Systems biology is based on the understanding that the behaviour of the whole is greater than would be expected from the sum of its parts. Thus the ultimate goal of systems biology is to predict the behaviour of the whole system on the basis of the list of components involved. Recent advances in ‘-omics’ technologies and the development of new computational techniques and algorithms have greatly contributed to progress in this field of biology. Among the main ‘-omics’ technologies, metabolomics is expected to play a significant role in bridging the phenotype–genotype gap, since it amplifies changes in the proteome and provides a better representation of the phenotype of an organism than other methods. However, knowledge of the complete set of metabolites is not enough to predict the phenotype, especially for higher cells in which the distinct metabolic processes involved in their production and degradation are finely regulated and interconnected. In these cases, quantitative knowledge of intracellular fluxes is required for a comprehensive characterization of metabolic networks and their functional operation. These intracellular fluxes cannot be detected directly, but can be estimated through interpretation of stable isotope patterns in metabolites. Moreover, analysis of these fluxes by means of metabolic control theories offers a potentially unifying, holistic paradigm to explain the regulation of cell metabolism. In this chapter, we provide an overview of metabolomics and fluxomics, highlighting stable isotope strategies for fluxome characterization. We also discuss some of the tools used to quantitatively analyse the control exerted by components of the network over both the metabolome and the fluxome. Finally, we outline the role and future of metabolomics and fluxomics in drug discovery.
With the beginning of the new century, it is clear that the detailed inventory of genes, proteins and metabolites is not sufficient to explain cell complexity. The central dogma of one gene, one protein, one function died in the last years with the discovery that one gene may give rise to several proteins, each of which may undergo a number of post-translational modifications, thereby conferring distinct functional activities, and that a cellular function is the result of the complex interaction of various proteins. This finding, together with the impossibility of handling the enormous amount of data generated by the reductionist approach, has led to the development of bioinformatics and the integration of concepts and techniques from many other scientific disciplines, such as physics and mathematics. Furthermore, the integration of all of these disciplines has brought about a change in our view of science. This change, which implies attempting to explain biological phenomena through the net interactions of all cellular and biochemical components within a cell or an organism, has resulted in the emergence of the new discipline of systems biology.
Systems biology can be described as the integration of experimental and computational approaches in order to explain and predict complex cellular behaviour of biological systems . On the basis of how the available information is handled computationally, there are at least two distinct branches of systems biology: knowledge discovery or data mining, and simulation-based analysis. Whereas the former combines distinct high-throughput data to build and analyse cell network models as an extension of the interpretation of genome-wide data, the latter attempts to create dynamic models to test hypotheses with in silico experiments, providing predictions to be tested by in vitro and in vivo studies, which could be of relevance for the development of preventive medicine and drug discovery .
Systems biology emerged from the need to manipulate, integrate and interpret the ever increasing large-scale data generated by the numerous ‘-omics’. Thus the development and evolution of systems biology and ‘-omics’ are mutually dependent.
Although the first ‘-omic’ was undoubtedly ‘genomics’, several other ‘-omics’ are now well established within functional genomics. Among these, the most relevant are transcriptomics, proteomics and metabolomics. Despite initial hopes placed on genomics, the knowledge of genes alone does not explain or predict the phenotype of a cell. Furthermore, changes observed in the transcriptome or in the proteome do not always correspond to phenotypic alterations. However, changes in the metabolome are the ultimate response of an organism to genetic alterations, disease or environmental influences. Consequently, the metabolome is the ‘-ome’ that best predicts phenotype . As such, there has been an exponential increase in research into metabolomics, since it is expected that this ‘-omic’ will solve many open questions with its integration into metabolic models through systems biology.
Metabolites form part of the metabolic processes that occur within a living cell. These metabolic processes are connected, thereby constituting an intricate system of metabolic pathways that work in unison to form a metabolic network. This network is not static but dynamic, depending directly on both the metabolome and the fluxome, the latter being the total set of fluxes in the metabolic network of a cell. Thus metabolomics and fluxomics complement each other, and their development is highly pertinent for fundamental systems biology of metabolic networks and for all the applications that focus on manipulating or monitoring metabolic behaviour, such as those in the fields of metabolic engineering, nutrition and medicine.
