Identifying fibroblast-derived sFRP2 as a therapeutic target and engineering siRNA therapy for uterine scarring.

An ionizable cationic lipid (DLin-MC3-DMA), distearoylphosphatidylcholine (DSPC), and cholesterol were purchased from Macklin (China). 1,2-Dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMG-PEG) was purchased from Suzhou Highfine Biotech Co., Ltd. RIPA lysis buffer and BCA protein assay kit were purchased from Wuhan Servicebio Technology Co. The companies and catalog numbers of different antibodies and recombinant proteins are as followed: sFRP2 antibody (Proteintech, 66328-1-lg), VEGFA antibody (Proteintech, 19003-1-AP), α-SMA antibody (Proteintech, 14395-1-AP), fibrinogen antibody (Proteintech, 20645-1-AP), Col1A1 antibody (Proteintech, 14695-1-AP), Col3A1 antibody (Proteintech, 22734-1-AP), vimentin antibody (Proteintech, 10366-1-AP), WNT5A antibody (Affinity, DF6856), Ki67 antibody (Proteintech, 27309-1-AP), cytokeratin 7 antibody (CK7, Proteintech, 15539-1-AP), Flag tag antibody (Proteintech, 66008-4-lg), HA tag antibody (ABclonal, AE105), FZD5 antibody (ABclonal, A12775), myeloperoxidase (MPO) polyclonal antibody (Proteintech, 22225-1-AP), F4/80 antibody (Servicebio, GB113373 ), CD3 antibody (Servicebio, GB150004 ), GAPDH antibody (Proteintech, 60004-1-lg), β-actin antibody (Proteintech, 66009-1-lg), FITC-conjugated affinipure goat anti-Rabbit IgG (Proteintech, SA00003-2), Cy3-conjugated affinipure goat anti-Mouse IgG (Proteintech, SA00009-1), recombinant human TGF-β protein (Proteintech, HZ1011), recombinant human sFRP2 protein (rhsFRP2, MedChemExpress, HY- P77835 ), recombinant human WNT5A protein (R&D, 645-WN). DMEM/F12K and DMEM were obtained from Gibco (U.S.A.). Collagenase type I, collagenase type II, collagenase type IV, and rat tail type I collagen were obtained from Solarbio (China). Fluo-4 AM was purchased from Beyotime (China). All experiments relevant to human myometrial tissues were approved by Human Research Ethics Committee of Chongqing Medical University (No. 20172601; Chongqing, China). Written informed consents were obtained from all participants before tissue collection. Female patients aged 20-35 years with singleton pregnancy, at least one previous cesarean delivery and diagnosed with uterine scarring (US) were enrolled in our study as the US group. Primiparas aged 20-35 years with term pregnancies scheduled for elective cesarean section were included in the normal group. Participants were excluded if they had pregnancy complications such as diabetes mellitus, cardiovascular disease, infectious disease, autoimmune disease, chronic renal disease, chronic hypertension, and metabolic diseases. Detailed clinical characteristics of all human participants are provided in Supplementary Table  1 and source data files. Myometrial tissues were collected during laparotomy after delivery of the infant but before the administration of oxytocin and removal of the placenta. A 0.5 × 2.0 cm 2 sample was excised with surgical scissors from the myometrium at the uterine scar site (the US group) or uterine incision site (the normal group) at the First Affiliated Hospital of Chongqing Medical University. After washing to remove blood, some collected samples were immediately frozen in liquid nitrogen and stored at -80 °C for biological analysis, while other samples were preserved in 4% formalin or optimal cutting temperature compound (OCT) for histological sections. Three myometrial samples from human uteruses in the US and normal groups were separately utilized for transcriptome analysis. The collected tissue was thoroughly washed to eliminate blood residues and preserved in RNAlater (LC Sciences). Total RNA was extracted from each sample using Trizol reagent (Invitrogen, U.S.A.) following the manufacturer’s instructions. The concentration of total RNA was measured using a NanoDrop ND-1000 spectrophotometer (NanoDrop, U.S.A.). The RNA integrity was assessed using an Agilent Bioanalyzer 2100 (Agilent Technologies, U.S.A.). RNA-sequencing (RNA-seq) and subsequent data analysis were conducted by LC-Bio Technology