aiko barsch metabolomics thessaloniki how to link hram...

METABOSCAPE – A METABOLITE PROFILING PIPELINE DRIVEN BY AUTOMATIC COMPOUND IDENTIFICATION

OR

HOW TO LINK HRAM QTOF PLANT METABOLOMICS DATA TO BIOLOGY

Aiko Barsch, Bruker Daltonics, Bremen, Germany

1

Outline

• If you want to automate data processingworkflows in Metabolomics you have to be ableto rely on the raw data.

• Introducing MetaboScape. Turning MS data into knowledge!

• An example studying the “war” between plants and insects!

Automatic data processing workflows in Metabolomics rely on the raw data quality

• The metabolome is an extremely complex mixture of chemically diverse small molecules covering a large dynamic range of concentrations.

• Robustness and Sensitivity and Dynamic range and High resolution, mass accuracy and isotopic fidelity in one scan mode at LC speed is absolute must in metabolomics.

• So, lets have a look at the capabilities of a QTOF instrument. Can this technology fulfill these needs?

Sample Injection #1

Sample Injection #100

selected BPCs from sequence of human urine sample injected 100 times

3 4 5 6 7 Time [min]0.0

0.2

0.4

0.6

0.8

6x10

Intens.

Phenylalanine peak

Peak Area

Average 730594

StDEV 19720

CV 2.70

Peak area for phenylalanine deviates less than 3% (n=100 injections).

Instrument robustness: >100 injections of complex human Urine sample

3 4 5 6 7 Time [min]0.0

0.2

0.4

0.6

0.8

6x10

Intens.

166.0863

1+

167.08971+

+MS, 2.89-2.96min #834-855

0.0

0.5

1.0

1.5

2.0

5x10

Intens.

166 167 168 169 m/z

Example - Phenylalanine: SmartFormula provides the correct molecular formula based on accurate mass and isotopic pattern fit: C9H12O2

-0.2 ppm mass accuracy

25882 Resolution

SmartFormula makes use of the perfect fit between measured and simulated spectrum

+MS, 2.88-2.97min #830-857

166.08631+

167.0894

1+

C₉H₁₂NO₂, 166.0863

0.0

0.5

1.0

1.5

5x10

Intens.

166 167 168 169 170 m/z

+MS, 2.88-2.97min #830-857

167.0894

1+

168.09171+

C₉H₁₂NO₂, 166.0863

0.0

0.5

1.0

1.5

2.0

4x10

Intens.

167.00 167.50 168.00 m/zzoom

Measured

Simulated

expressed in a low mSigma valueMass accuracy & isotopic fidelity

= confidence in ID

1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Time [min]

0

1

2

3

4

5x10

Intens.

Target compound spiked into the human urine sample can be detected by creating a high resolution EIC trace for the target mass 169.0495m/z.

EIC for vanillicacid

BPC

Reliable ID of target compoundsvanillic acid spiked into the urine sample

Detection by high resolution EIC

Basis for hrEIC=

mass stability across entire dynamic

concentration range

Human urine sample spiked with vanillic acid� Impact II� 4296 MS/MS spectra

acquired at this sample complexity level

200

400

600

800

m/z

2 4 6 8 10 Time [min]

200

400

600

800

m/z

2 4 6 8 10 Time [min]

Time

m/z

Survey View

Impact II - High dynamic range in MS/MS: up to 50Hz instrument scan speed combined with easy to use Instant Expertise software

All precursor ions fragmented in “one shot” acquisition.

MS/MS Spectrum acquired

200

400

600

800

m/z

2 4 6 8 10 Time [min]

120

140

160

180

200

m/z

4.0 4.2 4.4 4.6 4.8 Time [min]

zoom

105.0332

1+

125.05981+

180.0653

1+

169.0494

1+

1. +MS, 4.40-4.42min #1787-1795

39.0229

1+

65.03831+

81.03381+

93.0331

1+

111.0439

1+

125.05971+ 151.0388

1+

1. +MS2(169.0494), 20.0-50.0eV, 4.41-4.43min #1788-17960.0

0.5

1.0

1.5

2.0

2.5

4x10

Intens.

