WO2003080900A1

WO2003080900A1 - A crystallisation system and method

Info

Publication number: WO2003080900A1
Application number: PCT/GB2003/001230
Authority: WO
Inventors: Neil Isaacs; Jon Cooper; Gabriela Juarez-Martinez
Original assignee: University of Glasgow
Current assignee: University of Glasgow
Priority date: 2002-03-22
Filing date: 2003-03-24
Publication date: 2003-10-02
Anticipated expiration: 2004-09-22
Also published as: GB0206819D0; EP1490536A1; AU2003219289A1

Abstract

A system (10) for growing crystals, preferably bio macromolecular crystals (including protein crystals, and without excluding nucleic acids etc), having an array (14) of wells or channels (16) defined in a substrate, the wells or channels (16) being for receiving a crystallisation solution. Associated with the array (14) is a temperature controller (24) for creating a temperature differential across at least part thereof.

Description

A Crystallisation System and Method

The present invention relates to a system and method for optimising crystal growth and screening crystallisation conditions, the system and method being particularly useful for protein crystal growth.

Following significant industrial and academic activities in the field of genomics, the focus of attention has turned towards the study of proteins. There is particular interest in the relationship between the primary and secondary structure of a protein and its three-dimensional form and function. This area of study, known as post -genomics, includes the areas of structural determinations using nuclear magnetic resonance (NMR) and protein crystallography, as well as more generally the understanding of the function of the biomolecules either in vitro or in the cell. Recently, the scope of post- genomic studies has extended towards measuring interactions between proteins and other biological or bioactive compounds, including other proteins - as receptors, oligonucleotides and chemical compounds, such as drugs .

There are currently two major methods used to determine the three-dimensional structure of a given protein, namely NMR and X-ray diffraction. An advantage of NMR is that it can resolve the three-dimensional structure of proteins (with molecular masses of ca . 20 KDa) in solution. In contrast, X-ray diffraction methods can resolve the three dimensional structures of larger proteins or complexes. However, in either method, the generation of crystals of high structural quality is crucial in order to obtain robust, high resolution data. The crystallisation of such proteins is a multifactorial process that depends on the interplay of several independent parameters including pH, temperature, protein concentration, crystallisation agent concentration and the presence and nature of impurities or additives.

Although a number of proteins have been reported to have a temperature dependent solubility, which can influence the quantity, size and quality of the crystals, this parameter is still frequently overlooked in crystallisation studies (particularly as there is often a heat of crystallisation) . Whilst it has been reported that temperature variations are important in finding optimal crystallisation conditions, if these changes are not recorded and controlled, it is difficult to obtain reproducible results. By contrast with other crystallisation parameters, temperature is not invasive and thus can be readily and intentionally modified and controlled, once the crystallisation experiment has begun . In order to optimise crystallisation condition for a given protein, the parameter space defined by the physiochemical variables of temperature, pH, ionic strength and additional agents have to be explored. This often requires the use of large amounts of protein. Despite advances in molecular biology, many interesting proteins, e.g. the eukaryotic membrane bound proteins, are still only available in limited quantities. Therefore, there is a need to produce low volume crystallisation reactors, which use a reduced amount of material, whilst maintaining high protein concentration. A number of macroscale crystallisation reactors have been reported. In one particular example, an aluminium plate is provided, round which is circulated hot and cold water for creating a temperature gradient. Pipettes that contain protein in solution are positioned on the plate, in such a manner that the long axes of the pipettes run parallel to the temperature gradient. By positioning the pipettes in this way, crystal growth along the length of each pipette can be monitored. A disadvantage of this approach is, however, that the reproducibility of experimental conditions can be problematic. In addition, a relatively large quantity of solution has to be used to fill each pipette. Furthermore, removing crystals from the pipettes can be difficult.

Other studies have begun to investigate the role of microfabrication and micromachining in devices. In a significant collaboration between NASA and Caliper, research is aiming to investigate the control of protein crystallisation in lab-on-chip (closed) systems. However, difficulties in removing intact crystals and the non-conventional format of the assay both currently present practical hurdles.

An object of the present invention is to provide a system and method for improved screening and optimisation of protein crystallisation.

