The effect of preanalytical factors on stability of the proteome and selected metabolites in cerebrospinal fluid (CSF).
ABSTRACT To standardize the use of cerebrospinal fluid (CSF) for biomarker research, a set of stability studies have been performed on porcine samples to investigate the influence of common sample handling procedures on proteins, peptides, metabolites and free amino acids. This study focuses at the effect on proteins and peptides, analyzed by applying label-free quantitation using microfluidics nanoscale liquid chromatography coupled with quadrupole time-of-flight mass spectrometry (chipLC-MS) as well as matrix-assisted laser desorption ionization Fourier transform ion cyclotron resonance mass spectrometry (MALDI-FT-ICR-MS) and Orbitrap LC-MS/MS to trypsin-digested CSF samples. The factors assessed were a 30 or 120 min time delay at room temperature before storage at -80 degrees C after the collection of CSF in order to mimic potential delays in the clinic (delayed storage), storage at 4 degrees C after trypsin digestion to mimic the time that samples remain in the cooled autosampler of the analyzer, and repeated freeze-thaw cycles to mimic storage and handling procedures in the laboratory. The delayed storage factor was also analyzed by gas chromatography mass spectrometry (GC-MS) and liquid chromatography mass spectrometry (LC-MS) for changes of metabolites and free amino acids, respectively. Our results show that repeated freeze/thawing introduced changes in transthyretin peptide levels. The trypsin digested samples left at 4 degrees C in the autosampler showed a time-dependent decrease of peak areas for peptides from prostaglandin D-synthase and serotransferrin. Delayed storage of CSF led to changes in prostaglandin D-synthase derived peptides as well as to increased levels of certain amino acids and metabolites. The changes of metabolites, amino acids and proteins in the delayed storage study appear to be related to remaining white blood cells. Our recommendations are to centrifuge CSF samples immediately after collection to remove white blood cells, aliquot, and then snap-freeze the supernatant in liquid nitrogen for storage at -80 degrees C. Preferably samples should not be left in the autosampler for more than 24 h and freeze/thaw cycles should be avoided if at all possible.
- [Show abstract] [Hide abstract]
ABSTRACT: Early detection is the most effective way to improve the clinical outcome of malignancies. Although some tumor markers are now widely used in the clinic, their sensitivity and specificity are still not satisfactory. Thus, there is an urgent requirement for the discovery of new tumor markers. By measuring holistic endogenous metabolites, metabolomics can be used for delineating metabolic networks and discovering metabolic markers. Chromatography-mass spectrometry is the most widely used tool for metabolomics and has shown great potential for biomarker screening. In this review, the authors summarize: recent advances in the protocols and methodologies of chromatography-mass spectrometry-based metabolomics in the discovery of tumor markers; recently identified tumor metabolic markers for primary liver cancer, gynecologic cancer and genitourinary cancer and their applications; and commonly encountered problems in the translational research of metabolic markers.Expert Review of Molecular Diagnostics 05/2013; 13(4):339-48. · 4.09 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Information regarding the variability of metabolite levels over time in an individual is required to estimate the reproducibility of metabolite measurements. In intervention studies, it is critical to appropriately judge changes that are elicited by any kind of intervention. The pre-analytic phase (collection, transport and sample processing) is a particularly important component of data quality in multi-center studies. Reliability of metabolites (within-and between-person variance, intraclass correlation coefficient) and stability (shipment simulation at different temperatures, use of gel-barrier collection tubes, freeze-thaw cycles) were analyzed in fasting serum and plasma samples of 22 healthy human subjects using a targeted LC-MS approach. Reliability of metabolite measurements was higher in serum compared to plasma samples and was good in most saturated short-and medium-chain acylcarnitines, amino acids, biogenic amines, glycerophospholipids, sphingolipids and hexose. The majority of metabolites were stable for 24 h on cool packs and at room temperature in non-centrifuged tubes. Plasma and serum metabolite stability showed good coherence. Serum metabolite concentrations were mostly unaffected by tube type and one or two freeze-thaw cycles. A single time point measurement is assumed to be sufficient for a targeted metabolomics analysis of most metabolites. For shipment, samples should ideally be separated and frozen immediately after collection, as some amino acids and biogenic amines become unstable within 3 h on cool packs. Serum gel-barrier tubes can be used safely for this process as they have no effect on concentration in most metabolites. Shipment of non-centrifuged samples on cool packs is a cost-efficient alternative for most metabolites.PLoS ONE 01/2014; 9(2):e89728. · 3.53 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Discovery of clinically relevant biomarkers for diseases has revealed metabolomics has potential advantages that classical diagnostic approaches do not. The great asset of metabolomics is that it enables assessment of global metabolic profiles of biofluids and discovery of biomarkers distinguishing disease status, with the possibility of enhancing clinical diagnostics. Most current clinical chemistry tests rely on old technology, and are neither sensitive nor specific for a particular disease. Clinical diagnosis of major neurological disorders, for example Alzheimer's disease and Parkinson's disease, on the basis of current clinical criteria is unsatisfactory. Emerging metabolomics is a powerful technique for discovering novel biomarkers and biochemical pathways to improve diagnosis, and for determination of prognosis and therapy. Identifying multiple novel biomarkers for neurological diseases has been greatly enhanced with recent advances in metabolomics that are more accurate than routine clinical practice. Cerebrospinal fluid (CSF), which is known to be a rich source of small-molecule biomarkers for neurological and neurodegenerative diseases, and is in close contact with diseased areas in neurological disorders, could potentially be used for disease diagnosis. Metabolomics will drive CSF analysis, facilitate and improve the development of disease treatment, and result in great benefits to public health in the long-term. This review covers different aspects of CSF metabolomics and discusses their significance in the postgenomic era, emphasizing the potential importance of endogenous small-molecule metabolites in this emerging field.Analytical and Bioanalytical Chemistry 05/2013; · 3.66 Impact Factor
The Effect of Preanalytical Factors on Stability of the Proteome and
Selected Metabolites in Cerebrospinal Fluid (CSF)
Therese Rosenling,†Christiaan L. Slim,†Christin Christin,†Leon Coulier,‡Shanna Shi,⊥
Marcel P. Stoop,§Jan Bosman,†Frank Suits,#Peter L. Horvatovich,†Norbert Stockhofe-Zurwieden,∇
Rob Vreeken,|Thomas Hankemeier,|Alain J. van Gool,OTheo M. Luider,§and Rainer Bischoff*,†
Analytical Biochemistry, Department of Pharmacy, University of Groningen, Groningen, The Netherlands, TNO, Quality
of Life, Zeist, The Netherlands, Department of Neurology, Erasmus University Medical Center, Rotterdam, The
Netherlands, Analytical BioSciences, Leiden/Amsterdam Centre for Drug Research, Leiden University, Leiden, The
Netherlands, Netherlands Metabolomics Centre, Leiden/Amsterdam Centre for Drug Research, Leiden University, Leiden,
The Netherlands, IBM TJ Watson Research Centre, Yorktown Heights, New York 10504, Animal Sciences Group,
Wageningen UR, Division Infectious Diseases, Lelystad, The Netherlands, and Schering-Plough, Oss, The Netherlands
Received July 5, 2009
To standardize the use of cerebrospinal fluid (CSF) for biomarker research, a set of stability studies have
been performed on porcine samples to investigate the influence of common sample handling procedures
on proteins, peptides, metabolites and free amino acids. This study focuses at the effect on proteins and
peptides, analyzed by applying label-free quantitation using microfluidics nanoscale liquid chromatography
coupled with quadrupole time-of-flight mass spectrometry (chipLC-MS) as well as matrix-assisted laser
desorption ionization Fourier transform ion cyclotron resonance mass spectrometry (MALDI-FT-ICR-MS)
and Orbitrap LC-MS/MS to trypsin-digested CSF samples. The factors assessed were a 30 or 120 min time
delay at room temperature before storage at -80 °C after the collection of CSF in order to mimic potential
delays in the clinic (delayed storage), storage at 4 °C after trypsin digestion to mimic the time that samples
remain in the cooled autosampler of the analyzer, and repeated freeze-thaw cycles to mimic storage and
mass spectrometry (GC-MS) and liquid chromatography mass spectrometry (LC-MS) for changes of
metabolites and free amino acids, respectively. Our results show that repeated freeze/thawing introduced
changes in transthyretin peptide levels. The trypsin digested samples left at 4 °C in the autosampler showed
a time-dependent decrease of peak areas for peptides from prostaglandin D-synthase and serotransferrin.
Delayed storage of CSF led to changes in prostaglandin D-synthase derived peptides as well as to increased
levels of certain amino acids and metabolites. The changes of metabolites, amino acids and proteins in the
delayed storage study appear to be related to remaining white blood cells. Our recommendations are to
centrifuge CSF samples immediately after collection to remove white blood cells, aliquot, and then snap-
freeze the supernatant in liquid nitrogen for storage at -80 °C. Preferably samples should not be left in the
autosampler for more than 24 h and freeze/thaw cycles should be avoided if at all possible.
Keywords: Cerebrospinal Fluid (CSF) • stability • proteomics • metabolomics • mass spectrometry
In the search for molecular biomarkers for disorders of the
central nervous system (CNS), such as Alzheimer’s Disease (AD)
or Multiple Sclerosis (MScl), cerebrospinal fluid (CSF) is the body
fluid of choice because of its proximity to the diseased tissue and
its relative ease of sampling.1–9To obtain reliable results and
reduce the risk of detecting false biomarker candidates, it is
important to handle samples properly and to control for the effect
of preanalytical parameters on proteins or metabolites. Earlier a
cleavage product of cystatin C was reported as a possible biom-
This observation emphasizes that keeping control over factors
such as storage conditions, number of freeze/thaw cycles, and
the collection of CSF itself is critical.
Studies on urine and plasma or serum have shown that
internal (protease activity, blood contamination and protein
concentration) as well as external factors (storage, handling,
* To whom correspondence should be addressed. Prof. Dr. Rainer Bischoff,
Antonius Deusinglaan 1, Postbus 196, 9700 AD Groningen, The Netherlands.
Tel: +31 50 3633338. Fax: +31 50 3637582. E-mail: email@example.com.
†University of Groningen.
‡TNO, Quality of Life.
⊥Netherlands Metabolomics Centre, Leiden/Amsterdam Centre for Drug
Research, Leiden University.
§Erasmus University Medical Center.
