Placental endovascular extravillous trophoblasts (enEVTs) educate maternal T-cell differentiation along the maternal-placental circulation.

Abstract Objectives During human pregnancy, the endothelial cells of the uterine spiral arteries (SPA) are extensively replaced by a subtype of placental trophoblasts, endovascular extravillous trophoblasts (enEVTs), thus establishing a placental‐maternal circulation. On this pathway, foetus‐derived placental villi and enEVTs bath into the maternal blood that perfuses along SPA being not attacked by the maternal lymphocytes. We aimed to reveal the underlying mechanism of such immune tolerance. Methods In situ hybridization, immunofluorescence, ELISA and FCM assay were performed to examine TGF‐β1 expression and distribution of regulatory T cells (Tregs) along the placental‐maternal circulation route. The primary enEVTs, interstitial extravillous trophoblasts (iEVTs) and decidual endothelial cells (dECs) were purified by FACS, and their conditioned media were collected to treat naïve CD4+ T cells. Treg differentiation was measured by FLOW and CFSE assays. Results We found that enEVTs but not iEVTs or dECs actively produced TGF‐β1. The primary enEVTs significantly promoted naïve CD4+ T‐cell differentiation into immunosuppressive FOXP3+ Tregs, and this effect was dependent on TGF‐β1. In recurrent spontaneous abortion (RSA) patients, an evidently reduced proportion of TGF‐β1–producing enEVTs and their ability to educate Tregs differentiation were observed. Conclusions Our findings demonstrate a unique immune‐regulatory characteristic of placental enEVTs to develop immune tolerance along the placental‐maternal circulation. New insights into the pathogenesis of RSA are also suggested.


| INTRODUC TI ON
The healthy growth of a semi-allogeneic foetus in uterus requires the adaptive regulation of the maternal immune system to establish an immune-tolerant environment at the maternal-foetal interface.
Abnormal immune regulation is tightly associated with various pregnancy disorders including recurrent spontaneous abortion (RSA). [1][2][3][4] RSA affects approximately 1% of the childbearing families. 5 Around half of RSA cases are derived from unknown reasons, and immune factors have been suggested to be the most important causes in these patients. [6][7][8][9] Increasing evidence demonstrates the essential effects of T-cell differentiation in pregnancy maintenance. The immune-tolerant Th2 and Treg bias during pregnancy are critical to protect the foetus from maternal immune attack. [10][11][12][13] In vitro and in vivo studies have demonstrated that Tregs sustain immune homoeostasis by suppressing the activation of other leucocytes. 14 In placental mammals, the generation of Tregs can mitigate the maternal-foetal conflict, and Treg depletion leads to adverse pregnancy outcomes. 15 In humans, the proportion of Tregs in peripheral blood and decidua elevates during gestation and returns to non-pregnant status after delivery. [16][17][18] The RSA patients exhibit significantly lower numbers of Tregs in both peripheral blood and decidua than healthy pregnant ones. [19][20][21][22] Transforming growth factor-beta1 (TGF-β1) is a key factor that triggers the differentiation of the inducible Tregs through induction of transcription factor FOXP3. 23,24 In parallel with the change in Treg proportion, peripheral TGF-β1 level is higher in pregnant women compared with the non-pregnant individuals, 25 and its expression is sharply reduced in the peripheral blood and decidua of RSA patients. [26][27][28][29] It is most likely that TGF-β1-regulated differentiation of Tregs critically participates in maintaining a healthy pregnancy.
Physiologically, a placental-maternal circulatory system is established through the remodelling of uterine spiral arteries (SPA).
A subtype of extravillous trophoblasts (EVTs), the endovascular extravillous trophoblasts (enEVTs), invade into SPA and replace the endothelial cells. Another subtype of EVTs, interstitial extravillous trophoblasts (iEVTs), invade decidual stroma and eventually help the loss of vascular smooth muscle cells. The uterine SPA is therefore remodelled into low-resistance, high-capacity uteroplacental arteries to ensure the sufficient maternal blood perfusion from SPA to intervillous space (IVS). [30][31][32][33] It is currently unclear how the foetal-derived enEVTs in the remodelled SPA and the placenta villous trophoblasts immersing into maternal blood at IVS directly contact the maternal lymphocytes, while do not cause maternal immune rejection along the placental-maternal circulatory pathway.
Considering the localization of enEVTs and the route of maternal blood perfusion, we hypothesize that enEVTs in SPA may essentially educate maternal T-cell differentiation when maternal blood flows through the remodelled SPA, therefore contributing to local immune tolerance along the placental-maternal circulation.
To address this hypothesis, we comparatively examined the distribution of Tregs in SPA and IVS from healthy pregnant and RSA women. The TGF-β1 expression in enEVTs, iEVTs and decidual endothelial cells (dECs) was detected using in situ hybridization, flow cytometry analysis and specific ELISA . The naïve   CD4 + T cells isolated from healthy women or virgin female mice   were treated with the conditioned media from the primarily   cultured enEVTs, iEVTs or dECs, respectively, and their differentiation towards functional Tregs was analysed by flow cytometry analysis and carboxyfluorescein succinimidyl ester (CFSE) assays. Our data showed that enEVT is a unique trophoblast subpopulation that can mainly produce TGF-β1 and induce maternal Treg differentiation. The dysfunction of enEVTs in RSA patients correlates with the failure of their Treg differentiation.
Our findings reveal the physiological significance of replacing uterine blood vessel endothelial cells by placental enEVTs from the aspect of immune tolerance and provide a new understanding of the placental pathology of the pregnant disorders such as RSA.

