Defective O-fucosylation of EDIL3 attenuates invasion and vascular remodeling of trophoblasts via LIFR signaling pathway in preeclampsia

Abstract

Preeclampsia (PE) is a severe pregnancy disorder caused by placental dysfunction. Protein O-fucosylation is a type of protein post translational modification that is catalyzed by protein O-fucosyltransferases (poFUTs). However, the role and underlying mechanisms of O-fucosylation/poFUT1 in PE remain elusive. Here, we revealed a lower level of poFUT1 in the plasma and placental tissues of PE patients than in normal pregnancy (NP) women. Moreover, poFUT1 deletion induced PE-like phenotypes in a mouse model. Mechanistically, we globally screened O-fucosylated proteins and identified EDIL3 with an O-fucosylation site at threonine 88. Furthermore, O-fucosylation-EDIL3 can directly interact with LIFR on trophoblasts, consequently activating the STAT3 signaling pathway, promoting the invasion and vascular remodeling ability of trophoblasts. Conversely, de-O-fucosylation-EDIL3 aggravated PE-like phenotypes by attenuating placental development in vitro and in vivo. Our data elucidate the function of poFUT1/O-fucosylation EDIL3/LIFR axis during placental development, providing glycol-based target for diagnostic and therapeutic of preeclampsia.

Preeclampsia results from abnormal placentation, but the role of protein O-fucosylation is unclear. Here the authors show that poFUT1 loss reduces EDIL3 O-fucosylation, disrupts LIFR–STAT3 signaling, and causes defective placental remodeling and PE-like phenotypes.

Introduction

Preeclampsia (PE) is a multisystem pregnancy-specific complication, characterized by new-onset hypertension and proteinuria after week 20 of gestation^1,2. Preeclampsia affects 3–5% of all pregnancies worldwide, and is a leading cause of maternal and fetal morbidity and mortality³. The disorder of placenta, especially defective trophoblast invasion and disruption of spiral artery remodeling, are widely recognized as the main pathogenetic mechanism for preeclampsia^3,4. The placenta is composed by the chorionic villous, which act as the functional unit. In the villous, villous cytotrophoblast (vCTB) differentiate into syncytiotrophoblast (STB) and extravillous trophoblast (EVT). STBs are responsible for nutrient exchange between maternal and fetal circulations^5,6. EVTs invade the decidua to facilitate embryo implantation, EVTs also move to spiral arteries for transforming the arteries into high-conductance vessels during the first trimester of pregnancy, supporting placental development. Insufficient invasion and arterial transformation by EVTs lead to inadequate placental perfusion, and have been linked to the onset of preeclampsia⁷.

The Implantation of Glycobiology shows that protein glycosylation is closely related to embryo implantation and placental development⁸. Fucosylation is a common modification of glycoproteins. O-fucosyltransferases (poFUTs), which are composed of poFUT1 and poFUT2, are mainly responsible for adding GDP-Fucose directly to serine or threonine residues on proteins. poFUT1 transfers GDP-Fucose to epidermal growth factor-like (EGF-like) repeats containing the consensus sequence C²XXXX(S/T)C³, where C² and C³ are the second and third conserved cysteines of the EGF-like repeat, respectively. poFUT2 transfers GDP-fucose to thrombospondin type 1 repeats (TSRs) with the consensus sequence C¹XX(S/T)C² in group 1 TSRs and C²XX(S/T)C³ in group 2 TSRs⁹. In mice, pofut1 knockout leads to embryonic lethality¹⁰, and decreased pofut2 expression impairs gastrula formation in mouse embryos¹¹. However, the expression pattern and role of poFUT1 in human placental development and preeclampsia still remains elusive.

EGF Like Repeats and Discoidin I Like Domains-Protein 3 (EDIL3), also known as Del-1, is an extracellular matrix glycoprotein. EDIL3 is secreted by various cell types, including endothelial cells, macrophages and neuronal cells, and involved the angiogenesis, immune and anti-inflammatory responses^12–14. Studies have demonstrated that EDIL3 induces angiogenesis and increases neovascularization of ischemic tissues in a mouse model^15,16. In addition, EDIL3 prevent neutrophil adhesion and recruitment to sites of inflammation by interacting with integrin αLβ2 (LFA-1, lymphocyte function-associated antigen 1) on neutrophils¹³. Structurally, EDIL3 protein contains three N-terminal EGF-like domains and two Discoidin domains. Its second EGF-like domain has an evolutionarily conserved RGD motif (Arg-Gly-Asp), and O-fucosylation occurres in the second EGF-like domain¹². However, the expression properties and working mechanism of O-fucosylation on EDIL3 in preeclampsia remain largely unexplored.

In current study, we identified a decreased level of poFUT1 in the plasma and placental tissues of PE patients compared with normal pregnancy (NP) women. By globally screening O-fucosylation glycoproteins on trophoblast cells, we confirmed that EDIL3 has an O-fucosylation site at threonine 88. In addition, poFUT1 facilitates the biosynthesis of O-fucosylation EDIL3, which binds to LIFR, further promotes trophoblast cells invasion and vascular remodeling. In contrast, de-O-fucosylation EDIL3 hampers trophoblast function and contributes to the pathogenesis of PE. Taken together, our findings reveal a pivotal role for poFUT1/O-fucosylation EDIL3 in placental development and the pathogenesis of preeclampsia, and provide a glycol-based target for the diagnosis and treatment of preeclampsia.

Results

Decreased O-fucosyltransferase 1 (poFUT1) expression in preeclamptic placental tissues

We examined the differential expression levels of mRNAs in preeclampsia and normal pregnancy placental tissues derived from the Gene Expression Omnibus (GEO) database (GSE10588 and GSE102897, normal pregnancy = 29, preeclampsia = 20). A heatmap and volcano plot revealed that the expression of poFUT1 was significantly lower in preeclamptic placental tissues than in normal pregnancy placental tissues (Log₂fold change = −0.37, adjusted p value < 0.05) (Fig. 1A, B). Utilizing tissue samples, we further detected and validated the decreased level of poFUT1 in preeclamptic placental samples compared with that in normal pregnancy placental samples by q-PCR, immunofluorescence staining and Western blot. The results showed that poFUT1 expression was significantly decreased in preeclampsia placentas at the mRNA and protein levels (Fig. 1C, G, H). Immunofluorescence staining demonstrated that poFUT1 was expressed mainly in trophoblastic cells and vascular endothelial cells, and lower level of poFUT1 was expressed in placental tissues from preeclamptic patients than in normal control tissues (Fig. 1D–F). These data suggest that lower expression of poFUT1 is associated with the pathology of preeclampsia.

Fig. 1. Decreased poFUT1 expression in preeclamptic placental tissues.

A, B Heatmap and volcano plot analysis of poFUT1 expression in normal pregnancy and preeclamptic placental tissues from GEO database (GSE10588, GSE102897). C poFUT1 mRNA levels in normal pregnancy (n = 18) and preeclamptic (n = 19) placental tissues were detected by q-PCR. Data are presented as mean ± SEM (3 independent biological replicates). Two-tailed unpaired Student’s t test. ****p < 0.0001. D–F Immunofluorescence staining of Cytokeratin 7 (CK-7) and poFUT1 in normal and preeclamptic placental tissues. The bar represents 50 μm. Quantification of intensity of poFUT1 in trophoblast (E) or vascular endothelial cells (F). Data are presented as mean ± SEM (3 biological replicates). Two-tailed unpaired Student’s t test. ***p < 0.001 and *p < 0.05. G, H Expression of poFUT1 proteins in normal pregnancy (n = 5) and preeclamptic placental (n = 5) tissues by Western blot. Data are presented as mean ± SEM (3 independent biological replicates). Two-tailed unpaired Student’s t test. *p < 0.05. Source data are provided as a Source Data file.

Globally screens O-fucosylation proteins in trophoblasts

poFUT1 is an O-fucosyltransferase, which adds O-linked fucose to glycoproteins that contain epidermal growth factor like (EGF-like) repeats (C²XXXX(S/T)C³)⁹ (Fig. 2A). Silencing or overexpressing poFUT1 with specific sh-poFUT1 or poFUT1^WT inhibited or enhanced the O-fucosylation biosynthesis, respectively, in HTR8/SVneo cells (Supplementary Fig. 1). Subsequently, we screened O-fucosylation proteins globally in HTR8/SVneo cells by a chemical probing technique combined with liquid chromatography tandem mass spectrometry (LC-MS/MS) based proteomics. Briefly, alkynyl fucose analogs (6-Alk-fucose) were used for labeling O-fucosylated proteins in HTR8/SVneo cells, and O-fucosylated proteins were extracted with biotin-N3 and streptavidin, followed by analysis with LC-MS/MS (Fig. 2B). Based on the proteomic data, 11 highly abundant proteins with EGF-like repeat sequences were identified (Supplementary Table 1). Gene Ontology (GO) analysis revealed that these proteins were enriched in different biological processes, including cell adhesion, embryonic development, angiogenesis, protein secretion, epithelial cell differentiation, syncytium formation and developmental cell growth (Fig. 2C). Given that angiogenesis and vascular remodeling are considered the main factors involved in placental development¹⁷, we selected three O-fucosylation proteins related to angiogenesis by GO analysis (Supplementary Table 2). Combined with the LC-MS/MS results and the EGF-like repeat modification site, embryonic development and angiogenesis function, EDIL3 emerged as the prime candidate protein for further exploration (Supplementary Tables 2 and 3 and Fig. 2D). EDIL3 contain three EGF-like repeat sequences, and there is a potential O-fucosylation site (threonine 88, Thr88) located in the second EGF-like repeat sequence in different species (Fig. 2E). Furthermore, a comparison of H. sapiens EDIL3 with other vertebrate orthologues revealed high conservation of the Thr88 site by multiple-sequence alignment, implying an evolutionarily role for EDIL3 O-fucosylation modification (Fig. 2E). Moreover, to validate the glycosylation sites in EDIL3, His-tagged full-length EDIL3 was purified from HTR8/SVneo cells and analyzed by electron-transfer/higher-energy collisional dissociation mass spectrometry (EThcD-MS/MS), which identified Thr88 as a site of O-fucosylation (Fig. 2F). In addition, we cloned and purified peptide chains of EGF-like domain 2 wild type (EGF CA-2^WT) and mutation type (EGF CA-2^T88A), incubated them with crude enzyme extract and GDP-Fucose, then detected GDP and GDP-Fucose signals by high-performance liquid chromatography (HPLC). The results showed that the substrate EGF CA-2^WT bound GDP-Fucose to form O-fucose modification, whereas EGF CA-2^T88A group did not utilize GDP-Fucose. The results also indicated that EDIL3 contains O-fucosylation at the Thr88 site (Supplementary Fig. 2A, B).

