Author's personal copy
MicroRNA signatures of tumor-derived exosomes as
diagnostic biomarkers of ovarian cancer
Douglas D. Taylor⁎, Cicek Gercel-Taylor
Department of Obstetrics, Gynecology, and WomenTs Health, University of Louisville School of Medicine, Louisville, KY, USA
Received 21 March 2008
Objectives. Most ovarian cancer patients are diagnosed at an advanced stage (67%) and prospects for significant improvement in survival
reside in early diagnosis. While expression patterns of a recently identified biomarker family, microRNA, appear to be characteristic of tumor type
and developmental origin, microRNA profiling has been limited to tissue specimens. Tumors actively release exosomes into the peripheral
circulation and we now demonstrate the association of microRNAs with circulating tumor-derived exosomes.
Methods. Circulating tumor exosomes were isolated using a modified MACS procedure with anti-EpCAM. Initially, microRNA profiles of
ovarian tumors were compared to those of tumor exosomes isolated from the same patients. Levels of 8 microRNAs (miR-21, miR-141, miR-
200a, miR-200c, miR-200b, miR-203, miR-205 and miR-214) previously demonstrated as diagnostic, were compared in exosomes isolated from
sera specimens of women with benign disease and various stages of ovarian cancer.
Results. MicroRNA from ovarian tumor cells and exosomes from the same patients were positive for 218 of 467 mature microRNAs analyzed.
The levels of the 8 specific microRNAs were similar between cellular and exosomal microRNAs (exhibiting correlations from 0.71 to 0.90). While
EpCAM-positive exosomes were detectable in both patients with benign ovarian disease and ovarian cancer, exosomal microRNA from ovarian
cancer patients exhibited similar profiles, which were significantly distinct from profiles observed in benign disease. Exosomal microRNA could
not be detected in normal controls.
Conclusions. These results suggest that microRNA profiling of circulating tumor exosomes could potentially be used as surrogate diagnostic
markers for biopsy profiling, extending its utility to screening asymptomatic populations.
© 2008 Elsevier Inc. All rights reserved.
Keywords: MicroRNA; Ovarian cancer; Diagnosis; Screening; Exosomes
Despite progress made in the understanding and treatment of
ovarian cancer, it remains the sixth most common cancer in
women worldwide, causing approximately 125,000 deaths
annually . Most women with ovarian cancer are diagnosed
at an advanced stage, with 75% diagnosed with extra-ovarian
disease . In comparison with other cancers associated with
women, 73% of endometrial cancers, 55% of breast cancers and
50% of cervical cancers are diagnosed with Stage I disease .
While the 5-year survival of patients with Stage I ovarian cancer
exceeds 90%, only 21% of advanced-stage ovarian cancer
patients survive 5 years after initial diagnosis . Since long-
term survival has not changed significantly in the last two
decades, the best prospects for further improvement in ovarian
cancer survival reside in early diagnosis .
Over the last 5 years, expression profiling technologies have
biomarker group is a class of small noncoding RNAs, termed
microRNAs [4–6]. MicroRNAs, small (22–25 nucleotides in
length) noncoding RNAs, suppress the translation of target
mRNAs by binding to their 3′ untranslated region [7,8]. Post-
transcriptional silencing of target genes by microRNA can occur
Available online at www.sciencedirect.com
Gynecologic Oncology 110 (2008) 13–21
⁎Corresponding author. Division of Gynecologic Oncology, University of
Louisville School of Medicine, 511 S. Floyd Street, MDR 420, Louisville, KY
E-mail address: firstname.lastname@example.org (D.D. Taylor).
0090-8258/$ - see front matter © 2008 Elsevier Inc. All rights reserved.
Author's personal copy
cell death . All tumors analyzed by microRNA profiling have
exhibited significantly distinct microRNA signatures, compared
with normal cells from the same tissue [4,6,10]. Lu et al. 
performed an analysis of leukemias and solid cancers and deter-
mined that microRNA-expression profiles could classify human
cancers by developmental lineage and differentiation state. The
expressions of individual microRNAs and specific microRNA
many human cancers.
Using tissue specimens, Iorio et al.  demonstrated that, in
expressed in ovarian cancer, withmiR-141,miR-200a,miR-200b,
and miR-200c being the most significantly overexpressed. They
further demonstrated the hypomethylation in ovarian tumors
resulted in the up-modulation of miR-21, miR-203, and miR-205,
compared with normal ovary. Two of these up-modulated
microRNAs, miR-200a and miR-200c, were enhanced in all the
three histologic types examined (serous, endometrioid, and clear
cell), whereas miR-200b and miR-141 up-modulation was shared
by endometrioid and serous histologic types. In general, the
microRNA signatures obtained comparing different histologic
types of ovarian cancers (serous, endometrioid, clear cell, and
mixed) with the normal tissue were overlapping in most cases.
differentially expressed microRNAs in relation to tumor stage or
grade, which could have resulted from their set of samples being
primarily derived from advanced-stage tumors. However, they
also postulated that microRNAs might be critical for the
development of ovarian cancer, but not for its progression .
