Gut tumors in flies alter the taste valence of an anti-tumorigenic bitter compound

Summary

The sense of taste is essential for survival, as it allows animals to distinguish between foods that are nutritious from those that are toxic. However, innate responses to different tastants can be modulated or even reversed under pathological conditions. Here, we examined whether and how the internal status of an animal impacts taste valence by using Drosophila models of hyperproliferation in the gut. In all three models where we expressed proliferation-inducing transgenes in intestinal stem cells (ISCs), hyperproliferation of ISCs caused a tumor-like phenotype in the gut. While tumor-bearing flies had no deficiency in overall food intake, strikingly, they exhibited an increased gustatory preference for aristolochic acid (ARI), which is a bitter and normally aversive plant-derived chemical. ARI had anti-tumor effects in all three of our gut hyperproliferation models. For other aversive chemicals we tested that are bitter but do not have anti-tumor effects, gut tumors did not affect avoidance behaviors. We demonstrated that bitter-sensing gustatory receptor neurons (GRNs) in tumor-bearing flies respond normally to ARI. Therefore, the internal pathology of gut hyperproliferation affects neural circuits that determine taste valence postsynaptic to GRNs, rather than altering taste identity by GRNs. Overall, our data suggest that increased consumption of ARI may represent an attempt at self-medication. Finally, although ARI’s potential use as a chemotherapeutic agent is limited by its known toxicity in the liver and kidney, our findings suggest that tumor-bearing flies might be a useful animal model to screen for novel anti-tumor drugs.

eTOC Blurb

In this work Leung et al. demonstrate that fruit flies with gut tumors display an increased propensity to consume the normally aversive anti-tumorigenic compound, aristolochic acid. The change in taste valence for aristolochic acid may be a form of self-medication as feeding on aristolochic acid suppressed the gut tumors.

Graphical Abstract

Introduction

In animals, taste preference is controlled by related but distinct gustatory neural circuits that determine the identity (how does it taste?) and valence (is it good or bad?) of a tastant.¹ In most vertebrates and invertebrates, bitter taste typically signals potentially toxic chemicals that evoke innate aversive responses. Interestingly, effective compounds in medicines often taste bitter.² Animals normally reject bitter-tasting foods to avoid ingestion of toxins, but some animals ingest unpalatable or repulsive food under pathological conditions. For example, when experiencing parasitic infections, chimpanzees in the wild consume Vernonia amydgalina, a bitter plant with antimicrobial properties.³ Similar observations have also been made in invertebrates, as parasitized caterpillars are more likely than non-parasitized caterpillars to ingest bitter-tasting anti-parasitic pyrrolizidine alkaloids.⁴

Cancer patients often report alterations in taste, a common symptom attributed to damaged tissue important for discerning the identity of tastants.⁵ However, it is unclear whether animals, including humans, under pathological conditions exhibit altered taste valence due to changes in their internal status. Given that some animals self-medicate, it is possible that an animal with tumors might increase acceptance of an otherwise aversive chemical in food with potential anti-tumorigenic properties.

To test the idea that an animal with tumors may alter its taste proclivities, we turned to the fruit fly, Drosophila melanogaster. We generated flies with gut tumors by expressing proliferation-inducing transgenes in intestinal stem cells (ISCs), and found a profound increase in acceptance of aristolochic acid (ARI)–a normally aversive, botanical compound. Moreover, consumption of ARI resulted in suppression of midgut hyperproliferation in the fly tumor models, suggesting that the reduced rejection of ARI in these flies may represent a form of self-medication.

Results

Gut tumors increase gustatory acceptance of aristolochic acid

To address whether tumors affect taste preferences, we generated fly lines with gut tumors to determine whether the condition alters the gustatory preference of ARI. Although tumors have been found in the adult germline and gut of wild-type flies, the relatively low frequency and long time required for tumorigenesis makes it difficult to conduct this analysis using flies with spontaneous tumors.⁶ Therefore, we created gut tumor models by expressing proliferation-inducing transgenes in ISCs of adult flies using the Gal4/UAS system.⁷ These included Ras1^A,⁸ Yki^3SA,⁹ and insulin receptor (InR),^10–12 which we expressed under control of the esg-Gal4 to drive expression specifically in ISCs of adult fruit flies.¹³ We included tubGal80^ts in the genetic background to prohibit Gal4 activity unless shifted to the permissive temperature¹⁴ and refer to these flies harboring the esg-Gal4 and the UAS-GFP tubGal80^ts as EGT.^9,15 All three models exhibit gut tumor-like phenotypes including epithelial hyperproliferation, multilayering, and thickening, which is consistent with previous reports.^8,9

While a previous study found that gut tumors do not impair the animal’s capacity to ingest food,⁹ we were curious whether gut tumors alter taste preference or food choice. To test the effects of gut tumors on changes in taste preference, we used two-way choice assays in which flies selected between 2 mM sucrose alone versus 2 mM sucrose laced with aversive tastants. The two food options were mixed with either red or blue food dye, permitting us to inspect the color of their abdomens, determine which food option they consumed, and calculate a preference index (PI). A complete preference for sucrose only or sucrose plus the bitter compound results in a PI of 1.0 and −1.0, respectively, while indifference between the two options results in a PI = 0. The dyes did not influence food preference since there was no bias for 2 mM sucrose combined with either dye alone (Figure S1A).

Interestingly, the tumor-bearing flies, regardless of their specific genotypes, exhibited pronounced reductions in their rejection of ARI. While control flies (with the EGT driver only) preferred sucrose alone over sucrose plus 5 mM ARI with a PI = 0.47 ± 0.018, the PI values dropped significantly in flies with gut tumors (Figure 1A). Specifically, Ras1^A- and InR-expressing flies no longer rejected ARI with a PI = 0.01 ± 0.06 and 0.04 ± 0.04, respectively, showing no bias between 2 mM sucrose alone versus 2 mM sucrose plus 5 mM ARI (Figure 1A). Yki^3SA-expressing flies not only increased their tendency to ingest ARI, but also reversed their response to ARI food from avoidance to attraction with a PI = −0.42 ± 0.10 (Figure 1A). Moreover, tumor-bearing flies were more likely to ingest ARI compared to their corresponding UAS only control flies (Figure 1A). When examining ARI avoidance at a lower concentration (1 mM), we again observed a decrease in PI values for flies with gut tumors (Figure 1B), demonstrating an increased tendency to ingest 1 mM ARI. As an additional control, we overexpressed tdTomato, a non-tumorigenic reporter gene with the same EGT driver and did not observe any changes in ARI avoidance (Figure S1B). Moreover, the Gal4 driver in the EGT line used in this study (esg-Gal4) expresses UAS-GFP, which does not impact selection or rejection of ARI (Figures 1A, 1B, and S1B).

