Title: Metabolomics: What You See is What You Extract
Abstract: In recent years, life sciences have shifted towards a more holistic approach to the study of organisms, considering them as a whole rather than as a sum of individual parts. This approach, known as systems biology, is necessarily interdisciplinary, requiring the use of physiological, physical, molecular, biochemical and chemical methods to observe and measure the response of an organism to diverse conditions. The omics technologies, such as genomics, transcriptomics, proteomics and metabolomics, are particularly important tools for systems biology. Metabolomics is one of the most recent omics technologies. In the last couple of years, it has been increasingly applied to many fields of biology, such as functional genomics, physiology, toxicology, pathology, chemotaxonomy, chemical ecology and natural products research (Verpoorte et al., 2007; Nielsen and Oliver, 2012). The aim of metabolomics is the measurement of the changes in the relative concentrations of metabolites as a result of internal or external factors, including the deletion or overexpression of a gene and environmental stimuli (Oliver et al., 1998). The design of a metabolomics experiment includes four basic steps: sample harvesting and extraction, analysis of the extracts, data reduction including statistics and metabolite identification. After completing the pre-analytical procedures, a non-targeted analysis of the metabolites is performed in the resulting extracts. In this step, individual metabolites must be identified. There are many reports that compare the advantages and limitations of the analytical platforms used for metabolomics, such as mass- (MS), nuclear magnetic resonance (NMR)-, infrared (IR)- and ultraviolet (UV) spectroscopy, with or without combination with chromatography (reviewed by Sumner et al., 2003; Verpoorte et al., 2007; Kim et al., 2011; Wolfender et al., 2013). Robust protocols that use these analytical platforms have been published for NMR (Kruger et al., 2008; Kim et al., 2011), GC–MS (Lisec et al., 2006) and LC–MS (De Vos et al., 2007). In these protocols, sensitivity, selectivity, diversity of detected metabolites and suitability for target tissues have been maximised. The large data sets generated by the analytical instruments require the use of mathematical and statistical procedures to extract efficiently the maximum amounts of useful information from the data. A basic unsupervised data reduction method, principal component analysis (PCA), is commonly used as the first step of multivariate data analysis and is followed by a variety of supervised multivariate data analyses depending on the quality of data and the research interests (Berrueta et al., 2007). In the final step of metabolomics, selected signals must be interpreted to identify the metabolites, using, for example, computational NMR analysis (Holmes and Antti, 2002; Ludwig et al., 2011), databases of NMR spectra of common metabolites, or various MS databases in the case of MS or MS combined with other techniques (reviewed by Tohge and Fernie, 2009). However, although attention is paid to the analytical and statistical treatment of data, sample preparation and extraction are often neglected. Researchers generally choose pre-analytical methods simply based on their experience, but with little evidence of the suitability for metabolomics and no validation for a wide range of applications. The enormous number of metabolites with a wide range of polarities, different stabilities and chemical diversity present in a single organism pose a great challenge for the design of efficient pre-analytical protocols. An organism is thought to contain approximately as many metabolites as genes, that is somewhere in the range of 30,000 metabolites in the case of plants. As metabolomics deals with all metabolites in an organism, the failure to extract all metabolites or the transformation of metabolites during the processing of the material studied will provide unreliable information, resulting in a skewed view of the metabolome. There are a great number of factors that must be considered when preparing samples for a metabolomic study. The time of collection (diurnal, seasonal and developmental variation), possible contamination, particle size, exposure to light and temperature, and the immediate quenching of chemical and enzymatic reactions following collection, must all be considered to preserve the integrity of the metabolome at a given point in time. So far, studies on the extraction for metabolomics analyses have focused mostly on quenching methods, variation of extraction instruments and the effect of solvent polarity and selectivity. Quenching methods that interrupt cellular metabolism have been used for diverse tissues to obtain intact metabolites during sampling. The mechanism of extraction involves not only solute–solvent interactions but also solute–solute interactions, and in particular, the dissociation of the analytes from the biological matrix. The latter is greatly influenced by the chemical characteristics of the matrix, the solvent and the metabolites. The behaviour of the matrix itself, for example, its degree of swelling, is affected by pH, contact time and the localisation of metabolites in tissues. To achieve an efficient extraction of intracellular metabolites, the cell wall (when present) and membrane must be made permeable before the metabolites can be extracted (Villas-Bôas, 2007). All these issues and those referred to the actual extraction of metabolites are discussed extensively in this issue. Mechanical means are also employed to increase the efficiency of metabolite extraction. Ultrasound and microwave are commonly used in extraction methods to provide auxiliary energy to solid samples and speed up the solubilisation process (Luque de Castro and da Silva, 1997). The combination of mechanical grinding and ultrasound-supported solvent extraction has proved to produce high extract yields and chemical diversity (Jaki et al., 2006). Ultrasound-assisted extraction is also faster than classical methods such as Soxhlet extraction (Ruiz-Jiménex and Luque de Castro, 2004). Microwave technology is of considerable significance as an alternative extraction method. It can help to reduce the extraction time by improving transport rates of molecules, molecular agitation and heating of solvents above their boiling points (Luque-García and Luque de Castro, 2003). The choice of the extraction solvent is of utmost importance. Considering the chemical diversity of the metabolites, it is practically impossible to find one solvent that will extract them all. Other issues, such as solvent volume, sample:solvent ratio and the conditioning of the sample for its introduction into the analytical instrument, can change the outcome of a metabolomic study completely because they affect the solubility of the metabolites. Protocols should thus be meticulously followed for reproducible results. To avoid these problems, Yuliana et al. (2011) developed comprehensive extraction, in which a gradient of extraction solvents was used. Analysis of the compounds in each fraction obtained from this extraction show that the isolated metabolites are clustered in three groups that can be characterised as lipophyllic (n-hexane or ethyl acetate), medium polar (methanol) and polar or hydrophilic (water). This should thus be considered when choosing an extraction protocol. A two-phase extraction with chloroform:methanol:water systems could also cover most metabolites. Alternately, several extracts with solvents of different polarity or selectivity should be made to obtain the whole metabolome. The use of new classes of solvents is gaining ground: ionic liquids, deep eutectic solvents and supercritical solvents are being tested for metabolomics studies. Aside from the chemical aspects related to extraction mechanisms, physical factors such as pressure and temperature are known to have a great effect on extraction efficiency. For example, sample structures such as cell walls or membranes and degree of swelling are easily modified by both temperature and pressure, and so are the density and ability of the solvents to permeate into the matrix. Pressurised solvent extraction (PSE) is a method that optimises these variables using organic solvents at high pressure and temperature to enhance the recovery of organic compounds from environmental, food, pharmaceutical and industrial samples. This special issue of Phytochemical Analysis offers a number of examples of extraction methods of various metabolites. It also contains reviews of pre-analytical methods applied in metabolomic studies. The articles reveal the need to address the major challenge for metabolomics studies, that is, whether it is possible to agree on a standard protocol that ensures the extraction of the widest range of metabolites for metabolomic studies. In applying metabolomics it should thus always remember that: 'What you see is what you extract'.