0

3

-3

PC2 (18.6%)x107

x1070 2 4-2-4PC1 (38.0%)

Untargeted LC-MS/MS-based metabolic phenotyping applied to the CD248 knock out mouse modelNeil J Loftus1; Emily Armitage1; Alan Barnes1; Janak Bechar2; Ed Rainger3; Matthew Harrison3; Ian D Wilson4; Christopher D Buckley2; Amy J Naylor2

1 Shimadzu MS/BU, Manchester, United Kingdom; 2 Rheumatology Research Group, Institute of Inflammation and Ageing, University of Birmingham, Birmingham, United Kingdom; 3 Institute of Cardiovascular Sciences, University of Birmingham, Birmingham, United Kingdom; 4 Dept Metabolism, Digestion and Reproduction, Imperial College, London, United Kingdom

Overview Untargeted metabolomics utilizing HRAM LC-MS/MS analysis has been

applied to study the impact of a C57Bl/6J CD248-/- (knock out) on mouse serum profiles following administration of a high fat diet, relative to chow diet controls.

The response to diet has been compared to a C57Bl/6J CD248+/+ wild type phenotype.

1. IntroductionUntargeted LC-MS/MS is a powerful tool by which to identify metabolic phenotypes. Here we applied metabolic phenotyping analysis to the CD248 knockout mouse model. CD248 is a transmembrane glycoprotein, expression of which is markedly upregulated in a considerable number of disease models including tumor growth, inflammation and injury-induced fibrosis. In human clinical studies, CD248 expression is upregulated in the fat cells of patients with diabetes, conversely, CD248 expression reverted to a normal range when obesity-associated diabetes was reversed through weight loss. In this work, an untargeted LC-MS/MS metabolic phenotyping analysis, using a reverse-phase LC separation and high resolution accurate mass (HRAM), was applied to a CD248-/- mouse model following high fat diet (HFD) feeding to study the effects of diet on serum metabolite profiles.

2. MethodsSerum samples were taken from C57Bl/6J CD248-/- knock out mice and C57Bl/6J CD248+/+ wild type controls fed with high fat diet or a regular chow diet. Following protein precipitation with methanol and subsequent centrifugation, the precipitate was resuspended and re-extracted with water and methanol. Extracts were combined for each animal and analyzed, after randomization, using HRAM (high resolution accurate mass) mass spectrometry (LCMS-IT-TOF system and a QTOF LCMS-9030; Shimadzu Corporation, Japan). MS and MS/MS data were acquired in positive and negative ion mode on the Q-TOF with MS (m/z 100-1000; 100 msecs) and DDA-MS/MS (18 mass scans; 50 msec for each scan event with a collision energy spread 0-40V to acquire precursor and product ion data in each cycle). Cycle time of 1 second.

3. ResultsMetabolic features were extracted from raw HRAM LC-MS data and filtered based on QC criteria (ion signals are present in at least 50 % of the QC samples and at RSD < 30%) as well as groups (ion signals present in at least 75 % of at least one of the 4 groups).

3.1 Serum lipid profiling in the CD248+/+ knock out mouse model following a high fat dietIn the wild type control group different lyso-phospholipid species, mainly lysophosphatidylcholines (LPC) and phosphatidylcholine (PC) species wereidentified as being differentially changed following a HFD (Figure 4). Cholesterol was also differentially expressed. In this group the LPC and PC species were significantly elevated in the wild type control group C57Bl/6 CD248 +/+ following a HFD, with few exceptions.

In the C57Bl/6 CD248 knock out mice, LPC and PC species also differed from the wild type controls, but the magnitude of change was lower.

0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 32.5 35.00.00

0.25

0.50

0.75

1.00

1.25

1.50

1.75

x106

CholineL-carnitine

Pantothenic acid

FFAs FAMEsLPLs

PLsSLs

Cholesterol

Figure 1. Precursor HRAM mass chromatograms of 95 annotated significant components detected in the pooled QC serum extract. LPLs (lyso-phospholipids), SLs (sphingolipids), FFAs (free fatty acids) and FAMEs (fatty acid methyl esters), in addition to cholesterol, L-carnitine, choline and pantothenic acid.

Figure 2. PCA scores plot for 5596 features extracted using HRAM LC-MS in positive ion mode. Metaboanalyst software was used to analyze the data using multivariate (PCA, heat map analysis, pattern hunter) and univariate (ANOVA) techniques.

4. Conclusions A HRAM LC-MS and DDA-MS/MS method was applied to study the metabolic

effects of a mouse CD248 knock-out relative to a wild-type phenotype when a high fat diet was administered compared to a standard chow diet control. Significant differences in metabolite response were identified using multivariate and univariate statistics including PCA, ANOVA and heat-map analysis.

Based upon their metabolite profiles, differential effects of the high fat diet were observed between the knock-out and wild type, suggesting CD248 is involved in an altered response to high fat diets. The most marked differences were observed in cholesterol, choline, carnitine, pantothenic acid, lyso-phosphatidylcholines, free fatty acids and fatty acid methyl esters.

