JXB Advance Access originally published online on June 18, 2004
Journal of Experimental Botany 2004 55(404):1861-1870; doi:10.1093/jxb/erh177
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
RESEARCH PAPER |
Towards dissecting nutrient metabolism in plants: a systems biology case study on sulphur metabolism
1Max-Planck-Institut für Molekulare Pflanzenphysiologie, Department of Molecular Physiology, Am Mühlenberg 1, D-14476 Golm, Germany
2Timiryazev Institute of Plant Physiology, Russian Academy of Sciences, Botanicheskaya Str. 35, Moscow 127276, Russia
* To whom correspondence should be addressed. Fax: +49 331 5678 9 8205. E-mail: hoefgen{at}mpimp-golm.mpg.de
Received 28 January 2004; Accepted 26 March 2004
 |
Abstract |
|---|
Top
Abstract
IntroductionA genomics analysis on sulphur metabolism has been conducted at the level of transcriptomics and metabolomics. The analysis of these data after applying bioinformatic tools is to reveal novel findings. The findings are discussed and the knowledge obtained from comparable analyses on sulphur metabolism and other plant nutrient genomic studies is reviewed. The analysis of the response of the transcriptome and metabolome to sulphur deprivation in the growth medium provides a tool set for the analysis of comparable genomics studies of other nutrients. The goal of this ‘sulphobolomics’ (i.e. sulphur genomics and metabolome analysis) approach, and of other investigations, is to describe in a holistic way the biochemical, molecular, and physiological response of a plant to nutrient starvation, here sulphate, or, more generally, to alterations and imbalances in nutrient availability. Eventually, this analysis will provide a case study for a systems biology approach.
Key words: Bioinformatics, metabolomics, networks, nutrient, sulphate, sulphur metabolism, systems biology, transcriptomics
|
Introduction |
|---|
Immobile vascular land plants need to adapt to environmental conditions such as CO2 uptake from the atmosphere and uptake of water and nutrient minerals from the pedosphere. Nutrient availability is determined by the mineral composition resident in the soil, its accessibility, and the water supply conditions such as capillary, surface or groundwater carrying ions. Efficient constitutive and inducible uptake systems, i.e. ion transporters which are often specific for certain nutrients, the ability of explorative root growth, and mineral solubilization, for example, by citrate extrusion, allow uptake even against concentration gradients (Grossmann and Takahashi, 2001
). These mechanisms are then coupled with internal transport, nutrient ion assimilation, storage, and remobilization to provide a stable interior for the biosynthesis of organic molecules. Biotic factors might well contribute to this scenario, for example, symbiosis with nitrogen-fixing bacteria allows growth on low-nitrogen-containing soils or fungal mycorrhiza provide their partners with minerals, both in exchange for carbohydrates abundantly produced in the process of photosynthesis. Plants have developed a vast repertoire of physiologically adaptive mechanisms to exploit these resources optimally.
Plant biochemistry, molecular biology, and physiology have described the linear routes of ion uptake, assimilation, and the biosynthesis of the various biomolecules in great detail, predominantly the formation of the covalently bound ‘C–N’, ‘C–P’, and ‘C–S’ compounds or co-ordinated metal ions such as iron, magnesium, or zinc, the latter often contributing to the biological activity of proteins or their co-factors (for a review see Buchanan et al., 2000). A more holistic approach to understanding the interacting network or biochemical and informational web within plants has become available only recently with the development of multiparallel, highly sensitive, and high throughput techniques (Johnston, 1998; Fiehn et al., 2001; Minorsky, 2003). These approaches are directed towards obtaining profiles of cellular or tissue responses to varying conditions, either developmental or environmental. Such array technologies allow scoring of the transcriptome which is the full RNA complement at a given state, an approach which is termed transcriptomics. Transcripts are further translated to enzymes which exert their respective activities. Thus, the ultimate response of the plant cell is the resulting composition of metabolites as a result of complex biosynthetic processes and regulatory circuits. This pattern describes the homeostatic state of the plant's metabolism. Therefore, the metabolite composition of a cell or organism, the metabolome, is linked to the transcriptome and to the physiological state of the plant.
Transcriptomics and metabolomics, i.e. the analysis of the RNA and metabolite profiles, are therefore suitable for describing the web of plant metabolism and physiology and for describing the systems response of the plant. To complement this matrix in order to describe the system fully and to allow predictions of the systems response, further refinement and experimental data will be necessary. Proteomics, enzyme activity profiling, and flux analyses will contribute substantially to this endeavour. However, the combination of metabolomics and transcriptomics will describe and unravel complex biological system relations leading to a breakthrough in plant systems biology of plant nutrient metabolism.
