ORIGINAL PAPER

Down-modulation of TNFSF15 in ovarian cancer by VEGFand MCP-1 is a pre-requisite for tumor neovascularization

Weimin Deng • Xin Gu • Yi Lu • Chao Gu •

Yangyang Zheng • Zhisong Zhang •

Li Chen • Zhi Yao • Lu-Yuan Li

Received: 28 September 2011 / Accepted: 8 December 2011 / Published online: 31 December 2011

� Springer Science+Business Media B.V. 2011

Abstract Persistent inflammation and neovascularization

are critical to cancer development. In addition to upregu-

lation of positive control mechanisms such as overexpres-

sion of angiogenic and inflammatory factors in the cancer

microenvironment, loss of otherwise normally functioning

negative control mechanisms is likely to be an important

attribute. Insights into the down-modulation of such neg-

ative control mechanisms remain largely unclear, however.

We show here that tumor necrosis factor superfamily-15

(TNFSF15), an endogenous inhibitor of neovasculariza-

tion, is a critical component of the negative control

mechanism that operates in normal ovary but is missing in

ovarian cancer. We show in clinical settings that TNFSF15

is present prominently in the vasculature of normal ovary

but diminishes in ovarian cancer as the disease progresses.

Vascular endothelial growth factor (VEGF) produced by

cancer cells and monocyte chemotactic protein-1 (MCP-1)

produced mainly by tumor-infiltrating macrophages and

regulatory T cells effectively inhibits TNFSF15 production

by endothelial cells in vitro. Using a mouse syngeneic

tumor model, we demonstrate that silencing TNFSF15 by

topical shRNA treatments prior to and following mouse

ovarian cancer ID8 cell inoculation greatly facilitates

angiogenesis and tumor growth, whereas systemic appli-

cation of recombinant TNFSF15 inhibits angiogenesis and

tumor growth. Our findings indicate that downregulation of

TNFSF15 by cancer cells and tumor infiltrating macro-

phages and lymphocytes is a pre-requisite for tumor

neovascularization.

Keywords Inflammation � Ovarian cancer � Angiogenesis

switch � Tumorigenesis � Tumor infiltrating lymphocyte �Macrophage

Introduction

Chronic inflammation, fueled by periodic hypoxic condi-

tions, disrupted vascular integrity, lymphocyte infiltration,

and continual generation of dysfunctional blood vessels, is

a recognized risk factor for malignancies [1–6]. Cancer

cells in hypoxia produce a number of protein factors to

promote angiogenesis, including vascular endothelial cell

growth factor (VEGF) [7, 8]. VEGF, which was known

initially as vascular permeability factor [9], causes an

increase of permeability of the blood vessel. Blood vessel

leakage becomes prominent as a result, which greatly

Electronic supplementary material The online version of thisarticle (doi:10.1007/s10456-011-9244-y) contains supplementarymaterial, which is available to authorized users.

W. Deng � Y. Zheng � Z. Zhang � L.-Y. Li (&)

State Key Laboratory of Medicinal Chemical Biology,

Tianjin, China

e-mail: liluyuan@nankai.edu.cn

W. Deng � X. Gu � Y. Lu � C. Gu � Z. Yao (&)

Key Laboratory of Cellular and Molecular Immunology,

Tianjin Medical University, Tianjin, China

e-mail: yaozhi@tmu.cn

W. Deng � Y. Lu � C. Gu � Y. Zheng � Z. Yao

Key Laboratory of Ministry of Education, Tianjin, China

Y. Lu � L. Chen � L.-Y. Li

Department of Pathology, University of Pittsburgh School

of Medicine, Pittsburgh, PA, USA

Y. Lu � L. Chen � L.-Y. Li

University of Pittsburgh Cancer Institute, Pittsburgh, PA, USA

Y. Zheng � Z. Zhang � L.-Y. Li

Nankai University School of Pharmaceutical Science, Tianjin,

China

123

Angiogenesis (2012) 15:71–85

DOI 10.1007/s10456-011-9244-y

facilitates infiltration of macrophages and T cells [10, 11],

among others. These cells secrete a variety of inflammatory

cytokines, including monocyte chemotactic protein-1

(MCP-1), into the microenvironment, further driving

angiogenesis [12, 13]. MCP-1 is of particular interest to us

because it can be produced by a variety of cell types in the

cancer microenvironment, such as monocytes and macro-

phages in the case of ovarian cancer [14–17], either con-

stitutively or induced under oxidative stress or other

inflammatory conditions [18–21]. Malignant tumors are

thus likened to wounds that will not heal because of this

vicious cycle [22]. It is hypothesized that the maintenance

of vascular integrity is tightly controlled by a balance of

growth factors and cytokines of opposite functions to

impede or accelerate neovascularization and hyperplasia

[23]. Conceivably, while an overall antiangiogenesis effect

prevails in a normal vasculature because of the activities of

endogenous suppressors of angiogenesis, these activities

must be hampered prior to the initiation of a neovascular-

ization process. Insights into how this happens remain,

however, largely unclear.

The ovary is a unique organ where the initiation, for-

mation, and degeneration of blood vessels under physio-

logical conditions occur in a cyclic manner [24–26].

Similar to the rest of the human female reproductive tract,

the ovary is highly dependent on VEGF for normal func-

tions such as endometrial proliferation and corpus luteum

development [27]. The expression and activity of VEGF

influenced by female steroid hormones deems angiogenesis

an important facet of the development of ovarian cancer

when the harmonic balance is compromised [28]. VEGF is

constitutively expressed in both benign and malignant

human ovary tissue [29]. VEGF and its receptors are

present in both primary ovarian carcinomas and ascitic

fluid [30]. Comparison of normal ovarian tissue to ovarian

carcinoma shows that VEGF protein levels are significantly

higher in carcinoma compared with normal tissue [31].

Malignant ovarian cyst fluid has elevated VEGF levels

[32]. Formation of malignant ascites has been directly

linked to VEGF expression in ovarian cancers [33]. VEGF

overexpression is correlated with a shorter disease-free

survival [34]. Similarly, tumor microvessel density in

advanced ovarian carcinoma is associated with a poorer

overall survival [35]. VEGF thus is a main driving force of

ovarian cancer angiogenesis and inflammation.

Tumor necrosis factor superfamily-15 (TNFSF15;

VEGI; TL1A) is a unique endogenous inhibitor of neo-

vascularization [36–38]. TNFSF15 is produced largely by

vascular endothelial cells of established blood vessels and

is a specific inhibitor of endothelial cell proliferation.

TNFSF15 is able to enforce growth arrest on quiescent

endothelial cells but induce apoptosis of proliferating

endothelial cells [39]. TNFSF15 can prevent bone marrow-

derived endothelial progenitor cell (EPC) differentiation

into endothelial cells [40]. Systemically administered

recombinant TNFSF15 exhibits a highly potent inhibitory

effect on tumor angiogenesis and tumor growth in animal

models [41]. It also inhibits the incorporation of bone

marrow-derived EPC into tumor vasculature [42].

TNFSF15 expression is reported to be absent or marginal

in tumor vasculatures in breast cancer [43], prostate cancer

[44, 45], urothelial cancer [46], as well as in wound tissue

[47]. Together these findings support the view that

TNFSF15 is an important component of the negative

control mechanisms of neovascularization.

We report here that TNFSF15 is downregulated in ovarian

cancer by VEGF produced by cancer cells and MCP-1 pro-

duced mainly by tumor-infiltrating macrophage and

CD4?CD25? FOXP3? regulatory T (TReg) cells under the

influence of cancer cells. We show with an ovarian cancer

model that TNFSF15 gene silencing with shRNA leads to

markedly enhanced angiogenesis and tumor growth,

whereas systemic administration of recombinant TNFSF15

inhibits angiogenesis and tumor growth. The findings are

consistent with the view that downmodulation of TNFSF15

is a pre-requisite for tumor neovascularization.