Among the microRNAs most significantly up-modulated,
miR-200a and miR-141 belong to the same family, miR-200b is
localized on chromosome 1p36.33 in the same region as miR-
200a and miR-200c is localized on chromosome 12p13.31 in the
same region of miR-141 . This association would agree with
the findings of Zhang et al.  that proposed that the up-
modulation of specific microRNAs could be the amplification of
the microRNA genes. Using high-resolution array-based com-
parative genomic hybridization, an aberrantly high proportion of
loci containing microRNA genes exhibited DNA copy number
alterations. In ovarian cancer, 37.1% of the genomic loci con-
alterations . In breast cancer and melanoma, an even greater
proportion of these loci exhibit altered DNA copy numbers
(72.8% and 85.9%, respectively) . As a result, microRNA-
of the developmental origins of tumors than mRNA expression
patterns and may be associated with diagnosis, staging, progres-
sion, prognosis, and response to treatment. However, as cancer
diagnostictools,the analysesofmicroRNAsignaturesare limited
to tissue biopsies.
In 1979, we initially demonstrated the presence of tumor-
derived exosomes, small (50–100 nm) membrane vesicles of
endocytic origin, in the peripheral circulation of women with
been demonstrated to be capable of releasing exosomes, in-
cluding reticulocytes, dendritic cells, B cells, Tcells, mast cells,
epithelial cells, and embryonic cells [15,16]; however, their
accumulation in the peripheral circulation appears to be unique
to cancer and pregnancy [17,18]. While the primary source of
circulating exosomes in cancer patients is the tumor, other
normal cells within the peripheral circulation can contribute to
the level exosome population. Our recent work has focused on
the separation of tumor-derived exosomes from those derived
from normal lymphoid cells [19,20]. Utilizing adherence to
specific magnetic beads, exosomes of tumor origin can be
Since exosome functionality appears to be determined by its
specific protein content, proteomic analysis has been performed
on in vivo and in vitro derived exosomes. Analyses of exosomes
have demonstrated that all exosomes share certain common
characteristics, including structure (lipid bilayer), size, density
and general protein composition. Some proteins are commonly
associated with all exosomes, including cytoplasmic proteins
(such as tubulin, actin, actin-binding proteins, annexins and Rab
proteins), signal transduction proteins, (protein kinases, hetero-
trimeric G-proteins), MHC class I molecules, and heat-shock
proteins (such as Hsp70 and Hsp90) [21–24]. Tetraspanins,
including CD9, CD63, CD81 and CD82, are the protein family
as exosome markers [23,25]. While tumor-derived exosomes
exosome release can be demonstrated in many proliferating cell
types, their release is exacerbated in tumor cells, as evidenced by
their elevated presence in plasma, ascites and pleural effusions of
cancer patients [26,27]. This elevated presence in serum and
ascites fluids of cancer patients and the overexpression of certain
biomarkers has lead investigators to propose a role for exosomes
in diagnosis and biomarker analysis .
Exosomes have been postulated to play an important role in
cell–cell communication and appear to affect target cells either
by stimulating them directly by surface expressed ligands or by
transferring molecules between cells. Ratajczak et al.  dem-
onstrated the presence of exosomal RNA and provided evidence
for the horizontal transfer of genetic information between cells.
The biological effects of these exosomes were inhibited after
pretreatment with RNAse, indicating the involvement of RNA
components. RNA molecules, following translocation from the
nucleus to the cytoplasm, can bind to and be transported by
membranous organelles or vesicles to specific intracellular sites,
which may provide an explanation for the association of RNA
populations with exosomes. Valadi et al.  demonstrated that
released exosomes contain a subset of both cellular mRNA and
microRNA, which could be transferred to target cells. Our
preliminary results suggest that microRNA contained in tumor
exosomes is functional and can suppress the mRNA for signal
transduction components within T cells. Since released exo-
somes contain RNA populations, including microRNA, it is
possible that this exosomal microRNA reflects the microRNA
to determine whether the microRNAs contained within ovarian
cancer-derived exosomes mirrored that of the tumor and thus
could be used diagnostically.