Genomics approaches usually result in an enormous amount of primary data. Of key importance for a deeper understanding of the ‘data flood’, as provided by these high throughput multiparallel analytical approaches, is the use, adaptation, or development of mathematical and statistical tools starting from the set-up of the experimental design to data evaluation, mining, and model building. Bioinformatics encounters a number of problems inherent for biological systems analysis despite all efforts of standardizing the plant material or the experimental conditions such as inappropriate sample sizes, highly variable, and sometimes confusing data. Dealing with these problems is discussed in this paper. Owing to the high number of unknown and uncontrollable parameters, systems biology approaches must not only integrate biology, biochemistry, molecular biology, and mathematics, but also need to account for ecology, breeding, statistical experimental design, and might even be driven to include knowledge derived from systems analysis in the social sciences and economy.
The first genomic studies have been applied to the major mineral nutrients, such as N, S, P, Fe, Cu, Zc, K, and Ca (Wang R et al., 2000, 2003; Wang Y-H et al., 2001, 2002; Crawford and Forde, 2003; Hammond et al., 2003; Wu et al., 2003; Nikiforova et al., 2003; Hirai et al., 2003; Maruyama-Nakashita et al., 2003; Wintz et al., 2003; Thimm et al., 2001; Buckhout and Thimm, 2003; Negishi et al., 2002; Maathuis et al., 2003). Usually, plants grown under standardized conditions, such as climate-controlled greenhouses or growth chambers, are challenged with alterations in respect of nutrient ion concentrations. The plant responses to these alterations are monitored at certain time points of either depletion of the nutrient or resupply after starvation. As the whole plant system is exposed to this alteration, not only immediate responses have been observed, but also further effects are relayed through the network as a whole. Whether these effects are viewed as an integral part of the response to a given stimulus or as ‘downstream’, ‘secondary’, or ‘pleiotropic’ effects is more a philosophical or semantic question. One might distinguish between ‘sulphur-starvation-induced’ genes and ‘sulphur-response-related’ genes. The fact is that a cascade of responses, as identified at the transcriptome and metabolome level, will lead to a distortion of the biochemical–physiological network from a current to a new, adapted level of homeostasis, unless continued starvation drives the system to destruction. The entire development of the series of responses is viewed as the ‘sulphur-starvation response’.
In this study, a systems analysis of the response of Arabidopsis thaliana to sulphur starvation using transcriptome and metabolome profiling as a case study for attaining a deeper understanding of plant nutrient physiology is described.
|
Sulphur metabolism in plants |
|---|
Sulphur is one of the major plant nutrients amounting to about 10% of the total N content (Haneklaus et al., 2003
; Hell and Rennenberg, 1998). Sulphur metabolism has been characterized in detail at the biochemical and molecular levels (Miflin and Lea, 1990; Bryan, 1990; Schmidt and Jäger, 1992; Matthews, 1999; Leustek et al., 2000; Saito, 1999; Hell et al., 2002; Hesse and Hoefgen, 2003) and has been shown to be intimately interactive with many parts of plant metabolism and physiology (Fig. 1). As an example of general nutrient response in plants, the response of Arabidopsis thaliana to continued deprivation of sulphate in the growth medium is described here. Treated plants have been subjected to a multiparallel, high-throughput analysis on the transcriptome and metabolome level independently in three laboratories (Nikiforova et al., 2003; Hirai et al., 2003; Maruyama-Nakashita et al., 2003). Sulphur is usually taken up by the roots as sulphate through the activity of sulphate transporters (Saito, 2000; Takahashi et al., 2000; Hawkesford, 2000, 2003; Hawkesford and Wray, 2000; Hawkesford et al., 2003; Leustek and Saito, 1999; Grossmann and Takahashi, 2001) and then transported in the xylem stream to the leaves where it is reduced to sulphide. A side branch leads to sulpholipids (Benning, 1998). Eventually, the reduced sulphide is incorporated into activated serine, O-acetylserine, to form cysteine, thus linking glycine-serine metabolism to sulphur metabolism (Bryan, 1990; Hell, 1997; Saito, 2000). Cysteine itself serves as a building stone for all further derived reduced sulphur containing compounds such as methionine, proteins, glutathione, phytochelatines, biotine, thiamine, S-adenosylmethionine (SAM), glucosinolates, and others (Fig. 1). In particular, the methionine, SAM, branch is of high importance as SAM serves as the major methyl-group donor in plant metabolism and is, therefore, involved in numerous biosynthetic activities.
|
Sulphur availability is limited in certain soils and affects plant vigour and crop yield (Haneklaus et al., 1997). Sulphur starvation leads to various phenotypically visible phenotypical alterations (Fig. 2). The first of these is chlorosis of the younger leaves, accumulation of anthocyanins, bending of the leaf blade, and enhanced root growth (Nikiforova et al., 2003
; Lopez-Bucio et al., 2003). These symptoms are also partially shared by other nutrient stresses, such as nitrate (Zhang and Forde, 1998, 2000), or phosphate starvation, which already indicates that distinct stresses might, in part, use common signalling paths with specific response-control elements providing specificity. It is notable that owing to the storage of sulphate in the vacuole and the presence of a pool of thiol compounds, such as glutathione, the development of symptoms is a slow and gradual process. Therefore, it must be assumed that various local, systemic, and time-dependent regulation steps occur in a series of responses. As the major site of sulphate uptake is the root, but the major site of reduction and biosynthetic need is the leaf, it has further to be assumed that the evolution of the signalling mechanisms is in both directions. Thus, plants subjected to sulphur deprivation provide a good model to establish a systematic analysis of nutrient metabolism.
