ORIGINAL ARTICLE – TRANSLATIONAL RESEARCH AND BIOMARKERS

Overexpression of Activin A in Oral Squamous Cell Carcinoma:Association with Poor Prognosis and Tumor Progression

Kai-Ping Chang, MD, PhD1, Huang-Kai Kao, MD2, Ying Liang, MD3, Ming-Hui Cheng, MD, MHA2,

Yu-Liang Chang, DDS, PhD4, Shiau-Chin Liu, BS1, Yu-Chi Lin, MS5, Tzu-Yin Ko, BS1, Yun-Shien Lee, PhD5,6,

Chia-Lung Tsai, PhD5, Tzu-Hao Wang, MD, PhD5, Sheng-Po Hao, MD1, and Chi-Neu Tsai, PhD7

1Department of Otolaryngology—Head & Neck Surgery, Chang Gung Memorial Hospital & Chang Gung University

College of Medicine, Tao-Yuan, Taiwan; 2Department of Plastic & Reconstructive Surgery, Chang Gung Memorial

Hospital, Tao-Yuan, Taiwan; 3Molecular Medicine Research Center, Chang Gung University, Tao-Yuan, Taiwan;4Department of Oral & Maxillofacial Surgery, Chang Gung Memorial Hospital, Tao-Yuan, Taiwan; 5Genomic Medicine

Research Core Laboratory, Chang Gung Memorial Hospital, Tao-Yuan, Taiwan; 6Department of Biotechnology, Ming

Chuan University, Tao-Yuan, Taiwan; 7Graduate Institute of Clinical Medical Sciences, Chang Gung University,

Tao-Yuan, Taiwan

ABSTRACT

Background. Both activin A, a member of transforming

growth factor b superfamily, and its inhibitor follistatin

have been shown to be overexpressed in various cancers.

We examined the potential role of activin A and follistatin

in tissue and blood samples from patients with oral squa-

mous cell carcinoma.

Methods. For activin A and follistatin, the expression of

tissue samples from 92 patients was examined by immu-

nohistochemical study, and the serum levels of blood

samples from 111 patients and 91 healthy controls were

measured by enzyme-linked immunosorbent assay.

Results. We found that overexpression of immunohisto-

chemically detected activin A was correlated with positive

N stage, poor histological differentiation, and perineural

invasion (P = 0.029, 0.002, and 0.014, respectively). In

survival analyses, patients with oral squamous cell carci-

noma, whose tumors overexpressed activin A, had a worse

prognosis for overall survival and disease-free survival

(P = 0.009 and 0.007). However, expression of follistatin

in tumor was not correlated with overall survival or dis-

ease-free survival. Serum activin A and follistatin levels in

111 untreated patients were neither significantly different

from those of 91 control samples nor associated with any

clinicopathological manifestations. In vitro suppression

of activin A expression in OC3 cells using specific inter-

fering RNA-attenuated cell proliferation, migration, and

invasiveness.

Conclusions. These findings suggest that activin A over-

expression in oral squamous cell carcinomas is associated

with patients’ survival and may contribute to tumor pro-

gression and metastasis.

Oral squamous cell carcinoma (OSCC) is the most

common cancer of the head and neck and accounts for

approximately 3% of all newly diagnosed cancer cases.1,2

Despite recent advances in surgical, radiotherapy, and

chemotherapy treatment protocols, the long-term survival

of patients with OSCC has remained at approximately

50%–60% for the past three decades.3,4 These unsatisfac-

tory treatment results and lack of improvement despite

advances in treatment modalities may be explained by the

fact that OSCCs frequently present with extensive local

invasion and have a high probability of cervical lymph

node metastasis. OSCC is thought to develop from pre-

cancerous dysplastic lesions through multistep processes

of carcinogenesis involving activation of oncogenes

and loss of the tumor suppressor genes, and exposure of

Electronic supplementary material The online version of thisarticle (doi:10.1245/s10434-010-0926-2) contains supplementarymaterial, which is available to authorized users.

� Society of Surgical Oncology 2010

First Received: 20 May 2009;

Published Online: 23 March 2010

S.-P. Hao, MD

e-mail: shengpo@cgmh.org.tw

C.-N. Tsai, PhD

e-mail: pink7@mail.cgu.edu.tw

Ann Surg Oncol (2010) 17:1945–1956

DOI 10.1245/s10434-010-0926-2

environmental carcinogens such as tobacco, alcohol, and

betel quid chewing.5–8 Detecting carcinogenetic abnor-

malities in OSCCs might provide an important prognostic

indicator of patient prognosis and contribute to the future

development of tailored treatments based on the presence

of specific markers.

Activin A is a member of the transforming growth

factor b (TGF-b) superfamily of secreted signaling mol-

ecules, which mediate cell growth and differentiation.9

This protein was originally purified from ovarian fluid and

identified as a stimulatory factor for pituitary FSH

secretion, but was subsequently found to have diverse

regulatory functions in a wide range of tissues and

organs.10,11 Activin A is a homodimer of 2 disulfide-

linked bA subunits encoded by INHBA gene and has been

shown to be a crucial regulator of organ development and

wound healing.12 In adult mice and humans, abnormal

expression of activin A is associated with various dis-

eases, including different types of cancer.13 Although the

role of activin A in cancer progression remains a topic of

investigation, it has been reported to be overexpressed in

various cancers, including colon, esophageal, ovarian,

prostate, pancreatic, and lung cancers.14–19 Additionally,

high serum levels of activin A have been proposed as a

marker of human breast, hepatocellular, endometrial, and

cervical carcinomas.20–22 Activin A is also reported to

promote cancer cell growth, migration, and invasion in

some in vivo and in vitro studies.23–26 These results

suggest that activin A might be involved in the carcino-

genesis, progression, and metastasis of some types of

cancer. However, other studies have shown that overex-

pression of activin A resulted in antiproliferative and

apoptotic effects in different cancer cells.27–29 Thus, the

exact function of activin A in different human malig-

nancies remains to be elucidated.

The protein follistatin (FST), encoded by the FST gene,

is a secreted monomeric protein that is structurally unre-

lated to activins. FST binds to activin A with high affinity

and blocks its interaction with activin receptors.30 As an

inhibitor of activin A, FST also regulates a variety of

biological functions, including cell growth, differentiation,

and apoptosis. A recent study showed that cells transfected

with FST produced fewer metastatic colonies in a multiple-

organ metastasis murine model, predominantly by inhibit-

ing the angiogenesis.31 Furthermore, activin A secreted by

vascular endothelial cells promotes vascular endothelial

growth factor-induced angiogenesis, but these effects are

inhibited by the addition of FST.32 These data suggest that

FST expression in the tumor microenvironment might

influence cancer progression.

