Pro-Inflammatory Profile of Preeclamptic Placental Mesenchymal Stromal Cells: New Insights into the Etiopathogenesis of Preeclampsia

The objective of the present study was to evaluate whether placental mesenchymal stromal cells (PDMSCs) derived from normal and preeclamptic (PE) chorionic villous tissue presented differences in their cytokines expression profiles. Moreover, we investigated the effects of conditioned media from normal and PE-PDMSCs on the expression of pro-inflammatory Macrophage migration Inhibitory Factor (MIF), Vascular Endothelial Growth Factor (VEGF), soluble FMS-like tyrosine kinase-1 (sFlt-1) and free β-human Chorionic Gonadotropin (βhCG) by normal term villous explants. This information will help to understand whether anomalies in PE-PDMSCs could cause or contribute to the anomalies typical of preeclampsia. Methods Chorionic villous PDMSCs were isolated from severe preeclamptic (n = 12) and physiological control term (n = 12) placentae. Control and PE-PDMSCs’s cytokines expression profiles were determined by Cytokine Array. Control and PE-PDMSCs were plated for 72 h and conditioned media (CM) was collected. Physiological villous explants (n = 48) were treated with control or PE-PDMSCs CM for 72 h and processed for mRNA and protein isolation. MIF, VEGF and sFlt-1 mRNA and protein expression were analyzed by Real Time PCR and Western Blot respectively. Free βhCG was assessed by immunofluorescent. Results Cytokine array showed increased release of pro-inflammatory cytokines by PE relative to control PDMSCs. Physiological explants treated with PE-PDMSCs CM showed significantly increased MIF and sFlt-1 expression relative to untreated and control PDMSCs CM explants. Interestingly, both control and PE-PDMSCs media induced VEGF mRNA increase while only normal PDMSCs media promoted VEGF protein accumulation. PE-PDMSCs CM explants released significantly increased amounts of free βhCG relative to normal PDMSCs CM ones. Conclusions Herein, we reported elevated production of pro-inflammatory cytokines by PE-PDMSCs. Importantly, PE PDMSCs induced a PE-like phenotype in physiological villous explants. Our data clearly depict chorionic mesenchymal stromal cells as central players in placental physiopathology, thus opening to new intriguing perspectives for the treatment of human placental-related disorders as preeclampsia.


Introduction
Preeclampsia (PE) is a severe placenta-related syndrome exclusive of human pregnancy that represents the main cause of feto-maternal mortality and morbidity worldwide [1,2]. PE affects 5-10% of all pregnancies [1,2] and, despite intensive investigation over the past decade, its etiopathogenesis still remains elusive. Clinical features of preeclampsia are severe maternal hypertension accompanied by maternal and placental exacerbated inflammatory response and generalized endothelial damage [3,4]. Even though PE resolves with placenta removal, it may cause long term complications as hypertension, cardiovascular diseases, metabolic and neurological disorders for both the mother and the newborn [5]. PE clinical symptoms become evident in the third trimester of pregnancy, but it is widely believed that they originate from anomalies in placenta development earlier on during first trimester. The PE placenta is characterized by immature trophoblast phenotype with shallow invasion of maternal spiral arteries and impaired villous vasculogenesis [6]. These aberrations lead to reduced uteroplacental perfusion and placental ischemia, with consequent increased systemic release of pro-inflammatory cytokines and anti-angiogenic factors that promote endothelial cells activation and damage [6,7,8,9]. Maternal immune maladaptation towards the feto-placental district has been implicated as a possible cause for the defective trophoblast development and related maternalplacental pathological anomalies.
The trophoblast is generally considered the main placental site of the morphological and molecular alterations responsible for the onset of preeclampsia. Nevertheless, the human placenta is a heterogeneous organ composed of several cellular populations each contributing to the organ physiology. Indeed, beside the trophoblast, placental villi are composed of endothelial cells and progenitors, myofibroblasts and mesenchymal stromal cells forming a complex network that concertedly interacts to maintain placental functionality.
Recently, the human placenta has been identified as source of a particular type of mesenchymal stromal cells (MSCs) that retain stem cell-like properties. Placental MSCs (PDMSCs) have been successfully isolated from the placental basal plate, the amnion and the decidua. They are characterized by different degrees of plasticity and possess unique immunologic and immune-regulatory properties [10,11]. MSCs do not express HLA-DR [12] and have the ability to suppress the ''mixed lymphocyte reaction'', thus being non-immunogenic and exerting an immunosuppressive effect on T-cells [12]. Recently, amniotic MSCs have been shown to promote angiogenic growth and to possess anti-inflammatory and anti-fibrotic activities mediated by the release of specific trophic factors [13].
MSCs plasticity is of great interest for regenerative medicine and several groups are exploring the possibilities offered by the placenta as a new therapeutic tool. On the other hand, the investigation of placental MSCs immune-modulatory, proangiogenic and anti-inflammatory properties could open new perspectives into the understanding of placenta-related disorders due to maternal immune maladaptation and aberrations of trophoblast development as preeclampsia. The secretion of cytokines and chemoattractant by the first trimester placenta with consequent recruitment of lymphocytes, natural killer cells and macrophages is necessary to maintain proper pregnancy physiology [14,15,16,17]. Nevertheless, the aberrant production of inflammatory molecules by placental cells could result in excessive recruitment of macrophages and immune cells leading to the abnormal placental development, inflammation and endothelial dysfunction typical of PE. Indeed, pregnancies complicated by preeclampsia and Fetal Growth Restriction are characterized by increased presence of aberrantly-activated macrophages into the placental bed, data that were correlated with reduced trophoblast invasion [18].
Our hypothesis is that anomalies in PDMSCs anti-inflammatory and pro-angiogenic properties might cause or contribute to the defective placental development and/or immune maladaptation typical of preeclampsia''. In the present study, we characterized cytokine expression profiles of PDMSCs isolated from the basal plate of normal and severe PE placentae. Moreover, we investigated the effect of conditioned media from normal and PE PDMSCs on the expression of Macrophage migration Inhibitory Factor (MIF), Vascular Endothelial Growth Factor (VEGF), soluble FMS-like tyrosine kinase-1 (sFlt-1) and b-human Chorionic Gonadotropin (bhCG) in normal placental villous explants. MIF is a pro-inflammatory cytokine key player in PE inflammatory response [19], while VEGF is the main pro-angiogenic factor sequestered and down-regulated by sFlt-1, soluble variant of VEGF receptor 1 which aberrant expression is an hallmark typical of PE [20,21,22,23,24,25]. Thus, we will determine whether PE PDMSCs could negatively influence trophoblast inflammatory response and angiogenesis.