#IBM TJ Watson Research Centre.
∇Animal Sciences Group, Wageningen UR.
|Analytical BioSciences, Leiden/Amsterdam Centre for Drug Research,
10.1021/pr9005876 CCC: $40.75
2009 American Chemical Society
Journal of Proteome Research 2009, 8, 5511–5522 5511
Published on Web 10/22/2009
and analysis method) crucially affect the proteome profile.13–17
There are a few reports on the stability of single proteins in
CSF connected to AD18–21as well as on the effect of storage
temperature (4 or 23 °C) on the proteome profile of CSF by
analyzing intact proteins and peptides with surface-enhanced
laser desorption/ionization time-of-flight mass spectrometry
Two studies on canine CSF report the effect of storage
temperature on protein concentration, glucose level, pH and
enzyme activity23as well as the effect of storage temperature,
with or without addition of stabilizing agent, on the cell count.24
Furthermore, guidelines have been published concerning pre-
analytical factors that should be examined (e.g., cytological
examination and glucose concentration measurement) leading
to recommendations for proteomic CSF analysis.25–27
Many of the studies evaluating the influence of preanalytical
factors on protein composition in body fluids have been
performed by SELDI-TOF-MS,15,16,22,28a method that suffers
from rather poor concentration sensitivity and that may be
prone to mass spectrometric artifacts.29More sensitive ap-
proaches using enrichment of proteins and peptides on
magnetic bead separators or separation by Liquid Chromatog-
raphy (LC) followed by Matrix-Assisted Laser Desorption
Ionization Time-Of-Flight Mass Spectrometry (MALDI-TOF-
MS) have also indicated that conditions of sample handling
and preparation are critical.14,30–36
In a study conducted by Anesi et al., the level of amino acids
(Glu, Gly, Asp and Tau) was analyzed with respect to different
storage conditions (-20 and -80 °C) after a number of
pretreatments (deproteinization, neutralization, no treat-
ment).37This study indicated changes in the amino acid
concentrations in samples that were not treated prior to long-
term storage. Another study by Levine et al. reported the effect
of storage at room temperature (for 72 h) on CSF metabolites.31
In this study, the samples had been stored at -70 °C for 30
months prior to analysis by NMR. NMR analysis revealed
changes in metabolite concentration with both increasing and
decreasing concentration levels.
In this paper, we describe a study into the effects of
preanalytical factors on the porcine CSF proteome and me-
tabolome using a variety of techniques comprising liquid
chromatography mass spectrometry (LC-MS), gas chromatog-
raphy mass spectrometry (GC-MS) and matrix-assisted laser
desorption ionization Fourier transform ion cyclotron reso-
nance mass spectrometry (MALDI-FT-ICR-MS).
2. Material and Methods
2.1. Sample Set. Porcine CSF was collected at the Animal
Sciences Group, Wageningen University, Division of Infectious
Diseases, Lelystad, The Netherlands. Five conventional pigs (ca.
100 kg weight, males) were euthanized and CSF was sampled
from the cerebromedullary cistern of the subarachnoid space
in the cervical region directly after euthanasia. Euthanasia was
performed by intravenous injection of T61 (Embutramid) (pig
3 [P3] and pig 7 [P7]) or pentobarbital (pig 2 [P2], pig 4 [P4]
and pig 5 [P5]) followed by exsanguination. Samples (∼10 mL)
were taken under mild suction (using a syringe with a 22 G
needle), transferred to a 50 mL Falcon tube and immediately
transported to the laboratory where contamination with red
blood cells (RBC) and white blood cells (WBC) was assessed
with an F-800 cell counter (Sysmex). The CSF samples from
P2 contained 40 RBC/µL and 40 WBC/µL, P4 contained 40 RBC/
µL and 60 WBC/µL and P5 had 120 RBC/µL and 40 WBC/µL
before centrifugation (10 min, 1500g). RBC or WBC were
undetectable after centrifugation. The CSF from P3 and P7
showed no contamination with RBC but both contained 100
WBC/µL (samples were not centrifuged).
Aliquots of samples (1.5 mL) were directly snap-frozen in
liquid nitrogen (T0). Further aliquots were left for 30 min (T30)
or 2 h at room temperature (T120) before being frozen in liquid
nitrogen and stored at -80 °C. The aliquots were stored in 2
For studying the effect of WBC contamination on stability at
room temperature between CSF collection and freezing, samples
T0, T30 and T120 of P3, P7 (not centrifuged) and P2, P4 and
P5 (centrifuged) were compared. For the freeze/thaw study
sample, T0 from P3 was used, and for evaluating stability of
trypsin-digested CSF at 4 °C, T0 from P2 and P4 was used. The
total protein concentration of the samples was measured using
the Micro BCA Assay (Pierce, Rockford, IL). For proteomics
analysis, CSF was depleted of the six most abundant serum
proteins and digested with trypsin (see section 2.3). The mean
protein concentration in the nondepleted samples was 870 ng/
µL, and in the depleted samples, it was 50 ng/µL (Relative
Standard Deviation (RSD): <10% measured in triplicate).
A quality control (QC) sample was prepared to control for
analytical variability during the chipLC-MS analyses. This
sample consisted of pooled, depleted, trypsin-digested porcine
CSF spiked with digested horse heart cytochrome C (Fluka, part
# 30396) (final concentration 100 fmol/µL). Prior to the study,
all porcine CSF samples were exposed to two freeze/thaw cycles
(sampling/storage and depletion/storage). After this, the samples
were exposed to at most two further freeze/thaw cycles before
analysis (aliquoting and storage between digestion and analy-
After depletion, the CSF was stored, and all reactions were
carried out in 1.5 or 2 mL reaction tubes (Greiner Bio-One, part
# 616 201/623 201). After digestion, the samples were trans-
ferred to plastic inserts (Waters, Milford, MA, part # WAT094170)
inside glass vials with screw caps (Agilent Technologies, Santa
Clara, CA, part # 5187-0716 + 5182-0717).
2.2. Design of Study. In this study, three preanalytical
parameters were assessed: (1) Delayed storage after sample
collection; porcine CSF was frozen at three different time points
after sampling (0, 30, 120 min at room temperature without
centrifugation or after centrifugation). The proteome profile
was first assessed on the chipLC-MS platform and subsequently
by MALDI-FT-ICR-MS in a separate laboratory. Furthermore,
the metabolome profile was assessed using two different
analytical platforms in two separate laboratories (see sections
2.4-2.8). (2) Number of freeze/thaw cycles of depleted CSF;
the proteome profile of depleted porcine CSF exposed to
different numbers of freeze/thaw cycles (control, 1×, 2×, 3×,
4×, 5× and 10× cycles) was assessed by chipLC-MS. (3) Storage
of depleted/digested CSF at 4 °C; after depletion and digestion,
CSF was left at 4 °C for up to 1 month (control, 1 day, 1 week
and 1 month) and the effect was assessed by chipLC-MS.
2.3. Sample Preparation (Proteomic Analysis). For pro-
teomic analysis, all CSF samples were depleted of 6 highly
abundant proteins; albumin, IgG (immunoglobulin G), IgA
(immunoglobulin A), haptoglobin, transferrin and R-1-Anti-
trypsin by immunoaffinity chromatography. Prior to depletion,
200 µL of porcine CSF was concentrated 20 times on an
Ultrafree -0.5 centrifugal filter device (5 kDa cutoff, Millipore,
Billerica, MA) by centrifugation for 20 min at 12 000 rpm
(11 290g) at 4 °C. After centrifugation, ∼10 µL of concentrated
Rosenling et al.
5512Journal of Proteome Research • Vol. 8, No. 12, 2009
CSF (retentate of >5 kDa) remained on the filter. This retentate
was diluted in Buffer A (Agilent, part # 5185-5987) to a final
volume of 135 µL at 4 °C, transferred to a 0.22 µm cellulose
acetate spin filter (Agilent, part # 5185-5990), and centrifuged
for 20 min at 13 000 rpm (13 250g) (4 °C) to remove particulates.
The filtrate was transferred to a sample vial (Agilent, part #
5182-0716 + blue screw cap 5182-0717) with a 300 µL insert
(Waters, part # WAT094170) and 110 µL of this was injected on
a Multiple Affinity Removal column, Hu-6 (Human), 4.6 × 50
mm, (Agilent, part # 5185-5984) using an A¨KTA FPLC system
(Amersham Biosciences, Uppsala, Sweden) with detection at
280 nm. The system was connected to a cooled autosampler
(4 °C) and fraction collector. The following elution scheme was
used: 0-10 min, 100% buffer A (0.25 mL/min); 10-13.5 min,
100% buffer B (part # 5185-5988, Agilent) (1 mL/min); 13.5-20
min, 100% buffer A (0.25 mL/min). The flow-through fraction
(depleted CSF collected between 2.8-5.3 min) of a total volume
of ∼0.6 mL was collected. The depleted protein fraction
(containing the 6 depleted proteins) was also collected between
11.6 and 13.2 min in a total volume of ∼1.6 mL. Both fractions
were snap frozen in liquid nitrogen and stored at -80 °C until
The digestion of CSF for proteomics analysis was done
according to the following procedure: 30 µL of CSF and 30 µL
of 0.1% RapiGest (in 50 mM ammonium bicarbonate) (Waters)
were added to the same sample tube (Greiner Bio-One, Alphen
aan den Rijn, The Netherlands, part # 623201). The sample was
reduced by adding 0.6 µL of 1,4-dithiothreitol (DTT) (0.5 M)
followed by incubation at 60 °C for 30 min After cooling to
room temperature, the sample was alkylated with 3 µL of
iodoacetamide (IAM) (0.3 M) in the dark for 30 min at room
temperature and 0.6 µL of sequencing grade modified trypsin
(Promega, Madison, WI, part # V5111) (125 ng/µL) was added
to give a trypsin-to-protein ratio of 1:20 (w/w). The sample was
incubated for ∼16 h at 37 °C with slight vortexing (450 rpm) in
a thermomixer comfort (Eppendorf). Thereafter, 6 µL of
hydrochloric acid (0.5 M) was added to stop the digestion
followed by incubation for 30 min at 37 °C to hydrolyze the
RapiGest. The sample was centrifuged at 13 000 rpm (13 250g)
for 10 min at 4 °C to remove the insoluble part of the
hydrolyzed RapiGest and any particulates that might block the
LC-chip. A total of 60 µL of the liquid phase was transferred to
six screw cap vials (Agilent) with 300 µL inserts (Waters) in 10
µL aliquots for chipLC-MS analysis.
