Decidual tissues were collected from women with normal pregnancy (undergoing elective terminations, n=85) and spontaneous abortion (n=31) at 6-10 weeks of gestation. All subjects were confirmed by ultrasound. We excluded the spontaneous abortion with infection, chromosomal abnormality, anatomic defects, glucose metabolism disorders, thyroid dysfunction, or potential autoimmune diseases (where the anti-phospholipid antibodies or antinuclear antibodies were positive). The demographic and clinical characteristics of women with normal pregnancy and spontaneous abortion are summarized in Supplemental Table 1 . All decidual samples were immediately kept in ice-cold sterile DMEM/F12 medium (HyClone) and transported to the laboratory for further processing within 30 min after the operation.

Female CBA/J mice (RRID : IMSR_JAX:000656), male DBA/2 mice (RRID : IMSR_JAX:000671), male BALB/cJ mice (RRID : IMSR_JAX:000651) and wildtype C57BL/6NJ mice (RRID : IMSR_JAX:005304) were purchased from Jackson Laboratory. After adaptive feeding, female CBA/J mice were mated to male BALB/c or DBA/2 mice to establish the models of normal pregnancy (CBA/J × BALB/c, n=5) or spontaneous abortion (CBA/J × DBA/2, n=6). The fertilized eggs of C57BL/6 mice were microinjected with Cas 9/guide RNA complexes targeting at exons 2, 3, and 4 of the murine Pgf, incubated in a 5% CO₂ incubator at 37°C until the two‐cell stage, and then transferred to foster mothers. Then the F1 heterozygotes were generated by the backcrossing of genome‐modified F0 generation mice with wild‐type mice, and confirmed by qRT-PCR. To avoid the confounding effect of embryonic genotype, the heterozygotes were mated with wild-type mice to generate pregnant mice in the experimental group (Pgf^{- /+} females × wild-type males, n=24) and control group (wild-type females × Pgf^{- /+} males, n=26).

As previously described, human DICs and DSCs were isolated from decidual tissues, and murine uterine immune cells were isolated from uterine tissues at embryonic day 8.5 (24). Human dMφs were then purified from DICs using Anti-CD14 MicroBead Kit (Miltenyi, Cat #130-0490601). The purity of CD45⁺ DICs (above 98%), Vimentin⁺CD45^- DSCs (above 98%), CD45⁺CD14⁺ dMφs (above 95%), and CD45⁺ murine uterine immune cells (above 98%) were validated by flow cytometry (FCM) analysis. The primary isolated cells were cultured in RPMI 1640 (HyClone) or DMEM/F12 (HyClone) medium containing 10% fetal bovine serum (FBS, Gibco, Australia) in 5% CO₂ at 37°C.

The human epithelial cell line 293-T was purchased from American Type Culture Collection (ATCC, CRL-3216) and cultured in DMEM (HyClone) medium containing 10% FBS (Gibco, Australia) in 5% CO₂ at 37°C.

Human decidual tissues were labeled with rabbit anti-PlGF (Abcam) for immunohistochemical staining as previously described (24). Specifically, the formalin-fixed decidual tissues were dehydrated in ethanol, embedded with paraffin, and cut into slices. After the deparaffinization and rehydration, the sections were pretreated with sodium citrate buffer (0.01M, pH=6.0) for antigen retrieval, followed by blocking the endogenous peroxidase activity with 3% H₂O₂ and 5% bovine serum albumin. Then, the sections were incubated with rabbit anti-PlGF antibodies (1:100, Abcam), or rabbit IgG overnight at 4°C. After washing with PBS, the sections were incubated with secondary antibody for 30min at room temperature, labeled with 3,3’-diaminobenzidine, and counterstained with hematoxylin.

Total RNA was extracted using the TRIzol reagent (Invitrogen) and then reverse-transcribed into cDNA using the PrimeScript RT Master Mix (Takara). The qRT-PCR analysis was performed using the TB Green Premix EX Taq II (Takara) and the CFX96 Real-Time system (Bio-Rad). The primer sequences are shown in Supplemental Table 2 . Each sample was analyzed in triplicate. Relative mRNA levels were normalized to GAPDH.

After 48-hour culture of freshly isolated DSCs, the concentration of PlGF was measured by Human PlGF Quantikine ELISA kit (R&D Systems) according to the manufacturer’s instructions.

