Two-Dimensional Template Matching (2DTM) enables accurate identification and probing of macromolecular structure within in situ cryo-EM datasets

TL;DR

A new cryo-EM method called 2DTM helps identify and study macromolecules in their natural cellular environment with high accuracy.

Contribution

The paper introduces MOSAICS and Leopard-EM, which improve 2DTM for detecting low-abundance macromolecules in situ.

Findings

MOSAICS can detect structural differences of <10 kDa in populations of fewer than 150 molecules.

Leopard-EM is a GPU-accelerated software package for 2DTM analysis.

Abstract

Cryogenic electron microscopy (cryo-EM) and tomography (cryo-ET) are powerful tools to visualize biological structure and heterogeneity at the molecular level. Analyzing in situ data, as opposed to purified in vitro samples, is complicated by the crowded and low-contrast nature of cellular environments. Two-dimensional template matching (2DTM) is an emerging in situ cryo-EM data analysis technique sensitive enough to locate and orient molecules within cells based on a reference template. Importantly, 2DTM maintains contextual information allowing inference into biologically relevant structural heterogeneity, native interactions, and cellular localization. Here, I will cover 2DTM as a cryo-EM data analysis method focusing on improvements and extensions the Lucas Lab has recently made to 2DTM. For example, we have developed a new method called MOSAICS (Molecular in situ Atomic Coordinate Scanning) as a quantitative method for comparing reference structures with experimental cryo-EM data. We identify significant differences of <10 kDa with MOSAICS between populations of less than 150 molecules in the cell—two orders of magnitude fewer than required for classical 3D reconstruction— thereby enabling structural characterization of low abundance macromolecular states in situ. These advancements have been enabled by a new software package the lab has developed: Leopard-EM (Location and Orientation of Particles found using Two-Dimensional Template Matching), an extensible and open-source Python package for GPU-accelerated. Creating flexible and performant software tools like Leopard-EM and MOSAICS, which build upon 2DTM, will help advance the fields of cryo-EM and in situ structural biology.