7
Untargeted Metabolomics Workflow Using UHPLC/Quadrupole Orbitrap Mass Spectrometer and SIEVE 2.1 Software Junhua Wang, David A. Peake, Mark Sanders, Michael Athanas, and Yingying Huang Thermo Fisher Scientific, San Jose, CA

Untargeted Metabolomics Workfl ow Using …apps.thermoscientific.com/media/cmd/kl/resources/PN_ASMS13_T382... · Untargeted Metabolomics ... Workflow Using UHPLC/Quadrupole Orbitrap

  • Upload
    vunhi

  • View
    232

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Untargeted Metabolomics Workfl ow Using …apps.thermoscientific.com/media/cmd/kl/resources/PN_ASMS13_T382... · Untargeted Metabolomics ... Workflow Using UHPLC/Quadrupole Orbitrap

Untargeted Metabolomics Workfl ow Using UHPLC/Quadrupole Orbitrap Mass Spectrometer and SIEVE 2.1 Software Junhua Wang, David A. Peake, Mark Sanders, Michael Athanas, and Yingying HuangThermo Fisher Scientifi c, San Jose, CA

Page 2: Untargeted Metabolomics Workfl ow Using …apps.thermoscientific.com/media/cmd/kl/resources/PN_ASMS13_T382... · Untargeted Metabolomics ... Workflow Using UHPLC/Quadrupole Orbitrap

2 Untargeted Metabolomics Work� ow Using UHPLC/Quadrupole Orbitrap Mass Spectrometer and SIEVE 2.1 Software

Untargeted Metabolomics Workflow Using UHPLC/Quadrupole Orbitrap Mass Spectrometer and SIEVE 2.1 Software Junhua Wang, David A. Peake, Mark Sanders, Michael Athanas, Yingying Huang Thermo Fisher Scientific, San Jose, CA

Conclusion An efficient and robust workflow for untargeted metabolomics is presented here. The reliable high-resolution, accurate-mass (HR/AM) performance of the Q Exactive LC-MS system eliminates the need for technical replicates on biological samples. The superior S/N in Orbitrap data allows efficient data reduction in SIEVE 2.1 software, resulting in much reduced data analysis effort. KEGG pathway visualization allows quick access to biological pathway mapping. The MSn spectral library mzCloud facilitates accurate compound identification.

References 1. Wang S.Y., Kuo, C.H., Tseng, Y. J. Anal. Chem., 2013, 85 (2), 1037–1046. 2. Scripps Center For Metabolomics Home Page. https://xcmsonline.scripps.edu/index.php 3. m/zCloud Home Page. https://www.mzcloud.org/home.aspx 4. Muoio, D.M.; Newgard, C.B. Nature Rev. Mol. Cell. Biol. 2008, 9, 193–205.

Overview Purpose: Demonstrate a generic, integrated workflow for untargeted metabolomics study using a UHPLC/benchtop Thermo Scientific™ Orbitrap™ mass spectrometer and informatics software.

Introduction Metabolomics is a rapidly growing field of post-genomic biology, aiming to comprehensively characterize the small molecules in biological systems. Nonbiological systematic biases from instrument calibration or the order of sample injection account for the most significant errors in LC/TOF-MS data [1]. Here we present a workflow using a UHPLC/benchtop quardrupole Orbitrap platform and automated data analysis software for untargeted metabolomic profiling of plasma samples for biomarker discovery from the Zucker diabetes fatty (ZDF) rat model. The optimal conditions for sample preparation, liquid chromatography (LC), column, mass spectrometry (MS), and data processing parameters are explored.

Methods Sample Preparation

Plasma samples were deproteinized with organic solvent. Four extraction solvent systems including methanol (MeOH), acetonitrile (ACN), acetone, and 1:1:1 of the above were tested in this work. Endogenous metabolites were reconstituted in methanol/water (1:9) containing isotopically labeled internal standard (IS), d5-hippuric acid for LC-MS analysis. Solvent blank, pooled QC, and biological samples were analyzed in a randomized injection order.

FIGURE 2. UHPLC/MS Experimental Design and Run Sequence. Left, schematic showing the vials and sample names. Right, detailed content and overall time for each step.

FIGURE 3. Metabolomics data analysis. SIEVE 2.1 software, Thermo Scientific™ TraceFinder™ 3.1 software, and mzCloud support untargeted and targeted metabolomics workflows.

Component Detection

Differential (Statistical)

Analysis Component

Identification Quantitation Pathway Annotation

Advanced Statistical Analysis

• SIEVE • XCMS • mzMine

• SIEVE

• SIEVE • mzCloud • METLIN

• SIEVE • TraceFinder

• SIEVE • KEGG • Cytoscape

• SimcaP • Partek • GeneData • R

Targeted Untargeted

Results Challenges in Untargeted Metabolomics Study

Complexity of biological samples

Diversity of small molecule metabolites: polar and non-polar analytes

Ionization requires both positive and negative ion

Wide range of concentrations

No universal method for chromatographic separation

Multiple sources of variability

Structure elucidation of unknowns is expensive: lack of synthetic standards

Preparing for the UHPLC-MS Data Acquisition

Prior to the real samples analysis, a solvent blank with internal standard (IS) is injected at the beginning to check the solvent and the LC-MS status. The injections of the real samples should be randomized in order to eliminate systematic bias. Triplicate injections of the pooled plasma are intermittently repeated throughout the whole batch to validate consistent performance of the overall system. The experimental design and run sequence are shown in Figure 2.

UHPLC provides fast chromatography for high throughput analysis, the typical peak width is 4–6 seconds. Our method can baseline resolve Isoleucine and Leucine, generating peak width 1.2 s at FWHM. Refer to Figure 4. For such narrow peaks, Q Exactive mass spectrometer operating at 70,000 resolution acquires >15 point across the peak without losing sensitivity (A).

The QC of each run was performed by monitoring the intensity of IS, d5-hippuric acid (B), and the overall base peak (C). When all samples were finished, the selected ion chromatograms can be quickly viewed with RawMeat 2.1 (a free app with SIEVE 2.1 software). By inspecting the RT and intensity of the IS, runs with large retention time drift and bad injections can be excluded from the following data analysis.

SIEVE 2.1 Software XCMS Online Ratio 5.0 Fold change 4.8

p-value 3.44e-4 p-value 3.4e-4 RT 2.03 min RT 2.03 min

IDs: Norepinephrine

Pyridoxine Oxidopiamine, etc.

IDs: Norepinephrine

Pyridoxine (VB6) Hydroxydopamin, etc.

Table 1. Statics of component m/z 170.081 from SIEVE 2.1 software and XCMS Online

[M+H]+ Time Fatty/Lean P-value Δppm Formula Name 854.5657 7.25 2.81 3.65E-02 4.0 C50H80NO8P PC44:10 838.6292 8.84 2.48 6.45E-03 3.0 C48H88NO8P PC 40:4 836.6133 8.55 4.54 2.22E-02 0.0 C46H88NO8P PC38:2 812.6095 8.58 2.79 3.08E-02 4.0 C46H86NO8P PC 38:3 788.6127 9.04 2.13 3.56E-02 3.0 C44H86NO8P PC36:1 778.5351 7.20 4.15 3.38E-03 0.0 C42H78NO8P PC34:3 768.5498 8.01 2.14 3.17E-04 4.0 C43H78NO8P PC35:4 764.5196 7.44 3.71 6.37E-04 3.0 C43H74NO8P PC35:6 754.5347 7.29 2.67 1.35E-02 3.0 C42H76NO8P PC34:4 834.5936 8.10 2.83 7.13E-04 0.0 C44H83NO13 LacCer(d18:1/14:0) 836.6041 8.01 2.42 1.10E-03 4.0 C44H85NO13 LacCer(d18:0/14:0) 813.6790 9.72 2.40 3.84E-02 4.0 C47H93N2O6P SM(d18:2/24:0) 675.5412 7.20 2.29 3.34E-02 3.0 C37H75N2O6P SM(d16:1/16:0) 546.3525 5.84 2.33 9.23E-03 4.0 C28H52NO7P LysoPC(20:3) 526.2910 5.69 2.42 1.67E-04 3.0 C27H44NO7P LysoPE(22:6) 153.0521 1.79 2.46 1.56E-03 3.0 C4H12N2S2 Cystamine 230.0954 0.80 2.53 2.80E-02 1.0 C9H15N3O2S Ergothioneine 245.0917 3.56 3.93 2.16E-03 1.0 C13H12N2O3 Haematopodin 172.1693 5.10 3.18 4.19E-02 1.0 C10H21NO decanamide 302.3043 5.48 2.28 2.50E-02 3.0 C18H39NO2 Sphinganine 204.1228 3.75 3.10 1.83E-02 1.0 C9H17NO4 Acetylcarnitine 232.1534 2.26 4.49 1.99E-03 2.0 C11H21NO4 Butyryl-L-carnitine 344.2786 5.31 2.13 1.57E-02 2.0 C19H37NO4 Lauroylcarnitine 400.3409 5.53 2.10 3.49E-04 2.0 C23H45NO4 Palmitoyl-L-carnitine 466.3160 4.97 0.07 3.52E-03 0.0 C26H43NO6 Glycocholic Acid 357.2780 5.39 0.15 3.03E-03 2.0 C24H36O2 THA 355.2626 5.13 0.17 8.89E-03 1.0 C24H34O2 delta2-THA 170.0810 2.03 0.20 3.44E-04 0.0 C8H11NO3 Norepinephrine, 5-Hydroxydopamine 184.0968 2.69 0.23 2.70E-02 1.0 C9H13NO3 Normetanephrine; Methylnoradrenaline 212.1279 3.74 0.34 4.04E-03 1.0 C11H17NO3 Methoxamine 226.1433 4.20 0.29 8.10E-03 1.0 C12H19NO3 N-Methylmescaline; Terbutaline 161.0918 2.45 0.35 3.76E-03 2.0 C6H12N2O3 (2S)-4-Acetamido-2-aminobutanoic acid 224.0915 2.59 0.34 3.56E-02 1.0 C11H13NO4 Acetyl-L-tyrosine 146.1173 1.44 0.42 3.54E-02 1.0 C7H15NO2 DL-Aminoheptanoic acid

RT: 1.6471 - 1.8824 SM: 7B

1.65 1.70 1.75 1.80 1.85Time (min)

0

10

20

30

40

50

60

70

80

90

100

Rel

ativ

e Ab

unda

nce

1.8147132.1020

1.7255132.1020

NL:4.76E8m/z= 132.1012-132.1026 MS Urine_Junhua_1_10_4Isoleucine

Leucine FWHM 1.5 s;

7 Hz, Res 70K

Δppm 0.8

RT: 0.00 - 12.01 SM: 7B

0 2 4 6 8 10 12Time (min)

0

20406080

100

Rel

ativ

e Ab

unda

nce

02040

6080

100

Rel

ativ

e Ab

unda

nce

2.83185.0966

2.48205.0967

11.31149.01176.38

391.28235.88496.3373 8.35

732.5496

NL:1.94E8m/z= 185.0957-185.0975 MS ZDF_F1_2

NL:3.17E9m/z= 80.0000-1200.0000 MS ZDF_F1_2

IS=d5-hippuric acid

Base Peak

A

B

C

FIGURE 1. Untargeted metabolomics workflow

Comparing Differential Analysis Results from SIEVE 2.1 software and XCMS Online

The results from SIEVE 2.1 software are compared to XCMS Online, an academia-developed open-source software for metabolomics data analysis (Table 1) (2). As shown in Figure 5, the significant components (p-value <0.05, ratio > 2.0) identified by SIEVE 2.1 software and XCMS Online are similar, while the SIEVE software started with a much smaller (~20 less) number of components because of its capability to automatically remove solvent background, thus saving users time and labor in data review (Figure 6).

