DISCO-SCA and properly applied GSVD as swinging methods to find common and distinctive processes.

BACKGROUND: In systems biology it is common to obtain for the same set of biological entities information from multiple sources. Examples include expression data for the same set of orthologous genes screened in different organisms and data on the same set of culture samples obtained with different high-throughput techniques. A major challenge is to find the important biological processes underlying the data and to disentangle therein processes common to all data sources and processes distinctive for a specific source. Recently, two promising simultaneous data integration methods have been proposed to attain this goal, namely generalized singular value decomposition (GSVD) and simultaneous component analysis with rotation to common and distinctive components (DISCO-SCA). RESULTS: Both theoretical analyses and applications to biologically relevant data show that: (1) straightforward applications of GSVD yield unsatisfactory results, (2) DISCO-SCA performs well, (3) provided proper pre-processing and algorithmic adaptations, GSVD reaches a performance level similar to that of DISCO-SCA, and (4) DISCO-SCA is directly generalizable to more than two data sources. The biological relevance of DISCO-SCA is illustrated with two applications. First, in a setting of comparative genomics, it is shown that DISCO-SCA recovers a common theme of cell cycle progression and a yeast-specific response to pheromones. The biological annotation was obtained by applying Gene Set Enrichment Analysis in an appropriate way. Second, in an application of DISCO-SCA to metabolomics data for Escherichia coli obtained with two different chemical analysis platforms, it is illustrated that the metabolites involved in some of the biological processes underlying the data are detected by one of the two platforms only; therefore, platforms for microbial metabolomics should be tailored to the biological question. CONCLUSIONS: Both DISCO-SCA and properly applied GSVD are promising integrative methods for finding common and distinctive processes in multisource data. Open source code for both methods is provided.

Simplivariate models: uncovering the underlying biology in functional genomics data.

One of the first steps in analyzing high-dimensional functional genomics data is an exploratory analysis of such data. Cluster Analysis and Principal Component Analysis are then usually the method of choice. Despite their versatility they also have a severe drawback: they do not always generate simple and interpretable solutions. On the basis of the observation that functional genomics data often contain both informative and non-informative variation, we propose a method that finds sets of variables containing informative variation. This informative variation is subsequently expressed in easily interpretable simplivariate components.We present a new implementation of the recently introduced simplivariate models. In this implementation, the informative variation is described by multiplicative models that can adequately represent the relations between functional genomics data. Both a simulated and two real-life metabolomics data sets show good performance of the method.

A clone-based transcriptomics approach for the identification of genes relevant for itaconic acid production in Aspergillus.

Several Aspergillus species are well-known for the production of a variety of organic acids. In this study, a cloned based transcriptomics approach was used to identify genes crucial in the biosynthesis pathway for one of these acids, itaconic acid. From a number of different Aspergillus terreus controlled batch fermentations, those cultures with the largest difference in itaconic acid titer and productivity were selected for mRNA isolation. cDNAs derived from these mRNA samples were used for subsequent hybridization of glass slide arrays based on a collection of 5000 cDNA clones. A selection of 13 cDNA clones resulting in the strongest (>10-fold) differential hybridization signals were identified and subsequently the inserts of these clones were sequenced. Sequence analysis revealed the presence of in total five different gene inserts among the sequenced clones. From one of these sequences, encoding a gene of the MmgE-PrpD family, the full length coding region was shown to encode one of the crucial itaconic acid pathway enzymes cis-aconitate decarboxylase, by heterologous expression in Escherichia coli. Expression of this gene in Aspergillus niger, which is a natural citric acid producer, resulted in itaconate production. Genome analysis suggests that in A. terreus the cis-aconitate decarboxylase gene is part of an itaconate acid related gene cluster including genes encoding two pathway specific transporters and a Zinc finger protein. Interestingly, this cluster is directly linked to the large lovastatin gene cluster.

Metabolomics as a tool for target identification in strain improvement: the influence of phenotype definition.

