BMP6 as a therapeutic target for preeclampsia: enhancing trophoblast invasion and vascular mimicry

Abstract

Shallow trophoblast invasion and improper maternal spiral artery remodeling are the primary mechanisms underlying the development of preeclampsia (PE). Bone morphogenetic protein 6 (BMP6) is a proinvasive and proangiogenic factor in vitro; however, its regulatory mechanisms in trophoblast behavior and its role in PE development remain unclear. In this study, primary human trophoblasts and the HTR8/SVneo cell line were utilized asin vitrostudy models. Bulk RNA sequencing (RNA-seq) and single-cell RNA sequencing (scRNA-seq) data were analyzed to explore the expression patterns of BMP6-regulated genes. We found that BMP6 treatment significantly upregulated inhibitor of DNA-binding 1 (ID1) in human trophoblasts. ID1 depletion abolished both basal and BMP6-induced trophoblast invasion and vascular mimicry. Mechanistically, ID1-mediated upregulation of serpin family E member 2 (SERPINE2) and placental growth factor (PlGF) was essential for BMP6-induced trophoblast invasion. In third-trimester placentas, BMP6 mRNA and protein levels were significantly elevated in PE compared with controls. In the adenovirus-expressing fms-like tyrosine kinase-1 (Ad Flt1)-induced rat model of PE, both circulating BMP6 and placental Bmp6 expression were increased in PE rats in late pregnancy. Significantly, BMP6 supplementation during early pregnancy (gestational days 10–13) alleviated maternal hypertension and fetal growth restriction in the PE model. These findings suggest BMP6 promotes trophoblast invasion through ID1-mediated upregulation of SERPINE2 and PlGF. The late-gestation upregulation of BMP6 may represent a compensatory response to shallow trophoblast invasion in PE. Early BMP6 supplementation mitigates PE-related phenotypes in a rat model, highlighting BMP6 as a potential therapeutic target for the prevention and management of PE.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-025-06040-w.

Introduction

Preeclampsia (PE) is a pregnancy-specific disease that affects 2–8% of pregnancies worldwide [1, 2] and accounts for an estimated 70,000 maternal deaths and 500,000 neonatal deaths annually [3–5]. Its hallmark features are the de novo onset of hypertension (≥ 140/90 mm Hg) after 20 weeks of gestation, accompanied by proteinuria (≥ 300 mg/L per 24 h) or other signs of maternal organ dysfunctions [6, 7]. PE contributes to substantial maternal morbidity and adverse perinatal outcomes, including prematurity and growth restriction [8]. It also increases long-term cardiovascular and cerebrovascular risks for both mothers and their offspring, impacting long-term health [9–11]. Currently, delivery remains the only definitive treatment for PE, reinforcing the widely accepted theory that poor placentation is central to its pathogenesis [12, 13]. However, the precise underlying mechanisms and effective therapeutic targets of PE remain to be fully elucidated.

Impaired invasion of extravillous cytotrophoblasts (EVTs), which mediates spiral artery remodeling via interstitial and endovascular routes [14, 15], leads to placental hypoxia and disrupts the balance of trophoblast-derived angiogenic factors, notably increasing soluble fms-like tyrosine kinase-1 (sFlt-1) and decreasing placental growth factor (PlGF), contributing to PE pathogenesis [16]. Transforming growth factor-β (TGF-β) superfamily members have been implicated in regulating EVT differentiation and invasion [17, 18]. Bone morphogenetic proteins (BMPs), conserved members of the TGF-β superfamily, signal through type I and type II serine/threonine kinase transmembrane receptors [19]. BMPs primarily activate canonical SMAD1/5/9 pathways [18], although some BMPs can also signal via noncanonical SMAD2/3 pathways [20, 21]. Phosphorylated SMADs form complexes with SMAD4, translocate to the nucleus, and regulate target gene expression (Fig. 7) [18]. BMP2 has been reported to be upregulated in the third-trimester placenta of PE patients and serves as a compensatory factor for shallow trophoblast invasion. BMP6, a downstream target of BMP2, is elevated in the serum of third-trimester PE patients [22]. Although BMP6 is highly enriched in placental tissues, the molecular regulatory mechanisms by which BMP6 regulates trophoblast behavior and contributes to PE remain poorly understood.

Fig. 7.

BMP6 signaling enhancement compensates for shallow trophoblast invasion in PE. An examination of clinical samples from PE patients and their gestational age-matched controls revealed that BMP6 was upregulated in preeclamptic placentas. BMP6 enhances trophoblast invasion via ID1-mediated upregulation of SERPINE2 and PlGF. Moreover, BMP6 also intensifies trophoblast vascular mimicry through the ID1-mediated upregulation of PlGF. Both canonical (p-SMAD1/5/9) and noncanonical (p-SMAD2/3) pathways participate in BMP6-induced ID1 expression. Our findings indicate that the increase in BMP6 signaling during late gestation could serve as a compensatory response to shallow trophoblast invasion in PE, which in light of BMP6 and its downstream targets could serve as diagnostic markers and therapeutic target applications in the clinical management of PE. This schematic diagram was created with Biorender.com

Inhibitors of DNA-binding (ID) proteins, a family of highly conserved transcriptional regulators, are critical mediators of BMP signaling [23]. ID proteins lack a DNA-binding domain and instead dimerize with basic helix-loop-helix (bHLH) transcription factors to inhibit their binding to DNA, thereby modulating gene transcription [24]. Accumulating evidence suggests that ID proteins are targets for BMPs [25–28]. Among the four ID family members (ID1, ID2, ID3, and ID4), ID1, ID2, and ID3 are associated with enhanced cell invasion and angiogenesis, whereas ID4 functions as a tumor suppressor [23, 29]. Despite evidence linking ID proteins to BMP signaling, their roles in PE pathogenesis and the regulation of BMP6-induced trophoblast function remain poorly understood.

Serpin family E member 2 (SERPINE2), a serine protease inhibitor highly expressed at the maternal-fetal interface [30, 31], is predominantly localized in the trophoblast of placenta during early pregnancy [32]. Involved in tissue remodeling, SERPINE2 knockdown impairs the biological behaviors of trophoblasts, including migration and invasion [30, 32, 33]. According to a Mendelian randomization study, SERPINE2 has been identified as a pivotal protein in the development of PE [34]. However, whether SERPINE2 mediates BMP6-induced trophoblast invasion and vascular mimicry remains unknown.

In this study, we aimed to explore the regulatory mechanisms of BMP6 in trophoblast behavior and its potential as a therapeutic target for PE. Specifically, we investigated the role of ID1 as a critical mediator of BMP6-induced trophoblast invasion through the upregulation of SERPINE2 and PlGF. Using an adenovirus-expressing fms-like tyrosine kinase-1 (Ad Flt1)-induced rat model of PE and control rats injected with an adenovirus expressing the control IgG2a Fc fragment (Ad Fc), we further examined whether BMP6 supplementation could alleviate PE-related phenotypes. Our findings suggest that BMP6 and its downstream pathway represent promising therapeutic targets for the management of PE.

Results

BMP6 is expressed in the trophoblast of first-trimester villi and promotes trophoblast invasion and vascular mimicry

To investigate the potential role of BMP6 in trophoblast behavior and PE pathogenesis, we first conducted an immunohistochemistry analysis, and the results revealed that BMP6 was significantly expressed in first-trimester villi and decidua, as well as in the third-trimester placenta. Notably, BMP6 was expressed across all primary trophoblast cells, including villous cytotrophoblasts (VCTs), syncytiotrophoblasts (STBs), and EVTs in first-trimester villi (Fig. 1A), suggesting the regulatory role of BMP6 in trophoblast behavior. Although a prior study demonstrated the proinvasive effect of 50 ng/mL BMP6 [22], a comprehensive assessment of its dose-dependent effects is lacking. Therefore, we investigated a concentration range from 0 to 100 ng/mL. Our results confirmed that BMP6 enhances both cell invasion (Fig. 1B) and endothelial-like tube formation (Fig. 1C) in a dose-dependent manner, and 50 ng/mL was identified as the minimum effective concentration for enhancing trophoblast function. We subsequently used 50 ng/mL BMP6 to treat both HTR8/SVneo cells and human primary EVTs in all functional experiments.

Fig. 1.

