Identifying novel autoantibody signatures in ovarian cancer using
high-density protein microarrays
C. Geeth Gunawardanaa,b, Nader Memaria,b, Eleftherios P. Diamandisa,b,c,⁎
aDepartment of Pathology and Laboratory Medicine, Mount Sinai Hospital, 6th Floor, Room m6-201, 60 Murray Street, Toronto, ON, Canada
bDepartment of Laboratory Medicine and Pathobiology, The University of Toronto, Toronto, ON, Canada
cDepartment of Clinical Biochemistry, University Health Network, Toronto, ON, Canada
Received 25 June 2008; received in revised form 13 November 2008; accepted 13 November 2008
Available online 3 December 2008
Objectives: To identify autoantibody signatures in ovarian cancer using protein microarray technology.
Design and methods: Protein microarrays were screened using non-malignant peritoneal fluid (n=30) and ascites fluid pooled from ovarian
cancer patients (n=30).
Results: Fifteen potential tumour-associated antigens were discovered. AASDHPPT showed the strongest signal-to-noise ratio.
Conclusions: Protein microarrays are suitable for autoantibody discovery in ovarian cancer but the signatures are of low frequency.
© 2008 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
Keywords: Tumour-associated antigens; Protein microarray; Ovarian cancer
Epithelial ovarian cancer in particular, comprises more than
80% of the ovarian cancer cases . It is often diagnosed in the
late stages when the cancer has metastasized to other organs in
the peritoneum. The 5-year survival rate for patients with
advanced disease (stages III and IV) is 10–20%. In contrast, the
5-year survival rate for patients diagnosed with early-stage
disease can be high as 90% . These numbers clearly support
the need for early diagnosis.
CA-125 is the clinically accepted biomarker for ovarian
cancer and is used routinely to monitor patients' response to
therapy and recurrence of disease. However, the test is poor for
early detection and risk prediction due to frequent false-positive
and false-negative results . Thus, we need to discover new
biomarkers that perform better as screening tools. Autoantibody
responses to tumour-associated antigens (TAAs) can have both
diagnostic and prognostic value. Some documented autoanti-
body responses to tumours include ones against p53 , NY-
ESO-1 , MUC-1 , and Tyrosinase . In this study we
used protein microarray technology to identify autoantibody
signatures in ovarian cancer.
Materials and methods
Ascites fluids from ovarian cancer patients with primary and
recurrent disease were either collected during surgery or
withdrawn at paracentesis. Non-malignant peritoneal fluid
was collected from female patients with benign pathologies.
All samples were kept frozen at −20 °C until analysis. Our
protocols have been approved by the Institutional Review
Board of Mount Sinai Hospital, Toronto, Canada.
Protein microarray screen
ProtoArrays® were purchased from Invitrogen Canada Inc.
(Burlington, Ontario, Canada). Microarrays were screened
Available online at www.sciencedirect.com
Clinical Biochemistry 42 (2009) 426–429
⁎Corresponding author. Department of Pathology and Laboratory Medicine,
Mount Sinai Hospital 6th Floor, Room 6-201, Box 32, 60 Murray Street,
Toronto, Ontario, Canada M5T 3L9. Fax: +1 416 619 5521.
E-mail address: firstname.lastname@example.org (E.P. Diamandis).
0009-9120/$ - see front matter © 2008 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
according to the instructions provided by the manufacturer.
Briefly, the microarrays were blocked for 1 h with blocking
buffer. Ascites fluid, pooled from 30 patients with ovarian
carcinoma of the serous type, was used as the source of
primary antibodies at a dilution of 1:200 in probing buffer,
and 120 μl of sample was overlaid on the microarrays. For
the control, 30 specimens of non-malignant peritoneal fluid
were pooled together, and the dilution was identical to that
of the cancer pool. Arrays were incubated for 2 h at 4 °C,
followed by washes with a proprietary wash buffer. Bound
antibodies were detected using a mouse anti-human IgG con-
jugated to Alexa Fluor 647 (Molecular Probes, Carlsbad, CA,
USA). Signals were detected using a PE ScanArray Express
Signal analysis was performed using Invitrogen's Proto-
Array® Prospector v4. Briefly, the protein-protein interaction
mode was selected in the software. ProtoArray® Prospector
calculates the mean value and standard deviation of the signals
of each protein on the microarray, followed by the calculation of
the z-score for each protein.