Fig. 2. Globally screens O-fucosylation proteins in trophoblasts.

A Diagram of poFUT1 catalyze O-fucosylation on Ser/Thr residue of glycoproteins that contains EGF like repeats. B Chemical reporter-based metabolic labeling and click-reaction methods used to analyze the O-fucosylation proteins by LC-MS/MS. Created in BioRender. Yaqi Li. (2026) https://BioRender.com/nz0lnlg. C GO terms enriched for O-fucosylation proteins in HTR8/SVneo cells. D Secondary spectrum of EDIL3. E The comparison of H. sapiens EDIL3 with other vertebrate orthologues. F The site of EDIL3 O-fucosylation was mapped using mass spectrometry. G ELISA analysis of EDIL3 in plasma from the normal pregnancy women and preeclampsia patients. Data are presented as mean ± SEM (n = 20 in normal pregnancy group, n = 20 in preeclampsia group). Two-tailed unpaired Student’s t test. **p < 0.01. H Western blot analysis of EDIL3 in plasma from the normal pregnancy women and preeclampsia patients. I ELISA analysis of poFUT1 in plasma of normal pregnancy and preeclampsia patients. Data are presented as mean ± SEM (n = 18 in normal pregnancy group, n = 20 in preeclampsia group ). Two-tailed unpaired Student’s t test. ***p < 0.001. J Correlation analysis between poFUT1 and EDIL3 expression in the normal pregnancy women (n = 18). A Spearman’s rank-order correlation analysis was performed, revealing a statistically significant positive correlation between the two variables (ρ = 0.67, p < 0.05). Source data are provided as a Source Data file.

To assess the association between EDIL3 expression and preeclampsia, we queried publicly available transcriptomic data from the GEO databases (GSE66273, GSE143953 and GSE144776). However, the data showed that there was no significant difference in EDIL3 expression between normal and preeclamptic placental tissues (Supplementary Fig. 2C). The expression patterns were also confirmed by q-PCR of clinical samples (Supplementary Fig. 2D). Similarly, Western blot results showed that EDIL3 expression was decreased in some preeclamptic placental tissue samples, but there was no significant difference between the normal and preeclampsia groups (Supplementary Fig. 2E). Immunohistochemistry staining further revealed that EDIL3 was expressed in trophoblastic cells and vascular endothelial cells in placental tissues, but there was no distinct difference in EDIL3 expression between normal pregnancy and preeclamptic placental tissues (Supplementary Fig. 2F). Physiologically, EDIL3 functions as a secretory protein. Thus, we detected the plasma level of EDIL3 using ELISA and Western blot assays, and interestingly found that the circulating EDIL3 level was significantly lower in preeclampsia patients than in normal pregnancy women (Fig. 2G, H). Moreover, alterations in the levels of plasma poFUT1 and plasma EDIL3 were positively correlated in normal pregnancy (Fig. 2I, J). Therefore, we hypothesize that O-fucosylation is involved in EDIL3 secretion and its corresponding biological functions.

O-fucosylation at Thr88 strengthens the stability of EDIL3

To achieve a more accurate understanding of O-fucosylation at the Thr88 site of EDIL3, His-tagged EDIL3 cDNA (EDIL3^WT) and His-tagged EDIL3 with a threonine to alanine at Thr88 (EDIL3^T88A) mutation plasmid were constructed and transfected into HTR8/SVneo cells, respectively. Interestingly, despite elevated EDIL3 mRNA did not increase the intracellular EDIL3 protein level in HTR8/SVneo cells, as determined by Western blot (Fig. 3A, B). However, the level of EDIL3 in the culture medium was greater in the EDIL3^WT group than in the vector (pcDNA3.1) group, and the EDIL3^T88A group was significantly lower than EDIL3^WT group according to ELISA and Western blot results (Fig. 3B, C). These findings suggest that O-fucosylation facilitates the secretion of EDIL3 in HTR8/SVneo cells. Subsequently, confocal microscopy observations and Co-IP assays indicated that the O-fucosylation modification of EDIL3 (Fig. 3D, E). To determine the effects of poFUT1 on O-fucosylation synthesis on EDIL3, we also constructed sh-poFUT1, poFUT1 cDNA (poFUT1^WT) and poFUT1 enzymatic active site mutant plasmids with arginine 240 mutated into alanine (poFUT1^R240A). As expected, sh-poFUT1 inhibited O-fucose biosynthesis on EDIL3, and the poFUT1^WT group presented increased O-fucose synthesis on EDIL3, but poFUT1^R240A group abolished enzymatic activity and presented the inhibition effect (Fig. 3F).

Fig. 3. O-fucosylation at Thr88 strengthens the stability of EDIL3.

A q-PCR analysis of EDIL3 mRNA level. Data are presented as mean ± SEM (4 independent biological replicates). Two-tailed unpaired Student’s t test. ****p < 0.0001. B Western blot analysis of protein expression. C ELISA assay analysis of secreted EDIL3 in culture medium of HTR8/SVneo cells. Data are presented as mean ± SEM (3 independent biological replicates). Two-tailed unpaired Student’s t test. *p < 0.05, **p < 0.01. D Immunofluorescence staining detected the location of EDIL3 and O-fucose. The bar represents 25 μm. E, F Co-IP and Western blot analysis of O-fucosylation on EDIL3 in HTR8/SVneo cells. G Molecular dynamic simulation of O-fucosylation EDIL3^WT-4d90 and de-O-fucosylation EDIL3-4d90. Snapshots were shown from the simulations of two structures at 80 ns. H RMSD diagram of the EDIL3^WT-4d90 and de-O-fucosylation EDIL3-4d90 backbone atoms for 100 ns molecular dynamics simulation. I RMSF diagram of EDIL3^WT-4d90 and de-O-fucosylation EDIL3-4d90 backbone atoms. The RMSF for each residue  is shown. J Rg of C-α atoms of the EDIL3^WT-4d90 and de-O-fucosylation EDIL3-4d90 is shown. K The free energy landscape (FEL) characterizes the free energy changes of O-fucosylation and de-O-fucosylation EDIL3-4d90 during the simulation. Source data are provided as a Source Data file.

For function of EDIL3, previous studies have centered mostly on angiogenesis, inflammatory and immune processes^12–15, but little is known about the relationship between O-fucose and its molecular structure. Schürpf et al. proposed that fucose at the Thr88 site may help stabilize the conformation of the EGF domain of EDIL3¹⁸. To test this hypothesis, the structure of the EGF-like repeats of wild-type EDIL3 (EDIL3^WT) was downloaded from the Protein Data Bank (www.rcsb.org) and was coded 4d90. We then used molecular dynamics (MD) simulation methods to study the stability of O-fucosylation EDIL3^WT-4d90 and de-O-fucosylation EDIL3-4d90 (Supplementary Fig. 3A). EGF-like repeats (amino acids: 27 to 155) exhibited distinctly different stabilities, which were caused by differences in the O-fucosylation modification states (Fig. 3G). The root-mean square deviation (RMSD), root-mean-square fluctuation (RMSF), radius of gyration (Rg), covariance matrix and free energy landscape (FEL) were analyzed for O-fucosylation EDIL3^WT-4d90 and de-O-fucosylation EDIL3-4d90. A severe RMSD fluctuation represents motion intensity. The 100 ns MD simulations showed that the time evolutions of the RMSD for EDIL3^WT-4d90 and de-O-fucosylation EDIL3-4d90 are displayed in Fig. 3H. From the trajectories, EDIL3^WT-4d90 achieved equilibrium at 70 ns, whereas de-O-fucosylation EDIL3-4d90 was always in a small fluctuation state. The RMSF of individual amino acid residues of the protein were computed during the entire simulation process to ascertain the flexibility of the protein. The RMSF of EDIL3^WT-4d90 was mostly less than 0.5 nm. However, the RMSF of de-O-fucosylation EDIL3-4d90 was relatively large, ranging from 0.5 to 2.0 nm with significant fluctuations, indicating that the overall structure of de-O-fucosylation EDIL3-4d90 was not as stable as that of EDIL3^WT-4d90 (Fig. 3I). Small Rg values designate a compact structure, whereas higher Rg values indicate a loose structure. EDIL3^WT-4d90 reaches a stable value at 70 ns with a radius of 2.0 nm. However, de-O-fucosylation EDIL3-4d90 exhibited a divergent state within the simulated range, with significant fluctuations, indicating that the de-O-fucosylation EDIL3-4d90 is not as stable as EDIL3^WT-4d90 (Fig. 3J). The free energy landscape (FEL) showed that the structure of EDIL3^WT-4d90 was concentrated in smaller regions, whereas de-O-fucosylation EDIL3-4d90 covered a larger and more dispersed area during the simulation process, which indicating that EDIL3^WT-4d90 was more stable (Fig. 3K). The per-base-averaged covariance matrix showed that the related mobility in de-O-fucosylation EDIL3-4d90 was stronger than that in EDIL3^WT-4d90, which was around 75 and 120 amino acids, respectively (Supplementary Fig. 3B). The above results demonstrated that O-fucose at Thr88, catalyzed by poFUT1, enhances the structural stability of EDIL3.

