Kisspeptin-10, a KiSS-1/metastin-derived decapeptide, is a physiological invasion inhibitor of primary human trophoblasts

The invasion of trophoblast cells (cytotrophoblasts), epithelial cells of the placenta, into the maternal decidua in the first trimester of pregnancy is a key process for successful reproduction and embryonic development. As a result, the embryo will be anchored in the maternal uterus and utero-placental arteries will be invaded, leading to the dilation of the vessel lumina, which is necessary to enhance uterine and intervillous blood flow in pregnancy. There are striking similarities between invasive cytotrophoblasts and invasive cancer cells (Kliman and Feinberg, 1990; Murray and Lessey, 1999; Bischof and Campana, 2000; Bilban et al., 2000; Murray and Lessey, 1999; McMaster et al., 1998). However, in contrast to tumor invasion, interactions between trophoblasts and uterine cells are closely regulated both temporally and spatially by mechanisms that are largely unknown. If mechanisms controlling migration and invasion of both trophoblast and neoplastic cells are related, it is tempting to postulate that genes expressed in early pregnancy, a period necessitating a tight control of these processes, also play a key role in regulating migration and invasion of cancer cells.

KISS-1 is one of the genes involved in controlling cancer cell dissemination. Differential display and subtractive hybridization were used to isolate KiSS-1 from melanoma cells (Lee et al., 1996). Its expression is lost during cancer progression, and overexpression of KiSS-1 cDNA in metastatic human cancer cell lines suppressed metastasis in athymic nude mice (Lee and Welch, 1997; Shirasaki et al., 2001; Sanchez-Carbayo et al., 2003). The primary translation product of the KiSS-1 gene is a 145 amino acid polypeptide (Kp-145) (Ohtaki et al., 2001), but shorter `kisspeptins' (Kp) with 10, 13, 14 or 54 amino acid residues have been discovered (Fig. 1) (Kotani et al., 2001; Stafford et al., 2002; Harms et al., 2003). Kps are endogenous ligands for an orphan G-protein-coupled receptor (GPR54, hOT7T175, AXOR12; KiSS-1R as designated from here on) that couples primarily to Gq/11 (Ohtaki et al., 2001; Kotani et al., 2001; Muir et al., 2001). KiSS-1R and its cognate ligand KiSS-1 are expressed in a variety of tissues including pancreas, testes, central nervous system and placenta (Janneau et al., 2002; Harms et al., 2003). Kp-54, also termed `metastin', is the putatively secreted and biological active form of Kp-145 (Harms et al., 2003). It inhibits chemotaxis in vitro, enhances the expression and activity of focal adhesion kinase and inhibits melanoma cell metastasis (Ohtaki et al., 2001). Kp-10 stimulates phosphatidylinositol-4,5-bisphosphate hydrolysis, Ca²⁺ mobilization, arachidonic acid release, extracellular signal-regulated kinase (ERK)1/2 and p38 mitogen-activated protein (MAP) kinase phosphorylation, and stress fiber formation (Kotani et al., 2001). These effects have been found in cells overexpressing KiSS-1R, but have not been investigated in a physiological system so far. Human first trimester (FT) placenta enables the mechanisms of invasion and its regulation to be investigated in a physiological setting.

FT trophoblasts secrete collagenases MMP-2 and MMP-9 to degrade extracellular matrices, thus making migration and invasion possible (Bischof and Campana, 2000; Xu et al., 2001). Interestingly, the KiSS-1 transcript product represses MMP-9 activity (Yan et al., 2001).

Despite their putative importance in invasion regulation in a biological system, neither Kp-10 has been detected nor have the effects of Kps been investigated. Here we report for the first time the identification of Kp-10 as a naturally occurring peptide in conditioned medium of primary human trophoblasts, its ability to reduce trophoblast migration without altering proliferation, and to downregulate the activity of MMP-2, one of the key enzymes in trophoblast invasion.

