Cryo EM Flashcards
Advantages of cryo EM
• Allows molecules to be studied in near native environment
• Biochemically functional buffers
• Can look at membrane environment
• Functionally relevant conformations
• No crystal packing artefacts
• X ray crystallography can trap non relevant conformations as it needs to pack that way to make the crystal
• Relies on only a few micro litres of material
• Concs as low as 10s of nanomolar
• Can image large scale of complexes, including macromolecular assemblies
• Can get atomic distances
• Cells, bacteria, viruses, proteins, atoms
Why is EM the resolution revolution
• High energy e- are waves
• Wavelength is smaller than inter atomic distances
• Resolution is proportional to wavelength
• Diffraction limit for EM is not a problem like for light microscopy
• These e- can be focused by magnetic lenses
• Images more powerful than diffraction patterns (images contain amplitude and phase info)
• Don’t need to solve phase problem
Issues with EM from biologists perspective
• Em operate at high vacuum to avoid unwanted scattering of electrons – means all the water would be sucked out of sample
• High energy e- are not just short wavelength waves but also ionising radiation – means you destroy sample when you image it
• Forced to limit exposure of sample to the electron beam to avoid damage – means you collect noisy images
• CryoEM samples move when irradiated (beam induced motion) – blurry pictures
Dubochets vitrification method
• Sample prep method
• Sample transferred to metal mesh and excess material removed
• The sample forms a thin film across the holes in the mesh when its shot into ethane at about -190 deg C
• The water vitrifies around the sample, which is then cooled by liquid nitrogen during the measurements in the electron microscope
• Allows sample to be looked at in vacuum condition
Frank’s image analysis for 3D structures
• Randomly oriented proteins are hit by the e- beam, leaving a trace on the image
• The computer discriminates between the traces and the fuzzy background, placing similar ones in the same place
• Uses Fourier transforms
• Using thousands of similar traces, the computer generates a high-resolution 2d image
• The computer calculates how the different 2D images relate to each other and generates a high resolution structure in 3d
Henderson’s vision for high resolution cryo em
• Worked on 2d crystals of membrane proteins
• Pioneered low dose imaging
• Was aware of what damage the radiation was doing to samples
• Pioneer of direct electron detection for cryoEM
• Increased sensitivity of the image and reduced radiation dose
What is the Rayleigh criteria
Defines the theoretical resolution limit
Resolution is the ability to resolve 2 points
Distance between 2 points = (wavelength of radiation x 0.61) / numerical aperture of the lens
• Visible light has a wavelength of 400nm so would have theoretical resolution of 200nm
• E- have wavelength of 0.0025nm so have theoretical res of 0.0015nm
• E- display wave particle duality
Compare the types of interactions electrons can have with a sample
• Most (80%) e- pass straight through sample- transmitted
• Some are elastically scattered by the nucleus and give the signal for the image
• Some are inelastically scattered and release energy in the form of ionising radiation, this contributes to noise level and radiation damage
• Some e- are back scattered so don’t contribute any info to the image
• The fundamental challenge is balancing signal and noise
Major elements of an electron microscope
• Electrons are emitted from a source
• E- generated at top of microscope
• Can be from tungsten filament
• Accelerate e- and heat up the filament
• E- are extracted from the pointed tip under a vacuum
• Field emission guns are also used as they have a much better coherence of the e- beam than tungsten
• Lens is made up of a series of copper coils
• E- wave passes through the centre
• Focus the e- by applying different currents through the copper wire
3 lenses of the EM
• Condenser lens: lenses control intensity, brightness, coherence and convergence of e-
• Objective lens: important for generating contrast
• Projector lens: amplification, magnifies image before detector
Apertures of the EM
• Apertures remove highly scattered e-
• Blocks highly scattered e- from going further down the column
• 1 set after the condenser lens to reduce spherical adoration
• 1 after objective lens to increase amplitude contrast
How is EM image detected
• Used to use film camera
• Used CCD camera but had bad resolution
• Can now use DD sensor to collect quick frames and compile into a video
• ID sensors collected an e- signal which was converted to a light signal then back to an e- signal so there was a loss of info in the process
• DD sensors have no info loss as they detect signal from the scattering an a single e-, increase the sensitivity so you can detect with less dose and can make movies with short frames
Why do particles move in the ice and how do we fix this
• Freezing the sample changes the potential energy of the system
• The e- release some of the tension from the sample so some of the particles can move
• By aligning frames of the movie you can correct rotation and get a clearer image
What are the 2 types of contrast
• Contrast is the difference in intensity between scattered and unscattered waves
• 2 types:
• Amplitude contrast (due to particle properties of e-)
• Phase contrast (due to wave properties), very important carrier of info
