A Novel Lipidomics Workflow for Improved Human Plasma

Apr 30, 2018 - Institute of Bioanalytical Chemistry, Faculty of Chemistry and Mineralogy, ... With the advent of the ultrahigh-performance chromatogra...
1 downloads 4 Views 1MB Size
Subscriber access provided by UNIV OF NEW ENGLAND ARMIDALE

Article

A novel lipidomics workflow for improved human plasma identification and quantification using RPLC-MSn methods and isotope dilution strategies Evelyn Rampler, Angela Criscuolo, Martin Zeller, Yasin El Abiead, Harald Schoeny, Gerrit Hermann, Elena Sokol, Ken Cook, David A. Peake, Bernard Delanghe, and Gunda Koellensperger Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.7b05382 • Publication Date (Web): 30 Apr 2018 Downloaded from http://pubs.acs.org on May 2, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

A novel lipidomics workflow for improved human plasma identification and quantification using RPLC-MSn methods and isotope dilution strategies Evelyn Rampler1,2,3,*, Angela Criscuolo4,8, Martin Zeller4, Yasin El Abiead1, Harald Schoeny1, Gerrit Hermann1,5, Elena Sokol6, Ken Cook6, David A. Peake7, Bernard Delanghe4, Gunda Koellensperger1,2,3 1

Department of Analytical Chemistry, Faculty of Chemistry, University of Vienna, Währingerstr. 38, 1090

Vienna, Austria 2

Vienna Metabolomics Center (VIME), University of Vienna, Althanstraße 14, 1090 Vienna, Austria

3

Chemistry Meets Microbiology, Althanstraße 14, 1090 Vienna, Austria

4

Thermo Fisher Scientific (Bremen GmbH), Hanna-Kunath-Str. 11, 28199 Bremen, Germany

5

ISOtopic Solutions, Währingerstr. 38, 1090 Vienna, Austria

6

Thermo Fisher Scientific, 1 Boundary Park, Hemel Hempstead HP2 7GE, United Kingdom

7

Thermo Fisher Scientific, 355 River Oaks Parkway, 95134 San Jose, California, USA

8

Institute of Bioanalytical Chemistry, Faculty of Chemistry and Mineralogy, Universität Leipzig, Leipzig, Germany

Abstract Lipid identification and quantification are essential objectives in comprehensive lipidomics studies challenged by the high number of lipids, their chemical diversity and their dynamic range. In this work, we developed a tailored method for profiling and quantification combining (1) isotope dilution, (2) enhanced isomer separation by C30 fused-core reversedphase material and (3) parallel Orbitrap and ion trap detection by the Orbitrap Fusion Lumos Tribid mass spectrometer. The combination of parallelizable ion analysis without time loss together with different fragmentation techniques (HCD/CID) and an inclusion list led to higher quality in lipid identifications exemplified in human plasma and yeast samples. Moreover, we used lipidome isotope labeling of yeast (LILY) - a fast and efficient in vivo labeling strategy in Pichia pastoris - in order to produce (non-radioactive) isotopically labeled eukaryotic lipid standards in yeast. We integrated the

13

C lipids in the LC-MS

workflow to enable relative and absolute compound-specific quantification in yeast and human plasma samples by isotope dilution. Label-free and compound-specific quantification was validated by comparison against a recent international interlaboratory study on human plasma SRM 1950. In this way, we were able to prove accuracy of quantification enabled by the LILY approach in complex matrix such as human plasma. Excellent analytical figures of merit with enhanced trueness, precision and linearity over 4-5 orders of magnitude were 1 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 22

observed applying compound-specific quantification with

13

C-labeled lipids. We strongly

believe that lipidomics studies will benefit from incorporating isotope dilution and LC-MSn strategies. Keywords Lipidomics,

13

C labeling, quantification, Orbitrap Fusion Lumos, C30, MSn, LILY, SRM

1950, human plasma Abbreviations: Diglyceride

(DG),

triglyceride

(TG),

ceramide

(Cer),

cholesterol

ester

(CE),

phosphatidylethanolamine (PE), lysophosphatidylethanolamine (LPE), phosphatidylcholine (PC), lysophosphatidylcholine (LPC), phosphatidylinositol (PI), phosphatidylserine (PS), phosphatidic acid (PA), sphingomyelin (SM) acyl carnitine (AcCa), hexosyl ceramide (HexCer), dihexosyl ceramide (Hex2Cer), coenzyme (Co), high resolution mass spectrometry (HRMS), ultra-high pressure chromatography (UPLC), data-dependent mass spectrometry (ddMS), higher energy collisional dissociation (HCD), collision induced dissociation (CID)