| Sample collection
The placental villi and decidual tissues from healthy pregnant women (n = 65) or RSA patients (n = 10) at gestational weeks 7-9 were collected upon therapeutic termination of pregnancy at the Reproductive Medicine. In brief, these patients had a history of two or more failed pregnancies for unknown reasons. 34 Women who manifested an endocrine disorder, foetal chromosomal or congenital abnormalities, uterine anatomical disorders, renal disease or pregnancies conceived by fertility treatment were excluded from this study. All the enrolled patients had arrested foetal development for less than one week before the termination of pregnancy. The clinical characteristics of the pregnant women enrolled in this study are summarized in Table S1.
Mouse spleen tissues were obtained from virgin female SPF C57BL/6 (B6) mice aged 8 weeks (Beijing SPF Biotechnology Co. Ltd.). The experimental procedure was approved by the Animal Welfare and Ethics Committees of the Institute of Zoology, Chinese Academy of Sciences.

| Immunohistochemistry
Human decidual tissues were fixed in 4% paraformaldehyde and subjected to routine dehydration and embedding in paraffin wax.

| Immunofluorescence
Human decidual or villous tissues were embedded in OCT compound Image-Pro Plus 6.0 (Media Cybernetics). At least three random views were analysed for each section, and 10 cases each from healthy pregnancy or RSA group were randomly selected for analysis.

| ELISA for TGF-β1
Levels of TGF-β1 secretion in cell supernatants were analysed by using a sandwich ELISA according to the manufacturer's instruction (ProteinTech, KE00002). In brief, the conditioned media from cultured cells were pre-incubated with 1 N HCl followed by neutralization with 1 N NaOH and subjected to sandwich ELISA. The levels of TGF-β1 were determined based on the standard curve.

| Statistical analysis
The statistical analyses were performed with GraphPad Prism ver-

| Distribution pattern of Tregs along the placental-maternal circulation pathway
To illustrate the distribution of Tregs at the maternal-foetal interface, especially along the placental-maternal circulation pathway, we performed immunofluorescence staining for CK7 and FOXP3 in human decidual tissues at early pregnancy, which specifically marked trophoblasts and Tregs, respectively. In typical pregnant cases ( Figure 1A-E), FOXP3 + Tregs existed in the lumen of the remodelled SPA ( Figure 1A,B) and the IVS area ( Figure 1D,E). The area of SPA or IVS in one view was measured by Image-Pro, and the number of Tregs in unit area of SPA and IVS was statistically quantified. Data revealed that in RSA decidua ( Figure 1F,J), the proportion of FOXP3 + Tregs in the lumen of remodelled SPA ( Figure 1F,G) and IVS ( Figure 1I,J) were significantly lower than that in normal pregnancy decidua ( Figure 1M,N). Few Tregs were found in the non-remodelled SPA, either in normal ( Figure 1K,N) or in RSA ( Figure 1L,N) pregnancy. In addition, very few FOXP3 + Tregs were observed in the decidual stroma, where iEVTs were clustered ( Figure 1C,H).