Collagen type I (rat tail tendon), fibronectin and Matrigel were from Beckton Dickenson (Oxford, UK), Sigma (St Louis, MO) and Stratech Scientific (Luton, UK), respectively. Kp-10, Kp-13 and Kp-14, corresponding to KiSS-1 aa 112-121, 109-121 and 108-121, respectively (Ohtaki et al., 2001; Kotani et al., 2001), were chemically synthesized. Kp-54 (corresponding to aa 68-121 of KiSS-1) was purchased from California Peptide Research (Nappa, CA). A polyclonal antibody to Kp-54 was raised in rabbits. Polyclonal rabbit anti-KISS-1R (human) antibody (cl375-398) was from Phoenix Pharmaceuticals (Belmont, CA).

Two milligrams of Kp-54 (California Peptides) were dissolved in 0.1 M 2-(4-Morpholino)-ethanesulfonic acid buffer (pH 4.9) and coupled to 3 mg KLH (keyhole limpet hemocyanine) by use of a water-soluble derivative of EDCI (1-ethyl-3-(3-dimethylaminopropyl)-carbodiimidhydrochloride) as a condensation agent (Hermanson, 1996). After 2 hours EDCI and other low molecular weight molecules were removed by size exclusion chromatography (PD-10 columns, Sephadex G25 M, Amersham) using PBS as purification buffer. Two New Zealand White rabbits were immunized subcutaneously by the use of complete Freud Adjuvant and two boosts after 4 weeks and 2 weeks, respectively, with incomplete Freud Adjuvant according to standard protocols used by Charles River (Kisslegg, Germany). Globulins were precipitated from the isolated sera by ammonium sulfate and IgG were purified by protein-A affinity chromatography (Pierce, Rockford, IL).

All patients gave informed consent for collection and investigational use of tissues. This study was approved by the ethics committee of the Karl-Franzens University (Graz, Austria). FT placental tissue was obtained after termination of normal pregnancies by vacuum suction; term placentae were collected after uncomplicated pregnancy and vaginal delivery. Tissues were immediately used for experiments or fixed for 48 hours in 20% formalin buffered with 0.1 M phosphate buffer, pH 7.4, embedded in paraffin, and cut into 4 μm (immunohistochemistry) or 7 μm (in situ hybridization) sections. Five different placentae with at least ten sections of each were examined.

Mononucleated trophoblast cells were isolated by tissue digestion with trypsin and subsequent separation of released cells on a density gradient. Cells from the bands containing trophoblasts were further purified using immunomagnetic beads conjugated with either a combination of anti-CD45RB and fibroblast-specific antibodies (FT) or anti-HLA class-I antibodies (term) to remove nontrophoblast components. Isolated cells were cultured in Dulbecco's Eagle's Medium (DMEM; Gibco, Life Technologies, Paisley, UK) and rigorously characterized as described in detail elsewhere (Blaschitz et al., 2000; Cervar et al., 1999).

Preparation of cRNA, hybridization to human U95A GeneChips (Affymetrix, Santa Clara, CA) and scanning of the arrays were carried out at The Scripps Research Institute's Affymetrix Array Core facility (http://www.scripps.edu/services/dna_array/) according to the manufacturer's protocols (available online at https://www.affymetrix.com; (Su et al., 2002)). Images were analyzed with GeneChip software (Affymetrix, version 3.3). Genes were scored on the basis of the putative biological functions of the encoded proteins, as determined by database searches on PubMed, gene cards from the Weizmann Institute of Science (http://bioinfo.weizmann.ac.il/bioinfo.html), and a previously published classification scheme (`OntoExpress') for cellular functions (Khatri et al., 2002).

Using the software Primer3 (Whitehead Institute/MIT Center for Genome Research; http://www.genome.wi.mit.edu), sequence-specific primers were selected from the full-length cDNA sequences on the basis of published sequences of 13 invasion-associated and one housekeeping gene (Table 1). One microgram of total RNA was amplified using the One-Step RT-PCR kit (Qiagen,Valencia, CA). Reverse transcription was carried out at 55°C for 30 minutes and HotStarTaq DNA polymerase was activated at 95°C for 15 minutes followed by 25 (for all 13 invasion-associated genes) or 30 cycles (GAPDH) of amplification (94°C for 30 seconds, 55°C for 30 seconds and 72°C for 60 seconds), and final extension at 72°C for 10 minutes.

