HOME :: CHAPTER 11 :: Essay 11.2

PREVIOUS :: NEXT

Essay 11.2

Metabolic Profiling of Plant Cells

Phuc T. Do, Sonia Osorio, Alisdair R. Fernie, Max-Planck-Institut für Molekulare Pflanzenphysiologie, Germany

September, 2010

Introduction

Metabolomics has been carried out since the mid 1970s, but only became a standard laboratory technique in the past decade (Fernie et al. 2004). It has emerged as a valuable technology for the comprehensive profiling and comparison of metabolites in biological system of both primary and secondary metabolism. Given plants are especially rich in chemically diverse metabolites, which are usually present in a large range of concentrations, profiling the levels of a broad range of metabolites is highly challenging. In this essay we will introduce the methods currently used in plants, as well as the statistical tools used to analyze and interpret the large data sets obtained on the application of the same. We will then describe the general applicability of these methods in phenotyping genetically and environmentally diverse plant systems, illustrated with data from our laboratory on respiratory metabolism. We additionally concentrate on the growing role of metabolite profiling in plant systems biology. It will begin with a historical evaluation of holistic approaches to biochemistry. Thereafter, its major focus will be on the integration of broad-range metabolite profiling into genomics approaches, and we will briefly touch on the use of metabolic profiling in gene functional identity elucidation and in genomics-assisted breeding. Finally, we describe future perspectives for this rapidly expanding branch of plant biochemistry.

Methodology

Given the chemical diversity and dynamic range of the metabolites present in the plant kingdom, no single analytical method is currently capable of extracting and detecting all metabolites. However, iterative improvement of the coverage and accuracy of applied methods is ongoing.

Traditional analysis deals with a limited number of compounds that were expected to be of particular importance in a given situation. For example, in the case of carbohydrate metabolism, analyses usually focus on starch and sugars, and investigations of glycolysis and respiration tend to look at intermediates of these individual pathways. While these approaches have yielded important insights into metabolic regulation, they have been limited to the pathway under study. They are also labour intensive because the methods used provide information on only a single compound per assay (as typical in enzyme-based metabolite analyses). In the case of chromatographic separation coupled to nonspecific detection, the analysis can only be applied to relatively simple mixtures, which often require clean-up steps.

Over the past decade, several methods suitable for large-scale analysis and comparison of metabolites in plant extracts have been established. So far two main technologies, namely, nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry (MS), have been employed. MS-based platform are most widely used in plant metabolomics, where a separation technique such as gas chromatography (GC), capillary electrophoresis (CE), or liquid chromatography (LC) is coupled to different kinds of MS. Methods developed in our institute based on the coupling of gas chromatography to mass spectrometry (GC-MS), enable the identification and robust quantification of a few hundred primary metabolites within a single extract. For example 150 compounds were identified in potato (Solanum tuberosum) tuber tissue (Roessner et al., 2000) and 326 compounds were identified in Arabidopsis thaliana leaf (Fiehn et al., 2000). This method has recently been applied to other plant species such as tomato, wheat, strawberry, cucumber, poplar and eucalyptus (Schauer and Fernie, 2006). The technical reliability was tested by a wide range of control studies that determined both the variability of the procedure and the recovery of 25 defined metabolites through the process. These tests revealed that the biological variability in wild-type potato tubers was generally an order of magnitude higher than that associated with the analytical process (Roessner et al., 2000). GC-MS has a relatively broad coverage of non-volatile compound classes, mainly those involved in primary metabolism, including organic and amino acids, sugars, sugar alcohols, phosphorylated intermediates within the polar phase (Roessner et al., 2000; Schauer et al., 2006), as well as lipophilic compounds such as fatty acids, fatty alcohols, sterols, and aliphatics within the apolar phase (Fiehn et al., 2000). The preferred method for analyzing semi-polar metabolites is LC-MS. Compounds detected by LC-MS include the large and often economically important group of plant secondary metabolites such as phenylpropanoids, flavonoids, and alkaloids (Tohge et al., 2005; Moco et al., 2007; Iijima et al., 2008; Mintz-Oron et al., 2008). Depending on the type of column used, various primary metabolites including several polar organic acids and amino acids can be reliably analyzed by LC-MS. An alternative platform for broad scope metabolic profiling is nuclear magnetic resonance (NMR), which relies on the detection of para-magnetic nuclei of atoms following application of a constant magnetic field. These are well-developed and well-validated methods and the computer software associated with NMR instrument is advanced. Compared to MS-based approaches, NMR requires very limited sample preparation and offers potential structural and quantitative annotation of the measured compound. The main problem with respect to metabolomics using NMR remains its low sensitivity if applied to highly complex biological samples (Dunn et al., 2005; Eisenreich and Bacher, 2007). Although no single method provides a really comprehensive analysis of plant metabolites, since the last edition of this essay (2006) techniques developed allow the detection and quantification of a wide range of abundant metabolites (Fernie and Schauer, 2009).

