“Metabolic Profiling: Limitations, Challenges , Perspectives for the Analytical Chemist ”

1. “Metabolic Profiling: Limitations, Challenges, Perspectives for the Analytical Chemist ” Georgios Theodoridis Dept. Chemistry Aristotle University Thessaloniki

2. AUTh bioAnalytical group Fundamental/Developmental work ‘’Standardising Metabolomics’’ Excellence Grant GSRT • Validation • New Methods (Targeted, Untargeted) DOES IT WORK? • New Chromatographic Materials Clinical Studies • Rheumatoid Arthritis Fleming Institute Prof. G. Kollias • Physical Exercise Prof. V. MougiosAUTh • Frailty/Ageing Prof. V. MougiosAUTh • EmbryoMetabolomicshttp://www.embryometabolomics.eu/ • Sepsis/NEC newborns with Hippokrateion Hosp. Intens. Care Unit

3. Systems Biology and Metabolomics ‘’the systematic study of the uniquechemical fingerprints that specific cellular processesleave behind’’ Holistic Analysis of small molecules

4. Source: Considerations in the design of clinical and epidemiological metabolic phenotyping studies G Theodoridis et al 2013, ebook Metabolic profiling in clinical applications. doi:10.4155/EBO.13.487

5. Analytical focus study design sample collection sample prep analytical procedure analysis data extraction Data mining, chemometrics data analysis biomarkers IDs Develop specific assay

6. Bottlenecks in analytical procedure • Wide spectrum of analytes (unlike genomics) • Huge span in concentration: 7 orders of magnitude • MS: Different instrumentation architecture • Need for long analytical batches • clean up steps : when? Can I combine data? • Instrument calibration along the run: DISASTER ! • LC-MS instrumentation variability: Drifts in Rt, mass, sensitivity • Ionisation in Mass Spectrometry not controlled • Lack of LC-MS spectral libraries

7. Bottlenecks in data treatment • Big datasets • Impractical to correlate-combine data • Various peak picking and treatment algorithms • data repositories and databases still immature • metID(>4 years trying to identify candidate markers, G. Patti, Bioanalysis 2012)

8. Major Problems • Analytical Chemists, Informaticians, Chemometricians, Biochemists still speak different language • Fragmentation of research • Genomics labs can split tasks /Metabolomics labs can’t trust other peoples results

9. Way to go?Standardization & Harmonisation Establishing SOPs • Data quality, QC procedures • Instrument performance and maintenance • Sample collection/storage • Sample treatment • Data acquisition protocols • Data manipulation • Reporting

10. QC procedures • How can we validate a metabolic profiling method when we don’t know the analytes that will be analysed? • How can precision and reproducibility be assessed when we don’t know what we are measuring? • How can we report data quality? • What analytical protocols should be adapted ? • Which method is good?

11. QC procedures • Integration of “classical” analytical strategies • with unbiased data analysis • Implementation of QC • Pooled sample, Injected in-between samples • Synthetic mixturesinjections • Randomisation of injection order • Technical replicates and other measures…

12. QC strategy: example 1 Raw data, TIC across all samples QC samples Sensitivity drift

13. Example 2 day-to day effect Gika et al Bioanalysis 2012

14. QC roadmap Gika et al J Proteome Res 2007

15. Aim 1: New analytical methodologies • method robustness • extraction efficiency • metabolome coverage • Profiling methods with complementary/orthogonal selectivities Sampsonidis P2-04 • Protocols for sample extraction Optimization studies on extraction of samples, (e.g. different pH values, organic solvent composition, mass tovolume ratio) • HILIC/MS-MS for quantitative determination of ca. 140primary metabolites • Implementation of other HILIC chemistries eg zwitterionic, diol, RP-WAX • Computational approach for column selection for metabolic profiling

16. Ion Pair MS/MS (1) glutamine, (2) methionine (3)adenine, (4) thymine, (5) inosine, (6) glutamic acid, (7) phenylalanine, (8) aspartic acid, (9) glucuronic acid, (10) tryptophan, (11) lactic acid, (12) galactose 1P, (13) xylulose5P, (14) pyruvic acid, (15) NAD, (16) UMP, (17) GMP, (18) AMP, (19) maleic acid, (20) phosphocreatine, (21) malic acid, (22) a-ketoglutarate, (23) G3P, (24) NADP, (25)FBP,(26) isocitric acid, (27) dCTP, (28) ATP, (29) acetyl CoA and (30) butyryl CoA. Michopoulos et al J Chromatogr A 2014

