Automation of Protein Crystallography

Successful Identification of Structure Dependent on Producing High Quality Crystals

Protein crystallography has significantly aided the understanding of human disease at a molecular level. As the knowledge of accurate molecular structures is a prerequisite for rational drug design, protein crystallography has also had a substantial impact on the drug discovery pipeline.

Among the wealth of structures investigated are disease-specific targets whose further functional studies can considerably aid the development of effective therapeutics. Target validation has been a downstream area where protein crystallography has proved effective. The ability to study the structure of protein/drug complexes can provide more detailed information on their interactions.

As high-throughput screening has developed and increased in sophistication, protein crystallographers have been under considerable pressure to keep up with the leads produced. There has subsequently been substantial investment to develop high-performance protein crystallography instrumentation to maximize throughput and efficiency. This can allow scientists to develop high-throughput methodologies without sacrificing data quality while providing the flexibility to optimize experimental parameters for individual proteins.

Here, we describe how such a process can be facilitated using the Rhombix™ range of modular automated crystallization platforms from Thermo Electron.

Experimental Optimization

The successful identification of a protein structure is reliant on the production of high quality crystals. This requires the use of extremely pure protein, which is often in limited supply as protein production and purification can be costly and labor-intensive. Additionally, every protein provides its own crystallization challenge and specific protein groups, such as membrane proteins, are notoriously difficult to crystallize.

Furthermore, for structural determination using X-ray diffraction, not all crystals produced will diffract sufficiently. Thus, the crystallization process may have to be revisited; surface residues may have to be modified or crystallization conditions changed.

Crystal growth will occur when molecules are brought into a thermodynamically unstable state called supersaturation, achieved through the gradual removal of solvent. Nucleation, growth, and growth cessation are the three steps of crystal growth and each may require a change in experimental conditions to produce the best quality crystal.

Every protein crystallization process is hence designed and set up under a variety of conditions, adjusting experimental parameters to identify optimal crystallization conditions. These can include temperature, pH, precipitants, additives, protein concentration, and expression and purification conditions. Essentially each parameter variable becomes a crystallization experiment.

Frequently performed in 96- or 24-well plates, each sealed well will contain at least one individual experiment. This process is of paramount importance, since some proteins may only crystallize under a specific set of conditions.

In the past, screening and optimization of experimental conditions have relied on what is commonly regarded as the black art of protein crystallography, often incorporating more complex methods that were difficult to automate for high-throughput experimentation. Scientists now have the ability to automate most of these methods.

Automation, however, has resulted in a considerable amount of experimental data that must be logged. Providing a flexible and easy to use alternative to manual logs or isolated databases is necessary to fully exploit the data produced. All experimental conditions and results should be recorded, closely monitored, and easily tracked to ensure repeatability for future crystallizations. Additionally, many optimizations are often required before acceptable crystals are even visible, thus having the ability to data mine from previous experiments or initial screens is invaluable.

Automation for Higher Throughput

The protein crystallization process can be divided into several steps: experimental design and set up, plate storage and imaging, and data capture, analysis, and mining. Automated experimental configuration and execution must allow for the use of multiple plate types and ideally, crystallization techniques, i.e. sitting drop, hanging drop, or microbatch, often defined by the individual laboratory or the protein under investigation (Figure 1).

Fig. 1: Crystallization techniques: A) Sitting Drop; B) Hanging Drop; C) Microbatch

The typical plate designs for automation enable sitting drop experiments, comply to SBS standards, and contain 96 wells, each of which are sealed after the building of the screen and the placement of the drops are complete. Other formats are also used, the most popular for optimization experiments being the traditional Linbro format for hanging drop experiments.

One or all crystallization steps may require automating but for true high-throughput crystallization, integration into one fully automated platform is preferable. By incorporating the most advanced technologies into the working schedule, the crystallographer can optimize each application. Thermo’s Rhombix crystallization products offer a modular working solution, a tool for each process that can be fully integrated if required.