|
|
Plant material and sampling |
|---|
Transcript profiling provides a basis to describe one level of the complex responses of a plant to developmental or environmental changes (Johnston, 1998
; Schena et al., 1998; Desprez et al., 1998; Kehoe et al., 1999; Ohlrogge and Benning, 2000). In order to apply this technology to the investigation of plant sulphur metabolism, Arabidopsis seedlings have been subjected to four time points of sulphur starvation (6 d and 10 d after pregrowth on sulphate sufficient medium for 8 d, termed, consecutively, 6 and 10–1, and 10 d and 13 d when sown directly on a sulphate-depleted medium, termed 10–2 and 13) as described by Nikiforova et al. (2003). Whole seedlings were sampled in five repetitions of pooled plant material just before and shortly after the appearance of the visible symptoms, chlorosis, and growth retardation as determined in pre-experiments. This material was aliquoted and used for the profiling of transcript levels and of metabolite levels. |
Data acquisition: transcriptomics |
|---|
Transcriptome analysis was conducted using a cDNA EST collection of 15 442 clones comprising about 7200 individual genes spotted on nylon filters, and hybridization was done again in five repetitions each, as an aid for statistical accuracy (Nikiforova et al., 2003
). The data were used to calculate ratios of expression levels comparing steady-state transcript amounts under sulphur-deficient and sulphur-sufficient growth conditions. This resulted in a catalogue of the expression profile exhibiting statistically significant ratios of inductions or reductions of gene expression. Scatter-plot analysis proved the technical reliability of the experiment as shown by the perfect diagonal shape of the dot cloud of all experimental points (15 442) as shown, for example, in Fig. 3 for Arabidopsis seedlings after 10 d starvation on agarose medium with 8% residual sulphate. Residual sulphate originates from the iron-EDTA micronutrients. Although single outliers might be identified here already, the application of strict statistical selection criteria (with P<0.05) results in the identification of 2031 (13.5%) differentially expressed clones in this example. When thresholds of 2.5-fold for over-expression and 0.4-fold for under-expression were applied, 505 ESTs remain altered after applying these two consecutive selection criteria, of which 17% are up-regulated and 8% are down-regulated. Most surprising is the still high number of changes which corresponded with 3.3% of the tested gene complement, even after applying a strict selection parameter. When estimating this on the assumed number of 25 000 Arabidopsis genes, it corresponded with about 825 genes in total which are responsive to sulphur deprivation as the primary stimulus. The distribution between over-expression and reduction of gene expression (under-expression), though, is different for the various durations of starvation (see Fig. 5 below following paragraph). Thus, despite no visible alterations at this experimental point (cf. Nikiforova et al., 2003), an unexpectedly high number of genes show altered expression levels. Therefore, simple cataloguing to comprehend this response fully is inadequate. Comparable results have been obtained by Hirai et al. (2003), by applying 48 h sulphur depletion, and Maruyama-Nakashita et al. (2003), by analysing plant material starved for 24 h continuously for sulphate, both using A. thaliana.
|
|
Sorting the responding genes (ratio >2 or <0.5 for transcription factors and >2.5 or <0.4 for all the others) of all four experimental points into functional categories (MIPS Arabidopsis thaliana data base; MATDB) genes of various functional classes are pulled away from equilibrium in both up- and down-regulated genes; amongst them about 35% genes of hitherto unknown function. In summary, 5548 clones show significant alterations, 60% up and 40% down. The responses in the various categories, depicted as ratio up-to-down, show characteristic patterns for the four experimental time points within the classes (Fig. 4) or also in summary (Fig. 5). As a general conclusion, it can be deduced that, after 13 d starvation, genes of all functional categories showed a tendency for down-regulation. At the first three data points most categories show a tendency for over-expression, however, the greater number of genes in the classes ‘transcription’ and ‘energy’ are more down-regulated.