Introduction The chemical diversity of lipids as well as increasing proof of their biological relevance 1–5, demands high-throughput analytical methods. The combination of liquid chromatography and mass spectrometry has evolved as state of the art for lipid profiling and quantification due to the possibility to separate lipids by their chemical behaviour prior to mass spectrometric detection and monitoring of fragmentation patterns6–12. In order to investigate the role of lipids involved in biological processes, accurate lipid quantification approaches are essential. However, lipid quantification is challenged by their chemical diversity13, their abundance and high dynamic range14,15. Isomers and isobars, stability issues and lipid-specific effects such as concentration dependent micelle formation can lead to huge quantification problems6,16–18. All these factors make suitable lipid standards indispensable. Ideal standards are stableisotope labeled standards as they provide the same chemical behavior so that simultaneous treatment under identical experimental conditions is enabled19–21. So called isotope dilution approaches involving stable-isotope labeled standards can account for any error accumulated in the analytical chain of sample preparation, storage and measurement. This strategy is successfully applied in relative and absolute quantification of many omics disciplines22–26. 2 ACS Paragon Plus Environment

Page 3 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Although the field of lipidomics is rapidly growing, state of the art lipidomics quantification strategies do not involve isotope dilution due to the complexity and arising costs of labeled lipids produced by individual chemical synthesis6,16. Recently, our group introduced the LILY (Lipidome Isotope Labeling of Yeast) strategy to produce highly enriched (over 99%) 13

C lipids via yeast fermentation27. In this work, we aim to extend the approach to complex

biological matrices such as human plasma. Complex biological matrices demand powerful analytical methods in order to monitor as many lipid species as possible. Reversed-phase chromatography is commonly used for lipid profiling6,8,27–29 and enables the separation of lipids by hydrophobic interaction based on fatty acid chain length, degree of saturation and double bond position7,28,30–32. Lipid chromatographic behaviour can be very similar for lipids from different classes so that class-specific separation is often not possible. This can cause co-elution of lipid classes leading to identification and quantification problems with isobaric lipid species so that increased (chromatographic and mass spectrometric) resolution is demanded. With the advent of the UPLC technology and the downscaling of chromatographic filling material by sub-2µm particles or fused-core particles, increased plate numbers and enhanced chromatographic speed are enabled33–37. The longer carbon chain of C30 compared to the more common C18 chain length on reversed-phase column resins improves the interaction between longer lipid chains and the column material. The improved alignment provides greater shape selectivity between similar alkyl chains. This enhanced separation power allows better characterization of closely related lipid species38,39. In this study, the analytical challenge of high diversity and dynamic range in the lipidome will be met with enhanced chromatographic selectivity by the C30 reversed-phased material combined with high resolution accurate mass (HRAM) mass spectrometry (resolving power up to 500,000 FWHM at m/z 200) and the MSn and parallel higher energy collisional dissociation and collision induced dissociation HCD/CID fragmentation capabilities provided by the Orbitrap Fusion Lumos40. Moreover, the potential of labeled LILY lipids for merging targeted and non-targeted lipidomics workflows is assessed.

Material and Methods Lipid Standards Lipid Standards (Cer d36:1, DG 34:1, LPC 16:0, LPC 18:0, PC 34:0, PC 34:1, PC 34:2, PE 34:1, PE 36:2, PA 34:1, PS 34:1, PS 36:1, TG 52:2) were obtained from Avanti Polar Lipids, Inc. (Alabaster, Alabama, USA). The

13

C labeled lipid extract (99.5-99.8% enrichment 3

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

degree) for internal standardization was produced in-house from labeled yeast cells (ISOtopic Solution, Vienna, Austria).