According to one aspect of the present invention, there is provided a crystallisation system comprising an array of wells or channels defined in a substrate, the wells or channels being for receiving crystallisation solution, and means for providing a temperature differential across the array. Preferably, the system is a protein crystallisation system.

By providing an array of wells or channels in a single, uniform substrate, the transfer of heat thereto can be controlled and predicted. This means that the temperature at which solutions crystallise can be determined more accurately than before, which improves screening. This improved accuracy leads to a greater than expected reproducibility of experimental results.

The substrate may be any suitable medium that can be used to define wells or channels. The substrate may comprise silicon. The substrate may comprise a thermally conducting polymer, such as that currently marketed under the trade mark "Coolpoly" .

The array may comprise a 2D grid of wells or channels. The wells or channels may have dimensions in the millimeter range, for example 3mm by 3mm.

Means may be provided for sealing the solution within the wells or channels, thereby to avoid evaporation.

The means for providing a temperature differential may be operable to cool and/or heat the array relative to ambient temperature. The means for providing a temperature differential may comprise one or more peltier elements .

A controller may be provided for controlling the temperature across the array. The controller may include temperature sensors for measuring temperature across the array. The temperature sensors may be thermistors or thermocouples . The temperature sensors may be in direct thermal contact with the substrate. The controller may be operable to maintain a pre-determined temperature change or gradient across the array. This may be done using feedback from the peltier elements and the temperature sensors using one or more temperature controllers. The temperature may change linearly as a function of position across the array. According to another aspect of the present invention, there is provided a crystallisation system comprising an array of wells or channels for receiving crystallisation solution and means for providing a temperature differential across at least part of the array, the arrangement being such that at least some of the wells or channels are aligned substantially perpendicular to a direction in which the temperature is to be changed. Preferably, the system is a protein crystallisation system.

By aligning the wells or channels substantially perpendicular to a direction in which the temperature is to be changed, this improves the accuracy with which the average temperature of each well or channel can be determined.

The array may comprise a 2D grid of wells or channels. The wells or channels may have dimensions in the millimeter range, for example 3mm by 3mm. For microchannels the length could be in cm range but the width and depth in the micrometer or mm range. The array may be defined in a substrate. The substrate may comprise silicon. The substrate may comprise a thermally conducting polymer, such as that marketed under the trade mark "Coolpoly" . According to yet another aspect of the present invention, there is provided a method for determining optimal crystal growth conditions, the method comprising introducing into an array of wells or channels defined in a substrate a crystallisation solution, and maintaining a temperature differential across the array. In this way, many samples of the solution can be tested simultaneously at different temperatures.

The method may further comprise sealing or isolating the crystallisation solution within the wells or channels, thereby to limit evaporation. Various systems and methods in which the invention is embodied will now be described by way of example only and with reference to the accompanying drawings, of which:

Figure 1 is a system for crystallising proteins; Figure 2 is a flow diagram that shows schematically a method for fabricating an array of microchambers for use in the system of Figure 1;

Figure 3 is a photograph of a crystallisation array made using the method of Figure 2; Figure 4 is a plot of temperature variation along the centre of the device of Figure 1 ;

Figure 5 is a plot of 2D temperature distribution measured across half of the device of Figure 1, in which the cross-hatched boxes indicate the relative positions of two peltier elements; Figures 6 (A) to (D) are micrographs of protein crystals obtained with the crystallisation device using a first set of crystallisation conditions; Figures 7 (A) to (D) are micrographs of protein crystals obtained with the crystallisation device using a second set of crystallisation conditions; Figures 8 (A) to (D) are micrographs of protein crystals obtained with the crystallisation device using a third set of crystallisation conditions, and Figure 9 is a diffraction pattern from a lysozyme crystal .

Figure 1 shows an arrangement 10 for crystallising proteins having a silicon wafer 12 that includes an array

14 of discrete wells 16 for receiving protein in solution. By array, it is meant any ordered arrangement of wells, and could include a single row of such wells.

