Protein X-ray Crystallography

Protein XProtein X--ray Crystallographyray Crystallography--Method to study the 3D structure of Molecules. Method to study the 3D structure of Molecules. Allows understanding of biologicalprocesses at the most basic level: Interactions, Enzyme catalysis, Drug action, DrugDevelopment etc.

Steps in Structure Determination

1. Protein purification

2. Protein crystallisation

3. Data collection

4. Structure Solution(Phasing)

5. Structure determination(Model building and refinement)

Why use X-rays and crystals?

Optical microscopy vs. XOptical microscopy vs. X--ray diffractionray diffraction

What we see is limited by the wavelength of light we are using. You cannot image things that are muchsmaller than the Wavelength of light you are using.X-rays is in the order of atom diameter and bond lengths, allowing these to be individuallyresolved. No lenses available to focus X-rays. Crystal acts as a magnifier of the scattering of X-rays.

Step 1 Purification of Proteinminimum of 5 to 10 milligrams pure soluble protein are required with better than 95% purity

A21kDa

C

21kDa

B

21kDa

cyclophilin-3

Overexpression

Step 1SP Sepharose

Step 2Mono S

Step 2 Grow Crystals

Hanging Drop Method

1 to 5µl protein solution is suspended over a 1 ml reservoir containing precipitant solutione.g. ammonium sulfate solution or polyethylene glycol

Improving Crystal Quality

A: First crystals grown at 40C 30%MPEG, pH 6

B: Crystals grown at 170C30%MPEG, pH 6

C: Crystals grown at 170C30%MPEG, pH 5.6

D: Sitting drop at 170C31%MPEG, pH 5.6

Why use Crystals?

X-ray scattering from a single unit would be unimaginably weak. A crystal arranges a hugenumber of molecules in the same orientation. Scattered waves add up in phase and increaseSignal to a level which can be measured.

Growing useful crystals can be a serious bottleneck in structure solution.No crystal = no data = no structure

Crystals comprised of repeating units.

Unit cellthe basic crystal building block containing

at least one protein molecule

Unit cell may comprise numerousSymmetry elements.

Two ‘molecules’ per unit cell related by a 2-fold rotation:

Crystal packing

Protein crystals contain large solvent channels. Solvent content is usually40% to 60%.This may be utilized in structure solutionand is often used in introducing ligandsinto crystals

Step 3 Collect Data

RIGHT:A single crystal (0.1 to 0.5 mm)

is mounted in a loop on the end of a pin and frozen in a stream

of liquid nitrogen .

LEFT:The crystal is put in the X-ray beam produced by the rotating anode generator and the diffracted rays are measured on an image plate.

Production of X-rays

Synchrotron Electrons accelerated in a storage ring emit strong radiation at all wavelengths between about 0.5 and 4 Å.

Rotating Anode GeneratorElectrons accelerated onto a rotating copper target causing emission of monochromatic X-rays with a wavelength of 1.5 Å.

UK Synchrotron: Daresbury LaboratorySRLRC Synchrotron Radiation Laboratory Research Council

ESRF European Synchrotron Radiation Facility (Grenoble)

Crystal is rotated in X-ray beam and data collected through a broad rotation range

Why do we see spots?X-rays are light waves. These waves can either be in phase or out of phase with each other.

2 waves and their sum.

BraggBragg’’s s LawLaw

Diffraction is treated asDiffraction is treated as sselectiveelective rreflectioneflection::

OOnlynly certaincertain angles of reflection angles of reflection ((θθ)) areareselectedselected, , whenwhen XX--rays of rays of a a givengiven wavewave--lengthlength λλ areare reflectedreflected byby lattice planes lattice planes ((hklhkl) ) withwith a a characteristiccharacteristic interplanar distance dinterplanar distance dhklhkl

λλ = 2d= 2dhklhkl sinsin θθ

Bragg’s Law2dhkl sinθ = λdhkl = spacing between planesλ = wavelength (1.5 Å)

Different sets of planes can be arranged through the crystal (described using Miller indices (hkl)). Each Bragg reflection (black spot on the detector) is the result of a reflection from one set of planes. An X-ray data set consists of measuring the intensity (blackness) caused by the diffracted beams from all possible sets of planes (usually between 5,000 and 50,000 measurements)

Positions of diffraction spots are thusdetermined by the crystal packing.

The act of taking diffraction data resultsin the loss of phase information this cannotbe measured and has to be obtained separately so that the structure may besolved.The Miller indices of the (123) crystal lattice plane

3 Ways to Solve X-ray Protein Structures1. Molecular Replacement- Related model required

2. MAD (Multiple Anomalous Dispersion)- Need synchrotron radiation - Need anomalous scatterers e.g. Selenomethionine

3. MIR (Multiple Isomorphous Replacement)- Crystals soaked in heavy atom solutions (e.g. HgCl2)

Molecular Replacement

If two proteins have a similar sequence (30% identity) or share a similardomain, the known structure can be used to model the unit cell packing of the unknown structure.

Model structure‘Small Cyclophilin’

Structure to solve‘Large Cyclophilin’

Molecular Replacement: Not always straightforward

GCAP - trial model 100% identical to protein of crystalsNo molecular replacement solution has been forthcoming.Model is from an NMR structure.

Cytochrome C4 from Pseudomonas aeruginosa.76% identity with trial model - Pseudomonas stutzeri version.Molecular replacement solution not straightforward and wasobtained using fragments of input model.

Single stranded DNA binding protein from Aquifex aeolicus.25% identity with trial model - E. coli version.Molecular replacement solution was straightforward.

