Metabolomic Profiling of Coffees Metabolomic Profiling of Coffees

Metabolomic Profiling of Coffees Using UHPLCUHPLC-QTOF LC/MS and Multivariate Analysis Tool
ASMS 2011
WP 225
Aki H
Akio
Hayashi,
hi Hirokazu
Hi k
Sawada
S
d Agilent
A il t Technologies,
T h l i 99-1,
1 TTakakura,
k k
H
Hachioji,
hi ji Tokyo
T k 192-8510
192 8510 JJapan
I t d ti
Introduction
The coffee
Th
ff bean
b
i one off the
is
h most cultivated
li
d
andd traded
t d d agricultural
i lt l commodities
diti
i the
in
th
world.
ld In
I recentt years biotechnological
bi t h l i l researchh
h made
has
d attempts
tt
t to
t interfere
i t f
with
ith caffeine
ff i
production
d ti
b preventing
by
ti
th expression
the
i
off
genes that encode key enzymes in the caffeine
(1 2). In addition,
biosynthesis(1,
addition food companies
have tried to improve varieties containing highly
functional compounds by crossbreeding.
crossbreeding This
creates issues with the adoption of the
compound to be enhanced and the
determination of which variety naturally
contains a high concentration of the compound,
compound
thus requiring an efficient method for
interpretation To address this need,
interpretation.
need we show a
solution of metabolomic profiling for variance in
the coffee (eg. multiregional blend, decafe, dark
roasted)) as the multivariate sample
p sets.
E
Experimental
i
t l
Materials and Methods
All coffee beans ((Blend, Decafe, Dark Roast, Blue
Mountain, Kilimanjaro, Mocha)) were purchased as
the multivariate sample sets.
Indole 3 acetamide and chlorogenic acid were
Indole-3-acetamide
purchased from Sigma.
The coffee beans were ground and extracted with
deionized water at 90 Ԩ.
Ԩ The extracts were filtered
through a 0.2
0 2 um PVDF membrane (Whatman).
(Whatman) A
1290 UHPLC system coupled to an 6530 accurate
accuratemass Q
Q-TOF
TOF LC/MS system was used for analysis.
analysis
Water extracts were applied to a ZORBAX Eclipse
Plus C18 RRHD column (2.1
(2.1*150mm,
150mm, 1.8µm). The
mobile
ob e pphase
ase was
as 100 mM aammonium
o u formate
o ate / 00.1%
%
formic acid ((solvent A)) and acetonitrile ((solvent B)) .
The gradient program was as follows: 2% B kept for
5 min,
min 90% B to 20 min,
min and 90% B kept for 5 min.
min
The flow rate was set to 0.2
0 2 ml/min,
ml/min and 1 ul of each
diluted sample was injected.
injected
D t Processing
Data
P Results
Sc
Sdcheme
h
and
R lt
R l i and
Di
Discussion
i d Results
Molecular Feature Exxtractor
TOF
Q-TOF
Diandd Discussion
Results
R
ltDiscussion
Dii
i
Data
mining
i i
File
export
LC/MS data
Compounds
p
alignment
g
(RT, Mass)
Acquisition
Groupp
definition
METLIN DB search
search, Molecular
Formula Generation
& MS/MS identification
I d l 3
Indole-3-acetamide
id Std.
Sd
Statistical
filtering
Characterization
Figure 1. Schematic flow of the multivariate non-taarget analysis procedure.
The samples were ionized with a Dual ESI source,
source
and acquired mass range was m/z 50-1200.
50 1200
Acquisition rate was 3 MS spectra/sec.
spectra/sec
Figure 5.