Trophoblast cell lysates were separated by SDS-PAGE and blotted as described previously (Hahn et al., 1998). Blots were incubated with Kp-54 antiserum (1:10,000). To determine the apparent molecular weight of the band identified by the antiserum, a standard curve was generated by plotting the log10 of unstained molecular weight standards against their retention factor (Rf). Thus, a single band at ∼15 kDa could be identified as Kp-145.

Conditioned media from FT trophoblasts cultured for 48 hours in serum-free DMEM were precleared from excess protein over a Nucleosil RP-18 column (Macherey Nagel, Easton, PA). The peptides were eluted with a linear gradient of acetonitrile (10-90%) in 0.1% (v/v) trifluoroacetic acid (TFA) over 30 minutes with a flow rate of 0.4 ml/minute; fractions were collected from 10.7 to 13 minutes. The synthetic peptides (Kp-10, -13, -14 and -54) were analyzed in a separate run to determine their retention times. In a second purification step, purified fractions were reconstituted in 10% (v/v) acetonitril containing 0.1% (v/v) TFA and passed over a Vydac RP-18 column (Vydac, USA). Peptides were eluted over 15 minutes with the same linear gradient as above. Fractions containing Kisspeptins were collected according to the retention times of the synthetic Kisspeptins and subjected to MS-Analysis. MALDI-TOF spectra were obtained in positive linear mode on a KRATOS Axima CFR mass spectrometer (Manchester, UK) by integrating 20 to 50 laser shots per spectrum. Alpha-Cyano-4-hydroxycinnamate saturated in 30% (v/v) acetonitrile and 0.1% (v/v) TFA in water was used as a matrix. External calibration was performed with adrenocorticotropic hormone (ACTH) fragment 18-39; 2466.71 m/z [MH+ av.], and insulin 5734.6 m/z [MH+ av.] (used as the mass calibration standards).

Deparaffinized tissue sections were treated for 10 minutes with blocking solution, then incubated for 30 minutes with primary antibodies (rabbit polyclonal antiserum against Kp-54, 1:500; anti-KiSS-1R, 1:100) followed by an incubation with an appropriate secondary antibody for 15 minutes, and then 15 minutes with avidinbiotin-peroxidase (Labvision). Immunolabeling was visualized by a 5 minute exposure to 3-amino-9-ethylcarbazole. Sections were then counterstained for 3 minutes with hematoxilin and mounted in Faramount (DAKO).

FT or term placenta sections were deparaffinized and hybridized with digoxygenin (dig)-labeled riboprobes to KiSS-1 and KiSS-1R, according to a published protocol (Simeone, 1998). RNA probes for human KiSS-1 and KiSS-1R were prepared as follows: on the basis of the published sequence, a 457 bp fragment of the the human KiSS-1 cDNA sequence was amplified by PCR using a primer pair KiSS-1_EcoRI (5′-GGG AAT TCT AGA CCC ACA GGC CAG CAG CTA GAA-3′; bp 296-328) and KiSS-1_HindIII (5′-TTT ATT GCC TAA GCT TGG AAG CTC CAG CGC CCC-3′; bp 724-752). Similarly, a 781 bp fragment of the human KiSS-1R cDNA sequence was amplified by PCR using primer sets KiSS-1R_EcoRI (5′-TGG GGA ATT CGC TGG TCA TCT ACG TCA TCT G-3′;) and hOT75T175_HindIII (5′-CAG GTC TTA AGC TTG TAG GCG GCG TAG CTG-3′). The amplified sequences were confirmed at The Scripps Research Institute's DNA Sequencing core facility (http://www.scripps.edu/). Amplified fragments were subcloned into the EcoRI and HindIII sites of pGEM-3Z (Promega, Madison, WI), and the resultant vectors were used as templates for construction of the dig-labeled RNA probes according to manufacturer's instructions (Roche Diagnostic, Indianapolis, IN). After hybridization, the slides were washed, and incubated for 4 hours at RT with peroxidase-labeled goat-anti dig monoclonal antibody. Signals were detected overnight at RT in buffer 2 containing 3 μl/ml NBT (Nitroblue tetrazoliuim chloride), 50 mg/ml BZIP (5-bromo-4-chloro-3-indolyl-phosphate, 4-toluidine salt) and 1 mM Levamisol. Color development was terminated in TE (10 mM Tris/HCl; pH 8.0, 1 mM EDTA) for 5 minutes. Sections were mounted in gelatin-glycerol (Sigma).