Application of Metabolic Profiling Methods for the Characterization of Diverse Biological Systems

Having established that the GC-MS methods were sound, Fiehn et al. (2000) applied them to study the metabolic complement of four different Arabidopsis genotypes: The ecotypes Colombia (Col-2) and C24, the dgd1 mutant (deficient in galactosyl transferase activity displayed a reduction in digalactosyldiacylglycerol) in a Colombia background, and the sdd1-1 mutant (exhibiting enhanced stomatal density) in the C24 background. Statistical evaluation of the data set showed that there were 41 significant differences in the sdd1-1 mutant and 153 significant differences in the dgd1 mutant when compared to their respective parental ecotypes. Fiehn et al. (2000) also applied the statistical tool of principal component analysis, which reduces the dimensionality of the vast and complex data sets and allows a display of linear combinations of the original independent variables. Interestingly, it became clear that the metabolic phenotypes of the two ecotypes were more divergent than the mutants were from their respective parental ecotypes. This experiment shows that metabolic profiles can be used to define the metabolic phenotype of a system. Genotypes and ecotypes can be clearly distinguished on this basis and thus underscore the value of metabolic profiling for genomic studies.

Figure 1 Phenotyping of modified plant systems by metabolic profiling. The metabolic complement of these systems is analyzed by principal component analysis and is represented two-dimensionally based on the maximum differences between the metabolite contents of the various genetically modified potato tubers. Therefore, samples displaying similar metabolite complements are very close graphically, whereas those that are very distinct are very distant. (Reproduced from Roessner et al. 2001, with kind permission of the publisher. © American Society of Plant Biologists.) (Click image to enlarge.)

Furthermore, pair-wise metabolite-to-metabolite correlation analysis and construction of metabolic networks have been used and reveal novel correlations and alteration even if no apparent difference was seen in average metabolite levels of transgenic potato plants suppressing a sucrose synthase gene (Weckwerth et al., 2004). Metabolic alterations could be assigned to carbohydrate and amino acid metabolism. This indicates metabolic network based on correlation analysis is also a potential statistical tool for unravel the underlying regulatory metabolic pathway.

Recently, the improvement of coverage and accuracy of metabolite measurements have been reported. In the study of Kusano et al. (2007), two-dimensional GC x GC-MS was used to enhance the GC resolution with great advantages in increasing the resolution and peak capacity over the one dimensional separation method. The combination of 1D GC-MS and GC x GC-MS was applied to the metabolic phenotyping of 68 different natural variants of rice seeds from the NIAS Global Rice Core Collection and two additional varieties Dahongg and Pokkali. 1D-GC-MS analysis revealed approximately 430 MS peaks, while the GC x GC-MS analysis enabled the detection of approximately 620 peaks with higher intensity, indicating higher resolution ability of the later technique compared to the former one. Based on multivariate statistical analysis (principal component analysis and the partial least square-discriminate analysis) using 1D-GC-MS data, three varieties showing good separation from the others were chosen for further analysis by GC x GC-MS. Statistical comparison of ten representative metabolites obtained using two techniques showed relatively similar tendency. The authors concluded that the combination of 1D- and GC x GC-MS renders the profiling data more accurate than the 1D method alone, and is useful for the metabolic phenotyping of natural variants in rice for further studies in breeding programs.