BMP6 facilitates human trophoblast invasion and vascular mimicry and increases the expression of ID1, which is associated with PE. A, BMP6 is expressed at the maternal-fetal interface. Immunohistochemistry localization of BMP6 in first-trimester villi, first-trimester decidua, and third-trimester placenta. Scale bar, 100 μm. B-C, BMP6 facilitates human trophoblast invasion and vascular mimicry. B, HTR8/SVneo cells were treated with or without BMP6, followed by analysis of cell invasion. Scale bar, 100 μm. C, HTR8/SVneo cells were treated with or without BMP6, followed by analysis of endothelial-like tube formation. Scale bar, 100 μm. D-I, Bulk RNA-Seq analysis reveals that BMP6 treatment significantly upregulates ID1, ID2, and ID3 in HTR8/SVneo cells. D-F, Heatmap (D), volcano plots (E), and waterfall plots (F) obtained from RNA-Seq analysis of HTR8/SVneo cells with or without BMP6 treatment for 6 h. G-I, Heatmap (G), volcano plots (H), and waterfall plots (I) obtained from RNA-Seq analysis of HTR8/SVneo cells treated with or without BMP6 for 24 h. J, Purity of the primary human EVTs. Primary human EVTs were stained by CK7 (left panel) and HLA-G (right panel). Scale bar, 50 μm. K-P, single-cell analysis of the placenta shows ID1 is predominantly expressed in invasive EVTs. K and M, Analysis of previously published single-cell transcriptomes of human placentas collected during early pregnancy (6–12 gestational weeks, n = 5). L, N-P, Analysis of previously published single-cell transcriptomes of placentas collected during late pregnancy: three control placentas at 38 gestational weeks and three preeclamptic placentas at 34‒35 gestational weeks. K and L, UMAP visualization of all captured cell types in the placenta during early pregnancy (K) and late pregnancy (L), respectively. M and N, UMAP plot displaying the ID1 expression levels in all cell types in the early placenta (M) and the late placenta (N), respectively. O, UMAP visualization of all captured cell types in control women and PE patients. P, Violin plot displaying the expression level of ID1 in the EVT cell type. The quantitative results are expressed as the means ± SEMs of at least three independent experiments. One-way ANOVA was used for analyses in B and C, and Student’s t-test was used for comparisons between two groups in P. Groups without common letters are significantly different from each other (P < 0.05)BMP6, bone morphogenetic protein 6; GE, glandular epithelium; SC, stromal cell; CK7, cytokeratin-7; HLA-G, human leukocyte antigen G; UMAP, uniform manifold approximation and projection; EVT, extravillous cytotrophoblast; SCT, syncytiotrophoblast; VCT, villous cytotrophoblast; ID1, inhibitor of DNA-binding 1; PE, preeclampsia

BMP6 upregulates ID1, ID2, and ID3 expression in human trophoblast

To explore the molecular mechanism by which BMP6 promotes trophoblast invasion and vascular mimicry, we analyzed the bulk RNA sequencing (RNA-seq) data from HTR8/SVneo cells treated with or without BMP6 (50 ng/mL). Heatmaps, volcano plots, and waterfall plots revealed that BMP6 treatment upregulated key regulators involved in TGF-β signaling and epithelial-to-mesenchymal transition, such as SMAD9 and ZEB2. Notably, the ID family members (ID1, ID2, and ID3) presented the most pronounced differential expression after both 6 and 24 h of BMP6 treatment (Fig. 1D-I). Analysis of transcription factor activity revealed that multiple transcription factors from the ID and SMAD families were activated (Fig. S1A and B). The functional enrichment and gene ontology (GO) analyses revealed significant enrichment of terms related to the BMP signaling pathway, wound healing, and epithelial-to-mesenchymal transition (Fig. S1C and D). Pathway analysis showed that components of the TGF-β signaling pathway, the MAPK signaling pathway, and several cancer pathways were significantly enriched, indicating a proinvasive role for BMP6 (Fig. S1E and F). Gene set enrichment analysis (GSEA) further revealed that BMP6 primarily activated the TGF-β signaling pathway in HTR8/SVneo cells (Fig. S1G and H). RT‒qPCR analysis of HTR8/SVneo cells (Fig. S2A) and human primary EVTs (Fig. S2B) further validated that BMP6 upregulated ID1, ID2, and ID3 expression. The purity of primary EVT cultures was confirmed by immunofluorescence staining for the trophoblast marker CK7 and the EVT-specific marker HLA-G (Fig. 1J).

Single-cell RNA sequencing (scRNA-seq) analysis shows ID1 is predominantly expressed in invasive EVTs in the first-trimester placenta

To identify the cell-type-specific expression profiles of ID1, ID2, and ID3 in the placenta, we analyzed publicly available scRNA-seq datasets of placentas obtained from five women during early pregnancy [35] and six women (three controls and three with PE) during late pregnancy [36]. Cell types identified in these datasets are presented in Fig. 1K and L for the early pregnancy and late pregnancy placentas, respectively. ID1 (Fig. 1M and N) and ID3 (Fig. S2C and D) were predominantly expressed in invasive EVTs, SCTs, and VCTs, whereas ID2 exhibited a broader expression pattern across various cell types (Fig. S2E and F) during both early and late pregnancy. The placental cell type distributions in patients with PE and their controls during late pregnancy [36] are visualized in Fig. 1O. Interestingly, compared with control patients, PE patients presented elevated ID1 expression, specifically in EVTs, as visualized by violin plots (Fig. 1P). We further conducted RT‒qPCR analysis of human placental samples (10 PE vs. 10 control), the results confirmed that ID1 mRNA levels were significantly higher in the placentas of PE patients compared with their controls, whereas ID2 and ID3 mRNA levels were comparable between the two groups (Fig. S2G). These findings prompted us to focus on ID1, which has been identified as a mediator of BMP2-induced trophoblast invasion and vascular mimicry [26]. Immunohistochemistry assays revealed that ID1 was significantly expressed in VCTs, STBs, and EVTs in first-trimester villi, with the highest expression in VCTs (Fig. S2H). Because a specific subset of VCTs function as true progenitors capable of differentiating into both EVT and STB lineages [14], this expression pattern suggests that ID1 may regulate trophoblast differentiation and thereby modulate trophoblast behavior during early pregnancy.

ID1 mediates BMP6-promoted human trophoblast invasion and vascular mimicry

To validate the upregulation of ID1 by BMP6 treatment, RT‒qPCR and Western blot assays were employed. BMP6 treatment significantly increased ID1 mRNA and protein levels in HTR8/SVneo cells in a dose-dependent manner (Fig. 2A and B). Moreover, BMP6 treatment (50 ng/mL) significantly increased ID1 protein levels in both HTR8/SVneo cells and human primary EVTs in a time-dependent manner (Fig. 2C and D). To assess the functional role of ID1 in BMP6-promoted trophoblast invasion and vascular mimicry, we conducted transwell invasion assays and endothelial-like tube formation assays. The knockdown efficiency of siRNA targeting ID1 was confirmed by RT-qPCR in HTR8/SVneo cells and human primary EVTs, respectively (Fig. 2E and F). siRNA-mediated ID1 knockdown abolished both basal and BMP6-induced human trophoblast invasion in HTR8/SVneo cells and primary human EVTs (Fig. 2G and H), as well as basal and BMP6-induced vascular mimicry in HTR8/SVneo cells (Fig. 2I). By employing EDU, TUNEL, and flow cytometry-based apoptosis assays, we found there were no significant differences in the proliferation (Fig. S3A) or apoptosis (Fig. S3B and C) rates of HTR8/SVneo cells following BMP6 treatment or siRNA-mediated ID1 knockdown. These results suggest that ID1 mediates the explicit effects of BMP6 on trophoblast invasion and vascular mimicry, without altering cell viability or inducing apoptosis.

Fig. 2.

ID1 mediates BMP6-induced trophoblast invasion and vascular mimicry. A-D, BMP6 upregulated ID1 mRNA and protein levels in HTR8/SVneo cells. A and B, HTR8/SVneo cells were treated with different concentrations (0, 6.25, 12.5, 25, 50 or 100 ng/mL) of BMP6, and the ID1 mRNA levels after 6 h of treatment (A) and the ID1 protein levels after 24 h of treatment (B) were examined by RT‒qPCR and Western blot analysis, respectively. C-D, ID1 protein levels in HTR8/SVneo cells (C) and primary EVTs (D) after treatment with vehicle (Ctrl) or 50 ng/mL BMP6 for different durations. E‒I, ID1 mediates BMP6-promoted human trophoblast invasion and vascular mimicry. HTR8/SVneo cells or primary human EVTs were transfected for 48 h with 20 nM control non-targeting siRNA (si-Ctrl) or siRNA targeting ID1 (si-ID1) before treatment with or without 50 ng/mL BMP6. E and F, ID1 mRNA levels were examined by qPCR after BMP6 treatment for 6 h in HTR8/SVneo cells (E) and primary EVTs (F), with GAPDH used as the reference gene. G and H, Transwell assays were employed to examine the invasiveness of HTR8/SVneo cells (G) and primary EVTs (H) with or without BMP6 treatment for 36 h. Representative images from the invasion assay are displayed in the upper panel, while the summarized quantitative results of the invasion assay are shown in the lower panel. Scale bar, 100 μm. I, Endothelial-like tube formation assays were used to assess the acquisition of the endothelial-like phenotype of HTR8/SVneo cells with or without BMP6 treatment for 12 h. Representative images from the endothelial-like tube formation assay are displayed in the upper panel, while the summarized quantitative results of the endothelial-like tube formation assay are shown in the lower panel. Scale bar, 100 μm. The quantitative results are expressed as the means ± SEMs of at least three independent experiments. One-way ANOVA was used for analyses in A and B. Two-way ANOVA was used for grouped analyses in C-I. Groups without letters are significantly different from each other (P < 0.05). BMP6, bone morphogenetic protein 6; ID1, inhibitor of DNA-binding 1; Ctrl, control

Both canonical p-SMAD1/5/9 and noncanonical p-SMAD2/3 pathways mediate BMP6-induced ID1 protein upregulation in human trophoblast

BMPs can activate the canonical SMAD1/5/9-SMAD4 [18] and noncanonical SMAD2/3-SMAD4 pathways [20, 21]. To investigate the signaling pathways involved in BMP6-induced ID1 activation, we examined SMAD phosphorylation in trophoblasts. Western blot analysis revealed that BMP6 treatment significantly activated both p-SMAD1/5/9 and p-SMAD2/3 pathways in both HTR8/SVneo cells and human primary EVTs (Fig. S4A and B). Furthermore, siRNA-mediated knockdown of SMAD1/5/9, SMAD2/3, or SMAD4 attenuated BMP6-induced ID1 protein upregulation in HTR8/SVneo cells (Fig. S5A-C) and human primary EVTs (Fig. S5D-F). These results indicate that both the p-SMAD1/5/9 and p-SMAD2/3 pathways are involved in BMP6-induced upregulation of ID1 in human trophoblast.