14D.D. Taylor, C. Gercel-Taylor / Gynecologic Oncology 110 (2008) 13–21
Author's personal copy
Materials and methods
Patient samples and cell lines
This study utilized sera derived from women diagnosed with serous
papillary adenocarcinoma of the ovary (n=50; n=10 for Stage I, n=10 for
Stage II, n=20 for Stage III and n=10 for Stage IV), age-matched women with
benign ovarian adenoma (n=10), and age-matched women with no evidence of
ovarian disease (n=10). Controls, patients with benign ovarian disease and
Stages III and IVovarian cancer were selected based on age-matching to patients
with early stage ovarian cancer. This study also investigated primary tumor cell
cultures, established from 6 women with Stage IIIc cyst adenocarcinoma of the
ovary, and their corresponding pre-surgery sera samples. All of these materials
were obtained under an informed consent approved by the University Human
Studies Committee of the University of Louisville. The primary ovarian tumor
cell cultures were established in our laboratory and designated UL-1, UL-2,
UL-3, UL-6, UL-B, and UL-O. UL-2 and UL-3 were derived from hereditary
ovarian cancers, while UL-1, UL-6, UL-B, and UL-O were derived from spon-
taneous cancers. These ovarian tumor cells were grown in RPMI 1640 medium
supplemented with 10% exosome-free (by ultrafiltration) fetal bovine serum,
100 mg/ml streptomycin and 100 IU/ml penicillin in a humidified 5% CO2
utilized were N95% viable.
Isolation of circulating exosomes
Tumor-derived exosomes were specifically isolated by a modified magnetic
activated cell sorting (MACS) procedure, using anti-epithelial cell adhesion
molecule (EpCAM). Our previous studies have demonstrated that exosomes
from epithelial tumors express EpCAM on their surface and can be used for their
selective isolation. Serum samples (2.5 ml) from normal controls, patients with
benign disease, and patients with early stage ovarian cancer were incubated with
anti-EpCAM coupled to magnetic microbeads (50 μl). These were mixed and
MACS Separator and the column was rinsed with 500 μl Tris-buffered saline
(TBS). The magnetic immune complexes were applied onto the column and
unbound (unlabeled) material passed through and was discarded. The column
was washed four times with 500 μl of TBS. The specifically selected exosomes
were recovered by removing the column from the separator and placing it on a
collection tube. TBS (1 ml) was be added to the column and the magnetically
labeled exosomes were obtained by applying the plunger supplied with the
column. The isolated exosomes/microbeads were diluted in IgG elution buffer
(Pierce Chemical Co, Rockford, IL) and the complex was centrifuged at
10,000 rpm to separate the microbeads from the exosomes (supernatant). The
supernatant was then centrifuged at 100,000 g for 1 h at 4 °C. The pelleted
tumor-derived exosomes were assayed for total protein. The quantity of protein
was determined by the Bradford microassay method (Bio-Rad Laboratories,
Hercules, CA), using bovine serum albumin (BSA) as a standard.
Transmission electron microscopy
For transmission electron microscopy, the pelleted exosomes were fixed in
2.5% (w/v) glutaraldehyde in PBS, dehydrated and embeddedin Epon.Ultrathin
sections (65 nm) were cut and stained with uranyl acetate and Reynold's lead
citrate. The sections were examined in a Jeol 1210 transmission electron
Isolation and profiling of microRNA
microRNA isolation kit according to manufacturer's instructions (Ambion,
Austin, TX). The RNA quality, yield, and size of microRNA fractions were
analyzed using Agilent 2100 Bioanalyzer (Agilent Technologies, Foster City,
CA). The isolated microRNAs were 3′-end labeled with Cy3 using the mirVana
(Amersham Bioscience, Pittsburgh, PA). MicroRNA profiling was performed in
microRNA arrays covering the 467 microRNAs present in the Sanger Insitute
mirBASE v9.0, consisting of 35–44-mer oligonucleotides, manufactured by
Invitrogen and spotted in duplicate. After hybridization, the microRNA arrays
were scanned using a GenePix 4000A array scanner (Axon Instruments, Union
City, CA) and the raw data normalized and analyzed using GeneSpring 7.0
Software (Silicon Genetics, Redwood City, CA). Normalization was performed
by expressing each microRNA replicate relative to control microRNA (Ambion)
percentile of negative controls (TPT95) were calculated based on hybridization
signal from negative control probes including: 38 mismatch and shuffled con-
trol probes and 87 non-conserved Caenorhabditis elegans probes. To define
sensitivity, NCode synthetic microRNAwas spiked at 1/500,000 mass ratio into
labeling reactions and the signal intensity was detected. For specificity, perfect
match probes for miR-93, miR-27a, and miR-152 and 2 mismatches for each
were used. The 2 base pair mismatch probes demonstrated a signal below or at
TPT95 on all arrays.