Ethics Statement
This study was conducted according to the principles expressed in the Declaration of Helsinki. The study was approved by the Institutional Review Board of O.I.R.M. S.Anna Hospital and ''Ordine Mauriziano di Torino'' (n.209; protocol 39226/C.27.1 04/08/09) (Turin, Italy). All patients provided written informed consent for the collection of samples and subsequent analysis.

Tissue Collection
The study groups included singleton pregnancies complicated by severe preeclampsia (n = 12) and physiological control term pregnancies (n = 12). The diagnosis of PE was made according to the following criteria: presence of pregnancy-induced hypertension (systolic $140 mmHg, diastolic $90 mmHg) and proteinuria ($300 mg/24 h) after the 20 th weeks of gestation in previously normotensive women [2]. We excluded pregnancies with congenital malformations, chromosomal anomalies (of number and/or structure) or evident intra uterine infections. Physiological control placentae were obtained from normal pregnancies that did not show any signs of preeclampsia or other placental disease. Patients with diabetes, infections, kidney disease, congenital malformations and chromosomal anomalies (number and/or structure) were excluded. Placental villous samples were collected randomly from the central placental area and snap frozen immediately after delivery. Calcified, necrotic and visually ischemic areas were excluded from collection.

Placenta-derived Mesenchymal Stromal Cells (PDMSCs) Isolation
PDMSCs were isolated by using a protocol modified from Brooke and colleagues [26]. Immediately after delivery, the amniotic membranes were mechanically removed and the decidua was peeled off from the placental basal plate and discharged, in order to avoid maternal cells contamination. From each control and PE placenta, 30 grams of tissue were excised from the central part of placental cotyledons, thus avoiding decidual contamination from the placental septa, and washed several times with Hank's Buffered Salt Solution (HBSS, Gibco, Life Technoligies, Italy) to remove excess of blood. Next, the placental tissue was mechanically minced and incubated with 100 U/ml collagenase type I (Gibco, Life Technologies, Italy) plus 5 mg/ml DNAse I (Invitrogen by Life Technologies, Italy) for 3 hours. After enzymatic digestions, the cells were separated by gradient using 1.073 Ficoll Paque Premium (GE Healthcare Europe, Italy) and the mononuclear cells ring was collected, washed and cells were resuspended in Dulbecco's modified Minimum Essential Medium (DMEM, Gibco, Life Technoligies, Italy) supplemented with 10% Fetal Bovine Serum (FBS Australian origin, Italy) and seeded in T150 flask. Importantly, the culture media was not supplemented with basic fibroblast growth factor (bFGF) in order to maintain the original biochemical and molecular features of the placentaderived mesenchymal cells. The cultures were maintained at 37uC and 5% CO 2 .
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium Bromide (MTT) Assay Normal and preeclamptic PDMSCs proliferation ability was assessed by MTT assay (Sigma, Italy, Cat.no. M5655). At passage five, physiological (3 different cell lines) and PE (3 different cell lines) PDMSCs were seeded in 96-well plates at a density of 6000 cells per well. Each cell lines was plated in six different wells. At time 0, 48 h and 120 h of culture, cells were incubated with 5 mg/ ml of MTT for 4 h. Next, medium was removed and formazan salts were dissolved with 100 ml of dimethylsulfoxide (DMSO, Sigma, Italy Cat. no 000070). 570 nm absorbance at each time point was determined using a microplate spectrophotomether in order to establish PDMSCs cell proliferation curve.