Materials and methods

Antibodies and reagents

Recombinant VEGF and antibodies against human VEGF,

VEGFR1, VEGFR2, and goat isotype control were pur-

chased from R&D (Minneapolis, MN, USA). Recombinant

human MCP-1, anti-human MCP-1 antibody, and mouse

isotype control were from Sigma (St. Louis, MO, USA).

Anti-mouse CD31 antibody was from BD (Franklin Lakes,

NJ, USA). Antibodies against human CD31, CD4, MAC1

and mouse TNFSF15 were from Santa Cruz (Santa Cruz,

CA, USA). Alexa Fluor-647 anti-human CD25, Alexa

Fluor-488 anti-human CD4 and Pacific Blue anti-human

FOXP3 were from BioLegend (San Diego, CA, USA).

Recombinant human TNFSF15 and anti-human TNFSF15

and mouse monoclonal antibodies 3-12D and 1-8F against

human TNFSF15 were prepared as described (32). Endo-

toxin level in the TNFSF15 preparation was 25 ng/mg.

Cells

Human umbilical vein endothelial cell (HUVEC) and

human ovarian cancer cell line OVCAR3 were purchased

from American Type Culture Collection (Manassas, VA,

USA) and cultured under conditions recommended by the

vendor. Mouse ovarian epithelial papillary serous adeno-

carcinoma cell line ID8 was a gift from Dr. K. F. Roby

72 Angiogenesis (2012) 15:71–85

123

(Center for Reproductive Sciences, University of Kansas

Medical Center). ID8 cells were cultured in Dulbecco’s

modified Eagle’s medium (Invitrogen, CA, USA) supple-

mented with 4% fetal bovine serum, 100 units/ml penicil-

lin, 100 lg/ml streptomycin, 5 lg/ml insulin (which is

known to stimulate VEGF production [48]), 5 lg/ml

transferrin, and 5 ng/ml sodium selenite (Roche, IN, USA).

Mice

Six-week-old female C57BL/6 mice were purchased from

the Jackson Laboratory (Bar Harbor, ME, USA). All pro-

cedures involving experimental animals were performed in

accordance with protocols approved by the University of

Pittsburgh Institutional Animal Care and Use Committee.

Immunohistochemistry

Five-micrometer sections of formalin-fixed, paraffin-

embedded tumors of clinical specimens of human ovarian

cancer or an experimental model of mouse ID8 syngeneic

tumors, as indicated, were deparaffinized, microwaved for

15 min in citric acid solution (pH = 6.0), and then incu-

bated in 3% hydrogen peroxide for 15 min. The sections

were then incubated for 2 h at room temperature with

respective primary antibodies, followed by 50 min incu-

bation with secondary antibody. For the use of fluorescein-

labeled antibodies, the slides were incubated at room

temperature for 2 h in the dark. The sections were sub-

jected to microscopic analysis with a Nikon NIS-Elements

microscope or an Olympus Fluo-view 1000 confocal

microscope. A minimum of 10 fields per section was

analyzed. The intensity of the immunostaining on each

section was assessed by two clinical pathologists inde-

pendently and in blinded manner, using a four-step grading

system (-, ?, ??, ??? for negative, low, high, and very

high, respectively).

CD4? T cell, CD4?CD25? T cell and monocyte

isolation

One hundred milliliters of peripheral blood was collected

from a healthy donor. Procedures involving human donors

were carried out in accordance with a University of Pitts-

burgh Institutional Review Board approved protocol.

Peripheral blood mononuclear cells were prepared by using

density gradient centrifuge (Ficoll, Uppsala, Sweden), and

then used for the isolation of CD4? T cells and

CD4?CD25? T cells with respective immobilized anti-

bodies (Miltenyi, Auburn, CA, USA). The isolation of

monocyte was carried out by depleting non-monocytes

with a Biotin antibody cocktail against CD3, CD7, CD16,

CD19, CD56, CD123 and Glycophorin A (Miltenyi,

Auburn, CA, USA) according to manufacturer’s instruc-

tions. Cells from different donors (three females, one male)

were used for experimental repeats.

Enzyme-linked immunosorbant assay (ELISA)

VEGF, TNFSF15 and MCP-1 concentration was deter-

mined by ELISA respectively. For TNFSF15, mouse

monoclonal antibodies 3-12D and 1-8F were used as the

capturing and detecting reagents, respectively. Recombi-

nant TNFSF15 was used to construct the standard curve.

MCP-1 and VEGF were determined by using the respective

MCP-1 or VEGF Elisa Kit (R&D, Minneapolis, MN,

USA).

Analysis of VEGF, TNFSF15, and MCP-1 expression

OVCAR3 cells (1 9 106 cells in 5 ml DMEM containing

10% FBS in 25-cm2 flasks) were cultured at 37�C in an

atmosphere of 5% CO2 in either 21% O2 (normoxia) or 5%

O2 (hypoxia). The culture media were collected at indicated

time intervals and either analyzed for VEGF protein levels or

used as OVCAR3-conditioned media. Freshly isolated

CD4?CD25? T cells (1 9 106) were suspended in 1.5 ml

RPMI 1640 culture medium (10% FBS) and seeded onto

12-well plates, then treated with 0.5 ml cancer cell condi-

tioned medium that was prepared by culturing OVCAR3

cells (1 9 106) in 5 ml DMEM (10% FBS) in a 25-cm2 flask

for 72 h. The supernatants were subjected to analysis with a

RayBio Human Cytokine Antibody Array G Series-6 panel

that included MCP-1 (Raybiotech, Norcross, GA, USA)

according to the manufacturer’s instructions. Conditioned

media from OVCAR3 cultures or CD4?CD25? T cell

untreated with OVCAR3 were used as controls. HUVECs

(passages 5–7) were plated onto 12-well culture plate with

the density of 2 9 105 cells per well in 2 ml of EGM-2. Once

grown into confluent cultures in about 7 days, the cultures

were treated with VEGF, MCP-1, OVCAR3-CM, or the

conditioned media from OVCAR3-CM treated CD4?

CD25? T cell cultures. Neutralizing antibodies against

VEGF, VEGFR1, VEGFR2, or MCP-1 were added to some

of the culture media as indicated.

Treatment of ID8 tumor-bearing animals

with recombinant TNFSF15

Six-week-old female C57BL/6 mice were injected subcuta-

neously on the flank with ID8 cells (1 9 107) suspended in

PBS (Lonza, Basel, Switzerland). Tumor-bearing animals

were randomized on day 6 into two groups. The experimental

group received intraperitoneal injection of TNFSF15

(125 lg in 0.5 ml PBS) on the same day and every other day

thereafter. The control group received vehicle treatments.

Angiogenesis (2012) 15:71–85 73

123

The animals were observed and weighed daily prior to

treatment. Tumor sizes were measured with a dial caliper in a

blinded manner. Tumor volumes were determined using the

equation: volume = width 9 width 9 length 9 0.52. The

animals were euthanized on the end of the experiment as

indicated according to a University of Pittsburgh IACUC

approved protocol. The tumors were retrieved for pathologic

analysis.

Treatment of ID8 tumor-bearing animals

with TNFSF15 shRNA

Six-week-old female C57BL/6 mice were randomized into

two groups. Animals in the experimental group were

injected subcutaneously on the flank with TNFSF15-

shRNA (1.5 9 105 IFU in 100 ll PBS). The injection sites

were marked with ink. Animals in the control group were

treated with scrambled shRNA in a similar manner. After

3 days, all animals were inoculated subcutaneously with

5 9 106 ID8 cells at the marked sites. The shRNA treat-

ments were repeated on day-6 post-cancer cell inoculation.

The animals were observed, weighed, and the tumor sizes

measured as described above. The animals were euthanized

at the end of the experiment and the tumors were retrieved

for pathologic analysis as described above.

