Development of a robotic metabolomics sample preparation platform

Metabolomics is one of the key technologies that can contribute large-scale data to the new life sciences research area of Systems Biology and the MeT-RO project provides metabolomics data to a variety of UK research institutions that are working towards a more complete understanding of biological systems. The ability to quantify many metabolites as they change - as a result of genetic modification, disease or environment / requires multiple measurements on cells and tissues. Increasingly several thousands of tissue samples are involved in single metabolomics experiments. The chemical analysis itself is carried out by a number of spectroscopic techniques such as Nuclear Magnetic Resonance (NMR) and Mass Spectroscopy (MS). Modern instruments for this analysis run in automation with robotic sample changers. However, before samples are ready to be loaded into automated spectrometers, biological tissue and fluid samples need varying amounts of processing. The preparation of plant and microbial tissue for metabolomics data collection on these analytical instruments, in particular requires tedious manual processing. Many thousands of samples need to weighed, extracted with solvents and manipulated by processes such as centrifugation, pipetting and chemical derivatisation. This proposal aims to develop laboratory robotics and associated control and data capture systems to carry out these repetitive tasks and prepare the samples that can then be analysed by robotic spectrometers already in use in the MeT-RO project. In this way many more samples can be processed as the machines can be operated 24 hours a day. In addition the precision of robotics and use of bar-coding will minimise experimental errors that are introduced into the system by human processing and labelling. In summary we aim to carry out the research to design, and construct an automated laboratory, similar to a car assembly line, to carry out metabolomic analysis from sample weighing through data collection on spectrometers, to multivariate analysis of thousands of datasets. This will contribute significantly to large scale biology projects that are aiming to integrate metabolite, protein and gene expression profiles of organisms as they develop and react to environmental changes and stresses such as disease.

The project aims to provide robotic systems that can carry out all of the processes involved in sample preparation for metabolomics. For systems biology, metabolomics technology needs to be improved such that higher throughput and data quality can be assured. Metabolomics data is usually obtained by NMR and MS, which can run in high throughput using existing autosampler technology. Sample preparation is the key to quality. Here, we present a design for a robotic 'wet' laboratory that carries out all of the operations involved in randomised, triplicate, sample preparation for NMR, GC-MS and ESI-MS starting from freeze-dried, milled biological material. The system is built on RRes SOPs and is based on two XYZ pick and place robots that will weigh tissue in triplicate, randomising aliquots into different 96 tube arrays. Samples for NMR and ESI-MS analysis will be extracted by a process involving solvent addition, mixing, heating, and centrifugation steps. Solutions will then be dispensed into NMR and ESI-MS sample tubes for analysis. The whole process will be tracked by bar-coding. The system will also recycle NMR tubes. Samples for GC-MS will be further processed by a robotic derivatisation system that will prepare MOX-TMSi derivatives directly from freeze-dried tissue aliquots. This system based on a Gerstal MPS2 robot will interface with the LECO Pegasus III GC-TOF autosampler, such that these air sensitive samples are prepared 'just in time' for analysis. The system will be adaptable to other metabolomic samples such as clinical and culture fluids. It will be of immediate use in the MeT-RO service but will also be used to implement 'dynamic metabolomics', where more information can be obtained by following reaction of a biological system over time. To be statistically valid, such experiments require processing and data collection of much larger numbers of samples. This will only be possible with the proposed robotics.