2.4. ChipLC-MS(/MS) Proteomic Analysis. A quadrupole-
time-of-flight (QTOF) mass spectrometer (Agilent 6510) with a
liquid chromatography-chip electrospray ionization (LC-chip
ESI) interface was coupled with a nano LC system (Agilent 1200)
with an 8 µL injection loop. The instrument was operated under
the MassHunter Data Acquisition software (version B.01.02/
B.01.03; Build 188.8.131.52; Agilent Technologies, Santa Clara, CA).
To compare peptide profiles, samples were analyzed by LC-
MS and the discriminatory peptides were identified and as-
signed to proteins by targeted LC-MS/MS. For the freeze-thaw
and the 4 °C storage studies, an LC-chip with a 40 nL trap
column was used (Protein ID chip #2; G4240-62002; Agilent
Technologies; separating column: 150 mm × 75 µm Zorbax
300SB-C18, 5 µm; trap column: 40 nL Zorbax 300SB-C18, 5 µm).
For the delayed storage study, an LC-chip with a 160 nL trap
column was used (Protein ID chip #3; G4240-63001 SPQ110:
Agilent Technologies; separating column: 150 mm × 75 µm
Zorbax 300SB-C18, 5 µm; trap column: 160 nL Zorbax 300SB-
C18, 5 µm). For LC separations, the following solutions were
used: A, ultrapure water (conductivity 18.2 MΩ, Elga Labwater,
Ede, The Netherlands) with 0.1% formic acid (98-100%, pro
analysis, Merck, Darmstadt Germany); B, acetonitrile (HPLC-S
gradient grade, Biosolve, Valkenswaard, The Netherlands) with
0.1% formic acid. Eight microliters of depleted, digested sample
(section 2.3) was injected on the trap column at a flow rate of
3 µL/min (3% B). After ∼3 min, the trapped peptides were
transferred to the separation column at a flow rate of 250 nL/
min. For the freeze/thaw and the 4 °C storage studies, peptides
were eluted from the separation column using the following
program: 80 min linear gradient from 3 to 50% B, 5 min linear
gradient from 50 to 70% B, 20 min linear gradient from 70 to
3% B. For the delayed storage study, the gradient program was
set up as follows: 80 min linear gradient from 3 to 40% B; 10
min linear gradient from 40 to 50% B; 10 min linear gradient
from 50 to 3% B.
The MS settings were the following: mass range, 100-2000
m/z (delayed storage 50-2000 m/z); acquisition rate, 1 spec-
trum/s (delayed storage 0.9 spectra/s); data storage, profile;
fragmentor, 175 V; skimmer, 65 V; OCT 1 RF Vpp, 750 V. The
spray voltage was ∼1800 V; drying gas (N2) temp, 325 °C; drying
gas flow, 6 L/min. Mass correction was done during analysis
using internal standards (methyl stearate m/z, 299.294457 and
HP-1221 m/z, 1221.990637) continuously evaporating from a
wetted wick inside the spray chamber.
For protein identification, selected peptide ions were frag-
mented by collision-induced dissociation. The MS/MS settings
used were the following: fragmentor, 175 V; skimmer, 65 V; OCT
1 RF Vpp, 750 V; precursor ion selection, medium (4 m/z); mass
range, 200-2000 m/z; acquisition rate MS, 4 scans/s; acquisi-
tion rate MS/MS, 3 scans/s; collision energy, 4 V/100 Da; offset,
2 V. Collision energies for targeted ions: (4 °C storage) m/z
369.188, z ) 2, 16 V; m/z 681.858, z ) 2, 27 V; (delayed storage)
m/z 638.655, z ) 3, 25 V; m/z 921.443, z ) 4, 30 V; m/z 957.479,
z ) 2, 35 V. In the freeze-thaw study fragmentation spectra
were obtained by auto-MS/MS; precursor setting, maximum 3
precursors/cycle; absolute threshold, 1000; relative threshold,
0.01% of the most intense peak; active exclusion enabled after
2 selections, release of active exclusion after 0.15 min, precur-
sors sorted by abundance only. The MS/MS files were stored
in centroid mode with a threshold of MS, 300/0.01 (absolute/
relative threshold); MS/MS, 100/0.01.
To assess the reproducibility of the analytical platform and
exclude method-related changes in peak areas, each sample
was analyzed five times in a randomized order with blank runs
and QC samples between every fifth or 10th sample injection
(depending on the total number of samples analyzed in one
batch). Each sample was divided over 6 vials. Only one injection
was done from each vial to avoid changes caused by changing
surface to volume ratios Each sample was left in the autosam-
pler for no more than 24 h at 4 °C. The QC samples were
injected in order to calculate the repeatability of the method
over the course of the analysis in terms of mass accuracy,
retention time and peak area. Selected cytochrome C peptides
were smoothed (Gaussian function: width 15 points (1.08 s
between each data point) and integrated for calculation of the
relative standard deviation (RSD) with respect to peak area and
retention time. The peak area RSD was within (20% and the
retention time shifts were less than (0.5% (7 s) for the selected
peaks. Mass accuracy was calculated as a mean of five
measurements of each selected ion and compared to the
theoretical mass of the originating peptides and was below (7
Effect of Preanalytical Factors
Journal of Proteome Research • Vol. 8, No. 12, 2009
ppm. Overall these data confirm the performance of the
chipLC-MS system as reported before.39
Prior to starting the analyses, the linear range of the chipLC-
MS system was determined in order to define the maximal
amount of digested CSF that could be injected without column
overloading (extensive peak broadening) or saturation of the
detector of the mass spectrometer (ion signal does no longer
increase with increasing injected amount) (data not shown).
The injection volume was 8 µL (∼200 ng [4 pmol]) for the CSF
stability samples and 5 µL for the QC samples (∼125 ng [2.5
pmol], 500 fmol cytochrome C) (assuming a mean protein mass
of 50 kDa). Repeatability of the analysis is depicted in Figures
1, 4 and 5 in the form of box and whisker plots.
2.5. Data Processing of ChipLC-MS(/MS) Data. The raw LC-
MS data were preprocessed using the Agilent MassHunter
Qualitative Analysis software (version B.01.02/B.01.03; Build
184.108.40.206;) (Agilent). The data was analyzed both manually and
with two in-house developed data processing pipelines.40–42
With the manual method, conditions were compared to each
other by overlaying chromatograms (total ion chromatograms,
[TIC] and base peak chromatograms, [BPC]) in the MassHunter
software. In regions with visible differences between the
conditions, extracted ion chromatograms (EICs) were compared
for peaks that were multiply charged (trypsin digested proteins
analyzed on this instrument give rise to at least doubly charged
ions). Peaks that appeared to be discriminating were smoothed
(Gaussian function: width 15 points (1.08 s between each data
point)) and integrated. Peak areas were compared using a two-
tailed Student’s t test (Microsoft Office Excel 2007) and box
plots were created in Origin 7.0.
For more extensive data analysis, two different data process-
ing workflows were used. An in-house data processing method
developed in Matlab40and a data processing approach devel-
oped in C++.42For both approaches, raw data was converted
to MzData.XML in profile mode. These files were further
converted to ASCII format with the following parameter set-
tings: peak intensity, > 300 counts; mass range, 200-1600 m/z
(no multiply charged ions were found outside this range);
retention time range, 3-80 min (elution range of the peptides).
The Matlab workflow consists of the following steps: meshing
the raw data in the m/z dimension to 1 amu mass resolution
using Gaussian smoothing followed by peak picking based on
MN-rules with M ) 3 and N ) 8.43One dimensional peak
matching was performed using a sliding window along a given
m/z trace (1 min width and 0.1 min interval step) to create an
aligned peak matrix containing the intensities of the matched
peaks from the different samples where the row corresponds
to the peak number and the column to the sample number.
In the C++ programmed data processing workflow, the data
was meshed to an evenly spaced grid with a Gaussian algorithm
Figure 1. (A) Base peak chromatograms of the chipLC-MS proteomic analysis of depleted, trypsin-digested, noncentrifuged porcine
CSF with delay times of 0, 30, or 120 min at room temperature (T0, T30, T120) between CSF collection and freezing in liquid nitrogen.
The encircled regions show peaks at ∼39 min and ∼46 min retention time that change with increasing delay time. (B) Extracted ion
chromatograms (detection window (50 ppm) of three peaks, derived from two peptides, that change with increasing delay time (see
panel A). Peaks at m/z ) 638.655 (Mw: 1912.943) and 957.479 (Mw: 1912.943) have the same retention time and represent different
charge states of the same peptide that was assigned to prostaglandin D-synthase by MS/MS. The peak at m/z ) 921.443 (Mw: 3681.743)
was also assigned to prostaglandin D-synthase by MS/MS. (C) Univariate statistical analysis of peak areas of peptides that change
significantly with respect to delay time. Analysis is based on two-tailed students t tests of 5 repetitions of the LC-MS analyses. Data are
represented as box plots with p-values (T0 vs T30 and T0 vs T120, respectively). Peak areas between T30 and T120 were not significantly
different (p > 0.05).
Rosenling et al.
5514Journal of Proteome Research • Vol. 8, No. 12, 2009
of 0.01 amu mass resolution. From these “meshed” files, the
10 000 most intense peaks were picked using a slope-based
geometric peak picking algorithm. To compare chromatograms,
peaks were time-aligned using warp2D, which uses the over-
lapping peak volume of the extracted peaks as a two-
dimensional benefit function to drive the optimization proce-
dure to obtain optimal retention time correction. Peaks from
different LC-MS runs were matched after alignment resulting
in one aligned matrix containing the average height, the average
retention time and the average m/z value for each peak.
Columns correspond to the individual samples (chromato-
grams) and rows to the matched peaks.
With the use of the aligned peak matrices from both
workflows, a double cross-validated Nearest Shrunken Centroid
(NSC) algorithm44was applied to classify samples according
to the evaluated preanalytical conditions and to detect peaks
that changed due to changing preanalytical parameters. EICs
were subsequently extracted for these peaks from the raw data
focusing on those peaks that occurred at the highest shrinkage
and lowest cross validation error. All ions in the list of selected
peaks were assessed with respect to their charge state. Only
ions with a charge of +2 or higher were further analyzed as
described above for the manual method. Class separation based
on discriminatory peaks selected by the NSC algortithm was
visualized after Principal Component Analysis (PCA).