| enEVTs, but not iEVTs or dECs, could specifically produce TGF-β1
The above data showed specific distribution of FOXP3 + Tregs along the route of maternal blood perfusion. It is well known that TGF-β1 is the master regulator that triggers Treg differentiation. 23,24 To address whether the FOXP3 + Tregs in SPA and RSA were potentially induced by any TGF-β1-producing cells at the placental-maternal circulation, we first detected the expression of TGF-β1 in decidual tissues.
In normal decidua at early pregnancy, in situ hybridization for TGF-β1 and immunohistochemistry for an EVT marker, HLA-G, were conducted. We found strong positive signals for TGF-β1 in the majority of enEVTs in the remodelled SPA (Figure 2A), while few signals in iEVTs ( Figure 2B). The dECs that were marked by CD31 exhibited also scarce TGF-β1 signal ( Figure 2C).
We then isolated the primary enEVTs, iEVTs and dECs from normal decidual tissues at early pregnancy to measure TGF-β1 production. So far, the knowledge of enEVT properties is limited, and the known markers include HLA-G, NCAM1 and Jagged1. [35][36][37][38] Our immunofluorescence and immunohistochemistry in human decidual tissues clearly showed that enEVTs were positive for both HLA-G and NCAM1 ( Figure 2D, Figure S1A), while iEVTs were positive for HLA-G but negative for NCAM1 ( Figure 2E). Therefore, FACS was performed to obtain NCAM1 + HLA-G + enEVTs and NCAM1 -HLA-G + iEVTs from human decidual tissues, and their proportions were around 0.3% and 15% of whole decidual cells, respectively ( Figure 2F, Figure S2). Immunofluorescence for Jagged1 and CK7 further proved the high purity of these isolated cells ( Figure S1B).
dECs were isolated as NCAM1 -CD31 + cells in the decidual tissues ( Figure 2G). In these primary cells, flow cytometry analysis of TGF-β1 revealed that approximately 22% of enEVTs, 4% of iEVTs ( Figure 2F) and 1% of dECs ( Figure 2G) expressed TGF-β1. Statistical analysis showed that the percentage of TGF-β1-positive cells in enEVTs was significantly higher than that in iEVTs and dECs ( Figure 2H).
The isolated primary cells were cultured for 24 hours, and TGF-β1 concentration in the supernatants/conditioned media was measured by using specific ELISA. The cell-free medium was used as the negative control (NC). The concentration of TGF-β1 in iEVTs or dECs was comparable to NC, while that in enEVTs was approximately fourfold higher of NC ( Figure 2I).
The data revealed that enEVTs, but not iEVTs or dECs, were the unique cells being capable of producing TGF-β1, and their number, as well as TGF-β1 production, decreased remarkably in RSA pregnancy.

| enEVTs utilized TGF-β1 to promote functional Treg differentiation
To explore whether enEVTs are efficient in priming Treg differentiation, we cultured the primary enEVTs for 24 hours and collected their conditioned media (enEVT-CM) to treat naïve CD4 + T cells.
Human naïve CD4 + T cells were isolated from the peripheral blood of healthy non-pregnant women by using negatively selected magnetic sorting. Cell purity was more than 97%, as revealed by flow cytometry analysis for CD45RA and CD4 ( Figure S3A). The purified human naïve CD4 + T cells were activated with anti-CD3 and anti-CD28 antibodies, followed by treatment with 50% enEVT-CM for three days. The cells in the control group were treated with enEVT cell-free-conditioned media. The percentage of CD4 + CD25 + FOXP3 + Tregs in enEVT-CMtreated group was over 40% ( Figure 4B), being approximately 13-fold higher than that in control ( Figure 4A,F). The enEVT-CM was then pre-incubated with the blocking antibody against TGF-β1 ( Figure 4D To evaluate whether the enEVT-primed Tregs are functional or not, we isolated the enEVT-primed mouse Treg and co-cultured with mouse CD4 + CD25 − T cells to observe T-cell proliferation by CFSE assay. We found that CD4 + CD25 − T-cell proliferation was significantly reduced by more than 45% after co-culturing with the enEVTprimed Tregs ( Figure 5A,B), indicating that the enEVT-primed Tregs were immunosuppressive.

