SUPPLEMENT - Nature Research · Zeleznik, Kathrin A Zierler, Federico Torta, Amaury Cazenave-Gassiot, Markus R Wenk, Alexander Fauland, Craig E Wheelock, Aaron M Armando, Oswald Quehenberger,

SUPPLEMENT

Deciphering lipid structures based on platform-independent decision rule sets

Jürgen Hartler, Alexander Triebl, Andreas Ziegl, Martin Trötzmüller, Gerald N Rechberger, Oana A Zeleznik, Kathrin A Zierler, Federico Torta, Amaury Cazenave-Gassiot, Markus R Wenk, Alexander Fauland, Craig E Wheelock, Aaron M Armando, Oswald Quehenberger, Qifeng Zhang, Michael J O

Wakelam, Guenter Haemmerle, Friedrich Spener, Harald C Köfeler, Gerhard G Thallinger

Supplementary Note 1: Sample preparations, experimental settings and protocols for MS/MS. Supplementary Note 2: LipidBlast settings for file conversion and lipid annotation. Supplementary Note 3: Spectral denoising. Supplementary Note 4: Algorithm for retention time prediction.

Supplementary Figure 1: Frequency of lipid molecular species in different abundance classes detected in murine liver on Orbitrap Velos Pro in CID mode.

Supplementary Figure 2: Spectral evidence for novel lipid molecular species identified in murine liver. Supplementary Figure 3: Co-eluting regio-isomers of PE 16:0_20:4 observed on Orbitrap Velos Pro in

negative ion mode. Supplementary Figure 4: Correct annotation of chromatographically separated DG 16:0/18:1/0:0

(sn-1,2 DG) and DG 16:0/0:0/18:1 (sn-1,3 DG). Supplementary Figure 5: Lipid nomenclature based on the level of structural detail for the example of

PI 18:0/20:4. Supplementary Figure 6: Levels of structural information deducible from MS/MS spectra for the

example of PI 18:0/20:4. Supplementary Figure 7: Graphical user interface for decision rule set adaption/development. Supplementary Figure 8: Complete decision rule set for deprotonated PI. Supplementary Figure 9: Decision rule check for lipid head groups ([HEAD]). Supplementary Figure 10: Decision rule check for constituent acyl or alkyl/1-alkenyl chains ([CHAINS]). Supplementary Figure 11: Decision rule check for sn-positions of acyl or alkyl/1-alkenyl chains

([POSITION]). Supplementary Figure 12: Annotation of a mixed spectrum of a pair of structural isomers. Supplementary Figure 13: Graphical representation of data from pairs of structural isomers identified in

control experiment 3. Supplementary Figure 14: MS/MS spectra allow for MS1 peak splitting of overlapping isomeric lipid

subclasses/adducts.

Supplementary Table 1: Overview of experiments performed and mass spectrometric platforms used. Supplementary Table 2: Lipid standards for control experiment 1. Supplementary Table 3: Isomeric pairs of lipid standards for different lipid subclasses applied in

control experiment 2. Supplementary Table 4: Data evaluation of isomeric subclasses/adducts (control experiment 2). Supplementary Table 5: Pairs of structurally isomeric lipid standards for control experiment 3. Supplementary Table 6: Data evaluation of structurally isomeric pairs (described in Supplementary

Table 5) measured in positive ion mode (control experiment 3).

Nature Methods: doi:10.1038/nmeth.4470

Supplementary Table 7: Data evaluation of structurally isomeric pairs (described in Supplementary Table 5) measured in negative ion mode (control experiment 3).

Supplementary Table 8: Novel lipid molecular species identified in murine liver. Supplementary Table 9: Lipid species and lipid molecular species from murine liver detected by seven

MS/MS platforms and correctly identified by LDA (available in Excel format only).

Supplementary Table 10: Overview on cross-platform species detection of seven MS/MS platforms based on murine liver samples (available in Excel format only).

Supplementary Table 11: Sensitivity and positive predictive value (PPV) of LDA and LipidBlast in negative ion mode based on data acquired on Orbitrap Velos Pro in CID mode.

Supplementary Table 12: Sensitivity and positive predictive value (PPV) of LDA and LipidBlast in positive ion mode based on data acquired on 4000 QTRAP.

Supplementary Table 13: Sensitivity and positive predictive value (PPV) of LDA and LipidBlast in negative ion mode based on data acquired on 4000 QTRAP.

Supplementary Table 14: Data evaluation of regio-isomeric pairs of chromatographically separated sn-1,2- and sn-1,3-diacylglycerols (DG).


Supplementary Note 1: Sample preparations, experimental settings and protocols for MS/MS.

Sample preparation All solvents used were at least HPLC (high performance liquid chromatography) grade and purchased either from Sigma-Aldrich (St. Louis, MO, USA) or from Merck (Darmstadt, Germany). Deionized water was obtained from an in-house MilliQ Gradient A10 system (Millipore, Billerica, MA, USA).

Lipid standards were purchased from Avanti Polar Lipids (Alabaster, AL, USA) and Larodan Fine Chemicals AB (Malmö, Sweden). For each lipid molecular species, 1 mol/L stock solutions were prepared, gravimetrically or by dilution, in chloroform/methanol 1:1 (v/v). Working solutions at 5 µmol/L were stored at -20 °C. See Supplementary Tables 2, 3, 5 and 14 for the list of lipid standards used in control experiments 1, 2, 3 and the detection of regio-isomeric species, respectively. Prior to analysis, aliquots were thawed to room temperature and storage solvent was replaced by injection solvent.

Liver samples from wild-type mice (Mus musculus, MGI:5502402, male, age six months) with ad libitum access to water and a standard chow diet were excised as described by Zierler et al.27. Three liver aliquots (25-40 mg) from three different mice were homogenized using a BioPulverizer (BioSpec Products, Bartlesville, OK, USA) and extracted according to Matyash et al.28: 1.5 mL methanol and 5 mL MTBE (methyl-tert-butyl ether) were added to liver homogenate in 12 mL glass test tubes with PTFE-lined caps, and the mixture was incubated for 10 min in an overhead shaker at room temperature. After addition of 1.25 mL deionized water and 10 min of additional shaking, the mixture was centrifuged for 5 min at 2,000 x g and the upper phase was transferred to a new test tube. The lower phase was re-extracted with 2 mL of the upper phase of MTBE/methanol/deionized water 10:3:2.5 (v/v/v). The upper phases were combined, evaporated in a vacuum centrifuge and dissolved in 1 mL chloroform/methanol 1:1 (v/v) for storage at -20 °C. Animal procedures were approved by the Austrian Federal Ministry for Science and Research and the Ethics Committee of the University of Graz (BMWFW-66.007/0006-WF/3b/2014).

Control experiment 1 Orbitrap Velos Pro CID and HCD experiments carried out at Medical University of Graz LC-MS/MS analysis was essentially performed as described by Triebl et al.29 Chromatographic separation of lipids was performed on a Waters BEH C8 column (100 x 1 mm, 1.7 µm) (Waters, Milford, MA, USA), thermostated to 50 °C in a Dionex Ultimate XRS UHPLC (ultra-high performance liquid chromatography) system. Mobile phase A for chromatographic separation was deionized water containing 10 mmol/L ammonium formate and 0.1 vol% of formic acid. Mobile phase B was acetonitrile/2-propanol 5:2 (v/v) with the same additives. Gradient elution started at 50% mobile phase B, rising to 100% B over 40 min, 100% B was held for 10 min and the column was re-equilibrated with 50% B for 8 min before the next injection. Flow rate was 150 µL/min and injection volume was 2 µL.

The Orbitrap Velos Pro hybrid mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) was operated in Data Dependent Acquisition (DDA) mode using a HESI II (heated electrospray ionization) ion source. Prior to experiments, lens settings and source parameters were optimized to the signal of PE 34:1 (using protonated/deprotonated molecular ion in positive/negative ion mode, respectively) infused into the column effluent of 80% mobile phase B with a mixing tee. Ion source parameters for positive polarity were as follows: Source voltage 4.5 kV; source temperature 375 °C; sheath gas 25 arbitrary units (AU); aux gas 9 AU; sweep gas 0 AU; capillary temperature 300 °C. Ion source parameters for negative ion mode were: Source voltage 3.8 kV; source temperature 325 °C; sheath gas


30 AU; aux gas 10 AU; sweep gas 0 AU; capillary temperature 300 °C. Full scan spectra from m/z 400 to 1,000 for positive ion mode and from m/z 400 to 1,000 in negative ion mode were acquired at a target resolution of 100,000 (FWHM at m/z 400). From each full scan spectrum, ions were selected for sequential fragmentation either in the linear ion trap (collision-induced dissociation (CID) experiments) or in the C-trap (higher-energy collision dissociation (HCD) experiments) and detected in the ion trap or the Orbitrap analyzer, respectively. For CID experiments, the 10 most abundant ions of the full scan spectrum were fragmented using He as collision gas (isolation width 1.5 m/z; activation Q 0.2; activation time 10 ms). Normalized collision energies were 30, 40, 50, 60, and 70% in both positive and negative ion mode. For HCD experiments, fragment spectra of the 4 most abundant ions were collected in the Orbitrap mass analyzer at a target resolution of 7,500 (FWHM at m/z 400, isolation width 1.5 m/z; activation time 0.1 ms). Exclusion time (duration after fragmentation for excluding ions with similar m/z) was set to 10 s. Normalized collision energy settings were 15, 20, 25, 30, and 35% for positive ion mode, and 40, 50, 60, 70, and 80% for negative ion mode.

4000 QTRAP experiments carried out at Medical University of Graz The Agilent HPLC 1100 system (Agilent Technologies, Waldbronn, Germany) required a different chromatographic setup; therefore, a Kinetex C8 column (100 x 2.1 mm, 1.7 µm) (Phenomenex, Torrance, CA, USA) was used. Mobile phases were identical as used for Orbitrap Velos Pro. Gradient elution started at 50% mobile phase B, rising to 100% B over 30 min, 100% B was held for 15 min and the column was re-equilibrated with 50% B for 13 min. Flow rate was 250 µL/min and 2 µL of the sample were injected.

The 4000 QTRAP (AB Sciex, Concord, Canada) was controlled by Analyst v. 1.6.1 software and operated in Information Dependent Acquisition (IDA) mode, using an enhanced MS (EMS) scan from m/z 400 to 1,050 and 400 to 1,000 in positive and negative ion mode, respectively, at 1,000 Da/s as a survey scan. Prior to analysis, source parameters were optimized using flow injection analysis into an isocratic flow of 80% mobile phase B. Analytes were PC 36:1 (m/z 788.6-184) for positive ion mode and PE 36:2 (m/z 742-279) for negative ion mode. The optimized source parameters of the Turbo V ion source for positive ion mode were as follows: Curtain gas 15 AU; collision gas high; ionspray voltage 5,000 V; temperature 450 °C; gas 1 50 AU; gas 2 50 AU; declustering potential 60 V; collision energy spread 0; collision energies 30, 40, 45, 50, and 60 eV. Source parameters for negative ion mode were: Curtain gas 10 AU; collision gas high; ionspray voltage 4,500 V; temperature 600 °C; gas 1 20 AU; gas 2 50 AU; declustering potential 95 V; collision energy spread 0; collision energies 30, 40, 45, 50, and 60 eV. Isolation width was unit resolution, which corresponds to 0.6-0.8 Da. From each survey scan, the four most abundant ions exceeding 1,000 counts were chosen for enhanced product ion (EPI) scans at 4,000 Da/s. Previously detected target ions were excluded for 10 s.

SYNAPT G1 HDMS QTOF experiments carried out at University of Graz The LC-MS/MS method is based on Knittelfelder et al.30. Briefly, chromatographic separation was performed using a BEH C18 column (2.1 x 150 mm, 1.7 µm) thermostated to 50 °C in an AQUITY UPLC system (Waters, Manchester, UK). Mobile phases A and B were water/methanol 1:1 (v/v), and 2-propanol, respectively, both containing 8 µmol/L phosphoric acid, 10 mmol/L ammonium acetate and 0.1 vol% formic acid. Flow rate was 150 µL/min; the linear gradient starting at 45% B reached 100% B within 32 min, followed by 10 min at this level, and re-equilibration for 8 min.

The SYNAPT G1 HDMS QTOF mass spectrometer (Waters, Manchester, UK) equipped with an ESI (electrospray ionization) source was used for detection. Source parameters were as follows: Capillary temperature 100 °C; desolvation temperature 400 °C; nebulizer gas nitrogen. Capillary voltage was


2.6 kV in positive and 2.1 kV in negative ionization mode. Lock spray was achieved by an external pump (L-6200, Hitachi) at a flow rate of 200 µl/min with a 1:13 split. Leucine-enkephalin ([M+H]+: m/z 556.2771 and [M-H]-: m/z 554.2615) was used as lock mass and measured every 20 s with a scan time of 0.7 s independent of the other scan modes, allowing for continuous mass correction. This error was always below 5 ppm. For MS experiments, the Survey Scan Mode was used with the following settings: MS mass range V-mode; extended dynamic range 200 to 1,000 m/z in positive ion mode and 100-1,500 m/z in negative ion mode, respectively; scan time 0.5 s; isolation width 1 Da; MS/MS mass range 50 to 1,000 m/z in both ion modes; MS/MS threshold 20 counts/s, up to 8 ions selected for MS/MS; MS/MS scan time, 0.5 s. An inclusion list containing m/z values of the ions of interest was used for MS/MS experiments. Collision energies were 15, 20, 25, 30, and 35 V in positive ion mode and 25, 30, 35, 40 and 45 V in negative ion mode.

