10) Cryo-EM

what is transmission

no longer fourier space

phases are kept, distances are the same

what is a CT scanner

sum of transmission images, like XRD summing waves

the finer the angles, the better the resolution of the image you produce

adjust contrast

compare XRD and CT

both use x-rays

both use data collected from different angles
XRD rotates crystal, CT scanner rotates x-ray source

both add images to get final result
CT relies on transmission through a sample
XRD relies on diffraction by a crystal (waves)

CT scanner gives 2D cross-sections that can be joined to give 3D image
crystallography gives 3D map

how does electron microscopy of protein work

technology revolution for cryo-EM: richard henderson, jacque dubochet, joachim frank

beam of electrons fired at frozen protein solution
emerging scattered electrons pass through lens to create magnified image on detector, structure can be worked out from that

solution of proteins, tiny EM grid put into liquid ethane cooled solution

analyze data using microscopy

within the grid there are carbon foils, within that is coils, see if protein gets inside

collect transmission image, from random orientations, tomograms

most important is getting accurate angles, without that you can’t get structure

transmission EM (TEM) vs light microscopy

TEM involves magnets and a detector (not eye)
electrons poorly penetrate sample, bad resolution

what must you have for cryo-em

pure sample, otherwise you have to classify each shape into its own class and you must be able to identify your protein among others or contaminants

cryo-em map not electron density map

XRD vs single particle cryo-EM

very similar

x-ray → protein crystal → diffraction pattern (fourier space) → phase determination (using fourier transformation) → electron density map (real space) → protein atomic model

electron beam → vitrified protein sample (vitreous ice) → 2D projection (collect at different orientations from particle) → orientation determination of particles (3D reconstruction/EM density map) → pseudo/atomic model

what is the workflow for single particle cryo-EM

sample of pure protein

put on tiny grids, remove excess liquid, dunk in liquid ethan (freezes faster than liquid nitrogen)

particles will be in random orientation, single particle images generated by TEM, pick particles

align them in 2D, arrange in 3D

details on sample optimization

you lose 98% of protein to the carbon grid, very few actually make it into the holes

Ice thickness must be optimized: too thin, protein won’t go into holes
too thick, contrast will be reallly bad

accurate assignment of angles is crucial

details on ice thickness

can vary in a single hole (thinner or thicker in center)

take low resolution of protein, simulate what it should look like under different orientations

projection matching with reference model, but this can create biases