Antagonism of EG-VEGF Receptors as Targeted Therapy for Choriocarcinoma Progression In Vitro and In Vivo

Gestational choriocarcinoma (CC) is a malignant trophoblastic tumor that develops upon normal or abnormal pregnancy. The latter includes complete hydatidiform moles (CHM) or partial hydatidiform moles (PHM), spontaneous abortions, or ectopic pregnancies (1, 2). CHM develops when one or two spermatozoa fertilize an oocyte in the absence of the maternal nucleus, while PHM results from dispermic fertilization of a nucleated oocyte (1, 2). Both CHM and PHM patients are at high risk of developing CC, nevertheless this risk is much higher after CHM (20%) than after PHM (1.5%; refs. 1, 2). CC has an estimated incidence of 2 to 7 in 100,000 pregnancies in North America and Europe. This incidence is higher in Africa (3, 4) and even higher in Asia, where it reaches 5 to 202 in 100,000 pregnancies (5, 6). CC is highly metastatic due to the intrinsic invasive property of trophoblast cells (7). Most patients with non-metastatic gestational CC are successfully treated with single-agent chemotherapy (8–10). Metastatic cases are only curable using multi-agent chemotherapy, known to induce considerable adverse effects (10–12). CC diagnosis and progression is based on the measurement of the human chorionic gonadotropin (hCG) released by syncytiotrophoblast cells (1, 2). Nevertheless, hCG measurements have recently been reported to be associated with false positive diagnoses and to unnecessary invasive therapeutic procedures, including chemotherapy, hysterectomy, and other surgeries (13–17). Thus, it is necessary to prevent the occurrence of this cancer and develop novel and more specific diagnostic markers as well as less toxic therapeutic approaches.

We recently reported that a specific endocrine placental factor named endocrine gland-derived vascular endothelial growth factor (EG-VEGF) exhibits features of a potential marker of pathologic pregnancies, including preeclampsia (18). EG-VEGF is also called prokineticin-1 and represents the canonical member of a recently described family of cytokines, the prokineticin family (19, 20). EG-VEGF expression is grossly restricted to endocrine tissues, including the placenta (21). It exerts its actions via two G protein-coupled prokineticin receptors, called PROKR1 and PROKR2 (21, 22). We reported that EG-VEGF (i) is abundantly produced by the endocrine unit of the placenta, the syncytiotrophoblast with a peak of expression just before the establishment of the feto-maternal circulation, (ii) its expression is upregulated by hypoxia (a key parameter of tumor development) and by hCG, and that (iii) its receptors, PROKR1 and PROKR2 are highly expressed in trophoblast and placental endothelial cells (21, 23, 24). We have also demonstrated that EG-VEGF circulating levels are around 50 pg/mL in non-pregnant women and increase by 5-fold during the first trimester of pregnancy (≈250 pg/mL; ref. 24).

In relation to CC development and progression, it is proposed that trophoblast cells are excessively proliferative, a phenomenon that ultimately results in an increased pool of cells acquiring a migratory and invasive phenotype (25). Importantly, we have demonstrated that EG-VEGF increases trophoblast proliferation and ensures their survival (21, 26). Also, EG-VEGF has recently been reported to be associated with tumor development of multiple reproductive organs such as ovary (27), testis (28), and prostate (29, 30).

Altogether, these findings suggest that EG-VEGF could be a potential actor in CC development and progression with the perspective of its use as therapeutic target. However, no previous study has investigated its involvement in the development and/or progression of this cancer. Here, we conducted (i) a clinical study to investigate EG-VEGF involvement in to CHM and CC development, (ii) an in vitro study, using JEG3 cells to characterize EG-VEGF effects on their proliferation, migration, and invasion. Moreover, we developed for the first time an animal model of CC upon an orthotropic injection of JEG3 cells into the placenta of immunodeficient gravid mice. This model was used to characterize CC progression and to test the effects of EG-VEGF receptors antagonists on tumor development and progression.

