UC Irvine

Poxvirus molecular structure has been a long-standing problem in structural virology. Poxviruses are large, enveloped, double stranded DNA viruses that complete their entire replication cycle in the host cytoplasm. Despite >60 years of investigation into the ultra- and molecular structure of vaccinia virus, the prototypical poxvirus, various factors have rendered its molecular structure refractory to traditional structural and molecular biology approaches. This dissertation describes new insights into the vaccinia virion molecular structure, achieved by protein crosslinking mass spectrometry (XLMS) combined with protein structure prediction by deep learning. To examine the vaccinia molecular structure in situ, we developed an XLMS approach implementing a “strategy of variation” by which we were able identify protein-protein interaction interfaces for almost all of the ~75 packaged virion proteins. As a part of this process, we developed an updated virus purification protocol for high yield-high vaccinia virus with proteomic purity, along with new methods for full vaccinia protein solubilization for proteomics, that was able to significantly improve the completeness of digestion of virion proteins to peptides for mass spectrometry. We also implemented bioinformatic and other strategies to maximize the identification of crosslinked peptides, finally allowing us to achieve what we consider likely to be a saturating XLMS dataset. Alongside this work, we applied a structural homology prediction approach (HHsuite) to identify homologs of vaccinia proteins and to hopefully generate “placeholder” models that we could integrate with XLMS to generate higher order three dimensional models. We applied structural homology prediction to other viruses from the nucleocytoplasmic large DNA virus (NCLDV) phylum to the identify orthologous genes between the virus families that were undetectable by sequence homology alone. This contributed substantially to expanding annotations of previously uncharacterized proteins from NCLDV proteomes. Confident structural homologs for core structural proteins of the vaccinia virion, however, were not identified. Instead, we eventually pursued protein structure prediction by AlphaFold2 to generate high confidence models of vaccinia virion proteins to combine with XLMS data. Through the resulting combination of XLMS based structure validation of AlphaFold2 models and crosslink-guided protein docking, we describe previously unidentified higher order structural assemblies of vaccinia virion transmembrane and surface-associated proteins. The density of crosslinks within and between molecules of Vaccinia virion core structural P4a (the major component of the palisade layer of the core wall) allowed us to address in greater depth its structure, maturation, and possible assembly pathway. This work was able to identify a likely order of events in which P4a precursor is first proteolytically processed at a downstream site to release its P4a-3 fragment. Removal of P4a-3 releases a steric block to a major rearrangement between the two domains of P4a-1, resulting in a final conformation that is stabilized by disulfide-locking. Removal of a second proteolytic fragment, P4a-2 and trimerization of P4a-1 then allows the assembly of P4a-1 trimers into a higher order hexagonal lattice as the palisade layer of the Vaccinia virus core wall. Overall, these findings contributed to our understanding of the Vaccinia mature virion molecular architecture and open the door to future studies of virion dynamics and morphogenesis. This approach was then extended to the 23 transmembrane and virion surface proteins associated with the virion envelope.