ID1 mediates BMP6-induced SERPINE2 and PlGF upregulation in human trophoblast

SERPINE2, a key regulator of trophoblast invasion and a recognized hallmark of PE [33, 34], has been implicated in abnormal placental development. To further explore these established roles, we investigated the involvement of SERPINE2 in BMP6-induced trophoblast invasion. Bulk RNA-seq data from HTR8/SVneo cells were reanalyzed, revealing a molecular link between BMP6 signaling and SERPINE2 as well as PGF (Fig. 1D-I). To validate these findings, we performed RT-qPCR and Western blot assays in HTR8/SVneo cells and human primary EVTs after treatment with either vehicle or BMP6 at various dosages and durations. We observed that BMP6 treatment upregulated SERPINE2 (Fig. S6A-C) and PGF (Fig. S6D-F) mRNA levels in a both dose-dependent and time-dependent manner in HTR8/SVneo cells and human primary EVTs. The protein levels of SERPINE2 were also increased following BMP6 treatment, as confirmed by Western blot in both HTR8/SVneo cells (Fig. 3A and B) and human primary EVTs (Fig. 3C). PlGF, a secreted protein encoded by PGF, is predominantly expressed in the placenta [37]. To assess the regulatory effect of BMP6 on PlGF, we measured PlGF accumulation in conditioned medium to determine whether BMP6 enhances PlGF secretion. ELISA results showed PlGF accumulation in conditioned medium was significantly increased after BMP6 treatment in both HTR8/SVneo cells and human primary EVTs (Fig. 3D and E).

Fig. 3.

ID1 mediates BMP6-induced SERPINE2 and PlGF upregulation in human trophoblasts. A-C, BMP6 upregulates SERPINE2 protein levels in trophoblasts. A, HTR8/SVneo cells were treated with different concentrations (0, 6.25, 12.5, 25, 50, or 100 ng/mL) of BMP6, and the SERPINE2 protein levels after 24 h of treatment were examined by Western blot analysis. The upper panel shows a representative Western blot image, and the lower panel shows the summarized quantitative results. B, SERPINE2 protein levels in HTR8/SVneo cells after treatment with vehicle (Ctrl) or 50 ng/mL BMP6 for different durations (24, 48, and 72 h). The upper panel shows a representative Western blot image, and the lower panel shows the summarized quantitative results. C, SERPINE2 protein levels in human primary EVTs. The left panel shows a representative Western blot image, and the right panel shows the summarized quantitative results. D-E, BMP6 promotes PlGF accumulation in the conditioned medium of trophoblasts. D, HTR8/SVneo cells were treated with or without 50 ng/mL BMP6 for 24–48 h. PlGF accumulation in conditioned medium was measured using ELISA. E, PlGF accumulation in conditioned medium was assayed by ELISA 48 h after BMP6 treatment in primary EVTs. F-J, ID1 mediates BMP6-induced SERPINE2 and PlGF upregulation in human trophoblasts. HTR8/SVneo cells were transfected for 48 h with 20 nM control non-targeting siRNA (si-Ctrl) or siRNA targeting ID1 (si-ID1) before treatment with or without 50 ng/mL BMP6. F, ID1 mRNA levels were examined by qPCR 6 h after BMP6 (50 ng/mL) treatment in HTR8/SVneo cells, with GAPDH used as the reference gene. G and H, SERPINE2 and ID1 protein levels in HTR8/SVneo cells (G) and human primary EVTs (H) after transfection with siRNA targeting ID1, followed by treatment with or without BMP6 for 24 h, as assessed by Western blot. The left panel shows a representative Western blot image, and the right panel shows the summarized quantitative results. I, PGF mRNA levels were examined by qPCR 6 h after BMP6 treatment in HTR8/SVneo cells, with GAPDH used as the reference gene. J, PlGF accumulation in conditioned medium was assayed by ELISA 24 h after BMP6 treatment in HTR8/SVneo cells. K, SMAD4 mediates BMP6-induced upregulation of PlGF in human trophoblasts. HTR8/SVneo cells were transfected for 48 h with 20 nM control non-targeting siRNA (si-Ctrl) or siRNA targeting SMAD4 (si-SMAD4) before treatment with or without 50 ng/mL BMP6. PlGF accumulation in conditioned medium was assayed by ELISA 24 h after BMP6 treatment in HTR8/SVneo cells. The quantitative results are expressed as the means ± SEMs of at least three independent experiments. One-way ANOVA was used for analyses in A, and two-way ANOVA was used for comparisons in B-K. Groups without common letters are significantly different from each other (P < 0.05). BMP6, bone morphogenetic protein 6; SERPINE2, serpin family E member 2; EVT, extravillous cytotrophoblast; Ctrl, control; PlGF, placental growth factor; ID1, inhibitor of DNA-binding 1

To determine whether ID1 mediates BMP6-induced SERPINE2 and PlGF upregulation, HTR8/SVneo cells and human primary EVTs were transfected with siRNA targeting ID1, followed by treatment with or without BMP6. Knockdown of ID1 significantly attenuated BMP6-induced upregulation of SERPINE2 mRNA and protein levels (Fig. 3F-H). Moreover, siRNA-mediated ID1 knockdown suppressed BMP6-induced PGF mRNA upregulation (Fig. 3I) and PlGF accumulation in the conditioned medium of HTR8/SVneo cells (Fig. 3J). To further confirm the role of SMAD-dependent signaling in BMP6-upregulated SERPINE2 and PlGF expression, HTR8/SVneo cells and human primary EVTs were transfected with si-SMAD4, followed by treatment with vehicle or BMP6 (50 ng/mL). Western blot analysis revealed that knockdown of SMAD4 blocked BMP6-upregulated SERPINE2 expression in both HTR8/SVneo cells and human primary EVTs (Fig. S6G and H). Additionally, ELISA assays demonstrated that si-SMAD4 suppressed BMP6-induced PlGF accumulation in the conditioned medium from HTR8/SVneo cells (Fig. 3K). Collectively, these findings suggest that ID1 mediates BMP6-induced upregulation of SERPINE2 and PlGF.

SERPINE2 and PlGF mediate BMP6-induced human trophoblast invasion

Analysis of scRNA-seq datasets of placentas obtained from five women during early pregnancy [35] and six women (three control and three PE) during late pregnancy [36] indicated that both SERPINE2 (Fig. S7A and B) and PGF (Fig. S7C and D) are preferentially expressed in invasive EVTs compared with SCTs and VCTs. To determine whether SERPINE2 and PlGF are involved in BMP6-induced trophoblast invasion and vascular mimicry, we performed siRNA-mediated knockdown of SERPINE2 and PGF in HTR8/SVneo cells and human primary EVTs. The knockdown efficiencies of SERPINE2 and PGF in these cells are shown in Fig. 4A-D. Transwell invasion assays demonstrated that depletion of SERPINE2 (Fig. 4E and F) and PGF (Fig. 4G and H) abolished both basal and BMP6-induced HTR8/SVneo cell and human primary EVT invasion. Endothelial-like tube formation assays revealed that knockdown of SERPINE2 did not affect vascular mimicry (Fig. 4I), whereas knockdown of PGF abolished both basal and BMP6-induced vascular mimicry (Fig. 4J) in HTR8/SVneo cells. Furthermore, the results of the EdU (Fig. S7E) and TUNEL (Fig. S7F) assays did not reveal any significant effect of SERPINE2 or PGF knockdown on the rate of proliferation or apoptosis of HTR8/SVneo cells. Taken together, these results suggest that SERPINE2 and PGF mediate both basal and BMP6-promoted invasion of HTR8/SVneo cells. Furthermore, PGF is also involved in both basal and BMP6-induced vascular mimicry.

Fig. 4.