O-fucosylation influences EDIL3 secretion and degradation

Having shown increased EDIL3 levels in culture medium after transfection with the EDIL3^WT plasmid, but no changes in the EDIL3^T88A group, we speculated that O-fucosylation modification affects EDIL3 secretion ability. To verify our speculation, HTR8/SVneo cells were transfected with sh-poFUT1, poFUT1^WT or poFUT1^R240A plasmids, respectively. The results demonstrated that the intracellular EDIL3 expression level did not differ among the groups. However, sh-poFUT1 reduced, while poFUT1^WT improved the secreted level of EDIL3, but poFUT1^R240A did not increase EDIL3 biosynthesis or the secretion level of EDIL3 in culture medium (Fig. 4A, B). Thus, the above findings indicate that poFUT1 increases EDIL3 secretion via the O-fucosylation.

Fig. 4. O-fucose modification influences EDIL3 secretion and degradation.

A Western blot analysis of intracellular EDIL3 expression. B ELISA assay analysis of secreted EDIL3 levels. Data are presented as mean ± SEM (3 independent biological replicates). Two-tailed unpaired Student’s t test. **p < 0.01, ***p < 0.001, ****p < 0.0001. C The amino acid sequence of EDIL3 contains “KFERQ” like and “RIDKQ” like pentapeptide motifs. D HTR8/SVneo cells were treatment with Leu (10 μM), NH₄Cl (20 mM), or their combination for 12 h, expression of EDIL3 were detected by Western blot. E Western blot analysis of EDIL3 levels in HTR8/SVneo cells. F Western blot analysis of EDIL3 levels in EDIL3^WT and EDIL3^T88A HTR8/SVneo cells treated with Leu and NH₄Cl (N/L) for 12 h. G Lysotracker assay of HTR8/SVneo cell lysosome levels. The bar represents 25 μm. H Transmission electron microscopy showed the autophagosome-like structures in cells. Blue arrows showed the autophagolysosomes containing the cargo. Red arrow showed lysosome. The bar represents 0.5 μm. I Immunofluorescence staining of colocalization of EDIL3 with lysosomes. LAMP1 is used as a lysosomal marker. The bar represents 25 μm. J Co-IP analysis of interaction between endogenous EDIL3 and LAMP2A in HTR8/SVneo cells. K Immunofluorescence staining of the co-localization of EDIL3 and LAMP2A. The bar represents 20 μm. L Co-IP assay confirm the interactions of EDIL3 and HSPA8 in HTR8/SVneo cells. M Immunofluorescence staining analysis the co-localization of EDIL3 and HSPA8. The bar represents 20 μm. N, O q-PCR analysis of LAMP2A and EDIL3 mRNA after knockdown LAMP2A. Data are presented as mean ± SEM (3 independent biological replicates). Two-tailed unpaired Student’s t test. ***p < 0.001, ns no significance. P Western blot analysis of EDIL3 levels after knockdown LAMP2A. Q Western blot analysis of EDIL3 levels after knockdown LAMP2A in EDIL3^WT and EDIL3^T88A HTR8/SVneo cells. R Working model of EDIL3^T88A degradation by CMA. Created in BioRender. Yaqi Li. (2026) https://BioRender.com/1uaf3bl. Source data are provided as a Source Data file.

As mentioned earlier, O-fucose on EDIL3 is involved in maintaining the structural stability of EDIL3. Abnormal proteins are identified and degraded by the ubiquitin proteasome system or the autophagosome lysosomal pathway in cells¹⁹. Amino acid sequence analysis indicated that EDIL3 contains a “KFERQ” like motif (357-361) and a “RIDKQ” like motif (425-429), which are referred to as autophagy-like sequences (Fig. 4C)²⁰. In the chaperone-mediated autophagy (CMA) pathway, chaperone heat shock 70 kDa protein 8 (HSPA8) recognizes specific substrate proteins with autophagy-like motifs (“KFERQ” or “RIDKQ”), and under the synergistic action of lysosome-associated membrane glycoprotein 2 (LAMP2A), abnormal proteins are degraded in lysosomes²¹. Here, HTR8/SVneo cells were treated with leupeptin (Leu), NH₄Cl or a combination of both, and Western blot results showed that the levels of EDIL3 and SQSTM1 elevated after treatment (Fig. 4D, E), which suggested that EDIL3 was degraded by lysosomes. Then, HTR8/SVneo cells were transfected with EDIL3^WT or EDIL3^T88A plasmids, and treated with leupeptin (Leu) and NH₄Cl. Western blot results showed the inhibitory effects of de-O-fucosylation EDIL3 degradation were more obvious than those of O-fucosylation EDIL3 (Fig. 4F). Consistently, lysotracker staining and transmission electron microscopy (TEM) revealed greater autophagolysosome formation in the EDIL3^T88A group than in the EDIL3^WT group (Fig. 4G, H). Immunofluorescence staining also showed that EDIL3 colocalized with lysosomes (Fig. 4I). In the chaperone-mediated autophagy, LAMP2A and HSPA8 act downstream of chaperones, recognize and bind substrate proteins and mediate their recruitment to lysosomes, where target proteins bind LAMP2. Here, we demonstrated that more level of HSPA8 or LAMP2A interacted with de-O-fucosylation EDIL3 than with O-fucosylation EDIL3 in HTR8/SVneo cells by Co-IP and immunofluorescence (Fig. 4J–M). To functionally link LAMP2A to EDIL3 degradation, we used siRNA to knockdown LAMP2A. Notably, there was no significant change in the mRNA level of EDIL3 after LAMP2A siRNA transfection (Fig. 4N, O). But LAMP2A siRNA increased EDIL3 protein levels (Fig. 4P). Moreover, EDIL3^T88A accumulated inside HTR8/SVneo cells in co-treatment with EDIL3^T88A and LAMP2A siRNA group (Fig. 4Q). Taken together, these results support that EDIL3 is a target of CMA, and the degradation of de-O-fucosylation EDIL3 by CMA is highly dependent on LAMP2A and HSPA8 (Fig. 4R).

EDIL3 facilitates invasion, vascular remodeling and proliferation of trophoblasts via O-fucosylation

We carried out a series of experiments to probe the associations between EDIL3 O-fucosylation and trophoblast biological behaviors. The results of the Transwell assay demonstrated that EDIL3^WT promoted HTR8/SVneo cells migration and invasion ability compared with vector control group, while EDIL3^T88A had no effect. Moreover, the increase in migration and invasion potential induced by EDIL3^WT could be partially reversed by sh-poFUT1, implying that O-fucosylation determines the role of EDIL3 in trophoblasts (Fig. 5A). Consistent with the phenotypic changes, Western blot results revealed that EDIL3^WT led to a notable increase in the expression of epithelial mesenchymal transformation (EMT) markers (N-cadherin, Vimentin and Snail), while EDIL3^T88A did not. Similarly, poFUT1 silencing attenuated the EDIL3^WT-induced up-regulation of EMT markers (Fig. 5B). In addition, we evaluated the role of EDIL3 in vascular tube formation in trophoblastic cells. As shown in co-tube formation assay, EDIL3^WT markedly promoted the vascular network formation by HTR/SVneo cells and HUVECs, an effect that was diminished by poFUT1 knockdown (Fig. 5C). Correspondingly, similar changes were observed for vascular endothelial cell markers (CD31, CD34, VE-cadherin, VEGFR2 and VEGFA). This upregulation was not observed with EDIL3^T88A and was partially abolished by poFUT1 inhibition (Fig. 5D). Furthermore, EdU (5-ethynyl-2’-deoxyuridine) and Western blot assays showed that EDIL3^WT enhanced trophoblast proliferation and up-regulated proliferation-related proteins (Cyclin D1, Cyclin E1 and PCNA), but sh-poFUT1 weakened the proliferative ability of trophoblastic cells (Fig. 5E, F). Taken together, these findings indicate that EDIL3 promotes the migration/invasion, vascular remodeling and proliferation of trophoblasts via O-fucosylation modification.

Fig. 5. EDIL3 facilitates EMT, vascular remodeling and proliferation of trophoblast biological behaviors via O-fucosylation modification.

A Transwell assay detected migration and invading ability of HTR8/SVneo cells. B Western blot analysis of the expression of EMT related-proteins expression (E-cadherin, N-cadherin, Vimentin and Snail). C Co-tube formation of HTR8/SVneo cells and HUVECs. BF bright field, IF immunofluorescence field. D Western blot analysis of the expression of angiogenesis-related proteins (CD31, CD34, VE-cadherin, VEGFR2 and VEGFA) expression. E EdU assay evaluated the proliferation potential of HTR8/SVneo cells. F Western blot analysis of the expression of proliferation related proteins expression (Cyclin D1, Cyclin E1 and PCNA). The bar represents 50 μm. Data (A, C, E) are presented as mean ± SEM (3 independent biological replicates). Two-tailed unpaired Student’s t test. **p < 0.01, ***p < 0.001. Source data are provided as a Source Data file.

