Evolution of taste processing shifts dietary preference

Enrico Bertolini, Daniel Münch, Justine Pascual, Noemi Sgammeglia, Carlos Ribeiro, Thomas O. Auer

Posted on: 31 March 2025

Preprint posted on 12 October 2024

Flies learning to like 'vomit fruit': an example of bitter going from 'yuck' to 'yum.' New research in Drosophila sechellia reveals how taste circuits can evolve to make bitter better. 🍋🪰

Selected by T. W. Schwanitz

Categories: animal behavior and cognition, evolutionary biology, genomics, neuroscience

Introduction

Do you like dark chocolate? If so, you’re not alone in your penchant for bitter foods. One particular species of Drosophila has also evolved a taste for what would otherwise be a repulsive cue: Drosophila sechellia. Among the 1,600 or so species of Drosophila, this one stands out for its remarkable specialization—feeding and laying eggs primarily on noni fruit, which is native to the Seychelles. Often called “vomit fruit” due to its notably unpleasant smell (at least to humans), noni fruit is preferred by D. sechellia over more commonly appealing options like bananas, grapes, or apples (Fig. 1A). This raises an intriguing question: where did this peculiar preference come from, and what specific neural circuits might underlie it?

To investigate, Bertolini and colleagues harnessed the extensive genetic tools available for Drosophila melanogaster and conducted a series of behavioral experiments across three species of drosophilids—all to help unravel the evolutionary mechanisms behind bitter acceptance and noni specialization in D. sechellia.

Highlighted results

Even a cursory glance at the figures in this paper reveals that this is far from a straightforward story. With the growing arsenal of tools available to neuroscientists—especially those working with Drosophila—it has become increasingly apparent that biological neural circuits function in intricate and often seemingly paradoxical ways. Readers are encouraged to dig into the preprint for all the complexity of this story, as the scope of this preLight is limited to some of the especially interesting insights and experiments.

To begin, the authors demonstrate that the preference for noni is supported by changes in taste, even in the absence of olfactory input. Using a well-designed semi-natural four-choice assay with real fruit, they tested D. sechellia flies that were genetically modified to be unable to smell most compounds by knocking out two key olfactory co-receptor genes, Ir8a and Orco (Fig. 1A, rightmost plot). The authors then employed various preference assays, confirming that these effectively capture D. sechellia’s natural affinity for noni fruit under different conditions (Fig. 1B-E). One particularly compelling experiment involved D. sechellia flies with the Orco gene knocked out and their antennae surgically ablated, rendering them completely anosmic. While these flies did not exhibit a clear preference for noni juice over grape juice, they still showed a statistically significant difference in behavior compared to the generalist species Drosophila melanogaster and Drosophila simulans (Fig. 1C). This indicates that something has changed in the taste system of this species.

Figure 1 of the preprint: A) Illustrates the behavioral four-choice assay, highlighting the species’ preferences investigated in this study and showing that D. sechellia flies with limited olfactory capabilities still favor noni. B) shows several different types of assays the authors tested, all reproducing the loss of sweet preference. C) Flies with their antennae removed still choose noni over grape juice. D-F) More experiments and analysis showing that D. sechellia prefers noni (check out the preprint for all the details!). G) Schematic of different taste receptors on the fly’s labellum (its tongue, effectively). Sweet and bitter neurons when silenced (green, purple, i.e., those first two with asterisks over them; Ir25a, the final one with asterisks, is expressed in all taste neurons) change D. melanogaster’s preference for grape juice over noni juice.

To explore the genetic basis of these species differences, the authors leveraged the extensive genetic toolkit available in D. melanogaster, the six-legged workhorse of biology. They used several transgenic lines in D. melanogaster to drive the expression of Kir2.1 in neurons responsible for sensing bitter and sweet (Fig. 1G). Kir2.1 is an inward-rectifier potassium channel that hyperpolarizes neurons, effectively silencing them. Highlighting the intricacies of taste processing, they found that silencing neurons involved in sensing any of these modalities—sweet, bitter, or acidic—could alter D. melanogaster’s preference for noni juice (Fig. 1G, green, purple, and gray-blue boxplots).

