1. Crystallography: An introduction Harma Brondijk Crystal and Structural chemistry, Utrecht University

2. X-ray Diffraction Electron density Pure protein Construct Crystal Crystallography   Phase problem  3D structure

3. Why X-ray crystallography? (Light) microscope: Limitations: Object needs to be larger than the wavelength of the light (visible light 400-700 nm, atoms = 0.15 nm apart) X-rays(0.08-0.6 nm) cannot be focussed by lenses Molecules are very weak scatterers X-ray diffraction of a crystal A crystal contains many molecules in identical orientation Diffracted x-rays of individual molecules ‘add up’ (positive interference) to produce strong reflections Computers can simulate a lens and reconstruc the image (Fourier transform)

4. Growing crystals Hanging drop: 1 μl protein solution+ 1 μl reservoir solution Reservoir: precipitant solution eg. 1 M NaCl or 30% PEG-4k [precipitant] and [protein] slowly rise as drop equilibrates with reservoir precipitation Nucleation & growth growth Soluble protein [protein] [precipitant]

5. Getting your data X-ray data are measured on frozen crystals (~100K) Frozen crystal mounted in loop for X- ray data collection For high quality X-ray data collection extremely intense synchrotron beam lines - like here in Grenoble - are used In house X-ray data collection set up

6. Fourier transform (phase problem) Raw data: Thousands of intensities of reflections Electron density Each diffraction spot (reflection) contains information on the position of every atom!

7. The degree of order in the crystal determines the quality of the diffraction data and ultimately the quality of the final atomic model “low resolution” “high resolution”

8. The precision of the atomic model is mainly determined by the maximal resolution to which the crystal diffracts X-rays d = 4 Åd = 3 Å d = 2 Å d = 1 Å Atomic resolution

9. Some real life examples 1.8 Å structure: core vs surface loop at 2σ 3.1 Å resolution, well ordered core of the protein

10. What’s a crystal structure? Different representations of an Fab fragment of monoclonal antibody 82D6A3 bound to the collagen binding domain of human von Willebrand factor

11. Things you - as a potential user of crystallographic data - should know about crystals and crystal structures

12. Protein crystals contain a lot of solvent and are held together by a limited number of weak contacts between protein molecules  2 Glycoprotein I ~90% solvent (extremely high!) Acetylcholinesterase ~68% solvent Typical solvent content 40-60% Solvent channels allow diffusion of compounds into crystal Often these compounds can reach the active or binding site Often enzymes are active in crystalline state

13. Two types of solvent: ordered and disordered Ordered water molecules show up as discrete blobs of electron density in contact with the protein or with other ordered water molecules Disordered water regions show up as featureless (flat) electron density

14. PDB files: Basically just simple tekst files At the top: information about the crystal: –Which proteins/ligands etc –Crystalization conditons –How was the structure solved –The resolution –Some usefull statistics to judge the quality of the crystal –How to get from the structure to the biological unit –Remarks about missing bits etc. Crystal parameters: cell dimensions/space group A list of all atoms in the structure

15. A crystal structure according to the protein data bank (PDB) ATOM 25 N ASP A 928 19.062 9.157 35.067 1.00 4.73 N ATOM 26 CA ASP A 928 19.770 10.123 34.232 1.00 4.58 C ATOM 27 C ASP A 928 19.075 9.938 32.899 1.00 4.56 C ATOM 28 O ASP A 928 19.074 8.824 32.351 1.00 5.39 O ATOM 29 CB ASP A 928 21.259 9.776 34.071 1.00 3.13 C ATOM 30 CG ASP A 928 22.112 10.245 35.233 1.00 5.52 C ATOM 31 OD1 ASP A 928 21.693 11.114 36.025 1.00 5.42 O ATOM 32 OD2 ASP A 928 23.239 9.742 35.349 1.00 7.93 O ATOM 33 N VAL A 929 18.417 10.985 32.405 1.00 3.68 N ATOM 34 CA VAL A 929 17.726 10.864 31.125 1.00 4.63 C x,y,z coordinates (Å) occupancy Isotropic B-factor or temperature factor is a measure of the mobility of an atom B (Å 2 ) = 8  2, where is the mean square atomic displacement

