A novel stable isotope labelling assisted workflow

Authors: Christoph Bueschl Bernhard Kluger Marc Lemmens Gerhard Adam Gerlinde Wiesenberger Valentina Maschietto Adriano Marocco Joseph Strauss Stephan Bödi Gerhard G Thallinger Rudolf Krska Rainer Schuhmacher

Publish Date: 2013/12/04

Volume: 10, Issue: 4, Pages: 754-769

PDF Link

Abstract

Many untargeted LC–ESI–HRMS based metabolomics studies are still hampered by the large proportion of nonbiological sample derived signals included in the generated raw data Here a novel powerful stable isotope labelling SILbased metabolomics workflow is presented which facilitates global metabolome extraction improved metabolite annotation and metabolome wide internal standardisation IS The general concept is exemplified with two different cultivation variants 1 cocultivation of the plant pathogenic fungi Fusarium graminearum on nonlabelled and highly 13C enriched culture medium and 2 experimental cultivation under native conditions and use of globally U13C labelled biological reference samples as exemplified with maize and wheat Subsequent to LC–HRMS analysis of mixtures of labelled and nonlabelled samples twodimensional data filtering of SIL specific isotopic patterns is performed to better extract truly biological derived signals together with the corresponding number of carbon atoms of each metabolite ion Finally feature pairs are convoluted to feature groups each representing a single metabolite Moreover the correction of unequal matrix effects in different sample types and the improvement of relative metabolite quantification with metabolome wide IS are demonstrated for the F graminearum experiment Data processing employing the presented workflow revealed about 300 SIL derived feature pairs corresponding to 87–135 metabolites in F graminearum samples and around 800 feature pairs corresponding to roughly 350 metabolites in wheat samples SIL assisted IS by the use of globally U13C labelled biological samples reduced the median CV value from 71 to 36  for technical replicates and from 151 to 108  for biological replicates in the respective F graminearum samplesWhile full genome sequences have been determined for many organisms it is currently still not possible to measure the complete metabolite inventory of a biological system due to methodical limitations Complementary sensitive and generic techniques are required to cope with the large chemical diversity and wide dynamic range of low molecular weight metabolites Gas chromatography GC or liquid chromatography LC coupled to mass spectrometry MS as well as nuclear magnetic resonance NMR spectroscopy have emerged as key techniques in the field of metabolomics as recently reviewed by eg Zhang et al 2012 Patti et al 2012b and Zhou et al 2012 The combination of LC with electrospray ionisation ESI high resolution mass spectrometry HRMS has proven to be particularly powerful as this technique enables the detection of a large number of known and unknown metabolites simultaneously and requires only small amounts of the biological sample Hiller et al 2011 Patti et al 2012bTwo different metabolomics concepts can be distinguished targeted and untargeted approaches In targeted approaches a set of predefined known substances is determined thus absolute quantification of those metabolites which are available as authentic reference standards is feasible In contrast untargeted approaches try to find mass spectrometric features of all detectable metabolites including those unknown or at least unidentified at the time of measurement Therefore the untargeted approach has the advantage of probing the entire observable metabolic space and can obtain relative abundances of several hundreds to thousands of metabolites simultaneously Patti et al 2012b For the automated data processing of such LC–HRMS derived metabolomics datasets various workflows and software packages have been developed and are frequently used in untargeted metabolomics studies eg XCMS Smith et al 2006 MzMine Pluskal et al 2010 MetAlign Lommen and Kools 2012 or Maven Clasquin et al 2012 These software tools have in common that they extract as many features as possible from raw LC–HRMS derived metabolomics data sets In this respect the term feature has been defined to be a bounded two dimensional LC–HRMS signal consisting of a chromatographic peak ie retention time and a MS signal m/z value Kuhl et al 2012Despite the recent advances regarding both LC–HRMS instrumentation and data handling platforms the comprehensive annotation of the metabolome of a biological sample of interest and subsequent metabolite identification still remain the major bottlenecks in untargeted metabolomics especially for LCESIHRMS based studies Scalbert et al 2009 Castillo et al 2011 Patti et al 2012b Theodoridis et al 2012 Dunn et al 2013 This limitation can largely be attributed to the generic nature of the ESI process unavoidably leading to LCESIHRMS full scan chromatograms and spectra containing a large proportion of background and chemical noise compared to the signals originating from true metabolites Keller et al 2008 Covey et al 2009 Trotzmüller et al 2011 Further challenges arise from the fact that a single metabolite leads to more than one ion species eg isotopologue peaks different adducts insource fragments and even more complex combinations of the previous species In addition many metabolites cannot completely be separated in the chromatographic dimension and therefore LC–HRMS measurements result in mass spectra which contain signals from more than one metaboliteAnother obstacle of untargeted LCESIHRMS based metabolomics is related to relative quantification of the detected metabolite ions which is caused by so called matrix effects The composition of the evaporated sample at any time point of the LC–HRMS measurement can significantly influence the ionization efficiency and leads to ion suppression or ion enhancement in the ESI source of the mass spectrometer Tang and Kebarle 1993 King et al 2000 Matrix effects can seriously affect signal intensities as well as precision and even limit the coverage of the metabolome Vogeser and Seger 2010 Koal and Deigner 2010 They are difficult to overcome in global untargeted studies as the matrix is composed of the biological sample itself Thus except protein precipitation sample purification is generally not a suitable option as this would largely discriminate many sample constituents of interest Tulipani et al 2013 Moreover the availability of appropriate internal standards is often limited The detailed and comprehensive study of matrix effects is laborious and challenging thus only a few studies reported the systematic evaluation of matrix effects and their limitations on relative metabolite quantification in the field of LC–HRMS based metabolomics Böttcher et al 2007 Redestig et al 2011 Tulipani et al 2013With respect to the above mentioned limitations regarding global annotation of the metabolome and method performance evaluation there is a great demand for both innovative approaches for the analytical measurement of biological samples with LC–HRMS as well as the development of novel improved data processing algorithmsStable isotope labelling SIL is a technique which is becoming increasingly used in different areas of metabolomics research and it shows the potential to conquer many of the elucidated limitations in untargeted metabolomics research In this respect SIL assisted experiments employ stable isotopes of elements such as carbon 13C hydrogen 2H oxygen 18O nitrogen 15N and sulphur 34S Klein and Heinzle 2012 Nakabayashi et al 2013 respectively However 13C is used most commonly as the main labelling isotope since carbon is part of virtually any metabolite Nonlabelled partly labelled and highly 98  13C enriched U13C metabolites show the same physicochemical properties and therefore are not separated by chromatography but can easily be distinguished by their mass to charge ratio m/z using an MS instrument

Keywords: