Placenta (2003), 24, 155–163doi:10.1053/plac.2002.0890

Regulation of Trophoblastic Gelatinases by Proto-oncogenes

P. Bischofa, K. Truong and A. Campana

Department of Obstetrics and Gynaecology, University of Geneva, Maternite, 1211 Geneva 14, Switzerland

Paper accepted 13 September 2002

During the first trimester of pregnancy, certain cytotrophoblastic cells (CTB) of anchoring villi invade the underlying decidua.Regulation of this invasive behaviour depends on cytokines and growth factors secreted by decidua and trophoblast, whichmodulate metalloproteinase (MMP) secretion of CTB. Since MMP-9 expression by CTB is a prerequisite for matrigel invasionand since the promoter region of the MMP-9 gene contains two AP-1 binding sites, we hypothesized, that transient activation ofc-jun and c-fos oncogenes (which bind to form AP-1) by tumour necrosis factor (TNF�), or the phorbol ester TPA will promotethe invasive phenotype of CTB and induce the production of MMP-9.

TNF� or TPA when added to primary cultures of CTB increase MMP-9 activity and MMP-9 mRNA. This effect is inhibitedby cycloheximide indicating the necessity of protein synthesis. TPA or TNF� induces also the binding of nuclear proteins(extracted from treated CTB) to a radiolabelled oligonucleotide corresponding to the consensus sequence of the TPA responsiveelement. Antibodies to Jun and Fos can displace this binding. Transient transfection of antisense mRNA to jun or fos into CTBinhibits the immunoreactivity and gelatinolytic activity of MMP-9.

We conclude that AP-1 is necessary but may not be sufficient for transactivation of the MMP-9 gene in human CTB.Placenta (2003), 24, 155–163 � 2002 Elsevier Science Ltd. All rights reserved.

a To whom corresponding should be addressed. Tel.: 41 22 382 43 36;Fax: 41 22 382 43 10; Email: paul.bischof@hcuge.ch

INTRODUCTION

Cytotrophoblastic cells (CTB) are derived from the trophec-todermal cells of the blastocyst and represent a heterogeneouspopulation during early pregnancy. CTB follow one of twoexisting differentiation pathways. Villous CTB (vCTB) form amonolayer of polarized epithelial stem cells which proliferateand eventually differentiate by fusion to form a syncytiotro-phoblast (STB) covering the entire surface of the placentalvillous, or they can break through the STB at selected sites toform multilayered columns of non-polarized but invasiveCTB. These motile and highly invasive extravillous CTB(evCTB) are found as cytokeratin positive cells in the decidua,the intima of the uterine spiral arteries and the proximal thirdof the myometrium.

Trophoblast invasion, like tumour invasion is due to anactive secretion of proteolytic enzymes capable of digestingextra-cellular matrices (ECM) of the host’s tissues. Serineproteases, cathepsins and metalloproteinases have been impli-cated in invasive processes. Matrix metalloproteinases (MMP)also called matrixins form a family of about 26 humanzinc-dependent endopeptidases collectively capable of degrad-ing essentially all components of the ECM. According to theirsubstrate specificity and structure, members of the MMP genefamily can be classified into 5 subgroups (Bischof and

0143–4004/03/$-see front matter

Campana, 2000; Matrisian, 2000). Gelatinases (MMP-2 andMMP-9) digest collagen type IV (the major constituent ofbasement membranes) and denatured collagen (gelatin). Colla-genases (MMP-1, 8, 13) digest collagen types I, II, III, VII andX. They are thus appropriately designed for digesting thecollagens of the ECM of the interstitium. Stromelysins(MMP-3, 7, 10, 11 and 12) have a relatively broad substratespecificity and digest collagen type IV, V, VII as well aslaminin, fibronectin, elastin, proteoglycans and gelatin. Thesubstrate of the membrane-bound metalloproteinases (MMP-14, 15, 16, 17, 24, 25) is essentially proMMP-2 and theseenzymes allow activation of proMMP-2 at the cell surface onthe invasive front. It must be mentioned however that some ofthese MT-MMPs can also digest fibrillar collagen. There isalso a subfamily of MMPs provisionally classified as ‘others’(MMP-18 to MMP-23 and MMP-26) because their charac-terization is not yet completed. Most MMPs are secreted asinactive proenzymes (proMMPs) which become activated inthe extracellular compartments with the exception of MMP-11and MT-MMPs. Several enzymes are capable of activating thepromatrixins such as plasmin, stromelysins and MT-MMPs.The activity of MMPs in the extracellular space is specificallyinhibited by Tissue Inhibitor of Metalloproteinases (TIMP)which binds to the highly conserved zinc-binding site of activeMMPs at molar equivalence. The TIMP gene family consistsof four structurally related members: TIMP-1, -2, -3 and -4(Westermarck and Veli-Matti, 1999).

