03 (2006) 199–206www.elsevier.com/locate/ygyno

Gynecologic Oncology 1

Arsenic trioxide (As2O3) inhibits peritoneal invasion ofovarian carcinoma cells in vitro and in vivo

Jingjing Zhang ⁎, Bo Wang

Department of Obstetrics and Gynecology, Qilu Hospital of Shandong University, Jinan 250012, Shandong, China

Received 21 December 2005Available online 19 April 2006

Abstract

Objectives. To study the role of arsenic trioxide (As2O3) in regulating peritoneal invasive activity of ovarian carcinoma cells in vitro and in vivo.Methods. The effects of As2O3 on human ovarian cancer cell lines (3AO, SW626 and HO-8910PM) migration, invasion and adhesion with

tumor cells and human peritoneal mesothelial cells (HPMC) were observed by means of cell migration test, cell invasion test and cell adhesiontest. The effects of As2O3 on MMP-2, MMP-9, TIMP-1 and TIMP-2 gene expressions and protein expressions of tumor cells were determined byRT-PCR and ELISA, respectively. In animal experiments, ovarian tumor cells were implanted into abdominal cavity of nude mice and then thenude mice were treated by intraperitoneal injection of different doses As2O3. The foci on the surface of peritoneum were counted.

Results. As2O3 inhibited tumor cells migration, invasion and adhesion with HPMC in a dose-dependent manner, while the same treatmentenhanced tumor cell–tumor cell interactions. As2O3 inhibited mRNA and protein expressions of MMP-2, MMP-9 and TIMP-2 of tumor cells. Incontrast, As2O3 increased mRNA and protein expressions of TIMP-1. As2O3 could reduce tumor cells peritoneal metastasis in nude mice.

Conclusion. As2O3 inhibits in vitro and in vivo peritoneal invasive activity of ovarian carcinoma cells in a dose-dependent manner. Its anti-invasive activity may be the results of reduced cell motility, inhibited attachment of tumor cells to HPMC and enhanced tumor cell–tumor cellinteraction, as well as down-regulation of MMP-2 and MMP-9 levels and up-regulation of TIMP-1 level.© 2006 Elsevier Inc. All rights reserved.

Keywords: Arsenic trioxide; Ovarian neoplasms; Peritoneum; Metastasis

Introduction

Ovarian cancer is the second most frequent gynecologicalcancer. In 60–70% of patients, ovarian tumors are diagnosed atadvanced stages only, since symptoms are absent and a specifictumor marker is lacking. Ovarian cancer, which originates in thesurface epithelium in about 97% of the cases, spreadsprevalently by direct extension of the tumor to adjacent tissuesand by shedding of tumor cells from primary tumor into theperitoneal cavity. Although peritoneal disseminated metastasisis often associated with malignant ascites and a poor prognosis,no effective therapy exists at present.

As2O3 is an active ingredient of a traditional Chinese medicinethat has been used successfully for treating acute promyelocyte

⁎ Corresponding author. Present address: Department of Obstetrics andGynecology, Qingdao Municipal Hospital, Medical College of QingdaoUniversity, Qingdao 266011, Shandong, China.

E-mail address: xzmzjj@163.com (J. Zhang).

0090-8258/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.ygyno.2006.02.037

leukemia(APL) [1] and has been shown to be an effective inducerof apoptosis in some solid cancer cells, such as esophageal,prostate and ovarian carcinoma [2,3]. However the effects ofAs2O3 on the tumor cell peritoneal invasiveness has seldom beenreported. The steps required for ovarian carcinoma metastasisinclude the following: (1) separation of cells from the primarytumor mass on the ovarian surface; (2) passive movement of thetumor cells through the peritoneal fluid; (3) adherence of the cellsto the mesothelium, which lines the peritoneal cavity; (4)migration of tumor cells past the mesothelium into the stroma; (5)enzymatic digestion or remodeling of the underlying stroma byproteinases; (6) growth of the newly established tumor cells intheir new location. So adhesion, motility and proteolysis areimportant factors involved in ovarian carcinoma cells peritonealinvasion. Matrix metalloproteinases (MMPs) are a family ofstructurally related neutral metalloproteinases, which togethercan degrade all the components of the extracellular matrixproteins (ECM) and are involved in tumor invasion andmetastasis. Among the MMPs, the 72 kDa gelatinase A (or

200 J. Zhang, B. Wang / Gynecologic Oncology 103 (2006) 199–206

MMP-2) and 92 kDa gelatinase B (or MMP-9) are believed tobe the critical enzymes for degrading type IV collagen, a majorcomponent of basement membrane, which lies under peritonealmesothelial cells, thus playing a critical role in cancer cellinvasion and metastasis. Their catalytic activity is tightlycontrolled by endogenous inhibitors, including tissue inhibitorof metalloproteinase (TIMP)-1 and TIMP-2. In the present study,we investigated the effect of As2O3 on adhesion, motility,expressions of MMP-2, MMP-9, TIMP-1 and TIMP-2, andperitoneal invasive activity of ovarian carcinoma cells in vitroand in vivo, to clearly demonstrate the anti-invasive activity ofAs2O3 and to provide evidence for exploring drugs forcontrolling the peritoneal metastasis of ovarian tumor cells.

