Technical Highlight - May 2015
Short description: Two methods for macromolecular structure determination via near-atomic resolution cryo-EM help fill a resolution gap.
In just the past few years, cryo-electron microscopy (cryo-EM) technology has made amazing strides, largely spurred by the development of highly sensitive direct electron detector cameras. Researchers can now do what was previously unthinkable—solve macromolecular structures at resolutions as good as 3 Å without crystallization.
This so-called resolution revolution in cryo-EM brings a need for new tools for building, refining and validating macromolecular structure models. Current tools for structure determination using cryo-EM data are optimal for resolution in the 5–10 Å range, and are not adequate for higher-resolution data. Tools for structure determination using X-ray crystallography data are optimized for very high-resolution data, in the 1–3 Å range, and tend to perform poorly for medium-resolution data.
Baker, DiMaio and colleagues address this resolution gap with two methods that take advantage of the powerful Rosetta modeling platform and offer complete packages for model building, refinement and validation from 3–5 Å resolution cryo-EM data.
The method reported in DiMaio et al. was designed to solve structures starting from a previously solved template homologous structure. The approach is based on a Monte Carlo sampling algorithm that fits the electron density map with backbone fragments collected from the PDB. This method alternates between model rebuilding and Rosetta-based all-atom refinement, until an optimal fit of the structure model to the electron density map is achieved. The authors also provide a cross-validation metric for assessing model accuracy.
Wang et al., on the other hand, present a de novo algorithm for structure determination from near-atomic resolution cryo-EM data that does not require a starting model. This method includes three main steps: first, sequence-based local backbone conformations are matched into the electron density map and scored; second, a subset of these fragment matches that optimize the scoring function is identified; and third, a partial model is assembled. The process is iterated until 70% of the sequence is covered, and finally the model is completed through rebuilding and all-atom refinement.
DiMaio et al. demonstrated good performance of their approach for three systems, including a challenging PrgH/PrgK ring complex. Wang et al. were successful in solving six of nine structures, including a large heterodimeric complex, VipA-VipB. The software tools for performing both methods are available at https://www.rosettacommons.org. Together, these methods help close the resolution gap for solving near-atomic resolution cryo-EM structures, whether a good homology model is already available or not.
F. DiMaio et al. Atomic-accuracy models from 4.5-Å cryo-electron microscopy data with density-guided iterative local refinement.
Nat Methods. (2015). doi:10.1038/nmeth.3286
R.Y.-R. Wang et al. De novo protein structure determination from near-atomic-resolution cryo-EM maps.
Nat Methods. (2015). doi:10.1038/nmeth.3287