|
As the causal relationships are not immediately evident, various approaches can be applied to deduce functional relationships. Overlaying the results to known pathway architecture, for example, yielded the relationship between the sulphur-serine metabolism and the tryptophan–glucosinolate–auxin metabolism complex (Nikiforova et al., 2003
; Kutz et al., 2002). This provided a potential link and explanation of one of the major phenotypical symptoms: root growth as triggered by auxin increase. A specific nitrilase III, only induced under sulphur-limiting conditions, might provide specificity in this general stress-response pathway. This is further supported by the finding of nine auxin-induced genes among the differentially responding genes (Nikiforova et al., 2003). |
Data acquisition: metabolomics |
|---|
Transcript data, however, are only one manifestation of the response. Transcripts need to be converted to proteins to affect metabolism. The activity changes in the resulting enzymes eventually alter the metabolic state, adapting plant metabolism to the environmental challenge, here of reduced sulphur availability. Unlike transcriptomics, based on the analysis of a uniform molecule class, i.e. RNA, metabolite profiling obviously requires an arsenal of analytical tools owing to the diverse structures, properties, and sizes representing the plant's chemical composition, the metabolome (Fiehn et al., 2000
, 2001; Sumner et al., 2003; Stitt and Fernie, 2003). Experimental tools in use are element analysis via ICP-AES, ion analysis via HPLC or CE, specific HPLC analyses as, for example, for amino acids and thiols, and highly random, high-throughput approaches based mainly on mass spectrometry combined with various prior separation tools such as GC–MS, GC–TOF, LC–MS (Roessner et al., 2001; Fiehn, 2002; Fiehn and Weckwerth, 2003; Wagner et al., 2003), or others such as NMR techniques (Ward et al., 2003; Ott et al., 2003; Defernez and Colquhoun, 2003; Le Gall et al., 2003). Therefore, it is possible to catalogue, qualitatively and quantitatively, several hundred to a thousand metabolites in a given tissue. A high number of detectable peaks in chromatograms is not yet identified, in the same way that not all potential ORFs of the transcriptome are annotated. A further problem for general and broad application, especially of GC technologies, is the ‘matrix dependency’ of the sample analysis. Compounds in different tissues of a given species, or when compared in different species, show slight shifts in their peak positions, thus providing problems for automated analysis. In addition, there are problems with calibration and quantification for certain metabolites. Taken together, metabolomics has reached a resolution where it can serve to provide a system-oriented characterization of gene function, and cellular and physiological responses, although it still needs further improvement (Stitt and Fernie, 2003).
The metabolome of sulphur-starved plants of the above-mentioned experimental set of sulphur-deprived Arabidopsis thaliana seedlings, including shorter starvation times of 3, 6, and 48 h (unpublished data) was analysed. 315 peak-forming derivatized metabolites were detected using gas chromatography–mass spectrometry (GC–MS) analysis (Roessner et al., 2001), among which 110 derivatives of 82 non-redundant chemical compounds were identified.
A summary sketch depicts the response development of the transcript profile and the metabolite profile, distinguished for up- and down-regulated genes and metabolites for 2, 6, 10, and 13 d starvation (Fig. 5; for details of the experimental set up, see Nikiforova et al., 2003). Data points are sorted such that 1, 2, and 3 are plants, respectively, with seedlings not showing any macroscopically visible phenotype due to starvation, while data points 4 and 5 are derived from seedlings where extended starvation of sulphate led to chlorosis, anthocyanin accumulation, and growth reduction compared with the controls. The graph serves to illustrate that the response of the transcriptome depicted as percentages of altered gene expression is dynamic with few changes after 2 d starvation (data point 1, Fig. 5) and increasing response with a maximum at data point 3 (17% up-regulated, 8% of all genes down-regulated). With the increasing accumulation of phenotypical alterations, the number of up-regulated genes is reduced, while finally (data point 5) a deregulation of gene expression can be observed with about 20% of all genes down-regulated in expression and only 7% still up-regulated. Despite the different experimental background of the single data points (cf. text of Fig. 5) indicative trends, such as a general breakdown of transcription upon prolonged sulphate deprivation, can be deduced which will be exploited in future experimentation.
At the metabolite level the amount of altered metabolites at 2 d of starvation (data point 1, Fig. 5) is also low as is the case for the transcripts. However, the up-regulated metabolite contents reach a relatively constant level of about 35% quickly, while reduced metabolite concentrations stay constant over all time points at about 10%. The metabolite profile does not mirror the transcript profile and the observed breakdown of transcription. It was speculated that at still later time points this would also be the case. The timing of both events is of interest. While transcript response develops gradually, metabolites quickly reach a new, stable status. This might indicate that small alterations at the transcript level early in starvation-response development trigger a shift in the metabolite composition moving the plant homeostasis at the status of sulphur sufficiency to a new homeostatic state under sulphur deficiency which is being kept constant despite enduring starvation. Time-dependent alterations of transcript and metabolite levels are a result of continuous gene action for maintaining viability. To produce such a response via the simultaneous action of a large number of elements, a system needs to be highly co-ordinated, and elemental behaviour should be coherent rather than complex (Kitano, 2002).
Taking a subset of metabolites, for example, elements from sulphur metabolism, the aspartate family and linked pathways such as glutathione and putrescine/polyamine, the response development can be visualized by plotting the metabolite concentration versus time of starvation, here represented as a grid (Fig. 6). Metabolites directly dependent on reduced sulphur supply, such as cysteine and glutathione, and, unexpectedly, lysine are decreasing; others increase in amount (OAS and serine as precursors of cysteine, tryptophan, and putrescine due to the decrease of SAM (not shown here)). At the same time, metabolites of the aspartate family remain mainly unaffected (aspartate, threonine, isoleucine, and even methionine). Methionine seems to be kept constant despite the starvation of sulphur as an indispensable part of the SAM methylation cycle. It has been shown previously that at least two of the enzymes involved in the SAM recycling to methionine are induced under sulphur starvation, adenosylhomocysteinase, and SAM synthetase (Nikiforova et al., 2003). For the network construction and the analysis of the systems response these data indicate the reliability of the metabolite profiling data obtained from GC–MS, LC–MS, HPLC and ICP–AES analysis integrated to a joint matrix actually visualizing the cluster of co-regulated metabolites.