Amid the complexity of taste processing, it is remarkable that the authors identified a specific receptor change that probably plays a role in D. sechellia’s preference for noni fruit. The gustatory receptor Gr39a, previously identified as a caffeine sensor in D. melanogaster, was shown to have four alternatively spliced isoforms. One isoform, Gr39a.a, carries a three-amino-acid deletion specifically in D. sechellia (Fig. 4B of the preprint). While this deletion does not occur in the ligand-binding domain, it nonetheless has a significant impact on the function of the bitter receptor Gr39a.a, as demonstrated in the authors’ experiments. When the D. sechellia version of Gr39a.a was expressed in D. melanogaster flies lacking their own endogenous protein, neurons stimulated by caffeine responded similarly to those in complete knockouts (Fig. 4D of the preprint). These results indicate that the D. sechellia variant of Gr39a.a behaves like a non-functional protein in this assay.

Bertolini and colleagues conducted additional experiments to further probe the role of Gr39a.a in shaping neuronal response dynamics to bitter compounds like caffeine and coumarin. They demonstrated that expressing the D. sechellia version of Gr39a.a in D. melanogaster alters the neuronal response dynamics of bitter neurons, making them resemble those of D. sechellia. Conversely, expressing the D. melanogaster version of Gr39a.a in D. sechellia produces neuronal dynamics similar to D. melanogaster’s bitter neurons. It seems straightforward, right? Swap the Gr39a.a allele between the two species and then their bitter neurons swap properties too.

However, the plot thickens with behavioral experiments. While D. sechellia bitter neurons can be activated by caffeine when they carry the D. melanogaster version of the allele (Fig. 4F), the flies themselves are not repulsed by the taste of caffeine in the same way that D. melanogaster flies are (Fig. 4G-H). This striking observation suggests that higher-level neural circuits process bitter tastes differently in D. sechellia, which relishes the bitter noni juice. In the preprint’s final main figure (Fig. 5), the authors delve into these higher-order circuits, proposing that differences in interneuron activity may help explain the distinct taste processing in D. sechellia.

Why I liked this preprint

This work tackles an intricate neural circuit through an evolutionarily informed lens. One of my favorite aspects of this preprint is the semi-natural four-choice assay: an elegantly designed experiment that uses a real, complex stimulus to convincingly demonstrate the predilections of different fly species. It also persuasively shows that taste plays a central role in D. sechellia’s preference for noni. I particularly appreciated the authors’ efforts to unravel both the peripheral changes in the bitter-tasting circuit and the alterations in higher-order processing areas of the brain, offering a comprehensive view of how these adaptations may contribute to D. sechellia’s unique behavior.

Tags: chemoreception, circuits, drosophila, fly, fruit fly, taste

doi: https://doi.org/10.1242/prelights.40091

Read preprint

(No Ratings Yet)

Author's response

Thomas O. Auer shared

1. Given that there are taste neurons on the legs of the flies, how do you think that bitter sensing on the legs of D. sechellia has changed to make them more attracted to noni fruit? Do you expect the story to be similar to what you outline in your preprint, or is there likely to be some additional factor influencing leg stimulation?

That’s a great question. Sensory neurons on the leg tarsi are involved in the early steps of food evaluation and for sure play a role in decision making. Unfortunately though, how information from leg neurons is integrated into circuits in the subesophageal zone (where we imaged sensory inputs from the labellum, the mouth of the fly) is still largely unknown. Moreover, most of these neurons also innervate the ventral nerve cord, the spinal cord equivalent of the flies, and additional processing steps might happen there. For this preprint, we therefore focused on the direct link between the labellum and motor outputs. These circuits have recently been described in D. melanogaster through connectomic and imaging approaches so we have a chance to go down to comparative neuron level studies in the future. When working with multiple species in a comparative approach, I find the main challenge is often to focus on a tangible circuit to reduce complexity. In our case, the strong sucrose tuning of labellum-motor circuits in D. melanogaster and great recent circuit work made it a good target. Surprisingly, in D. sechellia this sucrose tuning is lost – but not at the periphery but somewhere downstream.

My guess would be that leg circuits also show differences across species; potentially with altered bitter sensitivity in D. sechellia. If this happens at the level of the sensory neuron or elsewhere remains open. It will be exciting to dive into this in the future and also to see how others will establish links between leg and labellar circuits in the D. melanogaster connectome.

2. You’ve extensively tested individual bitter compounds of noni juice, noni juice, and the whole fruit itself. How do you think these individual compounds are working together in taste processing? Do you think that repeating your experiments in Fig. 4I with noni juice as well would recapitulate the results you see with the single compounds caffeine and coumarin?

