The JCSG has published a study that shows the pH's used and the pI's of the proteins. There are clusters of proteins with similar pIs that are easier to crystallize. However the pH's used are clustered around the pH's that are most often used in crystallization screens. It was commented that even when the data are normalized, there seems to be two crystallizability peaks occurring (see: Slabinski, L. et al. 2007 Protein Science 16:2472-2482.). For proteins <345 aa, there is a crystallization peak for those with a pI of 5-6 and again at a pI of 8-9. For proteins >345 aa, there is only one crystallizability peak, and that is for the proteins with a pI of 6-7. Crystallization screens typically have nominal pH ranges of 4 to 9 (nominal meaning the pH of the buffer is given, NOT the pH of the final solution).

It was reported that most proteins in the PDB have low pIs. Is this a general tendency? (Is this really true? Most E. coli proteins have a pI around 4.5, but is this true for all the proteins in the PDB?). One person said that when they have proteins with a high pI (>8) that won't crystallize, they use reductive methylation. This reduces the pI and also reduces the solubility of the protein, which is often helpful. One person stated that the pI needs to be determined experimentally, one cannot trust the answer you get from ExPasy and similar web sites as they only calculate the theoretical pI. Others said that a good estimate can be obtained from the sequence. Set up the crystallization experiment far from the pI (because the protein is least soluble at its pI). One person uses low pH for proteins with high pI. It was mentioned that Molecular Dimensions has a screen called ZetaSol, that allows high, neutral or low pH to be used as appropriate, depending on the pI of the protein.

It was reported that freezing had always worked in several people's hands, in hundreds of cases. It worked even in cases where the crystals were unstable at 4 degrees Celsius. Another person reported that in one case seed crystals were stable at 22 degrees Celsius, but did not work when they were frozen. They seemed to freeze okay, but dissolved when they were defrosted. Two people commented that the potency of seed stocks could be increased by freezing. This was attributed to the freezing breaking up the seed crystals into smaller fragments. Cross-seeding can work when there is homology between the two proteins. However it was reported that they have never seen a case that could not be explained by the addition of chemical additives in the seed stock, e.g., a metal ion. One person reported a case where self-seeding doesn't work, but cross-seeding does (in this case it is the same protein, but different space groups).

A show of hands indicated that putting down the protein solution first is far more popular. One person reported they did the opposite, pipetting reservoir first then protein. One person put down the reservoir solution first because it could be viscous, and this order gives the best mixing (but others pointed out that they did not want mixing). One person reported that either order was okay, but that it is essential to be consistent, because the results will be different from the two sequences. In one case, dispensing a peptide first, then ether, gave crystals. Dispensing the ether first then the peptide gave precipitate. It was suggested that dispensing high molecular weight PEG first could result in drying out or the absorbance of water. This implies that PEG should be dispensed first, because protein will almost always dry rather than absorb water! One person reported it was shocking that there is so little agreement, and that there should be well-established Standard Operating Procedures (SOPs) so that outsiders could receive clear guidance. One person said we should look into this as a community. Editor noted what he was reacting to was not this discussion, but the one we had about how to handle the Deep Well blocks. Someone asked how often we should renew our Deep Well blocks (containing the crystallization solutions.) When they stand for any longer period of time, a concentration gradient forms. One person reported using a shaker to try to eliminate the gradient and it does not work. One person said to invert the Deep Well block ten timesx, then centrifuge it a few minutes at very low speed (500 rpm) to avoid droplets on the lid. You don't want to cross-contaminate the wells when you pull the lid or tape sealer off. But there was no good agreement on how to handle this question and then someone said we should establish standards.

There are many good examples of the benefits of dehydration of crystals after growing them. A show of hands showed that a third of the audience has used the method. Some people increased the concentration of PEG by a third. Others used a 4% higher concentration. Extra sodium chloride can be added to harvesting solution. Overnight soaks are also used. One person a technique where drops are dried out by evaporation until a thick glop is formed. Sometimes this causes twin crystals to fall apart. (Acta Cryst. (2005). D61, 1173-1180 [ doi:10.1107/S0907444905019451 ])

Post-crystallization treatments for improving diffraction quality of protein crystals, B. Heras and J. L. Martin Abstract: X-ray crystallography is the most powerful method for determining the three-dimensional structure of biological macromolecules. One of the major obstacles in the process is the production of high-quality crystals for structure determination. All too often, crystals are produced that are of poor quality and are unsuitable for diffraction studies. This review provides a compilation of post-crystallization methods that can convert poorly diffracting crystals into data-quality crystals. Protocols for annealing, dehydration, soaking and cross-linking are outlined and examples of some spectacular changes in crystal quality are provided. The protocols are easily incorporated into the structure-determination pipeline and a practical guide is provided that shows how and when to use the different post-crystallization treatments for improving crystal quality. One person reported they increased PEG concentration from 40% to 70%. This increased resolution from 10Å to 2.8Å. This person then dropped the concentration from 40% to 20%. This caused cracking of the crystals which then rapidly healed to give 2.5Å. This shows that it is hard to make predictions. There was a consensus that annealing often works. It is thought to cause shrinkage of the crystal, which increases the number of hydrogen bonds. Dehydration and increasing PEG concentration often works.

It was recommended 200 conditions (two plates) at two temperatures for an initial screen. However larger screens are also acceptable, but remember that this strategy generates a lot of images to look at. It was noted that large screens already have a lot of duplication in them, with identical or very similar conditions. One person reported larger screens do not generate significantly more hits.

Proteins with over 500 amino acids and those with fewer than 100 were said to be more difficult to crystallize (Slabinski, L. et al. 2007 Protein Science 16:2472-2482). It was said that a paper by Tom Peat and Janet Newman showed that high molecular weight proteins are crystallized with lower precipitant concentration on average (Peat et al. Acta Cryst. (2005). D61, 1662-1669).

Start at 20 degrees Celsius and only change to 4 degrees Celsius because you don't want to work in the cold room for optimization! However, at least one person does the opposite of everyone else, starting at 4 degrees Celsius then changes to 20 degrees Celsius. It was reported that data mining has shown that room temperature is the most popular, and also that the trend is towards higher temperatures. Light scattering can tell you whether solubility increases or decreases as the temperature is raised. The recommendation is to use the temperature at which the protein is most soluble. Jose Antonio Marquez' results indicate that one should use a temperature that is roughly 25 degrees Celsius below the melting temperature of the protein as shown in thermal shift (Thermofluor) experiments.

Use the Hampton Research Proti-Ace kit, then run a gel to see which proteases work well. If a protease gives good results, then run a screen with protein treated with this protease. Limited proteolysis can work better than in situ because with in situ you may have too many crystallization experiments. Others felt that this is not a problem, and that a combinatorial experiment can work well. Mass spec is another way to find which proteases work well.

Janet Newman has recommended the Granada Crystallization Box with counter-diffusion for this situation. A slight excess of RNA is recommended. RNases can be purified out of the protein sample, but the robots and plates must be kept clean. It's important to map the RNA against the protein to make sure that you use the right length of RNA

It was pointed out that there are three commonly used reducing agents in the context of crystallization: Beta-mercaptoethanol (2-mercaptoethanol) is the weakest, DTT (dithiothreitol) is the next strongest and TCEP hydrochloride is the strongest of all. Which one to use will depend on what pH you are working at (DTT is not so stable at pH 8) and what buffer is present (TCEP is not particularly stable in phosphate buffers, especially at neutral pH. Therefore, if TCEP is to be used in PBS buffers, prepare the working solution immediately before use). Read chapter 14 of the 2nd edition of book Protein Crystallization edited by Terese Bergfors, International University Line, 2009. This chapter contains a discussion of which reducing agent to use and when it is contra-indicated. Remember as well that if you have a 20,000 Da protein with 6 free cysteines, then the protein concentration is 1 mM and the free cysteines require at least 6 mM of antioxidant.

Buy screens in tubes, then transfer the solutions to deep-well blocks and freeze. Others said that freezing could cause precipitation. Replace when used up, roughly once every 6 weeks.

Use the oligomer, especially if it is the physiologically-relevant form. Use both, to double your chances. The monomer always has the option to form higher oligomeric forms under the crowded conditions in crystal packing

Yes definitely.

Yes, it is worth trying. A nice paper on this method: C. Abergel, Acta Cryst. (2004). D60, 1413-1416 Spectacular improvement of X-ray diffraction through fast desiccation of protein crystals.

Yes, it can work, " incredibly well". The best available reference for the method is a paper by Carol Lusty, J. Appl. Cryst. (1999). 32, 106-112 A gentle vapor-diffusion technique for cross-linking of protein crystals for cryocrystallography.

It was pointed out that this method is (like so many others) very protein crystal- dependent. Some protein crystals will effectively cross-link in 5 to 10 minutes, others require overnight. The methodology has been discussed in talks, but other than the paper by Lusty, no other manuscript was known to exist.

Treating the crystals for 30 minutes at 4 degrees Celsius was suggested by one researcher. Another person suggested adding 5 microliters of a 25% solution of glutaraldehyde to a Micro-Bridge in a 24 well plate. Let the crystal cross- link through the vapor phase. Periodically remove crystals to test diffraction and see if there is any evidence that the crystal is cross-linked by physical manipulation.

The bar graph shows the distribution of crystallization hits obtained at the Hauptman- Woodward Medical Research Institute for 167 NESG and SGPP targets. Each target was set up with 1536 different crystallization cocktails and produced 1 or more hits during microbatch-under-oil crystallization screening.

