© The Author 2005. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. 1For Permissions, please email: journals.permissions@oupjournals.org

Human Reproduction Page 1 of 9 doi:10.1093/humrep/dei272

Differential effects of inducers of syncytialization and apoptosis on BeWo and JEG-3 choriocarcinoma cells

S.Al-Nasiry1,2, B.Spitz1, M.Hanssens1, C.Luyten1 and R.Pijnenborg1

1Department of Obstetrics and Gynaecology, University Hospital Gasthuisberg, Katholieke Universiteit Leuven, 3000 Leuven, Belgium2To whom correspondence should be addressed: e-mail: salwan.al-nasiry@med.kuleuven.ac.be

BACKGROUND: The interactions of trophoblasts with the cytokine network at the fetomaternal interface determinethe pathway the cell undertakes, e.g. proliferation, differentiation and apoptosis. METHODS: We used cultures offusigenic BeWo and non-fusigenic JEG-3 choriocarcinoma cells to study the effects of inducers of syncytialisation(forskolin) and apoptosis [tumour necrosis factor-� (TNF�)] on differentiation, viability, proliferation and apoptosis.RESULTS: E-cadherin immunostaining showed that syncytium formation was confined to BeWo and not JEG-3cells, while secretion of hCG was promoted by forskolin in both cell types implying a ‘dissociation’ between morpho-logical and biochemical differentiation. Forskolin also had differential effects on cell viability (MTT reduction test)and proliferation (Ki67 immunostaining with MIB-1 monoclonal antibody), both decreasing in BeWo and increasing inJEG-3 cells. TNF� increased apoptosis (cytokeratin neo-epitope immunostaining with M30 monoclonal antibody) inboth cell types, an effect which was blocked by epidermal growth factor selectively in JEG-3 cells. CONCLUSION:Our results suggest that the differential responses of BeWo and JEG-3 cells to inducers of syncytialization and apop-tosis might be related to their fusigenic capacity. Caution is needed when extrapolating results obtained by thesemodels to normal trophoblast populations. However, we speculate that these models can help identify key factorsinvolved in trophoblast differentiation at the placental bed.

Key words: choriocarcinoma/differentiation/EGF/orskolin/TNFα

Introduction

A successful pregnancy is largely dependent on co-ordinatedevents taking place very early after implantation at the fetoma-ternal interface in which trophoblast cells are key players.These cells differentiate according to one of two distinct path-ways. In the villous pathway, mononuclear cytotrophoblastsfuse to form a specialized multinuclear syncytium on the outerlayer of placental villi with synthetic, immune and transportfunctions. In the extravillous pathway, cytotrophoblasts prolif-erate and differentiate into an invasive phenotype that pene-trates the maternal decidua and myometrium, and ultimatelyfuse into multinuclear giant cells.

While there is much literature on how the process of fusionis regulated in villous trophoblast, little is known about the reg-ulation of extravillous trophoblast fusion. Fusion of extravil-lous trophoblasts, as well as induction of apoptosis, has beensuggested to restrict extravillous trophoblast invasion (Pijnenborget al., 1981; Reister et al., 2001). Fusion and apoptosis in bothvillous and extravillous trophoblasts are regulated by differentcomponents of the cytokine network. Indeed, the cytokinetumour necrosis factor-α (TNFα) has been implicated in thecontrol of apoptosis of the invasive trophoblast of the placentalbed (Pijnenborg et al., 1998; Reister et al., 2001) as well as tro-phoblast proliferation and differentiation (Yang et al., 1993).Epidermal growth factor (EGF) on the other hand has been

shown to induce proliferation and differentiation into syncytio-trophoblast (Morrish et al., 1987), and to inhibit TNFα inducedapoptosis (Garcia-Lloret et al., 1996). Yet, the effects of anycytokine may differ according to the specific type of trophob-lastic cell and may be modulated by the presence of othercytokines and growth factors.

We therefore hypothesized that trophoblastic cells with dif-ferent fusigenic capacity respond differently to inducers ofsyncytialization and apoptosis. Although primary trophoblastcultures would be an obvious choice, they were not used in thisstudy because of the high variability between samples reflect-ing heterogeneous populations of isolated cyto- and syncytio-trophoblasts. Instead, we used two choriocarcinoma cell lines(Ringler and Strauss, 1990; King et al., 2000; Shiverick et al.,2001; Sullivan, 2004): (i) the fusigenic BeWo cells (Pattilloand Gey, 1968; Wice et al., 1990); and (ii) the non-fusigenicJEG-3 cells (Kohler and Bridson, 1971; Coutifaris et al.,1991).

We studied the effects of forskolin, an inducer of syncytiali-zation (Wice et al., 1990), of TNFα, an inducer of apoptosis,and of the combination of both (Fk/TNFα) on morphologicaldifferentiation (E-cadherin expression), biochemical differenti-ation (hCG levels in culture media), viability (MTT reductiontest), proliferation (Ki67 immunostaining using MIB-1 mono-clonal antibody) and apoptosis (cytokeratin immunostaining

Hum. Reprod. Advance Access published October 6, 2005

S.Al-Nasiry et al.

2

using M30 monoclonal antibody). Secondly, since pretreatmentwith EGF has been reported to protect trophoblast againstTNFα-induced cytotoxicity (Garcia-Lloret et al., 1996), we evalu-ated whether pretreatment with EGF would modulate the cellu-lar responses to TNFα and forskolin in the chosen model system.