Lipid extraction The yeast samples were fermented22 and extracted as previously described27. For the comparison of label-free versus compound-specific relative quantification, defined Pichia pastoris samples of known fold changes (0.002, 0.01, 0.05, 0.10, 0.20, 0.50, 1.00 = reference point corresponding to 500 µL of yeast extract, 2.00, 3.00, 5.00) were prepared once with and once without internal standardization by 500 µL of pooled 13C yeast extracts (corresponding to 107 cells and phospholipid concentrations in the low µM range). For the relative quantification of human plasma samples the certified reference material SRM 1950 was ordered at the National Institute of Standards and Technology (NIST). 200 µL aliquots were extracted by MTBE using the Matyash protocol and diluted 1:5 for a pooled human plasma sample41. The pooled human plasma was used to obtain dilutions of defined fold changes (0.002, 0.01, 0.05, 0.10, 0.20, 0.50, 1.00=reference point corresponding to 20 µL of undiluted human plasma, 3.00) and was prepared once with and once without internal standardization by 500 µL of pooled

13

C yeast extracts. The absolute quantification experiments were

performed using human plasma aliquots of the reference point (fold change of 1.00quantification of high abundant lipids) and fold change of 3.00 (quantification of low abundant lipids) with and without 500 µL of pooled 13C yeast extract (n=3)All samples were dried and reconstituted in 200 µL starting eluent conditions. Detailed description of the data evaluation workflow can be found in the Supporting information.

C30 Reversed-phase (RP) chromatography The RP-LC separation was performed with a Vanquish Flex UHPLC (Thermo Fisher Scientific, Germering, Germany) system and a Thermo ScientificTM AccucoreTM C30 column (150 x 2.1 mm) with fused-core particles of 2.6 µm diameter. The temperatures of the autosampler and the column oven were set to 10 and 40 °C, respectively. Solvent A was ACN/H2O (3:2, v/v), and solvent B was IPA/ACN (9:1, v/v). Both solvents contained 0.1% formic acid, and 10 mM ammonium formate. All mobile phase solvents were Optima LC-MS grade (Fisher Scientific, Loughborough, UK). The chromatographic separation was carried out at a flow rate of 300 µL min-1 with 2 µL injection volume using the following gradient: 0−18 min ramp (non-linearly applying slope curve 4) from 30% to 85% B, 85% B, 18−20 min ramp to 90% B, 20−24 min ramp to 90% B, 24.0 min to 30% B. The column was then 4 ACS Paragon Plus Environment

Page 4 of 22

Page 5 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

re-equilibrated for 4 min. The injector needle was washed with 75% IPA, 24.9% H2O, 0.1% formic acid prior to each injection.

Lipid profiling and quantification with RPLC-MS The UPLC-system was coupled to an ultra-high-resolution accurate mass Orbitrap Fusion Lumos Tribrid Mass Spectrometer (Thermo Fisher Scientific, San Jose, CA, USA) for lipid profiling and quantification. Identification runs of human plasma (SRM 1950) and yeast samples in the positive mode exploiting the MSn capabilities of the Orbitrap and parallel fragmentation in the linear ion trap were performed. Additionally, high-resolution MS1 quantification runs were performed comparing label-free and compound-specific relative and absolute quantification for human plasma and yeast samples. Detailed descriptions of the workflow and method parameters can be found in the Supporting Information.

Results and Discussions In this work, a novel RPLC-HRMS strategy for lipid profiling and quantification based on 13

C LILY lipids was developed.

Fast lipid profiling by C30 reversed-phase chromatography Most lipids are amphiphilic and possess strong hydrophobic properties enabling reversedphase separation by fatty acid chain length, degree of saturation and double bond position7,28,30–32. In this study, we exploited the enhanced shape selectivity of C30 reversedphase material39 with the advantage of fused-core particles. Fused-core particles lead to similar peak widths as fully porous silica-based sub-2µm particles and possess the advantage of less column backpressure due to shorter diffusion paths33–37. Chromatographic separation based on fused-core C30 reversed-phase enabled fast chromatographic profiling of complex human plasma samples in less than 30 min (Figure 1). Using C30 reversed-phase material enhanced separation of lipid isomers with different sn-positions or fatty acyl composition but same molecular species level can be achieved, as can be seen for the examples of LPC 18:0, PC 34:4 or TG 50:5 in human plasma (Supporting Table S1). Figure 1: RPLC-MS base peak of a human plasma lipid profile. Lipid profiling of human plasma (SRM 1950) was performed by C30 reversed-phase chromatography with a 28 min isopropanol gradient and high-resolution mass spectrometric detection (Orbitrap Fusion Lumos) in positive mode. Enhanced lipid identification was achieved using parallel Orbitrap 5 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

and ion trap detection. From 0-18 min, (1) data-dependent MS2 (ddMS2), (2) MS2 HCD/CID for enhanced PC and SM annotation and (3) inclusion list for ceramide detection was used. From 18-28 min, (1) ddMS2 in parallel to (2) MS2 CID and MS3 HCD for TG fatty acid annotation was performed. The chromatogram was created with Skyline (Version 3.7).