Loop access seems to be the big issue facing people when trying to mount directly from a plate. Some researchers suggested using an acupuncture needle to move the crystal to one side of the drop so that it would be easier to loop.

One group did a study of plates to see which ones were the best for retrieving crystals. The MRC plate, and the Intelli-Plate were both deemed best.

Increase the ratio of ligand to protein by 3 to 5x. Adding ligand to the cryo-preservative can help. It is difficult when you have a low binging affinity; however success was reported even with micromolar inhibitors.

In cases where the sample is crystallized from conditions such as 3.5M sodium chloride, you can sometimes transfer the crystals to a solution of PEG added to the original cocktail to improve soaking.

Soak for longer periods of time. In some cases no evidence of the ligand was observed in the structure after a 24 hour soak, but after a one week soak, the ligand was clearly observed in the electron density.

Try to transfer the crystals to a solution with a different pH before adding the ligand.

Try to cross-link the crystals before soaking the ligand.

Solubility of the ligands can be a problem. DMSO can be used to increase the solubility of the ligand, and works as a cryoprotectant at concentrations of 20-30%.

Part of the success is due to over sampling. A lot of people use PEG 3350 and so a lot of crystallization conditions come from it.

PEG breaks down into aldehydes and peroxides. These compounds change the solution pH.

PEG is used for many things other than crystallization.

FLUKA now assays PEG that is purchased by Hampton Research for crystallization reagents. A study was done at Hampton Research to look at the stability of PEG solutions. Hundreds of bottles of PEG solutions were prepared and separately bottled for 2 years. Among other measurements pH, and conductivity were recorded. A fresh bottle was sacrificed for each measurement. Conductivity increases as the PEG breaks down. PEG solutions will degrade over time. Exposure to oxygen and light increases the rate of degradation.

PEG 3350 is an ingredient in food products and medications. Hence, it is regulated by the FDA. This means that it is well-characterized and does not have batch to batch variation. PEG 3350 is also the 'most stable' PEG.

To prepare PEG and keep it stable for a longer period of time, you need to get oxygen out of the bottle, keep it in the dark, and keep it cold (4 degrees Celsius or freezer storage is recommended after preparing the solution).

Regardless of how the solution is stored, it will break down over time. Solid PEG will break down over time as well.

If you can obtain crystals only from an aged cocktail, dope the drop with the aged cocktail (cut the protein solution with the aged cocktail) and use fresh and easier to obtain cocktail as the reservoir solution. You can get a lot of crystallization experiments from a few milliliters of vintage PEG containing cocktail solution if you only use it in the experiment drop.

A very large and vocal majority cried out, "no"!

Attendees felt that dialysis is a good approach, but requires too much sample, and is just too difficult to set up using traditional methods. People would use the method if there was a better way to set up the experiments.

A metal frame that holds the membrane flat and eases set up was described by one researcher. This frame eliminates the bubbles that can form during microdialysis experiments and block the liquid diffusion across the membrane.

Another person described a method using Eppendorf microcentrifuge tubes to set up dialysis experiments using 10microL of sample.

DLS is widely used, however it seems to be the general opinion that it works best in conjunction with other methods. If you use: 1) SDS-PAGE, 2) size-exclusion chromatography, and 3) dynamic light scattering to check your sample's purity you are much better off. Any one of the methods can mislead you, but all three provide solid evidence that your sample is likely to crystallize. 70% of samples will crystallize if all three methods produce "good" data on a sample.

Many of the attendees routinely use mass spec (MALDI-TOF) to identify post- translational modifications. The size range of suitable samples was reported to be 2,000 to 90,000 Daltons. You can readily see contaminants. It was reported to be used with transmembrane proteins by one attendee. It only requires 15 minutes to run the experiment and "get the answer".

Other approaches included SDS-PAGE and size-exclusion chromatography, and dynamic light scattering.

There seems to be a growing trend away from very small drop volumes. The general feeling was that too small a drop would decrease the likelihood of nucleation. MCSG now uses 300 nanoliter or protein plus 300 nanoliter of cocktail solution for their screening drops. Approximately one half of the crystals for MCSG come right from those screening drops.

A range of drop volumes are used, from 100 nanoliter plus 100 nanoliter up to 1 microliter. Many people seemed to agree that 600 or 800 nanoliter drops would lead to the formation of more crystal hits than smaller nanoliter drops.

Adding an extra volume of protein solution to the drop was deemed an effective strategy by one of the attendees. They felt that there was a significant amount of protein lost on the surface of the small volume drops.

Using a deep UV light source and a couple of filters you can usually distinguish protein from salt crystals. It doesn't work very well when a protein precipitate covers a salt crystal.

Newport makes a 260-290 band pass filter. The camera is from PLS Design and is the most expensive component of the system. The cost of an appropriate UV light source is approximately 9000 Euro.

A case was discussed where the protein sample is formulated with 500mM ammonium sulfate which was not considered ideal for crystallization trials. Options were presented for reformulation including, neutralized organic acids, TACSIMATE, these may be more amenable to cryopreservation of the crystals.

Phosphate buffer has gone out of fashion as a protein buffer because of the false leads (salt crystals) that can form during crystallization screening trials in the presence of calcium and magnesium. It was pointed out that it is a great buffer and works well for some proteins.

Also, the following paper by Jancarik et al was discussed as offering options for formulating the protein for crystallization trials. Acta Cryst. (2004). D60, 1670-1673 [ doi:10.1107/S0907444904010972 ] Optimum solubility (OS) screening: an efficient method to optimize buffer conditions for homogeneity and crystallization of proteins J. Jancarik, R. Pufan, C. Hong, S.-H. Kim and R. Kim.

Another discussion took place that mentioned the stargazer (real-time PCR) approach to look at chemicals that stabilize proteins for crystallization trials.

PNAS October 24, 2006 vol. 103 no. 43 15835-15840 Chemical screening methods to identify ligands that promote protein stability, protein crystallization, and structure determination Masoud Vedadi, Frank H. Niesen, Abdellah Allali-Hassani, Oleg Y. Fedorov, Patrick J. Finerty, Jr., Gregory A. Wasney, Ron Yeung, Cheryl Arrowsmith, Linda J. Ball, Helena Berglund, Raymond Hui, Brian D. Marsden, Par Nordlund, Michael Sundstrom, Johan Weigelt, and Aled M. Edwards. 

It was reported that for PSI2 they tried the surface entropy reduction method on about 20 proteins, of those there was only 1 successful application of the method.

The online server that identifies which residues to modify for surface entropy reduction is considered to be fast and effective (http://nihserver.mbi.ucla.edu/SER/). It was reported that a modeler required an entire year of effort to come to the same conclusions about which residues to modify, verifying the effectiveness and time-savings that this software offers.

Controlled proteolysis works as another approach to improving crystallizability of proteins. Test 8 to 10 proteolytic enzymes and run an SDS-PAGE gel to check the sample purity post-treatment. Concentrations used range from 0.01 to 0.001% of the protein concentration. pH is important to control the enzymes functions. In some cases, it may be worth trying to add a proteolytic enzyme directly to the crystallization drop.

If you look under a microscope at about 40x magnification the suspension of crushed crystals will not have any discernable single crystals. There should be an opalescence, or sheen to the mixture with particles that are too small to clearly distinguish.

Very small crystals are difficult to crush. Larger crystals are easier to crush. Allan D'Arcy uses the reservoir solution to suspend the seeds. He recommends first using micro tools to crush the crystals, followed by the seed bead. If the crystals are cross- linked and gummy when you try to crush them with a needle or probe, they will not produce a good seed stock. They should crumble like parmesan cheese when pressure is applied with the probe.

Fairly concentrated seed stock suspensions should be used when trying to screen for initial crystallization conditions. Dilute suspensions are used when micro-seeding to try to produce a few larger crystals.

Should you crush crystals grown from different cocktails together or separately?

If the cocktails are chemically close, you can pool the crystals and crush them together, otherwise, crush them separately.

Can you seed using the horse hair crushed using liquid nitrogen using the robotic delivery system?

No, it isn't very effective. The crushed hair seed stock settles out from solution rather quickly.

The Nanodrop ND-1000 system (http://www.nanodrop.com/nd-1000-overview.html) was used by about one-third of the audience. According to attendees you can place about 2 microliters of your sample on the detector, lower an arm to read the concentration of the sample (from fairly dilute to about 80mg/mL) all in a few seconds. You can retrieve some of the solution after the read. There is a single and 8 probe version of the device available.

Bradford, Biorad colorimetric assays were the other popular choice.

It was noted that centrifuge-based ultrafiltration cells can form a concentrated layer of glycerol on the surface of the ultrafiltration membrane.

People are using the Centricon, Centriprep, Minicon and Zeta-spin centrifuge-based ultrafiltration cells. It was reported that people using these systems would often stop the centrifuge every five minutes to mix the solution and then begin the centrifugation process again to eliminate the formation of a gradient.

A number of researchers brought up the use of an ultrafiltration cell that is stirred, using pressurized nitrogen. An example of this is the Amicon. The stirred cell eliminates the concentrated layer of protein, glycerol, etc... that often forms in a centrifuge-based ultrafiltration cell. The smallest volume of these cells currently available is 3 mL and using this you can concentrate a solution down to ~ 100microL volume.