What is amplitude contrast
• Due to differences in thickness(not super important as sample is very thin), density (important in -ve staining microscopy) and lattice (for e- crystallography of 2D crystals)
• Enhanced by objective aperture as highly scattered e- contribute to noise
What is phase contrast
• E- move down the column of the microscope as a planar wave from the source
• Encounter atoms of a sample
• Causes phase shift of the planar waves
• When they are scattered they become complex waves which are a sum of planar waves and spherical waves
How do you focus with the objective lens
Incident wave is planar
• Interacts with scattering centres
• Resultant is an unscattered planar wave ( from the transmitted e-) and a complex wave
• Scattered wave is scattered 90 deg phase shift
• Lens causes an additional phase shift to focus the wave, this changes the amplitude of the wave
• Many waves are being scattered
• Higher scattering angle = higher resolution
• Waves superimpose to create a resultant wave
What is the effect of the lens
• Electrons are bent more strongly on the periphery
• Lens causes additional phase shift
• Strength of the lens is not uniform across the whole length – stronger at higher scattering angles
• Images that are in focus have 0 contrast
• Defocus = more contrast
• You want to under focus your image to get contrast
• Contrast is the difference between the unscattered and scattered waves
Too much defocus will give artefacts
What happens when you change the focus
• Changing the focus changes the wave function
• Oscillations between +ve and -ve contrast can leave artefacts in the image
• CTF (contrast transfer function) describes how the system modulates the contrast of various spatial frequencies in the specimen
• CTF is mainly interested by lens abberation and defocus
• These can alter the phase and amplitude of the e- waves that interact with the specimen
• Need to correct images for CTF-induced distortions
Phase plates as a way to generate contrast
Cause phase shift using a carbon disk in the column
Describe the characteristics of a biological sample that would lead you to choose cryoEM as a structural technique
• CryoEM was traditionally used for big objects e.g. viruses and ribosomes
• Since DD detectors we can do smaller and smaller sizes
• Comfortably we can image >150kDa
• We can use a very small amount of material
• 2.4-4.5 microL material
• 0.01-0.11mg/ml conc
• X ray crys would need 10mg/ml
• Can visualise things in a membrane environment
• Can visualise in native environment
Understand what limits resolution when imaging biological samples
• Heterogeneity in cryo EM is important
• We need to get pure samples for the image
• Usually prep by recombinantly expressing the protein
• Culture and lyse the cells
• Use affinity/size exclusion chromatography
• Assess purity in SDS-PAGE or activity assay
• Can negative stain (see later)
• Compositional heterogeneity is when distinct regions of the compound have different chemical compositions
• This can lead to variations in e- density which can lead to differences in image contrast
• Conformational heterogeneity – we want to limit number of different conformations of the compound to limit the signal to noise ratio
• Goal of sample prep to to balance preserving the native structure with limiting the signal: noise
What is negative staining
• Protein is adsorbed to a carbon support
• Blot excess liquid and add a heavy metal stain to give amplitude contrast (difference in density)
• Heavy metals have high atomic number so scatter the e- more
• Use tungsten or uranium
• We aren’t imaging the object, we are imaging the stain surrounding the object
• Usually use a copper grid with wire mesh and a thin carbon film
• Initially very hydrophobic
• Apply a charge to the carbon film with a glow discharge machine
• Applies a -ve charge so sample with adsorb and liquid will distribute over the surface of the mesh
• Sample is loaded into the tip of a room temp holder
• Tip goes inside the column
Advantages of negative staining
• Speed of screening – very quick image
• High contrast- because of high atomic numbers of heavy metals, doesn’t suffer from bad signal:noise
• Radiation hard- not sensitive to radiation as sample is dehydrated and gone so all that’s left is metal stain
Disadvantages of negative staining
• Resolution limited – imaging a cast so limited by the grain size of the stain (20A)
• Protein distortion – can get distortions in how thick the material is
Describe CryoEM as a sample preparation technique
• Sample transferred to metal mesh and excess material removed
• Sample forms a thin film across the holes in the mesh when its shot into ethane at -190 deg C
• Water vitrifies around the sample, which is then cooled by liquid nitrogen during the measurements in EM
Types of ice
• Not all ice is the same
• Vitrous ice (good) is water that’s frozen in a glass-like state (transparent)
• Obtained by temp and speed at which you freeze the substance
• Crystallin ice = bad, shows up like leopard spots and scatters e-
• Can get ice contaminants in bad freezing conditions
• Can also get surface contaminants due to moisture when loading the sample after freezing it
Issues with getting the particles into the ice
• Getting the particles into the ice is a major challenge
• Ideally want a random distribution of orientations
• Forces are created at the air/water interface as air is hydrophobic and water is hydrophilic