For FCM analyses, cells were incubated with monoclonal antibodies ( Supplemental Table 3 ) according to the manufacturers’ instructions. Intracellular and intranuclear staining was performed after the fixation and permeabilization with Fixation/Permeabilization Solution Kit (BD Pharmingen) and Transcription Factor Buffer Set (BD Pharmingen) respectively. FCM was performed using Beckman-Coulter CyAN ADP Analyzer (Beckman-Coulter) and analyzed by FlowJo software (version 10.0, TreeStar).

With the help of the JAS-PAR database, we predicted the binding site of the CEBPB promoter to STAT3 and introduced mutations in the predicted site. 293-T cells were seeded in 24-well plates (2×10⁵ cells per well) and cotransfected with STAT3-overexpression or vector plasmids, wild-type or mutant CEBPB promoter-reporter plasmids, and TK promoter-Renilla plasmids (GeneChem Co.) using Lipofectamine 2000 (Invitrogen). After 48 hours of transfection, dual-luciferase reporter assays were performed with the dual-luciferase reporter assay kit (Promega) according to the manufacturer’s instructions. The primer sequences are shown in Supplemental Table 4 .

Continuous variables are shown as mean ± SEM. The statistical comparison between two groups was performed using t-test when they are normally distributed or Wilcoxon test if not. The correlation analysis of two continuous variables was performed by the Spearman correlation analysis. All analyses were conducted with IBM SPSS statistics 25.0. A two-tailed P-value < 0.05 was considered statistically significant.

All procedures of this study were approved by All participants provided informed consent as part of the protocols approved by the Research Ethics Committee of the Second Affiliated Hospital, Chongqing Medical University (Chongqing, China, #2020-150). And all animal procedures were approved by the Animal Care and Use Committee of Chongqing Medical University.

To investigate the possible role of DSC-derived PlGF in the maintenance of early pregnancy, we firstly analyzed the expression of PGF in DSCs from normal pregnancy and spontaneous abortion during early pregnancy. In comparison with normal pregnancy, DSCs from spontaneous abortion showed a weaker staining of PlGF in the cytoplasm ( Figure 1A ). Furthermore, a markedly lower PlGF secretion in DSCs from spontaneous abortion was demonstrated by ELISA ( Figures 1B ). These results suggest that the decreased level of DSC-derived PlGF may be related to spontaneous abortion.

FLT-1 is the only known signaling receptor for PlGF in humans and expresses on macrophages (18). We then investigated the expression of FLT-1 on dMφs in the first trimester. Interestingly, with the help of FCM, FLT-1⁺ macrophages were found in the early decidua and a prominent reduction of FLT-1⁺ dMφs was observed in patients with spontaneous abortion when compared with normal pregnancy (16.65 ± 1.06% vs. 44.83 ± 1.47%) ( Figure 1C ). The mRNA level of FLT1 in dMφs from patients with spontaneous abortion was also lower than that of normal pregnancy ( Figure 1D ). Consistent with this, CBA/J×DBA/2 abortion-prone mice displayed fewer FLT-1⁺ uterine macrophages than did CBA/J×BALB/c control pregnant mice (25.63 ± 2.39% vs. 42.10 ± 1.63%) ( Figure 1E ). These findings indicate that the reduction of PlGF/FLT-1 expression in decidua may result in early spontaneous abortion.

We further characterized the profiles of macrophages in the early decidua from normal pregnancy and spontaneous abortion. A higher M1/M2 ratio of dMφs was demonstrated in spontaneous abortion, which manifested a higher expression of M1-associated markers (CD86, CD80, and IL-1β) and lower expression of M2-associated markers (CD209, CD206, IL-10, and TGF-β1 ( Figure 2A ). As previously described, a significant reduction of FLT-1⁺ dMφs was observed in early spontaneous abortion. Interestingly, compared to the FLT-1^- subtype, FLT-1⁺ dMφs in the early decidua exhibited an M2-like profile, with a higher expression of M2-associated markers (CD209 and IL-10) and a lower expression of M1-associated markers (CD86) ( Figure 2B ). Thus, we hypothesized that the reduction in FLT-1⁺ dMφs might be associated with the imbalance of M1/M2 in spontaneous abortion. To confirm this, we assessed the expression levels of M1- and M2-associated markers in primary dMφs treated with rhPlGF for 48 hours. As shown in Figure 2C , the rhPlGF-polarized (100 ng/mL) dMφs exhibited an M2-like profile, with an increased expression of CD209 and IL-10, as well as a decreased expression of CD86. Also, the enhancing effect on CD209 expression was found in primary dMφs treated with rhPlGF at the concentration of 50 ng/mL, and the suppressing effect on CD86 expression was found at the concentrations of 50 ng/mL and 200 ng/mL ( Figure 2C ). However, the expression of other M1- and M2-associated markers (CD206, CD80, TGF-β1, and IL-1β) remained unaltered after rhPlGF treatment (data not shown). These data suggest that the decreased PlGF/FLT-1 expression may participate in early spontaneous abortion by inducing the M1-like deviation of dMφs.

