Structure of catalase determined by MicroED

Abstract

MicroED is a recently developed method that uses electron diffraction for structure determination from very small three-dimensional crystals of biological material. Previously we used a series of still diffraction patterns to determine the structure of lysozyme at 2.9 Å resolution with MicroED (Shi et al., 2013). Here we present the structure of bovine liver catalase determined from a single crystal at 3.2 Å resolution by MicroED. The data were collected by continuous rotation of the sample under constant exposure and were processed and refined using standard programs for X-ray crystallography. The ability of MicroED to determine the structure of bovine liver catalase, a protein that has long resisted atomic analysis by traditional electron crystallography, demonstrates the potential of this method for structure determination.

Keywords: MicroED; biochemistry; biophysics; catalase; cryo EM; electron diffraction; electron microscopy; micro crystals; none; structural biology.

Figure 1.. Diffraction from catalase microcrystals.

Representative untilted still diffraction pattern of a catalase microcrystal that shows sharp reflections extending to approximately 3.0 Å. Crystals of this quality were used to collect a data set by continuous rotation. Inset shows an example catalase microcrystal as seen in over-focused diffraction mode. Scale bar is 2 µm. The dimensions of most microcrystals varied between 6 and 20 µm in length, 2 and 8 µm in width, and the thickness was approximately 100–200 nm, in agreement with previous catalase crystal sizes (Dorset and Parsons, 1975b). DOI: http://dx.doi.org/10.7554/eLife.03600.002

Figure 2.. Final structure of catalase at 3.2 Å resolution determined by MircoED.

(A) The complete refined catalase structure shown as ribbons. The corresponding 2mF_obs-DF_calc density map around the complete structure can be seen in Video 2. (B) The density map, contoured at 1.5 σ, around a representative region of the structure (residues 178–193 of chain A) shows well-defined density around the model. (C) The mF_obs-DF_calc difference map, contoured at +2.5 σ (green) and −2.5 σ (red), around the same region shows no interpretable densities near the model indicating no obvious differences between the observed data and the calculated model. (D) Views of the density map (contoured at 1.0 σ) around a large region of the crystal lattice shows there is no significant density in the in the solvent channels of the catalase crystal. DOI: http://dx.doi.org/10.7554/eLife.03600.007

Figure 2—figure supplement 1.. Molecular replacement validation tests.

(A) The results of the single monomer MR show that all four chains of the tetramer could be properly placed. The RMSD between this MR solution and the original MR solution from complete tetramer search model was only 0.381 Å. (B–C) The result of MR with a polyalanine model (B) shows well-defined density extending beyond the alanine model where the proper side-chains could be placed (C). (D) The model (gray) and corresponding 2mF_obs-DF_calc density map (contoured at 1.5 σ) resulting from the autobuild procedure shows good agreement with our final refined model (blue) indicating the high quality of the maps from our data. DOI: http://dx.doi.org/10.7554/eLife.03600.008

Figure 3.. Final Model validation tests.

(A–D) To check for model bias in the final MicroED data (A and C), and to compare with data obtained from synchrotron X-ray diffraction (B and D), residues 181–185 (A–B) or the heme groups (C–D) from all four chains were removed prior to simulated annealing and re-refinement. In all cases, significant positive mF_obs-DF_calc difference density (contoured at ±2.0 σ) can be seen for the missing regions, which are displayed in yellow for clarity, indicating no significant model bias in the final structure. (E–H) When a model (PDB ID: 3RGP; gray) lacking NADP was used to phase the MicroED data or the synchrotron x-ray data, both of which contained NADP, significant density in the mF_obs-DF_calc difference map was observed although it appeared fragmented even at a lower contour level. This density could be fit with the NADP and the conformational change of residue F197 present in our final refined model is apparent. Maps presented in panels (E–H) are all unrefined, following MR. DOI: http://dx.doi.org/10.7554/eLife.03600.010