Pathologic material consisting of placental tissue (n = 28) and sera from patients with CHM (n = 38) and CC (n = 9) were collected at Ibn Rochd Hospital in Casablanca, McGill University Health Centre Research Institute, and the French Reference Center for Gestational Trophoblastic Diseases. Normal sera were also collected from normal first trimester pregnant women (n = 20). Collection and processing were approved by the local hospital ethics committees, and informed patient consent was obtained in all cases. Supplementary Fig. S1 reports all clinical information for pathologic pregnancies.

JEG3 (ATCC HTB-36) is one of the six clonally derived lines isolated from the Woods strain of the Erwin-Turner tumor by Kohler and associates (31). JEG3 was used as CC cell line model. JEG3 cells were systematically tested for mycoplasma and used between 4 and 10 passages.

JEG3-Luc (Luciferase positive JEG3) were prepared using a lentivirus supernatant (pLenti-II-CMV-Luc-IRES-GFP control vector). The protocol was performed according to the company's instructions (Applied Biological Materials Inc.). Briefly, JEG3 cells were plated in DMEM/F-12 (1/1) medium supplemented with 10% fetal bovine serum (FBS). Cells were infected for 6 hours with lentivirus at a ratio of 1:1 in fresh culture medium. Infected cells were selected with G418 (200 μg/mL) for 7 days.

EG-VEGF was measured by ELISA (PeproTech) in the collected sera and conditioned media. Two separated standard curves were constructed to allow accurate readings of samples at upper and lower ranges of the assay. All samples were in the linear range of the standard curves. The detection limit of the assay was 16 pg/mL. The intra-assay coefficient of variability (CV) was 6.7%, and the interassay CV was 8.1%.

JEG3 cells were seeded into 96-well plate at the density of 3 × 10⁴ cells/well and incubated with increased concentrations of recombinant EG-VEGF for 24 hours. Proliferation was also assessed in the absence or presence PROKRs antagonists. PROKR1 antagonist was obtained from Dr G. Balboni (University of Cagliari, Italy) and PROKR2 antagonist was obtained from Dr. QY Zhou (University of California Irvine, CA). Both antagonists were used at 1 μmol/L. Proliferation was performed using a non-radioactive assay that measures cell viability and proliferation (WST-1, Roche Diagnostics GmbH, Mannheim, Germany). siRNA strategy for PROKR1 and PROKR2 was also used in this test (Supplementary Fig. S2B, S2C, and S2D).

Wound-healing (24) and invasion assays are reported in Supplementary materials

Confluent monolayers of JEG3 cells were trypsinized. A total of 1,000 cells were suspended in culture medium DMEM-F12 supplemented with 10% FBS, and seeded in non-adherent round-bottom 96-well plates (Greiner) coated previously with Poly-HEMA [Poly (2-hydroxyethyl methacrylate); Sigma-Aldrich]. Under these conditions, all suspended cells contribute to the formation of a single JEG3 cell spheroid specimen. The spheroids were harvested after 24 hours and transferred into a collagen gel (3.54 mg/mL, BD Biosciences). Spheroids were treated with different concentrations of recombinant EG-VEGF for 24 hours. These experiments were also performed in the absence or presence of PROKRs antagonists. Quantification of JEG3 cells spreading was assessed using ImageJ software (http://rsb.info.nih.gov/ij/) after 24 hours of culture by image analysis of microphotographs. At least three replicates were included within each experiment. The spreading was assessed by calculating the ratio of invaded area (the whole spheroid) over the non-invaded area (center of the spheroid).

JEG3 cells were incubated in the presence of increased concentrations of recombinant EG-VEGF (Peprotech) for 24 hours. Protein extracts were prepared as previously described (32). Both PROKR1 and PROKR2 antibodies were used at a final concentration of 0.45 μg/mL (24). β-Actin antibody (Sigma-Aldrich) was used as an internal control for protein loading. For signaling pathway analyses, membranes were incubated with phospho-MAPK (1:5,000; Promega), phospho-AKT (Ser473; 1:2,000; Cell Signaling Technology), phospho-Src (pY418; 1:1,000; Invitrogen). Blots were reprobed respectively with anti-MAPK (1:40,000; Sigma), anti-AKT (1:1,000; Cell Signaling Technology), or anti-Src (1:1,000; Millipore) antibodies to standardize for sample loading.

