Clinical Chemistry 52:9
1713–1721 (2006)
Identification of Novel Brain Biomarkers
Omar F. Laterza,†‡ Vijay R. Modur,† Dan L. Crimmins, Jitka V. Olander,
Yvonne Landt, Jin-Moo Lee, and Jack H. Ladenson*
Background: The diagnosis of diseases leading to brain
injury, such as stroke, Alzheimer disease, and Parkinson
disease, can often be problematic. In this study, we
pursued the discovery of biomarkers that might be
specific and sensitive to brain injury.
Methods: We performed gene array analyses on a mouse
model to look for biomarkers that are both preferen-
tially and abundantly produced in the brain. Via bioin-
formatics databases, we identified the human homologs
of genes that appeared abundant in brain but not in
other tissues. We then confirmed protein production of
the genes via Western blot of various tissue homoge-
nates and assayed for one of the markers, visinin-like
protein 1 (VLP-1), in plasma from patients after ischemic
stroke.
Results: Twenty-nine genes that were preferentially
and abundantly expressed in the mouse brain were
identified; of these 29 genes, 26 had human homologs.
We focused on 17 of these genes and their protein
products on the basis of their molecular characteristics,
novelty, and/or availability of antibodies. Western blot
showed strong signals in brain homogenates for 13 of
these proteins. Tissue specificity was tested by Western
blot on a human tissue array, and a sensitive and
quantitative sandwich immunoassay was developed for
the most abundant gene product observed in our search,
VLP-1. VLP-1 was detected in plasma of patients after
stroke and in cerebrospinal fluid of a rat model of
stroke.
Conclusions: The use of relative mRNA production
appears to be a valid method of identifying possible
biomarkers of tissue injury. The tissue specificity sug-
gested by gene expression was confirmed by Western
Department of Pathology and Immunology, Division of Laboratory Med-
icine, Washington University School of Medicine, St. Louis, MO.
†These authors contributed equally to this work.
‡Current affiliation: Merck & Co., Inc., Rahway, NJ 07065-0900.
*Address correspondence to this author at: Department of Pathology and
Immunology, Division of Laboratory Medicine, Washington University School
of Medicine, 660 S. Euclid Ave., Campus Box 8118, St. Louis, MO 63110. Fax
314-454-5208; e-mail ladenson@labmed.wustl.edu.
Received March 29, 2006; accepted June 21, 2006.
Previously published online at DOI: 10.1373/clinchem.2006.070912
1713
Lipids, Lipoproteins,
and Cardiovascular
Risk Factors
blot. One of the biomarkers identified, VLP-1, was
increased in a rat model of stroke and in plasma of
patients after stroke. More extensive, prospective stud-
ies of the candidate biomarkers identified appear
warranted.
© 2006 American Association for Clinical Chemistry
Stroke is the third most common cause of death world-
wide, after ischemic heart disease and malignancies, and
⬃700 000 patients in the US have a new or recurrent
stroke each year (1 ). The clinical diagnosis of stroke by
specialists is reasonably accurate. On clinical grounds but
in general practice, up to 20% of the patients suspected of
stroke have a different diagnosis (2 ).
The evaluation of patients with suspected stroke, an
acute brain injury, would be greatly aided by the avail-
ability of a readily measured biomarker such as is avail-
able for acute coronary syndrome and myocardial infarc-
tion, e.g., troponin I or troponin T (3 ). Such a brain
biomarker for brain injury might also be useful in chronic
neurodegenerative diseases.
There have been several studies of possible biomarkers
of acute brain damage such as stroke, with S-100B, neu-
ron-specific enolase (NSE),1 glial fibrillary associated pro-
tein (GFAP), and myelin basic protein (MBP) generating
the most interest.
S-100B (4 ) is found at high concentrations in glial and
Schwann cells. Numerous reports (4 ) show that S-100B is
increased in blood and cerebrospinal fluid (CSF) after
traumatic brain damage, stroke, and a variety of neuro-
degenerative diseases. The increase of S-100B in serum
after acute ischemic stroke peaks at 2–3 days, and it can be
observed at ⬃12 h (5 ). S-100B does not appear to be
specific for brain injury; high amounts of S-100B are
released in response to trauma of muscle, fat, and bone,
and values in trauma without head injury are also in-
creased (6–8 ).
1
Nonstandard abbreviations: NSE, neuron-specific enolase; GFAP, glial
fibrillary-associated protein; MBP, myelin basic protein; CSF, cerebrospinal
fluid; MMP, matrix metalloproteinase; MCAO, middle cerebral artery occlu-
sion.
1714 Laterza et al.: Identification of Novel Brain Biomarkers
NSE (9 ) has been found to increase after brain injury,
such as stroke (10, 11 ), but it is also increased in cancers of
neuroendocrine origin, such as small-cell lung cancer,
neurobastoma, carcinoid tumors, and melanoma (12 ). In
addition, hemolysis has been reported to cause positive
interference in at least one of the commercially available
procedures for its measurement (13 ).
GFAP is a monomeric intermediate filament protein
thought to be produced almost exclusively in astrocytes. It
has been reported to be increased after stroke, with peak
values around 3 days (14 ). Recently, promising results for
fatty acid binding proteins after ischemic stroke have been
reported as well (15 ).
Other efforts to identify stroke have included the
monitoring of hemostatic function. D-dimer appears to be
more sensitive than other hemostatic markers, such as
fibrinopeptide A and B-thromboglobulin, particularly for
cardioembotic stroke (16, 17 ).
Inflammatory markers have been studied as well, includ-
ing various cytokines (18 ) and matrix metalloproteinases
(MMPs), particularly MMP-9 (gelatinase B) (18–20 ); MMP-9
is associated with hemorrhagic transformation (20 ).