|
|
Data interpretation: biochemistry and bioinformatics |
|---|
Analysis of genomic data is still not a routine task. Many studies are confined to the identification of the most highly induced or repressed transcripts/metabolites. However, this strikingly neglects the wealth of information inherently contained in a full-profile analysis. Such a confined analysis will not help to unravel the response of a whole system, or to describe the complex network. It does not yet allow elucidation beyond that achieved so marvellously through biochemical and molecular analysis in the past decades. A step forward was to overlay the transcriptome and metabolome results on the known biochemical pathways in a biased, knowledge-driven approach. This enables research to focus on certain pathways and interactions of pathways (Nikiforova et al., 2003
; Buckhout and Thimm; 2003), a task that can be achieved, for example, using web tools such as AraCyc (http://www.arabidopsis.org/tools/aracyc/) or Mapman (Thimm et al., 2004) which visualize the results on biochemical pathways.
Bioinformatic tools allow biologists to move beyond cataloguing and simple linear interpretations to increasing the understanding of how network components interact (Fiehn et al., 2000, 2001; Girke et al., 2000, 2003; Jasny and Ray, 2003; Bray, 2003; Alon, 2003; Stitt and Fernie, 2003). Statistical tools are available, or being established, to exploit, extract, and mine raw data to perform correlation analyses and deduce matrices and networks. Furthermore, as both data sets rely on ratios between an experimental and a control state, it is possible to fuse metabolome and transcriptome databases. Combined analyses have been performed, however, mainly on a few metabolites to transcriptome data or on pairwise correlations (Askenazi et al., 2003; Urbanczyk-Wochniak et al., 2003). The analytical web-based tool that was applied to analyse sulphur-nutrient deprivation, which combines different statistical methods to analyse transcript and metabolic data, is MetaGeneAlyse (http://metagenealyse.mpimp-golm.mpg.de; Daub et al., 2003).
A huge number of changes could be observed, both on the transcript and the metabolite level, a number of them apparently ‘unrelated’ to sulphur metabolism as far as current knowledge stands (Nikiforova et al., 2003; Hirai et al., 2003; Maruyama-Nakashita et al., 2003). Yet, these effects, which used to be termed ‘pleiotropic’ describing the fact that they are observed, but their relationship to the original cause of the physiological response is unknown, also belong to the response development of sulphur deprivation in plants. As sulphur is integrated into a densely woven biochemical and regulatory network (Fig. 1), effects on what has been viewed previously as unrelated pathways are a logical consequence of network properties of biological systems.
In conclusion, an attempt was made to reconstruct a ‘hypo-sulphur’ response network, where similarities in the patterns of element behaviour were regarded as a measure of their coherence (Kitano, 2002). Simply put: the more two patterns are alike over time, the closer the correlation and their positioning in the response network. The way forward will be governed by the following notions. In this analysis of sulphur-nutrient deprivation, linear- and non-linear-response dependencies were studied. This can be expected for all types of nutrient-metabolic responses. Therefore, data analysis has to account for two data sets using on the one hand Pearson correlation analysis which is suited for linear-distributed data, and with Mutual Information (MI) content analysis, suited for non-linear correlations as well (Daub et al., 2003). From this the network features and elements may be deduced (Bray, 2003). Assuming such a network does not mirror biochemical pathways per se any longer (though it might in part), but rather describes families of co-behaving (coherent) transcripts and/or metabolites (vertices or nodes) and their correlation via connecting lines (edges) (Jasny and Ray, 2003). Typically, biological networks are expected to display inhomogeneous connectivity patterns distinct from a random network (Bray, 2003) with elements of highest connectivity (hubs), while other elements remain lowly connected. These hubs will be points of high interest for further investigations and often do not appear among the usually selected ‘top-ten’ responders. This will allow the deduction of functional relations from the network. Further, this approach can easily be applied to other nutrient and environmental stresses challenging plant adaptivity, or also to investigations of plant developmental programmes.
|
Conclusions for systems biology |
|---|
Plants as living beings are highly complex systems and cannot be described in full detail yet. Usually, cause-and-effect relationships, biochemical-synthesis pathways, developmental programmes, or responses to environmental conditions are described in a way consisting of linear relations or integration of few branches. Inherently, the complexity of biological systems which is due to the multitude of input, intermediate, and output signal steps, leads to a high degree of variations and ‘noise’, when data are collected and analysed. This, in turn, is also a prerequisite for the flexibility of plant metabolism.
