See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/225164314

Procedure for tissue sample preparation and metabolite extraction for high-
throughput targeted metabolomics

Article  in  Metabolomics · February 2011

DOI: 10.1007/s11306-011-0293-4

CITATIONS READS

190 8,082

6 authors, including:

Werner Römisch-Margl Karsten Suhre

Helmholtz Zentrum München Cornell University
97 PUBLICATIONS   3,392 CITATIONS    473 PUBLICATIONS   12,536 CITATIONS   

SEE PROFILE SEE PROFILE

Jerzy Adamski
Helmholtz Zentrum München
723 PUBLICATIONS   16,313 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Immediate reduction of serum citrulline but no change of steroid profile after initiation of metformin in individuals with type 2 diabetes View project

Targeted Metabolomics Identifies Reliable and Stable Metabolites in Human Serum and Plasma Samples View project

All content following this page was uploaded by Karsten Suhre on 29 May 2014.

The user has requested enhancement of the downloaded file.

Metabolomics
DOI 10.1007/s11306-011-0293-4
ORIGINAL ARTICLE
Procedure for tissue sample preparation and metabolite
extraction for high-throughput targeted metabolomics
Werner Römisch-Margl • Cornelia Prehn •
Ralf Bogumil • Cornelia Röhring • Karsten Suhre
Jerzy Adamski
Received: 28 November 2010 / Accepted: 18 February 2011
Ó Springer Science+Business Media, LLC 2011
Abstract Reproducible quantification of metabolites in
tissue samples is of high importance for characterization of
animal models and identification of metabolic changes that
occur in different tissue types in specific diseases. How-
ever, the extraction of metabolites from tissue is often the
most labor-intensive and error-prone step in metabolomics
studies. Here, we report the development of a standardized
high-throughput method for rapid and reproducible
extraction of metabolites from multiple tissue samples
from different organs of several species. The method
Werner Römisch-Margl and Cornelia Prehn contributed equally to
this work.
Electronic supplementary material The online version of this
article (doi:10.1007/s11306-011-0293-4) contains supplementary
material, which is available to authorized users.
W. Römisch-Margl  K. Suhre
Helmholtz Zentrum München, Institute of Bioinformatics and
Systems Biology, German Research Center for Environmental
Health, Neuherberg, Germany
C. Prehn  J. Adamski (&)
Helmholtz Zentrum München, Institute of Experimental
Genetics, Genome Analysis Center, German Research Center
for Environmental Health, Ingolstaedter Landstrasse 1,
85764 Neuherberg, Germany
e-mail: adamski@helmholtz-muenchen.de
R. Bogumil  C. Röhring
BIOCRATES Life Sciences AG, Innsbruck, Austria
K. Suhre
Faculty of Biology, Ludwig-Maximilians-Universität,
Munich, Germany
J. Adamski
Lehrstuhl für Experimentelle Genetik, Technische Universität
München, 85350 Freising-Weihenstephan, Germany
•
involves a bead-based homogenizer in combination with a
simple extraction protocol and is compatible with state-of-
the-art metabolomics kit technology for quantitative and
targeted flow injection tandem mass spectrometry. We
analyzed different extraction solvents for both reproduc-
ibility as well as suppression effects for a range of different
animal tissue types including liver, kidney, muscle, brain,
and fat tissue from mouse and bovine. In this study, we
show that for most metabolites a simple methanolic
extraction is best suited for reliable results. An additional
extraction step with phosphate buffer can be used to
improve the extraction yields for a few more polar
metabolites. We provide a verified tissue extraction setup
to be used with different indications. Our results demon-
strate that this high-throughput procedure provides a basis
for metabolomic assays with a wide spectrum of metabo-
lites. The developed method can be used for tissue
extraction setup for different indications like studies of
metabolic syndrome, obesity, diabetes or cardiovascular
disorders and nutrient transformation in livestock.
Keywords Mass spectrometry  Tissue extraction 
High-throughput  Quantification  Complex diseases 
Food quality
1 Introduction
Studies of the metabolome using kit-based high-throughput
flow injection (FIA) mass spectrometry in large population
cohorts and animal studies revealed new insights into the
interplay between genetics and environmental impact
(Altmaier et al. 2009,2008; Gieger et al. 2008; Illig et al.
2010; Wang-Sattler et al. 2008; Weikard et al. 2010;
Zhai et al. 2009). At present, hundreds of endogenous
123