For each peak that by univariate statistical analysis (two-
tailed Student’s t test) proved to be discriminatory between
the sample groups, based on its p-value being below 0.05
(confirmed in a second, independent set of analyses), frag-
mentation spectra were manually extracted from the obtained
MS/MS data files, and each selected spectrum was converted
to the Mz.Data.XML format and exported to the PhenyxOnline
database search tool (Geneva Bioinformatics, version 2.5/2.6;
for details about parameters see Figure S.1 in Supporting
Information). The databases searched were uniprot_sprot
(56.5_25_Nov_2008) and uniprot_sprot_rev (56.5_25_Nov_2008)
(the later database is a reverse version of uniprot_sprot and
serves to determine the false positive rate). Taxonomy, mam-
malian; scoring model, ESI-QTOF (QTOF); parent charge, +2,
+3, +4 (trust the charge). The search was done in two
subsequent rounds. The following search parameters were
common for both rounds: peptide/AC score, g5; peptide
length, g5; p-value <0.0001; enzyme, trypsin (KR); missed
cleavage, e 1; parent tolerance, 10 ppm. The following search
parameters differed between round 1 and round 2 of the search.
Round 1 amino acid modifications, Cys_CAM (carboxy methy-
lation, fixed), Oxidation_M (oxidation of methionine, variable,
e 2); Round 2, Cys_CAM (fixed), Oxidation_M (variable, e 2),
Oxidation_HW (oxidation of histidine and tryptophan, variable,
e 2), DEAMID (deamidation, variable, e 2), PHOS (phospho-
rylation, variable, e 2). The cleavage mode was set to ‘normal’
for round 1 and to ‘half cleaved’ for round 2. The collision
spectrum from each ion was manually evaluated, and only
fragments with intensities clearly above the background noise
(50-100 counts) were considered.
2.6. MALDI-FT-ICR-MS and Orbitrap LC-MS/MS Proteo-
mic Analysis. The samples were depleted and digested as
described in section 2.3. A matrix solution was prepared by
dissolving 10 mg of 2.5-dihydroxybenzoic acid (DHB) in 1 mL
0.1% TFA/water. All samples were desalted by loading them
on C18 material and washing the samples with 0.1% TFA/water.
Subsequently, the samples were eluted in 10 µL of 50%
acetonitrile/0.1% TFA in water. From each sample, 0.5 µL of
eluate was added to 0.5 µL of the matrix solution on a MALDI
target plate (600/384 AnchorChip with transponder plate,
Bruker Daltonics, Germany). All samples were manually spotted
in duplicate. After drying at room temperature, the samples
were measured manually on an APEX IV Qe 9.6 T MALDI-FT-
ICR mass spectrometer (Bruker Daltonics) equipped with the
first version of the vacuum Combisource, and a 20 Hz nitrogen
laser with an irradiation area of ∼200 µm. Multishot accumula-
tion was used as recommended in the literature.45–47Ten laser
shots were accumulated in the storage hexapole, transferred
to the FT-ICR cell, and scanned for 0.78 s. Fifty scans were
summed for each mass spectrum. The acquisition mass range
was m/z 800-4000. The mass spectra were subsequently
apodized and zerofilled twice. An external mass calibration
using Pepmix standard (Bruker Daltonics, Germany) was ap-
plied using a quadratic equation. All samples were measured
until the highest peak in the spectrum had attained a prede-
termined intensity of 107counts. For each spectrum, all peak
intensities were normalized relative to this peak. The mass
spectrometer was operated under Xmass version 7.0.8 and the
data analysis was done using the DataAnalysis version 3.4 (both
from Bruker Daltonics).
LC-MS/MS measurements, for identification of the peptide
peaks that by visual inspection showed differential abundance
by MALDI-FT-ICR-MS, were carried out on an Ultimate 3000
nano LC system (Dionex, Germany) online coupled to a hybrid
linear ion trap /Orbitrap mass spectrometer (LTQ Orbitrap XL;
Thermo Fisher Scientific, Bremen, Germany) (Orbitrap LCMS/
MS). Five microliters of digest was injected onto a C18 trap
column (C18 PepMap, 300 µm i.d. × 5 mm, 5 µm particle size,
100 Å pore size; Dionex, Amsterdam, The Netherlands) and
desalted for 10 min using a flow rate of 20 µL/min 0.1% TFA.
The trap column was then switched online to the analytical
column (PepMap C18, 300 µm i.d. × 5 mm, 3 µm particle and
100 Å pore size; Dionex, The Netherlands) and peptides were
eluted with the following binary gradient: 0-25% solvent B for
120 min and 25-50% solvent B for further 60 min, where
solvent A consists of 2% acetonitrile and 0.1% formic acid in
water and solvent B consists of 80% acetonitrile and 0.08%
formic acid in water. Column flow rate was set to 300 nL/min.
For MS detection, a data dependent acquisition method was
used starting with a high resolution survey scan from 400 to
1800 m/z. This was detected in the Orbitrap (target value of
automatic gain control (AGC) 106, resolution 30 000 at 400 m/z;
lock mass was set to 445.120025 u (protonated (Si(CH3)2O)6).
On the basis of this survey scan, the 5 most intense ions were
consecutively isolated (AGC target set to 104ions) and frag-
mented by collision induced dissociation (CID) applying 35%
normalized collision energy in the linear ion trap. Precursors
that had been selected for MS/MS were excluded for further
CID for 3 min. Peptides were identified using the Bioworks 3.2
software package (Thermo Fisher Scientific, Germany) and its
Sequest feature adhering to the HUPO criteria with XC scores
of 1.8, 2.2, and 3.1 for singly, doubly and triply charged ions,
respectively, in the uniprot_sprot (56.5_25_Nov_2008, tax-
onomy: mammalia) database. The cutoff for mass differences
with the theoretical mass of the identified peptides was set at
2 ppm. For identification of the peptides, the Sequest feature
of the Xcalibur software package was used (Thermo Scientific).
2.7. GC-MS Metabolomic Analysis. A nontargeted GC-MS
method including a derivatization step, which has frequently
been applied for metabolomics studies,48was used to analyze
various classes of (polar) metabolites simultaneously (e.g.,
Effect of Preanalytical Factors
Journal of Proteome Research • Vol. 8, No. 12, 2009
amino acids, organic acids, fatty acids, sugars). In short,
lyophilized CSF samples (100 µL) were derivatized using a
solution of ethoxyamine hydrochloride in pyridine as the
oximation reagent followed by silylation with N-methyl-N-
(trimethylsilyl)trifluoroacetamide principally as described ear-
lier.48One-microliter aliquots of the derivatized samples were
injected in splitless mode on an HP5-MS 30 mm × 0.25 mm ×
0.25 mm capillary column (Agilent Technologies, Palo Alto, CA)
using a temperature gradient from 70 to 320 °C at a rate of 5
°C/min. The GC-MS analysis was performed using an Agilent
6890 gas chromatograph coupled with an Agilent 5973 quad-
rupole mass spectrometer. Detection was carried out using MS
detection in electron impact ionization (70 eV) and full scan
monitoring mode (m/z 15-800). During the different steps in
the sample workup, that is, lyophilization, derivatization and
injection, different (deuterated) internal standards (leucine-d3,
glutamic acid-d3, phenylalanine-d5, glucose-d7, alanine-d4,
cholic acid-d4, dicyclohexylphthalate, Sigma-Aldrich Chemie
B.V. (Zwijndrecht, The Netherlands), typical purity of standards
>97% atom D) were added at a level corresponding to the
concentration of analytes in the QC sample. The samples were
derivatized in duplicate (excepted for t ) 0), and each deriva-
tized sample was injected twice in a random manner.
Data-preprocessing was carried out by automated peak
integration followed by a manual check. This resulted in a target
list of 79 peaks (both known and unknown metabolites)
characterized by a combination of retention time and m/z ratio.
All peak areas were subsequently normalized using the internal
standard dicyclohexylphthalate (DCHP). These normalized
target lists were used for further statistical analysis. Identities
were assigned based on the presence of library searching of
the identical mass spectra in the TNO database.
Multivariate data-analysis (PCA and PCDA) was performed
in the Matlab environment (version 6.5.1 Release 13, The
Mathworks, 2003) using in-house written routines and the PLS
toolbox (version 3.0.2, eigenvector Research, 2003). Student’s
t test was applied to all amino acids and known metabolites.
Metabolites that showed a significant difference in both
samples (p < 0.05) were considered as discriminatory and some
of them were represented as box plots. The analytical error was
within (20% based on replicated injections of the samples.
2.8. LC-MS Analysis of Amino Acids. Briefly, 10 µL of
internal standard (13C15N-labeled amino acids; Asp_C13N15,
Glu_C13_N15, Asn_C13N15, Gln_C13N15, Gly_C13N15, Thr_C13_N15,
Ala_C13_N15, Arg_C13N15, Tyr_C13N15, Val_C13N15, Met_C13N15,
Trp_C13N15, Phe_C13N15, ILe_C13N15, Leu_C13N15, Lys_C13N15,
Pro_C13_N15, Cambridge Isotope Laboratories, Inc.; Andover,
MA) with a concentration range of 70-930 ng/mL for different
amino acids was added to 10 µL of CSF followed by addition
of 100 µL of methanol. The mixture was vortexed for 10 s and
centrifuged at 10 000 rpm (9408 g) for 10 min at 10 °C. The
supernatant was dried under N2and the residue was dissolved
in 80 µL of borate buffer (pH 8.5). After 10 s vortexing, 20 µL of
AQC reagent (Waters part #: 186003836) was added and the
mixture was vortexed immediately. The sample was heated 10
min at 55 °C. Each sample (except P3: T0) was prepared in
duplicate. After cooling, 1 µL of the reaction mixture was
injected in duplicate on a UPLC-MS/MS system comprising an
ACQUITY UPLC system with autosampler (Waters Chroma-
tography B.V., Etten-Leur, The Netherlands) coupled with a
Quattro Premier Xe tandem quadrupole mass spectrometer
(Waters Corporation) and analyzed in positive-ion electrospray
mode. The instrument was operated under the MassLynx data
acquisition software (version 4.1; Waters). The samples were
analyzed on an AccQ-Tag Ultra 2.1 × 100 mm (1.7 µm particle
size) column (Waters) and a binary gradient system of water/
eluent A (10:1, v/v) (AccQ Tag, Waters) and 100% eluent B
(AccQ Tag, Waters). Analytes were eluted by ramping the
percentage of eluent B from 0.1 to 90% in approximately 9.5
min using a combination of both linear and convex profiles.