In the case of Figure 1, the array of wells 14 is arranged as a 2D grid. For clarity, a 5 by 5 grid is shown, although for the crystallisation experiments discussed later a 10 by 10 grid was used. The dimensions of each chamber are 3mm wide by 3 mm long by 1.1mm deep .

The separation of adjacent chambers 16 is 2mm, so that the centre to centre spacing is 5mm. These dimensions could of course be varied as desired. Mounted on opposing sides of the wafer 12 are two peltier elements 18 for creating a temperature gradient across the array 14. The direction of heat flow caused by the peltier elements 18 across the array 14 is shown in Figure 1. From this it can be seen that the array 14 is located relative to the peltier elements 18 in such a manner that rows of the wells 16 are aligned substantially perpendicular to the direction of heat flow. The wafer 12 and the peltier elements 18 are all carried on a heat sink 20.

In order to measure the temperature gradient, two thermistors 22 are provided in thermal contact with the wafer 12, one on either side of the array 14. Connected to each of the thermistors 22 and the peltier elements 18 is a control module 24. This is operable to set a desired temperature gradient by sending appropriate control signals to the peltier elements 18. The control module 24 is also adapted to monitor the temperature on either side of the array 14 using information from the thermistors 22.

Using signals fed back from the thermistors 22, the control module 24 is operable to identify if the predetermined gradient has been achieved and whether there are any changes in temperature across the array 14 over time. In the event that the desired temperature differential has not been achieved or the temperature does drift with time, the control module 24 is operable to send appropriate control signals to the peltier elements 18 in order to modify their output. In this way, heat flow and so the temperature differential can be maintained substantially constant over time.

The array of Figure 1 can be fabricated using standard photolithography techniques, reactive ion etching (RIE) and anisotropic wet etching. Figure 2 shows the process steps for fabricating the microchamber array used in Figure 1. The wafer 12 to be processed is, for example, a 3 inch diameter <100> single crystal silicon wafer 26, being 2 mm thick with 100 nm of low pressure chemical vapour deposited (LPCVD) silicon nitride 28 on both sides. A layer of positive photoresist 30 (S1818 from Microposit) is spun onto one side of the wafer at 1000 rpm. The template for the microchamber array is defined in the photoresist 30 using UN exposure through a corresponding photomask. Typically the template defines a 10 by 10 grid of squares, each being 3mm by 3mm and being separated by 2mm. The template or pattern is subsequently transferred into the Si₃Ν₄ layer by development of the photoresist, thereby to provide a masking layer 32. The masked substrate is then etched. The etch was done for 8 minutes using a reactive ion etch (RIE) that can be done using an Oxford Plasma RIE-80 with a C₂F₆ etch a flow meter reading of 80%, corrected flow 20 seem and an RF power of 100 W. This etch removes exposed areas of silicon nitride 28 to provide dry etched silicon nitride 36, but does not penetrate the silicon layer 26.

In order to transfer the array template into the silicon layer, the wafer is wet etched. This can be done using a 40% aqueous solution (v/v) potassium hydroxide (Micro Image Technology Ltd. 215-181-3) . The final etch depth can be controlled, and in the present case was selected to be 1.1mm. Care should be taken during the process of silicon micromachining to control the temperature of the wet etch bath in order to ensure uniform and reproducible depths for each of the chambers. After the microchambers 16 are formed in the silicon, the remaining Si₃N₄, on both sides of the wafer is removed using the RIE procedure described above. The end result is a 10 by 10 grid of chambers 16, each chamber having dimensions of substantially 3mm by 3mm by 2mm and a total

sample volume of about 5μl . A photograph of an exemplary

array is shown in Figure 3.

To determine optimal temperature for the growth of protein crystals, the microchambers 16 of the array 14 of Figure 1 are each filled with protein in solution, the solution for each chamber being the same. The chambers 16 are sealed to avoid evaporation. The peltier elements 18 are then used to establish a varying temperature across the array. This is done by setting the elements 18 to different temperatures, thereby causing a net heat flow across the substrate. As will be appreciated, the magnitude of the heat flow and so the temperature gradient depends on the temperatures at which the peltier elements 18 are set. The temperature may vary linearly across the array, as shown in Figure 4.