Both SERPINE2 and PlGF mediate BMP6-induced trophoblast invasion. A-J, HTR8/SVneo cells or primary EVTs were transfected for 48 h with 20 nM control nontargeting siRNA (si-Ctrl), 20 nM siRNA targeting SERPINE2 (si-SERPINE2) or PGF (si-PGF) before treatment with or without 50 ng/mL BMP6 for 24 h. A and B, The protein levels of SERPINE2 in HTR8/SVneo cells (A) and human primary EVTs (B) after 24 h of BMP6 treatment. The upper panel shows a representative Western blot image, and the lower panel shows the summarized quantitative results. C, PlGF accumulation in conditioned medium with or without BMP6 treatment for 24 h was assayed by ELISA in HTR8/SVneo cells. D, PGF mRNA levels in human primary EVTs with or without BMP6 treatment for 6 h were examined by RT‒qPCR, with GAPDH as the reference gene. E-H, Transwell assays were employed to examine the invasiveness of HTR8/SVneo cells (E and G) and primary EVTs (F and H) with or without BMP6 treatment for 36 h. I and J, Endothelial-like tube formation assays were used to assess vascular mimicry of HTR8/SVneo cells with or without BMP6 treatment for 12 h. Representative images from the endothelial-like tube formation assay are displayed in the above panel; the summarized quantitative results are displayed in the lower panel. Scale bar, 100 μm. The quantitative results are expressed as the means ± SEMs of at least three independent experiments. Two-way ANOVA was used for data comparison. Groups without common letters are significantly different from each other (P < 0.05). BMP6, bone morphogenetic protein 6; SERPINE2, serpin family E member 2; PGF, placental growth factor; Ctrl, control; EVT, extravillous cytotrophoblast

BMP6 is elevated in PE patients and PE model rats

A previous study reported that BMP2 was compensatorily upregulated in the placentas of PE patients during late pregnancy [22]. BMP6, a downstream target of BMP2, has also been suggested to be elevated in the serum of third-trimester PE patients [22]. To further explore the placental expression pattern of BMP6 in PE, we conducted RT‒qPCR analysis (10 Control vs. 10 PE) and Western blot analysis (4 Control vs. 4 PE) of placental samples. We found significantly increased BMP6 mRNA and protein levels in PE placentas compared with control placentas (Fig. 5A and B), providing further evidence that BMP6 is upregulated in PE during late pregnancy. Spearman correlation analysis was performed to examine placental BMP6 mRNA levels and SBP, showing that the placental levels of BMP6 were positively correlated with SBP (Fig. 5C). The baseline characteristics of the participants are summarized in Supplementary Table 1.

Fig. 5.

BMP6 is elevated in patients with PE and in PE model rats. A-C, BMP6 is elevated in patients with PE. A, RT‒qPCR analysis of BMP6 mRNA expression levels in the placentas of control women (n = 10) and PE patients (n = 10), with GAPDH as the reference gene. B, Western blot analysis of BMP6 protein expression levels in the placentas of control women (n = 4) and PE patients (n = 4). C, Spearman correlation analysis between the placental BMP6 mRNA levels and the value of log10 (SBP) of the corresponding patients. The gray area represents the 95% CI. Each dot represents one sample. D, The animal experimental protocol. E and F, BMP6 is elevated in the plasma of PE model rats. Rat plasma levels of BMP6 (E) and PlGF (F) in the Ad Fc + PBS group (n = 4) and Ad Flt1 + PBS group (n = 4). G-M, BMP6 is elevated in the placenta of PE model rats at G13. RNA-seq analysis of rat placentas at G13 in the Ad Fc + PBS group (n = 3) and Ad Flt1 + PBS group (n = 3). G, Heatmap depicting DEGs in the two groups. H, Dot plots of significantly enriched GO terms; the dot size represents the number of DEGs associated with a particular GO term. I, KEGG hierarchical network plot of pathways. J, GSEA-KEGG Ridge plot of pathways. K, GSEA plots of cytokine-cytokine receptor interaction pathway. L, Volcano plot of RNA-seq data showing DEGs between the Ad Fc + PBS group and the Ad Flt1 + PBS group. M, Circos graph displaying the coexpression networks of five genes in rat placenta samples. Each sector of the circle represents one gene, and its width indicates the total amount of co-occurrence that connects one gene to the other. The width of each link represents the total number of coexpressed genes among the linked genes. Student’s t-test was used for comparisons between two groups in A, B, E, and F. Groups without common letters are significantly different from each other (P < 0.05). SD, Sprague–Dawley; BMP6, bone morphogenetic protein 6; PlGF, placental growth factor; PBS, phosphate-buffered saline; Ad Flt1, adenovirus expressing fms-like tyrosine kinase-1; Ad Fc, adenovirus-expressing control IgG2a Fc fragment; FC, fold change; Serpine2, serpin family E member 2; Id1, inhibitor of DNA-binding 1

To explore whether BMP6 exhibits compensatory effects similar to those of BMP2, we established a PE rat model and corresponding control rats by injecting adenovirus encoding Flt1 (Ad Flt1) or Fc (Ad Fc) via the rat tail vein on gestational day 8. Rats were subsequently treated with PBS or recombinant BMP6 (10 µg/kg/day) from gestational days 10 to 13 (G10-13) as shown in Fig. 5D. We observed significantly increased plasma BMP6 levels in Ad Flt1-induced PE rats (Ad Flt1 + PBS group) compared with uninduced rats (Ad Fc + PBS group) during late pregnancy (G14 and G19) (Fig. 5E), aligning with the findings in humans. Interestingly, plasma PlGF levels were also elevated in the Ad Flt1 + PBS group compared with the Ad Fc + PBS group during middle pregnancy in G14 (Fig. 5F). To identify the dysregulated genes in the placenta of PE rats, placentas from G13 rats were collected and subjected to bulk RNA-seq analysis. The heatmap (Fig. 5G) revealed distinct differences in the placental transcriptome between the Ad Fc + PBS group and the Ad Flt1 + PBS group. Functional enrichment and GO analyses of DEGs identified several biological processes associated with PE, including those related to organ development, cell migration, female pregnancy, extracellular matrix organization, tube formation, as well as placenta and uterus development (Fig. 5H). The KEGG hierarchical network plot showed significant enrichment of pathways related to signal transduction, signaling molecules and interaction, and cardiovascular disease in the transcriptomes of Ad Flt1-induced PE rats (Fig. 5I). GSEA further identified the cytokine‒cytokine receptor interaction pathway as the most significantly enriched pathway (Fig. 5J), within which Bmp6 was prominently enriched (Fig. 5K). Volcano plots of gene expression revealed that placental levels of Bmp6, Flt1, and Serpine2 were significantly higher in the Ad Flt1 + PBS group compared with the Ad Fc + PBS group (Fig. 5L). Moreover, the Circos graph displays the coexpression networks of Bmp6, Flt1, Serpine2, Pgf, and Id1 in rat placenta samples, suggesting that both Bmp6 and Serpine2 are positively related to Flt1 and that Bmp6 is also positively associated with Serpine2 (Fig. 5M). These results indicate that BMP6 is compensatorily upregulated in the PE rat model during late pregnancy.

Supplementation with Recombinant BMP6 alleviates PE-related phenotypes in an ad Flt1-induced PE rat model

To explore whether BMP6 supplementation could alleviate PE-related phenotypes in vivo, we assessed a series of pathological changes in four experimental groups: the Ad Fc + PBS, Ad Fc + BMP6, Ad Flt1 + PBS, and Ad Flt1 + BMP6 group. We found that compared with the Ad Flt1 + PBS group, BMP6 supplementation during G10 to G13 rescued the increase in SBP and MAP in the Ad Flt1 + BMP6 group (Fig. 6A) and improved fetal growth restriction (Fig. 6B and C), placental efficiency (calculated as fetal weight/placental weight) (Fig. 6C), and the placental labyrinth/junction ratio (Fig. 6D). As revealed by immunohistochemistry of CD34 (the marker of endothelial cells) and α-SMA (the marker of smooth muscle), BMP6 supplementation increased blood vessel formation in the placenta (Fig. 6E) and blood vessel remodeling in the uterus (Fig. 6F) of rats in the Ad Flt1 + BMP6 group compared with the Ad Flt1 + PBS group. Additionally, RT‒qPCR assay of placentas at G19 revealed that Bmp6 and Serpine2 mRNA levels were higher in the Ad Flt1 + PBS group than in the Ad Fc + PBS group, and BMP6 supplementation reduced placental Bmp6 mRNA levels in the Ad Flt1 + BMP6 group compared with the Ad Flt1 + PBS group (Fig. S8A). Importantly, compared with the Ad Flt1 + PBS group, BMP6 supplementation reduced plasma sFlt-1 levels in the Ad Flt1 + BMP6 group (Fig. S8B). However, BMP6 supplementation did not affect the urine albumin/creatinine ratio (Fig. S8C) or glomerular damage (Fig. S8D) in the Ad-Flt1 + BMP6 group compared with the Ad-Flt1 + PBS group. Taken together, these data indicate the significant role of BMP6 in alleviating PE phenotypes.

