Whole-brain, all-optical interrogation of neuronal dynamics underlying gut interoception in zebrafish

Weiyu Chen, Ben James, Virginia M. S. Ruetten, Sambashiva Banala, Ziqiang Wei, Greg Fleishman, Mikail Rubinov, Mark C. Fishman, Florian Engert, Luke D. Lavis, James E. Fitzgerald, Misha B. AhrensbioRxiv 2025 · DOI

TL;DR

Researchers have developed a novel all-optical system to simultaneously image whole-brain neuronal activity and deliver nutrients to the gut in awake, behaving zebrafish. This breakthrough allows for unprecedented insight into how the brain processes internal bodily signals, integrating them with external sensory information to guide behavior.

What Problem Does This Paper Address?

Understanding how our brains process internal bodily signals (interoception) and how these signals influence behavior has been a major challenge. Traditional methods struggle to capture whole-brain activity at a cellular level while simultaneously perturbing internal physiology in a behaving animal. Existing techniques either lack the spatial or temporal resolution, or are too invasive. This paper tackles the need for a minimally invasive, high-resolution, and temporally precise method to study the complex interplay between the gut and the brain.

Methods

The core innovation is an all-optical system that combines:

  • Whole-brain light-sheet microscopy: This technique allows for rapid, single-cell resolution imaging of the entire zebrafish brain.
  • Optical uncaging: Chemically "caged" nutrients (like glucose and glutamate) are delivered to the zebrafish gut via microgavage. A targeted UV laser then precisely releases these nutrients within specific gut regions, triggering neuronal responses. This method avoids the mechanical disruptions of traditional feeding or surgical interventions.
  • Virtual reality environment: The zebrafish are placed in a virtual reality setup that allows for controlled visual stimulation and real-time tracking of their swimming behavior. This enables the simultaneous study of interoception, exteroception (external senses), and behavior.
  • Advanced regression analysis: A "fused distributed lag ridge regression" model is used to disentangle the contributions of gut stimulation, visual input, and motor activity to individual neuron activity. This method allows for the estimation of unique response kernels for each neuron and stimulus modality.

Datasets

Not explicitly discussed as "datasets" in the traditional machine learning sense. The study utilizes data from multiple larval zebrafish experiments (N=6 for glucose, N=4 for L-glucose and fish water controls). The generated data consists of whole-brain calcium imaging recordings and behavioral data from these experiments.

Key Insights and Ideas

  • Gut signals are broadly represented: Nutrient delivery to the gut elicits widespread neuronal activity across many brain regions, not just those traditionally associated with visceral processing.
  • Temporal dynamics matter: Neuronal responses to gut signals unfold across multiple timescales, suggesting complex processing and integration.
  • Integration of internal and external cues: Gut-sensitive neurons are not solely dedicated to processing internal signals; many also respond to visual and motor stimuli, highlighting the brain's need to integrate diverse information for adaptive behavior.
  • Regional specialization in integration: Different brain regions appear to specialize in integrating specific types of information. For example, brainstem areas seem to integrate gut and motor signals, while midbrain regions integrate gut and visual signals.
  • Topographic mapping of gut responses: The brain can distinguish between nutrient delivery to different parts of the gut (foregut vs. midgut), suggesting a topographic representation of gut signals.

Key Results

  • Nutrient-specific responses: Amino acids (glutamate) and D-glucose elicited strong, time-locked neuronal responses, while L-glucose (a metabolically inactive enantiomer) did not, confirming the specificity of the gut sensing mechanism.
  • Whole-brain mapping of gut-sensitive neurons: The study generated the first whole-brain map of neuronal dynamics underlying gut interoception at single-cell resolution.
  • Multimodal integration: Approximately 26% of gut-sensitive neurons showed improved prediction accuracy when visual and motor information were included in the analysis models, indicating multimodal integration.
  • Signal and noise correlations: High signal correlations were observed between gut-responsive neurons across distant brain regions, suggesting similar representations. Noise correlations were higher within specific brain regions (DVC, PBN, DHC) compared to control populations, hinting at tightly coupled functional networks.
  • Temporal diversity: Neurons exhibited a wide range of response decay times, from seconds to over a minute, with all brain regions containing neurons tiling this temporal spectrum.

Limitations and Future Work

  • Larval zebrafish model: While powerful, larval zebrafish may not fully recapitulate all aspects of adult mammalian interoception. The authors suggest extending this paradigm to adult fish (e.g., Danionella cerebrum) and potentially other vertebrates.
  • Chemical byproducts: While controls were performed, the potential for subtle artifacts from uncaging byproducts remains a consideration.
  • Future directions: The authors envision using this platform to study the developmental trajectory of interoception, investigate how internal states influence complex cognitive behaviors, and further explore the neural basis of brain-body communication. The ability to uncage a wide variety of compounds opens doors to studying responses to different nutrients, toxins, and hormones.

Summary updated

Abstract

Abstract Internal signals from the body and external signals from the environment are processed by brain-wide circuits to guide behavior. However, the complete brain-wide circuit activity underlying interoception—the perception of bodily signals—and its interactions with sensorimotor circuits remain unclear due to technical barriers to accessing whole-brain activity at the cellular level during organ physiology perturbations. We developed an all-optical system for whole-brain neuronal imaging in behaving larval zebrafish during optical uncaging of gut-targeted nutrients and visuo-motor stimulation. Widespread neural activity throughout the brain encoded nutrient delivery, unfolding on multiple timescales across many specific peripheral and central regions. Evoked activity depended on delivery location and occurred with amino acids and D-glucose, but not L-glucose. Many gut-sensitive neurons also responded to swimming and visual stimuli, with brainstem areas primarily integrating gut and motor signals and midbrain regions integrating gut and visual signals. This platform links body-brain communication studies to brain-wide neural computation in awake, behaving vertebrates.
BibTeX
@misc{chen2025whole,
  author = {Weiyu Chen and Ben James and Virginia M. S. Ruetten and Sambashiva Banala and Ziqiang Wei and Greg Fleishman and Mikail Rubinov and Mark C. Fishman and Florian Engert and Luke D. Lavis and James E. Fitzgerald and Misha B. Ahrens},
  title = {{Whole-brain, all-optical interrogation of neuronal dynamics underlying gut interoception in zebrafish}},
  year = {2025},
  doi = {10.1101/2025.03.26.645305},
  howpublished = {bioRxiv},
}
Figure