LC-MS-based Metabolomics: Workflows, Strategies and Challenges

Transcription

1 LC-MS-based Metabolomics: Workflows, Strategies and Challenges Presented by: Sponsored by:

2 Dr. Clary Clish Director of the Metabolite Profiling Platform at the Broad Institute of MIT and Harvard

3 LC-MS-based Metabolomics: Workflows, strategies, and challenges Clary B. Clish, Ph.D. June 19, 2013

4 Outline Introduction to metabolomics and LC-MS Challenges to measuring the metabolome Overview of an LC-MS metabolomics workflow - Sample collection and storage - Preparation of LC-MS samples - LC-MS measurements and quality monitoring

5 Metabolite Profiling Metabolite profiling (or metabolomics ): the systematic analysis of endogenous metabolites in biological specimens Metabolomics is relatively new term, 1 but the concept is not new Analysis of metabolites in body fluids using a bioassay dates back to BC 2 Chinese doctors used ants to detect high levels of glucose in urine as indicator for diabetes 1. Oliver SG et al. Trends Biotechnol 1998; 16: van der Greef and Smilde. J Chemometrics 2005; 19:

6 General approach to metabolomics Biological Samples Analytical Profiling Statistics & Data Analyses Scientific Discovery Human & Animal fluids tissues Yeast Signatures of disease/biomarkers Metabolic phenotypes Bacteria Drug activity and efficacy Cultured cells Growth media Discovery of new biological mediators Key features: multiple metabolites signal proportional to [metabolite] precision & robustness univariate and multivariate statistics correlations pathway analyses LC-MS

7 Liquid chromatography: separation of metabolites in complex mixtures Chromatography: separation of mixtures based on differential affinities of substances for a stationary phase and a moving phase Liquid chromatography: - stationary phase is solid material packed into a column - mobile phase is a liquid solvent Biological sample mobile phase inject mixture of metabolites stationary phase eluent Time

8 Varieties of chromatography (not comprehensive) Description Typical stationary Phase Typical mobile phase Analytes Gas Coated capillary column Gas (He, N 2 ) Thin layer silica Organic solvent blends Amino acids, dyes Broad range of small molecules Normal phase silica Organic solvents Broad range of small molecules Reversed phase Alkyl groups bonded to silica or polymeric beads Water/organic solvent blends Best for compounds with hydrophobic groups HILIC Polar functional groups Organic solvent/ water blends Polar analytes Ion exchange Polar functional groups Water Ionic analytes Size exclusion Porous, polymeric beads Water Macromolecules and complexes

9 Varieties of chromatography (not comprehensive) Description Typical stationary Phase Typical mobile phase Analytes Gas Coated capillary column Gas (He, N 2 ) Thin layer silica Organic solvent blends Amino acids, dyes Broad range of small molecules Normal phase silica Organic solvents Broad range of small molecules Reversed phase Alkyl groups bonded to silica or polymeric beads Water/organic solvent blends Best for compounds with hydrophobic groups HILIC Polar functional groups Organic solvent/ water blends Polar analytes Ion exchange Polar functional groups Water Ionic analytes Size exclusion Porous, polymeric beads Water Macromolecules and complexes Separation depends on the combination of stationary and mobile phases used

10 Mass Spectrometry: separation and measurement Example: Triple quadrupole MS (QQQ) LC Eluent Q1 Q2 Q3 Detector Ionized Molecules collision gas (N 2 ) separation selection of parent ion fragmentation generation of product ions selection of specific product ion monitoring parent/product ion transition provides: confidence in identity improvement in signal-to-noise

11 Examples of other MS analyzers Quadrupole Time-of-flight (QTOF) High resolution data (Orbitrap > QTOF >> QQQ) Fast data acquisition rates Orbitrap Not as sensitive as QQQ, but are superior for nontargeted analyses e.g. Exative Plus (Thermo Fisher Scientific) Image from:

12 LC-MS Chromatogram of a plasma extract Biological sample mixture of metabolites Features: Metabolite peaks separated over time Co-eluting peaks separated by mass