� 2002 Elsevier Science Ltd. All rights reserved.

156 Placenta (2003), Vol. 24

Although MMPs digest the extracellular matrix theyare probably not the only enzymes capable or regulatingtrophoblast invasion. Indeed, recent data obtained in humans(Graham et al., 1998) and in rhesus monkey attribute to theplasmin/plasminogen activator system a more importantrole than just activating MMPs. Besides regulating extra-cellular matrix degradation, these enzymes and their inhibitorsmodify also migration and invasive behaviour of extravillousCTB (Graham and Lala, 1992; Hu et al., 1999; Feng et al.,2000).

In vitro, CTB invade a reconstituted basement membrane(Matrigel). This invasive behaviour is due to the ability ofCTB to secrete MMPs since TIMP inhibits their invasiveness(Librach et al., 1991). All MMPs are not equally important fortrophoblast invasion. Gelatinase B (MMP-9) and gelatinase A(MMP-2) have been shown to mediate invasion of CTB or aCTB cell line (HTR-8) into Matrigel (Librach et al., 1991;Lala and Connelly, 1994; Bischof et al., 1995a). Although CTBbehave like metastatic cells, in vivo they are only transientlyinvasive (first trimester of pregnancy) and invasion is normallylimited to the endometrium and to the proximal third of themyometrium (Pijnenborg et al., 1980). This temporal andspatial regulation of trophoblast invasion is believed to bemediated in an autocrine way by trophoblastic factors and in aparacrine way by uterine factors (Bischof and Campana, 2000).Several factors were investigated: hormones, cytokines, growthfactors and ECM glycoproteins exert regulatory effects in vitroon trophoblastic MMP secretion and/or trophoblast invasion(Librach et al., 1994; Meisser et al., 1999a,b; Bischof andCampana, 2000; Chakraborty et al., 2002).

Expression of most MMPs is normally low in tissues and isinduced when ECM remodelling is required. MMP geneexpression is primarily regulated at the transcriptional level.The promoter regions of inducible MMP genes (MMP-1, -3,-7, -9, -10, -12, -13) show remarkable conservation of regulat-ory elements (Westermarck and Veli-Matti, 1999). One or two12-O-tetradecanoyl phorbol 13-acetate (TPA) responsiveelements (TRE) are found in the promoter region of eachinducible MMP gene. TRE is a DNA consensus sequence thatbinds the Activator Protein-1 (AP-1) transcription factorsfamily. AP-1 transcription factors are dimeric leucine zipperproteins usually formed by binding of members of the Junfamily (c-Jun, Jun B, Jun D) to members of the Fos family(c-Fos, Fra-1, Fra-2, Fos B), although Jun-Jun dimers alsooccur. Jun and fos are oncogene products and their genesbelong to the family of early response genes.

MMP-9 expression is a prerequisite for matrigel invasion byhuman CTB (Bischof et al., 1995a; Bischof and Campana,2000). Since the promoter region of the MMP-9 gene containstwo AP-1 binding sites and since TPA (Westermarck andVeli-Matti, 1999) and tumour necrosis factor alpha (TNF�,Brenner et al., 1989) are AP1 inducers, we hypothesized thattransient activation of c-jun and c-fos oncogenes in CTB byTNF�, or TPA will promote the invasive phenotype of CTBand induce the production of MMP-9. Here we show thatprimary cultures of human first trimester CTB secrete

MMP-9 but not MMP-2 in response to TPA or TNF� andthat this response occurs through an increased expression ofc-jun and c-fos oncogenes.

