A fully automated crystallisation pipeline to achive high-throughput with short crystallisation plate processing times, high reproducibility, and increased efficiency of the screening process.

Jean Cavarelli Jean Cavarelli IGBMC-CERBM email hidden
Orly Dym Orly Dym Weizmann Institute of Science email hidden
Karl Harlos Karl Harlos email hidden
Juergen Koepke Juergen Koepke Max Planck Society email hidden

Instruct has 8 centres offering Crystallisation across Europe. Navigate the map and click on the pins to discover centres near you.

View Centre
Instruct Centre - France 1
View Centre
ISPC (WIS) Israel
View Centre
EMBL Grenoble
View Centre
View Centre
View Centre
Protein facility - NKI
View Centre
View Centre

Crystallisation Details

Diffraction of single biological molecules is extremely weak* so amplification is required if X-ray diffraction from conventional sources is used. This amplification can be achieved when three dimensional arrays of periodically repeated molecules, commonly called crystals, irradiated instead. The total scattering then becomes proportionally stronger with the increasing number of repeating unit cells.

Crystallising the protein molecule is thus a crucial step during the course of X-ray structure determination. Many attempts to ease the process using theoretical approaches have not been very helpful in practice; protein crystallisation is still a matter of trial and error and some practitioners consider protein crystallisation more of an art than a science. To overcome the bottleneck of growing well diffracting protein crystals consistently for structure determination, structural biologists have, over the last decade, sought to put the crystallisation process onto a more systematic and rational base, by automating crystallisation and optimisation of screening conditions.

In the past crystallisation screening tended to consume large amounts of sample and time. However, this task has been notably advanced in the last years by the introduction of automated crystallisation systems. These systems can produce large numbers of experiments in parallel, but more importantly, they contribute to increase the efficiency of the screening process by reducing significantly the size of each individual experiment. It is possible to carry out very extensive crystallisation screening experiments even when samples are available in very limited amounts. Projects that were traditionally considered to be beyond the reach of the crystallography technique, like those involving protein production in mammalian cell expression systems or the study of large multi-protein complexes, are accessible today.

Crystals grow from supersaturated solutions. In macromolecular crystallography supersaturation is normally achieved by adding a precipitant to a concentrated solution of purified protein or macromolecule. A precipitant can be defined as any type of substance that would decrease the solubility of the sample without inducing denaturation. Neutral polymers, like different types of polyethylene glycol, salts or small organic acids are amongst the most common types of precipitants. However crystal growth is a rare phenomenon and is very sensitive to the chemical and physical environment. Since the conditions that would lead to the formation of crystals canĀ“t be predicted, it is necessary to perform extensive screening experiments testing a series of different precipitants at different concentrations, at different pH and in the presence and absence of salts and other type of additives. Given the large number of parameters to be explored the capacity to perform large screening experiments is a critical factor for success in crystallisation. However other parameters external to the crystallisation experiment itself, like construct definition, sample purity and homogeneity are equally important.

User Guide

Crystallisation is at least a four step process which starts with selection of the adequate screening conditions for the respective protein, followed by preparing a crystallisation plate according to this screen, incubation of the plate at a suitable temperature, and ending up with the scheduled imaging of the plate. All these steps have been automated, the first two by use of pipetting robots, the third by a bar-code controlled storage system, and the last by a computer controlled imaging system.

To achieve a high-throughput crystallisation system the whole pipeline has to be tied together by another robot which transfers screens and crystallisation plates to the different stations. In addition automation for sealing deep well blocks and crystallisation plates, mixing the screening solutions, and punching the deep well block seals are required. Only by this high degree of automation can a 96 well plate be processed in a few minutes to guarantee the highest possible reproducibility.

Crystal Screen Pipetting

Finding the optimal crystallisation conditions can be treated as a sampling problem in a multi dimensional parameter space. The distribution of successful conditions is thereby unknown, it can even be zero for a chosen combination of reagents. Since the number of trials is finite one has to find a method to distribute the initial conditions as effectively as possible in this parameter space. Different mathematical strategies with sound names have been proposed like: grid screen, sparse matrix sampling, random sampling, full factorial, incomplete factorial and factorial design. All of them vary the combination of reagents from the basic groups of precipitant, buffer, additives, and eventually detergent in the case of membrane proteins.