Although two previous transcriptomic studies using

cDNA microarray analysis revealed that the gene INHBA

was overexpressed at the mRNA level in the OSCC tissues

compared with matched normal epithelial tissue, to the best

of our knowledge, there are no published reports that fur-

ther address the potential role of activin A and FST in

OSCC.33,34 In the current study, we evaluated the expres-

sion of activin A and FST in human OSCC. Herein, we

provide the first evidence that activin A and FST are both

overexpressed in OSCC tumor cells and that activin A

overexpression is tightly associated with OSCC disease

status and treatment outcome. Finally, specific interfering

RNA (siRNA)-mediated knockdown of INHBA transcripts

in an OSCC tumor cell line suppressed the in vitro pro-

liferation, migration, and invasion ability of OSCC cells.

Based on these results, we discuss the biological signifi-

cance of activin A and FST overexpression in OSCC and

the potential role of these proteins in future targeted

therapies.

MATERIALS AND METHODS

Patient Characteristics and Clinical Specimens

Tumor specimens for immunohistochemical study were

obtained from 92 consecutive patients with OSCC diag-

nosed at the Chang Gung Memorial Hospital (CGMH, Tao-

Yuan, Taiwan) from 2003 to 2006. Serum samples were

collected from 91 healthy controls (88 men and 3 women)

and 111 patients with OSCC (103 men and 8 women) at the

CGMH from March 2005 to November 2006. Patients with

inoperable disease, synchronous cancers, recurrent cancers,

distant metastasis, or a preoperative history of another

malignancy were excluded from the analyses. All patients

provided informed consent prior to study participation.

Patients in the study underwent standard preoperative

work-ups according to institutional guidelines, including a

detailed medical history, complete physical examination,

computed tomography or magnetic resonance imaging

scans of the head and neck, chest radiographs, bone scan,

and abdominal ultrasound. Primary tumors were excised

with adequate margins under frozen section control intra-

operatively. Classic radical or modified neck dissections

(level I–V, about 50–60 lymph nodes) were performed in

patients with clinically positive lymph node disease. Su-

praomohyoid neck dissections (level I–III, about 30–40

lymph nodes) were performed in clinically node-negative

patients.35 Surgical defects were immediately recon-

structed by a free-flap or local-flap by plastic surgeons.

After surgical treatment, the pathological and nodal stage

of all tumors was established according to the AJCC

Cancer Staging Manual (2002). After discharge, all

patients had regular follow-up visits every 2 months for the

first year, every 3 months for the second year, and every

6 months thereafter.

1946 K.-P. Chang et al.

RNA Extraction and Quantitative Real-time RT-PCR

Detection of INHBA and FST

A total of 26 paired OSCC tumor and pericancerous

normal tissues were homogenized in liquid nitrogen with a

pestle and mortar, incubated with RNAzol B reagent (Tel-

Test Inc, Friendwood, TX), and total RNA was extracted

according to the manufacturer’s protocol. Total RNA was

further purified using the RNeasy Mini-Elute cleanup kit

(Qiagen Inc., Valencia, CA) according to the manufac-

turer’s protocol. First-strand cDNA was synthesized from

5 lg of total RNA, and then added to a PCR reaction

mixture consisting of commercially available primers

(INHBA, Hs00171410_m1, FST, Hs 00246256_m1, and

normalization control GAPDH, Hs99999905_m1; Assay-

on-Demand, Applied Biosystems, Foster City, CA),

RNase-free water, and TaqMan Universal PCR Master

Mix. Quantitative real-time RT-PCR was performed and

analyzed by 7900 HT Sequence Detection System and SDS

versions 2 (Applied Biosystems, Foster City, CA),

respectively. All experiments were repeated in triplicate

and mean fold-change was calculated for each sample.

Immunohistochemical Staining

For immunohistochemistry, formalin-fixed and paraffin-

embedded tissues were sectioned to 4 lm in thickness and

deparaffinized, rehydrated, and prepared for antigen

retrieval. The slides of consecutive sections were then

incubated with an appropriate dilution of antibodies: acti-

vin A (1:20) and follistatin (1:40) (R&D Systems

Minneapolis, MN) at room temperature for 1 hour. After

incubation, these slides were washed with phosphate buf-

fered saline (PBS) 3 times, incubated with horseradish

peroxidase (HRP) polymer antibody (Zymed) at room

temperature for 10 minutes, and developed by addition of

3,30-diaminobenzidine tetrahydrochloride (DAB) reagent

(DAKO) as the chromogen and hematoxylin as a coun-

terstain. Images of the stained slides were obtained using

the ScanScope computed tomography (CT) automated

slide-scanning system (Aperio Technologies, Vista, CA).

Expression of activin A and follistatin was scored using a

combined scoring method that accounts for both the

staining intensity and the percentage of stained cells.36

Strong, moderate, weak, and negative staining intensities

were scored as 3, 2, 1, and 0, respectively. For each of the

intensity scores, the percentage of cells that stained at the

specific level was visually estimated. The resulting com-

bined score was calculated as the sum of the percentage of

stained cells multiplied by the intensity scores. For exam-

ple, a case with 20% weak staining, 30% moderate

staining, and 50% strong staining would be assigned a

score of 230 (20 9 1 ? 30 9 2 ? 50 9 3 = 230) out of a

possible score of 300. All specimens were independently

evaluated by our pathologist (Liang Y.) without prior

knowledge of the clinical origin of the specimen. Immu-

nohistochemical scores 0, 1–100, 101–200, and 201–300

were classified as immunohistochemical levels 0, ?, ??,

and ???, respectively. Levels ?? and ??? were

defined as having protein overexpression.

ELISA Assays of Serum Activin A and FST Levels

Activin A and FST levels in the tested samples were

determined using Quantikine ELISA kits for human activin

A and FST (R&D Systems Minneapolis, MN). Human

recombinant activin A and FST proteins were used as assay

standards. Briefly, 100 lL of serum samples or standard

were added to microtiter plates coated with a murine

monoclonal antibody against human activin A or FST,

incubated for 2 hours at room temperature, and then

washed 3 times with wash buffer. A horseradish peroxi-

dase-conjugated polyclonal antibody was then added to the

individual wells, and the plates were incubated for 2 hours

at room temperature. The plates were then washed, and

hydrogen peroxide and tetramethylbenzidine were added

for color development at room temperature for 30 minutes.

The reaction was stopped by addition of 2 N sulfuric acid.

The color intensity in each well was measured as the

optical density using a microplate reader set to 450 nm. A

standard curve was constructed by plotting the optical

value of the standard and the amounts of Activin A and

FST in the respective samples. Each experiment was per-

formed in duplicate.