G6550A QTOF experiments carried out at National University of Singapore Chromatographic separation was performed using an Agilent C18 Zorbax RRHD 100 x 2.1 mm, 1.8 µm column thermostated to 50 °C in an Agilent 1290 Infinity UHPLC system (Agilent, Santa Clara, CA, USA). Mobile phases A and B were acetonitrile/water 40:60 (v/v) and 2-propanol:acetonitrile 90:10 (v/v), respectively, both containing 10 mmol/L ammonium formate. Gradient elution started at 10% mobile phase B, rising to 25% B in 1 min, to 60% B in 4.5 min, to 85% B in 8.5 min, and to 100% B in 4 min. 100% B was held for 1 min before the column was re-equilibrated for a further 1.5 min, leading to a total run time of 20.5 min. Flow rate was 400 µL/min, and 2 µL of sample were injected.

The Agilent G6550A QTOF mass spectrometer was operated in AutoMSMS mode. From the full scan spectra, acquired from m/z 100 to 1,700 at a rate of 4 spectra/s, tandem mass spectra of the 15 most abundant precursor ions were acquired at a scan rate of 20 spectra/s, isolation width 1.3 Da, and a collision energy setting of 30 V. Previously detected target ions were excluded after three spectra for 9 s. Source parameters for positive and negative ion mode were as follows: Gas temperature 200 °C; gas flow 14 AU; nebulizer 35 AU; sheath gas temperature 350 °C; sheath gas flow 11 AU; capillary voltage 3,500 V; nozzle voltage 1,000 V; fragmentor 175; skimmer 65; octopole RF peak 750.

Orbitrap Elite experiments carried out at Babraham Institute, Cambridge Chromatography was performed on a Gemini-NX C18 (150 x 2 mm) column (Phenomenex, Torrance, CA, USA) using a Shimadzu Prominence 20-AD system (Shimadzu, Kyoto, Japan). Mobile phase A was water and mobile phase B was acetonitrile/2-propanol 5:2 (v/v), both containing 10 mmol/L ammonium formate and 0.1 vol% formic acid as additives. Eight µL of sample were injected onto the column and eluted with a gradient of 50-100% B over 25 min, 100% B was held for 20 min before the column was re-equilibrated with 50% B. Flow rate was 450 µL/min, the column oven temperature was 50 °C and the samples were kept at 4 °C.

MS/MS data was acquired on an Orbitrap Elite mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) in DDA mode. Source parameters for both polarities were as follows: Capillary temperature 380 °C; source heater temperature 380 °C; sheath gas 60 AU; aux gas 20 AU; sweep gas 5 AU. Source voltage was 3.5 kV and 5 kV for positive and negative ion mode, respectively. Full scan spectra in the range of m/z 400 to 1,050 were acquired at a target resolution of 120,000 (FWHM at m/z 400). The four most abundant ions from each spectrum were subjected to fragmentation in the C-trap, and the tandem mass spectra were recorded at a target resolution of 15,000 (FWHM at m/z 400). Isolation width was 1.5 m/z, activation time was 0.1 ms, normalized collision energy was 25% in positive ion mode and 50% in negative ion mode, and the exclusion time was 15 s.


QTRAP 6500 experiments carried out at University of California San Diego A Waters Acquity UPLC system (Waters Technologies, Milford, MA, USA) with a Phenomenex Kinetex C18 column (150 x 2.1 mm, 1.7 µm) was used for chromatographic separation. Gradient elution started at 40% mobile phase B for 10 min, then rising to 100% B over 10 min. 100% B was held for 30 min and the column was equilibrated with 50% B for 8 min. Flow rate was 300 µL/min and 10 µL of sample was injected.

The QTRAP 6500 (AB Sciex, Concord, ON, Canada), controlled by Analyst v. 1.6.2 software, operated in IDA mode, using an EMS scan from m/z 400 to 1,000 at 10,000 Da/s as a survey scan. Source parameters were automatically optimized using flow injection analysis into an isocratic flow of 80% mobile phase B. Analytes were PC 34:1 (m/z 760-184) for positive ion mode and PE 34:1 (m/z 716-281) for negative ion mode. The optimized source parameters of the Turbo V ion source for positive ion mode were as follows: Curtain gas 20 AU; collision gas high; ionspray voltage 5,000 V; temperature 300 °C; gas 1 30 AU; gas 2 30 AU; declustering potential 100; collision energy spread 0; collision energy 45 V. Source parameters for negative ion mode were: Curtain gas 20; collision gas high; ionspray voltage 4,500 V; temperature 300 °C; gas 1 30 AU; gas 2 30 AU; declustering potential 150; collision energy spread 0; collision energy 10 V. Isolation width was unit resolution, which corresponds to 0.6-0.8 Da. From each survey scan, the eight most abundant ions exceeding 100,000 counts were chosen for EPI at 10,000 Da/s. Previously detected target ions were excluded for a duration of 6 s.

Q Exactive experiments carried out at Karolinska Institute Stockholm Chromatographic separation of lipids was performed on a Waters (Waters, Milford, MA, USA) BEH C8 column (100 x 1 mm, 1.7 µm), thermostated to 50 °C in a Thermo Ultimate 3000 UHPLC system. Mobile phase A for chromatographic separation was deionized water with 10 mmol/L ammonium formate and 0.1 vol% of formic acid as additives. Mobile phase B was a mixture of acetonitrile/2-propanol 5:2 (v/v) with the same additives. Gradient elution started at 50% mobile phase B, rising to 100% B in 40 min, 100% B were held for 8 min and the column was re-equilibrated with 50% B for 7 min before the next injection. The flow rate was 150 µL/min and injection volume was 2 µL.

The Q Exactive mass spectrometer (Thermo Fisher Scientific Inc., Waltham, MA, USA) was operated in DDA mode. Ion source parameters for positive polarity were as follows: Spray voltage 4.5 kV; aux gas heater temperature 375 °C; sheath gas 35 AU; aux gas 10 AU; sweep gas 0 AU; capillary temperature 320 °C. Ion source parameters for negative ion mode were: Spray voltage 3.8 kV; aux gas heater temperature 325 °C; sheath gas 40 AU; aux gas 20 AU; sweep gas 0 AU; capillary temperature 300 °C. For both positive and negative ion mode, full scan spectra from m/z 400 to 1,050 were acquired at a target resolution of 140,000 (FWHM at m/z 200) from which ions were selected for fragmentation. Automatic gain control (AGC) target was set to 1e6 ions and the maximum injection time to 500 ms. For the DDA MS/MS scan for both positive and negative ion mode, target resolution was set to 17,500 (FWHM at m/z 200), AGC target to 5e4 ions, maximum injection time to 50 ms and isolation width to 1.5 m/z. The 10 most abundant ions from the full scan spectra were selected for the MS/MS scan. Normalized collision energy was 25% for positive ion mode and 30% for negative ion mode.

Control experiment 2 Orbitrap Velos Pro CID and HCD experiments carried out at Medical University of Graz Chromatography on intentionally co-eluting isomeric pairs of PE/PC and LPE/LPC was performed on an Agilent Zorbax Extend C18 (50 x 2.1 mm, 3.5 µm) column (Agilent Technologies, Waldbronn, Germany). Chromatographic separation was done as described for control experiment 1. Elution started at 95% B for 15 s before switching to 85% B for 3 min. Prior to the next injection, the column was re-equilibrated


with 95% B for 5 min. Flow rate was 500 µL/min, the column heater was set to 50 °C and 2 µL of sample were injected.

The Orbitrap Velos Pro mass spectrometer was operated in positive DDA mode as described for control experiment 1, with an inclusion list containing protonated and sodiated adducts of the analytes under investigation. Full scan resolution was lowered to 60,000 (FWHM at m/z 400) to reduce cycle time. For CID experiments, a normalized collision energy of 50% and an exclusion time of 1 s were employed. For HCD experiments, a normalized collision energy of 25% for positive ion mode, and an exclusion time of 2.1 s were employed.

4000 QTRAP experiments carried out at Medical University of Graz Intentional co-elution of the investigated compounds was performed using the same column and mobile phases as described for control experiment 1, but with an isocratic flow of 93% mobile phase B at a flow rate of 250 µL/min for 5 min.

The 4000 QTRAP operated in IDA mode, using an EMS scan from m/z 400 to 900 at 1000 Da/s as survey scan. Source parameters were as follows: Curtain gas: 14 AU; collision gas high; ion spray voltage 5,000 V; temperature 550 °C; gas 1 60 AU; gas 2 60 AU; declustering potential 70 V; collision energy spread 0; collision energy 10 eV. The four most abundant ions from an inclusion list containing protonated and sodiated adducts of the analytes under investigation were chosen for EPI scans at 4,000 Da/s at a collision energy of 45 eV. Previously detected ions were excluded for a duration of 4 s.

SYNAPT G1 HDMS QTOF experiments carried out at University of Graz Parameters for chromatography and mass spectrometry were as described for control experiment 1 for SYNAPT G1 HDMS QTOF, except that the gradient time was shortened to 10 min and following MS/MS settings were used: MS/MS threshold 10 counts/s, up to 3 ions selected for MS/MS; MS/MS scan time 0.4 s.

Control experiment 3 carried out at Medical University of Graz and University of Graz In this control experiment, Orbitrap Velos Pro CID and HCD, 4000 QTRAP, and SYNAPT G1 HDMS QTOF were used as platforms. Chromatography and mass spectrometric detection parameters were identical to those of control experiment 1, except for the QTOF where a MS mass range of 200 to 1,000 m/z was used for full scans in both ion modes. Experiments were conducted at the following collision energies: Orbitrap CID positive ion mode 50%, negative ion mode 50%; Orbitrap HCD positive ion mode 25%, negative ion mode 50%; QTRAP positive ion mode 45 eV, negative ion mode 45 eV; QTOF positive ion mode 30 V, negative ion mode 30 V. For QTRAP and QTOF, an inclusion list containing protonated/ammoniated and sodiated adducts in positive ion mode, and deprotonated molecules of the analytes of interest for negative ion mode, respectively, was used for MS/MS experiments.

Biological experiment The three liver extracts were measured in quintuplicates for each polarity in the laboratories, as described for control experiment 3 for Orbitrap Velos Pro CID, Orbitrap Velos Pro HCD and 4000 QTRAP, and as described for control experiment 1 for G6550A QTOF, Orbitrap Elite, QTRAP 6500, and Q Exactive.

Demonstration of novel lipid molecular species derived from low-abundance MS/MS fragments carried out at Medical University of Graz Liver extract 1 was measured as described in control experiment 3 for the Orbitrap Velos Pro HCD platform in positive and negative ion mode. Full scan resolution, however, was lowered to 30,000


(FWHM at m/z 400), and fragment spectra of the 4 most abundant ions, selected from an inclusion list containing the m/z values of the novel lipid molecular species, were collected in the Orbitrap mass analyzer at a target resolution of 7,500 (FWHM at m/z 400, isolation width 1.5 m/z; activation time 0.1 ms). Exclusion time (duration after fragmentation for excluding ions with similar m/z) was set to 3 s. For each lipid molecular species and adduct, the optimum collision energy was used.

Demonstration of detection of regio-isomeric lipid molecular species carried out at Medical University of Graz Chromatographic separation was performed on a Waters BEH C18 column (150 x 1 mm, 1.7 µm) (Waters, Milford, MA, USA), thermostated to 50 °C in a Dionex Ultimate XRS UHPLC (ultra-high performance liquid chromatography) system. Mobile phase A for chromatographic separation was deionized water containing 10 mmol/L ammonium formate and 0.1 vol% of formic acid. Mobile phase B was acetonitrile/2 propanol 5:2 (v/v) with the same additives. Gradient elution started at 50% mobile phase B, rising to 100% B over 75 min, 100% B was held for 5 min and the column was re-equilibrated with 50% B for 15 min before the next injection. Flow rate was 100 µL/min and injection volume was 2 µL.

The regio-isomers were measured in triplicates as described in control experiment 3 for Orbitrap Velos Pro CID. However, an inclusion list was used containing the m/z value of the ions of interest (sodiated adducts of isobaric DG 16:0/0:0/18:1 and DG 16:0/18:1/0:0), and the exclusion time (duration after fragmentation for excluding ions with similar m/z) was lowered to 1 s.

Evaluation of mass spectrometric data carried out at Graz University of Technology, Medical University Graz and Brigham and Women's Hospital Boston All raw data elaborated by mass spectrometric groups were sent to our bioinformatics group and processed by LDA. Data of control experiment 1 and the biological experiment were also processed by LipidBlast. These data were acquired on Orbitrap Velos Pro in CID mode and 4000 QTRAP with collision energy settings of 50% and 45 eV, respectively, in both positive and negative ion mode. The correct assignment of lipid species and lipid molecular species identified in liver lipidomes was verified by manual inspection of spectra, and by aligning them with the respective retention time data.