metabolites can be quantified in body fluids like plasma and
serum in a high-throughput manner using targeted meta-
bolomics approaches (Altmaier et al. 2008; Bogumil et al.
2008; Denkert et al. 2008; Freedman et al. 2008; Griffin and
Kauppinen 2007; Koal and Deigner 2010; Sreekumar et al.
2009; Urban et al. 2010). For further studies of candidate
target genes in animal models comprehensive analyses are
required, similar to that offered in specialized phenotyping
centers (Fuchs et al. 2009; Saloniemi et al. 2009). Several
aspects of sample logistics and storage conditions (Kratzsch
and Ceglarek 2010; Zivkovic et al. 2009) are crucial for the
quality of large scale studies.
Tissue homeostasis is largely intracellular, while the
established matrices for metabolomic studies, like blood,
urine, or saliva, are typically extracellular. Metabolite
concentrations in body fluids therefore reflect systemic
effects as a result of various simultaneously occurring
processes over different places and cell types in a whole
organism. In both human and animal models not only blood
samples and other body fluids but also tissue samples or
biopsies are moving more and more into research focus.
The variety of metabolite classes, tissue types, organisms,
analytical methods and biological questions addressed so
far are a challenge in the development of standardized
high-throughput extraction procedures. On the other hand
pre-analytical procedures and requirements like collection
and storage conditions were addressed already in detail
(Deprez et al. 2002; Fiedler et al. 2007; Griffin and
Nicholls 2006). Nevertheless, simple and robust methods,
which are suitable for high-throughput analyses and
applicable at least to several tissue types and/or analytical
methods, are highly desirable. Accurate and reproducible
determination of metabolite concentrations in tissue sam-
ples is of high importance to characterize animal models
and to identify metabolic changes that occur in different
tissue types in specific diseases. However, the extraction of
metabolites from tissue is often one of the most labor-
intensive steps for metabolomic studies and these manual
approaches often lead to low reproducibility between
repeated samples. Nevertheless, the disturbed metabolic
processes specific for disease could be better analyzed in
target tissues for disorders rather than in body fluids.
We describe a high-throughput method for the rapid
extraction of metabolites from multiple tissue samples in
different species using a bead-based homogenizer in
combination with a simple extraction protocol. The method
is compatible with the AbsoluteIDQTM technology devel-
oped by Biocrates AG (Innsbruck, Austria), which has been
proven to be in conformance with FDA-Guideline (U.S.
Department of Health and Human Services 2001). The
assay procedures in human serum and plasma have been
described in our previous works (Altmaier et al. 2008;
Gieger et al. 2008; Illig et al. 2010). The metabolite panel
123
W. Römisch-Margl et al.
of the kit covers more than 150 analytes including amino
acids, hexoses, acylcarnitines, glycerophospholipids, and
sphingolipids representing metabolites with significantly
distinct lipophilic and hydrophilic properties. We tested
different extraction solvents for reproducibility and matrix
effects in different mouse and bovine animal tissues, such
as liver, kidney, muscle, brain, and adipose tissue. We have
chosen these tissue types to address target tissues in fre-
quent human diseases like metabolic syndrome, obesity,
diabetes or cardiovascular disorders and to support studies
for food quality. The developed method can be used for
tissue extraction setup for different indications now.
2 Materials and methods
2.1 Sample preparation
Male mice (C57BL/6 J), aged 41 weeks, were narcotized
and all available blood was drawn (retro orbital) as
described (Gailus-Durner et al. 2009). The mice were
sacrificed and the abdominal skin was moistened with 80%
ethanol, opened and pulled aside. The abdomen was
opened and the abdominal aorta and the inferior vena cava
were cut through while avoiding contact with organ sur-
faces. Organs were dissected within few minutes, carefully
patted with lint free tissue paper, cut into small pieces, and
placed into pre-labeled tubes. Tissue samples were snap-
frozen in liquid nitrogen and stored at -80°C until extrac-
tion. Snap-frozen bovine tissue samples (F2 Charolais x
German Holstein) were collected as described (Kühn
et al. 2002; Weikard et al. 2010) and provided by the
Leibniz Institute for Farm Animal Biology (Dummerstorf,
Germany).
2.2 Tissue homogenisation
Preliminary tests with a glass–glass potter and a micro
dismembrator (B. Braun Biotech International GmbH,
Melsungen, Germany) demonstrated in our hands their
limited applicability to high-throughput processing. We
therefore explored in detail a Precellys 24 homogenizer
(PEQLAB Biotechnology GmbH, Germany) equipped with
an integrated cooling unit. Different homogenization tubes
with various beads in various sizes are available for this
instrument. According to the manufacturer, tubes with
ceramic beads with a diameter of 1.4 mm are recom-
mended for soft tissue types like brain, liver, kidney and
more. In addition we have also tested homogenization
tubes with ceramic beads (2.8 mm diameter) and with
mixed beads of both sizes, which are recommended for
hard tissues like ear, lung, heart, spleen, tail, cornea, artery,
and plant material.
Tissue protocol for HTP metabolomics
In several pilot tests we tried different homogenization