Statistical analysis

At least three repeats were carried for each experiment.

The data were subjected to statistical analysis. The meth-

ods of analysis and P values are indicated in the text.

Results

TNFSF15 gene expression is downmodulated

in ovarian cancer

We determined TNFSF15 protein levels by immunohisto-

chemistry in normal ovary tissues and clinical specimens of

ovarian cancer, using a four-step grading system (-, ?,

??, ??? for negative, low, high, and very high,

respectively). TNFSF15 in normal ovary exhibited vascular

distribution in a pattern similar to that of the endothelial

cell marker CD31 (Fig. 1A). While TNFSF15 expression

was high or very high in most normal ovary specimens, it

was negative or low in the cancer specimens (Table 1).

Remarkably, low TNFSF15 expression (-/?) was found in

nearly all age groups of the patients, and the percentage of

low TNFSF15 expression specimens increased as the

patients aged between 40 and 69 (Table 1). When analyzed

on the basis of Federation of International Gynecology

Organizations (FIGO) stages, TNFSF15 protein levels

decreased markedly in stages I and II, and diminished in

stages III and IV, with the percentages of highly positively

stained specimens (??/???) being 35 and 16%,

respectively (Table 1). TNFSF15 protein levels in ovarian

cancer are thus strongly correlated inversely to the pro-

gression of the disease. No significant differences were

seen among the histological types of the disease or based

on histology staging. Microvessel density (MVD) values

increased significantly as the disease progressed based on

FIGO stages, as expected, with the median MVD values for

normal, stages I/II, and stages III/IV being 8.3, 15.2, and

22.1, respectively. However, of the 84 cancer specimens

that were informative for both TNFSF15 and CD31, MVD

values were significantly higher in low TNFSF15 speci-

mens than in high TNFSF15 specimens, with the median

values being 17.4 and 13.3, respectively (Fig. 1C), indi-

cating that TNFSF15 protein levels are inversely correlated

with the extent of tumor angiogenesis in ovarian cancer.

TNFSF15 downregulation by VEGF

We analyzed VEGF expression in normal ovary and ovarian

cancer specimens by immunohistochemistry. Of the 88

ovarian cancer and 12 normal specimens studied, we found

that, while VEGF staining was basically negative in normal

ovary, an increasing proportion of ovarian cancer cells were

positive for VEGF as the disease progressed (Fig. 2a). About

26% of the cancer cells in stages I and II and as much as 69%

of the cancer cells in stages III and IV were positive for

VEGF (Fig. 2b). Consistently, of the 84 cancer specimens

that were informative for CD31 immunostaining, MVD

values were about 70% higher in cancer specimens that also

exhibited high levels of VEGF (median = 19.8) than that in

those with moderate or no VEGF production (median =

14.1; Fig. 2c). Interestingly, enhanced VEGF production by

cancer cells was accompanied by diminished TNFSF15. Of

the 86 cancer specimens that were informative for both

TNFSF15 and VEGF, the percentage of endothelial cells

with low (-/?) TNFSF15 increased from about 70% ovarian

cancer specimens with moderate or no VEGF staining to

about 90% in specimens with high or very high VEGF levels

(Fig. 2d).

Additionally, we determined the impact of hypoxia on

VEGF production by ovarian cancer cells. OVCAR-3 cells

were cultured under either normoxic (20% O2) or hypoxic

conditions (5% O2). We found that VEGF in the conditioned

media of the cancer cells in hypoxia reached a half-maxi-

mum level in 12 h; in comparison, VEGF production

reached a similar level in about 72 h in cells cultured in

normoxia (Fig. 2e). We then treated HUVEC cultures with

recombinant human VEGF at various concentrations and

found that the treatments substantially inhibited TNFSF15

production (Fig. 2f). To determine whether the inhibition of

74 Angiogenesis (2012) 15:71–85

123

HUVEC production of TNFSF15 by OVCAR-3 cell-condi-

tioned media was specifically caused by VEGF, we treated

HUVEC with OVCAR-3 conditioned media in the presence

or absence of neutralizing antibodies against VEGF, VEG-

FR1, or VEGFR2. We found that ability of OVCAR-3 cell to

inhibit TNFSF15 expression in HUVEC was significantly

hindered by neutralizing antibodies against VEGFR1,

VEGFR2 or VEGF (Fig. 2g). Together these findings

indicate that VEGF produced by the cancer cells is to a large

part responsible for TNFSF15 downregulation.

Correlation between tumor-infiltration of T-cell

and TNFSF15 expression

We determined the correlations between the extent of

T-cell infiltration and TNFSF15 protein levels in clinical

Fig. 1 Relationship between TNFSF15 expression, microvessel den-

sity, and ovarian cancer disease progression in clinical settings.

A Typical images of TNFSF15 and CD31 immunostaining of blood

vessels (brown) of normal ovary and ovarian cancer specimens based

on FIGO staging. Magnification, 9400. B Box plots of MVD (CD31-

positive vessels per 9400 field) of ovarian cancer specimens of

various disease stages. C Box plots of MVD with regard to TNFSF15

protein levels. The number of cases analyzed is indicated for each

group. Horizontal lines indicate median values. **P \ 0.01;

***P \ 0.005; ANOVA (B) or Student t test (C)

Angiogenesis (2012) 15:71–85 75

123

settings. The degree of CD4? T-cell accumulation in

ovarian cancer specimens increased markedly as the dis-

ease progresses based on FIGO staging (Fig. 3A). Of the

86 cancer specimens we analyzed, 75% of stages III and IV

specimens were strongly CD4-positive (??/???),

whereas approximately 41% of FIGO stages I and II

samples were strongly CD4-positive (Fig. 3B). We then

analyzed the relationship between TNFSF15 expression

and T-cell infiltration. Of the specimens in which T-cell

infiltration was negative or marginal (CD4 -/?), 64% was

TNFSF15-negative, whereas of the specimens with high

degree of T-cell infiltration (CD4 ??/???), 83% were

TNFSF15-negative (Fig. 3C). We further determined the

identity of the tumor-infiltrating T-cells by carrying out

CD4-CD25-FOXP3 triple immunostaining of 15 randomly

selected cancer specimens of stage III and IV (Fig. 3D).

We found that only 20% of these specimens were TReg

negative (the ratio of CD4?CD25?FOXP3? cell/CD4? cell

was less than 10%; Fig. 3E). Among the 12 TReg positive

specimens, 30–60% (median = 45%) of the CD4-positive

cells were also CD25?FOXP3? (Fig. 3E). Further analysis

of these specimens indicated that about 95% of the

CD4?CD25? cells were CD4?CD25?FOXP3? (Fig. 3E).

These findings strongly suggest that TNFSF15 protein

levels are inversely correlated with the degree of T-cell

infiltration, especially that of a population of CD4?CD25?

FOXP3? TReg cells, in ovarian cancer with stages III

and IV.

TNFSF15 downregulation by MCP-1

Since a large number of the tumor-infiltrating T-cells in the

stages III&IV ovarian cancer specimens were CD4?CD25?

FOXP3? TReg cells, we determined what cytokine(s) pro-

duced by these cells was responsible for the downregula-

tion of TNFSF15. As more than 95% of tumor-infiltrating

CD4?CD25? cells were CD4?CD25?FOXP3? TReg cells

in the detected specimens, we first treated freshly isolated

CD4?CD25? T cells with OVCAR3 cell conditioned

media. We then treated HUVEC cultures with conditioned

media prepared from these CD4?CD25? T cell cultures.