Supplementary Note 2: LipidBlast settings for file conversion and lipid annotation.

For file conversion to mgf format, ProteoWizard MSConvert GUI package 3.0.4976 was used. For LipidBlast library search, the NIST MSPepSearch 64bit (release date June 4th 2016; download date April 27th 2016) was used as recommended in the LipidBlast user manual (lipidblast-handbook-v5.pdf).

Data conversion into mgf: Thermo RAW data files from Orbitrap Velos Pro and AB Sciex WIFF files from 4000 QTRAP were converted to mgf format by msconvert of the ProteoWizard package by e.g. the command:

msconvert.exe --mgf $FILE_TO_TRANSLATE$ --filter "peakPicking true 2-2" --filter “msLevel 2-2” --filter “activation CID” --filter "threshold count 100 most-intense" -o ./

Data annotation with LipidBLAST by MSPepSearch: The following parameters were set:

Search library positive data: LipidBlast-pos and Cer-d-Hpos used in combination (CustomPC+H and CustomPC+Na was not used, since it is for QTOF data according to the manual)

Search negative data: LipidBlast-neg, PC-Form-neg, and PC-Me-neg and Cer-d-Hneg used in combination

Presearch mode: Standard

m/z limits: Standard settings of 0-2,000 Precursor ion tolerance: 0.01 and 0.4 for Orbitrap Velos Pro and 4000 QTRAP,

respectively (same settings as in LDA)

Ignore peaks around precursor: true

Fragment peak m/z tolerance: 0.2 and 0.4 for Orbitrap Velos Pro and 4000 QTRAP, respectively (same settings as in LDA)

Min. match factor: 450 (default setting) or 10 (more relaxed setting)

Min. peak intensity: 1 (default setting)

Max. number of output hits: 10 for standards (higher chance for correct detection) 3 for biological data to decrease the number of false positives

Show spectra without matches: false

Include Hit-Unknown precursor m/z difference: true

Include precursor m/z in the output: true

Output the input spectrum number: true Set program priority above normal: true

Use number of replicates: false

Calculate rev-dot: true

QTOF: false Load libraries in memory: false (according to user manual: this will lead to errors, if true)


Supplementary Note 3: Spectral denoising.

Spectral noise varies from platform to platform and even from spectrum to spectrum. Some platforms already deliver denoised spectra (Orbitrap HCD), where further denoising would result in loss of information. Therefore, we deal in two ways with noise:

1. Spectra containing 100 or more signals are subjected to dynamic noise removal (applied to each individual spectrum before generation of consolidated spectrum).

2. Noise varies between different lipid subclasses/adducts due to varying fragmentation efficiencies. Thus, a noise limit can be set for a lipid subclass/adduct (“BasePeakCutoff” in the decision rule sets), to remove presumable noise from lipid subclasses producing noisy spectra.

Ad 1) To cope with considerable intensity variations between spectra we implemented dynamic noise level estimation for each individual spectrum as described in Xu et.al.31 (except SNRmin is 1.7 instead of 2.0) to allow for differentiation between noise and signal peaks. For the latter, intensities must be 1.5 times higher than the noise level. This filtering approach correctly assigns peaks in extremely noisy spectra; an example is given in Supplementary Note 3, Figure 1, below: The noise signals identified are still displayed, but no fragments will be assigned to them and consequently noise is excluded from calculation of the spectrum coverage.

Supplementary Note 3 Figure 1: Spectrum of the reference standard DG (18:1/0:0/18:1) measured in control experiment 1 on Orbitrap Velos Pro in CID 50%.


Supplementary Note 4: Algorithm for retention time prediction.

Retention times of lipid species in reverse phase chromatography are very specific for a certain run and can be used to filter incorrect species identifications. LDA supports this with a module for retention time modeling, which can be enabled by the “RetentionTimePostprocessing” parameter in the decision rule sets. Parameters for a retention time model are estimated by fitting identifications verified by decision rule sets to the non-linear model. The expected retention time of a species is predicted from the model for each lipid subclass based on the number of carbon atoms and double bonds. Identifications are discarded when the actual retention time is outside a certain range around the estimated retention time.

We discovered previously that retention times of lipids follow a certain pattern.23 However, the impact of additional carbon atoms and double bonds on retention time shift varies between different subclasses. Consequently, the model must be fitted to each individual lipid subclass. Since retention times vary from column to column, and are influenced by column degradation processes, retention times are predicted dynamically for each individual MS run. By testing different lipid classes on preliminary data (data not shown), we found the following equation that satisfied best the different lipid subclasses:

𝑅𝑅𝑅𝑅(𝑥𝑥,𝑦𝑦) = 𝐴𝐴 ∗ (1 − 𝐵𝐵 ∗ 𝑥𝑥−𝐶𝐶) + 𝐷𝐷 ∗ 𝑒𝑒(−𝐸𝐸∗𝑦𝑦+𝐹𝐹∗𝑥𝑥) + 𝐺𝐺 (1)

x and y correspond to the number of carbon atoms and the number of double bonds, respectively, and A, B, C, D, E, F, and G are the model parameters to be estimated for each lipid subclass. The model is fitted using species verified by decision rule sets and unambiguously assigned to a single lipid subclass as described in Application Note 3. These species are verified to be typically non 13C isotopic peaks (MS1 algorithm), their isotopes follow their theoretical isotopic pattern, and their fragmentation pattern agrees to the one defined in the corresponding decision rule set. As such, they form a reliable basis to predict retention times of hits where only a single MS1 signal is present. For model fitting, we apply the Levenberg-Marquardt algorithm32, 33, starting with a damping parameter of 10 and using a factor of 1.5 for increasing/reducing the damping factor according to Marquardt’s recommendation33. In our implementation, the iterative fitting procedure terminates as soon as χ2 does not improve anymore, or when 10,000 iterations were performed.

Due to the amount of parameters and the rather complex model, fitting can result in a singular matrix. To overcome this situation, we consecutively use the following simpler models as a fallback:

𝑅𝑅𝑅𝑅(𝑥𝑥,𝑦𝑦) = 𝐴𝐴 ∗ �1 − 𝐵𝐵 𝑥𝑥� � + 𝐶𝐶 ∗ 𝑒𝑒(−𝐷𝐷∗𝑦𝑦+𝐸𝐸∗𝑥𝑥) + 𝐹𝐹 (2)

𝑅𝑅𝑅𝑅(𝑥𝑥,𝑦𝑦) = 𝐴𝐴 ∗ ln(𝐵𝐵 ∗ 𝑥𝑥) + 𝐶𝐶 ∗ 𝑒𝑒(−𝐷𝐷∗𝑦𝑦+𝐸𝐸∗𝑥𝑥) + 𝐹𝐹 (3)

This simplification is feasible, since lipid subclasses containing fewer MS/MS identifications are typically accompanied by less correct detections based on MS1 evidence only. Thus, predictions do not need to be as precise as for subclasses such as TG where we observed ranges from 30-64 carbon atoms and 0-15 double bonds. For less than six verified MS/MS identifications, no retention time prediction is possible.

After the model is fitted, a mean retention time deviation is calculated by the sum of squares of the differences between the retention times of the measured MS/MS detections and the predicted ones. For passing the filter, a (MS1) detection must be inside the range of the predicted retention time ± four times the mean retention time deviation, or ± a fixed range parameter “RetentionTimeMaxDeviation” in the decision rule sets (not used in this study). By using four times the mean retention time deviation, we


designed the filter rather permissive to avoid removal of correct identifications. If more stringent criteria are required, the fixed range can be set upon user decision.

This approach was applied to the biological data of all seven platforms tested (different reverse phase columns were used). The approach delivers feasible results when sufficient numbers of MS/MS identifications are available (no specific evaluation; MS1 detection is beyond the scope of this paper).


Supplementary Note 5

Lipid Data Analyzer

Study Data

1 Study data download instructions ................................................................................................... 1 1.1 Instructions for download from the lab’s homepage .............................................................. 1 1.2 Instructions for download from MetaboLights ....................................................................... 2

2 Data description of benchmarking (MTBLS397) .............................................................................. 4 3 Data review and visualization .......................................................................................................... 4

3.1 Visualization by LDA (recommended) ..................................................................................... 4 3.2 Data reviewing in MetaboLights .............................................................................................. 8

4 Available data ................................................................................................................................ 11 4.1 Biological data (MTBLS396) ................................................................................................... 11 4.2 Control experiment 1 (MTBLS394) ........................................................................................ 12 4.3 Control experiment 2 (MTBLS391) ........................................................................................ 12 4.4 Control experiment 3 (MTBLS398) ........................................................................................ 12 4.5 HCD characterization and regio-isomers (MTBLS462) .......................................................... 12


1

1 Study data download instructions The example data is available from our homepage: http://genome.tugraz.at/lda2/lda_data.shtml or from MetaboLights: http://www.ebi.ac.uk/metabolights/ Data are available for three control experiments, one biological experiment, and the benchmarking of LDA against LipidBlast. Within the individual experiments, data is further subdivided to allow download of individual MS platforms, because we assume that users want to download only data relevant to their MS platforms. Details for downloading data from our homepage and from MetaboLights are given in the following two chapters.

1.1 Instructions for download from the lab’s homepage Study data is categorized in ‘BENCHMARKING LIPIDBLAST’, ‘BIOLOGICAL EXPERIMENT’, ‘CONTROL EXPERIMENT 1’, CONTROL EXPERIMENT 2’, ‘CONTROL EXPERIMENT 3’, ‘HCD CHARACTERIZATION’ and ‘IDENTIFICATION OF REGIO-ISOMERS’. Furthermore, the fragmentation rules and this description is provided in the sections ‘FRAGMENTATION RULE SETS’, and ‘STUDY DATA DESCRIPTION’, respectively.

For each platform in a category, a link is provided to download the corresponding data. In this figure, the link for downloading the biological results generated by Orbitrap Velos Pro in CID mode is highlighted. Data from various experiments for various platforms contain typically the three folders: massLists (quantitation Excels used as input for LDA), and positive and negative for the polarities (except for control experiment 2, where there is no negative folder, since the experiments were performed in positive ion mode only). In most of the cases, the folders positive and negative contain MS runs, comments, and


http://genome.tugraz.at/lda2/lda_data.shtml

http://www.ebi.ac.uk/metabolights/

2

evaluation results. Only for Control Experiment 1 for the platforms ‘Orbitrap Velos Pro’, ‘Orbitrap Velos Pro HCD’, ‘SYNAPT G1 HDMS QTOF’ and ‘4000 QTRAP’, these folders are further subdivided in the collision energies used. Further details about the individual files are given in chapters 2 and 4, and their visualization in chapter 3.

1.2 Instructions for download from MetaboLights Study data is organized in benchmarking (MTBLS397), biological experiment (MTBLS396), control experiment 1 (MTBLS394), control experiment 2 (MTBLS391), control experiment 3 (MTBLS398), and HCD characterization and regio-isomers (MTBLS462), and can be accessed by adding the identifier to the MetaboLights URL, e.g. http://www.ebi.ac.uk/metabolights/MTBLS394. In contrast to the homepage download, MetaboLights is organized much more granularly containing individual ZIP archives for each MS-run and data evaluation. For downloading data of a specific platform, click on the tab “Study Files”, and type in the name identifying the platform, e.g. ‘QTOF_G6550A’, and press enter:

This will select all files containing the corresponding identifier in the name:


http://www.ebi.ac.uk/metabolights/MTBLS394

3

The following study identifiers are used throughout the different categories:

• Orbitrap_velos_CID • Orbitrap_velos_HCD • QTOF_SYNAPTG1 • 4000QTRAP (for Control Experiment 2 use: ABSciex_QTRAP4000) • Orbitrap_elite • QTRAP6500 • QTOF_G6550A • QExactive

Depending on the experiment and platform, data download can be further confined by polarity and collision energies. For the LipidBlast benchmarking (MTBLS397), data of two platforms were used (Obitrap Velos Pro CID, 4000 QTRAP). The biological data was benchmarked on both platforms, and experiment 1 on the Orbitrap, only. Thus, data containing QTRAP in the name belongs to the 4000 QTRAP, the rest to the Orbitrap:

• Biological_positive • Biological_negative • Biological_QTRAP_positive • Biological_QTRAP_negative • Experiment1_positive • Experiment1_negative

Finally, click on the “Download selected files” button which is located at the bottom of the page. Attention: some studies might contain several GB. MetaboLights will create now one zipped archive containing the individual data files stored in zip files. Please note, depending on the size of data files, it might take a while to generate the zip file before the download starts. The zipped data files contain typically data in its vendor format, a ‘chrom’ directory (data in LDA format), and an Excel file containing the results produced by LDA. Data of the 4000QTRAP and the QTRAP6500 do not contain data in vendor format; the file can be found in the corresponding “Comments” or “Evaluation” file, since several MS runs from AB Sciex instruments are packed in one file. “Comments” and “Evaluation” files contain comments and overall analysis results. In order to have a complete data set of an experiment (all MS runs of one quintuplicate including comments or evaluation results), all files located in the individual ZIP archives (downloaded by selecting one platform as described above) have to be put in one directory. Further details about the individual files are given in chapters 2 and 4, and their visualization in chapter 3.


