Monitoring Membrane Lipidome Turnover by Metabolic 15N Labeling

Nov 7, 2017 - Lipidomes undergo permanent extensive remodeling, but how the turnover rate differs between lipid classes and molecular species is poorl...
0 downloads 14 Views 1MB Size
Subscriber access provided by READING UNIV

Article 15

Monitoring Membrane Lipidome Turnover by Metabolic NLabeling and Shotgun Ultra-High Resolution Orbitrap FT MS Kai Schuhmann, Kristina Srzenti#, Konstantin O. Nagornov, Henrik Thomas, Theresia Gutmann, Ünal Coskun, Yury O. Tsybin, and Andrej Shevchenko Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.7b03437 • Publication Date (Web): 07 Nov 2017 Downloaded from http://pubs.acs.org on November 9, 2017

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 free 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 accessible to all readers and 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.

Analytical Chemistry 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 28

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

Monitoring Membrane Lipidome Turnover by Metabolic 15N-Labeling and Shotgun Ultra-High Resolution Orbitrap FT MS

Kai Schuhmann1, Kristina Srzentić2, Konstantin O. Nagornov3, Henrik Thomas1, Theresia Gutmann4,5, Ünal Coskun4,5, Yury O. Tsybin3 and Andrej Shevchenko1,6

1

MPI of Molecular Cell Biology and Genetics, Pfotenhauerstr. 108, 01307 Dresden, Germany

2

Ecole Polytechnique Fédérale de Lausanne, 1015 Lausanne, Switzerland

3

Spectroswiss, EPFL Innovation Park, Building I, 1015 Lausanne, Switzerland

4

Paul Langerhans Institute Dresden of the Helmholtz Zentrum München at the University Hospital

and Faculty of Medicine Carl Gustav Carus of TU Dresden, TU Dresden, Fetscher Str. 74, 01307 Dresden, Germany 5

German Center for Diabetes Research (DZD e.V.), Ingolstädter Landstr. 1, 85764 Neuherberg,

Germany 6

corresponding author: [email protected]

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 28

2 Lipidomes undergo permanent extensive re-modelling, but how the turnover rate differs between lipid classes and molecular species is poorly understood. We employed metabolic 15N-labeling and shotgun ultra-high resolution mass spectrometry (sUHR) to quantify the absolute (molar) abundance and determine the turnover rate of glycerophospho- and sphingolipids by the direct analysis of total lipid extracts. sUHR performed on a commercial Orbitrap Elite instrument at the mass resolution of 1.35×106 (m/z 200) baseline resolved peaks of 13C isotopes of unlabeled and monoisotopic peaks of 15

N–labeled lipids (∆m = 0.0063 Da). Therefore, the rate of metabolic 15N-labeling of individual lipid

species could be determined without compromising the scope, accuracy and dynamic range of fulllipidome quantitative shotgun profiling. As a proof of concept we employed sUHR to determine the lipidome composition and fluxes of 62 nitrogen-containing membrane lipids in human hepatoma HepG2 cells.

Key words: shotgun lipidomics; ultra-high resolution mass spectrometry; metabolic lipid flux.

ACS Paragon Plus Environment

15

N-labeling;

Page 3 of 28

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

3 INTRODUCTION Cell membranes are asymmetric dynamic constellations of lipids and proteins subjected to permanent turnover and compositional changes in response to genetic, development and environmental challenges (reviewed in1,2). Lipid turnover is an important molecular mechanism that controls membranes compositional identity and regulates their critical physicochemical properties, such as lipid asymmetry, thickness and fluidity (reviewed in3). A broad palette of methods of highthroughput lipidome quantification by shotgun or LC-MS/MS analyses is available (reviewed in4-6). However, monitoring lipids turnover at the full-lipidome scale remains a highly challenging task that requires elaborate analytical procedures and software7. Mass spectrometry (MS) is commonly applied to determine the rate of metabolic incorporation of isotopic labels carried by precursors of acetyl CoA, most commonly 13C-glucose or 13

C-acetate, into newly synthesized fatty acid (FA) moieties (reviewed in8). This approach, however,

excludes lipids with essential fatty acids that are not synthesized de novo. Since lipids are carbon-rich molecules their metabolic

13

C-labeling is seldom complete and typically yields highly convoluted

overlapping isotopic profiles of partially labeled and unlabeled lipids whose interpretation requires high fidelity of measurements of abundances of isotopic peaks. Alternatively, polar head groups of glycerophospholipids could be labeled in a lipid class-specific way using poly-deuterated compounds, e.g. (trimethyl-d9)choline or 2-amino(ethanol-1,1,2,2-d4) (reviewed in9,10). The advantage of targeted labeling is that, independently of FA moieties, each species of the targeted lipid class will incorporate exactly one labeled head group and therefore labeled and unlabeled species could be readily distinguished by tandem mass spectrometry (MS/MS). One caveat is that poly-deuterated head groups are much bulkier compared to native molecules11 and it is thought that pronounced sterical hindrance might affect their interaction with membrane proteins and / or

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 4 of 28

4 clustering into structurally ordered lipid-protein microdomains (reviewed in12). Another downside of targeted labeling rests with its specificity: the individual source compound and quantification method are required for each lipid class and therefore it is difficult to design a comprehensive labeling experiment aiming at the full lipidome coverage. Therefore, it is highly desirable to identify a generic isotopic marker that enables direct, parallel and quantitative monitoring of metabolic labeling of many lipid classes associated within the same biological context. Dancy et al proposed that membrane lipids turnover could be monitored by metabolic

15

labeling13. This could be an attractive and cost-efficient alternative to using poly-deuterated or

