In vivo virus structures: Simultaneous classification, resolution enhancement, and noise reduction in whole-cell electron tomography

Sulfolobus Turreted Icosahedral Virus (STIV) experiences an extra-cellular environment of near boiling acid (80 °C, pH 3) and particles purified under these conditions were previously analyzed by cryo electron microscopy and image reconstruction. Here we describe cryo-tomograms of Solfolobus cells infected with STIV and the maximum likelihood algorithm employed to compute reconstructions of virions within the cell. Virions in four different tomograms were independently reconstructed with an average of 91 particles per tomogram and their structures compared with each other and with the higher resolution single-particle reconstruction from purified virions. The algorithm described here automatically classified and oriented two different particle types within each cell and generated reconstructions of full and empty particles. Because the particles are randomly oriented within the cell, the reconstructions do not suffer from the missing wedge of data absent from the reciprocal-space tomogram. The fact that the particles have icosahedral symmetry is used to dramatically improve the signal to noise ratio in the reconstructions. The reconstructions have approximately 60 Å resolution (based on Fourier Shell Correlation analysis among reconstructions computed by the algorithm described here from four different tomograms).