1.1 Introduction
Metabolomics (Dettmer and Hammock, 2004) aims at providing an in-depth view of chemical changes in cells, tissues, organs or organisms evoked by cellular processes in response to genetic and environmental causes. It is an integral part of systems biology and provides a direct link between an external stimulus and the phenotype or physiology of a biological system (Gygi et al., 1999, Sumner et al., 2003). As such, it contributes, for example, to our understanding of the role of nutrition in maintaining good health and in contributing to or causing disease (German et al., 2002).
Metabolome analysis comprises both the qualitative and the quantitative assessment of low-molecular mass compounds (< 1000 Da), which show tremendous diversity in their chemical and physical properties. Moreover, metabolite concentrations range over up to ten orders of magnitude. There are different strategies in metabolomics (Fiehn, 2002, Dettmer et al., 2007): metabolic profiling is a hypothesis-driven rather than hypothesis-generating approach, which aims at the quantitative analysis of sets of metabolites representing either a particular biochemical pathway or a specific class of compounds. It imposes particularly high requirements on the robustness and accuracy of an analytical method. A related approach is target analysis, which is directed at the measurement of only a few selected analytes, such as biomarkers. Metabolic fingerprinting, on the other hand, pursues the recognition of changes in metabolite patterns (so-called âfingerprintsâ) in response to disease, environmental or genetic alterations. Its ultimate goal is the identification of discriminating metabolites. Metabolic fingerprinting puts high requirements on precision of measurement and data alignment, as well as classification and multivariate analysis of large and high-dimensional data sets.
Mass spectrometry (MS) has become a prominent tool in metabolomics. It allows not only the sensitive quantitative determination of metabolites over a few orders of magnitude of dynamic range, but also their identification based on accurate mass measurements, isotopic distributions, and characteristic fragmentation patterns. Originally used in combination with gas chromatography to detect metabolites in urine and tissue extracts (Dalgliesh et al., 1966), the introduction of electrospray ionization (Fenn et al., 1989) popularized direct infusion mass spectrometry (Goodacre et al., 2003, Zahn et al., 2001). However, in the case of complex biological specimens, one has to cope with so-called matrix effects, in which the matrix co-extracted with the analytes can alter the signal response, resulting in poor analytical accuracy, linearity, and reproducibility. Furthermore, direct infusion MS cannot distinguish between isobaric and isomeric species, whose masses are identical, and which often yield similar fragmentation profiles. While still popular in the targeted analysis of metabolites in combination with extraction techniques for enrichment of analytes and the use of appropriate stable-isotope labeled analogues (Gieger et al., 2008), direct infusion MS has been increasingly replaced by the on-line coupling of high-performance separation techniques such as liquid chromatography (LC) and gas chromatography (GC) to MS using a variety of interfaces for the ionization of eluting analytes.
Mass spectrometry instrumentation has seen significant improvement over the years, including the development of robust front-end ionization technologies for a wide range of compound classes, increased sensitivity and resolution, ease of use, and hybrid instruments such as the linear ion trapâFourier transform ion cyclotron resonance mass spectrometer for fast multistage tandem MS experiments as well as ultra-high resolution and accurate mass determinations. Nevertheless, metabolite identification remains a major challenge. At present, identification can be achieved by means of mass spectral libraries in the case of commonly used electron ionization by GC-MS, but mass spectral library searches typically identify only a minority of signals (Almstetter et al., 2011, Almstetter et al., 2009). Additional strategies must be implemented, such as the generation of quasi-molecular ions by soft ionization methods (chemical ionization (CI), atmospheric pressure chemical ionization (APCI)) followed by calculating sum formulas and additionally considering the isotopic pattern (Kind and Fiehn, 2006) and chemical and heuristic rules (Kind and Fiehn, 2007). Accurate mass measurements for identification are commonly performed using LC-MS, and databases are searched afterwards. The lack of reference compounds, however, impedes unambiguous assignment of unknowns. MSn acquisition and interpretation of the fragmentation pattern might be of help, as it provides additional structural information. Recent years have seen a flurry of novel computational approaches for identifying metabolites, but systematic evaluations of their performance are still lacking, making a definitive judgment on their routine usefulness difficult.
Cross-platform approaches are utilized to cope with analyte diversity. Moreover, multiple metabolite databases can be searched (e.g. HMDB (Wishart et al., 2009), LIPID MAPS (Sud et al., 2007), and METLIN (Smith et al., 2005)). Once identified, metabolites may be visualized within their metabolic pathways (e.g. provided by KEGG database) to facilitate biological interpretation. Finally, metabolomics data can be incorporated with results obtained by the other -omics methods to obtain a global picture of the organism under study.
This chapter portrays front-end separation (LC, GC) and MS technologies for both targeted and untargeted metabolomics, with a special focus on compound ID approaches.
1.2 Liquid chromatography
Coupling of MS with a separation technique (GC, LC) reduces the complexity of the mass spectra of biological specimens due to metabolite separation in a time dimension, provides isomer and isobar separation, and delivers information on the physicochemical properties of metabolites. LC is preferred for semi- or nonvolatile analytes provided that they can be dissolved in a suitable solvent such as water, water/methanol or water/acetonitrile in the case of reversed phase chromatography. Ideally, the sample should be dissolved in the mobile phase of the starting gradient.
An LC assembly consists of a pumping system for one to four solvents (often joined by an on-line degasser) that constitute the mobile phase required for carrying the dissolved analyte mixture through the LC column. The analyte mixture is introduced in the majority of applications by an autosampler that uses a six-port valve to maintain an uninterrupted flow through the column. Finally, following separation according to interaction with the stationary phase, analytes are detected and a chromatogram is recorded. Commonly employed detection methods include UV absorbance at selected wavelengths or full UV/Vis spectra data acquisition by means of a diode-array detector, fluorescence and electrochemical detection, the latter being the most sensitive and selective mode of LC detection for the measurement of trace amounts of oxidizable or reducible compounds, as well as mass spectrometry. The latter is the most universal detector. Depending on the type of mass analyzer, it can be applied to both screening and selective detection provided that the analytes can be ionized. Selective detection as pr...