Knockdown of Activin A Using RNA Interference

(RNAi)

RNAi specifically targeting human INHBA (No. L-011701-

00-0005, Dharmacon) and scrambled control RNAi

(No. D-001810-10-05, Dharmacon) were purchased from

Thermo Fisher Scientific, Inc. RNAi (final concentration of

RNAi was 400 nM) was mixed with Oligofectamine

reagent (Invitrogen, Carlsbad, CA) and Opti-MEM med-

ium (Invitrogen, Carlsbad, CA) without serum, incubated

for 15 minutes at room temperature, and then added to OC3

cells that were seeded at a density of 2 9 105 cells per well

in six-well plates.37 After incubation for 6 hours at 37�C,

500 lL of 3X concentrated OC3 cell medium (Keratino-

cyte-SFM [Invitrogen, Carlsbad, CA]: Dulbecco’s

Modified Eagle’s Medium (DMEM) supplemented with

10% fetal bovine serum = 3:1) was added to each well.

After transfection for 24 hours, cells were harvested for

analysis of cell proliferation, migration, and invasive

capacity.

Activin a as a Prognostic Marker of Oscc 1947

Cell Proliferation Assay

Cell proliferation was measured using the 3-(4,5-

dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)

reagent. OC3 cells transfected with specific RNAi,

scrambled RNAi, or PBS were plated at a density of

1 9 104 cells/well in 96-well plates and subsequently

incubated in serum-free medium (supplemented with

DMEM) for 48 hours prior to the addition of MTT for 4

hours. Acidic isopropanol was then added and incubated

overnight. The optical density of color development was

measured at 595 nm using a SpectraMax M5 microplate

reader (Molecular Devices, Sunnyvale, CA). Each data

point was in the result of quadruplicate determinations

from 3 separate experiments.

Cell Migration Assay

Cell migration was evaluated using a chemotaxis

chamber (Corning Inc., Lowell, MA) with a polycarbonate

membrane (pore size of 8 lm) placed between the two

chambers. Transfected OC3 cells (3 9 105) in 300 lL of

culture medium were applied to the upper chamber and 600

lL of medium was added to the lower chamber. Chambers

were incubated at 37�C for 16 hours, and then the mem-

brane was fixed in methanol for 10 minutes and stained

with hematoxylin and eosin. Cells on the upper surface of

the filter were carefully removed with a cotton swab, and

then cells that had migrated through the membrane to the

lower surface of the filter were counted in 9 different visual

fields under a light microscope (magnification: 200 9).

Each migration assay was performed in triplicate during 3

independent experiments.

Cell Invasion Assay

The Cell Invasion Assay Kit (Chemicon, Temecula, CA)

was used to measure the invasiveness of the cancer cell

lines. Briefly, polycarbonate membranes (8 lm pore size)

in the upper compartments of the provided Transwell cul-

ture chambers were coated with the provided ECMatrix.

Cells (1 9 105) were suspended in 100 lL serum-free

medium and placed in the upper compartments. The lower

compartments were filled with 500 lL of cell growth

medium containing 10% FBS. Cells were incubated at

37�C for 48 hours, and then the membranes were fixed in

methanol and stained with hematoxylin and eosin. Cells on

the upper surface of each filter were carefully removed

with a cotton swab, and the cells that had migrated through

the membrane to the lower surface of the filter were

counted in 9 different fields under a light microscope

(magnification: 200 9). Each invasion assay was per-

formed twice in 3 independent experiments.

Statistical Analysis

All statistical data are expressed as the mean ± SD. The

Wilcoxon signed ranks test was used for comparison of the

relative signal intensity of quantitative real-time PCR and

immunohistochemical staining scores between the paired

tumor and pericancerous normal mucosa samples. Serum

protein levels, cell proliferation, cell growth, and migration

experiments were compared using unpaired t tests. The

associations between the various characteristics of patients

and activin A and FST overexpression were evaluated

using the chi-square test. For analyzing the correlation of

expression level of INHBA and FST transcripts, the average

Ct of GAPDH was subtracted from the raw Ct value to

obtain d-Ct (dCt). The experimentally normalized dCt

values were converted to 39 – Ct used to estimate the

expression level of mRNA transcripts. All statistical anal-

yses were performed using SPSS and SAS software. All

patients received follow-up evaluations at our outpatient

clinic until November 2008 or death. Survival analysis was

plotted using the Kaplan-Meyer method and differences

were evaluated using the log-rank test. Univariate and

multivariate regression analyses were used to define spe-

cific risk factors for overall survival and disease-free

survival. All P values were 2-sided, and statistical signifi-

cance was accepted when P \ .05.

RESULTS

Overexpression of Activin A and FST in Tumor Cells

of OSCC Tissues

To distinguish the specific cell types in the tumor mass

that expressed activin A and FST, consecutive tissue

sections from 92 patients were subjected to immunohisto-

chemical staining. Both Activin A and FST were highly

expressed in the cytoplasm of tumor cells without nuclear

or membranous staining. Both expression was largely

absent or faintly stained in the tumor-infiltrating lympho-

cytes and interstitial tissues except that activin A was also

expressed in some macrophages in the interstitial tissues

(Fig. 1). Moreover, the paired normal oral epithelium

samples showed either low or no expression of activin A

and FST (Fig. 1). Expression of INHBA and FST were

examined by quantitative real-time RT-PCR in 26 paired

OSCC tumor and adjacent normal tissues. Transcripts

for both INHBA and FST were significantly elevated in

OSCC tumor specimens compared with adjacent normal

tissue (INHBA: 20,100 ± 3160 vs. 2876 ± 585.0, P \.0001; FST: 1490 ± 1677 vs. 660.8 ± 221.1, P = .0003)

(Fig. 2a). Statistical analysis of the 80 paired samples

available from these 92 OSCC patients revealed that

immunohistochemical scores of both activin A and FST

1948 K.-P. Chang et al.

were significantly higher in tumor tissue than in nontumor

tissue (activin A: 141.5 ± 50.2 vs. 9.1 ± 28.0, P \ .0001;

FST: 126.5 ± 49.1 vs. 6.7 ± 22.7, P \ .0001) (Fig. 2b).

These quantitative real-time RT-PCR and immunohisto-

chemistry results demonstrated that both activin A and FST

are overexpressed in OSCC tumor cells, but not in nontu-

mor cells within the tumor specimen or in adjacent normal

tissue. Furthermore, activin A and FST mRNA expression

in OSCC tissues was analyzed by normalizing to GAPDH

expression. Activin A and FST protein expression was

performed using immunohistochemical scores from the

consecutive sections. Statistical significant correlations

between activin A and FST were observed in both mRNA

and immunohistochemical expression (Pearson’s correla-

tion r = 0.507, P = .008 and r = 0.354 and P = .0005,

respectively; see Fig. 2c).

