The sections of term placentas of normal pregnancy and PE patients were subsequently de-waxed, rehydrated, the antigen was retrieved by using a 1× Universal HIER antigen retrieval reagent (ab208572, Abcam, MA 02453, United States) and incubated with 3% H₂O₂. Sections were then blocked with 10% goat serum (ab7481, Abcam) for 1 h at room temperature. The sections were incubated in chemerin (1: 100, ab72965, Abcam) overnight in the cold room. Chemerin staining was visualized using a DAB substrate kit for peroxidase from Vector Laboratories and counterstained and mounted as described above. Digital photographs at 4× and 10× were taken, and Image J software was used to quantify the H-score.

This study was approved by the Institutional Review Board (IRB) of The University of Hong Kong/Hospital Authority Hong Kong West Cluster (IRB No: UW 17-057). Human dNK cells were isolated as described (Male et al., 2012; Lee et al., 2018) from late first-trimester decidual tissues collected from patients who had undergone surgical termination of pregnancy due to psychosocial reasons with written consent. The tissues were minced and digested with collagenase (300 U/ml) and DNase I (50 μg/ml), and the cells were passed through 100 and 40 μm filters followed by Ficoll-Paque (17,144,002, GE Healthcare Illinois, United States) density gradient centrifugation. The cells were then cultured on a plastic plate overnight. The non-adherent cells were collected, and dNK cells were further enriched using CD56 microbeads (Miltenyi Biotec, Germany). The isolated dNK were cultured with RPMI medium supplemented with 10% FBS and 1% penicillin-streptomycin under standard culture conditions in a 37°C and 5% CO₂ incubator.

The CyQUANT^® cell proliferation assay (CyQUANTTM NF Cell Proliferation assay kit, C35006, Invitrogen) was employed as a measurement of cell proliferation according to the manufacturer’s instructions. Primary dNK cells (5 × 10³ cells/well) were seeded in 96 well plate and cultured for 24 h with 0.5 µM chemerin (SRP6002, Sigma, St. Louis, UnitedStates), or 0.5 µM chemerin +1 µM 2-(alpha-Naphthoyl)ethyl trimethylammonium iodide (α-Neta, sc-221190A, Santa Cruz Biotechnology, United States) in RPMI1640 incomplete medium with 500 U/ml interleukin (IL)-2. The fluorescence intensity was measured using a fluorescence microplate reader (Infinite F Nano⁺) with an excitation wavelength at 485 nm and an emission wavelength at 530 nm. The concentrations of IL-8, IL-10 and TNF-α in the spent medium of dNK cells were determined using commercially available enzyme-linked immunosorbent assay (ELISA) kits (KHC3012, BMS215-2 and KHC0081, Abgent, San Diego, CA, United States). The samples were analyzed in quadruplicate.

JEG-3 cells (HTB-36, ATCC) were cultured in DMEM/Ham’s F-12 medium (Sigma, St. Louis, MO) containing 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin, at 37°C, in an atmosphere of 5% CO₂. The trophoblast invasion was measured by invasion assay (BD BioCoat™ Matrigel™ Invasion Assay, BD Biosciences). In brief, JEG3 cells (1 × 10⁴ cells/well) were placed in the upper invasion chamber, with the spent medium from dNK cells (1 × 10⁵ cells/well) treated with 0.5 µM chemerin, or 0.5 µM chemerin + 1 µM α-Neta treatment for 24 h were added into the upper chamber. FBS (10%) was added as a chemoattractive agent in the lower chamber. The invaded cells on the lower membrane of the chamber were stained with crystal violet, and images were acquired under light microscopy. The cells were then dissolved by 10% acetic acid, and the absorbance was measured at 595 nm.

