TO DOWNLOAD A COPY OF THIS POSTER, VISIT WWW.WATERS.COM/POSTERS ©2018 Waters Corporation

INTRODUCTION

Obesity is a risk factor for manifestation of metabolic

syndrome, a combination of metabolic abnormalities

such as type 2 diabetes, hyperlipidemia, hyper-

cholesterolemia and hypertension. It has previously

been demonstrated that excessive glycosphingolipids

play a role insulin resistance, a major driver of the

metabolic syndrome. Reduction of glycosphingolipids in

obese mice with small compound iminosugars has been

found to negate symptoms1. The work presented here is

to provide an integrated multi-omic and system biology

analysis of the results from the relative quantitative LC-

MS analysis of protein and lipid liver extracts from

control and obese mouse models undergoing treatment

to either prevent or revert obesity, providing

mechanistic insights.

INTEGRATED MULTI-OMICS AND SYSTEMS BIOLOGY ANALYSIS OF THE TARGETED TREATMENT OF OBESITY WITHIN MOUSE MODELS

Martijn van der Lienden1, Gertjan Kramer2, Mina Mirzaian2, Lee A Gethings3, Jimmy Yuk1, Johannes PC Vissers3, Johannes MFG Aerts1, Carmen Argmann4, Marco van Eijk1,2 1 Department of Medical Biochemistry, Leiden Institute of Chemistry, The Netherlands, 2 Academic Medical Centre, University of Amsterdam, The Netherlands, 3 Waters Corporation, Wilmslow, United Kingdom, 4 Department of Genetics and Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York, USA

Figure 1. Glycosphingolipid biosynthesis and pharmacological modula-

tion .

METHODS

Sample preparation

Proteins and lipids were extracted from liver tissue, which originated

from 3 control and 3 treated (MZ-21 inhibitor aka AMP-dnm) obese

mice. The protein extracts were prepared with 1% RapiGest SF prior to

reduction, alkylation and overnight digestion with trypsin.

Lipids were extracted by homogenizing liver tissue in chloroform-

methanol (2:1, v/v) and extracted according to the Bligh and Dyer

method. The extracts were centrifuged for phase separation and the

lower fraction collected for LC-MS analysis. An overview of the

experimental and analytical workflow is provided in Figure 1.

LC-MS conditions

Label-free LC-MS was used for qualitative and quantitative peptide

analyses. Experiments were conducted using a 90 min gradient from 5

to 40% acetonitrile (0.1% formic acid) at 300 nL/min using a nano-

ACQUITY system and a HSS 1.8 µm C18 reversed phase 75 µm x 15

cm nanoscale LC column. For lipid identification, the LC-MS

experiments consisted of a 20 min gradient from 3 to 40% isopropanol:

methanol (10 mM ammonium formate) at 500 µL/min using an

ACQUITY LC system. Here, a BEH 1.7 µm C8 reversed phase 2.1 x 10

cm LC column was used.

Proteomic data acquisition utilized data independent analysis (DIA) with

the LC system directly interfaced to a hybrid IMS-oaToF Synapt G2-Si

instrument. Ion mobility (IM) was used in conjunction with the proteomic

acquisition schema (HDMSE). Lipidomic measurements were conduc-

ted using a Xevo G2-S mass spectrometer, operating in broadband DIA

(MSE) mode.

References

1. Aerts et al. Glycosphingolipids and insulin resistance. Adv Exp Med Biol. 2011;721:99-119.

2. Aerts et al. Pharmacological inhibition of glucosylceramide synthase enhances insulin sensitivity.

Diabetes. 2007;56(5):1341-9.

3. Bijl et al. Reduction of Glycosphingolipid Biosynthesis Stimulates Biliary Lipid Secretion In Mice.

Hepatology. 2008;49(2):637-45

4. Lombardo et al. Correction of Liver Steatosis by a Hydrophobic Iminosugar Modulating

Glycosphingolipids Metabolism. PLoS One. 2012;7(10):e38520.

5. van Eijk et al. PLoS ONE. Reducing glycosphingolipid content in adipose tissue of obese mice restores

insulin sensitivity, adipogenesis and reduces inflammation. 2009;4(3):e4723.

6. Langeveld et al. Treatment of genetically obese mice with the iminosugar N-(5-adamantane-1-yl-

methoxy-pentyl)-deoxynojirimycin reduces body weight by decreasing food intake and increasing fat

oxidation.Metabolism. 2012;61(1):99-107

7. Gurzov et al. Hepatic oxidative stress promotes insulin-STAT-5 signaling and obesity by inactivating

protein tyrosine phosphatase N2. Cell Metab. 2014;20:85-102

Informatics

The LC-MS peptide data were processed and searched using

Progenesis QI for proteomics, whilst the lipid data was processed and

searched with Progenesis QI. In both cases normalized label-free

quantification was achieved from the Progenesis software. Additional

multivariate analysis was conducted with EZInfo. Pathway and network

analyses were performed with Ingenuity Pathway Analysis.

RESULTS

MZ21 inhibition has been studied previously2-6

and its efficacy

demonstrated as summarized in Figure 4. Here, multi-omics data and

results are presented, aiming to increase mechanistic insight in the

observed beneficial action of MZ-21.