JEG3 cells were incubated in the absence or presence of EG-VEGF for 24 hours. Protein concentration was determined in the culture medium of all conditions. The same amount of proteins was electrophoresed under non-reducing conditions in a 10% acrylamide gel containing 1 mg/mL gelatin (Sigma-Aldrich) according to the method of Xu and colleagues (33).

Human placental tissues were collected from normal pregnant women during the first trimester of pregnancy (n = 20) and from patients with complete hydatiform moles (n = 38) or CC (n = 9). Tissues were processed as described previously (34). Specific antibodies were used to detect EG-VEGF, PROKR1, and PROKR2 (Covalab; ref. 23). Prolactin (ThermoFisher Scientific), and MMP9 (Epitomics) antibodies were also used. These antibodies were used and characterized previously (23, 24, 26, 35). To quantify the intensity of the stainings, images were processed for morphometric analysis with ImageJ software. A macro command was edited to give the total stained areas after binarization, skeletonization, and pixel count.

All animal studies were approved by the institutional guidelines and those formulated by the European Community for the Use of Experimental Animals. Two- to 3-month-old SHO SCID female mice were mated in the animal facility. The presence of a vaginal plug was observed at 0.5 dpc. The gravid mice were randomly assigned to be injected by JEG3-luc cells (see flowchart for detailed protocol, Supplementary Fig. S2A). Five groups of female mice were established: At least five animals were assigned to each experimental group, which were defined as follows: Mice in group 1 are non-gravid mice that were injected in their uterine horns with JEG3-luc embedded in Matrigel (n = 4). In group 2, mice were injected at 7.5 days post coïtus (dpc) in two opposed placenta with Matrigel alone in either uterine horns or placentas (n = 4). In group 3, mice were injected with JEG3 cells embedded in Matrigel (n = 4). In group 4, mice were injected in two opposed placenta with JEG3 cells embedded in Matrigel and at days 8.5, 11.5, 14.5, and 17.5, these mice were injected with PC7 (36), the PROKR1 antagonist [500 μg/kg, intraperitoneally (i.p.; n = 5]. In group 5, mice were injected in two opposed placentas with JEG3 cells embedded in Matrigel and at 8.5, 11.5, 14.5, and 17.5 dpc, these mice were injected with the PROKR-2 antagonist (100 mg/kg, i.p.; n = 5; ref. 37). Two other control groups of mice were injected with vehicles of PC7 or PKR505 antagonists (n = 4).

At 19.5 dpc, bioluminescence imaging was performed with a highly sensitive, cooled CCD camera, mounted in a light-tight specimen box (IVIS. In Vivo Imaging System. PerkinElmer). Before imaging, animals were anesthetized in 2% isoflurane. Ten microliters per 10 g of body weight of luciferin (potassium salt, Xenogen) was injected to the mice 15 minutes before imaging. This dose and route of administration have been shown to be optimal for studies in rodents when images were acquired within 15 minutes after luciferin administration.

For imaging, mice were placed onto the warmed stage inside the light-tight camera box, with continuous exposure to 1% to 2% isoflurane. The animals were imaged, and data were acquired for 45 seconds; this imaging time was shown to yield superior results. The low levels of light emitted from the bioluminescent tumors were detected by the IVIS camera system and were then integrated, digitized, and displayed. The regions of interest (ROI) from displayed images were designated around the tumor area and were quantified as total photon counts or in photons/s, using Living Image software (Xenogen).

After imaging of the whole body of the mouse, a laparotomy was performed to collect blood and to expose and image the uterine horn containing embryos with their attached placentas, as well as the rest of metastatic organs. A second imaging of the organs was performed and quantified as described above. Placentas and metastatic organs were collected and stored at −80°C or collected in PFA for immunohistochemistry analyses.