Metabolite Identification Guided by mzCloud

The mzCloud is the first cloud-based MSn spectral trees library built for small molecule structural elucidation (3). It contains a standard reference database, an elucidated compound database with putative structures, and a database of virtual spectra. Q Exactive HCD MS/MS spectra of interested metabolites can be searched against m/zCloud. The matching compounds are scored and the fragment ions are annotated accordingly (Figure 7).

FIGURE 5. PCA by SIEVE 2.1 software (A), Numbers in SIEVE 2.1 software (B), Numbers in XCMS Online (C)

206

2,075 components

403 p<0.05

Ratio > 2 212

440 „real peaks‟

38,268 features

3,848 p<0.05

A B

C

FIGURE 6. SIEVE 2.1 software output of component m/z 170.081, showing alignment across different samples (A), adducts grouping (B), peak integration (C), and trend intensity view (D)

15 features grouped to one component

A C

D B

FIGURE 7. Searching MS/MS spectrum of component m/z 184.0607 in mzCloud. The entry of 4-pyridoxic acid matches the spectrum with the highest score.

TABLE 2. Representative metabolites that are significantly changed

4-Pyridoxic acid_pos_HCD_QE #141 RT: 1.26 AV: 1 NL: 2.70E8T: FTMS + p ESI Full ms2 [email protected] [50.00-205.00]

50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200m/z

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

80

85

90

95

100

Rel

ativ

e Ab

unda

nce MS/MS of m/z 184.0606 from sample

mzCloud search and score

4-pyridoxic acid

166.0500

148.0393 138.0551 184.0607

FIGURE 8. SIEVE 2.1 software with KEGG pathway visualization

p-value: 3.76e-3 Ratio: 2.86 Composition: C6H12N2O3

ChemSpider ID: 389841 (2S)-4-Acetamido-2- aminobutanoic acid Branched amino acids are known hallmarks of Type II Diabetes [3]

XIC

KEGG is a trademark of Kyoto University, Kanehisa Lab. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries.

This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others.

Take an aliquot of 200 µL plasma

Add 600 µL (3) of cold 4 C organic solvent and vortex (3 min)

Spin down at 4 C, 4000 g (5 min)

Take supernatant and divide into 3 portions (each 200 µL), dry down (30–60 min)

Reconstitute in 400 µL (9:1) H2O:CH3OH containing IS, vortex, and sonicate (2 min)

Liquid Chromatography

UHPLC separation was implemented on a Thermo Scientific™ Dionex™ UltiMate™ 3000 HPG (high-pressure gradient) pump using Thermo Scientific™ Hypersil GOLD™ RP C18 column at 450 µL/min, column temperature at 55 C. LC solvents were 0.1% FA (A) and 0.1% FA in MeOH (B). Apply linear gradient from 0.5–50% B for 5.5 min, followed by increasing to 98% at 6 min, hold 98% B for 6 min, then decrease to 0.5% at 13 min, then equilibrate for another 2 min.

Mass Spectrometry

The Thermo Scientific™ Q Exactive™ mass spectrometer was operated under electrospray ionization (ESI) positive, negative, and polarity switching modes. Full scan (m/z 67–1000) used resolution 70,000 with automatic gain control (AGC) target of 1106 ions and a maximum ion injection time (IT) of 35 ms. Data-dependent MS/MS were acquired on a “Top5” data-dependent mode using the following parameters: resolution 17,500; AGC 1105 ions; maximum IT 80 ms; 2.0 amu isolation window; normalized collision energy 35% ± 50%; underfill ratio 1.0%; Apex trigger 2–4 s, and dynamic exclusion 6 s. Source ionization parameters were: spray voltage, 3.8 kV; capillary temperature, 300 C; and S-Lens level, 45.

Data Analysis

Differential analyses of the obese versus lean plasma were performed using Thermo Scientific™ SIEVE™ 2.1 software, which executes background subtraction, component detection, peak alignment, differential analysis (Figure 1). Statistical results, putative IDs, and pathways were generated through searching ChemSpider and KEGG™. Metabolites of interest were identified via MS/MS mass spectral database matching. The *.raw files were converted to mzXML format using ProteoWizard and also analyzed by XCMS Online [2] to compare the results.

3h

14h

15m

14h

15m

Solvent blank

with IS

QC Mixture with IS

Solvent blank NO IS

Pooled plasma with IS

Control sample with IS

#2

Treated sample with IS

#1

Control sample with IS

#1

Treated sample with IS

#2

b a

I

II

III

IV

1

2

3

4

5

6

7

9

Control sample with IS

#2

Treated sample with IS

#2

8

Control sample with IS

# n

Treated sample with IS

# n

Step | Preconditioning LC under one polarity (+ or -) 1. Solvent blank spiked with IS, 5-10X | (check signal) 2. QC mixture with IS, 3-5X | (check LC system status) 3. Solvent blank with IS, 1X | (check carry over) • If carry over, wash followed by repeating step 2 & 3 | (clean) • If no carry over, proceed to next step

Step | Experiment I: Positive, R=70K 4. Solvent blank no IS, 1X | (for background subtraction) 5. Pooled plasma with IS, 3X | (check method and column) 6. Control sample with IS #1, 1X | 7. Treated sample with IS #1, 1X | 8. Run all biological replicates with a randomized order | (all)

Step | Experiment II: MS/MS Positive 9. Pooled plasma, top5 ddMS2 HCD | (dd-MS/MS)

Step | Experiment III: Negative, R=70K 4-8. Run all biological replicates as Experiment I | (all samples)

Step | Experiment IV: MS/MS Negative 10. Repeat as in Experiment II, step 9 | (dd-MS/MS)

FIGURE 4. Method validation, data quality control in metabolomics application

An example of SIEVE 2.1 software with KEGG pathway visualization demonstrated with component m/z 161.0918 is shown in Figure 8. All matching metabolites are labeled with black dots.

Table 2 shows some of the representative metabolites that are significantly changed. Up-regulated metabolites in fatty vs. lean rat plasma are shown in green while down-regulated metabolites are shown in blue (p-value <0.05, ratio >2).

Page 3: Untargeted Metabolomics Workfl ow Using …apps.thermoscientific.com/media/cmd/kl/resources/PN_ASMS13_T382... · Untargeted Metabolomics ... Workflow Using UHPLC/Quadrupole Orbitrap

3Thermo Scienti� c Poster Note • PN ASMS13_T382_JWang_E 07/13S

Untargeted Metabolomics Workflow Using UHPLC/Quadrupole Orbitrap Mass Spectrometer and SIEVE 2.1 Software Junhua Wang, David A. Peake, Mark Sanders, Michael Athanas, Yingying Huang Thermo Fisher Scientific, San Jose, CA

Conclusion An efficient and robust workflow for untargeted metabolomics is presented here. The reliable high-resolution, accurate-mass (HR/AM) performance of the Q Exactive LC-MS system eliminates the need for technical replicates on biological samples. The superior S/N in Orbitrap data allows efficient data reduction in SIEVE 2.1 software, resulting in much reduced data analysis effort. KEGG pathway visualization allows quick access to biological pathway mapping. The MSn spectral library mzCloud facilitates accurate compound identification.

References 1. Wang S.Y., Kuo, C.H., Tseng, Y. J. Anal. Chem., 2013, 85 (2), 1037–1046. 2. Scripps Center For Metabolomics Home Page. https://xcmsonline.scripps.edu/index.php 3. m/zCloud Home Page. https://www.mzcloud.org/home.aspx 4. Muoio, D.M.; Newgard, C.B. Nature Rev. Mol. Cell. Biol. 2008, 9, 193–205.

Overview Purpose: Demonstrate a generic, integrated workflow for untargeted metabolomics study using a UHPLC/benchtop Thermo Scientific™ Orbitrap™ mass spectrometer and informatics software.

Introduction Metabolomics is a rapidly growing field of post-genomic biology, aiming to comprehensively characterize the small molecules in biological systems. Nonbiological systematic biases from instrument calibration or the order of sample injection account for the most significant errors in LC/TOF-MS data [1]. Here we present a workflow using a UHPLC/benchtop quardrupole Orbitrap platform and automated data analysis software for untargeted metabolomic profiling of plasma samples for biomarker discovery from the Zucker diabetes fatty (ZDF) rat model. The optimal conditions for sample preparation, liquid chromatography (LC), column, mass spectrometry (MS), and data processing parameters are explored.

Methods Sample Preparation

Plasma samples were deproteinized with organic solvent. Four extraction solvent systems including methanol (MeOH), acetonitrile (ACN), acetone, and 1:1:1 of the above were tested in this work. Endogenous metabolites were reconstituted in methanol/water (1:9) containing isotopically labeled internal standard (IS), d5-hippuric acid for LC-MS analysis. Solvent blank, pooled QC, and biological samples were analyzed in a randomized injection order.

FIGURE 2. UHPLC/MS Experimental Design and Run Sequence. Left, schematic showing the vials and sample names. Right, detailed content and overall time for each step.

FIGURE 3. Metabolomics data analysis. SIEVE 2.1 software, Thermo Scientific™ TraceFinder™ 3.1 software, and mzCloud support untargeted and targeted metabolomics workflows.

Component Detection

Differential (Statistical)

Analysis Component

Identification Quantitation Pathway Annotation

Advanced Statistical Analysis

• SIEVE • XCMS • mzMine

• SIEVE

• SIEVE • mzCloud • METLIN

• SIEVE • TraceFinder

• SIEVE • KEGG • Cytoscape

• SimcaP • Partek • GeneData • R

Targeted Untargeted

Results Challenges in Untargeted Metabolomics Study

Complexity of biological samples

Diversity of small molecule metabolites: polar and non-polar analytes

Ionization requires both positive and negative ion

Wide range of concentrations

No universal method for chromatographic separation

Multiple sources of variability

Structure elucidation of unknowns is expensive: lack of synthetic standards

Preparing for the UHPLC-MS Data Acquisition

Prior to the real samples analysis, a solvent blank with internal standard (IS) is injected at the beginning to check the solvent and the LC-MS status. The injections of the real samples should be randomized in order to eliminate systematic bias. Triplicate injections of the pooled plasma are intermittently repeated throughout the whole batch to validate consistent performance of the overall system. The experimental design and run sequence are shown in Figure 2.