Determination of Cellular Senescence
Senescence-associated b-galactosidase (SA-b-gal) nuclear staining was performed using a SA-b-gal staining kit (Sigma, Cat.no.CS0030) following manufacturer instructions. At passage five, normal (two different cell lines) and PE (2 different cell lines) PDMSCs were plated in 6 well plates at a density of 2610 5 cells/well in DMEM LG supplemented with 10% FBS. Each cell line was seeded in six different wells. After 48 h of culture, the cells were fixed in 1X Fixation Buffer for 6 minutes and were stained with SA-b-gal-staining solution for 4 h. The senescent SA-b-gal-positive cells exhibited a blue nuclear color. Number of positive cells number was determined by phasecontrast microscopy.

Conditioned Media (CM) Preparation
At passage five, PDMSCs were plated in 6 well plates at a density of 2.0610 5 cells/ml (final volume: 2 ml, 4610 5 cells total). After 72 h of culture, conditioned media was collected, centrifuged, filtered to remove cellular debries and immediately stored at 280uC, while cells were processed for mRNA and protein isolation.

Cytokines Array
To investigate differences in cytokines production between normal and preeclamptic PDMSCs, the RayBio Human Cytokine Antibody Array (#AAH-CYT-5, RayBiotech Inc, GA, USA) was used. This specific array was chosen because it allows contemporary detection and quantification of eighty different human cytokines, chemokines and inflammation-related growth factors. Arrays were performed on unconditioned and conditioned culture media from normal and PE PDMSCs (prepared as described above) following manufacturer instructions. Cytokines levels were quantified by densitometric analysis using ImageQuant software and normalized by setting the positive controls as 100 and negative controls as zero percent. The arrays were performed in duplicate on CMs derived from four normal term PDMSCs lines, four preeclamptic PDMSCs lines and on unconditioned media (UCM) as baseline for molecules already present in the culture media. UCM data were used to normalize Normal and PE PDMSCs CM results. Cell lines used were all derived from different control and PE patients.

Human Chorionic Villous Explants Cultures and PDMSCs Conditioned Media Treatment
Placental biopsies from physiological term placentae were processed within 2 hours from delivery. Fetal membranes and decidua were mechanically removed and placental tissues were washed in phosphate-buffered saline (PBS) solution to remove excess of blood. Small portions of placental chorionic villi (35 mg, n = 48) were excised and placed in a 24-well culture dish. Explants were cultured in Ham's F12 media (Gibco, Invitrogen by Life Technologies, Italy) and incubated overnight at 37uC and 5% CO 2 to equilibrate. Next, culture media was removed and explants were treated for 72 hours with 500 ml of conditioned media from 4 different control PDMSCs lines (n = 16 explants) and 4 different PE PDMSCs lines (n = 16 explants). Explants treated by unconditioned culture media were used as controls (n = 16 explants). Finally, culture media were collected while control and treated explants were processed for mRNA and protein isolation. Each experimental condition was performed in duplicate.

RNA Isolation and Real Time PCR
Total RNA was extracted from placental MSCs and chorionic villous explants using TRIZOL reagent according to manufacturer instructions (Invitrogen by Life Technologies, Italy) and treated with DNAse I to remove genomic DNA contamination. Two mg of total RNA was reverse transcribed using random hexamers approach (Fermentas Europe, St. Leon-Rot., Germany). The resulting templates were quantified by Real-time PCR (StepOne TM Real-Time PCR System, Applied Biosystems, Carlsbad, California). TaqMan primers and probes for ribosomal 18S, MIF, VEGF, HCGB, IL-8, IL-6 and TNF-a were purchased from Applied Biosystems as TaqManH Gene Expression Assays. sFlt-1 primers and probe were designed as previously described by Nevo and colleagues [23] and purchased from Applied Biosystems as Custom Gene Expression Assays. For the relative quantitation, PCR signals were compared among groups after normalization using 18S as internal reference. Relative expression and fold change was calculated according to Livak and Schmittgen [27].

