A multi-scale map of cell structure fusing protein images and interactions

Area of Science:

• Cell biology
• Proteomics
• Systems biology

Background:

• Cells exhibit modular organization across multiple scales.
• Protein imaging and interaction data are typically analyzed separately.
• A unified approach is needed to map cellular architecture.

Purpose of the Study:

• To integrate diverse proteomic datasets for a hierarchical map of human cell architecture.
• To develop a machine learning-based calibration for integrating imaging and affinity purification data.
• To identify and validate novel subcellular systems and protein functions.

Main Methods:

• Integrated immunofluorescence images from the Human Protein Atlas with BioPlex affinity purification data.
• Developed a machine learning model to calibrate protein distance measures from imaging and interaction data.
• Performed additional affinity purifications to validate system associations and functions.

Main Results:

• Created the multi-scale integrated cell (MuSIC 1.0) map, resolving 69 subcellular systems.
• Identified approximately 35 previously undocumented subcellular systems.
• Discovered a pre-ribosomal RNA processing assembly, validated roles for SRRM1, FAM120C, and RPS3A in cellular processes.

Conclusions:

• Integrating multi-scale data enhances the resolution of cell imaging and provides spatial context for protein interactions.
• MuSIC 1.0 offers a framework for incorporating diverse proteomic data into comprehensive cell maps.
• This approach enables the discovery of novel cellular components and functions, advancing our understanding of cell biology.