Development of various high-throughput analytical procedures now provides an ever-exploding amount of data and information (Fiehn et al., 2001; Kitano, 2002). The amount, the variability of the data, and the incomparability of experimental conditions provides a challenge for analytical procedures and the data analysis using bioinformatics (Katagiri, 2003). It is to be expected that the body of cumulating data will give rise to a better understanding of biological systems as a whole and will allow the interpretation and forecast of the responses and manifestations of biological systems (Sweetlove et al., 2003; Alon, 2003; Minorsky, 2003; Bray, 2003). The goal, eventually, will be to describe the wiring scheme of metabolic and physiological processes in plants (Chong and Ray, 2002; Quackenbush, 2003) or even cross-species (Stuart et al., 2003). Approaches to the analysis of the response of Arabidopsis thaliana to nutrient starvation were described using sulphate as a model ion to provoke the systems response (Nikiforova et al., 2003; Hirai et al., 2003; Maruyama-Nakashita et al., 2003). This study, as presented here, provides clues on how to convert multiparallel analyses and genomics data into a consistent systems description of sulphur metabolism and physiology.
Plants are, however, exposed to multiple external and internal inputs, i.e. environment and development, which, in addition, vary in time and space. The obvious challenge will be to analyse and integrate the influence of these parameters on the plant system as a whole.
Further to this, systems biology might be helpful in paving new ways to biotechnology and application (Hesse and Höfgen, 2001; Fiehn, 2002; Galili and Höfgen, 2002).
|
Acknowledgements |
|---|
We thank Dr Malcolm Hawkesford for providing the sulphur lump in Fig. 1 which was mined in the volcanic region of the massif central in France. It has been emitting sulphurish odours for quite a while on my bookshelf, thus finally forcing me to return it. We thank the Greenteam of the MPI-MP headed by Dr Karin Köhl for taking care of our plants. We thank Josef Bergstein for photographic work. Dr Joachim Selbig and Carsten Daub of the BioInformatics group of the MPI-MP receive our thanks for supporting the mathematical analysis of our experimental data, and Dr Joachim Kopka and Oliver Fiehn for helping with the metabolite analysis to generate these data. The work was enabled by a grant of the European Commission in framework programme 5 (QLRT-2000-00103) and by the Max Planck Society.
|
References |
|---|
Alon U. 2003. Biological networks: the tinkerer as an engineer. Science 301, 1866–1867.
Askenazi M, Driggers EM, Holtzman DA, et al. 2003. Integrating transcriptional and metabolite profiles to direct the engineering of lovastatin-producing fungal strains. Nature Biotechnology 21, 150–156.[CrossRef][ISI][Medline]
Batagelj V, Mrvar A. 2003. Pajek—analysis and visualization of large networks. In: Jünger M, Mutzel P, eds. Graph drawing software. Berlin: Springer, 77–103.
Benning C. 1998. Biosynthesis and function of the sulpholipid sulphoquinovosyl diacylglycerol. Annual Review of Plant Physiology and Plant Molecular Biology 49, 53–75.[Medline]
Bray D. 2003. Molecular networks: the top-down view. Science 301, 1864–1865.
Bryan JK. 1990. Advances in the biochemistry of amino acid biosynthesis. In: Miflin BJ, Lea PJ, eds. The biochemistry of plants: amino acids and derivatives, Vol. 5. New York: Academic Press, 161–196.
Buchanan BB, Gruissem W, Jones RL. 2000. Biochemistry and molecular biology of plants. Rockville, USA Rockville, Md: American Society of Plant Physiologists.
Buckhout TJ, Thimm O. 2003. Insights into metabolism obtained from microarray analysis. Current Opinion in Plant Biology 6, 288–296.[CrossRef][ISI][Medline]
Chong L, Ray LB. 2002. Whole-istic biology. Science 295, 1661.[Abstract]
Crawford NM, Forde BG. 2002. Molecular and developmental biology of inorganic nitrogen nutrition. The Arabidopsis book. American Society of Plant Biologists. www.aspb.org/publications/arabidopsis/toc.cfm
Daub CO, Kloska S, Selbig J. 2003. MetaGeneAlyse: analysis of integrated transcriptional and metabolite data. Bioinformatics 19, 2332–2333.
Defernez M, Colquhoun IJ. 2003. Factors affecting the robustness of metabolite fingerprinting using H-1 NMR spectra. Phytochemistry 62, 1009–1017.[CrossRef][ISI][Medline]
Desprez T, Amselem J, Caboche M, Höfte H et al. 1998. Differential gene expression in Arabidopsis monitored using cDNA arrays. The Plant Journal 14, 643–652.[CrossRef][ISI][Medline]
Fiehn O. 2002. Metabolomics—the link between genotypes and phenotypes. Plant Molecular Biology 48, 155–171.[CrossRef][ISI][Medline]
Fiehn O. 2003. Metabolic networks of Cucurbita maxima phloem. Phytochemistry 62, 875–886.[CrossRef][ISI][Medline]
Fiehn O, Kloska S, Altmann T. 2001. Integrated studies on plant biology using multiparallel techniques. Current Opinion in Biotechnology 12, 82–86.[CrossRef][ISI][Medline]
Fiehn O, Kopka J, Dormann P, Altmann T, Trethewey RN, Willmitzer L. 2000. Metabolite profiling for plant functional genomics. Nature Biotechnology 18, 1157–1161.[CrossRef][ISI][Medline]
Fiehn O, Kopka J, Dormann P, Altmann T, Trethewey RN, Willmitzer L. 2001. Metabolite profiling for plant functional genomics. Nature Biotechnology 19, 173.
