Metabolic Profiling

Metabolic profiling is defined as identification and quantification of metabolites (Hall 2006). Approximately 90 primary metabolites (Carrari et al. 2006) and 320 volatile metabolites (Tikunov et al. 2005) of tomato have been routinely profiled by GC-MS. A limited number of secondary metabolites have been successfully profiled using LC-MS due to their intrinsic complexity and the lack of a database. Metabolic profiling often includes estimating correlations between metabolites and between metabolites and other measures, such as gene expression and phenotypes. A correlation profile can provide critical clues to generate hypotheses concerning metabolic pathways, for example, assigning gene function or understanding gene interactions.

Cultivated varieties and wild species of tomato display a wide range of morphological and physiological phenotypes. Substantial variation of metabolite content is therefore expected. Metabolite profiles of S. lycopersicum and five wild species (S. pimpinelli-folium, S. neorickii, S. habrochaites, S. chmielewskii, and S. pennellii) have been compared (Schauer et al. 2005). Extracts from leaves and fruits were analyzed using GC-MS, and levels of more than 80 metabolites (generally primary metabolites) were estimated. The leaf metabolite profile of S. lycopersicum was closest to S. pimpinellifolium, followed by S. pennellii,

Fig. 12. Concepts of metabolic profiling and metabolite annotation. Metabolic profiling aims to identify and quantify metabolites. Metabolite annotation aims to provide chemical information concerning the peaks, especially unknown peaks, to allow consistent identification of peaks across multiple samples. Annotated peaks and known metabolites are then used in metabolic profiling analysis. Metabolic profiling data is statistically analyzed using PCA, hierarchical clustering analysis (HCA), and network analysis

Fig. 12. Concepts of metabolic profiling and metabolite annotation. Metabolic profiling aims to identify and quantify metabolites. Metabolite annotation aims to provide chemical information concerning the peaks, especially unknown peaks, to allow consistent identification of peaks across multiple samples. Annotated peaks and known metabolites are then used in metabolic profiling analysis. Metabolic profiling data is statistically analyzed using PCA, hierarchical clustering analysis (HCA), and network analysis

Fig. 13. Schematic relationships between molecular breeding and functional genomics. Functional genomics resources that potentially support each step of breeding are indicated (gray boxes)

S. chmielewskii, and S. neorickii. S. habrochaites was the most distinct from S. lycopersicum. In contrast to leaves, fruit metabolite profiles showed higher degree of variation. In particular, accumulation levels of sucrose, isocitrate, chlorogenate, and shikimate were higher in the wild species than in S. lycopersicum. Accumulation levels of glucose, fructose, a-tocopherol, and puturesine were lower in the wild species than in S. lycopersicum. The contents of amino acids and their derivatives were generally higher in S. lycoper-sicum compared to wild species, with the exception of higher y-aminobutiric acid (GABA) in S. pennellii, and higher tryptophane in S. habrochaites.

Comprehensive metabolic profiling of S. lycoper-sicum x S. pennellii ILs using GC-MS was reported by Schauer et al. (2006). In this study 74 metabolites were quantified in ripe fruit pericarps from all 76 ILs. Metabolite contents tended to show increases in the ILs relative to S. lycopersicum. These increases were not limited to metabolites that were inherently more abundant in the wild species. For example, trichloroacetic acid cycle intermediates that were invariant across several species (Schauer et al. 2005) apparently increased in the ILs.

Comprehensive profiling of secondary metabolites using LC-Qtime offlightMS (LC-Q-TOF-MS) was applied to compare high pigment-2dg (hp-2dg) mutant and wild-type (wt) S. lycopersicum alleles (Bino et al. 2005). The hp-2dg allele is associated with increased levels of carotenoids, flavonoids, and ascorbic acid. The authors found 383 mass signals that were significantly higher in hp-2dg compared to wt, and 62 mass signals that were significantly lower in hp-2dg than wt in red ripe fruit. Only eight metabolites were successfully identified out of 445 mass signals.