Fig. 6.

Supplementation with recombinant BMP6 alleviates PE-related phenotypes in the Ad Flt1-induced PE rat model. An SD rat model of PE was established as described in Fig. 5A. Pregnant rats were randomly divided into four groups: the Ad Fc + PBS, Ad Fc + BMP6, Ad Flt1 + PBS, and Ad Flt1 + BMP6 groups. A, BMP6 supplementation rescues the increased blood pressure in the Ad Flt1-induced PE rat model. SBP and MAP of pregnant rats in each group (n = 4). B-C, BMP6 supplementation rescues fetal growth restriction and placental hypoefficiency in the Ad Flt1-induced PE rat model. B, Representative images of fetal rats and placentas from each group at G19 (n = 7). C, Weights of fetal rats and corresponding placental efficiency at G19 in the Ad Fc + PBS group (n = 33), Ad Fc + BMP6 group (n = 39), Ad Flt1 + PBS group (n = 32), and Ad Flt1 + BMP6 group (n = 47). D-F, BMP6 supplementation alleviates placental damage in the Ad Flt1-induced PE rat model. D, HE staining of rat placentas was performed to observe the placental labyrinth and junction areas. Representative images are presented in the left panel, and the placental labyrinth/junction ratios in each group were quantified and summarized in the right panel (n = 8). Scale bar, 3 mm. E, Immunohistochemistry localization of CD34 in rat placenta on G19. Representative images are presented in the left panel, and the ratios of CD34-positive area were quantified and are summarized in the right panel. Scale bar, 100 μm. F, Immunohistochemistry localization of α-SMA in the rat uterus on G19. Scale bar, 100 μm. Representative images are presented in the left panel, and the ratios of un-remodeled blood vessels were quantified and are summarized in the right panel. The quantitative results are expressed as the means ± SEMs of at least three independent experiments. Two-way ANOVA was used for data comparison. Groups without common letters are significantly different from each other (P < 0.05). SD, Sprague–Dawley; SBP, systolic blood pressure; MAP, mean arterial pressure; PBS, phosphate-buffered saline; Ad Fc, adenovirus-expressing control IgG2a Fc fragment; Ad Flt1, adenovirus expressing fms-like tyrosine kinase-1; LZ, labyrinth zone; JZ, junction zone

Discussion

The present study reveals that BMP6 promotes trophoblast invasion by upregulating ID1-mediated SERPINE2 and PlGF expression in a SMAD signaling-dependent manner and is thus crucial for appropriate placentation (Fig. 7). Supplementation with BMP6 during the early gestational stage significantly alleviates PE-related phenotypes (including maternal hypertension and fetal growth restriction) in an Ad Flt1-induced PE rat model. These findings underscore the potential of BMP6 as a novel target for the clinical diagnosis and treatment of PE.

TGF-β superfamily members, such as activin A [38, 39] and BMP2 [22], are known to promote trophoblast invasion and are upregulated during late pregnancy in patients with PE, possibly as compensatory responses to PE-related placental dysfunction. Similarly, we identified BMP6, a downstream target of BMP2, as a proinvasive and proangiogenic factor in human trophoblasts, with increased placental BMP6 expression in women diagnosed with PE compared with their gestational age-matched controls. While published bulk RNA-seq datasets report reduced or comparable placental BMP6 in early-onset PE [40], these apparently contradictory findings likely reflect gestational-stage specificity: BMP6 upregulation was exclusively detectable only in late pregnancy, when a compensatory effect could have manifested and become detectable. Previous studies have also reported the compensatory upregulation of BMP6 after myocardial infarction in mice, and the administration of recombinant BMP6 has been shown to alleviate abnormal ventricular remodeling after myocardial infarction [41]. BMP6 has been identified as a potential biomarker for predicting coronary heart disease [42], and supplementation with recombinant BMP6 has been shown to lower blood glucose and lipid levels in mice [43]. Its therapeutic potential also extends to bone health, BMP6 exhibits greater bone-forming capability than BMP2 [44]. Notably, while recombinant BMP2 alleviated PE-related phenotypes (including maternal hypertension and fetal growth restriction) in Ad Flt1-induced PE rats without affecting plasma sFlt-1 levels [22], BMP6 supplementation in our study not only improved these phenotypes but also reduced sFlt-1 levels, suggesting a broader mechanism of action of BMP6.

Previous studies have drawn analogies between cancer and placentation, particularly in the context of the invasive behavior of trophoblast [45, 46]. Our study showed that ID1 mediates BMP6-induced trophoblast invasion and vascular mimicry, aligning with its established proinvasive and proangiogenic functions in cancers [23, 47]. While ID1 has been suggested to be pivotal during stromal cell decidualization and is overexpressed in PE decidual stromal cells [48], its role in PE placentas has rarely been explored in previous studies. We observed elevated placental ID1 mRNA levels in PE patients compared with their gestational age-matched controls and higher ID1 levels in EVTs in PE placentas based on scRNA-seq data. However, no significant differences in placental Id1 mRNA levels were found between Ad Flt1-induced PE rats and uninduced rats, potentially due to high baseline Id1 transcription in rat placentas. These findings highlight the need for further validation in human samples and study models, given the limitations of animal models in fully recapitulating PE pathophysiology [49, 50].

In our investigation of crucial genes associated with invasion that are downstream of ID1, we found that placental Serpine2 mRNA levels were elevated in Ad Flt1-induced PE rats compared with uninduced controls. SERPINE2, a secreted protein with anti-serine protease activity that directly inhibits thrombin, is upregulated in various vascular pathologies to reestablish homeostasis following injury or stress [51, 52]. Similarly, this upregulation was observed in our PE model rats. PE shares pathological similarities with thrombotic microangiopathies and atherosclerosis [53, 54]. Women with PE exhibit increased thrombin levels compared with controls [55], suggesting that the upregulation of SERPINE2 in PE may be a compensatory mechanism to counteract elevated thrombin levels and maintain cardiovascular system homeostasis.

PlGF, a well-established proangiogenic factor critical for maternal cardiovascular adaptation and placental development [56], is suppressed in PE due to increased sFlt-1 levels [37]. While lower plasma PlGF levels are associated with PE, the cause-and-effect relationship remains unclear. It has been reported that PlGF KO in mice did not induce PE-like symptoms during late gestation, suggesting that low PlGF levels were unlikely to contribute to the development of PE [57]. Our study reveals increased plasma PlGF levels in Ad Flt1-induced PE rats compared with uninduced rats during middle pregnancy, possibly reflecting a compensatory response to angiogenic imbalance. Furthermore, the necrosis, apoptosis, autophagy, and excessive release of extracellular vesicles by SCT may also lead to an additional release of PlGF. Interestingly, there are four PlGF splice variants in humans, while only one variant is found in rats, indicating a more complex PlGF biology and function in humans, which deserves further study. Importantly, BMP6-induced PlGF upregulation contributed to trophoblast invasion and vascular mimicry, suggesting that BMP6 supplementation might provide a more targeted strategy for restoring angiogenic balance in PE. Notably, beyond the well-characterized defects in trophoblast invasion and spiral artery remodeling, impaired STB fusion has emerged as another hallmark pathological feature of preeclampsia. Even with adequate vascularization, proper placental function depends on vascular membrane formation. The role of BMP6 signaling in regulating STB fusion warrants further investigation.

Some limitations should be considered when interpreting this study. First, the number of clinical samples in our study was relatively small, and PE cases were not stratified as early- or late-onset. In early-onset PE, placental development is abnormal from the outset and fails to progress adequately, whereas in late-onset PE, placental growth is initially appropriate but later fails at term, reflecting a late maladaptation of the placenta to the growing fetus [58]. Further investigations with larger cohorts and gestational-age-stratified analyses are needed to unravel the complexity of PE pathogenesis. Furthermore, although chorionic villi and decidua tissues were obtained from healthy pregnant women, no prenatal or histopathological diagnostic data were available for these patients. Secondly, BMP6 circulates in the blood and thus has the potential to exert effects on distant tissues and organs. Therefore, the observed rescue effect of BMP6 in PE may not be limited to trophoblast invasion and vascular mimicry; BMP6 may regulate various aspects of pregnancy, including maternal cardiovascular maladaptation. In addition, BMP6 treatment primarily ameliorated maternal hypertension and placentation in rats but had limited effects on the repair of glomerular endotheliosis. Future studies should focus on developing novel targeted delivery strategies for BMP6 (e.g., utilizing nanotechnology) or other agents further to enhance therapeutic efficacy for PE beyond blood pressure control. Based on our advancing knowledge of the regulation of BMP6 in PE, we identify opportunities for therapeutic intervention in PE and hope to inspire further research into the physiological roles of BMP6 in PE development.

Conclusions

In conclusion, our study demonstrates that BMP6 promotes trophoblast invasion by upregulating SERPINE2 and PlGF expression in an ID1-dependent manner. Furthermore, BMP6 is compensatorily upregulated in the placentas of PE patients and Ad Flt1-induced PE model rats during late pregnancy. BMP6 supplementation during the early gestational stage alleviates PE-related phenotypes in PE model rats. Our findings provide novel insights into the molecular regulation of BMP6 in the placenta and foundations for future development of drug-based interventions targeting insufficient trophoblast invasion in PE.