Mouse placentas were processed similarly to human tissues. They were incubated with the following commercial antibodies, Ki-67 (Dako) and Podocalyxin (R&D systems), Proliferin (Santa Cruz Biotechnology), Nov (Covalab), and MMP-9 (Epitomics).

Antibody-array was used to compare the expression profile and levels of angiogenesis-related proteins in sera collected from the four different groups of female mice. Antibody array experiment was assessed using Mouse Angiogenesis Array kit (R&D Systems) as recommended by the manufacturer. The intensities of immunoreactive bands were measured by scanning the photographic film and analyzing the images on a desktop computer using ImageJ software. The Chemidoc analyzing system was also used (Image Lab version 4.0.1). Angiogenesis Array kit specific for human (R&D Systems) was also used to compare sera collected from control, complete hydatiform mole patients and choriocarcinoma patients.

Statistical comparisons were made using Student t test and one-way ANOVA. All data were checked for normality and equal variance. When normality failed, a nonparametric test followed by Dunn's or Bonferroni's test was used. (SigmaPlot and SigmaStat, Jandel Scientific Software). All data are expressed as means SEM (P < 0.001, 0.01, and 0.05).

We have previously shown that EG-VEGF is secreted by the syncytiotrophoblast and that its levels are elevated during the first trimester of pregnancy (23, 24). Here, we compared its circulating and placental levels in a cohort of normal (n = 20), CHM (n = 38), and CC (n = 9) patients. Figure 1A shows that EG-VEGF levels were significantly higher in the CHM group compared with the normal group (P < 0.05). The mean value was increased by 1.5-fold in CHM patients and almost 5-fold in patients who were diagnosed with CC. Figure 1B shows representative photographs of Control, CHM, and CC placental sections stained for EG-VEGF, PROKR1, and PROKR2. EG-VEGF protein expression was markedly increased in CHM placentas (b) and was mostly localized to the syncytiotrophoblast layer. CC (c) exhibited stronger staining for EG-VEGF compared with normal (a) and CHM tissues (b). We then compared the levels of expression of PROKR1 and PROKR2 proteins in the same placental samples. There were strong increases in the intensity of PROKR1 (f and g) and PROKR2 (j and k) stainings in CHM and CC placental sections as compared with normal ones (e and i). Quantification of the intensity of the staining is reported for EG-VEGF, PROKR1, and PROKR2 in CTL, CHM, and CC patients in Supplementary Fig. S3.

We have previously shown that EG-VEGF increases normal trophoblast proliferation (23, 26), suggesting its potential involvement in CC cell proliferation. First, we demonstrated that JEG3 expresses both PROKR1 and PROKR2 receptors and that their expression is upregulated by EG-VEGF (Fig. 2A and B). PROKR1 and PROKR2 belong to GPCR proteins family and appear on the blots as two to three glycosylated forms (H, highly glycosylated form; I, intermediate glycosylated form; U, unglycosylated form). EG-VEGF (10 and 25 ng/mL) increased the expression of the H and U forms for PROKR1 (Fig. 2A) and all three forms for PROKR2 (Fig. 2B). Figure 2C shows that EG-VEGF significantly increased JEG3 cell proliferation and that this effect was abolished in the presence of PROKR1 or PROKR2 antagonists, suggesting that both receptors are involved in the EG-VEGF-mediated-proliferative effect in JEG3.

Nevertheless, when antagonists were used alone, we observed a decrease in JEG3 proliferation. This observation was in line with reports from Ngan and colleagues 2007 (38) who demonstrated that PROKR2 knockdown in glioblastoma cells induced cell apoptosis, suggesting that tumor cells are sensitive to the absence of EG-VEGF receptors. To exclude the hypothesis that this effect was due to side effects of PROKR1 and PROKR2 antagonists, we performed the same experiment, but using siRNAs for both receptors. The results of this experiment are shown in Fig. 2D and substantiate the effects observed using the antagonists, suggesting the importance of both receptors, even in the absence of their exogenous ligand.

