Application of NMR-based Metabolomics in Obesity Research
1. Application of NMR and MS based
Metabolomics
in Natural Product Science
February, 2010
Choi, Hyung-Kyoon
hykychoi@cau.ac.kr
College of Pharmacy, Chung-Ang University
Republic of Korea
2. Metabolomics
Metabolomics
○ Metabolome
- total low molecular weight compounds in biofluid,
cells,
and tissue in living organism
○ Metabolomics
- comparative and non-targeted analysis of metabolome
using various analytical methods
3. Tools for metabolomics
Tools for metabolomics
Tools Pros Cons
Robustness and Metabolite
NMR
reproducibility overlapping
GC-MS Excellent sensitivity Need to derivatize
GC X GC TOF
Lower reproducibility
LC-MS Excellent sensitivity
than GC
4. Statistical methods (1)
Statistical methods (1)
○ Principal component analysis (PCA)
- Oldest and most widely used non-supervised
multivariate statistical technique
- Reduce the dimension of the original data set
○ Partial least squares-discriminant analysis (PLS-DA)
- Supervised method rendering class to each sample
- Clearer differentiation of each class and easier
investigation of marker compounds
5. Statistical methods (2)
Statistical methods (2)
○ Partial least squares-regression (PLS-R)
- Correlate the X variables (eg. NMR spectra data)
with Y variables (eg. Antioxidative activity)
- Prediction model can be developed
6. Schematic overview of NMR –based
metabolomics
Holmes et al. (2006) Planta medica 72:771-785
17. Metabolomic profiling and prediction model
Metabolomic profiling and prediction model
development of Citrus Fruit using NMR and
development of Citrus Fruit using NMR and
MVA
MVA
NMR and antioxidative activity analysis
Mature and immature fruit
Peel and flesh
24. Application of Metabolomics (1)
Application of Metabolomics (1)
•Biomarker development
Early biomarkers
Prognostic biomarkers
Diagnostic biomarkers
Late biomarkers of diseases
such as cancers, diabetes, Alzheimers etc.
25. Pharma perspective on metabolomics
• Looking for disease markers
Disease Conventional Ideal scenario Animal Metabolic
biomarker model profiling tools
Diabetes Increased Earlier marker High fat diet Lipid-MS,
plasma/urinary pre-disease mice NMR/MS
glucose onset profiling
Atherosclerosis Lipoprotein Earlier marker Watanabe Lipid-MS,
profiles pre-disease rabbits NMR/MS
onset profiling
Alzheimer Cognitive Markers of PS1 mice NMR/MS
function test disease onset, profiling
progression
Schizophrenia Behavioural Markers of Coloboma NMR/MS
test disease onset, mice profiling
progression
26. Consideration for Right Samples!
Consideration for Right Samples!
• Getting the right sample
- plasma, serum, urine, tissue, saliva
- Correlation with the disease
• Control group
- Gender
- Ethnic
- Age
- Lifestyle
- Nutritional and medical condition
27. Effect of acute dietary standardization on the urinary, plasma, and
salivary metabolomic profiles of healthy humans
Urine
Saliva
Plasma
Marianne et al. Am J Clin Nutr 2006;84:531–9.
28. Application of Metabolomics (2)
Application of Metabolomics (2)
Enhanced production of useful secondary metabolites by
M/O, plant cell and tissue culture
Metabolic engineering 의 기초자료로 metabolomics 이
용
stress 에 의해 유도된 metabolic change 의 monitoring
대사경로 중 rate-limiting step 의 규명
유전체 기능 연구 (functional genomics)
외래유전자 도입에 의해 유발된 metabolic changes 의
규명
knockout mutation 에 의한 metabolic effects 의 규명
29. Application of Metabolomics (3)
Application of Metabolomics (3)
• 천연물 신약개발 : 지표성분 탐색 , 원료 및제품 표준
화
elucidation of bioactivity correlated biomarker
약용식물 개체별 원산지 구분
천연물 함유 제품의 quality control
(batch to batch variation)
• 건강기능식품의 efficacy 조사
dietary effects
32. Introduction
The prevalence of obesity is increasing rapidly worldwide.
To reduce the associated risks, it is necessary to investigate the causes of
weight gain (e.g., lifestyle and behavior).
To prevent obesity, early diagnosis and treatment of obesity are important.
Obesity studies involving the administration of a high-fat diet (HFD) in
animal models are known to be applicable to human obesity.
33. Materials & Methods
Experimental Design
SD Male Rats
(n=20, 110-120 g)
Normal diet group
(ND, n=10)
ND low gainers ND high gainers HFD low gainers HFD high gainers
(n=5) (n=4) (n=5) (n=5)
visceral fat-pad urine
serum 1
H-NMR
liver
multivariate
statistical
Biological analysis
Analaysis
35. Results
Fig. 1. 1H-NMR spectra and assignment of urine metabolites
The signals assigned based on comparisons with the chemical shifts of standard
compounds using the Chenomx NMR software suite (version 5.1, Chenomx, USA).