For the optimization of microbial production processes, the choice of the quantitative phenotype to be optimized is crucial. For instance, for the optimization of product formation, either product concentration or productivity can be pursued, potentially resulting in different targets for strain improvement. The choice of a quantitative phenotype is highly relevant for classical improvement approaches, and even more so for modern systems biology approaches. In this study, the information content of a metabolomics dataset was determined with respect to different quantitative phenotypes related to the formation of specific products. To this end, the production of two industrially relevant products by Aspergillus niger was evaluated: (i) the enzyme glucoamylase, and (ii) the more complex product group of secreted proteases, consisting of multiple enzymes. For both products, six quantitative phenotypes associated with activity and productivity were defined, also taking into account different time points of sampling during the fermentation. Both linear and nonlinear relationships between the metabolome data and the different quantitative phenotypes were considered. The multivariate data analysis tool partial least-squares (PLS) was used to evaluate the information content of the datasets for all the different quantitative phenotypes defined. Depending on the product studied, different quantitative phenotypes were found to have the highest information content in specific metabolomics datasets. A detailed analysis of the metabolites that showed strong correlation with these quantitative phenotypes revealed that various sugar derivatives correlated with glucoamylase activity. For the reduction of protease activity, mainly as-yet-unidentified compounds correlated.

Identification of modules in Aspergillus niger by gene co-expression network analysis.

The fungus Aspergillus niger has been studied in considerable detail with respect to various industrial applications. Although its central metabolic pathways are established relatively well, the mechanisms that control the adaptation of its metabolism are understood rather poorly. In this study, clustering of co-expressed genes has been performed on the basis of DNA microarray data sets from two experimental approaches. In one approach, low amounts of inducer caused a relatively mild perturbation, while in the other approach the imposed environmental conditions including carbon source starvation caused severe perturbed stress. A set of conserved genes was used to construct gene co-expression networks for both the individual and combined data sets. Comparative analysis revealed the existence of modules, some of which are present in all three networks. In addition, experimental condition-specific modules were identified. Module-derived consensus expression profiles enabled the integration of all protein-coding A. niger genes to the co-expression analysis, including hypothetical and poorly conserved genes. Conserved sequence motifs were detected in the upstream region of genes that cluster in some modules, e.g., the binding site for the amino acid metabolism-related transcription factor CpcA as well as for the fatty acid metabolism-related transcription factors, FarA and FarB. Moreover, not previously described putative transcription factor binding sites were discovered for two modules: the motif 5'-CGACAA is overrepresented in the module containing genes encoding cytosolic ribosomal proteins, while the motif 5'-GGCCGCG is overrepresented in genes related to 'gene expression', such as RNA helicases and translation initiation factors.

Microbial renewable feedstock utilization: a substrate-oriented approach.

Increasingly lignocellulosic biomass hydrolysates are used as the feedstock for industrial fermentations. These biomass hydrolysates consist of complex mixtures of different fermentable sugars, but also contain inhibitors and salts that affect the performance of the product-generating microbes. The performance of six industrially relevant microorganisms, i.e., two bacteria (Escherichia coli and Corynebacterium glutamicum), two yeasts (Saccharomyces cerevisiae and Pichia stipitis) and two fungi (Aspergillus niger and Trichoderma reesei) were compared for their ability to utilize and grow on different feedstock hydrolysates (corn stover, wheat straw, sugar cane bagasse and willow wood). Moreover, the ability of the selected hosts to utilize waste glycerol from the biodiesel industry was evaluated. P. stipitis and A. niger were found to be the most versatile and C. glutamicum, and S. cerevisiae were shown to be the least adapted to renewable feedstocks. Clear differences in the utilization of the more abundant carbon sources in these feedstocks were observed between the different species. Moreover, in a species-specific way the production of various metabolites, in particular polyols, alcohols and organic acids was observed during fermentation. Based on the results obtained we conclude that a substrate-oriented instead of the more commonly used product oriented approach towards the selection of a microbial production host will avoid the requirement for extensive metabolic engineering. Instead of introducing multiple substrate utilization and detoxification routes to efficiently utilize lignocellulosic hydrolysates only one biosynthesis route forming the product of interest has to be engineered.

