Fragment-based screening is now well-established as a powerful approach to early drug ("lead") discovery.
See this step-by-step guide on how to prepare a good ground state model using XCE and PanDDA.
Thanks to Jola Kopec!
|16%EDO, 16%MPD, 50%1,2-propanediol, 16%DMSO|
|33%EDO, 33%glycerol, 33%1,2-propanediol|
|40%EDO, 20%MPD, 20%1,2 propanediol, 20%DMSO|
|50%EDO, 16%MPD, 16%1,2propanediol,16%glycerol|
|42%EDO, 14%MPD, 14%1,2propanediol, 14%glycerol, 14%DMSO|
|16%EDO, 16%MPD, 16%1,2propanediol,16%DMSO, 33% glycerol|
|90% propylene glycol|
|5M Li acetate|
|4M Li formate|
|3.M Mg acetate+0.5MNaCl|
|7M Na formate|
|20:30 glucose EDO|
|100% DMA (dimethylacetamide)|
Below are things we've learnt from now over 30 experiments with over 20 users. (And before that, 12 years of crystallization at SGC-Oxford...)
1. Crystals are not diffracting well, including controls, even though crystal system usually diffracts well.
|Crystals may go old faster than you realise.||••► Prepare fresh crystallisation trays before your visit at Diamond.|
2. Active site is blocked by a disulphide bond between two cysteins.
|If that is not physiologically relevant, protein is sensitive to oxidation.|| |
••► Add redox agent (DTT, TECEP) during sample preparation and crystallisation.
••► Add a redox agent to crystallisation drops via Echo before transferring the compounds.
3. Crystals are not growing reproducibly.
- Was your crystal system was optimised for larger drops / sitting drops?
- Did they really grow reproducibly in the past?
••► A few dozen crystals per plate may be enough
••► Do optimisation screens with very broad ranges of conditions (pH, salt, etc)
••► Definitely try increasing protein concentration and decreasing precipitant - might need a lot more than you expect! (E.g. for protein 8->30mg/ml, take PEG 25->3%.) Go look again carefully at basic crystallization theory and the phase diagram.
••► Definitely do seeding.
••► Look carefully at your purification protocol - is it really yielding consistent protein quality? Be sure gel filtration is the final step. Try additives (arginine) or deterge
••► Check these tips from Douglas Instruments on seed stocks.
4. Crystals are not diffracting reproducibly.
|- Is your crystallisation condition is straight from crystallisation screen? |
- How extensively have you optimised the crystallization?
••► Do more optimisation (see above)
••► try many more cryo-protection protocols, including quick-dip-and-freeze, and gradual increase in cryoprotectant (annoying but supportable). Don't forget sugars!
|5. Can't reproduce crystals in SwissCi 3-drop plates, or any low-volume sitting drop for that matter.||Was crystal system was optimised for larger hanging drop format?|| |
••► Conditions can change a LOT between formats: get SwissCi plates and optimise aggressively - all the things in row #3 above.
••► The good news: you know the protein likes to crystallise, so it's unlikely to be format specific - keep looking.
••► Take seeds from the hanging-drop crystals, and seed the sitting drop conditions.
••► We will soon support other sitting drop plate formats (it's taking time for complicated but boring reasons)
••► We have harvesting protocols that allow you to pick crystals from hanging drop slides and transfer them to the compounds pre-dispensed in SwissCi plates. It is a significantly slower, though, and crystals need to like it, so we suggest it to be last resort.
••► Protein engineering may be necessary: SER, crystallization epitopes, different construct lengths, crystallization chaperones... obviously a lot more work.
6. Drops dry very quickly once the seal is removed.
|Crystallisation condition may contain a volatile solvent or high salt.|| |
••► Place filter paper soaked with some solvent (e.g. iso-propanol, water) place around the shifter mounting aperture.
••► Add a small volume of reservoir solution to all drops before transferring compounds.
••► Search for easier conditions (see #3 above.)
7. Crystals dissolve after adding compounds (solvent) by Echo.
|Your crystal system may be sensitive to DMSO.|| |
••► Do a more extensive solvent tolerance test
••► Try even lower DMSO concentrations: the first priority is crystal survival; maximising compound concentration is much less important.
••► Try very short soaks - one user's system required <5secs.
••► Try different solvents. Ethylene glycol has been effective for several systems; and we have interesting preliminary results using hydrotropes.
8. Crystals look "dirty" or diffraction images are bad.
|Your crystals may be growing as sub-microscopic clusters or stacks of plates.|| |
••► Do all the optimisation stuff in #3, especially seeding (!!)
••► If data quality from processing is okay, then you may not have a problem: modern data collection and processing lets you get away with murder, and besides, data don't actually need to be particularly accurate (it's not SAD phasing!)
9. Crystals grow in several indistinguishable crystal form from the same condition.
|Some proteins are like that...|| |
••► If resolution is not affected, you don't have a problem: during processing, crystal forms can be clustered retroactively and handled individually.
••► Mount a few dozen crystals, and carefuly compare outcomes to morphology: it may be visible after all.
••► Explore more cryoprotection protocols: since the crystal packing evidently lies near a phase transition, certain prototols might drive the system to a particular phase. See #4 above.
••► Do more aggressive optimisation as in #3 - don't forget seeding with the best crystal form!
10. Showers of crystals quickly form after removing the seal.
|Your crystallisation condition remains meta-stable even after crystals have grown.|| |
••► If they're salt crystals: search for lower-salt conditions.
••► If they're protein crystals, do optimisation to find a lower-saturation condition - try seeding!
••► Try adding reservoir solution to drops before transferring compounds.
••► Characterise how few crystals you can have open at once, by seeing how quick the process is, and then modify the soak/harvest workflow accordingly. It makes everything a lot more tedious, though, so last resort.
|11. Crystals are long needles so they need large loops, meaning they're often missed by the beam despite loops being "properly" centred||Some crystals are like that...|| |
••► See if you can fish the crystals to lie consistently across the centre of the loop (tough!)
••► You may need to do assisted data collection: it requires you watching every crystal and fixing any mis-centring - it's a bit tedious but we rigged the software to make it very easy.
••► (We're working with Mitegen to develop new needle-supporting loops.)
••► (We're making X-ray centring more robust and fast - but it reduces beamline throughput 3x, so will always be plan B>)
|12. Crystals don't survive solvent for more than a few seconds.||Some crystals are like that...||••► We can customise the workflow so that compound addition happens straight before harvesting. It's more tedious, but hey, your crystals are boss!|
|13. Crystallisation drops and reservoir solutions are all mixed up at the end of the soaking time.||Might have been busy in the lab - blame someone else for mishandling your plate!||••►Always leave your crystallization plates out of the way on the designated shelves - we need you to work tidily anyway.|
|14. A skin forms in the drop and makes compouns dissolutions and/or crystal manipulation difficult.||It always grew like this.||••►Change your crystallisation volumes. For instance, a system which showed skin when grown from 500nl+500nl drops stopped showing skin when grown from 200nl+200nl (a preferred volume, anyway).|
Diamond Light Source is the UK's national synchrotron science facility, located at the Harwell Science and Innovation Campus in Oxfordshire.
Copyright © 2018 Diamond Light Source
Diamond Light Source Ltd
Harwell Science & Innovation Campus