Materials and methods

Cell culture

Human ovarian cancer cell lines 3AO and SW626 cells were obtained fromthe American Type Culture Collection (Rockville, MD, USA), and HO-8910PMcells were purchased from Shanghai Institute of Cell Biology, Chinese Academyof Science. Tumor cells were maintained in RPMI1640 medium supplementedwith 10% heat-inactivated fetal bovine serum (FBS) in a cell culture incubator at37°C in 5% CO2. Human peritoneal mesothelial cells (HPMC) were isolatedfrom the greater omentum in patients operated on for a nonperitonitis cause andwithout disseminated cancer and cultured as previously reported with minormodification [4]. Human omental tissue was obtained from surgically resectedspecimens from patients with informed consent. The omentum was washed inseveral changes of sterile phosphate-buffered saline (PBS), and finely dividedinto approximately 1-mm2 segments. These were washed in PBS twice toremove any contaminating red blood cells. The tissue was then treated with a0.25% trypsin-EDTA solution for about 20 min. The cell suspension wascentrifuged at 1000 rpm for 10 min, and collected cells were cultured inRPMI1640medium supplemented with 20% FBS, 0.4% dermacort and 10 μg/mlinsuline at 37°C in 5% CO2. HPMC were confirmed by immunohistochemistrywith antibodies against cytokeratin, vimentin, CD45 antigen and factor VIII(Signet, Dedham, MA, USA). Passage 2 cultures were used for experiments.

Cytotoxicity assays

In vitro cytotoxic effects of As2O3 (Sigma) on tumor cells was determinedby using CellTiter 96 Non-Radioactive Cell Proliferation Assay kit (Promega,Madison, WI). Briefly, cells growing in plates were dispersed in 0.05% trypsinsolution and resuspended in RPMI-1640 containing 10% FBS. Approximately5000 cells were added to each well of 96-well plate and incubated for 18 h. Afterchanging the media with or without 10% serum, various concentrations (0.5 μM,1.0 μM, 2.0 μM) of As2O3 were added to each well. After 48 h of incubation,cell cytotoxicity was estimated by MTT based assay following the supplier'sinstructions.

Calcein-AM solution and incubation

According to Van Rossen et al. [5], the dye solution, calcein-AM, used toquantify tumor cell adhesion, was prepared by dissolving 50 μg of calcein(Molecular Probes, Leiden, The Netherlands) in 5 μl of anhydrous dimethylsulphoxide (Sigma) and adding this solution to 5 ml of RPMI mediumsupplemented with 0.5% bovine serum albumin (Sigma). Trypsinized tumorcells (1 × 106 cells/ml) were incubated in RPMI/0.5% BSA at 37°C for 45 min,with occasional mixing.

Adhesion assay

To quantify tumor cells adhesion to the tumor cells (homotypic adhesion)and peritoneal mesothelial cells (heterotypic adhesion), a standardized celladhesion assay was developed according to methods described previously [5].

Tumor cells were labeled with calcein as described above. The labeled tumorcells were washed to remove free dye and incubated with indicatedconcentrations (0.5 μM, 1.0 μM, 2.0 μM) of As2O3 for 24 h. Mesothelial andtumor monolayers were established in a 96-well plate. The calcein-labeled,As2O3 -treated tumor cells (2.5 × 104) were seeded additionally. After plateswere incubated at 37°C for 60 min, the medium of each well was removed andwashed twice with RPMI-1640 medium. The remaining fluorescence per wellwas measured on a Perkin Elmer plate reader, using 485 excitation and 530emission filters. On each plate, a standard was prepared by adding differentnumbers of labeled tumor cells to the wells. The amount of adherent tumor cellswas determined by calibrating the measured fluorescence of the experimentalwells in relation to the standard.

Migration assays

To determine the effect of As2O3 on the migration of tumor cells, a wound-migration assay was performed. Tumor cells were cultured on a 12-well plate.When cells were 80% confluent, monolayers were wounded with a 200 μlmicropipette tip, washed twice with PBS, and incubated for 24 h with serum-freemedium in the presence or absence of indicated concentrations (0.5 μM, 1.0 μM,2.0 μM) of As2O3. After incubation, cells were washed with PBS. Migration wasquantified by counting the number of cells that migrated from the wound edgeinto the denuded area for a distance of 1 cm.