MATERIAL AND METHODS

Material

RPMI and DMEM media, Gentamicin, Amphoptericin-B,-glutamin, microplates, Maxisorb F16 (Nunc), foetal calfserum (FCS) and trypsin were from Life Technologies, Basel,Switzerland. Penicillin was from Hoechst-Pharma, Zurich,Switzerland, Streptomycin from Grunenthal, Stolberg,Germany. Clostridium histoliticum Collagenase (EC 3.4.24.3,330 U/mg), HEPES, 12-O-tetradecanoyl phorbol 13-acetate(TPA), cycloheximide, bovine serum albumin and trypan bluewere all from Sigma, Buchs, Switzerland. Antibodies to CD45or C3, were from Dako Diagnostics AG, Zug, Switzerlandwhereas polyclonal antibodies to Jun and Fos were fromOncogene Research, Stehelin AG, Basel, Switzerland. Oligo-nucleotides were purchased from Microsynth, Balgach,Switzerland. Recombinant tumour necrosis factor alpha(TNF�) was from R&D Systems, Buehlmann AG, BaselSwitzerland. Nylon Hybond M membranes were purchasedfrom Amersham, Buckinghamshire, UK. Restriction enzymes:Bam H1, xba I and EcoRI, Digoxigenin-UTP (DIG), DIGeasy Hyb, DIG nucleic acid detection kit were all from,Boehringer, Mannheim, Germany. X-OMAT AR films werefrom Kodak, Rochester, USA.

Cell preparation and cultures

CTB were isolated, purified and cultured as previouslydescribed (Bischof et al., 1991). Briefly, most trophoblastic villiobtained from legal abortions (6–12 weeks of pregnancy) werebetween 7 and 9 weeks of pregnancy (n=21, 2 were at 10 weeksand one at 11 weeks). The minced trophoblastic villi weredigested by trypsin and CTB were separated from blood cellsand syncytia on a discontinuous Percoll gradient. Contaminat-ing leukocytes were removed by immunopurification with anantibody to CD45 coupled to magnetic particles. These puri-fied CTB which represent a mixture of villous and extravillouscells (70 and 30 per cent respectively, Bischof et al., 1995b)were counted in a Neubauer cell in presence of Trypan Blueand diluted to 107 or 106 cells/ml.

Cells (2�105 or 106 cells/wells) were cultured overnight inDMEM containing 2 m -glutamin, 4.2 m magnesiumsulphate, 2.5 m HEPES, 1 per cent gentamycin, 1 per centamphoptericin-B, 100 �g/ml streptomycin and 100 U/ml.The next morning (day 0), medium was changed and the cellsincubated in the presence or the absence of 100 ng/ml ofTNF� or TPA with or without cycloheximide (50 �g/ml).Incubation was performed under a 5 per cent CO2 and 95 percent air atmosphere in a humid incubator at 37�C and the

Bischof et al.: Gelatinases and Proto-oncogenes 157

culture stopped at different periods of time (2–48 h). Thesupernatants and the cells were separated and stored at �20�Cuntil assayed. Each experiment was repeated at least 3 timeswith different CTB preparations and duplicates of each culturecondition were used throughout the study.

Table 1. Description of oligonucleotides used for transient transfection of CTB

Oligonucleotide Sequence Description

Jun193 GAGCCAATGGGAAGGCCT Non translated sequence 193 to 210: antisense JunJun contr CTCCGGTTACCCTTCCGGA Non translated sequence 193 to 210: sense JunJun 1261 ATGACTGCAAAGATGGAAAC Beginning of coding region: 1261–1280, antisense JunFos 1183 GACTTCTGCACGGACCTG Beginning of exon II, 1183–1200, antisense FosFos contr CTGAAGACGTGCCTGGAC Beginning of exon II, 1183–1200, sense FosFos 2088 GAGACAGACCAACTAGAAGATGA Beginning of exon IV, 2088–2110, antisense Fos

Figure 1. Upper panel: representative zymogram of CTB supernatants cultured for 40 h in presence or absence of TPA (100 ng/ml). Note the presence of proMMP-9 and pro MMP-2 (pMM-9 and pMMP-2 respectively). Molecular weight markers are given in kilo daltons (kDa). Lower panel: gelatinolytic activity ofCTB supernatants incubated for 24 h with increasing concentrations of TPA. Error bars represent of three different experiments run in duplicates. P valuesrefer to differences compared to cells not treated with TPA (ANOVA).