Figure 1. Fruit flies with gut tumors exhibit reduced avoidance of ARI but not other bitter tastants.

Two-way choice feeding assays testing flies expressing Ras1^A (blue), Yki^3SA (red), or InR (green) in ISCs under the control of the EGT driver for 3 days. EGT or UAS only flies were used as controls.

(A–E) Taste preference between 2 mM sucrose alone versus 2 mM sucrose plus the indicated compounds. (A) 5 mM ARI. (B) 1 mM ARI. (C) 0.5 mM quinine (QUI). (D) 0.1 mM QUI. (E) 30 mM L-canavanine (CAN).

(F) Taste preference between 2 mM sucrose plus 0.5 mM QUI versus 2 mM sucrose plus 5 mM ARI.

n = 10 (A–D) and n = 5 (E, F) groups of 35–50 flies per genotype. Means ± SEMs. p values were calculated using the Kruskal-Wallis test with Dunn’s multiple comparisons test. **, p < 0.01; ***, p < 0.001; n.s. (not significant), p ≥ 0.05.

See also Figure S1 for extensive controls.

Taste preference could be affected by satiety, as bitter sensitivity in fruit flies could decrease during severe starvation.¹⁶ To determine if the increased ARI acceptance was due to starvation, we compared wild-type flies that were not starved with those starved for 18 or 24 hours prior to the two-way choice assay. The flies exhibited similar levels of aversion to 5 mM ARI whether or not they were starved prior to the assay (Figure S1C), indicating that starvation does not increase ARI acceptance. Furthermore, we measured the total volume of food intake in flies using a capillary feeder (CAFE) assay.¹⁷ We found that flies overexpressing Yki^3SA or InR in ISCs exhibited similar intakes of 2 mM sucrose compared to control flies, whereas Ras1^A-expressing flies consumed more 2 mM sucrose compared to the EGT control, but not to its UAS-Ras1^A only control (Figure S1D). Moreover, flies overexpressing Ras1^A or InR in ISCs exhibited similar intakes of 2 mM sucrose plus 1 mM ARI or 1 mM caffeine compared to their controls, whereas Yki^3SA-expressing flies slightly increased ARI or caffeine consumption when compared to its UAS-Yki^3SA only control, but not to the EGT control (Figures S1E and S1F). Importantly, tumor-bearing flies exhibited similar levels of ARI and caffeine consumption (Figures S1E and S1F). Therefore, tumor-bearing flies were not defective in food intake, and starvation was unlikely to account for increased ARI acceptance in these flies.

Recent studies suggest that pathogenic infection might impair general olfactory function in fruit flies.¹⁸ To determine whether gut tumor-induced ARI ingestion is caused by a general, non-specific disruption of the Drosophila gustatory system, we compared avoidance of three other aversive plant-derived compounds, which we assayed at concentrations that were soluble and elicited similar levels of avoidance to 5 mM ARI in control flies. In contrast to the changes in ARI preference due to the expression of proliferation-inducing genes in ISCs (Figures 1A and 1B), we observed no increased acceptance of 0.5 mM or 0.1 mM quinine (QUI; Figures 1C and 1D), 30 mM L-canavanine (CAN; Figure 1E), or 1 mM or 0.1 mM caffeine (CAF; Figures S1G and S1H). We also tested the preference between similarly aversive levels of 5 mM ARI and 0.5 mM quinine in flies overexpressing either the Ras1^A or Yki^3SA oncogene in ISCs. We found that these flies showed a significant preference for ARI over quinine compared to their controls (Figure 1F). Therefore, gut tumors increased the tendency of fruit flies to ingest ARI but not to the other aversive tastants tested.

ARI exhibited broad and long-lasting anti-tumor effects in the fly gut

To examine how ARI ingestion impacts gut tumors, we stained fly guts with antibodies for the mitosis marker phosphohistone H3 (pH3) and for GFP (the EGT line includes UAS-GFP driven by the esg-Gal4), which labels ISCs and their immediate progeny.¹³ Strikingly, ARI feeding suppressed midgut hyperproliferation caused by overexpression of Ras1^A, Yki^3SA, or InR (Figures 2A–2C). Moreover, a 2-day ARI regimen caused sustained reduction of hyperproliferation after the flies are shifted back on normal food for another 3 days (Figures 2D–2F). As we kept monitoring the tumor-bearing flies for 6 days on normal food following the 2-day ARI regimen, ISC abundance and hyperproliferation were significantly suppressed in Ras1^A- and InR-expressing flies but not Yki^3SA-expressing flies (Figures 3A–3C). Although Yki^3SA-expressing flies receiving the ARI regimen eventually regained a level of gut hyperproliferation close to that of the same cohort reared on normal food (Figure 3A), our electron microscopy examination and quantification of epithelial thickness suggests that a 2-day ARI regimen effectively alleviates the gut epithelial multilayering and thickening phenotype in tumor-bearing flies of all three genotypes at 6 days post-ARI treatment (Figures 3D–3F). Furthermore, we examined the intestinal barrier function by feeding flies with a non-absorptive blue dye that is typically restricted to the digestive tract. A loss of intestinal integrity results in blue dye that leaks throughout the body (referred to as “Smurfs”).¹⁹ Using the Smurf assay, we found that a large fraction of Yki^3SA-expressing flies developed a leaky gut phenotype, which was not observed for Ras1^A- and InR-expressing flies, and this phenotype was rescued by a 2-day ARI regimen (Figures 4A–4C). Yki^3SA overexpression also resulted in the accumulation of immature progenitor cells lacking septate junctions (Figures 4D, 4E, S2A, and S2B), which is consistent with the loss of cell polarity in fly gut tumors²⁰ and might account for the leaky gut phenotype.²¹ Interestingly, at 6 days post-ARI treatment, Yki^3SA-expressing flies exhibited a partial restoration of septate junctions (Figures 4F, 4G, S2C, and S2D) and a reduction of esg-positive cells (Figure 4H) compared to the same cohort of flies that did not receive the ARI regimen. Therefore, ARI exhibited broad and long-lasting anti-tumor effects in different fly gut tumor models.

Figure 2. ARI suppresses gut tumor proliferation in fly midguts.