O-fucosylation mediates the binding of EDIL3 to LIFR on trophoblasts

Secretory proteins exert functions by binding to specific receptors. In current study, we globally screened receptors which recognized by EDIL3, and especially identified the O-fucosylation mediated receptor-ligand binding on trophoblastic cells. First, we expressed and purified the EDIL3^WT proteins and EDIL3^T88A proteins with a His-tag in HTR8/SVneo cells, and confirmed the EDIL3 protein by Western blot analysis (Fig. 6A). Then, recombinant proteins (50 μg/ml) were added to the culture medium of HTR8/SVneo cells for 24 h. Interacting proteins were isolated by a pull-down assay, and the resulting proteins were subjected to LC-MS/MS based proteomic analysis and immunoblotting (Fig. 6B). A total of 434 proteins were identified by LC-MS/MS. Among them, 336 proteins were identified in the recombinant EDIL3^WT (rEDIL3^WT) group, while 295 proteins were identified in recombinant EDIL3^T88A (rEDIL3^T88A) group. rEDIL3^WT and rEDIL3^T88A contain a total of 197 proteins that were commonly matched. There were 139 differentially expressed proteins in the rEDIL3^WT group, and 98 differentially expressed proteins were detected in the rEDIL3^T88A group (Fig. 6C). Subcellular localization of the 237 differential proteins in the two groups included intracellular proteins (61.51%), extracellular proteins (15.47%), multiple proteins (12.23%) and other unassigned proteins (1.08%). Additionally, 9.71% of the differential proteins in our dataset were located in the cell membrane (Supplementary Table 4 and Fig. 6D). Immunofluorescence staining confirmed that the rEDIL3^WT proteins was on the cell menbrane, but markedly reduced in rEDIL3^T88A group (Fig. 6E), which indicating that O-fucose on EDIL3 may mediate the binding of EDIL3 to its receptor on cell membrane. In addition, GO enrichment analysis revealed that these cell membrane protein interactors were involved in various biological processes, including the immune response, signal transduction, and embryonic development (Fig. 6F). Considering the differential membrane protein binding by rEDIL3^WT/rEDIL3^T88A, combined with embryonic development and signal transduction functions, two candidate proteins (LIFR and GNA13) were selected (Fig. 6G). Among these, the receptor of leukemia inhibitor (LIFR), an interleukin (IL)-6 receptor family member, is widely present in embryonic stem cells and is involved in reproduction processing²². However, how O-fucose on proteins influences intermolecular interactions, especially the binding between EDIL3 and LIFR, remains unclear. In our study, immunofluorescence staining and Co-IP demonstrated that the loss of O-fucose on EDIL3 hampered its binding with LIFR (Fig. 6H–J).

Fig. 6. O-fucosylation facilitates the interaction of EDIL3 and LIFR.

A Expression and purification of recombinant EDIL3 proteins. Left panel: Coomassie blue-stained SDS-PAGE analysis showing purification of recombinant EDIL3^WT and EDIL3^T88A protein; Lane 1: raw culture supernatant, Lane 2: Ni⁺-NTA purified EDIL3^WT protein; Lane 3: Ni⁺-NTA purified EDIL3^T88A protein; Middle panel: Western blot analysis of recombinant EDIL3^WT and EDIL3^T88A proteins using anti-EDIL3 antibody. Right panel: Western blot analysis of recombinant EDIL3^WT and EDIL3^T88A protein using anti-His antibody. B Schematic of analysis strategy for screening EDIL3 interacted proteins. C Venn diagrams showing the number of differential interacted proteins on recombinant EDIL3^WT and EDIL3^T88A groups. D Cellular location analysis of the differential proteins. E Immunofluorescence staining showing the location of recombinant protein in HTR8/SVneo cells. The bar represents 25 μm. F GO analysis of the differential cell membrane proteins. G Venn diagrams showing the number of differential proteins involved in embryonic development and signal transduction. H Immunofluorescence staining analysis the location of LIFR and rEDIL3-His proteins. The bar represents 25 μm. I, J Co-IP analyze the interaction of LIFR and rEDIL3^WT/rEDIL3^T88A. K Representative snapshot of the molecular docking of the complex of EDIL3^WT-4d90 and LIFR. LIFR is shown in green, and EDIL3^WT-4d90 is shown in purple, interacting amino acids in yellow and inset showed direct interaction between O-fucose on EDIL3^WT-4d90 with residues of LIFR in R365. Yellow dashed lines indicate hydrogen bonding, and scarlet dashed lines indicated salt bridge interactions. L Representative snapshot from the molecular docking of the complex of de-O-fucosylation EDIL3-4d90 and LIFR. LIFR is shown in green, and de-O-fucosylation EDIL3−4d90 is shown in purple, interacting amino acids in yellow and inset shows direct interaction between de-O-fucosylation EDIL3-4d90 and residues of LIFR. M RMSD diagram of the O-fucosylation and de-O-fucosylation system backbone atoms for 100 ns MD simulation. N Diagram of EDIL3 binding to LIFR/GP130. Created in BioRender. Yaqi Li. (2026) https://BioRender.com/q6llt7d. Source data are provided as a Source Data file.

Next, we employed molecular simulation to study the binding capabilities of O-fucosylation EDIL3/de-O-fucosylation EDIL3 with LIFR. The two systems showed distinctly different binding properties with LIFR, which were caused by different O-fucosylation modification states. According to the trajectories, the major groove structure (amino acids:150 to 360) of LIFR could bind tightly with EDIL3 in the O-fucosylation system but not in the de-O-fucosylation system. The hydrogen bonds (E124, R109, E90, S92, E93 and G97), salt bridge (E90) and the O-fucosylation site (T88) enhanced the interaction of EDIL3 with LIFR. In contrast, the de-O-fucosylated EDIL3 engaged LIFR through fewer hydrogen bonds (R109, E90 and S92) in the de-O-fucosylation system. Consistent with this, the calculated binding was significantly stronger for the O-fucosylation EDIL3 to LIFR complex (-64.6306 kcal/mol) than de-O-fucosylation complex (Fig. 6K, L and Supplementary Fig. 4). The O-fucosylation modification of Thr88 strengthened the intramolecular interaction between EDIL3 and LIFR, and stabilized the spatial relationship. Furthermore, molecular dynamics simulations showed that EDIL3^WT-4d90/LIFR group exhibited more conformational flexibility, which suitable for binding other molecules in the complex, such as GP130 (Fig. 6M). Together, our simulations suggest that the site-specific O-fucosylation leads to strong intermolecular interactions between EDIL3 and LIFR, which promotes the ability of EDIL3 to modulate trophoblast function (Fig. 6N). The molecular docking of another candidate protein, guanine nucleotide-binding protein subunit alpha-13 (GNA13), showed that O-fucose was not directly involved in the interaction between GNA13 and EDIL3 (Supplementary Fig. 5).

The binding of O-fucosylation EDIL3 and LIFR facilitates trophoblast invasion and vascular remodeling in autocrine and paracrine manners

Having confirmed the interaction between O-fucosylation EDIL3 and LIFR, we subsequently explored whether the binding of EDIL3 and LIFR affects trophoblast implantation potential. For trophoblastic tissues, matrigel-imbedded placental explants cultured with the rEDIL3^WT proteins presented extensive trophoblast column outgrowth at 24 h, compared with control group. In contrast, placental explants cultured with the rEDIL3^T88A protein exhibited impaired column outgrowth. Critically, there was a strong inhibitory effect in the rEDIL3^WT and EC359 (100 nM, LIFR inhibitor) cotreated group, suggesting that rEDIL3 stimulated trophoblastic cell outgrowth through LIFR (Fig. 7A). LIFR is known as the major upstream receptor of the STAT3 signaling pathway²³. Here, we further evaluated the activating state of the STAT3 cascade underlying the binding of rEDIL3^WT and rEDIL3^T88A to LIFR in trophoblastic cells. The results showed that p-STAT3 Tyr⁷⁰⁵ was increased in rEDIL3^WT group, whereas the levels p-STAT3 Tyr⁷⁰⁵ were significantly attenuated in both the rEDIL3^T88A and rEDIL3^WT/EC359 groups (Fig. 7B). The rEDIL3^WT protein promoted trophoblast EMT behaviors and tube formation ability, whereas the rEDIL3^T88A and rEDIL3^WT/EC359 groups abolished the effects (Fig. 7C–E). We further evaluated angiogenesis in vivo using a chick chorioallantoic membrane (CAM) assay. Compared with control group, rEDIL3^WT treatment resulted in thicker vascular tubes and a significant increase in terminal vascular branches, whereas rEDIL3^T88A had no effect (Fig. 7F). Furthermore, EDIL3 siRNA and LIFR siRNA were used to knockdown EDIL3 and LIFR, respectively. Silencing EDIL3 and LIFR impaired the potential of trophoblast cell migration, invasion and vascular remodeling (Supplementary Fig. 6). The above results demonstrated that O-fucosylation EDIL3 bound to LIFR and promoted trophoblast implantation and angiogenic potential by activating the STAT3 signaling pathway.

Fig. 7. The binding of O-fucosylation EDIL3 and LIFR facilitates trophoblast invasion and vascular remodeling in autocrine and paracrine manners.