13

NC-

isotopic markers since a broad palette of source compounds is already available for both targeted (15N-serine, 15N-choline or 15N-ethanolamine) and full-organism (e.g. 15N-ammonium salts) labeling, including complex foods suitable for in vivo experiments in flies14 and rats15. Major classes of membrane glycerophospholipids, such as PC and PE (along with corresponding ether-lipids PC Oand PE O-), PS and sphingolipids (e.g. Cer) contain one, while SM and some glycosphingolipids contain two atoms of nitrogen. Therefore, upon metabolic labeling, they are expected to yield relatively simple isotopic profiles. However, 15N-incorporation only shifts lipid masses by one or by two Daltons. Nominal masses of monoisotopic peaks of labeled lipids and of first isotopic peaks of corresponding unlabeled lipids (for presentation clarity we further term them as 15N- and 13C-peaks, respectively) are the same and could not be resolved by conventional mass spectrometry means. Since 13C-peaks are abundant in lipids, turnover measurements required impractically high degree of metabolic incorporation of 15N-isotopes. Here we employed ultra-high resolution (UHR) Fourier transform mass spectrometry (FT MS) on a commercial Orbitrap Elite instrument to achieve baseline separation of

ACS Paragon Plus Environment

15

N- /

13

C- peaks

Page 5 of 28

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

5 and enable the identification, absolute quantification and determination of the turnover rate of membrane lipids in a single shotgun analysis of total lipid extract.

EXPERIMENTAL SECTION Annotation of lipid species Lipid classes were annotated as follows16: FA, fatty acid; PC, phosphatidylcholines; PC O-, 1-Oalkyl–2-acyl-glycerophosphatidylcholines;

LPC,

lysophosphatidylcholines;

PE,

phosphatidylethanolamines; PS, phosphatidylserines; Cer, ceramides; SM, sphingomyelins. Species of glycerophospholipids were annotated by their lipid class and the total number of carbon atoms and double bonds in their fatty acid or fatty alcohol moieties. The molecular species were annotated by their fatty acid moieties separated by minus or slash symbols if their sn-1/sn-2 positioning was, respectively, ambiguous or certain. For example, PC 32:1 stands for PC molecules comprising 32 carbon atoms and one double bond in both FA moieties; PC 16:0-16:1 indicates that this PC comprises 16:0 and 16:1 FA moieties located at undefined positions of the glycerol backbone; PC 16:0/16:1 indicates that 16:0 and 16:1 FA moieties are located at the sn-1 and sn-2 positon, respectively. Sphingolipid species were annotated by the total number of carbon atoms, double bonds and hydroxyl groups in N-amidated fatty acid and long-chain base moieties.

Chemicals and lipid standards Synthetic lipid standards were all purchased from Avanti Polar Lipids (Alabaster, AL). Methanol, isopropanol, water and ammonium formate were purchased from Sigma-Aldrich or Merck (Darmstadt, Germany) and were of LC-MS or Chromasolv/LiChrosolv grade. Chloroform (HPLC

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 6 of 28

6 grade) was purchased from Rathburn Chemicals (Walkerburn, UK).

15

N-L-serine and

15

N-choline

chloride were purchased from Sigma (cat.no. 609005 and 609269, respectively).

Metabolic 15N-labeling of HepG2 cells Human hepatoma HepG2 cells were maintained in MEM medium (Gibco, 21090-022) supplemented with 2mM glutamine (Gibco, 25030-024), 1x non-essential amino acids (Gibco, 11140-035), and 15% FCS (Gibco, 10270098). For metabolic 15N-labeling cells were plated at 4.5×105 cells/well in a 12-well plate and allowed to adhere. They were washed and serum-starved overnight in FCS-free medium. Next day the medium was changed to serine- and choline-free DMEM (PAN-Biotech, P0401550S1) supplemented with 28.56 µM

14

N-choline chloride (Sigma, C7017) and 400µM

14

N-L-

serine (Sigma, S4500) (same concentrations as in DMEM medium) and cells were allowed to adapt for 1h. Next, choline chloride and/or L-serine were replaced by corresponding

15

N-containing

compounds with the same concentration. Cells were incubated for 3, 6, 9, 12, 18, 24, and 30h. For harvesting, cells were washed 3-times with ice-cold PBS, scratched from the plate and transferred to a 1.5ml Eppendorf tube, centrifuged (5min, 1200g, 4℃) and re-suspended in 100µl of ice-cold ammonium bicarbonate buffer. 200µl of ice-cold isopropanol was immediately added, vortexed and snap-frozen in liquid nitrogen. Samples were stored at -80℃ until analyzed.

Lipid extraction from HepG2 cells HepG2 cells were mixed with 300 µl isopropanol, homogenized using 1mm zirconium beads in 3 cycles of 1.5 min each at the frequency of 30 Hz. Prior to lipid extraction an aliquot of the homogenate equivalent to 6×105 cells was transferred to a new 2 ml Eppendorf tube and dried under vacuum. For lipid extraction 280 µl H2O and 1400 µl of methyl-tert-butyl ether (MTBE) / methanol

ACS Paragon Plus Environment

Page 7 of 28

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

7 (MeOH) (10:3 v/v) mixture that also included internal standards (Table S1) were added to the tube17. The mixture was shaken at 4℃ for 1.5 h and then centrifuged at 13400 rpm at 4℃. The upper (organic)

phase

was

collected,

dried

under

vacuum

and

re-dissolved

in

isopropanol/methanol/chloroform (4:2:1; v/v/v) mixture containing 7.5 mM ammonium formate or ethanol/chloroform (5:1; v/v) with 0.1 % trimethylamine before mass spectrometric analyses.