36. Results
Fig. 2. PLS-DA score plots of urine metabolites
• The PLS-DA score plot showed a separation
between ND low gainers and ND high gainers
• Although each rat of the two groups
comsumed the same normal diet, it was
possible to metabolically discriminate rat
groups with different physical constitutions.
• The PLS-DA score plot showed a separation
between ND low gainers and HFD high gainers
• The various endogenous metabolites changed
in rats comsuming the high-fat diet.
37. Results
Validation of PLS-DA models
Cross-validation
• Plastic cage
• R2: the goodness of fit (0<R2<1)
- 1 means perfect fit
• Q2: the goodness of prediction
- >0.5 means good prediction
- >0.9 means excellent prediction
• Plastic cage testing
Permutation
• Provided the statistical significance of the estimated
predicted power of the models
• Comparing R2Y and Q2Y values of original model with
them of re-ordered model
• Valid model
: R2Y intercept <0.3-0.4 & Q2Y intercept <0.05
38. Results
Table 4. The VIP values of the compounds
Generally, a cutoff for VIP around 0.7-0.8 works well.
The compounds with VIP>0.75
: influential compounds for separating each samples in PLS-DA models.
39. Results
Fig. 4. Intensity of the metabolites
Normalized relative to the creatinine
l cons titution intensity
Physica
An independent t test (*p < 0.025)
was performed to assess the statistical
significance between each group
The relative intensities of betaine,
taurine, acetone/acetoacetate,
t diet
High-fa phenylacetylglycine, pyruvate, lactate,
and citrate differed significantly
between ND low gainers and ND high
gainers/HFD high gainers.
40. VIP in Metabolomics
Dr.
Nicholson
Imperial Dr.
Col.
Verpoorte
Leiden
Dr.
Gonzalez
Univ. Dr. Tomita NIH/NCI
Keio Univ.
Dr. Kopka
Max-Planck
Institute
Dr. Fiehn Dr. Sumner
UC Davis Samuel
Roberts
Noble
Foundation
41. SWOT of Metabolomics
Strength Weakness
Robust and stable Analytical sensitivity
analytical platforms Analytical dynamic range
Minimally invasive Complexity of data sets
Real biological endpoint High capital cost
Whole system integration
Oppurtinities Threats
Much experience from Skepticism of non-
mammalian system studies hypothesis led studies
(e.g. pathways) Conservatism
Potential of multi-omics Lack of well trained scientists
integration
Web-based diagnotics
42. Acknowledgement
Prof. Rob. Verpoorte, Leiden University
Dr. Younghae Choi, Leiden University
Dr. Dae Young Kwon, KFRI
Prof. Young-Suk Kim, Ewha Womans University
Prof. Somi Cho, Kim, Cheju National University
Prof. Taesun Park, Yonsei University
Prof. Yeon-Soo Cha, Chunbuk National University
Prof. Jung-Hyun Kim, Chung-Ang University
Ph.D students
Seung-Ok Yang, Sun-Hee Hyun
MS students
So-Hyun Kim, Hee-su Kim, Yujin Kim
43. What is now proved
was once only
imagined.
- William Blake
Notas do Editor
Another key feature of metabolomics is that the metabolomics is generally performed with multivariate statistical analysis, such as principal component analysis or partial least squares-discriminant analysis. PCA is ---
Another key feature of metabolomics is that the metabolomics is generally performed with multivariate statistical analysis, such as principal component analysis or partial least squares-discriminant analysis. PCA is ---
I also conducted the metabolomic profiling and developed prediction model using NMR data for antioxidative andivities of Citrus fruits.
5 to 10 meters in height, with long, sharp, solitary spines. The leaflets are entire or nearly so, sparingly hairy beneath and on margins, ovate oblong to elliptic, and 8 to 12 centimeters long. The obovate petioles are broadly winged. The flowers are white, very fragrant, and crowded in axillary, short racemes. The fruit is large, nearly spherical or obovoid, up to 20 centimeters or more in diameter. The rind, which is very thick and spongy, is fairly easily removed from the segments of the fruit. The pulp is pale yellow to pink or red, and sweet or acid, with large distinct vehicles.
This is the representative spectrum of Citrus fruit.
This is the comparison between observed FRSA predicted FRSA derived from the prediction model I developed.
It was possible to confirm the predictibility of the developed model using test set validation. It shows good correlation between observed and predicted FRSA values.
I also conducted the metabolomic profiling and developed prediction model using NMR data for antioxidative andivities of Citrus fruits.