The flow-rate was 0.7 mL/min, the column temperature was
maintained at 60 °C, and the temperature of the autosampler
tray was set to 10 °C. After each injection, the injection needle
was washed with 200 µL of strong wash solvent (95% ACN),
and 600 µL of weak wash solvent (5% ACN). All analytes were
monitored by Selective Reaction Monitoring (SRM) using
nominal mass resolution (fwhm 0.7 amu). Next to the deriva-
tization reagent all amino acids were selectively monitored via
the transition from the protonated molecule of the AccQ-Tag
derivative to the common fragment at m/z 171. Collision energy
and collision gas (Ar) pressure were 22 eV and 2.5 mbar,
respectively. The complete chromatogram was divided into 6
time windows, restricting the number of SRM transitions to
follow and allowing quantitative information to be gathered
in each segment. Acquired data was evaluated using MassLynx
software (version 4.1; Waters). Quantification and predata
analysis was done using LC-QuanLynx (Waters) and Microsoft
Office Excel 2003 (Microsoft, Redmond, WA), respectively. Data
was analyzed by Student’s t test (T0 vs T30, T30 vs T120 and
T0 vs T120) and amino acids that showed significant differences
between the sample groups (p < 0.05) in samples from P3 and
P7 were considered as discriminating. The data from P3 and
P7 were combined and box plots were created for the discrimi-
natory peaks (Origin 7.0). PCA plots were created to visualize
classifications based on the discriminatory amino acids. The
repeatability of the measurement of significantly discriminatory
amino acids was (21% based on the replicated analysis of each
3. Results and Discussion
3.1. Effect of Delay Time between Sample Collection
and Freezing on the CSF Proteome and Metabolome
(Delayed Storage). In the clinical situation, it may not always
be possible to treat samples taken from patients immediately
by centrifugation in a refrigerated centrifuge and snap freezing.
It is therefore important to record changes that may occur
during this time period in order to discriminate them from
disease-related changes. To follow proteome and metabolome
profiles of CSF, samples were left at room temperature (30 and
120 min) directly after sampling without centrifugation and
after centrifugation before being frozen and stored at -80 °C.
The CSF proteome was analyzed by chipLC-MS and followed
up by MALDI-FT-ICR-MS. Noncentrifuged samples were also
analyzed for differences in metabolite profiles by LC-MS (amino
acid analysis) and GC-MS.
Noncentrifuged porcine CSF that was left at room temper-
ature after sampling showed differences in terms of peak areas
of certain peptides, although the sample was free of red blood
cells, a criterion that is often applied to CSF for acceptance or
rejection in proteomics.47,49,47
Figure 1A depicts a discriminatory region in the base peak
chromatograms (BPCs) from a sample (P7), where visible
changes were detected as a result of increasing the time delay
between sampling and freezing (chipLC-MS method). Three
peptide-related peaks increased strongly in CSF that was left
for 30 or 120 min at room temperature. Figure 1B shows the
Rosenling et al.
5516Journal of Proteome Research • Vol. 8, No. 12, 2009
extracted ion chromatograms (EICs, detection window (50
ppm) corresponding to these discriminatory peaks related to
two peptides (m/z, 638.655 [z, 3; Mw, 1912.943]; m/z, 957.479
[z, 2; Mw, 1912.943]; m/z, 921.443 [z, 4; Mw, 3681.743]). Univari-
ate statistical comparison of the peak areas from 5 repetitive
analyses (Figure 1C) showed that the observed differences are
highly significant with p- values below 0.001 (T0 vs T30 and
T0 vs T120) indicating that these peptides may serve as stability
markers especially in already existing CSF sample collections.
Differences in peak areas between T30 and T120 are not
statistically significant (p > 0.05) indicating that the changes
occur rather rapidly following CSF sampling. Multivariate
statistical analysis by PCA showed that classification of all
sample groups (T0, T30, T120) is possible based on these
discriminatory peaks (Figures S.6-S.8, Supporting Information).
To ensure that the observed differences were not animal-
specific, the discriminatory peaks were confirmed in CSF from
another animal (P3) supporting our findings (data not shown).
The discriminatory ions were subsequently identified as two
tryptic peptide fragments of prostaglandin-D synthase (PTGD-
S_PIG, Q29095, no hit in reverse database) by LC-MS/MS
analysis (see Figures S.2-S.5 in Supporting Information for
details). MALDI-FT-ICR-MS pointed to another peptide orig-
inating from prostaglandin-D synthase (m/z, 1350.705 [z, 1; Mw,
1349.697]) that changed with increasing delay time. However,
instead of increasing with time delay, this peptide decreased
in intensity (Supporting Information S.19-S.22). While both
analyses show that prostaglandin D-synthase-derived peptides
change in abundance upon Delayed Storage, further analyses
are needed to explain the changes of these peptides in terms
of structural alterations of the enzyme.
To assess the stability of the observed peptides in cell-free
porcine CSF, samples were centrifuged prior to incubation at
room temperature for 30 or 120 min, snap-freezing and storage
at -80 °C. Interestingly, no significant change in peak area of
the respective peptides was observed between the groups
indicating that changes might be related to remaining WBCs
that may release proteases upon Delayed Storage.
Our findings indicate that changes in the proteome profile
are introduced in CSF when left at room temperature after
collection. These changes are most notable within the first 30
min after sampling when CSF is not centrifuged. While changes
are clearly reduced or absent when removing cells through
centrifugation right after CSF collection, it is still recommended
to snap-freeze CSF after the centrifugation step and to store it
at -80 °C in aliquots to avoid freeze-thaw cycles (see below).
According to our results, prostaglandin D-synthase is particu-
larly sensitive. The detected peptides after trypsin digestion may
serve as indicators of sample quality in retrospective sample
collections. Similarly, changes in prostaglandin D-synthase in
CSF as a possible biomarker for disease-related processes must
be considered with care unless sample handling steps are
rigorously controlled in order to avoid false positives.
Amino acid analysis (LC-MS) of noncentrifuged CSF indi-
cated a trend to elevated concentrations with increasing time
delay. Ten (Thr, Glu, Gly, His, Ser, Ala, Phe, Tyr, Ile and Asn)
of the detected amino acids (Met, Trp, Leu, Val, Pro, Asp, Arg,
Lys, Thr, Glu, Gly, His, Ser, Ala, Phe, Tyr and Asn) showed a
significantly increased concentration when compared by univari-
ate statistics (only metabolites that were significantly different
in both of the samples were considered) (Figure 2). The rest of
the amino acids also showed a tendency to increase in level
with increased incubation time but either they were only
significantly discriminatory in the samples from one of the pigs
(Met, Leu, Val, Asp, Arg, and Lys) or the difference was not
significant in any of the samples (Pro, Trp). PCA analysis based
on the discriminatory amino acids showed a time-dependent
trend (Figure S.25 in Supporting Information), since groups T0
and T120 are clearly separated from each other with group T30
forming an intermediate state. This is reflected in box plots
(Figure 2), where the standard deviation is generally higher in
T30 samples than in T0 and T120.
PCA analysis of the GC-MS data considering all detected
metabolites shows also a tendency of evolving profiles between
T0, T30 and T120 (Figure S.24 in Supporting Information). The
GC-MS metabolite analysis detected in total 57 known and 22
unknown metabolites (only the identified metabolites were
considered in the statistical analysis). Out of the known
metabolites, 11 were amino acids (Leu, Ile, Lys, Val, Gln, Met,
Ala, Phe, Ser, Thr, Pro). All metabolites except eight (sucrose,
2,3,4-trihydroxybutanoic acid, 1-monostearinglycerol, lactic
acid, uric acid, 1-monopalmitinglycerol, C9:0 fatty acid and
proline) showed a significant increase in concentration over
time. All 49 metabolites that presented significant differences
between the time groups are listed in Table S.23. in Supporting
Information. In total, 24 metabolites showed a significant
increase (p < 0.05) in concentration between T0 and T30, while
46 metabolites were significantly increased after 120 min (T0
versus T120) (see Figure S.23 in Supporting Information). This
indicates that metabolic processes are ongoing during Delayed
Figure 2. Amino acids in noncentrifuged porcine CSF that show an increase in concentration with delay times of 0, 30, or 120 min at
room temperature (T0, T30, T120) between sampling and snap-freezing as assessed by LC-MS amino acid analysis. Only amino acids
with significant differences between the sample groups in both P3 and P7 are shown. The figure shows combined data from P3 and
P7. Symbols mark significant p-values according to students t test (T0 vs T30 and T0 vs T120) * )p < 0.05, 9 )p < 0.01, [ )p < 0.005,
b )p < 0.001, ≤ )p < 0.0005, (box with solid circle) )p < 0.0001.
Effect of Preanalytical Factors
Journal of Proteome Research • Vol. 8, No. 12, 2009
Storage and that rapid centrifugation and snap freezing are
indispensable for reliable metabolite analysis in CSF. Figure 3
shows the 11 amino acids that were significantly different
between time groups determined by GC-MS analysis. Seven of
them showed increased levels in LC-MS and GC-MS analyses
(Ala, Ser, Phe, Thr, Ile and Gly) and one (Pro) was unchanged
in both methods. There were four amino acids, detected by
both methods, that showed a significant difference by GC-MS
only (Leu, Lys, Val and Met).These four amino acids were
increased in only one of the samples in the LC-MS analysis
and were thus not considered significant. There was no
discrepancy between GC-MS and LC-MS analysis, since no
amino acid showed an opposite trend.
One of the compounds that increased upon delayed storage
(GC-MS analysis) was glucose, a metabolite that is routinely
measured in CSF25(Figure S.23, Supporting Information). It is
thus important to realize that sample handling after CSF
collection may affect glucose measurements and therefore the
interpretation of routine laboratory analyses. Centrifugation
after CSF collection followed by snap freezing and storage at
-80 °C is therefore also recommended for metabolite analysis.