A Representative images of placental villous explants cultured in rEDIL3^WT and rEDIL3^T88A groups. The bar represents 50 μm. B Western blot analysis of STAT3 and p-STAT3 expression, as well as the expression of EMT markers (E-cadherin, N-cadherin, Vimentin and Snail). C Transwell assay analysis of cell invasion and migration ability of HTR8/SVneo cells. D Bright field images of tube formation. E Western blot analysis of angiogenesis-related protein (CD31, CD34, VE-cadherin, VEGFR2 and VEGFA) expression. F CAM assay analysis of vascular formation. Data (A, C, D, F) are presented as mean ± SEM (3 independent biological replicates). Two-tailed unpaired Student’s t test. *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data file.

Given the critical role of trophoblasts in vascular remodeling at the maternal-fetal interface, we subsequently evaluated the paracrine effect of secreted EDIL3 on vascular endothelial cells. HUVECs were treated with the rEDIL3^WT promoted its binding to LIFR on HUVECs, activating the STAT3 signal pathway, and facilitated HUVECs migration, invasion and tube formation. However, rEDIL3^T88A protein and rEDIL3^WT/EC359 inhibited HUVEC tube formation (Supplementary Fig. 7). Thus, O-fucosylated EDIL3 facilitates vascular remodeling through autocrine and paracrine functions.

Loss of O-fucose on EDIL3 in the placenta during early pregnancy induced a PE-like phenotype in vivo

To provide further insights into the contribution of O-fucosylation of EDIL3 to placental development in vivo, pregnant mice were administered Ad-EDIL3^WT, Ad-EDIL3^T88A or Ad-sh-poFUT1 at E8.5. During the gestation period, mouse weight, blood pressure and 24 h proteinuria levels were monitored, and the data were collected every 2 days beginning at E7.5. Mouse plasma, placental tissue, and fetal samples were collected by cesarean section at E18.5 (Fig. 8A). Compared with those in vector and EDIL3^WT groups, EDIL3^T88A and poFUT1 deficiency led to PE-like symptoms in mice, including decreased fetal weight, elevated the systolic blood pressure, and exacerbated proteinuria and soluble FLT-1 (sFLT-1) levels (Fig. 8B–E). The pregnancy outcomes of the EDIL3^T88A and sh-poFUT1 groups also exhibited poor pregnancy outcomes, included placental dysplasia, fetal growth restriction and early embryo resorption, as manifested by decreases in placental weight, crown-rump length (CRL) and fetal weight (Fig. 8F–I).

Fig. 8. Loss of O-fucose on EDIL3 in the placenta during early pregnancy induced a PE-like phenotype in mouse model.

A Schematic of strategy for mouse model in vivo. Pregnancy mice treated with Ad-Vector, Ad-EDIL3^WT, Ad-EDIL3^T88A and Ad-sh-poFUT1. Each group include 10 mice. Created in BioRender. Yaqi Li. (2026) https://BioRender.com/bzc3vca. B Bodyweight of pregnancy mice treated with Ad-Vector, Ad-EDIL3^WT, Ad-EDIL3^T88A and Ad-sh-poFUT1 at E7.5, 9.5, 11.5, 13.5, 15.5 and 17.5. C Systolic blood pressure of pregnancy mice. Data (B, C) are presented as mean ± SEM (n = 10 each group). Two-way ANOVA. ****p < 0.0001, ns: no significance. D 24 h proteinuria of pregnancy mice. E sFLT-1 levels in pregnancy mice determined by ELISA. Data (D, E) are presented as mean ± SEM (n = 9 each group). Two-tailed unpaired Student’s t test. *p < 0.05, **p < 0.01, ****p < 0.0001. F–H Placenta weight, crown-rump length and fetus weight were measured in different groups at E18.5. Data (F–H) are presented as mean ± SEM (n = 20 each group). Two-tailed unpaired Student’s t test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. I Representative images of the uterus and fetuses of the Ad-Vector, Ad-EDIL3^WT, Ad-EDIL3^T88A and Ad-sh-poFUT1 groups. J Skeleton of E18.5 mice stained with alcian blue (cartilage) and alizarin red (bone). K H&E staining of maternal kidneys collected at E18.5. The bar represents 50 μm. L H&E staining of the placental histology at E18.5 in Ad-Vector, Ad-EDIL3^WT, Ad-EDIL3^T88A and Ad-sh-poFUT1 groups. Boxed areas in low magnification images (left panel) are shown at a higher magnification (right panel). Data are presented as mean ± SEM (3 independent biological replicates). Two-tailed unpaired Student’s t test. *p < 0.05. JZ junctional zone, LZ labyrinth zone, TGC trophoblast giant cells, SpT spongiotrophoblast, GlyT glycogen trophoblast, CT cytotrophoblast, MS maternal, sinusoid. The bar represents 50 μm. M Reduced spongiotrophoblasts and glycogen trophoblasts at the junctional zone. Immunohistochemistry staining of TPBPA in placental tissues from Ad-Vector, Ad-EDIL3^WT, Ad-EDIL3^T88A and Ad-sh-poFUT1 injected mice. SpT spongiotrophoblast, GlyT glycogen trophoblast, LT labyrinthine trophoblast. The bar represents 50 μm. N Immunostaining for CD34 in placental tissues from the Ad-Vector, Ad-EDIL3^WT, Ad-EDIL3^T88A and Ad-sh-poFUT1 groups. FC fetal capillaries, MBS maternal blood space, LT labyrinthine trophoblasts, S-TGC sinusoidal trophoblast giant cells. The bar represents 50 μm. O Immunostaining for trophoblasts (CK-7; red) or smooth muscle (α-SMA; magenta) in E18.5 placentas. VSMCs vascular smooth muscle cells. The bar represents 50 μm. Source data are provided as a Source Data file.

EDIL3 restrains osteoclastogenesis and inhibits inflammatory bone loss²⁴. Furthermore, staining of skeletons from fetal mice with alcian blue and alizarin red revealed that the mice injected with EDIL3^T88A showed significant fetal growth restriction due to placental defects, but the endochondral bone differentiation was normal, compared with that of the vector and EDIL3^WT groups. However, the skeletal elements of sh-poFUT1 mice derived from endochondral bone formation were shorter and displayed a straight back, lacking the physiological curvature of the spine. In particular, the small size of the rib cage was obvious. In fetal sh-poFUT1 mice, the cartilage of the posterior occipital region of the skull, limbs, rib cage, humerus and femur bones severely delayed endochondral bone formation, suggesting that decreased O-fucosylation was sufficient to perturb or delay fetal skeletal maturation (Fig. 8J). Maternal kidney histology showed a reduction in the glomerulus open capillary area and narrowing of the balloon gap in EDIL3^T88A and sh-poFUT1 treated mice (Fig. 8K). The structure of the mouse placenta had already formed at E10. The longitudinal section of the maternal facing the fetus can be divided from top to bottom into the decidual layer, trophoblast giant cells (TGCs), spongiotrophoblasts (SpTs) and labyrinth trophoblasts. The equivalent in the mouse to the human villous placenta, where the fetal-maternal exchange occurs, is known as the labyrinth zone (LZ), which has a complex arrangement of maternal and fetal vascular channels that are responsible for exchange. Trophoblast giant cells can secrete various factors to dilate blood vessels and promote placental blood perfusion. Above the labyrinth zone (LZ), there is a junctional zone (JZ) that consists of trophoblast giant cells (TGCs), spongiotrophoblasts (SpTs) and glycogen trophoblasts (GlyTs) that have endocrine functions²⁵. H&E staining of placentae collected at E18.5 revealed that the ratio of the labyrinth zone to the junctional zone was decreased and that the area of the labyrinth placenta was severely attenuated in the EDIL3^T88A and sh-poFUT1 placentas. Placental histological lesions were indicative of labyrinth vascular dysfunction, as evidenced by the presence of nuclear pyknosis, nuclear fragmentation, nuclear dissolution and decreased chromatin basophilia in EDIL3^T88A and sh-poFUT1 compared with vector and EDIL3^WT groups (Fig. 8L). In the mouse placenta, spongiotrophoblasts and glycogen trophoblasts are mainly consisted in the junctional zone (JZ), which expresses trophoblast specific protein alpha (TPBPA)²⁶. Here, immunohistochemistry staining revealed low expression of TPBPA in the placentas of the EDIL3^T88A and sh-poFUT1 groups at E18.5. Moreover, TPBPA expression was detectable in the labyrinth layer of sh-poFUT1, but absent from the junctional zone, demonstrated a reduction in spongiotrophoblasts and/or glycogen trophoblasts in this zone. Taken together, these results indicate that EDIL3 is required for the development of spongiotrophoblasts which marked by TPBPA, particularly at the junctional zone (Fig. 8M). In addition, placental angiogenesis was evaluated. The number of neovessels in EDIL3^T88A and sh-poFUT1-treated mice was significantly decreased, compared with that in the vector and EDIL3^WTgroups (Fig. 8N). Placental vascularization showed trophoblasts invasion and displacement of α-smooth muscle actin (α-SMA) positive vascular smooth muscle cells (VSMCs), indicating normal spiral artery remodeling. The vascular morphology is resembled in  EDIL3^WT and vector groups, however, a thicker α-SMA lining and narrow vessel morphology were observed in EDIL3^T88A and sh-poFUT1 groups at E18.5 (Fig. 8O). Together, these observations indicate that the O-fucosylation EDIL3 is crucial for placental development. Its deficiency maybe one of the factors for preeclampsia development in vivo.

