Metabolomic Profiling has emerged as a platform technology for quantitatively assessing the metabolic fingerprint of a biological system, being actually one of the fastest growing –“omics” technique to-date. In its short five year history Metabolomic analysis has already shown potential for commercial applications in Nutrition, Healthcare, Diagnostics, Toxicology, AgriBiotech and Industrial Biotech applications. In spite of the drastically increasing interest, investment and growth, however, there are still issues regarding accuracy of measurements, sample preparation and protocol standardization, speed and user-friendliness that need to be resolved. Further there is also a need for large dynamic metabolomic data-sets which can be used to develop data analysis methods for identifying regulation, metabolic network reconstruction and integrated metabolomic and transcriptomic analysis for building system level models.

In this context, we systematically perturbed the Arabidopsis thaliana liquid culture system by applying (1) Elevated CO₂ stress, (2) Osmotic (NaCl) stress, (3) Sugar (trehalose) signal, and (4) Hormone (Ethylene) signal, individually, and stress (1) in combination with stresses (2)-(4). The short-term response of the biological system to this plethora of perturbations was monitored in a high-throughput way at the metabolic level by harvesting plants at different time points throughout the first 30h period after the initiation of the perturbation and measuring their polar metabolomic profile using Gas Chromatography-Mass Spectrometry (GC-MS), which has been the most widely used platform for metabolomic analysis due to its low cost and technical advantages over other platforms.

GC-MS metabolomics, however, requires derivatization of the original sample, with potential for biases which distort the one-to-one proportionality between the original metabolite concentration and the derivative peak area profiles. It is imperative that the metabolomic profiles are corrected from these biases, because of the high risk of assigning biological significance to changes due only to chemical kinetics. A streamlined data correction, normalization and validation strategy [1] not jeopardizing the high-throughput nature of metabolomics analysis was developed which significantly increased the accuracy, reliability and reproducibility of GC-MS metabolomic analysis. These corrected profiles were than analyzed using various multivariate statistical / machine learning techniques incorporated in TIGR MEV. Significant metabolites at individual time points were determined using MiTimeS software [2] which was previously developed in our lab for transcriptomic analysis.

It has to be underlined that this is among first(if not the first) currently reported studies of plant physiology that concerns metabolic fingerprinting of dynamic plant response to such an extensive number of individual and simultaneously applied perturbations. Particularly, the role of CO₂, the primary source of carbon in plants, was elucidated in great extent since it was the common stress among all combined perturbations. In the context of plant physiology, this study has provided a vast amount of data that allow for a comprehensive understanding of the primary metabolism regulation, which of great interest for the emerging role of plants in large scale production in bio-refineries, bio-plastics and even some industrial chemicals.

Apart from contributing, however, significantly in plant physiology research, the present work materializes also an extensive metabolomic study that demonstrates the importance of metabolomics in deciphering metabolic regulation networks even in highly complex eukaryotic systems.

1. Kanani H and Klapa MI, 2006. “Data Correction Strategy for Metabolomics Analysis using Gas Chromatography-Mass Spectrometry”, under review (also US Letter patent application 11/362,717) .

2. Dutta B, Snyder R and Klapa MI, 2006. “Significance Analysis for time-series transcriptomic data." under review

* This work is funded by US NSF (QSB-0331312)