Automated HT crystallography transforms drug discovery
technical developments in which the development of automation in
protein expression, protein engineering and crystallisation will
provide the accurate screening data that has proved elusive to
date.
The emergence of HTP crystallography has been the result of a synergy of technical progress and the financial influx that has resulted in increased public funding. This has lead to the opportunity to finally address the flaws that have hampered this method so far.
Difficult tasks remain in the area of automated crystal harvesting and handling, feedback from crystal image analysis to custom optimisation, and in automated validation and ligand building.
As well as the development of high throughput automation technology the development of integrated crystallographic data collection and computation techniques will allow a viable solution to the refinement, validation and analysis of the screening data.
The financial influx from both the public and commercial sector has injected much needed investment within the sector, which in turn is set to massively reduce the cost per structure of the early crystallography screening process.
The National Institute of Health's (NIH) protein structure initiative (PSI-I) provides public funding for nine structural genomics centres at $3 million (€2.3 million) per year. The NIH roadmap for PSI-Phase II in 2005 calls for 3-5 large centres to be built at a cost of $75 million for five years, with plans for small specialised centres to be built. Organisations such as the Structural Proteomics In Europe (SPINE), the International Conference on Structural Genomics (ISGO) and the International Symposium on Grid Computing (ISGC) were also singled out as vital contributors to technique advancement.
In particular, the introduction of powerful 3rd generation synchrotrons, massive machines built to accelerate sub-atomic particles to almost the speed of light have opened up new techniques for advanced phasing. Third generation synchrotron radiation sources allow phase contrast imaging, hard X-Ray inelastic scattering and microdiffraction to be used in the crystallography approach that could vastly improve its analytical power.
Speaking at the Lab Automation Europe show in London, Bernhard Rupp, group leader of the maromolecular crystallography and structural genomics group at the University of California, also detailed the recent influx of large venture capital into structural genomics biotech, in particular, from drug discovery companies.
US biotechnology company, SGX, has invested a significant amount in X-ray diffraction data collection by building the SGX-CAT at the Advanced Photon Source (APS), located at the US Department of Energy Argonne National Laboratory in Illinois.
The company's own proprietary technology is a high-throughput X-ray crystallography technique that yields detailed three-dimensional protein structures that creates homology sets that travel through the pipeline together, which increases the odds of coming up with solved structures for a given target.
UK drug discovery company Astex Technologies has also embraced high throughput technology with AutoSolve, its HT protein-ligand crystallography software that generates protein-ligand complexes by the automated processing of X-ray data at an increased rate. The company claim the time taken to generate protein-ligand crystal structures from X-ray data has been reduced from days to minutes.
Its newest technology for drug discovery, the DCX, uses X-ray crystallography to detect potent small-molecules that are generated by chemical reactions between complimentary fragments binding in the active site of a target protein. These small molecules are synthesized in a combinatorial library in the presence of crystals of the protein target, and then detected by X-ray crystallography.
The investment shown by the commercial sector highlights a trend showing the cost of early target screening gradually reducing year by year. According to the NIH, current estimates show cost per structure falling from $250 000 (€190 000) in 2000 to $150 000 in 2002-03 with a 2005 price prediction of less than $50 000.
In customising automated HTP crystallography, the issue and demands on liquid handling becomes a priority. This is due to the large volumes (ml range and up) for the cocktail preparation and precise nanolitre dispensing for plate setup requiring different machine liquid handling capabilities.
The trend now emerging sees specific, cocktail preparation will become more prevalent than fixed, kit-based screening. Indeed the screening protocol must provide optimal screening flexibility. Rupp recommended that the inclusion of prior knowledge would allow design of maximally efficient screens.
Rupp emphasised it was important to recognise not every research environment required the same degree of automation and operations research. He went on to say the operations research and process analysis were necessary to optimise the robotic automation layout most suitable for maximum overall efficiency.
He summed up the major difficulty in scaling up from manual to automated processing by saying: "Simply copying human behaviour does not produce a smart robot. It produces an expensive, wasteful, inefficient, mechanised contraption."