We found that treatments of HUVEC with the T cell

conditioned media that were influenced by OVACR-3 cell

led to a decrease of TNFSF15 in HUVEC culture media by

about 40% (Fig. 4a). To identify any changes in cytokine

production pattern by CD4?CD25? T cells under the

influence of ovarian cancer cells, we carried out microarray

analysis of a panel of 60 cytokines known to be produced

by regulatory T cells (Figure S1). The cytokine profiling

results indicated that, among the 60 cytokines analyzed,

MCP-1 distinguished itself because its production by

CD4?CD25? T cells increased by as much as 4-fold when

these cells were treated with OVACR-3 conditioned media,

compared with that by untreated CD4?CD25? T cells

(Fig. 4b, c; data regarding the production of all cytokines

on the panel is given in supplemental Figure S1). Sus-

pecting that MCP-1 was responsible for downregulating

TNFSF15, we treated HUVEC cultures with recombinant

human MCP-1. We found that MCP-1 treatments led to a

significant decrease of TNFSF15 production by HUVEC in

a dose-dependent manner (Fig. 4d). A neutralizing anti-

body against MCP-1 specifically blocked the inhibitory

effect (Fig. 4e). MCP-1 treatment of HUVEC culture

did not result in discernible changes in the number of

HUVEC cells (Fig. 4f). These findings indicate that MCP-1

produced by tumor-infiltrating lymphocytes in ovarian

cancer is, besides VEGF, responsible for TNFSF15

downregulation.

Correlation between macrophage infiltration

and TNFSF15 expression

Since tumor-associated macrophages are a major source of

MCP-1 [49], we determined the extent of macrophage

infiltration by immunohistochemistry in normal ovary tis-

sue (n = 10) and ovarian cancer specimens (n = 69), using

macrophage marker MAC-1 (Fig. 5A). We also analyzed

Table 1 TNFSF15 protein levels in normal ovary and ovarian cancer

Number of

cases

%a P valueb

-/? ??/???

Normal ovary 12 16.7 (2) 83.3 (10)

Ovary cancer 94 73.4 (69) 26.6 (25) \0.001

Age

B39 9 66.7 (6) 33.3 (3) 0.06

40–49 28 67.9 (19) 32.1 (9) 0.003

50–59 33 75.8 (25) 24.2 (8) 0.01

60–69 13 92.3 (12) 7.7 (1) \0.001

C70 11 63.6 (7) 36.4 (4) 0.06

FIGO stage

I&II 51 64.7 (33) 35.3 (18)

III&IV 43 83.7 (36) 16.3 (7) 0.038c

Histological type

Serous 58 79.3 (46) 20.7 (12)

Mucinous 8 50.0 (4) 50.0 (4) 0.17c

Others 19 63.2 (12) 36.8 (7) 0.267c

Histology staged

Middle-high 63 74.6 (47) 25.4 (16)

Low 22 77.3 (17) 22.7 (5) 0.803c

a Numbers in parentheses are the number of casesb Compared with normal ovary unless otherwise indicated; Chi-

square testc Compared between the sub-categories; Chi-square testd Histology staging method was not applied to all cases studied

76 Angiogenesis (2012) 15:71–85

123

Fig. 2 VEGF produced by human ovarian cancer cells inhibits

TNFSF15 production by HUVEC. A Typical images of VEGF

immunostaining (brown) of normal ovary or ovarian cancer speci-

mens grouped on the basis of FIGO staging. B Percentage of VEGF

positive cells in ovarian cancer specimens grouped on the basis of

FIGO staging. Black -/?. White ??/???. C Box plots of MVD of

ovarian cancer specimens grouped on the basis of low (-/?) or high

(??/???) VEGF levels. D Percentage of TNFSF15 positive cells in

cancer specimens grouped on the basis of low (-/?) or high (??/

???) VEGF levels. E VEGF concentrations in conditioned media of

HUVEC cultured under either normoxic (White) or hypoxic (Black)

conditions. F TNFSF15 concentrations determined at different time

intervals in conditioned media of HUVEC cultures treated with VEGF

at indicated concentrations. White bars 24 h. Gray bars 48 h. Blackbars 72 h. G Concentrations of TNFSF15 in conditioned media of

HUVEC cultures treated with OVCAR3 conditioned media in the

absence or presence of various neutralizing antibodies as indicated.

All cell cultures were in triplicate for each experimental condition.

Each experiment was repeated two times. *P \ 0.05; **P \ 0.01;

***P \ 0.005, Chi-square (B, D); Student t test (C, E, F, G)

Angiogenesis (2012) 15:71–85 77

123

MVD in the cancer specimens by staining for CD31

(n = 67). We found that blood vessel density was posi-

tively correlated with the degree of macrophage infiltration

in ovarian cancer (Fig. 5B). We then determined the rela-

tionship between TNFSF15 expression and macrophage

infiltration. We found that, of the cancer specimens in

which macrophage infiltration was negative or marginal

(MAC-1 -/?), 66% were TNFSF15-negative, whereas of

the samples that exhibited high degree of macrophage

infiltration (Mac-1 ??/???), 90% were TNFSF15-neg-

ative (Fig. 5C). These findings indicate that TNFSF15

expression levels are inversely correlated with the degree

of macrophage inflammation in ovarian cancer. To deter-

mine potential influence of ovarian cancer cell on macro-

phage production of MCP-1, we isolated monocytes from

peripheral blood, placed them in cultures, then treated the

cell cultures with OVCAR3 conditioned media and deter-

mined MCP-1 concentration in the culture media

Fig. 3 Tumor infiltration of CD4?CD25? FOXP3? TReg cell in

human ovarian cancer. A Typical images of CD4 immunostaining of

normal ovary and ovarian cancer specimens based on FIGO staging;

magnification, 9400. B Percentage of CD4 positive cells in speci-

mens based on FIGO staging. Black low percentage of CD4? cells

(-/?). White high percentage of CD4 positive cells (??/???).

C Percentage of TNFSF15 positive cells grouped on the basis of the

extent of CD4? T cell infiltration. Black low percentage of TNFSF15

positive cells (-/?). White high percentage of TNFSF15 positive

cells (??/???). D Typical images of CD4-CD25-FOXP3 triple

immunostaining of cancer specimens. Green CD4? (a). Red CD25?

(b). Blue FOXP3? (c). Merged, CD4?CD25?FOXP3? (d). Magnifi-

cation, 9400. E Percentage of TReg-negative and TReg-positive

specimens in ovarian cancer of stages III&IV. Numbers on top ofboxes are the number of cases determined. *P \ 0.05; ***P \ 0.005,

Chi-square (B, C); Student t test (E)

78 Angiogenesis (2012) 15:71–85

123

(Fig. 5D). MCP-1 production by freshly isolated mono-

cytes was negligible. OVCAR3 produced a readily

detectable amount of MCP-1. Remarkably, when treated

with OVCAR3 conditioned media, MCP-1 production by

the monocytes sharply increased by approximately 5-fold

within 72 h. This suggests that, once infiltrated into ovarian

cancer tissues, macrophages could produce a significant

amount of MCP-1 under the influence of the cancer cells.

Changing TNFSF15 levels in tumor microenvironment

markedly impacts on tumor angiogenesis and growth

We used a murine ovarian cancer cell-line ID8 syngeneic

tumor model (44) to determine the effect of changing

TNFSF15 levels in cancer cell microenvironment on

angiogenesis and tumor growth. In order to mimic the high

level of VEGF we observed in ovarian cancer tissues under

Fig. 4 Inhibition of HUVEC

production of TNFSF15 by

MCP-1. A TNFSF15

concentrations in culture media

of HUVEC treated with various

conditioned media (CM) as

indicated. B Protein array

analysis of cytokines secreted

by freshly isolated human

CD4?CD25? T cells treated

with indicated conditioned

media (CM), using a panel of 60

cytokines (refer to

Supplemental Figure S1 for

detail). Circled MCP-1.

C Quantitative analysis of

MCP-1 production shown in

B. RFU, relative fluorescence

unit. D TNFSF15

concentrations in culture media

of HUVEC treated with

recombinant MCP-1.