Free bhCG Assay
Free b hCG released by physiological villous explants treated with normal or PE PDMSCs conditioned media was measured using the B?R?A?H?M?S Free bhCG KRYPTOR immunofluorescent assay (Thermo Scientific, Germany) following manufacturer instructions.

TNF-a and VEGF Enzyme-linked Immunosorbent Assays (ELISA)
Concentrations of TNF-a and VEGF proteins in normal and PE-PDMSCs CM and in protein lysates of CM-treated villous explants were evaluated by commercially available ELISA kits (Peprotech), according to the manufacturer's recommendations. Values are given as the mean concentration 6 SEM of fourteen (PDMSCs CM) and twelve (CM-treated villous explants) independent experiments, all repeated in duplicate.

Statistical Analysis
All data are represented as mean 6 SE. For comparison of data between multiple groups we used one-way analysis of variance (ANOVA) with posthoc Dunnett's test. For comparison between 2 groups we used paired and unpaired Student's t-test as appropriate. Statistical test were carried out using Graph Pad Prism 5 statistical software and significance was accepted at P,0.05.

Study Population
Clinical data of the study population are summarized in Table 1. Seventy-five percent of the babies from severe preeclamptic pregnancies were growth restricted (Table 1), while 44% of PE pregnancies were characterized by abnormal Doppler flow velocity waveforms in the umbilical arteries (AEDF or REDF) and 88% of them had pathological uterine Doppler.

Placenta-derived Mesenchymal Stem Cells Showed Proper MSC Antigens Profile
PDMSCs were isolated from both physiological term and preeclamptic placentae and, starting from passages 3 to 5, they were characterized for the expression of typical MSCs markers by flow cytometry. All PDMSCs lines investigated were positive for CD105, CD166, CD90 and CD73, while they were negative for HLAII, CD34, CD133, CD20, CD326, CD31, CD45 and CD14 (Fig. S1A), thus showing proper mesenchymal stem cell phenotype and excluding contamination from trophoblast/epithelial cells and hematopoietic progenitors. Moreover, all PDMSCs properly expressed both Oct4 and Nanog mRNA (Fig. S1B).

PDMSCs Isolated from PE Placentae Showed Decreased Proliferation and Increased Cellular Senescence Relative to Normal PDMSCs
PDMSCs cell proliferation at 48 h and 120 h was evaluated by MTT assay. We reported a significant cell proliferation increase in Normal PDMSCs at both 48 h (p,0.01, 2 Fold Increase) and 120 h (p,0.01, 1.4 Fold Increase vs 48 h, 2.7 Fold Increase vs time 0), whereas PE PDMSCs showed a slower and not significant proliferation rate at both 48 h and 12 h ( Figure 1). Importantly, despite Normal and PE PDMSCs had the same cell number at time 0, cell proliferation at 120 h, as determined by MTT assay, was significantly increased in Normal vs PE PDMSCs (p,0.01, 1.4 Fold Increase, Figure 1). We next performed senescenceassociated b-galactosidase (SA-b-gal) staining in normal and preeclamptic PDMSCs. We found significantly increased percentage of positive/senescent cells in PE relative to normal PDMSCs (p,0.01, 3.2 Fold Increase, Figure 2B).