Recently, several studies combining various biomark-
ers have been reported. McGirth et al. (21 ) reported that a
combination of serum von Willebrand factor, MMP-9, and
vascular endothelial growth factor could predict the onset
of cerebral vasospasm and subarachoid hemorrhage. An-
other study reported that S-100B, B-type neurotrophic
growth factor, von Willebrand factor, MMP-9, and mono-
cyte chemotactic protein-1 could be effectively used. The
authors stated that if any 3 of the 5 biomarkers exceeded
their reference range, then the sensitivity was 92% and
specificity was 93% for ischemic stroke for samples taken
within 6 h from symptom onset (22 ). Another study
screened 26 biomarkers and found that 4 were highly
correlated with stroke: S100-B, MMP-9, vascular cell adhe-
sion molecule, and von Willebrand factor (23 ).
Proteomic approaches have been used with only mod-
erate success to identify potential brain biomarkers (24–
27 ). A consortium has been formed, The Human Brain
Proteome Project, to facilitate the identification of brain
proteins that may be involved in disease (http://www.
hupo.org).
Despite a large amount of effort, there have been few
successful systematic approaches to the discovery of new
brain biomarkers for disease such as stroke.
Materials and Methods
gene array analysis and bioinformatics
We systematically approached the identification of brain
biomarkers by seeking genes expressed in brain, at least
10-fold higher than in other tissues. We elected to use
mice for our experiments so that we could obtain nonde-
graded mRNA. With the advent of mouse gene expres-
sion arrays and the complete DNA sequence of mice and
human available, we thought this approach well worth
exploring.
Brain, liver, spleen, kidney, skeletal muscle, lung, pan-
creas, heart, and small intestine from 3 (2 male and 1
female) C57BL/6 mice (Jackson Laboratory), ages 4 – 6
weeks, were obtained by careful dissection. The organ
samples were snap frozen in liquid nitrogen and pro-
cessed as described (28 ) to isolate RNA. The quality of the
RNA was confirmed by (a) spectrophotometry of RNA
with an A260/A280 ratio ⬎1.9, and (b) a 28S/18S ratio of
extracted RNA ⬎1.4, as observed by RNA LabChip (Agi-
lent 2100 Bioanalyzer RNA 6000 LabChip reagent set).
Ten micrograms of the total RNA was converted to
double-stranded cDNA with an oligo(dT) primer contain-
ing the T7 promoter and used to prepare biotinylated
cRNA with the Bioarray HighYield reagent set (Enzo)
according to the manufacturer’s directions. The biotinyl-
ated cRNA probes were fragmented and applied as de-
scribed (29, 30 ) to Mouse MU74A (Version 1) Genechip®
arrays (Affymetrix). The overall fluorescence intensity
across each chip was scaled to 1500 with Affymetrix
analysis software Microarray Suite. The data were trans-
ferred to a Microsoft Excel worksheet. One selection
criterion was gene expression, with mean difference val-
ues ⬎10 000 in the brain (i.e., signal intensity of each
mRNA as computed by Affymetrix software that calcu-
lates the difference between the perfect match and mis-
match oligonucleotides for each nucleotide sequence rep-
resented on the gene chip array). The mean difference
values as described above reflect an expression signal well
above “background” and suggest sufficient abundance of
the expression products to provide a suitable marker. The
second criterion was the preferential expression of the
gene in the brain. In this case, we considered the gene to
be preferentially expressed in the brain if its RNA expres-
sion value was 10-fold higher than spleen, kidney, skeletal
muscle, lung, pancreas, heart, and small intestine. For the
genes that met these 2 criteria, we then perused their human
homologs in the bioinformatics databases Entrez Gene
(http://ncbi.nlm.nih.gov/entrez/query.fcgi?db⫽gene), On-
line Mendelian Inheritance in Man (OMIM; http://ncbi.
nlm.nih.gov/entrez/query.fcgi?db⫽OMIM), and Unigene
(http://ncbi.nlm.nih.gov/entrez/query.fcgi?db⫽unigene).
preparation of antibodies
Nucleotide sequences encoding putative marker proteins
were inserted into pGEX or pET vectors for production of
protein in Escherichia coli. Proteins from pET vectors were
purified with Qiagen Ni-NTA according to the manufac-
turer’s protocol (The Qiaexpressionist 06/2003; Qiagen).
Those from the pGEX vector were purified with immobi-
lized glutathione from Pierce following the manufactur-
er’s protocol. In some cases, plasmids were sent to Gen-
Way Biotech, Inc., for large-scale production of protein
from E. coli. These included pET 100 GAD67, pET 102
GAD67, pET 28 Zygin, and Zygin I. Quality control of the
produced protein included analysis by sodium dodecyl
sulfate–polyacrylamide gel electrophoresis, N-terminal
Edman sequencing, and mass spectrometry as appropri-
 Clinical Chemistry 52, No. 9, 2006
ate (31 ). Protein concentration was estimated by absorbance
at 280 nm with calculated extinction coefficients from the
protein sequence plus construct obtained at the Swiss-Prot
web site (http://www.expasy.org/sprot) and agreed with
visual inspection of protein-stained bands in sodium dode-
cyl sulfate–polyacrylamide gel electrophoresis.
Antipeptide immunogens were prepared as described
by Li et al. (31 ). Rabbits were immunized at Harlan
Bioproducts for Science, Inc. Rabbit serum was immuno-
purified over an affinity column containing the cognate
peptide or protein antigen.
