When fruit flies were put on a low-protein diet, they did something unexpectedly specific: they walked right past sugar and headed for the protein. A study published in Science in 2025 now explains why. A small signaling molecule produced in the gut, called CNMamide (CNMa), surges when essential amino acids run low and rewires the brain’s reward circuits within minutes, suppressing the appeal of carbohydrates while leaving protein appetite fully intact.
The discovery, led by researchers at Korea Advanced Institute of Science and Technology (KAIST), offers the clearest biological explanation yet for a pattern nutritionists have observed for years: animals and people on protein-poor diets do not just eat less. They overeat other macronutrients, as if the body is searching for something it cannot find.
A peptide with two express lanes to the brain
The team focused on enterocytes, the absorptive cells lining the gut, which act as frontline nutrient sensors. When flies were deprived of essential amino acids, their enterocytes ramped up production of CNMa dramatically. Earlier work, including a 2021 study published in Nature, had already identified CNMa as a peptide that rises during protein deprivation. The new paper goes further by mapping exactly how the signal reaches the brain and what it does when it arrives.
CNMa travels along two parallel routes. A fast neuronal pathway delivers the protein-deficit message to the brain in seconds. A slower hormonal pathway reinforces that message over longer periods. The dual-channel design acts as a fail-safe: the brain registers the shortfall quickly and sustains the behavioral shift until the animal actually consumes enough protein.
Once CNMa reaches the brain, it dampens a cluster of neurons that express a neuropeptide called DH44. Previous research established that DH44-positive neurons serve as post-meal amino acid sensors in Drosophila, firing when they detect amino acids after eating. By quieting those neurons, CNMa effectively turns down the reward value of sugar-containing food. The behavioral result was stark: flies with intact CNMa signaling consistently chose protein-dense options over carbohydrate sources, even when both were freely available.
The researchers confirmed causation, not just correlation, using targeted genetic tools. They knocked out CNMa production in specific gut cells and separately silenced or activated DH44 neurons, then measured feeding choices and neural activity in real time. Flies that could not produce CNMa failed to shift their preferences under protein restriction. Flies whose DH44 neurons were artificially suppressed behaved as though they were protein-deprived even on a balanced diet.
The mammalian parallel: FGF21 from the liver
Flies are not humans, and the most pressing question is whether a similar mechanism operates in mammals. No direct CNMa homolog has been confirmed in the human gut, though the Science paper includes cross-species observations in mice suggesting the pathway may be conserved. Detailed mammalian behavioral and neural data on CNMa itself have not yet been published separately, so any claim that the exact same molecule drives human cravings would be premature.
A related line of evidence, however, points in the same direction. FGF21, a hormone produced mainly by the liver during protein restriction, has been shown in mouse studies to act on brain reward circuits and shift motivation toward protein-containing foods. Research published in Molecular Metabolism demonstrated that FGF21 signaling in glutamatergic neurons is required for the metabolic and weight-loss responses that accompany dietary protein dilution. Separate conditional-knockout experiments, documented in open-access mouse data, showed that removing FGF21 receptors from specific neuronal populations prevented the usual increase in protein-seeking behavior.
The conceptual overlap is striking: in both flies and mice, a peripheral organ detects low protein, releases a signal molecule, and changes how rewarding different foods feel in the brain. But no published experiment has tested whether FGF21 and CNMa operate in the same cascade or represent independent systems that happen to converge on the same behavioral outcome. That gap matters. If the pathways are redundant, blocking one might not fully suppress the drive to seek protein. If they are sequential, intervening at a single step could have broader effects than expected.
What the evidence can and cannot support
The fly data are mechanistically strong. Targeted genetics, real-time neural imaging, and controlled feeding assays together provide a clear causal chain: essential amino acids fall, enterocytes detect the deficit, CNMa production rises, and the peptide engages brain circuits that selectively lower the appeal of sugar. In the controlled setting of the experiments, this chain was sufficient to redirect feeding toward protein without changing overall food availability. The effect was about choice, not simple hunger.
The mammalian FGF21 studies are similarly rigorous within their scope. Conditional knockouts in mice provide strong causal evidence that protein restriction changes brain reward signaling through a defined molecular pathway. Removing FGF21 receptors from specific neuronal populations blunted both protein-seeking behavior and some metabolic adaptations to low-protein diets.
What remains missing is the bridge. No study has simultaneously measured gut peptide levels and brain dopamine responses in the same mammalian subjects under controlled protein restriction. A key experiment to watch for: whether simultaneous elevation of FGF21 and a CNMa-like peptide produces a stronger suppression of sugar motivation than either signal alone, measurable through dopamine recordings in reward centers like the ventral tegmental area. Until that work is published, the gut-peptide story in flies and the liver-hormone story in mice remain compelling but unjoined.
Long-term outcome data are also absent. No study has reported what happens to body weight, total caloric intake, or macronutrient balance over weeks or months when CNMa or FGF21 signaling is chronically blocked in mammals. It is possible that short-term shifts in food choice wash out as other regulatory systems compensate, or that sustained manipulation of these signals carries unintended effects on mood, metabolism, or cognition.
What this means at the dinner table
The practical takeaway is narrow but worth noting. The biology described here suggests that when protein intake drops, the body does not simply broadcast a generic “eat more” signal. It sends a more targeted message: seek protein. Part of that targeting involves dialing down the relative appeal of sugary options. In environments where high-protein foods are scarce or expensive, that signal may instead manifest as persistent, hard-to-satisfy cravings and a tendency to overconsume whatever calories are available.
This fits with the “protein leverage” hypothesis in nutrition science, which holds that when diets are diluted in protein, people eat larger quantities to reach a biological protein target, inadvertently overshooting on fats and carbohydrates. The CNMa and FGF21 findings offer candidate mechanisms for how that leverage might work at the cellular level. They do not, however, prove that manipulating a single peptide will be a safe or effective way to control appetite in humans.
One question the research does not yet address is whether protein quality matters as much as quantity. Essential amino acids, the specific trigger for CNMa release in flies, are found in varying concentrations across animal and plant protein sources. Whether a diet adequate in total protein but low in one or two essential amino acids (common in some plant-based eating patterns) would trigger the same gut signal remains untested.
Where the research goes from here
For now, the most grounded implication is behavioral rather than pharmaceutical. Ensuring that meals contain adequate high-quality protein, roughly 25 to 30 grams per meal based on existing dietary guidelines, may help align internal protein-sensing circuits with external food choices, potentially reducing the drive to snack on sugary foods between meals. That suggestion is consistent with the experimental data in flies and mice, even if the precise molecules differ in humans.
As of June 2026, several research groups are expected to be testing whether CNMa-like peptides exist in the mammalian gut and whether they interact with FGF21 signaling. Those results will determine whether this elegant fly circuit is a universal feature of animal biology or a specialized solution that evolution reinvented differently in vertebrates. Either answer would reshape how scientists think about the gut’s role in food choice, and whether the next generation of appetite-targeting therapies should look beyond the stomach and liver to the intestinal lining itself.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