UHPLC provides fast chromatography for high throughput analysis, the typical peak width is 4–6 seconds. Our method can baseline resolve Isoleucine and Leucine, generating peak width 1.2 s at FWHM. Refer to Figure 4. For such narrow peaks, Q Exactive mass spectrometer operating at 70,000 resolution acquires >15 point across the peak without losing sensitivity (A).

The QC of each run was performed by monitoring the intensity of IS, d5-hippuric acid (B), and the overall base peak (C). When all samples were finished, the selected ion chromatograms can be quickly viewed with RawMeat 2.1 (a free app with SIEVE 2.1 software). By inspecting the RT and intensity of the IS, runs with large retention time drift and bad injections can be excluded from the following data analysis.

SIEVE 2.1 Software XCMS Online Ratio 5.0 Fold change 4.8

p-value 3.44e-4 p-value 3.4e-4 RT 2.03 min RT 2.03 min

IDs: Norepinephrine

Pyridoxine Oxidopiamine, etc.

IDs: Norepinephrine

Pyridoxine (VB6) Hydroxydopamin, etc.

Table 1. Statics of component m/z 170.081 from SIEVE 2.1 software and XCMS Online

[M+H]+ Time Fatty/Lean P-value Δppm Formula Name 854.5657 7.25 2.81 3.65E-02 4.0 C50H80NO8P PC44:10 838.6292 8.84 2.48 6.45E-03 3.0 C48H88NO8P PC 40:4 836.6133 8.55 4.54 2.22E-02 0.0 C46H88NO8P PC38:2 812.6095 8.58 2.79 3.08E-02 4.0 C46H86NO8P PC 38:3 788.6127 9.04 2.13 3.56E-02 3.0 C44H86NO8P PC36:1 778.5351 7.20 4.15 3.38E-03 0.0 C42H78NO8P PC34:3 768.5498 8.01 2.14 3.17E-04 4.0 C43H78NO8P PC35:4 764.5196 7.44 3.71 6.37E-04 3.0 C43H74NO8P PC35:6 754.5347 7.29 2.67 1.35E-02 3.0 C42H76NO8P PC34:4 834.5936 8.10 2.83 7.13E-04 0.0 C44H83NO13 LacCer(d18:1/14:0) 836.6041 8.01 2.42 1.10E-03 4.0 C44H85NO13 LacCer(d18:0/14:0) 813.6790 9.72 2.40 3.84E-02 4.0 C47H93N2O6P SM(d18:2/24:0) 675.5412 7.20 2.29 3.34E-02 3.0 C37H75N2O6P SM(d16:1/16:0) 546.3525 5.84 2.33 9.23E-03 4.0 C28H52NO7P LysoPC(20:3) 526.2910 5.69 2.42 1.67E-04 3.0 C27H44NO7P LysoPE(22:6) 153.0521 1.79 2.46 1.56E-03 3.0 C4H12N2S2 Cystamine 230.0954 0.80 2.53 2.80E-02 1.0 C9H15N3O2S Ergothioneine 245.0917 3.56 3.93 2.16E-03 1.0 C13H12N2O3 Haematopodin 172.1693 5.10 3.18 4.19E-02 1.0 C10H21NO decanamide 302.3043 5.48 2.28 2.50E-02 3.0 C18H39NO2 Sphinganine 204.1228 3.75 3.10 1.83E-02 1.0 C9H17NO4 Acetylcarnitine 232.1534 2.26 4.49 1.99E-03 2.0 C11H21NO4 Butyryl-L-carnitine 344.2786 5.31 2.13 1.57E-02 2.0 C19H37NO4 Lauroylcarnitine 400.3409 5.53 2.10 3.49E-04 2.0 C23H45NO4 Palmitoyl-L-carnitine 466.3160 4.97 0.07 3.52E-03 0.0 C26H43NO6 Glycocholic Acid 357.2780 5.39 0.15 3.03E-03 2.0 C24H36O2 THA 355.2626 5.13 0.17 8.89E-03 1.0 C24H34O2 delta2-THA 170.0810 2.03 0.20 3.44E-04 0.0 C8H11NO3 Norepinephrine, 5-Hydroxydopamine 184.0968 2.69 0.23 2.70E-02 1.0 C9H13NO3 Normetanephrine; Methylnoradrenaline 212.1279 3.74 0.34 4.04E-03 1.0 C11H17NO3 Methoxamine 226.1433 4.20 0.29 8.10E-03 1.0 C12H19NO3 N-Methylmescaline; Terbutaline 161.0918 2.45 0.35 3.76E-03 2.0 C6H12N2O3 (2S)-4-Acetamido-2-aminobutanoic acid 224.0915 2.59 0.34 3.56E-02 1.0 C11H13NO4 Acetyl-L-tyrosine 146.1173 1.44 0.42 3.54E-02 1.0 C7H15NO2 DL-Aminoheptanoic acid

RT: 1.6471 - 1.8824 SM: 7B

1.65 1.70 1.75 1.80 1.85Time (min)

0

10

20

30

40

50

60

70

80

90

100

Rel

ativ

e Ab

unda

nce

1.8147132.1020

1.7255132.1020

NL:4.76E8m/z= 132.1012-132.1026 MS Urine_Junhua_1_10_4Isoleucine

Leucine FWHM 1.5 s;

7 Hz, Res 70K

Δppm 0.8

RT: 0.00 - 12.01 SM: 7B

0 2 4 6 8 10 12Time (min)

0

204060

80100

Rel

ativ

e Ab

unda

nce

02040

6080

100

Rel

ativ

e Ab

unda

nce

2.83185.0966

2.48205.0967

11.31149.01176.38

391.28235.88496.3373 8.35

732.5496

NL:1.94E8m/z= 185.0957-185.0975 MS ZDF_F1_2

NL:3.17E9m/z= 80.0000-1200.0000 MS ZDF_F1_2

IS=d5-hippuric acid

Base Peak

A

B

C

FIGURE 1. Untargeted metabolomics workflow

Comparing Differential Analysis Results from SIEVE 2.1 software and XCMS Online

The results from SIEVE 2.1 software are compared to XCMS Online, an academia-developed open-source software for metabolomics data analysis (Table 1) (2). As shown in Figure 5, the significant components (p-value <0.05, ratio > 2.0) identified by SIEVE 2.1 software and XCMS Online are similar, while the SIEVE software started with a much smaller (~20 less) number of components because of its capability to automatically remove solvent background, thus saving users time and labor in data review (Figure 6).

Metabolite Identification Guided by mzCloud

The mzCloud is the first cloud-based MSn spectral trees library built for small molecule structural elucidation (3). It contains a standard reference database, an elucidated compound database with putative structures, and a database of virtual spectra. Q Exactive HCD MS/MS spectra of interested metabolites can be searched against m/zCloud. The matching compounds are scored and the fragment ions are annotated accordingly (Figure 7).

FIGURE 5. PCA by SIEVE 2.1 software (A), Numbers in SIEVE 2.1 software (B), Numbers in XCMS Online (C)

206

2,075 components

403 p<0.05

Ratio > 2 212

440 „real peaks‟

38,268 features

3,848 p<0.05

A B

C

FIGURE 6. SIEVE 2.1 software output of component m/z 170.081, showing alignment across different samples (A), adducts grouping (B), peak integration (C), and trend intensity view (D)

15 features grouped to one component

A C

D B

FIGURE 7. Searching MS/MS spectrum of component m/z 184.0607 in mzCloud. The entry of 4-pyridoxic acid matches the spectrum with the highest score.

TABLE 2. Representative metabolites that are significantly changed

4-Pyridoxic acid_pos_HCD_QE #141 RT: 1.26 AV: 1 NL: 2.70E8T: FTMS + p ESI Full ms2 [email protected] [50.00-205.00]

50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200m/z

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

80

85

90

95

100

Rel

ativ

e Ab

unda

nce MS/MS of m/z 184.0606 from sample

mzCloud search and score

4-pyridoxic acid

166.0500

148.0393 138.0551 184.0607

FIGURE 8. SIEVE 2.1 software with KEGG pathway visualization

p-value: 3.76e-3 Ratio: 2.86 Composition: C6H12N2O3

ChemSpider ID: 389841 (2S)-4-Acetamido-2- aminobutanoic acid Branched amino acids are known hallmarks of Type II Diabetes [3]

XIC

KEGG is a trademark of Kyoto University, Kanehisa Lab. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries.

This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others.

Take an aliquot of 200 µL plasma

Add 600 µL (3) of cold 4 C organic solvent and vortex (3 min)

Spin down at 4 C, 4000 g (5 min)

Take supernatant and divide into 3 portions (each 200 µL), dry down (30–60 min)

Reconstitute in 400 µL (9:1) H2O:CH3OH containing IS, vortex, and sonicate (2 min)

Liquid Chromatography

UHPLC separation was implemented on a Thermo Scientific™ Dionex™ UltiMate™ 3000 HPG (high-pressure gradient) pump using Thermo Scientific™ Hypersil GOLD™ RP C18 column at 450 µL/min, column temperature at 55 C. LC solvents were 0.1% FA (A) and 0.1% FA in MeOH (B). Apply linear gradient from 0.5–50% B for 5.5 min, followed by increasing to 98% at 6 min, hold 98% B for 6 min, then decrease to 0.5% at 13 min, then equilibrate for another 2 min.

Mass Spectrometry

The Thermo Scientific™ Q Exactive™ mass spectrometer was operated under electrospray ionization (ESI) positive, negative, and polarity switching modes. Full scan (m/z 67–1000) used resolution 70,000 with automatic gain control (AGC) target of 1106 ions and a maximum ion injection time (IT) of 35 ms. Data-dependent MS/MS were acquired on a “Top5” data-dependent mode using the following parameters: resolution 17,500; AGC 1105 ions; maximum IT 80 ms; 2.0 amu isolation window; normalized collision energy 35% ± 50%; underfill ratio 1.0%; Apex trigger 2–4 s, and dynamic exclusion 6 s. Source ionization parameters were: spray voltage, 3.8 kV; capillary temperature, 300 C; and S-Lens level, 45.

Data Analysis

Differential analyses of the obese versus lean plasma were performed using Thermo Scientific™ SIEVE™ 2.1 software, which executes background subtraction, component detection, peak alignment, differential analysis (Figure 1). Statistical results, putative IDs, and pathways were generated through searching ChemSpider and KEGG™. Metabolites of interest were identified via MS/MS mass spectral database matching. The *.raw files were converted to mzXML format using ProteoWizard and also analyzed by XCMS Online [2] to compare the results.

