Lecture 9 - CRYEM-ET Basics, Part 2 (Guest - Luiza) PDF

CRYOEM: SEEING IS BELIEVING AMAZING Luiza Mendonça ([email protected]) PART 2 Oct 18th 2024  “PROJECTION” IMAGING IN EM An EM image is the 2D projection of a 3D object (plus...

CRYOEM: SEEING IS BELIEVING AMAZING Luiza Mendonça ([email protected]) PART 2 Oct 18th 2024  “PROJECTION” IMAGING IN EM An EM image is the 2D projection of a 3D object (plus aberrations, plus noise). One image is not sufficient to deduce the structure of a sample particle. How to get the 3D structure? CLASSIFYING, ALIGNING AND John O'Brien, The New Yorker Magazine (1991) AVERAGING CRYOEM IMAGES HAVE VERY LOW S/N (LOW DOSE TO REDUCE RADIATION DAMAGE OF ELECTRON BEAM + SHOT NOISE + ABERRATIONS) IMAGE PROCESSING – BRACE YOURSELVES! Show proof (https://thepythoncodingbook.com/ 2021/08/30/2d-fourier-transform- in-python-and-fourier-synthesis-of- images/) Images (in 3D or 2D) are made of the sum of sine waves Sine waves can be EQUALLY represented in “real-space” or “reciprocal space or Fourier space” (sometimes wrongly called “Power Spectrum”)  IMAGING PROCESSING IS COMPUTATIONALLY COSTLY Image transformations (translate, rotate, average, add, subtract, filter) are easier and faster in Fourier space FFT are also mightly useful in identifying issues with data collection in cryoEM REAL SPACE TO FFT CONVERSION IS LOSSLESS!  (AMPLITUDE, FREQUENCY, DIRECTION AND PHASE ARE PRESERVED) CRYOEM - 2 MAIN FLAVORS Single Particle Analysis Cryo-tomography Sample is purified, in solution and Sample can be purified or not monodisperse Tilt-series collected at a single location Thousands of images taken at different locations Tilt-series is back-projected into tomogram (3D) Orientations determined, 2D images  Usually low resolution aligned and averaged Subtomograms (3D) are aligned and 2D projections used to “build” original averaged (subtomogram averaging) 3D structure  High resolution It’s all about averaging! To increase signal, decrease noise and build the 3D structure. SINGLE PARTICLE ANALYSIS Single Particle Analysis Sample is purified, in solution and monodisperse (and at different orientations) Thousands of images taken at different locations Orientations determined 2D projections used to “build” original 3D structure SINGLE PARTICLE ANALYSIS The projection theorem class averages in “reciprocal space” (spatial frequency domain)  Electron microscopy map Coulomb potential map Electron scattering map 1.2A STRUCTURE OF APOFERRITIN BUT NOT AN ELECTRON MAP!!! (same goes for density) Nakane, Kotecha, Sente et al., 2020 “RESOLUTION” IN CRYO Still a “contentious” topic Different ways of estimating it (A sure minefield in email lists) Some hard limits:  Wavelength of wave  Sampling at the camera level (pixel size) We never MEASURE the resolution in cryoEM, we ESTIMATE it Automated Modeling and Validation of Protein Complexes in Cryo-EM Maps Tristan Cragnolini, Aaron Sweeney & Maya Topf 2020 FOURIER SHELL CORRELATION A measure of self consistency of the data and the reconstruction process We won’t go (much) into the rabbit hole, so here’s some premises: Images (in 3D or 2D) are made of the sum of sine waves Sine waves can be EQUALLY represented in “real-space” or “reciprocal space” (the FFT of a sine wave) Show proof (https://thepythoncodingbook.com/2021/08/30/2d-fourier- transform-in-python-and-fourier-synthesis-of-images/) Why “shell”? Low resolution (lower spatial frequency) info High res (higher spatial freq) info  FSC A measure of self consistency of the data and the reconstruction process 2 MAIN FLAVORS Single Particle Analysis Cryo-tomography Sample is purified, in solution and Sample can be purified or not monodisperse Tilt-series collected at a single location Thousands of images taken at different locations Tilt-series is back-projected into tomogram (3D) Orientations determined  Usually low resolution 2D projections used to “build” original Subtomograms (3D) are aligned and 3D structure averaged (subtomogram averaging)  High resolution  CRYOELECTRON TOMOGRAPHY (CRYOET) Cryoelectron Tomography ▪ Native (vs purified) ▪ Hydrated (vs resin- embedded) ▪ 3D (vs 2D) ▪ Con: penetration Cellular cryo-ET Woodward, 2015 CELLULAR CRYOET  UNLEASHING THE FULL POWER OF CRYOET Molecular resolution Complementing cellular tomography Subtomogram averaging with correlative imaging HIV-1 Gag A B 4.2 Å D C 6-helix bundle Mendonça et. al, 2021  UNLEASHING THE FULL POWER OF CRYOET Molecular resolution Subtomogram averaging HIV-1 Gag 4.2 Å 6-helix bundle Mendonça et. al, 2021 Mendonça et. al, 2021 UNLEASHING THE FULL POWER OF CRYOET Complementing cellular tomography with correlative imaging (cryoCLEM) A B D C UNLEASHING THE FULL POWER OF CRYOET Accessing the cell interior IRL cryoFIB/SEM (Aquilos) LAMELLA POSITIONING TEM Atlas TEM Medium-mag montage SEM Lamella positioning Use the positive cells identified in TEM to do mill lamella in FIB-SEM Mendonça et. al, 2021 VIRAL ASSEMBLY Mendonça et. al, 2021 INTEGRATIVE IMAGING HAVE YOUR CAKE AND EAT IT TOO Goal: Cover multiple spatial scales Problem: Resolution vs Field of View Solution: Combine multiple techniques  INTEGRATIVE IMAGING INDIVIDUAL PROTEINS AND CELLULAR CONTEXT HIV virological synapse 3A 50um Residues Molecules Cytoskeleton Viruses Organelles Cells EXAMPLE OF INTEGRATIVE IMAGING WORKFLOW Cryo Soft X-ray CryoET CryoFIB/SEM Tomography periphery Serial SXR Tomogram cryoFIB/SEM lamella internal CRYOEM: SEEING IS BELIEVING AMAZING Native Fully hydrated 3D Cellular context (cellular cryoET) Cryoelectron Molecular resolution microscopy Non-destructive (except FIB) Integration with other techniques Electron More to come… cryomicroscopy