7 Why take this approach – value? Improved prediction of sensory propertiesBetter product characterizationDiscovery - statistically linking stimuli to perceptionNew contributors to perceptionUnderstanding of pathways leading to stimuli
8 Why now?Developments in “omics” are driving the development (and availability) of instrumentation, approaches, and data handling and analysis.Our UniversityTwo new metabolomics facultyOver $2,000,000 in advanced MS and some nmr instrumentationStaffing with data handling/analysis experts from Super Computing Center
9 Presentation What is metabolomics bringing us? Sample preparation/isolation for analysisData collection (instrumentation)Data handlingData analysisMany similar challenges we face
10 Preparation/Isolation Volatiles – not much help.We recognize the limitations of any extraction/isolation methodHelp with instrumentation for volatiles
11 Non-volatiles – helping us Searching for “best” method for us.Going through a host of methods evaluating each for sensitivity and breath
12 Non-volatiles – common approach Solvent extraction of solid tissuesMechanical disruption of the tissue (grinding, vibration or other methods on a frozen sample.)Solvent selection varies widely with compounds of interest;polar compounds being best extracted with isopropanol, ethanol, methanol, acidic methanol, acetonitrile, water, and methanol:water.Non-polar compounds are most often extracted with chloroform or ethyl acetate(Dettmer et al., 2007).
13 Polar substances in potatoes Polar substances – (potatoes) best extraction methodmethanol and heating (for enzyme deactivation)Methoxylation of sugars and silylationGC-MS profiling resulted in 150 polar compounds, 77 of which were identified(Roessner et al. 2000)
14 Secondary plant metabolites Extraction with acidified aqueous methanol. (75% methanol, 0.1% formic acid, water from the sample or added as necessary).Key factor is simplicityDe Vos et al. (2007)
15 Analysis MS –GC/GC, LC (UPLC), direct analysis nmr Long history of application of basic techniques to foodsImpressive evolution in methodsnmrApplied to foods in related work for more than two decades.Historically, emphasis was to detect adulteration and fraudA small quantity of each sample was centrifuged and 0.1 ml D2O(for NMR ®eld/frequency lock) was added to 1ml ofeach supernatant. A 0.1% solution of TSP (sodium saltof 3-(trimethylsilyl)propionic-2,2,3,3-d4 acid) in D2Owas added to 0.6 ml of each supernatant in a 5mm o.d.NMR tube, to give a ®nal TSP concentration of 0.01%.
16 GCxGC Better resolution Better sensitivity – no back ground Better quantitative dataGoing from unique use to standard method
17 GC-MS Broad use of TOF instruments Improved sensitivity (4X vs. quad)Fast scans – permits peak deconvolution (peaks with 1 sec difference in peak apex)Also minor peaks not well separated from major peaksAccurate mass at fast scanLeco – GC/MS and GCxGC/MS - low resolution MS but peak deconvolution due to high sampling rateThermo – triple quad also high resolution magnetic sectorWaters – GCxGC with accurate mass TOF
18 UPLC or Capillary LCUPLC – sub -2μm stationary phase and high linear velocity of the mobile phaseFaster analysis (high throughput)Better peak resolutionHigher peak capacityMust be interfaced with fast MS – often TOF(Plumb et al., 2005).
