X-ray Crystallography

Question 1

Give an overview of X-ray crystallography?

Answer 1

In X-ray crystallography we shoot them at protein crystals
The output of x-ray diffraction is a 3D electron density map
Single proteins can be mapped up to viruses/ribosomes
Not a size limit
Will be used extensively in structure based drug design

Question 2

What is the history of x-ray crystallography?

Answer 2

Sir William Bragg at University of Leeds studied X-ray diffraction peaks
1913 - they solved the structures of NaCl and diamond - essentially invented x-ray crystallography
1915 awarded a Nobel prize (as father and son)
Astbury - first to define alpha/beta protein conformations (1931) and first x-ray structural analysis of DNA (1938)
He also defined molecular biology in print for the first time

Question 3

Describe x-rays?

Answer 3

They are part of the electromagnetic spectrum
The waves are vectors - wave amplitude and wave phase
When an electron is hit by an x-ray - it starts vibrating with the same frequency as the x-ray beam
As a result the secondary beams will be scattered in all directions

Question 4

Describe the scattering of a molecule

Answer 4

A molecule is composed of many electrons
Each electron will scatter secondary radiation upon exposure to X-rays
The scattered secondary beams will interact and cause interference
The scattering from a molecule is dependent on number of and distances between electrons
Scattering from molecule is dependent on its structure
If we were to know the amplitudes and phases of scattered X-rays, we could calculate the structure of molecule

Scattering from a single molecule is far too weak to be observed
But - if molecules are all oriented in the same way (like in a crystal), the scattering from individual molecules will be multiplied in certain directions

Question 5

How do we grow protein crystals?

Answer 5

A protein solution mixes with something that will reduce its solubility - it will either precipitate or crystallise (with an energetically conformation)

Types - sandwich drop, sitting drop or hanging drop
They are all sealed containers with a reservoir at the bottom
When the container is sealed - vapour diffusion pulls the water out of the protein solution into the reservoir (this takes 2 days)

We start with a protein that is undersaturated and as we reduce solubility we move into other zones of saturation
Metastable zone, nucleation zone and then precipitation zone

Question 6

How does energy change throughout crystal growth?

Answer 6

Initially this process needs energy towards non-specific aggregates and then specific aggregates
Once critical nuclei has been formed there is a decline in energy required to continue forming the crystals

Question 7

Once some proteins have crystalised, what happens?

Answer 7

We screen many different chemicals and sometimes magically find initial hits
They are then optimised and eventually form a large ‘perfect’ crystal - difficult to achieve
The crystal is plunged into liquid nitrogen to keep them frozen and protected against the x-rays

Question 8

Describe protein crystal packing?

Answer 8

The crystals are in a solvent environment
Typically - 50% but can be 30-80%
They can give artifacts but most of the crystal structures are very accurate
With these crystals we need to understand how they fill space - therefore they can have multiple rotations

Question 9

What happens once the diffraction pattern has emerged?

Answer 9

Once we have a diffraction pattern - we need to understand the relationship of the proteins in the crystal between each other
This is called how they pack together or the symmetry

A unit cell is the smallest unit of volume that contains all of the structural and symmetry information and that by translation can reproduce a pattern in all of space
An asymmetric unit is the smallest unit of volume that contains all of the structural information and that by application of the symmetry operations can reproduce the unit cell

In x-ray crystallography we solve the structure of the asymmetric unit

Question 10

What are the definitions of structural information, symmetry information and translation?

Answer 10

Structural information- the pattern (atoms) plus all surrounding space

Symmetry information- mirrors, glides, axes, and inversion centers (this is just rotation for proteins because all amino acids are l-isomers)

Translation- motion along a cell edge the length of the cell edge

Question 11

What is Braggs law?

Answer 11

When x-rays hit the crystals they can act constructively or destructively
Constructive - the scattered beams are in phase and they add up (the peaks and troughs line up on the diffraction side)
Destructive - the scattered beam is not in phase and they cancel each other (peaks and troughs don’t line up in the diffraction)

nλ - 2dsinθ

Question 12

Why do we see spots in the diffraction pattern?

Answer 12

Why we see spots - a combination of constructive and destructive interference, a bit like the double slit experiment

Question 13

What does the diffraction pattern show?

Answer 13

We can’t focus x-rays diffracted from a crystal
We use maths to put all the information of what the crystal has interacted with back together

Resolution - the further ‘out’ your data does the better the resolution (the circle shows the resolution shells)
We resolve every atom in the structure

Question 14

What is workflow of x-ray crystallography?

Answer 14

Express and purify (recombinant) protein 
Produce and optimise crystals 
Shoot with X-rays
Collect and complete data sets (multiple diffraction images - rotations) 
Phasing (more data sets)
Build a model
Refine
Validate
Deposit and write paper

Question 15

Describe the diffraction experiment?