Shotgun lipidomics by Orbitrap and ICR FT mass spectrometry Quantitative shotgun lipidomic profiling was performed on Q Exactive, Orbitrap Elite and LTQ FT mass spectrometers (all from Thermo Fisher Scientific, Bremen, Germany) equipped with a Triversa Nanomate nanoflow robotic ion source (Advion BioSciences, Ithaca NY) as described18. On a Q Exactive mass spectrometer lipids were quantified by data-independent acquisition (DIA) under automated gain control (AGC) values of 3×106 and 2×104 and maximum fill time of 500 ms and 650 ms in FT MS and FT MS/MS experiments, respectively. Both were acquired at the target mass resolution of Rm/z 200 = 1.4 ×105. Broad-band FT MS spectra were acquired for 1 min within m/z range of 400 to 1200 in each polarity mode. In FT MS/MS experiments using higher energy collision-induced dissociation (HCD) the normalized collision energy (NCE) was set to 25% in positive and 35% in negative ion mode19 with precursor ions isolation width of 1 Th. S-lense RF level was 50% and transfer capillary temperature 200℃. UHR analyses on an Orbitrap Elite were performed in a targeted selected ion monitoring (tSIM) mode. The mass resolution Rm/z 200 = 1.35×106 was reached by acquiring 3 s transients and processing them in enhanced FT (eFT) mode under AGC values of 3×103, 1.5×104 and 5×104 and maximum fill time of 500 ms. Extended (3 s vs default 1.5 s)20 transients acquisition time was enabled by developer’s kit functionality using Instrument Control Language (ICL)21,22. If required, it

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 28

8 could be further extended up to 6 s reaching Rm/z 650 = 1.2×106 (Figure S1). S-lense RF level and capillary temperature were as above. In t-SIM experiments the ions isolation width was 20 Th; spectra were successively acquired with 10 Th steps within m/z ranges of 490 to 890 and of 410 to 820 for samples sprayed from ammonium formate- and trimethylamine- containing buffers, respectively; in total, these m/z ranges were covered by 37 (or, respectively, 42) t-SIM mass windows. By default, spectra were acquired in reduced profile mode. Each t-SIM spectrum with elapsed scan time of 6 to 7 s was acquired at least 5 times. Analyses by the method of ion cyclotron resonance (ICR) FT MS were performed on a 10T LTQ FT mass spectrometer23. The instrument was equipped with narrow aperture detection electrodes (NADEL) ICR cell having four detection electrodes operating in the quadrupolar ion detection mode at the “true” cyclotron frequency24. The full profile mass spectra were acquired with 6 s detection period and processed in a magnitude FT (mFT) mode with the mass resolution of Rm/z 750

= 7.3×105. It was close to the resolution achieved during sUHR analyses on an Orbitrap Elite and

therefore spectra acquired on both machines could be directly compared (Figure S1). Spectra were acquired in t-SIM mode using AGC value of 3×105 and maximum fill time of 1000 ms. Each t-SIM spectrum with an elapsed scan time of 6.3 s was measured 6 times.

Data processing All raw files were pre-filtered by PeakStrainer25 and lipids identified by LipidXplorer software26-28. Isotopic peaks of identified lipid species were recognized by dedicated molecular fragmentation query language (MFQL) queries26,27.

ACS Paragon Plus Environment

Page 9 of 28

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

9 RESULTS AND DISCUSSION

Baseline separation of 15N- / 13C-peaks by ultra-high resolution Orbitrap FT MS. We proposed to analyze total lipid extracts of metabolically 15N-labeled cells by quantitative shotgun profiling to identify lipid species, quantify their absolute (molar) concentrations and, in parallel, to determine the relative abundances of matching pairs of further reasoned that the relative quantification of

14

15

N-labeled and unlabeled

N-/

15

14

N-lipids. We

N-lipids should rely on the direct

comparison of abundances of their intact molecular ions in FT MS spectra because both HCD FT MS/MS and collision induced dissociation (CID) MS/MS of some major nitrogen-containing lipids (e.g. PE, PE O- and PS, to mention only a few) produce no abundant head group fragments containing atom(s) of nitrogen. Since the exact masses of

13

C- and

15

N-peaks differ by 6.3 mDa (Figure 1) their base-line

separation would require the mass resolution exceeding R = 0.5×106 within a “typical” lipid m/z range of 400 to 1000 that we achieved by ultra-high resolution Orbitrap FT MS.

Figure 1. Resolving 15N- and 13C-peaks by UHR Orbitrap FTMS. (A) Low AGC settings diminished ion coalescence and enabled (B) to resolve pairs of 15N- and 13C-peaks of labeled and unlabeled PE

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 10 of 28

10 36:2 ([M-H]-; m/z 742.5392), respectively. Peaks detected at different AGC (shown in the inset) are annotated with m/z and mass deviation (in ppm). Panel B shows peak profiles acquired in five successive acquisitions (color coded) at Rm/z 750 = 0.68×106 and AGC of 5×103; 13C- and 15N-peaks of unlabeled and labeled PE 36:2 are annotated with m/z and relative abundances (in % to 12C-peak) averaged over all acquisitions.