MIR: Multiple Isomorphous Replacement

Native Crystal Heavy Atom Derivative CrystalSoak native crystals in 5-20 mM M+X- solutions

Steps in MIR

- Prepare heavy atom derivatives by soaking native crystals insolutions of heavy atom salts

- Collect native and derivative X-ray data sets- Solve the positions of the heavy atoms- Use this information to calculate phases of the native protein

The effect of added heavy atomThe effect of added heavy atom

The effect of added heavy atomThe effect of added heavy atom

Two diffraction photos superimposed, but displaced vertically.

The differences caused by the additionof a Hg2+ ion can be seen.

MAD: Multiwavelength Anomalous Dispersion

Requires:

• synchrotron radiation so that data sets can be collected at different wavelengths.

• anomalous scatterer in the crystal

At certain resonance frequencies some heavier atoms (e.g. Fe, Se, Hg, Pt, I) absorb radiation and this causes differences in diffraction intensities. Data sets measured at different wavelengths from one crystal can be collected and the differences in intensities used to calculate the positions of the anomalous atoms. This information can be used to calculate phases for the native protein and interpretable maps can be calculated.

XX--ray crystallographyray crystallography

Pure protein

Phases

CrystalsElectron density map

Diffraction Fourier transform

( ) )(2*),,(1,, lzkyhxi

h k l

elkhFV

zyx ++−∑∑∑= πρ

Fourier Summation

Each term is added in with the correct intensity (amplitude) and phase. The ‘object’ is

shown at the foot.

The effect of successive summations on the final

‘structure’: the more terms included, the higher the

‘resolution’.

Importance of correct phases.Trying to solve the structure of a duck.

Take magnitudes from duck(essentially diffraction pattern).

+

Add incorrect phases fromcat.

Result looks like the cat - incorrect

This can be a large problem with molecular replacement. Electron density calculated from an incorrectstructure looks like the model and so the model appears to be correct.

Improve phases and get a complete structureImprove phases and get a complete structure

Initial PhasesInitial Phases

Analysis of electron Analysis of electron density mapsdensity maps

Density improvement(ie. DM, Non crystallographic symmetry)

Building / correctionBuilding / correctionof modelof model RefinementRefinement

Structure analysis, Figure generation,Structure analysis, Figure generation,Paper writing, PresentationsPaper writing, Presentations

Electron density Electron density –– Not self explanatory.Not self explanatory.To obtain any useful structural information some form of intelliTo obtain any useful structural information some form of intelligencegence(Machine/Human) has to interpret the electron density in the for(Machine/Human) has to interpret the electron density in the form ofm ofa model that best fits the data.a model that best fits the data.

What can be interpreted is largely defined by resolution.What can be interpreted is largely defined by resolution.

22ÅÅ11ÅÅ 33ÅÅ 44ÅÅ

Phase errors and unidentifiable sections of density also play a Phase errors and unidentifiable sections of density also play a role role in restricting accurate model building. These can be overcome orin restricting accurate model building. These can be overcome ordecreased.decreased.

What does this mean?What does this mean?Ideal situation – very high resolution excellent starting phases.

Build starting Build starting modelmodel

Initial electron densityInitial electron density

Correct and Correct and add to modeladd to model

Not realistic example Not realistic example --ElastaseElastase essentiallyessentiallyjoining the dots.joining the dots.

More realistic More realistic –– tetramerictetrameric PFKPFK

Initial buildingInitial building With NCS averagingWith NCS averaging

Rebuild and refineRebuild and refineContinue building and refiningContinue building and refining

Once structure can no longer be improved?

Validate protein model for fit to data – R, Rfree.Check chemical geometry of protein model.Analyse structure for questions / answers / relevance.Publish / present.

1. Quality of dataResolution should be better than 3 Å.Completeness of data > 85%.R-merge < 10% (how well multiply measured reflections agree).

2. Quality of the final modelR-factor expresses the goodness of fit between the calculated and observed data.For a random (wrong) structure R= 0.59. A well determined structure has an R-factor between 0.15 and 0.2:

R = Σ ||Fhklobs| - |Fhkl

calc|| / Σ |Fhklobs|

Deviation from standard bond lengths and angles should be about 0.02 Å and 2°.

Φ (phi) and Ψ (psi) angles should fall in the allowed regions of the Ramachandran plot.

A reasonable number of water molecules should be included in the structure(about one water per amino acid residue).

Temperature factors (B-factors) measure the amount of motion in the atoms. They should normally be less than about 30 Å2 (equivalent to an rms vibration of 0.6 Å).

Assessment of Published Protein Structures

Examples of protein crystal structures.Examples of protein crystal structures.Type I dehyroquinase from S. typhi.Enzyme of the shikimate pathway, found in plants, bacteria and funghi. Antibiotic target

Crystal was soaked with substrate and structure subsequently solved.Crystal structure able to show importance of histidine in enzyme mechanism.

+ =

Inhibitors soaked into crystals reveal the importance of lysine in inhibitor design

Human Cyclin dependant kinase 2 (CDK2)

Involved in the cell cycle. Inhibitors have been shown to act as anti cancer drugs.

Linkerresidues

T loopY 15

30 - 41

Sometimes it’s very pretty

Suggested readingSuggested reading

Crystallography 101 http://www-structure.llnl.gov/Xray/101index.html

Crystallography coursehttp://www-structmed.cimr.cam.ac.uk/course.html

Iain McNae

INSTITUTE of CELL & MOLECULAR BIOLOGY

THE UNIVERSITY of EDINBURGHSwann 3

Email: iain@bch.ed.ac.uk