The explant culture method followed a standard protocol (Aplin et al., 2000). Dissected mesenchymal villous tissue was arranged radially on 80 μl drops of collagen type I gel, covered with 20 μl of serum-free medium (SFM; DMEM:Ham's F12 (Gibco) 1:1, supplemented with 100 μg/ml streptomycin/penicillin (Gibco) and incubated overnight at 37°C, 5% CO₂ to allow attachment. Wells were then carefully flooded with 1 ml SFM and incubated in the absence or presence of 0.3 or 1.0 μM Kp-10 at 37°C, 5% CO₂ and subsequent growth characteristics monitored by light microscopy. Newly synthesized DNA was detected with bromodeoxyuridine (BrdU; Roche) added to selected cultures at 1 μM for the periods specified. The extent of migration, i.e. the distance from the cell column base to the tip of the outgrowth, was measured at three defined positions (three explants per condition) and the mean of these values was used for comparing the Kp-10 effects.

Cytosolic free Ca²⁺ concentration ([Ca²⁺]_c) in single cells was measured using the conventional single-cell fura-2 technique with a customized fluorescence microscope as described previously (Graier et al., 1998), and expressed as ratio (F₃₄₀/F₃₈₀) units. Briefly, 1×10⁵ FT trophoblasts were plated on glass coverslips. The following day, adherent cells were loaded for 45 minutes at RT in the dark with 2 μM fura-2/AM (Molecular Probes, Leiden, The Netherlands), washed twice and equilibrated for a further 20 minutes. The coverslip was mounted into an experimental chamber and perfused (1 ml/min^-1) with buffer solution containing, in mM: 145 NaCl, 5 KCl, 2 Ca₂Cl, 1 MgCl₂ and 10 Hepes free acid (pH adjusted at 7.4).

Transwell migration assays were performed as described (Hintermann et al., 2001; Song et al., 2001) with filters coated with 30 μg/ml collagen I or 10 μg/ml fibronectin. After initial experiments to determine the temporal kinetics of migration, subsequent migration assays were performed for 48 hours. Cells that had migrated through the filters were viewed under bright-field optics and were counted in eight fields (using a 20× objective) from each of two filters for each condition, determining the mean number of cells counted per field.

Protein content of trophoblastic conditioned media was measured according to the method of Bradford (Bradford, 1976), and equal amounts of protein were subjected to gelatin zymography on nonreducing 10% polyacrylamide gels supplemented with 1% gelatin to determine the proteolytic activity of conditioned media as described previously (Xu et al., 2001). Prestained molecular mass standards (Invitrogen) were used to determine the molecular mass of proteolytic activity.

Migration indices and MMP-2 levels were analyzed using one-way ANOVA, followed by the Tukey test. As data were not normally distributed, the Mann-Whitney rank-sum test was employed to determine the level of significance of differences in pairs of various treatment groups. P<0.05 was considered significant.

DNA microarray-based large-scale gene expression profiling was applied to highly (FT) and poorly (term of gestation) invasive trophoblast cells. In all, 643 differentially expressed genes (more than twofold difference between FT and term trophoblast cells) were further compared with the gene expression profile of the poorly invasive but proliferative trophoblast-like cell line BeWo (Grummer et al., 1994; Morgan et al., 1998; Xu et al., 2001). Genes that showed a similar expression pattern in both FT and BeWo cells were assumed to control processes common to both cell types, including proliferation and endocrine function, but not invasion. Thus, BeWo cells served as a `filter' to identify genes regulating trophoblast invasion. Using this selection method, 220 invasion-associated genes were differentially expressed, some of which are shown in Table 1. A similar approach was used to identify prognostic biomarkers in prostate cancer (Dhanasekaran et al., 2001). In that study, distinct noninvasive prostate samples clustered together, differently from the metastatic samples. In order to identify novel key players regulating trophoblast invasion we focused on KiSS-1, a recently described tumor suppressor gene (Ohtaki et al., 2001), as it was one of the strongest differentially expressed genes (29.0-fold higher mRNA levels in FT compared with term trophoblasts).