intervals, extraction solvents and volumes. Among others,
methanol, 10 mm phosphate buffer (pH 7.5), and mixtures
of both with 30, 50, and 70% MeOH were tested on bovine
muscle and mouse liver samples. We also tested seven
solvent systems (10 mM phosphate buffer, methanol, eth-
anol, ethanol/phosphate buffer (85/15 v/v), methanol/
phosphate buffer (85/15 v/v), ethanol/dichloromethane (1/
1 v/v), and a two-step extraction with ethanol/dichloro-
methane (1/1 v/v) in the first step and 10 mM phosphate
buffer in the second step) on bovine liver samples with and
without additional shaking (5 min) of the homogenates at
4°C and room temperature.
As next step we investigated three types of bovine tissue
(liver, muscle, and fat) with the three best solvents from the
pilot tests: methanol, 10 mM phosphate buffer, and etha-
nol/phosphate buffer (85/15 v/v). All samples of one tissue
type came from the same tissue preparation from the same
animal. To explore the reproducibility of the homogeni-
zation and the kit based FIA-MS analysis, every tissue type
was homogenized three times with every solvent and each
homogenate was analyzed on three wells of the 96 well kit
plate. Finally, we applied the same protocol to 5 different
mouse tissues (liver, kidney, muscle, fat, and brain). Every
tissue type was homogenized with methanol and 10 mM
phosphate buffer and each homogenate was analyzed six
times.
The following optimized homogenization protocol was
used for all tissue types: Frozen samples (50–150 mg) were
placed into pre-cooled (dry ice) 2 ml homogenization tubes
containing ceramic beads (1.4 mm diameter). Pre-cooled
extraction solvents (methanol, 10 mM phosphate buffer)
were added (6 ll/mg for liver and kidney tissue, 3 ll/mg
for other tissues), and the tissue was homogenized three
times for 20 s at 5,500 rpm, with 30 s intervals (to ensure
freezing temperatures in sample vials) between the
homogenization steps. The tubes were subsequently cen-
trifuged for 5 min at 10,0009 gav at room temperature and
10 ll of the supernatants were loaded onto the kit.
2.3 MS measurements
The AbsoluteIDQTM kit was prepared as described by
manufacturer. Shortly, following sample preparation steps
were performed on a Hamilton ML Star robotics system
(Hamilton Bonaduz AG, Bonaduz, Switzerland): (a) pipet-
ting of 10 ll tissue extract onto the filter inserts of the 96
well kit plate (containing stable isotope labeled internal
standards), (b) drying of samples under nitrogen stream,
(c) derivatization of amino acids with 5% phenylisothio-
cyanate reagent (PITC), (d) drying of samples, (e) extrac-
tion of metabolites and internal standards with 5 mM
ammonium acetate in methanol, (f) centrifugation through
filter membrane, (g) dilution with MS running solvent. The
final extracts were analyzed using an API 4000TM triple
quadrupole mass spectrometer (ABSciex) equipped with an
Agilent 1200 Series HPLC and a HTC PAL auto sampler
from CTC controlled by the software Analyst 1.5. The
standard flow injection method comprising two 20 ll
injections (one for positive and one for negative electro-
spray ionisation mode) was applied for all measurements.
Quantification was achieved by multiple reaction moni-
toring (MRM) detection in combination with the use of
stable isotope-labeled and other internal standards as
described (Bogumil et al. 2008). Data evaluation for
quantification of metabolite concentrations was performed
with the MetIQTM software package (integral part of the
AbsoluteIDQTM kit). Concentrations of all metabolites in
the tissue extracts are initially calculated in lM. The
method has been proven to be in conformance with FDA-
Guidelines (U.S. Department of Health and Human Ser-
vices 2001), which imply proof of reproducibility within a
given error range. Analytical specifications for LOD and
evaluated quantification ranges, further LOD for semi-
quantitative measurements, identities of quantitative and
semiquantitative metabolites, specificity, potential inter-
ferences, linearity, precision and accuracy, reproducibility
and stability were described in Biocrates manual AS-P150.
The LODs were set to three times the values of zero
samples (for tissue assays methanol or 10 mM phosphate
buffer with internal standards was defined as zero sample).
The LLOQ (lower limit of quantification) and ULOQ
(upper limit of quantification) were determined experi-
mentally by Biocrates AG (Innsbruck, Austria).
Depending on the tissue type and the subsequent dilu-
tion factor, the concentrations received from MetIQ have to
be multiplied by 4 (muscle, brain, and fat) or 7 (liver and
kidney) to get the tissue specific concentration in nmol/g,
assuming that 1 mg tissue is equivalent to 1 ll volume.
2.4 Metabolite denomination
Quantification of total 163 metabolites was attempted: 40
acylcarnitines (Cx:y), free carnitine (C0), 14 amino acids, 1
sum of hexoses (H1), 92 glycerophospholipids (lys-
ophosphatidylcholines (lysoPC) and phosphatidylcholines
(PC)), and 15 sphingolipids (SM). The nomenclature used
for lipid metabolites refers to the Lipid Maps comprehen-
sive classification system (Fahy et al. 2007). Since only the
total composition of the lipid species are determined and
side chain and substitution regio- and stereochemistry is
unknown, abbreviations Cx:y are used where x and y refer
to the total number of carbons and double bonds of all
chains. Substitutions of side chains with hydroxy- (OH),
methyl- (M), and dicarboxy- (DC) are indicated. Acyl side
chains are denoted by an a, alkyl and alkenyl residues by
123