3h

14h

15m

14h

15m

Solvent blank

with IS

QC Mixture with IS

Solvent blank NO IS

Pooled plasma with IS

Control sample with IS

#2

Treated sample with IS

#1

Control sample with IS

#1

Treated sample with IS

#2

b a

I

II

III

IV

1

2

3

4

5

6

7

9

Control sample with IS

#2

Treated sample with IS

#2

8

Control sample with IS

# n

Treated sample with IS

# n

3h

Step | Preconditioning LC under one polarity (+ or -) 1. Solvent blank spiked with IS, 5-10X | (check signal) 2. QC mixture with IS, 3-5X | (check LC system status) 3. Solvent blank with IS, 1X | (check carry over) • If carry over, wash followed by repeating step 2 & 3 | (clean) • If no carry over, proceed to next step

Step | Experiment I: Positive, R=70K 4. Solvent blank no IS, 1X | (for background subtraction) 5. Pooled plasma with IS, 3X | (check method and column) 6. Control sample with IS #1, 1X | 7. Treated sample with IS #1, 1X | 8. Run all biological replicates with a randomized order | (all)

Step | Experiment II: MS/MS Positive 9. Pooled plasma, top5 ddMS2 HCD | (dd-MS/MS)

Step | Experiment III: Negative, R=70K 4-8. Run all biological replicates as Experiment I | (all samples)

Step | Experiment IV: MS/MS Negative 10. Repeat as in Experiment II, step 9 | (dd-MS/MS)

FIGURE 4. Method validation, data quality control in metabolomics application

An example of SIEVE 2.1 software with KEGG pathway visualization demonstrated with component m/z 161.0918 is shown in Figure 8. All matching metabolites are labeled with black dots.

Table 2 shows some of the representative metabolites that are significantly changed. Up-regulated metabolites in fatty vs. lean rat plasma are shown in green while down-regulated metabolites are shown in blue (p-value <0.05, ratio >2).

Page 4: Untargeted Metabolomics Workfl ow Using …apps.thermoscientific.com/media/cmd/kl/resources/PN_ASMS13_T382... · Untargeted Metabolomics ... Workflow Using UHPLC/Quadrupole Orbitrap

4 Untargeted Metabolomics Work� ow Using UHPLC/Quadrupole Orbitrap Mass Spectrometer and SIEVE 2.1 Software

Untargeted Metabolomics Workflow Using UHPLC/Quadrupole Orbitrap Mass Spectrometer and SIEVE 2.1 Software Junhua Wang, David A. Peake, Mark Sanders, Michael Athanas, Yingying Huang Thermo Fisher Scientific, San Jose, CA

Conclusion An efficient and robust workflow for untargeted metabolomics is presented here. The reliable high-resolution, accurate-mass (HR/AM) performance of the Q Exactive LC-MS system eliminates the need for technical replicates on biological samples. The superior S/N in Orbitrap data allows efficient data reduction in SIEVE 2.1 software, resulting in much reduced data analysis effort. KEGG pathway visualization allows quick access to biological pathway mapping. The MSn spectral library mzCloud facilitates accurate compound identification.

References 1. Wang S.Y., Kuo, C.H., Tseng, Y. J. Anal. Chem., 2013, 85 (2), 1037–1046. 2. Scripps Center For Metabolomics Home Page. https://xcmsonline.scripps.edu/index.php 3. m/zCloud Home Page. https://www.mzcloud.org/home.aspx 4. Muoio, D.M.; Newgard, C.B. Nature Rev. Mol. Cell. Biol. 2008, 9, 193–205.

Overview Purpose: Demonstrate a generic, integrated workflow for untargeted metabolomics study using a UHPLC/benchtop Thermo Scientific™ Orbitrap™ mass spectrometer and informatics software.

Introduction Metabolomics is a rapidly growing field of post-genomic biology, aiming to comprehensively characterize the small molecules in biological systems. Nonbiological systematic biases from instrument calibration or the order of sample injection account for the most significant errors in LC/TOF-MS data [1]. Here we present a workflow using a UHPLC/benchtop quardrupole Orbitrap platform and automated data analysis software for untargeted metabolomic profiling of plasma samples for biomarker discovery from the Zucker diabetes fatty (ZDF) rat model. The optimal conditions for sample preparation, liquid chromatography (LC), column, mass spectrometry (MS), and data processing parameters are explored.

Methods Sample Preparation

Plasma samples were deproteinized with organic solvent. Four extraction solvent systems including methanol (MeOH), acetonitrile (ACN), acetone, and 1:1:1 of the above were tested in this work. Endogenous metabolites were reconstituted in methanol/water (1:9) containing isotopically labeled internal standard (IS), d5-hippuric acid for LC-MS analysis. Solvent blank, pooled QC, and biological samples were analyzed in a randomized injection order.

FIGURE 2. UHPLC/MS Experimental Design and Run Sequence. Left, schematic showing the vials and sample names. Right, detailed content and overall time for each step.

FIGURE 3. Metabolomics data analysis. SIEVE 2.1 software, Thermo Scientific™ TraceFinder™ 3.1 software, and mzCloud support untargeted and targeted metabolomics workflows.

Component Detection

Differential (Statistical)

Analysis Component

Identification Quantitation Pathway Annotation

Advanced Statistical Analysis

• SIEVE • XCMS • mzMine

• SIEVE

• SIEVE • mzCloud • METLIN

• SIEVE • TraceFinder

• SIEVE • KEGG • Cytoscape

• SimcaP • Partek • GeneData • R

Targeted Untargeted

Results Challenges in Untargeted Metabolomics Study

Complexity of biological samples

Diversity of small molecule metabolites: polar and non-polar analytes

Ionization requires both positive and negative ion

Wide range of concentrations

No universal method for chromatographic separation

Multiple sources of variability

Structure elucidation of unknowns is expensive: lack of synthetic standards

Preparing for the UHPLC-MS Data Acquisition

Prior to the real samples analysis, a solvent blank with internal standard (IS) is injected at the beginning to check the solvent and the LC-MS status. The injections of the real samples should be randomized in order to eliminate systematic bias. Triplicate injections of the pooled plasma are intermittently repeated throughout the whole batch to validate consistent performance of the overall system. The experimental design and run sequence are shown in Figure 2.

UHPLC provides fast chromatography for high throughput analysis, the typical peak width is 4–6 seconds. Our method can baseline resolve Isoleucine and Leucine, generating peak width 1.2 s at FWHM. Refer to Figure 4. For such narrow peaks, Q Exactive mass spectrometer operating at 70,000 resolution acquires >15 point across the peak without losing sensitivity (A).

The QC of each run was performed by monitoring the intensity of IS, d5-hippuric acid (B), and the overall base peak (C). When all samples were finished, the selected ion chromatograms can be quickly viewed with RawMeat 2.1 (a free app with SIEVE 2.1 software). By inspecting the RT and intensity of the IS, runs with large retention time drift and bad injections can be excluded from the following data analysis.

SIEVE 2.1 Software XCMS Online Ratio 5.0 Fold change 4.8

p-value 3.44e-4 p-value 3.4e-4 RT 2.03 min RT 2.03 min

IDs: Norepinephrine

Pyridoxine Oxidopiamine, etc.

IDs: Norepinephrine

Pyridoxine (VB6) Hydroxydopamin, etc.

Table 1. Statics of component m/z 170.081 from SIEVE 2.1 software and XCMS Online

[M+H]+ Time Fatty/Lean P-value Δppm Formula Name 854.5657 7.25 2.81 3.65E-02 4.0 C50H80NO8P PC44:10 838.6292 8.84 2.48 6.45E-03 3.0 C48H88NO8P PC 40:4 836.6133 8.55 4.54 2.22E-02 0.0 C46H88NO8P PC38:2 812.6095 8.58 2.79 3.08E-02 4.0 C46H86NO8P PC 38:3 788.6127 9.04 2.13 3.56E-02 3.0 C44H86NO8P PC36:1 778.5351 7.20 4.15 3.38E-03 0.0 C42H78NO8P PC34:3 768.5498 8.01 2.14 3.17E-04 4.0 C43H78NO8P PC35:4 764.5196 7.44 3.71 6.37E-04 3.0 C43H74NO8P PC35:6 754.5347 7.29 2.67 1.35E-02 3.0 C42H76NO8P PC34:4 834.5936 8.10 2.83 7.13E-04 0.0 C44H83NO13 LacCer(d18:1/14:0) 836.6041 8.01 2.42 1.10E-03 4.0 C44H85NO13 LacCer(d18:0/14:0) 813.6790 9.72 2.40 3.84E-02 4.0 C47H93N2O6P SM(d18:2/24:0) 675.5412 7.20 2.29 3.34E-02 3.0 C37H75N2O6P SM(d16:1/16:0) 546.3525 5.84 2.33 9.23E-03 4.0 C28H52NO7P LysoPC(20:3) 526.2910 5.69 2.42 1.67E-04 3.0 C27H44NO7P LysoPE(22:6) 153.0521 1.79 2.46 1.56E-03 3.0 C4H12N2S2 Cystamine 230.0954 0.80 2.53 2.80E-02 1.0 C9H15N3O2S Ergothioneine 245.0917 3.56 3.93 2.16E-03 1.0 C13H12N2O3 Haematopodin 172.1693 5.10 3.18 4.19E-02 1.0 C10H21NO decanamide 302.3043 5.48 2.28 2.50E-02 3.0 C18H39NO2 Sphinganine 204.1228 3.75 3.10 1.83E-02 1.0 C9H17NO4 Acetylcarnitine 232.1534 2.26 4.49 1.99E-03 2.0 C11H21NO4 Butyryl-L-carnitine 344.2786 5.31 2.13 1.57E-02 2.0 C19H37NO4 Lauroylcarnitine 400.3409 5.53 2.10 3.49E-04 2.0 C23H45NO4 Palmitoyl-L-carnitine 466.3160 4.97 0.07 3.52E-03 0.0 C26H43NO6 Glycocholic Acid 357.2780 5.39 0.15 3.03E-03 2.0 C24H36O2 THA 355.2626 5.13 0.17 8.89E-03 1.0 C24H34O2 delta2-THA 170.0810 2.03 0.20 3.44E-04 0.0 C8H11NO3 Norepinephrine, 5-Hydroxydopamine 184.0968 2.69 0.23 2.70E-02 1.0 C9H13NO3 Normetanephrine; Methylnoradrenaline 212.1279 3.74 0.34 4.04E-03 1.0 C11H17NO3 Methoxamine 226.1433 4.20 0.29 8.10E-03 1.0 C12H19NO3 N-Methylmescaline; Terbutaline 161.0918 2.45 0.35 3.76E-03 2.0 C6H12N2O3 (2S)-4-Acetamido-2-aminobutanoic acid 224.0915 2.59 0.34 3.56E-02 1.0 C11H13NO4 Acetyl-L-tyrosine 146.1173 1.44 0.42 3.54E-02 1.0 C7H15NO2 DL-Aminoheptanoic acid