In order to assess the temperature stability of the system, the thermistor signals were logged using a data acquisition system. Analysis of this data confirmed a

long term stability of +/- 0.1°C between 0°C and 50°C. The time to reach equilibrium was dependent upon the ambient environment, which was maintained locally at 27°C with minimal airflow. Of particular interest in studying the design of the array is the steady state thermal behaviour of the system. To a first approximation, modelling has been carried out using the one-dimensional equation of heat flow per unit area, Q, accordingly:

dT_

Q = -K (1) dx

where K is the thermal conductivity.

As both sides of the device are held at a fixed temperature, it is reasonable to assume that the heat flow across the bulk of the device is constant, yielding the simple linear solution: T = C -—x (2) K

where C is a constant of integration subject to ambient conditions. Heat losses will occur at the perimeter of the device, which in a regulated environment will add a corrective constant term to the value of Q, in this case x. As expected, at the edges of the wafer the gradient is less uniform, and in practice, measurements are best made in the centre of the array. In a working device, such deviations could be readily compensated for either in the physical design, or experimentally.

In order to test the arrangement 10 of Figure 1, various protein solutions were prepared. In these experiments, the hen egg protein lysozyme was used. This protein has well defined and reproducible behaviour and provides a crystallisation process that is sensitive to both the concentrations of crystallisation agent and temperature .

The first solution tested was hen egg white lysozyme from Fluka (Rieden-de Haen) , sodium acetate (S-9513) and sodium nitrate (22,134-1), both purchased from Sigma Chemical Co. All other reagents, unless stated were from Aldrich. The protein/agent mix should be such that supersaturation is instantaneously reached. Where appropriate, the solutions were prepared with de-ionised water from a Millipore Elix 10 system and subsequently

filtered using a 0.22μm Whatman filter.

The samples were prepared in Eppendorf tubes by adding equal volumes of a solution of lysozyme (at a final concentration of 30 mg ml^"1) and the crystallisation agent, sodium nitrate, with a range of final concentrations varying in a step-wise manner between 33mM and 800mM. All solutions were equilibrated in an acetate buffer, pH 4.5. Once the final crystallisation solutions were made, each of the microchambers of the array was filled with

5μl of the solution. This was done using a micropipette, in a temperature and humidity control environment . The total amount of protein required for the specific array

described above is typically a maximum of 250 μl . This

may, of course, vary depending on the size of the chambers and/or the protein concentrations used.

Sealing of the device was performed immediately to minimise the effect of the evaporation. This was done using the industry-standard Crystal Clear™ sealing tape from Hampton Research Co. Alternative strategies to overcome evaporation include the use of polymers, such as moulded poly (dimethylsiloxane) overlayers. These alternatives were tested, although the self-sealing tape provided the most convenient and robust method to mitigate against evaporation and to enable the recovery of the crystals at the end of the experiment. Another option is to use oil, for example paraffin oil, to prevent evaporation. In this case, the chambers are filled almost completely with paraffin oil and the crystallisation solution is injected into it.

After the microchambers were sealed, the peltier elements were set up to maintain the temperature at one end of the array at 12C and the other end at 40C. The direction of heat flow across the device is shown in Figure 1. The resultant temperature gradient was maintained constant for twelve days, so that over this period the temperature of individual chambers was held at a substantially constant value. Of course, because of the variation in temperature across the array, neighbouring chambers may be at different temperatures.

The temperature in each microchamber was determined using the output of the thermistors. Before beginning the experiment, the thermistors were calibrated against temperature. The temperature of each chamber was then estimated using the calibration graph. In practice the temperature showed a close correlation with the model proposed in Equations (1) and (2) . As shown in Figure 5 there was a linear gradient across the bulk of the microarray, although, as expected there is a reduced heat loss from those chambers situated close to the peltiers . Figure 6 shows protein crystallisation obtained using crystallisation conditions where the protein concentration was 30mg ml^"1, crystallisation agent concentration (NaN0₃) was 0.4M and the pH was 4.50. Figures 6 (A) to (D) show protein crystals that were grown at 13.6, 17.6, 23.1, 25.8°C respectively. Of course, using the device of Figure 1, these crystals were grown simultaneously.