To assess the stability of the exosomal profiling with storage and manipu-
lations, sera from patients with ovarian cancer patients were obtained and
aliquoted into four 4 ml samples. Tumor exosomes were isolated from the first
aliquot by the MACS procedure immediately and total RNA was isolated and
stored at −70 °C until isolation of all samples. The remaining sera samples were
stored at 4 °C for subsequent exosome isolation. Tumor exosomes were isolated
from the second aliquot after 24 h, from the third aliquot after 48 h and from the
fourth sample after 96 h at 4 °C. RNA was isolated from each exosome
preparation and stored. In a similar study, 3 additional serum aliquots were
stored at −70 °C for 7 to 28 days, prior to exosome and RNA isolations to mimic
the use of banked specimens.
General statistical considerations
Data were analyzed using the statistical software package, SAS9.1 (SAS
Institute, Cary, NC). The levels of circulating exosomes for each group of
patients were defined as means±standard deviations from at least two separate
experiments performed in triplicate. Comparisons between these groups were
performed by one-way ANOVA, followed by the Tukey's multiple comparisons
post-test comparing each population. Relative quantification of microRNA
expression was calculated with the 2−ΔΔCt method (Applied Biosystems User
Bulletin No. 2) and data were analyzed as log10 of relative quantity (RQ) of the
target microRNA, normalized with respect to control microRNA added to each
sample, allowing comparisons between arrays. The microRNA distributions and
correlations along with confidence intervals were calculated for each subset.
Statistical significance was set as p≤0.05.
Presence of circulating EpCAM-positive exosomes in women
with benign and malignant ovarian disease
EpCAM-positive exosomes were specifically isolated using
anti-EpCAM magnetic beads and these circulating exosomes
were assayedfor total protein and plotted versus stage of disease
(Fig. 1A). The levels of EpCAM-positive exosomes in age-
matched normal volunteers (control) were 0.039±0.030 mg/ml
of exosomal protein, which represented the background of the
assay. Patients diagnosed with benign ovarian disease possessed
elevated over controls. Patients diagnosed with ovarian cancer all
exhibited significantly elevated levels of EpCAM-positive
exosomes (compared to benign disease or controls). Women
with Stage I ovarian cancer exhibited 0.320±0.056 mg/ml of
15 D.D. Taylor, C. Gercel-Taylor / Gynecologic Oncology 110 (2008) 13–21
Author's personal copy
both controls and benign disease (pb0.01). The levels of circu-
lating exosomes increased as the stage progressed, with Stage II
cancer having 0.640±0.053 mg/ml, Stage III possessing 0.995±
0.084 mg/ml and Stage IV presenting with 1.42±0.228 mg/ml.
Levels of exosomes associated with these three stages were
significantly greater than women with benign disease or controls
(pb0.001). The resulting fractions were further analyzed by
electron microscopy, which demonstrated vesicular structures
characteristic of exosomes (Fig. 1B). The exosomal nature of this
material was further confirmed by the presence of tetraspanins,
class I antigens, and placental-type alkaline phosphatase by
Western immunoblotting (data not shown).
Fig. 2. Presence of small RNA associated with circulating EpCAM-positive exosomes from ovarian cancer patients. Panel A: Representative analysis of the RNA
isolated from tumor exosomes using Agilent 2100 Bioanalyzer. Panel B: Agarose gel (1%) separation of total RNA from circulating exosomes and corresponding
tumors. This total RNA was used as the starting material for microRNA profiling.
Fig. 1. Panel A: The levels of circulating tumor-derived exosomes compared to stage of ovarian cancer. Exosomes were isolated from sera obtained from age-matched
female controls (n=10), age-matched women with benign ovarian disease (n=10), and women diagnosed with ovarian cancer (n=10 for each stage). Levels of
exosomes are presented as protein concentrations. Panel B: Electron micrograph of circulating exosomes isolated by magnetic beads. Ultrathin sections (65 nm) were
cut and stained with uranyl acetate and Reynold's lead citrate. The sections were examined in a Jeol 1210 transmission electron microscope.
16 D.D. Taylor, C. Gercel-Taylor / Gynecologic Oncology 110 (2008) 13–21
Author's personal copy
Association of small RNA with tumor-derived exosomes
To identify whether these isolated exosomes contained small
RNAs, they were examined using a Bio-Analyzer 2100 (Fig. 2).
This analysis identified the presence of a significant population
of small RNA in the absence of 18S and 28S RNA, generally
observed with cell-derived RNA. This material was subse-
quently used for microRNA profiling.