RT: 1.6471 - 1.8824 SM: 7B

1.65 1.70 1.75 1.80 1.85Time (min)

0

10

20

30

40

50

60

70

80

90

100

Rel

ativ

e Ab

unda

nce

1.8147132.1020

1.7255132.1020

NL:4.76E8m/z= 132.1012-132.1026 MS Urine_Junhua_1_10_4Isoleucine

Leucine FWHM 1.5 s;

7 Hz, Res 70K

Δppm 0.8

RT: 0.00 - 12.01 SM: 7B

0 2 4 6 8 10 12Time (min)

0

204060

80100

Rel

ativ

e Ab

unda

nce

02040

6080

100

Rel

ativ

e Ab

unda

nce

2.83185.0966

2.48205.0967

11.31149.01176.38

391.28235.88496.3373 8.35

732.5496

NL:1.94E8m/z= 185.0957-185.0975 MS ZDF_F1_2

NL:3.17E9m/z= 80.0000-1200.0000 MS ZDF_F1_2

IS=d5-hippuric acid

Base Peak

A

B

C

FIGURE 1. Untargeted metabolomics workflow

Comparing Differential Analysis Results from SIEVE 2.1 software and XCMS Online

The results from SIEVE 2.1 software are compared to XCMS Online, an academia-developed open-source software for metabolomics data analysis (Table 1) (2). As shown in Figure 5, the significant components (p-value <0.05, ratio > 2.0) identified by SIEVE 2.1 software and XCMS Online are similar, while the SIEVE software started with a much smaller (~20 less) number of components because of its capability to automatically remove solvent background, thus saving users time and labor in data review (Figure 6).

Metabolite Identification Guided by mzCloud

The mzCloud is the first cloud-based MSn spectral trees library built for small molecule structural elucidation (3). It contains a standard reference database, an elucidated compound database with putative structures, and a database of virtual spectra. Q Exactive HCD MS/MS spectra of interested metabolites can be searched against m/zCloud. The matching compounds are scored and the fragment ions are annotated accordingly (Figure 7).

FIGURE 5. PCA by SIEVE 2.1 software (A), Numbers in SIEVE 2.1 software (B), Numbers in XCMS Online (C)

206

2,075 components

403 p<0.05

Ratio > 2 212

440 „real peaks‟

38,268 features

3,848 p<0.05

A B

C

FIGURE 6. SIEVE 2.1 software output of component m/z 170.081, showing alignment across different samples (A), adducts grouping (B), peak integration (C), and trend intensity view (D)

15 features grouped to one component

A C

D B

FIGURE 7. Searching MS/MS spectrum of component m/z 184.0607 in mzCloud. The entry of 4-pyridoxic acid matches the spectrum with the highest score.

TABLE 2. Representative metabolites that are significantly changed

4-Pyridoxic acid_pos_HCD_QE #141 RT: 1.26 AV: 1 NL: 2.70E8T: FTMS + p ESI Full ms2 [email protected] [50.00-205.00]

50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200m/z

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

80

85

90

95

100

Rel

ativ

e Ab

unda

nce MS/MS of m/z 184.0606 from sample

mzCloud search and score

4-pyridoxic acid

166.0500

148.0393 138.0551 184.0607

FIGURE 8. SIEVE 2.1 software with KEGG pathway visualization

p-value: 3.76e-3 Ratio: 2.86 Composition: C6H12N2O3

ChemSpider ID: 389841 (2S)-4-Acetamido-2- aminobutanoic acid Branched amino acids are known hallmarks of Type II Diabetes [3]

XIC

KEGG is a trademark of Kyoto University, Kanehisa Lab. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries.

This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others.

Take an aliquot of 200 µL plasma

Add 600 µL (3) of cold 4 C organic solvent and vortex (3 min)

Spin down at 4 C, 4000 g (5 min)

Take supernatant and divide into 3 portions (each 200 µL), dry down (30–60 min)

Reconstitute in 400 µL (9:1) H2O:CH3OH containing IS, vortex, and sonicate (2 min)

Liquid Chromatography

UHPLC separation was implemented on a Thermo Scientific™ Dionex™ UltiMate™ 3000 HPG (high-pressure gradient) pump using Thermo Scientific™ Hypersil GOLD™ RP C18 column at 450 µL/min, column temperature at 55 C. LC solvents were 0.1% FA (A) and 0.1% FA in MeOH (B). Apply linear gradient from 0.5–50% B for 5.5 min, followed by increasing to 98% at 6 min, hold 98% B for 6 min, then decrease to 0.5% at 13 min, then equilibrate for another 2 min.

Mass Spectrometry

The Thermo Scientific™ Q Exactive™ mass spectrometer was operated under electrospray ionization (ESI) positive, negative, and polarity switching modes. Full scan (m/z 67–1000) used resolution 70,000 with automatic gain control (AGC) target of 1106 ions and a maximum ion injection time (IT) of 35 ms. Data-dependent MS/MS were acquired on a “Top5” data-dependent mode using the following parameters: resolution 17,500; AGC 1105 ions; maximum IT 80 ms; 2.0 amu isolation window; normalized collision energy 35% ± 50%; underfill ratio 1.0%; Apex trigger 2–4 s, and dynamic exclusion 6 s. Source ionization parameters were: spray voltage, 3.8 kV; capillary temperature, 300 C; and S-Lens level, 45.

Data Analysis

Differential analyses of the obese versus lean plasma were performed using Thermo Scientific™ SIEVE™ 2.1 software, which executes background subtraction, component detection, peak alignment, differential analysis (Figure 1). Statistical results, putative IDs, and pathways were generated through searching ChemSpider and KEGG™. Metabolites of interest were identified via MS/MS mass spectral database matching. The *.raw files were converted to mzXML format using ProteoWizard and also analyzed by XCMS Online [2] to compare the results.

3h

14h

15m

14h

15m

Solvent blank

with IS

QC Mixture with IS

Solvent blank NO IS

Pooled plasma with IS

Control sample with IS

#2

Treated sample with IS

#1

Control sample with IS

#1

Treated sample with IS

#2

b a

I

II

III

IV

1

2

3

4

5

6

7

9

Control sample with IS

#2

Treated sample with IS

#2

8

Control sample with IS

# n

Treated sample with IS

# n

Step | Preconditioning LC under one polarity (+ or -) 1. Solvent blank spiked with IS, 5-10X | (check signal) 2. QC mixture with IS, 3-5X | (check LC system status) 3. Solvent blank with IS, 1X | (check carry over) • If carry over, wash followed by repeating step 2 & 3 | (clean) • If no carry over, proceed to next step

Step | Experiment I: Positive, R=70K 4. Solvent blank no IS, 1X | (for background subtraction) 5. Pooled plasma with IS, 3X | (check method and column) 6. Control sample with IS #1, 1X | 7. Treated sample with IS #1, 1X | 8. Run all biological replicates with a randomized order | (all)

Step | Experiment II: MS/MS Positive 9. Pooled plasma, top5 ddMS2 HCD | (dd-MS/MS)

Step | Experiment III: Negative, R=70K 4-8. Run all biological replicates as Experiment I | (all samples)

Step | Experiment IV: MS/MS Negative 10. Repeat as in Experiment II, step 9 | (dd-MS/MS)

FIGURE 4. Method validation, data quality control in metabolomics application

An example of SIEVE 2.1 software with KEGG pathway visualization demonstrated with component m/z 161.0918 is shown in Figure 8. All matching metabolites are labeled with black dots.

Table 2 shows some of the representative metabolites that are significantly changed. Up-regulated metabolites in fatty vs. lean rat plasma are shown in green while down-regulated metabolites are shown in blue (p-value <0.05, ratio >2).

Page 5: Untargeted Metabolomics Workfl ow Using …apps.thermoscientific.com/media/cmd/kl/resources/PN_ASMS13_T382... · Untargeted Metabolomics ... Workflow Using UHPLC/Quadrupole Orbitrap

5Thermo Scienti� c Poster Note • PN ASMS13_T382_JWang_E 07/13S

Untargeted Metabolomics Workflow Using UHPLC/Quadrupole Orbitrap Mass Spectrometer and SIEVE 2.1 Software Junhua Wang, David A. Peake, Mark Sanders, Michael Athanas, Yingying Huang Thermo Fisher Scientific, San Jose, CA

Conclusion An efficient and robust workflow for untargeted metabolomics is presented here. The reliable high-resolution, accurate-mass (HR/AM) performance of the Q Exactive LC-MS system eliminates the need for technical replicates on biological samples. The superior S/N in Orbitrap data allows efficient data reduction in SIEVE 2.1 software, resulting in much reduced data analysis effort. KEGG pathway visualization allows quick access to biological pathway mapping. The MSn spectral library mzCloud facilitates accurate compound identification.

References 1. Wang S.Y., Kuo, C.H., Tseng, Y. J. Anal. Chem., 2013, 85 (2), 1037–1046. 2. Scripps Center For Metabolomics Home Page. https://xcmsonline.scripps.edu/index.php 3. m/zCloud Home Page. https://www.mzcloud.org/home.aspx 4. Muoio, D.M.; Newgard, C.B. Nature Rev. Mol. Cell. Biol. 2008, 9, 193–205.

Overview Purpose: Demonstrate a generic, integrated workflow for untargeted metabolomics study using a UHPLC/benchtop Thermo Scientific™ Orbitrap™ mass spectrometer and informatics software.

Introduction Metabolomics is a rapidly growing field of post-genomic biology, aiming to comprehensively characterize the small molecules in biological systems. Nonbiological systematic biases from instrument calibration or the order of sample injection account for the most significant errors in LC/TOF-MS data [1]. Here we present a workflow using a UHPLC/benchtop quardrupole Orbitrap platform and automated data analysis software for untargeted metabolomic profiling of plasma samples for biomarker discovery from the Zucker diabetes fatty (ZDF) rat model. The optimal conditions for sample preparation, liquid chromatography (LC), column, mass spectrometry (MS), and data processing parameters are explored.

Methods Sample Preparation

Plasma samples were deproteinized with organic solvent. Four extraction solvent systems including methanol (MeOH), acetonitrile (ACN), acetone, and 1:1:1 of the above were tested in this work. Endogenous metabolites were reconstituted in methanol/water (1:9) containing isotopically labeled internal standard (IS), d5-hippuric acid for LC-MS analysis. Solvent blank, pooled QC, and biological samples were analyzed in a randomized injection order.

