metabolomics technology development ldi-ms (uk epsrc/rsc); sers (uk bbsrc) imaging

Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)

• Metabolomics Technology Development– LDI-MS (UK EPSRC/RSC);

SERS (UK BBSRC)• Imaging

– MALDI imaging (Shimadzu); Raman, FT-IR imaging (ORS); SIMS (UK BBSRC)

• Bacterial Identification– SERS (UK HOSDB)

GenomicsGenome

GeneGenomicsGenome

Gene

ProteomicsProteome

ProteinProteomicsProteome

Protein

MetabolomicsMetabolome

MetaboliteMetabolomics

MetabolomeMetabolite

BioinformaticsIntegration

Holtorf et al. (2002)

TranscriptomicsTranscriptome

mRNATranscriptomics

TranscriptomemRNA

PhenotypePhenomicsPhenome

PhenotypePhenomicsPhenome

Surface-enhanced Raman Scattering for MetabolomicsRoger Jarvis & Roy GoodacreContact: [email protected]

Levels of functional genomics


MetabolomicsThe analysis of metabolites (typically low molecular weight molecules) in a biological organism at a given time, with the aim of elucidating gene function and defining biochemical pathways.The Metabolome“The total biochemical composition of a cell, tissue or organisms at any given time (Oliver et al., 1998).”

School of Chemistry & School of Chemistry & Manchester Interdisciplinary Manchester Interdisciplinary Biocentre, The University of Biocentre, The University of

ManchesterManchester

• Metabolomics– E. coli stress (BBSRC & AZ);

recombinant mammalian cells (BBSRC); Oral cancer (EPSRC); Psoriasis (Stiefel Labs); META-PHOR (EU FP6); Biotrace IP (EU FP6); Plants (BBSRC)

• Systems Biology– STREPTOMICS (EU FP6);

SYSMO (EU/BBSRC)


Why study the metabolome?

Functional Genomics aims to assign (new) functions to (uncharacterised) genes.

Mainly E. coli,S. cerevisieae

Measure cell components with MS, FTIR, GCMS

Grow mutant & WT cells under different conditions

Functional genomics

Current

Need

Ultimate Goal

Knowledge of (most) fundamental metabolic processes

Develop understanding to investigate metabolic network regulation

Determine gene function(including Bioinformatics)

Develop understanding of responses to genetic or environmental influences

“Genomics and proteomics tell you what might happen, but metabolomics tells you what actually did happen.”

Bill Lasley, University of California, Davis.


Four Approaches

Metabolic profilingQuantification of

pre-defined targets.(GC-MS, LC-MS, NMR,

HPLC, LC/MS/MS)

Metabolite target analysis

Analysis of specific metabolites.

MetabolomicsUnbiased identification of all metabolites in sample.

Metabolic FingerprintingCrude metabolite mixtures for classification. (FT-IR/Raman/DIMS)

(Fiehn, 2001)

Metabolite analysis.

Particular interest in low molecular weight compounds – the substrates and products in pathways.

Selection of technology is a compromise between speed, selectivity and sensitivity.

SER(R)S

SER(R)S

SER(R)S ??


SERRS Reproducibilty

• We want to use SERRS as a metabolic profiling and fingerprinting tool

• We know that there is a question mark over reproducibilty

• Metabolomics requires quantitatively accurate data

• Therefore we have been looking at strategies for assessing objectively, the reproducibility of our SERRS experiments


Ag citrate

Au citrate

EDTA Fructose Glucose Oleyl-amine

PVP Thiol400

450

500

m

ax.

50

100

150

FW

HH

1

2

3

Ext

inct

ion

Colloidal Batch-Batch Reproducibility

• 3 replicate absorbance measurements • (absorption) max. - larger value equates to a larger particle size• FWHH (full width at half height), a larger FWHH indicates wider particle size distribution. • Extinction - lower value for the extinction indicates greater aggregation

Colloids prepped by Emma Oleme and Arunkumar Paneerrselvam


SERRS spectra of Cresyl Violet

800 850 900 950 1000 1050 1100 1150 1200 1250 1300

81077

8

846 87

7

906

949 96

498

5

1038

1049 10

77

1102

1167

1186

1230

1277

Raman shift (cm -1)

Ram

an

phot

on

coun

t (a

.u.)

Au citrate

Ag citrate

PVP

EDTA

**

* *

* *

*

*

* *

* *

* *

*

*

Mean SERRS spectra of cresyl violet acquired using the four colloidal substrates that were found to be SERRS active.