Profiling of exosome-derived versus cell-derived microRNA
The presence and levels of specific microRNAs from both
cell-derived and exosome-derived microRNA were determined
using microarray analysis probing for 467 microRNAs. The
microRNA profiles of our ovarian tumors confirmed the altera-
tions, previously reported . Further, we demonstrated that of
the 467 microRNAs, 218 were above the normalized threshold,
calculated based on the 95th percentile of the negative control
positive microRNAs, the levels of 175 were not significantly
different between the ovarian tumor cells and their correspond-
ing exosomes. By comparison, 12 were present at a higher
proportion in the cells, while 31 were present at elevated levels
Previously, specific microRNAs were demonstrated to be
overexpressed in human ovarian cancer (miR-21, miR-141,
miR-200a, miR-200c, miR-200b, miR-203, miR-205, and miR-
214). To correlate these findings with exosomal-derived mate-
rial, RNA fractions were isolated from the original tumor cells
and circulating tumour exosomes of the same patients (Fig. 3).
Using microarray analysis, comparisons between tumor-derived
microRNA profiles and peripheral blood-derived exosomal
microRNAs indicated that they were not significantly different.
Further, the levels of tumor-derived microRNA profiles ex-
hibited a strong correlation with the levels of peripheral blood-
derived exosomal microRNAs (for miR-21, r=0.77; miR-141,
r=0.88; miR-200a, r=0.76; miR-200b, r=0.85; miR-200c,
r=0.83; miR-203, r=0.85; miR-205, r=0.91; and miR-214,
Exosomal microRNA correlation with presence and stage of
Our previous comparisons between tumor and circulating
pare the associations of specific microRNAs with the presence of
disease across various stages, the mean intensities of exosomal
microRNAs were determined. The presence of the 8 diagnostic
microRNAs among patients with Stages I, II and III were not
significantly different for most of these miRNAs (Fig. 4). miR-
200c and miR-214 were lower in patients with Stage I, compared
to Stages II and III. However, in all cases, these miRNAs were
significantly elevated over the levels detected in exosomes
derived from benign disease. The small RNA fraction could not
be demonstrated in normal controls and attempts to assess the
presence of miRNAs were negative.
Stability of exosomal microRNA profiles
Since the measurement of circulating exosomal microRNA
raised. When the microRNA profiles were performed on serum
intensities compared (Fig. 5A), no significant differences were
observed in the 3 diagnostic microRNAs analyzed. When the
Association of microRNA with peripheral blood-derived tumor exosomes compared with microRNA isolated from their corresponding tumors
Elevated in cells Equal between cells and exosomes Elevated in exosomes
miR-218, miR-196a, miR-195,
miR-15a, miR-519d, miR-382,
miR-503, miR-34b, miR-520d,
miR-29c, miR-135a, miR-155
miR-296, miR-20a, miR-28, miR-302a, miR-99a, miR-99b, miR-10a, let-7a, let-7b, let-7c,
let-7d, let-7f, let-7g, let-7i, miR-138, miR-23a, miR-183, miR-25, miR-107, miR-181a,
miR-125a, miR-222, miR-198, miR-16, miR-200a, miR-18a, miR-101, miR-136, miR-31,
miR-106b, miR-92, miR-342, miR-128a, miR-182, miR-663, miR-502, miR-500, miR-652,
miR-424, miR-130a, miR-429, miR-365, miR-29a, miR-550, miR-422a, miR-585, miR-92b,
miR-629, miR-671, miR-210, miR-26a, miR-454-5p, miR-769-3p, miR-765, miR-301,
miR-191, miR-93, miR-200b, miR-100, miR-324-5p, miR-220, miR-151, miR-186,
miR-128b, miR-130b, miR-125b, miR-122a, miR-30d, miR-203, miR-15b, miR-192,
miR-133a, miR-126, miR-98, miR-190, miR-137, miR-105, miR-96, miR-95, miR-519b,
miR-29b, miR-453, miR-23b, miR-517c, miR-625, miR-200c, miR-193a, miR-22, miR-224,
miR-369-3p, miR-106a, miR-181c, miR-17-5p, miR-19b, miR-24, miR-17-3p, miR-221,
miR-335, miR-126, miR-181a, miR-331, miR-188, miR-9, miR-34a, miR-30c, miR-19a,
miR-371, miR-10b, miR-21, miR-148a, miR-339, miR-187, miR-346, miR-146a, miR-185,
miR-328, miR-196b, miR-129, miR-522, miR-30a-5p, miR-27a, miR-30a-3p, miR-494,
miR-20b, miR-521, miR-181b, miR-423, miR-487b, miR-425-3p, miR-594, miR-532,
miR-512-3p, miR-526a, miR-578, miR-638, miR-422b, miR-484, miR-486, miR-645,
miR-146b, miR-571, miR-647, miR-637, miR-30b, miR-452, miR-361, miR-432, miR-375,
miR-766, miR-768-3p, miR-769-5p, miR-513, miR-362, miR-565, miR-30e-3p, miR-320,
miR-590, miR-152, miR-181d, miR-660, miR-584, miR-141, miR-18b, miR-582, miR-505,
miR-628, miR-425-5p, miR-421, miR-27b, miR-768-5p, miR-454-3p, miR-148b, miR-194,
miR-214, miR-140, miR-147,
miR-135b, miR-205, miR-150,
miR-149, miR-370, miR-206,
miR-197, miR-634, miR-485-5p,
miR-612, miR-608, miR-202,
miR-373, miR-324-3p, miR-103,
miR-593, miR-574, miR-483,
miR-527, miR-603, miR-649,
miR-18a, miR-595, miR-193b,
miR-642, miR-557, miR-801,
17 D.D. Taylor, C. Gercel-Taylor / Gynecologic Oncology 110 (2008) 13–21
Author's personal copy
intensities of these microRNAs on the microarrays were not
significantly different (Fig. 5B). These results indicate that the
levels of these exosomal microRNAs were stable and do not
significantly change with storage.