19 Ambient ionization – gas analysis PTR-MS or API-MSIdeally suited to providing an ion profile of a gaseous sampleMany current applications in flavor chemistry
20 Ambient Ionization (AI) Methods Ionization MethodIonizing AgentSample StateDesorption ElectroSpray Ionization (DESI)Charged droplets/ionsCondensed phase, gas phaseDesorption Atmospheric Pressure Chemical Ionization (DAPCI)Primary ions via corona discharge or charged dropletsDirect Ionization in Real Time (DART)Metastable atomsAtmospheric Pressure Matrix Assisted Laser Desortion Ionization (AP-MALDI)Laser irradiation/MatrixCondensed phaseAtmospheric Pressure Solids Ionization Probe (ASAP)Primary ions via corona dischargeElectrospray Laser Desorption Ionization (ELDI)Laser irradiation/Charged dropletsIonization methods allow for analysis of samples under ambient conditionsMany AI methods require no sample preparationCurrently, AI methods are receiving considerable attention
21 inlet of mass spectrometer DESI InstrumentationImplementation of DESIDESI spray head – generation of charged droplets/primary ionsAtmospheric pressure ionization compatible mass spectrometer – mass analysis of generated secondary ionsSome ancillary equipment/supplies – either necessary for experiment or convenienceTo date, majority of DESI data has been generated using linear quadrupole ion trapsDESI has been demonstrated on nearly all mass analyzersSeveral different API interfacesabsurfacedt-sgas jetspray capillaryspraynebulizer capillaryinlet of mass spectrometersampledesorbed ionssolventN2HV
22 Example - Detection of lysozyme DESI spectrum of lysozyme present on PTFE surface; average surface concentration 50 ng/cm2Microbial contamination on food processing surfaces?
23 nmrMS offers sensitivity and capacity to detect compounds in mixtures but are limited to ionizable species, have difficulties resolving isomers, and usually require standard compounds for quantification.NMR has the capacity to characterize chemical structure and quantity but is limited to the most abundant compounds in a given sample without isotope labeling. (several other limitations)Quantitative and reproducible – statistical analysisSensitivity not limited by same factors as MSHigh throughput – 500 samples per dayHegeman et al. Anal. Chem. 2007, 79, , Hegeman for isotope labeling studies; (Pan and Raftery, 2007; (Lindon et al., 2004There are some issues with the evaluation of biofluids with NMR. Solvent and pH of the samples have an effect on the chemical shift seen in the NMR signal and thus consistent sample preparation is extremely important. Intermolecular interactions such as hydrogen bonding and the presence of metal ions also contribute to shifts in the spectra (Defernez and Colquhoun, 2003). An excessive amount of water present in a biological sample can result in a large water peak that hides other compound peaks. A water suppression pulse sequence can be used to circumvent this problem (Lindon et al., 2004). A problem with NMR that is also seen with MS methods is that the large number of compounds present in the sample may lead to over-lap in some areas with certain compounds being neglected in the analysis as a result. NMR has the additional disadvantage that it is less sensitive than other methods; it has a lower detection limit of 10μM (Pan and Raftery, 2007).
24 Simple sample preparation Freeze sampleFreeze dryReconstitute in 80:20 D2O:CD3OD containing 0.05% w/v TSP-d4 (sodium salt of trimethylsilylpropionic acid)Sample heated (50C – 10 min)Micro centrifugationWard J, Harris C, Lewis J, Beale MH, Phytochemistry 62 (2003) 949–957
25 Data handling/Analysis Multivariate statistical projection methods (partial least squares, principle component analysis) are commonly used (as starting point)Lend themselves well to biological data because of their ability to correlate multiple variables in a robust and easily interpretable fashion(Jonsson et al., 2005).Partial least squares (PLS) is a data analysis technique that can relate the independent variables of the samples (type of sample, specific sample characteristics, etc.) to the instrumental response. Clusters of responses seen from PLS can be further differentiated using a technique called discriminant analysis (DA) (Lindon et al., 2004). DA is a categorical technique which uses an algorithm to project data into clusters and can give an indication of which parameters exhibit importance (Hollywood et al., 2006).Principle component analysis (PCA) is a multivariate data handling technique that is valuable in the analysis of large amounts of instrumental data; it serves to locate patterns and trends in large data sets (Jonsson et al., 2005). The amount of variability in a sample is explained by principle components. The first principle component accounts for as much variability as possible and each principle component thereafter further accounts for the variability (Lindon, Holmes, et al., 2004). Jonsson et al. (2005) were able to use PCA in conjunction with PLS to differentiate different types of mice using LC-MS. They developed predictive models to screen large sets of samples. Chen et al. (2006) used PCA to combine NMR and DESI-MS to obtain a 3D score plot which potentially could help differentiate large data sets. Belton et al. (1998) used PCA of NMR data to differentiate apple juice from different cultivars.