E TNFSF15 concentrations in

culture media of HUVEC with

recombinant MCP-1 in the

presence or absence of a

neutralizing antibody against

MCP-1. F Analysis of the

number of HUVEC in cultures

treated with recombinant MCP-

1 at indicated concentrations.

All cell cultures were in

triplicate for each experimental

condition. Each experiment was

repeated two times. *P \ 0.05,

Student t test

Angiogenesis (2012) 15:71–85 79

123

clinical conditions, we treated the ID8 cells with insulin

prior to implantation as insulin was known to stimulate

VEGF production in these cells [48]. We first examined the

effect of raising TNFSF15 levels. ID8 cells were subcuta-

neously implanted (5 9 106 cells per injection) on C57BL/

6 mice. The animals were treated by intraperitoneal

injection of recombinant human TNFSF15 (10 mg/Kg;

control group was treated with vehicle) on day 6 post-

inoculation when the tumors were palpable, then repeated

daily. The treatments markedly inhibited the growth of ID8

tumors (Fig. 6A). The tumor volumes of recombinant

TNFSF15-treated group on average were about 40%

smaller than those of the control group.

We then determined the effect of downregulating

TNFSF15 in cancer cell microenvironment on tumor

growth. We subcutaneously injected TNFSF15 shRNA

(1.5 9 105 IFU per injection) 72 h prior the implantation

of ID8 cells on the same site, then repeated shRNA

injection on day 6 post cancer cell inoculation. Monitoring

tumor growth rate, we found that the cancer cells implanted

at TNFSF15 shRNA-treated sites exhibited significantly

facilitated ability to form tumors as compared with the

cells injected at control shRNA-treated sites (Fig. 6B). The

tumor volumes on average were approximately 53% larger

in TNFSF15 shRNA-treated group than those in control

group. Additionally, we determined TNFSF15 protein

levels in the skin of shRNA-treated sites by immunohis-

tochemical analysis. We found that TNFSF15 shRNA-

treatments led to nearly complete elimination of TNFSF15

in the blood vessels (Fig. 6C, a, b). In sharp contrast, the

number of blood vessels identified by CD31-positive

staining increased markedly (Fig. 6C, c, d). Quantitative

analysis of the skin sections revealed that the median MVD

value in TNFSF15 shRNA-treated skin increased by nearly

100% compared with that in the control group (Fig. 6C, e).

We compared the extent of angiogenesis in the tumors

by carrying out CD31 immunostaining of tumor sections

from vehicle- or recombinant TNFSF15-treated animals

Fig. 5 Relationship between

macrophage infiltration, MVD,

and TNFSF15 protein levels in

human ovarian cancer.

A Typical images of tumor-

associated macrophages (BrownMAC-1 positive cells).

Magnification, 9400. B Box

plots of MVD of cancer

specimens grouped on the basis

of low (-/?) or high (??/

???) degree of macrophage

infiltration. The number of cases

in each group is indicated.

C Percentage of TNFSF15

positive cells in cancer

specimens grouped on the basis

of the degree of macrophage

infiltration. White TNFSF15

positive. Black TNFSF15

negative. The number of

informative specimens in each

group is indicated. D MCP-1

concentrations in the culture

media of freshly isolated

monocytes untreated or treated

with OVCAR3 conditioned

media, and OVCAR3

conditioned media alone.

*P \ 0.05; **P \ 0.01;

***P \ 0.005, Chi-square (C);

Student t test (B, D)

80 Angiogenesis (2012) 15:71–85

123

(Fig. 6D, a, b). The median MVD value in tumors of

recombinant TNFSF15-teated group decreased by 75%

compared with that of vehicle-treated group (Fig. 6D, c).

On the other hand, CD31 immunostaining showed that the

degree of angiogenesis was remarkably higher in tumor

sections of TNFSF15 shRNA-treated group than that in the

control group (Fig. 6E, a, b). The median MVD value of

TNFSF15 shRNA-treated group increased by approxi-

mately 120% compared with that of control shRNA-treated

group. Together these findings indicate that increasing

TNFSF15 levels in the cancer cell microenvironment

results in significant inhibition of tumor angiogenesis and

growth, whereas decreasing TNFSF15 levels leads to

greatly facilitated tumor angiogenesis and growth.

Discussion

It is well known that the process of malignant tumor

development resembles abnormal wound healing charac-

terized by uncontrolled inflammation and neovasculariza-

tion driven by highly upregulated pro-angiogenesis growth

factors and cytokines. The initiation of angiogenesis,

however, requires a concomitant downregulation of the

activities of endogenous inhibitors of angiogenesis, as

these inhibitors normally play a critical role to maintain the

stability of an established vasculature. In this study we

focus on how the role of a key endogenous suppressor of

angiogenesis is removed from the normal control mecha-

nism. TNFSF15 is a unique cytokine that functions as an

endogenous inhibitor of neovascularization to maintain

vascular homeostasis [36, 38–41]. It is produced largely by

endothelial cells in an established vasculature and inhibits

the proliferation of endothelial cells themselves. We

hypothesize that downmodulation of TNFSF15 is a pre-

requisite for the initiation of neovascularization such as in

cancers. We determined whether TNFSF15 is downmod-

ulated in ovarian cancer under clinical conditions, what

factors are responsible for the action, and what are the

sources of these factors.

Our data demonstrate in clinical settings that TNFSF15

is present at high levels in the vasculature of normal ovary

but declines sharply at early stages of ovarian cancer

development, and nearly completely diminishes in the

tumor tissues as the disease progresses to later stages. We

found that the downmodulation of TNFSF15 is attributable

to a significant extent to VEGF produced by the cancer

cells. The human female reproductive tract is highly

dependent on VEGF for normal functions such as endo-

metrial proliferation and corpus luteum development [27].

The unique influence of female sex hormones on the

expression and activity of VEGF deems angiogenesis an

important facet of the development of ovarian cancer.

VEGF is highly upregulated under periodical hypoxic

conditions in a malignant tumor. By increasing vascular

permeability in endothelial cells, it facilitates the recruit-

ment of immune cells, mostly noticeably macrophages and

T cells, to the immediate surroundings of the cancer cells,

making tumor formation a process of abnormal wound

healing characterized by uncontrolled inflammation and

angiogenesis. Showing a strong, inverse correlation

between VEGF and TNFSF15 expression in ovarian cancer

in clinical settings, sharp upregulation of ovarian cancer

cell production of VEGF under hypoxic conditions, and

effective downregulation of endothelial cell production of

TNFSF15 by VEGF, our data indicate that TNFSF15 as an

important endogenous inhibitor of angiogenesis is targeted

by VEGF for downmodulation, highly likely as an initial

step of the process of neovascularization.

Our data also reveal a strong, inverse correlation

between the extent of TReg cells and macrophage infiltra-

tion and TNFSF15 protein levels. Additionally, we show

that TNFSF15 expression in human endothelial cells in

culture can be downmodulated by MCP-1 produced by

TReg cell and monocyte (a circulating counterpart of mac-

rophage) under the influence of ovarian cancer cell.

Importantly, the production of MCP-1 by freshly isolated

TReg cells and monocytes is initially insignificant, but is

highly upregulated when treated with ovarian cancer cell

conditioned media. It has been reported MCP-1 under

certain conditions can be produced by a variety of cell

types, including endothelial, fibroblast, epithelial, smooth

muscle, mesangial, astrocytic, monocytic, and microglial

cells, either constitutively or after induction by oxidative

stress, cytokines, or growth factors [18, 20, 21, 50]. MCP-1

has been reported to have an important role in various

disease states, including immunodeficiency, cardiovascu-

lar, and cancer [51–54]. Monocytes and macrophages are

the major source of MCP-1, however, in many cancers,

including ovarian cancer [14–17]. The role of MCP-1 in

cancers appears ambiguous. It stimulates host antitumor

responses by attracting and activating lymphocytes [55],

and is attributed to cancer progression because of its

angiogenic activities [56]. We show that, promoted by

ovarian cancer cells, monocytes and TReg cells produce

high level of MCP-1, and that MCP-1 directly downregu-

lates TNFSF15 expression in endothelial cell. The ability to

downregulate an endogenous suppressor of angiogenesis

suggests that MCP-1 is capable of shifting the equilibrium

between pro- and anti-angiogenesis forces in cancer

microenvironment in favor of inflammation and neovas-

cularization, making MCP-1 a more potent promoter of

cancer progression than simply driving endothelial cell

migration. Further investigation on TNFSF15 downmodu-

lation by tumor infiltrating macrophages and TReg cells is

likely to reveal critical insights into the loss of negative

Angiogenesis (2012) 15:71–85 81

123

control mechanisms in cancers which otherwise would

function to suppress inflammation and angiogenesis.