To get more insight into the mechanism by which EG-VEGF induces the proliferation and survival of CC cells, we investigated its effects on key signaling pathways involved in these processes. Figure 2E shows that EG-VEGF induces the phosphorylation of AKT and MAPKs upon 10 minutes of treatment. Because Src phosphorylation is associated with tumor cell aggressiveness, we investigated EG-VEGF effect on Src phosphorylation in JEG3 cells. Figure 2E shows that EG-VEGF induces Src phosphorylation after 10 minutes of treatment. Standardization was also performed using β-actin.

Given the importance of cell migration and invasion in the process of tumor progression and metastasis, we investigated the effects of EG-VEGF on JEG3 migration and invasion in the absence or presence of its receptor antagonists. Supplementary Fig. S4 shows representative photographs of JEG3 at (T0) and (T24) after wounding and subsequent incubation in the absence or the presence of EG-VEGF (5, 10, and 25 ng/mL). Following a 24-hour incubation, EG-VEGF induced JEG3 cell migration at all concentrations tested. The closure of the wound reached 80% in cells treated with 10 ng/mL of EG-VEGF as compared with 60% of closure in the controls. Supplementary Fig. S4B and S4C shows that EG-VEGF effect on JEG3 migration was abolished in the presence of PROKR1 or PROKR2 antagonists. Quantification is reported in Supplementary Fig. S4D. The effect of EG-VEGF on JEG3 invasion was also tested in the presence of the PROKR1 and PROKR2 antagonists. Figure 3A shows that EG-VEGF significantly increased JEG3 invasion and that this effect was reversed in the presence of PROKR1 or PROKR2 antagonists, suggesting that EG-VEGF mediates its effect on JEG3 invasion through PROKR1 and PROKR2. Quantification of the degree of invasion is reported in Fig. 3B.

It was particularly relevant to determine EG-VEGF effect on JEG3 differentiation and invasion in a system in which the tumor architecture is maintained. Tumor cells cultured as spheroids form a topology similar to the one observed in vivo (39). Hence, we determined the effect of EG-VEGF on the degree of JEG3 spreading in the spheroid system. Figure 3C shows representative photographs of JEG3 spheroids at the time of their incubation with EG-VEGF (a–f) and 24 hours later (g–l). EG-VEGF treatment resulted in a significant increase in JEG3 spreading within the collagen. EG-VEGF effect was abolished in the presence of PROKR1 and PROKR2 antagonists. Spreading ratio for each spheroid was calculated as the ratio of the size of the whole spheroid and the size of its central core. Interestingly, while EG-VEGF treatment decreased the size of the core of the spheroid, treatments with antagonists increased the core size and decreased both basal and EG-VEGF–induced cell spreading. Quantification of the degree of JEG3 cell spreading in each condition is reported in Fig. 3D.

A critical component of tumor progression is the remodeling of extracellular matrix (ECM) by matrix metalloproteinases (MMPs). Trophoblast cells are known to express high levels of MMP-2 and MMP-9. We measured MMP-2 and MMP-9 activity in culture media collected from JEG3 cultures after their treatment with EG-VEGF (10 and 25 ng/mL). Zymography was used to assess the MMP-2 and MMP-9 activity. Figure 3E shows that MMP2 was more abundant than MMP9 in JEG3 cells and that EG-VEGF treatment significantly increased both activities. Quantification of EG-VEGF effect on MMP2 and MMP9 levels is reported in Fig. 3F.

The above clinical and in vitro studies strongly suggested that EG-VEGF might contribute to the overall development and progression of CC in vivo. To test this hypothesis, we first developed an animal model of CC that mimics CC development and progression in humans.

To mimic to the best manner CC development and progression from its primary site, the placenta, we injected JEG3-Luc cells orthotopically within the placenta of SCID mice at 7.5 dpc. Mice were monitored for tumor development and for tumor progression in other organs until 19.5 dpc (see flowchart in Supplementary Fig. S2A). To determine whether placental environment contributes to the metastasis of CC cells, we compared JEG3-Luc cells injection into the placenta versus their injection into the uterine horn of non-gravid mice. Figure 4A shows that JEG3 cell injection in the gravid placentas leads to more aggressive tumor development as compared with their injection within the uterine horn, Fig. 4B. There was an important tumor development and metastasis in multiple organs including the placenta, intestine, mammary glands, spleen, vagina, and liver, Fig. 4Aa and b. These data suggest that placental environment and vascularization contribute to CC metastasis. More importantly, we have also observed that gestation of all gravid mice injected with JEG3-luc cells was not successful, as all fetuses were dead at 19.5 dpc upon resorption of their placentas, Supplementary Fig. S5.