In the NMR study by Levine et al., glucose, alanine, myo-
inositol and acetate showed no change over 72 h at room
temperature.31Acetate was not found in our analysis, but the
three other metabolites were detected and showed an increased
level after 120 min at room temperature. They also showed a
decreased citrate level, while in our study, this metabolite (citric
acid) increased. These variances might depend on the differ-
ences in the experimental designs. Levine et al. assessed
samples that were stored frozen for 30 months prior to
incubation at room temperature, which may have affected the
outcome. The fact that the incubation time was significantly
longer (72 h compared to maximally 2 h in our study) may have
also affected the results. Some metabolites may decrease in
concentration because of adsorption to the tube walls,50while
others might increase due to enzymatic processes. Further-
more, Levine et al. revealed that lactate, creatine, glutamine
and creatinine were found to increase. While lactate and
creatine were not detected in our GC-MS analysis, both
glutamine and creatinine showed the same result in our
analysis as in the NMR study.
3.2. Effect of Freeze/Thaw Cycles on the CSF Proteome.
Freeze/thaw cycles are often unavoidable when analyzing
biological samples. Freeze/thawing is known to affect biological
samples and notably to induce conformational changes in
To assess the influence of freeze/thawing on the proteome
profile, depleted noncentrifuged porcine CSF was exposed to
1×, 2×, 3×, 4×, 5× or 10× freeze/thaw cycles and compared
to the starting material. A freeze-thaw cycle comprised 30 min
thawing on ice and 2 min freezing in liquid nitrogen. All
samples were left on ice throughout the study until the
overnight trypsin digestion (the control sample was thus left
on ice throughout the entire process of freeze/thawing (∼5.5
h until digestion) and the 5× freeze/thaw sample was left on
ice ∼2.5 h). The experiment was performed twice on two
separate dates on CSF from the same animal (P3, T0). Only
peaks that showed the same pattern in both the experiments
were considered as discriminatory. The chromatograms of
freeze/thaw-treated samples did not show differences when
inspected visually at the TIC/BPC level. However, detailed
statistical analysis after data processing (see section 2.4)
revealed four tryptic peptide peaks out of 41 peaks on the list
that changed significantly in peak area when comparing the
starting material with a sample having undergone 10 freeze/
thaw cycles. PCA analysis based on the discriminatory peaks
showed a clear discrepancy between the control group and the
10× freeze/thaw group (Figure S.14 in Supporting Information).
Figure 4A depicts the EICs (detection window (50 ppm) of the
four peptide peaks that were identified by LC-MS/MS as tryptic
Figure 3. Amino acids showing a significant increase in concentration in noncentrifuged porcine CSF with increased delay time when
analyzed by GC-MS metabolomic analysis. The amino acids were significantly different between the groups in both sample P3 and P7
and the figure shows a combination of the results from both samples. The symbols mark significant p-values according to Students
t test (T0 vs T30 and T0 vs T120) * )p < 0.05, [ )p < 0.005, ≤ )p < 0.0005, (box with solid circle) ) p < 0.0001.
Rosenling et al.
5518Journal of Proteome Research • Vol. 8, No. 12, 2009
fragments of transthyretin (TTY_PIG, P50390, no hit in reverse
database) (m/z, 428.234 [z, 2; Mw, 854.453]; m/z, 464.267 [z, 2;
Mw, 926.519]; m/z, 555.265 [z, 3; Mw, 1662.7732]; m/z, 681.838
[z, 2; Mw, 1361.661]) indicating that transthyretin is particularly
sensitive to freeze/thawing (Supporting Information, Figures
S.9-S.13). Univariate statistical analysis of the peak areas
(Figure 4B) resulted in highly significant differences. Analysis
of these peptides for an increasing number of freeze/thaw
cycles showed that 3 of the 4 peptides were already increased
after a single freeze/thaw cycle while one peptide (m/z: 428.234)
increased only significantly after 10 freeze/thaw cycles (Figure
4C). These results emphasize that the number of freeze-thaw
cycles should be controlled when comparing different sets of
Transthyretin is a homotetrameric protein present in plasma
and CSF that is known to be potentially unstable and prone to
fibril formation leading to amyloidosis causing plaque forma-
tion in Alzheimer’s disease.54Mutants of transthyretin are
presently being screened with respect to the risk of developing
Familial Amyloidotic Polyneuropathy.55This propensity for
aggregation might explain the sensitivity of transthyretin to
repeated freeze/thawing. The strong increase in tryptic peptides
after freeze/thaw cycles indicates that cleavage sites become
more accessible possibly due to a change in conformation. This
finding underlines that transthyretin must be considered with
caution as biomarker candidate due to its structural instability
and that the number of freeze/thaw cycles should be carefully
controlled, since changes in transthyretin levels may simply
be due to freeze-thawing rather than to biological processes
related to the disease of interest.
3.3. Effect of Storage at 4 °C after Digestion on CSF
Proteome Analysis. Earlier results from our group have shown
that peptide concentrations may change during storage in the
autosampler of the LC-MS system at 4 °C. We therefore
investigated whether this is also the case for depleted, trypsin-
digested CSF, since different samples might stay for various
time periods in the autosampler prior to injection, which may
lead to differences in peptide profiles. While randomization of
the order of injection can alleviate this problem to some extent,
it will lead to an increased variability of the results making
observation of minor changes in peptide concentrations related
to disease difficult. For the discovery of peptide peaks that are
susceptible to this effect, we compared freshly prepared
(depleted, trypsin-digested) porcine CSF (P4) with the same
samples stored for 1 month at 4 °C. PCA analysis based on the
discriminatory peaks showed a distinct difference between the
control group and the 1 month sample group (Supporting
Information, Figure S.18.). Figure 5A displays EICs (detection
window (50 ppm) of two peptide peaks out of 45 peaks that
decrease significantly after storage of the digest at 4 °C for 1
month (m/z, 681.858 [z, 2; Mw, 1361.701]; m/z, 369.188 [z, 2;
Mw, 736.361]). One of the peptides was derived again from
prostaglandin D-synthase (369.188) (PTGDS_PIG Q29095, no
hit in reverse database) while the other belongs to serotrans-
ferrin (681.858) (TRFE_PIG P09571, no hit in reverse database)
(Figures S.15-S.17 in Supporting Information). Comparison of
profiles after 1 day or 1 week showed that the serotransferrin
peptide (m/z ) 681.858) decreased already significantly after
1 week at 4 °C (Figure 5B). A repetition of the experiment on
samples from two different animals (P2/P4) confirmed these
results. Both peptides showed a significant decrease in peak
area after 1 week or even after a single day (data not shown)
in the autosampler at 4 °C. Although major changes in the
overall peptide profile were not observed, the results show that
samples should not be left for more than 1 day at 4 °C in the
autosampler to avoid effects that might confound disease-
Figure 4. (A) Extracted ion chromatograms (detection window
(50 ppm) of four peptide peaks that increase after 10× freeze/
thaw cycles. All peaks originate from different tryptic peptides
of transthyretin detected by the chipLC-MS proteomic analysis.
(B) Univariate statistical analysis of peak areas of the peptides
(see Figure 5A) based on two-tailed Students t tests of 5
repetitions of the chipLC-MS runs. The same experiment was
performed twice on CSF from the same pig confirming the
results. Data are represented as box plots with p-values (control
vs 10× freeze/thaw cycles). (C) Univariate statistical analysis of
peak areas of the peptides that change significantly with respect
to the number of freeze/thaw cycles (see Figure 5A). Analysis is
based on two-tailed Students t tests of 5 repetitions of the chipLC-
MS runs of depleted, trypsin-digested, porcine CSF. The same
experiment was performed twice confirming the results. Data are
represented as box plots with p-values (control vs 1× freeze/thaw
cycle, control vs 5× freeze/thaw cycles, control vs 10× freeze/
thaw cycles, respectively). b p < 0.05, [ p < 0.005, 9 p < 0.0001;
n.s, nonsignificant with respect to the control CSF sample.
Effect of Preanalytical Factors
Journal of Proteome Research • Vol. 8, No. 12, 2009
In this study, we have evaluated a number of preanalytical
factors related to sample handling of CSF to assess their effect
on the protein, peptide and metabolite profiles using different
analytical platforms. The time delay between CSF collection,
centrifugation and snap-freezing proved to affect notably
prostaglandin D-synthase as revealed by a significant increase
of two peptides. Metabolite analysis showed that the level of a
number of metabolites, including free amino acids, increased
during delayed storage. These changes indicate that remaining
enzymatic activity in CSF after collection may lead to altered
metabolite and protein levels, emphasizing that CSF should be
centrifuged and snap frozen as soon as possible after collection
for proteomics or metabolomics studies. The prostaglandin
D-synthase-derived peptides and the identified metabolites
may be used to assess consistent sample handling notably
when considering analysis of existing CSF sample collections
The addition of stabilizing agents or specific pretreatments
(protease inhibitors, deproteinization) was not assessed in this
study. The addition of stabilizing agents alters the samples by
introducing additional compounds, which lead to new peaks
in the mass spectra and chromatogram. It was our intention
to handle CSF in a way to avoid difficulties with subsequent
analyses. In this study, we wanted to assess how CSF is affected
during a possible delay of sample handling in the clinic.
CSF samples should ideally be centrifuged immediately after
collection followed by storage at -80 °C in order to remove all
cellular elements and to avoid protease activation and other
kinds of degradation of macromolecules. Unless sample han-
dling in the clinic is carefully controlled, we recommend studies
targeting certain proteins (e.g., prostaglandin D-synthase,
transthyretin) or metabolites to be cautious with interpreta-
tions. We have indications that even minor contamination of
CSF with white blood cells accelerates the observed changes
and that it is not sufficient to assess that CSF is free of
erythrocytes. Chow et al.56and Steele et al.57pointed out that
the major part of the WBCs lysed when CSF samples were
exposed to room temperature for 22 h and that this process
continues at 4 °C even if the rate is slower. This indicates that
leaving cell-containing CSF samples on ice is not recommended.
We have also shown that widely used sample handling
routines such as freeze/thawing or keeping digested CSF in an
autosampler at 4 °C prior to LC-MS analysis can induce changes
in proteome-derived peptide profiles. To avoid the introduction
of unnecessary freeze/thaw cycles, samples should be aliquoted
after centrifugation in volumes that avoid freeze-thawing. CSF
samples that have been exposed to repeated freeze/thaw cycles
are not ideal for the assessment of transthyretin, since peptides
of this protein showed an increased level with increasing
numbers of freeze/thaw cycles. Samples remaining for exten-
sive time periods (>24 h) in a cooled autosampler are not
reliable when it comes to the quantitative analysis of for
example, serotransferrin and prostaglandin D-synthase.