Discussion

Preeclampsia is a pregnancy complication that poses a serious threat to the health of women and infants. Accumulating evidence indicates that preeclampsia progression is accompanied by plasma molecule alternations, and those molecules are used as biomarkers for evaluating the prognosis of preeclampsia, such as the angiogenesis-related molecules sFlt-1, PIGF and sEng, VEGF, and the placental source PAPP-A^27–29. Glycoproteins and glycotransferases are also as potential biomarkers for evaluating the pregnancy related diseases. For example, PSG1 (Pregnancy-Specific beta-1 Glycoprotein 1) and hCG (human Chorionic Gonadotropin) are widely used for detecting and monitoring pregnancy progression in clinic^30,31. Decreases in α1,3-fucosyltransferase 4 (FUT4) and α1,3-fucosyltransferase 7 (FUT7) are associated with miscarriage^32,33. In the present study, we comprehensively evaluated the expression trait of poFUT1 and EDIL3 in tissues and serum from preeclampsia patients and normal pregnancy individuals, and found that decreased plasma levels of poFUT1 and EDIL3 in preeclampsia patients, and the alternation of them present positively correlated. Furthermore, O-fucosylation EDIL3 promoted, while de-O-fucosylation EDIL3 dramatically inhibited the invasion and vascularization potential of trophoblasts, and induced a PE-like phenotype in vivo. These results suggest that poFUT1 and O-fucosylation EDIL3 may serve as serum biomarkers for evaluating placental function and auxiliary diagnoses of preeclampsia.

Accumulated evidence has reinforced that the roles of protein O-fucosylation in embryo implantation and placental development process. In mice, pofut1 knockout leads to embryonic lethality¹⁰, and decreased pofut2 expression impairs gastrula formation in mouse embryos¹¹. Our previous findings demonstrated that poFUT1 not only facilitates endometrial decidualization³⁴, but also promotes trophoblast migration and invasion at the fetal-maternal interface during early pregnancy³⁵. In current study, we further revealed that aberrant decreased poFUT1 in placental tissues of preeclampsia patients. And the decreased O-fucosylation by poFUT1 induced a PE-like phenotype in mouse model. O-fucose plays functions by specific scaffolding proteins. Although many proteins are modified by O-fucose, there is a lack of systematic research for O-fucosylation proteins in trophoblasts. Herein we provided a global screening dataset of O-fucosylation proteins in human placental trophoblastic cells. Among 11 different O-fucosylation modified proteins, EDIL3 was confirmed to contain O-fucose at Thr88. Moreover, we clarified that the O-fucose increases the spatial conformation stability and secretory function of EDIL3 in trophoblasts, then linking trophoblast invasion and tube formation ability. Besides, de-O-fucosylation EDIL3 resulted in molecular instability and degradation via the lysosomal pathway in trophoblasts. Although de-O-fucosylation can lead to the degradation of most EDIL3, there are still a small amount of de-O-fucosylation EDIL3 can still be secreted into the extracellular space. This result suggest that O-fucosylation may not be the sole factor regulating EDIL3 secretory ability, and the complex molecular mechanism needs further study.

EDIL3 is a multi-functional protein mediate many biological processes in tissue microenvironment, including cancer invasion, angiogenesis, and maintaining homeostasis of the immune system^12–16,36. For cell invasion ability, EDIL3 enhances the hepatocellular carcinoma (HCC) cell invasion and spreading behaviors, and promotes capillary tube formation^37,38. Trophoblast motility potential is crucial for embryo implantation and placentation. Here, we displayed that the de-O-fucosylated EDIL3 impaired trophoblast invasion ability in villus tissue culture model and cell line, which also resulted in reduced LZ region in placenta in Ad-EDIL3^T88A mice. For angiogenesis, EDIL3 serves as a potent factor for cardiac remodeling in deoxycorticosterone acetate (DOCA)-salt-induced hypertension, and angiogenesis in ischemic hindlimbs^39,40. EDIL3 has potent angiogenic activity and promotes vessel remodeling in the yolk sac cells^41,42. During placenta development, multiple capillaries break down, and new vasculature remodels to increase vessel volume and match normal fetal development. In this process, trophoblast cells migrate towards the spiral arteries and transform into endovascular trophoblast cells (eEVTs), invade the endothelium of maternal arteries, resulting in the original vessel change into high-conductance vessels. Defective trophoblast invasion is the primary underlying cause of impaired spiral artery remodeling and dysfunctional placentation, and cause many obstetric syndromes, including preeclampsia⁴³. In this study, we revealed that O-fucosylated EDIL3 plays a pro-angiogenic role, which is important for the early stage of placenta formation. De-O-fucosylated EDIL3 impaired vascular remodeling and placenta function, leading to preeclampsia. Recently research also showed that apoptotic VSMCs are internalized and degraded by EDIL3 mediated macrophage efferocytosis in TAD⁴⁴. The process of trophoblast invasion and placental vascular remodeling is accompanied with cell apootosis⁴⁵. Hofbauer cell, a population of tissue-resident macrophages in placental villi, can efficiently clear the debris of apoptotic cells and aid in placenta development⁴⁶. Therefore, we speculate that DEL-1/EDIL3 facilitates placental development through regulating Hofbauer cell efferocytosis of apoptotic cells. However, the specific function of EDIL3 in macrophages during placentation needs further study.

EDIL3, as a ligand, has been reported to bind with the receptor integrins through an RGD motif in the second EGF repeat¹⁸. EDIL3 interacts with αVβ3 integrin and promotes epithelial-mesenchymal transition and paclitaxel resistance in breast and prostate cancer cells¹⁶. It also stimulates osteoblast differentiation through the integrin α5β1/ERK/Runx2 pathway⁴⁷. In our study, we observed that O-fucose affected the structural stability of EDIL3, and we wanted to determine whether alterations in structure influence ligand-receptor interaction. Employing pull-down combined proteomics analysis, we enriched and screened the proteins that bind to the EDIL3^WT and EDIL3^T88A proteins. Intriguingly, in addition to integrins, leukemia inhibitory factor receptor (LIFR) has been shown to interact with EDIL3^WT, but not with EDIL3^T88A. Notably, we also presented evidence that the EDIL3^WT/LIFR complex interacts at a lower binding energy than does EDIL3^T88A. These observations pointed to a crucial role for O-fucosylation in regulating protein interactions. When linked to ligands such as interleukin-6 (IL-6) or leukemia inhibitor factor (LIF), LIFR is activated and subsequently regulates a variety of cellular processes, including cell proliferation, embryonic development and signal transduction^48,49. Studies have shown that lLIF binds to LIFR, and subsequently interacts with GP130 to form a complex, activating the STAT3 signaling cascade to carry out a series of biological activities²³. However, lifr knockout mice had a normal implantation history, but newborns died within 24 h of birth due to impaired placental function⁵⁰. These findings confirmed the critical role of LIFR mediated signaling during placenta development. Our study further demonstrated that reducing the O-fucosylation on EDIL3 decreased its binding potential with LIFR, limited the activation of the STAT3 signaling pathway and inhibited trophoblast invasion and vascular remodeling. These data supply a deep understanding of the EDIL3/LIFR axis during placental development.

In summary, our study uncovers the function of poFUT1 and EDIL3 in placental development. Moreover, poFUT1 increased O-fucosylation EDIL3 biosynthesis, which further interact with LIFR, activate the STAT3 signaling pathway, promote the invasion and tube formation ability of trophoblasts and are involved in placental development (Fig. 9). Furthermore, these findings broaden our understanding of the effect of O-fucose on EDIL3 and suggest that poFUT1/O-fucosylation could be a strategy for the auxiliary diagnosis of preeclampsia.

Fig. 9. The schematic diagram of O-fucosylation EDIL3 in trophoblast invasion and vascular remodeling in preeclampsia.

O-fucosylation EDIL3 interacts with LIFR, and activates STAT3 signal pathway, consequently, promotes trophoblasts invasion and vascularization during placental development both in vitro and in vivo. Aberrant EDIL3 O-fucosylation drives PE development. Created with BioRender. Yaqi Li. (2026) https://BioRender.com/ymh5ghk.

Methods

Tissue collection and ethics statement

Human placental tissues and plasma samples were collected in the Second Affiliated Hospital of Dalian Medical University. Placental tissue samples from normal pregnancy (NP) and preeclampsia (PE) patients were collected from pregnant women who were hospitalized for delivery via cesarean section. PE was diagnosed according to the guidelines of the American College of Obstetrics and Gynecology (ACOG), with a systolic blood pressure to ≥140 mmHg or a diastolic blood pressure to ≥90 mmHg on two separate occasions in a patient who was previously normotensive, as well as proteinuria of ≥300 mg in a 24 h collection, occurring after 20 weeks of pregnancy (Supplementary Table 5). Human placental villus tissue (6–12 weeks of pregnancy) was collected from pregnancy women who underwent artificial abortion in the Department of Obstetrics and Gynecology of the Second Affiliated Hospital of Dalian Medical University. All patients provided written informed consent before sample collection. The procedures were conducted in accordance with the Guidelines of Dalian Medical University and approved by the Ethics Committee of Dalian Medical University (2023-231).