To get more insight into the role of EG-VEGF in CC development and progression in vivo, we treated gravid JEG3-luc mice by PROKR1 or PROKR2 antagonists at three time points during tumor progression. Figure 4C shows that mice injected with either antagonist exhibited significant decrease in tumor development and progression. These data suggest a remarkable potential of both antagonists to reduce tumor development and metastasis. Quantification of at least three mice per condition shows that PROKR1 or PROKR2 antagonists significantly decreased the intensity of quantified region of interest (ROI) compared with their vehicle, respectively, Fig. 4D. More importantly, in both PROKR1 and PROKR 2 antagonists-treated JEG3-luc mice, we observed a maintenance of the gestation with few resorbed placentas (Supplementary Fig. S5D and S5E).

Histologic comparison of placentas collected from mice injected with JEG3-Luc, or JEG3-Luc+PROKR1 antagonist or +PROKR2 antagonist is reported in Fig. 5A. Placentas collected from mice injected with JEG3-luc cells exhibited strong histologic changes with loss of all placental structures and zones (Fig. 5Ad, Ae, Af), compared with placenta collected from mice injected with matrigel (Fig. 5Aa, Ab, Ac). Interestingly, histologic analyses of placentas collected from mice injected with JEG3-Luc +PROKR1 or +PROKR2 antagonists exhibited minor placental changes, as all placental zones and structures were conserved, (Fig. 5Ag, Ah, Ai) and (Fig. 5Aj, Ak, Al). To better characterize these changes, we stained the same placentas with an antibody that recognizes human Ki67, a nuclear protein coded by MKI67 and a uniquitous marker of cellular proliferation. No nuclear staining was observed in placenta collected from mice injected with matrigel (Fig. 5Ba and Bb). In placentas collected from mice injected with JEG3-luc cells, almost all cells were positively stained with Ki67, confirming the presence of increased number of proliferative JEG3 cells. (Fig. 5Bc and Bd). Placentas collected from JEG3-Luc +PROKR1 and +PROKR2 exhibited low Ki67 staining that was limited to some placental vasculature (Fig. 5Be and Bf) and (Fig. 5Bg and Bh). Quantification of Ki67 staining in all four conditions is reported in Fig. 5C.

To gain more insights into placental changes upon tumor development, we stained placentas collected from the four groups by podocalyxin, a membrane protein that maintains vascular integrity. This protein is known to orchestrate interactions between the extracellular matrix components and basement membranes (40). Supplementary Fig. S5F shows representative placental photographs for each group. In Control mice (Supplementary Fig. S5Fa, 5Fb), podocalyxim exhibited homogeneous staining of all the vascular system within the labyrinth zone. Placentas of JEG3 injected mice showed an important disorganization of the vasculature within the labyrinth layer (Supplementary Fig. S5Fc and S5Fd). Large vessels filled with inflammatory cells could be observed within this layer. Interestingly, placentas collected from mice treated with PROKR1 (Supplementary Fig. S5Fe and S5Ff) or PROKR2 (Supplementary Fig. S5Fg and S5Fh) antagonists showed less vascular abnormalities.