Cytokines Expression Profiles in Conditioned Media from Normal and Preeclamptic PDMSCs
Mesenchymal stem cells are a promising tool for regenerative medicine because of their unique immunomodulatory, provasculogenic and anti-inflammatory properties. Since preeclamp- sia is characterized by an exacerbated feto-maternal inflammatory response accompanied by generalized endothelial damage and aberrant placental vascular development, we investigated whether PDMSCs derived from normal and PE placentae showed differences in cytokines production profiles. These data would help us to understand whether or not PDMSCs could contribute to physiological placenta development and/or to the aberrant placentation and inflammatory response typical of PE. To reach our goal, we performed a cytokine array able to detect 80 different cytokines (Table S1) present in the conditioned media of both control and PE PDMSCs. Cytokines expression levels in Normal and PE PDMSCs conditioned media were compared to positive and negative standards (used as 100% and 0% expression ranges respectively) ( Figure 3A). The results reported in Table S1 and in  Table S1).
Furthermore, we performed Real Time PCR analysis on normal and PE-PDMSCs in order to confirm increased expression of IL-8, IL-6, MIF, TNF-a and VEGF as detected by Cytokine Array analysis ( Figure S2). We selected this specific pool of molecules because they represent key players in PE pathogenesis (MIF, TNF-a and VEGF) and the main pro-inflammatory cytokines over-expressed by PE-PDMSCs (IL-8 and IL-6). We found significantly increased expression of MIF (p,0.001, 2.9 Fold Increase), VEGF (p,0.01, 1.9 Fold Increase), TNF-a (p,0.01, 2 Fold Increase), IL-8 (p = 0.02, 2.3 Fold Increase) and IL-6 (p = 0.02, 2.2 Fold Increase), thus confirming Cytokine Array results ( Figure S2). Finally, we further corroborated our findings on PE-PDMSCs increased TNF-a and VEGF protein release by using specific ELISA assays. We found that PE-PDMSCs CM contains significantly higher concentrations of both TNF-a (p,0.01, 34 Fold Increase) and VEGF (p = 0.02, 1.55 Fold Increase) molecules relative to normal PDMSCs CM ( Figure S3A).

Conditioned Media from Preeclamptic PDMSCs Induced MIF, VEGF, sFlt-1 and Ki67 Over-expression in Physiological Human Villous Explants
Since we determined that preeclamptic PDMSCs released augmented concentrations of pro-inflammatory cytokines and chemotactic factors, we next investigated whether they were able to negatively influence inflammation and/or vasculogenesis on the placental tissue. To reach our goal, we treated term physiological chorionic villous explants with conditioned media from normal and PE PDMSCs. As markers of inflammation and vasculogenesis, we investigated MIF and VEGF expression. We found significantly increased MIF mRNA expression levels (p = 0.033, 1.75 Fold Increase) in PE PDMSCs CM explants relative to both untreated controls and explants treated with normal PDMSCs conditioned media ( Figure 4A, left panel). Results were confirmed at the protein levels where we described significantly increased MIF protein levels (p = 0.027, 2.4 Fold Increase) in preeclamptic PDMSCs media explants relative to both untreated controls and explants treated with normal PDMSCs CM ( Figure 4B, left upper and lower panels). VEGF mRNA expression was significantly increased in both PE and normal PDMSCs CM explants relative to untreated controls (p,0.01, 2.5 Fold Increase, Figure 4A, left panel). In stark contrast to mRNA data, only normal PDMSCs conditioned media was able to induce a significant VEGF protein accumulation (p,0.01, 35 Fold increase) relative to both untreated and PE PDMSCs CM explants ( Figure 4B, right upper and lower panels). These findings were confirmed also by ELISA assay. Indeed, we found significantly increased VEGF protein concentration in normal PDMSCs CM-treated explants relative to both untreated controls (p,0.01, 1.4 Fold Increase) and PE-PDMSCs CM-treated explants (p,0.01, 1.6 Fold Increase) ( Figure S3B, right panel). Anti-angiogenic sFlt-1 is the main responsible for VEGF down regulation during preeclampsia. Herein, we found significantly increased sFlt-1 mRNA expression in PE PDMSCs CM explants relative to both normal PDMSCs CM explants and untreated controls (p,0.001, 2.4 and 1.5 Fold Increase respectively; Figure 4C left panel). Data were confirmed at the protein levels, where we reported significantly increased sFlt-1 levels in explants treated by PE PDMSCs CM relative to both normal PDMSCs CM and untreated control explants (p = 0.003, 2.3 and 1.4 Fold Increase respectively; Figure 4C right upper and lower panels). Finally, we reported significantly increased expression of Ki67, marker of cell proliferation, in chorionic villous explants treated by PE PDMSCs media (p = 0.044, 4.8 Fold Increase) compared to normal PDMSCs CM explants and untreated controls ( Figure 5B, left panel). The pro-inflammatory profile of PE-PDMSC CM explants was further confirmed by significantly increased TNF-a protein levels relative to both normal PDMSCs CM explants (p,0.01, 4.25 Fold Increase) and untreated controls (p = 0.014, 1.87 Fold Increase) as detected by ELISA assay (Figure S3B left panel).
Villous explants are representative of the whole placental tissue and composed of trophoblast, mesenchymal and endothelial cells. Thus, we investigated the contribution of these cell populations to the above mentioned results by determining the expression of Cyt7 (trophoblast marker), CD45 (haematopoietic/endothelial cell marker) and CD166 (MSCs marker) in our experimental model. Cyt7 was strongly expressed without differences in untreated, normal and PE PDMSCs CM explants (p,0.05, Figure 5B right upper panel). In contrast, CD45 and CD166 were barely detectable ( Figure 5B right middle and lower panels), thus indicating that trophoblast cells were the main villous explants cellular component.