The BeWo cell line has been widely used as an in vitro model for studying trophoblast intercellular fusion and differentiation (Heaton et al., 2008). BeWo cells (CCL-98, ATCC) were cultured in DMEM/Ham’s F-12 medium (Sigma, St. Louis, MO) containing 10% FBS and 1% penicillin-streptomycin, at 37°C, in an atmosphere of 5% CO₂. BeWo cells (1 × 10⁶ cells/well) were plated in 6 well multi-dishes and maintained in culture until the cells were 70–80% confluent. Differentiation was induced by the addition of 20 µM forskolin (F6886, St. Louis, MO, as positive control) with or without 0.5 µM chemerin, or 0.5 µM chemerin +1 µM α-Neta, in a serum-free medium for 48 h. All experiments involving the cell treatment were accompanied in parallel by vehicle controls. The quantification of BeWo cell fusion under different stimulatory conditions was performed by calculating the fusion index in 20 randomly selected microscopic fields of each condition. The fusion index = number of nuclei in syncytia/total number of nuclei x 100%. The cells were harvested to measure the expression of relevant differentiation markers by qPCR.

Total RNA was extracted from placentas or BeWo cells after 48 h of treatment using Illustra RNAspin Mini RNA isolation kit (25-0500-71, GE Healthcare, United States). RT-PCR was performed following the protocol for Taqman Gold RT-PCR (Manufacturer). For cDNA amplifications, highly specific forward and reverse primers were used with initial heating at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. The TaqMan™ Gene Expression Assay primers were purchased from Life Technologies, including mouse Interleukin (IL)-6 (Mm00446191_m1); IL-8 (Mm04208136_m1); IL-10 (Mm01288386_m1); Tumor necrosis factor-alpha (TNF-α) (Mm00443258_m1); Matrix metalloproteinase-2 (MMP)2 (Mm01253624_m1); MMP9 (Mm00600157_g1); Human Glial Cells Missing-1 (GCM1) (Hs00961601_m1); syncytin-1 (Hs00835189_CE); Caudal-type homeobox gene 2 (CDX2) (Hs01078080_m1); Human leukocyte antigen G (HLA-G) (Hs00365950_g1); and CK7 (Hs00559840_m1).

C57BL/6 mice (8 weeks) were obtained from the Guangdong Medical Laboratory Center (Guangdong, China). Cmklr1 ^−/− mice were obtained from The Jackson Laboratory (Bar Harbor, ME United States). Estrous Cmklr1-deficient females were paired with Cmklr1 ^−/− males to obtain timed pregnancy, and copulation plug detection the next morning was designated as gestation day (GD) 1. The same procedure was performed between wildtype mice. Animals were housed in a controlled environment where constant temperature, humidity, and a 12-h light-dark cycle with free access to chow diet and water were given. All animal usage has complied with the procedures approved by the Committee on the Use of Live Animals for Teaching and Research, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences (Permit Number: SIAT-IRB-120223-A0009).

Undissected GD 12 mice enveloped within uterine tissue were collected. Tissue samples were fixed in fixative solution (Pyrocitric acid: formaldehyde: glacial acetic acid 15: 5: 1) and processed for embedding in paraffin. Paraffin-embedded tissues were sectioned into 5 µm for hematoxylin and eosin staining. Digital photographs of three central sections of each placenta were taken. Image J software was used to measure the relevant stained or unstained regions of each section. For each placental section, the area of the whole placenta, labyrinth, junctional, and decidua zone was measured.

The sections of GD12 placenta from both wildtype and Cmklr1-deficient pregnant mice mounted on glass slides were subsequently de-waxed, rehydrated, and incubated with 3% H₂O₂. It was then washed three times with PBS, and the antigen was retrieved by using a 1× Universal HIER antigen retrieval reagent (ab208572, Abcam, MA 02453, United States). Sections were then blocked with 10% goat serum (ab7481, Abcam) for 1 h at room temperature. The sections were incubated in isolectin B4 (BSI-B4)(1: 100 in PBS, L-5391, Sigma-Aldrich, St Louis, United States) for fetal blood vessels and biotinylated-Dolichos biflorus (DBA) lectin (1: 1,000 in 1% BSA/PBS, L6533, Sigma Aldrich, Saint Louis, MO, United States) for dNK. The sections were incubated for 1 h and then washed three times in PBS. The lectin binding was detected by 30 min incubation with streptavidin-peroxidase (Extraditing-peroxidase; E2886, Sigma Aldrich, Saint Louis, MO, United States) and peroxidase-labeling kit (Vector Laboratories, city, country). The slides were then counterstained with hematoxylin, and then mounted followed by dehydration and clearing through a graded series of ethanol and xylene washes.