To determine whether circulating angiogenic factor(s) were deregulated upon JEG3 cells injection in the placentas of the gravid mice and whether treatments with PROKR1 and PROKR2 antagonists reverse any of these parameters, we used an antibody array specific for a selection of angiogenic factors (Supplementary Fig. S6). Densitometric analysis showed that the levels of three angiogenic factors, MMP-9, Nov, and proliferin, were significantly affected upon JEG3-luc cells injection. MMP-9 was significantly increased while both NOV and proliferin were significantly decreased. Importantly, the expression of the three proteins was significantly reversed when mice were treated by PROKR1 or PROKR2 antagonists (Fig. 6A and B). These data are in line with the roles for MMP-9 and NOV proteins in tumor development. MMP-9 is well established as pro-tumoral protein (41), while Nov is known as an inhibitor of cancer cell proliferation (42). Nov overexpression resulted in reduced tumor size in glioma cell xenografts (43). The decrease in the angiogenic hormone proliferin is in line with the observation that mice injected by JEG3-luc cells had an arrested gestation. In fact, proliferin is secreted specifically by the trophoblast giant cells and ensures the establishment of the fetomaternal circulation early on during gestation. Interestingly, treated mice with either antagonist exhibited higher levels of proliferin secretion, confirming the beneficial effect of these antagonists on the pregnancy maintenance and progression. In situ analysis to search for deregulation of MMP9, Nov, and proliferin was also performed in placental tissue collected from all groups of mice. Data reported in Fig. 6C–F show that MMP9 and proliferin proteins exhibited the same changes as observed in the sera. However, Nov protein was only decreased in JEG3-Luc injected mice with no changes upon antagonists' treatments, Fig 6G and H.

As for sera analysis from the in vivo study, we have performed a human angiogenesis antibody array with sera collected from Control, CHM, and CC patients. The analysis showed that prolactin levels, the equivalent hormone of proliferin in mice, were significantly reduced both in CHM and CC compared with Control pregnant women, and MMP9 levels were increased, Supplementary Fig. S7A—S7B. These changes were confirmed by immunohistochemistry when comparing tissues collected from Control and CC patients. In the CC tissues, we have also observed a decrease in the intensity of the staining of Nov protein, Supplementary Fig. S7C–S7F. The human antibody array did not contain Nov protein, however, reductions in comparable proteins, involved in tumor suppression, were observed in the CHM and CC patients, Supplementary Fig. S8.

The present work demonstrates the direct involvement of EG-VEGF in the development and progression of CC and brings evidences for the Control of CC progression through the use of EG-VEGF receptor antagonists. Hence, we propose EG-VEGF as a key placental growth factor that should be ranked among the important etiological factors of CC. These statements are based on three observations. First, EG-VEGF circulating and placental levels are increased in CHM and in CC patients. These results are of great interest in terms of the potential use of this molecule as a prognostic biomarker of placental cancer development. EG-VEGF specificity for the placenta, its Control of trophoblast proliferation, and its regulation by hCG further support these statements (21, 23, 35). Second, EG-VEGF is a survival factor for CC cells and a strong activator of their migration and invasion. These results were obtained using the JEG3 cell line, but clearly reflect what might occur in vivo during CC development and metastasis, as JEG3 cells represent the extravillous trophoblastic cells that drive tumor progression in gestational CC. Third, we took advantage of the animal model we developed and of the EG-VEGF receptors antagonists to demonstrate that this placental factor contributes to CC development and progression in the gravid mouse.

Since its identification, EG-VEGF has been associated with the development of multiple types of cancer. It has been proposed as a potential prognostic factor in colorectal, gastrointestinal, and neuroblastoma cancer (38, 44). In the reproductive system, EG-VEGF has been reported as an indicator of ovarian and prostate cancer progression (27, 30). The discovery of EG-VEGF association to CC development and progression was predictable as this factor exhibits all features of a placental growth factor that Controls key developmental aspects of human placentation, such as trophoblast proliferation (21, 23, 35). Under physiologic conditions, EG-VEGF-mediated-proliferative effects were only observed during the first trimester of pregnancy, as EG-VEGF levels decrease by the time of the establishment of the fetomaternal circulation (23, 24, 26). In CHM and CC, maintenance of increased EG-VEGF levels might contribute to an excessive proliferation rate of trophoblastic cells, a phenomenon that ultimately results in an increased pool of cells acquiring a migratory and invasive phenotype.

