Accelerated Discovery of Cell Migration Regulators Using Label-Free Deep Learning-Based Automated Tracking.

Area of Science:

• Cell Biology
• Biophysics
• Computational Biology

Background:

• Cell migration is crucial for development and disease (e.g., metastasis) but is challenging to study at scale.
• Existing cell migration assays lack throughput, single-cell resolution, or require cell labeling, limiting mechanistic insights.
• High-throughput analysis is needed to understand the complex molecular and environmental factors influencing cell motility.

Purpose of the Study:

• To develop a high-throughput, label-free method for analyzing single-cell migration.
• To establish a machine-vision and deep-learning workflow (DeepBIT) for automated cell tracking.
• To investigate molecular and extracellular regulators of cancer cell migration using the new assay.

Main Methods:

• Developed a 96-well plate imaging workflow for real-time cell migration observation.
• Created and validated DeepBIT, a deep-learning model for automated cell detection and tracking in time-lapse videos.
• Applied the assay to screen small-molecule inhibitors, extracellular matrix variations, and CRISPR knockouts affecting cancer cell motility.

Main Results:

• DeepBIT enabled tracking of ~1.3 million cells across 840 conditions in 70 hours, a task requiring ~5.5 years of manual tracking.
• The assay successfully identified previously unknown molecular regulators of cancer cell migration.
• Demonstrated that collagen content can alter the regulatory role of cytoskeletal molecules in cell migration.

Conclusions:

• The DeepBIT workflow significantly enhances throughput and reduces bias in cell migration studies.
• This method facilitates the discovery of novel regulators of cell motility and their interactions with the microenvironment.
• The findings provide new mechanistic insights into cancer cell migration and metastasis.