Enhanced lipid identification by parallel Orbitrap and ion trap detection As detailed in Figure 1, a tailored measurement method was implemented together with the C30 based separation with the aim of increasing lipid identifications. This enhanced lipid profiling was enabled using the novel detection and ion transfer capabilities of the Fusion Lumos Tribrid. The Fusion Lumos has not only high resolving power properties (resolving power up to 500,000 FWHM at m/z 200) but also an additional ion trap for MSn capabilities with both detection in the Orbitrap for HRAM spectra as well as in the ion trap on secondary electron multipliers for low resolution and nominal mass (but very sensitive, fast and parallelizable for high duty cycles)40,42. The additional ion trap enables parallel ion accumulation and multistage fragmentation of a compound of interest. Aiming at high ionization coverage of the lipidome, while reducing the number of analytical runs, all 6 ACS Paragon Plus Environment

Page 6 of 22

Page 7 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

measurements were performed in positive mode, where most lipids can be ionized (e.g. TG and DG cannot be analyzed in negative mode). In the first 18 minutes of the method when predominately glycerophospholipids and sphingolipids elute, we specifically addressed phosphatidylcholine, sphingomyelin and ceramide profiling (Figure 1). Phosphatidylcholine (PC) and sphingomyelin (SM) profiling is challenging using HCD fragmentation due the high abundance of the head group ions at m/z 184. Therefore, each precursor showing the ion at m/z 184 upon HCD fragmentation was subjected to repeated precursor ion isolation, CID fragmentation and ion trap detection. CID provided complementary information as (1) CID is a more gentle fragmentation technique compared to HCD40,43 and (2) the low mass cut-off of the ion trap leads to the loss of the abundant 184 head group ion. In that way predominantly the lysophospholipid backbone ions were produced and additional structural fatty acyl information of phosphatidylcholine and sphingomyelin lipids was obtained as exemplary shown for PC (16:0_18:1) and SM (d18:1_16:0) (Supporting Figure S1). Hence, using the combination of parallel HCD fragment ion generation and fragment ion detection in the Orbitrap with CID precursor fragmentation in the ion trap, additional species level annotations of PCs and SMs were possible. A significant increase in PC and SM identifications (n=6) by a factor of 2-3 (MS2: PCs- 59, SMs-13; MS2 with HCD/CID: PCs- 137, SMs-39) was observed with higher confidence level annotations (Lipid Search 4.1 A-grade IDs of MS2: PCs- 12, SMs- 0; Agrade IDs of MS2 with HCD/CID: PCs: 90, SMs: 20) compared to classical data-dependent MS2 runs (n=6) (Supporting Table S1, Table S2). To increase the number of identifications of the low abundant class of ceramides, an inclusion list (containing suspect ceramides for human plasma and yeast samples deduced from literature) was added and identification by data-dependent HCD MS2 with detection in the Orbitrap was performed. In order to prevent information loss of analyte ions that were not on the inclusion list, data-dependent HCD MS2 scans with detection in the linear ion trap were carried out in parallel. Analysis of the human plasma reference material SRM 1950 showed enhanced lipid identifications of this rather low abundant lipid class. Comparing ceramide identification numbers from the standard ddMS2 method (n=6) in the Orbitrap (ceramides are only fragmented by chance) to the MS2 method using ddMS2 with inclusion list (n=6, all potential ceramides of the list including phosphorylated and glycosylated ceramides are fragmented) and parallel linear ion trap detection led to 11 versus 20 ceramides in human plasma, respectively (Supporting Table S1, Table S2). 7 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 22