For resolving

15

N- /

13

C-peaks it was critical to optimize the number of ions filling the Orbitrap

analyzer: for damping ions coalescence it should be maintained at sufficiently low levels29, although larger number of trapped ions provides better ion statistics required for the accurate measurement of abundances of isotopic peaks30. In shotgun FT MS spectra of total lipid extracts the abundance of peaks of bona fide lipids differs by more than 1000-fold31,32. Therefore, direct acquisition of broad-band FT MS spectra under unusually low AGC required for separating

15

N- /

13

C-peaks could impact both the sensitivity and

dynamic range by failing to detect low abundant species. Instead, we chose to acquire a large number of t-SIM spectra that, taken together, covered the entire range of lipid masses. Next, we “stitched” them together into a full “master” FT MS spectrum33,34 by post-acquisition processing using in-house developed software. In t-SIM experiments we isolated ions within a broader m/z range and, for further processing, only used ions from a narrower m/z range centered in the middle of the isolation window. In this way, ions isolation by a linear ion trap (LTQ) was less perturbed by the proximity of their m/z to the isolation window edges. To determine the optimal size of both “broad” and “narrow” m/z ranges we used an equimolar mixture of PE 36:4, PE 36:2 and PE 36:0 that, together with their isotopic peaks, covered m/z range of 11 Th. We used its broad-band FT MS spectrum as a reference and monitored the profile of ions transmitted in negative ion mode t-SIM experiments targeting m/z 743.0 that is located at the center of the m/z range of these PE standards. We found that on an Orbitrap Elite even isolation and transmission of ions within the “narrow” range of 10 Th required at least 20 Th

ACS Paragon Plus Environment

Page 11 of 28

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

11 isolation window (Figure S2). At the same time, using isolation windows larger than 20 Th together with low AGC reduced the detection sensitivity and dynamic range. Next, we estimated the maximum number of ions that could be filled into an Orbitrap analyzer such that 15N - and 13C-peaks remained base-line resolved. To this end we used a total lipid extract of fully 15N-labeled E. coli that contained several PE molecules (including abundant PE 36:2), spiked it with the synthetic standard of PE 36:2 and analyzed their mixture by t-SIM. The isotopic cluster consisting of the labeled (from E.coli) and unlabeled (spiked standard) PE 36:2 was isolated within 20 Th window and spectra were acquired on an Orbitrap Elite by collecting 3 s transients that enabled the target mass resolution of Rm/z 200 = 1.35×106 and Rm/z 750 = 0.68×106. By varying AGC values within the range of 2×103 to 12×103 we determined the highest number of charges still compatible with baseline separation of 15N- / 13C- peaks and delivering the correct abundance of 13Cpeak relative to the monoisotopic (12C) peak of the unlabeled PE 36:2. We observed that

15

N-/

13

C-

peaks were resolved under AGC settings below 8×103, otherwise they collapsed into a single peak (Figure 1). Spectra acquired under AGC of 6×103 faithfully represented the isotopic profile of both labeled and unlabeled lipids: the mass difference between 15N- and 13C-peaks was 4.3 mDa (calc. 6.3 mDa) and the ratio of intensities of 13C- / 12C-peaks of unlabeled PE 36:2 was 42.7% (calc. 44.7%; 4.3% rel.difference). The slightly reduced mass difference between 15N- / 13C- peaks suggested that ion-ion interactions and peaks interference still occurred under these low AGC values, but ions coalescence did not 29. Altogether, for shotgun UHR (sUHR) profiling on an Orbitrap Elite we adopted the following settings: t-SIM spectra were acquired using 20 Th transmission windows, from which only the “middle” range of 10 Th was further used for post-acquisition reconstruction of the full FT MS spectrum. Each t-SIM spectrum was acquired at three AGC settings of 5×103, 1.5×104 and 5×104 and

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 28

12 the same transient accumulation time period of 3 s delivered the mass resolution of Rm/z

200

=1.35×106.

Shotgun lipidomics by sUHR Orbitrap FT MS. The workflow of sUHR FT MS analyses of total lipid extracts is exemplified in Figure 2.

Figure 2 The workflow of sUHR. (A) t-SIM spectra were acquired with 10 Th steps using the isolation window of 20 Th at three different AGC values (for clarity, panels A and B show spectra acquired at the same AGC). (B) Noise was eliminated by PeakStrainer software; m/z ranges of 5 Th were removed from both edges and spectra acquired at the same AGC were stitched together into a full FT MS spectrum. (C) Lipids identified by LipidXplorer were selected from these FT MS spectra by considering the abundance ratios of 12C- and 13C-peaks and the separation of 13C- / 15N-peaks. Here the three panels show the same m/z range in FT MS spectra acquired at different AGC (values are shows above the spectra) and comprising peaks of two labeled lipids (A and B); IS stands for the internal standard. Lipids A, B and IS belong to the same lipid class.

ACS Paragon Plus Environment

Page 13 of 28

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

13

To diminish the impact of isolation window edges we acquired t-SIM spectra such that adjacent 20 Th – wide windows overlapped by 10 Th. During post-acquisition stitching the software disregarded peaks that were offset by less than 5 Th from each edge (Figure 2A, B) and only the “middle” 10 Th ranges were subsequently stitched. Figure 2A schematically shows three successively acquired t-SIM spectra targeting m/z 730, 740 and 750 and covering m/z ranges of 720 to 740; 730 to 750 and 740 to 760, respectively. Next, the range of 5 Th was cut from both edges of each spectrum and the three 10-Th wide ranges m/z 725 to 735, 735 to 745 and 745 to 755 were stitched together (Figure 2B). Prior to stitching, noise peaks (symbolically shown in red in Figure 2A) were eliminated by PeakStrainer software25 that drastically reduced spectra processing time and RAM load. Altogether, the full range of expected lipid masses (m/z 490 to 890) was covered by 37 t-SIM spectra each of which was acquired 5 times at three AGC values. In this way, a typical sUHR acquisition required ca 35 min, while a single broad-band FT MS with ca 10-fold lower mass resolution could be acquired in ca 60 s. Extended acquisition times required stable nanoflow electrospray (Figure S3) that was delivered by a chip-based robotic ion source. Acquisition of seemingly redundant spectra is critical for sUHR workflow. In contrast to conventional broad-band FT MS, for each t-SIM spectrum AGC settings should be maintained within an optimal range that depends on the abundance of peaks falling into the same mass isolation window. It is not possible to predict which AGC value allows to resolve 13C- and 15N-peaks that are only spaced by 6.3 mDa and, at the same time, produces sufficiently abundant signals supporting their robust quantification. Therefore, from each target m/z we acquired three t-SIM spectra at different AGC (Figure 2A). t-SIM spectra acquired at the same AGC were pre-processed and stitched together (Figure 2B) such that the three full FT MS spectra, each related to the same AGC value,