Cell invasion assay

An invasion assay was performed according to the method reported byHirashima et al. [6]. A millicell (Sigma) with an 8-μm porosity cell-permeablepolycarbonate filter was coveredwith recombinant basementmembraneMatrigel(Sigma) and was placed on a 24-cell culture plate. HPMC were cultured fully onMatrigel, and washed, and 2 × 105 tumor cells pretreated with or without As2O3

were then added into the millicell. As a chemoattractant, NIH3T3 fibroblastconditioned medium was used outside of the millicell. After cultivation for 12 h,the millicell was immersed in 100%methanol for 1 min for fixation, and all cellswere then stained by hematoxylin. The cells remaining on the top surface of thefilter were completely removed with a cotton swab, and the filter was removedfrom the millicell and mounted on a glass slide. These preparations wereexamined under a microscope at ×400 magnification. The number of infiltratingtumor cells was counted in five regions selected at random, and the extent ofinvading tumor cells was determined by the mean count.

RT-PCR

Total RNA was extracted from tumor cells, pretreated with or withoutindicated concentrations of As2O3 for 24 h, by Trizol (Invigen) according to themanufacturer's protocol. Reverse transcription of RNAwas performed in a finalreaction volume of 20 ml containing 5 μg of total RNA in Moloney murineleukemia virus (MMLV) reverse transcriptase buffer (Promega, Madison, WI,USA). A 1 μl portion of the reaction mixture was then amplified by PCR withthe pairs of primers (Table 1). A total of 30 cycles were performed, with eachcycle comprising 1 min at 94°C, 1 min at 50–60°C depending on each gene, and1.5 min at 72°C with a final extension of 8 min at 72°C. Human β-actinreactions were amplified for 25 cycles. The reaction products were separated on2% agarose gel, stained with 1 μg/ml ethidium bromide, and visualized using aDigital imaging system. Semi-quantitative analysis was performed using theratio of the genes tested and β-actin.

Enzyme-linked immunosorbent assays (ELISA)

After tumor cells were pretreated with or without As2O3 for 24 h, thesupernatant was collected. The levels of MMP-2, MMP-9, TIMP-1 and TIMP-2were measured in the supernatants using the ELISA kit (R&D Systems)according to the manufacturer's instructions.

Animal experiments

Male, 4- to 6-week-old BALB/C nu/nu mice (provided by Animal researchcenter, Chinese Academy of Medical Sciences, SCXT11-00-0006) were used.

Table 1Sequences of PCR primers and length of PCR product

Genes PCR primers Length ofPCR product(bp)

MMP-2 5′-CTGAGATCTGCAAACAGGACATTGT-3′ 4815′-CCGCCCTGCAGGTCCACGACGGCAT-3′

MMP-9 5′-GACTCGGTCTTTGAGGAGCC-3′ 3505′-GAACTCACGCGCCAGTAGAA-3′

TIMP-1 5′-ATCCTGTTGTTGCTGTGGCTG-3′ 5205′-GACTGGAAGCCCTTTTCAGA-3′

TIMP-2 5′-CTCGCTGGACGTCGTTCGAGGAAAGAA-3′ 1555′-AGCCCATCTGGTACCTGTGGTTCA-3′

β-actin 5′-GCATGGAGTCCTGTGGCAT-3′ 3205′-GGTGGCGTTTACGAAGATC-3′

201J. Zhang, B. Wang / Gynecologic Oncology 103 (2006) 199–206

Tumor injections containing 4 × 106 (0.2 ml) tumor cells were made i.p. intoabdominal cavity. The animals were randomized into one of four groups: (1)normal saline (control); (2) 0.5 mg/kg As2O3; (3) 1.0 mg/kg As2O3; (4) 2.0 mg/kg As2O3. At 4 days after implantation, normal saline or various concentrationsof As2O3 was administered i.p. once a day for 7 days. Another course oftreatment was given 3 weeks later. The animals were killed (n = 8 per group).The peritoneum was stained with H and E, and the metastasis foci werecounted.

Statistics

Data are expressed as mean ± standard deviation (x̄ ± SD). Statisticalevaluation of the data was performed with analysis of variance and chi-squaretest. P < 0.05 was considered statistically significant.

Fig. 1. Effect of As2O3 on the cytotoxicity of tumor cells. Exponentially growing3AO (A), SW626 (B) and HO-8910PM cells (C) were treated with indicatedconcentrations of As2O3 for 2 days in the presence or absence of serum.Cytotoxicity was determined by MTT-based assay as described.