The increased interest of research groups and industry in the attainable attractive results of metabolomics in recent years has promoted the advancement of this ‘-omic’, and there has been a huge increase both in fields of application and the number of metabolites that can be measured. Although the word ‘metabolome’ was first proposed in 1998 to designate the set of all low-molecular-mass compounds present in cells in a particular physiological or developmental state , it was some years until Fiehn  introduced the word metabolomics. Thus metabolomics or ‘metabolome analysis’ addresses the identification and quantification of all the metabolites of an organism . Metabolite analysis is complicated by the number of analytes, and by their diversity and dynamic ranges. It is estimated that the range of metabolite concentrations varies by approx. nine orders of magnitude (pmol–mmol), and the number of metabolites differs from ~600 metabolites in yeast to up to 200000 in humans and plants [6,7]. Consequently, the isolation and measurement of all metabolites together, what is currently known as ‘true metabolomics’, has been impossible to date, since a unique strategy that allows the simultaneous determination of all the metabolites present in a sample is not available.
Although there are no universally accepted metabolomics strategies, the most popular ones can be summarized. From a methodological point of view, Nielsen and co-workers  divided these strategies into target analysis and metabolite (or metabolic) profiling. Target analysis refers to the quantitative analysis of a class of compounds that are related to a specific pathway or to intersecting pathways. In contrast, metabolite profiling consists of the identification of the specific metabolic profile that characterizes a given sample, i.e. the set of all of the metabolites or derivative products (identified or unknown) detected by analysing a sample using a particular analytical technique. Furthermore, metabolite profiling was subdivided into metabolic fingerprinting and metabolic footprinting. This subdivision is essentially based on the distinct nature of the samples studied. Thus fingerprinting covers the scanning of intracellular metabolites, whereas footprinting measures all of the extracellular metabolites present in a culture medium. These two approaches of metabolite profiling are complementary and provide crucial information for the physiological characterization of the system analysed. Although many authors do not agree with this subdivision, the definitions presented here coincide in general terms.
To date, there is no single analytical platform that is capable of covering all of the abovementioned metabolomic strategies and that guarantees the maximum quality of results. Instead, we have at our disposal a battery of strategies for sampling and extracting metabolites as well as different analytical techniques, all of which permit us to obtain metabolic information. When the present chapter was written, analysis for metabolomics was mainly performed using spectroscopic techniques such as NMR spectroscopy, direct infusion MS, FT-IR (Fourier transform IR spectroscopy), and separation-based techniques such as GC and LC (gas and liquid chromatography) or CE (capillary electrophoresis), coupled with MS as a means of detection and identification (see Table 1 for further details). These topics have been extensively covered in several recent reviews [6,8–10].
However, despite the huge battery of protocols and analytical techniques available, the main problem now affecting metabolomics research is the lack of reliable and standardized databases that contain all of the existing spectral information and also allow the correct identification of unknown detected metabolites. In this regard, some databases have been generated that are especially useful in human metabolomics research: the so-called BiGG (biochemically, genetically and genomically structured) database developed by Bernhard Palsson and co-workers at the University of California, San Diego, CA, U.S.A. (freely accessed at http://bigg.ucsd.edu) and the HMDB (human metabolome database) compiled by David Wishart and co-workers at the University of Alberta, Alberta, Canada (available at http://www.hmdb.ca).
Metabolomics refers to an analytical technology that seeks to identify and measure all metabolites present in a biological sample at any given point in time. However, metabolism is a dynamic phenomenon. The metabolites constituent of the metabolome are continuously transformed as they are part of the set of metabolic processes that occur in cells. These processes are combined among them, constituting an intricate system of metabolic pathways that work in unison to form the metabolic network, which depends directly on both the fluxome and the metabolome.
The fluxome, or the total set of fluxes in the metabolic network of a cell, represents integrative information on several cellular processes, and hence there is a unique phenotypic characteristic of cells. Flux analysis provides a true dynamic picture of the phenotype because it captures the metabolome in its functional interactions with the environment and the genome . Although it is now common to talk about fluxomics as a science in its own right, it should not be forgotten that the fluxome is a function of at least the proteome and the metabolome, and the study of the fluxome is always accompanied by knowledge of the metabolome.
