Automated classification of synaptic vesicles in electron tomograms of C. elegans using machine learning

Area of Science:

• Neuroscience
• Cell Biology
• Biophysics

Background:

• Synaptic vesicles (SVs) are crucial for neuronal communication, with distinct types like clear core vesicles (CCVs) and dense core vesicles (DCVs) identified morphologically.
• Dense core vesicles (DCVs) store neuropeptides and are implicated in various C. elegans behaviors, but their precise functions and roles in synaptogenesis require further elucidation.

Purpose of the Study:

• To develop and validate an automated machine learning approach for classifying synaptic vesicles (CCVs and DCVs) in electron tomograms.
• To investigate differences in CCV and DCV distribution and proximity to active zones (AZs) between C. elegans dauer larvae and young adult hermaphrodites.

Main Methods:

• Utilized machine learning combined with an extended ImageJ macro workflow (3D ART VeSElecT) for vesicle segmentation and classification.
• Analyzed electron tomograms of C. elegans neuromuscular junctions (NMJs) to distinguish CCVs and DCVs based on image features.
• Quantified the fraction of DCVs and their mean distance to AZs in different developmental stages.

Main Results:

• Successfully developed an automated method to reliably distinguish CCVs and DCVs in C. elegans NMJ electron tomograms.
• Observed a higher proportion of DCVs and a greater mean distance between DCVs and AZs in dauer larvae compared to young adult hermaphrodites.
• Demonstrated the adaptability of the machine learning tools for studying diverse synaptic vesicle pools across different model organisms.

Conclusions:

• The automated classification of synaptic vesicles provides a powerful tool for quantitative analysis of vesicle populations.
• Developmental stage significantly impacts the distribution and localization of dense core vesicles relative to active zones.
• This machine learning-based approach facilitates the study of synaptic vesicle dynamics and function in various biological contexts.