A–C) Flies expressing GFP only (EGT), Ras31^A, Yki^3SA, or InR in ISCs were fed either normal food for 5 days (blue arrow) or normal food for 3 days (blue arrow), followed by feeding on food containing 10 mM ARI during the last 2 days (red arrow). (A) Quantification of mitosis by tabulating the number of pH3-positive cells in midguts. (B, C) Midguts stained with anti-pH3 (red), anti-GFP (green; driven by the esg-Gal4 in EGT), and DAPI (blue). The GFP labels ISCs and their immediate progeny. The white arrowheads highlight examples of pH3-positive cells. (B) Staining of midguts from flies fed normal food for 5 days. (C) Staining of midguts from flies fed normal food for 3 days and then food with 10 mM ARI for 2 days.

(D–F) Flies expressing GFP only (EGT), Ras1^A, Yki^3SA, or InR in ISCs were fed either normal food for 8 days (blue arrow) or normal food for 3 days (blue arrow), followed by feeding on food containing 10 mM ARI for 2 days (red arrow), followed by normal food for the last 3 days (blue arrow). (D) Quantification of mitosis by tabulating the number of pH3-positive cells in midguts. (E, F) Midguts stained with anti-pH3 (red), anti-GFP (green; driven by the esg-Gal4 in EGT), and DAPI (blue). The GFP labels ISCs and their immediate progeny. The white arrowheads highlight examples of pH3-positive cells. (E) pH3 staining of midguts from flies fed normal food for 8 days. (F) pH3 staining of midguts from flies fed normal food for 3 days, then food with 10 mM ARI for 2 days, followed by normal food for the last 3 days.

n ≥ 9 midguts for each genotype/treatment (A) and n ≥ 8 midguts for each genotype/treatment (D). Means ± SEMs. p values were calculated using the two-tailed Mann-Whitney U test. ***, p < 0.001; n.s., p ≥ 0.05. Scale bars: 50 μm (B, C, E, and F).

See also Figures S2–S4.

Figure 3. A single dose of ARI feeding causes long-lasting tumor suppression in tumor-bearing flies.

Flies expressing GFP only (EGT), Ras1^A, Yki^3SA, or InR in ISCs were fed either normal food for 11 days (blue arrow) or normal food for 3 days (blue arrow), followed by feeding on food containing 10 mM ARI for 2 days (red arrow), followed by the last 6 days on normal food.

(A) Quantification of mitosis by tabulating the number of pH3-positive cells in midguts.

(B, C) Midguts stained with anti-pH3 (red), anti-GFP (green; driven by the esg-Gal4 in EGT), and DAPI (blue). The GFP labels ISCs and their immediate progeny. The white arrowheads highlight examples of pH3-positive cells. (B) Staining of midguts from flies fed normal food for 11 days. (C) Staining of midguts from flies fed normal food for 3 days, then food with 10 mM ARI for 2 days, followed by normal food for the last 6 days.

(D) Quantification of average intestinal epithelial thickness.

(E, F) Electron micrographs of midguts. The red arrows in the electron micrographs highlight examples of epithelial thickness measured from the basement membrane side to the gut lumen side. (B) Midguts from flies fed normal food for 11 days. (C) Midguts from flies fed normal food for 3 days, then food with 10 mM ARI for 2 days, followed by normal food for the last 6 days.

n ≥ 7 midguts for each genotype/treatment (A) and n ≥ 10 midguts for each genotype/treatment (D). Means ± SEMs. p values were calculated using the two-tailed Mann-Whitney U test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; n.s., p ≥ 0.05. Scale bars: 50 μm (B, C), 10 μm (E, F).

See also Figures S2–S4.

Figure 4. ARI improves tissue organization and integrity in tumor-bearing flies.

Flies expressing GFP only (EGT), Ras1^A, Yki^3SA, or InR in ISCs were fed either normal food for 11 days (blue arrow) or normal food for 3 days (blue arrow), followed by feeding on food containing 10 mM ARI for 2 days (red arrow), followed by the last 6 days on normal food.

(A) Examples of Smurf assays performed on flies expressing GFP only (EGT) or Yki^3SA in ISCs, and fed normal food for 11 days.

(B) Examples of Smurf assays performed on flies expressing GFP only (EGT) or Yki^3SA in ISCs, and fed normal food for 3 days, then food with 10 mM ARI for 2 days, followed by normal food for the last 6 days.

(C) Smurf/total flies scored for each genotype/treatment.

(D–G) Anti-Tsp2A staining of midguts of flies expressing GFP only or Yki^3SA in ISCs. The upper row shows anti-Tsp2A (red), anti-GFP (green), and DAPI (blue). The bottom row shows anti-Tsp2A-only staining in a gray scale. Scale bars: 50 μm. (D) Flies expressing GFP only and fed normal food for 11 days. (E) Flies expressing Yki^3SA and fed normal food for 11 days. (F) Flies expressing GFP only and fed normal food for 3 days, then food with 10 mM ARI for 2 days, followed by normal food for the last 6 days. (G) Flies expressing Yki^3SA and fed normal food for 3 days, then food with 10 mM ARI for 2 days, followed by normal food for the last 6 days.

(H) Quantification of the percentage of esg-positive cells (labeled by esg-driven GFP) in the midguts of flies expressing GFP or Yki^3SA in ISCs. n = 6 midguts for each genotype/treatment. Means ± SEMs. p values were calculated using the two-tailed Mann-Whitney U test. **, p < 0.01.

See also Figures S2 and S5.

For both control and Yki^3SA tumor flies, ARI feeding caused a decrease in the percentage of esg-positive cells (Figure 4H) and increased expression of the septate junction components, Tsp2A and Ssk (Figures 4D–4G and S2A–S2D), which is a hallmark of enterocyte maturation.²² Consistent with a role of ARI in promoting ISC differentiation, our ultrastructural analysis of wild-type midguts revealed a loss of basally localized progenitor cells after 2 days of ARI ingestion (Figures S2E and S2F). Moreover, the electron micrographs of ARI-fed flies captured massive cell blebbing towards the gut lumen, indicative of cell death (Figure S2F). We confirmed the induction of ARI-induced cell death by immunostaining for cleaved caspase 3 (CCasp3), a marker of apoptosis (Figures S2G and S2H).