Setu-up and Analysis

Rhombix Screen systems can integrate and control robust liquid-handling instrumentation for drop-making and screen preparation. Manual plate set-up can be laborious, and multiple protein and reagent titrations are time-consuming and low throughput. With the increasing sophistication of low-volume pipettors available, protein crystallographers can dispense a range of viscosities and volatilities with speed, minimizing evaporation. Thus it is now possible to screen numerous crystallization variables using nanoliter volumes, significantly reducing waste of valuable protein or expensive reagents and reducing set-up time from weeks to minutes.

Crystals may take weeks or months to grow, with a shelf life of around 6–12 months. To maintain reproducible experimental conditions, plates must be stored in a stable environment, at a constant temperature. Thus, reliable plate storage that meets these criteria while still allowing easy access is essential.

Rhombix Vision is a series of efficiently controlled storage hotels integrated with imaging capabilities spanning a range of throughput and storage requirements that allow the robotic handling and imaging of bar-coded plates one at a time when needed with smooth movement and random access. This is an efficient alternative to standard plate stacking, where retrieval can be disruptive, arduous, and detrimental to crystal growth.

Used to monitor crystal formation and growth, crystal imaging is fundamental to the crystallization process. Image scheduling is essential, as early intervention can be critical to adjust drop solutions for enhanced growth. Each imaging event should be scheduled over a period of days, weeks, and months at specific intervals, with each drop imaged around 5–10 times.

Image Capture and Data Management

Crystals, especially in early growth, often remain elusive but their birefringent properties can be exploited through the use of polarized light for easier visualization (Figure 2). As a result, advanced imaging technology that can be programmed for automatic execution can provide the crystallographer with a valuable tool for experimental monitoring and subsequent optimization.

Fig 2: Crystal visualization without polarization (left) and with (right)

The Rhombix Vision hardware incorporates a specialized camera and lens configuration that enables true color detection with multimode imaging capacity, such as polarized extinction and variable angles, for enhanced imaging. Advanced software algorithms may be used to automatically zoom into the drop and extend the depth of focus for better crystal visualization, with one or both of these being employed. Additionally, diffuse LED light sources are used, reducing heat generation and eliminating artifacts, such as shadows around drops.

Employed not only for the efficient control of plate storage and handling, Thermo’s Rhombix Software Suite has components for dynamic image scheduling and real-time experimental monitoring of automation and experimental status. Furthermore, each crystallization tool can be controlled individually or integrated to pass relevant information between applications, allowing crystallographers to design, view, and refine experiments.

All experimental details, process parameters, and automation settings can be recorded and stored in one centralized database for advanced data mining and experimental monitoring (Figure 3). Such extended functionality enables the crystallographer to manage data from hundreds or thousands of experiments, tracking parameters such as protein purification or crystallization conditions. Linking crystallization success or failure to a specific set-up increases the effectiveness of experimental optimization, reproducibility, and scale up.

Fig.3: Rhombix Software allows all experimental details to be efficiently stored for advanced data mining and experimental monitoring


Automated, high-throughput crystallization is essentially a screening process. For true high throughput, the laboratory must employ a strategy for effective process integration, incorporating fast and accurate pipetting instrumentation, high-capacity crystal storage linked to automated imaging facilities, and the resources to capture, record, and export all experimental data.

Bringing together all experimental components into one platform can offer scientists great benefits, not just for increased throughput but also for effective data management and time efficiency. Additionally, as automation requirements differ considerably between institutions, modular systems can allow a smooth transition to greater throughput with the addition of instrumentation when required.

Fig 4: The crystallization process workflow

Thermo’s Rhombix series of crystallization products can be fully integrated into one platform, the Rhombix Opus, allowing crystallographers to design and set-up experiments, automatically store and image crystallizations, and capture data for analysis using an intuitive interface (Figure 4).

Having one system means data capture and experimental monitoring is easy and efficient and the risk of human error reduced, allowing the effective optimization of the crystallization process.


Paige Vinson, Ph.D., is applications scientist, protein crystallization at Thermo Electron. Phone: (615) 263-9406. E-mail: [email protected].