This proof of concept of metabolic profiling as a diagnostic tool suggests that it should also be valuable for the comparison of genetically modified food with unmodified ones. A study by Noteborn et al. (2000) suggested that non-modified and genetically modified food are indistinguishable by metabolic profiling using a combination of ¹H-NMR and liquid chromatography (LC-NMR). They took different genetically modified tomato plants and compared the chemical fingerprints obtained by analysis of these genotypes. The fingerprints included profiles of lipids, fatty acids, carotenoids, indole-like compounds, glycoalcohols, aromatic and aliphatic compounds, sugars, fruit acids, and Krebs cycle intermediates. With this protocol they were able to detect 3000 peaks and found that transgenic tomatoes exhibiting antisense expression of exogalactinose displayed significantly different amplitudes than the unmodified control at approximately 250 of these peaks. However, analyses of more commercial varieties indicate that many of these differences fall within values already seen in cultivated tomato varieties. The authors conclude that such methods could be used in routine screening of genetically modified foodstuffs in which there is an indication from other traits that there may be unintended effects of the modification. These methods could be important to test whether genetically modified plants have any nutritional deficiency, or if they are hazardous from a toxicological standpoint. Negative results would help to allay fears over the use of genetic engineering technologies.

From Diagnostics to System Biology

Several recent studies have illustrated the utility of combining data from metabolomics with that from other genomics platforms to provide new insights on both gene annotation (Goossens et al., 2003; Achnine et al., 2005; Fridman et al., 2005; Hagel et al., 2008) and regulation in complex biological systems (Urbanczyk-Wochniak et al., 2003; Hirai et al., 2004; Alba et al., 2005). These approaches have resulted in the identification of numerous candidate genes including several in which expression correlated strongly with the levels of metabolites with important nutritional or organoleptic (taste) properties. In this vein, metabolite:transcript correlations from large data sets collected throughout development in wild-type lines and transgenic tubers engineered to have enhanced sucrose metabolism, allows the identification of candidate genes for biotechnology (Urbanczyk-Wochniak et al., 2003). In this study, the transcript levels of approximately 280 transcripts that showed reproducible changes with respect to control samples were compared to changes in metabolite levels in paired samples. A total of 517 out of the 26,616 possible pairs showed significant correlation (at the P < 0.01 level). Although some of these relationships were already known, most were new and contained several strong relations between transcript levels and nutritionally important metabolites. The studies using combination of metabolomic and expression profiling (Keurentjes et al., 2007; West et al., 2007) on two Arabidopsis populations, the recombinant inbred line (RIL) resulting from a Bayreuth-0 (Bay) x Shahdara (Sha) cross and Landsberg erecta (Ler) x Cape Verde Islands (Cvi) cross, and also analysis by enzymatic profiling for Ler x Cvi (Keurentjes et al., 2008), reveled the full complexity of interaction across the various levels of cellular organization and, thus, the full scale of the challenge of engineering plants by targeted methods. Both analyses indicated that, although natural variation in transcripts can significantly impact phenotypic variation, the natural variation in primary metabolism or the enzymatic loci that correspond to them can feedback to affect the transcripts (Wentzell et al., 2007). The use of metabolomics to assign gene function has also been carried out in several other studies with profiling being used in conjunction with knockout mutagenesis to identify the specific function of PAL1 and PAL2 genes of phenylpropanoid metabolism (Rohde et al., 2004), the Myb-like transcription factor, PAP1 (Tohge et al., 2005), the flavonol arabinosyltransferase, UGT78D3, and UDP-rhamnose, RHM, in flavonoid pathway (Yonekura-Sakakibara et al., 2008), and the UDP-glucose pyrophosphorylase3, UGP3, in sulfolipid synthesis (Okazaki et al., 2009).