Cell culture and transfection

Human trophoblast cell line HTR8/SVneo and human umbilical vein endothelial cells (HUVECs) were purchased from the American Type Culture Collection (Manassas, VA, USA). HTR8/SVneo cells and HUVECs were cultured in DMEM/F-12 medium (SC103, SEVEN, China) supplemented with 10% FBS and placed in a 37 °C cell incubator with in humidified air containing 5% CO₂. Then, 100 μL of serum-free culture medium was used to dilute 2 μg of DNA to form DNA diluent, and add 4 μL of transfection reagent-PEI (Mine Bio, Shanghai, China) was gently added to 100 μL of the DNA for dilution. The samples were subsequently incubated at room temperature for 10–20 min to form DNA-PEI nucleic acid transfection reagent complexes. When the optimal cell density reached 70–80% for transfection, removed the culture medium and added 2 mL freshly preheated complete culture medium to each well, and the samples were placed in an incubator for 6–8 h.

q-PCR

Total RNA was isolated from transfected HTR8/SVneo cells using Trizol Reagent (T9108, Takara, Tokyo, Japan) according to the manufacturer’s instructions. One microgram of total RNA was used for reverse transcription to generate template cDNA (Evo M-MLV RT Kit, AG11601, AGBio, China). Quantitative real-time PCR reactions were performed using SYBR Green (Evo M-MLV RT Kit, AG11701, AGBio, China). Primer of poFUT1 (5’-3’) were as follows: Forward: CAGCGAAGCCCAGATAAGAAG; Reverse: CTGTAGGAAGCTCTGAAGGAA. Primer of EDIL3 (5’-3’): Forward: AGCATACCGAGGGGATACATT; Reverse: CAAGGCTCAACTTCGCATTCA. Primer of GAPDH (5’-3’): Forward: GGAGTCCACTGGCGTCTTCAC; Reverse: GCTGATGATCTTGAGGCTGTTGTC (Sangon Biotech, Shanghai, China).

Western blot assay

The proteins from cells or placental tissue were collected for quantification (P0012S, Beyotime, China). Equal amounts of proteins were separated by 10% SDS-PAGE, followed by electrophoretic transfer to nitrocellulose membranes. The membranes were incubated with blocking buffer containing 5% BSA (CAS:9048-46-8, Solarbio, China) and probed with the following primary antibodies at 4 °C overnight: poFUT1 (Cat No.14929-1-AP, Proteintech, 1:1000), EDIL3 (Cat No.12580-1-AP, Proteintech, 1:1000), His (Cat No.66005-1-Ig, Proteintech, 1:1000), CD31 (Cat No.11265-1-AP, Proteintech, 1:1000), CD34 (Cat No.BS45128, Bioworld, 1:1000), E-cadherin (Cat No.20874-1-AP, Proteintech, 1:1000), N-cadherin (Cat No.22018-1-AP, Proteintech, 1:1000), Vimentin (Cat No.10366-1-AP, Proteintech, 1:2000), Snail (Cat No.13099-1-AP, Proteintech, 1:1000), VE-cadherin (Cat No.sc-515467, SantaCruz, 1:500), VEGFR2(Cat No. 26415-1-AP, Proteintech, 1:1000), VEGFA (Cat No.sc-7269, SantaCruz, 1:500), PCNA (Cat No.AF0261, Beyotime, 1:1000), Cyclin D1 (Cat No.AF0126, Beyotime, 1:1000), Cyclin E1 (Cat No.AF6384, Beyotime, 1:1000), SQSTM1 (Cat No.18420-1-AP, Proteintech, 1:1000), LAMP2A (Cat No.66301-1-Ig, Proteintech, 1:1000), HSPA8 (Cat No.66442-1-Ig, Proteintech, 1:1000), ACTB (Cat No.20536-1-AP, Proteintech, 1:3000), p-STAT3^Tyr705(Cat No.AP0705, Abclonal, 1:1000), STAT3 (Cat No.A19566, Abclonal, 1:1000), LIFR (Cat No.22779-1-AP, Proteintech, 1:1000) and GAPDH (Cat No.10494-1-AP, Proteintech, 1:1000). An incubation with an IgG-HRP conjugated secondary antibody at room temperature for 1 h was used for detection system. Second antibodies used were HRP Goat Anti-Rabbit IgG (Cat No.AS014, Abclonal, 1:1000) and HRP Goat Anti-Mouse IgG (Cat No.AS003, Abclonal, 1:1000), Streptavidin-HRP (Cat No.A0303, Beyotime, 1:3000).

Immunofluorescence

Cell climbing films were fixed with 4% paraformaldehyde for 20 min followed by blocking with goat serum for 30 min at room temperature. The cells were then incubated with primary antibodies in PBS at 4 °C overnight, followed by incubation with the corresponding secondary antibody in the dark at room temperature for 40 min. The primary antibodies were used: poFUT1 (Cat No.14929-1-AP, Proteintech, 1:100), Cytokeratin 7 (Cat No.15539-1-AP, Proteintech, 1:100), EDIL3 (Cat No.12580-1-AP, Proteintech, 1:100), His (Cat No.66005-1-Ig, Proteintech, 1:100), LAMP1 (Cat No. 21997-1-AP, Proteintech, 1:100), LAMP2A (Cat No. 66301-1-Ig, Proteintech, 1:100), HSPA8 (Cat No.66442-1-Ig, Proteintech, 1:100), LIFR (Cat No.22779-1-AP, Proteintech, 1:100), CD34 (Cat No.BS45128, Bioworld, 1:100), α-SMA (Cat No.67735-1-Ig, Proteintech, 1:100). The secondary antibody as follows: FITC-conjugated Goat anti-Rabbit IgG (H + L) (Cat No.AS011, Abclonal, 1:100), ABflo® 647-conjugated Goat anti-Mouse IgG (H + L) (Cat No.AS059, Abclonal, 1:100), Cy3-conjugated Goat anti-Mouse IgG (H + L) (Cat No.AS008, Abclonal, 1:100), PE-conjugated Streptavidin (Cat No.D110513, BBI, 1:100). The cells were subsequently washed with PBS three times. The nuclei were counterstained with DAPI (C1002, 1:4000, Beyotime, China) before visualization and image acquisition under a microscope (Olympus, BX83, Japan).

Transwell assay

HTR8/SVneo cells were adjusted to 5 × 10⁴ cells/well, and the cells were resuspended in serum-free medium and seeded into the upper chamber, which was coated with matrigel (REF:354262, Corning, NY, USA) at a ratio of 1:9 with serum-free medium DMEM/F-12. DMEM/F-12 medium containing 10% FBS was added to the lower chamber, which was then placed in a 37 °C incubator. After 24 h, non-invading cells on the upper surface of the membrane were removed with cotton-tipped swabs, and the chambers were fixed with methanol for 15 min and stained with 0.1% crystal violet dye for 15 min. The number of cells was determined by imaging with a microscope (Olympus, BX83, Japan).

Co-tube formation assay

Matrigel was used to assess the formation of capillary-like structures. HTR8/SVneo cells were labeled with CellTracker Green dye (Yeasen, Shanghai, CN) and cultured for 1 h in an incubator. After being washed three times with PBS, the cells were collected and counted. HTR8/SVneo cells (1 × 10⁴) and HUVECs cells (1 × 10⁴) (without labels) were mixed and cultured in 96-well plates. After 2 h of incubation, the tube-forming structures were photographed and analyzed under a microscope (Olympus, BX83, Japan).

Immunoprecipitation (IP)

According to the instructions for Dynabeads Protein A/G (B23201, Selleck, Houston, TX, USA), Protein A/G beads were added and incubated with antibody (anti-EDIL3, anti-His or anti-LIFR) for 30 min at room temperature, followed by washing three times. Subsequently, 1000 μg of total protein lysate was immunoprecipitated by incubation with beads at 4 °C overnight, followed by washing buffer and centrifugation at 1 × 10⁴ g at 4 °C. The beads were subsequently resuspended in SDS-PAGE sample buffer and subjected to SDS-PAGE and Western blotting. The antibodies used for IP were EDIL3 (Cat No.12580-1-AP, Proteintech, 1:50), His (Cat No.66005-1-Ig, Proteintech, 1:50), LIFR (Cat No.22779-1-AP, Proteintech, 1:50).

Enzyme-linked immunosorbent assay (ELISA)

Human poFUT1 ELISA Kit (Langton, China) was used to determine plasma poFUT1. Human EDIL3 ELISA Kit (Langton, China) was used to determine secretory EDIL3 levels in plasma and conditioned medium according to the manufacturer’s instructions. Mouse sFTL-1 ELISA Kit (Langton, China) was used to determine sFTL-1 levels in mouse plasma. The concentrations were calculated by the ELISACalc system.

Culture of villus explant

Matrigel was diluted with serum-free medium DMEM/F-12 at a ratio of 1:2, and 50 μL of matrigel was added to each well of a 96-well plate. Villi tissue was cut into 2–5 mm segments and carefully anchored in the center of the matrigel. Transfection, observation and imaging by microscope were performed, and the exogenic migration distance of EVTs was measured with Photoshop software.