To confirm differential expression of KiSS-1 mRNA and to include KiSS-1R in our studies, we performed semi-quantitative RT-PCR and found 17.4±3.4-fold (KiSS-1) and 23.9±5.4-fold (KiSS-1R) higher mRNA levels in FT than term trophoblasts (Fig. 2A). In western blot analyses, the antiserum generated against Kp-54, which is within the Kp-145 sequence, detected Kp-145, but not Kp-54, in lysates from FT trophoblasts (Fig. 2B) but not in FT trophoblast conditioned medium, even after 30-fold enrichment (data not shown). This suggests the following scenarios: (1) intracellular Kp-54 pools are rapidly depleted by secretion; (2) intracellularly generated Kp-54 is rapidly degraded because of the presence of PEST sequences (rich in proline, glutamic acid, serine, threonine and aspartic acid residues predisposing proteins to ubiquitination and proteosome degradation (Harms et al., 2003). The presence of this motif (aa 68-87) may allow rapid degradation of cytosolic Kp-145 resulting in a short half-life (<30 seconds) (Harms et al., 2003); (3) levels of intra- or extracellular Kp-54 are below the detection limit of western blotting. To exclude the latter, conditioned medium of cultured FT trophoblasts was analyzed by MALDI-TOF, which revealed the presence of Kp-10, -13 and -14, as well as Kp-54 in conditioned medium, with masses consistent with the theoretical masses of the amidated forms for Kp-10, -13, -14 and Kp-54 (Fig. 2C). Secreted Kp-54 may be the source for the shorter Kps Kp-10, -13 and -14.

Next, we tested which of the various Kps can activate KiSS-1R by monitoring cytosolic Ca²⁺ levels after stimulation of isolated FT trophoblasts with the respective Kps (Fig. 3). This treatment increased intracellular Ca²⁺ in KiSS-1R-overexpressing cells (Ohtaki et al., 2001; Muir et al., 2001; Kotani et al., 2001). Addition of Kp-10 to FT trophoblasts increased cytosolic Ca²⁺ levels with an EC₅₀ of 21 nM (Fig. 3A,B). By contrast, Kp-54 as well as Kp-13 and Kp-14 failed to produce this response (Fig. 3A).

Both KiSS-1 mRNA and protein were found at the syncytiotrophoblast of anchoring (Fig. 4A,B) and floating villi, whereas it was undetectable in villous and extravillous cytotrophoblasts. Maternal decidua was devoid of KiSS-1 mRNA staining (data not shown). On the basis of these results, expression of KiSS-1 as well as KiSS-1 encoded Kps in FT placenta appears to correlate with cells that have fused and become noninvasive. KiSS-1R mRNA and protein was localized to syncytio-, villous- and extravillous cytotrophoblast at similar levels (Fig. 4C,D and insets). Thus, syncytiotrophoblast expresses both the ligand Kp-145/Kp-54 and its receptor, whereas the extravillous cytotrophoblast only expresses the receptor KiSS-1R (Fig. 7). This is in line with the hypothesized inhibitory role of Kp-54 in trophoblast invasion.

Given the anti-metastatic function of KiSS-1, we investigated whether Kp-10 can inhibit migration of FT placental tissue. Tissue cultures of anchoring villi explanted from early gestation (6 to 9 weeks) placentas onto an extracellular matrix substrate allow the study of trophoblast outgrowth, migration and invasion, thus mimicking the key processes that occur in anchoring villi during the FT of gestation (Genbacev et al., 1992; Aplin et al., 2000). This assay has the significant advantage of allowing migratory behaviour to be observed in living trophoblasts, because migration occurs largely across the surface of the gel. Control explants displayed prominent outgrowth of extravillous trophoblast from the distal end of the villous tip and an increased number of cells migrating into the surrounding matrix (Fig. 5A,B) similar to previous reports (Aplin et al., 2000). Kp-10 strongly affected the pattern of cell migration (ANOVA P<0.001): explants exposed to Kp-10 for 48 or 72 hours displayed about 70% diminished extravillous trophoblast outgrowth (Fig. 5C-F,K).

