A Rigorously Validated De Novo Atomic Structure from a DE Camera

Posted on September 7, 2014 in News

How can I get a rigorously-validated atomic model of a previously-unknown structure using cryo-EM?

Using a DE camera from Direct Electron, Wah Chiu and colleagues have recently published a rigorously validated, de novo cryo-EM atomic structure of brome mosaic virus (BMV), a small ssRNA virus. The paper was released on 4 September 2014 in Nature Communications.

According the paper, “advances in electron cryo-microscopy have enabled structure determination of macromolecules at near-atomic resolution. However, structure determination, even using de novo methods, remains susceptible to model bias and overfitting… This study demonstrates a practical approach to obtain a rigorously validated atomic resolution electron cryo-microscopy structure.”

Data for the study was collected using a DE-12 camera (Direct Electron) on a 300 kV JEM-3200FSC (JEOL) transmission electron cryo-microscope (cryo-TEM). Using “movie-mode” data acquisition from the DE-12, the study’s authors used a new protocol to acquire and computationally filter various subsets of “movie” frames to generate optimized (high-contrast and high-resolution) images, with electron exposures beyond the conventionally-accepted radiation damage limit for a biological specimen. Direct Electron pioneered this technique (called “damage compensation”) in early 2013, and the present paper is its first detailed description and high-resolution application. The high-quality data from the DE-12 yielded a density map with a reported resolution of 3.3 and 3.8 Å, with and without subunit averaging, respectively.

The resulting density map was rigorously validated using various statistical measures, including the gold-standard Fourier shell correlation (FSC) and image phase randomization of Fourier components beyond 10 Å. Furthermore, the resulting atomic model (which was generated using a novel real-space optimization procedure) was validated using MolProbity statistics scores, a universally-accepted metric for assessing the quality of atomic models deposited in the PDB. Impressively, the final atomic model of BMV ranks in the 99th percentile when compared to PDB structures at an equivalent resolution. In fact, the cryo-EM map and model exceeds the resolution and quality of an earlier X-ray crystal structure of the same specimen.

According to Corey Hyrc, one of the study’s lead authors, “The DE-12 Camera System allowed for us to visualize our specimen at a level that we had not seen before. Obtaining the protein fold of the capsid was straight-forward and the technology allowed us to optimize a molecular model at the amino acid level. When compared to the previously resolved X-ray crystallographic structure, the high-resolution features in the cryo-EM map are comparable, and in some areas, significantly better. With these structural features, we could optimize the molecular model to the cryo-EM density using a new refinement protocol. The resulting model had stereochemical scores that were vastly better than the X-ray crystal structure.”

The paper concludes by stating that “the ultimate goal of a high-resolution cryo-EM study is to derive all-atom models for the macromolecular components of the assembly that pass the rigorous validation metrics routinely applied to X-ray crystallographic structures. This study integrates a number of recently developed technologies, including a direct electron detector, image processing using data beyond conventional radiation damage limits, map validation and resolution assessment using multiple indices, and de novo all-atom modeling, refinement and validation.”

The methods described in the paper promise to be very useful for cryo-EM researchers studying a wide-variety of specimens with unknown structure. The paper describes a protocol to derive an atomic model of a macromolecular assembly with minimal prior knowledge of the structure throughout the entire workflow. Moreover, it will be of considerable interest to the many groups now considering investment in direct detection technology due to the substantially reduced cost and ease-of-use of DE integrating-mode cameras compared to alternatives.

Regarding the impact of the DE camera, Wah Chiu, the study’s principle investigator, said, “The Direct Electron camera allows us to record high-throughput and high-resolution data of beam-sensitive particles in an electron microscope. This new technology enables biologists to quickly derive atomic resolution structures of biological nanomachines. The open collaboration between our NIH-NIGMS supported cryo-EM center and Direct Electron has made this development possible, benefiting the scientific community.”

The open-access publication is available for free from
Nature Communications.

The study was a collaborative effort between Baylor College of Medicine, Direct Electron, Lawrence Berkeley National Laboratory, Indiana University, and University of California at Berkeley. The study was supported by the US National Institutes of Health, the US Department of Energy, the Gulf Coast Consortia, and the Robert Welch Foundation.