26 Statistical heterospectroscopy (SHY) New technique used to correlate nmr data with UPLC-MS results by cross-assigning the signals.(Crockford et al., 2006). (Pan and Raftery, 2007).Technique used to combine nmr with DESI-MS to correlate known biomarkers with specific metabolites(Pan et al., 2007).CARA
27 MetAlignTMUsed in numerous publications for data extraction from GC-MS and LC-MS data sets.Pulls out all of the masses and sorts which are the same in all of the data sets and which could differentiate the data.Allows the user to look for unique peaks in the sample set above a chosen noise threshold(Lomen et al., 2006).
28 MetAlign“MetAlign is a software program for full scan LC-MS and GC-MS comparisons and was designed and written by Arjen Lommen of RIKILT-Institute of Food Safety. It has been extensively tested in collaboration with Plant Research International.”
29 Commercial packagesFor example – Marker Lynx (Waters), Xaminer™ Thermo and ?
30 Umetrics SIMCA-PSelect compounds to focus efforts on – not try to identify everything
31 IdentificationsMost existing metabolite libraries are either proprietary, insufficiently comprehensive, collected under non-standardized conditions or unsearchable by computers.Exceptions:Human Metabolome Database (HMDB; (>20,000 metabolites)and
32 Madison Metabolomics Consortium (MMC) Database (MMCD;nmrfam.wisc.edu/), a web-based tool thatcontains data pertaining to biologicallyrelevant small molecules from a variety ofspecies.
33 ConclusionsNot much help in isolation methods for volatiles – information on semi or non—volatilesNew instrumentation – GC/GC, MS and sample interfaces, nmr, LC-MSNew tools for data handling and analysisAlso combining instrument data e.g. nmr w/ MSEstablishing public databases
39 Figure 1. GC-MS-based metabolomics Figure 1. GC-MS-based metabolomics. A, Analytical approach used B, Conventional approach. C, Alternative, unbiased approach to GCMS data analysis.Tikunov Y, Lommen A, Ric de Vos CH, Verhoeven HA, Bino RJ, Hall RD, Bovy AJ Plant Physiology, November 2005, Vol. 139, pp. 1125–1137
47 Strawberry metabolites Fourier Transform Ion Cyclotron Mass Spectrometry (FTMS)FTMS - only MS system capable of routinely achieving ultra high resolution at high acquisition rate (100–1,000 amu scan/sec) allowing multiple scans in (1–2 min).Separation of the metabolites achieved solely by ultra-high mass resolution, eliminating the need for time consuming chromatography and derivatization.Aarón A, Ric De Vos, Verhoeven HA, Maliepaard CA, Kruppa G, Bino R, Goodenowe DB. Omics 6(3), 2002 ( )
48 Method 4 stages of ripeness for berries Extraction –50/50 MeOH/0.1% formic acid or 100% acetonitrile (AN)Direct injection into FTMSMS ionization - electrospray (ESI + or -) or atmospheric pressure chemical ionization (APCI + or -)
49 Strawberry metabolite 1ppm mass accuracy ppm mass accuracy(2 possible only 1 is logical)
50 Resulttotal of 5,250 unique 12C masses were obtained from extracts of the four different developmental fruit stages, ionized in the four ionization modes.In the red stage extract, 55% of the masses were assigned a single empirical formula, 10% two formulae, and 35% three or more formulae.Went to data databases for help - 159,000 natural products (Chapman and Hall, Dictionary of Natural Products)
52 Next stepsLook for changes or relationships of data to some attribute via multivariate statisticsIdentify the compounds of interest.HPLC, mass spectrometry (MS/MS) and/or nuclear magnetic resonance (NMR) must be employed.Evaluate result for value