Mice were immunized with 25 ␮g protein/mouse
immunogen in MPL-TDM adjuvant (Sigma-Aldrich) fol-
lowed by at least 2 boosts in adjuvant and 1 boost in
phosphate-buffered saline (150 nm NaCl, 50 nm sodium
phosphate pH 7.2) 3 days before fusion. Armenian ham-
sters were immunized with 100 ␮g protein/hamster in
complete Freund’s adjuvant followed by boosts in incom-
plete Freund’s adjuvant and a final boost in phosphate-
buffered saline. All fusions were performed at the Wash-
ington University School of Medicine Hybridoma Center.
Monoclonal antibodies were purified from culture media
on protein A-agarose or goat antimouse IgG-agarose or
produced in ascites by Maine Biotechnology. All purified
antibodies were dialyzed vs phosphate-buffered saline/
azide, pH 7.2, and protein concentration was estimated
from absorbance at 280 nm with an absorptivity of 1.4
(L 䡠 g⫺1 䡠 cm⫺1). Subclass determination for monoclonal
antibodies was performed with the IsoStrip reagent set
from Roche. Minimum epitope assignment was based on
immunostaining of ABIMED spot peptide arrays pre-
pared at the MIT Biopolymers Facility. Each spot com-
prised a 10-mer contiguous peptide, and depending on
the number of residues in the antigen of interest, either a
1-, 2-, or 3-residue offset was used to cover the entire
antigen sequence. For example, for a 1-residue offset, spot
1 contains sequence 1–10, spot 2 contains sequence 2–11,
spot 3 contains sequence 3–12, etc. Epitope mapping was
attempted in some cases without success (noted as not
established), and in others, it was not experimentally
determined (noted as not determined).
Monoclonal antibodies were raised against visinin-like
protein 1 (VLP-1) with a combination of DNA and protein
injections. The vector VR1012 (Vical Inc.) and certain
sequences from CTLA4Ig were used for the DNA injec-
tions. VR1012 has been optimized for protein production
in mouse skeletal muscle, whereas sequences contained in
CTLA4Ig were previously shown to greatly increase the
antibody response in mice (32 ).
Standard procedures used at the Hybridoma Center at
Washington University in St. Louis for the creation and
maintenance of the fusions were followed.
Antibodies against some of the candidate biomarkers,
used to further characterize the utility of the protein
products of candidate genes in the detection of brain
injury, were purchased through commercial sources
1715
(Chemicon, Santa Cruz Biotechnology, Synaptic Systems,
or Abnova Corp.).
western blot analysis
To confirm the presence of the proteins encoded by the
candidate genes, we performed Western blots on unaf-
fected human brain homogenate (Geno Technology, Inc.).
Likewise, to assess the specificity of the coded protein for
brain, we used a human tissue array (Geno Technology,
Inc.; 50 ␮g per tissue) and followed the manufacturer’s
specifications for the Western blot. The human tissue
array included liver, brain, lung, kidney, spleen, testis,
ovary, heart, pancreas, uterus, breast, cervix, rectum,
prostate, thyroid, laryngopharynx, stomach, and skin. For
some of the Western blots, different antibodies were
mixed 1:1 before use so that more epitopes would be
detected.
rat stroke model
A rat stroke model was used to assess the appearance of
VLP-1 in CSF as a function of time. Ischemia was induced
in femoral vein– cannulated male Sprague–Dawley rats
(Charles-River) by the previously described permanent
(p) or transient (t) middle cerebral artery occlusion
(MCAO) intraluminal filament method with some modi-
fications (33 ). Both pMCAO and tMCAO methods were
used so that different degrees of injury could be obtained.
As controls, some animals were subjected to sham oper-
ations (no filament placed). In brief, a midline incision
was performed, and the right common, internal, and
external carotid arteries were exposed. The external ca-
rotid and occipital arteries were ligated. The common
carotid artery was ligated, and the internal carotid artery
was temporarily closed. A small incision was made in the
common carotid artery and a filament (3.0 Ethilon; heat-
blunted tip) was inserted into the internal carotid artery
through the common carotid artery. The filament was
advanced 17.5 mm to occlude the origin of the middle
cerebral artery, either permanently or transiently (60 –90
min). The filament was secured in place in the right
common carotid artery with a surgical nylon suture. After
surgery, anesthesia (isoflurane) was discontinued, and the
animals were allowed to recover.
CSF was collected by cisternal puncture when the
animals were killed at either 2 h [sham (n ⫽ 4), pMCAO
(n ⫽ 5), tMCAO (n ⫽ 3)] or 24 h [sham (n ⫽ 5), pMCAO
(n ⫽ 4), tMCAO (n⫽ 3)] postsurgery. Groups were
compared by an unpaired t-test (GraphPad Prism 4 soft-
ware) at a significance of P ⫽ 0.05. Blood (EDTA plasma)
was collected, when possible, at 0, 2, 4, 8, and 24 h
postsurgery (sham- and MCAO-operated).
quantitative assays
Sandwich immunoassays were developed for VLP-1
(monoclonal antibody 3A8 –1 as a capture antibody and
rabbit antibody R3471 for detection), GAD67 (rabbit anti-
body Rb4043 as capture and Rb4610 for detection), and
zygin (mouse monoclonal IG4.4 as capture and hamster
1716 Laterza et al.: Identification of Novel Brain Biomarkers
monoclonal 4G3.1 as detection) with ruthenium-labeled
detection antibodies and electrochemiluminescent detec-
tion (Meso Scale Discovery). The lower limit of detection
for the human plasma VLP-1 assay was set to be the first
point on the calibration curve (0.04 ␮g/L), which was at least
2 SDs above the background. In addition, 3 quality control
samples were prepared by adding to heparinized plasma
various amounts of postmortem CSF (high concentration of
VLP-1), and were measured in every run. The interassay
CVs for these samples were ⱕ15% (n ⫽ 10).
