overcoming limitations of Icosahedral reconstructions of viruses
Open Access
- Author:
- Dinunno, Nadia
- Graduate Program:
- Biomedical Sciences (PHD)
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- November 16, 2023
- Committee Members:
- Jianming Hu, Co-Chair & Major Field Member
Aron Lukacher, Major Field Member
Matthew Swulius, Outside Unit & Field Member
Joseph Wang, Co-Chair & Dissertation Advisor
Lisa Shantz, Program Head/Chair - Keywords:
- structural biology
- Abstract:
- The work that is presented in this thesis dissertation is composed of investigations of viruses of both icosahedral, enveloped, and helical composition which were purified directly from serum and in vitro media. Though structural virology is typically restricted to icosahedral investigations, which are efficient and unproblematic, this dissertation attempts to elucidate the structural details of protein-bound capsids, naked capsids, enveloped particles, and helical organizations that require tools beyond the typical 60-fold symmetry imposition. The following obstacles of viral reconstructions were faced including 1) icosahedral averaging artifacts, 2) density maps lacking sequence information, 3) and helical symmetry. Each problem was overcome in the following aims using a combination of 1) sub-volume classification and refinement 2) machine learning prediction and sequence blast 3) helical indexing and tomography. Multiple cryoEM software packages were employed to attempt to overcome key limitations of each approach. Towards AIM 1, the structure of Parvovirus B19 was isolated from patient serum for the first time to elucidate sub-3 angstrom resolution details while revealing novel protein-protein interactions and a stabilized N-terminal region of the major capsid protein. The protein-protein interactions with the capsid were identified as two acute phase reactants, alpha-1 chymotrypsin and ITIH4. These densities were identified directly from the density maps, without sequence information, and then supported with mass spectrometry of the biological sample. Upon structural interrogation, the ITIH4 residue was found to have bound directly to the N-terminal region of the major capsid protein, VP2. For the first time, all 20 amino acids were visualized and built into the sub-3 angstrom density map. Investigations of inter-subunit interactions demonstrated that inter-subunit hydrogen bonding is essential for the exposed state of the N-terminal region. This N-terminal region, furthermore, is internalized upon heating which leads to genome exposure. Towards AIM 2, the structure of two insect viruses, Aedes Birnavirus and Densovirus, were identified from contaminated CHIKV cultures. Due to their unexpected presence, the maps were lacking sequence information, which was overcome with new software developments in machine learning prediction of amino acids directly from density maps. The sub-3 angstrom resolution maps reveal both genetic material and deviations from close structural homologs. Towards AIM 3, the structures of WHV were solved directly from serum samples for the first time. To solve the capsid structure, particle subtraction and symmetric reconstructions were employed in Cryosparc. Unique surface contacts were identified as putative pre-S1 protein interactions that may be conserved across HBV and WHV. Helical indexing demonstrates inconsistencies with published structures and sub-tomogram averaging shows variable spike number per turn. The research presented in this thesis took place over 4 years and combined software across Relion, Cryosparc, and Phenix-Coot-Isolde packages.