Of all genomics tools, metabolite profiling offers arguably the best combinations of practical performance and cost per sample. The expression of almost every gene from the yeast and E. coli genomes in Arabidopsis and subsequent metabolite profiling with GC-MS and LC-MS/MS has recently been achieved (Fernie et al., 2004; Figure 2). This approach was deliberately nonbiased with respect to both the choice of genes and metabolites measured, because of twin objectives; that is, to explore gene function at the level of protein activity and to explore the consequences of introducing new proteins into the metabolic network of Arabidopsis.

Figure 2 Overexpression and metabolic profiling at the transgenomic level. An example of a heat map of the metabolite profiles of the leaves of around 19,000 mature plants, including plant lines that each overexpresses essentially every gene of the yeast genome. Most of this map is white, which reflects the fact that overexpression does not result in a change in metabolite content compared with control plants. Regions of red and blue indicate that the metabolite content is either increased or decreased, respectively, following overexpression. The color scale is nonlinear and the maximum increases and decreases detected are around one hundredfold. A total of 158 metabolite peaks that have been derived from GC-MS and LC-MS analyses are shown; the chemical identity is known for around 60% of them. The chemical classes covered include amino acids, organic acids, sugars, sugar alcohols, vitamins, and pigments. Although the individual metabolite columns can be visually distinguished, the pixel resolution of the image is not sufficient to distinguish the rows (which represent the plants and plant lines). The software that is used to generate the images uses smoothening software algorithms to circumvent this limitation. Such datasets provide a rich resource for the identification of novel gene-function relationships and provide a foundation for systems-biology approaches. (Reprinted with permission from Fernie et al. 2004). (Click image to enlarge.)

The approach of focussing on individual genes can be easily extended to explore the phenotypic relevance of genome regions. Recently, GC-MS profiling of breeding populations of wild tomato species on leaf and mature fruit were analysed (Schauer et al., 2005). Changes in metabolite content were identified in these species that are potentially important with respect to both stress responses and nutritional benefits. Subsequently metabolites profiling was applied to the identification of pericarp metabolite quantitative trait loci (QTL) (Schauer et al., 2006), using a set of well characterized tomato introgression lines of the wild species S. pennellii (ILs) (Eshed and Zamir, 1995). This study resulted in the identification of 889 QTL that were stable over two independent harvests and governed the accumulation of 74 metabolites, including important primary metabolites, such as sugars, organic acids, essential amino acids, intermediate metabolites and vitamins. Furthermore, many of the metabolites for which QTL were detected are important nutritionally, agronomically or organoleptically; therefore, the incorporation of genetic material from wild species represents an attractive alternative to transgenic approaches for crop improvement. To follow up, heritability of these traits was investigated by analyzing an additional year’s harvest and evaluating the metabolite profiles of lines heterozygous for the introgression (ILHs) (Schauer et al., 2008). These studies revealed that most of the metabolic QTL (174 of 332) were dominantly inherited, with relatively high proportions of additively (61 of 332) or recessively (80 of 332) inherited QTL and a negligible number of QTL displaying overdominant inheritance. Interestingly, the mode of inheritance was quantitatively different between diverse classes of compounds (i.e. between sugars and acids), whilst several metabolite pairs belonging to the same pathway displayed a similar mode of inheritance at the same chromosomal loci. However, evaluation of the association between morphological and metabolic traits in the ILHs revealed that this correlation was far less prominent than in the ILs. Thus, the possibility of uncoupling enhanced metabolite content from any penalties with respect to plant performance and fecundity and redevelopment of hybrid genetic material could prove an important advance in the use of genomics-driven breeding approaches. Stress responses in plants in their own right are also starting to be evaluated by multiple genomics tools with recent studies revealing important information on responses to sulfur starvation, nitrogen, and low temperature stresses (Saito et al., 2006).