Microbial production host selection for converting second-generation feedstocks into bioproducts.

BACKGROUND: Increasingly lignocellulosic biomass hydrolysates are used as the feedstock for industrial fermentations. These biomass hydrolysates are complex mixtures of different fermentable sugars, but also inhibitors and salts that affect the performance of the microbial production host. The performance of six industrially relevant microorganisms, i.e. two bacteria (Escherichia coli and Corynebacterium glutamicum), two yeasts (Saccharomyces cerevisiae and Pichia stipitis) and two fungi (Aspergillus niger and Trichoderma reesei) were compared for their (i) ability to utilize monosaccharides present in lignocellulosic hydrolysates, (ii) resistance against inhibitors present in lignocellulosic hydrolysates, (iii) their ability to utilize and grow on different feedstock hydrolysates (corn stover, wheat straw, sugar cane bagasse and willow wood). The feedstock hydrolysates were generated in two manners: (i) thermal pretreatment under mild acid conditions followed by enzymatic hydrolysis and (ii) a non-enzymatic method in which the lignocellulosic biomass is pretreated and hydrolyzed by concentrated sulfuric acid. Moreover, the ability of the selected hosts to utilize waste glycerol from the biodiesel industry was evaluated. RESULTS: Large differences in the performance of the six tested microbial production hosts were observed. Carbon source versatility and inhibitor resistance were the major discriminators between the performances of these microorganisms. Surprisingly all 6 organisms performed relatively well on pretreated crude feedstocks. P. stipitis and A. niger were found to give the overall best performance C. glutamicum and S. cerevisiae were shown to be the least adapted to renewable feedstocks. CONCLUSION: Based on the results obtained we conclude that a substrate oriented instead of the more commonly used product oriented approach towards the selection of a microbial production host will avoid the requirement for extensive metabolic engineering. Instead of introducing multiple substrate utilization and detoxification routes to efficiently utilize lignocellulosic hydrolysates only one biosynthesis route forming the product of interest has to be engineered.

Characterizing the precision of mass-spectrometry-based metabolic profiling platforms.

In conventional analytical chemistry it is customary to report figures of merit - such as precision, analysis time, limit of detection - for the chemical analysis performed. Usually, such figures of merit are reported for each analyte separately, generating a list of figures of merit. In metabolomics such a listing is not informative, since very many compounds are measured. An ANOVA-based strategy is proposed for a global measure of precision of the whole experiment, broken down in components of variation contributing to the total variation. This strategy uses well established statistical techniques and can be used easily. It was implemented to study the reproducibility of different comprehensive GC- and LC-MS methods for the analysis of tobacco aerosols. The results give insight into different sources of variation contributing to the total variation, such as biological variability, sampling variability and repeatability. For the specific example, median CV values ranged from 4.6% to 12.5% for repeatability; from 14.7% to 18.0% for sampling variability and, finally, from 24.2% to 26.8% for biological variability. Such a breakdown of sources of variability gives clues for improving the methods.

Metabolomics data exploration guided by prior knowledge.