5 Overlayed EICs (175.0866,
(175 0866 Indole-3-acetamide,
Indole 3 acetamide
C10H10N2O,
O CAS 879
879-37-8)
37 8) from Kilimanjaro,
Kilimanjaro Blue
Mountain, Mocha extracts. The extraction width of EICs
were ±10ppm. It shows that the content of indole-3acetamide
t id in
i Kilimanjaro
Kili
j
extract
t t increases
i
significantly.
i ifi tl
Data Analysis
y
Figure 2. TICs of coffee extracts (Blend; red, Decafe: blue, Dark Roast; brown, Blue Mountain; pink,
Kilimanjaro; orange, Mocha; green). All extracts sh
how almost same chromatograms except “Decafe”,
i di i that
indicating
h feature-finding
f
fi di tooll is
i necessary for
f thes
h e comparative
i analyses.
l
I i i l entities
Initial
i i (993)
Filt d entities
Filtered
titi (61)
F
Figure
3 PCA 3D score plot of extracted features
3.
(Blend; red,
red Decafe; blue,
blue Dark roast; grey,
grey Blue
M
Mountain;
brown, Kilimanjaro; green, Mocha; pink).
By filtering
By
g out the non-specific
p
common features,,
the profile of each variety could be discriminated.
Figure
g
7. Profile pplot of chrologenic
g
acid ((RT 10.35min,
C16H18O9, CAS 327-97-9) for all coffee beans.
MPP showed significant decrease of chlorogenic
acid known as one of the polyphenols,
acid,
polyphenols in the dark
roasted coffee extract. Chlorogenic acid is a thermal
thermalunstable compound,
p
and the roastingg pprocess should
be circumvented nutritionally.
Conclusions
Caffeinne (RT 9.75min,
9 75min
C8H10N4O2, CAS 58
58-08-2)
08 2)
All extracted
t t d coffee
ff samples
l were injected
i j t d 3 times
ti
f QC.
for
QC
The groupings of this study were varieties of coffee
beans roasting process,
beans,
process and presence of caffeine.
caffeine
The procedure for data analysis comprised five
sequential steps. First, the raw data was processed
by a molecular feature extractor (MFE) for non
non-target
target
miningg and reconstitution of a compound-centric
p
list
((RT,, neutral mass,, and intensity)
y) equipped
q pp
for
MassHunter data analysis
y
software. Mass Profiler
Professional ((MPP)) software for the multivariate
analysis aligned the compounds among the data
sets, and QC was performed using the intensity of
20% CV in each sample condition. To discover the
quantitative variation of compounds, we performed
ANOVA (p < 0.05) and fold change (FC ≥ 2.0)
filtering.
Figure
g
4. K-means clustering
g of Blue Mountain,
Kilimanjaro, and Mocha extracts. Indole-3-acetamide was
id tifi d by
identified
b METLIN personall DB in
i the
th KilimanjaroKili
j
specific cluster.
cluster
Indole-3-acetamide
Indole
3 acetamide std.
F ti from
Fraction
f
Kilimanjaro
Kili
j
extract
Figure 6.
6 MS/MS spectra of indole-3-acetamide
indole 3 acetamide standard
and extracted fraction (RT 12.00 min, m/z 175.0866) from
Kilimanjaro beans. Accurate-Mass, RT, and MS/MS
spectrum of this fraction corresponded to indole-3acetamide
acetamide.
• Because visual inspection of the data from each
coffee type was indistinguishable, data mining
and statistical tools were required. It was shown
that the Kilimanjaro bean contained indole-3acetamide
id muchh more than
h other
h beans,
b
M h
Mocha,
Bl Mountain.
Blue
M
i
• Chlorogenic
Chl
i acid
id was thermally
th
ll decomposed
d
d by
b
th roasting
the
ti process, therefore
th f
D k Roast
Dark
R t beans
b
showed
h
d lower
l
contents
t t off chlorogenic
hl
i acid
id than
th
any other ones.
ones
• This study showed multivariate analysis and
UHPLC coupled to Accurate-Mass
Accurate Mass QTOF LC/MS
system is powerful solution for high-throughput
high throughput
screening to make improvements in agricultural
varieties and nutritional food processing more
efficiently
efficiently.
References
1)
2)
M. Silvarolla et al, Nature, 2004, 429, 826.
S. Ogita et al, Nature, 2003, 423, 823.