To recapitulate, the integration of metabolite profiling with other genomics tools is starting to prove very effective in gene-functional annotation and identification of candidate genes for biotechnology and /or breeding strategies.

Conclusions and Perspectives

Current technologies employed for metabolic profiling clearly are showing remarkable value for comprehensive phenotyping strategies and within functional genomics strategies. In addition, they will most probably have an important role to play in safety-testing of genetically modified foodstuffs and in metabolomics-assisted breeding. While the exact number of chemical compounds found in plants is unknown, it has been estimated to be between 90,000 and 200,000. A single species such as Arabidopsis has in excess of 5000 different compounds. A large proportion of this enormous diversity results from compounds of secondary metabolism. Although current protocols do not cover the full complement of the plant cell, improvement in the coverage of metabolomics techniques has been made. With the combination of next-generation technologies and isotope labeling applied in metabolite profiling, it would appear likely that a multitude of important information concerning the regulation of primary metabolism will become accessible (Tohge and Fernie, 2009). The further development and combination of many analytical techniques will additionally allow a fuller description of the metabolome status of a plant. When this is achieved, global analyses of RNA, protein, and metabolites will allow us to obtain a full picture of the complexity of the system under study.

References

Achnine, L., Huhman, D. V., Farag, M. A., Sumner, L. W., Blount, J. W., and Dixon, R. A. (2005) Genomics-based selection and functional characterization of triterpene glycosyltransferases from the model legume Medicago truncatula. Plant J. 41: 875-887

Alba, R., Payton, P., Fei, Z., McQuinn, R., Debbie, P., Martin, G. B., Tanksley, S. D., and Giovannoni, J. J. (2005) Transcriptome and selected metabolite analyses reveal multiple points of ethylene control during tomato fruit development. Plant Cell 17: 2954-2965

Dunn, W. B., Bailey, N. J., and Johnson, H. E. (2005) Measuring the metabolome: current analytical technologies. Analyst. 130: 606-625

Eisenreich, W. and Bacher, A. (2007) Advances of high-resolution NMR techniques in the structural and metabolic analysis of plant biochemistry. Phytochemistry 68: 2799-2815

Eshed, Y. and Zamir, D. (1995) An introgression line population of Lycopersicon pennellii in the cultivated tomato enables the identification and fine mapping of yield-associated QTL. Genetics 141: 1147-1162

Fernie, A. R. and Schauer, N. (2009) Metabolomics-assisted breeding: a viable option for crop improvement? Trends Genet. 25: 39-48

Fernie, A. R., Trethewey, R. N., Krotzky, A. J., and Willmitzer, L. (2004) Metabolite profiling: from diagnostics to systems biology. Nat. Rev. Mol. Cell Biol. 5: 763-769

Fiehn, O., Kopka, J., Dormann, P., Altmann, T., Trethewey, R. N., and Willmitzer, L. (2000) Metabolite profiling for plant functional genomics. Nat. Biotechnol. 18: 1157-1161

Fridman, E., Wang, J., Iijima, Y., Froehlich, J. E., Gang, D. R., Ohlrogge, J., and Pichersky, E. (2005) Metabolic, genomic, and biochemical analyses of glandular trichomes from the wild tomato species Lycopersicon hirsutum identify a key enzyme in the biosynthesis of methylketones. Plant Cell 17: 1252-1267

Goossens, A., Hakkinen, S. T., Laakso, I., Seppanen-Laakso, T., Biondi, S., De Sutter, V., Lammertyn, F., Nuutila, A. M., Soderlund, H., Zabeau, M., Inze, D., and Oksman-Caldentey, K. M. (2003) A functional genomics approach toward the understanding of secondary metabolism in plant cells. Proc. Natl. Acad. Sci. U S A 100: 8595-8600