MMPs assays

Zymography was performed as previously described (Martelliet al., 1993). Zymograms were scanned in an ‘Apple One-scanner’ and the surface of the digestion bands measured by

158 Placenta (2003), Vol. 24

the NIH Image 1.60 program on a Power Macintosh 7100/66computer. All zymograms were evaluated using the samepre-set standards. Quantitative estimation of total (MMP-2+MMP-9) gelatinolytic activity was performed by measuringthe degradation of heat-denatured 3H-collagen type IV using amethod already reported by us (Bischof et al., 1995b). Thestandard curve was built by using collagenase from Clostridiumhistolyticum and ranged from 0.8 to 50 ng/ml (0.26–16.5 U/ml). Concentrations of immunoreactive MMP-2 and MMP-9were measured by ELISA as described previously (Meisseret al., 1999a).

Figure 2. Left panel: representative Northern blot of MMP-2, MMP-9 and GAPDH. CTB were incubated (or not) for 24 h with TPA (100 ng/ml). Lanes Aand B were hybridized with MMP-2 cDNA and lanes C and D were hybridized with MMP-9 cDNA. Right panel: measurements of messenger RNA (in per centof GAPDH) for MMP-2 (upper) and MMP-9 (lower) in CTB incubated in presence (dotted lines) or absence (full line) of TPA. Error bars represent of fiveexperiments run with different CTB preparations.

Northern blots

Total RNA was extracted from 10 million purified andcultured CTB by guanidinium thiocyanate followed bycentrifugation in cesium chloride as described elsewhere(Sambrook et al., 1989). Extracted RNA (10 �g/culture con-dition) was denatured in Glyoxal/Dimethyl Sulfoxide mix andelectrophoresed in sodium phosphate (10 m, pH 7.0) at3 V/cm on an 1 per cent agarose gel as described (Sambrook

et al., 1989). Denatured RNA was transferred immediatelyafter electrophoresis to nylon Hybond N membranes bycapillary elution overnight at room temperature.

Human MMP-2 and MMP-9 cDNA (pK-191 and pHG-1,2654 bp, 1.7 kb respectively) were a generous gifts from DrKarl Tryggvason (University of Oulu, Finland, Huhtala et al.,1990, 1991). MMP-2 cDNA was cloned into the EcoRI site ofa pBR322 vector whereas MMP-9 cDNA was cloned into theEcoRI site of a pUC19 vector. Mouse glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA (1220 bp) clonedin a bluescript KS vector was a generous gift from Prof. P.Sappino (Dept Oncology, University of Geneva). MMP-2,MMP-9 and GAPDH plasmids were linearized with restric-tion enzymes (Bam H1, xba EcoRI and I respectively), purifiedby phenol/chlorofom extraction and ethanol precipitation andlabelled with Digoxigenin-UTP according to the manufactur-er’s protocol. After prehybridization with DIG easy Hybsolution (1 h at 68�C), the nylon membranes were hybridizedin the same solution with the respective probes overnight at68�C. The next morning, the nylon membrane was washed andprocessed with the DIG nucleic acid detection kit according tothe manufacturer’s instructions, and exposed 8–10 h to an

Bischof et al.: Gelatinases and Proto-oncogenes 159

X-OMAT AR film. After processing the film, hybridizationbands were scanned in the same way as for zymograms (seeabove) and MMP-2 or MMP-9 band intensity expressed as percent of GAPDH.

Band shift assays

Nuclear extracts (5 �g, Schreiber et al., 1989) of treated(TNF� or TPA) or untreated CTB were incubated for 20 minat room temperature with 32P labelled (40 000 cpm TREoligonucleotides (5�-ATTGACCCCTGAGTCAGC-3�). Anti-bodies to complement factor 3 (C3) or a mixture (1 : 1) ofpolyclonal antibodies to Jun and Fos (3 �g) were added or not.Reaction products were separated on a non-denaturing 5 percent polyacrylamide gel (150 V, 15 mA, 3 h) in 0.25�TBEbuffer (22.5 m Tris-Borate, 0.5 m EDTA, pH 8.0) at roomtemperature and the gel exposed to an X-OMAT AR film.