How many people use ammonium sulfate to concentrate their samples? Only one or two people raised their hands. It was pointed out that Allan D'Arcy does this quick process and that it worked well for a number of samples. How many people place a dialysis tube filled with dilute protein solution over a bed of high molecular weight PEG to concentrate their samples?

Only a few people raised their hands.

(Howard Einspahr) It’s a wonderful idea but a nightmare to implement. Information in the literature is not well described. There are reported crystallization conditions with a missing key ingredient. The BMCD is not longer current.

(Onkar Singh) With the variation in protein batches, different conditions and ways to measure protein concentration being used in all the different laboratories minimal information may be adequate.

(Martin Caffrey) Journals have page limits for articles, limiting the amount of detail you can use to describe your crystallization experiments.

(Howard Einspahr) Not any more with Acta cryst F.

(Shane Atwell) We would need a database with clean interfaces. It would have to be made available to the public. Someone would have to decide what data should be entered.

(Patrick Shaw Stewart) There isn’t much progress. The suites sold by companies are not modular. You buy it and have to use the entire suite, not just the sections you may like.

(Bob Cudney) There will be a newsgroup/bulletin board on the new Hampton web site that will provide a forum for communication amongst crystallizers.

(Bob Cudney) The original Jancarik and Kim Crystal Screen was the result of a survey of the literature for successful crystallization conditions.

The low ionic strength screens use lower buffer concentrations to keep the ionic strength down.

You need a high enough buffer concentration in your crystallization screens to affect the pH of the final experiment drop. If the protein solution has an initial buffer concentration of 25-50mM you need to have a higher buffer concentration to effectively shift the pH of the crystallization experiment.

(Onkar Singh) Yes, vary the concentration of the buffer.

The majority of people polled work in the 5 to 10 mg/ml range of protein concentration.

(Bob Cudney) A continually updated survey of the literature shows an average protein concentration of 14.5 mg/ml.

(Patrick Shaw Stewart) This is going to be technique dependent. Batch experiments will require a higher protein concentration than vapor diffusion experiments. Looking at 800 experiments in the literature 4 were set up at < 2mg/ml and 4 were set up at >300mg/ml. The majority are set up in between that wide range of concentrations.
Add a larger volume of protein than crystallization cocktail to your experiment drop to increase the amount of protein in the experiment.

(Shane Atwell) Commercial screens seem to be targeted to a protein concentration of 10 mg/ml. Wizard screens seem to work best at a slightly higher protein concentration than the other screens.
The Hampton pre-crystallization test (PCT) was reported to work well.

(Shane Atwell) Surface mutations are okay, but avoid active site mutations.

(Allan D’Arcy) Be sure to check the biological activity of the mutants.

(Allan D’Arcy) Use superglue.

(Shane Atwell) The pins oxidize when using superglue to seal them. Use the thicker glue-like superglue.

(Howard Einspahr) Old school use sealing wax, it won’t oxidize the pins and repels liquid.
Mounted CryoLoops produced by Hampton Research are now sealed.

(Shirley Roberts) How does the wax behave in liquid nitrogen? No one knew the answer to this question.

What has been people’s experience with the litholoops?Mixed results.

(Onkar Singh) No luck with them.

(Shane Atwell) The litholoops tended to pull more liquid away from the sample than traditional loops.

Both.

The general consensus seemed to be that you were better off not mixing the drops. The higher concentrations of protein and crystallization cocktail at the boundary between the two unmixed drops would set up a highly supersaturated boundary zone. This high level of supersaturation is more likely to initiate nucleation of the crystals.

(Allan D’Arcy) Round wells provide better optics.

(Shirley Roberts) Image processing algorithms do not work in her group with round wells.

Flat bottom plates do not help localize the drop.

The sample may be contaminated with something that you can’t resolve by the methods of analysis that you are using.

(Allan D’Arcy) Roughly 30% of proteins are soluble to a high concentration but do not produce crystals.

(Zygmunt Derewenda) Disorder. Mutation does nothing to improve crystallization. Using NMR analysis the structure was shown to be disordered. Look from a number of angles including secondary structure, tertiary structure structure prediction.

(Patrick Shaw Stewart) Weitzman Institute web service program to predict crystallization of a sample. Prilusky J., Felder C.E., Zeev-Ben-Mordehai T., Rydberg E., Man O., Beckmann J.S., Silman I. and Sussman J.L. (2005) FoldIndex© predicts whether a given protein is intrinsically disordered. Bioinformatics (in press).
http://bioportal.weizmann.ac.il/fldbin/findex
See also http://www.disprot.org/predictors.php

Joel Sussman emphasized that the choice of predictor may depend on the question that you are asking. The Wiesmann interface is the best for identifying unfolded proteins and large unfolded regions. Some of the other predictors are better at identifying smaller floppy regions.

(Aengus MacSweeney) Heard of two people who used it and had good results.

(Howard Einspahr) Ligands can help. Purification protocols sometimes carry along with the protein natural ligands that can help the protein fold correctly.

(Aengus MacSweeney) If you have no ligand try 50mM substrate, anything can help.

(Shane Atwell) Ligands are as important as the protein. If the concentration of the ligand is high enough, anything from a related family will bind and may help to stabilize the protein.

(Howard Einspahr) Less than 100 nanoliters is small. Going from 20 nanoliter drops containing crystals to larger drops is difficult. The literature contains all we need to bridge that gap. We should run experiments to figure this out.

(Shirley Roberts) Our group likes to use smaller drops. 100 – 150 nL drops do produce more salt crystals than larger drops. It can be difficult to retrieve crystals with so little mother liquor.

(Heather Ringrose) Use vapor diffusion vary the size of the protein portion of the experiment drop 200nL protein + 100nL precipitating agent.

(Beat Blattmann) Make it part of your routine with a robot to check the delivery on a daily basis. Make sure the drops are delivered to the correct location and are of the correct volume.

(Howard Einspahr) Are robotic experiments more reproducible than those set up by Humans?

(Shane Atwell) Side by side robot/technician experiments did not show a significant difference in reproducibility between humans and robots. Smaller drop volumes seem to be better. Smaller drops have a higher pressure than larger drops. The parameters of the drop change with drop volume.

Janet Newman) If the robot is not suitable for the process you end up with an expensive dust collector.

You need to make regular use of robotics. It’s not just the initial cost of the robot(s) but also the cost of technician time (for use, repair, and periodic maintenance), extended warrantees and disposables.

Nanodrops achieved with robotics were considered a breakthrough for one group. It lowered the cost per experiment and the sample requirements.

Robots can reduce the amount of sample required for screening.

Some industry participants report they are backing off on their use of robots and primarily use them for screening, doing optimization manually.

If you have a choice of two proteins, one with a His-Tag and one without a His-Tag look at the level of expression before making a decision. If expression is the same for both then use the one without the Tag. It is best to cleave the Tag before concentrating the sample. An example presented described a protein that would not refold with the Tag but would refold without it.

(Janet Newman) We should try to develop a standard, constant method to report crystallization information. Publish it in a standard way using the same method developed by the PDB for structures. Inconsistencies in the literature include the reporting of ammonium sulfate concentrations as both percent of saturation and as molarity. Molarity was agreed to be the correct way to report this. Other issues discussed included the reporting of PEG 3350 as PEG 4000 in the literature.

(Shane Atwell) The initial hits are never reported in the literature. You only see the final ‘optimized’ crystallization conditions that produced the crystal(s) used in the diffraction experiment. (Martin Caffrey) Data mining needs all of the conditions, those that fail as well as those that succeed in producing crystals.

(Howard Einspahr) Commercial crystallization screens are standardized, they are fixed in time. If you report conditions that gave you hits in these screens, then by default you also are reporting those that did not produce hits. Acta F will publish this in the supplementary information.

(Bob Cudney) Some vendors sell PEG in large batches using paper bags as a container. The PEG is not as stable when it is stored in paper bags. PEG 3350 is pharmaceutical grade, monodisperse, and has batch to batch consistency. PEG 4000 will have batch to batch variation.

(Allan D’Arcy) The Index screen was developed by Allan D’Arcy and Bob Cudney. Each of them surveyed the literature and came up with 48 solutions that they separately decided would be the best for an initial screen of crystallization conditions. Allan D’Arcy’s 48 were called ‘Hammer’. Bob Cudney’s 48 were called USA3. The genomics centers did not screen a number of different crystallization variables and may have been settled in a local minima for conditions.

(Patrick Shaw Stewart) Acta cryst has a macro that is available for formatting papers. Would it be possible for Acta F to do something similar?

(Howard Einspahr) Yes, great idea. A database like the PDB is only a receiver of information. It can only say, “Stay out”. There is no enforcement power. The PDB could not exclude deposition of coordinates until journals began to say, “No”. This was an expression of the will of the community. We as researchers have the power to help the databases get the data. “Make your voices heard”!

A poll of the meeting participants showed that ~ 1/3 routinely screen temperature as a crystallization variable.

(Neil Grodsky) Use DLS to determine the effect of temperature on macromolecular samples before beginning crystallization trials. Place your sample in the DLS and make readings at 4, 13 and 21oC. Look at the results of these measurements to determine the ‘best’ temperature to use for crystallization trials. You should screen at least two temperatures. Based on experience in the laboratory he feels that for ~50% of the samples they work with temperature is an important variable.