HRAM Q-TOF (LCMS-9030 Shimadzu Corporation) acquired MS and DDA-MS/MS data with a cycle time of 1 second over the MS/MS mass range of 40-1000 Da was performed to support metabolite and lipid identification.

High Fat Diet CD248-/- mouse group

High Fat Diet CD248+/+ control group

Regular chow diet Both CD248+/+ control and

CD248-/- mouse group

Figure 3. Boxplots generated using Metaboanalyst presenting 4 metabolites that significantly differed in response (peak area) to high fat diet that were specific to the CD248 knock out phenotype. When fed a high fat diet, the serum concentration of cholesterol was significantly increased in wildtype control mouse group whereas L-carnitine and pantothenic acid were reduced in the CD248 knock out mouse group.

L-Carnitinex106

-10

-5

0

5

Cholesterolx106

3

2

1

0

-1

-2

Pantothenic acidx106

2

1

0

-1

CD248 +/+ CD248 -/-Chow HFDChow HFD QCs

Wildtype CD248-/-

Chow HFDChow HFDClass

LPC 14:0 sn-1LPC 14:0 sn-2

LPC 16:0 sn-2

LPC 16:1 sn-2

LPC 18:0 sn-2

LPC 18:1 sn-2

LPC 18:2 sn-2

LPC 20:2 sn-2

LPC 20:3 sn-2

LPC 20:4 sn-2

LPC 20:5 sn-2

LPC 22:6 sn-2

LPC 16:0 sn-1

LPC 16:1 sn-1

LPC 18:0 sn-1

LPC 18:1 sn-1

LPC 20:1 sn-1LPC 20:2 sn-1

LPC 18:3 sn-1

LPC 18:2 sn-1

LPC 20:3 sn-1

LPC 20:4 sn-1

LPC 20:5 sn-1

LPC 22:4 sn-1LPC 22:5 sn-1LPC 22:6 sn-1

PC 32:1PC 32:2PC 34:2PC 34:3PC 36:1PC 36:2PC 36:4PC 36:5PC 38:3PC 38:4PC 38:5PC 38:6PC 40:6

CD248+/+ CD248-/-

Chow HFDChow HFD QCs

Cholinex106

-4

0

4

8

CD248 +/+ CD248 -/-Chow HFDChow HFD QCs

Distribution of LPC and PC Influence of high fat diet The HFD caused either no, or lower, elevation in LPCs in the CD248 -/- compared to the wild type, while others (LPC 18:2 for example) were significantly reduced only in the CD248 -/-phenotype after HFD exposure.

CD248 +/+ ChowCD248 +/+ HFDCD248 -/- ChowCD248 -/- HFD

Figure 4. Heatmap generated using Metaboanalyst highlighting the differential expression of LPC and PC species

LPC 14:0 sn-1x106

2

1

0

LPC 20:5 sn-1x106

2

1

0

-1

3

LPC 18:2 sn-1x106

0

-5

5

3.2 High resolution accurate mass metabolite identification in metabolic phenotyping An untargeted LC-MS and MS/MS based metabolic phenotyping workflow was

applied to the CD248 knock out mouse model following a high fat diet. For putatively annotating precursor ion metabolic features, data were acquired

using a QTOF DDA-MS/MS method with a collision energy spread of 0-40V and 18 DDA-MS/MS scans with a mass scan time of 50 msecs.

Metabolite identification was in agreement with published databases. Targeted DDA-MS/MS was further validated using a cross platform approach

with different mobile phases to help enhance positive and negative ion data.

Figure 5. DDA-MS/MS spectra shown for the sn-1 isoform in positive ion mode and the [M+HCOOH-H]- adduct in negative ion mode (FWHM). As the complexity of the lipidome includes 8 major categories of lipids, over 80 major classes, 300 sub-classes and thousands of lipid species, acquiring targeted DDA-MS/MS in both ionization modes helped to provide high confidence in the identification of lipids.

Disclaimer: The products and applications in this presentation are intended for Research Use Only (RUO). Not for use in diagnostic procedures.

104.10616

184.07296

285.23977450.29691

468.30964

m/z50 100 150 200 250 300 350 400 450 500

LPC 14:0 sn-1 Positive ion data

152.99765224.06744

227.20135

452.28032

512.30055

m/z50 100 150 200 250 300 350 400 450 500

LPC 14:0 sn-1 Negative ion data

HRAM Metabolite IdentificationPositive and negative ion data in lipid identificationReported mass accuracy Positive ion data m/z 468.30964 2.498 ppmm/z 450.29691 -2.199 ppmm/z 184.07296 -1.956 ppm

Negative ion data m/z 512.30055 2.264 ppmm/z 452.28032 4.555 ppmm/z 227.20135 -1.320 ppm