Materials and methods

Cell culture

BeWo and JEG-3 choriocarcinoma cells were purchased from theEuropean Collection of Cell Cultures (ECACC, Salisbury, UK) andwere maintained in 75 cm2 Falcon culture flasks (BD Biosciences,Erembodegem, Belgium) under standard culture conditions of 5%CO2 in air (containing 20% O2) at 37°C with medium renewal every2–3 days. The culture media used were Ham F12 (Sigma-Aldrich,Bornem, Belgium) for BeWo cells and Dulbeco’s modified Eagle’smedium (DMEM) (Gibco, Life Technologies, Belgium) for JEG-3cells containing 25 mM glucose, 25 mM HEPES, 200 mM glutamine,1/100 penicillin-streptomycin solution and 1/1000 gentamycin solu-tion, and supplemented with 10% heat-inactivated fetal calf serum(FCS).

Cell concentrations were adjusted to 4 × 104 cell/ml (BeWo) and2 × 104 cell/ml (JEG-3) after counting the cells with a haemocytometer.For the experiments, cells passaged for 10–16 times after purchasewere used. Cell suspensions (500 μl) were seeded onto eight-wellLabTek slides (Nunc, VEL, Leuven, Belgium) and 250 μl suspen-sions onto 96-well Nunc plates (Nunc). Cells were kept for 4 days instandard culture conditions of 5% CO2 in air (containing 20% O2) at37°C. After the initial 4 h of culture, which allowed for cell attach-ment, the medium was changed with the addition of EGF (5 ng/ml)(Gibco BRL, Life Technologies, Merelbeke, Belgium) into half thewells. After 24h, the medium was removed and kept at –20°C forhCG determination and fresh serum-free medium was added con-taining 100 μM of forskolin (Sigma-Aldrich, Bornem, Belgium)with or without 1000 IU/ml of human recombinant TNFα (Invitro-gen, Life Technologies). After preliminary testing, it was decided tofix forskolin treatment at 100 μM concentration. Serum-freemedium was used to avoid possible stimulatory effects of growthfactors present in FCS. Cultures were stopped after 48 and 72 h, via-bility was tested by an MTT test on the 96-well Nunc plates, andLabTek slides were fixed for immunohistochemistry. Conditionedmedia were collected from LabTek slides for protein and hCGquantification.

MTT cell viability assay

An MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bro-mide) test was performed on cultures at 48 h and 72 h after initialplating as described by Mosmann (1983). The MTT test is based onthe ability of viable cells to produce formazan from the cleavage ofthe tetrazolium salt by functional mitochondria. After a 2-hour incu-bation with MTT (Sigma-Aldrich, Bornem, Belgium) at 37°C, thecells were lysed with dimethyl sulphoxide (DMSO) in Sorensen’sglycine buffer and the formazan crystals solubilized. Absorbancewas read at 550 nm using a spectrophotometric microplate reader.The experiments were performed in triplicate wells and repeated sixtimes. The intraassay and interassay coefficients of variation were3.8% (SD 1.5) and 8.2% (SD 3.1), respectively.

Immunohistochemistry

LabTek slide cultures were stopped at 48 h and 72 h after initial plat-ing by washing in phophsate-buffered saline (PBS) solution and

fixation in 50/50 (v/v) ethanol/paraformaldehyde solution for 1 min.Cells were kept in 70% ethanol at 4°C until immunostaining. All incu-bations were carried out at room temperature. Briefly, the cells werewashed twice with 0.01M Tris-buffered saline pH 7.6 (TBS) andblocked with bovine serum albumin (BSA) 2% containing Tween 80(0.5%) for 15 min. The following primary antibodies were used:

(i) monoclonal mouse anti-human E-cadherin, clone HECD-1(R&D Systems, Abingdon, UK) in a concentration of 2 μg/ml for 30 min;

(ii) monoclonal mouse Ab (MIB-1 clone) against Ki67 epitope(DakoCytomation, Glostrup, Denmark) in a concentration of 4 μg/mlfor 30 min

(iii) monoclonal mouse Ab against the M30 epitope of cytokeratin18 (Boehringer Mannheim Gmbh, Mannheim, Germany) in a concen-tration of 4 μg/ml for 30 min.

After another blocking step, peroxidase-conjugated goat antimouseIgG 1/100 (v/v) in TBS (Jackson ImmunoResearch, West Grove,Pennsylvannia, USA), with the addition of normal human serum 1/25(v/v) in TBS, was used as secondary antibody. Peroxidase activity wasdemonstrated by adding 3-amino-9-ethylcarbazole (AEC) for 15 min ina dark chamber. The cells were counterstained with Mayer’s haematox-ylin. Finally, the slides were mounted with glycerine jelly. As negativecontrols, we used irrelevant mouse antibody as the primary antibody.

