The determination of subnanometer resolution structures of macromolecules and complexes is now possible by cryo electron microscopy (cryoEM) single particle reconstruction. I will present three examples of cryoEM structures with resolutions in the range of 6 to 10Å. These are structures of human adenovirus, which is a 150MDa icosahedral virus; an engineered small heat-shock protein assembly of 850kDa with octahedral symmetry; and the DNA-dependent protein kinase catalytic subunit (DNA-PKcs), which is a large asymmetric protein of 469kDa.
In the case of adenovirus a pseudo atomic capsid can be built by fitting the crystal structures of the two main capsid proteins, hexon and penton base, into the cryoEM density. The visualization of a-helices as density rods is confirmed by comparison of the cryoEM density with the positioned atomic resolution structures of hexon and penton base. Alpha-helical density regions are observed for other capsid proteins for which we do not have atomic resolution structures or good homology models. The current challenge is to build an atomic model for these additional proteins using de novo protein structure prediction methods guided by the cryoEM density.
For the engineered small heat-shock protein variant HSP16.5-P1, the cryoEM structure revealed an expanded octahedral protein shell with twice the number of subunits compared to the wild-type form of HSP16.5. The N-terminal region of both HSP16.5 and HSP16.5-P1 is important for substrate binding; however structural disorder in this region has prevented building atomic models for these regions from x-ray crystallographic data. EPR provides distance and solvent accessibility information for specific labeled residues and a structural “fingerprint” for a particular protein fold. The challenge we are currently facing is to build atomic models for the N-terminal region based on the observed cryoEM density and EPR spin labeling data.
DNA-PKcs regulates the non-homologous end joining pathway for repair of double-stranded DNA breaks. It is predicted to have stretches of a-helical HEAT (huntingtin-elongation-A-subunit-TOR), TPR (tetratricopeptide repeats), and PFT (protein farnesyl transferase) repeats, as well as a PI3K-like kinase domain. The present challenges are how to dock homology models for a-helical motifs that may not have the correct superhelical arc; and how to evaluate the docking of kinase domain homology models when there may significant differences in the folds of the DNA-PKcs kinase domain and the closest available homology models. Solutions for these computational challenges would greatly expand our ability to interpret near atomic resolution cryoEM structures and would provide a wealth of information on biological macromolecules and functional domains that may not be amenable to traditional atomic resolution structure determination.
Back to Image Analysis Challenges in Molecular Microscopy