(Bob Cudney) Routinely use microbatch under oil for screening experiments. They are well suited for screening temperature without the condensation issues normally associated with using vapor equilibration techniques. Experiments are set up at room temperature and scored after one day, the samples are moved to an incubator set at a temperature of 16 oC stored for one day and scored. The process is continued at temperatures of 4 and 30 oC.

(Alexander McPherson) Temperature is very important when working with detergents. Protein solubility is not nearly as affected by temperature as detergent solubility. If your protein is hydrophobic and tends to aggregate there is good reason to try to work with it at 37 oC.

(Martin Caffrey) We have designed lipids that will form a lipidic cubic phase in temperatures ranging from 4 to 60 oC. The primary lipid, mono olein forms a solid at 17 oC and so cannot be used for crystallization trials at 4 oC.

(Aengus MacSweeney) You can get condensation on the oil that could harm your crystallization experiments when changing temperatures.

(Bob Cudney) This condensation is climate dependent. It doesn’t occur in southern California.

(Lesley Haire) Temperature can be used to separate nucleation from growth phases. Set up the experiments at 4 oC where they are more highly supersaturated and nucleate (note this is case specific) and then transfer the experiments to 18 oC for the growth phase of the crystallization trials to prevent over-nucleation. Use the effect of temperature on protein solubility to control the level of supersaturation in your experiments.

(Allan D’Arcy) Be sure to test crystals for X-ray diffraction at room temperature. Loop mount your crystal and push it into an X-ray capillary sealing the ends of the capillary with clay (plasticine). If the crystal diffracts then, take the same crystal and use it in a cryogenic diffraction experiment. Make sure the crystal diffracts X-rays at room temperature before spending time trying to optimize cryo conditions. Someone brought up the fact that it is not uncommon to see significant differences in the quality of X-ray diffraction from ‘identical’ crystals taken from the same experiment plate or even the same drop.

(Howard Einspahr) Flash freezing is a big insult to the crystals. There is an increase in mosaic spread after flash freezing. Robert Thorne has shown that even when annealing works to lower mosaic spread it does not occur throughout the crystal. There are still regions of high mosaicity within the crystal. Freezing hurts the crystal mosaicity but pays off in the time you have to collect data before radiation damage becomes a problem. Elspeth Garman’s tip ; “Be the crystal”

Joe Luft reported his studies had shown there was an optimal speed of growth for a crystal but it varied from protein to protein.

Generally, no one reported having seen a correlation between speed of growth and the diffraction quality of the crystal.

Margaret O'Gara, Pfizer reported having successfully used silica hydrogels for crystal growth. Most people had been unsuccessful and didn't like using it.

Examples were reported where reaction of protein solutions with heavy metals (Mercury compounds in particular) and then analysis with mass spectrometry had revealed binding of mercury to the protein. This was then confirmed when the protein was crystallised and the structure solved.

Yes and no!

Detergents may help. Additives may help.

A change of buffer can often make the protein "happier".

A high speed spin (100,000 x g in an airfuge) can remove aggregates and leave a monodisperse supernatant.

One suggestion was to run the screen and look for positive leads and then add those components back into the protein and then rerun the light scattering experiment. No examples could be given of this having worked!

The preferred buffer for the protein was generally agreed to be one containing low salt, low buffer concentration and not phosphate. Glycerol is OK at low concentrations.

Observing the drops at higher magnification than normal (100 to 1,000x) can sometimes be useful in deciding when a precipitate is amorphous or has some structure.

One comment was made that sometimes more buffer can be added directly to the drop and if its crystalline material then it would probably dissolve. If it is precipitate it will most likely not dissolve.

Also, adding a chromophore to the protein prior to crystallisation may help to make the distinction.

It was felt that other disciplines now appreciated the need for a molecular biology effort to work closely with protein crystallisation.

Industrial groups felt there was a definite need to have a molecular biologist within their structural group and the reason where this was not the case was mainly due to organisational problems.

A show of hands among the audience revealed 20 (out of 100) people were working in groups which had a dedicated molecular biology effort.

Hampton Research have microbridges made from polypropylene which can be used for sitting drop experiments and give a nice drop even when they contain detergent.

Several people apparently have unwanted cross-linking occurring in their crystals.

Anil Mistry reported this happening in on of his proteins and he believed it to be a problem associated with the cysteine residues in the protein. One solution would be to keep the crystals in a solution containing 200mM DTT which was kept fresh.

A recommendation was made for TCEP (sold by Pierce) as a reducing agent. It doesnt polymerise like DTT does and can be used successfully at 1/10th the concentration of DTT.

Treat the protein with a moderate concentration (30 mM) of DTT for a certain period of time (30 minutes) before setting the crystallization experiment.

People would like a standard set of proteases to be readily available.

Immobilised proteases (beads or plates) would be preferred.

People suggested there be readily available standard proteins for crystallization.

Some people were in favour of a fairly limited number of screens before reassessing the protein they were using.

Others reported using screens customised for an individual protein.

Screen for a while then use molecular biology to change the protein solubility and repeat screens.

A comment was made that statistically, a total of 400 solutions was a reasonable number.

No one liked to throw away plates and people held onto them as long as possible.

One comment was "give it back".

Other suggestions were:

Place at 4°C , add 15-20% glycerol, alter the buffer conditions. It was felt that if the cloudiness was reversible then this may well be a good sign.

One suggestion was to put the protein on a cover slip over a well solution of acetic acid or ammonium hydroxide and see if the cloudiness disappears due to pH change.

It was noted that several of the posters showed the application of Microsoft Access software to the handling of data generated from crystallisation trials.

The key problem here was generally felt to be getting people to input the data to keep the information up to date. One factor here would be to have the PC right next to the microscope. Ease of use would be the determining factor for a successful program. Everyone would like to have such a program and the consensus was that a figure of $1000 would be a reasonable sum to expect to pay for such a program. Again Margaret OGara pointed out that Leica sell a digital camera which comes with useful software which is aimed at general biology use but could be adapted for crystallisation applications.

There was a high degree of interest in which digital cameras were suitable for collecting images of the set-ups. One recommended was the Pixera system sold by Leica. (Details from Margaret O'Gara at Pfizer) See web site for additional information: http://www.pixera.com/

Most people try to crystallise their protein with and without the tag.

A minority were quite happy to pursue crystallisation without looking at removing the tag and about 1/3 would remove the tag before attempting crystallization.

It was reported that the presence of the his tag can lead to protein insolubility at a pH of around 6.57.0.

The comment was made that with modern chromatographic techniques it should not be necessary to introduce tags purely as a purification tool.

Thrombin was the most popular cleavage site enzyme.

One comment advised using short strands of DNA in complex with the protein and ensuring that the DNA was not too long to "overhang" the folded protein.

Also the annealing processes mentioned in the poster and talk by Clare Williams may be of benefit.

No correlation was noted between birefringence and diffraction.

There was an approx. 50:50 split between those who go for soaking and those who prefer co-crystallisation. The consensus seemed to be that given the choice then co-crystals were preferred. Some groups go for both apo crystals and co-crystals to maximise the chances of success.

It was pointed out that often it is feasible to use apo crystals as seeds to obtain complex crystals.

An instance was noted when soaking in the crystallisation conditions using ammonium sulfate was not possible since the sulfate ion was found to block the active site of the protein. So, soaking experiments required the crystal to be transferred to PEG conditions prior to inhibitor soaking experiments.

The issue of conformational chance and crystal contacts may lead to problems during soaking. Tightly bound inhibitors can crack crystals during soaking in contacts are restricted. Consider the size of the inhibitor when soaking (i.e. might be better to co-crystallize than soak with a large inhibitor).

Dissolve the highest concentration of inhibitor into 100% DMSO and mix with protein. If anything precipitates out, spin and set supernatant in hopes that some bound inhibitor stayed in the supernatant.

Mix dilute protein with dilute inhibitor then concentrate the mixture. Or mix dilute protein with dilute inhibitor and run over a size exclusion column, then concentrate.

Soak crystal in solution with a solid inhibitor. Try this with heavy atoms too.

Most people would use a robot for initial screening trials.

A smaller number would want a robot for optimisation studies.

Most people said no, they would not be interested in a robot capable of screening in a 96-well format mainly due to the problems that would raise in results analysis. Those that would use it would only do so for screening.

A useful innovation would be if automation was capable of "screening" the trials and dismissing large numbers of useless drops and perhaps picking out "interesting" ones.

One problem with automated result analysis arises from the need to scan up and down through a drop to get the full "result" from that experiment.

One possible innovation would involve a joystick controlled microscope stage and a microscope connected to a PC-linked camera.

When doing set-ups manually it is often possible to make changes to the design of the experiment "on-the-fly"; this is not be possible with automated screening robotics.

One person suggested using color schemes to score crystals.

One person reported using a mini-incubator with the capacity to hold 6-12 plates. This will allow a wider variety of temperatures to be screened on a smaller scale. This person reported obtaining an increased number of initial screen hits by cycling the same screen across several temperatures using standard proteins. Also it was his experience with Ribonuclease A that cycling through a temperature gradient lead to a larger crystal. For example use temperature T1 to nucleate, increase temperature to T2 to etch crystal back, reduce temperature to T1 and continue to grow the crystal and repeat until the desired results are achieved.

About 2/3 of the audience routinely screen at two temperatures. About 2/3 screen at 40°C and most of the audience use 20°C. A few people screen at intermediate temperatures and some screen at temperatures greater than 20°C.

Subtle differences in temperature such as the difference between 4 and 8 degrees Celsius can be significant.