Slides were evaluated by light microscopy at 20× magnification. Indifferent slides, the number of entities (nuclei or cells) positive forM30 or MIB-1, or negative for E-cadherin was counted in 16 centralfields. This number was normalized to a total number of 500 nucleiper slide and the mean of values from three experiments was calcu-lated. The quality of the staining was generally adequate for cytologi-cal evaluation with <10% of tested areas showing unsatisfactory cellattachment or fixation/staining errors, which were excluded fromevaluation.

hCG and protein measurement in culture media

Conditioned media were collected from LabTek wells at the timeintervals specified and were kept at 4°C until the time of assay. Pro-tein concentration in culture media was quantified based on the Biuretreaction using the BCA kit (Pierce, Rockford, Illinois, USA) accord-ing to the manufacturer’s instructions. For hCG quantification, anenzyme immunoassay kit was used (DRG Diagnostics, Marburg,Germany), which is based on the sandwich principle and detects bothwhole hCG molecule and the free β-hCG subunit present in spentmedia. The assay was carried out according to the manufacturer’sinstructions. To minimize the effects of medium colour on photomet-ric reading, samples were diluted 1/10 (v/v) in the sample diluentsolution provided with the kit. The optical density was read at 450 nmwith a microplate reader. The intensity of colour developed from theperoxidase-substrate reaction is proportional to the concentration ofhCG in the sample. The measurement was carried out in duplicatewells from three experiments. The intraassay and interassay coeffi-cients of variation were 6.6% (SD 3.0) and 7.8% (SD 4.9), respec-tively. The results are expressed as hCG concentrations (aftercorrection for background optical density) in mIU/100 mg protein inculture media.

Statistical analysis

The experiments were done in duplicate or triplicate wells andrepeated 3–6 times, depending on the analysis. We used one-wayANOVA (analysis of variance) with either Bonferroni’s post hoc testto measure statistical differences between the means of treatment con-ditions versus control conditions, or with Dunnett’s post hoc test forEGF pretreatment versus non-pretreatment. P values of < 0.05 wereconsidered significant.

Effects of forskolin, TNF� and EGF on BeWo and JEG-3

3

Results

Morphological differentiation assay: E-cadherin immunostaining

E-cadherin is a cell adhesion molecule involved in trophoblastdifferentiation and invasion. During the process of differentia-tion of isolated trophoblast cells (as well as BeWo cells treatedwith cAMP agonists), E-cadherin mRNA and protein aredown-regulated in association with loss of E-cadherin stainingfrom the surface of fusing cells (Coutifaris et al., 1991; Getsioset al., 2000). E-cadherin staining was evaluated by two meth-ods: (i) by percentage of E-cadherin-negative cells in a total of500 nuclei; and (ii) by the number of nuclei per syncytium.Cells were considered E-cadherin-negative when total disap-pearance of E-cadherin from cell surfaces was present, indicatingsyncytializing cells. Cells still in the process of syncytializa-tion exhibiting partial E-cadherin staining were not consideredas E-cadherin-negative.

In BeWo cells

Under control conditions, >90% cells aggregated and showedstrong E-cadherin staining at cell boundaries (Fig. 1C). Occa-sionally cells underwent spontaneous syncytialization at 72h—manifested by formation of small multinuclear masses withenlarged clear cytoplasm. Fk alone or in combination withTNFα resulted in disappearance of E-cadherin from cell–cellcontacts, i.e. higher number of E-cadherin-negative nuclei (P <0.01 versus control conditions) (Table IA), together withformation of large, sometimes giant, syncytia containing up to35 nuclei per syncytium (Table IB). Neither TNFα (alone) norEGF pretreatment had any effect on the number of E-cadherin-negative cells or number of nuclei per syncytium.

In JEG-3 cells

In contrast to BeWo cells, JEG-3 cells remained mononuclearwith strong expression of E-cadherin at cell boundaries underall conditions tested (Fig. 1D). Data relating to E-cadherin-negative cells or number of nuclei per syncytium are notshown.

Biochemical differentiation assay: hCG secretion in culture medium

Trophoblast differentiation is associated both in vitro and invivo with increased synthesis and secretion of hCG (Klimanet al., 1986; Morrish et al., 1987; Kao et al., 1988). Choriocar-cinoma cells secrete increasing amounts of hCG into culturemedium in response to cAMP agonists (Wice et al., 1990;Hohn et al., 1998). We measured both whole hCG moleculeand the free β-hCG subunit released in culture media. Resultsare expressed in mIU of hCG per 100 mg protein in collectedmedia.

In BeWo cells

Culture of BeWo cells over time resulted in almost two-foldincrease in hCG secretion (Fig. 2A and B). Addition of Fk sig-nificantly raised hCG levels 2.6 and 3.7 fold at 48 and 72h,respectively, compared with control conditions (P < 0.01).TNFα alone had no effect on hCG secretion, while the combi-nation of Fk/TNFα had a less pronounced and less sustainedeffect than Fk alone. EGF pretreatment had a noticeable stimu-latory effect on hCG secretion in all treatment groups (versusEGF non-pretreatment).

In JEG-3 cells

Secretion of hCG increased over time in JEG-3 cell cultures(1.6 fold) (Fig. 2C and D). Similar to its effect on BeWo cells,Fk increased hCG secretion in JEG-3 cells, as did the combi-nation Fk/TNFα. TNFα alone had no effect. These effectswere noticed in the group not pretreated with EGF at 48 and 72 h.EGF pretreatment increased hCG levels (versus EGF non-pretreated cells) in all treatment groups, but no additional stim-ulatory effect was noticed with TNFα, Fk or the combinationFk/TNFα.