4

2 Data description of benchmarking (MTBLS397) The final evaluation results are stored in the following Excel files: LDA-LipidBlast_positive.xlsx and LDA-LipidBlast_negative.xlsx (biological) LDA-LipidBlast_40000QTRAP_positive.xlsx and LDA-LipidBlast_40000QTRAP_negative.xlsx (biological 4000 QTRAP) LDA-LipidBlast_positive_standards.xlsx and LDA-LipidBlast_negative_standards.xlsx (control 1). For MetaboLights, such a file can be found e.g. in the Biological_positive_Evaluation_Summary.zip. These Excel files consist of a ‘Summary’ tab containing the final results, and five more tabs for the evaluation results of the individual MS runs. Detected lipid species and lipid molecular species are counted in two ways: adduct sensitive, where every species/adduct combination is one hit; adduct insensitive, where every detected species is one hit. E.g. PC 18:0/20:4 is found in protonated and sodiated form; in the first evaluation scheme, these are counted as two hits, in the second one, this corresponds to one hit only. Tables in the manuscript are based on the adduct insensitive numbers. For the biological experiment, data is evaluated for sensitivity (percent of correctly identified species relative to the total number of manually identified species) and positive predictive value (percent of correctly identified species relative to the total number of manually identified species) on the lipid species and lipid molecular species level. For control experiment 1, the sensitivity is evaluated in four aspects: 1) lipid species detection rate; 2) detection rate of correct chain combinations; 3) detection rate of correct chain combinations without reporting additional wrong chain combinations; 4) is the same as 3) including a correct position assignment. Data from control experiment 1 clearly shows that LipidBlast is not able to correctly assign sn-positions. The folders containing the particular final result file contain 10 more Excel files (at MetaboLights, these files are zipped in one file which is e.g. called Biological_positive_Evaluation_Details.zip), where the names end by “LB10_comp.xlsx” or “LB450_comp.xlsx”. These files contain details about the individual comparisons of LDA and LipidBlast, where 10 and 450 correspond to LipidBlast using a matching factor of 10 and 450, respectively. The “Comment” columns contain detailed reasons for counting a hit as false positive, or why LDA missed a hit. Furthermore, the folders contain “mgf.tsv” (in the output folder of the home page download) and “mgf” files which are the original LipidBlast output files and the mgf files used for the searches, respectively. When downloaded from MetaboLights, these files can be found e.g. in Biological_positive_LipidBlast.zip. For the sake of completeness, the folders contain the original Thermo RAW files. For MetaboLights download, the LDA’s chrom and Excel results are provided too.

3 Data review and visualization This chapter covers data description and visualization of all experiments, except for the benchmarking study which is described in chapter 2. A complete data set (one polarity, one collision energy) consists of five MS runs (fifteen for biological data) and some txt or Excel files containing comments and/or evaluation results. One MS run comprises typically three files, i.e., data in its vendor format, a ‘chrom’ directory (data in LDA format), and an Excel file containing the results produced by LDA. This chapter contains two sections for the visualization of results, i.e. by LDA (recommended) and by MetaboLights.

3.1 Visualization by LDA (recommended) In order to install the LDA, use the provided installation wizards. During the installation process, it does not matter which MS platform you choose, since this can be changed easily in the application itself. Usually the default settings in the wizard should do.


5

In order to visualize data of a single MS run, start the LDA, click on “Settings” tab, select the corresponding MS platform and the corresponding collision energies, and click on “Save as default”:

For study data, use the following settings: Platform MS settings Frag. Selection 1 Frag. Selection 2

• Orbitrap_velos_CID Orbitra_velos_pro +50 -50 • Orbitrap_velos_HCD Orbitra_velos_pro_HCD +25 -50 • QTOF_SYNAPTG1 QTOF_centr +30V -30V • 4000QTRAP QTRAP_4000 +45V -45V • Orbitrap_elite OrbiTrap_Elite +25 -50 • QTRAP6500 QTRAP_6500 +45V -45V • QTOF_G6550A QTOF_Agilent_G6550 +30V -30V • QExactive OrbiTrap_exactive_pos/ 25

OrbiTrap_exactive_neg -30 For control experiment 1, depending on the data set, different collision energy settings might apply for the platforms Orbitrap_velos_CID, Orbitrap_velos_HCD, QTOF_SYNAPTG1, and 4000QTRAP. Then, click on the “Display Results” tab, select the corresponding chrom and LDA Excel result file, and click on “Start Display”:

A menu on the left side will appear now containing a select box at the top, where you can switch between the different lipid subclasses. Then a table follows, whereupon the name of the analyte is in the first column, and the integrated mass-chromatographic area in the second. The display name consists primarily of “$name$:$double_bonds$_$retention_time$” (e.g. 38:5_24.98 lipid species name). If there are more than one adduct/modification, the modification name is added with _$modification_name$ (e.g. 38:5_24.98_H).


6

The color encoding of the table entries has the following meaning:

• White: MS1 identification without any MSn evidence (not relevant for the publication, and not exported to MetaboLights).

• Green: MS1 identification verified by MSn spectra. • Yellow: The MSn algorithm detected more than one lipid subclass in the MS1 peak → the peak was

split on the MS1 level in a part for the one and a part for the other lipid subclass. The split is based on information derived from MSn spectra.

• Orange: The MSn algorithm detected more than one lipid subclass in this peak, but a split on the MS1 level is not possible. The intensities of the intact MS1 peak are split according to the distinct fragments of MSn spectra. This way of splitting is rather unreliable, since usually fragment intensities do not correlate with MS1 abundances.

• Blue: Current selection for display.


7

The “Show MSn” checkbox switches from the lipid species to the lipid molecular species view of the table where possible:

The table displays now information about the obtained structure according to Liebisch, G. et al. Shorthand notation for lipid structures derived from mass spectrometry. J Lipid Res 54, 1523-1530 (2013). I.e., the sum of carbon atoms and double bonds in the chains if no structural information can be obtained (e.g. 38:5), the fatty acid chains separated by an underscore if fatty acids are detectable (e.g. 20:4_18:0), or the fatty acid chains separated by a slash if the position of the fatty acid chains is assignable (e.g. 18:1/20:4). Attention: For the manual inspection of MSn, there is a difference in the displayed annotation of the spectra, if the lipid species consists of more than one lipid molecular species:

• The lipid species view is selected: The fragments of all detected lipid molecular species are annotated.

• The lipid molecular species view is selected: Only the fragments of the particularly selected lipid molecular species are annotated.

When a row of the table is activated by the left mouse button, the peaks are displayed in the 3D and the chromatogram viewer. When a row is activated by the right mouse button, a popup appears:

By clicking “Show MS/MS”, MS/MS spectra are shown as annotated by the used MSn decision rules at the time of quantitation (see figure on next page). “Edit MSn rule” will start the interface for decision rule set definition/adaption/generation.


8

The big red box shows an annotated spectrum, and the smal red box shows the buttons +/- for scrolling through spectra. Further details regarding LDA visualization options are to be found in the user manual which is accessible in the “Help” tab of the LDA, or can be downloaded from the institute’s homepage: http://genome.tugraz.at/lda2/lda_documentation.shtml.

3.2 Data reviewing in MetaboLights MetaboLights does not allow visualization of spectra, but metabolite identifications are stored in tabular format in so-called MAF (metabolite annotation/assignment file) within assays. However, you can always download the MAF directly by selecting the m_…txt from “Study files” download, and view it using Excel. An assay consists of a name describing the respective MS platform, a table comprising information about the identified metabolites, and a table containing information about the data files and their generation.


http://genome.tugraz.at/lda2/lda_documentation.shtml

9


10

The data columns contain a column called “Raw Spectral Data File” which indicates which files belong to one particular data set.

Files belonging to one data set start typically by the same file name to ease selective download of the files (see chapter 1.2). In this case, “Orbitrap_velos_HCD_pos_Exp3_1to1” has to be entered in the text box as described in chapter 1.2 for selection of the five data files. Additionally, the file called Orbitrap_velos_HCD_pos_Exp3_Evaluation.zip has to be selected manually. This file is not called Orbitrap_velos_HCD_pos_Exp3_1to1_Evaluation.zip, since it is an evaluation file for all five ratios. The “Metabolites” table shows the identified species in a collapsed form where each row corresponds to one lipid (molecular) species identified by one adduct in any of the associated 5 MS runs (or 15, in the case of the biological data).

The identifier for one species is shown in the “Metabolite identification” column in brackets at the end of the entry. The identifier starts with the lipid species (e.g. “PC 32:0”), and if present, followed by the lipid molecular species (e.g. “PC 14:0/18:0”), separated by a comma. Since data presentation is up to now not completely standardized in MetaboLights, and it was important for this study to reflect the actual identifications in each MS run, a pipe "|" was used to discriminate between the individual MS runs, while comma "," was used to account for an enumeration of identifications (e.g. several retention times) and


11

ambiguous identifications (e.g. DG 20:0/18:0/-,20:0/-/18:0 corresponds to a diacylglyceride where 20:0 is identified to be on the sn-1 position, while 18:0 might be at sn-2 or sn-3). For the first row in this example, PC 14:0/18:0 was found in all MS runs except for the second one where the missing detection is indicated by a hyphen “-“. For the column “Fragmentation”, m/z values indicative for these identifications are presented as comma separated values. The review of the control experiments by the MetaboLights interface might work quite well, however, for the biological data, we strongly recommend the LDA approach (see chapter 3.1). The reason is that the “Metabolites” column does not allow for sorting by the species identifier. Sorting by the column “Metabolite Identification” always uses the whole content of the cell. If the first MS run contains a missing detection (a hyphen), the hyphen is regarded as alphabetically first, and will be at the start of the table. Since data is provided as derived by LDA including all false positives, by this mechanism, rare species and false positives are shown first:

Nevertheless, class specific sorting is partially possible by entering the lipid subclass name in the “Search” text box which will filter the entries for the contained text. But e.g. entering of “PC” will always include LPC and P-PC too.

4 Available data This chapter covers all datasets except for the benchmarking (MTBLS397) study which is explained in chapter 2. Data sets or assays consist of 5 or 15 MS runs (stored in ZIP files in MetaboLights), and 1 or more txt or Excel files containing results of detailed analysis, comments about information content that can be derived from the various adduct ions, and suggestions about optimal collision energies for subclasses/adducts (stored in one or more ZIP files in MetaboLights). Each MS run consists of a ‘chrom’ directory (raw data in LDA format), an Excel file containing the results produced by LDA, and the data in its vendor format, except for QTRAP data, where only one wiff is available for several files. The following chapters explain in detail the amount of additional information provided for each study.

4.1 Biological data (MTBLS396) Each data set in biological data consists of 15 MS runs, and several Excel files containing manual evaluation results for both positive and negative ion mode. In the case of MetaboLights, these Excel files are packed in a file ending with “Evaluation.zip”. These files contain at least one “SpeciesDetectable …” Excel file, reporting the identified species for mouse liver sample one, and one file containing “FP” in the name, reporting the identified false positives on the MS2 level (there is no evaluation of the MS1 level, since this study covers the fragmentation spectra only). For some platforms, an Excel document starting with “FNs …” is available, reporting missed identifications, and several Excel documents containing the lipid subclass and the “species” in the file name, reporting details which lipid species and lipid molecular species were found in the individual MS runs. For 4000 QTRAP data, the wiff file is provided in the


12

4000QTRAP_Liver_Evaluation.zip, and for QTRAP 6500 data, the wiff fie can be found in the Exp1_QTRAP6500.zip of control experiment 1 (MTBLS394 – contains raw data for both the control experiment 1 and the biological experiment).

4.2 Control experiment 1 (MTBLS394) Each data set consists of 5 MS runs and several text files containing details about the detected species, missed species and false positives. In MetaboLights, these files are packed in one ZIP file whose name ends by Exp1_Comments.zip. For the platforms where the data was collected at more than one collision energy (‘Orbitrap Velos Pro’, ‘Orbitrap Velos Pro HCD’, ‘SYNAPT G1 HDMS QTOF’ and ‘4000 QTRAP’), additional comments are available providing general information about the individual subclasses and their optimal collision energies. In the download from the homepage, these files are located in the parent directory of the data set, while at MetaboLights, an additional Exp1_Comments.zip file is provided for these platforms. At MetaboLights, the used mass lists for the particular experiment are added to the “Exp1_Comments.zip” archives. For QTRAP data, the wiff files are provided in the Exp1_Comments.zip archives.

4.3 Control experiment 2 (MTBLS391) This study was performed in positive ion mode only. Each data set consists of 5 MS runs and one text file containing information about unsuccessful splits. At MetaboLights, this file is provided in a file ending by “Exp2_Comments.zip” together with the actually used mass list, and the wiff file in the case of QTRAP data.