In metabolomics research, it is often important to focus the data analysis to specific areas of interest within the metabolome. In this paper, we describe the application of consensus principal component analysis (CPCA) and canonical correlation analysis (CCA) as a means to explore the relation between metabolome data and (i) biochemically related metabolites and (ii) an amino acid biosynthesis pathway. CPCA searches for major trends in the behavior of metabolite concentrations that are in common for the metabolites of interest and the remainder of the metabolome. CCA identifies the strongest correlations between the metabolites of interest and the remainder of the metabolome. CPCA and CCA were applied to two different microbial metabolomics data sets. The first data set, derived from Pseudomonas putida S12, was relatively simple as it contained metabolomes obtained under four environmental conditions only. The second data set, obtained from Escherichia coli, was much more complex as it consisted of metabolomes obtained under 28 different environmental conditions. In case of the simple and coherent P. putida S12 data set, CCA and CPCA gave similar results as the variation in the subset of the selected metabolites and the remainder of the metabolome was similar. In contrast, CCA and CPCA yielded different results in case of the E. coli data set. With CPCA the trends in the selected subset--the phenylalanine biosynthesis pathway--dominated the results. The main trends were related to high and low phenylalanine productivity, and the metabolites showing a similar behavior in concentration were metabolites regulating the phenylalanine biosynthesis route in the subset and metabolites related to general amino acid metabolism in the remainder of the metabolome. With CCA, neither subset truly dominated the data analysis. CCA described the differences between the wild type and the overproducing strain and the differences between the succinate and glucose grown cells. For the difference between the wild type and the overproducing strain, metabolites from the beginning and the end of aromatic amino acid pathways like erythrose-4-phosphate, tryptophan, and phenylalanine were important for the selected metabolites. CCA and CPCA proved to be complementary data analysis tools that enable the focusing of the data analysis on groups of metabolites that are of specific interest in relation to the remainder of the metabolome. Compared to an ordinary PCA, focusing the data analysis on biologically relevant metabolites lead especially for the complex E. coli data to a better biological interpretation of the data.

A structured overview of simultaneous component based data integration.

BACKGROUND: Data integration is currently one of the main challenges in the biomedical sciences. Often different pieces of information are gathered on the same set of entities (e.g., tissues, culture samples, biomolecules) with the different pieces stemming, for example, from different measurement techniques. This implies that more and more data appear that consist of two or more data arrays that have a shared mode. An integrative analysis of such coupled data should be based on a simultaneous analysis of all data arrays. In this respect, the family of simultaneous component methods (e.g., SUM-PCA, unrestricted PCovR, MFA, STATIS, and SCA-P) is a natural choice. Yet, different simultaneous component methods may lead to quite different results. RESULTS: We offer a structured overview of simultaneous component methods that frames them in a principal components setting such that both the common core of the methods and the specific elements with regard to which they differ are highlighted. An overview of principles is given that may guide the data analyst in choosing an appropriate simultaneous component method. Several theoretical and practical issues are illustrated with an empirical example on metabolomics data for Escherichia coli as obtained with different analytical chemical measurement methods. CONCLUSION: Of the aspects in which the simultaneous component methods differ, pre-processing and weighting are consequential. Especially, the type of weighting of the different matrices is essential for simultaneous component analysis. These types are shown to be linked to different specifications of the idea of a fair integration of the different coupled arrays.

Analyzing longitudinal microbial metabolomics data.

A longitudinal experimental design in combination with metabolomics and multiway data analysis is a powerful approach in the identification of metabolites whose correlation with bioproduct formation shows a shift in time. In this paper, a strategy is presented for the analysis of longitudinal microbial metabolomics data, which was performed in order to identify metabolites that are likely inducers of phenylalanine production by Escherichia coli. The variation in phenylalanine production as a function of differences in metabolism induced by the different environmental conditions in time was described by a validated multiway statistical model. Notably, most of the metabolites showing the strongest relations with phenylalanine production seemed to hardly change in time. Apparently, potential bottlenecks in phenylalanine seem to hardly change in the course of a batch fermentation. The approach described in this study is not limited to longitudinal microbial studies but can also be applied to other (biological) studies in which similar longitudinal data need to be analyzed.

The effect of environmental conditions on extracellular protease activity in controlled fermentations of Aspergillus niger.