Recombinant AASDHPPT production
The protocol and materials for the production of recombinant
protein has been described previously . Primer sequences
used for PCR amplification and cloning of AASDHPPT was as
follows: Forward primer, 5′ CAC CGT TTT CCC TGC CAA
ACG GTT CTG 3′; Reverse primer, 5′ AGG GAA TCA TCA
TGA CTT TGT ACC A 3′ (ACGT Corporation, Toronto, ON,
Canada). Tandem mass spectroscopy (Thermo Finnigan LTQ)
was used to confirm the identity of recombinant AASDHPPT.
Enzyme-linked immunosorbent assay
Briefly, 96-well plates were coated with sheep anti-mouse
immunoglobulin (Jackson Immunoresearch Laboratories
Inc., West Grove, PA, USA). The ELISA plates were then
incubated with mouse antibodies specific to N-terminal His
tags, followed by incubations with either recombinant
AASDHPPT or BCZ4. The latter protein has no homology
to AASDHPPT but was produced in the same system and has
a poly-His tag. It was used as a negative control. The wells
were then incubated with ascites fluid from patients with
serous ovarian carcinoma; 4 wells per ascites sample (2 wells
for AASDHPPT and 2 wells for BCZ4). Binding events were
detected using mouse anti-human IgG antibodies conjugated
to alkaline phosphatase (Jackson Immunoresearch Labora-
Antibody isolation and purification
Antibodies were purified from the pooled ascites fluid used
to screen the protein microarrays. Protein A affinity purifica-
tion was performed using the kit system, MAPS (Bio-Rad
Laboratories, Hercules, CA, USA), according to the manufac-
Western blots for AASDHPPT were performed using
recombinant AASDHPP. Membranes were blocked for 1 h at
room temperature using 5% non-fat milk in PBS-Tween (0.1%
v/v). Antibodies isolated from ascites fluid pooled from 30
patients were used as the source of primary antibodies (1:500
dilution in PBS containing 0.1% Tween) and membranes were
incubated overnight at 4 °C. Individual ascites samples were
also diluted 1:500 in PBS containing 0.1% Tween. Membranes
were washed using PBS-Tween (0.1% v/v). Goat anti-human
IgG coupled to alkaline phosphatase (1:10000 dilution in PBS
containing 0.1% Tween and 5% non-fat milk) was used as the
secondary antibody. Signals were developed and captured on
film by using a chemiluminescent substrate (Diagnostics
Product Corporation, Los Angeles, CA, USA).
List of proteins found on microarray screen
Identified on both microarrays
E3 ubiquitin-protein ligase
Homo sapiens cDNA clone
Regulator of G-protein
signalling 13 (RGS13)
Homo sapiens lymphocyte
cytosolic protein 2 (LCP2)
Serine/threonine kinase 25
O14921Increases GTPase activity
of G-protein alpha
T-cell antigen receptor
Oxidative stress activated
regulation of transcription
from RNA polymerase II
modification of target
Vacuolar sorting protein SNF8Q96H20
High mobility group nucleosomal
binding domain 3 (HMGN3)
Chromosome 18 open reading
Q15651Thyroid receptor binding
Identified on one microarray
Large proline-rich protein BAT3
F-box/WD repeat protein 1B
Homo sapiens cDNA
‡, which are NCBI identifiers.