16. At typical resolutions (1.8 Å or worse) The electron density of hydrogen atoms is not resolved (and no hydrogen atoms are present in the pdb file) The electron densities of C, N, and O atoms are all rather similar

17. Position of N and O atoms in Gln (and Asn) side chain must be inferred from hydrogenbonding network H H H ?? Asp Main chain carbonyl Main chain amide

18. The same holds for the orientation and protonation of the imidazole ring of histidines ?

19. A pdb file may contain residues for which no, or only limited electron density is visible No density for amide N of glutamine ? ? Break in side chain density of glutamate

20. Sometimes the electron density suggests two side chain conformations but often only one is modeled in the pdb file Threonine side chain conformation present in pdb file Alternative conformation that is also compatible with electron density

21. The interpretation of dynamic loops in the pdb file may be tentative Well defined  -strand in the core of a protein: atomic positions are reliable Flexible loop at the surface of a protein: atomic positions are not well defined

22. Look at B-factor distribution! Protein coloured by B-factor: Well defined regions have low B-factors (blue/green) Poorly defined/more mobile regions have high B-factors ( yellow/orange/red)

23. A protein molecule is dynamic The electron density is a spatial average over all molecules in the crystal and a time average over the duration of the X-ray data measurement Multiple discrete conformations of a residue in different molecules are superimposed. Damage caused by X-rays may change the protein (mainly breaking of disulfide bonds) A crude description of dynamics is provided in the pdb file as the isotropic B-factor Some dynamical aspects evident in the electron density are lost in the pdb file

24. Reading a crystallography paper: Judge the quality of the data:  R merge : 0.05-0.10 good, 0.1-0.15 acceptable  I/σ = signal/noise >2.0  Completeness  Redundancy  R work /R free :  difference < 0.05,  R work ≈ resolution/10  Deviations of known geometry  waters: at 2 Å ≈ 1 water/residue, at > 3Å usually none

25. More and more structures: learn how to use them!

26. Crystallography and drug design and lead optimization

27. The crystal structure of a protein-substrate complex can serve as starting point for structure-based drug design Relenza was developed starting from the crystal structure of influenza virus neuraminidase with bound sialic acid Guanidino group provides additional interactions

28. Structures of bird-flu neuraminidase reveal new cavity that could be exploited in drug design Russell et al., Nature 443, 45-49 (2006)

29. Can X-ray crystallography contribute to lead discovery? -Development of high through-put (HTP) methods in crystallography has considerably reduced the time needed to solve a crystal structure while minimizing the need for human intervention -This now allows for screening of medium sized compound libraries (~1000 compounds) Library used in traditional HTP screen –10 6 compounds –Mw 300-500 Da Library used in X-ray based screen –10 3 compounds –Mw 100-250 Da

30. Technical advances enabling HTP crystallography Automated set-up of crystallization in 96-well format (100 + 100 nl drops Automated imaging of 96- well crystallization plates Automated crystal transfer from liquid nitrogen to X-ray beam  Stronger in house X-ray sources  Automated beam lines at synchrotrons  Improved software for automated interpretation of ligand density

31. Expanding hits into larger and higher affinity compounds -Joining fragments by an appropriate scaffold (C) -Grow a fragment to fill neighboring pockets (d) Electron densities of initial fragment (left) and expanded fragment (right)

32. Simultaneous binding of two compounds from a mixture can be detected Top: Electron density of two compounds bound simultaneously together with the two automatically built compounds. Bottom: Electron densities from individual soaks shown in different colors

33. Drug design cycle Crystallise complex Design better compound in silico Synthesize inhibitors Test inhibitor

34. Contribution of X-ray crystallography to drug design and discovery Lead optimization –Well established –All major pharmaceutical companies do it –Numerous drugs on the market and in pipeline Lead discovery –Promising, but quite recent –Performed in small companies (that have collaboration agreements with large pharmaceutical companies –Must still prove its value