Figure 3. Measurements of messenger RNA (in per cent of GAPDH) for MMP-9 (upper) and MMP-2 (lower) in CTB incubated with medium alone (0), TPAor TNF� (100 ng/ml) in presence or absence of 50 �g/ml cycloheximide (cyclo). Error bars represent of three experiments run with different CTBpreparations.

Transient transfection

Immunopurified CTB (0.5�106 cells/well) were incubatedovernight in DMEM without FCS, in presence of TPA orTNF� (100 ng/106 cells) and in presence or absence of 5 � ofsense or antisense oligonucleotides (Table 1). After incubation,supernatants were collected and stored at �20�C untilassayed.

Efficiency of transfection was checked after overnight incu-bation in three different CTB preparations (6�106 cells/ml)using 32P labelled (400 000 cpm/well, Sambrook et al.,1989) Jun control oligonucleotides (Table 1). After col-lecting the supernatants, cells were lysed in 500 �l lysis buffer(10 m Tris, pH 7.5, 150 m NaCl, 1 per cent SDS),extracted in phenol/chloroform/isoamyl alcohol (25/24/1respectively). Radioactivity in the upper phase was expressedas a percentage of the lower phase plus the respective cellsupernatants.

160 Placenta (2003), Vol. 24

Statistical analysis

Arbitrary units of scanned zymograms as well as immuno-reactive concentrations of MMPs (ELISA) were expressedrelatively to untransfected CTB obtained from the same cellpreparation and run at the same time under the same con-ditions. Statistical evaluation was performed on these relativeunits by ANOVA using the Statview program (Abascus) onIMac. Statistical significance was considered for a P valuebelow 0.05.

Figure 4. Band-shift assays of nuclear extracts of CTB treated or not with TPA or TNF� (24 h, 100 ng/ml) and incubated with 32P labelled TREoligonucleotides. A, in absence of antibodies; B, in presence of antibodies to jun and fos; C, in presence of antibodies to jun and fos or to C3.

RESULTS

Increasing concentrations (0, 1 and 5 ng/ml) of TPA induce adose dependant stimulation (linear regression P=0.003) of thetotal gelatinolytic activity secreted by CTB (Figure 1). Higherconcentrations of TPA such as 20 and 100 ng/ml do notincrease the gelatinolytic activity any further. The increasedgelatinolytic activity observed in presence of TPA is due toa stimulation of MMP-9 and not MMP-2 as shown byzymography (Figure 1).

As measured by Northern blot, TPA had no effect onMMP-2 mRNA (Figure 2 right panel, lanes A and B). Incontrast, the mRNA of MMP-9 increases in CTB when TPA(100 ng/ml) is added to the cells (Figure 2 left panel, lanes Cand D). This increase reaches a maximum after 20 to 30 h ofincubation (Figure 2, lower left panel). The significant stimu-

latory effect that TPA or TNF� exert on MMP-9 mRNA(P=0.007 and P=0.021 respectively) could be significantlyinhibited by addition of cycloheximide to the cells (P=0.009and P=0.028 for cycloheximide added to TPA and TNF�respectively, Figure 3, upper panel). TPA or TNF� did notincrease MMP-2 mRNA and cycloheximide remained withouteffects (Figure 3 lower panel).

TPA or TNF� increased the amount of nuclear proteinsbinding to the radiolabelled TRE consensus sequence asevidenced by the band shift assay (Figure 4A). A mixture ofantibodies to Jun and Fos decreased the amount of nuclearproteins binding to the TRE consensus sequence (Figure 4B)whereas an unrelated antibody raised against the humancomplement factor C3 was ineffective (Figure 4C).

Culture supernatants of TNF� or TPA treated CTB tran-siently transfected with sense and antisense oligonucleotides tofos and jun mRNA were analysed by zymography and immu-noassays for MMP-2 and MMP-9. Results obtained followingtransfection of antisense probes Jun193 and Jun 1261 or Fos1183 and Fos 2088 (Table 1) were pooled as antisense Jun orantisense Fos and compared to the respective sense probes.Using radiolabelled control Jun probes, a transfection ef-ficiency of 5.3+1.6 per cent (mean+, n=3) was found. Asshown in Figure 5A, antisense mRNA to Fos or to Fos and Jun(but not to Jun) significantly (P=0.019 and P=0.013 respect-ively) decreased the zymographic activity of pro MMP-9. Theactivity of MMP-9 was significantly (P=0.013) decreased bytransfection with Fos antisense whereas Jun antisense dereased