Human heparinized plasma samples were retrospec-
tively collected and frozen from excess clinical laboratory
specimens that were obtained for routine care from a group
of patients who presented to Barnes-Jewish Hospital with an
acute neurological deficit between March and November of
2002. The medical histories of these patients were reviewed
by one of us (J.-M.L.) without knowledge of the VLP-1 results,
and patients with a discharge diagnosis of ischemic stroke
and a clear time of onset of stroke were identified. Eighteen
patients met this criterion, and their samples were analyzed.
Human heparinized samples were also obtained from
healthy males and females and used as controls. These
samples were not age- or sex-matched to the stroke patients.
Excess plasma samples from the Clincial Chemistry Lab-
oratory were obtained with the approval of the Washington
University IRB committee. Human CSF postmortem sam-
ples were obtained from Analytical Biological Services.
Results
identification of brain-specific genes
Twenty-nine genes were identified in the mouse as hav-
ing reasonable abundance and specificity. Twenty-six of
the 29 had human homologs that were confirmed to be
enriched in the brain in humans by the abundance of
expressed sequence tags derived from a brain source in
Unigene. Two of the gene products, MBP and NSE, have
been tested previously as brain injury markers and were
excluded from further study. S-100B and GFAP, also
previously looked at as brain injury biomarkers, did not
meet the requirement of having a signal intensity ⬎10 000
in the brain. They had signal intensities of 2500 and 5000,
respectively. Of the 24 remaining human genes,2 20 had a
predicted protein sequence chain length of Mr ⬍70 000.
2
Human genes: VSNL1, visinin-like 1; SNAP25, synaptosomal-associated
protein, Mr 25; GAD1, glutamate decarboxylase 1 (brain, Mr 67); MOBP,
myelin-associated oligodendrocyte basic protein; SYT1, synaptotagmin I;
TUBB4, tubulin ␤4; FEZ1, fasciculation and elongation protein ␨1 (zygin I);
GLRB, glycine receptor ␤; VMP, ;Vesicular membrane protein p24; OLFM1,
olfactomedin 1; ZIC1, zic family member 1 (odd-paired homolog, Drosophila);
PACSIN1, protein kinase C and casein kinase substrate in neurons 1; PLP1,
proteolipid protein 1 (Pelizaeus-Merzbacher disease, spastic paraplegia 2,
uncomplicated); INA, internexin neuronal intermediate filament protein ␣;
SLC32A1, solute carrier family 32 (␥-aminobutyric acid vesicular transporter),
member 1; SERPINI1, serine (or cysteine) proteinase inhibitor, clade I (neuro-
serpin), member 1; NNAT, neuronatin; GABRG2, ␥-aminobutyric acid A
receptor ␥2; VAMP2, vesicle-associated membrane protein 2 (synaptobrevin 2);
NRGN, neurogranin (protein kinase C substrate, RC3).
The Mr 70 000 cutoff was selected because albumin, a
protein abundant in the plasma, is known to enter the brain
after injury because of damage to the blood– brain barrier
(34 –36 ), suggesting that this cutoff value for the egress of
proteins from the brain might also be similar. The 20 genes
and their gene products are listed in Table 1 in order of the
mRNA production of their mouse homologs in brain.
Antibodies for the gene products of SLC32A1, VMP,
and GABRG2 are not commercially available, nor had we
generated antibodies as of the time of preparation of this
report, and they could not be evaluated further. Thus, 17
possible brain-specific biomarkers were selected for fur-
ther investigation. A summary of the antibodies used to
further assess the candidate biomarker is shown in Table 2
(antibodies prepared internally) and Table 3 (antibodies
purchased commercially). The immunogen was in the
protein form unless otherwise indicated.
western blot analysis of gene products
Of the 17 gene products for which antibodies were avail-
able, Western blots showed strong signals from brain ex-
tracts for 13 of them. We did not find any signal for the
gene products of the genes family member 1, proteolipid
protein 1, neuronatin, or olfactomedin 1 (ZIC1, PLP1, NNAT,
or OLFM1) with the antibodies described in Table 3. We
then tested the reactivity of the 13 gene products found in
brain in different tissues by Western blot. The tissue
Western blot for VLP-1 is shown in Fig. 1. We scored the
intensity of the Western blots of the tissue arrays for the
13 candidate gene products as follows: 5, very strong; 4,
strong; 3, moderate; 2, weak; 1, very weak. Results are
summarized in Table 4. Western blots were also per-
formed on human postmortem CSF at various dilutions,
and the intensity of the bands compared visually with
standards of the respective proteins. Zygin, neuroserpin,
and GAD67 were below the limit of detection of this
method (⬍2.5 ␮g/L), and the intensity of the VLP-1 band
was between the 25 and 250 ␮g/L VLP-1 standards. This
range was in agreement with the quantitative assay deter-
mination of 17 ␮g/L for VLP-1. The quantitative assays
detected GAD67 at 0.3 ␮g/L and zygin at 0.6 ␮g/L in the
postmortem spinal fluid. No quantitative assay for neuro-
serpin was available at the time of these experiments.
rat stroke model
VLP-1 was not detected in the CSF of sham-operated rats,
whereas VLP-1 was found in high abundance in the CSF
of MCAO-operated rats 24 h post injury (Fig. 2), suggest-
ing neuronal damage. The concentration [mean (SE)]
of VLP-1 in the CSF of pMCAO rats was higher [16.8
(6.5 ␮g/L)] than in tMCAO [6.0 (2.4 ␮g/L)], although the
difference was not statistically different (P ⫽ 0.23, un-
paired t-test). No VLP-1 was detected in the CSF of sham-
or tMCAO-operated rats at 2 h postinjury. Interestingly,
no VLP-1 were detected in EDTA plasma of either sham-
or MCAO-operated rats at any of the time points tested
up to 24 h postinjury.