Previous studies revealed that the stimulation of FLT-1 by PlGF leads to the activation of downstream signaling molecules, including STAT3, PI3 kinase (PI3K), and p38 MAP kinase (p38 MAPK) (25, 26). To identify the downstream pathway through which PlGF drives the M2-like polarization of macrophages in the early decidua, we analyzed the profiles of rhPlGF-polarized dMφs following the pretreatment with selective inhibitors of STAT3 (AG490), PI3K (LY294002) or p38 MAPK (PD169316). Notably, the expression changes of CD209, IL-10, and CD86 caused by rhPlGF treatment (100 ng/mL) were only almost abrogated by AG490, a STAT3 inhibitor ( Figure 3A ). As is known, STAT3 forms homodimers by tyrosine phosphorylation at position 705, which is essential for the transcriptional function of STAT3 (27). Moreover, the transcriptional activity of STAT3 can also be enhanced by serine phosphorylation at position 727 (28). We further assessed the phosphorylation status of STAT3 at tyrosine 705 (Tyr 705) and serine 727 (Ser727) in primary dMφs treated with rhPlGF. Compared with the control, the rhPlGF-polarized (100 ng/mL) dMφs showed a significantly higher tyrosine phosphorylation level of STAT3 at position 705 ( Figure 3B ), while a similar serine phosphorylation level at position 727 (data not shown). In addition, the rhPlGF-promoted STAT3 phosphorylation was profoundly inhibited by AG490 ( Figure 3B ). These results suggested that PlGF may drive the M2-like polarization of dMφs by up-regulating the phosphorylation level of STAT3 at Tyr 705 and enhancing the transcriptional activity of STAT3.

To investigate the mechanism underlying PlGF-mediated dMφ polarization, we explored whether rhPlGF regulated the expression of key transcription factors involved in the M2-like polarization of macrophages. The mRNA level of CEBPB remarkably increased in rhPlGF-treated (100 ng/mL) dMφs compared to the control ( Figure 4A ). However, the transcription level of other transcription factors remained unaltered, including PPARGC1A, PPARGC1B, IRF4, IRF5, STAT1, and STAT6 ( Figure 4B ). Then we observed that the rhPlGF treatment (100 ng/mL) also elevated the C/EBPβ level in dMφs ( Figure 4C ). Furthermore, a positive correlation was found between the expression level of C/EBPβ and the phosphorylation level of STAT3 at Tyr 705 in rhPlGF-treated (100 ng/mL) dMφs ( Figure 4D ; Table 1 ). It was noteworthy that the promoting effect of rhPlGF on the expression of CEBPB at both mRNA and protein levels was abrogated by AG490 pretreatment ( Figures 4A, C ). Finally, the dual-luciferase reporter assay revealed an increased luciferase activity with the co-transfection of STAT3 overexpression and wild-type CEBPB promoter-reporter plasmids, which was completely blocked by the mutations of the CEBPB promoter in its predicted binding site to STAT3 ( Figure 4E ). These data indicated that rhPlGF promotes the M2-like polarization of dMφs via the STAT3-C/EBPβ signaling pathway.

To validate the role of PlGF in the M2-like polarization of dMφs and the maintenance of early pregnancy in vivo, we constructed Pgf knockout mice. Unfortunately, we didn’t obtain Pgf^{- /-} mice by mating heterozygotes, which was proved by the genotype detection. The genotypes of wild type (7/17), heterozygote (8/17) and homozygote (2/17) were detected at embryonic day 3.5, while only wild type (6/19) and heterozygote (13/19) at embryonic day 9.5. These results suggested the homozygous Pgf defect conferred embryo lethality. Compared to the control, the Pgf knockout pregnant mice showed a prominently higher embryo resorption rate at embryonic day 13.5 ( Figure 5A ), as well as a lower IL-10 production ( Figure 5B ), a lower STAT3 phosphorylation at Tyr 705 ( Figure 5C ), and a lower C/EBPβ expression ( Figure 5D ) in uterine macrophages at embryonic day 8.5. These data proved that the maternal PlGF deficiency may lead to the dysfunction of murine uterine macrophages and increase the risk of fetal loss. These findings further supported that PlGF promotes the M2-like polarization of dMφs via the STAT3-C/EBPβ signaling pathway during early pregnancy.