427 C.G. Gunawardana et al. / Clinical Biochemistry 42 (2009) 426–429
Screening microarays with ascites fluid
Two identical human protein microarrays were screened
with ascites fluid pooled from 30 patients with ovarian cancer.
The first experiment identified 10 proteins that were
potentially antigenic. Replicate protein features whose mean
signals were 3 standard deviations or greater than the mean
signals of all protein features were considered as true binding
events. Table 1 lists these proteins, their corresponding
database identifiers, and the normalized signals (in arbitrary
units). In the second identical experiment, the same 10 proteins
plus 5 other proteins were identified. As a control, we screened
a microarray with non-malignant peritoneal fluid. All proteins
identified in the control experiment were eliminated from the
compiled list of potential antigenic proteins. Of the 15
phosphopantetheinyl transferase (AASDHPPT) had the highest
Verification of AASDHPPT as an immunogenic protein
To confirm that AASDHPPT was indeed immunogenic,
we conducted Western blots with recombinant AASDHPPT,
and the pooled ascites fluid as the primary antibody source
(Fig. 1). BCZ4 (His-tagged protein produced in E. coli) was
used as a negative control to verify that non-specific
binding to the His-tag was not occurring. The estimated size
on the Western blot. There was no binding to BCZ4 (∼37 kDa
Ascites samples used to create the pooled sample were used
in Western blots to examine which samples contained anti-
AASDHPPT antibodies. One sample contained a high-titre of
anti-AADHPPT antibodies relative to the other samples, as
indicated in Fig. 1B. A dose-response experiment was
conducted using this particular ascites, to verify that the binding
was specific (Fig. 1C). For comparison, the same experiment
was repeated using an anti-AASDHPPT negative ascites.
Binding to recombinant AASDHPPT increased with increasing
amount of protein. This was not seen in the Western blot with
the anti-AASDHPPT negative ascites.
Validation of AASDHPPT by ELISA
We screened 100 ascites samples from patients with serous
ovarian carcinoma by ELISA, for anti-AASDHPPT activity.
Raw counts for AASDHPPT were compared with the
corresponding counts for BCZ4. Raw counts for AASDHPPT
that were at least twice that of the corresponding counts for
BCZ4 indicated presence of anti-AASDHPPTantibodies. From
the 100 ascites samples screened, only one was positive. This
was the same sample that showed positive reactivity in the
Western blots (Fig. 1B). The average raw count for the anti-
AASDHPPT positive ascites was 8 times greater than that
measured for corresponding BCZ4 control (49637 versus 6199
arbitrary units respectively).
Our screens yielded 15 proteins that were candidates for
further study as TAAs, 10 of which were reproducible in the
cancer set. To date, none of these proteins have been studied in
ovarian cancer. AASDHPPT was a good candidate for further
study, given the high signal-to-noise ratio. However, testing
individual ascites samples revealed that only one sample out of
the 100 ascites tested positive for anti-AASDHPPT antibodies.
Fig. 1. (A) Western blot analysis for anti-AASDHPPT antibodies in the pooled ascites sample from ovarian cancer patients. Lane 1: human IgG (control); Lane 2:
recombinant BCZ4 produced in E. coli; Lane 3: recombinant AASDHPPT produced in E. coli. (B) Western blot analysis for anti-AASDHPPTantibodies in individual
ascites samples. Recombinant AASDHPPT was used as the antigen. Panels 1–9 represent individual Western blots. (C) Western blot for AASDHPPT using an anti-
AASDHPPT antibody positive and negative ascites. Lanes 1–3 have increasing amounts of AASDHPPT. Top panel is the Western blot using the anti-AASDHPPT
positive ascites. The bottom is the Western blot using anti-AASDHPPT negative ascites.