 Clinical Chemistry 52, No. 9, 2006 1717

Table 1. Human genes with high mRNA abundance in brain relative to other tissues.a
Gene symbol Gene name Entrez gene ID Unigene cluster Microarray units
VSNL1 Visinin-like 1 7447 Hs0.444212 25 837.6
SNAP25 Synaptosomal-associated protein, Mr 25 000 6616 Hs0.167317 25 438.5
GAD1 Glutamate decarboxylase 1 (brain, Mr 67 000) 2571 Hs0.420036 20 930.6
MOBP Myelin-associated oligodendrocyte basic protein 4336 Hs0.121333 20 079.4
SYT1 Synaptotagmin I 6857 Hs0.310545 15 650.1
TUBB4 Tubulin ␤4 10382 Hs0.110837 15 537.5
FEZ1 Fasciculation and elongation protein ␨1 (zygin I) 9638 Hs0.224008 14 864.1
GLRB Glycine receptor ␤ 2743 Hs0.32973 14 726.1
VMP Vesicular membrane protein p24 140767 Hs0.49230 14 185
OLFM1 Olfactomedin 1 10439 Hs0.522484 13 884.6
ZIC1 Zic family member 1 (odd-paired homolog, Drosophila) 7545 Hs0.41154 13 456.4
PACSIN1 Protein kinase C and casein kinase substrate in 29993 Hs0.520087 13 142.6
neurons 1
PLP1 Proteolipid protein 1 (Pelizaeus-Merzbacher disease, 5354 Hs0.1787 13 009.3
spastic paraplegia 2, uncomplicated)
INA Internexin neuronal intermediate filament protein ␣ 9118 Hs0.500916 12 581.7
SLC32A1 Solute carrier family 32 (GABAb vesicular transporter), 140679 Hs0.179080 12 533.7
member 1
SERPINI1 Serine (or cysteine) proteinase inhibitor, clade I 5274 Hs0.478153 12 309.3
(neuroserpin), member 1
NNAT Neuronatin 4826 Hs0.504703 12 077.8
GABRG2 GABA A receptor ␥2 2566 Hs0.7195 10 620
VAMP2 Vesicle-associated membrane protein 2 6844 Hs0.25348 10 586.5
(synaptobrevin 2)
NRGN Neurogranin (protein kinase C substrate, RC3) 4900 Hs0.524116 10 321.5
a
Microarray units are arbitrary units calculated by Microarray Suite software. The values range from 32 672.2 (for mouse cytoskeletal ␥ actin mRNA) to null (negative
values for expression were converted to null values).
b
GABA, ␥ -aminobutyric acid.

determination of vlp-1 in stroke patients
Thirty-nine plasma samples from unaffected donors were
analyzed, and 37 of them were below the limit of detec-
tion of 0.04 ␮g/L; the others had low values of 0.08 and
0.11 ␮g/L. Of the 18 patients with confirmed stroke, only
2 had no samples with detectable VLP-1; most were
considerably higher. Shown in Fig. 3 are the results at
various times after stroke onset for 18 patients.
Discussion
Here, we describe the successful identification of brain-
specific biomarkers with a systematic approach that in-

Table 2. Antibodies prepared to 4 possible brain biomarkers.

Antibody designation Species Isotype Immunogen Epitope(s)
R3471 Rabbit IgG GST-VLP-1 Major: aaa 3–11, 16–23, 139–155
4399 3A8.1 Mouse IgG1K VLP-1 DNA and GST-VLP-1, boost Not established
4403 2B9.3 Mouse IgG2aK VLP-1 DNA and GST-VLP-1, boost Not established
R4609 Rabbit IgG His-GAD 67 Multiple epitopes
R4610 Rabbit IgG His-GAD 67 Multiple epitopes
R4043 Rabbit IgG GAD 67 peptide(61–79) Presumed aa 61–79
R4044 Rabbit IgG GAD 67 peptide(79–97) Presumed aa 79–97
4554 1G4.4 Mouse IgG1K His-Zygin aa 23–28
4563 4G3.1 Hamster IgG1 His-Zygin aa 7–12
R4726 Rabbit IgG His-Zygin Multiple epitopes
R4727 Rabbit IgG His-Zygin Multiple epitopes
4421 2G10.2 Mouse IgG1K Neuroserpin DNA and GST-neuroserpin aa 145–154
protein boost for all
4421 5B5.1 Mouse IgG1K aa 370–379
4421 7D6.3 Mouse IgG2aK Not established
4505 2F1.1 Mouse IgG1K aa 193–202
a
aa, amino acid(s).
1718 Laterza et al.: Identification of Novel Brain Biomarkers

Table 3. Antibodies purchased to assess possible brain biomarkers.

