A COMPARISON OF MICROBATCH AND VAPOR DIFFUSION FOR INITIAL SCREENING OF CRYSTALLIZATION CONDITIONS
Peter Baldock, Vaughan Mills, Patrick Shaw Stewart
Douglas Instruments Ltd., Douglas House, East Garston,
Hungerford, Berks RG17 7HD, UK
Six commercially available proteins were screened using the "sparse matrix" solutions of Jancarik and Kim (with modifications by Hampton Research Inc.). The screens were performed using the most common vapor diffusion method and three variants of the microbatch crystallization method, including a novel evaporation technique.
Out of 58 crystallization conditions identified, 43 (74%) were identified by
microbatch, while 41 (71%) were identified by vapor diffusion. 26 conditions were found by
both methods, and 17 (29%) would have been missed if microbatch had not been used at all.
The evaporation technique provided the best microbatch method finding a total of 34
The microbatch method  has become well established for the crystallization of biological macromolecules. Small droplets (around 2µl) of protein and precipitant are dispensed under oil, using a fine multichannel dispensing tip. The method is used for screening  where a large experimental space is searched for crystals, and for optimization  where large, well formed crystals are produced for x-ray analysis.
Screens based on the sparse matrix approach introduced by Jancarik and Kim  are widely used by workers in the field. However, at least 73% of screening in Europe is done using the vapor diffusion method alone, although microbatch has the advantage of using less protein (0.5 to 1 µl instead of about 4µl per trial) and of being less labor intensive.
We wished to establish whether microbatch is as effective as vapor diffusion in
identifying new crystallization conditions. It is often assumed that because vapor
diffusion allows the concentration of the droplet to change slowly, a larger area of the
experimental space is searched. Therefore this study also investigated a variation of the
microbatch method introduced by D'Arcy , where crystallization conditions might be
scanned in a similar way to vapor diffusion. In this case the droplets were dispensed
under a mixture of silicone oil and paraffin liquid which allowed slow evaporation of
Materials and Methods
Six proteins readily available from Sigma Chemical were selected to be screened. 48 screening solutions were chosen from Hampton Research's Crystal Screen I. The two solutions not used were the two least successful solutions in a survey conducted by Hampton Research .
Four different methods of protein crystallization were compared:
The protein screening conditions are shown in table 1. The protein solution was first filtered with 0.45 mm Millipore filters. All four methods were performed using the same initial protein solution, and within 24 hours.
The hanging drop vapor diffusion method used siliconized cover slides and Linbro tissue culture plates, both from Hampton research. 750 µl of screening solution was added to each reservoir well. Glass microcapillary pipettes from Sigma Chemical were used to transfer 4 µl of protein and 4 µl of screening samples to the cover slides. The cover slides were sealed to the reservoir wells using silicone grease.
The three microbatch screens were dispensed into three Nunc HLA plates under a thin layer of n-decane, using the Douglas Instruments IMPAX 1-5 machine. Plates (ii) and (iv) were then topped up with 5 ml of paraffin oil, while a 50% mixture of silicone oil and paraffin oil was used for plate (iii).
The plates were stored at 4° C, and were observed every few days. The numbers and
sizes of crystals were re-corded.
Results and Discussion
A total of 133 wells produced protein crystals over a period of 10 weeks. These observations are summarized in Table 2 and are shown more completely in the Venn diagram in Figure 1. Where a condition occurs in more than one microbatch method it is counted once only.
Table 2: The number of
crystallization conditions found by the four methods
Using all three microbatch variants together proved slightly more successful than the single vapor diffusion method. This might be seen as an unfair comparison, as three microbatch screens are being compared to one vapor diffusion screen, but the total use of protein and operator time is still less to perform the three microbatch screens.
Of the individual microbatch methods, (iii) is the most successful, using low protein concentration and silicone oil/paraffin mixture to allow evaporation. However each microbatch method found between 10 and 12 conditions not found by vapor diffusion.
The total number of wells containing crystals as a function of time is shown in Figure 2. The line labeled "All VD" represents the cumulated number of crystallization conditions found in the vapor diffusion method and "All MB" represents the cumulated number found with all three microbatch methods. Again, where a condition occurs in more than one microbatch method it is counted once only.
Fig 2: Graph of number of conditions observed against time. The lowest line indicates the number of crystallization conditions for all the microbatch methods combined. Note that around 40% of microbatch conditions were not found by vapor diffusion and vice versa.
In the first 3 days microbatch and vapor diffusion produced the same number of crystallization conditions. This period gives the highest production rate in the experiment. In the period from 3 days to 4 weeks after dispensing, vapor diffusion method is most successful because the drops have quickly increased in concentration to a suitable nucleation point. The evaporation of water through the oil in the microbatch plates takes longer at 4°C, but after 10 weeks the number of crystallization conditions found by the two methods was similar. Both methods were still producing conditions at an undiminished rate at the end of the experiment.
Crystal quality was assessed by observing crystals and scoring them by appearance. No significant trend was apparent, and no method could be said to give better quality crystals.
Crystal size was found by measuring the two horizontal dimensions visible in the
microscope. In general crystals found with the same precipitant solution were larger with
vapor diffusion than with microbatch. This may be due to the larger quantity of protein in
the vapor diffusion drops, 4 µl compared to 1 µl.
The best microbatch method used low protein concentration and allowed evaporation with the silicone/paraffin oil mixture. This produced 34 crystallization conditions, which is nearly as successful as the vapor diffusion method. The three microbatch methods combined found more crystallization conditions (43 conditions) than did vapor diffusion. Each of the four methods found a number of crystallization conditions that were unique to that method, indicating that no single screening method will find all the protein's crystallization conditions.
During the first four weeks, vapor diffusion was significantly more successful than all the microbatch methods combined. This might not have been true if the experiments had been performed at room temperature, which would have given a higher evaporation rate, particularly from the trials with the mixture of paraffin and silicone oil. The high protein concentration method was the fastest microbatch method, presumably because the protein concentration was already high enough to produce crystals without further concentration. After 10 weeks the microbatch methods (especially the evaporation method) had caught up with vapor diffusion.
There are four principle factors which determine the priorities of a crystallization project: availability of protein, availability of labor, urgency (based on the time available to get first crystals for optimization), and the need for thoroughness (this increases if the project is very important or if very few crystallization leads are found). The data collected so far suggests the adoption of different screening strategies depending on these factors :
Because the most time consuming step for microbatch is setting up the IMPAX, which takes 5-10 minutes, it is especially useful when project time is limited.
If protein supply is limited then using multiple microbatch methods gives a thorough search of the experimental space with minimum protein consumption. If the protein supply is so limited that only one microbatch screen can be performed then the evaporation method is the best one to use.
If project time is limited then use vapor diffusion as it will generally reveal crystallization conditions more quickly than microbatch. However it is still a good idea to do the microbatch screens as well, in case vapor diffusion does not produce crystals. In this case start the microbatch screen using high protein concentration as soon as possible to minimize the time spent waiting for first conditions.
 N.E. Chayen, P.D. Shaw Stewart, D.M. Blow, J. Cryst. Growth 122 (1992). 176