Methods

Patients and sample collection

Placental tissues were obtained from women diagnosed with PE (PE group, n = 10) and their gestational age-matched controls (control group, n = 10) who achieved singleton deliveries at the Department of Obstetrics, Shandong Provincial Hospital. Gestational age-matched controls included women with spontaneous preterm birth due to cervical insufficiency and women with term delivery, all without pregnancy complications or infectious disease. PE samples from women who developed other pregnancy complications, such as gestational diabetes mellitus, placental abruption, placenta previa, and postpartum hemorrhage, were excluded. PE was defined as the de novo development of concurrent high blood pressure (≥ 140/90 mm Hg) after 20 weeks of gestation and proteinuria (≥ 300 mg/L per 24 h) or other maternal organ dysfunction, such as renal or liver involvement, neurological or hematological complications, or uteroplacental dysfunction (e.g., fetal growth restriction, abnormal umbilical artery Doppler waveform analysis, or stillbirth) [59]. For placental sample collection, the central maternal side of the placenta was dissected within 5 min after placental separation, washed with cold sterile saline, aliquoted into frozen pipes, quickly frozen, and stored in liquid nitrogen. The detailed clinical characteristics of the participants are summarized in Supplementary Table 1.

Human first-trimester chorionic villi and decidua from healthy pregnant women who underwent elective terminations due to unintended pregnancy (6–8 weeks of gestation) were obtained from the Department of Obstetrics and Gynecology, the Second Hospital of Shandong University. Human EVT isolation was performed within 1 h of artificial abortion-vacuum aspiration.

This study was approved by the School of Medicine Ethics Board of Shandong University, Jinan, China (Approval no. SDULCLL2021-1–15), for human sample collection. All participants provided informed consent for the use of their samples.

Immunohistochemistry

Tissues were fixed with 4% PFA, dehydrated, and embedded in paraffin wax. Serial Sect. (5 μm thick) were prepared for immunohistochemistry using the Rabbit Two-Step Assay Kit (ZSGB-BIO, PV-9001) or the Mouse Two-Step Assay Kit (ZSGB-BIO, PV-9002) according to the manufacturer’s instructions. Briefly, sections were first deparaffinized in xylene and rehydrated through a graded ethanol series, followed by antigen retrieval for 20 min, endogenous peroxidase block for 15 min, and 10% goat serum containing 0.3% (vol/vol) Triton X-100 block for 60 min. The sections were subsequently incubated overnight at 4 °C with primary antibodies against BMP6 (1:1000), ID1 (1:200), CD34 (1:200), and α-SMA (1:400), or a negative control (rabbit or mouse immunoglobulin G isotype control). The following day, sections were incubated with secondary antibodies for 20 min and stained with a diaminobenzidine (DAB) substrate kit (Vector Laboratories, SK-4100, Burlingame, CA, USA) and hematoxylin. Images were taken by an Aperio VERSA scanner (Leica Microsystems, Nanterre, France). For the quantitative analysis, five randomly selected fields of view were analyzed for each section, and the results are presented as the means.

Culture of immortalized HTR8/SVneo human trophoblast

The HTR8/SVneo immortalized human EVT cell line (ATCC, CRL-3271, Manassas, VA, USA) was cultured in complete medium: Dulbecco’s modified Eagle medium nutrient mixture F-12 Ham (DMEM/F12; Gibco, 11320033, Grand Island, NY, USA) supplemented with 10% (vol/vol) heat-inactivated fetal bovine serum (FBS; ExCell Bio, FSP500, Suzhou, China), 100 U/mL penicillin, and 100 µg/mL streptomycin (KeyGEN BioTECH, KGY0023, Nanjing, China). The cells were incubated in a 37 °C humidified incubator with 5% CO₂. To avoid possible interference of growth factors from FBS, the cells were starved in DMEM/F12 containing 0.1% (vol/vol) FBS, 100 U/mL penicillin, and 100 µg/mL streptomycin for 24 h before treatment with BMP6 (R&D, 507-BP-020, Minneapolis, MN, USA).

Matrigel-coated transwell invasion assays

Twenty-four-well transwell inserts (pore size 8 μm; Corning, 3422, NY, USA) that separated the upper and lower chambers were used for cell invasion assays. Growth factor-reduced Matrigel (Corning, 354230) was thawed overnight on ice in a 4 °C refrigerator and diluted at a 1:9 (vol/vol) ratio to achieve a final concentration of 1 mg/mL in DMEM/F12 containing 0.1% (vol/vol) FBS. The upper chambers of the inserts were coated with 50 µL of diluted Matrigel and incubated for 2 h at 37 °C in a humidified incubator until the Matrigel solidified. HTR8/SVneo cells and primary EVTs were pretreated with vehicle or BMP6 (50 ng/mL) for 20 min, after which 8 × 10⁴ cells in 250 µL of DMEM/F12 (supplemented with 0.1% FBS and 50 ng/mL vehicle or BMP6) were seeded in the upper chambers of the inserts, and 750 µL of DMEM/F12 medium supplemented with 10% FBS was added to the lower chambers. Then, the transwell inserts were incubated at 37 °C in a humidified incubator with 5% CO₂, and the cells in each insert were retreated with vehicle or BMP6 (50 ng/mL) 24 h later and incubated for an additional 12 h. After a total of 36 h of incubation, the noninvaded cells in the upper side of the membrane were removed, and the cells on the lower side were fixed with 4% PFA, followed by staining with Hoechst 33,258. Digital images were taken with an Olympus IX73 inverted microscope (Olympus, Tokyo, Japan) and analyzed with ImageJ.

Endothelial-like tube formation assay

The growth factor-reduced Matrigel was thawed on ice overnight and diluted 1:1 (vol/vol) to a final concentration of 5 mg/mL with DMEM/F12 containing 0.1% FBS. Then, 50 µL of diluted Matrigel was added to each well of a 96-well plate and incubated at 37 °C for 2 h until the Matrigel solidified. HTR8/SVneo cells were pretreated with vehicle or BMP6 (50 ng/mL) for 20 min and then were resuscitated with DMEM/F12 (supplemented with 0.1% FBS and 50 ng/mL vehicle or BMP6). Then, 100 µL of cell suspension containing 4 × 10⁴ cells was seeded in wells coated with Matrigel and incubated at 37 °C for 12 h. Digital images were captured using an Olympus IX73 inverted microscope, and the total tube length was measured using ImageJ software.

Isolation and culture of human primary EVTs

Chorionic villous tissue samples were collected from healthy pregnant women who underwent elective terminations via artificial abortion-vacuum aspiration during the first trimester of pregnancy (6–8 weeks). The isolation approach and culture of human primary EVTs have been described previously [60]. Briefly, the chorionic villous tissue was rinsed several times with cold phosphate-buffered saline (PBS) to remove blood, and then the villous tissue was dissected into 1–2 mm fragments using ophthalmic scissors. The tissue sections were planted and incubated in 60 mm dishes containing 3–4 mL complete medium for 3 days in a 37 °C humidified incubator with 5% CO₂. Subsequently, PBS was used to wash the nonattached tissue sections, and adherent explants and migrated EVTs were cultured for an additional 10–14 days until they reached 90% confluence. During this time, the medium was discarded and replenished every two days. The purity of the primary trophoblast culture was verified by cytokeratin-7 (CK7) and human leukocyte antigen G (HLA-G) antibody staining, yielding 99% positive results.

Immunofluorescence

Human primary EVTs were seeded in 15 mm glass-bottom cell culture dishes (NEST.801002, Wuxi NEST Biotechnology Co., Ltd., Wuxi, China) and cultured in a 37 °C humidified incubator with 5% CO₂. After reaching 70% confluence, primary EVTs were fixed with 4% paraformaldehyde (PFA) for 20 min, permeabilized with 0.3% (vol/vol) Triton X-100 for 40 min, and blocked with 10% goat serum (ZSGB-BIO, ZLI-9056, Beijing, China) for 1 h. Then, the cells were incubated overnight at 4 °C in the dark with CK7-specific rabbit polyclonal antibody (1:200) and HLA-G-specific mouse monoclonal antibody (1:100). The next day, the cells were incubated with a fluorescent goat anti-rabbit secondary antibody (1:800) and a fluorescent goat anti-mouse secondary antibody (1:800) for 1 h. Primary EVT nuclei were counterstained with Hoechst 33,258 (Abcam, ab228550, Cambridge, MA, USA). Images were captured under a confocal laser scanning microscope and analyzed with ImageJ (NIH, USA).

scRNA-seq data analysis

In this study, the scRNA-seq datasets of placentas obtained from 5 women during early pregnancy [35] and 6 women during late pregnancy [36] were reanalyzed using the Seurat R package (version 4.4). The cells were first filtered with the following criteria: >500 genes and a percentage of mitochondrial genes < 20%. To remove batch effects caused by background contamination of cell-free RNA, we removed a set of genes that tended to be expressed in ambient RNA (PAEP, HBG1, HBA1, HBA2, HBM, AHSP, and HBG2). A filtered gene‒barcode matrix of all the samples was subsequently integrated to remove batch effects across different donors. Finally, 3000 highly variable genes were identified via Seurat’s ‘FindVariableFeatures()’ function, and the data were then scaled via ‘ScaleData()’. Principal component analysis (PCA) was performed via the ‘harmony’ package. The ‘RunHarmony()’ function was used to integrate the data. Uniform Manifold Approximation and Projection (UMAP) dimension reduction with 50 principal components was performed. A nearest-neighbor graph using the 20 principal components of the PCA reduction was calculated by ‘FindNeighbors()’, followed by clustering using ‘FindClusters()’ with a resolution of 0.20. Conserved markers for each cluster were identified via ‘FindAllMarkers()’. The cells were reannotated according to the markers identified in the original publication. To visualize ID1, ID2, ID3, SERPINE2, and PGF expression among different cell types in the placenta, ‘FeaturePlot()’ and ‘VlnPlot()’ were used.

