nERdy: network analysis of endoplasmic reticulum dynamics.

Area of Science:

• Cell Biology
• Biophysics
• Computational Biology

Background:

• The endoplasmic reticulum (ER) has a complex morphology crucial for cellular function.
• Understanding ER network dynamics and the role of ER shaping proteins requires advanced reconstruction methods.
• Current methods for ER network reconstruction are limited by parameter sensitivity or extensive data requirements.

Purpose of the Study:

• To develop novel computational tools for accurate reconstruction and analysis of the dynamic ER network.
• To investigate the role of ER shaping proteins in tubular matrix formation and junction dynamics.

Main Methods:

• Introduction of nERdy, an image processing approach for ER network extraction.
• Development of nERdy+, a D4-equivariant neural network for enhanced ER network representation.
• Analysis of live-cell confocal and STED microscopy time-series data.

Main Results:

• nERdy and nERdy+ accurately extract and represent ER networks and junction dynamics, outperforming existing methods.
• The methods distinguish tubular matrices from peripheral ER networks and analyze tripartite junction movement.
• Atlastin and Reticulon 4 were identified to promote dynamic tubular matrix formation and enhance junction dynamics.

Conclusions:

• nERdy and nERdy+ provide powerful tools for studying ER structure and dynamics.
• Novel roles for Atlastin and Reticulon 4 in regulating ER morphology and dynamics were uncovered.
• This work advances our understanding of ER shaping proteins and their contribution to cellular function.