17. Sample Preparation and Stability • Blood: • Dried Blood Spots (F. MichopoulosBioanalysis 2012, F. Michopoulos J Proteome Res 2011) • Turbulent Flow Chromatography (J. Sep. Sci. 2010), • protein precipitation (J Proteome Res 2011) … • Urine: • Urine stability over freezing and freeze thaw cycles (J Chromatogr A 2008) • Protocols: Nature Protocols Want et al 2010 urine, 2013 tissue

18. Sample prep: An example

19. Num of features detected extraction mixtures(MeOH-H2O-CHCl3) RP TOF-MS Organic extract features detected in 3 extracts for 5 mixtures HILIC TOF-MS Aqueous extract

20. Analyte Distributionin fractions a) CHCl3b) H2Oc)MeOH

21. Aim 2: Data extraction • Evaluationof various data extraction software free and commercial: XCMS, MarkerLynx, MarkerView, Profiler and others in metabonomicsstudies • Spikingexperiments (comparisonof sensitivity and reliability of thedata treatment software) A. Pechlivanis, MSc Study 2009, AUTh • Intranet platform for the extraction ofinformation from MS-profiling data (rules for monitoring and reporting thevarious alterations and parameter selection to improve standardization indata extraction and reporting)

22. Aim 3: Quality Control and standardisation protocols • Scripts for QC in holistic MS data • Examine data in depth and applying rules by automated scripts (Matlab and R) • Correction for sensitivity loss (?scaling?) • Correction for retention timedrift to improve peak alignment in feature detectionZelena et al Anal Chem 2009

23. Aim 4: Data fusion • Software tools to fuse data from different methods LC-MS/MS+ GC-MS LC-MS/MS + NMR HILIC-MS + RPLC-MS +evi ESi/ -evi ESI • link data • combine into one table of features or metabolites (?)

24. Aim 5: Metabolite Identification MetID the major bottleneck in LC-MS metabonomics • scripts for adduct identification to reduce the number of detected features : +Na+, + NH4+ , dimers etc • MS spectra by analysis of standards (in-house MS database). • Scripts for automated searchesin local and internet-based spectral/biochemistry libraries. • Compare isotope patterns between peaks in samples and standards

25. Aim 6 : Retention Time Prediction • IncorporatingRt data to assists MetID • Use of data from orthogonal chromatographic systems: chemical information (polarity, LogP etc) • Rule out candidate IDs • Retention time predictionalgorithm in HILIC Gika et al Anal Bioanal Chem 2012, Gika et al J. Sep. Sci 2011 Fasoula OP12, P2-03 • software to organise the necessary analyses and data treatment for metID within an easy to use platform.

26. Disease Drug efficacy Toxicity What do metabolomics offer ? Bio-Markers Biochemistry insight Clinical symptoms Onset of disease Diagnosis/ therapeutic intervention Potential for the discovery of biomarkers Additional knowledge of the biochemical pathways

27. Perspective • Metabolites downstream the biochemical pathway compared to genes, proteins, closer to phenotype • Can describe effects of xenobiotics (e.g. pharmaceuticals) and host-guest interactions (e.g mammals with gut microflora) • Describes ongoing phenomena

28. An expanding field

29. Call • Metabolomics is analytically dependent • Metabolomis grows and provides openings for analytical chemists

30. The group Auth • Dr. G. Theodoridis • Dr. H. Gika • Prof. A. Papa • Dr. N. Raikos • C. Virgiliou MSc • O. Deda MSc • Dr. C. Zisi • S. Fasoula MSc • A. C. Hatzioannou MSc • D. Palachanis MSc • I. Sampsonidis MSc External collaborators • I. D. Wilson Imperial college London UK • P. Vorkas Imperial college London UK • P. Francheshi IASMA Trento Italy Funding