Viability assay: MTT test

In a homogenous sample of cells like the choriocarcinoma celllines, MTT reduction should be proportional to the number ofmetabolically active cells, but is frequently used as a crude

Table I. E-cadherin expression in BeWo cells after 48 h and 72 h of culture

The effects of tumour necrosis factor-α (TNFα), forskolin (Fk) and the combination of both (Fk/TNFα) (versus control conditions) and of epidermal growth factor (EGF) pretreatment (versus EGF non-pretreatment). (A) on the percentage of E-cadherin negative nuclei in a total of 500 nuclei and (B) on the number of nuclei per syncytium. Results are expressed as mean (SD). Statistical significance is indicated by: *P < 0.05, **P < 0.01. It refers to comparisons made with respect to the control condition in the same row using ANOVA with Dunnett’s post hoc test. NS = not significant.

A 48 h 72 h

Control TNFα Fk Fk/TNFα Control TNFα Fk Fk/TNFα

EGF– 7.1(2.5) 11.5(3.0) 32.2(7.8) ** 39.6(11.9) ** 15.4(6.8) 13.7(6.8) 42.1(9.4) ** 39.6(9.4) **EGF+ 15.8(7.0) 19.7(3.1) 28.4(13.3) ** 31.7(11.9) ** 23.4(10.1) 19.7(6.8) 41.8(16.2) ** 36.6(10.3) *EGF+ (versus EGF–) NS NS NS NS NS NS NS NS

B 48h 72h

Control TNFα Fk Fk/TNFα Control TNFα Fk Fk/TNFα

EGF– 2.8 (0.8) 4.0 (2.0) 10.9(3.9) ** 20.6(7.9) ** 6.7(2.5) 5.8(3.1) 15.2(7.7) ** 19.4(7.2) **EGF+ 3.8 (1.6) 5.4 (2.1) 24.7(7.4) ** 27.1(9.6) ** 9.2(3.8) 7.8(4.3) 18.5(6.5) ** 21.4(7.6) **EGF+ (versus EGF–) NS NS NS NS NS NS NS NS

S.Al-Nasiry et al.

4

indicator of cell proliferation. However, this does not take intoaccount the spontaneous rates of proliferation and apoptosis,and the possible effects of cytokines on mitochondrial function.Therefore, we prefer to use the term ‘viability’ and not ‘prolif-eration’ in interpreting results obtained with the MTT test.

In BeWo cells

When BeWo cells are kept in control conditions (i.e. serum-freemedium is added), a significant increase in MTT absorbance isseen over culture time (72 h versus 48 h). Conversely, MTTabsorbance remained basically unchanged over culture time whenFk, TNFα or the combination Fk/TNFα was added to cells (datanot shown). Fk alone or in combination with TNFα decreasedMTT absorbance compared with control conditions both at 48and 72 h (Fig. 3A and B). TNFα alone had a similar deleteriouseffect only at 72 h. In all tested conditions, MTT absorbance wasunaffected by EGF– pretreatment at either points in time.

In JEG-3 cells

In contrast to BeWo cells, JEG-3 cells showed important dif-ferences in their response to the same effectors. Treatment ofJEG-3 cells with Fk alone or in combination with TNFαincreased MTT absorbance compared with cells receiving controlmedium. This effect was very significant both at 48 and 72h(Fig. 3C and D). Addition of TNFα alone did not significantly

affect cell viability, neither did pretreatment with EGF in alltested conditions.

Proliferation assay: MIB-1 immunostaining

The use of the monoclonal antibody MIB-1 to determine cellproliferation is well established. This antibody reacts withrecombinant parts of the Ki-67 nuclear antigen present in allactively proliferating cells including trophoblasts (Cheunget al., 1994). Positive cells showed generally a red nuclearstaining together with frequent occurrence of mitotic figures(Fig. 1E and F).

In BeWo cells

In control conditions, 4% of cells showed MIB-1 staining(Fig. 4A and B). Treatment with TNFα reduced the number ofMIB-1 positive cells (P < 0.01 versus control conditions). Thenumber of MIB-1 positive cells was also severely reduced withFk and Fk/TNFα treatment with <0.7% and 0.4%, respec-tively, of cells still showing MIB-1 staining at 72 h. EGF pre-treatment had no effect on the number of MIB-1 positive cells.

In JEG-3 cells

In sharp contrast to its effects on the BeWo cells, Fk increasedthe number of MIB-1 positive cells in JEG-3 cells (Fig. 4C and D).

Figure 1. Immunocytochemical staining of BeWo (A,C,E) and JEG-3 (B,D,F) cells. Primary monoclonal antibodies were used against the M30epitope of cytokeratin 18 (A,B), against human E-cadherin (C,D) and against Ki67 epitope (MIB-1 clone) (E,F). Red staining represents the3,amino-9-ethylcarbazole (AEC) detection of peroxidase activity, counterstained with Mayer’s haematoxylin.

Effects of forskolin, TNF� and EGF on BeWo and JEG-3

5

However, this effect was only evident at 72 h of culture. Nei-ther treatment with TNFα or Fk/TNFα nor pretreatment withEGF appeared to alter the proliferation of JEG-3 cells.

Apoptosis assay: M30 immunostaining

During the early stages of apoptosis in epithelial cells, a neo-epitope on cytokeratin 18 is formed by caspase 3 cleavage and

reacts with the monoclonal antibody M30 (Caulin et al., 1997;Leers et al., 1999; Morsi et al., 2000). M30 immunostaining issuperior to the TUNEL method in detecting apoptotic cells(Kadyrov et al., 2001; Austgulen et al., 2002). Non-epithelialand necrotic cells show a negative M30 staining (Morsi et al.,2000; Kadyrov et al., 2001). M30-positive cells showeda homogeneous red staining of the cytoplasmic filaments(Fig. 1A and B). They also showed other morphological signs

Figure 2. hCG secretion in conditioned media of BeWo (A,B) and JEG-3 (C,D) cells at 48 h (A,C) and 72 h (B,D), showing the effects of addi-tion of forskolin (Fk, 100 μM), tumour necrosis factor-α (TNFα, 1000 IU/ml) and the combination Fk/TNFα and pretreatment with EGF. Resultsare expressed as mIU of hCG/100 mg protein (mean ±SD, n = 3). §P < 0.05, §§P < 0.01 (versus control);*P < 0.05, **P < 0.01,***P < 0.001(versus EGF–). Note that different scales in the y-axis are used for the two cell types.