At the retention time 18-28 min in parallel to data-dependent HCD MS2 fragmentation with detection in the Orbitrap, multistage fragmentation in the ion trap was implemented with the aim of fatty acyl level annotation of triglycerides (TGs) (Figure 1). Generally, MS2 fragmentation provides insufficient structural information to assign all three fatty acids of a given triglyceride e.g.: if only one neutral loss fragment is observed, this can be due to low abundance of other fragments or the presence of the same three fatty acid chains. In order to gain additional structural information and meet the requirement of fast duty cycles, MS2 CID followed by MS3 HCD in the ion trap was performed. MS2 CID in the ion trap produced for all TGs predominantly fatty acid backbone fragments with only two remaining fatty acids. Hence, CID fragmentation followed by isolation between 33-95% of the precursor mass and MS3 HCD fragmentation enabled to resolve the complete fatty acid composition for TGs. Using the enhanced isomer separation capacity of the C30 column combined with the detection properties of the MSn method, detailed fatty acid information was gained (increasing A type identifications in Lipid Search 4.1 by 30 %) as observed for the example of TG 54:4 in human plasma (Supporting Figure S2). Overall, one third more triglyceride annotations were possible using MS2 CID and MS3 HCD multistage fragmentation (TG annotations: 101) compared to classical ddMS2 (TG annotations: 61) in the investigated plasma samples (Supporting Table S2). The optimized profiling approach was applied to a panel of lipid standards, human plasma samples and non-labeled yeast extracts, which had been prepared in analogy to the LILY workflow without

13

C labeling. All standards could be unambiguously identified based on

fully data-dependent measurements and Lipid Search 4.1 annotation (Supporting Table S2) with retention time precisions of less than 1% for all lipids (measurement period over 72h). Overall, 390 lipids in human plasma (n=6) and 212 lipids in the non-labeled yeast extract (n=6)

were

annotated

from

the

lipid

categories

of

fatty acids,

glycerolipids,

glycerophospholipids, sterols and sphingolipids (Figure 2). Finally, all lipids found in this study were in accordance with previous studies on human serum17,44–47 and yeast samples27,48– 51

. These results highlight the potential of (1) MSn multistage fragmentation methods, (2) the

parallelizable ion trap capabilities of the Fusion Lumos including CID and HCD fragmentation as well as (3) lipid inclusion lists for lipidomics profiling tasks.

Application of 13C yeast lipids to human plasma

8 ACS Paragon Plus Environment

Page 9 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

To evaluate the potential of LILY lipids regarding internal standardization of different biological matrices, we compared lipid profiles of Pichia pastoris yeast and human plasma (SRM 1950) samples. The lipid identifications from human plasma (390 lipids) and yeast (212 lipids) were matched revealing which yeast lipids can be used for compound-specific quantification of human plasma under the applied experimental conditions. Moreover, the identified natural (non 13Clabeled) Pichia pastoris lipids were compared to their

13

C analogs in labeled yeast extract.

All lipids present in the non-labeled yeast lipid extract were also identified in 13C yeast lipid extracts with isotopic enrichments of 99.5-99.7% as previously shown27. Summarizing, 114 LILY lipids from the classes of PC, LPC, PE, PI, DG and TG werefound in both sample types, hence are available for compound-specific quantification. It has to be noted, that we only reported annotated lipid species with available fragmentation spectra obtained by datadependent acquisition of the profiling runs. Hence in addition to these 114 LILY lipids, low abundant LILY lipids are present and suitable for compound-specific quantification as well. E.g. PC 34:0 and LPC 18:0 were low abundant in the 13C-labeled yeast extract, but could be identified by authentic standards, retention time and MS2 fragmentation (with corresponding pattern from natural and 13C enriched yeast extracts) (Supporting Figure S3 A-D). Furthermore, all 212 LILY lipids (from the classes: Cer, TG, HexCer, PC, DG, PE, PA, LPC, PG, PI, Hex2Cer, PS, Co, LPE) identified are amenable for class-specific or retention-timebased quantification in human plasma (Figure 2). Figure 2: Lipid profiling by RPLC-MSn in Pichia pastoris yeast and SRM 1950 human plasma samples. A. Number of lipid species per lipid class and presence in LILY yeast extract and human plasma samples. B. Comparison of individual lipid species present in both human plasma and 13C-labeled LILY lipid extract.