Difficulties are still encountered in measuring intracellular fluxes. Several methods have been developed for flux quantification, such as flux balance analysis or stoichiometric metabolic flux analysis. However, the most reliable are based on the use of isotope-labelled precursors of metabolic pathways, mainly 13C-labelled substrates . In these methods, cells, tissues or animals are fed 13C-labelled substrates that are metabolized, thereby resulting in metabolites containing 13C atoms. Depending on the metabolic pathway driven by the tracer, 13C atoms are incorporated into the newly formed metabolites in distinct numbers and at different positions (see Figure 1 as an example). Thus for each metabolite there may be several isotope isomers (isotopomers), i.e. several molecules of the same metabolite with distinct labelling states. The determination of concentration and isotopomeric distribution (or labelling pattern) of these metabolites is performed by means of metabolomic analytical platforms. This variant of metabolomic analysis, in which the labelling patterns of metabolites are analysed, is starting to be known as tracer-based metabolomics .
Estimation of fluxes using tracer-based metabolomic data requires a priori knowledge of possible distributions of the tracer used within the network. This limits the working possibilities in two ways. First, the size of the network should be restricted; most fluxome analysis is limited to flux distribution within the central carbon metabolism . Secondly, only a few 13C-labelled substrates can be used as tracers: most of the 13C-labelled metabolites (even when they produce 13C-labelled metabolic intermediates) do not allow researchers to distinguish through which metabolic pathways they were metabolized, and consequently estimated fluxes involve considerable errors. Among 13C-labelled substrates, those that have been most widely used as tracers are glucose labelled in the first position ([1-13C]glucose) and the uniformly labelled glucose molecule ([U-13C]glucose). However, these tracers do not allow simultaneous differentiation of the contribution of some metabolic pathways (see Figure 1A for details of [1-13C]glucose). Consequently, on many occasions combinations of several isotopes are used to reliably measure the metabolic fluxes of the network, which requires parallel incubation of cells using distinct tracers. To avoid these parallel incubations, over the last few years several single-tracer approaches have been developed. One tracer used in these attempts is [1,2-13C]glucose, which has proved to be versatile and useful [14–16], as it permits researchers to distinguish the contribution of the main central carbon metabolic pathways to glucose metabolism (see Figure 1B for details).
Depending on the information level required and particularly the bioinformatics resources available, tracer-based metabolomic data can be used to analyse metabolic flux distribution by a number of methods, the three main ones being: the comparative approach, the analytical approach and the integrative approach.
The comparative and the analytical approaches: valid methods for estimating flux ratios
Comparative analysis of tracer-based metabolomic data, known as MIDA (mass isotopomer distribution analysis), is a useful tool for the characterization of cell metabolic flux distribution . This approach involves comparing the labelling patterns obtained with the feasible labelling distributions reached depending on the metabolic pathway followed by the tracer. MIDA provides reasonably reliable estimates of relative metabolic flux distributions. Furthermore, this approach requires no bioinformatics resources and only minimal statistics knowledge. The intuitive nature of this approach has contributed to its extensive use in the interpretation of tracer-based experiments.
However, MIDA does not provide estimates of individual absolute flux values. Analytical methods consisting of simple formulas to calculate fluxes from 13C isotopomeric distributions have been developed [15,18]. Although the estimated flux values are approximate, these formulas often offer a suitable approach to estimating flux distribution in a metabolic network and involve very little computational effort . These methods have long been used not only to identify biochemical pathways and reactions, but also to quantify individual flux partitioning ratios [15,18]. These analytically deduced flux ratios are also successfully used as constraints for metabolic flux analysis.
The integrative approach: modelling metabolic flux distribution inside the network
As described above, the interpretation of tracer-based metabolomic data by means of MIDA or using analytical formulas does not allow the full characterization of the fluxome. Like all the ‘-omics’, fluxomics is the estimation of the whole set of fluxes present in a cell. The identification of the unique distributions of intracellular fluxes is performed through the integration of tracer-based metabolomic data into software packages, which have the capacity to estimate the flux distribution by fitting the experimental data  into the framework of more complete metabolic models.