an e (in analogy to the O- and P- prefixes in the Lipid
Maps system, when individual side chains are known).
For example ‘‘PC ae C34:1’’ denotes a glyceropho-
sphatidylcholine with an acyl (a) and an ether (e) side
chain, with 34 carbon atoms in both side chains and a
single double bond in one of them. A complete list of
metabolites and the type of quantification (quantitative or
semiquantitative) is given in Supplementary Table 1.
3 Results and discussion
3.1 Experiment design
To develop a broadly applicable method, a selection of
commonly used tissue types from mice and bovine was
tested (Table 1). Several optional parameters for homoge-
nization and extraction were evaluated in different exper-
iments which are described under material and methods. To
ensure comparability, tissue samples for a given experi-
ment were always taken from the same animal. Therefore,
the development of the homogenization method and the
tests for reproducibility as well as for different extraction
solvents were first carried out on three types of bovine
tissue (liver, muscle, and fat) with sufficient material from
one animal available. The optimized method was then
applied to mouse tissue. We tested five mouse tissues
(liver, kidney, muscle, fat, and brain) of major interest in
research mechanisms of frequent human diseases.
To minimize the alteration of the metabolome by
ongoing biochemical reactions in the sample, frozen sam-
ples have to be homogenized and directly extracted. A
frequently used approach is to grind the frozen tissue to
powder using a liquid-nitrogen cooled pestle and mortar or
a ball-mill and to extract the powder subsequently. Another
option is to homogenize a tissue solvent mixture with a
potter or a disperser. Both methods have been tested by us
and turned out to be not applicable to high-throughput
methods because of serial processing of samples and
laborious manual intervention. There are two additional
Table 1 Analyzed animal tissue types
Species Tissue ll solvent/mg tissue
Mouse Liver 6 other and fall in the range of the method’s general variance.
Mouse Kidney (without capsula) 6 The median CV over all metabolites lies at 11% with
Mouse Muscle (M. biceps femoris) 3
Mouse Fat (visceral) 3
Mouse Brain (cerebrum) 3
Bovine Liver 6
Bovine Muscle (M. semitendinosus) 3
Bovine Fat (subcutaneous) 3
123
W. Römisch-Margl et al.
drawbacks of the ‘‘classical’’ homogenization method by
grinding the frozen tissue without solvent. The first is the
hardly reproducible way of weighting the frozen powder
due to water condensation and balance scale instability.
The second is the clotting of the already powdered and still
frozen tissue when solvent is added. Even when taking
aqueous solutions, the clots were very inhomogenously
resuspended which made a further homogenization step
necessary. The homogenizer used in the current work was a
Precellys 24, which is capable of processing 24 samples
simultaneously as recently described for MALDI-TOF–MS
and NMR (Hettick et al. 2008; Verollet 2008; Wu et al.
2008). The use of pre-cooled extraction solvent, cooling by
blowing cold air (-50°C) around the tubes and short
homogenization time ensured that sample intrinsic metab-
olism was arrested during the procedure. The settings with
three repeated cycles (20 s with 30 s breaks in between),
2 ml tubes with ceramic beads (1.4 mm diameter), and
moderate agitation speed (5,500 rpm) were applicable to
all tissue types under investigation. For soft tissue types
milder conditions (e.g. kidney 2 9 20 s, 5,000 rpm) also
resulted in a smooth homogenate. Further tests showed that
the use of 2.8 mm or mixed sized ceramic beads and
subsequent shaking steps at 4°C or at room temperature do
not result in higher extraction yields.
3.2 Reproducibility of homogenization method
The reproducibility of homogenization was tested in three
independent homogenizations using bovine tissue samples
of the same animal. Each homogenate was analyzed three
times resulting in nine technical replicates. In addition, the
experiments were performed using three different extrac-
tion solvents (methanol, 10 mM phosphate buffer, and 85%
ethanol/15% 10 mM phosphate buffer) and three different
tissue types (bovine fat, bovine muscle, bovine liver) to
check potential differences due to the different textures of
the tissues. For selected metabolites from different com-
pound classes, Fig. 1 shows the values of the nine repli-
cates from methanolic tissue extraction. Mean results and
variation coefficients for all metabolites with the three
tested extraction solvents are given in supplementary
Table 2. It can be seen that results obtained for indepen-
dent homogenization experiments are very similar to each
methanolic extraction. For fat, muscle and liver tissue it is
11, 10, and 12%, respectively. We also calculated separate
CVs for repeated homogenization and repeated measure-
ment, for both effects in the three tissue types the median
CV lies in the range of 6–9%. The CVs for homogenization
are all under 20%, for measurement there were some noisy
metabolites with CV [20%, especially in the muscle
Tissue protocol for HTP metabolomics

