MAX PL ANCK INSTITUTE OF BIOCHEMISTRY MAX PL ANCK SOCIETY 1 Image processing workflows for Tomography Winter School on Structural Cell Biology Julio Ortiz Wolfgang Baumeister MPI of Biochemistry Dept. of Structural Molecular Biology Martinsried, Germany Brno,12th February 2015 Summary/Outlook 2 Dual-axis tilt PyTomIn situ Plunge freezing -180 °C Single-axis tilt Matlab toolbox FEI Titan KriosFEI Quanta 3D Sample Preparation Electron Cryotomography Instrumentation Software In vitro Amira Chimera Adobe CS63ds Max Direct Electron Detectors FIB TM Phase Plate Cryo-electron tomography (CET) 3 CET combines the power of 3-D imaging with a close-to-life preservation of cellular structures and a allows to study macromolecular structures in situ. Low-level Techniques for Data Analysis in Tomography Segmentation Electron-dense labeling Template Matching Presynaptic vesicles Fernández-Busnadiego, R. et al., J. Cell. Biol. (2010) Boehm, J. et al. Proc Natl Acad Sci U S A. (2000) NPC substrate localization Beck, M., et al., Nature (2007) Detection and identification strategy A.S. Frangakis, J. Böhm, F. Förster, S. Nickell, D. Nicastro, D. Typke, R. Hegerl, W. Baumeister: Proc Natl Acad Sci USA 99 (2002) 14153-14158 Identification of macromolecular complexes in cryoelectron tomograms of phantom cells Frangakis, A. et al., PNAS (2002) Thermosomes and 20S proteasomes encapsulated in liposomes Schematic flow diagram showing the detection and identification strategy Hibernating ribosomes (in vitro & in vivo) Ortiz, J., et al. J. Cell. Biol. (2010) Spiroplasma Ortiz, J., et al. J. Struct. Biol. (2006) Polysomes in vitro Brandt, F., et al. Cell. (2009) Visual Proteomics - Leptospira Beck, M., et al. Nat. Methods (2009) Visual Proteomics - Mycoplasma Kuerner, S., et al. Science (2009) 2002-2012: Template Matching Achievements Polysomes in situ 8 Polysomes E. coli E. coli Aim: To detect ribosomal supramolecular arrangements in the cytoplasm of fast growing E. coli cells. Brandt, F. et al. Cell. 2009 23:261-71 Tomogram of thin whole E. coli cell (slow growing cell; CCD- camera) Cryo-ultramicrotomy- Diamond Knife Cryo-ultramicrotomy NAD filtered slice from tomogram ~ 30 % compression Polara 300 KV, Mag: 55 KX, Pixel size: 0.55 nm/pixel, Def: -14 nm, Thickness: ~100 nm E. coli cellCryo-section 12 Focused Ion Beam milling coupled to CET Advanced Transmission Electron Microscopy Dual-tilt axis CET with K2 camera and Energy filter 13 Polysomes in situ 14 Polysomes E. coli In situ Plunge freezing -180 °C FIB 15 E. coli BL21 MtlA385-SecM (non-stalling conditions) Dual-axis tiltFIB DDC K2 Overview 16 E. coli BL21 MtlA385-SecM (non-stalling conditions) Dual-axis tiltFIB DDC K2 Detail: 70S ribosomes and other complexes Methods 17 Template Matching Cryoelectron Tomography (CET) Positions (x,y,z) Orientation (,,) for putative ribosomes Lysates or Intact cells Relative Orientation Constrained Correlation Maximum- Likelihood 3D averaging Classification The art of averaging: Missing Wedge effect Combining reconstruction by averaging Reconstructions from identical objects in different orientations 2D Model Combined reconstruction The art of averaging Combining objects in different orientations… The art of averaging: The “Mona Lisa” effect – Bias caused by reference The art of averaging The “Vitruvian man” effect – Flexibility and occupancy problems The art of averaging The “Vitruvian man” effect – Flexibility and occupancy problems Average of 8 different images“Vitruvian man” – Leonardo Da Vinci Constrained Correlation Classification Förster F., et al. (2007) J. Struct. Biol. Constrained Correlation Classification (3D-average selected ribosomes) 25 mRNA entry and exit site Elongation factor binding sites Nascent chain exit site 70S ribosome Average of ~ 6000 part. selected by 30S subunit similarity resolution: ~2.6 nm Areas with associated densities: Hibernating Ribosomes in Lysates from Starved E. coli Cells Maximum- Likelihood Constrained Correlation Relative Orientation 3D Alignment and Averaging 27 27K p. iteration 1 FSC 0.5=3.1 nm 27K p. iteration 3 FSC 0.5=2.6 nm 8 K p. FSC 0.5=2.8 nm 292 p. ; 0.68 nm/p 292 p. ; 0.342 nm/p FSC 0.5=3.8 nm 3’-neighbor ribosome attached in ca. 1% of analyzed particles A B C D E Align. 