Increased Release of Free bhCG by Villous Explants Treated with PE PDMSCs Conditioned Media
In order to further characterize the preeclamptic-like phenotype induced by PE PDMSCs conditioned media, we compared the amount of free bhCG, well recognized PE biomarker [28,29], released by normal and PE PDMSCs CM explants. We found significantly increased bhCG mRNA levels in PE relative to normal PDMSCs CM explants (p = 0.01, 7.45 Fold Increase,   Figure 5A left panel). Importantly, immunofluorescent assay confirmed mRNA results revealing significantly increased free bhCG levels (p = 0.03, 1.6 Fold Increase) in media from PE PDMSCs CM explants relative to controls ( Figure 5A right panel).

Discussion
Preeclampsia is a multi-factorial placenta-related syndrome [3] which anomalies originate early on during the first stages of placenta development. One of the most recognized theories describes the poor placentation typical of PE as the result of deficient mother-trophoblast immune acceptance [30]. This pathological condition will lead to diminished placenta perfusion and oxidative stress [31,32,33,34,35], increased trophoblast apoptosis and turnover [36,37,38,39,40], release of pro-inflammatory cytokines and syncitiotrophoblast debris into the maternal circulation that directly damage the endothelium [41] However, the precise pathogenic events leading to preeclampsia still remain unclear.
In the present study, we reported aberrant release of proinflammatory cytokines by mesenchymal stromal cells derived from preeclamptic placentae relative to normal PDMSCs. Moreover, we described a slow proliferation rate accompanied by increased senescence in PE-PDMSCs, suggesting a compromised self-renewal capacity of these cells resident in the villous stroma. Such a defect could significantly contribute to the aberrant villous architecture typical of PE placentae and it is in stark contrast with the well established hyper-proliferative phenotype of the neighboring PE trophoblast.
Importantly, we demonstrated for the first time to our knowledge that conditioned media from PE-PDMSCs induces a preeclamptic-like phenotype in normal placental villous explants. PE-PDMSC media promoted the expression of pro-inflammatory MIF and anti-angiogenic sFlt-1 in normal term villous explants while, at the protein level, it induced VEGF down regulation.
The discovery of such unique cells as MSCs resident in the placental tissues (basal plate, amnion and decidua) brought us to the hypothesis that MSCs themselves could contribute to the pathophysiology of placentation in virtue of their immuneregulatory and anti-inflammatory properties. MSCs exert an immunosuppressive effect on T-cells [11] and secrete paracrine factors that act on vasculogenesis and protect against ischemic injury [42]. Herein, we analyzed by cytokine array technology the expression of eighty different cytokines and growth factors in MSCs isolated from the chorionic villous portion of normal and PE placentae. We investigated this specific MSCs population because it localizes in the villous mesenchyme, thus being able to interact with the neighboring trophoblast, mesenchyme and endothelial cells.
Hwang and colleagues recently characterized cytokine production profiles of MSCs isolated from amnion and decidua of healthy versus preeclamptic placentae [43]. Interestingly, they didn't find differences in cytokines expression between normal and PE amnion-derived MSCs [43] while they reported higher levels of SDF-1 and no differences in MCP-1 expression in normal relative to PE decidual MSCs. Their results differ from our data most likely because we analyzed a different MSCs population isolated from the placental villi instead of the amnion or the decidua. In PE-PDMSCs media, we found increased levels of MCP-1 and MCP-3, type I inflammatory chemokines able to recruit monocytes, memory T cells, eosinophils and dendritic cells [44,45,46,47]. During physiological early pregnancy, MCP-1 is produced by first trimester extravillous trophoblast cells following TNF stimulation [48], whereas MCP-3 is down-regulated through peri-and post-implantation stages to let the blastocyst invade and get acquainted to the uterine environment [49]. The aberrant release of Th1 cytokines (cell-mediated immune response) MCP-1 and/or MCP-3 by PE-PDMSCs is indicative of a failure in the Th2 (humoral immunity) switch typical of physiological pregnan-cies, necessary for the correct placental immune-acceptance by the maternal system.