Several factors affect the stability of proteins, peptides and
metabolites in CSF. We focused on the effect of delay time
between CSF sampling and freezing, the number of freeze/thaw
cycles and the stability of trypsin-digested CSF in the autosam-
pler. Other factors that might be interesting to assess are the
material of the sample tubes, long-term storage at different
temperatures and the effect of adding stabilizing agents.
Acknowledgment. The authors would like to thank
Bas Muilwijk for the GC-MS analysis and Adrie Dane for
the targeted amino acid LC-MS data analysis support.
This study was financially supported by and performed
within the framework of the Top Institute Pharma project
D4-102. The work was further supported by the BioRange
2.2.3 projectfrom TheNetherlands Proteomicsand
Figure 5. (A) Extracted ion chromatograms (detection window (50 ppm) of two discriminatory peptide peaks that decrease upon storage
of a tryptic digest of depleted porcine CSF at 4 °C. The peaks originate from tryptic peptides of serotransferrin (m/z, 681.858; Mw,
1361.701) and prostaglandin D-synthase (m/z, 369.188; Mw, 736.361)) detected by the chipLC-MS proteomic analysis. (B) Univariate
statistical analysis of peak areas of peptides that change significantly upon storage of a tryptic digest of depleted porcine CSF at 4 °C
(see panels A). Analysis is based on two-tailed Students t tests of 5 repetitions of the chipLC-MS runs of depleted, trypsin-digested,
porcine CSF. Data are represented as box plots with p-values for cases that were significantly different (control vs 1 month and control
vs 1 week).
Rosenling et al.
5520Journal of Proteome Research • Vol. 8, No. 12, 2009
Bioinformatics Centers (Christin). The image used for the
Table of Contents was obtained from http://www.brain-
Supporting Information Available: Database search
criteria and results, MS/MS spectra, amino acid sequences of
identified proteins and sequence coverage, PCA plots, lists of
discriminatory metabolites in GC-MS analysis. This material
is available free of charge via the Internet at http://pubs.acs.org.
(1) Belogurov, A. A., Jr.; Kurkova, I. N.; Friboulet, A.; Thomas, D.;
Misikov, V. K.; Zakharova, M. Y.; Suchkov, S. V.; Kotov, S. V.; Alehin,
A. I.; Avalle, B.; Souslova, E. A.; Morse, H. C., III; Gabibov, A. G.;
Ponomarenko, N. A. Recognition and Degradation of Myelin Basic
Protein Peptides by Serum Autoantibodies: Novel Biomarker for
Multiple Sclerosis. J. Immunol. 2008, 180 (2), 1258–1267.
(2) Dekker, L. J.; Burgers, P. C.; Kros, J. M.; Smitt, P. A.; Luider, T. M.
Peptide Profiling of Cerebrospinal Fluid by Mass Spectrometry.
Expert Rev. Proteomics 2006, 3 (3), 297–309.
(3) Lutz, N. W.; Viola, A.; Malikova, I.; Confort-Gouny, S.; Ranjeva,
J. P.; Pelletier, J.; Cozzone, P. J. A Branched-Chain Organic Acid
Linked to Multiple Sclerosis: First Identification by NMR Spec-
troscopy of CSF. Biochem. Biophys. Res. Commun. 2007, 354 (1),
(4) Myint, K. T.; Aoshima, K.; Tanaka, S.; Nakamura, T.; Oda, Y.
Quantitative Profiling of Polar Cationic Metabolites in Human
Cerebrospinal Fluid by Reversed-Phase Nanoliquid Chromatog-
raphy/Mass Spectrometry. Anal. Chem. 2009, 81 (3), 1121–1129.
(5) Noben, J. P.; Dumont, D.; Kwasnikowska, N.; Verhaert, P.; Somers,
V.; Hupperts, R.; Stinissen, P.; Robben, J. Lumbar Cerebrospinal
Fluid Proteome in Multiple Sclerosis: Characterization by Ultra-
filtration, Liquid Chromatography, and Mass Spectrometry. J.
Proteome Res. 2006, 5 (7), 1647–1657.
(6) Rejdak, K.; Petzold, A.; Stelmasiak, Z.; Giovannoni, G. Cerebrospinal
Fluid Brain Specific Proteins in Relation to Nitric Oxide Metabolites
During Relapse of Multiple Sclerosis. Mult. Scler. 2008, 14 (1), 59–
(7) Stoop, M. P.; Dekker, L. J.; Titulaer, M. K.; Burgers, P. C.; Sillevis
Smitt, P. A.; Luider, T. M.; Hintzen, R. Q. Multiple Sclerosis-Related
Proteins Identified in Cerebrospinal Fluid by Advanced Mass
Spectrometry. Proteomics 2008, 8 (8), 1576–1585.
(8) Walter, A.; Korth, U.; Hilgert, M.; Hartmann, J.; Weichel, O.; Hilgert,
M.; Fassbender, K.; Schmitt, A.; Klein, J. Glycerophosphocholine
Is Elevated in Cerebrospinal Fluid of Alzheimer Patients. Neurobiol.
Aging 2004, 25 (10), 1299–1303.
(9) Zhang, J.; Goodlett, D. R.; Montine, T. J. Proteomic Biomarker
Discovery in Cerebrospinal Fluid for Neurodegenerative Diseases.
J. Alzheimer’s Dis. 2005, 8 (4), 377–386.
(10) Del Boccio, P.; Pieragostino, D.; Lugaresi, A.; Di, I. M.; Pavone, B.;
Travaglini, D.; D’Aguanno, S.; Bernardini, S.; Sacchetta, P.; Federici,
G.; Di, I. C.; Gambi, D.; Urbani, A. Cleavage of Cystatin C Is Not
Associated With Multiple Sclerosis. Ann. Neurol. 2007, 62 (2), 201–
(11) Hansson, S. F.; Simonsen, A. H.; Zetterberg, H.; Andersen, O.;
Haghighi, S.; Fagerberg, I.; Andreasson, U.; Westman-Brinkmalm,
A.; Wallin, A.; Ruetschi, U.; Blennow, K. Cystatin C in Cerebrospinal
Fluid and Multiple Sclerosis. Ann. Neurol. 2007, 62 (2), 193–196.
(12) Irani, D. N.; Anderson, C.; Gundry, R.; Cotter, R.; Moore, S.; Kerr,
D. A.; McArthur, J. C.; Sacktor, N.; Pardo, C. A.; Jones, M.; Calabresi,
P. A.; Nath, A. Cleavage of Cystatin C in the Cerebrospinal Fluid
of Patients With Multiple Sclerosis. Ann. Neurol. 2006, 59 (2), 237–
(13) Chaigneau, C.; Cabioch, T.; Beaumont, K.; Betsou, F. Serum
Biobank Certification and the Establishment of Quality Controls
for Biological Fluids: Examples of Serum Biomarker Stability After
Temperature Variation. Clin. Chem. Lab. Med. 2007, 45 (10), 1390–
(14) Hsieh, S. Y.; Chen, R. K.; Pan, Y. H.; Lee, H. L. Systematical
Evaluation of the Effects of Sample Collection Procedures on Low-
Molecular-Weight Serum/Plasma Proteome Profiling. Proteomics
2006, 6 (10), 3189–3198.
(15) Papale, M.; Pedicillo, M. C.; Thatcher, B. J.; Di, P. S.; Lo, M. L.;
Bufo, P.; Rocchetti, M. T.; Centra, M.; Ranieri, E.; Gesualdo, L.
Urine Profiling by SELDI-TOF/MS: Monitoring of the Critical Steps
in Sample Collection, Handling and Analysis. J. Chromatogr. B:
Anal. Technol. Biomed. Life Sci. 2007, 856 (1-2), 205–213.
(16) Schaub, S.; Wilkins, J.; Weiler, T.; Sangster, K.; Rush, D.; Nickerson,
P. Urine Protein Profiling With Surface-Enhanced Laser-Desorp-
tion/Ionization Time-of-Flight Mass Spectrometry. Kidney Int.
2004, 65 (1), 323–332.
(17) West-Nielsen, M.; Hogdall, E. V.; Marchiori, E.; Hogdall, C. K.;
Schou, C.; Heegaard, N. H. Sample Handling for Mass Spectro-
metric Proteomic Investigations of Human Sera. Anal. Chem. 2005,
77 (16), 5114–5123.
(18) Kaiser, E.; Schonknecht, P.; Thomann, P. A.; Hunt, A.; Schroder, J.
Influence of Delayed CSF Storage on Concentrations of Phospho-
Tau Protein (181), Total Tau Protein and Beta-Amyloid (1-42).
Neurosci. Lett. 2007, 417 (2), 193–195.
(19) Schoonenboom, N. S.; Mulder, C.; Vanderstichele, H.; Van Elk, E. J.;
Kok, A.; Van Kamp, G. J.; Scheltens, P.; Blankenstein, M. A. Effects
of Processing and Storage Conditions on Amyloid Beta (1-42) and
Tau Concentrations in Cerebrospinal Fluid: Implications for Use
in Clinical Practice. Clin. Chem. 2005, 51 (1), 189–195.
(20) Bibl, M.; Esselmann, H.; Otto, M.; Lewczuk, P.; Cepek, L.; Ruther,
E.; Kornhuber, J.; Wiltfang, J. Cerebrospinal Fluid Amyloid Beta
Peptide Patterns in Alzheimer’s Disease Patients and Nonde-
mented Controls Depend on Sample Pretreatment: Indication of
Carrier-Mediated Epitope Masking of Amyloid Beta Peptides.
Electrophoresis 2004, 25 (17), 2912–2918.
(21) Ramont, L.; Thoannes, H.; Volondat, A.; Chastang, F.; Millet, M. C.;
Maquart, F. X. Effects of Hemolysis and Storage Condition on
Neuron-Specific Enolase (NSE) in Cerebrospinal Fluid and Serum:
Implications in Clinical Practice. Clin. Chem. Lab. Med. 2005, 43
(22) Ranganathan, S.; Polshyna, A.; Nicholl, G.; Lyons-Weiler, J.;
Browser, R. Assessment of Protein Stability in Cerebrospinal Fluid
Using Surface-Enhanced Laser Desorption/Ionization Time-of-
Flight Mass Spectrometry Protein Profiling. Clin. Proteomics 2006,
2 (1-2), 91–101.
