11.5.2016, 13:15, 15 + 5 min UHPLC-ESI-QTF-MS/MS based Lipidomics by Data-Independent Acquisition Michael Lämmerhofer, Jörg Schlotterbeck Institute of Pharmaceutical Sciences, Pharmaceutical (Bio-)Analysis, University of Tübingen, Germany Analytica 2016, May 11, 2016, Munich (Germany) Sciex Innovation Day
Endogenous sources Mitochondria Peroxisomes Lipoxygenases NADPH oxidases Cytochrome P450 Antioxidant defences Enzymatic systems CAT, SD, GPx Non-enzymatic systems Glutathione Vitamins (A,E,C) Sources and Cellular Responses to Reactive xygen Species (RS) Exogenous sources Ultraviolet light Ionizing radiation Chemotherpeutics Inflammatory cytokines Environmental toxins less. N2 2. R. N RS. 2 -.H H 2 2 more The etiology of every major disease has been linked to oxidative stress Protein Modification Impaired physiological function Homeostasis Impaired physiological function DNA Damage Decreased cellular proliferation Defective host defences Normal growth and metabolism Random cellular damage Aging Disease Specific signalling pathways Cell death Lipid Peroxidation Source: Toren Finkel and Nikki J. Holbrook, Nature 408, 239-247(9 November 2000) doi:10.1038/35041687
Molecular Complexity of xidized Phospholipids (xpls) sn-2 PUFA R CH 3 +R. R CH 3 R CH 3 Arachidonate + 2 Peroxyl radical Ȯ Hydroperoxide H R = Lyso-PC, lysope etc xidized fatty acyl Cyclization, xidation, rearrangement xidative fragmentation (Hock cleavage, -scission) -H -H PLs with multiple oxidation sites / functional groups H H Isoprostanes Isolevuglandins Isothromboxanes H Isofurans H H H H H H H, -Unsaturated H Saturated Furan-containing PC, Phosphocholine Advanced lipidoxidation endproducts (ALE): Malondialdehyde Non-fragmented xpls Fragmented xpls 4-Hydroxynonenal Bochkov et al. ANTIXIDANTS & REDX SIGNALING 2010, 12, 1009-10053
Lipid Analysis Workflows (Biological) Lipid Sample Sample prep Lipid Extraction (e.g. Bligh & Dyer) (SPE) (group selective) TripleTF 5600+ (Sciex) Untargeted Targeted Combined Targeted / Untargeted Shotgun MS/MS All DDA (IDA) DIA (SWATH) MRM HR SWATH LipidView IDA Explorer MS-DIAL MasterView (Target List) PeakView / MultiQuant MasterView MS-DIAL PeakView MultiQuant MarkerView (Sciex) (PCA, t-test) SIMCA (Umetrics) (PCA, PLS, PLS-DA, PLS-DA) 4
I n t e n s i t y Data-Dependent Acquisition Data-Dependent Acquisition (DDA) Most intensive precursor ions detected in survey scan are selected for isolation and fragmentation in a serial manner Performance declines with complexity of samples Limited reproducibility (stochastic precursor selection) Bias towards high abundant analytes Low sensitivity Missed trigger for fragmentation Limited analyte coverage XIC of PNPC 650.439 +/- 0.010 Da MS/MS triggered at RT = 12.1281 100% 50% 0% 12.1 10.0 10.5 11.0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 Experiment 16/20 +TF MS^2 (30-1000) from 12.128 min Precursor: 650.4 Da, CE: 20.0 600 400 200 0 Time, min C5H15N4P+ / Phosphocholine (Δm = 0.2 ppm) 184.0733 200 400 600 800 Mass/Charge, Da 650.4428 184.07332 C33H64N9P / PNPC (Δm = 5 ppm) Information-dependent acquisition (IDA) 5 2016 Universität Tübingen
m/z UHPLC-MS/MS Method with DIA Data-Independent Acquisition (DIA) Fragmentation of all detectable precursor ions systematically parallelized 1000 Q1 width = 25 Da 50 Cycle time ~ 1.s SWATH Retention time Improved reproducibility for identification Better sensitivity Improved analyte coverage Composite/multiplexed fragment ion spectra DIA (SWATH): Quantification in MS/MS mode (sensitivity and specificity gain) Retrospective data processing Data processing more elaborate, but SW for automated metabolite ID is being developed SWATH: Sequential windowed acquisition of all theoretical fragment ion mass spectra
min min Heatmap IDA 0 ESI - Untargeted Lipidomics ESI + 0 5 5 10 10 15 15 20 20 25 25 30 30 1000 800 600 400 200 0 200 400 600 800 1000 m/z Intensity 500-2000 Intensity 2000-5000 Intensity 5000-10000 Intensity 10000-30000 Intensity >30000 Intensity 500-2000 Intensity 2000-5000 Intensity 5000-10000 Intensity 10000-30000 Intensity > 30000 7 Jörg Schlotterbeck 2016 Universität Tübingen