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 14 of 28

14 were produced. Next, we used LipidXplorer to identify lipids and recognize their isotopic peaks independently in each of these three FT MS spectra. For each identified lipid we further checked if the ratio of its

12

same time, its

13

C- and C- and

13

C-peaks matched the expected value with +/- 20% accuracy and, at the

15

N- peaks were resolved. If so, these peaks were further taken for the

quantification. If these criteria were met in two or more spectra, we only took peaks detected at the higher AGC. Peaks selection is further exemplified at the scheme in Figure 2C: peaks of an abundant lipid “A” were taken from the spectrum in the middle because of ions coalescence that occurred in the spectrum at the left and lower AGC of the spectrum at the right. At the same time, peaks of a low abundant lipid “B” were taken from the spectrum at the left because in the spectrum at the right 12C- / 13

C-peaks ratio mismatched and the spectrum in the middle was acquired at the lower AGC. Once a

lipid was picked from one out of three FT MS spectra it was quantified using the internal standard of the same lipid class detected at the same AGC. We next asked if FT MS spectra acquired by sUHR from a total lipid extract accurately reflected its composition? We adopted a FT ICR MS as a reference method because of its higher ion coalescence threshold and further validated sUHR protocol in three ways (Figure 3). First, we checked if sUHR provides a linear response within nM to low µM range of lipid concentrations that are typical for shotgun experiments. Second, we checked if sUHR faithfully reflects the relative abundances of pairs of 15N- and 14N-lipids. And third, we tested if under the similar mass resolution sUHR and FT ICR produced quantitatively concordant lipid profiles.

ACS Paragon Plus Environment

Page 15 of 28

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

15

Figure 3 Validation of the sUHR protocol. (A) Linear dynamic range of quantification by sUHR. The plot was acquired by analyzing a dilution series of PE 36:2 standard ([M-H]-; m/z 742.5392) under three AGC settings (target values are shown in the inset). (B) Correlation of relative abundances of peaks of unlabeled and 15N-labeled PE 36:2 quantified by sUHR on an Orbitrap Elite and by FT ICR MS on a LTQ FT. (C) Correlation of relative abundances (in mol%) of PC and LPC species in a total extract of unlabeled HepG2 cells analyzed by broad-band FT MS (x-axes) and sUHR (y-axes). The plot presents data independently acquired from 3 total extracts (data points are color coded). (D) Correlation of relative abundances (in mol%) of unlabeled and 15N-labeled PC, LPC, and SM species in a HepG2 extract quantified by sUHR on an Orbitrap Elite and by FT ICR, both using stitched t-SIM acquisitions. Mass resolution Rm/z 200 was 1.35×106 and 2.45×106 at the Orbitrap Elite and LTQ FT, respectively.

To check the linear dynamic range of sUHR quantification we analyzed a dilution series of three lipid standards PE 36:4, PE 36:2 and PE 36:0 within the concentration range of 1250 to 0.125 nM

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 16 of 28

16 while PE 25:0 at the concentration of 0.5 µM was used as an internal standard (for presentation clarity Figure 3A only shows the calibration plot for PE 36:2). The three quantified PEs and PE 25:0 were detected in two 20 Th windows centered at m/z of 592 and 743, respectively. From these two tSIM spectra edge mass ranges were trimmed and lipids were quantified by LipidXplorer. We found that calibration plots of all three standards were linear within better than 104-fold concentration range and showed no marked dependence on AGC values (Figure 3A). Next, we checked if sUHR accurately determined full lipid profiles and, under the same settings, could quantify the relative abundance of matching pairs of unlabeled (14N-) and 15N–labeled lipids. To this end, we spiked different amounts of unlabeled PE 36:2 into a total lipid extract of fully 15

N-labeled E.coli, analyzed the mixtures by sUHR and, in parallel, by FT ICR MS at Rm/z

2.45×106 such that

13

C- /

15

200

=

N- peaks were completely resolved in both analyses and correlated the

relative abundances of monoisotopic peaks of the labeled and unlabeled forms of the lipid (Figure 3B). We found that sUHR and FT ICR MS determinations were highly concordant (R2=0.998) and were not affected by ions coalescence if relative abundances of 15N-/ 13C- peaks differed by less than 20-fold. Lipidome-wide quantification capabilities of sUHR were tested in two ways. We acquired tSIM spectra of total lipid extracts from HepG2 cells first by sUHR on an Orbitrap Elite mass spectrometer (20 Th isolation window; Rm/z 200 = 1.35×106), stitched them into a full spectrum as explained above (Figure 2) and relative abundances (in mol%) of PC and LPC species were quantified by LipidXplorer. The same extract was analyzed by broad-band FT MS on a Q Exactive mass spectrometer under its maximal mass resolution (Rm/z 200 = 1.4×105). PC and LPC species were quantified in the same way and relative abundances of 26 species independently determined in broadband FT MS and sUHR experiments were compared (Figure 3C). Excellent concordance of broad-