Bischof et al.: Gelatinases and Proto-oncogenes 161

the activity of MMP-9 without reaching statistical significance(Figure 5B). The zymographic activity of pro MMP-2 (Figure5C) was significantly (P=0.042) decreased by Fos antisensealone whereas MMP-2 activity (Figure 5D) was not modifiedby transfection of CTB with Jun or Fos antisenses. MMP-9immunoreactivity was significantly (P=0.033) decreased bytransfection with Jun antisense probes but not with Fosantisense (Figure 5E). In contrast, none of the antisense probesdecreased the immunoreactivity of MMP-2 (Figure 5F). Simi-lar results were obtained when TNF� was used instead of TPA(results not shown).

Figure 5. Zymographic activity of pro MMP-9 (pMMP-9, A), activated MMP-9 (MMP-9, B), pro MMP-2 (pMMP-2, C), activated MMP-2 (MMP-2, D) andimmunoreactivity of MMP-9 (E) and MMP-2 (F) in supernatants of CTB incubated with TPA and jun or fos antisense or jun and fos sense oligonucleotides. Errorbars represent of three experiments run in duplicates with different CTB preparations. Statistical analysis was performed by ANOVA and given P values arefor comparison with sense probes.

DISCUSSION

The tumour promoter TPA is a known activator of proteinkinase C. Activated protein kinase C phosphorylates proteinswhich lead to the transcription of several genes among whichare the early response genes c-jun and c-fos. Proteins from theJun family dimerise with proteins from the Fos family to formthe Activator Protein-1 (AP-1) complex, a well-studied tran-scription factor. TPA induced transactivation of the MMP-9gene occurs in part through AP-1 binding to a consensus TREsequence located in the promoter region of the MMP-9 gene

162 Placenta (2003), Vol. 24

(Westermarck and Veli-Matti, 1999). TPA induces MMP-9expression in HT1080 fibrosarcoma cells (Sato and Seiki,1993) and plating of rabbit synovial fibroblasts onto fibronectininduces MMP-9, a response which can be blocked by c-fosantisense (Tremble et al., 1995). Since AP-1 is also involved inthe MMP’s response to TNF� (Brenner et al., 1989) and sinceTNF� stimulates the secretion of MMP-9 by CTB (Meisseret al., 1999a) we included this cytokine in the present study.

The effects of TPA and TNF� on inducible MMPs are cellspecific: most tumour cell lines would respond to TPA orTNF� by an increased secretion of MMP-9 whereas someothers remain unresponsive (Mackay et al., 1992). This illus-trates that the regulation of MMP-9 expression is not onlydependent on AP-1 but that other transcription factors such asNFkB or ets-1 are also involved (Himmelstein et al., 1998).

CTB isolated according to the described methodology rep-resent a mixture of villous and extravillous cells (as measuredby the expression of histocompatibility antigens HLA-G,Bischof et al., 1995b). Since only extravillous CTB are invasiveone should use these cells to study the regulation of MMP-9.It is however possible to use the cells as prepared here becausein vitro these CTB behave all like extravillous cells: both typessecrete MMP-9 (Bischof et al., 1995b) and both types invadean extracellular matrix (Matrigel, Bischof et al., 1995b).

Here we observed, as we did previously with TNF�(Meisser et al., 1999a), that TPA stimulates the gelatinolyticactivity of CTB by increasing the activity of MMP-9 but notMMP-2. This effect is not due to an increased activation ofproMMP-9 since activated MMP-9 (Mr 85 kDa) was not seenon the zymogram, but is the result of an increased synthesis ofMMP-9 since the mRNA of this protease is increased by TPA.The fact that cycloheximide inhibited TPA and TNF�induced increase in MMP-9 message indicates that proteinsynthesis is required for MMP-9 mRNA to be transcribedsuggesting that early response gene products are involved inMMP-9 expression in CTB. This is confirmed by the band