4.4 Control experiment 3 (MTBLS398) Each data set (ratio) consists of 5 MS runs, one text file containing comments, and one Excel file containing evaluation of the performance of all five ratios for each platform and polarity. At MetaboLights, this file is provided in a file ending by “Exp3_Evaluation.zip” together with the actually used mass list, and the wiff file in the case of QTRAP data.

4.5 HCD characterization and regio-isomers (MTBLS462) The HCD characterization data consists of 4 MS runs for the positive ion mode (longer and shorter inclusion list), and two for negative ion mode. Furthermore, two additional directories are provided containing the spectra of the novel species in SVG and MGF format. At MetaboLights, these files are provided in Novel_species_SVGs.zip and Novel_species_MGFs.zip. The data of the regio-isomers consists of 3 MS runs for each sample, i.e. a mix of both regio-isomeric standards, and the individual standards. The results of the evaluation are stored in the Evaluation.xlsx. At MetaboLights, this file can be found in the PositionIsomers_Evaluation.zip.


Supplementary Figure 1: Frequency of lipid molecular species in different abundance classes detected in murine liver on Orbitrap Velos Pro in CID mode.

Lipid molecular species are shown in total and divided into three abundance groups: (i) more intense than 1% of the most abundant species (red); (ii) between 1‰ and 1% (yellow); (iii) less intense than 1‰ (green). a and b show the number of different molecular species in these groups in positive and negative ion mode, respectively. Abundance values are based on the mean value (n=5). In positive ion mode, more than 87% of the molecular species are less intense than 1% of the base beak, and more than 64% are less intense than 1‰. In negative ion mode, more than 71% are less intense than 1%, and more than 30% are less intense than 1‰.


Supplementary Figure 2: Spectral evidence for novel lipid molecular species identified in murine liver.

Species consist primarily of novel fatty acyl and/or 1-alkenyl combinations. For DG, novel stereochemistry is shown, and for Cer, a previously unreported lipid species was identified. For TG, fragments of all three fatty acid moieties have to be present for valid identification. These figures highlight the annotated experimental spectra for novel lipid molecular species identified. Evidences rely on spectra at full mass range and are supported by zoom in spectra where necessary. The majority are low resolution CID spectra acquired on an Orbitrap Velos Pro. Where evidence relies on fragments of very small intensity, high resolution HCD mode was used for final characterization. Such high-resolution spectra are flagged as “HCD characterization”. For verification of the 24:2 fatty acyl chain of the novel Cer, QTOF data was used, since on the Orbitrap Velos Pro, [M+HCOO]- adducts did not reveal any fatty acyl chain information and the abundance of [M-H]- adducts for this species was too low to trigger MS/MS spectra. Such spectra are flagged as “G6550A QTOF characterization”.


Novel species: PE P-18:1/18:1 Detected as [M-H]- (precursor m/z: 726.5461; retention time: 28.62 min; file:

003_liver2-1_Orbitrap_CID_neg)

Comment: detected in a mixed spectrum together with PE P-18:0/18:2

HCD characterization: PE P-18:1/18:1 Detected as [M+H]+ (precursor m/z: 728.5591; retention time: 26.42 min; file: 05_pos_incl_long)

300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

265.127

279.177

281.227

307.273437.225444.293

462.455

464.292

658.468708.7712.163

Carboxy (18: 2)

NL_Ketene (18: 2)

Alkenyl (P-18: 1)NL_Carboxy (18: 1)

Carboxy (18: 1)

NL_Ketene (18: 1)

100 200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

114.179123.487 179.273 261.939299.672

337.274

339.29

366.369389.662390.282

392.294

429.264 551.586

728.597

729.547plasm alogen (P-18: 0)

sn2 (18: 2)

plasm alogen (P-18: 1)

sn2 (18: 1)



HCD characterization: PE P-18:1/18:1 Detected as [M-H]- (precursor m/z: 726.5441; retention time: 26.33 min; file: 04_neg_incl)


100 200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

100.437103.985117.991122.727150.252154.272183.598216.095234.965

279.233

281.248

284.602462.299

588.87

Carboxy (18: 2)

Carboxy (18: 1)

NL_Ketene (18: 1)


Novel species: PE P-18:1/20:4 Detected as [M+H]+ (precursor m/z: 750.5436; retention time: 26.46 min; file:

002_liver2-1_Orbitrap_CID_pos)

Novel species: PE P-18:1/20:4 Detected as [M–H]– (precursor m/z: 748.5281; retention time: 26.69 min; file:



200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

184.031292.228

361.238

390.268

604.467609.633

626.57

707.43

732.512NL_H2O

PEhead_WRONGPChead_WRONG NL_Ethanolam ine_43_WRONG


plasm alogen_H3PO4 (P-18: 1)

sn2 (20: 4)

200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

259.226

303.172

329.291436.311

444.234

462.289

NL_Carboxy (20: 4)

Carboxy (20: 4)

NL_Ketene (20: 4)

Carboxy (22: 5)NL_Ketene (22: 5)




HCD characterization: PE P-18:1/20:4 Detected as [M–H]– (precursor m/z: 748.529; retention time: 24.52 min; file: 04_neg_incl)


100 200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

184.074

203.18

292.3

361.274

364.26

387.29

390.278

750.594PChead_WRONG



sn2 (20: 4)


sn2 (22: 5)

100 200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

259.242

303.231

329.247444.289

462.298

748.534Precursor

NL_Carboxy (20: 4)

Carboxy (20: 4)

NL_Ketene (20: 4)

Carboxy (22: 5)


Novel species: PE P-18:1/22:4 Detected as [M–H]– (precursor m/z: 776.5603; retention time: 28.59 min; file:




200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

285.23313.301

327.472

329.228

331.355

357.293436.312446.329462.481

464.324

466.365

640.627 692.922702.13

718.035732.512756.075

Carboxy (22: 4)

NL_Ketene (22: 4)NL_Carboxy (22: 5)

Carboxy (22: 5)

NL_Ketene (22: 5)

100 200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

292.303

389.306

390.277 637.56

779.499

PEhead_WRONGplasm alogen (P-18: 1)


sn2 (22: 4)


HCD characterization: PE P-18:1/22:4 Detected as [M–H]– (precursor m/z: 776.56; retention time: 26.17 min; file: 04_neg_incl)

100 200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

327.232

331.263

444.288

462.297NL_Carboxy (22: 4)

Carboxy (22: 4)

NL_Ketene (22: 4)


Novel species: PE 17:0/17:0 Detected as [M–H]– (precursor m/z: 718.5392; retention time: 28.6 min; file:


Comment: detected in a mixed spectrum together with PE 18:0/16:0

HCD characterization: PE 17:0/17:0 Detected as [M–H]– (precursor m/z: 718.5391; retention time: 26.39 min; file: 04_neg_incl)

Comment: detected in a mixed spectrum together with PE 18:0/16:0

300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

255.249

269.253

283.284

452.365

462.452466.373

480.449

566.723650.273

658.324

700.755

NL_PChead_60_WRONG

Carboxy (18: 0)

NL_Ketene (18: 0)

NL_Carboxy (16: 0)

Carboxy (16: 0)

NL_Ketene (16: 0)

Carboxy (17: 0)

NL_Ketene (17: 0)

100 200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

115.063126.968150.174 208.005226.775

255.232

261.956

269.248

272.928

283.263

295.487333.198 498.878

718.537Precursor

Carboxy (18: 0)

Carboxy (16: 0)

Carboxy (17: 0)


Novel positional isomer: DG 16:0_0:0/18:2 Detected as [M+Na]+ (precursor m/z: 615.495; retention time: 28.98 min; file:





Comment: detected in a mixed spectrum together with DG 16:1/0:0/18:2

Novel positional isomer: DG 16:0_0:0/18:3 Detected as [M+Na]+ zoomed

Comment: detected in a mixed spectrum together with DG 16:1/0:0/18:2




200 300 400 500 600 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

267.548

305.296

309.301

339.352

341.387

361.392

363.268

389.421 459.44

460.37

470.579509.292 560.772

577.071

585.437587.509599.663623.919

626.827

627.675

628.585

NL_Carboxy (18: 1)NL_Carboxy_Na (18: 1)

Na_Carboxy (18: 1)

NL_Carboxy (18: 0)

NL_Carboxy_Na (18: 0)sn -3


Novel positional isomer: DG 18:1_0:0/18:1 and DG 18:0_0:0/18:2 Detected as [M+Na]+ (precursor m/z: 643.5273; retention time: 31.15 min; file:


Novel positional isomer: DG 18:1_0:0/18:1 and DG 18:0_0:0/18:2 Detected as [M+Na]+, zoomed




200 300 400 500 600 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

310.285

339.28

351.228

353.307361.358

385.428

407.273

432.172 621.101 671.558

NL_Carboxy (22: 6)NL_Carboxy_Na (22: 6)

Na_Carboxy (22: 6)

NL_Carboxy (18: 1)

NL_Carboxy_Na (18: 1)

sn-3


Novel species: PC 16:0_19:2 Detected as [M+H]+ (precursor m/z: 772.5846; retention time: 25.47 min; file:


Comment: detected in a mixed spectrum together with the species PC 17:0/18:2

Novel species: PC 16:0_19:2 Detected as [M+H]+, zoomed

460 480 500 520 540 m / z0

1

2

3

Rel

ativ

e ab

unda

nce

[%]

492.42

496.503

502.442

510.512

516.435

520.308

NL_Carboxy (17: 0)

NL_Ketene (17: 0)

NL_Carboxy (18: 2)

NL_Ketene (18: 2)

NL_Ketene (19: 2)

NL_Carboxy (16: 0)


HCD characterization: PC 16:0_19:2 Detected as [M+HCOO]- (precursor m/z: 816.576; retention time: 24.09 min; file: 04_neg_incl)


HCD characterization: PC 16:0_19:2 Detected as [M+HCOO]-, zoomed


100 200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

168.042224.067255.232267.232

269.247

279.232

281.248293.246476.307

494.323756.552

NL_PChead_60

PChead_168PChead_224

Carboxy (17: 0)

NL_Carboxy (18: 2)

Carboxy (18: 2)

NL_Ketene (18: 2)Carboxy (17: 1)Carboxy (18: 1)

Carboxy (16: 0)Carboxy (19: 2)






200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

183.955

512.468

585.534

627.62

709.456

PChead_184

NL- t r iMeth_WRONG

NL-Na_wrong

NL_PEhead_141_WRONGNL_Carboxy (16: 0)

200 300 400 500 600 700 m / z0

1

2

Rel

ativ

e ab

unda

nce

[%]

478.359496.219

512.468

627.62NL_PEhead_141_WRONG

NL_Carboxy (16: 0)

NL_Carboxy (19: 4)NL_Ketene (19: 4)



Novel species: PC 16:0_19:4 Detected as [M+HCOO]– (precursor m/z: 823.5452; retention time: 23.59 min; file:



HCD characterization: PC 16:0_19:4

300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

255.197289.19

725.571

752.556NL_PChead_60

NL_serine_wrong

Carboxy (19: 4)Carboxy (16: 0)

100 200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

168.044

241.216

255.229289.217

303.231

466.296 752.524NL_PChead_60PChead_168

Carboxy (15: 0)

Carboxy (20: 4)

NL_Ketene (20: 4)Carboxy (19: 4)Carboxy (16: 0)


Detected as [M+HCOO]– (precursor m/z: 812.545; retention time: 21.68 min; file: 04_neg_incl)







200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

184.019

502.422506.473516.347520.461524.509534.414

538.522617.575659.54

676.67

741.53

757.543

PChead_184

NL- t r iMeth_WRONG

NL_PEhead_141_WRONGNL-Na_wrongNL_Carboxy (19: 0)NL_Ketene (19: 0)

NL_Ketene (18: 2)

NL_Carboxy (18: 0)NL_Ketene (18: 0)NL_Carboxy (19: 2)NL_Ketene (19: 2)

200 300 400 500 600 700 800 m / z0

2.5

5

7.5

10

Rel

ativ

e ab

unda

nce

[%]

502.422

506.473

516.347

520.461

524.509

534.414

538.522

617.575659.54NL_PEhead_141_WRONGNL-Na_wrong

NL_Carboxy (19: 0)

NL_Ketene (19: 0)

NL_Ketene (18: 2)

NL_Carboxy (18: 0)NL_Ketene (18: 0)

NL_Carboxy (19: 2)

NL_Ketene (19: 2)






006_liver2-1_orbitrap_CID_pos)



200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

478.34496.297 550.254578.292627.578

649.534

651.567

747.478766.377

775.48

788.494

NL- t r iMeth_WRONG

NL-Na_wrong

NL_Carboxy (18: 0)NL_Carboxy (24: 6)NL_Ketene (24: 6)NL_Carboxy (16: 0)

400 450 500 550 m / z0

1

2

3

Rel

ativ

e ab

unda

nce

[%]

478.34

496.297

550.254

578.292

NL_Carboxy (18: 0)

NL_Carboxy (24: 6)

NL_Ketene (24: 6)

NL_Carboxy (16: 0)


Novel species: PC 16:0_24:6 Detected as [M+HCOO]– (precursor m/z: 878.591; retention time: 25.86 min; file:


Novel species: PC 16:0_24:6 Detected as [M+HCOO]–, zoomed

200 300 400 500 600 700 800 m / z0

10

20

30

Rel

ativ

e ab

unda

nce

[%]

255.326 355.308391.204522.141

546.183

568.151

791.634NL_serine_wrongNL_Carboxy (24: 6)Carboxy (24: 6)Lyso_PA-H2O (16: 0)

Carboxy (16: 0)


HCD characterization: PC 16:0_24:6 Detected as [M+H]+ (precursor m/z: 834.6011; retention time: 23.9 min; file: 05_pos_incl_long)

HCD characterization: PC 16:0_24:6 Detected as [M+H]+ zoomed



100 200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

255.232

303.232

307.263

355.262

442.545 532.34 718.524764.497

818.566878.49

NL_PChead_60

Carboxy (20: 2)

Carboxy (20: 4)

NL_Ketene (20: 4)Carboxy (16: 0)

Carboxy (24: 6)




HCD characterization: PC 18:1_24:6 Detected as [M+H]+ (precursor m/z: 860.6161; retention time: 24.3 min; file: 07_pos_incl_short)

200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

504.348506.384550.451

576.328

578.326 653.56674.593

675.569

676.614

677.597

740.557762.55

773.527792.344

800.53

801.537

808.631814.555

824.693

842.618

NL- t r iMeth_WRONG

NL-Na_wrong

NL_Carboxy (24: 6)NL_Carboxy (18: 1)

100 200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

125.001

184.073

504.334860.63

PChead_184

PrecursorPChead_125 NL_Carboxy (24: 6)




HCD characterization: PC 18:2_24:6 Detected as [M+H]+ (precursor m/z: 858.6011; retention time: 23.38 min; file: 04_pos_incl_long)

200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

502.426576.338578.325

673.584675.539

738.563

760.627

771.525784.577

799.508

800.556

806.516812.508

NL- t r iMeth_WRONG

NL-Na_wrong

NL_Carboxy (24: 6)NL_Carboxy (18: 2)

100 200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

184.073

520.35858.613

PChead_184

PrecursorNL_Ketene (24: 6)


HCD characterization: PC 18:2_24:6 Detected as [M+Na]+ (precursor m/z: 880.5831; retention time: 23.39 min; file: 04_pos_incl_long)


100 200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

106.532112.022146.979202.235249.707252.014287.647295.615312.402332.747 397.076 465.236 545.817

697.515

704.973745.976

793.473

837.342858.124

NL_PChead_183

PChead_147 NL_Carboxy_Head (24: 6)


Novel species: Cer 24:2; Cer d18:1/24:2 Detected as [M+H]+ (precursor m/z: 646.6132; retention time: 31.19 min; file:


Comment: This is a novel lipid species (m/z value of this ceramide is not reported in LipidMaps); the

SPH_fragment_264 verifies d18:1 to be the long chain base of the ceramide

HCD characterization: Cer 24:2; Cer d18:1/24:2

200 300 400 500 600 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

187.026221.144245.295252.267259.173

264.255

269.292

283.268

287.239

288.282

297.289

311.4

325.291

341.379

342.437

362.269

403.457433.573441.369446.491448.433460.324461.37

510.786

520.894525.369

546.829

578.471

579.139

585.559586.558588.602

627.173

628.654

629.586

633.055

NL_H2O

SPH_fragm ent_264

100 200 300 400 500 600 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

252.269

264.269

282.279

598.594610.592

628.603

SPH_fragm ent_264

PrecursorSPH_fragm ent_282

SPH_fragm ent_252NL_H2O

NL_form aldehyde_30


Detected as [M+H-H2O]+ (precursor m/z: 628.6031; retention time: 28.95 min; file: 07_pos_incl_short)

Comment: This is a novel lipid species (m/z value of this ceramide is not reported in LipidMaps); the

SPH_fragment_264 verifies d18:1 to be the long chain base of the ceramide.

HCD characterization: Cer 24:2; Cer d18:1/24:2 Detected as [M+HCOO]- (precursor m/z: 690.6041; retention time: 28.99 min; file: 04_neg_incl)

Comment: This is a novel lipid species (m/z value of this ceramide is not reported in LipidMaps).

G6550A QTOF characterization: Cer 24:2; Cer d18:1/24:2 Detected as [M+HCOO]- (precursor m/z: 690.608; retention time: 13.22 min; file: Sample2.1_neg_03)

Comment: This spectrum verifies the 24:2 fatty acyl chain.

100 200 300 400 500 600 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

105.622120.62127.764154.206177.242178.953191.918204.444237.22

248.937263.239 324.031325.667362.342

388.357

398.666426.16452.901453.425 585.063614.569626.613

644.596NL_form ic_acid_46

NL_sphingosine_302

NL_form aldehyde+ form ic_acid_76SPH_fragm ent2SPH_fragm ent1 NL_sphingosine_328


G6550A QTOF characterization: Cer 24:2; Cer d18:1/24:2 Detected as [M+HCOO]- zoomed

Comment: This spectrum verifies the 24:2 fatty acyl chain.


Novel species: TG 14:0_12:0_8:0 TG 17:0_15:0_2:0 TG 16:0_16:0_2:0 TG 16:0_14:0_4:0 TG 18:0_14:0_2:0 TG 18:0_12:0_4:0 TG 18:0_10:0_6:0 TG 14:0_14:0_6:0

Detected as [M+NH4]+ (precursor m/z: 628.5503; retention time: 32.01 min; file:


Comment: detected in a mixed spectrum with co-eluting structural isomers. Evidence is based on

detected fatty acyl chains and combinations thereof.

200 300 400 500 600 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

309.236327.315

341.342342.389

355.236

369.292

383.324

411.325

415.238

439.362

467.43

495.468

516.405

523.456

551.589560.145561.01

610.6

611.533NL_NH3

NL_Carboxy_NH3 (16: 0)












Novel species: TG 18:0_16:0_2:0 TG 18:0_14:0_4:0 TG 18:0_12:0_6:0 TG 18:0_10:0_8:0 TG 14:0_14:0_8:0 TG 16:0_16:0_4:0 TG 16:0_14:0_6:0 TG 16:0_12:0_8:0





200 300 400 500 600 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

355.331

383.316

411.412

439.355

467.402495.482

523.481

551.561

579.616588.706638.784639.565NL_NH3




NL_Carboxy_NH3 (8: 0)NL_Carboxy_NH3 (10: 0)






Novel species: TG 18:0_10:0_8:0 TG 17:0_15:0_4:0





200 300 400 500 600 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

257.288

355.283

369.381

383.35

397.454

411.39

439.439

467.555495.452

509.52

523.57

537.628

551.572

579.64638.588639.615NL_NH3



MG_fragm ent (12: 0)











200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

411.313

439.367467.43

495.442

496.048

523.48

551.51

694.744

695.495NL_NH3








Novel species: TG 16:0_16:0_11:0 Detected as [M+NH4]+ (precursor m/z: 754.6907; retention time: 38.39 min; file:




200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

467.412

481.411

495.444

509.478

523.473

537.496

551.541736.712








Novel species: TG 18:0_16:1_10:0 and TG 17:0_16:1_11:0 Detected as [M+NH4]+ (precursor m/z: 766.691; retention time: 38.16 min; file:




Novel species: TG 18:0_16:1_10:0 and TG 17:0_16:1_11:0 Detected as [M+NH4]+ zoomed

200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

465.394

467.427

469.286479.518

493.475

495.48

507.478509.471

521.502

523.503

535.548

549.507

563.682

577.528605.669

706.662749.783NL_NH3















200 300 400 500 600 700 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

285.207299.283313.288 479.551

481.441493.534495.441

507.455

509.473

521.467523.473

535.5

537.49

549.479563.55

577.716 763.588NL_NH3



















200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

263.209285.244

519.438

521.475

523.458

545.447

547.46

549.418

573.449

575.488

601.512

783.759

801.635NL_NH3

NL_NH3_H2O


Carboxy_OH (18: 2)















200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

243.172261.133285.16 337.239378.193

412.221

495.477

523.513

541.49543.495

545.463

547.456

549.462569.493

571.443

573.512

575.484

597.47

599.546

601.619

738.571

783.55

807.648

825.703NL_NH3

NL_NH3_H2O

Carboxy_H2O_OH (18: 3)


Carboxy_OH (18: 3)










NL_Carboxy_NH3 (18: 1)NL_Carboxy_NH3 (18: 0)NL_Carboxy_NH3 (14: 2)



Detected as [M+NH4]+ zoomed






Novel species: TG 20:3_16:2_14:0 Detected as [M+NH4]+ zoomed






200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

263.4 313.35 547.536549.499

561.514

563.535

573.554

575.531

577.591

587.553

589.662

601.563

825.855

843.745NL_NH3

NL_NH3_H2O


Carboxy_OH (18: 2)





MG_fragm ent (16: 0)NL_Carboxy_NH3 (17: 0)




Novel species: TG 24:0_16:0_13:0 TG 26:0_14:0_13:0 TG 23:0_17:0_13:0 TG 23:0_15:0_15:0 TG 25:0_15:0_13:0 TG 24:0_17:0_12:0 TG 23:0_16:0_14:0





200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

481.462

495.441

509.444

523.525537.482

551.506565.588

579.559

593.516

599.576

607.596

621.587

635.564

649.596

663.673

677.744

826.217877.665NL_NH3





















Novel species: TG 19:2_17:1_17:0 Detected as [M+NH4]+, zoomed

200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

263.09 327.344339.221

575.509

577.558587.607

589.503591.553

601.532

603.564

615.543

853.794

871.814NL_NH3

NL_NH3_H2O


Carboxy_OH (18: 2)









560 580 600 620 640 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

575.537

577.524 587.472

589.52

591.556

601.51

603.533 615.552







Novel species: TG 19:3_17:1_17:0 Detected as [M+Na]+ (precursor m/z: 891.7417; retention time: 40.16 min; file:




Novel species: TG 19:3_17:1_17:0 Detected as [M+Na]+ zoomed

200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

575.576577.5

587.507

589.526

597.539

599.461601.522

609.48

611.475

613.54

621.475623.504

635.505

NL_Carboxy (18: 2)


NL_Carboxy (18: 1)


NL_Carboxy (17: 1)


NL_Carboxy (17: 0)


NL_Carboxy_Na (19: 3)NL_Carboxy (19: 2)

NL_Carboxy_Na (19: 2)NL_Carboxy (16: 0)








200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

325.272337.321 573.588

585.587

587.574

589.61

597.625

599.591

601.574849.835

867.851NL_NH3

NL_NH3_H2O








560 570 580 590 600 610 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

573.588

585.587

587.574

589.61

597.625

599.591

601.574











200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

261.183 325.291337.286339.266561.559

575.581

585.547

587.47

589.491

597.546

599.454

601.494

625.536

849.72

867.72NL_NH3

NL_NH3_H2O









Carboxy_OH (18: 3)








200 300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

285.25293.24311.295337.235

547.548

571.544

595.522

619.632621.641623.671

647.571

857.796

875.815NL_NH3

NL_NH3_H2OCarboxy_H2O_OH (22: 6)


Carboxy_OH (22: 6)




MG_fragm ent (14: 0)NL_Carboxy_NH3 (16: 1)NL_Carboxy_NH3 (16: 2)NL_Carboxy_NH3 (16: 0)


Novel species: TG 24:0_17:0_14:0 TG 23:0_16:0_16:0 TG 26:0_16:0_13:0 TG 23:0_17:0_15:0





200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

495.517509.534

523.565

537.516

551.563

565.648577.641579.658

621.633

635.694

649.648

663.689

677.699

691.73705.745 905.766NL_NH3


















200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

577.495

591.584

603.537

605.511

617.522

619.58

621.648

631.588645.568

883.795901.921NL_NH3

NL_NH3_H2O















200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

589.612591.598

601.55

603.563

605.642

615.554

617.558

619.617

629.572631.669881.938

899.858NL_NH3

NL_NH3_H2O





NL_Carboxy_NH3 (20: 2)NL_Carboxy_NH3 (17: 0)NL_Carboxy_NH3 (20: 1)NL_Carboxy_NH3 (17: 1)








300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

339.316589.564591.671

599.555

601.582

603.639

615.626

617.633

619.584

627.644629.617

631.745641.624829.888

879.871

897.875NL_NH3

NL_NH3_H2O








570 580 590 600 610 620 630 640 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

589.564591.671

599.555

601.582

603.639

615.626

617.633

619.584

627.644629.617

631.745 641.624






NL_Carboxy_NH3 (17: 2)NL_Carboxy_NH3 (18: 3)NL_Carboxy_NH3 (20: 3) NL_Carboxy_NH3 (16: 0)







300 400 500 600 700 800 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

261.206575.489587.515

589.522

597.49

599.529

601.52

613.519

615.499

616.196

617.577625.517627.493

639.567 827.701

854.718

877.78

895.772NL_NH3

NL_NH3_H2O


















200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

523.513

537.467

551.498

565.616579.595

635.6649.669

663.609

677.651

691.636NL_Carboxy_NH3 (25: 0)