For immunostaining, antigen retrieval was performed in the deparaffinized sections as described above. The sections were permeabilized by eBioscience™ Permeabilization Buffer (00-8333-56, Thermo Fisher Scientific, Massachusetts, United States), followed by blocking in 10% goat serum (ab7481, Abcam) for 1 h. The sections were incubated with anti-monocarboxylate transporter 4 (MCT4, 1: 100, AB3314P, Merck Millipore) or anti-trophoblast specific protein alpha (TPBPA, 1: 100, ab104401, Abcam) antibodies overnight. After incubation, sections were washed three times in PBS. Primary antibodies were detected with appropriate fluorescence-conjugated secondary antibodies (1: 500 dilutions in PBS) for 1-h incubation. After incubation, sections were washed by PBS three times for 15 min. Nuclei were then counterstained with 4’,6-Diamidino-2-phenylindole dihydrochloride (DAPI, 1:500 dilution in PBS, D9542, MERCK, United States) for 2 min followed by washing with PBS five times. The sections were then mounted with a fluorescence mounting medium (S3023, DAKO, United States) and digital images were captured under a fluorescent microscope.

The sections were incubated with, anti-CK8 antibody (ab53280, Abcam). CK8 staining was visualized using a DAB substrate kit for peroxidase from Vector Laboratories and counterstained and mounted as described above. Digital photographs at 4× and 10× were taken, and Image J software was used to quantify the ratios of the invaded vessel to the amount vessels in wildtype and Cmklr1 deficient decidual zone.

The blood from wildtype and Cmklr1^−/− mice in GD 12 were collected. Serum was collected and frozen at −20°C in aliquots. According to the manufacturer’s protocol, serum IL-15 levels were measured using commercially available and sandwich enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems, CA, United States). The samples were analyzed in triplicate.

All values were expressed as mean ± SD. One-way ANOVA on Rank test was used to test the statistical differences between groups. Parametric Student’s t-test or non-parametric Mann Whitney U test was used where appropriate as the post-test. GraphPad Prism 9.0 (GraphPad Software, La Jolla, CA, United States) was used for all statistical analyses. A p-value of less than 0.05 was considered significant.

Chemerin immunoreactivities were positively stained in the trophoblast layer of human placental tissues. Compared with normal pregnancy, the expression of chemerin was reduced in the term pregnancy placental villi from women who presented with PE (Figure 1A). The biological effects of chemerin are mediated by CMKLR1. By using flow cytometry and immunohistochemistry, we demonstrated CMKLR1 expressions in CD16⁻CD56⁺ dNK (Figure 1B) cells and chorionic villi (Figure 1C) of first-trimester pregnancies respectively. To determine the role of CMKLR1 in regulating dNK cell proliferation, dNK cells were treated with chemerin with or without α-Neta, a specific antagonist of CMKLR1 for 24 h. Our results showed that while 0.5 µM chemerin suppressed the proliferation of the dNK cells, the inclusion of 1 µM α-Neta abolished the suppressive effect of chemerin (Figure 1D). The results indicated that the interaction of chemerin-CMKLR1 inhibited dNK cell proliferation.

The dNK cells are in close proximity to EVTs and regulate EVT functions by secretory factors. For example, dNK cell-derived IL-6, IL-8, hepatocyte growth factor, granulocyte-macrophage colony-stimulating factor (GM-CSF), and interferon-inducible protein-10 have been demonstrated to regulate human EVT motility and/or invasion (Hanna et al., 2006; Wallace et al., 2012). Therefore, we investigated the possible roles of chemerin on IL-8, IL-10 and TNF-α secretions by dNK cells. Our data revealed that IL-8 (Figure 1E) production was downregulated, while IL-10 (Figure 1F) and TNF-α (Figure 1G) were upregulated in dNK cells after chemerin treatment. The inclusion of 1 µM α-Neta abolished the effect of chemerin, indicating the involvement of CMKLR1.