FIGURE 2. UHPLC/MS Experimental Design and Run Sequence. Left, schematic showing the vials and sample names. Right, detailed content and overall time for each step.

FIGURE 3. Metabolomics data analysis. SIEVE 2.1 software, Thermo Scientific™ TraceFinder™ 3.1 software, and mzCloud support untargeted and targeted metabolomics workflows.

Component Detection

Differential (Statistical)

Analysis Component

Identification Quantitation Pathway Annotation

Advanced Statistical Analysis

• SIEVE • XCMS • mzMine

• SIEVE

• SIEVE • mzCloud • METLIN

• SIEVE • TraceFinder

• SIEVE • KEGG • Cytoscape

• SimcaP • Partek • GeneData • R

Targeted Untargeted

Results Challenges in Untargeted Metabolomics Study

Complexity of biological samples

Diversity of small molecule metabolites: polar and non-polar analytes

Ionization requires both positive and negative ion

Wide range of concentrations

No universal method for chromatographic separation

Multiple sources of variability

Structure elucidation of unknowns is expensive: lack of synthetic standards

Preparing for the UHPLC-MS Data Acquisition

Prior to the real samples analysis, a solvent blank with internal standard (IS) is injected at the beginning to check the solvent and the LC-MS status. The injections of the real samples should be randomized in order to eliminate systematic bias. Triplicate injections of the pooled plasma are intermittently repeated throughout the whole batch to validate consistent performance of the overall system. The experimental design and run sequence are shown in Figure 2.

UHPLC provides fast chromatography for high throughput analysis, the typical peak width is 4–6 seconds. Our method can baseline resolve Isoleucine and Leucine, generating peak width 1.2 s at FWHM. Refer to Figure 4. For such narrow peaks, Q Exactive mass spectrometer operating at 70,000 resolution acquires >15 point across the peak without losing sensitivity (A).

The QC of each run was performed by monitoring the intensity of IS, d5-hippuric acid (B), and the overall base peak (C). When all samples were finished, the selected ion chromatograms can be quickly viewed with RawMeat 2.1 (a free app with SIEVE 2.1 software). By inspecting the RT and intensity of the IS, runs with large retention time drift and bad injections can be excluded from the following data analysis.

SIEVE 2.1 Software XCMS Online Ratio 5.0 Fold change 4.8

p-value 3.44e-4 p-value 3.4e-4 RT 2.03 min RT 2.03 min

IDs: Norepinephrine

Pyridoxine Oxidopiamine, etc.

IDs: Norepinephrine

Pyridoxine (VB6) Hydroxydopamin, etc.

Table 1. Statics of component m/z 170.081 from SIEVE 2.1 software and XCMS Online

[M+H]+ Time Fatty/Lean P-value Δppm Formula Name 854.5657 7.25 2.81 3.65E-02 4.0 C50H80NO8P PC44:10 838.6292 8.84 2.48 6.45E-03 3.0 C48H88NO8P PC 40:4 836.6133 8.55 4.54 2.22E-02 0.0 C46H88NO8P PC38:2 812.6095 8.58 2.79 3.08E-02 4.0 C46H86NO8P PC 38:3 788.6127 9.04 2.13 3.56E-02 3.0 C44H86NO8P PC36:1 778.5351 7.20 4.15 3.38E-03 0.0 C42H78NO8P PC34:3 768.5498 8.01 2.14 3.17E-04 4.0 C43H78NO8P PC35:4 764.5196 7.44 3.71 6.37E-04 3.0 C43H74NO8P PC35:6 754.5347 7.29 2.67 1.35E-02 3.0 C42H76NO8P PC34:4 834.5936 8.10 2.83 7.13E-04 0.0 C44H83NO13 LacCer(d18:1/14:0) 836.6041 8.01 2.42 1.10E-03 4.0 C44H85NO13 LacCer(d18:0/14:0) 813.6790 9.72 2.40 3.84E-02 4.0 C47H93N2O6P SM(d18:2/24:0) 675.5412 7.20 2.29 3.34E-02 3.0 C37H75N2O6P SM(d16:1/16:0) 546.3525 5.84 2.33 9.23E-03 4.0 C28H52NO7P LysoPC(20:3) 526.2910 5.69 2.42 1.67E-04 3.0 C27H44NO7P LysoPE(22:6) 153.0521 1.79 2.46 1.56E-03 3.0 C4H12N2S2 Cystamine 230.0954 0.80 2.53 2.80E-02 1.0 C9H15N3O2S Ergothioneine 245.0917 3.56 3.93 2.16E-03 1.0 C13H12N2O3 Haematopodin 172.1693 5.10 3.18 4.19E-02 1.0 C10H21NO decanamide 302.3043 5.48 2.28 2.50E-02 3.0 C18H39NO2 Sphinganine 204.1228 3.75 3.10 1.83E-02 1.0 C9H17NO4 Acetylcarnitine 232.1534 2.26 4.49 1.99E-03 2.0 C11H21NO4 Butyryl-L-carnitine 344.2786 5.31 2.13 1.57E-02 2.0 C19H37NO4 Lauroylcarnitine 400.3409 5.53 2.10 3.49E-04 2.0 C23H45NO4 Palmitoyl-L-carnitine 466.3160 4.97 0.07 3.52E-03 0.0 C26H43NO6 Glycocholic Acid 357.2780 5.39 0.15 3.03E-03 2.0 C24H36O2 THA 355.2626 5.13 0.17 8.89E-03 1.0 C24H34O2 delta2-THA 170.0810 2.03 0.20 3.44E-04 0.0 C8H11NO3 Norepinephrine, 5-Hydroxydopamine 184.0968 2.69 0.23 2.70E-02 1.0 C9H13NO3 Normetanephrine; Methylnoradrenaline 212.1279 3.74 0.34 4.04E-03 1.0 C11H17NO3 Methoxamine 226.1433 4.20 0.29 8.10E-03 1.0 C12H19NO3 N-Methylmescaline; Terbutaline 161.0918 2.45 0.35 3.76E-03 2.0 C6H12N2O3 (2S)-4-Acetamido-2-aminobutanoic acid 224.0915 2.59 0.34 3.56E-02 1.0 C11H13NO4 Acetyl-L-tyrosine 146.1173 1.44 0.42 3.54E-02 1.0 C7H15NO2 DL-Aminoheptanoic acid

RT: 1.6471 - 1.8824 SM: 7B

1.65 1.70 1.75 1.80 1.85Time (min)

0

10

20

30

40

50

60

70

80

90

100

Rel

ativ

e Ab

unda

nce

1.8147132.1020

1.7255132.1020

NL:4.76E8m/z= 132.1012-132.1026 MS Urine_Junhua_1_10_4Isoleucine

Leucine FWHM 1.5 s;

7 Hz, Res 70K

Δppm 0.8

RT: 0.00 - 12.01 SM: 7B

0 2 4 6 8 10 12Time (min)

0

204060

80100

Rel

ativ

e Ab

unda

nce

02040

6080

100

Rel

ativ

e Ab

unda

nce

2.83185.0966

2.48205.0967

11.31149.01176.38

391.28235.88496.3373 8.35

732.5496

NL:1.94E8m/z= 185.0957-185.0975 MS ZDF_F1_2

NL:3.17E9m/z= 80.0000-1200.0000 MS ZDF_F1_2

IS=d5-hippuric acid

Base Peak

A

B

C

FIGURE 1. Untargeted metabolomics workflow

Comparing Differential Analysis Results from SIEVE 2.1 software and XCMS Online

The results from SIEVE 2.1 software are compared to XCMS Online, an academia-developed open-source software for metabolomics data analysis (Table 1) (2). As shown in Figure 5, the significant components (p-value <0.05, ratio > 2.0) identified by SIEVE 2.1 software and XCMS Online are similar, while the SIEVE software started with a much smaller (~20 less) number of components because of its capability to automatically remove solvent background, thus saving users time and labor in data review (Figure 6).

Metabolite Identification Guided by mzCloud

The mzCloud is the first cloud-based MSn spectral trees library built for small molecule structural elucidation (3). It contains a standard reference database, an elucidated compound database with putative structures, and a database of virtual spectra. Q Exactive HCD MS/MS spectra of interested metabolites can be searched against m/zCloud. The matching compounds are scored and the fragment ions are annotated accordingly (Figure 7).

FIGURE 5. PCA by SIEVE 2.1 software (A), Numbers in SIEVE 2.1 software (B), Numbers in XCMS Online (C)

206

2,075 components

403 p<0.05

Ratio > 2 212

440 „real peaks‟

38,268 features

3,848 p<0.05

A B

C

FIGURE 6. SIEVE 2.1 software output of component m/z 170.081, showing alignment across different samples (A), adducts grouping (B), peak integration (C), and trend intensity view (D)

15 features grouped to one component

A C

D B

FIGURE 7. Searching MS/MS spectrum of component m/z 184.0607 in mzCloud. The entry of 4-pyridoxic acid matches the spectrum with the highest score.

TABLE 2. Representative metabolites that are significantly changed

4-Pyridoxic acid_pos_HCD_QE #141 RT: 1.26 AV: 1 NL: 2.70E8T: FTMS + p ESI Full ms2 [email protected] [50.00-205.00]

50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200m/z

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

80

85

90

95

100

Rel

ativ

e Ab

unda

nce 60

65

70

MS/MS of m/z 184.0606 from sample mzCloud search and score

4-pyridoxic acid

166.0500

148.0393 138.0551 184.0607

FIGURE 8. SIEVE 2.1 software with KEGG pathway visualization

p-value: 3.76e-3 Ratio: 2.86 Composition: C6H12N2O3

ChemSpider ID: 389841 (2S)-4-Acetamido-2- aminobutanoic acid Branched amino acids are known hallmarks of Type II Diabetes [3]

XIC

KEGG is a trademark of Kyoto University, Kanehisa Lab. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries.

This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others.

Take an aliquot of 200 µL plasma

Add 600 µL (3) of cold 4 C organic solvent and vortex (3 min)

Spin down at 4 C, 4000 g (5 min)

Take supernatant and divide into 3 portions (each 200 µL), dry down (30–60 min)

Reconstitute in 400 µL (9:1) H2O:CH3OH containing IS, vortex, and sonicate (2 min)

Liquid Chromatography

UHPLC separation was implemented on a Thermo Scientific™ Dionex™ UltiMate™ 3000 HPG (high-pressure gradient) pump using Thermo Scientific™ Hypersil GOLD™ RP C18 column at 450 µL/min, column temperature at 55 C. LC solvents were 0.1% FA (A) and 0.1% FA in MeOH (B). Apply linear gradient from 0.5–50% B for 5.5 min, followed by increasing to 98% at 6 min, hold 98% B for 6 min, then decrease to 0.5% at 13 min, then equilibrate for another 2 min.