• This can change the orientations/distribution of objects
• Can introduce a support to adsorb the particles but the particles may aggregate/ have preferred orientations/ be denatured by the hydrophobic air
• Using a support helps control protein adsorbption
• Used to be a carbon film it carbon caused extra scattering so bad signal:noise
• Now use graphene support (single crystal of carbon lattice) which is transparent to the e- mean so wont scatter e-
• Changing the strength/time of the glow discharge machine controls the charge of the graphene so controls how much protein adsorbs and can control conc of protein on grid
Understand how to reduce impact of radiation damage to the sample
• Inelastic scattering causes ionising radiation which destroys bonds in biological material
• Increasing exposure/ number of exposures increases the radiation damage so high resolution info is destroyed
• We do low dose imaging
• Low e- doses at low magnification to find the area we want to take pictures of
• Focus area is near where the info we want is
• Radiation is only happening at high magnification (over a small area) near, but not on, where we want the image
• We apply these parameters to our image
• Parameters include refocus and roundness of beam
• We don’t want too much refocus or we get artefacts
• Then we use this info when we shoot the beam at our sample
• Direct e- detectors detect subtle scattering so we can use low doses, also takes fast exposures
• Take many frames
• Earlier frames have less damage so can have more weighting when we align
Describe how 2D EM images are related to the 3D sample observed
• 2D images are projections of a 3D object
• 2D image can be ambiguous so you need many orientations and projections to be collected
• Goal of single particle analysis is to collect lots of single particle images
• Need 10s to 100s of thousands to get high resolution 3D structure
• Individual particles are very noisy due to transmitted and in elastic e-
• To harvest high resolution info and enhance the signal:noise you average all the pictures
• You also need different orientations to get accurate 3D structure
• Using Fourier transforms is essential in computer image development
• It is a conventient mathematical representation of the image
• Can use it to do mathematical computations in Fourier space then use the inverse to go back to the image
• Projection theorem
• 2D projection is a central slice through thr 3D Fourier transform of the object
• Each 2D Fourier transform represents a unique particle as they are all in random orientations
• We need to relate each orientation to each other to compile them
• Then you can inverse Ft to get the structure
• Relating the images to each other relies on knowing the orientation
• 3 parameters: initial position, translation, rotation
• If you misalign them you get an average which isnt a true representation of the sample
• Heterogeneity in the sample would give artefacts and composites in the image
• Noise in the image is lowered by averaging
• Averaging can also cause errors if you have different conformations of the protein
• Need to separate and classify
Protocol for single particle analysis to solve a structure
• Protein purification – want homogeneous sample
• Negative stain
• Generate initial model
• Do cryoEM on pure protein sample
• Collect images on microscope
• Many frames on fast exposure to create a stack of images
• Need to align the images in the stack
Motion correction
• Flash freezing puts strain on the ice
• Tension is released when the sample is irradiated so particles can rotate
• Need to correct for it or image will be blurry
Defocus estimation and CTF correction
• Objective lens introduces artefacts which are taken into account in the CTF
• CTF is oscillations of positive and negative contrast
• We find focus then take the image under focus
• Need to know CTF of images and individual particles to correct so everything has +ve contrast
• If we didn’t correct we would have distortions
Defocusing - when the microscope is deliberately defocused, the objective lens is adjusted to intentionally place the image plane in front of or behind the specimen. This results in a blurred image
Particle selection
• We use lots of different automated software to separate out noise and contaminants from actual particles
• You will always have some contamination you need to clean up
Particle classification
• Happens in 2D
• Don’t need to know relative orientation yet
• Separate real particles from ice contaminants/broken/denatured particles
Projection matching
• We use the initial model/ reference image
• Can rotate reference by a certain angular increment and can calculate a projection with a known rotation
• Then you find the best matching raw 2D image and can assign the angle
Reconstruction of the image
• Compile the 2D images to make 3D image
• Iterative process to always improve the model that feeds into the next cycle
Classification
• Refined model still could be a mixture of conformations
• Can classify different states in the computer using 3D classification methods from machine learning tools
Map validation
• You will get a lot of different structures
• DD improved signal: noise
• Detergent micelles contribute a lot to noise
• CCD scintillator was a big source of noise
• Should see secondary structure in the map that matches the model
• The iterative refinement process can lead to model bias
• Aligning noise to the model to match what it was
• Can create a phantom shape