Association of Activin A and FST Expression

with Various Clinicopathological Manifestations

We next evaluated whether increased expression of

activin A or FST was correlated with various clinicopath-

ological characteristics in 92 OSCC patients. Over-

expression of activin A was significantly correlated with

positive N stage, poorer OSCC histological differentiation,

and perineural invasion (P = .029, .002, and .014,

respectively). Nevertheless, there were no correlations

related to patient age, T stage, or overall tumor stage

(Table 1). Of the various clinicopathological characteris-

tics evaluated, overexpression of FST was only

significantly correlated with overall tumor stage (P = .03;

Table 1). There was also no association between overex-

pression of activin A or FST in OSCC tumors and the

habitual behaviors of betel nut chewing, smoking, or

alcohol consumption (Table 1).

Correlation of Activin A or FST Overexpression

and Patient Overall Survival

Using data from the 92 patients in the prospective

cohort, we further evaluated whether overexpression of

activin A or FST was correlated with patient overall sur-

vival (OS). Based on survival analysis on 88 patients

undergoing complete standardized treatment and regular

follow-up, we found that the long-term OS for patient

subgroups stratified by the absence or presence of activin A

overexpression were 91.3% and 55.5%, respectively. This

difference in OS was statistically significant when com-

pared using a log-rank test (P = .009) (Fig. 3a).

Conversely, long-term OS shown by Kaplan-Meier plots of

patients stratified by the absence or presence of FST

overexpression was not significantly different (72.3% and

60.1%, respectively; P = 0.179) (Fig. 3b). Besides, activin

A overexpression was also a significant predictor of OS in

univariate analysis of Cox proportional regression models

(Table 2), but FST overexpression was not. To determine

whether activin A overexpression is an independent pre-

dictor of OS, a multivariate analysis was carried out using

tumor stage, nodal stage, overall stage, perineural invasion,

tumor differentiation, and activin A overexpression as

parameters. The results indicated that only patients with

positive nodal stages had a significantly lower OS than

patients without cervical metastasis (hazard ratio: 3.234,

P = .034, Table 2), suggesting that nodal stage is an

independent predictor of OS in patients with OSCC.

Although activin A overexpression was not a key

FIG. 1 Overexpression of activin A and follistatin (FST) in OSCC

cells. Immunohistochemical staining of activin A and FST in paired

pericancerous adjacent normal epithelium (AN) and tumor tissues

from two representative cases (scale bar = 100 lm). Expression

(brown staining) of activin A and FST indicated that the proteins were

localized in the cytoplasm of OSCC tumor cells. Both expressions

were largely absent or faintly stained in the tumor infiltrating

lymphocytes and interstitial tissues

Activin a as a Prognostic Marker of Oscc 1949

independent factor of OS in our multivariate analysis, we

suggested that activin A might be associated with nodal

metastatic processes in OSCC patients according to the

results of Table 1.

Activin A and FST Expression and Disease-Free

Survival

As shown by Kaplan-Meier plots, long-term disease-free

survival (DFS) for patients stratified by the absence or

presence of activin A overexpression was 86.5% and

49.5%, respectively. This difference in DFS was statisti-

cally significant using a log-rank test (P = .007) (Fig. 3c).

Long-term DFS for patients stratified by the absence

or presence of FST overexpression was not significantly

different using a log-rank test (70.0% and 52.4%,

respectively; P = .098) (Fig. 3d). Besides, activin A

overexpression was also a significant predictor of DFS in

univariate analysis of Cox proportional regression models

(Table 3), but FST overexpression was not. To determine

whether activin A overexpression is an independent pre-

dictor of DFS, a multivariate analysis was carried out using

tumor stage, nodal stage, overall stage, perineural invasion,

tumor differentiation, and activin A overexpression as

parameters. The results indicated that tumor stage, nodal

stage, and tumor differentiation are independent predictors

of DFS (P = .010, .033, and .002, respectively; Table 3),

but activin A is not (P = .070, Table 3). Although activin

A overexpression was also not a key independent factor of

DFS in our multivariate analysis, it still also associated

with nodal metastatic processes and tumor differentiation

in OSCC patients according to the results of Table 1.

a

25,000

20,000

15,000

10,000

5,000

Copy per106 GAPDH

Normal Tumor

Activin A

p < 0.0001

Activin A

p < 0.0001

b

300

250

200

150

50

100

ImmunohistochemicalScore

Normal Tumor

FST

p < 0.0001

300

250

200

150

50

100

ImmunohistochemicalScore

Normal Tumor

c

18

16

14

12

Activin AQuantitative Real-Time RT-PCRr = 0.507p = 0.008

6 8 10 12 14 16FST

2,000

1,500

1,000

500

Copy per106 GAPDH

Normal Tumor

FST

p = 0.0003

300

200

100

Activin AImmunohistochemistryr = 0.354p = 0.0005

0 50 100 150 200 250FST

FIG. 2 Expression and

correlation of INHBA and

follistatin (FST) in OSCC cells.

a The overexpression of INHBAand FST transcripts (P \ .0001

and P = .0003, respectively) in

OSCC tumor tissues was

demonstrated using quantitative

real-time RT-PCR. Expression of

both genes was normalized using

endogenous GAPDH expression.

b Box plot analysis of the

immunohistochemical staining

scores for activin A and FST

expression in the paired

pericancerous AN and tumor

tissues. Both activin A and FST

were highly overexpressed in

OSCC tissues (P \ .0001 and

P \ .0001, respectively).

c Correlation between activin A

and FST in OSCC tissues using

quantitative real-time RT-PCR

and immunohistochemical

analyses. Pearson’s correlation

coefficient and P values for

individual analysis are shown in

the inserts

1950 K.-P. Chang et al.

No Correlation between Serum Activin A and FST

Levels and Disease Status

Since activin A and FST are secretory proteins, we

further evaluated whether serum protein levels might serve

as biomarkers for OSCC. Serum activin A and FST levels

in untreated OSCC patients (n = 111) were 744.3 ±

2118 pg/mL and 695.9 ± 343.5 pg/mL, respectively, and

were not significantly different from control samples

(n = 91) (468.4 ± 1246 pg/mL and 760.3 ± 769.0 pg/

mL; P = .251 and P = .487; Fig. 4). Similarly, no corre-

lation was found between the serum levels of activin A and

FST and various clinicopathological characteristics of 111

patients with OSCC, including age, gender, T stage, N

stage, overall stage, or histological differentiation. These

data may indicate that overexpression of both proteins is a

local change within OSCC tumors rather than a systemic

change detectable in the circulation.

