Cryo-EM Flashcards
Give an overview of cryo-EM?
Cryogenic Electron Microscopy
‘Single particle’ methods - How to solve structures to high resolution using electron microscopy
Tomographic methods - How we can study unique structures/events
The technique involves flash-freezing solutions of proteins/biomolecules and bombarding them with electrons to produce microscope images of individual molecules
They are used to reconstruct the 3D shape, or structure, of the molecule
These structures are useful for understanding how proteins work, how they malfunction in disease and how to target them with drugs
What is an image?
The image contains all sorts of different resolutions simultaneously
Low frequency/resolution can approximate the structure - but poorly shows the level of detail
By adding in multiple higher frequency waves to sum up the object
= all objects can be represented by a series of waves at various frequencies
Describe some concepts required fro cryo-EM - fourier transform and wave particle duality?
Fourier transform:
This is a maths transform that decomposes a function (often function of time or asignal) into its constituentfrequencies
High resolution information is at the edge of the pattern and low resolution in the centre
Wave particle duality
Electron behave as both waves and particles
Electrons have mass, charge, position and momentum and so behave like particles
Electrons can be focused and create interference patterns and so behave like waves
What are the fundamentals of electron microscopy?
Electrons are charged ‘particles’ - it is their interaction with atoms in the sample that deflects them and generates ‘information’ in an image
Contrast = Signal = Information
Resolution = Spatial frequency
99.999% of atoms found Biology are LIGHT (H,C,N,O,P,S)
Weak interaction with beam = poor signal
What are the conditions needed for cryo-EM?
The column is under vacuum and has a layer of ice with the proteins suspended within it
Speed of freezing is key:
Slow freezing - generates hexagonal or cubic ice - this dehydrates and kills proteins/cells
Fast freezing - generates vitreous ice - a glass-like form of ice that is essentially unstructured, frozen water - leaves proteins in a native-like state
Really fast - a rate of temperature change between 10^5 and 10^7 °C.s-1 is required (just needs freezing in the right solvent)
Describe vitrification and automation within cryo-EM?
Thin films only - the thinner the specimen the better chance for vitrification
Coolant must have a high thermal conductivity (typically liquid ethane - cooled by liquid N2)
Aim to generate a thin, frozen film with your macromolecular complex suspended in random orientations in a layer of unsupported vitreous ice
How could you increase contrast within cryo-EM?
Negative stain
We can embed the sample in a layer of a heavy metal stain – typically Uranium salts
Strong interaction between electrons and stain, so we “see where the sample is not” – hence negative staining
Positives:
Stable in beam
High contrast
Negatives
Not exactly native conditions
Sample is dried and flattened onto surface
Stain does not penetrate – surface only
Stain has a grain (or crystal) size – limits resolution
Describe the image produced in cryo-EM?
Contains the projected density of multiple copies of the same assembly
Information to atomic resolution
Has amplitudes and phases
Not much contrast i.e. lots of noise and not much signal
Projected density
Electrons are a penetrating radiation - the images we get are a projection of the 3-D density of our object onto the 2-D plane of the image
3D information is lost so to regain the third dimension we get different views of the specimen
E.g. Like a CT scan
How can we increase sound/noise ratio?
Increase dose - electrons are a penetrating, ionizing radiation - radiation damage destroys biological material
Fourier filtration - removing noise improves contrast - but we don’t know where noise stops and information starts
Masking - can remove influence of neighbouring particles without losing resolution
Averaging - aligned images improves S/N (essential technique)
It needs the right images to be aligned together otherwise it can make it worse
Describe alignment of the different images?
Starting model -> project in all possible directions
Makes reference images of different views of your structure
Raw images - get as many as possible (randomly orientated particles)
Align every raw image to each of the reference views
Average - the S/N goes up and particles are sorted into views
Reconstruct
What is significant about images take in cryo-EM?
All images in EM are taken out of focus
Light atoms interact weakly with the beam (very low amplitude contrast)
We can introduce phase contrast (an interference pattern around each particle) by defocussing the microscope
The contrast you see in EM is related to the resolution you look at, and the amount of defocus you used to record the image
Describe contrast transfer and correction?
We get oscillations between positive/negative contrast
Therefore defocusing your microscope includes contrast
This combines data recorded at different defocuses
Any individual image of a particle will have information gaps – because of the oscillations in the CTF
These gaps can be filled in by combining information from lots of different defocuses
Describe data sampling, storage and computer power associated with cryo-EM?
The smaller the Å/pixel (the real size that the pixel represents), the greater the pixel*pixel and therefore the more storage it takes up
E.g. 0.5 Å/pixel = 512x512 pixels = 4 Mb
Whereas, 8 Å/pixel = 32x32 pixels = 16 kb
The Nyquist-Shannon sampling theory - it depicts the maximum theoretical resolution, which is 2x sampling
We can’t always use fine sampling as it takes too long as well as the computer/disks needed for storage aren’t big enough
Computational time increases with the square of image size in 2D and the cube of image size in 3D
We don’t always use coarse sampling as resolution is important
How can we get better stuctures?
Resolution revolution
Better Stability – Better Microscopes - Automation
Optically more perfect e.g. Titan Krios
Improved Signal/Noise ratio in every image
Describe the requirements of a ‘better’ microscope?
The exposure time is generally 2s
We aspire to get 2Å resolution
We need the sample to move less than 5% of the desired resolution during the exposure time - otherwise our images will be blurred
= a drift rate of ~ 0.05Å/s (or 5x10-12 m/s)
Therefore the newer microscopes don’t allow the sample to move