Fiehn O, Weckwerth W. 2003. Deciphering metabolic networks. European Journal of Biochemistry 270, 579–588.[ISI][Medline]
Forde B, Lorenzo H. 2001. The nutritional control of root development. Plant and Soil 232, 51–68.[CrossRef][ISI]
Galili G, Höfgen R. 2002. Metabolic engineering of amino acids and storage proteins in plants. Metabolic Engineering 4, 3–11.[CrossRef][ISI][Medline]
Girke T, Ozkan M, Carter D, Raikhel NV. 2003. Towards a modeling infrastructure for studying plant cells. Plant Physiology 132, 410–414.
Girke T, Todd J, Ruuska S, White J, Benning C, Ohlrogge J. 2000. Microarray analysis of developing Arabidopsis seeds. Plant Physiology 124, 1570–1581.
Grossman A, Takahashi H. 2001. Macronutrient utilization by photosynthetic eukaryotes and the fabric of interactions. Annual Review of Plant Physiology and Plant Molecular Biology 52, 163–210.[CrossRef][ISI][Medline]
Hammond JP, Bennett MJ, Bowen HC, Broadley MR, Eastwood DC, May ST, Rahn C, Swarup R, Woolaway KE, White PJ. 2003. Changes in gene expression in Arabidopsis shoots during phosphate starvation and the potential for developing smart plants. Plant Physiology 132, 578–596.
Haneklaus S, Bloem E, Schnug E. 2003. The global sulphur cycle and its links to plant environment. In: Abrol YP, Ahmad A, eds. Sulphur in plants. Dordrecht, The Netherlands: Kluwer Academic Publishers, 1–28.
Hawkesford MJ. 2000. Plant responses to sulphur deficiency and the genetic manipulation of sulphate transporters to improve S-utilization efficiency. Journal of Experimental Botany 51, 131–138.
Hawkesford MJ. 2003. Transporter gene families in plants: the sulphate transporter gene family—redundancy or specialization? Physiologia Plantarum 117, 155–163.[CrossRef]
Hawkesford MJ, Buchner P, Hopkins L, Howarth JR. 2003. The plant sulphate transporter family: specialized functions and integration with whole plant nutrition. In: Davidian J-C, Grill D, De Kok LJ, Stulen I, Hawkesford MJ, Schnug E, Rennenberg H, eds. Sulphur transport and assimilation in plants regulation, interaction and signalling. ?? 1–10.
Hawkesford MJ, Wray JL. 2000. Molecular genetics of sulphate assimilation. In: Callow JA, ed. Advances in botanical research incorporating advances in plant pathology, Vol. 33. London, England: Academic Press Ltd. 159–223.
Hell R. 1997. Molecular physiology of plant sulphur metabolism. Planta 202, 138–148.[CrossRef][ISI][Medline]
Hell R, Jost R, Berkowitz O, Wirtz M. Molecular and biochemical analysis of the enzymes of systeine biosynthesis in the plant Arabidopsis thaliana. Amino Acids 22, 245–257.
Hell R, Rennenberg H. 1998. The plant sulphur cycle. Sulphur in Agroecosystems 2, 135–173.
Hesse H, Höfgen R. 2001. Application of genomics in agriculture. In: Hawkesford MJ, Buchner P, eds. Molecular analysis of plant adapatation to the environment. Dordrecht, The Netherlands: Kluwer AP. 61–79.
Hesse H, Hoefgen R. 2003. Molecular aspects of methionine biosynthesis in Arabidopsis and potato. Trends in Plant Science 8, 259–262.[CrossRef][ISI][Medline]
Hirai YM, Fujiwara T, Awazuhara M, Kimura T, Masaaki N, Saito K. 2003. Global expression profiling of sulphur-starved Arabidopsis by DNA macroarray reveals the role of O-acetyl-L-serine as a general regulator of gene expression in response to sulphur nutrition. The Plant Journal 33, 651–663.[CrossRef][ISI][Medline]
Jasny BR, Ray LB. 2003. Life and the art of networks. Science 301, 1863.[Abstract]
Johnston M. 1998. Gene chips—array of hope for understanding gene regulation. Current Biology 8, 171–174.
Katagiri F. 2003. Attacking complex problems with the power of systems biology. Plant Physiology 132, 417–419.
Kehoe DM, Villand P, Somerville S. 1999. DNA microarrays for studies of higher plants and other photosynthetic organisms. Trends in Plant Science 4, 38–41.[Medline]
Kitano H. 2002. Systems biology: a brief overview. Science 295, 1662–1664.