The 13C-based metabolic flux analysis, which is the term commonly used to define fluxome characterization, has been used mainly to quantify the intracellular fluxes of a broad range of micro-organisms . The great interest the biotech industry has shown in these metabolic characterizations has led to the development of numerous software packages able to calculate sets of fluxes, such as the software package 13C-FLUX . However, most of these software applications require the presence of a metabolic and isotopic steady state throughout the 13C-labelling experiment. Isotopic steady states in extracellular metabolites, which are the easiest to analyse, take too long to reach , thus limiting the application of steady-state analysis to long incubation times and several cellular types.
To overcome this limitation, a number of tools have recently been developed to perform isotopically non-stationary or dynamic flux analysis [22,23]. These tools can combine the kinetic information of the enzymes involved as well as concentrations and isotopomeric distributions of intermediary metabolites. Thanks to these methods, flux estimation in non-growing cells, short transient biological processes or simply short incubation times are now possible, thereby increasing the range of application of fluxome characterization.
Despite advances, fluxomics is still in its infancy. However, it is expected to progress and provide new and unexpected results, which are likely to be especially significant for the development of our knowledge of biological systems and systems biology itself.
Tools for quantitatively analysing the components that control the metabolome and metabolic flux distribution
A classical question when analysing metabolic networks is whether a rate-limiting reaction is present. Rate-limiting steps in a pathway have important effects, as varying the activity of that step alone would change the flux in that pathway. However, present evidence suggests that most pathways are affected by the activities of several steps rather than one single step .
In metabolic networks, fluxes and metabolite concentrations are controlled by parameters such as enzyme concentrations and kinetic constants. To date, no general theory has been developed to predict the effect of large parameter changes in a network. However, the knowledge of detailed kinetic parameters of all the enzymes involved in the network, together with metabolite concentrations and fluxes, allow the construction of detailed kinetic models that can describe the dynamics of metabolic networks. These models have the capacity to identify rate-limiting steps or whether several enzymes share the control of flux distribution and metabolite homoeostasis. In recent years, intensive effort has been devoted to the generation of databases that contain values for enzyme kinetic parameters measured following standards for an increasing number of biological systems. However, we are still far from achieving detailed ‘catalogues’ of kinetic parameters for all enzymes measured in situ. Obtaining all the data required to construct a kinetic model is time consuming and in some cases too costly.
To address these limitations, several approaches have been developed to predict the effects of changes in enzyme concentrations or parameters in a metabolic network. Among these, MCA (Metabolic Control Analysis) [24,25] and BST (Biochemical Systems Theory)  are two powerful methods of studying the genetic-, enzymatic- and substrate-level regulatory mechanisms in metabolic networks. Both approaches are equivalent in essential parts, and mathematically relate local properties of individual components (such as the activity of an enzyme) to global or systemic properties of a network of biochemical reactions (such as metabolic fluxes or metabolite concentrations). Given the more specific orientation of MCA toward the description of flux control in metabolic networks, in this chapter we introduce the basic concepts of this analytical technique.
MCA relates and determines the sensitivity of the variables of a metabolic system to its system parameters. These sensitivities are described by a set of coefficients related in terms of summation and connectivity theorems . The most important of these coefficients are the control coefficients, which are a relative measure of the degree to which a perturbation in enzyme activities affects a system variable [24,25]. These control coefficients have been defined as the fractional change in the system variable, such as metabolite concentrations or metabolic fluxes, over the fractional change in the enzyme activity. Depending on the system variable analysed, two control coefficients are defined: the flux-control coefficient and the concentration-control coefficient. The greater the control coefficient of an enzyme, the greater its contribution to pathway regulation, being the flux or the metabolite concentration more sensitive to changes in the concentration of this enzyme.