Alex McPherson does his initial screens at 20°C then moves the plates around after about 10 days to 37°C, 4°C etc. He also pointed out that he finds hydrophobic proteins (or proteins which oil-out) work better at 37°C and therefore temperatures above room temperature should not be ignored.

It was pointed out that when using an incubator to control temperature the vibration/mechanical movement of the motor within the incubator adds another factor to the crystallisation. One person reported removing the compressor unit from the incubator to remove the vibration source.

Try different host cells in a bauculovirus system to see different levels of glycosylation. Or use inhibitors to prevent glycosylation. Or remove the glycosylation for enzymes.

Note that 50% v/v glycerol is required as a cryoprotectant in 0.2 M magnesium formate.

Try beta-mercaptoethanol as a crystallization additive.

What is the best temperature for sample storage? Try an ammonium sulfate precipitation and store at 4°C. When ready to set up sample, spin and resuspend in the crystallization buffer. Others suggest flash freezing in liquid nitrogen and then freezing the sample at -80°C.

Experiencing and increase in mosaic spread? Try optimizing the cryoprotectant conditions, try optimizing the crystallization reagent, consider the speed of transfer, the speed of freezing, and the size of the crystal as variables.

Block the cryostream for a few seconds or even minutes to remove ice when you have poor cryo conditions.

Ice on crystal in cryostream? Poor liquid nitrogen onto crystal in cryostream.

Try a variety of buffers effective at the same pH when optimizing crystallization conditions since sometimes the buffer will bind the protein or the buffer may prevent a desired ligand from binding.

His tags were discussed again. It appears there is no consensus on whether to leave the tail on or chop it off. Everyone seemed to agree to leave the His tag one, try a screen and if no crystals are obtained, chop off the His tag and repeat the screen.

When working with a hydrophobic protein add the detergent during the purification procedure. Also try adding sodium chloride along with the detergent during purification.

Try using agarase to dissolve agarose gels when growing crystals in agarose gels.

When using polyamines for RNA and DNA crystallization keep in mind that polyamines effect both sample solubility and crystal quality.

Looking for new additives to try? Iodoacetic acid, iodoacetamide, and trifluoroethanol.

People who prefer to filter the protein sample prior to crystallization experiments suggest trying one of the following: 300 kD, 0.02 micron, 0.1 micron, or 0.2 micron filters.

During the RAMC 97 Meeting there was discussion about the production of glycoproteins with predefined glycosylation. Annie Hassell provided the following reference after the meeting: Davis, SJ, Puklavec, MJ, Ashford, DA, Harlos, K, Jones, EY, Stuart, DI, & Williams AF. Expression of soluble recombinant glycoproteins with predefined glycosylation: application to the crystallization of the T-cell glycoprotein CD-2. Protein Engineering 6: 229-232 (1993).

Poisoning to abolish excess nucleation. Problem: You have lots of showers of crystals in lots of drops no matter what you do. Solution: Do an additive screen on one or more of your hits. Out of 96 additives, chances are good that at least one will abolish all nucleation, i.e. the drops stay clear. This is now your poison additive. Step 2: Seed into the poisoned drop. Now it is you that controls the amount of nucleation added. Step 3: Vary the seed concentration to titrate exactly the amount of nuclei needed in the poisoned drop. Note: There is a published example of this idea in Acta D where they seeded a clear drop, hence my idea for a poison additive screen (I have to look up the reference on my computer).

Incubate your crystallization experiments at a temperature at least 25 degrees Celsius below the Tm of your protein.

Use protein seeds from an unrelated protein to seed a new protein that has not crystallized through standard screening and no crystals of the protein of interest are available.

Use grease on the end of a tool to pop bubbles in a drop.

If a room temperature soaking system results in loss of diffraction, adapt to 4 degrees Celsius to decrease the diffusion rate.

To enhance reproducibility for the scale-up alays use the same polymer type of your crystallization plates.

Boil for two minutes your low soluble protein binders (small molecules) before complexing with your proteins.

If you seal wells with grease and coverslips don't make a complete circle of grease as when you place the slide on top of the grease the pressure in the well - make a small gap with a pipette tip so that when you press the coverslip down it seals quickly and firmly - the grease is squishes in to fill the gap.

For protein complex : Pt substrate for screen I got one hit after six days (small crystal). I used the same condition as screening but instead of set up I incubated Pt substrate at 4 degrees Celsius for 6 days and set up, I got crystal the next day, twenty times bigger.

Try different ways of mixing the crystallization drop when setting up. For example, protein plus precipitant or vice versa, mixing, etc. Might yield different amount or size of crystals.

I noticed the evaporation drying rate of certain additives in the Hampton screen and also detergent screen. I started pre-plating with different volumes into regularly used plates them putting them into the freezer to preserve the deepwell blocks and future plates from evaporation.

If you want to crystallize your protein don't use precipitants. Use crystallants. And please use the term crystallant or crystallizing agent in your publications and lectures.

Don't rely exclusively on automated screening and don't limit yourself to commercial crystallization kits. Try to mix yourself protein solution with a small set of crystallant solutions and see what happens under binoculars. You will learn a lot about your protein's behavior. First clue: All drops remain clean, protein concentration probably too low. Precipitate everywhere, protein concentration probably too high. Mixed results, go ahead with screening.

Use acupuncture needles as a micro tool to manipulate crystals in a drop or to remove skin.

Use acupuncture needles as a micro tool to manipulate crystals in a drop or to remove skin.

First read McPherson's book or any other reasonable source of knowledge before you start your crystallization experiments.

Check the homogeneityof your protein sample with as many methods as available.

If you use PEGs for crystallization experiments be aware they easily oxidize and can change behavior (in case repeated trials do not work anymore).

BME (2-mercaptoethanol) does it again. I will routinely add 1 microliter of 14 M BME (2-mercaptoethanol is typically made available as a 14.3 M solution) into the reservoir after drop set up. It worked in multiple cases to get biggers better diffracting crystals or even to get crystals of proteins that never crystallized before. If not hits appear in initial screening after months, seed the trays with seed stock, precpiptate, spherulites or even phase separation containing drops. Seed, seed and seed again. Works often and well. For controlled seeding (when seed stock concentration matters) add seed stock in the reservoir before mixing it with the protein in a tray.

There is a parameter which is rarely identified crystallization screening (vapor diffusion) it is the crystallization plate. Despite testing different crystallization kits, different technique, temperatures, protein concentration, ratios, etc., try at least different plates. With standard proteins in different plates we observed different results with one plate compared to another. In the lab we found one protein which gave crystals in one plate but not in others.

Too many small crystals or needles? If you seed with another form of crystals you may have at the end drops with nice big crystals from two forms.

Fill in the reservoir not with the condition of crystallization but with a high concentration of Sodium chloride, Ammonium sulfate (1M to 3M), PEG 3350, PEG 6000 (20 to 60%) (cf Newman Acta Cryst (2005) D61, 490-493). To optimize a hit, ask every people in the lab to make few drops. Just give them the materials (solution of crystallization, protein, coverslip, etc.) and no more information (not the size of the drop, not the order of adding the components, etc.) Let's see what happens!

No hits from a screen, what next? You can always set up another screen! Or, you can make a list of all drops that have precipitate and use the precipitate as a potential seed stock. Streak seed from the precipitate into new drops that have been set up at 50% and 75% of the original screening solution. There may be nucleation sites or microscopic seeds in the precipitates that may grow at lower precipitate saturation. Better still, streak seed from the precipitate to a clear zone into the same drop to recycle the drop.

For reproducible micro-seeding by hand use a cryoloop to fish out your seed from the seed stock and transfer them to the drop. Use a 0.3-0.4 mm cryoloop.

For bigger crystals try to add 0.5-1 microliter of 14 M beta-mercaptoethanol to the reservoir after the protein drop was set up.

Problem: Can grow crystal but no protein-ligand crystals. Tip: Take the conditions from the apo crystals and develop a focused optimization screen (24 well maximum). Screen complexes using cross seeding and the focused screen and three drop ratios (1:1, 2:1 and 2:3).

Problem: Poor crystal quality. Tip: Change tray type or crystallization method. For example, initial screens done in sitting drop tray and crystal quality improved in 96 well hanging drop tray.

Every project, every protein, every construct is unique. Be careful of knowing too much. Just because things did or did not work in the past does not mean things will work that way for the next project.

Problem: Poor diffraction. Tip: Heavy atom soaks to stabilize floppy regions of the protein.

Don't focus all of your optimization efforts on a single crystallization condition. If you have several different crystallization conditions identified for a sample go after them. Crystals of the same protein produced from different chemical conditions and/or temperatures will have unique physical properties. These properties will determine how easy the crystal can be looped (physical stability), cryoprotected and ultimately how well the crystal diffracts X-rays. Avoid single points of failure, go after several hits.

If ligands can't be soaked into crystal or co-crystallization is ligand specific try seeding into drops that contain ligand of interest. When soaking crystals with insoluble ligands try adding the cryoprotectant to the soaking solution. This can help solubilize the ligand and also cryoprotect (so less handling). Doesn't work so well when salt is the cryoprotectant, but may well when it's glycerol, DMSO or ethylene glycol.

Problem: Drops are all/mostly clear. Tip: Remove stabilizing agents (salt, glycerol, etc) from the protein buffer. Then do crystallization screens.