Results

Effect of As2O3 on cytotoxicity of tumor cells

3AO, SW626 and HO-8910PM cells were cultured andcytotoxicity of As2O3 on cells in the presence or absence ofserum was measured. As shown in Fig. 1, significant cyto-toxic effects were observed in the serum-supplemented mediafollowing the As2O3 treatment for 48 h. In the serum-supplemented condition, 2 μM As2O3 exhibited approxi-mately 40% growth inhibition of 3AO, SW626 and HO-8910PM cells, respectively (P < 0.01) (Figs. 1A–C). Therewas no significant difference of growth inhibition betweendifferent cell lines (P > 0.05). In the serum-free condition,2 μM As2O3 did not affect 3AO cell population (P > 0.05),and significant reduction of the cell population was observedat high concentration of As2O3 (10 μM) (P < 0.01) (Fig.1A). Likewise, As2O3 in absence of serum did not inhibitgrowth of SW626 and HO-8910PM cells until the concen-tration reached 10 μM (P < 0.01) (Figs. 1B and C).Therefore noncytotoxic concentrations of As2O3 (below2 μM) in absence of serum were used in the following invitro experiments.

Effects of As2O3 on cell–cell adhesion

Adhesion of tumor cells was calculated by ratio of thenumber of adhesive tumor cells to the total amount of tumor

cells added. As shown in Fig. 2A, pre-incubation of 3AO,SW626 and HO-8910PM cells with As2O3 resulted indecreased tumor cell adhesion to HPMC (heterotypic adhesion)(P < 0.01). These effects were dose-dependent (P < 0.01). Onthe other hand, As2O3 treatment dose-dependently increasedtumor cell–tumor cell adhesion (homotypic adhesion) (P < 0.05)(Fig. 2B).

Fig. 2. Effects of As2O3 on the heterotypic and homotypic adhesion of tumorcells. Labeled tumor cells incubated with or without various concentrations ofAs2O3 were seeded onto the 96-well plates coated with HPMC(A) or tumor cells(B). After 60 min, the remaining fluorescence per well was measured.

Fig. 3. Effects of As2O3 on the migration and peritoneal invasion of tumor cellsin vitro. (A) Cell migration was assessed by counting the number of cells thatmigrated into the denuded area over a 1 cm distance along the wounded edge.(B) Invasion through a layer of Matrigel covered by HPMC monolayer wasdetermined by a Boyden chamber method.

202 J. Zhang, B. Wang / Gynecologic Oncology 103 (2006) 199–206

Effects of As2O3 on migration and invasion of tumor cellsin vitro

The results showed that As2O3 not only reduced themigration ability of three kinds of tumor cells (P < 0.01), butalso significantly inhibited tumor peritoneal invasion potentialusing the invasion assay in vitro (P < 0.01), and the inhibitiveeffects were both dose-dependent (P < 0.01) (Figs. 3A and B).

Effects of As2O3 on the expression of proteases and theirinhibitors

This study next explored whether As2O3 regulated theexpressions of MMP-2, MMP-9, which were involved increating and maintaining an environment that contributesperitoneal metastasis of tumor cells, and their inhibitors(TIMP-2 and TIMP-9) in tumor cells. RT-PCR was performedto determine whether the proteases and their inhibitors weremodulated by As2O3 at mRNA levels. ELISA was used todetermine the protein levels of MMP-2, MMP-9, TIMP-1 andTIMP-2 in the supernatants of tumor cells before or after thestimulation of As2O3. As shown in Figs. 4A–C, 3AO, SW626and HO-8910PM cells expressed MMP-2, MMP-9, TIMP-1and TIMP-2. As2O3 significantly reduced mRNA levels ofMMP-2, MMP-9, and TIMP-2 of all three kinds of tumor cellsdose-dependently (P < 0.05), but it increased TIMP-1 levels in adose-dependent manner (P < 0.01). Consistent with the mRNA

level, As2O3 markedly dose-dependently reduced protein levelsof MMP-2, MMP-9, and TIMP-2 of tumor cells (P < 0.05), butincreased TIMP-1 levels in a dose-dependent manner (P < 0.01)(Figs. 4D–F).

Effects of As2O3 on peritoneal metastasis in vivo

Furthermore, this study examined the effect of As2O3 ontumor peritoneal metastasis by the animal models. 3AO, SW626and HO-8910PM cells were injected into abdominal cavity ofmice, respectively, then different concentrations of As2O3 wereused i.p. During or after the administration, the mice did notshow abnormality, such as anorexy, loss of weight or nervoussystem disorder. Fig. 5 demonstrated that the injection of As2O3

to abdominal cavity reduced the number of foci on theperitoneum dose-dependently (P < 0.05), indicating thatAs2O3 may control peritoneal metastasis of ovarian carcinomacells.