In principle the experimentor has to choose methods and reagents which fit his respective protein best, but in practice the number of free parameters is reduced even more to the choice of one of the commercially available screen kits. In most cases, use of a commercial screen is probably the most promising strategy to start with. Only in cases where all available presampled screens fail or when hits of initial screens have to be optimized does the preparation of customized screens become necessary.

Robots to set-up screen solutions in deep well blocks for later use in crystallisation screening are available as stand alone machines or integrated into fully automatic crystallisation pipelines. They can be used to copy any commercial screen or to set-up customized screens with any type of gradient. The generated well conditions can be stored in a laboratory information management system (LIMS) for later correlation with conditions which led to growth of crystals. Growth conditions of one or more crystallisation hits can be used to generate fine screens.

Crystallisation Plate Pipetting

Once a screen has been chosen the next step is to decide which crystallisation technique to use. Available are vapor-diffusion techniques like hanging- or sitting-drop, batch methods, dialysis, and free-interface diffusion. The outcome of the first methods overlap relatively strongly, while free-interface diffusion or batch methods may have success at conditions where the others fail. For initial crystallisation trials the crystallisation technique is therefore a key part of this decision.

For technical reasons most crystallisation robots work with sitting-drop vapor-diffusion, since this method is the easiest to automate. Today there are commercial solutions available where the robots for the different steps can be independently automated, needing manual intervention or fully automatic pipelines which also automate the feeding, emptying and release of the pipetted and sealed plates. The preparation of the plate itself is only a transfer of screen conditions into a reservoir and well of the crystallisation plate and the addition of protein to the drop.

Fully automated crystallisation plate setup has the advantage of improving reproducibility and reduction of the time between setting the individual drop and sealing of the plate. Thus evaporation can be minimised and human intervention, which might cause errors, is eliminated.

Crystallisation Plate Incubation

In small molecule crystallisation the temperature is used instead of the concentration of the precipitating agent to gently move the crystallisation conditions from seed to growth conditions. In protein crystallography this parameter is not used in the same manner. After the crystallisation plate is sealed it is placed in an incubator with preselected temperature. In difficult cases it may be neccessary for soluble proteins to deposit crystallisation plates with exactly the same conditions at different temperatures to find the optimum for this additional parameter.

However, especially for crystallisation of membrane proteins, it is very important to select the incubation temperature carefully. There exist cases where the correct temperature was essential if the membrane protein was to crystallise at all. In this special case it can be very helpful for quick success to initially test already all different temperatures in parallel. Automation eases setup of plates with identical conditions and to distribute them to a broad number of different temperatures.

Incubators for crystallisation plate storage are available with integrated imagers and temperature control. Storage of the crystallisation plates can be automated and bar codes controlled. In that case plate loading is only possible via a computer controlled entrance. Temperature stability is usually within a degree or better. It is important to protect the stored crystallisation plates from vibrations generated either inside or outside of the incubator.

Crystallisation Plate Imaging

During the incubation time the droplets of the crystallisation plates have to be inspected on a regular basis. This can be more frequent in the beginning and in larger intervals towards the end. The imager feeds the images into a lab information management system (LIMS) which stores the conditions for each well of the crystallisation plate. Since the automatic detection of crystals is still not secure enough, manual inspection cannot be avoided. The scientist must mark the hits on each plate by inspecting plates drop by drop. When hits are marked it is easy to identify conditions which were successful and one is able to reproduce such conditions. In addition automatic optimisation screens set up around one or more hits is available.

There are different attempts underway, e.g. with edge detection algorithms, to automate the indentification of the crystals, but none of them is reliable enough yet to fully relinquish manual inspection. The use of UV-light will help to distinguish, on an automatic basis, drops with strong signals generated by protein crystals from droplets exposing no UV signal at all. But the experimentor must then decide if the crystals have a proper shape and are big enough to be harvested. Harvesting is another step that needs automation. With the advent of micro focus beamlines smaller crystals can be used which are more difficult to harvest. High throughput would also benefit from automatic harvesting.

Images of whole crystallisation plates or single crystallisation wells can either be inspected locally or, if a web-interface is available, via a web-browser. If connected to a laboratory information management system (LIMS), together with the crystallisation images, chemical information about the respective crystallisation conditions can be provided. Together with the setup of the crystallisation plate an inspection rhythm has to be defined. Optical images of the crystallisation wells can be combined with UV exposures and if the depth of field of the optics is not sufficient slices of the crystallisation drop at different heights can be demanded. These adjustments are usually predefined but can be adjusted if necessary in special cases.