(23) Rosato, P. M.; Gama, F. G. V.; Santana, A. E. Physical-Chemical
Analysis of the Cerebrospinal Fluid of Healthy Dogs Submitted to
Different Storage Periods and Temperatures. Cienc. Rural 2006,
36 (6), 1806–1810.
(24) Fry, M. M.; Vernau, W.; Kass, P. H.; Vernau, K. M. Effects of Time,
Initial Composition and Stabilizing Agents on the Results of Canine
Cerebrospinal Fluid Analysis. Vet. Clin. Pathol. 2006, 35 (1), 72–
(25) Deisenhammer, F.; Bartos, A.; Egg, R.; Gilhus, N. E.; Giovannoni,
G.; Rauer, S.; Sellebjerg, F. Guidelines on Routine Cerebrospinal
Fluid Analysis. Report From an EFNS Task Force. Eur. J. Neurol.
2006, 13 (9), 913–922.
(26) Hale, J. E.; Gelfanova, V.; You, J. S.; Knierman, M. D.; Dean, R. A.
Proteomics of Cerebrospinal Fluid: Methods for Sample Process-
ing. Methods Mol. Biol. 2008, 425 (2), 53–66.
(27) Jerrard, D. A.; Hanna, J. R.; Schindelheim, G. L. Cerebrospinal Fluid.
J. Emerg. Med. 2001, 21 (2), 171–178.
(28) Timms, J. F.; rslan-Low, E.; Gentry-Maharaj, A.; Luo, Z.; T’Jampens,
D.; Podust, V. N.; Ford, J.; Fung, E. T.; Gammerman, A.; Jacobs, I.;
Menon, U. Preanalytic Influence of Sample Handling on SELDI-
TOF Serum Protein Profiles. Clin. Chem. 2007, 53 (4), 645–656.
(29) Ekblad, L.; Baldetorp, B.; Ferno, M.; Olsson, H.; Bratt, C. In-Source
Decay Causes Artifacts in SELDI-TOF MS Spectra. J. Proteome Res.
2007, 6 (4), 1609–1614.
(30) Aristoteli, L. P.; Molloy, M. P.; Baker, M. S. Evaluation of Endog-
enous Plasma Peptide Extraction Methods for Mass Spectrometric
Biomarker Discovery. J. Proteome Res. 2007, 6 (2), 571–581.
(31) Levine, J.; Panchalingam, K.; McClure, R. J.; Gershon, S.; Pettegrew,
J. W. Stability of CSF Metabolites Measured by Proton NMR.
J. Neural Transm. 2000, 107 (7), 843–848.
(32) Pan, S.; Wang, Y.; Quinn, J. F.; Peskind, E. R.; Waichunas, D.;
Wimberger, J. T.; Jin, J.; Li, J. G.; Zhu, D.; Pan, C.; Zhang, J.
Identification of Glycoproteins in Human Cerebrospinal Fluid With
a Complementary Proteomic Approach. J. Proteome Res. 2006, 5
(33) Strawn, J. R.; Ekhator, N. N.; Geracioti, T. D., Jr. In-Use Stability
of Monoamine Metabolites in Human Cerebrospinal Fluid. J. Chro-
matogr., B: Biomed. Sci. Appl. 2001, 760 (2), 301–306.
(34) Tammen, H.; Schulte, I.; Hess, R.; Menzel, C.; Kellmann, M.;
Mohring, T.; Schulz-Knappe, P. Peptidomic Analysis of Human
Blood Specimens: Comparison Between Plasma Specimens and
Serum by Differential Peptide Display. Proteomics 2005, 5 (13),
(35) Villanueva, J.; Philip, J.; Entenberg, D.; Chaparro, C. A.; Tanwar,
M. K.; Holland, E. C.; Tempst, P. Serum Peptide Profiling by
Magnetic Particle-Assisted, Automated Sample Processing and
MALDI-TOF Mass Spectrometry. Anal. Chem. 2004, 76 (6), 1560–
(36) Villanueva, J.; Philip, J.; Chaparro, C. A.; Li, Y.; Toledo-Crow, R.;
DeNoyer, L.; Fleisher, M.; Robbins, R. J.; Tempst, P. Correcting
Effect of Preanalytical Factors
Journal of Proteome Research • Vol. 8, No. 12, 2009
Common Errors in Identifying Cancer-Specific Serum Peptide
Signatures. J. Proteome Res. 2005, 4 (4), 1060–1072.
(37) Anesi, A.; Rondanelli, M.; d’Eril, G. M. Stability of Neuroactive
Amino Acids in Cerebrospinal Fluid Under Various Conditions of
Processing and Storage. Clin. Chem. 1998, 44 (11), 2359–2360.
(38) van Midwoud, P. M.; Rieux, L.; Bischoff, R.; Verpoorte, E.; Nied-
erlander, H. A. Improvement of Recovery and Repeatability in
Liquid Chromatography-Mass Spectrometry Analysis of Peptides.
J. Proteome Res. 2007, 6 (2), 781–791.
(39) Horvatovich, P.; Govorukhina, N. I.; Reijmers, T. H.; van der Zee,
A. G.; Suits, F.; Bischoff, R. Chip-LC-MS for Label-Free Profiling
of Human Serum. Electrophoresis 2007, 28 (23), 4493–4505.
(40) Christin, C.; Smilde, A. K.; Hoefsloot, H. C.; Suits, F.; Bischoff, R.;
Horvatovich, P. L. Optimized Time Alignment Algorithm for LC-
MS Data: Correlation Optimized Warping Using Component
Detection Algorithm-Selected Mass Chromatograms. Anal. Chem.
2008, 80 (18), 7012–7021.
(41) Kemperman, R. F.; Horvatovich, P. L.; Hoekman, B.; Reijmers, T. H.;
Muskiet, F. A.; Bischoff, R. Comparative Urine Analysis by Liquid
Chromatography-Mass Spectrometry and Multivariate Statistics:
Method Development, Evaluation, and Application to Proteinuria.
J Proteome Res. 2007, 6 (1), 194–206.
(42) Suits, F.; Lepre, J.; Du, P.; Bischoff, R.; Horvatovich, P. Two-
Dimensional Method for Time Aligning Liquid Chromatography-
Mass Spectrometry Data. Anal. Chem. 2008, 80 (9), 3095–3104.
(43) Radulovic, D.; Jelveh, S.; Ryu, S.; Hamilton, T. G.; Foss, E.; Mao,
Y.; Emili, A. Informatics Platform for Global Proteomic Profiling
and Biomarker Discovery Using Liquid Chromatography-Tandem
Mass Spectrometry. Mol. Cell. Proteomics 2004, 3 (10), 984–997.
(44) Tibshirani, R.; Hastie, T.; Narasimhan, B.; Chu, G. Diagnosis of
Multiple Cancer Types by Shrunken Centroids of Gene Expression.
Proc. Natl. Acad. Sci. U.S.A. 2002, 99 (10), 6567–6572.
(45) Mize, T. H.; Amster, I. J. Broad-Band Ion Accumulation With an
Internal Source MALDI-FTICR-MS. Anal. Chem. 2000, 72 (24),
(46) Moyer, S. C.; Budnik, B. A.; Pittman, J. L.; Costello, C. E.; O’Connor,
P. B. Attomole Peptide Analysis by High-Pressure Matrix-Assisted
Laser Desorption/Ionization Fourier Transform Mass Spectrom-
etry. Anal. Chem. 2003, 75 (23), 6449–6454.
(47) O’Connor, P. B.; Costello, C. E. Application of Multishot Acquisition
in Fourier Transform Mass Spectrometry. Anal. Chem. 2000, 72
(48) Koek, M. M.; Muilwijk, B.; van der Werf, M. J.; Hankemeier, T.
Microbial Metabolomics With Gas Chromatography/Mass Spec-
trometry. Anal. Chem. 2006, 78 (4), 1272–1281.
(49) You, J. S.; Gelfanova, V.; Knierman, M. D.; Witzmann, F. A.; Wang,
M.; Hale, J. E. The Impact of Blood Contamination on the
Proteome of Cerebrospinal Fluid. Proteomics 2005, 5 (1), 290–296.
(50) Kraut, A.; Marcellin, M.; Adrait, A.; Kuhn, L.; Louwagie, M.; Kieffer-
Jaquinod, S.; Lebert, D.; Masselon, C. D.; Dupuis, A.; Bruley, C.;
Jaquinod, M.; Garin, J.; Gallagher-Gambarelli, M. Peptide Storage:
Are You Getting the Best Return on Your Investment? Defining
Optimal Storage Conditions for Proteomics Samples. J Proteome
Res. 2009, 8 (7), 3778–3785.
(51) Bhatnagar, B. S.; Bogner, R. H.; Pikal, M. J. Protein Stability During
Freezing: Separation of Stresses and Mechanisms of Protein
Stabilization. Pharm. Dev. Technol. 2007, 12 (5), 505–523.
(52) Dong, A.; Prestrelski, S. J.; Allison, S. D.; Carpenter, J. F. Infrared
Spectroscopic Studies of Lyophilization- and Temperature-Induced
Protein Aggregation. J. Pharm. Sci. 1995, 84 (4), 415–424.
(53) Klibanov, A. M.; Schefiliti, J. A. On the Relationship Between
Conformation and Stability in Solid Pharmaceutical Protein
Formulations. Biotechnol. Lett. 2004, 26 (14), 1103–1106.
(54) Sekijima, Y.; Kelly, J. W.; Ikeda, S. Pathogenesis of and Therapeutic
Strategies to Ameliorate the Transthyretin Amyloidoses. Curr.
Pharm. Des. 2008, 14 (30), 3219–3230.
(55) Bergen, H. R., III; Zeldenrust, S. R.; Naylor, S. An on-Line Assay
for Clinical Detection of Amyloidogenic Transthyretin Variants
Directly From Serum. Amyloid 2003, 10 (3), 190–197.
(56) Chow, G.; James, W.; Schmidley, M. D. Lysis of Erythrocytes and
Leukocytes in Traumatic Lumbar Punctures. Arch. Neurol. 1984,
41 (10), 1084–1085.
(57) Steele, R. W.; Marmer, D. J.; O’Brien, M. D.; Tyson, S. T.; Steele,
C. R. Leukocyte Survival in Cerebrospinal Fluid. J. Clin. Microbiol.
1986, 23 (5), 965–966.
Rosenling et al.
5522Journal of Proteome Research • Vol. 8, No. 12, 2009