We used a mouse model of ovarian cancer to determine

the impact of raising or eliminating TNFSF15 levels on

tumor angiogenesis. The facts that cancer cells and tumor-

infiltrating immune cells, most prominently macrophages

and TReg cells, cooperate to downmodulate TNFSF15

suggest that the downmodulation of TNFSF15 is a

82 Angiogenesis (2012) 15:71–85

123

prerequisite for the initiation of angiogenesis in tumors.

ID8 is a mouse ovarian cancer cell-line derived from

spontaneous malignant transformation in vitro of C57BL/6

mouse ovarian surface epithelial (MOSE) cell, an equiva-

lent of human epithelial ovarian cancer (EOC) cell

[57–59]. Previously we showed that TNFSF15, delivered

either by various methods of gene-transfer or systemic

administration of a recombinant protein, is able to inhibit

tumor growth in models of breast, prostate, colon cancers

[36, 37, 41]. Our study with the ID8 ovarian cancer model

demonstrates the importance of TNFSF15 as a negative

modulator of to guard against the initiation of angiogenesis.

Our data indicate that systemic application of recombinant

TNFSF15 inhibits angiogenesis and tumor growth, whereas

silencing TNFSF15 with shRNA topically prior to cancer

cell inoculation greatly facilitates angiogenesis and tumor

growth. A well-controlled angiogenesis cycle under phys-

iological conditions is characteristic of normal ovaries,

synchronized with ovulation and corpus luteum formation.

Up- and downregulation of VEGF and other angiogenesis-

promoting factors are implicated in this cycle. However,

insights into the down- and upregulation of endogenous

suppressors of angiogenesis in this cycle are largely

missing. Our findings indicate that the function of

TNFSF15 is likely to be important in the control of angi-

ogenesis cycle in ovary, and that the loss of this negative

control mechanism could be a critical attribute to ovarian

cancer development. It is interesting to notice in this regard

that we found a significant decline of TNFSF15 in ovarian

cancer occurs with patients in age group 40–49 and persists

in older age groups. This is consistent with the observation

that ovarian cancer incidence in Asian populations rises

sharply in age 40–59 and plateaus in older ages [60].

In summary, our data are consistent with the view that

downmodulation of TNFSF15 as an endogenous suppressor

of angiogenesis is a prerequisite for neovascularization in

ovarian cancer. This is achieved by the inhibition of

TNFSF15 expression by VEGF and MCP-1 produced,

respectively, by the cancer cells and tumor-infiltrating

macrophages and TReg cells under the influence of the

cancer cells. Keeping an elevated TNFSF15 level would

thus allow the maintenance of a normal vasculature and

prevention of uncontrolled inflammation and angiogenesis

common to malignant cancers.

Acknowledgments This study is supported in part by grants from

Ministry of Science and Technology of China (2009CB918901 to

L.Y.L), National Institute of Health of the United States (R01CA11

3875 to L.Y.L), and Natural Science Foundation of China (30670801

to W.M.D).

References

1. Rosmorduc O, Housset C (2010) Hypoxia: a link between

fibrogenesis, angiogenesis, and carcinogenesis in liver disease.

Semin Liver Dis 30(3):258–270. doi:10.1055/s-0030-1255355

2. Aggarwal BB, Shishodia S, Sandur SK, Pandey MK, Sethi G

(2006) Inflammation and cancer: how hot is the link? Biochem

Pharmacol 72(11):1605–1621. doi:10.1016/j.bcp.2006.06.029

3. Coussens LM, Werb Z (2002) Inflammation and cancer. Nature

420(6917):860–867. doi:10.1038/nature01322

4. Fiedler U, Augustin HG (2006) Angiopoietins: a link between

angiogenesis and inflammation. Trends Immunol 27(12):552–558.

doi:10.1016/j.it.2006.10.004

5. Coulon S, Heindryckx F, Geerts A, Van Steenkiste C, Colle I, Van

Vlierberghe H (2011) Angiogenesis in chronic liver disease and its

complications. Liver Int 31(2):146–162. doi:10.1111/j.1478-

3231.2010.02369.x

6. Balkwill F, Mantovani A (2001) Inflammation and cancer: back to

Virchow? Lancet 357(9255):539–545. doi:10.1016/S0140-6736

(00)04046-0

7. Folkman J (1995) Angiogenesis in cancer, vascular, rheumatoid

and other disease. Nat Med 1(1):27–31

8. Casanovas O, Hicklin DJ, Bergers G, Hanahan D (2005) Drug

resistance by evasion of antiangiogenic targeting of VEGF sig-

naling in late-stage pancreatic islet tumors. Cancer Cell 8(4):

299–309. doi:10.1016/j.ccr.2005.09.005

9. Dvorak HF, Detmar M, Claffey KP, Nagy JA, van de Water L,

Senger DR (1995) Vascular permeability factor/vascular endo-

thelial growth factor: an important mediator of angiogenesis in

malignancy and inflammation. Int Arch Allergy Immunol

107(1–3):233–235

10. Jeon BH, Jang C, Han J, Kataru RP, Piao L, Jung K, Cha HJ,

Schwendener RA, Jang KY, Kim KS, Alitalo K, Koh GY (2008)

Profound but dysfunctional lymphangiogenesis via vascular

endothelial growth factor ligands from CD11b? macrophages in

advanced ovarian cancer. Cancer Res 68(4):1100–1109. doi:

10.1158/0008-5472.CAN-07-2572

11. Duyndam MC, Hilhorst MC, Schluper HM, Verheul HM, van

Diest PJ, Kraal G, Pinedo HM, Boven E (2002) Vascular endo-

thelial growth factor-165 overexpression stimulates angiogenesis

and induces cyst formation and macrophage infiltration in human

ovarian cancer xenografts. Am J Pathol 160(2):537–548

Fig. 6 Impact of changing TNFSF15 levels in cancer microenviron-

ment on mouse ovarian cancer ID8 syngeneic tumor angiogenesis and

growth. A Growth rates of ID8 tumors on animals treated with

intraperitoneal administration of recombinant TNFSF15 or vehicle.

Closed circles vehicle treated. Open circles TNFSF15 treated. Values

are mean ± SD (number of animal per group, n = 5). B Growth rates

of tumors on animals treated with topical injection of TNFSF15

shRNA or control shRNA. Black bars control shRNA treated. Whitebars TNFSF15 shRNA treated. Values are mean ± SD (number of

animals per group, n = 5). C (a–d) Typical images of TNFSF15 or

CD31 immunostained skin sections collected from the sites of topical

TNFSF15 shRNA or control shRNA treatments, as indicated; (e) Box

plots of MVD values of tumors treated with control or TNFSF15

shRNA, as indicated (number of animals per group, n = 5). D Typical

images of CD31 immunostaining of tumor sections in (a) vehicle

treated or (b) recombinant TNFSF15 treated groups; (c) Box plots of

MVD values of vehicle or TNFSF15 treated tumors (number of

animals per group, n = 5). E Typical images of CD31 immunostain-

ing of tumor sections in (a) control shRNA treated or (b) TNFSF15

shRNA treated groups; (c) Box plots of MVD values of control

shRNA or THFSF15 shRNA treated tumors (number of animals per

group, n = 5). Horizontal bars indicate median values. The exper-

iment was repeated once and the results were reproducible.