Mass Spectrometry

The Thermo Scientific™ Q Exactive™ mass spectrometer was operated under electrospray ionization (ESI) positive, negative, and polarity switching modes. Full scan (m/z 67–1000) used resolution 70,000 with automatic gain control (AGC) target of 1106 ions and a maximum ion injection time (IT) of 35 ms. Data-dependent MS/MS were acquired on a “Top5” data-dependent mode using the following parameters: resolution 17,500; AGC 1105 ions; maximum IT 80 ms; 2.0 amu isolation window; normalized collision energy 35% ± 50%; underfill ratio 1.0%; Apex trigger 2–4 s, and dynamic exclusion 6 s. Source ionization parameters were: spray voltage, 3.8 kV; capillary temperature, 300 C; and S-Lens level, 45.

Data Analysis

Differential analyses of the obese versus lean plasma were performed using Thermo Scientific™ SIEVE™ 2.1 software, which executes background subtraction, component detection, peak alignment, differential analysis (Figure 1). Statistical results, putative IDs, and pathways were generated through searching ChemSpider and KEGG™. Metabolites of interest were identified via MS/MS mass spectral database matching. The *.raw files were converted to mzXML format using ProteoWizard and also analyzed by XCMS Online [2] to compare the results.

3h

14h

15m

14h

15m

Solvent blank

with IS

QC Mixture with IS

Solvent blank NO IS

Pooled plasma with IS

Control sample with IS

#2

Treated sample with IS

#1

Control sample with IS

#1

Treated sample with IS

#2

b a

I

II

III

IV

1

2

3

4

5

6

7

9

Control sample with IS

#2

Treated sample with IS

#2

8

Control sample with IS

# n

Treated sample with IS

# n

Step | Preconditioning LC under one polarity (+ or -) 1. Solvent blank spiked with IS, 5-10X | (check signal) 2. QC mixture with IS, 3-5X | (check LC system status) 3. Solvent blank with IS, 1X | (check carry over) • If carry over, wash followed by repeating step 2 & 3 | (clean) • If no carry over, proceed to next step

Step | Experiment I: Positive, R=70K 4. Solvent blank no IS, 1X | (for background subtraction) 5. Pooled plasma with IS, 3X | (check method and column) 6. Control sample with IS #1, 1X | 7. Treated sample with IS #1, 1X | 8. Run all biological replicates with a randomized order | (all)

Step | Experiment II: MS/MS Positive 9. Pooled plasma, top5 ddMS2 HCD | (dd-MS/MS)

Step | Experiment III: Negative, R=70K 4-8. Run all biological replicates as Experiment I | (all samples)

Step | Experiment IV: MS/MS Negative 10. Repeat as in Experiment II, step 9 | (dd-MS/MS)

FIGURE 4. Method validation, data quality control in metabolomics application

An example of SIEVE 2.1 software with KEGG pathway visualization demonstrated with component m/z 161.0918 is shown in Figure 8. All matching metabolites are labeled with black dots.

Table 2 shows some of the representative metabolites that are significantly changed. Up-regulated metabolites in fatty vs. lean rat plasma are shown in green while down-regulated metabolites are shown in blue (p-value <0.05, ratio >2).

Page 6: Untargeted Metabolomics Workfl ow Using …apps.thermoscientific.com/media/cmd/kl/resources/PN_ASMS13_T382... · Untargeted Metabolomics ... Workflow Using UHPLC/Quadrupole Orbitrap

6 Untargeted Metabolomics Work� ow Using UHPLC/Quadrupole Orbitrap Mass Spectrometer and SIEVE 2.1 Software

Untargeted Metabolomics Workflow Using UHPLC/Quadrupole Orbitrap Mass Spectrometer and SIEVE 2.1 Software Junhua Wang, David A. Peake, Mark Sanders, Michael Athanas, Yingying Huang Thermo Fisher Scientific, San Jose, CA

Conclusion An efficient and robust workflow for untargeted metabolomics is presented here. The reliable high-resolution, accurate-mass (HR/AM) performance of the Q Exactive LC-MS system eliminates the need for technical replicates on biological samples. The superior S/N in Orbitrap data allows efficient data reduction in SIEVE 2.1 software, resulting in much reduced data analysis effort. KEGG pathway visualization allows quick access to biological pathway mapping. The MSn spectral library mzCloud facilitates accurate compound identification.

References 1. Wang S.Y., Kuo, C.H., Tseng, Y. J. Anal. Chem., 2013, 85 (2), 1037–1046. 2. Scripps Center For Metabolomics Home Page. https://xcmsonline.scripps.edu/index.php 3. m/zCloud Home Page. https://www.mzcloud.org/home.aspx 4. Muoio, D.M.; Newgard, C.B. Nature Rev. Mol. Cell. Biol. 2008, 9, 193–205.

Overview Purpose: Demonstrate a generic, integrated workflow for untargeted metabolomics study using a UHPLC/benchtop Thermo Scientific™ Orbitrap™ mass spectrometer and informatics software.

Introduction Metabolomics is a rapidly growing field of post-genomic biology, aiming to comprehensively characterize the small molecules in biological systems. Nonbiological systematic biases from instrument calibration or the order of sample injection account for the most significant errors in LC/TOF-MS data [1]. Here we present a workflow using a UHPLC/benchtop quardrupole Orbitrap platform and automated data analysis software for untargeted metabolomic profiling of plasma samples for biomarker discovery from the Zucker diabetes fatty (ZDF) rat model. The optimal conditions for sample preparation, liquid chromatography (LC), column, mass spectrometry (MS), and data processing parameters are explored.

Methods Sample Preparation

Plasma samples were deproteinized with organic solvent. Four extraction solvent systems including methanol (MeOH), acetonitrile (ACN), acetone, and 1:1:1 of the above were tested in this work. Endogenous metabolites were reconstituted in methanol/water (1:9) containing isotopically labeled internal standard (IS), d5-hippuric acid for LC-MS analysis. Solvent blank, pooled QC, and biological samples were analyzed in a randomized injection order.

FIGURE 2. UHPLC/MS Experimental Design and Run Sequence. Left, schematic showing the vials and sample names. Right, detailed content and overall time for each step.

FIGURE 3. Metabolomics data analysis. SIEVE 2.1 software, Thermo Scientific™ TraceFinder™ 3.1 software, and mzCloud support untargeted and targeted metabolomics workflows.

Component Detection

Differential (Statistical)

Analysis Component

Identification Quantitation Pathway Annotation

Advanced Statistical Analysis

• SIEVE • XCMS • mzMine

• SIEVE

• SIEVE • mzCloud • METLIN

• SIEVE • TraceFinder

• SIEVE • KEGG • Cytoscape

• SimcaP • Partek • GeneData • R

Targeted Untargeted

Results Challenges in Untargeted Metabolomics Study

Complexity of biological samples

Diversity of small molecule metabolites: polar and non-polar analytes

Ionization requires both positive and negative ion

Wide range of concentrations

No universal method for chromatographic separation

Multiple sources of variability

Structure elucidation of unknowns is expensive: lack of synthetic standards

Preparing for the UHPLC-MS Data Acquisition

Prior to the real samples analysis, a solvent blank with internal standard (IS) is injected at the beginning to check the solvent and the LC-MS status. The injections of the real samples should be randomized in order to eliminate systematic bias. Triplicate injections of the pooled plasma are intermittently repeated throughout the whole batch to validate consistent performance of the overall system. The experimental design and run sequence are shown in Figure 2.

UHPLC provides fast chromatography for high throughput analysis, the typical peak width is 4–6 seconds. Our method can baseline resolve Isoleucine and Leucine, generating peak width 1.2 s at FWHM. Refer to Figure 4. For such narrow peaks, Q Exactive mass spectrometer operating at 70,000 resolution acquires >15 point across the peak without losing sensitivity (A).

The QC of each run was performed by monitoring the intensity of IS, d5-hippuric acid (B), and the overall base peak (C). When all samples were finished, the selected ion chromatograms can be quickly viewed with RawMeat 2.1 (a free app with SIEVE 2.1 software). By inspecting the RT and intensity of the IS, runs with large retention time drift and bad injections can be excluded from the following data analysis.

SIEVE 2.1 Software XCMS Online Ratio 5.0 Fold change 4.8

p-value 3.44e-4 p-value 3.4e-4 RT 2.03 min RT 2.03 min

IDs: Norepinephrine

Pyridoxine Oxidopiamine, etc.

IDs: Norepinephrine

Pyridoxine (VB6) Hydroxydopamin, etc.

Table 1. Statics of component m/z 170.081 from SIEVE 2.1 software and XCMS Online

[M+H]+ Time Fatty/Lean P-value Δppm Formula Name 854.5657 7.25 2.81 3.65E-02 4.0 C50H80NO8P PC44:10 838.6292 8.84 2.48 6.45E-03 3.0 C48H88NO8P PC 40:4 836.6133 8.55 4.54 2.22E-02 0.0 C46H88NO8P PC38:2 812.6095 8.58 2.79 3.08E-02 4.0 C46H86NO8P PC 38:3 788.6127 9.04 2.13 3.56E-02 3.0 C44H86NO8P PC36:1 778.5351 7.20 4.15 3.38E-03 0.0 C42H78NO8P PC34:3 768.5498 8.01 2.14 3.17E-04 4.0 C43H78NO8P PC35:4 764.5196 7.44 3.71 6.37E-04 3.0 C43H74NO8P PC35:6 754.5347 7.29 2.67 1.35E-02 3.0 C42H76NO8P PC34:4 834.5936 8.10 2.83 7.13E-04 0.0 C44H83NO13 LacCer(d18:1/14:0) 836.6041 8.01 2.42 1.10E-03 4.0 C44H85NO13 LacCer(d18:0/14:0) 813.6790 9.72 2.40 3.84E-02 4.0 C47H93N2O6P SM(d18:2/24:0) 675.5412 7.20 2.29 3.34E-02 3.0 C37H75N2O6P SM(d16:1/16:0) 546.3525 5.84 2.33 9.23E-03 4.0 C28H52NO7P LysoPC(20:3) 526.2910 5.69 2.42 1.67E-04 3.0 C27H44NO7P LysoPE(22:6) 153.0521 1.79 2.46 1.56E-03 3.0 C4H12N2S2 Cystamine 230.0954 0.80 2.53 2.80E-02 1.0 C9H15N3O2S Ergothioneine 245.0917 3.56 3.93 2.16E-03 1.0 C13H12N2O3 Haematopodin 172.1693 5.10 3.18 4.19E-02 1.0 C10H21NO decanamide 302.3043 5.48 2.28 2.50E-02 3.0 C18H39NO2 Sphinganine 204.1228 3.75 3.10 1.83E-02 1.0 C9H17NO4 Acetylcarnitine 232.1534 2.26 4.49 1.99E-03 2.0 C11H21NO4 Butyryl-L-carnitine 344.2786 5.31 2.13 1.57E-02 2.0 C19H37NO4 Lauroylcarnitine 400.3409 5.53 2.10 3.49E-04 2.0 C23H45NO4 Palmitoyl-L-carnitine 466.3160 4.97 0.07 3.52E-03 0.0 C26H43NO6 Glycocholic Acid 357.2780 5.39 0.15 3.03E-03 2.0 C24H36O2 THA 355.2626 5.13 0.17 8.89E-03 1.0 C24H34O2 delta2-THA 170.0810 2.03 0.20 3.44E-04 0.0 C8H11NO3 Norepinephrine, 5-Hydroxydopamine 184.0968 2.69 0.23 2.70E-02 1.0 C9H13NO3 Normetanephrine; Methylnoradrenaline 212.1279 3.74 0.34 4.04E-03 1.0 C11H17NO3 Methoxamine 226.1433 4.20 0.29 8.10E-03 1.0 C12H19NO3 N-Methylmescaline; Terbutaline 161.0918 2.45 0.35 3.76E-03 2.0 C6H12N2O3 (2S)-4-Acetamido-2-aminobutanoic acid 224.0915 2.59 0.34 3.56E-02 1.0 C11H13NO4 Acetyl-L-tyrosine 146.1173 1.44 0.42 3.54E-02 1.0 C7H15NO2 DL-Aminoheptanoic acid

