Locating macromolecular assemblies in cells by 2D template matching with cisTEM

Area of Science:

• Cellular and Molecular Imaging
• Structural Biology
• Cryo-Electron Microscopy

Background:

• Understanding molecular mechanisms requires studying macromolecules within the cellular context.
• Electron cryo-tomography (cryo-ET) visualizes cellular structures at nanometer resolution but faces throughput limitations.
• Existing 3D template matching in cryo-ET struggles with high-resolution signal and false positives.

Purpose of the Study:

• To improve the throughput and accessibility of 2D template matching for analyzing macromolecular complexes in situ.
• To present a GPU-accelerated implementation of 2D template matching within the cisTEM software.
• To evaluate the performance of 2D template matching against 3D template matching for identifying cellular structures.

Main Methods:

• Developed a GPU-accelerated 2D template matching algorithm implemented in cisTEM.
• Applied the method to identify ribosomes in cryo-images of Mycoplasma pneumoniae.
• Benchmarked 2D template matching against 3D template matching using identical sample locations.

Main Results:

• The GPU-accelerated 2D template matching significantly improves computational efficiency and accessibility.
• High precision and sensitivity were achieved in identifying ribosomes within cellular images.
• Complementary strengths and weaknesses of 2D and 3D template matching were identified for localization and identity.

Conclusions:

• GPU-accelerated 2D template matching is a versatile tool for in situ visual proteomics and structure determination.
• This approach enhances the ability to study macromolecules in their native cellular environment.
• The findings offer valuable insights into the complementary nature of 2D and 3D template matching techniques in cryo-ET analysis.