*P \ 0.05; **P \ 0.01; ***P \ 0.005, Student t test

b

Angiogenesis (2012) 15:71–85 83

123

12. Ueno T, Toi M, Saji H, Muta M, Bando H, Kuroi K, Koike M,

Inadera H, Matsushima K (2000) Significance of macrophage

chemoattractant protein-1 in macrophage recruitment, angiogen-

esis, and survival in human breast cancer. Clin Cancer Res 6(8):

3282–3289

13. Jin G, Kawsar HI, Hirsch SA, Zeng C, Jia X, Feng Z, Ghosh SK,

Zheng QY, Zhou A, McIntyre TM, Weinberg A (2010) An

antimicrobial peptide regulates tumor-associated macrophage

trafficking via the chemokine receptor CCR2, a model for

tumorigenesis. PLoS One 5(6):e10993. doi:10.1371/journal.

pone.0010993

14. Boelte KC, Gordy LE, Joyce S, Thompson MA, Yang L, Lin PC

(2011) Rgs2 mediates pro-angiogenic function of myeloid derived

suppressor cells in the tumor microenvironment via upregulation of

MCP-1. PLoS One 6(4):e18534. doi:10.1371/journal.pone.0018534

15. Saenz-Lopez P, Carretero R, Cozar JM, Romero JM, Canton J,

Vilchez JR, Tallada M, Garrido F, Ruiz-Cabello F (2008) Genetic

polymorphisms of RANTES, IL1-A, MCP-1 and TNF-A genes in

patients with prostate cancer. BMC Cancer 8:382. doi:10.1186/

1471-2407-8-382

16. Monti P, Leone BE, Marchesi F, Balzano G, Zerbi A, Scaltrini F,

Pasquali C, Calori G, Pessi F, Sperti C, Di Carlo V, Allavena P,

Piemonti L (2003) The CC chemokine MCP-1/CCL2 in pancreatic

cancer progression: regulation of expression and potential mech-

anisms of antimalignant activity. Cancer Res 63(21):7451–7461

17. Negus RP, Stamp GW, Relf MG, Burke F, Malik ST, Bernasconi S,

Allavena P, Sozzani S, Mantovani A, Balkwill FR (1995) The

detection and localization of monocyte chemoattractant protein-1

(MCP-1) in human ovarian cancer. J Clin Invest 95(5):2391–2396.

doi:10.1172/JCI117933

18. Cushing SD, Berliner JA, Valente AJ, Territo MC, Navab M,

Parhami F, Gerrity R, Schwartz CJ, Fogelman AM (1990) Min-

imally modified low density lipoprotein induces monocyte che-

motactic protein 1 in human endothelial cells and smooth muscle

cells. Proc Natl Acad Sci USA 87(13):5134–5138

19. Standiford TJ, Kunkel SL, Phan SH, Rollins BJ, Strieter RM

(1991) Alveolar macrophage-derived cytokines induce monocyte

chemoattractant protein-1 expression from human pulmonary

type II-like epithelial cells. J Biol Chem 266(15):9912–9918

20. Brown Z, Strieter RM, Neild GH, Thompson RC, Kunkel SL,

Westwick J (1992) IL-1 receptor antagonist inhibits monocyte

chemotactic peptide 1 generation by human mesangial cells.

Kidney Int 42(1):95–101

21. Barna BP, Pettay J, Barnett GH, Zhou P, Iwasaki K, Estes ML

(1994) Regulation of monocyte chemoattractant protein-1

expression in adult human non-neoplastic astrocytes is sensitive

to tumor necrosis factor (TNF) or antibody to the 55-kDa TNF

receptor. J Neuroimmunol 50(1):101–107

22. Dvorak HF (1986) Tumors: wounds that do not heal. Similarities

between tumor stroma generation and wound healing. N Engl J

Med 315(26):1650–1659. doi:10.1056/NEJM198612253152606

23. Folkman J, Watson K, Ingber D, Hanahan D (1989) Induction of

angiogenesis during the transition from hyperplasia to neoplasia.

Nature 339(6219):58–61. doi:10.1038/339058a0

24. Fraser HM, Bell J, Wilson H, Taylor PD, Morgan K, Anderson

RA, Duncan WC (2005) Localization and quantification of cyclic

changes in the expression of endocrine gland vascular endothelial

growth factor in the human corpus luteum. J Clin Endocrinol

Metab 90(1):427–434. doi:10.1210/jc.2004-0843

25. Davis JS, Rueda BR, Spanel-Borowski K (2003) Microvascular

endothelial cells of the corpus luteum. Reprod Biol Endocrinol

1:89. doi:10.1186/1477-7827-1-89

26. Fraser HM, Wulff C (2001) Angiogenesis in the primate ovary.

Reprod Fertil Dev 13(7–8):557–566

27. Delli Carpini J, Karam AK, Montgomery L (2010) Vascular

endothelial growth factor and its relationship to the prognosis and

treatment of breast, ovarian, and cervical cancer. Angiogenesis

13(1):43–58. doi:10.1007/s10456-010-9163-3

28. Lebovic DI, Shifren JL, Ryan IP, Mueller MD, Korn AP, Darney

PD, Taylor RN (2000) Ovarian steroid and cytokine modulation of

human endometrial angiogenesis. Hum Reprod 15(Suppl 3):67–77

29. Zhang L, Yang N, Conejo-Garcia JR, Katsaros D, Mohamed-

Hadley A, Fracchioli S, Schlienger K, Toll A, Levine B, Rubin

SC, Coukos G (2003) Expression of endocrine gland-derived

vascular endothelial growth factor in ovarian carcinoma. Clin

Cancer Res 9(1):264–272

30. Boocock CA, Charnock-Jones DS, Sharkey AM, McLaren J, Barker

PJ, Wright KA, Twentyman PR, Smith SK (1995) Expression of

vascular endothelial growth factor and its receptors fit and KDR in

ovarian carcinoma. J Natl Cancer Inst 87(7):506–516

31. Orre M, Rogers PA (1999) VEGF, VEGFR-1, VEGFR-2,

microvessel density and endothelial cell proliferation in tumours

of the ovary. Int J Cancer 84(2):101–108. doi:10.1002/(SICI)1097-

0215(19990420)84:2\101:AID-IJC2[3.0.CO;2-5

32. Hazelton D, Nicosia RF, Nicosia SV (1999) Vascular endothelial

growth factor levels in ovarian cyst fluid correlate with malig-

nancy. Clin Cancer Res 5(4):823–829

33. Kuwahara K, Sasaki T, Kobayashi K, Noma B, Serikawa M,

Iiboshi T, Miyata H, Kuwada Y, Murakami M, Yamasaki S,

Kariya K, Morinaka K, Chayama K (2004) Gemcitabine sup-

presses malignant ascites of human pancreatic cancer: correlation

with VEGF expression in ascites. Oncol Rep 11(1):73–80

34. Bolat F, Gumurdulu D, Erkanli S, Kayaselcuk F, Zeren H, Ali

Vardar M, Kuscu E (2008) Maspin overexpression correlates with

increased expression of vascular endothelial growth factors A, C,

and D in human ovarian carcinoma. Pathol Res Pract 204(6):

