Protein Crystallization Hits

One of Naomi Chayen's best advice on crystallization optimization she's given me was "always filter your sample for optimization". In more detail: she advocates filtering protein samples with filters of increasingly smaller mesh size (such as 0.22 micron, 0.1 micron and 300 kDa MWCO). The filtered solutions are then each used to set up optimization crystallization trials. That's simple to do when sample quantity is ample since you'll lose ca.5-10 ul in filter dead volume for each filtration run. Naomi has described the procedure in several of the talks that I had the pleasure to attend. And more recently she's condensed her filtration results and published it in a laboratory note in the Journal of Applied Crystallography (2009, 42) fittingly titled: "Rigorous filtration for protein crystallization". I like the word 'rigorous' in this context - sounds like: forcing out the rubbish. She's making the point that "Filtration is relevant to all methods of crystallization for both screening and optimization".

The thinking here is of course to clear up the sample by removing fines from chromatography resin, dust, aggregated protein and to reduce the number of nuclei, hence suppress heterogenous nucleation. The latter of course comes in handy when the goal is to convert showers of crystals into one single well diffracting crystal. The table she shows is impressive. The number (reduced from >1000 to 0), size (10 um to 600 um) of crystals grown from aliquots of the same protein solution under otherwise identical conditions depended on filter mesh size used (she compared unfiltered, 0.22 um, 0.1um, 0.1 um and 300 kDa MWCO).
The advantages of filtration are clear (sic!): growing better crystals right away and creating solutions that are ready for seeding. The biggest advantage of rigorous filtering in my mind however is that this is an additional tool to optimize without having to change the crystallization conditions.

Big thanks to Naomi!

Peter

Thinking ahead and setting some protein aside for optimization screening? Good idea, but how do you store this precious liquid? Keeping it 'on the bench' may be a bad idea. Even a minor contamination of the sample with proteases can wreak havroc with the target protein in a matter of days or weeks. While cooling on ice will slow down proteolytic fragmentation quite a bit it will not prevent it. So you want to freeze the protein for long term storage. And of course the protein sample ought not to be modified by addition of cryoprotectants such as glycerol. So - what's the best practice for storing protein samples?
Freezing as drops dripping in liquid nitrogen and harvesting the 'pearls' is the method of choice in many labs. Drop freezing has its issues though, so Deng et al. (An improved protocol for rapid freezing of protein samples for long-term storage. Acta Crystallogr D Biol Crystallogr. (2004) 60:203-204) researched this topic systematically and came up with these recommendations: 

1. Use thin-walled PCR tubes for shock-freezing - PCR plates work as well
2. Fill with volumes less than 50 ul / tube; for example 20 ul of protein sample in each tube
3. Thaw fast: thaw in 'the investigator's hand'

   

Take note that thawing seems to be just as important to get right as the cooling process. Not *on ice* but in 'the investigator's hand'.

(I don't know why, but the latter somewhat reminds me of that 189X chemistry paper that stated [translated from German] "... the laboratory servant shall stir the mixture until he is exhausted". We've come a long way...)

Cheers,
Peter

What's a good systematic way to vary all components of a crystallization hit in a single 96-well crystallization tray? For a typical 3 component formulation, Paul Reichert a crystallographer at Schering-Plough, has worked out a remarkably useful schema. Here's what he came up with: At first divide the 96 well tray into 4 quadrants, with 24 wells in each quadrant: 

Reichert systematic optimization quadrant schema for optimizing protein crystallization. Starting point is a single, 3 component hit.

The image above tells much of the story: In quadrant II (beige) the pH and salt concentration are varied systematically in a gradient-like fashion while the precipitation concentration is kept constant. In quadrant III (green) the precipitation and salt concentration is varied while the pH is kept at its original pH and, in quadrant IV (orange) the precipitation concentration and pH are varied while the salt concentration is kept constant. What about quadrant I? Here only the precipitation concentration is changed, keeping pH and salt constant. These are not conditions to be tested in triplicates but allow to assay three different protein concentrations.

Applying this scheme to single crystallization hits provides a lot of meaningful data because it isolates 4 different crystallization factors: (i) pH, (ii) protein, (iii) salt and (iv) precipitant concentration. While not perfect, this is an efficient schema that has often resulted in greatly improved crystal quality.

Since this has worked so well, Emerald provides this type of customized optimization screen as "Reichert Optimization Screen". This is one of several optimization screen services we offer. Just tell us about your hit condition and we'll design, prepare and send you the screen.

Enjoy your crystals,
Peter

P.S. We have recently incorporated this optimization schema into Escreen Builder, a free online tool to calculate an optimization screen using this tool.

Have you ever heard somebody claim that "Protein Crystallization is half art and half science"? I have, many times. This notion that there's something artistic or even magical about growing protein crystals has always bugged me. A lot actually. In my mind protein crystallization is a science. 100% reproducible science that is, with no dependence on the position of the planets or the person setting up the trial. I suspect that poor control over experimental parameters combined with the critical effects that come into play during nucleation and crystal growth are to blame for difficulties in reproducing crystals. Some people may cope with this issue by labeling crystallization as an 'art'.

A complicating factor has its roots in the many possible crystallization parameters as well as their priorities in crystallization success for one protein target vs. another. For instance, one crystallizer may find that a specific pH and temperature are important to grow crystals of a particular target. A second crystallizer, working on a different protein target, however finds that the purity of the sample prep is the only parameter that determines crystallization success.

Tracking down these main crystallization parameters for each protein we work on is at the heart of what we do when we optimize protein crystallization. Most of the time these are protein purity, concentration and a particular composition of a crystallization cocktail. For others, sometimes seemingly exotic parameters such as precise drop mixing regime, surface/volume drop ratio and the kinetics of evaporation, exposure to light, etc. (long list) are key to reproducibility. Nobody claims that this is a simple task. But it's a science, not an art.

 Protein Crystallization Art :)      [courtesy of Jeff Christensen]

1. No need to get excited about crystals in the drop if there are salt crystals the 'mother liquor'.
2. Those clear but irregularly shaped objects without facets that look like glass shards that show up in some hanging drops are indeed small pieces of glass that broke off when the cover slides were made. 
3. Those soft-looking, clear objects on the bottom of the sitting drop wells are not 'pre-crystalline material', they're plastic tray manufacturing artefacts.
4. Shiny precipitate is a good start for crystal growth optimization.
5. The color that's associated with very small crystals is not the chromophore of the protein but stems from chromatic aberration of the microscope lens.
6. Better save a small quantity of shock-frozen protein solution and use as a positive control in follow-up optimization trials than making another prep.
7. Take an image of the first crystals right when I see them for the first time. Take any image - use my phone if necessary. Otherwise nobody will believe that there were indeed crystals at 16C that 'melted' after tansferring the crystallization tray to the 21C lab with the high-resolution digital camera mounted on the microscope.
8. It doesn't hurt to take crystals out of the drop, solubilize them in SDS and run them out on a gel. If the main band does not run with the target protein, something's fishy - or something interesting is going on.
9. The 'flittery precipitate' that formed in the protein solution right before setting up the crystallization trial may be showers of microcrystals. Better check a small sample with a high-magnification microscope.
10. Do label the crystallization tray, not the cover (of the batch-under-oil crystallization experiment).
11. I will forget the meaning of "optimized hit condition #4" when it's paper writing time.