Cryo-electron microscopy and single particle analysis

Starting in 2004, we established a state-of-the-art cryo-EM facility within the Berlin-Brandenburg research consortium “UltraStructure Network” (USN), which was at first initiated in close collaboration with Prof. Christian Spahn (Charité Berlin) and Prof. Roland Beckmann (Charité Berlin, now LMU Munich). The main focus of our research is the structure determination of macromolecular protein complexes using cryo-EM in combination with the single particle approach. Our facility provides a technology platform for sample screening, semi-automated sample vitrification and data acquisition as well as intense computing resources for image processing. Core instrument is a 300 kV Tecnai G2 Polara electron microscope (FEI), which is now equipped with a 2kx2k Orius CCD camera and a k2summit direct electron detector (Gatan).

Figure 1: Cryo-EM reconstructions of sub-states I (TIPRE) (a) and II (TIPOST) (b) of the T. thermophilus 70S•EF-G•GDP•FA complex shown from the L7-stalk-site (yellow: 30S subunit; blue: 50S subunit; red: EF-G; green: tRNA). The comparison of sub-states I and II of the 30S subunit with the maps aligned at the 50S subunit indicates the difference in terms of ratcheting (c). The Alignment to the body/platform domains of the 30S subunit highlights differences in head swiveling (d). The 30S of sub-state I (TIPRE) is rendered in transparent orange, while the 30S of sub-state II (TIPOST) is in solid yellow (from Ratje et al., 2010, Nature).

In single particle cryo-EM, 3D information is derived by averaging over thousands if not hundred thousands of projection images of individual molecules (referred to as particle images), assuming that they all identical. In practice, however, preparations of protein complexes rather have to be considered as heterogeneous samples due variable complex assembly and conformational flexibility. Moreover, it is now commonly accepted that these conformational states often represent intermediate states within a metastable energy landscape; and tuning this energy landscape e.g. by ligand binding and/or hydrolysis of nucleotides is the fundamental basis for the biological function of these complexes (Munro et al., 2009, Trends Biochem Sci).

Within the collaborative research centre SFB740 (project Z1) our group aims at implementing new technologies to overcome three main hurdles in single particle cryo-EM, namely sample heterogeneity, the large amount of data required and the poor contrast and high noise-level of cryo-EM data. Together with C. Spahn (IMPB, Charité Berlin), we implemented and tested multiparticle refinement strategies developed in his lab to analyse sample heterogeneity in-silico. Introducing 3D variance analysis and variability mapping (Loerke et al., 2010, Methods Enzymology) we could split heterogeneous data sets of bacterial as well as eukaryotic ribosomal complexes into sub-populations representing different conformational and/or functional states. This allowed us to identify new functionally important conformational modes occurring during the elongation cycle of protein biosynthesis such as a novel intra-subunit pe/E hybrid state showing a partly translocated tRNA (Fig.1, Ratje et al., 2010, Nature), the large swivel movement of the 30S head (Ramrath et al., 2012, Nature), new translocational intermediate states (Ramrath et al., 2013, PNAS), spontaneous movements of tRNAs on the eukaryotic PRE 80S ribosome (Budkevich et al., 2011, Mol. Cell) and subunit rolling between the mammalian PRE and POST complexes (Budkevich et al., 2014, Cell). Furthermore, we could determine several structures of ribosomal complexes involved in IRES-mediated initiation (Yamamoto et al., 2014, Nat Struct Mol Biol) and termination (Muhs et al., 2015, Mol. Cell).

Figure 2: High-resolution cryo-EM map of the human ribosome in the POST state filtered to 3.5 Å (A) and corresponding atomic models of the ribosomal subunits (B). (C–F) Enlarged regions of the cryo-EM map showing well-resolved alpha-helices (C) or beta-strands (D) including individual side-chains, strong p-stacking interactions (E) and individual nucleotides (F) with nearby ions (from Behrmann et al., 2015, Cell).

Multiparticle refinement, however, requires even larger data sets comprising up to a million of particles and more. We therefore implemented the Leginon system (Suloway et al., 2005, J Struct Biol) for automated data collection on our Spirit and Polara microscopes, which allows us to acquire up to 20.000 digital micrographs per week at routine level. Leginon is also used for automatic acquisition of tilt-pairs required for random conical tilt analysis as well as tomographic tilt series. Another major break-through in cryo-EM was the recent introduction of direct electron detector devices (DEDs). In contrast to scintillator-based camera systems, DEDs directly register primary electrons and thus provide a much higher sensitivity and dramatically improved signal-to-noise ratio. Combining multiparticle refinement, automatic data collection and the potential of new DED detectors, we could identify more than 10 different functional states from ex-vivo derived human polysomes and refine the highest-occupied state, the 80S POST-complex, to near-atomic resolution (Fig. 2, Behrmann et al., 2015, Cell). In May 2015, we installed a k2summit DED on our Polara, which will now allow us to obtain high-resolution structures also from smaller protein complexes such as RNA-polymerase (collaboration with M. Wahl, FU-Berlin) or the DPE2-complex from plants (DGF project Mi 940/1-1), amongst others.

Go to Editor View