0.00

0.25

0.50

0.75

1.00

1.25

1.504x10

20 40 60 80 100 120 140 160 180 m/z

The target mass 169.0495m/z of the vanillic acid precursor was fragmented successfully.

Instant Expertise selected vanillic acid spiked into the sample for MS/MS fragmentation

0

1

2

3

4

5x10

Intens.

0

1

2

3

4

5

5x10

3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6Time [min]

0

2

4

6

4x10

Intens.

0

1

2

3

4x10

4.30 4.40 4.50 4.60 Time [min]10x less spike level for vanillic acid

EIC for Vanillic acid

BPC

438.2341

169.04951+

1. +MS, 4.45-4.50min #1796-1812

0.0

0.5

1.0

1.5

2.0

4x10

Intens.

100 200 300 400 500 600 700 800 900 m/z

Precursor was picked for fragmentation by the Instant Expertise method!

10x lower concentration is still fragmented-> high dynamic range in MS & MS/MS data acquisition

Confidence in the data.Even low abundant precursor ions are fragmented in complex samples.

105.0336

131.9741158.9623

180.0656

202.0480

169.04951+

1. +MS, 4.45-4.50min #1796-1812

111.04381+

125.0591

151.0391

1+

1. +MS2(169.0495), 20.0-50.0eV, 4.46-4.50min #1797-1813

111.0440

125.0597

136.0155

151.0394

169.0495

O

OH

O

OH

1. HMDB Library: Vanillic acid - HMDB00484, oTOF ESI +MS2(169.049), 20.0eV (P: 789, F: 833, R: 790, M: 1000)

0

1000

2000

3000

4000

5000

Intens.

0

250

500

750

1000

1250

0

100

200

300

100 120 140 160 180 200 m/z

0

1

2

3

4

5x10

Intens.

0

1

2

3

4

5

5x10

3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6Time [min]

10x less spiked level for vanillic acid

EIC for Vanillic acid

BPC

MS

MS/MS

Library MS/MS

MS/MS spectrum enables identification via Bruker HMDB Metabolite Library query

MS/MS data acquired permits confident ID via spectral library query.In this case via the Bruker HMDB Metabolite Library.

Acquired in collaboration with the “HMDB team“ at the University of Alberta. Lead by Profs. Liang Li and David Wishart.

• ~800 pure reference standards , measured on an impact HD QTOF• up to 5 different collision energies, each mass spectrum manually curated

Perfect matching between Library and experimental spectrum.

Germany

Canada

Deeper insights with the Bruker HMDB Metabolite Library“Human Metabolome Data Base” (HMDB)

April 18, 2016 14

Targeted Metabolomics:

• “Think” hrEICs – if you know what you are looking for

Non-targeted Metabolomics:

• “Think” extract all Features first

MetaboScape

Non- targeted & targeted Metabolomics

Both can be addressed using one ESI-TOF-MS data set

Acknowledgement

Max Planck Society

The study presented was conducted in collaboration with

Sven Heiling, Klaus Gase, Ian T. BaldwinMax Planck Institute for Chemical Ecology, MolecularEcology Department, Jena, Germany

Emmanuel GaquerelCentre for Organismal Studies Heidelberg,Plant Defense Metabolism Research Group, Heidelberg,Germany

wounding by herbivory

jasmonic acid

(conjugates)

- Specialist Manduca sexta – mayor pest on various

solanaceae species

- feeds on green leaf material

activation of defense

metabolites like nicotine, rutin,

chlorogenic acid and

HGL-DTGs

© Danny Kessler

Introduction – tobacco plants have enemies…Manduca sexta induces the Jasmonic acid pathway which activates plant defense mechanisms