SWATH Size / Da m/z SWATH 2.0 Calculation (swathtuner) Calculation of variable SWATH windows Enhanced peak finding 90 80 70 60 50 40 30 20 10 0 ptimized SWATH Windows 8 PGPC PAzPC PVPC PNPC 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 Experiment No. + Targets 5 Da Two methods for generating variable windows: Equal precursor ion population (PIP) Equalized precursor total ion current (TIC) G. Hopfgartner et al., J. Proteome Res., 2015 Window size normalized to peaks per window 1000 900 800 700 600 500 400 300 200 100 0 LC-MS LPC PC 0 5 10 15 20 25 30 35 min Improvement of MS/MS spectra quality, esp. of compounds with low intensity Improved identification score Improved quantitative reproducibility
MS-DIAL: Data-independent MS/MS deconvolution pen-source software pipeline (for data processing & automated metabolite identification) a) b) c) d) 9 liver Fiehn & Masanori Arita, et al., Nature Methods, 2015, doi:10.1038/nmeth.3393
Mass accuracy in all samples within 4 mda Benefits of MS-DIAL User-friendly display of results Alignment results can be viewed Deconvoluted spectra with better spectral quality Deconvoluted MS/MS chromatogram with enhanced assay specificity Automated lipid identification (of lipids in LipidBlast database) Supporting documentation (no black box system) Peak in all samples within 0.04 min 10 Joerg Schlotterbeck 2016 Universität Tübingen
Acute Coronary Stable Angina Syndrom (STEMI) Pectoris Control ACS 1 SAP K2 R1 ACS 2 SAP K3 R2 ACS 4 SAP K4 R3 ACS 5 SAP K5 R4 ACS 6 SAP K6 R5 ACS 7 SAP K7 R6 ACS 8 SAP K8 R7 ACS 9 SAP K9 R8 ACS 10 SAP K10 R9 ACS 11 SAP K11 R10 ACS 12 ACS 13 ACS 14 Analysis in randomized run order QC sample every 5th injection to validate stable method performance Analysis in positive and negative ionisation mode Software based database search with over 40000 lipid entries (LipidMaps) or MS- DIAL (LipidBlast) 7000-8000 features per sample 11 Jörg Schlotterbeck Platelet Lipidomics QC QC QC ACS 8 SAP K11 ACS 4 ACS 10 ACS 9 QC R10 ACS 12 ACS 11 R8 R1 QC R7 ACS 1 SAP K3 ACS 6 SAP K2 R9 QC SAP K8 R4 R5 SAP K10 SAP K6 QC ACS 5 SAP K9 ACS 14 ACS 13 SAP K7 QC R2 R6 SAP K4 ACS 7 SAP K5 ACS 2 R3 QC QC QC 300 cps S/N > 100 Data pre-processing: Retention time correction to internal standards Normalization to internal standards Normalisazion to total area sum of peaks Statistical evaluation: Principal component analysis (PCA) Supervised PCA t-test -- > vulcano plot Statistics used to find significant molecular features varying between groups of sample set
Assay stability as assessed by QC a b CVs of retention times: 0.4 % for LPC(17:1) 0.2 % for PC(17:0/20:4) c CVs of raw signal intensities: 12 % for LPC(17:1) 10 % for PC(17:0/20:4) 12 Jörg Schlotterbeck
Platelet Lipidomics Score Plot Responses are log transformed and autoscaled Multivariate statistics: PCA-DA Platelets from patients with SAP and ACS can be distinguished from healthy ones by their lipid profiles 13 Jörg Schlotterbeck
Statistical Significance Volcano Plots p-value vs log Fold Change t-test: ACS vs SAP p < 0.05 t-test: ACS vs Control (Healthy) p < 0.05 t-test: SAP vs Control (Healthy) p < 0.05 Decreased compared to control Increased compared to control 14 Joerg Schlotterbeck 2016 Universität Tübingen
Area Identification of Significant Features Hits with p < 0.05 were checked for identification (MS-DIAL) Positive identification were boxplotted Matching reference spectrum for PE 38:4; (18:0/20:4(5Z,8Z,11Z,14Z)) 627.535 MS/MS Identification Neutral loss of Phosphoethanolamine 40000 35000 30000 25000 20000 15000 10000 5000 0 m/z 768.5538 Control ACS SAP 15 Jörg Schlotterbeck
Metabolite name TG 53:7; TG(16:2/18:0/19:5); [M+NH4]+ SQDG 35:5; SQDG(16:2/19:3); [M+NH4]+ PC 38:3; PC(18:3(6Z,9Z,12Z)/20:0); [M+H]+ lysope 16:0; PE(-16:0/0:0); [M+H]+ R1 R10 R2 R3 R4 R5 R6 R7 R8 R9 SAP K10 SAP K11 SAP K2 SAP K3 SAP K4 SAP K5 SAP K6 SAP K7 SAP K8 SAP K9 ACS 1 ACS 10 ACS 11 ACS 12 ACS 13 ACS 14 ACS 2 ACS 4 ACS 5 ACS 6 ACS 7 ACS 8 ACS 9 Heatmap of Lipids Identified with MS-DIAL Response 0.02000 0.4000 TG 1.000 1.750 SM PE 4.000 PC 200.0 PlasmenylPC MGDG MG LysoPE LysoPC DGTS DGDG DG Cer CE ESI(+) 7,500 molecular features 664 lipids with p < 0.05 (ACS vs control) 162 of them unidentified by MS-DIAL Shown are identified ones (502) Sample no. 16 Jörg Schlotterbeck 2016 Universität Tübingen