EG-VEGF association with the development and progression of CC was confirmed using the newly developed animal model that mimics CC development via JEG3-luc cell injection within the placenta. In this model, the route of injection emphasizes the role of the placenta in the tumor progression, as mice injected in the uterine horn did not exhibit metastasis. The difference in the degree of metastasis might be due to the high degree of the vascularization of the placenta which might contribute to the hasty dissemination of CC cells into other organs. Also, because CC rises from the placenta during its hypoxic developmental period, one can speculate that this environment might promote tumor cells development and metastasis. Interestingly, recent data demonstrated that hypoxia generates microenvironments that can have a long-lasting influence on the dissemination of epithelial cancer cells (45, 46).

Previous studies reported the development of animal models of CC, mainly through subcutaneous injection of JEG3 cells (47). JEG3 dissemination and colonization of blood vessels were compared with non-trophoblastic tumor cells (48, 49). Nevertheless, these studies were performed in non-gravid mice and cells were not injected orthotopically within the primary site of CC development, for example, the placenta.

Gravid mice treated with PROKR1 and PROKR2 antagonists showed significant reduction in the size of the tumor development as well as a reduction in the tumor metastasis. Nevertheless, treatment with PROKR2 antagonist exhibited stronger inhibitory effects compared with PROKR1 antagonist. This discrepancy might be due to the fact that in JEG3 PROKR2 protein is increased even at lower concentrations of EG-VEGF, and that PROKR2 signaling has been reported to be associated with the control of invasive processes in trophoblast and endothelial cells (24, 26). Interestingly, in vivo use of PROKR2 antagonist has been reported to alleviate pain and to exhibit strong analgesic effects (50). Hence, these data comfort the potential use of these antagonists in the treatment of CC and/or other EG-VEGF–related cancer.

Altogether, these studies report the development of a novel preclinical animal model of CC that served to validate both the clinical observations and the in vitro studies demonstrating the role of EG-VEGF in the development and progression of CC. Our study proposes the antagonism of EG-VEGF effects as a new therapeutic approach for CC. This may provide safe and less toxic therapeutic options compared with the currently used multi-agent chemotherapies.

No potential conflicts of interest were disclosed.

Conception and design: W. Traboulsi, H. Boufettal, S. Brouillet, P. Hoffmann, V. Onnis, J.J. Feige, M. Benharouga, N. Alfaidy

Development of methodology: W. Traboulsi, F. Sergent, H. Boufettal, S. Brouillet, A. Salomon, T. Aboussaouira, M. Benharouga, N. Alfaidy

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.):W. Traboulsi, F. Sergent, R. Slim, M. Benlahfid, V. Onnis, P.A. Bolze, A. Salomon, P. Sauthier, T. Aboussaouira

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis):W. Traboulsi, S. Brouillet, M. Benlahfid, P.A. Bolze, J.J. Feige, N. Alfaidy

Writing, review, and/or revision of the manuscript:W. Traboulsi, H. Boufettal, S. Brouillet, P. Hoffmann, V. Onnis, P.A. Bolze, F. Mallet, T. Aboussaouira, J.J. Feige, M. Benharouga, N. Alfaidy

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases):W. Traboulsi, F. Sergent, S. Brouillet, Q.Y. Zhou, A. Salomon, F. Mallet, N. Alfaidy

Study supervision:S. Brouillet, P. Hoffmann, M. Benharouga, N. Alfaidy

Other (provided prokineticin antagonists): G. Balboni

The authors thank Pr N. Samouh (Ibn Rochd Hospital, Casablanca) and Pr F. Golfier, J. Massardier, M.C. Carlier. and T. Hajri (Centre des maladies gestationelles, Lyon, France) for their collaboration.

The authors acknowledge the following sources of funding: Institut National de la Santé et de la Recherche Médicale (U1036), University Grenoble-Alpes, Commissariat à l'Energie Atomique (DSV/iRTSV/BCI), Région Auvergne-Rhône-Alpes “CLARA”, Ligue Nationale contre le Cancer, and Ligue Départementale (Isère) contre le Cancer. W. Traboulsi was supported by Ligue Nationale contre le Cancer.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.