MCA has had a considerable impact on the study of regulation processes in central carbon metabolism. The estimation of control coefficients can facilitate the correlation between changes at the molecular level and pathological alterations at the physiological level, as well as helping to explain the molecular mechanisms underlying metabolic homoeostasis . However, what is of greater interest is the potential application of MCA to the rational design of drug therapies in the treatment of diseases such as cancer [28,29]. According to MCA, the enzymes with the highest control coefficients in a metabolic process are the most suitable sites to target drugs at. Thus, although each enzyme in a sequence is essential for the metabolic process to work, the effects on metabolic fluxes are likely to be obtained with lower concentrations of drugs when an enzyme with a higher control coefficient is inhibited rather than one with a lower coefficient . The use of lower drug concentrations is of great importance, as toxicity and secondary effects of the agent will be reduced to a minimum.
Impact of metabolomics and fluxomics on drug discovery
One of the fields in which metabolomics has had a great influence and has received considerable investment is drug discovery. The chemical diversity of the metabolome offers major opportunities for the discovery of novel drugs and bioactive molecules that could have a significant impact on the development of new medicines and ‘functional foods’ . However, what makes metabolomics especially interesting for pharmaceutical research and development is its application in identifying disease biomarkers. Thus metabolite profiling in biofluids (such as urine and blood) is being used to generate quantitative lists of metabolites, whose chemometric analysis will determine those that discriminate for a particular disease and those that could be used in predictive medicine. Metabolic biomarkers are closely identifiable with real biological endpoints and provide a global systems interpretation of biological effects. Furthermore, metabolic biomarkers can be more easily used across species than transcriptomic or proteomic biomarkers, a feature that is very important for pharmaceutical studies .
In addition to the identification of metabolite biomarkers, fluxomics is also opening other research avenues of interest in drug discovery. Although pharmaceutical investigation has made great progress in the development of targeted drugs for a huge variety of diseases, these kinds of drugs are effective mainly in the treatment of diseases involving a single, or a very narrow range of, genetic or protein targets. The problem arises in the treatment of multifactorial diseases such as cancer, in which it is usual to find more than one mutated protein and the pattern tends to vary from one patient to the next. Particularly in the treatment of these kinds of diseases, the contribution of metabolomics is expected to be considerable. Although metabolic networks vary in their substrate utilization patterns and flux distribution, their control points have been strictly preserved throughout evolution and are reliable drug targets considering the limited number of enzyme isoforms and the limited number of major alternative metabolic routes, which are mostly known . Thus it is expected that fluxomics will significantly contribute to the recognition and development of new treatment strategies for cancer and in cases of drug resistance.
In summary, metabolomics and fluxomics are essential components of systems biology, and systems biology can contribute significantly to our knowledge of the metabolome and metabolic networks. Despite the clear impact of metabolomics in pharmaceutical research and development, several issues require immediate attention, including the standardization of procedures and databases, and the range of sensitivities detected depending on the analytical platform used. However, good biological reproducibility and low cost per sample and per analyte, together with the possibility of performing in vivo analysis of intact samples, make metabolomics and fluxomics the ‘-omics’ likely to develop the fastest and the most. Finally, we anticipate that these two ‘-omics’ will shortly become leading technologies and the order-of-the-day for science and the development of systems biology.
• Metabolomics is an emergent ‘-omic’ that is experiencing rapid development and the knowledge gained from this field will contribute significantly to systems biology.
• Fluxome characterization, based on the use of isotope-labelled precursors, can be performed using three different approaches: MIDA, the use of analytical formulas and the integration of all tracer-based data into software packages that simulate metabolic networks.
• MCA allows the identification of rate-limiting steps in metabolic networks without prior knowledge of all of the enzyme kinetic parameters
• Metabolomics and fluxomics will greatly contribute to the development of new medicines and functional foods, and the identification of new treatment strategies for multifactorial diseases respectively.
This work was supported by grants from the European Commission (FP6) BioBridge (LSHG-CT-2006-037939), the Ministerio de Educación y Ciencia (SAF2005-01626) and ISCIII-RTICC (RD06/0020/0046) from the Spanish government and the European Union ERDF funds, and the Government of Catalonia (2005SGR00204).
- © The Authors Journal compilation © 2008 Biochemical Society