13 HILIC, Ion Pairing C18 Polarity C4 Challenge: Differences in physical properties among metabolites Metabolites triglycerides cholesterol esters diacylglycerols sphingomyelins phosphatidylcholines phosphatidylethanolamines other phospholipids fatty acids eicosanoids & metabolites bile acids bilirubin amino acids, amines organic acids sugars Other polar: e.g. purines & pyrimidines oily water soluble Chromatography MS Ionization & or

14 Log MS Signal Intensity Challenge: Range of endogenous metabolite concentrations LC-MS System Analytical Response 10 s-100 s of metabolites measurement over 4 orders of magnitude concentration Concentration Endogenous metabolite concentrations A single analysis can t provide full coverage

15 Biomass, metabolite coverage, and the impact of new technology more Biomass requirement & Sample work up effort less low Endogenous Concentration high

16 Biomass, metabolite coverage, and the impact of new technology more Effect of improved Instrument sensitivity: Biomass requirement & Sample work up effort less low Endogenous Concentration high

17 Challenge: Choosing targeted vs. nontargeted metabolomics Targeted Profiling specific lists of metabolites tuned MS settings QQQ MS optimal sensitivity metabolite IDs are known results limited to specified list no data on novel compounds Nontargeted Profiling (also: discovery, global profiling) Goal: measure all metabolite signals Full scan MS QTOF, FT-ICR, Orbitrap enables serendipity broad metabolite coverage less sensitive analysis Careful QC required for global peak integration and alignment metabolite ID can be incomplete

18 Example of an LC-MS-based metabolomics platform Biological Samples Sample Preparation Chromatography Mass Spectrometry Metabolite Profiles Sample Aliquot Amines & cationic metabolites MeOH/ACN extracts Sugars, organic acids, purines, pyrimidines, etc MeOH extracts Atlantis-HILIC + mode MS NH2-HILIC - mode MS Targeted Nontargeted 100 s of metabolites of known ID validation one 5500 QTRAP QQQ Two 4000 QTRAP QQQ Lipids IPA extracts RP/C4 + mode MS discovery metabolic tracers flux Free fatty acids, bile acids, lipid mediators MeOH extracts RP/C18 - mode MS two Q Exactive Orbitraps one Exactive Plus Orbitrap Targeted: Hundreds of metabolites of known ID Nontargeted: Thousands of LC-MS peaks measured

19 Metabolomics Workflow Study Design IRB/IACUC Approval Experiment & Sample collection Sample Storage & Shipment Sample preparation LC-MS Analyses Data Processing QC Analyses Data Export Data Analyses & Statistics

20 Metabolomics Workflow Study Design IRB/IACUC Approval Experiment & Sample collection Sample Storage & Shipment Sample preparation LC-MS Analyses Data Processing QC Analyses Data Export Data Analyses & Statistics

21 Sample Collection Study Design IRB/IACUC Approval Experiment & Sample collection Sample Storage & Shipment Sample preparation LC-MS Analyses Data Processing QC Analyses Data Export Data Analyses & Statistics

22 Sample collection Body fluids and tissues Goal is to reduce the time between collection and freezing E.g. for plasma: centrifuge blood within 30 minutes of collection, collect and snap freeze plasma; store at -80 C Reference SOPs for urine, serum, and plasma at: Tuck MK et al. Standard Operating Procedures for Serum and Plasma Collection: Early Detection Research Network Consensus Statement. J Proteome Res 2009; 8: Cell culture protocols generally need some degree of customization Matched to experimental protocol Matched to growth characteristics (e.g. adherent vs. suspension culture) Account for cell number and cell size

23 Scaling Adherent Cell Culture 10 cm dish 6 well plate 96 well plate # of cells Volume of Extraction solvent 2e6-5e6 2.5e5 6e ,000 4 ml 800 µl µl Cells/µL 1, Direct analysis? yes yes No

24 Sample Storage & Shipment Study Design IRB/IACUC Approval Experiment & Sample collection Sample Storage & Shipment Sample preparation LC-MS Analyses Data Processing QC Analyses Data Export Data Analyses & Statistics