Clear morphological differences can be observed in Figure 6 (A) to (D) as a function of the temperature gradient across the device at constant concentration of the crystallisation. Figure 6 (A) shows growth at relatively low temperatures. At this level, rod-like clusters appear. Figures 6 (B) to (D) show crystals that were grown at increasing temperatures. As can be seen from these, the rod-like clusters gradually transform to single crystals as the temperature gets higher. Hence, by running a single experiment the temperature dependence of the crystal growth can be determined, within a reasonable degree of accuracy.

In addition to monitoring temperature dependence, the device of Figure 1 was used to monitor the effects of variations in concentration. Figures 7 (A) to (D) show micrographs of protein crystals, which were grown at 14.4+ 0.8°C using different concentrations of crystallisation agent (NaN0₃) . More specifically, Figure 7 (A) shows growth at a concentration of 0.30M; Figure 7

(B) shows growth at a concentration of 0.40M, Figure 7

(C) shows growth at a concentration of 0.50M and Figure 7

(D) shows growth at a concentration of 0.60M. It should be noted that the micrograph of Figure 7 (D) was photographed at higher magnification for a clearer observation of the crystals, which were extremely small.

Figure 8 (A) to (D) show micrographs of the protein crystals grown with the same concentrations as for those of Figures 7 (A) to (D) , except in this case the crystals were grown at 18.2± 0.6°C.

At low concentrations of sodium nitrate (33 to 200mM) , there was no crystallisation. This is explained by the high solubility of the protein under these conditions. As the concentrations of crystallisation agent are increased between 300 to 600mM, crystals of decreasing size are formed, as can be seen from Figure 7 and 8.

In order to verify diffraction quality of the crystals obtained with this device of Figure 1, a crystal from the condition shown in Figure 7 was chosen for X-ray diffraction analysis. The crystals obtained were from the monoclinic system, space group P21, with the cell dimensions: a=28.010 A, b= 62.947 A, c=60.512 A, and β=90.698 deg. Diffraction data was collected at room temperature for the crystal mounted in the capillary. This was done using an X-ray source, e.g. Nonius FR591 Rotating Anode Generator and an image plate detector MacScience DIP2000.

Figure 9 shows an example of the diffraction pattern from the selected lysozyme crystal. The resolution of this is 1.78A at the edge of the detector. 180 frames were collected (1 deg, 20min exposure per image), which produced 20163 unique reflections. The average I/σ for all data was 7.8, the average redundancy 3.6, the overall completeness 99.7% and Rmerge 8.3%. Processing of the data was performed with the programs from the HKL suite of programs. This data demonstrates that the crystals produced are suitable for high quality X-ray diffraction analysis . Other experiments have been conducted to optimise the growth of non-model proteins, e.g. the C-terminal fragment of the tetanus toxin. In one particular case, a protein was found to crystallise at between 18 and 19C but not at 21.5, at which temperature needle clusters appeared, these being unsatisfactory for structure determination. Using the array of Figure 1 demonstrated quickly that this particular protein was very sensitive to changes in temperature, more so than had been previously realised. Prior attempts to crystallise the protein in a reproducible manner had failed. Identifying that the temperature sensitivity was greater than anticipated and finding the crystallisation temperature, enabled optimal growth conditions to be identified relatively quickly. Once the conditions were determined, it was possible to grow high quality protein crystals in a reproducible manner. This is advantageous.