Antibody designationa Speciesb Isotype Immunogen Epitope(s)
SC-20038 Mouse IgG1K Crude brain extract containing synaptosomal- Not determined
associated protein, Mr 25 000
Chemicon MAB 5406 Mouse IgG2aK r-Rat GAD 67 aa 13–25
SC-14520 Goat IgG Myelin-associated oligodendrocyte basic protein Internal peptide
SC-7753 Goat IgG Synaptotagmin I N terminus
SC-5274 Mouse IgG2b Tubulin ␤4 aa214–444
SC-17283 Goat IgG Glycine receptor ␤ N terminus
SC-17285 Goat IgG Glycine receptor ␤ C terminus
ABN-H10349 Mouse IgG1 GST-olfactomedin 1 Not determined
SC-28148 Goat IgG Zic family member 1 Near N
terminus
SC-30127 Rabbit IgG Mouse protein kinase C and casein kinase Internal peptide
substrate in neurons 1
SC-23570 Goat IgG Proteolipid protein 1 N terminus
SC-18529 Goat IgG Proteolipid protein 1 Internal peptide
SC-7571 Goat IgG Internexin neuronal intermediate filament protein ␣ N terminus
SC-7570 Goat IgG Internexin neuronal intermediate filament protein ␣ C terminus
SC-23437 Goat IgG Neuronatin C terminus
SS-104211 Mouse IgG1 Rat synaptobrevin 2 N terminus
SC-18336 Goat IgG Neurogranin Internal peptide
a
Chemicon, antibodies bought from Chemicon International; SC, antibodies obtained commercially from Santa Cruz Biotechnologies; SS, antibodies obtained
commercially from Synaptic Systems GmBH; ABN, antibodies obtained commercially from Abnova Corp.
b
Mouse antibodies are monoclonal, whereas the rabbit and goat antibodies are polyclonal and affinity-purified.

volved the use of a mouse model and RNA expression
analyses. We chose this approach as a sensitive way to
identify proteins in brain that are not likely to be abun-
dant in other tissues. We used this approach rather than a
proteomic one because the mouse and human genome
were known, means of rapidly obtaining tissue RNA and
measuring the RNA expression were readily available,
and interpretation of the data could be performed via
bioinformatic analysis of public databases. We have com-
pared this genomic approach with proteomic approaches
in public databases for biomarkers of myocardial infarc-
tion and found the genomic approach superior (our

Fig. 1. Western blot of human tissue array (Geno Technology, Inc.; 50 ␮g per tissue) for VLP-1. SBTI, soybean trypsin inhibitor Mr 20 000; Lys, hen
egg white lysozyme Mr 14 400.
 Clinical Chemistry 52, No. 9, 2006 1719

Table 4. Western blot of normal human tissue arrays for protein products from 13 candidate genes.a
Liver Brain Lung Kidney Spleen Testis Ovary Heart Pancreas Uterus Breast Cervix Rectum Prostate Thyroid Laryngo Stomach Skin
Protein
VLP-1 1 5 1 2 4 2 2 1 3
Synaptosomal- 5 1 1 1 1 1 1
associated protein,
Mr 25 000
GAD67 3
Myelin-associated 4 2 3 2 2 2 2
oligodendrocyte
basic protein
Synaptotagmin I 3 5 2 3 3 2 2 5 2 3 2 3 3 2 1 1 3 3
Tubulin ␤ 4 2 5 2 2 1 2 2 1 3 3 2 3 4 4 3 1 4 3
Zygin 3
Glycine receptor ␤ 1 3 4 3 3 5 3 2 5 2 3 5 5 1 1 5 3
Protein 1 kinase C 5 1 1 1 1 1 1 1
and casein kinase
substrate in
neurons 1
Internexin neuronal 4
intermediate
filament protein ␣
Neuroserpin 3 1 1 2 2
Synaptobrevin 2 5 1 1 1
Neurogranin 1
a
Scoring system: 5, very strong; 4, strong; 3, moderate; 2, weak; 1, very weak. No score, not detected.

unpublished data). The governing principle of our exper-
iments was to search for markers of brain cell death,
following the rationale that, upon death of the cells, their
contents will enter into the extracellular matrix and fur-
ther into the CSF. Because the blood– brain barrier is
compromised during a stroke, we hypothesized that these
biomarkers of brain cells injury will ultimately find their
way into the systemic circulation.
Through our analysis, we were able to identify genes
that are preferentially and abundantly expressed in the
mouse brain, and through the use of bioinformatics, were
able to confirm the presence of the homologs of these
genes in the human brain and gained further insight into
their molecular characteristics. Western blot analyses
showed expression in the brain with reasonable specific-
ity for 10 of 13 of the gene products.
Furthermore, we developed sandwich immunoassays
to detect the presence of some of these proteins in
postmortem CSF and the CSF and sera of rats that had
undergone stroke. Postmortem CSF was tested because it
should reflect a condition in which the brain has been

Fig. 3. VLP-1 concentration in 55 plasma samples from 18 different

Fig. 2. VLP-1 concentration in CSF of sham, transient (t), or permanent
(p) MCAO-operated rats 24 h postsurgery.
patients with ischemic strokes, obtained at different time periods after
the onset of stroke.
The bottom panel shows the number of samples obtained within a given time
period and the percentage of these samples in which VLP-1 was detected.
Positive was considered to be a detectable value ⱖ0.04 ␮g/L. VLP-1 was
undetected in 36 of 39 samples from unaffected donors.
1720 Laterza et al.: Identification of Novel Brain Biomarkers
deprived of oxygen for a period of time long enough to
cause extensive cell death. We put special emphasis on the
development of a sensitive assay for VLP-1, given its ideal
molecular characteristics (cytoplasmic protein of low mo-
lecular weight) and its high abundance in brain (most
abundant mRNA in our search). Also, VLP-1, a calcium-
sensor protein found in mammalian CNS neurons, has
been found to have a widespread distribution in the brain
and to be abundant in all brain areas except the caudate–
putamen by in situ mRNA hybdridization studies (37 ).