ACS Paragon Plus Environment

Page 17 of 28

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

17 band FT MS (Q Exactive; Rm/z 200 = 1.4×105) and sUHR (Orbitrap Elite; stitched acquisition at Rm/z 200

= 1.35×106) quantification of species whose relative abundance differed by ca 100-fold suggested

that sUHR faithfully represented a total lipid profile. To check if lipid profiles acquired by sUHR on an Orbitrap Elite also corroborated the stitched t-SIM acquisition by FT ICR on a LTQ FT, we analyzed them under the similar target mass resolution of Rm/z 200 = 1.35×106 and Rm/z 200 = 2.45×106, respectively. We interpreted sUHR and FT ICR mass spectra by LipidXplorer and found that the relative abundances of, in total, 56 peaks of unlabeled and

15

N-labeled lipids showed good

(R2=0.975) correlation (Figure 3D). Hence, also under the similar mass resolution, sUHR corroborated FT ICR quantification within better than 1000-fold dynamic range. Therefore, we concluded that the method of sUHR supported reliable quantitative profiling of total lipid extracts and, in the same experiment, accurately determined the relative abundances in pairs of unlabeled and 15N-labeled lipid species.

sUHR monitors the kinetics of 15N-incorporation into membrane lipids. We applied sUHR to monitor the rate of

15

N-incorporation into membrane lipids of a human

hepatocellular carcinoma HepG2 cell line. We were interested if the membrane lipidome is subjected to extensive turnover and how the turnover kinetics differs between lipid classes and molecular species having the unique composition of fatty acid moieties. We cultured HepG2 cells in minimal DMEM media supplemented with

15

N-serine,

15

N-

choline or with their equimolar mixture. At the specified time points lipids were extracted and quantified lipid by sUHR (Figure 4). In total, the analysis encompassed 305 species from 13 lipid classes that were quantified over 8 time points. From those, 137 species belong to 9 classes of membrane lipids; head groups of 6 out

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 18 of 28

18 of 9 classes contained nitrogen atoms and the turnover of 62 species was monitored by comparing the relative abundance of pairs of their unlabled and

15

N-labeled forms. The total lipid composition of

HepG2 cells remained practically unchanged during the entire incubation period of 30 hours (data not shown). However, sUHR revealed that membrane lipids were subjected to rapid turnover in lipid class- and

15

N-substrate–dependent manner (Figure 4). To compute the labeling rate for each lipid

class the total abundance of 15N-labeled lipids was normalized to the total abundance of lipids of the same class. Kinetic plots (Figure 4A,B) encompassed 31 and 3 species of PC and LPC classes, respectively. If cells were labeled with 15N-choline hours maxing at ca 45 mol% of

15

15

N-incorporation reached a plateau after ca 20

N-labeled PCs. Expectantly, neither maximal

15

N-incorporation

rate nor its kinetics was affected by 15N-serine – either alone or in the mixture with 15N-choline.

ACS Paragon Plus Environment

Page 19 of 28

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

19

Figure 4: Kinetics of 15N-labeling of different classes of membrane lipids: (A) PC, (B) LPC, (C) PS, (D) PE and (E) SM. (F) Isotopic profile of SM 34:1:1 ([M+HCO2]-) after 30h of 15N-labeling as detected by sUHR. The expected profile of the unlabeled lipid computed at the assumed resolution of Rm/z750 = 6.5 ×105 is in red.

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 20 of 28

20 Similarly, ca 45% labeling of PS was achieved by adding 15N-serine and was unaffected by 15

N-choline. Interestingly, its kinetic plot did not plateau: the labeling rate reached its maximum at ca

13 h and then declined, presumably because the pool of 15N-serine was depleted by on-going proteins and lipids biosynthesis and also diluted with free 14N-serine released by their degradation. PEs were labeled by

15

N-serine (but not by

15

N-choline) presumably by successive

decarboxylation of 15N-labeled PS in mitochondria (reviewed in35), which corroborated their kinetic curves (Figure 4 C, D). Overall, PE were labeled at ca 4-fold lower rate compared to PC and SM and the labeled species were only detectable in ca 5 hours after adding

15

N-serine, once a significant

amount of 15N-labeled PS was produced. The kinetics of

15

N-labeling of SM was affected by both

15

N-serine and

reached the highest rate if both were present in the medium. Labeling with lowest (15%) incorporation rate and kept steady, while adding

15

15

15

N-choline and

N-serine reached the

N-choline increased the total

15

N-

incorporation. At the starting phase (< 10 h) labeling with 15N-choline / 15N-serine mixture followed the kinetics of labeling with 15N-serine and later (>10 h) – with 15N-choline. Interestingly, sUHR resolved the isotopic peaks of SM species that incorporated one and two

15

N atoms (Figure 4F)

confirming that both phosphocholine head group and the sphingosine backbone could be labeled. Therefore, we concluded that the time course of

15

N-labeling corroborated known SM biosynthesis

pathways (reviewed in36) and suggested specific and quantitative readout of

15

N-incorporation

kinetics by sUHR. sUHR determined

15

N-incorporation rate for isobaric species of each lipid class and it was

possible to estimate their total abundance (in mol%) and turnover (as 15N-incorporation rate) (Figure 5A). While there was no marked difference in the completeness of 15N-incorporation – for all PCs it

ACS Paragon Plus Environment

Page 21 of 28

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

21 reached a plateau at approximately 45%, the kinetics of labeling strongly differed between the species (Figure 5B).