Discussion

In vitro growth inhibition of As2O3 has previously beenreported in various human hematologic cancers and solidtumors. According to Park et al. [7], the cytotoxicity of As2O3

Fig. 4. Effects of As2O3 on the expression of various proteases and their inhibitors in tumor cells. 3AO (A), SW626 (B) and HO-8910PM cells (C) were exposed toindicated concentrations of As2O3 for 24 h. Total RNA of the cells was extracted and RT-PCR was used for semiquantitation. The PCR products (MMP-2, MMP-9,TIMP-1, TIMP-2) and β-actin were separated by 2% agarose gel electrophoresis. B: 3AO (D), SW626 (E) and HO-8910PM cells (F) were pretreated with indicatedconcentrations of As2O3 for 24 h. The levels of MMP-2, MMP-9, TIMP-1 and TIMP-2 in the supernatant were determined using the ELISA kit.

203J. Zhang, B. Wang / Gynecologic Oncology 103 (2006) 199–206

was influenced by the presence of serum. Likewise, remarkablein vitro inhibition of cell growth was induced in all three ovariancancer cell lines at low or high doses As2O3 in the presence ofserum. It may be the results of that some ingredient in serummay influence the action of As2O3. But in the absence of serum,only high dose As2O3 had pronounced cytotoxic effects ontumor cells. In order to avoid the possibility that As2O3 mayreduce tumor invasiveness by influencing the growth of tumorcells, noncytotoxic dosage of As2O3 in absence of serum wasused in vitro experiments.

In order for the ovarian tumor cells to spread to peritoneum,they need to separate from the primary site and attach to HPMC

which cover the surface of peritoneum, involving the loss oftumor cell–tumor cell adhesion (homotypic adhesion) and gainof tumor cell–host cell (HPMC) adhesion (heterotypic adhe-sion). Our present study showed that As2O3 enhancedhomotypic adhesion and inhibited heterotypic adhesion dose-dependently. Homotypic adhesion of epithelial carcinoma ismediated for the most part by the E-cadherin/catenin complex[8]. In most primary foci of advanced ovarian carcinoma, theexpression of E-cadherin is down-regulated and instable, whichlooses the adherent junctions and benefits for the tumor cells todetach from each other [9]. Whether As2O3 enhances tumorcells homotypic adhesion through changing the expression of

Fig. 5. Effects of As2O3 on peritoneal metastasis of tumor cells in vitro. Tumorcells were injected into the abdominal cavity of nude mice and the mice weretreated with various concentrations of As2O3 as indicated. The animals werekilled. The peritoneum was stained with H&E, and the foci were counted.

204 J. Zhang, B. Wang / Gynecologic Oncology 103 (2006) 199–206

E-cadherin/catenin remains further research. It has been re-ported that the CD44 molecule, which is a major receptor forhyaluronic acid, is frequently expressed by ovarian cancer cellsand is partly responsible for mediating mesothelial bindingthrough recognition of mesothelial-associated hyaluronic acid[10]. Liu et al. found that As2O3 induced decrease of theexpression of CD44 in stomach cancer cells [11]. As2O3 mayinhibit heterotypic adhesion of ovarian tumor cells with HPMCby the same mechanisms. Our results suggest that down- andup-regulation of heterotypic and homotypic adhesion, respec-tively, by As2O3 may be a crucial event in the As2O3-inducedsuppression of invasion of ovarian tumor cells.

Tumor cell motility is required for the cells to invade andmetastasize [12]. The tumor cells with high invasive abilityusually have active locomotivity. Although many factors caninfluence cellular motility through different signal transductioncascades, cellular movement requires reorganization of cyto-skeleton at last, such as microtubule and microfilament. Thefactors which can affect the reorganization of cytoskeleton mayinfluence cell motility. It has been found that microtubule-affecting agents, such as Paclitaxel, Vincristine and Vindesine atnon-cytotoxic doses can inhibit the migration of tumor cells[13]. It has been reported that As2O3 markedly inhibits GTP-induced polymerization and microtubule formation in vitro inmyeloid leukemia cells [14]. So it may have some effects onmigration of tumor cells. Our results proved that As2O3

inhibited migration of ovarian carcinoma cells in vitro. It maybe the results of the action on ovarian tumor cell's cytoskeleton,such as microtubule formation.