Based on the experiments we present in Figure 1, I am sure, the dose-response curve you refer to in Fig. 4I for individual compounds would look similar for noni juice. However, working with the juice itself introduces the additional challenge of its strong olfactory attraction requiring testing of smell-blind flies. Moreover, it is hard to make dilution curves for natural, complex stimuli. Individual compounds offer several advantages here. First, again they help to simplify experimental parameters as they only activate individual taste populations. Second, they have been used extensively by other researchers and we know a lot about how flies react to them. In our study, we focused for example on caffeine. There is a clear species difference in behaviour and physiology towards this single chemical and it helped us to identify and test the role of the Gr39a locus.

Overall, we aimed to test noni juice in most of our experiments as we were really interested in ecologically relevant stimuli. It activates sweet and bitter neurons so for sure processing and integration of both modalities is important. That’s also why we went for volumetric imaging to see how this could potentially differ in the SEZ.

3. It looks like in Figure 5 (panel C especially) that the change in preference/neural activity in D. sechellia is more a function of a loss of activation by sweet stimuli rather than a direct increase in activity for bitter compounds like noni juice (though maybe I misunderstand?). In your schematic, however, you show an increase in the bitter pathway as driving specialization. Could you clarify where the preference for bitter arrow comes from?

Yes, indeed. The strongest phenotype we detected is the loss of sucrose tuning in D. sechellia. Both other species show more sucrose-evoked activity in the SEZ which is reduced in this species. Noni activates bitter neurons more strongly in D. sechellia (which is counter-intuitive) but then an increase in downstream activity – at the level of detail we could achieve in our imaging approach – is harder to spot.

In the schematic we tried to combine the information from behaviour and physiology and contrast the two species. The specific arrow you refer to represents not only bitter but rather the output of integration from peripheral sensation. We should add more details here to avoid confusion. We mainly wanted to highlight that D. melanogaster motor circuits are really responsive to sucrose; this is lost in D. sechellia and, inferred from behaviour and a tendency in physiology, noni is preferred by this fly.

Have your say Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up to customise the site to your preferences and to receive alerts

Also in the animal behavior and cognition category:

Responses to conflicting binocular stimuli in mouse primary visual cortex

Daniel P. Montgomery, Daniel A. Bowen, Jin Wu, et al.

Selected by 28 October 2025

Maitri Manjunath

Discussion

Effects of transcranial photobiomodulation on peripheral biomarkers associated with oxidative stress and complex IV activity in the prefrontal cortex in rats subjected to chronic mild stress

Luciana Bortoluzzi, Rafael Colombo, Karoline Motta Pinto, et al.

Selected by 21 October 2025

Rickson Ribeiro, Marcus Oliveira

Discussion

Psychedelics Align Brain Activity with Context

Devon Stoliker, Leonardo Novelli, Moein Khajehnejad, et al.

Selected by 01 September 2025

Loïk Holdrinet et al.

Discussion

Also in the evolutionary biology category:

A high-coverage genome from a 200,000-year-old Denisovan

Stéphane Peyrégne, Diyendo Massilani, Yaniv Swiel, et al.

AND

A global map for introgressed structural variation and selection in humans

PingHsun Hsieh, Natthapon Soisangwan, David S. Gordon, et al.

Selected by 02 December 2025

Siddharth Singh

Discussion

Dissecting Gene Regulatory Networks Governing Human Cortical Cell Fate

Jingwen W. Ding, Chang N. Kim, Megan S. Ostrowski, et al.

Selected by 14 November 2025

Manuel Lessi

Beyond venomous fangs: Uloboridae spiders have lost their venom but not their toxicity

Xiaojing Peng, Ludwig Dersch, Josephine Dresler, et al.

Selected by 13 November 2025

Daniel Fernando Reyes Enríquez, Marcus Oliveira

Also in the genomics category:

A high-coverage genome from a 200,000-year-old Denisovan

Stéphane Peyrégne, Diyendo Massilani, Yaniv Swiel, et al.

AND

A global map for introgressed structural variation and selection in humans

PingHsun Hsieh, Natthapon Soisangwan, David S. Gordon, et al.

Selected by 02 December 2025

Siddharth Singh

Discussion

Human single-cell atlas analysis reveals heterogeneous endothelial signaling

Zimo Zhu, Rongbin Zheng, Yang Yu, et al.

Selected by 11 November 2025

Charis Qi

Discussion