Antibodies and reagents

SMAD1 (D59D7) rabbit monoclonal antibody (#6944), phospho-SMAD1/5/9 (D5B10) rabbit monoclonal antibody (#13820), SMAD2 (D43B4) rabbit monoclonal antibody (#5339), phospho-SMAD2 (E8F3R) rabbit monoclonal antibody (#18338), SMAD3 (C67H9) rabbit monoclonal antibody (#9523), phospho-SMAD3 (C25A9) rabbit monoclonal antibody (# 9520), SMAD4 (D3R4N) rabbit monoclonal antibody (#46535), horseradish peroxidase (HRP)-linked anti-mouse IgG (#7076), and anti-rabbit IgG (#7074) were purchased from Cell Signaling Technology (Danvers, MA, USA); the mouse monoclonal antibody HLA-G (#11–499-C100) was purchased from EXBIO (Vestec, Czech Republic); CK7-specific rabbit polyclonal antibody (#17513-1-AP), BMP6 rabbit polyclonal antibody (#55421-1-AP), SERPINE2 rabbit polyclonal antibody (#11303-1-AP), and α-Tubulin mouse monoclonal antibody (#66031-1-Ig) were purchased from Proteintech (Wuhan, China); ID1 mouse monoclonal antibody (#sc-133104) was obtained from Santa Cruz Biotechnology (Santa Cruz, CA); CD34 rabbit monoclonal antibody (#A19015) and α-SMA rabbit monoclonal antibody (#A2235) were obtained from ABclonal (Wuhan, China); Goat anti-Rabbit IgG secondary antibody, Alexa Fluor 594 (#A-11012) and goat anti-Mouse IgG secondary antibody, Alexa Fluor 488 (#A-11011) were purchased from Thermo Fisher (NY, USA).

Reverse transcription real-time quantitative PCR (RT‒qPCR)

Total RNA from tissue samples was extracted via TRIzol reagent (15596026CN, Thermo Fisher), while total RNA from HTR8/SVneo cells and primary human EVTs was obtained via an RNA-Quick Purification Kit (YiShan Biotech, RN001, Shanghai, China) according to the manufacturer’s instructions. The concentration and purity of the total RNA were assayed via UV spectrophotometry, and the optical density ratio at 260/280 nm was between 1.8 and 2.1. A PrimeScript™ RT Reagent Kit (TaKaRa, RR047A, Kusatsu, Japan) was used for complementary DNA synthesis, and mRNA expression levels were quantified using TB Green™ Premix Ex Taq™ (TaKaRa, RR420A). RT‒qPCR was performed on a Roche LightCycler 480 instrument (Roche, Basel, Switzerland) equipped with 96-well (Roche, 04729692001) or 384-well (Roche, 04729749001) optical reaction plates. The relative quantification of the mRNA levels was performed via the 2^− ΔΔCT method, and the levels were normalized to those of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) or Gapdh. Gene-specific primers were synthesized by BioSune (Shanghai, China), and their sequences are listed in Supplementary Table 2.

Western blot

The cells and rat placental tissue were washed with cold PBS and lysed on ice with RIPA lysis buffer (Beyotime, p0013b; Shanghai, China) supplemented with a protease/phosphatase inhibitor cocktail (Cell Signaling Technology, 5872). The extracts were centrifuged at 14,000 g for 20 min at 4 °C, and the supernatants were collected for protein concentration determination via a BCA protein assay kit (Thermo Fisher, 23227). The samples were subsequently treated with SDS-PAGE protein loading buffer (Beyotime, P0015) and boiled at 100 °C for 10 min to reduce and denature the proteins. Equal amounts of protein (20 µg) in each lane were electrophoresed on SDS-polyacrylamide gels (Smart Life Science, Changzhou, China) and then electrotransferred to PVDF membranes (Millipore, IPVH00010, Burlington, MA, USA). The membranes were blocked with 5% nonfat milk or 5% bovine serum albumin (BSA, Solarbio, A8020, Beijing, China) at room temperature for 1 h and then incubated with primary antibodies against SMAD1 (1:1000), phospho-SMAD1/5/9 (1:1000), SMAD2 (1:1000), phospho-SMAD2 (1:1000), SMAD3 (1:1000), phospho-SMAD3 (1:2000), SMAD4 (1:1000), BMP6 (1:500), ID1 (1:500), SERPINE2 (1:1000), and α-Tubulin (1:10000) at 4 °C overnight. On the second day, the appropriate HRP-conjugated secondary antibodies were incubated at room temperature for 1 h, and signals were detected using the Bio-Rad ChemiDoc MP Imaging System and Image Lab Software (Bio-Rad, USA) with a chemiluminescent HRP substrate kit (Millipore, WBKLS0500). Densitometric quantification of the protein bands was performed using ImageJ software with α-tubulin as the normalization standard.

Small interfering RNA (siRNA) transfection

The cells were seeded in 6-well plates and were transfected when they reached approximately 50%−60% confluence. Transfection was performed by using Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher, 13778-150) and Opti-MEM I (Gibco, 31985070) according to the manufacturer’s instructions. The cells were transfected for 48 hours with 20 nM siRNA. ON-TARGETplus NON-targeting control siRNA or ON-TARGETplus SMARTpool siRNA targeting human ID1, SERPINE2, PGF, and SMAD4 were purchased from Dharmacon (Lafayette, CO, USA). siRNAs targeting human SMAD1, SMAD5, SMAD9, SMAD2, and SMAD3 and control siRNAs were purchased from GenePharma (Shanghai, China). SMAD1 siRNAs (5’ to 3’): GGUGCUCUAUUGUCUACUATT; UAGUAGACAAUAGAGCACCTT. SMAD5 siRNAs (5’ to 3’): CCAGCAGUAAAGCGAUUGUTT; ACAAUCGCUUUACUGCUGGTT. SMAD9 siRNAs (5’ to 3’): CUCUAGUGAAGAAGUUAAATT; UUUAACUUCUUCACUAGAGTT. SMAD2 siRNAs (5’ to 3’): GGUGUUCGAUAGCAUAUUATT; UAAUAUGCUAUCGAACACCTT. SMAD3 siRNAs (5’ to 3’): GCGUGAAUCCCUACCACUATT; UAGUGGUAGGGAUUCACGCTT. The knockdown efficiency was assessed via RT‒qPCR, Western blotting, or enzyme-linked immunosorbent assays (ELISAs).

EdU proliferation assay

Cell proliferation was assessed via a Cell-Light EdU Apollo 567 In Vitro Kit (RiboBio, C10310-1, Guangzhou, China) according to the manufacturer’s instructions. After BMP6 treatment or siRNA transfection, the cells were seeded in 96-well plates until they reached 60%−70% confluence. They were then incubated with 100 µL of 50 µM EdU medium per well for 2 h at 37 °C. Then, the cells were fixed with 4% PFA for 30 min. After that, the cells were incubated with 2 mg/mL glycine for 5 min and 0.5% Triton X-100 for 10 min, and subsequently stained with Apollo solution for 30 min, followed by Hoechst 33,342 for an additional 30 min. An Olympus IX73 inverted microscope was used to capture images, and ImageJ software was used for image analysis. The percentage of EDU-positive cells was calculated by dividing the total number of EDU-positive nuclei by the total number of nuclei. Triplicate wells were used for each experiment, and at least five random fields per well were counted.

Apoptotic assay

For the TUNEL apoptosis assay, a TUNEL BrightRed Apoptosis Detection Kit (Vazyme, A113, Nanjing, China) was used according to the manufacturer’s instructions. In brief, BMP6- or siRNA-treated cells were seeded in 96-well plates until they reached 80%−90% confluence. Afterward, the cells were fixed with 4% PFA for 30 min, permeabilized with 0.3% Triton X-100 for 5 min, and then equilibrated with equilibration buffer for 15 min. Cell nuclei were stained with TdT buffer and Hoechst 33,258. Images were taken with an Olympus IX73 inverted microscope, followed by analysis with ImageJ software. The percentage of TUNEL-positive cells was calculated by dividing the total number of TUNEL-positive nuclei by the total number of nuclei. Triplicate wells were used for each experiment, and at least five random fields per well were counted.

