How MIT cracked the code linking brain and behavior in a simple animal

Massachusetts Institute of Technology Researchers model and map how neurons in the tiny brain of C. elegans encode its behaviors, revealing many new insights into the strength and plasticity of its nervous system.

To understand the complex relationship between brain activity and behavior, scientists needed a way to map this relationship to all neurons in the entire brain. So far this has been an insurmountable challenge. But after inventing new techniques and methods for this purpose, a team of scientists at MIT’s Picauer Institute for Learning and Memory has produced an accurate computation of neurons in the tiny, controllable brain of a humble person. C. elegans A worm, showing how its brain cells encode nearly all of its basic behaviors, such as locomotion and feeding.

in the journal cell on August 21 The team presented new brain-level recordings and a mathematical model that accurately predicts the diverse ways neurons represent the worm’s behaviours. Applying this model to each cell specifically, the team produced an atlas of how most cells encode and the circuits in which the animal’s actions are involved. Thus, the atlas reveals the ‘logic’ behind how the worm brain produces a sophisticated and flexible repertoire of behaviours, even as its environmental conditions change.

Insights from the research

“This study provides a global map of how an animal’s nervous system is regulated to control behavior,” said senior author Stephen Flavell, associate professor in MIT’s Department of Brain and Cognitive Sciences. “It shows how many of the specific ganglia that make up an animal’s nervous system encode subtle behavioral features, and how this depends on factors such as the animal’s recent experience and current condition.”

Graduate students Jungsoo Kim and Adam Atanas, who each received their PhDs this spring for research, are the study’s lead authors. They’ve also made all of their data, model results, and atlas freely available to fellow researchers on a website called WormWideWeb.

A 2-minute excerpt from a typical neurological/behavioral dataset. The blue, orange and green dots are targets for tracking, which allowed the team to locate the worm’s head and keep the animal centered. A separate microscope view (not shown) that tracks the synchronous activity of each brain cell. Image credit: Flavell Lab/MIT Picware

Advanced techniques and notes

To make the measurements needed to develop their model, Flavell’s lab invented a new microscope and software system. This setup automatically tracks almost all of the worm’s behavior (moving, feeding, sleeping, laying eggs, etc.) and the activity of every neuron in its head (the cells are designed to flash when calcium ions build up). Reliably distinguishing and tracking discrete neurons as the worm wriggles and bends requires writing custom software, using state-of-the-art tools from machine learning. It has been shown to be 99.7 percent accurate in sampling the activities of individual neurons with a significant improvement in signal-to-noise compared to previous systems, the scientists report.

The team used the system to record the synchronized behavior and neural data from more than 60 worms as they walked around their dishes, doing whatever they wanted.

Data analysis revealed three new observations about neural activity in the worm: Neurons track behavior not only in the present moment, but also in the recent past. They adjust their coding of behaviors, such as movement, based on a surprising variety of factors; And many neurons simultaneously encode many behaviors.

For example, while the behavior of wriggling around a small lab dish may seem like a very simple act, neurons account for factors such as speed, orientation, and whether or not the worm is eating. In some cases, they represented an animal’s movement spanning time by about a minute. By encoding recent movement, not just current movement, these neurons can help the worm calculate how its past actions affect its current results. Many neurons also combine behavioral information to perform more complex maneuvers. Just as a human driver must remember to steer the car in reverse when going in reverse versus going forward, some neurons in the worm’s brain integrated the direction of movement of the animal and the direction of steering.

By carefully analyzing these kinds of patterns of how neural activity relates to behaviors, scientists have developed C. elegans Probabilistic neural coding model. Encapsulated in a single equation, the model explains how each neuron accounts for different factors to accurately predict whether and how neural activity reflects behavior. Approximately 60 percent of the neurons in the worm’s head are already responsible for at least one behavior.

When fitting the model, the research team used a probabilistic modeling approach that allowed them to understand how sure they were of each appropriate model parameter, an approach pioneered by co-author Vikash Mansingka, a principal research scientist who leads the Probabilistic Computing Project at MIT.

Atlas construction

By creating a model that could quantify and predict how any given brain cell would represent behaviour, the research team first collected data from neurons without tracking the specific identities of the cells. But the main goal of studying worms is to understand how each cell and circuitry contributes to behaviour. So to apply the model’s ability to each of the worm’s specific neurons, all of which had been previously mapped, the team’s next step was to correlate neural activity and behavior for each cell on the map. Doing so requires naming each neuron a unique color so that its activity can be linked to its identity. The team did this in dozens of freely moving animals, providing them with information about how nearly all of the specific neurons in the worm’s head relate to the animal’s behaviour.

The atlas resulting from this work revealed many insights, complete mapping of the neural circuits that control every animal behaviour. Flavell said these new findings will enable a more comprehensive understanding of how to control these behaviors.

“We were allowed to complete circles,” he said. “We hope that as our colleagues study aspects of neural circuit function, they can refer to this atlas to get a fairly complete view of the key neurons involved.”

Neuroplasticity

Another key result of the team’s work was the intriguing finding that while most neurons always obeyed the model’s predictions, a smaller group of neurons in the worm’s brain—about 30 percent of those encoding behavior—were able to flexibly remap their behavior. , performing fundamentally new roles. Neurons in this group were reliably similar across animals, and were well connected to each other in the worm’s synaptic wiring diagram.

Theoretically, these remapping events could happen for any number of reasons, so the team conducted more experiments to see if they could cause neurons to remap. As the worms wriggled around their plates, the researchers applied a laser zap that heated the agar around the worm’s head. The heat was harmless but enough to disturb the worms for a while, resulting in a change in the animal’s behavioral state that lasted for minutes. From these recordings, the team was able to see that many neurons correctly remapped their behavioral coding when the animals switched behavioral states.

“Behavioral information is richly expressed across the brain in many different forms — with distinct tunings, temporal scales, and levels of plasticity — that fit the specific classes of neurons in the brain. C. elegans neural network,” the authors wrote.

Reference: “Brain-level representations of behavior spanning multiple time scales and states C. elegansWritten by Adam A. Atanas, Gongsu Kim, Xiu Wang, Eric Bueno, McCoy Baker, Dae Kang, Jungyeon Park, Talia S. Kramer, Flossie K. Wan, Saba Pasquillo, Ugur Dag, Elbiniki Kalogeropoulou, Matthew A. Gomez, Casey Estrem, Nita Cohen, Vikash K. Mansingka and Stephen W. Flavell, August 21, 2023, Available Here. cell.
doi: 10.1016/j.cell.2023.07.035

In addition to Atanas, Kim, Mansingka, and Flavell, other authors of the paper are Xu Wang, Eric Bueno, McCoy Baker, Dee Kang, Jeongyeon Park, Talia Kramer, Flossy Wan, Saba Pasquello, Ugur Dag, Ilbeneke Kalogeropoulou, Matthew Gomez, Casey Estrem and Nita Cohen.

Research funding sources include National Institutes of Healththe National Science Foundation, the McKnight Foundation, the Alfred P. Sloan Foundation, the Picauer Institute for Learning and Memory, and the GPP Foundation.