Kutz A, Muller A, Hennig P, Kaiser WM, Piotrowski M, Weiler EW. 2002. A role for nitrilase 3 in the regulation of root morphology in sulphur-starving Arabidopsis thaliana. The Plant Journal 30, 95–106.[CrossRef][ISI][Medline]
Le Gall G, Colquhoun IJ, Davis AL, Collins GJ, Verhoeyen ME. 2003. Metabolite profiling of tomato (Lycopersicon esculentum) using H-1 NMR spectroscopy as a tool to detect potential unintended effects following a genetic modification. Journal of Agricultural and Food Chemistry 51, 2447–2456.[CrossRef][ISI][Medline]
Leustek T, Martin MN, Bick JA, Davies JP. 2000. Pathways and regulation of sulphur metabolism revealed through molecular and genetic studies. Annual Review of Plant Physiology and Plant Molecular Biology 51, 141–165.[CrossRef][ISI][Medline]
Leustek T, Saito K. 1999. Sulphate transport and assimilation in plants. Plant Physiology 120, 637–643.
López-Bucio J, Cruz-Ramírez A, Herrera-Estrella L. 2003. The role of nutrient availability in regulating root architecture. Current Opinion in Plant Biology 6, 280–287.[CrossRef][ISI][Medline]
Maathuis FJM, Filatov V, Herzyk P, et al. 2003. Transcriptome analysis of root transporters reveals participation of multiple gene families in the response to cation stress. The Plant Journal 35, 675–692.[CrossRef][ISI][Medline]
Maruyama-Nakashita A, Inoue E, Watanabe-Takahashi A, Yamaya T, Takahashi H. 2003. Transcriptome profiling of sulphur-responsive genes in Arabidopsis reveals global effects of sulphur nutrition on multiple metabolic pathways. Plant Physiology 132, 597–605.
Matthews B. 1999. Lysine, threonine and methionine biosynthesis. In: Singh BK, ed. Plant amino acids. New York: Marcel Dekker, Inc., 205–225.
Miflin BJ, Lea PJ. 1990. The biochemistry of plants. Vol. 16. Intermediary nitrogen metabolism. USA: Academic Press.
Minorsky PV. 2003. Achieving the in silico plant. Systems biology and the future of plant biological research. Plant Physiology 132, 404–409.
Negishi T, Nakanishi H, Yazaki J, et al. 2002. cDNA microarray analysis of gene expression during Fe-deficiency stress in barley suggests that polar transport of vesicles is implicated in phytosiderophore secretion in Fe-deficient barley roots. The Plant Journal 30, 83–94.[CrossRef][ISI][Medline]
Nikiforova V, Freitag J, Kempa S, Adamik M, Hesse H, Hoefgen R. 2003. Transcriptome analysis of sulphur depletion in Arabidopsis thaliana: interlacing of biosynthetic pathways provides response specificity. The Plant Journal 33, 633–650.[CrossRef][ISI][Medline]
Ohlrogge J, Benning C. 2000. Unraveling plant metabolism by EST analysis. Current Opinion in Plant Biology 3, 224–228.[ISI][Medline]
Ott K-H, Aranibar N, Singh B, Stockton G. 2003. Metabonomics classifies pathways affected by bioactive compounds. Artificial neural network classification of NMR spectra of plant extracts. Phytochemistry 62, 971–985.[CrossRef][ISI][Medline]
Quackenbush J. 2003. Microarrays—guilt by association. Science 302, 240–241.
Roessner U, Luedemann A, Brust D, Fiehn O, Linke T, Willmitzer L, Fernie AR. 2001. Metabolic profiling allows comprehensive phenotyping of genetically or environmentally modified plant systems. The Plant Cell 13, 11–29.
Saito K. 1999. Biosynthesis of cysteine. In: Singh, BK ed. Plant amino acids: biochemistry and biotechnology. New York, NY: Dekker, 267–291.
Saito K. 2000. Regulation of sulphate transport and synthesis of sulphur-containing amino acids. Current Opinion in Plant Biology 3, 188–195.[ISI][Medline]
Schena M, Heller RA, Theriault TP, Konrad K, Lachenmeier E, Davis RW. 1998. Microarrays: biotechnology's discovery platform for functional genomics. Trends in Biotechnology 16, 301–306.[CrossRef][ISI][Medline]
Schmidt A, Jäger K. 1992. Open questions about sulphur metabolism in plants. Annual Review of Plant Physiology and Plant Molecular Biology 43, 325–349.[CrossRef][ISI]
Stitt M, Fernie AR. 2003. From measurements of metabolites to metabolomics: an ‘on the fly’ perspective illustrated by recent studies of carbon-nitrogen interactions. Current Opinion in Biotechnology 14, 136–144.[CrossRef][ISI][Medline]
Stuart JM, Segal E, Koller D, Kim SK. 2003. A gene-coexpression network for global discovery of conserved genetic modules. Science 302, 249–255.
Sumner LW, Mendes P, Dixon RA. 2003. Plant metabolomics: large-scale phytochemistry in the functional genomics era. Phytochemistry 62, 817–836.[CrossRef][ISI][Medline]
Sweetlove LJ, Last RL, Fernie AR. 2003. Predictive metabolic engineering: a goal for systems biology. Plant Physiology 132, 420–425.