MicroRNA-expression profiling can be used as diagnostic
tools for cancers that currently lack reliable molecular markers,
such as ovarian cancer. While previous studies have indicated
that microRNA signatures could serve as diagnostic and prog-
nostic markers for ovarian cancer, these data were based on their
expression in tissue specimens. Here, we report the association
of microRNA with circulating tumor-derived exosomes. In
previous studies, microRNAs have been demonstrated to be
aberrantly expressed in human ovarian cancers and the overall
microRNA expression could differentiate normal versus cancer
tissues . The study of Lu et al.  demonstrated the use of
microRNA signatures as an important advance in cancer diag-
nosis. Their work indicated that microRNA-based identification
of cancers was superior in correctly diagnosing cancer of
unknown primaries than mRNA classification. However, it is
currently not possible to use microRNA profiling in the absence
of a mass to be biopsied.
Our original electron microscopic characterization of exo-
somes indicated that they were hollow (i.e. absence of viral-like
structures) . As a result, our group, together with others,
focused on external protein components of exosomes and the
biologic consequences of exosome exposure. In the present
study, we demonstrate for the first time the presence of small
RNA species associated with circulating tumor exosomes
(Fig. 2). This small RNA lacks the 18S and 28S associated
that at least part of the small RNA identified is microRNA.
Fig. 3. Intensities for specific microRNAs derived from the advanced-staged ovarian tumors (□) and from EpCAM-positive exosomes (■) isolated from the sera of
these patients. miR-21, miR-141, miR-200a, miR-200b, miR-200c, miR-203, miR-205, and miR-214 have previously been demonstrated to be upregulated markers
for ovarian cancer. Each bar presents the average intensities of duplicate samples with the results of four representative patients presented.
18 D.D. Taylor, C. Gercel-Taylor / Gynecologic Oncology 110 (2008) 13–21
Author's personal copy
The microRNA-expression profiles of our ovarian tumor
cells confirmed the microRNA aberrations reported in previous
studies. Analyses of both circulating tumor exosomes and the
tumor cells from the same patients demonstrated that both were
positive for 46% of the tested microRNAs (218/467). When the
intensities of the microRNA were normalized, most of these
microRNAs were expressed at similar levels between the cells
and exosomes or were elevated within the exosomes (175 were
not significantly different and 31 were elevated within
exosomes). Thus, the aberrantly expressed microRNAs, used
to establish cancer-specific signatures, appear in both cellular
and exosomal compartments of ovarian cancer patients.
Our comparison of specific microRNAs, previously demon-
strated to be diagnostic, indicated a high degree of correlation
between the microRNA from the tumor and its corresponding
exosomes (ranging from 0.71 to 0.90). This high correlation even
holds for microRNAs that appeared to be present at higher pro-
some microRNAs, is an active (selective) process. Such a process
could be mediated by components, such as nucleolin or nucleo-
phosmin, which are aberrantly expressed on tumor exosomes.
Since these results suggest thatexosomalmicroRNAprofiling
could be used as a surrogate for tissue microRNA and the goal in
screening would be the identification of early stage disease, the
ability to detect circulating exosomal microRNAs in early stage
disease was examined. The exosomal microRNA expressions of
stage ovarian cancers were not significantly different for most of
these microRNAs (Fig. 4). miR-200c and miR-214 werelower in
cases, these microRNAs were significantly elevated over the
levels detected in exosomes derived from benign disease. The
Thus, the absence of exosomes and/or exosomal small RNA is
associated with normal, non-cancer-bearing individuals and
the stages of ovarian cancer is likely the result of standardization
of starting exosomal small RNA quantities and the normalization
of the resulting array data. Despite this standardization and
normalization, the profiles obtained with exosomal microRNA
from patients with benign disease remained distinct. There are
several issues that are not analyzed in this pilot study. Since
previous studies examining the microRNA signatures obtained
comparing different histologic types of ovarian cancers (serous,
endometrioid, clear cell, and mixed) failed to demonstrate
differentially expressed microRNAs in terms of tumor stage or
adenocarcinomas and did not address differences in grade.