Activin A Promotes OSCC Cell Proliferation,

Migration, and Invasion In Vitro

To evaluate the biological significance of activin A

overexpression using an in vitro system, OSCC cell

endogenous expression of activin A was knocked down

using specific RNAi. The effects of RNAi were deter-

mined using quantitative real-time RT-PCR in OC3 cells

transfected with either INHBA-specific RNAi (si-INHBA)

TABLE 1 Clinicopathological characteristics related to the expression of activin A and follistatin in 92 samples of OSCCs

Activin A overexp (%) P value Follistatin overexp (%) P value

No Yes No Yes

Age

Range 30.1–72.0 30.7–82.5 31.0–82.5 30.1–72.1

Mean ± SD 47.4 ± 11.7 52.0 ± 11.4 .088 50.4 ± 13.0 51.0 ± 10.7 .812

T stage

T1–T2 13 (28) 34 (72) .914 20 (43) 27 (57) .363

T3–T4 12 (27) 33 (73) 15 (33) 30 (67)

N stage

N = 0 20 (35) 37 (65) .029a 25 (35) 32 (65) .142

N [ 0 5 (14) 30 (86) 10 (29) 25 (71)

Overall stage

I–II 11 (34) 21 (66) .257 17 (53) 15 (47) .030a

III–IV 14 (23) 46 (77) 18 (30) 42 (70)

Differentiation

W-D 16 (47) 18 (53) .002a 16 (47) 18 (53) .176

M-D 9 (16) 41 (82) 18 (36) 32 (64)

P-D 0 (0) 8 (100) 1 (12) 7 (88)

Perineural invasion

Negative 22 (35) 41 (65) .014a 28 (44) 35 (56) .062

Positive 3 (10) 26 (90) 7 (24) 22 (76)

Alcohol consumption

No 11 (29) 27 (71) .748 17 (45) 21 (55) .267

Yes 14 (26) 40 (74) 18 (33) 36 (67)

Betel nut chewing

No 4 (36) 7 (66) .482 6 (55) 5 (45) .229

Yes 21 (26) 60 (74) 29 (29) 52 (52)

Smoking

No 2 (18) 9 (82) .721 4 (36) 7 (64) .902

Yes 23 (28) 58 (72) 31 (38) 50 (62)

Total 25 67 35 57

overexp overexpression, W-D well-differentiated squamous cell carcinoma, M-D moderately differentiated squamous cell carcinoma, P-D poorly

differentiated squamous cell carcinomaa Statistically significant

Activin a as a Prognostic Marker of Oscc 1951

or a scrambled sequence control RNAi. As shown in

Fig. 5a, expression of endogenous INHBA was signifi-

cantly reduced (approximately 32-fold reduction) in

si-INHBA-transfected cells compared with control RNAi

transfectants.

Control and si-INHBA-transfected cells were further

analyzed for cell proliferation, migration, and invasiveness.

As shown in Fig. 5b, the cell proliferation ability in OC3

cell proliferation was also significantly reduced by si-IN-

HBA (37% reduction, P \ 0.0001). Also, both cell

migration and cell invasiveness were attenuated by addi-

tion of si-INHBA (63% and 67% reduction, P \ 0.001 and

P \ .0001, respectively; Fig. 5c, d). Similar decreases cell

proliferation, migration, and invasiveness were observed

when SCC4 cells, another OSCC cell line, were transfected

with si-INHBA (data not shown). Collectively, these

1.0

0.8

0.6

0.4

0.2

0 60Months

30 4010 20 50

0 60Months

30 4010 20 50

Activin A (0, +)Activin A (++, +++)

p = 0.009

OverallSurvival

a

1.0

0.8

0.6

0.4

0.2

Activin A (0, +)Activin A (++, +++)

p = 0.007

Disease-FreeSurvival

c

1.0

0.8

0.6

0.4

0.2

0 60Months

30 4010 20 50

0 60Months

30 4010 20 50

FST (0, +)FST (++, +++)

p = 0.179

OverallSurvival

b

1.0

0.8

0.6

0.4

0.2

FST (0, +)FST (++, +++)

p = 0.098

Disease-FreeSurvival

d

FIG. 3 Overexpression of

activin A was associated with a

poorer prognosis with regard to

OSCC patient overall survival

and disease-free survival.

a, b Kaplan-Meier plot for

overall survival indicated that the

4-year overall survival for

patient subgroups stratified by

overexpression of activin A and

FST was 91.3% versus 55.5%

(P = .009) and 72.3% versus

60.1% (P = .179), respectively.

c, d Kaplan-Meier plot for

disease-free survival indicated

that the 4-year disease-free

survival for patient subgroups

stratified by overexpression of

activin A and FST was 86.5%

versus 54.0% (P = .007) and

70.0% versus 57.5% (P = .098),

respectively

TABLE 2 Cox proportional hazard models on overall survival

Univariate crude HR (95% CI) P value Multivariate adjusted HR (95% CI) P value

T stage

T3–T4 vs. T1–T2

1.632 (0.784–3.394) .190 1.807 (0.718–4.546) .208

N stage

N [ 0 vs. N = 0

4.262 (1.971–9.216) \.001a 3.234 (1.090–9.595) .034a

Overall stage

III–IV vs. I–II

3.312 (1.262–8.691) .015a 0.689 (0.146–3.263) .639

Differentiation

P-D vs. W-D ? M-D

6.308 (1.905–20.892) .002a 3.421 (0.940–12.456) .062

Perineural invasion

Positive vs. negative

2.686 (1.288–5.598) .008a 1.357 (0.628–2.934) .437

Activin A overexpression

Yes vs. no

5.533 (1.315–23.281) .019a 2.610 (0.579–11.760) .211

HR hazard ratio, CI confidence interval, W-D well-differentiated squamous cell carcinoma, M-D moderately differentiated squamous cell

carcinoma, P-D poorly differentiated squamous cell carcinomaa Statistically significant

1952 K.-P. Chang et al.

findings indicate that overexpression of activin A in vitro

can mediate cell proliferation, migration, and invasion

processes in OSCC cells.

DISCUSSION

Overexpression of activin A has been reported for var-

ious cancers, but its mechanistic roles in cancer cells and

contributions to the carcinogenic process have remained to

be elucidated. The results of the present study strongly

suggest that overexpression of activin A in OSCC is

associated with poorer cell differentiation, positive cervical

nodal status, and perineural invasion of OSCC tumors and

therefore may be a risk factor for OSCC tumor progression.