At suboptimal protein concentrations the interface between protein solution and crystallization screen solution may exhibit excessive precipitation. To avoid this, before adding screen solution add 1 microliter of water. Downside of this is equilibration will take slightly longer. Upside of this is decreased osmotic shock for protein and less precipitation.

Put one conditions of 40% TCA into your standard screen. This should precipitate out all of your protein, so that you have an idea of what heavy precipitate should look like.

Don't consider a crystallization result in isolation. Look for neighbors in chemical space and use those results to provide chemical directions for optimization. If a cryocooled crystal does not diffract well. You cannot tell if it is the crystal, cryoprotectant or cooling that is causing the problem. Look at room temperature data before moving on.

Consider the case of poor nucleation and seeding did not work. Tip: Mix protein and well solution then use pipette tip to cross the drop into branched shape. Crystal may grow in the branches of the drop.

To increase you choices of producing more optimal crystal condition or conditions using seeds, try the following. Program a small volume liquid handler to dispense your protein and seed solution directly into 96 well commercial screens and/or an additive screen.

Recent success with loop deletions. Sequence alignments reveal either charged loops and/or loop insertions relative to homologues. I have removed 3 to 34 amino acids and retained high expression soluble protein and novel crystals.

Problem: Unstable protein. Tip: Add 1-5% low molecular weight (200 to 1,000) PEG directly to the protein and then screen.

Start with big crystals! Add at least 5% glycerol to everything.

Stabilize crystals in cryo by adding protein buffer components into the cryo. For example, most commonly I add 100 to 150 mM NaCl plus reservoir components plus cryoprotectant(s).

Play with protein-mother liquor ratio, especially with low solubility proteins. Try different concentration of protein coupled with streak seeding.

After looking at the results of initial screening or of additive screen, pick several of the best "hits" and screen in 96 well format, 6 or 8 identical drops of each favorite before scaling up. Helps to eliminate "one offs" and save time.

Apo protein is monidisperse but won't crystallize. Complex it! Complex it! Complex it!

If your crystal seeds withstand large serial dilutions, try matrix seeding via the reservoir by doing the following. 1) Create crystal seeds as described by Allan D'Arcy. 2) Dispense seed into reservoirs containing reservoir. 3) Aispirate/dispense to mix seed in reservoir. 4) Dispense mother liquor droplet containing seed onto protein drop.

If you do not see any crystal growth in several days after set-up (more than 1 to 2 weeks) and the drop are not all clear, add salt (such as ammonium sulfate) to 0.5 M to the drops. Even though ammonium sulfate salt crystals might form, you might actually get protein crystals. This worked for me recently.

Non denaturing (less than 3 M) concentrations of urea can be helpful to solubilize your protein.

Try stoichiometric levels of multivalents, cations as additives. They may be necessary for crystallization but sometimes the levels found in commercial screens are too high and toxic.

Heat treatment of protein complex to obtain diffraction quality crystals. Original complex has no initial crystal hits. Heat treat protein complex (25 to 80 degrees Celsius) for various times (5 to 30 minutes). Centrifuge to get rid of aggregated protein and screen again.

When using pre greased trays, take a toothpick and remove a bit of grease from each well. Now as you push down on your cover slip, you turn it a few degrees. This will allow the air to escape and the turn will form an airtight seal over each well.

Thermostability. Monitor the effect of additives, buffers, ligands, etc. on melt temp of your protein. We have seen in multiple cases that the most thermostable construct, buffers, additive yields the best or only crystals. How? We use Bio-Rad's iQ5 iCycler (a PCR instrument) as it has 5 sets of filters for excitational emission and hydrophobic dyes that fluoresce upon binding (protein unfolding).

Problem: Crystals adhering to plastic of sitting drop plate, and mechanical dislodging (by cryoloop, tool, etc) does not work. Solution: Stan a fine gauge syringe needle into the plastic, near but not into the crystal. This often distorts/disrupts the plastic near the crystal and breaks the seal.

Getting bigger crystal by low tech / low cost counter diffusion. If you are faced with either no nucleation or showers of crystals and the usual tricks including seeding do not work, try this: On a cover slide, set the protein and precipitant drop (example 1 microliter plus 1 microliter) separate, but very close to each other. Then, with a whisker or pipette tip streak through the drops to form a connecting bridge between the protein and precipitant solution. Invert cover slide and place over well. Crystals will form along the gradient and "self screen" for best conditions.

When working with compounds for co-crystallization, if the compounds are highly insoluble in protein buffer (50 micromolar or less) we often employ low concentration complexing. We dilute the protein and then add in diluted compound, so tha the compound is added close, or at least closer to a concentration where it is soluble and the content of DMSO in the protein sample remains less than 2%. The protein-compound complex is then concentrated for crystallization trials. This has helped us with several projects with highly insoluble compounds.

In experiments using model proteins we found that Ionic Liquids (IL's) specifically 1-Butyl-2-methyl imidazolium chloride gave increased numbers of crystallization outcomes compared to the IL controls. A large number of the crystals obtained had precipitated outcomes in the IL controls. In many other cases the IL and crystal had an improved morphology (needles to plates, plates to 3D crystals) over the IL controls. Tip: Using an IL such as 1-Butyl-2-methyl imidazolium chloride as an additive to improve chances of getting a crystal from conditions which otherwise would give precipitate. Marc Pusey, MI Research, Inc. Our favorite cryoprotectant. 1x UCP (Ultimate Cryo Protectant) 8% Glycerol, 8% Ethylene glycol, 9% Sucrose, 2% Glucose. We make a 2x solution. Generally add this 1:1 with reservoir. The ratio can be modified, for example 1.2 microliter 2x UCP : 0.8 microliter reservoir, or 1.4 microliter 2x UCP : 0.6 microliter reservoir, or 0.6 microliter 2x UCP : 1.4 microliter reservoir, etc. I believe this has been successful in cryoprotecting some 70% of all of our systems, resulting in more than 50 solutions of these targets regardless of previous cryogenic treatments. Author in unknown to me, but the credit is published in a singled Hencrickson paper. I was tipped off 6 years ago.

If you can't get your Se-met protein to crystallize try leaving ou the DTT/TCEP during purification and crystallization.Sometimes there are disulfide bridges near crystallization contacts that need preserving for crystallization to take place. Collected the Se edge is still possible.

Try 20 to 30% DMSO as cryo. This has worked well in a number of cases for me and I've added this to a very short list of cryos that I personally use. Note: If your mother solution contains a high salt concentration the DMSO will cause it to precipitate out of solution. So beware!

4 degrees Celsius crystallization plates prepared at room temperature always have a condensation problem on the plate seal. To avoid condensation cover the finished plates with two lids and plate it in the cold room for 20 minutes or on top of a cold metal block, The will reduce the temperature of the reservoir solution while the 2 lids delay the temperature change from the top long enough to avoid condensation.

Ever been manually sealing a plate only have it slip out from under you? The result is usually death for hanging drop plates, and with sitting drop plates your best bet is hoping the drop is not splashed onto the seal above. To ensure the plate stays put when applying pressure, we use a "grip pad" in our lab. Simply place the grip pad onto the bench, set plate on top and seal as usual. The grip pad prevents the plate from slipping out from under the compression tool, usually a brayer, used for ensuring the please seal is applied correctly. The grip pad can be cut from the material commercially available for lining tool shop drawers. In addition, we created a fixed plate holder that encloses the entire plate to guarantee the brayer does not slip off the plate when sealing, a common occurrence when manually sealing many plates. Our plate holder is custom cut from a hard rubber to fit both thee 24 well and 96 well plates. The plates sit slightly above the platform to ensure both ends are sealed and for easy removal. The sides of the platform are rounded to ensure the brayer has a smooth path of travel.

If your protein is not very concentrated, set up microbatch screening experiments with 0.1 microliter screening solution plus 0.5 microlier protein. Use Al’s oil to allow concentration in the drop. That way you are less likely to get salt crystals before you get protein crystals.

Patrick Shaw Stewart, Douglas Instruments

When optimizing leads obtained from the Fluidigm chips in a vapor diffusion format, screen different ratios of protein to well volumes in drops.

Neil Grodsky, Pfizer Global R&D

Wrap some lead tape (available from golf shops) around the lid of a cryo cap then the cap sits the appropriate way in a dewar of liquid nitrogen for easy storage of your frozen crystal.

Janet Newman, In Stilla Consulting

When manipulating crystals from their growing drop, and soaking them, try to imagine that you are the crystals, and how you would feel being poked with a loop or a needle, or what it would be like to have the pH temperature, osmotic pressure around you suddenly change without warning. It might help your procedures!

Elspeth Garman, Oxford University

If cryo-cooling is not giving satisfactory results, check how you are making up the cryoprotectant solution. The water in the mother liquor should be replaced by the cryoprotectant agent, rather than diluting the mother liquor. This factor is the single most common factor causing trouble which is easily rectified.

Elspeth Garman, Oxford University

Seeding with pieces of glass, broken cover slides. The protein crystals sometimes grow along the end of the glass.

Gaby, Novartis

If you are having problems repeating or transferring crystallizations from 96 well to 24 well plates, for optimization stick with the 96 well plates. The single well low profile Greiner plate case easily be set up by hand or robot. There is plenty of room for larger drops and crystals are accessible for fishing. There is even a ledge for a small drop of cryo.

Shirley Roberts, University of York

Use seeding for screening when you have only precipitate in your first screens.