Histidine PC aa C36:1 SM C18:0 Table 2 Extraction yield using methanol in comparison to phosphate
buffer
Metabolite Liver Kidney Muscle Brain Fat
80
Carnitine 32 62 126 57 93
Arg n. d. 10 48 25 11
Gln 58 27 96 105 49
60 Gly 60 39 89 59 43
His 49 23 79 63 31
Conc. [nmol/g]

Met 26 23 77 57 30
Orn 6 17 45 75 21
Phe 32 18 68 47 25
40 Pro 58 11 83 48 37
Ser 18 11 76 83 26
Thr 19 17 77 65 45
Trp 82 54 95 94 80
20 Tyr 32 24 76 57 40
Val 48 24 75 53 30
xLeu 38 20 75 36 24
Hexose 52 75 39 87 103

fat muscle fat muscle fat muscle n.d. not determined

The extraction yield is given in %. The medians of the concentration
Histidine PC aa C36:1 SM C18:0 values obtained in 10 mM phosphate buffer were set to 100%

Fig. 1 Scatter plot of technical replicates. Three independent

homogenizations using bovine tissue samples of the same animal
are shown (each extract loaded on three wells). Example results are
shown for the amino acid histidine, one phosphatidylcholine (PC aa
C36:1) and one sphingomyelin (SM C18:0). Explanations for
abbreviations are given in Materials and methods, empty symbols
muscle, filled symbols fat, median and average values are shown as
solid or dotted bold lines, respectively. Data points from the same
homogenization are indicated by the same color. Methanol was used
for tissue extraction. The variances for bovine liver tissue were
similar (not shown)
extracts. For distribution histograms of metabolite CVs see
supplementary Figure 1. In conclusion, a reproducible
homogenization can be achieved using this method.
3.3 Extraction solvents
For plasma sample analyses, 10 ll are loaded onto a filter
paper of the kits0 sandwich plate and dried in a stream of
nitrogen. Extraction of the metabolites is then achieved
using methanol containing 5 mM ammonium acetate.
Therefore, methanol was also the first choice for tissue
experiments. However, we also tested other solvents like
ethanol, 10 mM phosphate buffer (pH 7.5), different mix-
tures of ethanol and methanol with 10 mM phosphate
buffer, and a mixture of ethanol/dichloromethane. The
often used methanol/chloroform/water extraction (Bligh
and Dyer 1959; Folch et al. 1957; Wu et al. 2008) method
was not tested, since this method results in a biphasic
separation where each fraction needs to be dried separately.
Our aim was to develop a method suitable for high-
throughput applications thus avoiding any phase separation
and drying steps. We observed that no single solvent
worked best simultaneously for all kinds of tissues and all
metabolite classes. Furthermore, not all tissues behaved
similarly with regard to the different extraction solvents.
However, for the majority of metabolites to be analyzed
with the kit (phosphatidylcholines, lysophosphatidylcho-
lines, sphingomyelins and acylcarnitines) methanol gave
the best overall results (Supplementary Table 2). For
amino acids, free carnitine and for the sum of hexoses, the
use of 10 mM phosphate buffer resulted in clearly higher
yields. For these metabolites the extraction yields of mouse
tissues with methanol in comparison to phosphate buffer
are shown in Table 2. Although some general trends are
seen, it is also evident that the yield depends on tissue type.
Other tested solvents, like mixtures of ethanol or methanol
and phosphate buffer (85/15 or 70/30 v/v), did not result in
higher yields for the amino acids compared to the methanol
extraction, but gave lower yields for many lipids and
usually showed a greater variance.
For most tissue types, 3 ll solvent per mg tissue was
used, but for liver and kidney 6 ll per mg tissue gave better
results and 9 ll did not further improve the method. The
reason for that could be solvent saturation or ion suppres-
sion in the mass spectrometric experiment. For better
comparison, the values in the figures have been normalized
to nmol/g tissue—the simplified assumption was made that

123

1 mg frozen tissue is equivalent to a volume of 1 ll, since
the exact density of tissue samples is often unknown.
In conclusion, methanolic extraction is the preferred
method for most kit metabolites, although the yield of
amino acids, hexoses and free carnitine could be improved
by a second extraction step using 10 mM phosphate buffer.
A solvent:tissue ratio of 3:1 is considered as a good starting
point for many different tissue types, as long as matrix
effects play only a minor role.
3.4 Internal standards/ion suppression effects
To analyze the impact of different tissue types on ion
suppression and other matrix effects, tissue samples from
mouse liver, kidney, muscle, fat, and brain were homoge-
nized as described above with 3 or 6 ll/mg methanol and
10 mM phosphate buffer as extraction solvents. Each of the
extracts was analyzed six times.
The internal standards (IS) of the kit are essential for
quantification. Therefore, the signal intensities of the IS in
tissue extracts were compared to the values obtained for
human plasma and to the values of the preferred extraction
solvents (methanol and 10 mM phosphate buffer), which
were applied as zero samples (Supplementary Table 3).
The comparison provides information about ion suppres-
sion effects in tissue samples. The observed suppression
was dependent on the metabolite class. All IS for acyl-
carnitines, glycerophospholipids and sphingolipids in the
tissue samples gave higher intensities compared to plasma
samples, but lower intensities than the zero samples. A
significant increase of IS intensities of a specific MRM pair
compared to the zero samples would indicate potential
interferences of unknown compounds in the tissue extract,
but such an effect was generally not observed in the tissue
samples, only the values of free carnitine (in liver and
muscle) and of some amino acids in adipose tissue were
moderately increased.
For amino acids, the intensities of the IS were mainly in
a similar range as in plasma. However, some of the
intensities were considerably lower both in phosphate
buffer and in methanol. Especially the IS for the amino
acids arginine (Arg) and ornithine (Orn) showed signifi-
cantly lower values. Therefore, the obtained results for
these two amino acids in tissue samples should be evalu-
ated with caution. In case of liver tissue the MRM pair of
the internal standard for Arg exhibited unusually high ion
suppression. Thus, arginine concentrations in liver extracts
cannot be measured using this method.
3.5 Coefficient of variation (CV)
For the analyzed mouse tissue types, the intra-day CVs of
all metabolites above the limit of detection (LOD) were
123
W. Römisch-Margl et al.
compared. LOD is here defined by the median value of the
zero samples multiplied by three. For the amino acids and
hexoses all CVs were below 15%. The CVs of several
acylcarnitines and phospholipids were above 20%, how-
ever this was not observed consistently over all tissue types
(see Supplementary Table 4). Therefore, when starting
with tissue experiments it is recommended to perform
preliminary tests using identical tissue samples to evaluate
the CVs for the given tissue and preparation.
3.6 Comparison of different tissue types
Here, some general remarks are given regarding the met-
abolomics analysis of different tissues. In Fig. 2 the con-
centrations of all measured metabolites in the mouse and
bovine tissue samples are given in a heatmap illustration
with a logarithmic color scale. Generally, in most tissue
types the sum of hexoses was the most abundant analyte
followed by amino acids, free carnitine and some phos-
pholipids. Only in mouse kidney and brain the amino acids
glycine and glutamine, respectively, were the most abun-
dant metabolites. Several acylcarnitines showed very low
concentrations in the range or even below LOD (see Fig. 2
and supplementary Table 4). On the other hand distinct
high concentrations of free carnitine and acetylcarnitine
were found in bovine muscle in comparison to other tissues
or to mouse muscle. In Fig. 3 and 4, concentrations of
some representative metabolites in the five different mouse
tissues are shown. It can be seen that the concentration
ranges of the chosen metabolites vary considerably in
different tissues.
3.7 Amino acids
For most of the amino acids, highest concentrations were
found in liver and kidney tissue, and lowest concentrations
were found in fat tissue, as shown for phenylalanine. One
exception was glutamine, for which the concentration was
highest in brain tissue, where glutamate generated from
glutamine plays an important role as a neurotransmitter
(Fig. 3).
3.8 Acylcarnitines
In general, a high number of acylcarnitines were above
LOD in the investigated tissue samples with the exception
of fat tissue, in which only few of them were above LOD.
For free carnitine (C0) and acetylcarnitine (C2), relative
high concentrations were found in muscle tissue (Fig. 4).
This molecules are used in fatty acid transport between
cytosol and mitochondria where b-oxidation takes place
(Stephens et al. 2007). In bovine muscle, the concentrations
were much higher compared to mouse muscle, and lay in
Tissue protocol for HTP metabolomics