Figure 5: 15N-labeling of molecular species of PC. (A) PC profile of HepG2 cells: 15N-labeled and unlabeled species are in red and blue, respectively. The inset shows the time course of a total molar fraction of unlabeled PC (assuming 100% at t=0). (B) Kinetics of 15N-incorporation into selected isobaric PC whose molecular composition did not alter during labeling. Relative abundance of molecular species (in %) was calculated from the abundances of acyl anion fragments of fatty acid moieties produced from their precursor ions by HCD FT MS/MS in negative mode.

However, one caveat is that the molecular composition of isobaric species could change during the labeling experiment and therefore the apparent labeling rate could reflect both the turnover of existing and biosynthesis of novel isobaric species having different fatty acid moieties7. To

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 22 of 28

22 determine how the length and unsaturation of fatty acid moieties affected the turnover, in a separate experiment we subjected PC precursors to HCD FT MS/MS19, identified their fatty acid moieties and estimated the relative abundances of individual species within each pool of isobaric lipids26 (Table S2). For the interpretation consistency, we selected a few isobaric PC that were predominantly represented by one major molecular species whose composition did not change over 30 h of

15

N-

labeling and compared their kinetic curves (Figure 5B). We observed that fatty acid moieties had a major impact on the turnover rate and that seems more dependent on their length, rather than unsaturation. PC having shorter (C14 – C16) fatty acid moieties incorporated

15

N-label at ca 30%

faster rate compared to PC with longer (C18 – C20) fatty acids. Therefore, we concluded that sUHR enabled absolute (molar) quantification and turnover rate measurements of nitrogen-containing membrane lipids at the full-lipidome scale in the course of the same experiment. It also emerged that simultaneous monitoring of labeling kinetics of many species within each lipid class is critical for understanding molecular mechanisms underlying lipid turnover because individual rates strongly depend on the properties of fatty acid moieties.

CONCLUSIONS AND OUTLOOK Cells engage a complex protein machinery and channel substantial chemical energy into maintaining the lipid homeostasis. A broad palette of lipid species and classes is being continuously synthesized de novo or remodeled by poorly understood molecular mechanisms. There is a need for generic and accessible analytical tools that combine the full-lipidome quantification with simultaneous monitoring of the turnover of individual lipids and we argue that sUHR method is capable of filling this gap.

ACS Paragon Plus Environment

Page 23 of 28

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

23 Although the scope of sUHR analyses reported in this work was limited to nitrogencontaining lipids, it nevertheless covered the majority of membrane lipid classes and, arguably, provided ample support for studying the lipid membrane composition in dynamics.

15

N-labeling

could be performed in the targeted (like described here) or fully untargeted way by supplying

15

N-

enriched media to cells or even model organisms that opens up an intriguing possibility of monitoring protein15 and lipid turnover during the same labeling experiment. Ultra-high mass resolution together with minimal ion coalescence is a foundation of sUHR. Baseline resolution of 13C- / 15N- peaks should allow to quantify 15N-labeled lipids at relatively low rates of isotope incorporation, which simplifies biological experiments, enables to carry them under more physiological conditions and using a broader scope of cells and model organisms. Importantly, sUHR lipidomics workflow utilized a commercial Orbitrap Elite mass spectrometer. Higher than 7×105 mass resolution (now also delivered by a recently marketed Orbitrap mass spectrometer) within the range of common lipid masses contributed to high fidelity of deconvolution of their isotopic profiles and it seems promising to combine sUHR with alternative approaches relying on less targeted 13C- and 18O-metabolic labeling in the future.

ACKNOWLEDGEMENTS Work in AS laboratory was supported by Max Planck Gesellschaft; Liver System Medicine (LiSyM) program and Lipidomics & Informatics for Life Sciences (LIFS) Unit of de.NBI Consortium, both funded by Bundesministerium f. Bildung u. Forschung (BMBF). Work in AS and ÜC laboratories was supported by TRR83 grant (TP17, AS; TP18, ÜC) from Deutsche Forschungsgemeinschaft (DFG). Work in ÜC laboratory was supported by German Federal Ministry of Education and Research grant to the German Center for Diabetes Research (DZD e.V.). Work in YOT laboratory

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 24 of 28

24 was supported by European Research Council (ERC Starting Grant 280271). TG was supported by Dresden International Graduate School for Biomedicine and Bioengineering funded by DFG (GS97). The authors are indebted to Thermo Fisher Scientific for expert technical support of UHR mass spectrometry on an Orbitrap Elite instrument.

SUPPORTING INFORMATION Supporting Information Available: Figures S1, S2, S3 and S4 detail technical aspects of t-SIM and sUHR; Table S1 describes internal standards used for lipids quantification; Table S2 presents changes of the molecular composition of isobaric PC species during

15

material is available free of charge via the Internet at http://pubs.acs.org.

