Correlation Network Analysis reveals a sequential reorganization of metabolic and transcriptional states during germination and gene-metabolite relationships in developing seedlings of Arabidopsis
1 School of Biological Sciences, Bangor University, Bangor, Gwynedd LL57 2UW, UK
2 INRA, Université de Bordeaux, UMR619 Fruit Biology Unit, BP 81, F-33140 Villenave d Ornon, France
3 Division of Surgery, Oncology, Reproductive Biology and Anaesthetics, Sir Alexander Fleming Building, Imperial College London, London SW7 2AZ, UK
4 Plateforme Métabolome du Centre de Génomique Fonctionnelle Bordeaux, IFR103 BVI, BP 81, F-33140 Villenave d'Ornon, France
BMC Systems Biology 2010, 4:62 doi:10.1186/1752-0509-4-62Published: 13 May 2010
Holistic profiling and systems biology studies of nutrient availability are providing more and more insight into the mechanisms by which gene expression responds to diverse nutrients and metabolites. Less is known about the mechanisms by which gene expression is affected by endogenous metabolites, which can change dramatically during development. Multivariate statistics and correlation network analysis approaches were applied to non-targeted profiling data to investigate transcriptional and metabolic states and to identify metabolites potentially influencing gene expression during the heterotrophic to autotrophic transition of seedling establishment.
Microarray-based transcript profiles were obtained from extracts of Arabidopsis seeds or seedlings harvested from imbibition to eight days-old. 1H-NMR metabolite profiles were obtained for corresponding samples. Analysis of transcript data revealed high differential gene expression through seedling emergence followed by a period of less change. Differential gene expression increased gradually to day 8, and showed two days, 5 and 7, with a very high proportion of up-regulated genes, including transcription factor/signaling genes. Network cartography using spring embedding revealed two primary clusters of highly correlated metabolites, which appear to reflect temporally distinct metabolic states. Principle Component Analyses of both sets of profiling data produced a chronological spread of time points, which would be expected of a developmental series. The network cartography of the transcript data produced two distinct clusters comprising days 0 to 2 and days 3 to 8, whereas the corresponding analysis of metabolite data revealed a shift of day 2 into the day 3 to 8 group. A metabolite and transcript pair-wise correlation analysis encompassing all time points gave a set of 237 highly significant correlations. Of 129 genes correlated to sucrose, 44 of them were known to be sucrose responsive including a number of transcription factors.
Microarray analysis during germination and establishment revealed major transitions in transcriptional activity at time points potentially associated with developmental transitions. Network cartography using spring-embedding indicate that a shift in the state of nutritionally important metabolites precedes a major shift in the transcriptional state going from germination to seedling emergence. Pair-wise linear correlations of transcript and metabolite levels identified many genes known to be influenced by metabolites, and provided other targets to investigate metabolite regulation of gene expression during seedling establishment.