We found similar intensity for Western blots of various
human brain regions for VLP-1 (TB57 brain tissue region
blot; Geno Technology) (data not shown). We were able to
demonstrate that VLP-1 was present in the CSF of an
animal model of stroke and in the blood of stroke patients,
as well as in CSF obtained postmortem. It was interesting
that we were not able to detect VLP-1 in the serum of a rat
model of stroke. This could be caused by several factors;
we observed a decrease in the detection limit of our
electrochemiluminescence assay with rat EDTA plasma
compared with human heparinized plasma. Also, 24 h
postinjury may not be long enough to allow VLP-1 to
dissipate from the CSF into the bloodstream of the injured
rats. These rats showed no increase in VLP-1 in CSF 2 h
postinjury but did at 24 h. Similarly, 24 h may not be
sufficient time for VLP-1 to appear in the blood. It would
be interesting to have blood measurements in these ani-
mals beyond 24 h, but it will be difficult because there are
complications in keeping these animals alive beyond 24 h
postinjury. One study supports the idea of delayed ap-
pearance of neuronal injury biomarkers into the CSF, as
has been reported for other biomarkers (38 ). This study
looked at CSF markers after acute stroke and saw no
increase of total Tau in CSF at day 0 –1 after acute
ischemic stroke, but we did see increase at days 2–3,
peaking after 1 week and returning to unaffected levels
after 3–5 months. Our studies of VLP-1 samples from
patients after stroke are promising, but they were from
samples obtained retrospectively from excess samples
obtained for their routine clinical care and therefore were
not systematic as regards time after stroke.
Our results demonstrate that the identification of pos-
sible cell damage biomarkers via a genomic approach is a
viable one and likely generally applicable for other cell
damage markers. This approach has led to the identifica-
tion of several potential biomarkers of brain injury. We
are currently developing sensitive assays for these mark-
ers and further systematically assessing their ability to
assess brain damage in human disease.
We thank Lorrie P. Daggett of Merck for assistance with
the rat model of stroke; we thank Michael Tanen and
Derek Chappell of Merck for assistance with the VLP-1
immunoassay. We also thank Nancy Brada for recombi-
nant protein production and Matt Ohlendorf and Mary
Jane Eichenseer for help with immunoassay development.
O.F.L., V.R.M., Y.L., and J.H.L. are named as co-inventors
on pending patents filed by Washington University con-
cerning brain biomarkers.
References
1. Schwamm LH, Pancioli A, Acker JE III, Goldstein LB, Zorowitz RD,
Shepard TJ, et al. Recommendations for the establishment of
stroke systems of care: recommendations from the American
Stroke Association’s Task Force on the Development of Stroke
Systems. Circulation 2005;111:1078 –91.
2. Warlow C, Sudlow C, Dennis M, Wardlaw J, Sandercock P. Stroke.
Lancet 2003;362:1211–24.
3. The Joint European Society of Cardiology/American College of
Cardiology Committee. Myocardial infarction redefined—a consen-
sus document of The Joint European Society of Cardiology/
American College of Cardiology committee for the redefinition of
myocardial infarction. J Am College Cardiol 2000;36:959 – 69.
4. Rothermundt M, Peters M, Prehn, JHM, Arolt V. S100B in brain
damage and neurodegeneration. Microsc. Res. Tech. 2003;60:
614 –32.
5. Foerch C, Otto B, Singer OC, Neumann-Haefelin T, Yan B,
Berkefeld J, et al. Serum S100B predicts a malignant course of
infarction in patients with acute middle cerebral artery occlusion.
Stroke 2004;35:2160 – 4.
6. Vaage J, Anderson R. Biochemical markers of neurologic injury in
cardiac surgery: the rise and fall of S100␤. J Thorac Cardiovasc
Surg 2001;122:853–5.
7. Anderson RE, Hansson LO. Nilsson O, Dijlai-Merzoug R, Settergren
G. High serum S100B levels for trauma patients without head
injuries. Neurosurgery 2001;48:1255– 60.
8. Anderson RE, Hansson LO, Nilsson O, Liska J, Settergren G,
Vaage J. Increase in Serum S100A1-B and S100BB during cardiac
surgery arises from extracerebral sources. Ann Thorac Surg 2001;
71:1512–7.
9. Schmechel D, Marangos PJ, Zis AP, Brightman M, Goodwin FK.
Brain enolases as specific markers of neuronal and glial cells.
Science 1978;199:313–5.
10. Persson L, Hardemark HG, Gustafsson J, Rundstrom G, Mendel-
Hartvig I, Esscher T, et al. S-100 protein and neuron-specific
enolase in cerebrospinal fluid and serum: markers of cell damage
in human central nervous system. Stroke 1987;18:911– 8.
11. Anand N, Stead LG. Neuron-specific enolase as a marker for acute
ischemic stroke: a systematic review. Cerebrovasc Dis 2005;20:
213–9.
12. Muley T, Ebert W, Stieber P, Raith H, Holdenrieder S, Nagel D, et
al. Technical performance and diagnostic utility of the new Elec-
sys® neuron-specific enolase enzyme immunoassay. Clin Chem
Lab Med 2003;41:95–103.
13. Beaudeux J, Léger P, Dequen L, Gandjbakhch I, Coriat P, Foglietti
M. Influence of hemolysis on the measurement of S-100␤ protein
and neuron-specific enolase plasma concentrations during coro-
nary artery bypass grafting. Clin Chem 2000;46:989 –90.
14. Herrmann M, Vos P, Wunderlich MT, de Bruijn CH, Lamers KJB.
Release of glial tissue-specific proteins after acute stroke: a
comparative analysis of serum concentrations of protein S-100B
and glial fibrillary acidic protein. Stroke 2000;31:2670 –7.