To further confirm the implication of chemerin-CMKLR1 interaction in the regulatory activities of dNK cells on EVT functions, the spent medium of dNK cells after chemerin treatment was collected and applied to the EVTs. The dNK cells without chemerin treatment or with chemerin treatment in the presence of α-Neta were served as control. Our results showed that the spent medium of chemerin-treated dNK cells significantly decreased EVT invasion when compared to the controls, and such effect was abolished by α-Neta (Figure 1H).

Fusion of the cytotrophoblasts forms the multinucleated syncytiotrophoblast responsible for hormone production and maintenance of homeostasis of pregnancy. By using BeWo cells, an established model of forskolin induced syncytialization of cytotrophoblasts (Supplementary Figure S1), we demonstrated that chemerin-CMKLR1 interaction reduced E-cadherin expression, and thereby promoted syncytialization (Figures 2A,B). The inclusion of 1 µM α-Neta abolished the stimulatory effect of chemerin on syncytialization, indicating the involvement of CMKLR1 in cytotrophoblast syncytialization. The observations were supported by the upregulation of GCM1 (a trophoblastic differentiation marker) and Synctin-1 (a syncytiotrophoblast marker), and downregulation of CDX2 (a trophectoderm marker) in BeWo cells after chemerin treatment.

To assess the role of Cmklr1 in placental development in vivo, a Cmklr1-deficient mouse model was generated (Supplementary Figure S2A). Successful ablation of Cmklr1 was confirmed by PCR analysis (Supplementary Figure S2B). Cmklr1 knockout had no effect on chemerin expression (Supplementary Figure S2C). The Cmklr1 deficient mice exhibited larger litter size (Figure 3A) and more implantation sites (Figures 3B,C) but reduced fetal weight at birth (Figure 3D) when compared to the wildtype mice.

Cellular morphology of the wildtype and the Cmklr1-deficient placentas were compared (Figure 4A) which showed a significant reduction of the junctional zone thickness as indicated by TPBPA staining, and thus a proportionally increased labyrinth zone in the Cmklr1-knockout placentas. Significant enlargement of the luminal diameter of the spiral artery was also observed in the Cmklr1 ^−/− mice (Figure 4B) which was accompanied by accumulation of the CK8 positive trophoblast (Figure 4C). In contrast, there was no difference in size of the decidua basalis between the Cmklr1 ^−/− and the wildtype pregnant mice (Figure 4A).

To determine syncytialization in the Cmklr1-knockout placenta, we stained for MCT4, a monocarboxylate transporter specifically expressed in the syncytiotrophoblast II layer. The thickness of the syncytiotrophoblast II layer and the MCT4 staining were significantly reduced in the Cmklr1-deficient labyrinth. As the syncytiotrophoblast II layer enclosed fetal blood vessels in the labyrinth, the reduced MCT4 staining suggested that the fetal blood vessels were not properly developed in the region (Figure 4D), consistent with the diminished BSI-B4 staining that outlines the matrix surrounding the fetal vessels in the Cmklr1-deficient placenta when compared to the wildtype (Figure 4E).

Successful placental development and spiral arterial remodeling require dNK cells. The Cmklr1 ^−/− mice had a significant increase in the number of DBA-lectin positive dNK cells (Figure 5A) in the decidual zone when compared to the wildtype. IL-15 is involved in activating the proliferation of the CD56^Bright NK cells (Felices et al., 2018). Therefore, we analyzed the serum IL-15 level of the Cmklr1 ^−/− pregnant mice at GD 12. We detected a 5-fold increment in serum IL-15 level in the Cmklr1 ^−/− mice when compared to the wildtype (Figure 5B). We also investigated the downstream factors responsible for the altered placental development in the Cmklr1-knockout placenta. These factors, including IL-6 (Champion et al., 2012), IL-8 (Jovanović et al., 2010), IL-10 (Pang et al., 2008), MMP2, MMP9 (Staun-Ram et al., 2004), and TNF-α have been associated with the regulatory activities of dNK cells on trophoblast functions. Our data revealed that IL-8 (Figure 5C), MMP2 (Figure 5D), and MMP9 (Figure 5E) were upregulated while IL-10 (Figure 5F) and TNF-α (Figure 5G) were downregulated. The expression of IL-6 was not significantly affected (Figure 5H).