428C.G. Gunawardana et al. / Clinical Biochemistry 42 (2009) 426–429
This sample was also part of the pooled ascites sample used in
the two discovery screens. Thus, the relatively high signal seen
on the microarrays for AASDHPPT is likely due to the
contribution from this single ascites sample, rather than a
cumulative contribution from several ascites. The Western blots
for AASDHPPT using individual ascites fluids confirm this
observation. The strong signal present in our two microarrays
was also validated by the strong signal detected by ELISA.
Although the experiments did not succeed in demonstrating
AASDHPPT as a frequent TAA of ovarian cancer, we
demonstrated that protein microarrays are suitable for uncover-
ing autoantibody responses, as also reported by others .
The genesis of the autoantibody response in malignancy is
still poorly understood. This is complicated further by the rarity
of autoantibody responses to tumours. For example, anti-p53
antibodies are seen in only 18% of ovarian cancer patients 
and anti-NY-ESO-I antibodies are detected in no more than 20%
of lung cancer patients . Our results substantiate this
Based on the low frequency of autoantibody responses to
tumours and the difficulties in detecting new responses, it is
unlikely that an autoantibody response to a single TAA will
provide the superior sensitivity and specificity needed for an
early detection tool. This was clearly seen with AASDHPPT.
The prevailing view is the use of multiparametric analysis in
microarray format. Success with a multiparametric detection
assay has been demonstrated in prostate cancer where a
peptide microarray constructed from 22 putative TAAs,
detected prostate cancer with a specificity of 88.2% and a
sensitivity of 81.6% . These results were better than using
To conclude, protein microarrays are suited for detecting
autoantibody responses. Moreover, their ease of use, the low
amounts of reagents and sample required and the number of
molecules that can be tested per array make them ideal for
multiparametric testing. Although we did not succeed in finding
a TAA that was a good biomarker, this is not due to the
limitations of the protein microarray, but rather due to the nature
of autoantibody responses in cancer.
This work was supported by a Collaborative Research and
Development grant from the Natural Sciences and Engineering
Research Council of Canada (NSERC) and Sanofi Pasteur
Cancer Vaccine Program.
 Williams TI, Toups KL, Saggese DA, Kalli KR, Cliby WA, Muddiman
DC. Epithelial ovarian cancer: disease etiology, treatment, detection, and
investigational gene, metabolite, and protein biomarkers. J Proteome Res
 Cannistra SA. Cancer of the ovary. N Engl J Med 2004;351:2519–29.
 Moss EL, Hollingworth J, Reynolds TM. The role of CA125 in clinical
practice. J Clin Pathol 2005;58:308–12.
 Angelopoulou K, Diamandis EP. Detection of the TP53 tumour suppressor
gene product and p53 auto-antibodies in the ascites of women with ovarian
cancer. Eur J Cancer 1997;33:115–21.
 Tureci O, et al. Humoral immune responses of lung cancer patients against
tumor antigen NY-ESO-1. Cancer Lett 2006;236:64–71.
 von Mensdorff-Pouilly S, et al. Humoral immune response to polymorphic
epithelial mucin (MUC-1) in patients with benign and malignant breast
tumours. Eur J Cancer 1996;32A:1325–31.
 Fishman P, Merimski O, Baharav E, Shoenfeld Y. Autoantibodies to
tyrosinase: the bridge between melanoma and vitiligo. Cancer 1997;79:
 Memari N, Grass L, Nakamura T, Karakucuk I, Diamandis EP. Human
tissue kallikrein 9: production of recombinant proteins and specific
antibodies. Biol Chem 2006;387:733–40.
 Hudson ME, Pozdnyakova I, Haines K, Mor G, Snyder M. Identification
of differentially expressed proteins in ovarian cancer using high-density
protein microarrays. Proc Natl Acad Sci U S A 2007;104:17494–9.
 Bradford TJ, Wang X, Chinnaiyan AM. Cancer immunomics: using auto-
antibody signatures in the early detection of prostate cancer. Urol Oncol
429C.G. Gunawardana et al. / Clinical Biochemistry 42 (2009) 426–429