Cellular function is governed by the structure, location, and dynamics of its densely packed molecular machines. While cryo-electron tomography (cryoET) provides three-dimensional snapshots of these native environments, their robust annotation remains challenging. I will show how high-confidence 3D template matching (hcTM) for cryoET, coupled with multiscaling, enables us to model and simulate biological landscapes directly from cryoET data. The hcTM reliably localizes a wide range of macromolecular complexes in crowded cytoplasm and nuclei, producing spatial maps that are the bases to build simulation-ready models of subcellular environments. Using this framework, we uncovered vault particles loaded with ribosomal cargo and resolved the molecular architecture of human chromatin in situ. We then applied it to an autophagy-related condensate directly in cells, demonstrating that a single amino-acid substitution strengthens intermolecular interactions, drives a liquid-to-glass mesoscale transition, and abolishes autophagic clearance. We thereby link a molecular alteration to changes in condensate material properties and biological function. Integrating in situ imaging with multiscale simulation provides a pipeline for visual proteomics and sets the stage for simulating cellular processes in their native context.
Cruz-León et al. (Sun,) studied this question.