15. Pelsers MM, Glatz JFC. Detection of brain injury by fatty acid-
binding proteins. Clin Chem Lab Med 2005;43:802–9.
16. Fon EA, Mackey A, Cote R, Wolfson C, McIlraith DM, Leclerc J, et
al. Hemostatic markers in acute transient ischemic attacks.
Stroke 1994;25:282– 6.
17. Feinberg WM, Erickson LP, Bruck D, Kittelson J. Hemostatic
markers in acute ischemic stroke: association with stroke type,
severity, and outcome. Stroke 1996;27:1296 –1300.
 Clinical Chemistry 52, No. 9, 2006
18. Castillo J, Rodriguez I. Biochemical changes and inflammatory
response as markers for brain ischaemia: molecular markers of
diagnostic utility and prognosis in human clinical practice. Cere-
brovasc Dis 2004;17(Suppl 1):7–18.
19. Montaner J, Alvarez-Sabı́n J, Molina C, Anglés A, Abilleira S,
Arenillas J, et al. Matrix metalloproteinase expression after hu-
man cardioembolic stroke: temporal profile and relation to neuro-
logical impairment. Stroke 2001;32:1759 – 66.
20. Castellanos M, Leira R. Serena J, Pumar JM, Lizasoain I, Castillo
J. et al. Plasma metalloproteinase-9 concentration predicts hem-
orrhagic transformation in acute ischemic stroke. Stroke 2003;
34:40 – 6.
21. McGirt MJ, Lynch JR, Blessing R, Warner DS, Friedman AH, et al.
Serum von Willebrand factor, matrix metalloproteinase-9, and
vascular endothelial growth factor levels predict the onset of
cerebral vasospasm after aneurysmal subarachnoid hemorrhage.
Neurosurgery 2002;51:1128 –35.
22. Reynolds MA, Kirchick HJ, Dahlen JR, Anderberg JM, McPherson
PH, Nakamura KK, et al. Early biomarkers of sroke. Clin Chem
2003;49:1733–9.
23. Lynch JR, Blessing R, White WD, Grocott HP, Newman MF,
Laskowitz DT. Novel diagnostic test for acute stroke. Stroke
2004;35:57– 63.
24. Lubec G, Krapfenbauer K, Fountoulakis M. Proteomics in brain
research: potentials and limitations. Prog Neurobiol 2003;69:
193–211.
25. Gauss C, Kalkum M, Lowe M, Lehrach H, Klose J. Analysis of the
mouse proteome. (I) Brain proteins: separation by two-dimen-
sional electrophoresis and identification by mass spectrometry
and genetic variation. Electrophoresis 1999;20:575– 600.
26. Lescuyer P, Allard L, Zimmermann-Ivol CG, Burgess JA, Hughes-
Frutiger S, Burkhard PR, et al. Identification of post-mortem
cerebrospinal fluid proteins as potential biomarkers of ischemia
and neurodegeneration. Proteomics 2004;4:2234 – 41.
27. Prados J, Kalousis A, Sanchez JC, Allard L, Carrette O, Hilario M.
1721
Mining mass spectra for diagnosis and biomarker discovery of
cerebral accidents. Proteomics 2004;4:2320 –32.
28. Chomczynski, P, Sacchi, N. Single-step method of RNA isolation
by acid guanidinium thiocyanate-phenol-chloroform extraction.
Anal Biochem 1987;162:156 –9.
29. Lockhart DJ, Dong H, Byrne MC, Follettie MT, Gallo MV, Chee MS,
et al. Expression monitoring by hybridization to high-density oligo-
nucleotide arrays. Nat Biotechnol 1996;14:1675– 80.
30. Lipshutz RJ, Fodor SP, Gingeras TR, Lockhart DJ. High density
synthetic oligonucleotide arrays. Nat Genet 1999;21(Suppl 1):
20 – 4.
31. Li A, Crimmins DL, Luo Q, Hartupee J, Landt Y, Ladenson JH, et al.
Expression of a novel regenerating gene product, Reg IV, by high
density fermentation in Pichia pastoris: production, purification,
and characterization. Protein Expr Purif 2003;31:197–206.
32. Boyle JS, Brady JL, Lew AM. Enhanced responses to a DNA
vaccine encoding a fusion antigen that is directed to sites of
immune induction. Nature 1998;392:408 –11.
33. Longa EZ, Weinstein PR, Carlson S, et al. Reversible middle
cerebral artery occlusion without craniectomy in rats. Stroke
1989;20:84 –91.
34. Paterlini M, Revilla V, Grant AL, Wisden W. Expression of the
neuronal calcium sensor protein family in the rat brain. Neuro-
science 2000;99:205–16.
35. Cornford EM, Hyman S. Blood-brain barrier permeability to small
and large molecules. Adv Drug Deliv Rev 1999;36:145– 63.
36. Hampel H, Kötter HU, Möller HJ. Blood-cerebrospinal fluid barrier
dysfunction for high molecular weight proteins in Alzheimer dis-
ease and major depression: indication for disease subsets.
Alzheimer Dis Assoc Disord 1997;11:78 – 87.
37. Gingrich MB, Traynelis SF. Serine proteases and brain dam-
age—is there a link? Trends Neurosci 2000;23:399 – 407.
38. Rosengren HC, Vanmechelen E, Vanderstichele H, Jensen C,
Davidsson P, Blennow K. Cerebrospinal fluid markers for Alzhei-
mer’s disease evaluated after acute ischemic stroke. J Alzhei-
mers Dis 2000;2:199 –206.