Proteolytic degradation by host proteases is one of the key issues in the application of filamentous fungi for non-fungal protein production. In this study the influence of several environmental factors on the production of extracellular proteases of Aspergillus niger was investigated systematically in controlled batch cultures. Of all factors investigated in a series of initial screening experiments, culture pH and nitrogen concentration in particular strongly affected extracellular protease activities. For instance, at a culture pH of 4, protease activity was higher than at pH 5, and protease activity increased with increasing concentrations of ammonium as nitrogen source. Interestingly, an interdependence was observed for several of the factors studied. These possible interaction effects were investigated further using a full factorial experimental design. Amongst others, the results showed a clear interaction effect between nitrogen source and nitrogen concentration. Based on the observed interactions, the selection of environmental factors to reduce protease activity is not straightforward, as unexpected antagonistic or synergistic effects occur. Furthermore, not only were the effects of the process parameters on maximum protease activity investigated, but five other protease-related phenotypes were studied as well, such as maximum specific protease activity and maximum protease productivity. There were significant differences in the effect of the environmental parameters on the various protease-related phenotypes. For instance, pH significantly affected final levels of protease activity, but not protease productivity. The results obtained in this study are important for the optimization of A. niger for protein production.

Simplivariate models: ideas and first examples.

One of the new expanding areas in functional genomics is metabolomics: measuring the metabolome of an organism. Data being generated in metabolomics studies are very diverse in nature depending on the design underlying the experiment. Traditionally, variation in measurements is conceptually broken down in systematic variation and noise where the latter contains, e.g. technical variation. There is increasing evidence that this distinction does not hold (or is too simple) for metabolomics data. A more useful distinction is in terms of informative and non-informative variation where informative relates to the problem being studied. In most common methods for analyzing metabolomics (or any other high-dimensional x-omics) data this distinction is ignored thereby severely hampering the results of the analysis. This leads to poorly interpretable models and may even obscure the relevant biological information. We developed a framework from first data analysis principles by explicitly formulating the problem of analyzing metabolomics data in terms of informative and non-informative parts. This framework allows for flexible interactions with the biologists involved in formulating prior knowledge of underlying structures. The basic idea is that the informative parts of the complex metabolomics data are approximated by simple components with a biological meaning, e.g. in terms of metabolic pathways or their regulation. Hence, we termed the framework 'simplivariate models' which constitutes a new way of looking at metabolomics data. The framework is given in its full generality and exemplified with two methods, IDR analysis and plaid modeling, that fit into the framework. Using this strategy of 'divide and conquer', we show that meaningful simplivariate models can be obtained using a real-life microbial metabolomics data set. For instance, one of the simple components contained all the measured intermediates of the Krebs cycle of E. coli. Moreover, these simplivariate models were able to uncover regulatory mechanisms present in the phenylalanine biosynthesis route of E. coli.

Comprehensive analysis of the metabolome of Pseudomonas putida S12 grown on different carbon sources.

Metabolomics is an emerging, powerful, functional genomics technology that involves the comparative non-targeted analysis of the complete set of metabolites in an organism. We have set-up a robust quantitative metabolomics platform that allows the analysis of 'snapshot' metabolomes. In this study, we have applied this platform for the comprehensive analysis of the metabolite composition of Pseudomonas putida S12 grown on four different carbon sources, i.e. fructose, glucose, gluconate and succinate. This paper focuses on the microbial aspects of analyzing comprehensive metabolomes, and demonstrates that metabolomes can be analyzed reliably. The technical (i.e. sample work-up and analytical) reproducibility was on average 10%, while the biological reproducibility was approximately 40%. Moreover, the energy charge values of the microbial samples generated were determined, and indicated that no biotic or abiotic changes had occurred during sample work-up and analysis. In general, the metabolites present and their concentrations were very similar after growth on the different carbon sources. However, specific metabolites showed large differences in concentration, especially the intermediates involved in the degradation of the carbon sources studied. Principal component discriminant analysis was applied to identify metabolites that are specific for, i.e. not necessarily the metabolites that show those largest differences in concentration, cells grown on either of these four carbon sources. For selected enzymatic reactions, i.e. the glucose-6-phosphate isomerase, triosephosphate isomerase and phosphoglyceromutase reactions, the apparent equilibrium constants (K(app)) were calculated. In several instances a carbon source-dependent deviation between the apparent equilibrium constant (K(app)) and the thermodynamic equilibrium constant (K(eq)) was observed, hinting towards a potential point of metabolic regulation or towards bottlenecks in biosynthesis routes. For glucose-6-phosphate isomerase and phosphoglyceromutase, the K(app) was larger than K(eq), and the results suggested that the specific enzymatic activities of these two enzymes were too low to reach the thermodynamic equilibrium in growing cells. In contrast, with triosephosphate isomerase the K(app) was smaller than K(eq), and the results suggested that this enzyme is kinetically controlled.

