Deep Learning Reveals Neural Hierarchy in C. elegans Mating

By Soham NandiReviewed by Susha Cheriyedath, M.Sc.Mar 15 2024

In an article published in the journal PNAS Nexus, researchers explored decision-making in neuromuscular circuits during mating behavior in Caenorhabditis elegans (C. elegans) males. Using deep learning and calcium imaging, the authors analyzed posture and muscle activities to identify behavioral modules.

They uncovered previously unknown subtypes of copulatory behavior and highlighted the role of serotonergic neurons in muscle coordination. The findings suggested a hierarchical structure in mating behavior, revealing insights into the neural basis of reproductive actions.

Background

Understanding the neural basis of mating behavior in C. elegans poses challenges due to individual variability and the complexity of simultaneous neuronal activity assessment. Previous studies extensively characterized male mating behavior, revealing distinct steps involving tail movements and spicule insertion. However, while pioneer work identified essential genes and neural structures governing mating, significant individual behavioral variability remained largely unexplored.

Despite controlled lab conditions, mating behavior in wild-type males exhibited wide statistical variance, suggesting responsiveness to external and internal factors. The connectome's extensive connectivity provides a structural basis for this variability, but tracking neuronal activity remains challenging. While calcium imaging shows promise, identifying over 300 tiny neurons presents technical hurdles. Unlike neurons, C. elegans male muscles are distinguishable, offering an alternative approach to understanding mating behavior.

The present study bridged previous gaps by generating transgenic worms expressing calcium sensors in muscles and developing software to automatically quantify posture and muscle activities during mating. By modeling these data, the authors identified stereotyped motor patterns and revealed unexpected behavioral modules, including those affected by serotonergic neurons. This comprehensive analysis uncovered the hierarchical organization of mating behavior and shed light on the role of serotonergic neurons in decision-making during mating, addressing previous limitations and advancing our understanding of neural mechanisms underlying reproductive actions in C. elegans.

Materials and Methods

In this study, researchers aimed to dissect the complex mating behavior of C. elegans males using advanced imaging techniques and computational analysis. They first generated transgenic worms expressing calcium sensors in specific muscles and structures involved in mating. These transgenic strains allowed simultaneous recording of muscle activities and posture changes during mating.

To accurately quantify these data, the researchers developed a software package called neural network-based automatic worm analyzer (NAWA), which automatically extracted posture and muscle activity information from recorded videos. To train NAWA, a large dataset of annotated images depicting worm postures and muscle activities was created. This dataset was used to construct an artificial deep neural network that accurately predicts worm posture and muscle activity from recorded videos.

Additionally, an algorithm was developed to reconstruct the skeleton of the worms from the predicted body parts labels, allowing for precise analysis of worm movements. The researchers then applied NAWA to analyze mating behaviors in wild-type and genetically modified C. elegans males. They identified stereotyped motor patterns and behavioral modules involved in different steps of mating. They found that multiple behavioral modules were involved in certain mating steps, and these modules were organized into bi-module repeats that form the routines used by males during mating.

Overall, the authors provided a comprehensive analysis of C. elegans mating behavior, shedding light on the neural mechanisms underlying reproductive actions in this model organism. Combining transgenic strains, advanced imaging techniques, and computational analysis tools offered a powerful approach to studying complex behaviors in C. elegans and other organisms. 

Results

Transgenic strains expressing fluorescent markers in muscles allow for the monitoring of muscle activity and posture changes during mating. NAWA automated posture and muscle activity analysis from fluorescent images, achieving high accuracy in modeling wild-type and mutant males' datasets. The hierarchical model, segmented behavior into nondescriptive modules, using an autoregressive-hidden Markov model (AR-HMM), revealing nine major behavioral modules, including locomotion, backing/turning, and prodding.

Statistical analyses highlighted distinct characteristics of each module, such as locomotion direction, posture dynamics, and variability within and among module blocks. Notably, two prodding modules exhibited differential usage between wild-type and hypo-ser mutant males, with module 10 showing coordinated bending between segments and module 23 demonstrating higher muscle activity. Further investigation revealed males executed mating behavior by transitioning among bi-module blocks, with specific patterns observed in locomotion, backing, and prodding behaviors. Hypo-ser males exhibited altered usage patterns of bi-module repeats, indicating changes in locomotion and backing behaviors.

Discussion

The study presented a method to dissect the highly variable mating behavior of C. elegans males using posture and muscle activity analysis. By manually tracking fluorescently labeled animals, a library of more than 100 recordings was generated, enabling the extraction of posture and muscle activity data. Utilizing the NAWA software package, posture and muscle activity were automatically analyzed, demonstrating scalability and robustness.

The hierarchical AR-HMM model identified distinct behavioral modules and revealed the prevalence of bi-module repeats in mating behavior. Additionally, hypo-ser mutant males exhibited altered behavior, highlighting the role of serotonin in mating. Despite the complexity of C. elegans mating, the approach offered a systematic means of dissecting behavior, potentially uncovering novel features such as the "Molina maneuver." The system's flexibility allowed for adaptation to different genetic backgrounds or research questions, promising further insights into C. elegans behavior.

Conclusion

In conclusion, the researchers advanced our understanding of mating behavior in C. elegans males by revealing its hierarchical organization and the role of serotonergic neurons. Through deep learning and calcium imaging, behavioral modules were identified, shedding light on decision-making in neuromuscular circuits.

The developed methodology offered a scalable and robust approach to dissecting complex behaviors, promising further insights into the neural mechanisms underlying reproductive actions in C. elegans. This comprehensive analysis laid the foundation for future studies exploring additional features of mating behavior and its modulation by genetic and environmental factors.