25 Sample Storage & Shipment Storage Store samples at -80 C or colder Body fluids and tissues can be kept as intact samples Store as intact sample without freeze-thaw Good idea to prepare specific aliquots needed for analyses at time of collection Cell extracts Short term storage: store as extracts in solution Long term storage: evaporate solvents and store dried material Shipment Ship on dry ice overnight If using a courier, schedule pickup for Monday or Tuesday 25

26 Sample Preparation Study Design IRB/IACUC Approval Experiment & Sample collection Sample Storage & Shipment Sample preparation LC-MS Analyses Data Processing QC Analyses Data Export Data Analyses & Statistics

27 LC-MS Sample Preparation Key features of a good protocol: Uses the biological sample efficiently Generates samples in which metabolite concentrations fall within the linear response range of the analytical method The final sample solvent composition is matched to the chromatography method Preserves integrity of metabolites Has as few steps as possible to minimize variance Uses internal standards (IS) to enable monitoring of recoveries and analytical variance generally non-biological small molecules or synthetic, stable isotopelabeled metabolites

28 Protein Precipitation General method: Add an accurately measured volume of organic solvent containing one or more IS to an accurately measured amount of biological sample Allow proteins to precipitate Centrifuge and collect supernatant Analyze supernatant directly (a concentration step can be added if necessary) Features Only two measurement steps introduces little variance Samples are in organic solvent ideal for HILIC methods Highest throughput Lowest cost Limitations Yields diluted metabolite concentrations Minimal selectivity Samples are in organic solvent not ideal for RP chromatography 28

29 Liquid-Liquid Extraction Phase separation of two immiscible solvents Typically used for lipids Vary ratios of chloroform/methanol/water Bligh EG, Dyer WJ. A rapid method of total lipid extraction and purification. Can J Biochem Physiol 1959; 37: Folch J, Lees M, Sloane Stanley GH. A simple method for the isolation and purification of total lipides from animal tissues. J Biol Chem 1957; 226: Features Polar metabolites and lipids from one sample Potentially superior separation of metabolites relative to protein precipitation Limitations Metabolites at interface of separation are missed Solvents not compatible with all plastics Additional care required when pipetting sample from each phase

30 Solid Phase Extraction conditioning sample loading washing elution Used for preparation of specific metabolites Samples loaded onto a solid matrix, washed, and selectively eluted Metabolite Interferences Features Can significantly reduce interferences Enables analyses of low abundance metabolites Reduces impact on analytical performance of LC-MS system Limitations Low throughput Narrows range of metabolites Higher cost Greatest variance Recoveries seldom 100% 30

31 Tissue homogenization Many options: Pulverization Grinding: Mortar and pestle Probe sonication Focused acoustic waves Bead beating Etc. Bead homogenizer Example: one sample at a time high-speed shaking in plastic tubes stainless steel, tungsten carbide, or glass beads 48 samples in 1.5 ml tubes tissue Add 4 volumes of water 4 µl/mg tissue Homogenize Fixed amount of time Aliquot for each method Fixed volume Store or extract mg

32 LC-MS-based metabolomics platform Biological Samples Sample Preparation Chromatography Mass Spectrometry Metabolite Profiles Sample Aliquot Amines & cationic metabolites MeOH/ACN extracts Sugars, organic acids, purines, pyrimidines, etc MeOH extracts Atlantis-HILIC + mode MS NH2-HILIC - mode MS Targeted Nontargeted 100 s of metabolites of known ID validation one 5500 QTRAP QQQ Two 4000 QTRAP QQQ Lipids IPA extracts RP/C4 + mode MS discovery metabolic tracers flux Free fatty acids, bile acids, lipid mediators MeOH extracts RP/C18 - mode MS two Q Exactive Orbitraps one Exactive Plus Orbitrap Targeted: Hundreds of metabolites of known ID Nontargeted: Thousands of LC-MS peaks measured