9 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

These results reveal that the eukaryotic properties of yeast enable the production of interesting labeled lipid standards that can be applied for isotope dilution approaches in more complex biological systems such as human plasma. Moreover, stable isotope yeast lipids represent an interesting alternative for the state of the art lipidomics quantification procedure using non-endogenous class-specific lipid standards for quantification14,16,18,48. Labeled yeast lipids can be used for compound-specific quantification as well as for endogenous and nonendogenous class-specific quantification in different biological matrices. Biotechnological engineering of yeasts represents a promising approach to modify yeast lipid synthesis in order to produce compound-specific lipid standards from different classes52–58. Impressive examples are the engineering of the sphingosine lipid pathway of Saccharomyces cerevisiae in order to produce human ceramides57 or the genetic modification in S. cerevisiae to produce cholesterol instead of ergosterol58. Moreover, Pichia pastoris is optimally suited for genetic engineering even easier to manipulate than the standard yeast S. cerevisiae due to enhanced vector integration59.

Label-free versus compound-specific quantification Finally, we addressed quantification relying on 13C LILY lipids. More specifically, label-free versus compound-specific quantification was compared for relative and absolute 10 ACS Paragon Plus Environment

Page 10 of 22

Page 11 of 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

quantification of lipids in both yeast and human plasma samples. The quantification exercise relied on C30 reversed-phase chromatography and Full-MS quantification (resolution setting 120,000 for MS1). In a first step, we investigated the potential of LILY for relative quantification using the 13C yeast based lipid extract as compound-specific scalar to address fold changes in human plasma and yeast. In order to assess analytical figures of merit such as trueness bias and repeatability, dilutions series of human plasma and yeast lipid extracts were used to prepare samples of known fold changes with and without internal standard. Compound-specific FullMS relative quantification based on LILY lipids enabled enhanced analytical figures of merit for both yeast27 and human plasma samples. As exemplary shown in Figure S4 and S5 for 40 medium to high-abundant human plasma lipids from 6 classes (DG, LPE, PC, PE, PI, TG), trueness (given as fold change recovery: label-free relative quantification: 57-110%; compound-specific relative quantification: 86-115%) and precision (label-free relative quantification: 9-26% RSD; compound-specific relative quantification: 5-15%, considering all investigated concentrations levels of 40 lipids) over 4-5 orders of magnitude were significantly improved when

13

C LILY lipids were applied for compound-specific

quantification. Similar HRAM mass spectrometric studies on metabolites show a comparable linear dynamic range (depending on the matrix and sample complexity)60,61. Improved trueness, precision and linearity were also previously observed for the relative quantification of lipids in yeast samples spiked with

13

C LILY lipids using both PRM and Full-MS

quantification approaches27. In the second step, we investigated the potential of LILY for absolute quantification by comparing label-free versus compound-specific absolute quantification in human plasma and yeast. We quantified a panel of target lipids (Cer d36:1, DG 34:1, LPC 16:0, LPC 18:0, PC 34:0, PC 34:1, PC 34:2, PE 34:1, PE 36:2, PA 34:1, PS 34:1, PS 36:1, TG 52:2), by C30 reversed-phase chromatography and Full-MS quantification using external calibration (by endogenous standards) with and without internal standardization by LILY derived 13C lipids. The quantification was performed in analogy to established targeted metabolomics strategies23,24,62. Lipid concentrations determined in the yeast samples were in the nmol mL-1 range for 107 cells (Supporting Table S3) as already observed in previous studies on Pichia pastoris or other yeasts27,48–51. Lipid concentrations obtained in the human plasma sample were also in the nmol mL-1 range (Table 1, Supporting Table S3) as determined in different studies on SRM 195047,63. Overall, precision (label-free absolute quantification: 5-25% RSD; 11 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 22

compound-specific absolute quantification: 1-5%) and linearity (label-free absolute quantification: 0.8985-0.9986 R2; compound-specific absolute quantification: 0.9904-0.9999 R2) were improved applying compound-specific absolute quantification in both yeast and human plasma samples. Quantification limits (LOQ) for the analyzed lipids were in the low nM range corresponding to fmol absolute on column (calculated from 10 σ of the lowest detected 50 nM calibration standards64) and were significantly improved in compound-specific quantification compared to label-free quantification (Table 1). Table 1: Summary of the measurement accuracy of label-free versus compound-specific quantification compared against human plasma SRM 195065 Lipid Species DG 34:1 TG 52:2 LPC 16:0 LPC 18:0 PC 34:0 PC 34:1 PC P-35:1/34:2 PE 34:1 PE 36:2 CER d36:1

Label-free* 0.59 ± 0.51 1.1 ± < 0.1 48 ± 2 12 ± 2 1.2 ± 0.5 179 ± 1 393 ± 16 0.51 ± 0.11 0.30 ± 0.02