ARI can be metabolized in animal cells into reactive intermediates that bind with DNA and form aristolactam-DNA adducts.²³ To investigate potential mechanisms for the anti-tumor activity of ARI, we first examined whether ARI affects the expression activity of the esg driver. We did not observe changes in esg-driven GFP intensity between normally fed flies and those that received 18 hours of ARI feeding (Figures S2I–S2K). Thus, it is unlikely that ARI suppressed tumorigenesis by switching off activity of the esg-Gal4 and shutting down the expression of proliferation-inducing transgenes. Moreover, when we compared flies fed on normal food for 5 days with those that were fed on normal food for 4 days and then starved for 24 hours, we observed a decrease of mitotic ISC number in the EGT control flies (Figure S2L), which is consistent with previous studies reporting that fasting restricts ISC proliferation.¹² While starvation had no significant impact on the hyperproliferation of Yki^3SA-tumors, it caused some reduction of midgut proliferation in Ras1^A- or InR-expressing flies (Figure S2L). However, the reduction was far less dramatic relative to the impact of ARI feeding (Figures 2A and 2D). Based on these data, along with our observations using CAFE assays that addition of ARI did not cause significant changes in the total food intake of tumor-bearing flies (Figure S1E), we do not attribute the anti-tumor activity of ARI to starvation.

ARI is a mutagenic and nephrotoxic phytochemical found in the widely distributed plant family of Aristolochiaceae. In traditional herbal medicine, Aristolochiaceae extracts have been used to treat inflammation and cancer since antiquity, and ARI was reported to inhibit the growth of mouse mammary adenocarcinoma in vivo.²⁴ Interestingly, the impact of ARI on ISCs appears to be conserved in mammals as 1 μM and 5 μM ARI inhibits the growth and proliferation of human ISC organoids (Figures S3A–S3E) while inducing massive apoptosis and enterocyte differentiation (Figures S3F and S3G). However, the therapeutic potential of ARI as an anti-tumor reagent is not feasible due its severe toxicity in the kidney and liver. Laboratory mice with short-term exposure to ARI develop acute lethal renal tubular necrosis and serious liver injuries.²⁵ Consistent with the widely documented nephrotoxicity, 2 days of feeding on ARI caused swelling of Malpighian tubules (Figures S3H–S3J), the fly counterpart of the mammalian kidney. The toxicity might also account for our observation that 2 days post ARI feeding reduced the lifespan in control flies (EGT, UAS-Ras1^A, UAS-Yki^3SA, or UAS-InR; Figures S4A–S4E). Moreover, even if the tumor-bearing flies might benefit from the alleviation of gut tumors by ARI, we did not observe ARI-induced lifespan extension in these flies (Figures S4A and S4F–S4H), likely due to the counteracting effects of ARI toxicity.

Increased gustatory preference in tumor-bearing flies is specific for chemicals with anti-tumor effects

In addition to ARI, we also investigated whether other bitter tastants affected gut tumors. None of the other aversive tastants tested (quinine, caffeine, and L-canavanine) exhibited a broad impact on cell proliferation in different gut tumors (Figure 5) nor did they have long-lasting anti-tumor effects (Figures S5A–S5C). The only exception is caffeine, which resulted in a statistically significant suppression of proliferation in control and InR tumor flies (Figures 5A and 5C). Caffeine is an antagonist for the adenosine receptor (AdoR).²⁶ The observation of caffeine’s anti-proliferation effects in control and InR tumor flies is largely consistent with our previous findings that AdoR activation induces ISC proliferation via Ras/MAPK and PKA signaling.²⁷ Moreover, according to our previous finding that AdoR activity is also required for ISC differentiation,²⁷ caffeine is expected to have an opposite effect on ISC differentiation compared to ARI, which might account for its much more restricted anti-tumor activity. Furthermore, unlike ARI, the other bitter tastants did not reduce the percentage of Yki^3SA-expressing flies exhibiting gut leakiness in a Smurf assay (Figure S5D).

Figure 5. Testing the impact of multiple aversive tastants on cell proliferation in fly gut tumors.

Flies expressing GFP only (EGT), Ras1^A, Yki^3SA, or InR in ISCs were fed either normal food for 5 days (blue arrow) or normal food for 3 days (blue arrow), followed by feeding on food containing 1 mM quinine (QUI), 10 mM caffeine (CAF), or 30 mM L-canavanine (CAN) during the last 2 days (red arrow).

(A) Quantification of mitosis by tabulating the number of pH3-positive cells in midguts.

(B–D) Midguts stained with anti-pH3 (red), anti-GFP (green; driven by the esg-Gal4 in EGT), and DAPI (blue). The GFP labels ISCs and their immediate progeny. The white arrowheads highlight examples of pH3-positive cells. The flies were fed normal food for 3 days and then food laced with the indicated chemical for 2 days. (B) 1 mM QUI. (C) 10 mM CAF. (D) 30 mM CAN.

n ≥ 8 midguts for each genotype/treatment. Means ± SEMs. p values were calculated using the Kruskal-Wallis test with Dunn’s multiple comparisons test. ***, p < 0.001; n.s., p ≥ 0.05. Scale bars: 50 μm (B–D).

See also Figure S5.

We examined the effects of mitomycin C (MMC), a fungus-derived DNA crosslinking agent widely used for cancer chemotherapy,²⁸ on ISC hyperproliferation as well as on gustatory preference and food intake. Among the three gut tumor models we tested, MMC is most potent in suppressing Yki^3SA-driven hyperproliferation (Figure S5E). Interestingly, while MMC elicited neither attraction nor aversion in control flies, the Yki^3SA tumor flies exhibited a dramatic increase in taste preference for MMC (Figure S5F). Moreover, Yki^3SA-expressing flies, but not control flies or other tumor-bearing flies, exhibited elevated total food consumption when MMC was added (Figure S5G). Therefore, the increased acceptance of MMC in Yki^3SA tumor flies is likely associated with its anti-tumor effects.

Increased ARI acceptance in tumor-bearing flies is not due to impaired ARI detection in gustatory receptor neurons

Taste preference and food choice involve neurons in both the peripheral and central nervous systems, and the increase in ARI acceptance observed in tumor models could be due to changes in peripheral or central mechanisms, or both. Therefore, we set out to address whether gut tumor-bearing flies exhibit changes in peripheral detection of ARI in the gustatory receptor neurons (GRNs) of peripheral taste organs.

