Structural Biology Flashcards
Typical features of integral membrane proteins? (3 features) (hint - environments)
Amphiphilic
Involved in transport; Form channels/gates
Contain α-helices and β-barrels
What is used to extract membrane proteins and how?
Why do we gentle detergents over harsh ones?
Detergents are used to solubilise hydrophobic regions of membrane proteins so they can be extracted
Harsh detergents are more likely to denature the protein
Why are techniques like X-ray crystallography and cryo-EM favoured over light microscopy when visualising proteins?
What would happen if X-ray was fired at a single protein?
Can’t use light microscopes, as wavelength of visible light is too long to visualise protein
X-ray would destroy single protein
What makes up a protein crystal? Spaces?
Why do we use protein crystals in X-ray crystallography?
Protein crystals are formed by a sparse network of protein with weak intermolecular interactions; They are not tightly packed and water and buffer fills the spaces
By firing X-rays at multiple copies of the protein, we get some information from each one which we then piece together to make a larger picture
Wavelength of X-rays is just about the same as atoms. WHat happens to the X-rays when they hit atoms?
How did this affect our ability to generate an image?
What is Bragg’s law?
X-ray radiation is scattered
Cant focus the X-rays to make an image
There is reflection from multiple layers which adds to the scattering
How do generate images from X-rays scattering patterns? (hint - align)
We fire in parallel X-rays, which are reflected off the lattice plains and are scattered
Scattered rays only reinforce each other at certain angles, so if you position the angle of the X-rays right, the scattered X-rays align (in register) so that their peaks reinforce eachother and we get a strong signal
What is generated when the scattered x-ray signal is reinforced and detected?
How do we get information from this?
Diffraction pattern of dark spots coming off the different planes of the crystal
Each spot gives information about types of atoms in crystal, way the protein is sitting in the crystal, relative arrangement of atoms
When we scatter X-rays against a protein crystal, what are we actually measuring?
The density of the electrons and their arrangement in space
What do distance and amplitude of a spot in X-ray diffraction tell us?
Spots that are further from the centre, the finer the details we can get from it
Amplitude of every diffraction spot tells us what types of atoms there are
Explain the phase problem
What would help?
Without the phase, we don’t know relative positions of the atoms
This is where having a lens would help (maintains phase information so it isn’t lost); Cant use a lens in X-ray crystallography
What is the main way of solving the phase problem?
When is it more or less straightforward?
Molecular replacement
Use our known structure as search model to match diffraction patterns and figure out orientation of unknown model
Easier if known structure has similar function and the structures are similar
Harder if its a new structure that’s unrelated to known models
If molecular replacement is unsuccessful, how else can we solve the phase problem? (2 techniques)
Anchor points? (hint - reference)
Isomorphous replacement - Bind heavy metal atoms which scatter X-rays strongly to the protein crystal and compare diffraction patterns
Express proteins using seleno-methionine
Scattering from the selenium can sometimes be analysed to start the phasing
By comparing diffraction patterns pre and post treatment we can use altered diffraction points as a reference for all the protein atoms
Vapour diffusion is the usual way to grow protein crystals
What is it?
What are the 2 common methods?
Bring conditions to point were protein solution is supersaturated; Protein precipitates in ordered way and forms a crystal
Hanging (drop at top) and sitting (drop in seat) drop both contain protein, detergent and buffer
Reservoir of buffer is typically salt solution which is more concentrated than it is in drop
By diffusion, the concentrations equilibrate between drop and reservoir; Protein drop concentration increase, and protein drops out of solution to form crystal
What are detergents?
Difference between gentler and harsher detergents?
Bulky polar head group and long hydrocarbon tail
Amphiphiles - Good for solubilisation of membrane components
Gentlest detergents have polar headgroups, not ionic (these are harsher)
What is a trait all detergents have that is undesirable for protein solubilisation?
What does this then mean?
All detergents have slight denaturing (unfolding) activity
Detergent for crystallisation may not be the best for purification
How do detergents work?
What is CMC?
Detergents form micelles in water
- Tail points inwards to form hydrophobic environment
Critical Micellar Concentration (CMC) - Concentration at which detergent goes from monomer and aggregates into micelle
Micelles allow us to extract proteins from a membrane
What happens if detergent concentration is too low?
Why is it important to stay above CMC?
How do detergents and protein interact?