In line with our results, Szarka and colleagues [50] reported increased levels of pro-inflammatory IL-8, IL-6, IP10, TNF-a and MCP-1 in serum from preeclamptic patients. This cytokines profile was indicated as an hallmark of abnormal pro-inflammatory systemic environment [50]. The increased release of these molecules by PE-PDMSCs could have important pathological implications. IL-8 is a potent chemoattractant for neutrophils while IL-6 has a key role in the acute-phase response [51]. IP-10 is a pro-inflammatory and anti-angiogenic cytokine that has been proposed to link inflammation and anti-angiogenesis in preeclampsia [52]. TNF-a is a type I cytokine promoter of cellularmediated immune response that was previously reported as overexpressed in PE placentae and maternal serum [53,54]. TNF-a inhibits trophoblast cell mobility in vitro [55] and it reduces trophoblast invasive capacity [56,57,58]. Our data suggest that PE-PDMSCs could directly affect trophoblast invasivity via TNFa release. Moreover, TNF-a promotes the expression of several inflammatory molecules, thus inducing a vicious pro-inflammatory loop detrimental for pregnancy physiology.
Other molecules abundantly over-expressed in PE-PDMSCs media are TGF-b2, TIMP-2, Osteoprotegerin and Osteopontin. Transforming Growth Factor-bs have been indicated as main regulators of trophoblast invasion [59,60]. Moreover, TGF-b1 [61,62], TGF-b2 [63] and TGF-b3 [60] have been found increased in the maternal circulation and placental tissue of PE and FGR pregnancies, suggesting a role for TGF-bs in the trophoblast dysfunction typical of placenta-related pathologies. Herein, we found elevated levels of TGF-b2 and, to a lesser extent, TGF-b1 and TGF-b3 in the conditioned media of PE-PDMSCs, thus indicating a contribution of mesenchymal TGF-bs, TGF-b2 in particular, in PE dysfunction.
TIMPs are matrix metalloproteinase inhibitors that suppress proliferation of endothelial cells, thus exerting an anti-angiogenic activity. TIMP-2 has a key role during pregnancy in maintaining the homeostasis of trophoblast invasivity. Thus, TIMP-2 expression is low during the first stages of pregnancy, to increase with advancing gestation [64]. In accordance to our results, TIMP-2 immunoreactivity was previously found increased in syncitiotrophoblast and mesenchymal cells of PE placentae [65]. Data obtained in COMT knockout mice model, suggested that overproduction of both TIMP-2 and TGF-b3 may result in shallow utero-placental circulation and in the onset of a preeclamptic-like phenotype [65,66].
Osteoprotegerin (OPG) inhibit bone resorption, regulates vascular integrity [67,68], stimulates proliferation, inflammation and fibrogenesis in vascular smooth muscle cells [69]. The human placenta is considered the main source of OPG during gestation [70]. A significant increase in serum OPG concentration was observed during pregnancy [71] and, according to our data obtained on PE PDMSCs, placental and circulating OPG levels further increase in preeclampsia [72]. Osteopontin (OPN) is involved in adhesion and signal transduction at the uterineplacental interface during implantation and placentation [73]. Importantly, OPN acts as a pro-inflammatory Th1 type cytokine in vivo [74] and preeclamptic patients with extensive endothelial injury are characterized by increased plasma levels of OPN [75]. Thus, the increased OPN production by PE-PDMSCs that we reported could cause and/or further contribute to the aberrant placental inflammation underlying preeclampsia.
PE-PDMSCs expressed also higher levels of pro-angiogenic growth factors VEGF, Angiogenin and PlGF. Besides being pivotal for placentation, these angiogenic molecules can be strong endothelial cell activators. VEGF role in PE pregnancy has been controversially discussed, as its levels in PE maternal serum have been reported as increased [76,77,78], decreased [79,80], or even unchanged [81] Angiogenin expression in the human placenta increases during gestation [82] and it is further up-regulated in fetal growth restricted pregnancies relative to controls [83]. Placental Growth Factor (PlGF) expression is limited to the placental tissue where it promotes villous angiogenesis during early gestation. PlGF plasma levels and function are dramatically compromised in PE patients, resulting in endothelial injury and vascular permeability [22]. Indeed, increased PlGF expression by PE-PDMSCs could also indicate a compensatory mechanisms acted to induce new angiogenesis in the compromised PE placental vasculature.
Leptin, LIF, PDGF-BB and BDNF are as well augmented in PE-PDMSCs relative to control PDMSCs media. During pregnancy, these cytokines are involved in critical processes as placental invasion [84], trophoblast cell proliferation, fetal growth and development [84,85]. Moreover, they stimulates cytokines production [84] and are directly implicated in inflammation [85], suggesting a pathophysiological role in preeclampsia. Leptin and LIF have been previously reported to be over-expressed at both placental and plasma levels in severe preeclamptic pregnancies [86,87]. Leptin, in particular, could aggravate hypertension and endothelial damage by promoting catecholamine production [86]. On the other hand, PDGF-BB is a potent chemoattractor that drives mesenchymal cells migration [88], thus its increase in PE-PDMSCs media could be interpreted as an attempt to recruit MSCs and repair placental damage.