For cancer cells to invade the interstitial tissue, variousproteinases such as MMPs are required. MMP-2 and MMP-9are well known as decomposing enzymes of type IV collagen,which are important components of basement membraneunderlying HPMC. Tumor cells can dissolve type IV collagenand penetrate basement membrane of peritoneum to formperitoneal metastasis foci through secreting MMP-2 and MMP-9, and the contents of MMP-2 and MMP-9 of tumor cells arerelated to their peritoneal invasive ability. TIMP-2 and TIMP-1are tissue inhibitors of MMP-2 and MMP-9, respectively, whichcan modulate activity of MMP-2 and MMP-9. Keepingdynamic balance of MMPs/TIMPs expressions and maintaining

integrity of basement membrane are important mechanisms toanti-metastasis therapy [15–17]. So we detect whether As2O3

has influence on the expressions of MMP-2, MMP-9, TIMP-1and TIMP-2 of ovarian tumor cells. This study demonstratedthat As2O3 decreased MMP-2 and MMP-9 expressions atmRNA and protein levels. As2O3 also decreased mRNA andprotein expression levels of TIMP-2, but it increased TIMP-1expression at mRNA and protein levels. The increase of TIMP-1 expression inhibits the enzyme activity of MMP-9. Therefore,the down-regulation of MMP-2 and MMP-9, and up-regulationof TIMP-1 by As2O3 attenuated the dissolution and destructionof basement membrane and inhibited the peritoneal invasionability of ovarian carcinoma cells. The disagree of expression ofTIMP-1 and TIMP-2 may be the results of different regulationsin cells. TIMP-2 mRNA transcript levels were differentlyregulated from TIMP-1 levels, both in cell culture as evidencedby studies using 12-O-tetradecanoylphorbol-13-acetate (TPA)and transforming growth factor beta1 and in vivo as evidencedby comparison of transcript levels for these inhibitors in humancolon adenocarcinoma tissue and adjacent normal colonicmucosa [18]. Additional evidence supporting independentregulation of TIMP-1 and TIMP-2 has been provided byexperiments showing their differential responses to cytokines orother agents. For example, in macrophages, lipopolysaccharideshave been reported to down-regulate TIMP-1 and up-regulateTIMP-2 [19]. TIMP-1 was up-regulated by interleukin-1 andinterleukin-6, whereas TIMP-2 was not affected by both agents[20]. With characterization and comparison of the structure ofthe genes, differences and divergence of these genes accountedfor the fact that although there was a marked homology betweenTIMP-1 and TIMP-2 at the protein level, they were differentlyregulated [18,21]. MMPs and TIMPs expressions are regulatedby several transcription factors, such as nuclear factor-κB(NF-κB) and activator protein-1 (AP-1), because their promotersequences contain the putative binding sites for the transcriptionfactors. Although MMP-2 promoter itself does not contain aNF-κB binding site, several reports show that NF-κB canaugment MMP-2 activation [22,23]. It was reported that As2O3

blocked the promoter stimulating activity and DNA bindingactivity of NF-κB and suppressed the transcription of MMP-2and MMP-9 genes [7]. Moreover, arsenite inhibited IκB kinase,which was essential for NF-κB activation [24]. So As2O3 mayinhibit the expression of MMP-2 and MMP-9 by influencing thefunction of NF-κB.

In vitro invasive assay was used to verify the effects ofAs2O3 on tumor cell peritoneal invasive potential. It was foundthat when all three kinds of ovarian tumor cells were pretreatedwith As2O3, the number of filter-infiltrating cells was decreased,demonstrating that As2O3 inhibited the tumor cell peritonealinvasion ability in vitro.

Recent clinical studies have shown that As2O3 givenintravenously is effective and relatively safe in the treatmentof APL [25]. However, the clinical use of As2O3 for treatmentof ovarian cancer as well as tumor peritoneal metastasis by waysof abdominal administration has not been reported. Here weobserved the effect of As2O3 given i.p. on ovarian cancerperitoneal metastasis using animal experiments to provide basis

205J. Zhang, B. Wang / Gynecologic Oncology 103 (2006) 199–206

for further clinical application. LD50 of As2O3 for abdominaladministration in mice is 12 mg/kg [26]. In previous reports, itwas found that large doses As2O3 (4 mg/kg, 8 mg/kg, 16 mg/kg)could cause brain necrosis, liver necrosis, loss of appetite, highexcitability and even death of mice in seven days, buttherapeutic doses As2O3 (1 mg/kg, 2 mg/kg) did not damagethe brain, liver or kidney [27]. Dosages chosen in our in vivoexperiments did not cause abnormal reaction of mice, showingsafety of these dosages too. Such dosages were also applied byother investigators [28,29]. Our results revealed that when micewere treated with various concentrations of As2O3 after tumorinjections, the number of colonies formed on peritoneumreduced in a dose-dependent mode, which was agree withresults in vitro and proved As2O3 could inhibit the peritoneummetastasis of ovarian carcinoma. Huang et al. also found thatAs2O3 reduced the 3AO cell tumor formation rate in nude miceand concluded As2O3 could inhibit the abdomino-plantation ofhuman ovarian carcinoma in nude mice [28], which was agreewith our results. When different doses As2O3 were injected intoabdominal cavity of nude mice, the appearance of peritonealmesothelial cells on the surface of organs in abdominal cavitydid not change, indicating that As2O3 could act on carcinomacells by choice but did not damage the normal host cells, whichproved the safety of As2O3 [30].