Further, beyond the 8 microRNAs identified by Iorio et al. ,
some microRNAs appear to be differentially expressed between
scale study including additional confounding factors will need to
be performed to define their significance. Patients with benign
ovarian disease weresymptomatic and referredto the division for
resection of the ovarian mass. Thus, it is unclear what the profile
of women with asymptomatic masses would be. The women
constituting the control group, in addition to not having cancer,
the populations mostly to be screened would include such
individuals. Thus, it is unclear whether these two “control”
populations would be identical. Despite these limitations, these
results serve as a proof of concept that the analyses of specific
microRNAs associated with circulating exosomes can be applied
to all stages of ovarian cancer and that benign and malignant
diseases appear distinguishable based on the levels of these 8
Fig. 4. Intensities for specific microRNAs derived from EpCAM-positive exosomes isolated from the peripheral blood (2.5 ml) of the patients with benign ovarian
disease and patients with ovarian cancer. Patients with ovarian cancer were separated between Stages I, II, and III. The bars represent the mean±standard deviation of
the normalized intensities of each group of patients (n=10 for each group).
19 D.D. Taylor, C. Gercel-Taylor / Gynecologic Oncology 110 (2008) 13–21
Author's personal copy
The microRNA signatures of exosomes parallel that of the
microRNA-expression profiles of the originating tumor cells,
indicating that microRNA profiling can be performed in the
absence of tissue and accurately reflect the tumor's profile. We
also have observed that tumor-derived exosomes from lung
cancer patients contain microRNA that is similar to the
corresponding tumor microRNA signatures. Circulating
tumor-derived exosomes can be easily isolated using tumor
markers, such as EpCAM, followed by analysis of exosome-
associated microRNA. Since this approach is non-invasive, in
that it does not require a mass to be biopsied, exosomal
microRNA profiling has the potential to be used as a screening
tool for the detection of cancer. As specific microRNAs
associated with tumor tissues are identified that predict
prognosis, including therapeutic resistance (such as let-7i,
miR-16, miR-21 and miR-214) [31,32], their presence in
tumor exosomes can also be assessed to further define the
utility of exosomal microRNA profiling as a prognostic
indicator. While validation studies will be necessary prior to
bypassing the use with tumor mass biopsies, the use of
exosomal microRNA profiling could extend this approach to
screening of asymptomatic individuals, as well as for
monitoring disease recurrence.
Conflicts of interest statement
The authors have no conflicts of interest to declare.
We acknowledge the assistanceof Dr. Shesh Rai ofthe Brown
Cancer Center of the University of Louisville in the design and
statistical analyses of this study and Ms. Leta Weedman, Medical
tion of this manuscript.
Grant/funding support: This work was supported by grants
from the National Cancer Institute (CA98166), the Kentucky
Science and Technology Corporation and the Kentucky Science
and Engineering Foundation (KSEF-1025-RDE008).
Fig. 5. Comparison of specific exosomal microRNAs derived from the serum of an ovarian cancer patient, immediately after blood draw or 24, 48, and 96 h later with
sera stored at 4 °C (panel A) or after 7 to 28 days, stored at −70°C (panel B). Tumor exosomes were isolated by MACS using anti-EpCAM.
20D.D. Taylor, C. Gercel-Taylor / Gynecologic Oncology 110 (2008) 13–21
Author's personal copy Download full-text
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.ygyno.2008.04.033.
 Sankaranarayanan R, Ferlay J. Worldwide burden of gynaecological
cancer: the size of the problem. Best Pract Res Clin Obstet & Gynaecol
 Berek JS, Schultes BC, Nicodemus CF. Biologic and immunologic therapies
for ovarian cancer. J Clin Oncol 2003;21(s10):168–74.
 Menon U, Jacobs IJ. Recent developments in ovarian cancer screening.
Curr Opin Obstet Gynecol 2000;12:39–42.
 Iorio MV, Visone R, Di Leva G, Donati V, Petrocca F, Casalini P, et al.
MicroRNA signatures in human ovarian cancer. Cancer Res 2007;67:
S, et al. Gene expression profiling of advanced ovarian cancer: charac-
terization of a molecular signature involving fibroblast growth factor 2.
 Calin GA, Croce CM. MicroRNA-cancer connection: the beginning of a
new tale. Cancer Res 2006;66:7390–4.
 Esquela-Kerscher A, Slack FJ. Oncomirs — microRNAs with a role in
cancer. Nature Rev Cancer 2006;6:259–69.
 Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function.
 Miska EA. How microRNAs control cell division, differentiation, and
death. Curr Opi Genet Dev 2005;5:563–8.