Jasm

onic

acid

path

wa

y

CH3

CH3

OH

CH3

CH3

OH

DMAPP3 × IPPCH3

CH3

OPP

NaGGPPS

Introduction 17-HGL-DTG biosynthesis

CH2

CH3

OPP

MEP

DXP

HMBPP

GA3P +

pyruvate

GGPP

17-Hydroxygeranyllinalool

NaGLS

CYP71

Chloroplast

Cytosol

Lyciumoside I

CH3

CH3

CH3

CH3

O

O

OH

OH

OH

OH

OO

OH

OH

OH

OH

NaGT1/

NaGT2

CH3

CH3

CH3

CH3

O

O

OH

OH

OH

OH

OO

OH

OH

O

OH

O

OH

OH

OH

OH

Lyciumoside II

NaGT1/

NaGT2

NaRT1 CH3

CH3

CH3

CH3

O

O

OH

OH

O

OH

OO

OH

OH

OH

OH

O CH3

OH

OH

OH

Lyciumoside IV

NaRT1CH3

CH3

CH3

CH3

O

O

OH

OH

O

OH

OO

OH

OH

O

OH

O

OH

OH

OH

OH

O CH3

OH

OH

OH

Attenoside

MATMAT

MATMAT

- 17-HGL is glucosylated at C-3 and C-17 hydroxy groups- HGL-DTGs can be further glycosylated by either glucose or

rhamnose- HGL-DTGs can be malonylated up to three times- Until now we identified 46 HGL-DTGs in N. attenuata

NaRT1

NaGT1

These 2 genes will be important later in this talk

Databases of HRAM MS features

In-house database

Alkaloids

Polyamides Flavonoids

HGL-DTGs>46

Acyl sugars

0 5 10 15 20 25 30 35 40 Time [min]0

1

2

3

Inte

nsity ×

10

5

+MS, BPC of N. attenuata

- high resolution mass spectrometry database

- 137 HGL-DTGs in 35 different solanaceous species

- 46 HGL-DTGs in N. attenuata

- detailed fragmentation information (IS-CID, MS, MS/MS) leading to 955 non-redundant explained fragments

- 30 polyamides- 15 acyl sugars- 15 flavonoids- 7 amino acids- 4 alkaloids- others

In-house database

• Silencing NaGGPPS decreases the total HGL-DTG concentration• Total HGL-DTGs have a significant effect on the growth of the specialist

herbivore M. sexta

Ecological function of HGL-DTGs

Heiling et al. 2010, The Plant Cell, 22, 273-292

Chloroplast

Cytosol

DMAPP3 × IPPCH3

CH3

OPP

NaGGPPS

CH2

CH3

OPP

MEP

DXP

HMBPP

GA3P +

pyruvate

GGPP

CH3

CH3

OH

CH3

CH3

OH


NaGLS

CYP71

CH3

CH3

OH

CH3

CH3

OH

DMAPP3 × IPPCH3

CH3

OPP

NaGGPPS

What is the biological effect if NaGT1 or NaRT1 are silenced?

CH2

CH3

OPP

MEP

DXP

HMBPP

GA3P +

pyruvate

GGPP


NaGLS

CYP71

Chloroplast

Cytosol

Lyciumoside I

CH3

CH3

CH3

CH3

O

O

OH

OH

OH

OH

OO

OH

OH

OH

OH

NaGT1/

NaGT2

CH3

CH3

CH3

CH3

O

O

OH

OH

OH

OH

OO

OH

OH

O

OH

O

OH

OH

OH

OH

Lyciumoside II

NaGT1/

NaGT2

NaRT1 CH3

CH3

CH3

CH3

O

O

OH

OH

O

OH

OO

OH

OH

OH

OH

O CH3

OH

OH

OH

Lyciumoside IV

NaRT1CH3

CH3

CH3

CH3

O

O

OH

OH

O

OH

OO

OH

OH

O

OH

O

OH

OH

OH

OH

O CH3

OH

OH

OH

Attenoside

MATMAT

MATMAT

NaRT1

NaGT1

NaGT1

NaRT1

WT

NaRT1

Pathway designed by Sven Heiling with PathVisio

XX

X

Metabolites

Proteins

“Silencing” NaRT1 has no visible effect on the plant phenotype

“Silencing” NaRT1

– no visible effect on Phenotype, i.e. the “look of the plant”

- what about the metabolome?