Ovarian cancer commonly spreads within the peritonealcavity. Combined chemotherapy based on platinum analogue isimportant therapy at present. But drug-resistance, toxic and sideeffects restrict the effects of chemotherapy. As a majorcomponent of a traditional Chinese medicine, As2O3 has wideanticancer spectra, significant anticancer effects and littleadverse reaction in experiment and clinic. The most importantis that drug-resistance or cross drug-resistance with other anti-tumor drugs has not been found. So As2O3 shows favorableperspective in the future. This study showed that As2O3

inhibited the peritoneal invasion of human ovarian carcinomacells in vitro and in vivo, mainly by suppressing migration,heterotypic adhesion with HPMC and down-regulating MMP-2and MMP-9 expressions as well as enhancing homotypicadhesion with tumor cells and up-regulating TIMP-1 expres-sion. The rationale for intraperitoneal administration of As2O3 isthat tumor cells in abdominal cavity receive sustained exposureto high concentrations of agents while normal tissues, such asthe bone marrow, are relatively spared. With more research inpharmacodynamics and clinic application of As2O3, it might beconsidered as a possible agent for controlling the peritonealmetastasis of ovarian cancer.

References

[1] Gallagher RE. Arsenic: new life for an old potion. N Engl J Med 1998;339:1389–91.

[2] Shen ZY, Shen WY, Chen MH, Shen J, Cai WJ, Zeng Y. Mitochondria,calcium and nitric oxide in the apoptotic pathway of esophageal carcinomacells induced by As2O3. Int J Mol Med 2002;9:385–90.

[3] Uslu R, Sanli UA, Sezgin C, Karabulut B, Terzioglu E, Omay SB, et al.Arsenic trioxide-mediated cytotoxicity and apoptosis in prostate andovarian carcinoma cell lines. Clin Cancer Res 2000;6:4957–64.

[4] Mizutani K, Kofuji K, ShirouzuK. The significance ofMMP-1 andMMP-2

in peritoneal disseminated metastasis of gastric cancer. Surg Today 2000;30:612–21.

[5] Van Rossen ME, Hofland LJ, van den Tol MP, van Koetsveld PM, Jeekel J,Marquet RL, et al. Effect of inflammatory cytokines and growth factors ontumor cell adhesion to the peritoneum. J Pathol 2001;193:530–7.

[6] Hirashima Y, Kobayashi H, Suzuki M, Tanaka Y, Kanayama N, Terao T.Transforming growth factor-β1 produced by ovarian cancer cell line HRAstimulates attachment and invasion through an up-regulation of plasmin-ogen activator inhibitor type-1 in human peritoneal mesothelial cells. J BiolChem 2003;278:26793–802.

[7] Park MJ, Lee JY, Kwak HJ, Park CM, Lee HC, Woo SH, et al. Arsenictrioxide (As2O3) inhibits invasion of HT1080 human fibrosarcoma cells:role of nuclear factor-κB and reactive oxygen species. J Cell Biochem2005;95:955–69.

[8] Hsu MY, Meier FE, Nesbit M, Hsu JY, Van Belle P, Elder DE, et al.E-cadherin expression in melanoma cells restores keratinocyte-mediatedgrowth control and down-regulates expression of invasion-related adhesionreceptors. Am J Pathol 2000;156:1515–25.

[9] Davidson B, Gotlieb WH, Ben-Baruch G, Nesland JM, Bryne M,Goldberg I, et al. E-cadherin complex protein expression and survival inovarian carcinoma. Gynecol Oncol 2000;79:362–71.

[10] Lessan K, Aguiar DJ, Oegama T, Siebenson L, Skubitz AP. CD44 and β1integrin mediate ovarian carcinoma cell adhesion to peritoneal mesothelialcells. Am J Pathol 1999;154:1525–37.

[11] Liu TF, Cheng BL, Guan YG, Liang T. Study on relationship betweenexpressions of CD44 and inhibition effect of arsenic trioxide on carcinoma.J Harbin Med Univ 2001;35:111–2.