379–387. doi:10.1016/j.prp.2008.01.011

35. Juric G, Zarkovic N, Nola M, Tillian M, Jukic S (2001) The value

of cell proliferation and angiogenesis in the prognostic assess-

ment of ovarian granulosa cell tumors. Tumori 87(1):47–53

36. Zhai Y, Ni J, Jiang GW, Lu J, Xing L, Lincoln C, Carter KC,

Janat F, Kozak D, Xu S, Rojas L, Aggarwal BB, Ruben S, Li LY,

Gentz R, Yu GL (1999) VEGI, a novel cytokine of the tumor

necrosis factor family, is an angiogenesis inhibitor that sup-

presses the growth of colon carcinomas in vivo. FASEB J

13(1):181–189

37. Zhai Y, Yu J, Iruela-Arispe L, Huang WQ, Wang Z, Hayes AJ,

Lu J, Jiang G, Rojas L, Lippman ME, Ni J, Yu GL, Li LY (1999)

Inhibition of angiogenesis and breast cancer xenograft tumor

growth by VEGI, a novel cytokine of the TNF superfamily. Int J

Cancer 82(1):131–136. doi:10.1002/(SICI)1097-0215(19990702)

82:1\131:AID-IJC22[3.0.CO;2-O

38. Chew LJ, Pan H, Yu J, Tian S, Huang WQ, Zhang JY, Pang S, Li

LY (2002) A novel secreted splice variant of vascular endothelial

cell growth inhibitor. FASEB J 16(7):742–744. doi:10.1096/

fj.01-0757fje

39. Yu J, Tian S, Metheny-Barlow L, Chew LJ, Hayes AJ, Pan H, Yu

GL, Li LY (2001) Modulation of endothelial cell growth arrest

and apoptosis by vascular endothelial growth inhibitor. Circ Res

89(12):1161–1167

40. Tian F, Liang PH, Li LY (2009) Inhibition of endothelial pro-

genitor cell differentiation by VEGI. Blood 113(21):5352–5360.

doi:10.1182/blood-2008-08-173773

41. Hou W, Medynski D, Wu S, Lin X, Li LY (2005) VEGI-192, a

new isoform of TNFSF15, specifically eliminates tumor vascular

endothelial cells and suppresses tumor growth. Clin Cancer Res

11(15):5595–5602. doi:10.1158/1078-0432.CCR-05-0384

42. Liang PH, Tian F, Lu Y, Duan B, Stolz DB, Li LY (2011)

Vascular endothelial growth inhibitor (VEGI; TNFSF15) inhibits

bone marrow-derived endothelial progenitor cell incorporation

into Lewis lung carcinoma tumors. Angiogenesis 14(1):61–68.

doi:10.1007/s10456-010-9195-8

84 Angiogenesis (2012) 15:71–85

123

43. Parr C, Gan CH, Watkins G, Jiang WG (2006) Reduced vascular

endothelial growth inhibitor (VEGI) expression is associated with

poor prognosis in breast cancer patients. Angiogenesis 9(2):

73–81. doi:10.1007/s10456-006-9033-1

44. Zhou J, Yang Z, Tsuji T, Gong J, Xie J, Chen C, Li W, Amar S,

Luo Z (2011) LITAF and TNFSF15, two downstream targets of

AMPK, exert inhibitory effects on tumor growth. Oncogene

30(16):1892–1900. doi:10.1038/onc.2010.575

45. Zhang N, Sanders AJ, Ye L, Jiang WG (2009) Vascular endo-

thelial growth inhibitor in human cancer (review). Int J Mol Med

24(1):3–8

46. Zhang N, Sanders AJ, Ye L, Kynaston HG, Jiang WG (2010)

Expression of vascular endothelial growth inhibitor (VEGI) in

human urothelial cancer of the bladder and its effects on the

adhesion and migration of bladder cancer cells in vitro. Anti-

cancer Res 30(1):87–95

47. Conway KP, Price P, Harding KG, Jiang WG (2007) The role of

vascular endothelial growth inhibitor in wound healing. Int

Wound J 4(1):55–64. doi:10.1111/j.1742-481X.2006.00295.x

48. Bermont L, Lamielle F, Lorchel F, Fauconnet S, Esumi H, Weisz

A, Adessi GL (2001) Insulin up-regulates vascular endothelial

growth factor and stabilizes its messengers in endometrial ade-

nocarcinoma cells. J Clin Endocrinol Metab 86(1):363–368

49. Yoshimura T, Robinson EA, Tanaka S, Appella E, Leonard EJ

(1989) Purification and amino acid analysis of two human

monocyte chemoattractants produced by phytohemagglutinin-

stimulated human blood mononuclear leukocytes. J Immunol

142(6):1956–1962

50. Standiford TJ, Rolfe MR, Kunkel SL, Lynch JP III, Becker FS,

Orringer MB, Phan S, Strieter RM (1993) Altered production and

regulation of monocyte chemoattractant protein-1 from pulmon-

ary fibroblasts isolated from patients with idiopathic pulmonary

fibrosis. Chest 103(2 Suppl):121S

51. Cinque P, Vago L, Mengozzi M, Torri V, Ceresa D, Vicenzi E,

Transidico P, Vagani A, Sozzani S, Mantovani A, Lazzarin A,

Poli G (1998) Elevated cerebrospinal fluid levels of monocyte

chemotactic protein-1 correlate with HIV-1 encephalitis and local

viral replication. AIDS 12(11):1327–1332

52. Marini E, Tiberio L, Caracciolo S, Tosti G, Guzman CA, Schi-

affonati L, Fiorentini S, Caruso A (2008) HIV-1 matrix protein

p17 binds to monocytes and selectively stimulates MCP-1

secretion: role of transcriptional factor AP-1. Cell Microbiol

10(3):655–666. doi:10.1111/j.1462-5822.2007.01073.x

53. Krishnaswamy G, Smith JK, Mukkamala R, Hall K, Joyner W,

Yerra L, Chi DS (1998) Multifunctional cytokine expression by

human coronary endothelium and regulation by monokines and

glucocorticoids. Microvasc Res 55(3):189–200

54. Salcedo R, Ponce ML, Young HA, Wasserman K, Ward JM,

Kleinman HK, Oppenheim JJ, Murphy WJ (2000) Human

endothelial cells express CCR2 and respond to MCP-1: direct role

of MCP-1 in angiogenesis and tumor progression. Blood 96(1):

34–40

55. Krensky AM, Clayberger C (2009) Biology and clinical rele-

vance of granulysin. Tissue Antigens 73(3):193–198. doi:

10.1111/j.1399-0039.2008.01218.x

56. Soria G, Ben-Baruch A (2008) The inflammatory chemokines

CCL2 and CCL5 in breast cancer. Cancer Lett 267(2):271–285.

doi:10.1016/j.canlet.2008.03.018

57. Janat-Amsbury MM, Yockman JW, Anderson ML, Kieback DG,

Kim SW (2006) Comparison of ID8 MOSE and VEGF-modified

ID8 cell lines in an immunocompetent animal model for human

ovarian cancer. Anticancer Res 26(4B):2785–2789

58. Holtz DO, Krafty RT, Mohamed-Hadley A, Zhang L, Ala-

gkiozidis I, Leiby B, Guo W, Gimotty PA, Coukos G (2008)

Should tumor VEGF expression influence decisions on combin-

ing low-dose chemotherapy with antiangiogenic therapy? Pre-

clinical modeling in ovarian cancer. J Trans Med 6:2. doi:

10.1186/1479-5876-6-2

59. Li Z, Huang H, Boland P, Dominguez MG, Burfeind P, Lai KM,

Lin HC, Gale NW, Daly C, Auerbach W, Valenzuela D, Yanco-

poulos GD, Thurston G (2009) Embryonic stem cell tumor model

reveals role of vascular endothelial receptor tyrosine phosphatase

in regulating Tie2 pathway in tumor angiogenesis. Proc Natl Acad

Sci USA 106(52):22399–22404. doi:10.1073/pnas.0911189106

60. Kim K, Zang R, Choi SC, Ryu SY, Kim JW (2009) Current status

of gynecological cancer in China. J Gynecol Oncol 20(2):72–76.

doi:10.3802/jgo.2009.20.2.72

Angiogenesis (2012) 15:71–85 85

123