Signal-to-noise ratios (S/N) observed in the median SERRS spectra of cresyl violet

SubstrateRaman shift (cm-1)

Mean877 1049 1186 1277

Au citrate 1.24 1.20 1.58 1.88 1.47

846 877 985 1277

Ag citrate 1.23 1.28 1.81 2.10 1.60

EDTA 1.09 1.11 1.44 1.72 1.34

PVP 1.29 1.28 1.93 2.03 1.63


MANOVA on the S/N ratios calculated from the SERRS bands identified in spectra of cresyl violet, from four

active substrates

Ag citrate Au citrate EDTA PVP

Raw SERRS spectra

Wilks' L[a] 0..429 0.061 0.122 0.300

~ F[b] 1.187 6.890 4.187 1.856

P[c] NS 0.000 0.006 NS

Row normalised SERRS spectra

Wilks' L[a] 0.543 0.141 0.577 0.560

~ F[b] 0.803 3.749 0.712 0.757

P[c] NS 0.009 NS NS


Quantification of Cresyl Violet using SERRS

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90

500

1000

1500

2000

2500

3000

3500

Correlation coefficient

Nu

mb

er

of

mo

de

ls

0.781187

0.811145

0.841045

0.84999

0.84945

0.84855

0.84830

RRaman band (cm-1)

0.781187

0.811145

0.841045

0.84999

0.84945

0.84855

0.84830

RRaman band (cm-1)

-5.5 -5 -4.5 -4 -3.5 -3 -2.5 -22.5

3

3.5

4

Concentration cresyl violet (log10M)

log 1

0of

are

a un

der

830

cm-1

R = 0.84

• Bootstrapped correlation analysis for the log-log relationship to area under the cresyl violet SERRS band at 930 cm-1

• Dilution series from 5 x 10-6 M to 5 x 10-2 M, using the

• PVP capped colloidal silver substrate.


5 10 15 20 40 55 60 700

200

400

600

800

1000

1200

I732c

m-1

% colloidal silver

Next question – we can find colloids that give statistically reproducible batch to batch SERS – but

what happens when we start playing with chemistry?

5 10 15 20 40 55 60 70

0

500

1000

1500

2000

2500

3000

I732c

m-1

% colloidal silver

5 10 15 20 40 55 60 70

0

500

1000

1500

2000

2500

I732c

m-1

% colloidal silver

5 10 15 20 40 55 60 70

0

500

1000

1500

2000

2500

3000

I732c

m-1

% colloidal silver

Sodium nitrate

Potassium choride

Sodium chloride

Potassium nitrate

Optimisation of cytosine SERS


Cytosine

-7 -6.5 -60

0.05

0.1

0.15

0.2

log10 Concentration (M)

log 10

S/N

599

cm

-1

Power fit

R = 0.79295

Batch 1Batch 2

-7 -6.5 -60

0.05

0.1

0.15

0.2


log 10

S/N

599

cm

-1

Power fit

R = 0.79295

-7 -6.5 -60.5

1

1.5

2

2.5


log 10

Are

a un

der

599

cm-1

Power fit

R = 0.86377Batch 1Batch 2

-7 -6.5 -60.5

1

1.5

2

2.5


log 10

Are

a un

der

599

cm-1

Power fit

R = 0.86377


Optimisation of surface-enhanced Raman scattering (SERS)

experiments

Roger Jarvis, William Rowe, Nicola Yaffe, Sven Evans, Joshua Knowles,

Ewan Blanch & Roy Goodacre


Experimental

Pseudo Full-Factorial Experiment• 3 colloidal silver preps at 25, 50 & 75% v/v

– hydroxylamine, citrate, borohydride• 6 aggregating agents at 1, 10 & 100 mM

– NaCl, KCl, Na2SO4, K2SO4, NaNO3, KNO3

• 785 nm NIR Raman probe, 3 s integrations with ~ (Goodness knows what!!) mW power a source, spectral range (150 - 2900 cm-1)

• Single analyte – L-cysteine (100 mM)• Total of 162 experiments,5 replicate measurements for

each giving 810 SERS spectra

This allows us to determine the “optimal” experimental conditions


Cont…

Multiobjective optimisation• Questions

1.Can we use this experiment to determine the utility of an directed search algorithm for optimising these conditions more rapidly?

2.Could some form of interpolation be used to derive further experiments that yield superior results?

• Objective functions 1.Reproducibility: standard deviation of the Mahalanobis

distance between principal component scores recovered from replicate spectra

2.Signal intensity: peaks areas calculated for 4 major bands and meaned across replicates


Published results: GC-TOF mass spectrometer optimization via PESA-II

Yeast supernatant Pareto front after 114 generations

Rtime10 12 14 16 18 20 22 24

150

200

250

300

350

400

450

500

O’Hagan,S., Dunn, W.B., Brown, M., Knowles, J.D. and Kell, D.B. (2005) Closed-loop, multiobjective optimization of analytical instrumentation: gas chromatography/time-of-flight mass spectrometry of the metabolomes of human serum and of yeast fermentations. Analytical Chemistry 77(1): 290-303.