[12] Wells A. Tumor invasion: role of growth factor-induced cell motility. AdvCancer Res 2000;78:31–101.

[13] Hayot C, Farinelle S, De Decker R, Decaestecker C, Darro F, Kiss R, et al.In vitro pharmacological characterization of the anti-angiogenic and anti-tumor cell migration properties mediated by mirotubule-affecting drugs,with special emphasis on the organization of the actin cytoskeleton. Int JOncol 2002;21:417–25.

[14] Li YM, Broome JD. Arsenic targets tubulins to induce apoptosis inmyeloid leukemia cells. Cancer Res 1999;59:776–80.

[15] Stamenkovic I. Matrix metalloproteinases in tumor invasion andmetastasis. Semin Cancer Biol 2000;10:415–33.

[16] Fan YZ, Zhang JT, Yang HC, Yang YQ. Expression of MMP-2, TIMP-2protein and the ratio of MMP-2/TIMP-2 in gallbladder carcinoma and theirsignificance. World J Gastroenterol 2002;8:1138–43.

[17] Zucher S, Cao J, Chen WT. Critical appraisal of the use of the matrixmetalloproteinase inhibitors in cancer treatment. Oncogene 2000;19:6642–50.

[18] Stetler-Stevenson WG, Brown PD, Onisto M, Levy AT, Liotta LA. Tissueinhibitor of metalloproteinases-2 (TIMP-2) mRNA expression in tumorcell lines and human tumor tissues. J Biol Chem 1990;265:13933–8.

[19] Shapiro SD, Kobayashi DK, Welgus HG. Identification of TIMP-2 inhuman alveolar macrophages. Regulation of biosynthesis is opposite tothat of metalloproteinases and TIMP-1. J Biol Chem 1992;267:13890–4.

[20] Zafarullah M, Su S, Martel-Pelletier J, DiBattista JA, Costello BG, Stetler-Stevenson WG, et al. Tissue inhibitor of metalloproteinase-2 (TIMP-2)mRNA is constitutively expressed in bovine, human normal, andosteoarthritic articular chondrocytes. J Cell Biochem 1996;60:211–7.

[21] Hammani K, Blakis A, Morsette D, Bowcock AM, Schmutte C, Henriet P,et al. Structure and characterization of the human tissue inhibitor ofmetalloproteinases-2 gene. J Biol Chem 1996;271:25498–505.

[22] Kim H, Koh G. Lipopolysaccharide activates matrix metalloproteinase-2in endothelial cells through an NF-kappaB-dependent pathway. BiochemBiophys Res Commun 2000;269:401–5.

[23] Yoon SO, Park SJ, Yoon SY, Yun CH, Chung AS. Sustained production ofH2O2 activates pro-matrix metalloproteinase-2 through receptor tyrosinekinases/phosphatidylinositol 3-kinase/NF-kappaB pathway. J Biol Chem2002;277:30271–82.

[24] Miller Jr WH, Schipper HM, Lee JS, Singer J, Waxman S. Mechanisms ofaction of arsenic trioxide. Cancer Res 2002;62:3893–903.

[25] Liu Q, Hilsenbeck S, Gazitt Y. Arsenic trioxide-induced apoptosis inmyeloma cells: p53-dependent G1 or G2/M cell cycle arrest, activation of

206 J. Zhang, B. Wang / Gynecologic Oncology 103 (2006) 199–206

caspase-8 or caspase-9, and synergy with APO2/TRAIL. Blood 2003;101:4078–87.

[26] Kreppel H, Reichl FX, Szinicz L, Fichtl B, Forth W. Efficacy of variousdithiol compounds in acute As2O3 poisoning in mice. Arch Toxicol 1990;64:387–92.

[27] Wang Y, Hou LM, Dong LM, Wang XJ. Effects of As2O3 on transplantingovarian cancer in the normal immunity mouse. China J Mod Med 2003;13:18–22.

[28] Huang SG, Kong BH, Yang RF, Jiang S. Primary study of arsenic trioxide

inhibits abdomino-metastatic tumor formation of human ovarian carcino-ma in nude mice and its mechanisms. Ai Zheng 2002;21:401–4.

[29] Wei LH, Lai KP, Chen CA, Cheng CH, Huang YJ, Chou CH, et al. Arsenictrioxide prevents radiation-enhanced tumor invasiveness and inhibitsmatrix metalloproteinase-9 through downregulation of nuclear factorkappaB. Oncogene 2005;243:390–8.

[30] Cai HP, Deng ZH, Li S, Zhang XR, Shen JW. Study of the inhibitory effectof As2O3 on ascites production in nude mice with colorectal cancer. Chin JDig 1999;19:378–80.