The largest taste organs are two bilaterally symmetrical labella at the end of the proboscis, each of which is decorated with 31 taste hairs (sensilla).^29,30 Sensilla house the dendrites of GRNs, which are categorized into small (S-type), intermediate (I-type) and large (L-type).^29,30 Since gut tumors suppressed the avoidance to ARI but not to quinine, caffeine, or L-canavanine, we focused on the two S-type sensilla responsible for sensing ARI (S3 and S6)^31,32 and measured their responses to ARI by performing extracellular field recordings (tip recordings). We found that the frequency of ARI-induced action potentials tested at both 1 mM and 5 mM were not affected in tumor-inflicted flies compared to control flies (Figure 6). In addition, we examined S7 sensilla, which also responds to ARI,³¹ and did not observe any differences in spike frequencies triggered by ARI between control and gut tumor-bearing flies (Figure S6A). To determine whether there were any differences in the kinetics of ARI-induced action potentials, we calculated the spike frequencies over 100 millisecond bins between 200–1200 milliseconds after application of the recording pipet. We observed similar adaptation kinetics in both the control and tumor-inflicted flies (Figures S6B–S6D). Collectively, our data suggest that the sensilla in the labella of tumor-bearing flies exhibit normal peripheral sensing of ARI.

Figure 6. Gut tumors do not affect peripheral ARI sensing in the fly labellum.

Action potentials assayed in taste sensilla on the labellum using tip recordings.

(A, B) Representative tip recording traces from S3 sensilla from flies expressing Ras1^A, Yki^3SA, or InR in ISCs for 3–6 days. The flies were stimulated with the indicated concentration of ARI. Ctrl^w and EGT flies were used as controls. (A) 1 mM ARI. (B) 5 mM ARI.

(C, D) Quantification of action potentials from S3 sensilla between 200–1200 ms following application of the indicated concentration of ARI. (C) 1 mM ARI. (D) 5 mM ARI. n = 6 flies for each genotype/dose. p values for 1 mM ARI versus 5 mM ARI for S3 sensilla were calculated using the two-tailed Mann-Whitney U test: w¹¹¹⁸, 0.0079; EGT, 0.045; EGT>Ras1^A, 0.0087; EGT>Yki^3SA; 0.0022; EGT>InR 0.013.

(E, F) Representative tip recording traces from S6 sensilla from flies expressed Ras1^A, Yki^3SA, or InR in ISCs for 3–6 days. The flies were stimulated with the indicated concentration of ARI. Ctrl^w and EGT flies were used as controls. (A) 1 mM ARI. (B) 5 mM ARI..

(G, H) Quantification of action potentials from S6 sensilla between 200–1200 ms following application of the indicated concentration of ARI. (G) 1 mM ARI. (H) 5 mM ARI. n = 12 flies for each genotype/dose. p values for 1 mM ARI versus 5 mM ARI for S6 sensilla were calculated using the two-tailed Mann-Whitney U test: w¹¹¹⁸, 0.0052; EGT, 0.0077; EGT>Ras1^A, 0.066; EGT>Yki^3SA, 0.086; EGT>InR, 0.0024.

Means ± SEMs. p values were calculated using the Kruskal-Wallis test with Dunn’s multiple comparisons test. n.s., p ≥ 0.05.

See also Figure S6.

Another gustatory organ in the fly that harbors GRNs is the terminal segments of their legs, referred to as tarsi.³³ Taste sensilla on tarsi contain GRNs that are activated by sugars, amino acids, and bitter chemicals including ARI.^33,34 To address if gut tumors impact the responses of tarsi to ARI, we performed proboscis extension response (PER) assays. We stimulated the forelegs of control, Yki^3SA-expressing, and Ras1^A-expressing flies with 10 mM sucrose alone or sucrose laced with 1 mM or 5 mM ARI. If a fly is motivated to consume the food, it extends its proboscis. We scored a full extension as 1.0, a partial extension as 0.5, and no extension as 0. As expected, the PERs exhibited by control flies (w¹¹¹⁸, EGT, or UAS-Yki^3SA) in response to stimulation with 10 mM sucrose were reduced when the sugar was laced with either 1 mM or 5 mM ARI (Figures S6E). The flies with Yki^3SA tumors exhibited similar aversion to ARI as the controls, with near identical PERs to sucrose plus 5 mM ARI (Figure S6E). Therefore, the differences in tarsal sensitivity, if any, do not account for the increased ARI acceptance observed in tumor-inflicted flies. We could not assess the impact of the Ras1^A tumors since flies harboring just the UAS-Ras1^A transgene, which does not produce tumors, exhibited a reduced aversion to ARI; although, there was no further reduction in flies with the Ras1^A tumors (Figure S6F).

Discussion

In this work, we demonstrate that flies with gut tumors increase their acceptance of an otherwise repulsive bitter chemical, ARI. The effect is profound since a high concentration of this chemical either became neutral or even attractive, depending on the tumor model. Thus, the gut tumors had a significant impact on the valence of ARI. We suggest that this may represent a form of self-medication since ARI suppressed midgut hyperproliferation, gut epithelial multilayering, and thickening phenotype in all three tumor gut models. Moreover, the gut tumor-bearing flies did not display increased acceptance of three other aversive tastants (quinine, L-canavanine, and caffeine), which either had no or much less anti-tumorigenic effects on the fly tumor models.

In addition to the antitumorigenic effects of ARI on fly gut tumors, we found that ARI also inhibited the growth and proliferation of human ISC organoids. However, ARI does not have application as a therapeutic due to its severe side effects, including kidney toxicity.³⁵ Nevertheless, our findings that tumor-inflicted flies alter their taste preference to seek compounds to alleviate tumor burdens in the gut suggest that gut tumor-bearing flies could be useful models to screen for new chemotherapeutic agents for gastrointestinal carcinomas.

One key question is whether the increased consumption of ARI is due to a change in peripheral or central mechanisms. Previous reports have documented examples in which modification of the internal status of flies, due to infection or nutrient deprivation, affect the peripheral sensing of odor or taste.^18,36 However, our findings indicate that changes in the flies’ internal status due to tumors likely affect taste valence through a central mechanism, since ARI-induced action potentials in GRNs were mostly indistinguishable between the control and gut tumor-bearing flies. Axons of different classes of GRNs project to discrete regions of the subesophageal zone (SEZ) in the brain.^37,38 The projection neurons from the SEZ then extend axons to higher order brain regions. In the future, it will be intriguing to investigate how gut tumors affect taste circuits postsynaptic to GRNs. Finally, our work establishes a new model for studying tumor-induced dietary alterations and modulation of gustatory valence in a genetically tractable model organism.

STAR Methods

RESOURCE AVAILABILITY

Lead Contact

The lead contact is Craig Montell (cmontell@ucsb.edu).

Materials Availability

This study did not generate new unique reagents.