Figure 3. MTT reduction in BeWo (A,B) and JEG-3 (C,D) cells at 48h (A,C) and 72h (B,D), showing the effects of addition of forskolin (Fk,100 μM), tumour necrosis factor-α (TNFα, 1000 IU/ml) and the combination Fk/TNFα and pretreatment with EGF. Results are expressed as per-centage of MTT reduction to control conditions (mean ± SD, n = 6). §P < 0.05, §§P < 0.01 (versus control). Note that different scales in the y-axisare used for the two cell types.

S.Al-Nasiry et al.

6

of apoptosis including chromatin condensation and nuclearfragmentation.

In BeWo cells

Under control conditions, the cells showed ∼0.4% rate of apop-tosis, based on the number of M30-positive cells (Fig. 5A and B).Apoptosis rates were slightly increased over culture time (72 hversus 48 h). When Fk alone or in combination with TNFα wasadded to the cells, the apoptosis rates increased up to 2.3 and6.6%, respectively, at 72 h (P < 0.01 versus control condi-tions). TNFα alone increased apoptosis rates only at 72 h(2.6%). EGF had no effect.

In JEG-3 cells

JEG-3 cells had 10× higher apoptosis rates (4%) than BeWocells (Fig. 5C and D). At 72 h of culture, TNFα increasedapoptosis rates to 6.4% (P < 0.05 versus control conditions).The combination Fk/TNFα increased apoptosis rates further to10% (P < 0.01 versus control conditions). Fk alone had noeffect. Unlike BeWo cells, EGF pretreatment completelyblocked the TNFα- and Fk/TNFα-induced apoptosis at bothtime points (P < 0.001 versus EGF non-pretreated).

The results are summarized in Table II.

Discussion

Placental cytokines and growth factors acting at the fetoma-ternal interface including EGF, TNFα, transforming growthfactor (TGF)-β, insulin-like growth factors (IGF) and placen-tal growth factor (PlGF) are suggested to play a role in regu-lating various trophoblast functions. The pathways ofdifferentiation followed by placental trophoblasts are at least

partly regulated by these cytokines (Morrish et al., 1998).Our knowledge on the subject has been largely based onexperiments with isolated trophoblast cells from early andterm placentas.

However, this widely used model has certain limitations.The obtained cells represent a heterogeneous population ofcyto- and syncytiotrophoblasts with possible contaminationwith fibroblasts and macrophages. The use of established celllines in the study of trophoblast function have been advocatedby many researchers (Ringler and Strauss, 1990; King et al.,2000; Shiverick et al., 2001) and recently reviewed by Sullivan(2004). The homogeneity of these cells and the practicality ofin vitro propagation procedures are important advantages.Particularly in the study of apoptosis, immortalized cell linescould be preferred to primary trophoblasts because of their lowspontaneous rate of apoptosis (Gaus et al., 1997). Here, weutilized two choriocarcinoma cell lines with differing fusigeniccapacity as a model to study trophoblast proliferation, differ-entiation and apoptosis in response to various biologicalinducers.

The disappearance of E-cadherin protein from cell–cellboundaries, parallel to down-regulation of its mRNA, has beenused as an indication of syncytium formation in normal tro-phoblasts and in BeWo cells (Wice et al., 1990; Coutifariset al., 1991). This has been confirmed by immunoneutraliza-tion studies using function-perturbing antibodies to E-cadherin.In concordance with these studies, E-cadherin immunostainingshowed that BeWo cells differentiate morphologically to formmultinuclear syncytia in response to forskolin. Forskolin treat-ment also increased hCG secretion in culture medium indicatingbiochemical differentiation. In contrast, JEG-3 cells showed nosigns of morphological differentiation with forskolin treatment,

Figure 4. Quantification of MIB1-positive nuclei in BeWo (A,B) and JEG-3 (C,D) cells at 48 h (A,C) and 72 h (B,D), showing the effects ofaddition of forskolin (Fk, 100 μM), tumour necrosis factor-α (TNFα, 1000 IU/ml) and the combination Fk/TNFα and pretreatment with EGF.Results are expressed as number of MIB1-positive nuclei in a total of 500 nuclei (mean ±SD, n = 3). §P < 0.05, §§P < 0.01 (versus control). Notethat different scales in the y-axis are used for the two cell types.

Effects of forskolin, TNF� and EGF on BeWo and JEG-3

7

although they showed biochemical differentiation evident byincreased hCG secretion. This dichotomous effect of forskolinon differentiation suggests that at least some aspects of tro-phoblast differentiation can be dissociated. A similar dissocia-tion was also obtained in trophoblast cultured in the absence ofserum (Kao et al., 1988). It is interesting to speculate that analo-gous mechanisms for this dissociation exist in the two models.