 Calin GA, Croce CM. MicroRNA signatures in human cancers. Nature
Rev Cancer 2006;6:857–66.
 Lu J, Getz G, Miska EA, Alvarez-Saavedra E, Lamb J, Peck D, et al.
MicroRNA expression profiles classify human cancers. Nature 2005;435:
 Zhang L, Huang J, Yang N, et al. microRNAs exhibit high frequency
genomic alterations in human cancer. Proc Natl Acad Sci USA 2006;103:
 Taylor DD, Doellgast GJ. Quantitation of eroxidise-antibody binding to
membrane fragments using column chromatography. Anal Biochem
 Taylor DD, Homesley HD, Doellgast GJ. Binding of specific peroxidise-
labeled antibody to placental-type alkaline phospohatase on tumor-derived
membrane fragments. Cancer Res 1980;40:4964–9.
 Raposo G, Tenza D, Mecheri S, Peronet R, Bonnerot C, Desaymard C.
Accumulation of major histocompatibility complex class II molecules in
mast cell secretory granules and their release upon degranulation. Mol Biol
 Heijnen HFG, Schiel AE, Fijnheer R, Geuze HJ, Sixma JJ. Activation
platelets release two types of membrane vesicles: microvesicles by surface
shedding and exosomes derived from exocytosis of multivesicular bodies
and alpha granules. Blood 1999;94:3791–9.
 Taylor DD, Black PH. Shedding of plasma membrane fragments:Neoplastic
and developmental importance. In: Steinberg M, editor. Developmental
Biology, vol. 3; 1986. p. 33–57.
 Taylor DD, Bohler HC, Gercel-Taylor C. Pregnancy-linked suppression of
TcR signaling pathways by a circulating factor absent in recurrent
spontaneous pregnancy loss. Molecular Immunology 2006;43:1872–80.
 Sabapatha A, Gercel-Taylor C, Taylor DD. Specific isolation of placental-
derived exosomes from the circulation of pregnant women and their
 Taylor DD, Gercel-Taylor C. Tumour-derived exosomes as mediates of
T-cell signaling defects. Brit J Cancer 2005;92:305–11.
 Olver C, Vidal M. Proteomic analysis of secreted exosomes. Subcell
 Mears R, Craven RA, Hanrahan S, et al. Proteomic analysis of melanoma-
derived exosomes by two-dimensional polyacrylamide gel electrophoresis
and mass spectrometry. Proteomics 2004;4:4019–31.
 Bard MP, Hegmans JP, Hemmes A, et al. Proteomic analysis of exosomes
isolated from human malignant pleural effusions. Am J Respir Cell Mol
 Choi DS, Lee JM, Park GW, et al. Proteomic analysis of microvesicles
derived from human colorectal cancer cells. J Proteome Res 2007;6:
 Escola JM, Kleijmeer MJ, Stoorvogel W, Griffith JM, Yoshie O, Geuze HJ.
Selective enrichment of tetraspan proteins on the internal vesicles of multi-
vesicular endosomes and on exosomes secreted by human B-lymphocytes.
J Biol Chem 1998;273:20121–7.
 Andre F, Schartz NE, Movassagh M, et al. Malignant effusions and
immunogenic tumour-derived exosomes. Lancet 2002;360:295–305.
 Valenti R, Huber V, Filipazzi P, Pilla L, Sovena G, Villa A, et al. Human
tumor-released microvesicles promote the differentiation of myeloid cells
with transforming growth factor-beta-mediated suppressive activity on
T lymphocytes. Cancer Res 2006;66:9290–8.
 Koga K, Matsumoto K, Akiyoshi T, Kubo M, et al. Purification, character-
ization and biological significance of tumor-derived exosomes. Anticancer
 Ratajczak J, Miekus K, Kucia M, et al. Embryonic stem cell-derived
microvesicles reprogram hematopoietic progenitors: evidence for hori-
zontal transfer of mRNA and protein delivery. Leukemia 2006;20:847–56.
 Valadi H, Ekstrom K, Bossius A, Sjostrand M, Lee JJ,Lotvall JO. Exosome-
mediated transfer ofmRNA and microRNA isa novel mechanismofgenetic
exchange. Nature Cell Biol 2007;9:652–9.
 Yang H, Kong W, He L, Zhao JJ, O'Donnell JD, Wang J, et al. MicroRNA
expression profiling in human ovarian cancer: miR-214 induces cell
survival and cisplatin resistance by targeting PTEN. Cancer Res 2008;68:
 Paul E, Blower PE, Chung JH, Verducci JS, Lin S, Park JK, et al.
MicroRNAs modulate the chemosensitivity of tumor cells. Mol Cancer
21 D.D. Taylor, C. Gercel-Taylor / Gynecologic Oncology 110 (2008) 13–21