70S Class. 3p Class. 3p Align. 3p Polysomal organization in situ: 3D-Average “top-to-top” 28 • 103 particles • (256)3p, 0,342 nm/p • Average filtered at 3nm res. • Particles present in all analyzed tomograms 50S - central ribosome 30S - central ribosome 50S – neighbor ribosome 30S – neighbor ribosome Methods: Software 29 Matlab TOM toolbox The toolbox supports a wide range of functions for tomography that extend the capability of the MATLAB® numeric computing environment. Aim: Customable procedures for tomogram reconstruction and analysis. Nickell. S. et al. (2005) J. Struct. Biol. 149:227-34 Matlab TOM toolbox Methods: Software 30 PyTom PyTom Open-source platform that unifies standard tomogram processing steps in a single python-based toolbox. Aim: 3D alignment, averaging and classification of subtomograms. Implementation of new algorithms for image processing in ECT. Hrabe, T., et al. (2012) JSB 178: 177-188 Recent Improvements in Data Processing Workflow 1. Reconstruction • Phase Flipping (‘CTF correction’) • Projection Alignment with Tilt-Specific Image Rotation + Magnification 2. Localization • Template Matching (+Support Vector Machines) • Parallel processing on LINUX cluster • Better angular sampling 3. Subtomogram Alignment • Parallel processing on LINUX cluster • Resolution-dependent lowpass: low bias • Resolution-dependent sampling • Direct reconstruction from projections 4. Classification • Multi-reference correlation with simulated annealing Hrabe, T., et al. J. Struct. Biol. 2012 PyTom 50 nm Sample: Hippocampal primary cultured neuronal cell 26S Proteasome in situ! CET: Direct detection and Phase Plates Asano, S. et al. (2014) Science 23:347 Proteasomes in neurons cytoplasm Asano, S. et al. (2015) Science 23:347 40 m 10 m Cryo-FIB: Chlamydomonas reinhardtii Engel, BD. et al. (2015) Elife 13;4 Cryo-FIB: Chlamydomonas Chloroplast (thylakoids) Pyrenoid (RuBisCO) Starch ER Golgi Lipid Droplets Nucleus Contractile Vacuoles Centrioles Thylakoids Pyrenoid Starch Globules Cellular CET: ’fibbed’ Chlamydomonas Engel, BD. et al. (2015) Elife 13;4 RuBisCO CCD Cryo-FIB: Chlamydomonas reinhardtii Pyrenoid Engel, BD. et al. (2015) Elife 13;4 RuBisCO PyrenoidCCD K2 Summit pixel size: 0.424 nm; 10,000 particles out of ~300,000 from ONE tomogram CRYO-FIB: Chlamydomonas reinhardtii Robust membrane segmentation based on tensor voting Original tomogram Filtered tomogram Segmentation by thresholding Materials Volume-based labeling Martinez-Sanchez, A., et al., (2014) J. Struct. Biol. in press Tensor voting Robust membrane segmentation based on tensor voting Original tomogram Filtered tomogram Segmentation by thresholding Materials Volume-based labeling Martinez-Sanchez, A., et al., (2014) J. Struct. Biol. in press Tensor voting Robust membrane segmentation based on tensor voting 41 segmented outher membrane segmented inner membrane Organization of the mitochondrial translation machinery studied in situ by cryoelectron tomography Distribution of mitoribosomes in translationcompetent yeast mitochondria. Pfeffer, S. et al., (2015) Nature Communications Schematic representation of the cryo-ET workflow Lučić, V. et al. JCB (2014) Johnson,G and Hertig, S. Nat Rev Mol Cell Biol. (2014) 10:690-8. 44 Biomolecular Visualization Tools Acknowledgements 45 Present and former members of the Structural Biology at the MPI of Biochemistry Juergen Plitzko Friedrich Foerster MPI of Biochemisty Wolfgang Baumeister MPI for Biochemistry