Answer 15

From a rotating anode the primary x-ray beam is fired
Focusing mirrors (monochromator) position the x-rays toward the crystal
The crytal in contained in a 4-circle gonoimeter - which rotates the crystal
The x-rays are diffracted and caught by the detector

Question 16

How do we solve the structure of a molecule in X-ray diffraction?

Answer 16

Phasing
Fourier Transform
Electron density
Building and refining

In order to build up the position and intensity of the spots from the diffraction pattern we have to add the position intensity and phase

Question 17

What is fourier transform?

Answer 17

The electron density distribution of molecular structure and its produced diffraction pattern
are Fourier transforms with respect to each other

Question 18

What are phases?

Answer 18

The phase part of the wave often contains valuable information on the studied specimen
From the amplitudes of the scattered waves - we can measure the intensities of the diffracted waves from many planes of the molecule
The phase is how we offset all the waves of each plane when we add them together to reconstruct an image of our molecule

Question 19

What is the phase problem?

Answer 19

With detector you can measure only the intensity of reflections
The information about phases is lost – there is no such thing as “phase meter”
This means, you must obtain phase information in some other way
For small molecules (<100 atoms), direct methods exist
Meaning you can calculate phases from amplitudes without any extra information
Proteins are far too big to use direct methods, so other tools are developed

The missing phases are supplied by additional phasing experiments or in the form of model phases via molecular replacement

Question 20

What are some phasing techniques?

Answer 20

Multiple Isomorphous Replacement (MIR)
Multiwavelength Anomalous Dispersion (MAD)
Single wavelength Anomalous Dispersion (SAD)
Molecular Replacement (MR)
[Direct Methods]

Question 21

Describe isomorphous replacement?

Answer 21

By introducing heavy atoms in protein crystal (by soaking), the diffraction pattern can be altered
It is possible to determine positions of heavy atoms and from them the phases
You need to use at least 2 different heavy atom soaks

Problems:

1. Unit cell dimensions of crystal might change upon soaking
2. Crystal might get destroyed upon soaking and not diffract at all
3. Heavy atom ions might not bind in well defined places

Question 22

Describe multiple isomorphous replacement?

Answer 22

Using more than one isomorphous replacement is needed because of the structure factor and its direction change
Where they all overlap is the correct point

Question 23

Describe multi-wavelength anomalous dispersion?

Answer 23

By replacing the S in methionine to selenium - this cause different diffraction pattern
We can look at the difference between the diffraction patterns of the two elements

Question 24

Describe molecular replacement?

Answer 24

Currently the most common technique
Applicable only if a similar structure already exists (at least 25% sequence identity)
The phases of known structure are combined with intensities of unknown
Before that, the known model has to be in silico placed in an artificial unit cell in the same orientation and translation from origin as in the structure of interest
We need to rotate an image and then translate in order to get different models to overlap
Once we have these position models we can do a Fourier transform and determine the intensity of the spots

Problems:
May not work, if unknown structure is less than 30 % identical to the known structure
Model bias

Question 25

What is model building?

Answer 25

Easy in molecular replacement
More difficult if no initial model is available
Unambiguous if resolution is high enough (better than 3.0 Å)
Can be automated, if resolution is close to 2Å or better

Resolution is essential as it can be hard to tell where the atoms go

Question 26

What can we focus on looking at electron densities during model building?

Answer 26

Electron densities of the amino acid side chains
We build proteins, water, ligands, ions, alternative side chain conformations etc…
Anything that makes biological/chemical sense

We don’t see - hydrogens, unstructured regions, protonation states of ligands/cofactors

Question 27

Describe refinement?

Answer 27

(Semi)-automated improvement of the model, so it explains the observed data better (experimental diffraction data - intensity of the spots)
The phases get improved as well, so the electron density maps get better
We can use the electron density maps to guide us - using Fourier transform we can see areas that should/shouldn’t be there between the electron density map and the model that we originally produced
Or we can use out knowledge of chemistry

Correlation coefficient between the electron density map and the model is much higher through refinement

Question 28

What is the R-factor?

Answer 28

AKA residual factor or agreement factor
This is a measure of the difference between the observed and computed intensities
Note that the structure factor F is related to intensities from the diffraction pattern
We want low scores for the R value

Question 29

What is validation?

Answer 29

Assessment of the final(?) model quality

How the geometry of peptide bonds look (Ramachandran plot)
Anything in the disallowed region?
Are non-covalently bound atoms far enough from each other? (no atom clashes)
Are residues “happy” in their environment? (hydrophobic in core, polar on surface)
Are the hydrogen donors/acceptors satisfied?

Clashes, rotamers, bond length/angle etc…

Question 30

What happens once you think you have a final structure?

Answer 30

To really test we need to test the biological implications, make predictions (and mutants) and test them
If you understand the structure and the function - write the paper and deposit the protein in the PDB
Useful so you can go back when there are new refinement techniques and others can use your work

High impact journals have not as good structures - the are newer structures that don’t necessarily have other similar structures and they are rushed so not as refined