Detergent can intercalate (incorporate) into bilayer
Stay above CMC to extract proteins and maintain protein solubility
Detergent tails interact with hydrophobic protein surface, protecting surface from hydrated environment
- In 3D, detergents form belts around the proteins
How are detergents visualised? (hint - not X-ray, not proton ;) )
Neutron diffraction of crystals soaked in D2O can reveal the detergent at low resolution
Heavy water scatters neutrons strongly allowing for neutron diffraction
How can we improve crystallisation? (hint - fusions)
Increasing crystal contacts by combining proteins
Contacts have to be through polar regions not covered by detergents which is harder for smaller proteins
Increased contacts and more space for detergent micelles is formed through protein/domain fusions or partnering with small hydrophobic proteins
Why is Cryo-EM better than X-ray crystallography?
What is needed?
Cryo-EM doesn’t require crystallised proteins
Just need enough detergent solubilised protein in solution; High protein concentration
How are cryo-EM samples prepared? (4 steps)
- Deposit droplet of solubilised protein onto grid with carbon film (has many holes)
- We then blot the sample, so we have thin film of solution coating the grid; Hopefully with copies of protein in some of the holes in several orientations (top right)
- Rapidly freeze grid in liquid ethane; Creates vitrified water (glass/ice like non-volatile structure; Not crystallised)
- Fire electrons at frozen sample; Can use lenses which is good
- Need a vacuum; Vitrified water won’t evaporate in microscope vacuum
- Low temperature extends lifetime of protein in the electron beam
What are the advantages of using electrons rather than X-rays for imaging? (hint - focused, scattered)
Electron are scattered much more greatly and therefore samples can be very small - Radiation damage is still a problem
Electrons are charged and can be focused with a magnetic lens to form image
- Lens lets us keep the phase so we get a direct image
How are radiation sensitive proteins imaged to high resolution without staining proteins?
Low-dose microscopy
Gives very high detail
How can we get around potential protein damage in cryo-EM?
How do direct electron detectors correct for beam-induced motion?
When electrons are fired at proteins the proteins are damaged
By using many copies of a protein we only damage each one a little bit to get a little bit of information so we don’t need to damage them as much
Still the proteins get damaged and then they move which causes blurring in images
As detectors now are so sensitive we can create a ‘movie’ of multiple frames, allowing us to correct for the motions to get better contrast and detail
How are 3D images reconstructed in cryo-EM?
- How is this better than X-ray crystallography? (hint - rotate)
Why can this be hard?
In X-ray we must rotate the crystal to view it in different directions
However in cryo-EM we have multiple 2D projections of a protein
If we have projections of many different orientations we can piece them together into 3D structure
Difficult as we need to figure out relative angles
What can be done with high-resolution maps generated with cryo-EM?
Can see amino acid chains and build atomic models
We can have a mix of proteins in different conformations and easily separate them out
What is single particle averaging?
Computational technique that allows us to get all the information from lots of different molecules and combine them through image averaging
Supresses random noise and enhances the signal
When does single particle averaging work best?
When we have a molecule in lots of random orientations
How we improve signal/noise? (hint - class average)
By averaging we improve signal/noise of projection images
This is a class average
How are proteins aligned and classified?
Raw images are randomly orientated so we centre and align them into classes based on density distribution
Members of each class are averaged to get a better view with enhanced signal:noise
What are Euler angles and what do they define?
Level of care needed?
How many Euler angles?
Each class average is a 2D projection of the 3D volume defined by Euler angles
It requires great care, especially for small particles with little or no symmetry
3 Euler angles:
- α, β, γ
What is back projection?
Extrude 2D view into 3D column towards origin
Wherever those back projections interact, certain features are strengthened
Inclusions of other views at different angles improves map and you can reconstruct 3D object from the 2D views
How does AlphaFold work?
Training?
Uses deep-learning neural networks to look for patterns
Has a confidence rating, addressing its potential errors
Trained on > 10^5 known structures and sequences in PDB
What does AlphaFold look for when making predictions?
What are co-evolving amino acids?
Looks for “co-evolving” amino acids in related sequences as these are likely to be close in 3D space
Looks for amino acid changes between related sequences and tries to see if there’s a correlation between 1 amino acid changing and another; If there’s a correlation they are “co-evolving”
What are some limitations of AlphaFold? (hint - confidence, interactions, large)
Can’t predict the effect of making mutations
Not so well trained on protein-ligand interactions
Still a bit limited on protein complexes, particularly large ones
Varying degrees of confidence in its predictions
AlphaFold3 is improving these
How did AlphaFold help make sense of previously uninterpretable X-ray or cryo-EM data? (hint - replacement, low reolsution maps)
Making a search model with AI for molecular replacement in X-ray crystallography; Allowed us to solve structures
Fitting low resolution cryo-EM maps against AlphaFold models
AlphaFold models fit nicely within structures, allowing us to see details; Can also see where AlphaFold is slightly wrong and needs tweaking