Jens-Christian Poulsen, University of Copenhagen

Seeding drops with silicon carbide. Silicon carbide whiskers used to deliver DNA into cell nucleus. Microscopic in size but contain suitable surface for protein molecules to aggregate and nucleate for crystal growth. Uniform size of whiskers will allow reproducible crystallization set ups. Thereby could be used in initial screening trials to induce nuclei for crystals to grow.

Christopher Browning, IGBMC

What about the production of protein crystals in transgenic yeast containing the genes of Bacillus thuringiensis. Bacteria too small. Yeaste cells of reasonable size. Bacillus thuringiensis produce protein crystals in vivo, thus utilize this technique to produce crystals in yeast with overexpression of target protein. Or oocytes.

Christopher Browning, IGBMC

Use luminal in protein crystal identification. Luminol is used in forensics to stain blood. Uses UV to light up blood stains. Add luminal to drops with crystals. Luminol binds to protein crystals. Addition of UV light and protein crystal will fluoresce. Salt crystals should not fluoresce.

Christopher Browning, IGBMC

Smooth soaking in agarose gel. Grow your crystals under the usual conditions with 0.1 to 0.3% agarose gel and let the soaking solution (ligands, heavy atoms) diffuse smoothly through the gel.

Claude Sauter, IBMC-CNRS-Strasbourg

In setting up nanoliter screens use 2:1 protein:reagent. Higher hit rate. Three examples so far.

Heather Ringrose, Pfizer Global Research & Development

Original suggestion from Steve Swerdon, NIMR. For crosslinking crystals, put glutaraldehyde into the moatt of a Douglas Instruments VaporBatch plate to crosslink crystals in droplets dispensed under Al's oil.

Lesley Haire, NIMR

When doing microbatch screening, briefly centrifuge the plate at low speed. That not only ensures protein and precipitants mix and form a single drop, but also separates particulate matter. Any immediate precipitate is pelleted softly allowing the majority of the drop to be easier to visualize under the microscope. This might also generate two concentration phases ofprotein. Crystals may form from the precipitate or the clear phase.

Dean Devonshire, Medivir

Give yourself plenty of time to retrieve crystals from a hanging or sitting drop experiment. Add a drop of paraffin oil over the experiment drop. You will have plenty of time to mount crystals, test them with dye (remove a crystal to another drop of oil to dye so other crystals in the drop can still be used), crushed, etc. This works especially well when you find crystals that may be protein in an old dried out plate that cannot tolerate much additional evaporation.

Joe Luft, Hampton Woodward Institute

Have you experienced grease blow out with your robot? Large globs of grease on your tray? Try this: Make a disc about 3/8" wide, the same diameter of the grease cartridge. In the middle 1/8", make a slot that will hold a 0-ring in place. Drill a small threaded hole in one end ~1/8" deep. This allows you to retrieve the disc when the cartridge is empty.

Bryan Prince
Pharmacia & Upjohn Company: d.bryan.prince@am.pnu.com

Greasing Your Own Plates, the Easy Way

Make a 5 ml grease cartridge with a P200 tip. The size of the cartridge reduces hand strain & fatigue.

Bryan Prince
Pharmacia & Upjohn Company: d.bryan.prince@am.pnu.com

After you have crystals open the coverslip, remove the mother liquid in the droplet, dissolve the crystal with 3 ul of H₂0 or buffer and add 3 ul of well solution. Close the coverslip. The crystal will appear again. My crystals diffract ~ 4 A, sometimes I get twins that diffract ~2.5 A.

Nham Nguyen
University of British Columbia: nham@laue.biochem.ubc.ca

Pickle juice was added as an additive to a mutant form of a crystallized native protein. The mutant could not be crystallized in near similar conditions of the native. By chance, the components of pickle juice were read and found to contain compounds used in crystallization (i.e. Glycerol, PEG 400, citric acid, acetic acid, alum and a few vitamins). This juice (Sweet & Snappy Vlassic brand) was filtered (0.45 µl and pH¹ed to neutrality. It was then added to various PEG¹s (that crystallized the native form) and set up with the mutant. Crystals formed after 1 week! Trying to "add back" single components of the pickle juice to determine which component was responsible gave no crystals, the pickle juice (~1%) was necessary. Hint/ Tip: Commercially available food/ detergent solutions ought not to not be discounted as additives for crystallization!

Michael Hickey
Agouron Pharmaceuticals: micheal.hickey@agouron.com

Generally, people concentrate a protein to 10 mg/ml or higher, then dilute 2-fold when setting up their hanging drops by doing a 1:1 minx. What if in concentrating to 10 mg/ml aggregation has been induced which is irreversible, such when pipetting your 1:1 mix hanging drop, it already contains aggregates‹a bad starting point.

(For monodisperse protein samples)
Using DLS test the limit of concentrated protein, i.e., the maximum concentration that can be achieved before a polydisperse signal is obtained. Then test samples, up to this limit, by concentrating up to this limit and test for irreversible aggregation by diluting a concentrated sample to a number of levels and test for monodispersity. With the sensitivity of current DLS equipment even samples at 10 mg/ml should be measurable. In this way and within the limits of your DLS machine it should be possible to find out whether you will have aggregates when you dilute your concentrated protein 2-fold when setting up a hanging drop.

Anil Mistry
Parke-Davis Pharmaceuticals: anil.mistry@wl.com

When setting up drops with a protein which has low solubility a low starting concentration has to be used. Setup larger drops (5-10 ul) and leave them to stand "dry" for 3-5 minutes at room temperature /4°C prior to inverting over a well of a Linbro plate, this should allow some pre-concentration.

Anil Mistry
Parke-Davis Pharmaceuticals: anil.mistry@wl.com

To grow crystals at different temperatures around room temperature search the lab for spots that are consistently at higher or lower temperatures. A difference of several degrees can be found. Temperature shifts can be easily made by moving crystals to a different place. (Check office shelves too!) [Discovered by Charles Reed in my lab]

Irene Weber
Thomas Jefferson University: weber@asterix.jci.tju.edu

Use TCEP instead of DTT as a reducing agent in your protein solution. It isn¹t oxidized as quickly as DTT. Be sure to watch the pH of your solutions because it is very acidic.

Barbara Brandhuber
Array BioPharma: bbrandhuber@arraybiopharma.com

If higher DMSO does not damage your protein, try higher concentrations of DMSO (10-20%) for crystallization. It can help when dealing with insoluble compounds and is an excellent cryo-protectant. Freeze directly from drop!

Melissa Harris
Pharmacia & Upjohn: melissa.s.harris@am.pnu.com

Ching Yuan
Baylor College of Medicine: ching@emily.bioch.bcm.tmc.edu

It has been long accepted that crystallizers with an abundance of facial hair have highly successful careers (Leeds University‹personal observation). This was thought to stem from the fact that matter found its way from the hair into the trials and acted as a nucleation centre-BUT SERIOUSLY; Addition of small grains of sand to a crystallization drop that is "close to producing" crystals can aid in nucleation and crystal growth.

Jonathon Hadden
University of Leeds: bmbjmh@leeds.ac.uk

If your protein or protein complex fails to crystallize try seeding from crystals of the same protein (if it's a protein you want to crystallize) or a related protein for apo protein crystals. Serial micro seeding works best, make sure you look at the drops after seeding to identify any visible crystal seeds in there (i.e. not new crystals)

Margaret O'Gara
Pfizer Central Research: margaret_ogara@pfizer.sandwich.com

Single dehydrant/reservoir Hanging Drop/ Sitting Drop vapor diffusion using microbatch plate in suitable container.

Application for Hampton Screening:
- Low Ionic Strength Screen
- All Additive Screens
- Nucleic Acid Mini Screen
- All Detergent Screens

Betty Yu
NCI-FCRDC/SAIC: yub@mail.ncifcrf.gov

Prepare trays for crystallisation, leave at 4°C, and fill polystyrene box or flat container with ice. Imbed a metal plate in the ice and set out cover slips. When you¹re ready to set out crystallisations place the trays in the bed of the ice and prepare drops, when finished transfer to 4°C. Simple but it works.

Marie Anderson
ASTRA: marieanderson@hassle.se.astra.com

As I said in my talk, give it a go. You might be surprised!!

Clare Stevenson
Nitrogen Fixation Laboratory: clare.williams@bbsrc.ac.uk

If you have crystals that are very sensitive to being touched (they break) or stick to the glass or dish, use a pointed strip of parafilm to move them. Otherwise, grow the crystals on parafilm and punch "wells" around the crystals to move it.

Allan D'Arcy
Hoffmann-La Roche: allan.darcy@roche.com

To find the best buffer system which will keep your protein happy, stable, and soluble for concentration prior to crystallisation, setup your protein (at 1-2 mg/ml) in hanging drops over 1 ml well solutions containing a series of buffers at various pH¹s, with various additives/stabilizers, etc. (but no precipitant!). Checking for drops which are clear will give an idea of which solutions keep the protein happy. In addition, as the system attains equilibrium some in situ concentration of protein will be induced due to the bulk difference between the drop and well volumes, hence an idea of how the protein behaves upon concentration in this solution will also be observed. Modifying a clear or slightly clear drop by adding a higher concentration of buffer to the well may even produce crystals. However, the main piece of information this method can produce is an idea of which buffer system and additives to put your protein into prior to concentration a crystallisation screening.