Induction of cAMP is associated with decreased cell prolif-eration in normal trophoblast as well as BeWo cells (Taylor

et al., 1991). The decreased viability of BeWo cells in ourstudy could also be explained by rapid syncytialization inducedby high dose (100 μM) forskolin, which would imply accelera-tion of the life cycle ending by apoptosis of syncytializedBeWo nuclei. The presented data show that forskolin, aninducer of cAMP, had also differential effects on cell viability(MTT reduction) and proliferation (MIB-1 immunostaining),both decreasing in BeWo and increasing in JEG-3 cells. Ourdata suggest that the divergent effect of cAMP on BeWo and

Figure 5. Quantification of M30-positive nuclei in BeWo (A,B) and JEG-3 (C,D) cells at 48h (A,C) and 72 h (B,D), showing the effects of addi-tion of forskolin (Fk, 100 μM), tumour necrosis factor-α (TNFα, 1000 IU/ml) and the combination Fk/TNFα and pretreatment with EGF. Resultsare expressed as number of M30-positive nuclei in a total of 500 nuclei (mean ±SD, n = 3). §P < 0.05, §§P < 0.01 (versus control); ***P < 0.001(versus EGF–). Note that different scales in the y-axis are used for the two cell types.

Table II. Summary of results with BeWo and JEG-3 cells

The differential effects of tumour necrosis factor-α (TNFα), forskolin (Fk) and the combination of both (Fk/TNF) (versus control conditions) and of EGF pretreat-ment (versus EGF non-pretreatment) in BeWo and JEG-3 cells after 72 h of culture. Single, double and triple arrows refer to magnitude of probability and represent P < 0.05, P < 0.01 and P < 0.001, respectively. Formation of E-cadherin negative cells was confined to BeWo cells, while secretion of hCG was promoted by for-skolin in both cell types. Forskolin also had reversed effects on MTT reduction, MIB-1 positive cells and M30 positive cells in BeWo and JEG-3 cells. TNFα decreased MIB-1 positive cells in BeWo cells only. TNFα also increased M30 positive cells in both cell types, an effect which was blocked by EGF selectively in JEG-3 cells. NS = not significant.

BeWo JEG

Control TNFα Fk Fk/TNFα Control TNFα Fk Fk/TNFα

E-cadherin (–) cells EGF– (versus control) NS ↑↑ ↑↑ NS NS NSEGF+ (versus control) NS ↑↑ ↑ NS NS NSEGF+ (versus EGF–) NS NS NS NS NS NS NS NS

hCG secretion EGF– (versus control) NS ↑↑ NS NS ↑↑ ↑EGF+ (versus control) NS ↑↑ NS NS NS NSEGF+ (versus EGF–) ↑↑ ↑ ↑↑↑ ↑↑↑ NS NS NS NS

MTT reduction EGF– (versus control) ↓↓ ↓↓ ↓↓ NS ↑↑ ↑↑EGF+ (versus control) ↓ ↓↓ ↓↓ NS ↑↑ ↑↑EGF+ (versus EGF–) NS NS NS NS NS NS NS NS

MIB-1 (+) cells EGF– (versus control) ↓↓ ↓↓ ↓↓ NS ↑ NSEGF+ (versus control) ↓↓ ↓↓ ↓↓ NS ↑ NSEGF+ (versus EGF–) NS NS NS NS NS NS NS NS

M30 (+) cells EGF– (versus control) ↑↑ ↑↑ ↑↑ ↑ NS ↑↑EGF+ (versus control) ↑↑ ↑↑ ↑↑ NS NS NSEGF+ (versus EGF–) NS NS NS NS NS ↓↓↓ NS ↓↓↓

S.Al-Nasiry et al.

8

JEG-3 cell lines could be related to their fusigenic capacity. Itis tempting to speculate that the lack of fusigenic capacity inJEG-3 cells may be partly responsible for the altered responseto cAMP observed in our study and may result in switching thecellular programmes to a more proliferative type.

TNFα is a pro-inflammatory cytokine with a wide range ofcellular actions affecting proliferation, apoptosis and differen-tiation of various cell types. In trophoblast cell lines, theseeffects depend on cell type, dose of TNFα and culture condi-tions. In JEG-3 cells, TNFα had a less wide-spectrum effectthan in BeWo cells, confined to increasing apoptosis in theEGF non-pretreated group. This concurs with findings by Yanget al. (1993) that JEG-3 cells produce less TNFα and are lessreliant on this cytokine for their growth than other choriocarci-noma cells. On the other hand, BeWo cells showed decreasedviability and proliferation with TNFα treatment in addition toincreased apoptosis rates resembling in this manner theresponse to TNFα in the normal trophoblast culture (Garcia-Lloret et al., 1996). Since both normal trophoblast and BeWocells are fusigenic, one can speculate that the full-rangeresponse to TNFα particularly on viability and proliferationmay be related to the fusigenic capacity. Likewise, the fusi-genic capacity of invasive trophoblasts in the placental bedmight determine their response to TNFα induced-apoptosis asa possible mechanism of restricting their invasion. Currently,we are studying the role of fusion proteins in formation of tro-phoblast giant cells and the interaction with TNFα at the pla-cental bed of normal and pathological pregnancies.

Pretreatment of cell cultures with EGF increased hCGsecretion in culture media in both cell lines implying biochem-ical differentiation. However, EGF pretreatment was able toblock the TNFα-induced apoptosis only in JEG-3, but notBeWo cells, resembling in this aspect its effect on normal tro-phoblasts (Pijnenborg et al., 2000). Our experimental designdid not allow us to investigate whether this block was a directeffect of EGF on TNFα binding/signal transduction or second-ary to its differentiation-enhancing effect.

