Structure of catalase determined by MicroED.

Catalase crystallization and sample preparation

Catalase was recrystallized from a commercial aqueous suspension of catalase (C100; Sigma–Aldrich, St. Louis, MO) by first centrifuging the crystalline suspension and dissolving the pellet in 1.7 M NaCl. The solubilized catalase was then centrifuged and the supernatant was dialyzed against 50 mM sodium phosphate pH 6.3 overnight at 4°C. Crystals were removed from dialysis, stored in an Eppendorf tube, and incubated an additional 24 hr at 4°C. Catalase crystals were stored at 4°C and were washed with water prior to sample preparation. To prepare samples for the TEM, crystals were resuspended and the undiluted catalase crystal suspension was applied, blotted and vitrified in liquid ethane as described previously ( Shi et al., 2013 ).

Collection of electron diffraction data

All electron diffraction was performed on a FEI Tecnai F20 TEM operated at 200 kV with a selected area aperture (6 μm in diameter at the specimen) and data were collected with 4k × 4k TVIPS F416 CMOS cameras (15.6 μm pixel size). Diffraction data were collected with a frame rate of 1 frame per 6 s as the sample was continuously rotated from high to low tilt angle at ∼0.09° s −1 (0.54°/frame) as described previously ( Nannenga et al., 2014 ). A data set of approximately 61° was collected from a single crystal. Crystal thickness was estimated by measuring the intensity of the crystal (I) relative to the intensity of a hole in the carbon film (I 0 ) from an image and using Beer's law: l n I I 0 = − ε c t where ε is the molar absorptivity, c is the molar concentration, and t is the crystal thickness. As an approximation, the value of εc for catalase was assumed to be the same as those for calculated for lysozyme. The lysozyme coefficients were determined using images of lysozyme microcrystals with a known thickness as described previously ( Nannenga et al., 2014 ).

Data processing and structure refinement Raw

TEM diffraction data were converted and processed using MOSFLM v7.1.0 ( Leslie and Powell, 2007 ) and it's graphical interface iMOSFLM v1.0.7 ( Battye et al., 2011 ), POINTLESS ( Evans, 2006 ), and AIMLESS ( Evans and Murshudov, 2013 ) as described in previous work ( Nannenga et al., 2014 ). MOLREP ( Vagin and Teplyakov, 1997 ) was used to perform molecular replacement using catalase PDB ID: 3NWL ( Foroughi et al., 2011 ) as a search model (MOLREP contrast score = 40.8), and the molecular replacement solution was refined in PHENIX ( Adams et al., 2010 ) and REFMAC ( Murshudov et al., 1997 ) using a 5% free data set. Maps in Figure 3E,F,G and H were calculated using BUSTER-TNT ( Blanc et al., 2004 ). Maps and models were displayed using the UCSF Chimera package ( Pettersen et al., 2004 ).

Additional files Major dataset

The following previously published datasets were used: Foroughi LM , Kang YN , Matzger AJ , 2011 , The crystal structure of the P212121 form of bovine liver catalase previously characterized by electron microscopy , http://www.pdb.org/pdb/explore/explore.do?structureId=3nwl , Publicly available at RCSB Protein Data Bank. Ko TP , Day J , Malkin AJ , McPherson A , 1999 , The structure of orthorhombic crystals of beef liver catalase , http://www.pdb.org/pdb/explore/explore.do?structureId=4blc , Publicly available at RCSB Protein Data Bank. Purwar N , McGarry JM , Kostera J , Pacheco AA , Schmidt M , 2011 , Structural and kinetic analysis of the beef liver catalase complexed with nitric oxide , http://www.pdb.org/pdb/explore/explore.do?structureId=3rgp , Publicly available at RCSB Protein Data Bank.