For the flow cytometry-based apoptosis assay, an Annexin V-APC/7-AAD apoptosis kit (MULTISCIENCES, AP105, Hangzhou, China) was used according to the manufacturer’s instructions. The cells were seeded in 6-well plates and subjected to BMP6 treatment or siRNA transfection. Then, the cells were collected and centrifugally washed with cold PBS. The resulting precipitate was finally resuspended in 500 µL of binding buffer, followed by staining with Annexin V-APC and 7-AAD for 5 min in the dark at room temperature. Flow cytometry-based apoptosis analysis was performed via a flow cytometer (BD FACSCalibur, BD Biosciences, CA, USA).

Measurement of PlGF accumulation in conditioned medium

The cell culture medium was centrifuged to remove particulates and then either assayed immediately or stored at − 80 °C until further analysis. PlGF accumulation in the cell culture supernatant was measured via a Human PlGF Quantikine ELISA Kit (R&D Systems, DPG00) according to the manufacturer’s instructions (both intra-assay and inter-assay with coefficients of variation less than 10%) and was normalized to the total cellular protein concentration.

Animal study

Pregnant Sprague–Dawley (SD) rats on day 4 of gestation were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. All the rats, aged 8–10 weeks, were fed a standard laboratory diet and housed in temperature- and humidity-controlled houses with constant light: dark cycles of 12 h: 12 h. Pregnant rats were randomly divided into 4 groups: Ad Fc + PBS (N = 7), Ad Fc + BMP6 (N = 4), Ad Flt1 + PBS (N = 7), and Ad Flt1 + BMP6 (N = 4). A total of 1 × 10⁹ PFU of adenovirus (Ad Fc or Ad Flt1) was injected into the tail vein at G8. Adenoviruses expressing sFlt-1 or control Fc have been described elsewhere [61, 62] and were produced by Genechem (Shanghai, China). Recombinant BMP6 protein (10 µg/kg/day) (R&D, 10420-BM-020) or vehicle was injected into the tail vein from G10 to G13. On G13, three rats in the Ad Fc + PBS group and three rats in the Ad Flt1 + PBS group were sacrificed by cervical dislocation, and one placenta was collected from each rat for bulk RNA-seq. The other rats were sacrificed at G19 for tissue collection, and the harvested fetal rats and placentas were subjected to morphological assessment and weight measurement. After initial assessment, the placentas were fixed with 4% PFA for histology or quickly frozen and stored in liquid nitrogen until RT‒qPCR analysis. Kidneys were also collected on G19 for histology.

The systolic blood pressure (SBP), diastolic blood pressure (DBP), and mean arterial pressure (MAP) of G7, G10, G14, G16, and G18 rats were recorded via intelligent noninvasive blood pressure monitoring (BP-2010 A, Softron Biotechnology, Beijing, China) via tail-cuff plethysmography. Before the measurement, each rat was wrapped in a bag and restrained in a rat net. The net was placed in a thermal insulation cylinder to maintain a constant temperature of 32 °C for 15–20 min, which helped calm the rat down. A dark and warm environment helps the rat be quiet and promotes blood circulation, thus achieving good measurement. At least three consecutive measurements were performed with an interval of 1 min, and the averaged blood pressure indices were used for the final analyses. Rat blood from G7, G10, G14, and G19 was obtained from the inner canthal vein and collected in anticoagulant tubes containing ethylenediaminetetraacetic acid (EDTA) for plasma isolation. Rat plasma was measured with a sFlt-1 ELISA kit (Proteintech, KE10069), a BMP6 ELISA Kit (Novus Biologicals, NBP2-69999, Centennial, CO, USA), and a PlGF ELISA kit (CUSABIO, CSB-E07400r, Wuhan, China). Rat urine samples from G7, G14, and G19 were analyzed with an albumin ELISA kit (Abcam, ab108789) and a creatinine assay kit (Abcam, ab65340), respectively. This study was approved by the School of Medicine Ethics Board of Shandong University, Jinan, China (Approval no. SDULCLL2021-2–13).

Bulk RNA-seq and data processing

The RNA-Seq data of HTR8/SVneo cells treated with BMP6 were downloaded from the Genome Sequence Archive for Humans (HRA001422). Rat placental RNA samples from the Ad Fc + PBS group (n = 3) and Ad Flt1 + PBS group (n = 3) were extracted via TRIzol reagent and sent to Novogene (Beijing, China) for RNA-seq analysis. The raw sequencing data are available in the Sequence Read Archive (SRA) database (PRJNA1173398). The expression levels of relevant differentially expressed genes were verified by RT‒qPCR and/or western blotting.

The analysis of differentially expressed genes (DEGs) was conducted via the DESeq2 (version 1.44.0) R package. Volcano plots and waterfall plots were generated via the ggplot2 R package (version 3.5.1). Heatmaps were generated via the ComplexHeatmap R package (version 2.20.0). Functional enrichment analysis (including GO, pathway, and GSEA) was performed via the clusterProfiler R package (version 4.12.6). Transcription factor activity deduction analysis was conducted by the decoupleR package (version 2.14.0). A Circos plot was generated via the circlize R package (version 0.4.16). A KEGG hierarchical network plot was produced employing the ggraph R package (version 2.2.2). R packages were obtained from the Bioconductor website (http://www.bioconductor.org/), and the results were visualized using R Studio.

Statistical analysis

For continuous variables, normality was assessed using graphical methods, including histograms and Q–Q plots, as well as the Shapiro-Wilk test. For normally distributed data, group comparisons were performed via Student’s t test, one-way ANOVA, or two-way ANOVA. The two factors used in the two-way ANOVA were BMP6 condition (with or without BMP6 treatment) and siRNA treatment or adenoviral treatment (Ad Fc vs. Ad Flt1). For non-normally distributed data, the Mann–Whitney U test or Kruskal-Wallis test was performed. Correlation analysis was performed using Spearman correlation analysis. Categorical variables were compared via either chi-square analysis or Fisher’s exact test. Normally distributed variables are shown as the means ± SDs/SEMs, whereas nonnormally distributed data are shown as the medians (25th percentile to 75th percentile). Categorical variables are expressed as the number of cases (n) and the percentage of occurrence (%). Data comparisons were performed via SPSS 26.0 (IBM, Armonk, NY, USA) and GraphPad Prism 8.0 (GraphPad Software, San Diego, CA, USA). A P value of < 0.05 was considered to indicate a statistically significant difference.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

Ad Flt1 adenovirus

expressing fms-like tyrosine kinase-1

Ad Fc adenovirus

expressing control IgG2a Fc fragment

BMP

bone morphogenetic protein

Bulk RNA-seq

bulk RNA sequencing

CK7

cytokeratin 7

DEG

differentially expressed genes

ELISA

enzyme linked immunosorbent assay

EVT

extravillous cytotrophoblast

FBS

fetal bovine serum

GSEA

Gene set enrichment analysis

HLA-G

human leukocyte antigen G

MAP

mean arterial pressure

PBS

phosphate buffered saline

PFA

paraformaldehyde

PlGF

placental growth factor

RT‒qPCR

Reverse transcription real-time quantitative PCR

scRNA-seq 

single-cell RNA sequencing

SCT

syncytiotrophoblast

SERPINE2

serpin family E member 2

siRNA 

small interfering RNA

TGF-β

transforming growth factor-β

VCT

villous cytotrophoblast

Author contributions

PZ and DMW conceived and designed the project; YN performed the experiments and analyzed the data; HTY, SWH, HYX, and MXL performed bioinformatics analysis; YHY and XXL collected the clinical samples and information; YN and CK wrote the manuscript. HTY, YL and DMW critically revised the manuscript. All authors have been involved in interpreting the data and approved the final version.

Funding

This study was supported by the National Key Research and Development Program of China (2022YFC2703502, 2023YFC2705502, 2023YFC2706005), the National Natural Science Foundation of China (82495194, 82101784, and 82470359), and the Natural Science Foundation of Shandong Province (ZR2023MH141).

Data availability

All study data are included in the article and supporting information. All bioinformatic analyses were performed using publicly available software as described in Methods. The bulk RNA-seq data are available in the Sequence Read Archive (SRA) database with the accession number PRJNA1173398. Further information and requests for resources and reagents should be directed to and will be fulfilled by Daimin Wei (sdweidaimin@163.com).

Declarations

Ethics approval and consent to participate

The animal study protocol received approval from the Ethics Committee and was performed in accordance with the Animal Care and Use Committee Guidelines of Shandong University (No. SDULCLL2021-2-13). Collection of human tissues was approved by the School of Medicine Ethics Board of Shandong University, Jinan, China (No. SDULCLL2021-1-15) and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from participants who donated tissues.

Competing interests

All authors declare that they have no competing interests.