X-ray crystallography – an overview(based on Bernie Brown’s talk, Dept. of Chemistry, WFU)

• Protein is crystallized (sometimes low-gravity

atmosphere is helpful e.g. NASA)

• X-Rays are scattered by electrons in molecule

• Diffraction produces a pattern of spots on a film that

must be mathematically deconstructed

• Result is electron density (contour map) – need to

know protein sequence and match it to density

• Hydrogen atoms not typically visible (except at very

high resolution)

X-ray Crystallography – in a nutshell

REFLECTIONS

h k l I σ(I) 0 0 2 3523.1 91.30 0 3 -1.4 2.80 0 4 306.5 9.60 0 5 -0.1 4.70 0 6 10378.4 179.8 . . .

? Phase Problem ?

MIRMADMR

Electron density:(x y z) = 1/V |F(h k l)| exp[–2i (hx + hy + lz) + i(h k l)]

Bragg’s

law

Fourier transform

Crystal formation

• Start with supersaturated solution of protein

• Slowly eliminate water from the protein

• Add molecules that compete with the protein for water (3 types: salts, organic solvents, PEGs)

• Trial and error• Most crystals ~50% solvent• Crystals may be very fragile

Visible light vs. X-raysWhy don’t we just use a microscope to look at proteins?

• Size of objects imaged limited by wavelength. Resolution ~ /2– Visible light – 4000-7000 Å (400-700 nm) – X-rays – 0.7-1.5 Å (0.07-0.15 nm)

• It is very difficult to focus X-rays (Fresnel lenses) • Getting around the problem

– Defined beam– Regular structure of object (crystal)

• Result – diffraction pattern (not a focused image).

Diffraction pattern – lots of spots

X-ray beam

crystal

Film/Image plate/CCD camera

~1015 molecules/crystalDiffraction pattern is amplified

Bragg’s Law:2d sin = n

End result – really!Fourier transform of diffraction spots electron density fit a.a. sequence

Protein

DNA pieces

(Dimer of dimers)

Interference of waves

• In crystallography, get intensity information only, not phase information

• Need to deconvolute and obtain phase information:

• THE PHASE PROBLEM

How to get from spots to structure?

• Fourier synthesis• Getting around phase problem

– Trial and error– Previous structures– Heavy atom replacement – make a landmark– Ex: Selenomethionine

• Plenty of computer algorithms now

Electron density with incorrect phases

• Red is true structure

The effect of resolutionMore extensive diffraction pattern gives more structural

information = higher resolution

• 6.0-4.5 Å – secondary structure elements

• 3.0 Å – trace polypeptide chain

• 2.0 Å – side chain, bound water identification

• 1.8 Å – alternate side chain orientations

• 1.2 Å – hydrogen atoms

With computational tools, spots become density

Flexible regions give smeared density, often2-3 conformations visible, more than that invisible

Density becomes structure

Need to know protein sequence to trace backbone

Co-crystal structures

• Because of relatively high solvent content, can often “soak in” substrate

• Then can solve structure of protein with substrate bound

• If crystal cracks, good sign that substrate binding or enzyme catalysis results in conformational change in protein

• No longer has same crystal arrangement

NMR vs. crystallography

• Useful for different samples• Generally good agreement• E. coli thioredoxin:

X-rayNMR

Note missing region

Known protein structures

• ~17,000 protein structures since 1958• Common depository of x,y,z coordinates:

Protein data bank (http://www.rcsb.org) • Coordinates can be extracted and viewed • Comparisons of structures allows identification

of structural motifs• Proteins with similar functions and sequences =

homologs

Growth in structure determination

• Might identify a pocket lined with negatively-charged residues

• Or positively charged surface – possibly for binding a negatively charged nucleic acid

• Rossmann fold – binds nucleotides

• Zinc finger – may bind DNA

Function from structure

Domain organization

• Large proteins have polypeptide regions that fold in isolation

• May have distinct functional roles – Example:

glyceraldehyde-3-phosphate dehydrogenase

Protein families

• Similar function and overall structure• But amino acid sequence may or may not be

highly conserved• Limited number of protein domains• Homologs versus structural motifs

SCOP Classification Statistics

Class Folds Superfamilies Families

All 171 286 457

All 119 234 418

Alpha & beta () 117 192 501

Alpha & beta () 224 330 532

Multi-domain proteins 39 39 50Membrane /cell-surface proteins 34 64 128

Small proteins 61 87 135

Total 765 1232 2164

  Structural Classification of Proteins18946 PDB Entries, 49497 Domains (1 March 2002)

(excluding nucleic acids and theoretical models)

http://scop.berkeley.edu/ or http://scop.mrc-lmb.cam.ac.uk/scop/

Have all folds been found?

Red = Old foldsBlue = New folds