Because de novo cell column formation is accompanied by cytotrophoblast proliferation peaking after an initial period of 24 hours and ceasing after 48 hours (Caniggia et al., 1997; Genbacev et al., 1992; Aplin et al., 2000), we investigated whether the Kp-10-induced inhibition of trophoblast outgrowth was due to a reduction in cell division. When tissue was explanted in the presence of Kp-10, viability and proliferative potential of cytotrophoblast was maintained for the first 24 hours as indicated by the incorporation of BrdU into cytotrophoblasts beneath the villous syncytium and in the proximal cell column (Fig. 5). Because cytotrophoblast proliferation was unaffected in the critical phase, these results indicate that Kp-10 does not affect trophoblast proliferation and that the reduced trophoblast outgrowth is most likely due to a blockade of trophoblast migration. This was tested in a Transwell migration assay, where inhibition of FT trophoblast migration was readily apparent by visual inspection (Fig. 6A). The number of migrating cells scored per field was reduced by 37% and 46% in the presence of 0.5 μM Kp-10 on fibronectin or collagen I, respectively (Fig. 6B). Like in the explant assay, Kp-10 did not alter FT trophoblast proliferation (Fig. 6B).

Regulation of cell motility is controlled by a multitude of factors (Lauffenburger and Horwitz, 1996), among which proteases play an evident role (Bischof and Campana, 2000; Chang and Werb, 2001). To determine whether Kp-10 triggers a decrease in collagenase activity, resulting in reduced trophoblast migration, FT trophoblast-conditioned medium was analyzed for collagenase activity by zymography. Addition of 0.3 μM or 1.0 μM Kp-10 to FT trophoblast cultures suppressed MMP-2 activity by 32% and 62% after 24 hours, and by 28% and 42% after 48 hours, respectively (Fig. 6C,D). Secreted MMP-9 levels were too low to allow accurate quantification (data not shown).

In the present paper we show that kisspeptin Kp-10, a shorter version of the recently described metastasis suppressor gene product KiSS-1, inhibits trophoblast migration and invasion, presumably by downregulating protease activity. Our hypothesis is based on the finding that KiSS-1 was among the strongest differentially expressed genes in invasive FT versus noninvasive term trophoblast preparations. Furthermore, Kp-10 blocked explant invasion and Transwell migration and decreased MMP-2 activity. These results make KiSS-1 a strong candidate as a gene that regulates trophoblast invasion in vivo.

Differential expression of KiSS-1 was similar in microarray and RT-PCR experiments. These results confirm published data on the relative expression levels of the well-known invasion-specific trophoblast markers integrins α1 and α5, as well as HLA-G, which are almost exclusively expressed by the invasive cytotrophoblast phenotype (Damsky et al., 1994; McMaster et al., 1998; Cervar et al., 1996; Blaschitz et al., 2000), thus confirming our microarray strategy. Differential expression of KiSS-1 was paralleled by the expression and location of its receptor KiSS-1R, suggesting autocrine interactions. Additional receptor expression on villous and extravillous cytotrophoblast also suggests paracrine or endocrine mechanisms for receptor activation (Fig. 7, Fig. 3). This is supported by the findings that Kp-10 stimulates oxytocin secretion in rats and that pregnant women have high Kp-54 plasma levels (Horikoshi et al., 2003; Kotani et al., 2001). However, the cellular origin of secreted Kp-54 has not been shown so far. Here, Kp-54 and its fragments Kp-10, -13 and -14 were identified by MALDI-TOF analysis in serum-free conditioned medium of isolated FT trophoblasts. Kp-54 is predicted to result from the proteolytic processing of Kp-145 by furin or prohormone convertases, as its sequence is surrounded by pairs of basic residues in the full-size protein (Harms et al., 2003). At present, the mechanism resulting in the generation of Kp-10, -13 and -14 is unclear as no obvious cleavage sites have been found (Harms et al., 2003), but scenarios such as spontaneous decomposition as well as constitutive or regulated processing can be envisioned.