shift assay demonstrating that both TPA and TNF� increasethe binding of nuclear proteins to a TRE consensus sequenceand that among these nuclear proteins Fos and Jun play apivotal role. These last oncogenes are involved in the trans-activation or transrepression of many genes, and thus the bandshift experiment does not allow us to conclude that MMP-9 isone of them. However, studies using transfection of c-fos ortransfection of c-jun and c-fos antisenses into CTB, demon-strate that at least the synthesis of c-fos proto-oncogene isnecessary for TPA or TNF induced MMP-9 expression.Transfetion of antisenses to c-jun alone did not inhibitMMP-9 expression as measured by zymography (pMMP-9and MMP-9) or ELISA. This might be due to the fact thatAP-1 transcription factors are heterodimers of members of theJun and Fos family. The fact that MMP-2 expressionremained unaffected by TPA or TNF� treatment shows thatthese gelatinases are differently regulated largely becauseMMP-2 has no AP-1 binding site in the regulatory domain ofits gene (Westermarck and Veli-Matti, 1999). It was howeverunexpected to see that c-fos antisense nucleotides (but not junor a mixture of jun and fos antisenses) did inhibit the activityof MMP-2 (Figure 5C). We have no good explanation for thisobservation but this effect of c-fos antisense might indicate thattranscription factors that regulate MMP-2 transcription ortranscription of MMP2 activators could partially be dependentupon the transcription of c-fos proto-oncogenes (Van Straatenet al., 1983).

The present data strongly suggest that AP-1 is necessary forMMP-9 transactivation in human first trimester CTB but ourresults do not mean that AP-1 is the only transcription factorinvolved. Indeed it has been shown in different human tumourcell lines that AP-1 synergistically cooperates with eitherNFkB or SP-1 sites for full transactivation of MMP-9 (Satoand Seiki, 1993; He, 1996). The role that other transcriptionfactors play in the expression of trophoblastic MMP-9 remainsto be elucidated.

REFERENCES

Bischof P, Friedli E, Martelli M & Campana A (1991) Expression ofextracellular matrix- degrading metalloproteinases by cultured humancytotrophoblast cells: Effects of cell adhesion and immunopurification. Am JObstet Gynecol, 65, 1791–1801.

Bischof P, Martelli M, Campana A, Ithoh Y, Ogata Y & Nagase H(1995a) Importance of metalloproteinases (MMP) in human trophoblastinvasion. Early Pregnancy Biology and Medicine, 1, 263–269.

Bischof P, Haenggeli L & Campana A (1995b) Gelatinase and oncofetalfibronectine secretion is dependent upon integrin expression on humancytotrophoblasts. Hum Reprod, 10, 734–772.

Bischof P & Campana A (2000) Molecular mediators of Implantation. InImplantation and Miscarriage (Eds) Bischof P.. Bailliere’s Clinical Obstetricsand Gynaecology, vol. 14.

Brenner DA, O’Hara M, Angel P, Chojkier M & Karin M (1989)Prolonged activation of jun and collagenase genes by tumour necrosisfactor-alpha. Nature, 37, 661–663.

Chakraborty C, Gleeson LM, McKinnon T & Lala PK (2002) Regulationof trophoblast migration and invasiveness. Can J Physiol Pharmacol, 80,116–124.

Feng Q, Liu Y, Liu K, Byme S, Liu G, Wang X, Li Z & Ockleford CD(2000) Expression of urokinase, plasminogen activator inhibitors and uroki-nase receptor in pregnant rhesus monkey uterus during early placentation.Placenta, 21, 184–193.

Graham CH & Lala PK (1992) Mechanisms of placental invasion of theuterus and their control. Biochem Cell Biol, 70, 867–874.

Graham CH, Fitzpatrick TE & McCrae KR (1998) Hypoxia stimulatesurokinase receptor expression through heme protein-dependent pathway.Blood, 91, 3300–3307.

He C (1996) Molecular mechanism of transcriptional activation of gelatinase Bby proximal promoter. Cancer Letter, 106, 185–192.

Himmelstein BP, Lee EJ, Sato H, Seiki M & Muschel RJ (1998) Tumour cellcontact mediated transcriptional activation of the fibroblast matrix metallo-proteinase-9 gene: involvement of multiple transcription factors including Etsand an alternating purine-pyrimidine repeat. Clin Exp Metastasis, 16, 169–177.

ACKNOWLEDGEMENT

This work was supported by grant 32-49257.96 from the Swiss National Science Fond to P.B.