Microbial metabolomics: toward a platform with full metabolome coverage.

Achieving metabolome data with satisfactory coverage is a formidable challenge in metabolomics because metabolites are a chemically highly diverse group of compounds. Here we present a strategy for the development of an advanced analytical platform that allows the comprehensive analysis of microbial metabolomes. Our approach started with in silico metabolome information from three microorganisms-Escherichia coli, Bacillus subtilis, and Saccharomyces cerevisiae-and resulted in a list of 905 different metabolites. Subsequently, these metabolites were classified based on their physicochemical properties, followed by the development of complementary gas chromatography-mass spectrometry and liquid chromatography-mass spectrometry methods, each of which analyzes different metabolite classes. This metabolomics platform, consisting of six different analytical methods, was applied for the analysis of the metabolites for which commercial standards could be purchased (399 compounds). Of these 399 metabolites, 380 could be analyzed with the platform. To demonstrate the potential of this metabolomics platform, we report on its application to the analysis of the metabolome composition of mid-logarithmic E. coli cells grown on a mineral salts medium using glucose as the carbon source. Of the 431 peaks detected, 235 (=176 unique metabolites) could be identified. These include 61 metabolites that were not previously identified or annotated in existing E. coli databases.

Standard reporting requirements for biological samples in metabolomics experiments: microbial and in vitro biology experiments.

With the increasing use of metabolomics as a means to study a large number of different biological research questions, there is a need for a minimal set of reporting standards that allow the scientific community to evaluate, understand, repeat, compare and re-investigate metabolomics studies. Here we propose, a first draft of minimal requirements to effectively describe the biological context of metabolomics studies that involve microbial or in vitro biological subjects. This recommendation has been produced by the microbiology and in vitro biology working subgroup of the Metabolomics Standards Initiative in collaboration with the yeast systems biology network as part of a wider standardization initiative led by the Metabolomics Society. Microbial and in vitro biology metabolomics is defined by this sub-working group as studies with any cell or organism that require a defined external medium to facilitate growth and propagation. Both a minimal set and a best practice set of reporting standards for metabolomics experiments have been defined. The minimal set of reporting standards for microbial or in vitro biology metabolomics experiments includes those factors that are specific for metabolomics experiments and that critically determine the outcome of the experiments. The best practice set of reporting standards contains both the factors that are specific for metabolomics experiments and general aspects that critically determine the outcome of any microbial or in vitro biological experiment.

Simultaneous quantitative analysis of metabolites using ion-pair liquid chromatography-electrospray ionization mass spectrometry.