The lack of protective effects of EGF against TNFα-inducedcytotoxicity in BeWo cells is a novel finding. Although bothnormal and malignant trophoblasts produce and express recep-tors for EGF (Duello et al., 1994), BeWo cells have at least 10-fold higher levels of EGF receptor (EGF-R) than benign tro-phoblasts with functional tyrosine kinase activity (Filla andKaul, 1997). The over-expression of EGF-R in BeWo cells canrender them unresponsive to exogenous EGF (Filla and Kaul,1997), which might explain the lack of protective effect ofEGF in our BeWo model.

The differential response of BeWo and JEG-3 cell to otherinducers has been described previously (Mandl et al., 2002). Inaddition, differences in gene expression profiles between thesechoriocarcinoma cell lines and normal trophoblasts cells havebeen described recently using the technique of differential dis-play RT-PCR (Garcia and Castrillo, 2004). These findings sup-port the notion that these cell lines can serve as ‘specific’models to study certain defects in trophoblast function and thatcaution is necessary in extrapolating results obtained with onetrophoblast cell line to other cell lines or to the ‘general’ tro-phoblast population.

It is tempting to view the observed differential responses ofchoriocarcinoma cell lines in the context of trophoblast aberra-tions described in pre-eclampsia. The behaviour of BeWo cellsin our model resembles that of cultured placental villi frompre-eclamptic pregnancies described by Crocker et al. (2004)in at least two aspects. First, both BeWo cells and pre-eclampticvilli showed higher hCG secretion and exaggerated apoptoticresponse to TNFα. compared with JEG-3 cells and normalvilli. Secondly, while both JEG-3 cell in our model and villifrom normotensive placentas showed increased proliferationrates in response to forskolin and hypoxia respectively, BeWocells and preeclamptic villi showed restricted proliferationresponse to forskolin and hypoxia respectively. The linkbetween fusion defects in the trophoblast and the pathologyof pre-eclampsia is still speculative. Further studies are neededto clarify mechanisms of altered cell kinetics in choriocarci-noma cell lines before relating the observed changes to clinicalsituations.

In conclusion, this study demonstrates the differentialresponses of two choriocarcinoma cell lines with different fusi-genic capacity to inducers of syncytialization and apoptosis,suggesting that fusigenic capacity might play a role in deter-mining the responses to biological inducers in the selection ofcellular pathways. Whilst highlighting the need for cautionwhen extrapolating results obtained by these models to normaltrophoblast populations, this study endorses the use of thesecell lines as specific models in the search for novel defects insyncytialization. This could give us insight into the mecha-nisms controlling the selection of different pathways of tro-phoblast differentiation in normal and pathologic pregnancies.

References

Austgulen R, Chedwick L, Vogt I C, Vatten L and Craven C (2002) Trophoblastapoptosis in human placenta at term as detected by expression of a cytokeratin18 degradation product of caspase. Arch Pathol Lab Med 126,1480–1486.

Caulin C, Salvesen GS and Oshima RG (1997) Caspase cleavage of keratin 18and reorganization of intermediate filaments during epithelial cell apoptosis.J Cell Biol 138,1379–1394.

Cheung ANY, Ngan HYS, Collins RJ and Wong YL (1994) Assessment of cellproliferation in hydatidiform mole using monoclonal antibody Mib1 to Ki-67 antigen. J Clin Pathol 47,601–604.

Coutifaris C, Kao LC, Sehdev HM, Chin U, Babalola GO, Blaschuk OW andStrauss JF III (1991) E-cadherin expression during the differentiation ofhuman trophoblasts. Development 113,767–777.

Crocker IP, Tansinda DM and Baker PN (2004) Altered cell kinetics in cul-tured placental villous explants in pregnancies complicated by pre-eclampsiaand intrauterine growth restriction. J Pathol 204,11–18.

Duello TM, Bertics PJ, Fulgham DL and Vaness PJ (1994) Localization ofepidermal growth factor receptors in first trimester and third trimesterhuman placentas. J Histochem Cytochem 42,907–915.

Filla MS and Kaul KL (1997) Relative expression of epidermal growth factorreceptor in placental cytotrophoblasts and choriocarcinoma cell lines.Placenta 18,17–27.

Garcia J and Castrillo JL (2004) Differential display RT-PCR analysis ofhuman choriocarcinoma cell lines and normal term trophoblast cells: identi-fication of new genes expressed in placenta. Placenta 25,684–693.

Garcia-Lloret MI, Yui J, Winkler-Lowen B and Guilbert LJ (1996) Epidermalgrowth factor inhibits cytokine-induced apoptosis of primary humantrophoblasts. J Cell Physiol 167,324–332.

Gaus G, Funayama H, Huppertz B, Kaufmann P and Frank HG (1997) Parentcells for trophoblast hybridization I: Isolation of extravillous trophoblastcells from human term chorion laeve. Trophoblast Res 10,181–190.

Getsios S, Chen GT and MacCalman CD (2000) Regulation of beta-cateninmRNA and protein levels in human villous cytotrophoblasts undergoing

Effects of forskolin, TNF� and EGF on BeWo and JEG-3

9

aggregation and fusion in vitro: correlation with E-cadherin expression.J Reprod Fertil 119,59–68.