These three different clinicopathological factors have all

been suggested as poor prognostic factors for locoregional

control in OSCC and may explain why activin A expres-

sion has a vital impact on long-term OS and DFS.38 These

findings are in agreement with the previous reports

that indicated that activin A overexpression was associated

with advanced stages of colon cancer and lymph node

metastasis and poor prognosis of esophageal squamous cell

carcinomas.14,15 Although activin A and FST are secretory

proteins and serum levels of activin A have been proposed

as a potential tumor marker in some cancers, our results

clearly demonstrated that overexpression of activin A was

only found in cancer tissues.20–22 In the present patient

cohort, neither activin A nor FST serum levels could be

used to differentiate OSCC patients from healthy controls

or as a stratification marker of the clinicopathological

manifestations of OSCC patients. In other words, activin A

secreted from OSCC cells might have pivotal roles in

tumor biology through autocrine or paracrine mechanisms,

but it does not have systemic effects and is not a potential

serum marker for OSCC.

In vivo and in vitro studies of various types of malig-

nancies have demonstrated that activin A promotes cancer

cell growth, migration, and invasion.23–26 In addition,

studies of prostate and lung cancers revealed that inhibition

of activin A expression by its specific RNAi or antibody

neutralization attenuated cancer cell proliferation and cell

migration, respectively, which was similar to the present

TABLE 3 Cox proportional hazard models on disease-free survival

Univariate crude HR (95% CI) P value Multivariate adjusted HR (95% CI) P value

T stage

T3–T4 vs. T1–T2

2.701 (1.309–5.573) .007a 4.219 (1.404–12.673) .010a

N stage

N [ 0 vs. N = 0

3.651 (1.805–7.381) \.001a 2.639 (1.080–6.449) .033a

Overall stage

III–IV vs. I–II

5.254 (1.844–14.966) .002a 0.893 (0.188–4.242) .886

Differentiation

P-D vs. W-D ? M-D

4.970 (2.118–11.663) \.001a 6.038 (1.886–19.334) .002a

Perineural invasion

Positive vs. negative

2.225 (1.112–4.452) .024a 0.921 (0.393–2.158) .849

Activin A overexpression

Yes vs. no

4.336 (1.322–14.218) .015a 3.058 (0.911–10.260) .070

HR hazard ratio, CI confidence interval, W-D well-differentiated squamous cell carcinoma, M-D moderately differentiated squamous cell

carcinoma, P-D poorly differentiated squamous cell carcinomaa Statistically significant

Activin A

p = 0.251

a

14,000

12,000

10,000

250

500

Concentration(pg/ml)

OSCC Control

FST

p = 0.487

b

7,000

6,000

5,000

1,000

2,000

3,000

4,000

Concentration(pg/ml)

OSCC Control

FIG. 4 Serum activin A and

FST levels in untreated OSCC

patients (n = 111) were

744.3 ± 2118 pg/mL and

695.9 ± 343.5 pg/mL,

respectively, and were not

significantly different from

control samples (n = 91)

(468.4 ± 1246 pg/mL and

760.3 ± 769.0 pg/mL; P = .251

and P = .487)

Activin a as a Prognostic Marker of Oscc 1953

study using OSCC cells.19,39 Furthermore, RNAi targeting

of an activin A downstream effector, Smad3, reduced cell

proliferation and invasion.40 Thus, these results highlight

the role of activin A and its downstream signaling on

cancer cell malignancies. Another possible mechanism by

which activin A overexpression can promote malignant

characteristics of various cancers might be through stimu-

lation of vascular endothelial growth factor (VEGF)

expression.41,42 Because VEGF is a key regulator of

angiogenesis and its overexpression is associated with poor

prognosis of cancer patients, several therapeutic agents,

such as bevacizumab, that inhibit the activity of VEGF or

its receptors have been developed for the treatment of solid

malignancies.43,44 Therefore, inhibition of activin A and its

downstream signaling effectors might be of clinical rele-

vance for OSCC treatment in the future.

In the current study, we observed statistically significant

correlations between activin A and FST mRNA and protein

expression. Although FST is an antagonist of activin A, it

is also overexpressed in OSCC. Other negative regulators

of activin A, including another binding antagonist, FLRG,

and cell membrane antagonistic coreceptors, such as Cripto

or BAMBI, have also been discovered.45 The interaction

between activin A and its antagonists in different tumor

microenvironment will need to be further evaluated in the

future studies. To the best of our knowledge, the correla-

tion between activin A and FST has not been well

addressed to date. In a previous study of hepatocellular

carcinoma, deregulation of the FST-activin A system was

reported.27 FST overexpression was observed in both of

human and rat liver tumors, stimulating DNA synthesis

preferentially in preneoplastic rat hepatocytes.27 Moreover,

another study demonstrated that differential antagonism of

activin A occurred through binding with wild-type or

mutant FST.46 These findings might help explain why FST

overexpression was also associated with poor stages of

MigrationAssay

p < 0.001

c

e

1,500

1,000

500

Cells(No. /Field)

Scramble si-INHBA

Scramble

CellMigration

CellInvasion

si-INHBA

InvasionAssay

p < 0.0001

d

250

200

150

50

100

Cells(No. /Field)

Scramble si-INHBA

MTT Assay

b

20

0

–20

–40

Cell Growth(Percent)

PBS Scramble

50,000

40,000

30,000

20,000

10,000

0 40Cycle

15 205 10 25 30 35

Scramble-GAPDHsi-INHBA-GAPDHScramble-INHBAsi-INHBA-INHBA

Fluorescence(dR)

a

si-INHBA

FIG. 5 OSCC cell proliferation,

migration, and invasiveness was

attenuated by INHBA-specific

RNAi. a A plot of the threshold

PCR cycle showing expression

of endogenous GAPDH and

INHBA in OC3 cells transfected

with si-INHBA or a control

scrambled RNAi shown in a log

scale. The relative transcripts of

INHBA in si-INHBA (as orangeline shown) and scramble

transfected cells (as green lineshown) were normalized using

GAPDH endogenous expression

levels (as blue and light orangelines indicated). INHBAtranscripts were reduced about

32-fold in si-INHBA-transfected

OSCC cells as comparing

scramble transfectant. b MTT

assays showed that OC3 cell

proliferation was significantly

decreased following si-INHBA

transfection (37%, P \ 0.0001).

c, d Both OC3 cell migration and

invasiveness were attenuated by

INHBA-specific RNAi (63% and

67%, P \ 0.001 and P \ 0.0001,

respectively). e Representative

example images of cell

migration and invasion assays

are shown for control RNAi and

si-INHBA-treated groups. Data

are the mean ± SD of 3

independent experiments

1954 K.-P. Chang et al.

OSCC patients. Nevertheless, the true mechanism remains

to be elucidated and awaits further investigation in the

future.