Hagel, J. M., Weljie, A. M., Vogel, H. J., and Facchini, P. J. (2008) Quantitative 1H nuclear magnetic resonance metabolite profiling as a functional genomics platform to investigate alkaloid biosynthesis in opium poppy. Plant Physiol. 147: 1805-1821

Hirai, M. Y., Yano, M., Goodenowe, D. B., Kanaya, S., Kimura, T., Awazuhara, M., Ari/ta, M., Fujiwara, T., and Saito, K. (2004) Integration of transcriptomics and metabolomics for understanding of global responses to nutritional stresses in Arabidopsis thaliana. Proc. Natl. Acad. Sci. U S A 101: 10205-10210

Iijima, Y., Nakamura, Y., Ogata, Y., Tanaka, K., Sakurai, N., Suda, K., Suzuki, T., Suzuki, H., Okazaki, K., Kitayama, M., Kanaya, S., Aoki, K., and Shibata, D. (2008) Metabolite annotations based on the integration of mass spectral information. Plant J. 54: 949-962

Keurentjes, J. J., Sulpice, R., Gibon, Y., Steinhauser, M. C., Fu, J., Koornneef, M., Stitt, M., and Vreugdenhil, D. (2008) Integrative analyses of genetic variation in enzyme activities of primary carbohydrate metabolism reveal distinct modes of regulation in Arabidopsis thaliana. Genome Biol. 9: R129

Keurentjes, J. J., Fu, J., Terpstra, I. R., Garcia, J. M., van den Ackerveken, G., Snoek, L. B., Peeters, A. J., Vreugdenhil, D., Koornneef, M., and Jansen, R. C. (2007) Regulatory network construction in Arabidopsis by using genome-wide gene expression quantitative trait loci. Proc. Natl. Acad. Sci. U S A 104: 1708-1713

Kusano, M., Fukushima, A., Kobayashi, M., Hayashi, N., Jonsson, P., Moritz, T., Ebana, K., and Saito, K. (2007) Application of a metabolomic method combining one-dimensional and two-dimensional gas chromatography-time-of-flight/mass spectrometry to metabolic phenotyping of natural variants in rice. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 855: 71-79

Mintz-Oron, S., Mandel, T., Rogachev, I., Feldberg, L., Lotan, O., Yativ, M., Wang, Z., Jetter, R., Venger, I., Adato, A., and Aharoni, A. (2008) Gene expression and metabolism in tomato fruit surface tissues. Plant Physiol. 147: 823-851

Moco, S., Capanoglu, E., Tikunov, Y., Bino, R. J., Boyacioglu, D., Hall, R. D., Vervoort, J., and De Vos, R. C. (2007) Tissue specialization at the metabolite level is perceived during the development of tomato fruit. J. Exp. Bot. 58: 4131-4146

Noteborn, H. P., Lommen, A., van der Jagt, R. C., and Weseman, J. M. (2000) Chemical fingerprinting for the evaluation of unintended secondary metabolic changes in transgenic food crops. J. Biotechnol. 77: 103-114

Okazaki, Y., Shimojima, M., Sawada, Y., Toyooka, K., Narisawa, T., Mochida, K., Tanaka, H., Matsuda, F., Hirai, A., Hirai, M. Y., Ohta, H., and Saito, K. (2009) A chloroplastic UDP-Glucose Pyrophosphorylase from Arabidopsis is the committed enzyme for the first step of sulfolipid biosynthesis. Plant Cell 21(3):892-909

Roessner, U., Luedemann, A., Brust, D., Fiehn, O., Linke, T., Willmitzer, L., and Fernie, A. (2001) Metabolic profiling allows comprehensive phenotyping of genetically or environmentally modified plant systems. Plant Cell 13: 11-29

Roessner, U., Wagner, C., Kopka, J., Trethewey, R. N., and Willmitzer, L. (2000) Simultaneous analysis of metabolites in potato tuber by gas chromatography-mass spectrometry. Plant J. 23(1):131-142