| Neither iEVTs nor dECs could induce differentiation of Tregs
We cultured the primary iEVTs and dECs and collected their conditioned media (iEVT-CM and dEC-CM) at 24 hours of culture. Either human or mouse naïve CD4 + T cells were treated with 50% iEVT-CM or dEC-CM for three days. As shown, neither iEVT-CM ( Figure S4e,g) nor dEC-CM ( Figure S4f,h) had any effect on human or mouse T-cell differentiation towards CD4 + CD25 + FOXP3 + Tregs ( Figure S4i,j).
The results indicated that enEVTs were functionally different from iEVTs and dECs in inducing differentiation of Tregs.

| D ISCUSS I ON
The The interaction between enEVTs and Tregs, as revealed in this study, provides novel evidence to understand why the uterine endothelial cells need to be replaced by enEVTs in the remodelled SPA. It has been believed that iEVTs and enEVTs function together to make the decidual spiral arteries less resistant and of higher capacity.
However, the properties of enEVTs have not been well recognized.
Evidence indicated the specific expression of VE-Cadherin, NCAM1 and Jagged1 in enEVTs. In addition, our data revealed the strong and specific ability of enEVTs to produce TGF-β1. In vitro co-culture study by Tilburgs et al 44  into CD16 -NK cells with similarity to dNK phenotype. 47 We assume that enEVTs may also participate in educating pNKs into dNKs by producing TGFβ1 and other cytokines along the placental-maternal circulation. These evidences convincingly explain the physiological significance of replacing the spiral artery endothelial cells by enEVTs from the aspect of immune tolerance. Further characterization of enEVTs in regulating immune cell composition at the maternal-foetal interface is of key importance for understanding the pathogenesis of severe pregnancy disorders such as RSA and preeclampsia.
It has been suggested that the increase in Tregs in decidual tissue was due to the local expansion or a selective recruitment of Tregs to the maternal-foetal interface. 48 Figure S1A). It has been suggested that enEVTs floating in the lumen may function to reduce the blood flow velocity. [50][51][52] We did not find any convincing correlation between TGFβ1 expression in enEVTs and their localization.
Further investigation is needed to clarify the subtype properties of enEVTs, which will help to expand our understanding of pregnancy maintenance, as well as the aetiology of pregnancy disorders, such as RSA and preeclampsia.
It has been reported that the spiral artery remodelling is insufficient in RSA patients, 53 in line with our observations that the number of remodelled SPA and enEVTs in SPA significantly reduced in RSA decidua. In addition, the proportion of TGF-β1producing enEVTs and the production of TGF-β1 in enEVTs were much less in RSA decidua. This is in parallel with the evidence that circulating TGF-β1 concentration in RSA cases was significantly lower than normal pregnant controls. [26][27][28][29] Interestingly, TGF-β1 can promote SPA remodelling through inducing HIF-1α expression and subsequently stimulating VEGF expression in trophoblasts. 54,55 It is feasible to suggest that the reduction in TGF-β1 in RSA decidua may, at least partially, lead to insufficient SPA remodelling, which further impairs the immune cell differentiation.
Therefore, the multiple roles of TGF-β1 in regulating trophoblast function and maintaining immune tolerance indicate its central character in maternal-foetal communication.
In summary, the findings in this study demonstrate a unique immune-regulatory characteristic of placental enEVTs to educate maternal CD4 + T-cell differentiation into Tregs. This is an important cellular mechanism to develop an immune-tolerant environment along the placental-maternal circulation pathway. The study also provides new insights into revealing the mechanisms of immune imbalance in the development of pregnancy complications such as RSA.

ACK N OWLED G EM ENTS
We appreciate the technical support from Ms Shiwen Li, Xili Zhu,

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data of this study are available from the corresponding author upon reasonable request.