In the current study, when endogenous activin A

expression was inhibited by specific RNAi, the prolifera-

tive, migratory, and invasive capabilities of OSCC cells

were significantly reduced. To understand the role of

activin A in carcinogenesis, many previous studies have

reported their in vitro findings illustrating potential mech-

anisms. Activin A has been proposed to induce a RhoA-

ROCK-MEKK1-JNK pathway and a MEKK1-p38 path-

way as Smad-independent mechanisms controlling actin

cytoskeleton reorganization and promoting the migratory

and invasive potential in human epithelial cells during

epithelial-to-mesenchymal transitions.47 Also, activin A

induction of N-cadherin and MMP-7 overexpression are

reported to be associated with the invasiveness of esoph-

ageal cancer cells.25,48 Recently, overexpression of activin

A has been suggested to mediate ovarian carcinogenesis

through activation of Akt and repression of GSK, resulting

in increased cell proliferation.49 Collectively, inhibition of

activin A might be a potential method for attenuating

OSCC tumor progression and aggressiveness.

In conclusion, this study provides the first evidence that

activin A is aberrantly overexpressed in OSCC cells rela-

tive to histologically noncancerous tissue. In addition,

overexpression of activin A is associated with several

important clinicopathological factors and patient prognosis

in survival analyses for OSCC, suggesting its potential as

an immunohistochemical marker of OSCC. Furthermore,

inhibition of activin A expression can attenuate OSCC cell

proliferation, migration, and invasiveness, supporting

activin A as potential target of future OSCC therapeutic

agents.

ACKNOWLEDGMENT This study was supported by Chang

Gung Memorial Hospital (Grant No. CMRPG360212 and

CMRPG340481) and National Science Council (Grant No. NSC96-

2314-B-182A-109-MY3), Taiwan.

REFERENCES

1. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics,

2002. CA Cancer J Clin. 2005;55:74–108.

2. Reid BC, Winn DM, Morse DE, Pendrys DG. Head and neck in

situ carcinoma: incidence, trends, and survival. Oral Oncol.2000;36:414–20.

3. Boring CC, Squires TS, Tong T, Montgomery S. Cancer statis-

tics, 1994. CA Cancer J Clin. 1994;44:7–26.

4. Silverman S, Jr. Demographics and occurrence of oral and pha-

ryngeal cancers. The outcomes, the trends, the challenge. J AmDent Assoc. 2001;132 Suppl:7S–11S.

5. Kodani I, Shomori K, Osaki M, Kuratate I, Ryoke K, Ito H.

Expression of minichromosome maintenance 2 (MCM2), Ki-67,

and cell-cycle-related molecules, and apoptosis in the normal-

dysplasia-carcinoma sequence of the oral mucosa. Pathobiology.2001;69:150–8.

6. Lippman SM, Sudbo J, Hong WK. Oral cancer prevention and the

evolution of molecular-targeted drug development. J Clin Oncol.2005;23:346–56.

7. Silverman S Jr., Gorsky M, Lozada F. (1984) Oral leukoplakia

and malignant transformation. A follow-up study of 257 patients.

Cancer. 53:563–8.

8. Sudbo J. Novel management of oral cancer: a paradigm of pre-

dictive oncology. Clin Med Res. 2004;2:233–42.

9. Kingsley DM. The TGF-beta superfamily: new members, new

receptors, and new genetic tests of function in different organ-

isms. Genes Dev. 1994;8:133–46.

10. Ling N, Ying SY, Ueno N, Shimasaki S, Esch F, Hotta M, et al.

Pituitary FSH is released by a heterodimer of the beta-subunits

from the two forms of inhibin. Nature. 1986;321:779-82.

11. Vale W, Rivier J, Vaughan J, McClintock R, Corrigan A, Woo

W, et al. Purification and characterization of an FSH releasing

protein from porcine ovarian follicular fluid. Nature. 1986;

321:776–9.

12. Risbridger GP, Mellor SL, McPherson SJ, Schmitt JF. The

contribution of inhibins and activins to malignant prostate

disease. Mol Cell Endocrinol. 2001;180:149–53.

13. Risbridger GP, Schmitt JF, Robertson DM. Activins and inhibins

in endocrine and other tumors. Endocr Rev. 2001;22:836–58.

14. Wildi S, Kleeff J, Maruyama H, Maurer CA, Buchler MW, Korc

M. Overexpression of activin A in stage IV colorectal cancer.

Gut. 2001;49:409–17.

15. Yoshinaga K, Mimori K, Yamashita K, Utsunomiya T, Inoue H,

Mori M. Clinical significance of the expression of activin A in

esophageal carcinoma. Int J Oncol. 2003;22:75–80.

16. Zheng W, Luo MP, Welt C, Lambert-Messerlian G, Sung CJ,

Zhang Z, et al. Imbalanced expression of inhibin and activin

subunits in primary epithelial ovarian cancer. Gynecol Oncol.1998;69:23–31.

17. Thomas TZ, Wang H, Niclasen P, O’Bryan MK, Evans LW,

Groome NP, et al. Expression and localization of activin subunits

and follistatins in tissues from men with high grade prostate

cancer. J Clin Endocrinol Metab.1997;82:3851–8.

18. Kleeff J, Ishiwata T, Friess H, Buchler MW, Korc M. Concom-

itant over-expression of activin/inhibin beta subunits and their

receptors in human pancreatic cancer. Int J Cancer. 1998;77:860–

8.

19. Seder CW, Hartojo W, Lin L, Silvers AL, Wang Z, Thomas DG,

et al. Upregulated INHBA expression may promote cell prolif-

eration and is associated with poor survival in lung

adenocarcinoma. Neoplasia. 2009;11:388–96.

20. Reis FM, Cobellis L, Tameirao LC, Anania G, Luisi S, Silva IS,

et al. Serum and tissue expression of activin a in postmenopausal

women with breast cancer. J Clin Endocrinol Metab. 2002;87:

2277–82.

21. Pirisi M, Fabris C, Luisi S, Santuz M, Toniutto P, Vitulli D, et al.

Evaluation of circulating activin-A as a serum marker of hepa-

tocellular carcinoma. Cancer Detect Prev. 2000;24:150–5.

22. Petraglia F, Florio P, Luisi S, Gallo R, Gadducci A, Vigano P,

et al. Expression and secretion of inhibin and activin in normal

and neoplastic uterine tissues. High levels of serum activin A in

women with endometrial and cervical carcinoma. J Clin Endo-crinol Metab. 1998;83:1194–200.

23. Coerver KA, Woodruff TK, Finegold MJ, Mather J, Bradley A,

Matzuk MM. Activin signaling through activin receptor type II

causes the cachexia-like symptoms in inhibin-deficient mice. MolEndocrinol. 1996;10:534–43.