Taken together, our data clearly indicate that preeclamptic MSCs resident in the chorionic placental villi over-produce a repertoire of cytokines, chemokines and growth factors that could directly contribute to the anomalies of trophoblast development and placental angiogenesis and to the exacerbated inflammatory response typical of PE and FGR. These molecules, crucial for proper pregnancy outcome under normal conditions, when abnormally expressed are detrimental for placenta physiology. In line with our results, Waterman and colleagues described two different immune-modulating profiles in human MSCs isolated from bone marrow. They found pro-inflammatory Toll-like Receptor 4 (TLR4) primed MSCs, that expressed high levels of IL-8, IL-6 and TGF-b2 molecules, and immunosuppressive TLR3-primed MSCs, characterized by marked expression of IL-4. IL-10, RANTES and IP-10 [89]. We reported significantly increased levels of IL-8, IL-6 and TGF-b2 cytokines in preeclamptic placental MSCs relative to controls, thus indicating that PE PDMSCS resemble a pro-inflammatory TLR4-primed phenotype.
Therefore, we next investigated whether PE-PDMSCs could alter the physiological placental equilibrium. We treated normal chorionic villous explants with media conditioned by normal or PE PDMSCs and investigated the expression of MIF, VEGF, sFlt-1, Ki67 and bhCG molecules. MIF is a central regulator of inflammation and immune response [90] described in both villous and extravillous trophoblast in humans [91,92]. Emerging evidences emphasize its importance in early pregnancy, embryonic development and preeclampsia [19,91,92,93]. Herein, we found that normal villous explants treated with PE PDMSCs conditioned media over-expressed MIF at both gene and protein levels, while CM from normal PDMSCs did not affect physiological MIF production. Our group previously reported that MIF serum levels were significantly increased in women affected by severe PE [19,93], thus entailing a role for MIF in the etiopathogenesis of PE. Importantly, TNF-a, beside being directly induced by MIF, promotes MIF production by a positive feedback [94]. Indeed, the high TNF-a levels secreted by PE PDMSCs could explain MIF mRNA and protein accumulation. This PE-PDMSCs induced alteration in MIF expression further confirm the contribution of pathological mesenchymal stromal cells to the pathogenesis of preeclampsia.
As discussed above, VEGF orchestrates angiogenesis and it is vital for early embryonic development [95]. VEGF is involved also in the pathogenesis of preeclampsia. The preeclamptic placenta over-expresses VEGF in response to the hypoxic environment [21,96], but free and biologically active VEGF protein is sequestered and down-regulated by sFlt-1, soluble variant of VEGF receptor 1 released in excess by the PE placenta [22,23,24,25]. In the present study, both normal and preeclamptic PDMSCs's media promoted VEGF mRNA expression compared to untreated chorionic villous explants, while only PE PDMSCs media induced sFlt-1 mRNA accumulation. In accordance to mRNA data, normal PDMSCs's media treatment was accompanied by VEGF protein accumulation, thus confirming the well documented pro-angiogenic activity of mesenchymal stromal cells [42]. In stark contrast, treatment with PE PDMSCs's media resulted not in VEGF but in sFlt-1 protein increase. Our results resemble the in vivo condition typical of preeclampsia, where the VEGF placental mRNA increase induced by the hypoxic and pro-inflammatory environment is antagonized by the over-expression of anti-angiogenic sFlt-1.
Moreover, preeclamptic placentae are characterized by immature hyper-proliferative trophoblast phenotype. Thus, we investigated Ki67 expression in villous explants treated by normal or PE PDMSCs conditioned media. We reported significant Ki67increase in PE-PDMSCs CM explants, indicating the induction of an hypertrophic PE-like status by preeclamptic PDMSCs.
As a further confirmation of the PE-like environment provoked by PE-PDMSCs, we reported augmented release of free bhCG by normal villous explants treated with preeclamptic PDMSCs media. hCG is normally secreted by trophoblast cells to maintain maternal vascular supply during pregnancy and its increased circulating levels have been associated with a dramatically higher risk of developing preeclampsia [28,29].
We are currently investigating whether normal placental MSCs are able to revert the pathological inflammatory conditions typical of PE placentae.
In conclusion, our data clearly depict chorionic placental mesenchymal stromal cells as central players in placental physiopathology, thus opening to new intriguing perspectives for the treatment of human placental-related disorders as preeclampsia and fetal growth restriction.