200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

311.293425.397 521.462

535.477

549.509

551.557563.548

577.587605.58

633.585

635.64

647.728

649.65

661.639

663.646

664.346

675.663

677.631

689.699

691.788 916.889























300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

311.203 535.562

549.608

551.627

563.589

577.652

579.651

605.628

607.598

633.655

635.661

647.702

649.775

661.632663.67

675.758

677.74

689.702

703.603899.321

916.958

917.9NL_NH3





















200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

293.21311.308335.287337.342

573.52

619.537

621.503

623.517645.516

647.578

649.68883.708

901.764NL_NH3

NL_NH3_H2OCarboxy_H2O_OH (22: 6)


Carboxy_OH (22: 6)




NL_Carboxy_NH3 (18: 3)MG_fragm ent (18: 3)



Novel species: TG 25:0_17:0_15:0 TG 24:0_17:0_16:0 TG 23:0_18:0_16:0





300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

523.566

537.513

551.619

565.602

579.552

649.683

663.666

677.668

691.674

705.67NL_Carboxy_NH3 (24: 0)











Novel species: TG 25:0_17:0_15:0 TG 24:0_17:0_16:0 TG 23:0_18:0_16:0



Novel species: TG 24:0_18:0_16:0 TG 27:0_16:0_15:0 TG 26:0_16:0_16:0 TG 28:0_16:0_14:0 TG 25:0_17:0_16:0





200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

523.584537.583

551.497

565.571579.6

607.683635.68

663.677677.699

691.735

705.782

719.748



















Novel species: TG 24:2_18:1_16:0

200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

341.376

575.583577.603

601.572

603.582

605.607629.575

631.653

633.679

657.632

659.638

661.682

685.638

687.652 923.978

941.997NL_NH3

NL_NH3_H2O












TG 24:1_18:1_16:1 Detected as [M+NH4]+ zoomed







200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

339.315341.355575.552577.648

599.56

601.596

603.57

605.665627.705

629.644

631.65

633.65

655.645

657.656

659.674

683.68

685.689

921.912

939.952NL_NH3

NL_NH3_H2O












560 580 600 620 640 660 680 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

575.59577.57

599.558

601.547

603.545

605.545

607.646

627.55

629.6

631.575633.7

655.587

657.608

659.633

661.575

683.675

685.624











NL_Carboxy_NH3 (20: 3) NL_Carboxy_NH3 (16: 0)

NL_Carboxy_NH3 (24: 3) NL_Carboxy_NH3 (18: 3)NL_Carboxy_NH3 (20: 0)







200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

297.246315.312

339.312341.306343.377

577.564

603.519

605.545

627.559

629.63

631.563

653.572

655.595

657.622681.631919.805

937.821NL_NH3

NL_NH3_H2O



Carboxy_OH (22: 4)







Carboxy_OH (24: 4)







200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

295.286311.285

313.256315.32337.304

339.325

341.347367.407

575.569

577.562

601.582603.647

623.529625.607

629.602

649.533

651.575

653.592

677.594

915.909

933.819NL_NH3

NL_NH3_H2O


Carboxy_OH (22: 4)







Carboxy_OH (22: 5) NL_Carboxy_NH3 (18: 0)MG_fragm ent (18: 0)Carboxy_OH (22: 6)



NL_Carboxy_NH3 (20: 2)NL_Carboxy_NH3 (20: 1)MG_fragm ent (20: 1)



Novel species: TG 24:5_18:1_18:0 TG 26:5_18:1_16:0 TG 24:4_18:2_18:0 TG 24:4_18:1_18:1 TG 24:4_20:2_16:0 TG 24:5_20:1_16:0





300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

315.423339.377343.279367.423

577.526

603.533

605.589629.574

631.585

651.669

653.583

655.667

679.561

681.581

683.624

707.625 895.828

945.853

963.88NL_NH3

NL_NH3_H2O


Carboxy_OH (22: 4)






Carboxy_OH (24: 4)















200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

315.311321.288323.267

337.345

339.276341.335

343.364601.575

603.579

605.621

629.651

631.682651.581653.693655.631

657.619

677.607

679.638

681.635

705.728

943.84

960.914

961.901NL_NH3

NL_NH3_H2O


Carboxy_OH (22: 4)NL_Carboxy_NH3 (20: 2)








NL_Carboxy_NH3 (24: 4)Carboxy_OH (24: 4)


MG_fragm ent (18: 2) NL_Carboxy_NH3 (16: 0)










200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

313.28321.268323.279337.305

339.356

341.375

601.554

603.529

627.586

629.656

631.702649.604651.674

653.589655.625

675.643

677.58

679.589

681.707703.723

941.881

959.83NL_NH3

NL_NH3_H2O



Carboxy_OH (24: 6)




Carboxy_OH (24: 5)



Carboxy_OH (22: 5) NL_Carboxy_NH3 (20: 2)







Detected as [M+NH4]+ (zoomed)











Same spectrum as before, but zoomed for 17:3 and 18:3




Comment: detected in a mixed spectrum with co-eluting structural isomers. Evidence is based on detected fatty acyl chains and combinations thereof.

200 300 400 500 600 700 800 900 m / z0

25

50

75

100

Rel

ativ

e ab

unda

nce

[%]

269.303293.3311.3313.236

337.269339.406341.314

599.551

623.583625.58

626.713

627.574

647.572

649.595

651.586

673.659

675.628

699.589

937.857

955.866NL_NH3

NL_NH3_H2O



Carboxy_OH (22: 6)Carboxy_H2O_OH (20: 4)


MG_fragm ent (18: 0)NL_Carboxy_NH3 (20: 3)NL_Carboxy_NH3 (18: 1)MG_fragm ent (18: 1)





NL_Carboxy_NH3 (16: 0)MG_fragm ent (16: 0)



Supplementary Figure 3: Co-eluting regio-isomers of PE 16:0_20:4 observed on Orbitrap Velos Pro in negative ion mode.

Reference standard PE 16:0/20:4 is contaminated with PE 20:4/16:0: Partial chromatographic separation is shown in extracted ion chromatogram a. Intensity relations of position-indicative Carboxy and NL_Ketene fragments are different in the spectra in b. In the spectrum at left of the pre-eluting local maximum (first blue line; precursor m/z 738.508; retention time 24.20 min; file: 017_Ex1_Orbitrap_CID_neg_50), the high intensities of fragments shown in red indicate 16:0 to be at the sn-2 position. In the spectrum at the right side of the main peak (second blue line; precursor m/z 738.508; retention time 24.58 min; file 017_Ex1_Orbitrap_CID_neg_50), fragments shown in blue reveal stronger intensities than those in red, indicating 20:4 to be at the sn-2 position.


Supplementary Figure 4: Correct annotation of chromatographically separated DG 16:0/18:1/0:0 (sn-1,2 DG) and DG 16:0/0:0/18:1 (sn-1,3 DG). Samples with the individual or both standards were measured on an Orbitrap Velos Pro in CID mode.

a Extracted ion chromatogram of DG 16:0/0:0/18:1. b extracted ion chromatogram of DG 16:0/18:1/0:0. Both reference standards shown in a and b are contaminated by the other regio-isomer (small peaks). c Extracted ion chromatogram of 1:1 mixture of DG 16:0/0:0/18:1 and DG 16:0/18:1/0:0. Both regio-isomers are almost baseline separated. d Partial MS/MS spectra (m/z 250-400) of two sodiated DGs are shown. Fragments indicative for each fatty acyl chain are colored in red and blue, respectively: d left spectrum (precursor m/z 617.510; retention time 68.64 min; file: Mix_5uM-95min-1), sn-1,3-DG: Fatty acyls at sn-3 are characterized by Na_Carboxy fragments whose intensity is slightly less or higher than the neutral loss fragment (NL_Carboxy_Na) of the same chain; d right spectrum (precursor m/z 617.510; retention time 69.11 min; file: Mix_5uM-95min-1), sn-1,2-DG: Na_Carboxy fragments at sn-1 and sn-2 are less intense than their corresponding neutral loss fragment. In d left spectrum, the Na_Carboxy fragment at sn-1 is not detected.


Supplementary Figure 5: Lipid nomenclature based on the level of structural detail for the example of PI 18:0/20:44.

MS1 data report potential candidates for lipid species because of a matching mass of the intact lipid in the sum nomenclature PI 38:4. MS/MS spectra can provide several levels of structural evidence. First, head group fragments verify the lipid class and differentiate the identification from other lipid subclasses/adducts. Second, chain fragments reveal the constituent fatty acyl chains (PI 20:4_18:0). Third, intensity relationships of chain fragments indicate the sn-positions of the chains at the glycerol backbone (PI 18:0/20:4).

PI 38:4 MS1

Head group fragmentsPI 38:4

PI 20:4_18:0

Intensity of chain fragmentsPI 18:0/20:4

Lipid nomenclature Structural level

Lipid species

MS evidence

Lipid species

Lipid molecular species (fatty acyl level)

Lipid molecular species (fatty acyl position level)

Chain fragments


Supplementary Figure 6: Levels of structural information deducible from MS/MS spectra for the example of PI 18:0/20:4.

An annotated MS/MS spectrum (precursor m/z 885.55) is shown of deprotonated PI 18:0/20:4 detected in a murine liver sample acquired on an Orbitrap Velos Pro in CID mode. The fragments in brown color (1-4) are characteristic for the head group of deprotonated PI. The fragments in blue (5, 9, 10, 12) and red color (6-8, 11) characterize the constituent fatty acyl chains 18:0 and 20:4, respectively. Intensity relationships (5>6, 7>9, 8>10, 11>12) indicate 18:0 and 20:4 to be located on sn-1 and sn-2, respectively8.

PI 18:0/20:4 – negative ion mode 1-4: head group

1

2

3

4

PI5, 6: carboxylates of chains

5

12

9

10

18:0 7, 9: neutral loss of chain as fatty acid

8, 10: neutral loss of chain as ketene body

11, 12: neutral loss of chain and head group

6

11

7

8

20:4

• charge driven processes favor NL at sn-2

• secondary fragmentation processes favor the formation of carboxylates of FA at sn-1

/


Supplementary Figure 7: Graphical user interface for decision rule set adaption/development.

LDA software provides a graphical user interface (GUI) for adaption of existing rules and creation of novel rules. Spectra are annotated in real time while rules are entered, and detailed arguments are provided for refused annotations. See detailed information in the user manual: http://genome.tugraz.at/lda2/lda_documentation.shtml. The annotated MS/MS spectrum shows a deprotonated PI 18:0/20:4 detected in a murine liver sample acquired on an Orbitrap Velos Pro in CID mode (precursor m/z 885.549; retention time 24.86 min; file 002_liver2-1_Orbitrap_CID_neg).


http://genome.tugraz.at/lda2/lda_documentation.shtml

[GENERAL] AmountOfChains=2 ChainLibrary=fattyAcidChains.xlsx CAtomsFromName=\D*(\d+):\d+ DoubleBondsFromName=\D*\d+:(\d+) BasePeakCutoff=0.1% SpectrumCoverage=10% RetentionTimePostprocessing=true

[HEAD] !FRAGMENTS Name=PL153 Formula=C3H6O5P Charge=1 MSLevel=2 mandatory=false Name=PIhead_223 Formula=C6H8O7P Charge=1 MSLevel=2 mandatory=false Name=PIhead_241 Formula=C6H10O8P Charge=1 MSLevel=2 mandatory=true Name=PIhead_259 Formula=C6H12O9P Charge=1 MSLevel=2 mandatory=false Name=PIhead_297 Formula=C9H14O9P Charge=1 MSLevel=2 mandatory=false Name=PIhead_315 Formula=C9H16O10P Charge=1 MSLevel=2 mandatory=false Name=NL_Inositol_162 Formula=$PRECURSOR-C6H10O5 Charge=1 MSLevel=2 mandatory=false

!INTENSITIES Equation=PIhead_241*0.9>PIhead_223 mandatory=false Equation=PIhead_241*0.9>PIhead_297 mandatory=false

[CHAINS] !FRAGMENTS Name=Carboxy Formula=$CHAIN-H Charge=1 MSLevel=2 mandatory=false Name=NL_Carboxy Formula=$PRECURSOR-$CHAIN Charge=1 MSLevel=2 mandatory=true Name=NL_Ketene Formula=$PRECURSOR-$CHAIN+H2O Charge=1 MSLevel=2 mandatory=false Name=NL_Carboxy_Ino Formula=NL_Carboxy-C6H10O5 Charge=1 MSLevel=2 mandatory=false Name=NL_Ketene_Ino Formula=NL_Ketene-C6H10O5 Charge=1 MSLevel=2 mandatory=false

!INTENSITIES Equation=NL_Carboxy*0.8>NL_Ketene mandatory=true Equation=Carboxy+NL_Carboxy_Ino+NL_Carboxy*0>0*$BASEPEAK mandatory=true

[POSITION] !INTENSITIES Equation=Carboxy[1]*0.9>Carboxy[2] mandatory=true Equation=NL_Carboxy[2]*0.5>NL_Carboxy[1] mandatory=true Equation=NL_Ketene[2]*0.8>NL_Ketene[1] mandatory=false Equation=NL_Carboxy_Ino[2]*0.8>NL_Carboxy_Ino[1] mandatory=true Equation=NL_Ketene_Ino[2]*0.8>NL_Ketene_Ino[1] mandatory=false

Supplementary Figure 8: Complete decision rule set for deprotonated PI.