CONCLUSION

A multi-omic study using DIA, label-free strategies has been applied to the study of obese mice which have been treated with an inhibitor of glucosylceramide metabolism

Over 1200 proteins were identified with 30% of the data showing differential expression

Phosphatidylcholines, sphingomyelins, triglycerides and lysophosphatidylcholines are identified as contributing towards the lipid variance

Carbohydrate and lipid metabolism were identified as pathways significantly influenced by MZ-21 treatment of mice.

A number of the noted metabolic changes induced by MZ-21 in the mice are consistent with literature on effects of stimulated STAT signaling

Reduced, alkylated& tryptic digestion overnight

Chloroform/water & centrifugation (lower phase analysed)

1D LC-DIA-IMS-MS

Label-freeQualitative/Quantitative Analysis

1D LC-DIA-MS

Proteins Lipids

Control(3 subjects)

MZ-21 treated(3 subjects)

Control(3 subjects)

MZ-21 treated(3 subjects)

Obese mice liver extracts(3 control and 3 treated)

Label-freeQualitative/Quantitative Analysis

Figure 2. Experimental design study for proteins and lipids extracted

from liver tissue of control and MZ-21 treated mice.

control

proteomics

MZ-21 treated

control

lipidomics

MZ-21 treated

Figure 6. Log-fold change significance (ANOVA (p)) vs. log2 protein fold

change distribution (top) allows highly probable, differentially expressed

proteins to be detected for control (grey) and treated (blue) subjects.

Discriminant analysis (OPLS-DA) results (bottom) illustrating the covari-

ance (x-axis) and correlation (y-axis) relationship between control and

MZ-21 treated lipid liver extract samples to identify, blue shaded, signifi-

cantly regulated lipids.

Figure 5. Unsupervised PCA scores plots for proteomic (upper) and lipi-

domic (lower). Inhibitor treated subjects are highlighted in red whilst

controls are black.

Figure 9. STAT5B and STAT5A transcription factors predicted by to be

upstream regulators of several proteins affected by MZ21 treatment.

Figure 3. Low energy precursor (light grey) and inverted elevated en-

ergy fragment (dark grey) ion m/z values as a function of retention time

(min) and drift time (bins) of liver control proteomic sample analyzed by

IM enabled DIA LC-MS.

Figure 8. Pathway analysis highlights ‘Lipid Metabolism’ as most the af-

fected function reflected in the differentially expressed proteins (bottom/

left) with the predicted metabolite changes, based on the proteomics re-

sults, confirmed by the lipidomics data (top/right).

Figure 4. MZ21 inhibition efficacy demonstrators/examples.

Figures 7. Qualitative results summary and identification examples.

The bubble plots are sized by log (1/p) and number of peptides for

the lipidomics and proteomics data sets, respectively (grey = control

and red = MZ-21 treated).

lipidomics

control

MZ-21

proteomics

MZ-21 control

MZ21 (AMP-dnm) is known to acutely increase inuslin sensitivity in

obese rodents2. Upon administration of the drug, as shown in Figure

10, insulin signaling in tissue such as liver is increased. The pathway

analysis of AMP-drug treated mouse liver points to a change in STAT5

signaling. This finding is consistent with the recent observation that ab-

normal insulin-STAT-5 signaling occurs in liver of obese mice7; appar-

ently the AMP-dnm treatment corrects this .

Small amounts of the purified liver extracts were analyzed to identify,

quantify and investigate the proteomic and lipidomic variance between

control and inhibitor treated subjects. PCA was initially used to identify

significant changes between control and inhibitor treated samples of

which an example is shown in Figure 5. Similar, complementary cluster-

ing patterns were observed for both the protein and lipid data sets.

Next, shown in Figure 6, MVA was conducted to detected differentially

regulated species. Following quantitative analysis, the proteomics and

lipidomics data were appended with identifications.

Searches were conducted within the data analysis software environ-

ments. In total, over 1200 protein groups , representing 2100 proteins,

were detected and quantified with ~ 30% of the proteins differentially

expressed. In the instance of the lipidomic LC-MS data, 73 regulated

lipid were tentatively identified of which ~ 20% were annotated using

rule-based logic. Shown in Figures 7 are typical examples of key regu-

lators.

The results of the integrated systems biology analysis of the LC-MS DIA

data are summarized in Figures 8 and 9. In summary, the combined

pathway analysis of the proteomic and lipidomic data identified ‘Lipid

Metabolism’ as the most affected function and transcription factors to be

predicted as upstream regulators of a signification portion of the de-

tected proteins demonstration chances in concentration/abundance as a

function of inhibitor treatment.

Figure 10. Improved insulin signaling in the liver of ob/ob mice treated

with MZ-21. Animals were fed for 2 weeks with 0, 5, or 25 mg AMP-

DNM/kg. To animals starved overnight, insulin (0.75 units/kg) was ad-

ministered via the vena porta. Livers were collected after 5 min. Insulin

receptor protein, and its autophosphorylation, and mTOR protein, and

its serine phosphorylation, were visualized and quantified.