Data and Code Availability

This study did not generate any unique datasets or code.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Drosophila stocks

The following stocks were obtained from the Bloomington Drosophila Stock Center: w¹¹¹⁸ (Ctrl^w, BL5905), UAS-CD8-GFP (BL32186), UAS-Yki^3SA (BL28817), UAS-TdTomato (BL36327). Stocks from the Perrimon lab collection are: esg-Gal4 UAS-GFP tubGal80^ts (EGT), UAS-Ras1^A, and UAS-InR.⁴²

Drosophila husbandry

Flies were reared on standard corn meal/agar medium under 12 hr light/12 hr dark cycles. For consistency, female adult flies, whose midguts are larger and easier to dissect, were analyzed in this paper. For the midgut experiments, we added 15 females and 5 males into each vial and transferred them to fresh vials every other day. For conditional expression using tubGal80^ts, flies were grown at 18°C until eclosion, maintained at 18°C for an additional ~3–5 days to allow for post-eclosion midgut development, and then shifted to 29°C to induce expression before midgut dissection and immunostaining. Crosses for the behavior and electrophysiology experiments were performed by placing 20–30 females and 20–30 males in bottles. ~50 female and ~10 male progeny were transferred to vials 1 day post-eclosion, and moved to 29°C at 4 days post-eclosion to inhibit the Gal80ts, and allow expression of the UAS transgenes. The behavioral assays and tip recordings were performed 3 days and 3–6 days after moving the flies to 29°C, respectively. Prior to performing the binary choice taste assays, the flies were pre-conditioned on 1% agarose for 18 hours unless noted otherwise as in Figure S1C. For the PER assays, the flies were starved on a wet Kimwipe^™ in vials for 24 hours prior to performing the assays.

Chemicals for fly food and organoid culture

The concentrations of chemicals used for the experiments were based on their solubility and previous publications.^32,43 The following compounds were purchased from Sigma-Aldrich: sucrose (S0389), ARI (A9451), quinine (Q1125), caffeine (C0750), L-canavanine (C1625), and mitomycin C (M0440). Fly food was melted and mixed thoroughly with a final concentration of 1–10 mM ARI, 1 mM quinine, 10 mM caffeine, and 30 mM L-canavanine. For tissue culture, food choice assays, and tip recordings, the stock solutions of all chemicals were dissolved in water.

METHOD DETAILS

Two-way choice feeding assays

Two-way choice feeding assays were performed with mated females as previously described⁴⁴ with slight modifications. 35–50 flies per assay were starved in 1% agarose (Thermo Fisher Scientific) for 18–20 hours in a humidified chamber. The fly genotypes were blinded to the experimenter. 4–6 hours after onset of the light cycle, flies were tested in 72-well plates (Nunc) that contained tastant/dye mixtures, and allowed to feed for 90 minutes in the dark so that the possible visual appeal of one food color over the other did not influence feeding. The tastant/dye mixtures were added to low melting point agarose (Thermo Fisher Scientific) to avoid overheating of the compounds.

The preference index (PI) was calculated according to the following equation: (N_red − N_blue)/(N_red + N_purple + N_blue). We previously determined that the concentrations of the red (Sulforhodamine B; Sigma-Aldrich) and blue (Brilliant Blue FCF; Wako Chemical) food dyes had no significant effect on food selection by assaying the preference of control flies to each dye with 2 mM sucrose.⁴⁵ Using concentrations that lead to indifference between the two food colorings (Figure S1A), we used red dye for sucrose-only food and blue dye for food with sucrose plus the bitter compound. A PI = 1.0 and −1.0 indicates complete preference for food with sucrose only and sucrose plus the bitter compound, respectively. A PI = 0 indicates no preference for either food. Trials in which <70% of the flies participated were discarded. We did not notice any difference in the ratio of flies participating in food intake between healthy flies and flies with gut tumors.

Extracellular tip recordings

Extracellular tip recordings were performed using 7–10 day-old flies as previously described,⁴⁴ with slight modifications. Bitter compounds were dissolved in 1 mM KCl, which served as the electrolyte, and backfilled into the recording electrodes (World Precision Instruments, 1B150F-3) with 20 μm openings (Sutter Instrument, P-97 puller). Tastant-induced signals were amplified and digitized with an IDAC-4 data acquisition device and Autospike software (Syntech). Spike sorting was used to identify spike amplitudes that correspond to the action potentials of bitter-responsive neurons. Neuronal responses were quantified by counting the number of action potentials between 200–1200 milliseconds following contact with the stimulus.

Proboscis extension response (PER) assays

Flies were starved for 24 hours in vials with a wet Kimwipe^™, anesthetized with CO₂, and affixed to a glass slide using nail polish with their backs facing upward. To aid their recovery, the flies were placed in a humid chamber for 1.5–2 hours. Before the commencement of the experiments, each fly was satiated with water, and they were also allowed to drink water in between each stimulus. To perform the PER assays, we touched the forelegs with a 1.5 μL droplet for 2 seconds. The response elicited within the next 5 seconds was recorded. Full extensions were assigned a score of 1, partial extensions received a score of 0.5, and no extension was scored as 0. The flies were tested with 100 mM sucrose both before and after the experiment to ensure that the tarsal sensilla were not damaged during the course of the experiment. Any fly that failed to respond to the 100 mM sucrose either before or after the assay was discarded. The flies were first tested with 10 mM sucrose followed by 10 mM sucrose laced with 1 mM ARI followed by 5 mM ARI. To confirm that the response was caused by the tastant, the flies were assessed with water following each positive response to ensure that water alone did not elicit a response.

Immunostaining and fluorescence imaging

Drosophila midguts or human intestinal organoids were fixed in 4% paraformaldehyde in PBS for 1 hour, incubated for 1–2 hours in Blocking Buffer [5% normal donkey serum, 0.3% Triton X-100, 0.1% BSA in PBS], and stained with primary antibodies overnight at 4°C in PBST [0.3% Triton X-100, 0.1% BSA in PBS]. Female midguts were analyzed in this study for convenience of dissection because of their larger size (we did not notice any differences in drug responses between male and female midguts). The primary antibodies used were: rabbit anti-pH3 (Millipore #06–570; 1:3000), mouse anti-GFP (Invitrogen A11120; 1:300), rabbit anti-cleaved-caspase 3 (Cell Signaling #9661S; 1:500), rabbit anti-Tsp2A (from Yasushi Izumi; 1:2000), rabbit anti-Ssk (from Yasushi Izumi; 1:2000). After primary antibody incubation, the midguts or human intestinal organoids were washed three times with PBST, stained with DAPI (1:2000) and Alexa Fluor^™-conjugated donkey-anti-mouse or donkey-anti-rabbit secondary antibodies (1:1000; Molecular Probes) or Alexa Fluor^™ 488 Phalloidin (1:1000; Molecular Probes) in PBST at 22°C for 2 hours, washed three times with PBST, and mounted in Vectashield medium. Due to the green fluorescence of ARI that might accumulate in the midgut lumen, GFP staining of tissue from flies feed ARI, and the corresponding control groups were performed using Alexa Fluor^™ 647-conjugated secondary antibodies and pseudo-colored in green.