Bischof et al.: Gelatinases and Proto-oncogenes 163

Hu ZY, Liu YX, Byrne S, Ny T, Feng O & Ockleford CD (1999)Expression of tissue type and urokinase type plasminogen activators as wellas plasminogen activator inhibitor type-1 and type-2 in human and rhesusmonkey placenta. J Anat, 194, 183–195.

Huhtala P, Eddy RL, Fan YS, Byers MG, Shows TB & Tryggvason K(1990) Completion of the primary structure of the human type IVcollagenase proenzyme and assignment of the gene (CLG4) to the q21region of chromosome 16. Genomics, 6, 554–559.

Huhtala P, Tuuttila A, Cow LT, Lohi J, Keski-Oja J & Tryggvason K(1991) Complete structure of the human gene for 92-kDa, type IVcollagenase. Divergent regulation of expression for the 92- and 72-kilodaltonenzyme gene in HT-1080 cells. J Biol Chem, 266, 16485–16490.

Lala PK & Connelly IH (1994) Effect of type IV collagenase antisenseoligonucleotides on invasiveness of normal and malignant trophoblast cells.Proceedings of the Annual Meeting of the American Association for CancerResearch, 35, A3658.

Librach CL, Werb Z, Fitzgerald ML, Chiu K, Corwin NM, Esteves RA,Grobelny D, Galardy R, Damsky CH & Fisher SJ (1991) 92 kDa typeIV collagenase mediates invasion of human cytotrophoblasts. Journal of CellBiology, 113, 437–449.

Librach CL, Feigenbaum SL, Bass KE, Cui TY, Verastas N, SadovskyY, Quigley JP, French DL & Fisher S (1994) Interleukin-1 beta regulateshuman cytotrophoblast metalloproteinase activity and invasion in vitro.J Biol Chem, 269, 17125–17131.

Mackay AR, Ballin M, Pelina MD, Farina AR, Nason AM, Hartzler JL& Thorgeirsson UP (1992) Effect of phorbol ester and cytokines on matrixmetalloproteinases and tissue inhibitor of metalloproteinase expression intumour and normal cell lines. Invasion Metastasis, 12, 168–184.

Martelli M, Campana A & Bischof P (1993) Secretion of matrix metallo-proteinase by human endometrial cells in vitro. J Reprod Fertil, 98, 67–76.

Matrisian LM (2000) Matrix metalloproteinases. Current Biology, 10, R692.Meisser A, Chardonnens D, Campana A & Bischof P (1999a) Effects of

tumour necrosis factor alpha, interleukin-1 alpha, macrophage colonystimulating factor and transforming growth factor beta on trophoblasticmatrix metalloproteinases. Molec Hum Reprod, 5, 252–260.

Meisser A, Cameo P, Islami D, Campana A & Bischof P (1999b) Effectsof interleukin-6 (IL-6) on cytotrophoblastic cells. Molec Hum Reprod, 5,1055–1058.

Pijnenborg R, Dixon G, Robertson WB & Brossens I (1980) Tropho-blastic invasion of human decidua from 8 to 18 Weeks of pregnancy.Placenta, 1, 3–19.

Sambrook J, Fritsch EF & Maniatis T (1989) Molecular Cloning. NewYork, NY: Cold Spring Harbor Laboratory Press.

Sato H & Seiki M (1993) Regulatory mechanism of 92 kDa type IVcollagenase gene expression which is associated with invasiveness of tumourcells. Oncogene, 8, 395–405.

Schreiber E, Matthias P, Muller MM & Schaffner W (1989) Rapiddetection of octamer binding proteins with mini-extracts prepared fromsmall numbers of cells. Nucleic Acid Res, 17, 6419.

Tremble P, Damsky CH & Werb Z (1995) Components of the nuclearsignalling cascade that regulate collagenase gene expression in response tointegrin-derived signals. J of Cell Biol, 129, 1707–1720.

Van Straaten F, Muller R, Curran T, Van Beveren C & Verna IM (1983)Complete nucleotide sequence of human c-oncogene: deduced aminoacid sequence of the main c-fos protein. Proc Natl Acad Sci USA, 80,3183–3187.

Westermarck J & Veli-Matti K (1999) Regulation of matrix metalloprotei-nase expression in tumour invasion. FASEB J, 13, 781–792.