RT: 1.6471 - 1.8824 SM: 7B

1.65 1.70 1.75 1.80 1.85Time (min)

0

10

20

30

40

50

60

70

80

90

100

Rel

ativ

e Ab

unda

nce

1.8147132.1020

1.7255132.1020

NL:4.76E8m/z= 132.1012-132.1026 MS Urine_Junhua_1_10_4Isoleucine

Leucine FWHM 1.5 s;

7 Hz, Res 70K

Δppm 0.8

RT: 0.00 - 12.01 SM: 7B

0 2 4 6 8 10 12Time (min)

0

20406080

100

Rel

ativ

e Ab

unda

nce

02040

6080

100

Rel

ativ

e Ab

unda

nce

2.83185.0966

2.48205.0967

11.31149.01176.38

391.28235.88496.3373 8.35

732.5496

NL:1.94E8m/z= 185.0957-185.0975 MS ZDF_F1_2

NL:3.17E9m/z= 80.0000-1200.0000 MS ZDF_F1_2

IS=d5-hippuric acid

Base Peak

A

B

C

FIGURE 1. Untargeted metabolomics workflow

Comparing Differential Analysis Results from SIEVE 2.1 software and XCMS Online

The results from SIEVE 2.1 software are compared to XCMS Online, an academia-developed open-source software for metabolomics data analysis (Table 1) (2). As shown in Figure 5, the significant components (p-value <0.05, ratio > 2.0) identified by SIEVE 2.1 software and XCMS Online are similar, while the SIEVE software started with a much smaller (~20 less) number of components because of its capability to automatically remove solvent background, thus saving users time and labor in data review (Figure 6).

Metabolite Identification Guided by mzCloud

The mzCloud is the first cloud-based MSn spectral trees library built for small molecule structural elucidation (3). It contains a standard reference database, an elucidated compound database with putative structures, and a database of virtual spectra. Q Exactive HCD MS/MS spectra of interested metabolites can be searched against m/zCloud. The matching compounds are scored and the fragment ions are annotated accordingly (Figure 7).

FIGURE 5. PCA by SIEVE 2.1 software (A), Numbers in SIEVE 2.1 software (B), Numbers in XCMS Online (C)

206

2,075 components

403 p<0.05

Ratio > 2 212

440 „real peaks‟

38,268 features

3,848 p<0.05

A B

C

FIGURE 6. SIEVE 2.1 software output of component m/z 170.081, showing alignment across different samples (A), adducts grouping (B), peak integration (C), and trend intensity view (D)

15 features grouped to one component

A C

D B

FIGURE 7. Searching MS/MS spectrum of component m/z 184.0607 in mzCloud. The entry of 4-pyridoxic acid matches the spectrum with the highest score.

TABLE 2. Representative metabolites that are significantly changed

4-Pyridoxic acid_pos_HCD_QE #141 RT: 1.26 AV: 1 NL: 2.70E8T: FTMS + p ESI Full ms2 [email protected] [50.00-205.00]

50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200m/z

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

80

85

90

95

100

Rel

ativ

e Ab

unda

nce MS/MS of m/z 184.0606 from sample

mzCloud search and score

4-pyridoxic acid

166.0500

148.0393 138.0551 184.0607

FIGURE 8. SIEVE 2.1 software with KEGG pathway visualization

p-value: 3.76e-3 Ratio: 2.86 Composition: C6H12N2O3

ChemSpider ID: 389841 (2S)-4-Acetamido-2- aminobutanoic acid Branched amino acids are known hallmarks of Type II Diabetes [3]

XIC

KEGG is a trademark of Kyoto University, Kanehisa Lab. All other trademarks are the property of Thermo Fisher Scientific and its subsidiaries.

This information is not intended to encourage use of these products in any manners that might infringe the intellectual property rights of others.

Take an aliquot of 200 µL plasma

Add 600 µL (3) of cold 4 C organic solvent and vortex (3 min)

Spin down at 4 C, 4000 g (5 min)

Take supernatant and divide into 3 portions (each 200 µL), dry down (30–60 min)

Reconstitute in 400 µL (9:1) H2O:CH3OH containing IS, vortex, and sonicate (2 min)

Liquid Chromatography

UHPLC separation was implemented on a Thermo Scientific™ Dionex™ UltiMate™ 3000 HPG (high-pressure gradient) pump using Thermo Scientific™ Hypersil GOLD™ RP C18 column at 450 µL/min, column temperature at 55 C. LC solvents were 0.1% FA (A) and 0.1% FA in MeOH (B). Apply linear gradient from 0.5–50% B for 5.5 min, followed by increasing to 98% at 6 min, hold 98% B for 6 min, then decrease to 0.5% at 13 min, then equilibrate for another 2 min.

Mass Spectrometry

The Thermo Scientific™ Q Exactive™ mass spectrometer was operated under electrospray ionization (ESI) positive, negative, and polarity switching modes. Full scan (m/z 67–1000) used resolution 70,000 with automatic gain control (AGC) target of 1106 ions and a maximum ion injection time (IT) of 35 ms. Data-dependent MS/MS were acquired on a “Top5” data-dependent mode using the following parameters: resolution 17,500; AGC 1105 ions; maximum IT 80 ms; 2.0 amu isolation window; normalized collision energy 35% ± 50%; underfill ratio 1.0%; Apex trigger 2–4 s, and dynamic exclusion 6 s. Source ionization parameters were: spray voltage, 3.8 kV; capillary temperature, 300 C; and S-Lens level, 45.

Data Analysis

Differential analyses of the obese versus lean plasma were performed using Thermo Scientific™ SIEVE™ 2.1 software, which executes background subtraction, component detection, peak alignment, differential analysis (Figure 1). Statistical results, putative IDs, and pathways were generated through searching ChemSpider and KEGG™. Metabolites of interest were identified via MS/MS mass spectral database matching. The *.raw files were converted to mzXML format using ProteoWizard and also analyzed by XCMS Online [2] to compare the results.

3h

14h

15m

14h

15m

Solvent blank

with IS

QC Mixture with IS

Solvent blank NO IS

Pooled plasma with IS

Control sample with IS

#2

Treated sample with IS

#1

Control sample with IS

#1

Treated sample with IS

#2

b a

I

II

III

IV

1

2

3

4

5

6

7

9

Control sample with IS

#2

Treated sample with IS

#2

8

Control sample with IS

# n

Treated sample with IS

# n

Step | Preconditioning LC under one polarity (+ or -) 1. Solvent blank spiked with IS, 5-10X | (check signal) 2. QC mixture with IS, 3-5X | (check LC system status) 3. Solvent blank with IS, 1X | (check carry over) • If carry over, wash followed by repeating step 2 & 3 | (clean) • If no carry over, proceed to next step

Step | Experiment I: Positive, R=70K 4. Solvent blank no IS, 1X | (for background subtraction) 5. Pooled plasma with IS, 3X | (check method and column) 6. Control sample with IS #1, 1X | 7. Treated sample with IS #1, 1X | 8. Run all biological replicates with a randomized order | (all)

Step | Experiment II: MS/MS Positive 9. Pooled plasma, top5 ddMS2 HCD | (dd-MS/MS)

Step | Experiment III: Negative, R=70K 4-8. Run all biological replicates as Experiment I | (all samples)

Step | Experiment IV: MS/MS Negative 10. Repeat as in Experiment II, step 9 | (dd-MS/MS)

FIGURE 4. Method validation, data quality control in metabolomics application

An example of SIEVE 2.1 software with KEGG pathway visualization demonstrated with component m/z 161.0918 is shown in Figure 8. All matching metabolites are labeled with black dots.

Table 2 shows some of the representative metabolites that are significantly changed. Up-regulated metabolites in fatty vs. lean rat plasma are shown in green while down-regulated metabolites are shown in blue (p-value <0.05, ratio >2).

Page 7: Untargeted Metabolomics Workfl ow Using …apps.thermoscientific.com/media/cmd/kl/resources/PN_ASMS13_T382... · Untargeted Metabolomics ... Workflow Using UHPLC/Quadrupole Orbitrap

ASMS13_T382_JWang_E 07/13S

Africa-Other +27 11 570 1840Australia +61 3 9757 4300Austria +43 1 333 50 34 0Belgium +32 53 73 42 41Canada +1 800 530 8447China +86 10 8419 3588Denmark +45 70 23 62 60

Europe-Other +43 1 333 50 34 0Finland/Norway/Sweden +46 8 556 468 00France +33 1 60 92 48 00Germany +49 6103 408 1014India +91 22 6742 9434Italy +39 02 950 591

Japan +81 45 453 9100Latin America +1 561 688 8700Middle East +43 1 333 50 34 0Netherlands +31 76 579 55 55New Zealand +64 9 980 6700Russia/CIS +43 1 333 50 34 0South Africa +27 11 570 1840

Spain +34 914 845 965Switzerland +41 61 716 77 00UK +44 1442 233555USA +1 800 532 4752

www.thermoscientific.com©2013 Thermo Fisher Scientific Inc. All rights reserved. ISO is a trademark of the International Standards Organization. KEGG is a trademark of Kyoto University, Kanehisa Lab. All other trademarks are the property of Thermo Fisher Scientific, Inc. and its subsidiaries. Specifications, terms and pricing are subject to change. Not all products are available in all countries. Please consult your local sales representative for details.

Thermo Fisher Scientific, San Jose, CA USA is ISO 9001:2008 Certified.