Various different methods could be employed to fabricate the array of Figure 1. In order to make a 100 element array 14 under industrial standards, for use in a 1536 well format, a 7.62 cm diameter (100) single crystal silicon wafer 1 mm thick silicon wafer was used. A layer of negative photoresist (SU8-50 from MicroChem Corp.) was spun onto one side of the wafer at 450 rpm for 60 sec. The template was defined in the photoresist using UV exposure through a corresponding photomask. The pattern was subsequently transferred into the silicon wafer by development of the photoresist, followed by inductive couple plasma, for 250min (ICP-Surface Technology System, Gases: SF6/C4F8/02, flow meter reading: 13s/7s/13s, corrected flow 130/85/10 seem, RF power: 600/12W) . The wafer was cleaned with a solution of hydrogen peroxide and sulphuric acid in a ratio 1:7. This method etches the silicon quite fast. The mask made of SU8-50 survives the process but needs to be removed after the ICP process with a solution of hydrogen peroxide and sulfuric acid 1:7. As will be appreciated, other fabrication options exist . Whilst the experiments described above were conducted manually, it will be appreciated that they could be automated and carried out under computer control , thereby to further improve throughput . In order to do this a processor is needed, together with a computer program having code or instructions causing the introduction of the crystallisation solution into the array of wells. This could be done by, for example, controlling an array of micropipettes to inject simultaneously solution into the wells. Once this is done, the computer program would control the temperature differential across the array, so that heat flow is maintained substantially constant. The program may also be adapted to causes the automatic sealing or isolating of the crystallisation solution within the wells or channels .

As an extension of the proposed automation, the computer program could also be operable to load the substrate into a heat control unit, such as an incubator, and then use the incubator to control the temperature differential. In a commercial situation, the incubator may be adapted to receive a large number of substrates and may be able to create a range of temperature differentials across these. The system described with reference to the drawings uses a silicon substrate for the protein crystallisation array. The use of silicon has some clear advantages, including the fact that there are established procedures for its micromachining, enabling a robust and reproducible fabrication process. Silicon has a thermal conductivity of -150W m^"1 K^"1 at 20°C, which makes it a good substrate for devices that require a rapid and efficient heat transfer. In addition, a silicon microarray can be readily cleaned and re-used, although care must be taken in the cleaning protocol to avoid contamination between proteins (commercial solutions for removing proteins are readily available) .

The use of silicon does however present some disadvantages, namely the high cost of non-standard wafers and the lack of optical transparency (which makes observation of the protein crystals difficult, without the addition of a dye) . Hence, other material may be used for fabricating the microarray. One option would be to use thermally conducting polymers, for example the polymer sold under the trade mark "Coolpoly" . In another variation, although Figure 1 shows wells or chambers that are defined in a substrate, microfluidic channels could also be used for receiving crystallisation solution.

The dimensions of the chambers described above are not limited by the fabrication methods used, but rather by considerations of the user-interface in a standard laboratory, where robots may not be available. In contrast to other devices for thermal manipulation of fluids, such as those used for polymerise chain reaction

(where volumes may be <1 μL, but reaction times are

fast) , protein crystallisation experiments may take several days or months. Consequently, one of the major problems was the sealing of the crystallisation chambers (to prevent evaporation) consequently leading us to use a larger volume that might be the case at the limits of miniaturisation . Notwithstanding the problem of evaporation, reduction in the size and increasing numbers of the crystallisation chambers will be possible in future devices, providing a route for high throughput screening of crystallisation conditions for post -genomics . Indeed future experiments currently underway and including those involving vapour diffusion and counter diffusion crystallisation methods, can more readily be formatted in closed microsystems, and may be more amenable to further reductions in size (for example using "industry-standard" formats of array element densities, such as 1536 or 3456 footprints) .

The systems and methods in which the invention is embodied provide an improved mechanism for screening protein crystallisation. The ability to screen the conditions of crystallisation and define the optimal parameter space increases the probability of finding new crystallisation conditions. The systems and methods described herein are particularly valuable in optimising the conditions for intractable biological systems, including those such as membrane-bound proteins which are not only difficult to purify in large amounts, but also are difficult to crystallise.

A skilled person will appreciate that variations of the disclosed arrangements are possible without departing form the invention. Accordingly, the above description of a specific embodiment is made by way of example only and not for the purposes of limitation. It will be clear to the skilled person that minor modifications may be made without significant changes to the operation described above .

Claims

1. A crystallisation system comprising an array of wells or channels defined in a substrate, the wells or channels being for receiving a crystallisation solution, and a temperature controller for creating a temperature differential across at least part of the array.

2. A system as claimed in claim 1, wherein the temperature controller is operable to vary the temperature of the array relative to an ambient temperature .

