Metabolomics Technology Development LDI-MS (UK EPSRC/RSC); SERS (UK BBSRC) Imaging
description
Transcript of 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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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)
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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.
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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 ??
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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.
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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.
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
log10 Concentration (M)
log 10
S/N
599
cm
-1
Power fit
R = 0.79295
-7 -6.5 -60.5
1
1.5
2
2.5
log10 Concentration (M)
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
log10 Concentration (M)
log 10
Are
a un
der
599
cm-1
Power fit
R = 0.86377
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
Optimisation of surface-enhanced Raman scattering (SERS)
experiments
Roger Jarvis, William Rowe, Nicola Yaffe, Sven Evans, Joshua Knowles,
Ewan Blanch & Roy Goodacre
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Mahalanobis distance
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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
Mahalanobis distance
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
Laboratory for Bioanalytical Spectroscopy (http://biospec.net/)
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.