Project Summary
The HIV-1 capsid protects the viral genome from host innate sensors and mediates the nuclear import of pre-
integration complexes. At some point during intra-cellular trafficking the capsid must disassemble to release the
reverse transcribing vDNA for its integration into host-genes. The underlying mechanisms and the cellular
location of capsid disassembly, which is commonly referred to as uncoating, remain unclear. Recent evidence
points to the possibility to transport a few apparently intact capsids (dimensions ~30 x 60 nm) through a dilated
(~60 nm) nuclear pore complex and into the nucleus1-3. The relevance of these few capsids to infection remains
unclear, and their prior dynamics inside cells unknown. An unbiased and detailed structural analysis of relevant
intra-cellular HIV-1 uncoating structures by cryo-EM remains highly challenging. Notably, in situ cryo-EM
necessitates the use of cryo-focused ion beam (cryo-FIB) milling to generate a thin (100-300nm) sheet of vitreous
ice referred to as a ‘lamella’ inside rapidly frozen cells. Procedures to precisely control the location in a cell where
the lamella is prepared remains underdeveloped and a bottleneck for in situ structural investigation, especially of
sparsely located HIV-1 specimen. The scientific premise of this proposal is to develop a robust 3D correlative light
and cryo-EM (3D-CLEM) workflow to facilitate cryo-FIB ‘lamella’ preparation at precise intra-cellular sites containing
HIV-1, and to characterize relevant capsid structures and correlate to their functional dynamics by live-cell imaging.
This project will leverage, (1) fluorescence live-cell imaging of HIV-1 infection to locate relevant capsids that are
respectively poised for nuclear entry and integration4, 5 and correlate their dynamics to capsid structures determined
by 3D-CLEM, cryo-FIB lamella preparation and high-resolution cryo-electron tomography (cryo-ET); (2) develop
automated methods to locate HIV-1 macrostructures on a lamella using low-resolution cryo-ET maps via a
convoluted neural network (CNN)-trained software6 to un-biasedly pick capsid templates and 3D-HIV structures on
a lamella, for high-resolution cryo-ET structural imaging and classification. The development and validation of the
3D-CLEM workflow and automated macrostructure localization procedure will improve throughput in cryo-EM
imaging of HIV-1 structures inside cells, and improve our understanding of virus cell biology, which is of high
public health significance.