Rohde, A., Morreel, K., Ralph, J., Goeminne, G., Hostyn, V., De Rycke, R., Kushnir, S., Van Doorsselaere, J., Joseleau, J. P., Vuylsteke, M., Van Driessche, G., Van Beeumen, J., Messens, E., and Boerjan, W. (2004) Molecular phenotyping of the pal1 and pal2 mutants of Arabidopsis thaliana reveals far-reaching consequences on phenylpropanoid, amino acid, and carbohydrate metabolism. Plant Cell 16: 2749-2771

Saito, K., Dixon, R. A., and Willmitzer, L. (eds.) (2006) Plant Metabolomics, Springer, Heidelberg, Germany

Schauer, N., Semel, Y., Balbo, I., Steinfath, M., Repsilber, D., Selbig, J., Pleban, T., Zamir, D., and Fernie, A. R. (2008) Mode of inheritance of primary metabolic traits in tomato. Plant Cell 20: 509-523

Schauer, N., Semel, Y., Roessner, U., Gur, A., Balbo, I., Carrari, F., Pleban, T., Perez-Melis, A., Bruedigam, C., Kopka, J., Willmitzer, L., Zamir, D., and Fernie, A. R. (2006) Comprehensive metabolic profiling and phenotyping of interspecific introgression lines for tomato improvement. Nat. Biotechnol. 24: 447-454

Schauer, N. and Fernie, A. R. (2006). Plant metabolomics: towards biological function and mechanism. Trends Plant Sci. 11: 508-16

Schauer, N., Zamir, D., and Fernie, A. R. (2005) Metabolic profiling of leaves and fruit of wild species tomato: a survey of the Solanum lycopersicum complex. J. Exp. Bot. 56: 297-307

Tohge, T. and Fernie, A. R. (2009) Web-based resources for mass-spectrometry-based metabolomics: A user’s guide. Phytochemistry 70: 450-456

Tohge, T., Nishiyama, Y., Hirai, M. Y., Yano, M., Nakajima, J., Awazuhara, M., Inoue, E., Takahashi, H., Goodenowe, D. B., Kitayama, M., Noji, M., Yamazaki, M., and Saito, K. (2005) Functional genomics by integrated analysis of metabolome and transcriptome of Arabidopsis plants over-expressing an MYB transcription factor. Plant J. 42: 218-235

Urbanczyk-Wochniak, E., Luedemann, A., Kopka, J., Selbig, J., Roessner-Tunali, U., Willmitzer, L., and Fernie, A. R. (2003) Parallel analysis of transcript and metabolic profiles: a new approach in systems biology. EMBO Rep. 4: 989-993

Weckwerth, W., Loureiro, M. E., Wenzel, K., and Fiehn, O. (2004) Differential metabolic networks unravel the effects of silent plant phenotypes. Proc. Natl. Acad. Sci. U S A. 101: 7809-7814

Wentzell, A. M., Rowe, H. C., Hansen, B. G., Ticconi, C., Halkier, B. A., and Kliebenstein, D. J. (2007) Linking metabolic QTLs with network and cis-eQTLs controlling biosynthetic pathways. PLoS Genet. 3: 1687-1701

West, M. A., Kim, K., Kliebenstein, D. J., van Leeuwen, H., Michelmore, R. W., Doerge, R. W., and St Clair, D. A. (2007) Global eQTL mapping reveals the complex genetic architecture of transcript-level variation in Arabidopsis. Genetics 175: 1441-1450

Yonekura-Sakakibara, K., Tohge, T., Matsuda, F., Nakabayashi, R., Takayama, H., Niida, R., Watanabe-Takahashi, A., Inoue, E., and Saito, K. (2008) Comprehensive flavonol profiling and transcriptome coexpression analysis leading to decoding gene-metabolite correlations in Arabidopsis. Plant Cell 20: 2160-2176