3. A system as claimed in claim 1 or claim 2, wherein the temperature controller comprises one or more peltier elements.

4. A system as claimed in any one of the preceding claims, wherein the temperature controller includes a temperature sensor for measuring temperature across the array.

5. A system as claimed in claim 4, wherein the temperature sensor is a thermistor or a thermocouple.

6. A system as claimed in claim 4 or claim 5, wherein the temperature sensor is in direct thermal contact with the substrate.

7. A system as claimed in any one of the preceding claims, wherein the temperature controller is operable to maintain a pre-determined temperature change or gradient across the array.

8. A system as claimed in any one of the preceding claims, wherein at least some of the wells or channels are aligned substantially perpendicular to a direction in which the temperature is to be changed.

9. A system as claimed in any one of the preceding claims, wherein the array comprises a 2D grid of wells or channels.

10. A system as claimed in any one of the preceding claims, wherein means are provided for sealing the crystallisation solution within the wells or channels .

11. A system as claimed in any one of the preceding claims, wherein the substrate comprises silicon.

12. A system as claimed in any one of claims 1 to 10, wherein the substrate comprises a thermally conductive polymer.

13. A crystallisation system comprising an array of wells or channels for receiving a crystallisation solution and a temperature controller for creating a temperature differential across at least part of the array, the arrangement being such that at least some of the wells or channels are aligned substantially perpendicular to a direction in which the temperature is to be changed.

14. A system as claimed in claim 13, wherein the temperature controller is operable to vary the temperature of the array relative to an ambient temperature .

15. A system as claimed in claim 13 or claim 15, wherein the temperature controller comprises one or more peltier elements.

16. A system as claimed in any one of claims 13 to 16, the temperature controller includes a temperature sensor for measuring temperature across the array.

17. A system as claimed in claim 16, wherein the temperature sensor is a thermistor.

18. A system as claimed in claim 16 or claim 17, wherein the temperature sensor is in direct thermal contact with the substrate.

19. A system as claimed in any one of claims 13 to 18, wherein the temperature controller is operable to maintain a pre-determined temperature change or gradient across the array.

20. A system as claimed in any one of claims 13 to 19, wherein the array comprises a 2D grid of wells or channels.

21. A system as claimed in any one of claims 13 to 20, wherein sealing means are provided for sealing the crystallisation solution within the wells or channels.

22. A system as claimed in any one of claims 13 to 21, wherein the array of wells or channels is defined in a substrate.

23. A system as claimed in claim 22, wherein the substrate comprises silicon.

24. A system as claimed in any one of claims 13 to 23, wherein the substrate comprises a thermally conductive polymer.

25. An array of wells or channels defined in a substrate adapted for use in the system of any one of the preceding claims .

26. A method for determining optimal crystal growth conditions, the method comprising introducing into an array of wells or channels defined in a substrate a crystallisation solution, and maintaining a temperature differential across the array.

27. A method as claimed in claim 26 further comprising sealing or isolating the solution within the wells or channels, thereby to limit evaporation.

28. A computer program for incorporation on a data carrier or a computer readable medium, the computer program comprising code or instructions for: introducing into an array of wells or channels defined in a substrate a crystallisation solution, and maintaining a temperature differential across the array.

29. A computer program as claimed in claim 28 comprising code or instructions for sealing or isolating the solution within the wells or channels.

30. A computer program as claimed in claim 28 or claim 29 comprising code or instructions for causing the substrate to be placed into an incubator.

31. A computer program as claimed in claim 30, wherein the incubator is adapted to receive a plurality of substrates and the computer program has code or instructions for causing substrates to be placed in the incubator.

32. A temperature control system adapted to control a temperature differential across an array of wells or channels.

33. A computerised temperature control system adapted to control a temperature differential across one or more arrays of wells or channels.

34. A system comprising an array of wells or channels defined in a substrate and a temperature controller for creating a temperature differential across at least part of the array.

35. A crystallisation system substantially as described hereinbefore with reference to the accompanying drawings and as shown in Figure 1.

36. A method for crystallising proteins substantially as described hereinbefore with reference to the accompanying drawings .

37. A computer program substantially as described hereinbefore with reference to the accompanying drawings .