PESA-II used to optimize the settings of a mass-spectrometer to improve the chromatograms.

Optimized:Optimized:- Number of true peaks - Number of true peaks - Signal-to-noise ratio- Signal-to-noise ratio- Sample analysis time - throughput- Sample analysis time - throughput


Typical SERS spectrum of L-cysteine and Raman bands for which peak areas were calculated

400 600 800 1000 1200 1400

100

200

300

400

500

600

700 647

795

911

1034

Raman shift (cm-1)

Ram

an p

hoto

n co

unts

C-SRed –

shifted due to binding at

silver surface


0.2 0.4 0.6 0.8 1 1.2 1.4 1.60

2

4

6

8

10

12

14

Mahalanobis distance

Fre

quen

cySummary of metrics calculated to quantify signal reproducibility and

intensity of enhancement

0 100 200 300 400 500 600 700 8000

20

40

60

80

100

120

peak area

Fre

quen

cy

Homogeneous distribution Skewed Distribution


Summary of Results

0.2 0.4 0.6 0.8 1 1.2 1.4 1.60

100

200

300

400

500

600

700

800

1 2

3

4 5

6

7 8

9

10 11

12

13 14

15

16 17

18

19

20

21

22

23

24

25

26

27

28 29

30

31

32

33

34

35

36

37 38

39

40 41

42

43 44

45

46 47

48

49 50

51

52

53

54

55 56

57

58 59

60

61 62

63

64 65

66

67 68

69

70

71

72

73 74

75

76 77

78

79 80

81

82 83

84

85 86

87

88 89

90

91 92

93

94 95

96

97 98

99

100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154

155

156 157

158

159 160

161

162


A

rea

unde

r pe

aks

Pareto front

400 600 800 1000 1200 14000

500

1000

1500

2000

Raman shift (cm-1)

Ram

an p

hoto

n co

unt

Experiment #45

400 600 800 1000 1200 14000

500

1000

1500

Raman shift (cm-1)

Ram

an p

hoto

n co

unt

Experiment #54

Exp. Colloid Amount (% v/v) Agg. Agent Conc. (mM) Enhancement M. dist.45 Hydroxylamine 75 K2SO4 100 662.0311 0.853936 Hydroxylamine 75 NaNO3 100 779.4253 0.764254 Hydroxylamine 75 KNO3 100 675.0239 0.6618


Multiobjective Pareto optimisation using the PESA II algorithm

0.2 0.4 0.6 0.8 1 1.2 1.4 1.60

100

200

300

400

500

600

700

800

1 2

3

4 5

6

7 8

9

10 11

12

13 14

15

16 17

18

19

20

21

22

23

24

25

26

27

28 29

30

31

32

33

34

35

36

37 38

39

40 41

42

43 44

45

46 47

48

49 50

51

52

53

54

55 56

57

58 59

60

61 62

63

64 65

66

67 68

69

70

71

72

73 74

75

76 77

78

79 80

81

82 83

84

85 86

87

88 89

90

91 92

93

94 95

96

97 98

99

100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154

155

156 157

158

159 160

161

162


A

rea

unde

r pe

aks

Pareto front

• Find solutions which give best trade-off between 2 objectives

• PESA II is a region based Pareto selection algorithm– Select a region or

hypercube– Randomly select individual

from this subset

• Problem!! Our solution space is quite sparse and disperse!!

Analysis to be completed, however:• Directed search optimises

experimental conditions in 60 iterations

• Interpolation attempted but hasn’t improved SERS


www.biospec.net

Group Leader: Professor Roy GoodacrePostdocs: Dr Will Allwood, Dr Robert Cormell, Dr Elon Correa, Dr Roger Jarvis, Dr Yankuba Kassama, Dr Iggi Shadi, Dr Catherine Winder, Dr Yun Xu.With Collabs: SERS (4), Metabolomics (2), ToF-SIMS (2)Research Technicians: Steffi Schuler, Richard O’ConnorPhD Students: Felicity Currie, Katherine Hollywood, Nicoletta Nicolaou, Soyab Patel, Ketan Patel, Emma Wharfe, Nicola Wood, Dong Hyun Kim, Will Cheung, Robert Coe.

metabolomics technology development ldi-ms (uk epsrc/rsc); sers (uk bbsrc) imaging

Documents