33 Body Fluid/Homogenate/Media LC-MS Samples Body fluid/ Homogenate/media Extraction solvent Lipids Isopropanol extract 10 µl 190 µl isopropanol vortex Free fatty acids/ Bile acids Methanol extract 30 µl 90 µl methanol Nitrogenous compounds, cationic species MeOH/ACN extract 10 µl 90 µl 75/25 acetonitrile/methanol centrifuge Sugars, organic acids, purines, pyrimidines, etc MeOH extract 30 µl 120 µl 80% methanol transfer supernatant to AS vial

34 LC-MS Analyses and QC Study Design IRB/IACUC Approval Experiment & Sample collection Sample Storage & Shipment Sample preparation LC-MS Analyses Data Processing QC Analyses Data Export Data Analyses & Statistics

35 LC-MS Analyses Method throughput: typically minutes per analysis Chromatography Timeline elution wash Re-equilibration time Queue size Based on the robustness of a particular method Our methods: injections per column Samples processed in batches (generally samples/batch) Queue paused periodically to clean the MS Monitor internal standards and reference samples during the queue Samples randomized in queue Reference samples: synthetic mixtures and pooled samples

36 Mobile Phase Composi on (%) Hydrophilic Interaction Liquid Chromatography (HILIC)-MS Time (min) % A % B Flow rate (µl/min) Time (min) % A % B mobile phase A: 10 mm ammonium formate and 0.1% formic acid in water mobile phase B: acetonitrile with 0.1% formic acid 150 x 2 mm Atlantis HILIC column (Waters) Samples in 10/67.5/22.5/0.2 v/v/v/v water/acetonitrile/methanol/formic acid containing stable isotope-labeled internal standards (valine-d8 and phenylalanine-d8) MS: targeted analysis using an 4000 QTRAP QQQ

37 Mean LC-MS Peak Area 2830-B 5365-A 5308-A 2682-B 6474-B 3627-A 2939-A 7191-B pooled plasma A-a 1181-B 7198-B 2941-B 4041-A 4509-B 4920-A 3393-A 4160-A 1980-A 2324-A 4871-B 7306-B 1558-B 7292-A 2127-A 6129-A 7740-B 78-A 3988-B 172-B 3849-A 7560-A 930-A 3763-B 4229-B 3213-B 7830-A 5244-A 6878-A pooled plasma B 7206-A 6602-B 6659-B 3675-A 1659-B 1133-B 662-B 7319-B Mean LC-MS Peak Peak Area Area plasma_ B 7100-B 5365-B 3436-A 2682-A 1864-A 3627-B 7005-B 7191-A 1610-B 1624-A 1181-A 6906-A 2941-A 671-B 4509-A 2660-B 3393-B 246-B 1980-B 2324-B 920-B 7306-A 7434-A 7292-B 3759-A 6129-B 144-A 78-B 8075-B plasma_ B 2331-B 930-B 5261-A 4229-A 7029-B 7830-B 4315-B 6878-B 4024-A 2826-A 7206-B 1280-B 6659-A 7255-B 1659-A 7177-A 662-A 1060-A plasma_6 LC-M 1.50E+08 C12:0,C13:0 PC (IS) 1.00E plasma samples 5.00E+07 measured in five batches of 150 samples: 0.00E+00 Internal standard performance within batch and among five batches Batch 5 CV Valine-d8 9% 4.50E E E E E E E E E E E E E E+05 Lipid Method Internal Standards HILIC Method >750 Internal samples, Standards 5 weeks Batch 5, approx. 150 samples C17:1 LPC (IS) C12:0,C13:0 Valine d8 (IS) PC (IS) 1.00E E E E E+00 Batch1 Batch2 Batch3 Batch4 Batch5 Mean CV of Val-d8 over five batches (five weeks) 9.00E E E E E E E+05 HILIC Method Internal Standard >750 samples, 5 weeks Valine d8 (IS) 2.00E E E+00 Batch1 Batch2 Batch3 Batch4 Batch5 Drift in instrument sensitivity over time Normalize data to IS values = mean IS peak area per batch ± s.d.