Fig. 2 Heat map: For each metabolite and tissue type the mean concentrations (nmol/g) are color coded on a logarithmic scale. Most
concentrated metabolites in all tissue types are hexoses, amino acids, free carnitine and some phospholipids

the lmol/g range. In such cases, if the concentrations were
outside of the evaluated quantification ranges (defined in
the analytical kit specifications for plasma), values would
be considered as semi-quantitative. Alternatively, dilution
experiments can be performed to test if the assay is linear
up to very high concentration values. Therefore extracts
from bovine muscle tissue were diluted to the normal
concentration in human plasma in 4 steps (50, 25, 12.5, and
6.25% of original concentration) and were analyzed sepa-
rately. The concentrations of both carnitine and acetylcar-
nitine showed linear behavior up to lmol/g level as shown
in Fig. 5.

123
 W. Römisch-Margl et al.

Glutamine C2
Muscle Muscle

Liver Liver

Kidney Kidney

Fat Fat

Brain Brain

1000 2000 3000 4000 5000 6000 7000 20 40 60 80 100

Phenylalanine Muscle PC aa C34:2

Muscle

Liver Liver

Kidney Kidney

Fat Fat

Brain Brain

100 200 300 400 0 500 1000 1500 2000 2500 3000

Concentration [nmol/g]
Fig. 3 Box plots showing concentration values of the amino acids
glutamine and phenylalanine in different mouse tissues. The median
value is shown as a solid line (n = 6). The whiskers extend to the
extreme data points, which are no more than 1.5 times the
interquartile range from the box. Outliers beyond that range are
shown as dots. 10 mM phosphate buffer was used as extraction
solvent
3.9 Lipids
Generally, the concentrations of phosphatidylcholines,
lyso-phosphatidylcholines and sphingomyelins were high-
est in kidney and liver tissue as shown for two represen-
tative examples in Fig. 4. However, there were several
exceptions, in which certain lipids were higher in brain,
e.g. SM C18:1, lyso PC C18:1, or muscle tissue, e.g. SM
C18:0, PC aa C32:1–3, PC ae C38:0. The concentrations of
many lipids were lowest in fat tissue since most of the
lipids in fat are stored in the form of triglycerides, a class of
metabolites that is not analyzed by the used kit.
4 Concluding remarks
The tissue extraction protocol that we developed and ver-
ified here is compatible with the high-throughput targeted
metabolomics AbsoluteIDQTM kit. We show that the
metabolite panel of the kit, originally validated for plasma,
is also well applicable to animal tissue. In the investigated
tissue types, the majority of the metabolites were found in
concentrations clearly above LOD and such data should
Muscle lysoPC a C16:0
Liver
Kidney
Fat
Brain
50 100 150 200
Concentration [nmol/g]
Fig. 4 Box plots showing concentrations of selected metabolites in
different mouse tissues. The median value is represented by a solid
line (n = 6). For descriptions of whiskers and box area see legend to
Fig. 3. Methanol was used as extraction solvent
give an excellent overview of the metabolic status in dif-
ferent tissues. Methanol as extraction solvent for tissues is
recommended due to the best results for most kit metabo-
lites. In cases where the yield of amino acids/hexoses is of
high interest, a second separate tissue extraction step using
10 mM phosphate buffer should be applied.
After some experiment-specific tests regarding homog-
enization reproducibility and matrix effects, the provided
extraction setup could easily be adapted to other tissue
types and/or other species. The obtained tissue extracts
may also be used for complementary high-throughput
analyses involving LC–MS, GC–MS or NMR measure-
ments. Extraction of a few hundred samples a day becomes
possible if tissue samples are directly provided in the