The decision rule set shown is for annotation of deprotonated PI at a collision energy setting of -50%. Decision rule sets can contain up to four sections: [GENERAL], section containing information about lipid subclass/adduct; [HEAD], [CHAINS] and [POSITION] sections for verifying levels of structural details, i.e., [HEAD] for verification of lipid subclass, [CHAINS] for constituent fatty acyl and/or alkyl/1-alkenyl chains, and [POSITION] for assignment of sn-positions. The latter three sections provide fragment and intensity rules whose subsections are headed by the terms ‘!FRAGMENTS’ and ‘!INTENSITIES’. Depending on lipid structure, the latter three sections may not apply.


Supplementary Figure 9: Decision rule check for lipid head groups ([HEAD]).

Decision rule check for head fragments of PI 38:4. a shows fragment definitions and detected peaks in a real MS/MS spectrum (precursor m/z 885.549; retention time 24.86 min; file 002_liver2-1_Orbitrap_CID_neg). The fragment PIhead_241 is mandatory. If this fragment is not detected, the algorithm decides that this spectrum does not fit lipid class PI. Consequently, the prior MS1 hit is discarded. b shows the follow-up intensity rules, and the comparison of mandatory and optional fragments. This example is for illustration only, as deprotonated PI intensity rules are not required due to command “mandatory=false”.

Mandatory head fragments found

Check head intensity rules

Discard MS1 hit

yes no

[HEAD]!FRAGMENTSName=PL153 Formula=C3H6O5P Charge=1 MSLevel=2 mandatory=falseName=PIhead_223 Formula=C6H8O7P Charge=1 MSLevel=2 mandatory=falseName=PIhead_241 Formula=C6H10O8P Charge=1 MSLevel=2 mandatory=trueName=PIhead_259 Formula=C6H12O9P Charge=1 MSLevel=2 mandatory=falseName=PIhead_297 Formula=C9H14O9P Charge=1 MSLevel=2 mandatory=falseName=PIhead_315 Formula=C9H16O10P Charge=1 MSLevel=2 mandatory=falseName=NL_Inositol_162 Formula=$PRECURSOR-C6H10O5 Charge=1 MSLevel=2 mandatory=false

a

!INTENSITIESEquation=PIhead_241*0.9>PIhead_223 mandatory=falseEquation=PIhead_241*0.9>PIhead_297 mandatory=false

PI 38:4

Mandatory intensityrules fulfilled

Continue with chains

Discard MS1 hit

yes no

b


Supplementary Figure 10: Decision rule check for constituent acyl or alkyl/1-alkenyl chains ([CHAINS]).

The decision rule check and its effects are depicted for the 20:4 fatty acyl chain in PI 18:0/20:4. a, fragment definitions and their detected peaks in a real spectrum (precursor m/z 885.549; retention time 24.86 min; file 002_liver2-1_Orbitrap_CID_neg). The ‘NL_Carboxy’ fragment is mandatory. If this fragment is not detected, acyl chain 20:4 is found to be inappropriate for this detection. b, defined intensity rules refer to affected spectra. The second intensity rule is an OR connection requiring the ‘Carboxy’ or the ‘NL_Carboxy_Ino’ fragment to be found in addition to mandatory ‘NL_Carboxy’ fragment. When a mandatory chain intensity rule is violated, the corresponding chain is discarded.

a

b

Check chain (e.g. 20:4)

NL-Carboxy 20:4?

NL_Carboxy_Ino 20:4?

Carboxy 20:4?

NL-Ketene_Ino 20:4?

yes

yes

yes

Discard 20:4

no

no

Check chain intensity rules of 20:4

yes

no

no

NL_Ketene 20:4?

yes

no

[CHAINS]!FRAGMENTSName=Carboxy Formula=$CHAIN-H Charge=1 MSLevel=2 mandatory=falseName=NL_Carboxy Formula=$PRECURSOR-$CHAIN Charge=1 MSLevel=2 mandatory=trueName=NL_Ketene Formula=$PRECURSOR-$CHAIN+H2O Charge=1 MSLevel=2 mandatory=falseName=NL_Carboxy_Ino Formula=NL_Carboxy-C6H10O5 Charge=1 MSLevel=2 mandatory=falseName=NL_Ketene_Ino Formula=NL_Ketene-C6H10O5 Charge=1 MSLevel=2 mandatory=false

!INTENSITIESEquation=NL_Carboxy*0.8>NL_Ketene mandatory=trueEquation=Carboxy+NL_Carboxy_Ino+NL_Carboxy*0>0*$BASEPEAKmandatory=true

Mandatory intensity rules fulfilled

Chain is valid

yes no

Discard 20:4


Supplementary Figure 11: Decision rule check for sn-positions of acyl or alkyl/1-alkenyl chains ([POSITION]).

The decision rule check for position assignment is presented for PI 18:0/20:4. a shows intensity rules, their corresponding fragments in the spectrum (precursor m/z 885.549; retention time 24.86 min; file 002_liver2-1_Orbitrap_CID_neg), and position recommendations derived. b shows flow chart for processing recommendations. If mandatory rules apply, all of them must be fulfilled for sn-position assignment. In the case of optional rules only, the sn-position is assigned according to the majority of rules fulfilled. In case of a tie, no positional assignment is possible. For subclasses containing more than two acyl and/or alkyl/1-alkenyl chains, several candidates might vie for the same sn-position. In that case, the chain with the most recommendations is assigned.

a

b

[POSITION]!INTENSITIESEquation=Carboxy[1]*0.9>Carboxy[2] mandatory=trueEquation=NL-Carboxy[2]*0.5>NL-Carboxy[1] mandatory=trueEquation=NL-Ketene[2]*0.8>NL-Ketene[1] mandatory=falseEquation=NL_Carboxy_Ino[2]*0.8>NL_Carboxy_Ino[1] mandatory=trueEquation=NL_Ketene_Ino[2]*0.8>NL_Ketene_Ino[1] mandatory=false

1 = 18:02= 20:4

2= 20:4 1 = 18:0

2= 20:4 1 = 18:0

2= 20:4 1 = 18:0

No NL_Ketene_Ino (18:0) found – no effect

Assign multiple recommendations (higher evidence)

PI 20:4_18:0

Are there anymandatory rules?

yes

No position assignment possible

no

PI 20:4_18:0

PI 18:0/20:4

Mandatory rules fulfilled?

yes

Check for optional rules

noMore optional rule evidence?

no

Assign distinct identifications

yes


Supplementary Figure 12: Annotation of a mixed spectrum of a pair of structural isomers.

A mixed spectrum for co-eluting PI 18:1/20:4 and 18:0/20:5 is shown. The spectrum was detected in a murine liver sample acquired on an Orbitrap Velos Pro in CID mode (precursor m/z 883.534, retention time 23.09 min; file 002_liver2-1_Orbitrap_CID_neg). a shows the extracted ion chromatogram where a single peak for PI 38:5 is detected and the area under the curve (AUC) is assigned. b shows the spectrum at the retention time indicated by the thin blue line in a. The algorithm unambiguously detects both lipid molecular species in the mixed spectrum in b. Each molecular species obtains its share of the MS1 AUC calculated from the intensities of the distinct fragments.

m/z

PI 38:5

PI 18:1/20:4PI 18:0/20:5∑ int(18:1 frags) + ∑ int(20:4 frags)

∑ int(18:0 frags) + ∑ int(20:5 frags)

a

b


Supplementary Figure 13: Graphical representation of data from pairs of structural isomers identified in control experiment 3.

Overview of measured average abundance ratios of pairs of structural isomers present in known ratios in a sample (from Supplementary Table 5) in positive (a-e) and negative (f-j) ion mode (control experiment 3). Data were evaluated on four platforms in quintuplicates. The average ratio obtained on each platform is displayed and the expected ratio is indicated by the red dashed line. Error bars denote the standard deviation.

PC 32:0 PC 34:0 PC 36:2 PE 36:2 PE 36:4 PG 36:2 PS 36:2 TG 50:0

Lipid Species

Av

erag

e R

ati

o

0.0

0.5

1.0

1.5

2.0

2.5

3.0

22

.92

■■■■-

Orbitrap Velos Pro CIDOrbitrap Velos Pro HCD4000 QTRAPSYNAPT G1 HDMS QTOFExpected ratio

a Expected ratio: 1

PC 32:0 PC 34:0 PC 36:2 PE 36:2 PE 36:4 PG 36:2 PS 36:2

Lipid Species

Av

erag

e R

ati

o

0.0

0.5

1.0

1.5

2.0

2.5

3.0f Expected ratio: 1


Lipid Species

Av

erag

e R

ati

o

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.8

0

0.6

3

0.7

6

0.6

3

b Expected ratio: 0.2

PC 32:0 PC 34:0 PC 36:2 PE 36:2 PE 36:4 PG 36:2 PS 36:2

Lipid Species

Av

erag

e R

ati

o

0.0

0.1

0.2

0.3

0.4

0.5

0.6g Expected ratio: 0.2


Lipid Species

Ave

rage

Ra

tio

0

2

4

6

8

10

12

14

27.5

8

39.4

3

28.5

5

c Expected ratio: 5

PC 32:0 PC 34:0 PC 36:2 PE 36:2 PE 36:4 PG 36:2 PS 36:2

Lipid Species

Ave

rage

Ra

tio

0

2

4

6

8

10

12

14

h Expected ratio: 5


Lipid Species

Av

erag

e R

atio

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.3

10

.22

0.3

0

0.4

1

d Expected ratio: 0.05

PC 32:0 PC 34:0 PC 36:2 PE 36:2 PE 36:4 PG 36:2 PS 36:2

Lipid Species

Av

erag

e R

atio

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.140

.16

0.1

8

0.2

2

i Expected ratio: 0.05


Lipid Species

Av

erag

e R

ati

o

0

10

20

30

40

50

60

80.

98

97.

27

e Expected ratio: 20

PC 32:0 PC 34:0 PC 36:2 PE 36:2 PE 36:4 PG 36:2 PS 36:2

Lipid Species

Av

erag

e R

ati

o

0

10

20

30

40

50

60j Expected ratio: 20


Supplementary Figure 14: MS/MS spectra allow for MS1 peak splitting of overlapping isomeric lipid subclasses/adducts.

Spectra are obtained from control experiment 2 acquired on an Orbitrap Velops Pro in CID mode. a 3D view of an MS1 peak consisting of overlapping sodiated PC 17:0/14:1 and sodiated PE 16:0/18:1; b extracted ion chromatogram shows that the MS/MS spectrum at an earlier time point (first blue line; precursor m/z 740.520; retention time 0.38 min; file 001_Ex2_Orbitrap_CID) accounts for PC species, whereas the MS/MS spectrum at a later time (second blue line; precursor m/z 740.520; retention time 0.47 min; file 001_Ex2_Orbitrap_CID) point accounts for PE species; c MS1 peak is split according to weighted peak maxima of distinct fragments.

PC spectrum PE spectrum

PC fraction

PE fraction

One MS1 peak containing PC 17:0/14:1 and PE 16:0/18:1

2 MS/MS spectra at different time points

a

b

c


References

27. Zierler, K.A. et al. Functional cardiac lipolysis in mice critically depends on comparative gene identification-58. J. Biol. Chem. 288, 9892-9904 (2013).

28. Matyash, V., Liebisch, G., Kurzchalia, T.V., Shevchenko, A. & Schwudke, D. Lipid extraction by methyl-tert-butyl ether for high-throughput lipidomics. J. Lipid Res. 49, 1137-1146 (2008).

29. Triebl, A., Trötzmüller, M., Hartler, J., Stojakovic, T. & Köfeler, H.C. Lipidomics by ultrahigh performance liquid chromatography-high resolution mass spectrometry and its application to complex biological samples. J Chromatogr. B Analyt. Technol. Biomed. Life Sci. 1053, 72-80 (2017).

30. Knittelfelder, O.L., Weberhofer, B.P., Eichmann, T.O., Kohlwein, S.D. & Rechberger, G.N. A versatile ultra-high performance LC-MS method for lipid profiling. J Chromatogr. B Analyt. Technol. Biomed. Life Sci. 951-952, 119-128 (2014).

31. Xu, H. & Freitas, M.A. A dynamic noise level algorithm for spectral screening of peptide MS/MS spectra. BMC Bioinformatics 11, 436 (2010).

32. Levenberg, K. A method for the solution of certain non-linear problems in least squares. Quart. Applied. Math. 2, 164-168 (1944).

33. Marquardt, D.W. An Algorithm for Least-Squares Estimation of Nonlinear Parameters. J. Soc. Indust. Appl. Math. 11, 431-441 (1963).


Documents

SUPPLEMENT - Nature Research · Zeleznik, Kathrin A Zierler, Federico Torta, Amaury Cazenave-Gassiot, Markus R Wenk, Alexander Fauland, Craig E Wheelock, Aaron M Armando, Oswald Quehenberger,