To link metabolic profiling data to functional ge-nomics, metabolites must be assigned to biological pathways and functions of genes involved in syntheses and degradation of metabolites must be elucidated. These goals can be achieved by combining metabolic profiling data with genetic, genomics, and phenotypic data. A correlation-based analysis is a powerful approach in analyzing putative relationships between metabolites and genes. In the absence of gene expression or genotypic data, a metabolite-to-metabolite correlation pattern can provide insights into a metabolic pathway. Tikunovet al. (2005) calculated pair-wise correlations between 322 volatiles detected by GC-MS. From the correlation matrix, the authors found that volatiles within a cluster were derived from a common biochemical precursor. Volatiles in five major clusters were derived from phenylalanine, leucine, isoleucine, lipid, and isoprenoid, respectively. This result suggested that the volatiles within the same metabolic pathway tended to accumulate in a coordinated manner.

Integration of metabolic profiling data with phe-notypic traits was performed on 76 S. lycopersicum x S. pennellii ILs (Schauer et al. 2006). Profiles of 74 metabolites obtained by GC-MS were combined with data from nine yield-associated traits, and correlation coefficients between all possible pairs of metabolic and phenotypic traits were estimated. The authors illustrated the correlation patterns in the form of a network that was divided into three large modules. Module 1 comprised mainly phenotypic traits and sugar phosphates; module 2 comprised amino acids; and module 3 included sugars and organic acids. In the phenotypic traits module, harvest index and Brix° had relatively high numbers of total connections both within the module and to nodes in the two external modules, i.e., they served as network hubs. Brix° had high numbers of connections, probably because it is an integrator trait whose variance depends on the variance in many other traits. Interestingly, all other phenotypic traits were highly connected within module 1 and showed sparse connections to external modules. Most metabolites associated with phenotypic traits belonged to pathways of central metabolism. However, whether observed correlations directly reflected causal relationships cannot be concluded by a correlation-based study alone. Nevertheless, integrated analysis of metabolic profiling and phenotypic traits is potentially a powerful tool in building the foundation for molecular breeding.

Metabolite profiles are sometimes studied in conjunction with transcript profiles (Alba et al. 2005; Carrari et al. 2006). Carrari et al. (2006) performed a comprehensive parallel analysis of transcript and metabolites during tomato fruit development. Their study included the profiling of 92 tomato fruit metabolites by combined GC-MS and photodiode array high-performance liquid chromatography (PDA-HPLC) analyses, transcriptome analysis using a TOM1 microarray (Alba et al. 2004), and estimating metabolite-to-metabolite, transcript-to-transcript, and metabolite-to-transcript correlations. Correlations showed that metabolites belonging to the same class were regulated in a highly coordinated fashion, while transcripts belonging to the same functional group were relatively less coordinated.

Moreover, metabolite-to-transcript correlations demonstrated that levels of metabolites sometimes displayed low correlations with transcript levels of genes within the pathway. For example, cell wall polysaccharides and genes associated with cell wall metabolism showed little correlation. These results suggested that metabolism was largely regulated at the post-transcriptional level. However, several high correlations between metabolites and transcripts were identified. Sugar phosphates, organic acids, and pigments (carotenoids, xanthophylls, chlorophylls) were highly correlated with ripening-related transcripts including ACC oxidase, ethylene receptor, ripening-inducible transcription factors, and ethylene response genes. The authors assessed the correlation between primary metabolism and pigment metabolism. Although the metabolite levels of TCA cycle intermediates (organic acids) showed few correlations to pigment levels, TCA cycle intermediates and pigments had significant correlations to the same ripening-associated transcripts.

The integration of metabolomic data with ge-nomic data offers a promising approach to prioritize candidate genes for molecular breeding. To aid in understanding correlations between transcripts and metabolites by visualization of transcript changes within the context of metabolic pathways, the software MapMan, originally developed for Arabidopsis (Thimm etal. 2004), has been revised for Solana-ceous plants as Solanaceous MapMan (Urbanczyk-Wochniak et al. 2006).

Was this article helpful?

0 0
Addiction To Nutrition

Addiction To Nutrition

Get All The Support And Guidance You Need To Be A Success At Beating Addictions With Nutrition. This Book Is One Of The Most Valuable Resources In The World When It Comes To A Definitive Guide To Unchain Addiction The Smarter And Healthy Way.

Get My Free Ebook


Post a comment