123
Tissue protocol for HTP metabolomics
Carnitine
4000
3200
2400
nmol/g
1600
800
0 and Physiology, 37, 911–917.
0 25% 50% 75% 100%
percent dilution of starting extract
Acetylcarnitine
1200
1000
800
nmol/g
600
400
200
0 Thiery, J., et al. (2007). Standardized peptidome profiling of
0 25% 50% 75% 100%
percent dilution of starting extract
Fig. 5 Linearity of quantification. Even at high concentrations the
analytes can be measured. Example for very high concentration of
free carnitine and acetylcarnitine in bovine muscle is shown. Dilution
experiment of muscle extract shows good linearity up to these high
concentrations. The dashed lines indicate the upper limit of quanti-
tation as given in the analytical kit specifications
homogenization vials, due to parallelization of 24 prepa-
rations in a single run on a single instrument.
The developed method can be used for extraction setup
for different indications like studies of metabolic syn-
drome, obesity, diabetes or cardiovascular disorders where
tested tissues are targets for metabolic dysfunction of
homeostasis.
Acknowledgments We thank Dr. Christa Kühn from the Leibniz
Institute for Farm Animal Biology (Dummerstorf, Germany) for
bovine tissue samples. We thank Gabriele Zieglmeier for mouse tis-
sue preparations, and Tamara Halex and Arsin Sabunchi for the
excellent assistance in metabolomic assays. We are thankful to
Dr. Gabriele Möller and Dr. Michael Urban for their advice in
experimental design and tissue homogenization procedures. This
study was supported in part by a grant from the German Federal
Ministry of Education and Research (BMBF) to the German Center
Diabetes Research (DZD e.V.) and to the research project Greifswald
Approach to Individualized Medicine (GANI_MED).
References
Altmaier, E., Kastenmüller, G., Römisch-Margl, W., Thorand, B.,
Weinberger, K. M., Adamski, J., et al. (2009). Variation in the
human lipidome associated with coffee consumption as revealed
by quantitative targeted metabolomics. Molecular Nutrition &
Food Research, 53, 1357–1365.
Altmaier, E., Ramsay, S. L., Graber, A., Mewes, H. W., Weinberger,
K. M., & Suhre, K. (2008). Bioinformatics analysis of targeted
metabolomics—Uncovering old and new tales of diabetic mice
under medication. Endocrinology, 149, 3478–3489.
Bligh, E. G., & Dyer, W. J. (1959). A rapid method of total lipid
extraction and purification. Canadian Journal of Biochemistry
Bogumil, R., Koal, T., Weinberger, K. M., & Dammeier, S. (2008).
Massenspektrometrische Analyse von Blutplasma im Kitformat.
Laborwelt, 2, 17–23.
Denkert, C., Budczies, J., Weichert, W., Wohlgemuth, G., Scholz, M.,
Kind, T., et al. (2008). Metabolite profiling of human colon
carcinoma—Deregulation of TCA cycle and amino acid turn-
over. Molecular Cancer, 7, 72.
Deprez, S., Sweatman, B. C., Connor, S. C., Haselden, J. N., &
Waterfield, C. J. (2002). Optimisation of collection, storage and
preparation of rat plasma for 1H NMR spectroscopic analysis in
toxicology studies to determine inherent variation in biochemical
profiles. Journal of Pharmaceutical and Biomedical Analysis,
30, 1297–1310.
Fahy, E., Sud, M., Cotter, D., & Subramaniam, S. (2007). LIPID
MAPS online tools for lipid research. Nucleic Acids Research,
35, W606–W612.
Fiedler, G. M., Baumann, S., Leichtle, A., Oltmann, A., Kase, J.,
human urine by magnetic bead separation and matrix-assisted
laser desorption/ionization time-of-flight mass spectrometry.
Clinical Chemistry, 53, 421–428.
Folch, J., Lees, M., & Sloane Stanley, G. H. (1957). A simple method
for the isolation and purification of total lipides from animal
tissues. Journal Biological Chemistry, 226, 497–509.
Freedman, D. M., Chang, S. C., Falk, R. T., Purdue, M. P., Huang, W.
Y., McCarty, C. A., et al. (2008). Serum levels of vitamin D
metabolites and breast cancer risk in the prostate, lung,
colorectal, and ovarian cancer screening trial. Cancer Epidemi-
ology, Biomarkers and Prevention, 17, 889–894.
Fuchs, H., Gailus-Durner, V., Adler, T., Pimentel, J. A., Becker, L.,
Bolle, I., et al. (2009). The German mouse clinic: A platform for
systemic phenotype analysis of mouse models. Current Phar-
maceutical Biotechnology, 10, 236–243.
Gailus-Durner, V., Fuchs, H., Adler, T., Aguilar Pimentel, A., Becker,
L., Bolle, I., et al. (2009). Systemic first-line phenotyping.
Methods in Molecular Biology, 530, 463–509.
Gieger, C., Geistlinger, L., Altmaier, E., Hrabe de Angelis, M.,
Kronenberg, F., Meitinger, T., et al. (2008). Genetics meets
metabolomics: A genome-wide association study of metabolite
profiles in human serum. PLoS Genetics, 4, e1000282.
Griffin, J. L., & Kauppinen, R. A. (2007). A metabolomics perspective
of human brain tumours. FEBS Journal, 274, 1132–1139.
Griffin, J. L., & Nicholls, A. W. (2006). Metabolomics as a
functional genomic tool for understanding lipid dysfunction in
diabetes, obesity and related disorders. Pharmacogenomics, 7,
1095–1107.
Hettick, J. M., Green, B. J., Buskirk, A. D., Kashon, M. L., Slaven, J.
E., Janotka, E., et al. (2008). Discrimination of Aspergillus
isolates at the species and strain level by matrix-assisted laser
desorption/ionization time-of-flight mass spectrometry finger-
printing. Analytical Biochemistry, 380, 276–281.
123

