Structural Information obtained

Transcription

1 Structural Information obtained from Electron Microscopy Christiane Schaffitzel, Including slides from John Briggs, Bettina Boettcher, Nicolas Boisset, Andy Hoenger, Michael Schatz, and more colleagues

2 Outline Single Particle Analysis why and when? Sample Preparation for Electron Microscopy Cryo-Electron Microscopy Image Processing How can EM and SAXS/SANS be combined? examples from the literature. re 2

3 3D EM structures across range of resolutions and samples 13 Å membranes molecules atoms organisms/tissue cells organelles vesicles 1 Å m mm μmm viruses nm 3.5 Å pm Human Eyes Light Microscopy Scanning Electron Microscopy Transmission Electron Microscopy X-Ray MRI NMR Spectroscopy 3

4 Single Particle EM why and when? Solve structure in situ Zhang X, et al., Cell 2010: 3.3 Å cryo-em structure of a nonenveloped virus reveals a priming mechanism for cell entry. 4

5 Single Particle EM why and when? Solve multiple conformations from same sample Fischer et al., Nature

6 Single Particle EM Different Techniques Negative Stain EM Slide: John Briggs 6

7 Negative Stain EM Single Particle EM Negative Stain Advantage: nice contrast of the molecules. Scale bar : 20 nm Disadvantage: the resolution is limited by the stain. staining artefacts, flattening 7

8 Single Particle EM Cryo-Electron Microscopy Freezing Grids Holey carbon film vitreous ice ( Å) thick carbon (150 Å) copper grid 8

9 Advantages of Cryo-EM over Negative Staining The sample is in solution in the physiological buffer. There is no stain to distort the sample. When using holey carbon films, preferential orientations are minimized. High resolution can be reached (3 Å). Disadvantages: antages Very low signal to noise ratio. Tilt imaging is difficult: thick ice and charging for the tilted image. More time consuming to generate samples.

10 Cryo-Electron Microscopy Tecnai F20 Microscope 100 nm

11 Cryo-Electron Microscopy Polara Microscope 300 kv Titan Krios 300 kv

12 What is a Cryo-EM Image? It is a projection, containing 3D information This projection does not contain 3D information! 12

13 What is a Cryo-EM Image? 13 Slide: John Briggs

14 What is a Cryo-EM Image? 14 Slide: John Briggs

15 15

16 Beam Damage, an other example

17 What is a Cryo-EM Image? 17 Slide: John Briggs

18 What is a Cryo-EM Image? 18 Slide: John Briggs

19 The Contrast Transfer Function atoms will appear bright on a dark background no contrast envelope function Information limit atoms will appear dark on a bright background Consequences: 1. Images need to be phase-flipped to obtain high resolution. 2. To fill the gap of information, images taken over a range of defocus values need to be combined. g This curve (among others) depends on: Cs (the quality of objective lens defined by spherical aberration coefficient) l (wave-length defined by accelerating voltage) Df (the defocus value) Slide adapted from : Henning Stahlberg 19

20 Other Information present in the Fourier Transformed Image Astigmatism, detected from Thon rings Drift, detected from Thon rings Sort out these images Slide adapted from : Carsten Sachse 20

21 Image processing What s next? How do we gain resolution? How to get from here to a 3D reconstruction? 21 Slide: John Briggs

22 Single Particle EM Image Processing Two Key Concepts: 1. We need to average over many images. 2. We need different views of our object. Signal to noise ratio growths with n Slide: Bettina Boettcher 22

23 Single Particle EM Alignment Averaging is going to require alignment: rotational and translational 2D alignment, rot, trans The alignment is done based on cross correlation. The cross correlation coefficient also indicates the similarity of images. Slide: Bettina Boettcher 23

24 Variance of Average(s) reveals Differences (variable parts) Slide: Bettina Boettcher 24

25

26 Alignment-Classification-Averaging Classify and average similar images: Example: F-type ATP-Synthase, Bettina Boettcher And then use to improve the alignment iteration How do these views relate to each other? 26

27 Single Particle EM Image Processing Two Key Concepts: 1. We need to average over many images. 2. We need different views of our object. Possible Approaches: Single particle: (ideally) the particles are randomly oriented, and are all the same. These correspond to different views of the same object. Helical: Within one helix, we see the basic building block, (the unit cell), from many directions. The image of one helix contains many different views of the unit cell. 2D crystallography: All of the unit cells are in the same orientation we collect images of multiple different crystals, each tilted to different angles within the microscope. Tomography: We only have one copy of the object. We need to tilt it within the microscope and image it again to get different views. 27

28 Tomography vs. Single Particle Analysis One object is turned in the electron beam many 2D images with known orientation Many particles with unknown orientation Determine orientation 3D Reconstruction Split the electron dose. Use the max.electron dose for one image.

29 Angle Assignment, Euler Angles theta θ: Defines the elevation above or below the equator. phi φ: Defines the rotation (azimuth) around the equator. psi ψ: in plane rotation 29

30 Random Conical Tilt Reconstruction Viewing directions Fourier space Radermacher, M. (1987) Boisset, N. (1990) Initial 3D model 30

31 Refinement using Projection Matching Starts from some initial guess about the structure (e.g. structure determined by random conical tilt reconstrucution), which serves as the reference structure projections Alignment against each reference projection by rotation and shift and determination of the best cc value reference projection data Euler angles obtained from the initial 3D reference structure which is used to generate reference projections.

32 When we know the Euler angles we can calculate the 3D structure: Slide: Michael Schatz

33 Workflow Refinement 33 3D Starting Model e.g. from random conical tilt reconstruction Projections of the starting model Euler Angles known Re-project 3D reconstruction used as reference for the next round of projection matching Back-project Use the known angular relationships Data Individual Picked Particles Alignment rotation, shift Classification based on crosscorrelation Average Generation of 2D classes 2D classes Euler angles assigned

34 Challenges: Dealing with conformational and compositional heterogeneity requires advanced image processing Several starting models Sorting strategies one EM dataset is used to solve several structures. 34

35 Example1: Combination of EM and SAXS/SANS Šulák O. et al. (2011) Burkholderia cenocepacia BC2L-C C Is a Super Lectin with Dual Specificity and Proinflammatory Activity. PLoS Pathog 7(9): e doi: /journal.ppat SAXS ab initio model EM reconstruction Lectins: Bacterial adhesion proteins to host cells Projections 35 2D classes

36 Example 2: Combination of EM and SAXS/SANS Martin Alcorlo et al. Unique structure of ic3b resolved at a resolution of 24 Å by 3D-electron microscopy PNAS 2011 vol. 108 no Negative stain EM: 2D class averages of different C3 convertase activation states TED: thioester containing domain MG: macroglobulin li ring SAXS: global shapes of the same C3 convertase activation states 36 C3 is a complement factor which is cleaved to generate the activated fragment (C3b). Proteolysis of C3B leads to formation of ic3b which is targeting pathogens for clearance by phagocytosis.

37 Single Particle Reconstructions - Summary Requirements for Object: Objects must exist in multiple identical copies (but heterogeneity can be taken into account!) Limited number of conformations (ideally 1) Mass for unstained specimens should be >300 kda Typical Objects: Large Proteins and Complexes How much do we need? ( μg) Mode of Data-Collection: Many low dose micrographs Achievable Resolution: 3 Å possible (depends on number of correctly merged images, image quality) Resolution is isotropic (if no preferential views) 37