Hohn HP, Linke M, Ugele B and Denker HW (1998) Differentiation markersand invasiveness: discordant regulation in normal trophoblast and chorio-carcinoma cells. Exp Cell Res 244,249–258.

Kadyrov M, Kaufmann P and Huppertz B (2001) Expression of a cytokeratin18 neo-epitope is a specific marker for trophoblast apoptosis in humanplacenta. Placenta 22,44–48.

Kao LC, Caltabiano S, Wu S, Strauss JF III and Kliman H J (1988) The humanvillous cytotrophoblast: interactions with extracellular matrix proteins,endocrine function, and cytoplasmic differentiation in the absence of syncy-tium formation. Dev Biol 130,693–702.

King A, Thomas L and Bischof P (2000) Cell culture models of trophoblast II:trophoblast cell lines–a workshop report. Placenta 21 Suppl A, S113–S119.

Kliman HJ, Nestler JE, Sermasi E, Sanger JM and Strauss JF III (1986) Purifi-cation, characterization, and in vitro differentiation of cytotrophoblasts fromhuman term placentae. Endocrinology 118,1567–1582.

Kohler PO and Bridson WE (1971) Isolation of hormone-producing clonallines of human choriocarcinoma. J Clin Endocrinol Metab 32,683–687.

Leers MP, Kolgen W, Bjorklund V, Bergman T, Tribbick G, Persson B,Bjorklund P, Ramaekers FC, Bjorklund B, Nap M, et al. (1999) Immunocy-tochemical detection and mapping of a cytokeratin 18 neo- epitope exposedduring early apoptosis. J Pathol 187,567–572.

Mandl M, Haas J, Bischof P, Nohammer G and Desoye G (2002) Serum-dependent effects of IGF-I and insulin on proliferation and invasion ofhuman first trimester trophoblast cell models. Histochem Cell Biol117,391–399.

Morrish DW, Bhardwaj D, Dabbagh LK, Marusyk H and Siy O (1987) Epider-mal growth factor induces differentiation and secretion of human chorionicgonadotropin and placental lactogen in normal human placenta. J Clin Endo-crinol Metab 65,1282–1290.

Morrish DW, Dakour J and Hongshi L (1998) Functional regulation of humantrophoblast differentiation. J Reprod Immunol 39,179–195.

Morsi HM, Leers MP, Jager W, Bjorklund V, Radespiel-Troger M, el Kabarity H,Nap M and Lang N (2000) The patterns of expression of an apoptosis-related CK18 neoepitope, the bcl-2 proto-oncogene, and the Ki67 prolifera-tion marker in normal, hyperplastic, and malignant endometrium. Int JGynecol Pathol 19,118–126.

Mosmann T (1983) Rapid colorimetric assay for cellular growth and survival:application to proliferation and cytotoxicity assays. J Immun Meth 55–63.

Pattillo RA and Gey GO (1968) The establishment of a cell line of humanhormone-synthesizing trophoblastic cells in vitro. Cancer Res 28,1231–1236.

Pijnenborg R, Bland JM, Robertson WB, Dixon G and Brosens I (1981) Thepattern of interstitial trophoblastic invasion of the myometrium in earlyhuman pregnancy. Placenta 2,303–316.

Pijnenborg R, McLaughlin PJ, Vercruysse L, Hanssens M, Johnson PM, KeithJC Jr and Van Assche FA (1998) Immunolocalization of tumour necrosisfactor-alpha (TNF-alpha) in the placental bed of normotensive and hyper-tensive human pregnancies. Placenta 19,231–239.

Pijnenborg, R, Luyten C, Vercruysse L, Keith JC Jr and Van Assche FA(2000) Cytotoxic effects of tumour necrosis factor (TNF)-alpha and inter-feron- gamma on cultured human trophoblast are modulated by fibronectin.Mol Hum Reprod 6,635–641.

Reister F, Frank HG, Kingdom JC, Heyl W, Kaufmann P, Rath W and Huppertz B(2001) Macrophage-induced apoptosis limits endovascular trophoblast inva-sion in the uterine wall of preeclamptic women. Lab Invest 81,1143–1152.

Ringler GE and Strauss JF III (1990) In vitro systems for the study of humanplacental endocrine function. Endocr Rev 11,105–123.

Shiverick KT, King A, Frank H, Whitley GS, Cartwright JE and Schneider H(2001) Cell culture models of human trophoblast II: trophoblast cell lines–aworkshop report. Placenta 22 Suppl A, S104–S106.

Sullivan MHF (2004) Endocrine cell lines from the placenta. Mol Cell Endo-crinol 228,103–119.

Taylor RN, Newman ED and Chen SA (1991) Forskolin and methotrexateinduce an intermediate trophoblast phenotype in cultured human choriocar-cinoma cells. Am J Obstet Gynecol 164,204–210.

Wice B, Menton D, Geuze H and Schwartz AL (1990) Modulators of cyclicAMP metabolism induce syncytiotrophoblast formation in vitro. Exp CellRes 186,306–316.

Yang Y, Yelavarthi KK, Chen HL, Pace JL, Terranova PF and Hunt JS (1993)Molecular, biochemical, and functional characteristics of tumor necrosisfactor-alpha produced by human placental cytotrophoblastic cells. J Immu-nol 150,5614–5624.

Submitted on May 9, 2005; resubmitted on June 27, 2005; accepted on July 20,2005