Cryo-electron tomography (CET) provides unprecedented views into the native cellular environment at molecular resolution. While subtomogram analysis yields high-resolution native structures of molecular complexes, it also determines the precise positions and orientations of these macromolecules within the cell. Analyzing the geometric relationships between adjacent macromolecules can offer structural insights into molecular interactions and identify supramolecular ensembles. However, computational tools must be developed for quantitative analysis of this dense geometric information.
This thesis presents a statistical method for analyzing the local neighborhoods around macromolecules, with the aim of identifying 3D configurations of macromolecule pairs. This method of local geometric analysis emphasizes generality; it can be applied to any type of macromolecule and incorporates molecular symmetry. Here, the method was used to study the 3D organization of Ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) enzymes within the pyrenoid of C. reinhardtii cells. Subtomogram analysis identified RuBisCO complexes within pyrenoid tomograms, producing an in situ structure at 16 Angstrom resolution. Local geometric analysis of RuBisCO complexes suggested a fluid-like pyrenoid matrix. Predominant configurations of RuBisCO pairs were identified and combined into a geometric model of the unit cell of RuBisCO neighbors, showing a 3D configuration similar to closely packed spheres.
Next, this thesis progresses from local geometric analysis to identification of higher-order structures by presenting a method to detect polysomes, flexible supramolecular ensembles of ribosomes and messenger RNA. Local geometric analysis extracted the 3D arrangements of neighboring ribosomes in polysomes. This prior information is then used in the detection method, where ribosomes are represented as nodes in a graph and clustered by a Markov random field to reveal polysomes. Performance evaluation on synthetic and experimental tomograms of bacterial lysate indicated a 96% prediction accuracy. Finally, the method was applied to tomograms of rough microsomes derived from the endoplasmic reticulum (ER) of mouse myeloma cells, with the aim of detecting cytosolic and ER-associated polysomes.
This thesis presents geometry-based methods for analyzing the local organization of macromolecules and detecting supramolecular structures in CET. While these methods operate on macromolecule distributions of a single type, they could be expanded to incorporate multiple classes of macromolecules, enabling quantitative analysis for visual proteomics. Thus, they represent a step towards the spatial dissection of the complete macromolecular sociology of cells.
«
Cryo-electron tomography (CET) provides unprecedented views into the native cellular environment at molecular resolution. While subtomogram analysis yields high-resolution native structures of molecular complexes, it also determines the precise positions and orientations of these macromolecules within the cell. Analyzing the geometric relationships between adjacent macromolecules can offer structural insights into molecular interactions and identify supramolecular ensembles. However, computation...
»