For mitosis quantification, the numbers of pH3-positive cells in the entire midgut were counted with an epi-fluorescence microscope. Confocal images of midguts (the posterior region) or human intestinal organoids are captured with a Zeiss LSM780 confocal microscope. Z-stacks of images covering one layer of the epithelium from the apical to the basal side were obtained, adjusted and assembled using NIH Fiji (ImageJ), and shown as a maximum projection.

For measurement of average GFP intensity in ISCs, we showed z-stack maximum projections and filtered the images with a binary threshold by GFP staining, and measured the mean intensity of the GFP positive areas in the resulting images. Values from the 18 hours ARI fed were normalized to the control (18 hours normal food), which was fixed to an average of 1.

For quantification of the diameter of Malpighian tubules, we measured the diameter of the same region (lower tubule) in different groups. The lower tubules are the first segments following the branching of Malpighian tubules.

Electron microscopy

Tissues of the posterior midgut were fixed in the fixative solution [2.5% glutaraldehyde 1.25% paraformaldehyde and 0.03% picric acid in 0.1 M sodium cacodylate buffer (pH 7.4)] for 2 hours at room temperature, washed in 0.1 M cacodylate buffer, postfixed with 1% osmium tetroxide (OsO₄) and 1.5% potassium ferrocyanide (KFeCN₆) for 1 hour, washed twice in water, once in maleate buffer (MB), incubated in 1% uranyl acetate in MB for 1 hour, washed twice in water, and dehydrated in grades of alcohol (10 minutes each: 50%, 70%, 90%, 100%, 100%). The samples were then put in propylene oxide for 30 minutes and infiltrated overnight in a 1:1 mixture of propylene oxide and TAAB Epon (TAAB Laboratories Equipment Ltd). The following day the samples were embedded in TAAB Epon and polymerized at 60°C for 48 hours. Ultrathin plastic sections (~60 nm) were cut on a Reichert Ultracut-S microtome, picked up onto copper grids, and contrasted with 0.3% lead citrate. Grids with the plastic sections were imaged using a JEOL 1200-EX transmission electron microscope operating at 80 kV with an AMT 2k CCD camera. Representative images from more than three different sections are presented. For quantification of epithelial thickness, we examined electron micrographs with the structural hallmarks of visceral muscle, basement membrane, enterocyte nucleus, and microvilli, and measured the distance between the basement membrane and the tip of microvilli by drawing a line that crossed the enterocyte nucleus.

Human organoid culture and quantification

Normal human duodenum epithelial cells (H362D) were embedded in Matrigel (Corning^®, 356231) on ice, seeded at the center of tissue culture dishes, incubated at 37°C for ≥10 minutes to polymerize the Matrigel, and grown in basal organoid media [Advanced DMEM/F12 (Gibco, 12634010), 2 mM GlutaMAX (Gibco, 35050061), 10 mM HEPES (Gibco, 15630080), 1x penicillin-streptomycin (Gibco, 10378016), 1x B-27 supplement (Gibco, 17504044), 1x N-2 supplement (Gibco, 17502048), 100 μg/mL Primocin (Invivogen, Ant-pm-1), 100 μg/mL Normocin (Invivogen, Ant-nr-1), 10 μg/mL Fungin (Invivogen, Ant-fn-1)] supplemented with 10 mM nicotinamide (Sigma, N0636), 1 mM N-acetylcysteine (Sigma, A9165), 50% homemade L-WRN conditioned media⁴⁰ and 50 ng/mL EGF (Peprotech, 315–09). The medium was changed every 4 days. To passage the organoids, they were removed from Matrigel using Cell Recovery Solution (Corning^®, 354253) and mechanically dissociated before being embedded in fresh Matrigel. For maintenance of organoid cultures, passages were performed every 1–2 weeks with a 1:5 split ratio.

For drug treatment, 75 μL of organoids embedded in Matrigel were seeded in each well of a 24-well plate and incubated with media containing the indicated concentrations of ARI. For cell counting, each organoid was recovered from Matrigel with Cell Recovery Solution, digested into a single cell suspension with trypsin-EDTA (Gibco, 25200056), rinsed with PBS, centrifuged at 4 k RPM for 3 minutes, and resuspended in 250 μL ice-cold PBS. 10 μL of the resuspended cells were mixed with 10 μL of a 0.4% solution of trypan blue (G-Biosciences, 786–1383) to visualize dead cells and analyzed on a Countess^™ automated cell counter (Invitrogen, C10227) with the following settings: range, 5–40 μm; sensitivity, 5; and circularity, 8.

Capillary Feeder (CAFE) assays

CAFE assays were performed as described previously.⁴⁶ Briefly, groups of 10–20 mated female flies were collected 3–5 days post-eclosion, and then maintained at 29°C for 3 days. Prior to the assay, flies were starved for 16–20 hours at 25°C and moved to a food source where their food intake was recorded after three hours (ZT 2–5). To measure food intake, flies were allowed to feed from two 5 μL capillary tubes (VWR, 53432–706), which were secured to the top of a Drosophila vial using a cut pipette tip. To visualize food intake, blue food coloring (Brilliant Blue; 0.01% w/v) was added to each food source. To control for changes in food volume that resulted from evaporation, food consumption was normalized to the average change in food volume from three empty vials.

Measurement of fly lifespan

To test the effect of ARI treatment on the survival of flies with gut tumors, we collected mated female flies 3–5 days post eclosion at 18°C at a density of 10–15 flies per vial. We then shifted the flies to 30°C to allow for the expression of tumor-causing genes. On the third day at 30°C, we fed the flies for 2 days standard food laced with 1 mM ARI. On day 5, the flies were shifted back to normal fly food and placed in an incubator set to 25°C with 60% humidity, where they remained for the duration of their lifespan. Every 2–3 days, we shifted each group of flies to a new vial with fresh fly food and counted the number of dead flies. The hazard ratios shown in Figure S4A, were calculated using the “survival” and “survivalAnalysis” packages in R.