38 Quantitation Raw LC-MS data are unitless, peak area values Can be transformed to units of concentration using calibration curves (best) or by calculating ratios relative to reference standards (single concentration points) Challenges to absolute quantitation for all metabolites: Lack of availability of reference compounds for all metabolites Cost/time constraints Our lab does mainly do relative quantitation Works very well for projects completed in single period Measured values on the instrument can change over time due physical changes in the MS detector: need to mitigate such changes for long term projects Question: How can we address differences in instrument sensitivity across batches?

39 Pooled reference samples used for standardization Generate a reference sample that is of the same type as the study samples Body fluids: create a pooled sample and make aliquots Cell extracts: pool cell extracts, aliquot, and dry down samples for future use Analyze an aliquot of the pooled sample after every 10 or 20 study samples Normalize relative to nearest pooled reference: peak area(study sample) peak area(nearest pooled reference) One may quantify the concentration of significant metabolites in the pooled reference as needed

40 Pooled reference samples can also be used to assess overall quality Example: study of 865 plasma samples >950 samples including pooled plasma reference samples QA/QC: Reference mixtures analyzed before and after study samples to assure system performance Internal standard added in first step of sample extraction monitored during analyses may be used to normalize data Pooled study samples analyzed every 20 study samples used to normalize data across datasets Second pooled plasma reference samples analyzed every 20 study samples used to assess overall reproducibility assess impact of normalization procedures PP study PP Ref Study samples (20) PP study PP Ref Study samples (20) PP study PP Ref Study samples (20) PP study PP Ref

41 Mobile Phase Composi on (%) Lipid LC-MS 120 Time (min) % A % B Flow rate (µl/min) Time (min) % A % B mobile phase A: 95:5:0.1 vol/vol/vol 10mM ammonium acetate/methanol/acetic acid mobile phase B: 99.9:0.1 vol/vol methanol/acetic acid 150 x 3.0 mm Prosphere HP C4 column (Grace, Columbia, MD) Samples in 1/19 v/v water/isopropanol containing 1-dodecanoyl-2-tridecanoyl-snglycero-3-phosphocholine internal standard MS: full scan analysis using an Orbitrap

42 Coefficients of variation for 133 identified lipids in Reference Pooled Plasma Raw Normalized 43 PP samples Measured over ~ 4weeks CV RAW Normalized > 5% 2% 15% > 10% 71% 74% > 15% 90% 93% > 20% 96% 97% > 100% 100% 100%

43 PP1 PP2 PP3 PP4 PP5 PP6 PP7 PP8 PP9 PP10 PP11 PP12 PP13 PP14 PP15 PP16 PP17 PP18 PP19 PP20 PP21 PP22 PP23 PP24 PP25 PP26 PP27 PP28 PP29 PP30 PP31 PP32 PP33 PP34 PP35 PP36 PP37 PP38 PP39 PP40 PP41 PP42 PP43 LC-MS Peak Area PP1 PP2 PP3 PP4 PP5 PP6 PP7 PP8 PP9 PP10 PP11 PP12 PP13 PP14 PP15 PP16 PP17 PP18 PP19 PP20 PP21 PP22 PP23 PP24 PP25 PP26 PP27 PP28 PP29 PP30 PP31 PP32 PP33 PP34 PP35 PP36 PP37 PP38 PP39 PP40 PP41 PP42 PP43 LC-MS Peak Area Variance in an endogenous lipid signal in pooled plasma over ~4 weeks changed ion transfer capillary 1.4E E E E E E E E+00 Pre-norm C16:0 C16:0 LPC CV = 8.0% 1.2E E E E E E E+00 Post-norm CV = 3.1% C16:0 C16:0 LPC LPC

44 Summary and concluding remarks Metabolites have a broad range of physical characteristics and broad coverage of the metabolome requires multiple LC-MS methods The dynamic range of a typical LC-MS instrument spans four orders of magnitude so the coverage of metabolites depends on the amount of sample available and the specific work up Multiple instrumentation options are available for LC-MS: - Triple quadrupole MS are typically used for targeted analyses - High resolution instruments like QTOFs and Orbitraps may be used for nontargeted profiling - Consequence of varied options: methods are generally not standardized across labs Use of internal standards and reference samples is very important for assessing quality and normalizing data in larger studies Methods and capabilities continue to evolve with the development of new technologies