NaRT1

„HGL-DTG biosynthesis“

WT

NaGT1


X

X X

X

X

“Silencing” NaGT1

– severe phenotype, buds mostly stalled, multiple branching



NaGT1

“Silencing” NaGT1 causes a severe plant phenotype, buds mostly stalled, multiple branching

Metabolic profiling of WT, GT1 & RT1 tobacco plants

RT1

WT

GT1

Seamless data acquisition of LC-MS/MS measurements for the three N. attenuata lines via impact II & Instant ExpertiseTM

• Combining extracted FMF features resulted in 3539 buckets for further analysis in novel MetaboScape software

Comprehensive feature extraction by “Find Molecular Features” algorithm

•Combines adducts, charge states and isotopes belonging to the same compound detected within a sample•~3300 FMF compounds extracted per sample

753.32961+

775.3103

1+

M+H

M+Na

1588. 544_09_01_BC1_01_22955.d: +MS, MolFeature, 24.0-24.4min

0.0

0.2

0.4

0.6

0.8

1.0

4x10Intens.

745 750 755 760 765 770 775 780 785 790 m/z

Metabolic profiling of WT, GT1 & RT1 tobacco plants Seamless data evaluation by MetaboScape

Automatic identification of known compound >550 bucketsFast and confident de-replication using in-house DB

• 3539 buckets from extracted FMF features in HRAM LC-MS data of samples from WT, GT1 and RT

• Database in simple CSV format of 561 fragments consists of molecular formula, name and retention time

• MetaboScape – with one push of a button: Automatically & confidently annotate 561 buckets

Seamless identification of unknowns Automated molecular formula generation using SmartFormula

• Remaining ~ 3000 features automatically assigned with candidate molecular formula with SmartFormulaTM

Quickly identify relevant information in complex data sets by statistical evaluation PCA analysis separates WT, GT1, and RT1

• PCA scores plot separates metabolites extracts from wild tobacco plants and the plants transformed to alter the expression of rhamnosyltransferase (RT1) and glycosyltransferase (GT1) genes.

• Loadings plot points to relevant compounds for further investigation.

Seamless identification of relevant compounds Automated molecular formula generation using SmartFormula & structural assignment through DB query

• ~ 3000 features automatically assigned with candidate molecular formula with SmartFormulaTM

• Followed by structural assignment through public database (KEGG*, CheBI, PubChem) queries.

SmartFormulaTM

*

WT

NaRT1


XX

X

Metabolites

Proteins

“Silencing” NaRT1

– no visible effect on Phenotype, i.e. the “look of the plant”


NaRT1

MetaboScape 1.0Pathway mapping of identified target compoundslinks data to the underlying biology



log

2 –

fold

ch

an

ge t

oW

T

+10

-10

WT

NaRT1


• Silencing rhamnosyltransferase (NaRT1) leads to complete reduction of all HGL-DTGs with rhamnose

• Silencing rhamnosyltransferase (NaRT1) leads to increase in precursor metabolites


WT

NaGT1


X

X X

X

X

“Silencing” NaGT1

– severe phenotype, buds mostly stalled, multiple branching




log

2 –

fold

ch

an

ge t

o W

T

+7.5

-7.5

WT

NaGT1


• Intermediate products of HGL-DTGs biosynthesis appear after silencing the glucosyltransferase NaGT1



Outline -> Summary

• If you want to automate data processing workflows in Metabolomics you have to be able to rely on the raw data.

• Robustness and Sensitivity and Dynamic range and High resolution, mass accuracy and isotopic fidelity in one scan mode at LC speed is absolute must in Metabolomics: Impact II

• Introducing MetaboScape. Turning MS data into knowledge!

• MetaboScape combines statistical data mining, automatic compound identification routines, and pathway mapping functionalities for linking MS data to biology in discovery metabolomics.

• Here, enabling to detected and link diverse biological effects in genetically silenced lines of N. attenuata impaired in the biosynthetic steps of HGL-DTGs.

aiko barsch metabolomics thessaloniki how to link hram...

Documents