Smurf assays

Smurf assays were performed as previously described.¹⁹ 5% sucrose solution was mixed with blue dye no. 1 at a concentration of 2.5% (wt/vol). Flies were kept on paper towels for 12 hours before analysis. A fly was counted as Smurf positive when we observed blue dye outside of the digestive tract.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses were performed using GraphPad Prism 8. For comparison of any two data groups of different samples, p values were calculated using the non-parametric two-tailed Mann-Whitney U test. For comparison of multiple parallel data groups, p values were calculated using the non-parametric Kruskal-Wallis test with Dunn’s multiple comparisons test. *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001; n.s., p > 0.05. All error bars represent standard error of the mean (SEM). Sample sizes (n) are listed in the figure legends. For the two-way choice assays each ‘n’ represents a single test performed with 35–50 animals. For the CAFE assays, each ‘n’ represents a single test performed with 10–20 animals. Each ‘n’ for the other experiments represents an analysis of a single, independent biological sample.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti-GFP (mouse)	Invitrogen	Cat# A-11120; RRID:AB_221568
anti-pH3 (rabbit)	Millipore	Cat# 06-570; RRID:AB_310177
anti-cleaved-caspase 3 (rabbit)	Cell Signaling	Cat# 9661S
anti-Tsp2A (rabbit)	{Izumi, 2016 #7871}	NA
anti-Ssk (rabbit)	{Izumi, 2016 #7871}	NA
DAPI	Thermo Scientific	Cat# D1306
Donkey anti-mouse (Alexa Fluor 488)	Molecular Probes	Cat# A21202; RRID:AB_141607
Donkey anti-mouse (Alexa Fluor 647)	Molecular Probes	Cat# A21236; RRID:AB_2535805
Donkey anti-rabbit (Alexa Fluor 555)	Molecular Probes	Cat# A31572; RRID:AB_162543
Phalloidin (Alexa Fluor 488)	Molecular Probes	Cat# A12379; RRID:AB_2759222
Chemicals
Agarose	Life Technologies	Cat# 16500500
Sucrose	Sigma-Aldrich	Cat# S0389
Aristolochic acid	Sigma-Aldrich	Cat# A9451
Quinine	Sigma-Aldrich	Cat# Q1125
Caffeine	Sigma-Aldrich	Cat# C0750
L-canavanine	Sigma-Aldrich	Cat# C1625
Mitomycin C	Sigma-Aldrich	Cat# M0440
Sulforhodamine B (red dye)	Sigma-Aldrich	Cat# 230162
Brilliant Blue FCF	Wako Chemicals	Cat# 027-12842
Paraformaldehyde 16% solution	Electron Microscopy Sciences	Cat#15710
Donkey serum	Jackson Laboratories	Cat# 017-000-121
Triton X-100	Sigma	Cat# T9284
Bovine serum albumin	Sigma	Cat# A3912
Formaldehyde/glutaraldehyde 2.5% each in 0.1 M sodium cacodylate buffer pH 7.4	Electron Microscopy Sciences	Cat# 15949
Picric acid	Electron Microscopy Sciences	Cat# 19550
Paraformaldehyde (powder)	Electron Microscopy Sciences	Cat# 19208
Osmium tetroxide 2% solution	Electron Microscopy Sciences	Cat# 19150
Potassium ferrocyanide	Electron Microscopy Sciences	Cat# 25154
Maleic Acid	Electron Microscopy Sciences	Cat# 18150
Uranyl acetate	Electron Microscopy Sciences	Cat# 22400
Propylene oxide	Electron Microscopy Sciences	Cat# 20412
TAAB 812 Resin	TAAB Laboratories Equipment Ltd	Cat# T022
Lead citrate	Electron Microscopy Sciences	Cat# 17800
Matrigel	Corning	Cat# 356231
Cell Recovery Solution	Corning	Cat# 354253
DMEM/F12	Gibco	Cat# 12634010
GlutaMAX	Gibco	Cat# 35050061
HEPES	Gibco	Cat# 15630080
Penicillin-streptomycin	Gibco	Cat# 10378016
B-27 supplement	Gibco	Cat# 17504044
N-2 supplement	Gibco	Cat# 17502048
Primocin	InvivoGen	Cat# Ant-pm-1
Normocin	InvivoGen	Cat# Ant-nr-1
Fungin	InvivoGen	Cat# Ant-fn-1
Nicotinamide	Sigma	Cat# N0636
N-acetylcysteine	Sigma	Cat# A9165
Homemade L-WRN conditioned media	{Miyoshi, 2013 #7867}	NA
EGF	Peprotech	Cat# 315-09
Trypsin-EDTA	Gibco	Cat# 25200056
Trypan blue	G-Biosciences	Cat# 786-1383
Blue dye no. 1	Spectrum Chemical	Cat# 3844-45-9
Experimental Models: Cell Lines
Human: duodenum epithelial cells	HDDC Organoid Core	Cat# H362D
Experimental Models: Organisms/Strains
Drosophila: w1118	Bloomington Drosophila Stock Center	Cat# BL5905
Drosophila: UAS-CD8-GFP	Bloomington Drosophila Stock Center	Cat# BL32186
Drosophila: UAS-Yki3SA	Bloomington Drosophila Stock Center	Cat# BL28817
Drosophila: UAS-TdTomato	Bloomington Drosophila Stock Center	Cat# BL36327
Drosophila: esg-Gal4 UAS-GFP tubGal80ts	Perrimon Lab	NA
Drosophila: UAS-Ras1A	{Xu, 2017 #5951}	NA
Drosophila: UAS-InR	{Xu, 2008 #7872}	NA
Software
Prism8	Software	https://www.graphpad.com/scientific-software/prism/; RRID:SCR_002798
Fiji	Software	https://imagej.net/Fiji; RRID:SCR_002285
Autospike	Software	https://www.ockenfels-syntech.com/products/signal-acquisition-systems-2/
R Project	Software	http://www.r-project.org/; RRID:SCR_00195
Other
Glass capillaries	World Precision Instruments	Cat# 1B150F-3
Capillary tubes	VWR	Cat# 53432-706

Highlights.

• Gut tumors cause flies to increase their ingestion of anti-tumorigenic compounds

• Gut tumors do not affect avoidance of aversive compounds devoid of anti-tumor effects

• Increased consumption of anti-tumorigenic compounds suppresses gut tumors

• Shift in taste valance due to tumors via a mechanism postsynaptic to taste neurons