Note: I wrote this essay on my postdoctoral work for a competition (I didn’t win). I wanted to share because I think it’s a nice summary of the work I’ve done over the last several years, written for a general scientific audience. Non-scientists may have a hard time in some spots, but please comment or email me if you have any questions!

 

Few sensations are more viscerally satisfying than the taste of a delicious meal. Imagine that first savory bite of Thanksgiving dinner or the last sweet morsel of a rich chocolate cake.

But taste is more than a hedonistic luxury: it is critical to our survival. Across the animal kingdom, taste ensures that we consume calorie-rich foods such as sugar and essential nutrients such as salt and protein. Taste also prevents us from ingesting potential toxins, which we perceive as bitter. Mistakenly swallow a poisonous bitter root, and in a split second you might drop dead. That’s why taste is perhaps the most hardwired of our senses: we’re all born with a fondness for sugar and distaste for overly bitter or sour foods.

How are these taste perceptions encoded in the brain? Most research supports a “labeled-line” model, in which each type of taste activates a single neural pathway to evoke specific responses.¹ For example, sugar and bitter compounds activate separate populations of neurons from the tongue to the brain. The sugar-sensing pathway urges us to eat, whereas the bitter-sensing pathway suppresses this urge.

As a postdoc in Richard Axel’s lab at Columbia University, I have investigated taste processing using the fruit fly Drosophila melanogaster, which shares most of our taste preferences and is highly amenable to genetic manipulations. Early observations led me in unexpected directions that have revealed new principles in taste coding, ultimately suggesting that taste processing is much more complex than a simple labeled-line system.

A hunger-dependent switch in behavior

My first unexpected discovery arose when I examined how flies respond to the taste of acetic acid, a natural product of fruit fermentation that we encounter as vinegar. Acetic acid is a potential energy source but can also be toxic.^2,3 I found that unlike other chemicals, acetic acid evokes opposing responses depending on hunger state. Fed flies are repulsed by the taste of acetic acid, likely due to its potential toxicity, whereas hungry flies show strong attraction that may reflect their overriding need for calories.⁴

This observation immediately challenged the conventional labeled-line model. In a labeled-line system, each tastant activates a single pathway corresponding to a specific valence—attraction or aversion. But acetic acid can elicit either attraction or aversion depending on hunger state. How does the taste system accommodate this flexibility?

One stimulus, two pathways

To address this question, I used genetically targeted neuronal silencing and calcium imaging to identify which taste neurons mediate the responses to acetic acid. These experiments revealed that acetic acid elicits opposing, hunger-dependent behaviors by simultaneously activating two different classes of taste neurons. Bitter-sensing neurons mediate aversion in fed flies, whereas sugar-sensing neurons drive attraction in hungry flies. These two pathways compete to determine the fly’s behavior, and hunger shifts the response from aversion to attraction by enhancing the attractive sugar pathway as well as suppressing the aversive bitter pathway.⁴

 

In contrast to the labeled-line model, these results demonstrate that a single tastant can activate multiple pathways. Other recent studies also support this notion.^5,6 Moreover, our findings reveal remarkable flexibility in the taste system: animals can show dramatically different responses to the same stimulus depending on their internal needs. In fact, most of us are familiar with this phenomenon. The same dish that seemed irresistible at the first bite can viscerally repel us after eating too much. The simultaneous activation of opposing taste pathways that are differentially gated by hunger explains how even a “hardwired” system can accommodate a behavioral switch.

Temporal dynamics in the taste system

While using calcium imaging to examine taste responses, I made a second unexpected observation: temporal dynamics in taste coding. Temporal dynamics in the peripheral taste system have been largely ignored, perhaps due to the prevalence of the labeled-line model. After all, if taste perception purely depends on which pathways are activated, the exact timing of the response isn’t so important. I was therefore surprised to discover striking dynamics in taste responses that depend on the taste modality, organ, and stimulus.

Most taste sensory neurons, including sugar- and water-sensing neurons, show a sustained response during the taste presentation that rapidly diminishes upon taste removal. In contrast, some bitter-sensing neurons show strong, transient responses at both the onset (“ON”) and offset (“OFF”) of the taste stimulus. Intriguingly, the OFF response is often stronger than the ON response, meaning that these neurons are most robustly activated after the stimulus has disappeared. These bitter neurons thus encode taste in a fundamentally different way than other taste neurons.

 

I first investigated the mechanisms underlying these ON/OFF dynamics. Genetic manipulations revealed that the ON and OFF responses are generated cell-intrinsically, likely through the same bitter receptors. This suggests that bitter receptors have unusual transduction properties: they seem to elicit transient depolarization upon ligand binding, followed by a second wave of depolarization upon ligand unbinding.

Taste response dynamics influence synaptic plasticity

I next asked how these dynamics impact downstream neural circuits. Bitter taste signals are conveyed to a set of dopaminergic neurons in the higher brain that regulate aversive learning. Interestingly, these neurons showed the same ON/OFF dynamics as bitter sensory neurons.

Previous studies had shown that these dopamine neurons mediate aversive learning by inducing synaptic depression at specific synapses.^8-10 Using optogenetic stimulation, I found that dopaminergic activation can in fact depress or potentiate these synapses depending on the timing of activity. When bitter taste is used for aversive learning, the ON and OFF responses can thus drive plasticity in opposite directions. Unexpectedly, the OFF response is the dominant signal that determines the final synaptic state. These surprising results demonstrate that taste response dynamics have a dramatic impact on neural circuits for learning.

Complexity and flexibility in taste coding

From context to timing, my work has uncovered new dimensions of taste processing. Despite its visceral impact and hardwired underpinnings, taste defies our simple expectations and reveals a fascinating complexity. The flexibility and nuance inherent in taste coding may create the rich palette of flavors that we so enjoy while equipping us to adapt to changing internal and external worlds.

 

Interested in learning more? Read the acetic acid study here and the temporal dynamics paper here.

 

References:

1. E.R. Liman et al., Neuron (2014).
2. P.A. Parsons, Experientia (1980).
3. A.A. Hoffmann, P.A. Parsons, Biological Journal of the Linnean Society (1984).
4. A.V. Devineni et al., eLife (2019).
5. A.H. Jaeger et al., eLife (2018).
6. J-E. Ahn et al., eLife (2017).
7. A.V. Devineni et al., bioRxiv (2020).
8. T. Hige et al., Neuron (2015).
9. R. Cohn et al., Cell (2015).
10. J.A. Berry et al., Cell Reports (2018).