Anil Mistry
Parke-Davis Pharmaceuticals: anil.mistry@wl.com

As a graduate student, I spent countless hours and quantities of protein trying to get crystals of a single construct. We never got crystals. Another group got the structure of a homologous protein that was auto-digestive. Has we stopped after 400 trials and altered our construct, perhaps we would have faired betterŠ we couldn¹t have faired any worse. There is a reasonable statistical argument to demonstrate that 400 trials are a good limit.

Brent Segelke
Lawrence Livermore Laboratories: segelke1@llnl.gov

When crystals are fragile or you want to transfer them to a different mother liquor e.g. for heavy atom soaking, why not try crosslinking your crystals with 0.1% glutaraldehyde in your mother liquor. This can be done by adding the glutaraldehyde directly to the drop or placing it next to the drop and allowing for vapour diffusion. This crosslinking enabled us to solve Mod A at 1.2 A resolution.

Clare Stevenson
Nitrogen Fixation Laboratory: clare.williams@bbsrc.ac.uk

In a sitting drop well, cryoprotectant (20% Glycerol in crystallization buffer) should be dribbled down the side of the well in the following manner: See drawing.

Bryan Prince & Melissa Harris
Pharmacia & Upjohn Company: d.bryan.prince@am.pnu.com

I tried adding 1-10 mg/ml BSA in the cryoprotectant when soaking crystals that would crack. The one time it was used, the crystal did not crack and froze nicely. I don¹t know if the BSA was the reason for successful freezing. I was wondering if anyone else has tried this.

Laura Pelletier
Agouron Pharmaceuticals: pell@agouron.com

Try (set up) drop no. 6 first (of Screen I). It should produce light precipitation‹if ppt appears too heavy, reduce the protein concentration by 1/2 and try again. If there is no precipitation in drop 6 try drop 4. If there is no precipitation in either of the drops concentrate the protein 2 fold and try again.

Jaru Jancarik
UC Berkeley/Chemistry: j_jancarik@lbl.gov

Make a thin needle out of a glass capillary. Fish out crystal needles from drop with the capillary by holding the capillary parallel to the crystal. This will pick up less mother liquor (reduce background) and put less stress on the crystal.

Frederick de Mare
Pharmacia & Upjohn: fredrick.demare@eu.pnu.com

If your crystal does not freeze well in the cold stream try liquid nitrogen or vice versa.

Hans Parge
Agouron Pharmaceuticals: hans.parge@agouron.com

When optimizing or making solutions for a random scan, omit the buffer (@ final O.1 M) so that your stock is 90%. Prior to putting your stock in the well, add 100 uL of 1M buffer of your choice. Next, add 900 uL of 90% stock and mix. This reduces the number of tubes for crystallization (precipitant) stocks and allows flexibility in buffer identity and pH range. Be sure to make plenty of (~20mL) of precipitant, you¹ll need 900 uL per buffer.

Anna Stevens
Monsanto Company: anna.m.stevens@monsanto.com

If your protein refolding reaction has a low yield and produces lots of precipitate try collecting the precipitate, resolubilize in GuHCl, and refold again. This material is sometimes more pure than the washed inclusion bodies.

Neali Armstrong
Columbia University: naa15@columbia.edu

To preserve Hampton solution when you or a co-worker gets a "hit" and suspect the Hampton solution may be "magic" or you cannot reproduce crystals with lab-made solution‹make the reservoir solution from lab ingredients and use your homemade solution for reservoirs. Use the Hampton "magic" solution only for the drops thus using a few ul per experiment rather than O.5 ml. Saves having a 48 well screening solution set with one tube empty and the rest at 8 ml and still allows the superstitions or in Alex's case, the contaminated solution to reproduce crystals.

Cheryl Janson
SmithKline Beecham: cheryl_a_janson@sbphrd.com

Use centipreps (millipore) for concentrating protein >O.5 ml. Protein concentrating away from membrane. (No micro high concentration, ppt on membrane), works very nicely for a number of proteins.

Paul Reichert
Schering-Plough Research Institute: paul.reichert@spcorp.com

Look closely at your old test tubes when cleaning the place. Proteins do crystallize on the walls of the tube when stored in a cold room. I picked up my tubes from the wastebasket and an X-ray was made from an old supersaturated protein tube.

Kalevi Visuri
Macrocrystal Oy: management@macrocrystal.com

Keep an HPLC (Reverse phase) profile of your protein before crystallisation and after crystal formation. It can be used as a quality control and tells you if any modifications have occurred.

Glenn Dale
Hoffmann-La Roche: glenn.dale@roche.com

When only small amounts of protein are available, it is not feasible to screen many compounds which promote monodispersim using a dynamic light scattering machine. To detect aggregation, we use a pseudo-native gel approach: 1l of protein is mixed with 1 l additive from Hampton Additive/ or Detergent Screens. These samples are then incubated at room temperature for 20-30 minutes and then placed in 2x sample buffer containing no DTT, no SDS and are not boiled. These samples are run on a standard SDS-PAGE gel. We have screened many additives using this approach and it has given us leads for subsequent optimization of protein buffers.

Tom Zarembinski
Argonne National Laboratory: tomz@igg.anl.gov

Try soaking poorly diffracting crystals in higher concentration of precipitate‹ammonium sulfate or PEG. It may take several weeks so test after 1 or 2 months.

Irene Weber
Thomas Jefferson University: weber@asterix.jci.tju.edu

Add a small amount of (~ 1%) iodoacetic acid to buffer solutions. This helps prevent aggregation by carboxymethylation of cys. Also, iodoacetic acid seems to help form salt bridges and aid in crystallization.

Bernie Santarsiero
NIFG: bds@adrenaline.berkeley.edu

Glycerol has many benefits but also some drawbacks. We found it to be beneficial with one protein we were working with; this protein is a transpeptidase called Mur A. The protein is quite soluble and could concentrate to 20 mg/ml but it lost activity over time when stored at ­80°C. We therefore dialysed it into 50% v/v Glycerol to see if the activity could be retained for longer, this would allow us to make a large batch rather than regular smaller batches, for crystallisation. On dialysis we found a significant reduction in the volume of protein, so much so that the protein had concentrated from 5-10 mg/ml to almost 50 mg/ml. Activity was also found to be retained with no significant loss after 6 months at ­80°C. This gave us a method for storing large batches of Mur A at ­80°C, without losing activity and also resulted in a sample pre-concentrated for crystallisation and containing a cryo-protectant. Other sugars gave similar effects; sucrose, sorbitol, etc., but none were as effective as glycerol in achieving the 50 mg/ml final concentration.

Anil Mistry
Parke-Davis Pharmaceuticals: anil.mistry@wl.com

For soaking crystals with compounds with limited solubility I have tried two "extreme ways" (although not much- and more experiments should be tried):

- Leave some solid compound in the soaking solution.
- Dissolve some compound in n-octanol and layer the octanol solution on top of the soaking experiment.

These two provide (hopefully) more or less constant concentration of the compound in the soaking solution. The octanol layer may help reduce air oxidation by preventing direct contact of soaking solution to air.

I found that 200 µl of soaking solution in a well of the 24 well Linbro plate is a good volume to work with:
--Not too little that possibly causes concentration change due to evaporation (Sealed well) and not too much that the crystals get lost in the solution.

Jirundon Yuvaniyama
Mahidol University: scjyv@mahidol.ac.th

When removing crystals from a hanging drop, I sometimes find that the biggest crystals fall against the coverslip and are impossible to resuspend without damage. I had our glass shop make a stand to transfer the coverslip to that enabled me to manipulate the crystals more easily.

Dennise Dombroski
University of British Columbia: dombroski@byron.biochem.ubc.ca

One of my proteins produced only zillions of tiny useless crystals. When I mixed the drops the conventional way-mixing well, overlaying, etc. the protein with precipitant solution. Large, gorgeous crystals were produced when I crossed the drop, creating a gradient within the drop. This worked best, setting up sitting drops with vapor diffusion.  Ursula Kamlott Hoffmann-La Roche: ursula.kammlott@roche.com

We found mass spectrometry like ESMS and MALDI highly efficient in determining impurities and/or microheterogeneities in our protein sample/batch. In most cases it is a simple, straight forward method which requires a minimum amount of sample. In some cases it has shown to detect impurities/microheterogeneities when other techniques did not.

When making up cryoprotectant solutions containing glycerol, put a test tube of glycerol in a beaker of warm water. The viscosity falls and it is easier to pipette accurately.

A protein which was purified and showed some faint bands of contaminants on a native gel was crystallized and solved successfully. The same protein, purified by HPLC and resulting in a single band native gel did not crystallize.

When screening with low molecular weight PEGs try microbatch. Crystals appear rapidly with PEG 400-2000. To convert to vapor diffusion use 0.2 M buffer in the well and a 1:1 drop ratio. Try using a positive displacement pipette such as the Anachem Microman 1 - 10 microliter. These are much more accurate.

When working with crystals that grow fairly rapidly (one day) try the following. Pipette multiple protein drops (2 to 4 works best) onto the cover slide. Using a single pipette tip, get the reservoir solution to mix with the first drop. Now, go back to into the reservoir with the same tip to get the reservoir solution for the second drop. Continue for the remaining drops with the same tip. In certain cases, seeding starts very quickly, so by using the same tip one can introduce minute seeds to successive drops. Use the same cover slide with multiple drops to minimize evaporation.

To reduce the number of crystals and increase their size, try filtering the protein solution prior to setting up the experiment. Try the following filter sizes: 0.22, 0.1 micron and 300 kD molecular weight cut-off. Try the Millipore centrifugal filters.