Area Area Area Area Platelets of SAP and ACS Patients Show Altered xpl Levels PLPC-(H) PAPC-(H) Not identified by MS-DIAL 500000 400000 300000 200000 100000 m/z 790.5593 80000 60000 40000 20000 m/z 814.5593 Target list of 40,000 lipids from LipidMaps db complemented by target list of xpls Loaded into MasterView 0 Control ACS SAP 0 Control ACS SAP Level of metabolite Id: o accurate mass PLPC-(H) 2 PC 33:1-(H) o isotope pattern o fragmentation 70000 60000 50000 40000 30000 20000 10000 0 m/z 822.5491 Control ACS SAP 500000 400000 300000 200000 100000 0 m/z 762.5643 Control ACS SAP Standards for verification not available currently PLPC: 1-Palmitoyl-2-linoleoyl-sn-glycero-3-phosphocholine; PAPC, 1-palmitoyl-2-arachidonoyl-sn-glycero-3-phosphocholine
Targeted and nontargeted analysis in single workflow supported by SWATH Quantitative analysis based on precursor or fragments Improved selectivity and sensitivity Combined Targeted / Untargeted Analysis H H P - PNPC 1-palmitoyl-2-(5 -oxovaleroyl)-sn-glycero-3-phosphocholine N + P - N + H H PVPC 1-palmitoyl-2-(9 -oxononanoyl)-sn-glycero-3-phosphocholine P - N + H H PAzPC 1-palmitoyl-2-azelaoyl-sn-glycero-3-phosphocholine LDs (matrix): 0.2-2.0 ng/ml LQs (matrix): 1-6 ng/ml Good repeatabilities P - N + H H PGPC 1-palmitoyl-2-glutaroyl-sn-glycero-3-phosphocholine 18 Jörg Schlotterbeck 2015 Universität Tübingen
% I n t e n s i t y ( o f 1 9 1 4. 4 ) Combined Targeted / Untargeted Analysis IDA (20 MS/MS per cycle) XIC of PNPC 650.439 +/- 0.010 Da MS/MS triggered at RT = 12.1281 100% 12.1 50% H H P - PNPC 1-palmitoyl-2-(5 -oxovaleroyl)-sn-glycero-3-phosphocholine IDA: Quantification by TF- MS (via precursor ion only) N + 0% SWATH 80% 60% 40% 20% 0% 10.0 10.5 11.0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 Time, min XIC of PNPC 650.439 +/- 0.010 Da Fragment XIC of Phosphocholine 184.073 +/-0.01 Da 100% 12.1 10.0 10.5 11.0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 Time, min SWATH: Supports quantification in MS/MS mode (via fragment ions) Enough data points across peak SWATH without deconvolution: Limited assay specificity Several lipids with same/similar t R may yield the same fragment (interference) 19 Jörg Schlotterbeck 2016 Universität Tübingen
% I n t e n s i t y ( o f 1. 3 e 5 ) % I n t e n s i t y ( o f 1. 1 e 5 ) Quantification based on Deconvoluted MS/MS Chromatogram lysopc 18:1 IDA SWATH, deconvoluted Lyso-PC(18:1), 522.3554 +/-0.01 Da MS/MS triggered at RT = 10.3210 100% 80% 60% 40% 20% 10.157 10.466 0% SWATH 9.0 9.5 10.0 10.5 11.0 11.5 12.0 12.5 13.0 Time, min Lyso-PC(18:1), 522.3554 +/-0.01 Da Fragment XIC 184.073 +/- 0.010 Da 100% 10.47 80% 10.16 60% 40% 20% 0% 9.0 9.5 10.0 10.5 11.0 11.5 12.0 12.5 13.0 Time, min 20 Jörg Schlotterbeck 2016 Universität Tübingen
Summary / Conclusions Data-independent acquisition (DIA) with SWATH Better coverage of analytes as compared to DDA (IDA) profiling MS-DIAL a powerful SW for automated lipid identification xidized PLs not successfully identified, therefore target list approach used In future, focus on combined targeted/untargeted profiling Ion mobility and 2D-LC as further selectivity dimensions to resolve complexity Platelet lipidomics SAP and ACS show significant differences in lipid profiling (regulation via signaling cascades) Several oxidized PLs showed increased levels in SAP and ACS patients (oxidative stress or controlled regulation upon platelet activation)
Acknowledgements Jörg Schlotterbeck Jeannie Horak Adrian Sievers-Engler Stefan Neubauer Stefan Polnick Heike Gerhardt Markus Höldrich Siyao Liu Corinna Sanwald Bernhard Drotleff Ulrich Woiwode Stefanie Bäurer Carlos Calderon Mike Kaupert Aleksandra Zimmermann Prof. Meinrad Gawaz Dr. Madhumita Chatterjee Financial Support: Struktur- und Innovationsfonds für die Forschung (SI-BW)