24. Di Simone N, Crowley WF, Jr., Wang QF, Sluss PM, Schneyer

AL. Characterization of inhibin/activin subunit, follistatin, and

activin type II receptors in human ovarian cancer cell lines: a

Activin a as a Prognostic Marker of Oscc 1955

potential role in autocrine growth regulation. Endocrinology1996;137:486–94.

25. Yoshinaga K, Inoue H, Utsunomiya T et al. N-cadherin is regu-

lated by activin A and associated with tumor aggressiveness in

esophageal carcinoma. Clin Cancer Res. 2004;10:5702–7.

26. Yoshinaga K, Yamashita K, Mimori K, Tanaka F, Inoue H, Mori

M. Activin a causes cancer cell aggressiveness in esophageal

squamous cell carcinoma cells. Ann Surg Oncol. 2008;15:96–103.

27. Grusch M, Drucker C, Peter-Vorosmarty B, Erlach N, Lackner A,

Losert A, et al. Deregulation of the activin/follistatin system in

hepatocarcinogenesis. J Hepatol. 2006;45:673–80.

28. Matsuo SE, Leoni SG, Colquhoun A, Kimura ET. Transforming

growth factor-beta1 and activin A generate antiproliferative sig-

naling in thyroid cancer cells. J Endocrinol. 2006;190:141–50.

29. Reis FM, Luisi S, Carneiro MM, Cobellis L, Federico M, Cam-

argos AF, et al. Activin, inhibin and the human breast. Mol CellEndocrinol. 2004;225:77–82.

30. de Winter JP, ten Dijke P, de Vries CJ, van Achterberg TA,

Sugino H, de Waele P, et al. Follistatins neutralize activin bio-

activity by inhibition of activin binding to its type II receptors.

Mol Cell Endocrinol. 1996;116:105–14.

31. Ogino H, Yano S, Kakiuchi S, Muguruma H, Ikuta K, Hanibuchi

M, et al. Follistatin suppresses the production of experimental

multiple-organ metastasis by small cell lung cancer cells in nat-

ural killer cell-depleted SCID mice. Clin Cancer Res. 2008;

14:660–7.

32. Maeshima K, Maeshima A, Hayashi Y, Kishi S, Kojima I.

Crucial role of activin a in tubulogenesis of endothelial cells

induced by vascular endothelial growth factor. Endocrinology.2004;145:3739–45.

33. Shimizu S, Seki N, Sugimoto T, Horiguchi S, Tanzawa H, Ha-

nazawa T, et al. Identification of molecular targets in head and

neck squamous cell carcinomas based on genome-wide gene

expression profiling. Oncol Rep. 2007;18:1489–97.

34. Ye H, Yu T, Temam S, Ziober BL, Wang J, Schwartz JL, et al.

Transcriptomic dissection of tongue squamous cell carcinoma.

BMC Genomics. 2008;9:69.

35. Liao CT, Chang JT, Wang HM, Ng SH, Hsueh C, Lee LY, et al.

Surgical outcome of T4a and resected T4b oral cavity cancer.

Cancer. 2006;107:337–44.

36. De Marzo AM, Knudsen B, Chan-Tack K, Epstein JI. E-cadherin

expression as a marker of tumor aggressiveness in routinely

processed radical prostatectomy specimens. Urology. 1999;53:

707–13.

37. Lin SC, Liu CJ, Chiu CP, Chang SM, Lu SY, Chen YJ. Estab-

lishment of OC3 oral carcinoma cell line and identification of

NF-kappa B activation responses to areca nut extract. J OralPathol Med. 2004;33:79–86.

38. Liao CT, Huang SF, Chen IH, Chang JT, Wang HM, Ng SH,

et al. Risk stratification of patients with oral cavity squamous cell

carcinoma and contralateral neck recurrence following radical

surgery. Ann Surg Oncol. 2009;16:159–70.

39. Simon DP, Vadakkadath Meethal S, Wilson AC, Gallego MJ,

Weinecke SL, Bruce E, et al. Activin receptor signaling regulates

prostatic epithelial cell adhesion and viability. Neoplasia. 2009;

11:365–76.

40. Kang HY, Huang HY, Hsieh CY, Li CF, Shyr CR, Tsai MY, et al.

Activin A enhances prostate cancer cell migration through acti-

vation of androgen receptor and is overexpressed in metastatic

prostate cancer. J Bone Miner Res. 2009;24:1180–93.

41. Wagner K, Peters M, Scholz A, Benckert C, Ruderisch HS,

Wiedenmann B, et al. Activin A stimulates vascular endothelial

growth factor gene transcription in human hepatocellular carci-

noma cells. Gastroenterol. 2004;126:1828–43.

42. Poulaki V, Mitsiades N, Kruse FE, Radetzky S, Iliaki E, Kirchhof

B, et al. Activin a in the regulation of corneal neovascularization

and vascular endothelial growth factor expression. Am J Pathol.2004;164:1293–302.

43. Shih T, Lindley C. Bevacizumab: an angiogenesis inhibitor for

the treatment of solid malignancies. Clin Ther. 2006;28:1779–

802.

44. Ferrara N, Davis-Smyth T. The biology of vascular endothelial

growth factor. Endocr Rev. 1997;18:4–25.

45. Kelber JA, Panopoulos AD, Shani G, Booker EC, Belmonte JC,

Vale WW, et al. Blockade of Cripto binding to cell surface

GRP78 inhibits oncogenic Cripto signaling via MAPK/PI3 K and

Smad2/3 pathways. Oncogene. 2009;28:2324–36.

46. Schneyer AL, Sidis Y, Gulati A, Sun JL, Keutmann H, Krasney

PA. Differential antagonism of activin, myostatin and growth and

differentiation factor 11 by wild-type and mutant follistatin.

Endocrinology 2008;149:4589–95.

47. Zhang L, Deng M, Parthasarathy R, Wang L, Mongan M,

Molkentin JD, et al. MEKK1 transduces activin signals in

keratinocytes to induce actin stress fiber formation and migration.

Mol Cell Biol. 2005;25:60–5.

48. Yoshinaga K, Mimori K, Inoue H, Kamohara Y, Yamashita K,

Tanaka F, et al. Activin A enhances MMP-7 activity via the

transcription factor AP-1 in an esophageal squamous cell carci-

noma cell line. Int J Oncol. 2008;33:453–9.

49. Do TV, Kubba LA, Antenos M, Rademaker AW, Sturgis CD,

Woodruff TK. The role of activin A and Akt/GSK signaling in

ovarian tumor biology. Endocrinology 2008;149:3809–16.

1956 K.-P. Chang et al.