We have developed an analytical method, consisting of ion-pair liquid chromatography coupled to electrospray ionization mass spectrometry (IP-LC-ESI-MS), for the simultaneous quantitative analysis of several key classes of polar metabolites, like nucleotides, coenzyme A esters, sugar nucleotides, and sugar bisphosphates. The use of the ion-pair agent hexylamine and optimization of the pH of the mobile phases were critical parameters in obtaining good retention and peak shapes of many of the above-mentioned polar and acidic metabolites that are impossible to analyze using standard reversed-phase LC/MS. Optimum conditions were found when using a gradient from 5 mM hexylamine in water (pH 6.3) to 90% methanol/10% 10 mM ammonium acetate (pH 8.5). The IP-LC-ESI-MS method was extensively validated by determining the linearity (R2 > 0.995), sensitivity (limit of detection 0.1-1 ng), repeatability, and reproducibility (relative standard deviation <10%). The IP-LC-ESI-MS method was shown to be a useful tool for microbial metabolomics, i.e., the comprehensive quantitative analysis of metabolites in extracts of microorganisms, and for the determination of the energy charge, i.e., the cellular energy status, as an overall quality measure for the sample workup and analytical protocols.

Centering, scaling, and transformations: improving the biological information content of metabolomics data.

BACKGROUND: Extracting relevant biological information from large data sets is a major challenge in functional genomics research. Different aspects of the data hamper their biological interpretation. For instance, 5000-fold differences in concentration for different metabolites are present in a metabolomics data set, while these differences are not proportional to the biological relevance of these metabolites. However, data analysis methods are not able to make this distinction. Data pretreatment methods can correct for aspects that hinder the biological interpretation of metabolomics data sets by emphasizing the biological information in the data set and thus improving their biological interpretability. RESULTS: Different data pretreatment methods, i.e. centering, autoscaling, pareto scaling, range scaling, vast scaling, log transformation, and power transformation, were tested on a real-life metabolomics data set. They were found to greatly affect the outcome of the data analysis and thus the rank of the, from a biological point of view, most important metabolites. Furthermore, the stability of the rank, the influence of technical errors on data analysis, and the preference of data analysis methods for selecting highly abundant metabolites were affected by the data pretreatment method used prior to data analysis. CONCLUSION: Different pretreatment methods emphasize different aspects of the data and each pretreatment method has its own merits and drawbacks. The choice for a pretreatment method depends on the biological question to be answered, the properties of the data set and the data analysis method selected. For the explorative analysis of the validation data set used in this study, autoscaling and range scaling performed better than the other pretreatment methods. That is, range scaling and autoscaling were able to remove the dependence of the rank of the metabolites on the average concentration and the magnitude of the fold changes and showed biologically sensible results after PCA (principal component analysis).In conclusion, selecting a proper data pretreatment method is an essential step in the analysis of metabolomics data and greatly affects the metabolites that are identified to be the most important.

Microbial metabolomics with gas chromatography/mass spectrometry.

An analytical method was set up suitable for the analysis of microbial metabolomes, consisting of an oximation and silylation derivatization reaction and subsequent analysis by gas chromatography coupled to mass spectrometry. Microbial matrixes contain many compounds that potentially interfere with either the derivatization procedure or analysis, such as high concentrations of salts, complex media or buffer components, or extremely high substrate and product concentrations. The developed method was extensively validated using different microorganisms, i.e., Bacillus subtilis, Propionibacterium freudenreichii, and Escherichia coli. Many metabolite classes could be analyzed with the method: alcohols, aldehydes, amino acids, amines, fatty acids, (phospho-) organic acids, sugars, sugar acids, (acyl-) sugar amines, sugar phosphate, purines, pyrimidines, and aromatic compounds. The derivatization reaction proved to be efficient (>50% transferred to derivatized form) and repeatable (relative standard deviations <10%). Linearity for most metabolites was satisfactory with regression coefficients better than 0.996. Quantification limits were 40-500 pg on-column or 0.1-0.7 mmol/g of microbial cells (dry weight). Generally, intrabatch precision (repeatability) and interbatch precision (reproducibility) for the analysis of metabolites in cell extracts was better than 10 and 15%, respectively. Notwithstanding the nontargeted character of the method and complex microbial matrix, analytical performance for most metabolites fit the requirements for target analysis in bioanalysis. The suitability of the method was demonstrated by analysis of E. coli samples harvested at different growth phases.