In cells overexpressing KiSS-1R, Kp-54 was capable of activating the receptor and increasing intracellular Ca²⁺ levels (Kotani et al., 2001; Muir et al., 2001; Ohtaki et al., 2001). In contrast to the present system with a physiological receptor level, only Kp-10 gave rise to increasing intracellular Ca²⁺ with an EC₅₀ similar to other systems (Kotani et al., 2001; Muir et al., 2001). Because the Kp-54 used here caused a rapid Ca²⁺ response in KiSS-1R-overexpressing cells (Muir et al., 2001) we conclude that the chemically synthesized Kp-54 can induce effects in cells overexpressing KiSS-1R, whereas Kp-10 is the activator in physiological systems. The special position of Kp-10 is highlighted not only because of its threefold-to-tenfold higher receptor affinity than other Kps (Hori et al., 2001; Ohtaki et al., 2001), but also by the high conservation of the Kp-10 sequence between mouse and human with one conserved amino acid replacement, whereas other regions of the Kp-145 protein display only low homology (Stafford et al., 2002).

The extent of trophoblast invasion also depends on the provision of trophoblasts from the cell column that will differentiate and acquire an invasive phenotype. Explant and Transwell migration assays showed that Kp-10 signals reduced trophoblast motility without affecting trophoblast proliferation or viability. The absence of Kp-10 effect on trophoblast proliferation shows that the mechanism underlying invasion inhibition must be different. In fact, suppressed motility was associated with suppressed MMP-2 proteolytic activity. The high in vitro invasive capacity of FT trophoblasts has been shown to depend on their secretion of collagenases and gelatinases (Bischof et al., 1995; Bischof and Campana, 2000). Indeed, extravillous cytotrophoblasts secreting high amounts of gelatinases are highly motile and invasive, whereas extravillous cytotrophoblasts secreting low amounts of gelatinases have become immobile. Thus, syncytium-derived Kp-10, by suppressing gelatinolytic activity of extravillous cytotrophoblast located in proximity to maternal circulation, might be an endocrine mechanism of fine-tuning trophoblast motility (Fig. 7). In addition, syncytial Kp-10 may also act in a paracrine manner on the underlying villous cytotrophoblast by suppressing gelatinolytic activity in a coordinated fashion with other signals, thus rendering villous cytotrophoblasts immobile and anchored to the villous basement.

Although the ability of Kp-10 to inhibit migration has been documented in a few other cell types with overexpression of KiSS-1R (Ohtaki et al., 2001), this is the first study to report Kp-10 motility regulation in a human physiological setting. Our conclusion of an invasion inhibiting function of the KiSS-1/KiSS-1R system in human trophoblasts is corroborated by reduced expression of both in trophoblast-derived choriocarcinoma (Janneau et al., 2003). Invasion inhibition will lead to excessive invasion that is a characteristic of this tumor.

Downregulation of MMP-2 activity may be one potential mechanism underlying the Kp-10-induced invasion inhibition and this may be tissue specific (Yan et al., 2001). The possibility that Kp-10 affects MMP-9 expression or activity later in gestation cannot be excluded because trophoblast-derived MMPs are developmentally regulated throughout pregnancy and MMP-9 secretion increases between week 6 and 11 of gestation (Xu et al., 2001; Huppertz et al., 1998). In addition to MMP regulation, probably by direct interaction with MMPs (Takino et al., 2003), other mechanisms may be operative, such as activation of focal adhesion kinase (Kotani et al., 2001; Ohtaki et al., 2001; Ilic et al., 2001), alteration of tissue inhibitors of MMPs (TIMPs) or effects on the plaminogen activator/inhibitor system. The absence or low levels of invasion in situ at term despite low KiSS-1 levels suggests that other anti-invasive factors prevail at the end of gestation. Hence, the anti-invasive effect of KiSS-1/KiSS-1R system may only be important in the first trimester.

We thank Abdulaziz Moqrich for expert technical advice for in situ hybridization, Steven Head for assistance with DNA microarrays and Gabi Hagendorfer, Renate Michlmaier and Nicole Prutsch for assistance with trophoblast preparations. This work was supported by grants GM46902 and CA47858, National Institutes of Health to V.Q. and grants 7535 and 7793, Jubilee Fund, Austrian National Bank, Vienna to G.D.