Illig, T., Gieger, C., Zhai, G., Römisch-Margl, W., Wang-Sattler, R.,
Prehn, C., et al. (2010). A genome-wide perspective of genetic
variation in human metabolism. Nature Genetics, 42, 137–141.
Koal, T., & Deigner, H. P. (2010). Challenges in mass spectrometry
based targeted metabolomics. Current Molecular Medicine, 10,
216–226.
Kratzsch, J. & Ceglarek, U. (2010). Preclinical challenges in steroid
analysis of human samples. Journal of Steroid Biochemistry and
Molecular Biology, in press.
Kühn, C., Bellmann, O., Voigt, J., Wegner, J., Guiard, V., & Ender,
K. (2002). An experimental approach for studying the genetic
and physiological background of nutrient transformation in cattle
with respect to nutrient secretion and accretion type. Archives
Animal Breeding, 45, 317–330.
Saloniemi, T., Welsh, M., Lamminen, T., Saunders, P., Makela, S.,
Streng, T., et al. (2009). Human HSD17B1 expression mascul-
inizes transgenic female mice. Molecular and Cellular Endo-
crinology, 301, 163–168.
Sreekumar, A., Poisson, L. M., Rajendiran, T. M., Khan, A. P., Cao,
Q., Yu, J., et al. (2009). Metabolomic profiles delineate potential
role for sarcosine in prostate cancer progression. Nature, 457,
910–914.
Stephens, F. B., Constantin-Teodosiu, D., & Greenhaff, P. L. (2007).
New insights concerning the role of carnitine in the regulation of
fuel metabolism in skeletal muscle. Journal of Physiology, 581,
431–444.
Urban, M., Enot, D. P., Dallmann, G., Korner, L., Forcher, V., Enoh,
P., et al. (2010). Complexity and pitfalls of mass spectrometry-
123
View publication stats
W. Römisch-Margl et al.
based targeted metabolomics in brain research. Analytical
Biochemistry, 406, 124–131.
U.S. Department of Health and Human Services, F. a. D. A., Center
for Drug Evaluation and Research (CDER), Center for Veter-
inary Medicine (CVM) (2001). Guidance for Industry. Bioana-
lytical Method Validation.
Verollet, R. (2008). A major step towards efficient sample preparation
with bead-beating. Biotechniques, 44, 832–833.
Wang-Sattler, R., Yu, Y., Mittelstrass, K., Lattka, E., Altmaier, E.,
Gieger, C., et al. (2008). Metabolic profiling reveals distinct
variations linked to nicotine consumption in humans—First
results from the KORA study. PLoS ONE, 3, e3863.
Weikard, R., Altmaier, E., Suhre, K., Weinberger, K. M., Hammon,
H. M., Albrecht, E., et al. (2010). Metabolomic profiles indicate
distinct physiological pathways affected by two loci with major
divergent effect on Bos taurus growth and lipid deposition.
Physiological Genomics, 42A, 79–88.
Wu, H., Southam, A. D., Hines, A., & Viant, M. R. (2008). High-
throughput tissue extraction protocol for NMR- and MS-based
metabolomics. Analytical Biochemistry, 372, 204–212.
Zhai, G., Wang-Sattler, R., Hart, D. J., Arden, N. K., Hakim, A. J.,
Illig, T., et al. (2009). Serum branched-chain amino acid to
histidine ratio: A novel metabolomic biomarker of knee osteo-
arthritis. Annals of the Rheumatic Diseases, 69, 1227–1231.
Zivkovic, A. M., Wiest, M. M., Nguyen, U. T., Davis, R., Watkins, S.
M., & German, J. B. (2009). Effects of sample handling and
storage on quantitative lipid analysis in human serum. Meta-
bolomics, 5, 507–516.