Animal experiments

Seven-week-old KM female mice from the Experimental Animal Center of Dalian Medical University were mated with age-matched male mice. Mice were housed in specific pathogen-free (SPF) barrier facility under controlled conditions (20–25 °C, 50% humidity, 12 h light/dark cycle), and ad libitum access to food and water. Forty pregnant female mice were randomly divided into four groups: Ad-Vector pregnant mouse group (n = 10), Ad-EDIL3^WT pregnant mouse group (n = 10), Ad-EDIL3^T88A pregnant mouse group (n = 10), and Ad-sh-poFUT1 pregnant mouse group (n = 10). Adenoviral-carrying mouse EDIL3^WT overexpressing plasmid, mouse EDIL3^T88A overexpressing plasmid and shRNA-targeting mouse poFUT1 were purchased from Applied Biological Materials Inc. (British Columbia, Canada). Upon observation of vaginal plug, the day of mating was considered Day 0.5. Body weight and blood pressure were measured on Day 7.5. Systolic blood pressure (SBP) was monitored by mouse noninvasive blood pressure monitor BP-2010A Series (Softron Biotechnology, Beijing, China) daily from E7.5 to E18.5. Ad-green fluorescent protein (GFP)-Vector, Ad-GFP-EDIL3^WT, Ad-GFP-EDIL3^T88A and Ad-GFP-sh-poFUT1 (1 × 10⁹ PFU, 100 µL) were injected into the tail vein at E8.5. The mice were delivered by cesarean section on Day 18.5 (E18.5) to collect placentas and the weights of fetuses were examined. Animal models were approved by the Institute Animal Care and Use Committee of Dalian Medical University (Liaoning, China) (No. AEE20069), and the experiments were carried out in accordance with established institutional guidelines and approved protocols.

H&E staining

Placenta and kidney paraffin sections were de-paraffinized, rehydrated, and then the nuclei were stained with hematoxylin for 5 min and then with eosin cytoplasm for 2 min. The samples were dehydrated, sealed, and subjected to image acquisition and analysis.

Alcian blue and alizarin red staining

The mice were fixed with 95% alcohol for 1 week, degreased with acetone for 3 days, purified with 15% hydrogen peroxide for 2 days, and then stained with alcian blue and alizarin red. Finally, the sections were transparent in gradient glycerin (25%, 50%, 75%).

Molecular dynamic (MD) simulation

MD simulation was performed for EDIL3 (4d90) with a molecular force field by Gromacs. A cubic TIP3P water box was used to enclose structure models with a 10 Å buffering in all three orthogonal dimensions. The system was neutralized with an appropriate amount of sodium ions by replacing water molecules. First, to maintain the protein structure close to the native conformation, the heavy atoms of proteins (and small molecules) are constrained, and perform 5000 steps of energy minimization on are performed on water molecules. To maintain the constraints, a 50,000 step NVT ensemble simulation was subsequently performed on the entire system, with a temperature of 298.15 K and a time step was set to 2 fs. Maintaining a 50,000 step NPT ensemble simulation was maintained for the entire system, with a temperature of 298.15 K and a time step of 2 fs. Finally, a molecular dynamics simulation of 100 ns was performed on the system under NPT ensemble simulation, with a time step of 2 fs and trajectory data saved every 10 ps interval. All relevant data parameters were analyzed using the related software package’s built-in module.

Protein expression and purification

The CD sequence of human EDIL3 was obtained and ligated into the vector, together with an N-terminal signal peptide and a C-terminal 6×His tag. HTR8/SVneo cells were transfected with His-tagged EDIL3^WT or His-tagged EDIL3^T88A (GenePharma, Shanghai, China), and culture supernatant from mock-transfected cells was used as a control. His-tagged EDIL3^WT and EDIL3^T88A were purified from cell culture supernatant using high-affinity Ni-NTA resin (L00250, GeneScript) in a gravity column equilibrated with binding buffer (20 mM sodium phosphate, 0.5 M sodium chloride, 20 mM imidazole, pH 7.4), followed by elution buffer (20 mM sodium phosphate, 0.5 M sodium chloride, 500 mM imidazole, pH 7.4). The protein mixture was collected, and concentrated, and then detected by SDS-PAGE and Western blot with anti-EDIL3 and anti-His antibodies.

O-fucose assay

6-Alkyne-fucose was added to the cells for metabolic labeling for 48 h. Then, the cells were lysed in lysis buffer (1% SDS, 50 mM Tris-HCl, pH 8.0) for 30 min and sonicated on ice. The expression level of O-fucose was detected according to the instructions of the Click-iT Protein Reaction Buffer Kit (C10276, Invitrogen, Carlsbad, CA, USA). Click reactions were performed to collect O-fucosylation proteins, which were analyzed by Western blot.

LC–MS/MS analysis

O-fucosylation proteins were collected and treated with 1000 µL 10 mM DTT for 40 min at 56 °C and subsequently alkylated with 1000 µL 50 mM IAM for 30 min in the dark. The gel was then washed with destaining buffer and treated with ACN as described above. A total of 100 μL with 100 mM TEAB, 1 μg trypsin and 1/300 CaCl₂ were added, and the mixture was digested overnight at 37 °C. After centrifugation in low speed, the supernatant was collected, 200 µL of acetonitrile was added, put under vortex and mixed. And the supernatant was extracted in 100 μL of 0.1% formic acid (FA). Combined the supernatant and centrifuged at 12,000 × g for 5 min at room temperature and then lyophilized. The powder was dissolved and mixed in 0.1% formic acid. MS data acquisition was performed on Q ExactiveTM series mass spectrometer (Thermo Fisher Scientific, USA). The proteomes were analyzed by Scale Biomedicine Technology Co., LTD (Beijing, China).

Glycoproteomic site mapping of O-fucosylated peptides

To identify O-fucosylation sites, His-tagged human EDIL3 was expressed in HTR8/SVneo cells and purified using the His-tag. The purified EDIL3 samples were reduced with 10 mM DTT in water at 65 °C for 15 min and alkylated with 20 mM iodoacetamide in water at 35 °C in the dark for 30 min. Following a buffer exchange with 200 μL of 2 M urea in PBS, the protein was digested with Glu-C (Promega; 0.5 μg/μL) in the presence of 2 μL of 100 mM CaCl₂ at 37 °C for 16 h. The resulting peptides were analyzed using a nano-LC system coupled to an Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher Scientific). Peptides were loaded onto an in-house packed analytical column (10 cm length, 3 μm C18 resin; Dr. Maisch GmbH, Germany) with mobile phase A (0.1% formic acid). Separation was performed using a gradient of mobile phase B (ACN containing 0.1% formic acid) at a flow rate of 300 nL/min: 0–35% B for 30 min, 35–80% B for 10 min, followed by a hold at 80% B for 10 min. The mass spectrometer was operated in data-dependent acquisition (DDA) mode. MS1 spectra were collected in the Orbitrap with a top-speed method and a dynamic exclusion time of 20 s after two repeated activation events. HCD fragmentation was performed with a normalized collision energy (NCE) of 28%. EThcD spectra were acquired in the high-pressure linear ion trap with an optimized reaction time of 50 ms for ETD (2 × 10⁵ reagent AGC) and 25% supplemental activation energy. The identification and quantification of O-fucosylated peptides were performed using MaxQuant software integrated with the Andromeda search engine (version 1.5.3.28). Glu-C was selected as the proteolytic enzyme with a maximum of two missed cleavages allowed. Peptides were identified with ≥95% confidence and filtered at a 1% false discovery rate (FDR). The mass spectrometry proteomics data have been deposited to the iProX repository with the dataset identifier PXD072416.

CAM assay

Fertilized eggs were purchased from Boehringer Ingelheim Vital Biotechnology (Beijing, China). The eggs were incubated in an incubator at 37.8 °C with 70–75% relative humidity. For the first 3 days, eggs were left undisturbed without turning. From day 4 onward, they were turned manually 3 times daily. On embryonic day 8, eggs were checked to assess vascular development and mark the position of the air sac. A small window was then carefully cut in the eggshell over the air sac, and a silicone ring was placed on the surfaces of the growing CAMs. Recombinant proteins, inhibitor or human placental villi pre-treated with EDIL3 siRNA or poFUT1 siRNA were added to the CAM. The window was sealed with preservative film, and the egg was incubated for another 2 h. After 2 h of incubation, the area around the silicone ring was photographed, and the number of vascular density was analyzed.

HPLC detection

The consumption of the substrate GDP-fucose and the production of the GDP product were detected by HPLC to determine whether EDIL3 could be modified by O-fucosylation at Thr88. EDIL3 CA-2^WT and EDIL3 CA-2^T88A were used as substrates, and 5 µg of cell lysate as the enzyme source was mixed. The mixture was incubated at 37 °C for 10 h. The reaction was terminated by heating the mixture in a boiling water bath for 10 min. Following centrifugation at 10,000 × g, the supernatant was filtered through a 0.22 μm membrane. The separation of GDP and GDP-Fucose was achieved on an Elite C-18 column using a mobile phase of 2 mM Tetra-n-butylammonium hydrogen sulfate phosphate buffer (pH 6.2). The analysis was performed at 26 °C with 1 mL/min flow rate, achieving baseline separation within 15 min.

Statistics and reproducibility

ImageJ 1.52a (National Institutes of Health) and GraphPad Prism (GraphPad Software 6.0.1) were used to process and analyze images and data. All data are shown as the mean ± SEM (standard error of mean). Analysis of different groups was performed by unpaired two-tailed Student’s t test and two-way ANOVA. Statistical significance is indicated as the follows: *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001. The experiments were repeated at least three times independently with similar results.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Shuai Liu, Yubo Liu, and Yaqi Li designed the study. Yaqi Li, H.W., Yuyu Lei, S.B., and J.W. conducted experiments. Yaqi Li, Yuyu Lei, and H.W. analyzed the data. Shuai Liu, Yaqi Li, and B.Y. edited and revised the manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

The raw mass spectral data in our study are available via iProX with identifier PXD072416. The raw LC-MS/MS data is available via iProX with identifier IPX0011377000 (https://www.iprox.cn/page/PCV010.html). All data supporting the findings described in this manuscript are available in the article and in the Supplementary Information and from the corresponding author upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-73139-4.