutant, in terms of the failure to decrease sugar preference following feeding on the sugar-enriched diet (Fig. 7E, balanced vs sugar-enriched diet, dilp2-GAL4 p < 0.0001, UAS-dilp5 RNAi = 0.0007, dilp2>dilp5 RNAi= 0.4489, UAS-TNTG p = 0.0105, dilp2>TNTG p = 0.2572; Sidak's post hoc multiple comparisons test; see also Table 3). Inhibition of vesicle secretion in dilp2-GAL4 neurons by expressing tetanus toxin (TNTG) had the same effect (Fig. 7E), consistent with the expectation that secretion of Dilps from the IPCs is important for their functional roles. We next tested whether the increase in amino acid preference following exposure to the sugar-enriched/protein-depleted diet was dependent on dilp5. We observed similar increases in feeding preference for the three-amino acid mixture in dilp5 mutants and control flies (Fig. 7F, balanced vs sugar-enriched, w¹¹¹⁸ p = 0.0109, dilp5 p = 0.0021; Sidak's post hoc multiple comparisons test; see also Table 3), suggesting that Dilp5 may be specifically required for altering behavioral responses to sugar, but not to amino acids, following dietary macronutrient imbalance.

dilp5 acts via InR in serotonergic neurons to control sugar preference

Since InR is the only known receptor for Dilp5 we hypothesized that loss of InR would phenocopy the dilp5 mutant. We examined feeding preference of trans-heterozygous flies of two different InR mutant alleles, namely InR⁰⁵⁵⁴⁵ (InR¹) and InR^E19 (InR²), such flies have been reported to have ∼90% reduction in InR transcript expression (Tatar et al., 2001). The baseline sucrose preference of the InR mutant flies was significantly lower than that of the control flies (Table 3), somewhat reminiscent of the trend observed for dilp5 mutants (Fig. 7A,F). As predicted, InR mutants failed to register a decrease in sugar preference after feeding on the sugar-enriched diet, whereas control flies that were tested in parallel did so (Fig. 7G, balanced vs sugar enriched, w¹¹¹⁸ p < 0.0001, InR¹/InR² = 0.9753; Sidak's post hoc multiple comparisons test; see also Table 3). Pan-neuronal expression of InR RNAi was able to recapitulate the defect in sugar-enriched diet-induced alteration of sugar preference (Fig. 7H, balanced vs sugar-enriched, elav-GAL4>InR RNAi p = 0.999; Sidak's post hoc multiple comparisons test; see also Table 3). Like wild-type flies, both control lines were able to reduce sugar preference after feeding on the sugar-enriched diet (Fig. 7H, UAS-InR RNAi p = 0.0423, elav-GAL4 p = 0.0031; Sidak's post hoc multiple comparisons test; see also Table 3).

To identify the class of neurons in which InR activity is required for modulating sugar preference, we expressed InR RNAi in different subsets of neurons and examined feeding preference of the resulting flies after feeding on balanced or sugar-enriched diets. Similar to dilp5 and InR mutants, flies in which InR RNAi was expressed in Trh-GAL4 (Extended Data Fig. 7-5, brain InR p = 0.070619, independent samples t test) serotonergic neurons exhibited a lower baseline sugar preference when prefed on the balanced diet, and sugar preference was found at that same level regardless of whether the flies were given a balanced or sugar-enriched diet (Fig. 7I, balanced vs sugar-enriched, Trh>InR RNAi p = 0.8235; Sidak's post hoc multiple comparisons test; see also Table 3). UAS control flies, which were treated and tested in parallel, displayed a diet-induced shift in sugar preference as expected (Fig. 7I, balanced vs sugar-enriched, UAS-InR RNAi p < 0.0001, Sidak's post hoc multiple comparisons test; see also Table 3). As did flies in which InR RNAi was expressed in glutamatergic (v-Glut-GAL4), GABAergic (Gad1-GAL4), dopaminergic (Ple-GAL4), octopaminergic (Tdc2-GAL4), or npF (npF-GAL4) neurons (Fig. 7I, balanced vs sugar-enriched, v-Glut>InR RNAi p < 0.0001, Gad1>InR RNAi p = 0.0022, Ple>InR RNAi p = 0.0002, Tdc2>InR RNAi p < 0.0001, npF>InR RNAi p < 0.0001; Sidak's post hoc multiple comparisons test; see also Table 3). A previous study found that serotoninergic (Trh-GAL4) neurons and the serotonin receptor 5HT2A mediate increased protein feeding on protein starvation by altering relative value of protein in higher order brain neurons (Ro et al., 2016). Serotonergic neurons have also been shown to regulate IPCs via 5-HT(1A) receptors present on the IPCs (Luo et al., 2012). Combining our observations with these previous findings suggests the presence of a feedback loop between serotonergic neurons and IPCs to regulate protein feeding in flies.