ACS Paragon Plus Environment

N-metabolic labeling. This

Page 25 of 28

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

25

REFERENCES

(1) van Meer, G.; Voelker, D. R.; Feigenson, G. W. Nat Rev Mol Cell Biol 2008, 9, 112-124. (2) Holthuis, J. C.; Menon, A. K. Nature 2014, 510, 48-57. (3) Coskun, U.; Simons, K. Structure 2011, 19, 1543-1548. (4) Han, X.; Yang, K.; Gross, R. W. Mass Spectrom Rev 2012, 31, 134-178. (5) Wang, C.; Wang, M.; Han, X. Mol Biosyst 2015, 11, 698-713. (6) Yang, K.; Han, X. Trends Biochem Sci 2016, 41, 954-969. (7) Li, J.; Hoene, M.; Zhao, X.; Chen, S.; Wei, H.; Haring, H. U.; Lin, X.; Zeng, Z.; Weigert, C.; Lehmann, R.; Xu, G. Anal Chem 2013, 85, 4651-4657. (8) Brandsma, J.; Bailey, A. P.; Koster, G.; Gould, A. P.; Postle, A. D. Biochim Biophys Acta 2017, 1862, 792-796. (9) Ecker, J.; Liebisch, G. Prog Lipid Res 2014, 54, 14-31. (10) Postle, A. D.; Wilton, D. C.; Hunt, A. N.; Attard, G. S. Prog Lipid Res 2007, 46, 200-224. (11) Jacquot, C.; Wecksler, A. T.; McGinley, C. M.; Segraves, E. N.; Holman, T. R.; van der Donk, W. A. Biochemistry 2008, 47, 7295-7303. (12) Lingwood, D.; Simons, K. Science 2010, 327, 46-50. (13) Dancy, B. C.; Chen, S. W.; Drechsler, R.; Gafken, P. R.; Olsen, C. P. PLoS One 2015, 10, e0141850. (14) Gouw, J. W.; Pinkse, M. W.; Vos, H. R.; Moshkin, Y.; Verrijzer, C. P.; Heck, A. J.; Krijgsveld, J. Mol Cell Proteomics 2009, 8, 1566-1578. (15) Toyama, B. H.; Savas, J. N.; Park, S. K.; Harris, M. S.; Ingolia, N. T.; Yates, J. R.; Hetzer, M. W. Cell 2013, 154, 971-982. (16) Liebisch, G.; Vizcaino, J. A.; Kofeler, H.; Trotzmuller, M.; Griffiths, W. J.; Schmitz, G.; Spener, F.; Wakelam, M. J. J Lipid Res 2013, 54, 1523-1530. (17) Matyash, V.; Liebisch, G.; Kurzchalia, T. V.; Shevchenko, A.; Schwudke, D. J Lipid Res 2008, 49, 1137-1146.

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 26 of 28

26 (18) Schuhmann, K.; Almeida, R.; Baumert, M.; Herzog, R.; Bornstein, S. R.; Shevchenko, A. J Mass Spectrom 2012, 47, 96-104. (19) Schuhmann, K.; Herzog, R.; Schwudke, D.; Metelmann-Strupat, W.; Bornstein, S. R.; Shevchenko, A. Anal Chem 2011, 83, 5480-5487. (20) Zhurov, K. O.; Kozhinov, A. N.; Tsybin, Y. O. Energy & Fuels 2013, 27, 2974-2983. (21) Ghislain, T.; Molnarne Guricza, L.; Schrader, W. Rapid Commun Mass Spectrom 2017, 31, 495502. (22) Vetere, A.; Schrader, W. ChemistrySelect 2017, 2, 849-853. (23) Miladinović, S. M.; Kozhinov, A. N.; Gorshkov, M. V.; Tsybin, Y. O. Analytical Chemistry 2012, 84, 4042-4051. (24) Nagornov, K. O.; Kozhinov, A. N.; Tsybin, Y. O. J Am Soc Mass Spectrom 2017, 28, 768-780. (25) Schuhmann, K.; Thomas, H.; Ackerman, J. M.; Nagornov, K. O.; Tsybin, Y. O.; Shevchenko, A. Anal Chem 2017, 89, 7046-7052. (26) Herzog, R.; Schwudke, D.; Schuhmann, K.; Sampaio, J. L.; Bornstein, S. R.; Schroeder, M.; Shevchenko, A. Genome Biol 2011, 12, R8. (27) Herzog, R.; Schwudke, D.; Shevchenko, A. Curr Protoc Bioinformatics 2013, 43, 14 12 11-30. (28) Herzog, R.; Schuhmann, K.; Schwudke, D.; Sampaio, J. L.; Bornstein, S. R.; Schroeder, M.; Shevchenko, A. PLoS One 2012, 7, e29851. (29) Gorshkov, M. V.; Fornelli, L.; Tsybin, Y. O. Rapid Commun Mass Spectrom 2012, 26, 17111717. (30) Kaur, P.; O'Connor, P. B. Analytical Chemistry 2007, 79, 1198-1204. (31) Sales, S.; Graessler, J.; Ciucci, S.; Al-Atrib, R.; Vihervaara, T.; Schuhmann, K.; Kauhanen, D.; Sysi-Aho, M.; Bornstein, S. R.; Bickle, M.; Cannistraci, C. V.; Ekroos, K.; Shevchenko, A. Sci Rep 2016, 6, 27710. (32) Schwudke, D.; Schuhmann, K.; Herzog, R.; Bornstein, S. R.; Shevchenko, A. Cold Spring Harbor Perspect Biol 2011, 3, a004614. (33) Weber, R. J. M.; Southam, A. D.; Sommer, U.; Viant, M. R. Analytical Chemistry 2011, 83, 3737-3743. (34) Southam, A. D.; Weber, R. J.; Engel, J.; Jones, M. R.; Viant, M. R. Nat Protoc 2016, 12, 310328. (35) Vance, J. E.; Tasseva, G. Biochim Biophys Acta 2013, 1831, 543-554.

ACS Paragon Plus Environment

Page 27 of 28

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

27 (36) Gault, C. R.; Obeid, L. M.; Hannun, Y. A. Adv Exp Med Biol 2010, 688, 1-23.

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 28 of 28

28

Image for the table of content (TOC):

ACS Paragon Plus Environment