Descending neuron population dynamics during odor-evoked and spontaneous limb-dependent behaviors

Florian Aymanns, Chin‐Lin Chen, Pavan P Ramdya

TL;DR

This study uses advanced imaging techniques to record the activity of nearly 100 descending neurons (DNs) in behaving fruit flies, revealing how these neurons encode complex behaviors like walking and grooming. The findings suggest that while a large population of DNs is involved in locomotion, with specific subsets encoding turning, fewer DNs are dedicated to grooming, and some even encode odor information directly, offering a new perspective on how the brain controls movement.

What Problem Does This Paper Address?

Understanding how the brain orchestrates complex behaviors, like walking, is a fundamental challenge in neuroscience. In Drosophila melanogaster, this control is thought to be mediated by approximately 1100 descending neurons (DNs) that project from the brain to motor circuits. While activating a few DNs can trigger specific actions, the role of the larger DN population during natural behaviors remains unclear. Do they simply modulate existing commands, act as parallel pathways depending on sensory input, or perhaps encode sensory information directly? This paper aims to bridge this gap by investigating the population-level activity of DNs during spontaneous and odor-evoked behaviors, providing insights into their functional organization and contribution to motor control.

Methods

The researchers employed a sophisticated combination of techniques:

  • Genetically Encoded Calcium Imaging: They used a genetic strategy to restrict the expression of a calcium indicator (GCaMP6s) to descending neurons originating from the brain. This allowed them to optically record neural activity.
  • Two-Photon Microscopy: This high-resolution microscopy technique was used to image the axons of DNs as they passed through the thoracic cervical connective, a crucial junction between the brain and the ventral nerve cord.
  • Behavioral Tracking: Flies were tethered and walked on a spherical treadmill. Their movements were captured from multiple camera angles and analyzed using DeepFly3D for 3D pose estimation and FicTrac for inferring locomotor trajectories. A dilated temporal convolutional neural network (DTCN) was trained to classify various behaviors (walking, resting, grooming, etc.).
  • Olfactometry: A custom olfactometer was built to deliver controlled pulses of humidified air and specific odors (attractive ACV and aversive MSC) to the flies.
  • Data Analysis: Advanced statistical methods, including linear regression, principal component analysis (PCA), and kernel density estimation, were used to correlate neural activity with behavior and sensory input. They also developed methods to identify individual DNs based on their functional and spatial properties.

Datasets

  • Newly Generated Datasets: The study generated new datasets of descending neuron population activity and associated behavioral data from tethered, behaving Drosophila melanogaster. These datasets are available via Harvard Dataverse and GitHub.
  • Existing Datasets: The study also leveraged existing electron microscopy datasets of the Drosophila VNC and neck connective for morphological analysis and identification of specific neurons.

Key Insights and Ideas

  • DNs as Behavior Encoders: The study provides strong evidence that DN populations are not just simple command neurons but actively encode complex behaviors, with walking being the most prominently represented.
  • Spatial Segregation of Function: DNs encoding different behaviors, particularly walking and head grooming, appear to be spatially segregated within the cervical connective, suggesting a functional organization.
  • Turning vs. Speed Encoding: While many DNs are active during walking, their activity dynamics are more strongly linked to turning than to variations in walking speed, suggesting a primary role in steering.
  • Odor Encoding Independent of Behavior: A subset of DNs were found to encode odor identity (attractive vs. aversive) directly, even in the absence of overt behavioral responses, indicating a direct sensory-to-motor pathway for olfactory information.
  • Identifying Individual Neurons: The research demonstrates a powerful workflow for identifying specific DNs from population recordings by combining functional (activity patterns), topological (location in the connective), and morphological (shape) data, paving the way for detailed circuit dissection.

Key Results

  • Behavioral Encoding: Approximately 60% of recorded DNs encoded walking, while about 15% encoded head grooming. Very few DNs encoded resting, and none were found to encode posterior movements or front leg rubbing (though some overlap was observed due to sequential behaviors).
  • Turning Encoding: DNs could explain about 60% of the variance in turning behavior, significantly more than the ~30% they explained for walking speed.
  • Odor Encoding: Linear discriminant analysis showed that neural residuals could predict odor presence significantly better than behavioral information alone, indicating raw odor encoding. A few DNs uniquely encoded specific odors (MSC or ACV).
  • Spatial Organization: Walking-encoding DNs were distributed dorsally, while head grooming-encoding DNs were concentrated ventrally in the cervical connective. Turn-encoding DNs showed a clear ipsilateral enrichment.

Limitations and Future Work

  • Temporal Resolution: The authors acknowledge that the temporal resolution of calcium imaging (around 16 Hz) is lower than the stride frequencies of fly locomotion (up to 20 Hz) and leg movements during grooming (up to 7 Hz). This limits the ability to precisely link neural activity to individual joint angles and limb positions. They suggest using faster calcium indicators or voltage sensors, or advanced imaging techniques like light-sheet microscopy, or forcing slower movements.
  • SEZ DNs: The current genetic driver line did not express in the subesophageal zone (SEZ), which is known to contain DNs involved in grooming. Including these DNs would provide a more complete picture of grooming control.
  • Odor Specificity: While some DNs encode odors, the number of DNs is far smaller than the olfactory space, suggesting they encode odor categories (e.g., attractive vs. aversive) rather than individual odors with high specificity.
  • Future Work: The authors propose future studies to investigate synaptic connectivity of identified DNs, explore the role of SEZ DNs, and test whether the observed dichotomy between flexible locomotion and stereotyped grooming holds for other behaviors like flight. They also suggest exploring the potential for DNs to encode classes of odors rather than specific ones.

Summary updated

Abstract

Deciphering how the brain regulates motor circuits to control complex behaviors is an important, long-standing challenge in neuroscience. In the fly, <i>Drosophila melanogaster</i>, this is coordinated by a population of ~ 1100 descending neurons (DNs). Activating only a few DNs is known to be sufficient to drive complex behaviors like walking and grooming. However, what additional role the larger population of DNs plays during natural behaviors remains largely unknown. For example, they may modulate core behavioral commands or comprise parallel pathways that are engaged depending on sensory context. We evaluated these possibilities by recording populations of nearly 100 DNs in individual tethered flies while they generated limb-dependent behaviors, including walking and grooming. We found that the largest fraction of recorded DNs encode walking while fewer are active during head grooming and resting. A large fraction of walk-encoding DNs encode turning and far fewer weakly encode speed. Although odor context does not determine which behavior-encoding DNs are recruited, a few DNs encode odors rather than behaviors. Lastly, we illustrate how one can identify individual neurons from DN population recordings by using their spatial, functional, and morphological properties. These results set the stage for a comprehensive, population-level understanding of how the brain's descending signals regulate complex motor actions.
BibTeX
@inproceedings{aymanns2022descending,
  author = {Florian Aymanns and Chin‐Lin Chen and Pavan P Ramdya},
  title = {{Descending neuron population dynamics during odor-evoked and spontaneous limb-dependent behaviors}},
  booktitle = {eLife},
  year = {2022},
  doi = {10.7554/elife.81527},
}
Figure