Co. Ltd. (Zhejiang, China). The RNA-seq data are available from the corresponding authors upon reasonable request. The raw sequencing reads have been deposited in the NCBI Gene Expression Omnibus (GEO) database under accession number GSE293910 . The myometrial tissues were lysed using RIPA lysis buffer (Servicebio, Wuhan, China). A BCA protein assay kit (Servicebio, Wuhan, China) was used to determine the protein content. Proteins were denatured in loading buffer (Bio-Rad, U.S.A.), separated on sodium dodecyl sulfate-polyacrylamide gels, and subsequently transferred onto PVDF membranes (Roche, Germany). After blocking with 5% bovine serum albumin (BSA) in TBST (TBS with 0.1% Tween-20), the membranes were incubated with primary antibodies (sFRP2 1:1000, VEGFA 1:1000, α-SMA 1:1000, fibrinogen 1:1000, Col1A1 1:1000, Col3A1 1:1000, GAPDH 1:1000, β-Actin 1:1000) overnight at 4 °C. After washing with TBST, secondary antibodies were added and incubated at room temperature for 1 h. The bands were detected by enhanced chemiluminescence using a ChemiDoc Touch Imaging System (Bio-Rad, U.S.A.). The signal intensity was calculated using ImageJ. Protein expression was normalized to β-actin or GAPDH. Total RNA was extracted from the myometrial tissue using Trizol according to the manufacturer’s protocol (Invitrogen, U.S.A.) and quantified using NanoDrop (Thermo Fisher Scientific, U.S.A.). RNA was converted to cDNA using RT Master for PCR (MedChemExpress, U.S.A.). Quantitative real-time PCR (qPCR) was performed using SYBR Green Mater Mix (Bimake, U.S.A.) with a Bio-Rad CFX-96 system (Bio-Rad, California, U.S.A.). Results were normalized to GAPDH expression and analyzed using the 2 -(ΔΔCT) method. The primers used for human tissues are as follows: sFRP2 (forward, ACCGAGGAAGCTCCAAAGGTA; reverse, GAGCCACAGCCACCGATTTCT), VEGFA (forward; CCCAGTTTTGGGAACACCGA; reverse, CCCCAAAGCACAGCAATGTC), α-SMA (forward, TGGCTTGTCAGGGCTTGTC; reverse, TTGATGCGAAGTGCTGACCC), fibrinogen (forward, GAGGGCCAAGACGAAGACATC; reverse, CAGATCACGTCATCGCACAAC), Col1A1 (forward, AGTGATTCAGAACCGTCAAGAC; reverse, CATCCTGGTAAGCTGGCTAATTT), Col3A1 (forward, GGAGCTGGCTACTTCTCGC; reverse, GGGAACATCCTCCTTCAACAG), GAPDH (forward, GGAGTCCACTGGCGTCTTCA; reverse, GTCATGAGTCCTTCCACGATACC). Human myometrial tissues obtained from each group were fixed in 4% formalin and embedded in paraffin. Paraffin sections (5-μm thick) were cut, deparaffinized, and rehydrated following standard procedures. Subsequently, the sections were stained with hematoxylin-eosin (H&E) or Masson’s trichrome. For immunohistochemistry analysis, the rehydrated paraffin sections were subjected to citric acid retrieval and incubated in 3% H 2 O 2 . The sections were incubated with primary antibodies (sFRP2 1:500, VEGFA 1:500, α-SMA 1:500), followed by incubation with HRP-conjugated secondary antibody (PV9000, ZSGB-Bios, China). Visualization was achieved using diaminobenzidine (ZLI-9019, ZSGB-Bios, China) as a substrate and counterstaining with hematoxylin. Microscopic images were acquired using Slideview (Olympus VS200). Immunofluorescence analysis was carried out on frozen sections (7 μm) using primary antibodies of sFRP2, fibrinogen, Col1A1, and Col3A1, followed by incubation with FITC- or Cy3-labeled secondary antibodies and nuclear staining with 4’,6-diamidino-2-phenylindole dihydrochloride (DAPI, Servicebio, Wuhan, China). Images were captured by confocal microscopy (Carl Zeiss AG, LSM900) with laser intensity at 0.2%, master gain at 300 V, and digital gain at 1.0. The images were processed with ZEN software (Zeiss, Germany). Myometrial tissues were excised with a surgical scissor during laparotomy and immediately placed in saline containing 100 U/mL penicillin and 100 mg/mL streptomycin. To isolated hUSMCs, the tissue was cut into 5-10 mm 3 pieces and digested with DMEM containing 0.1% collagenase type I (Solarbio, China) and 20% fetal bovine serum (FBS, Gibco, U.S.A.). After 1 h of digestion to remove endometrial and remaining perimetrical tissues, the tissue was incubated in fresh enzyme solution at 37 °C. Following rinsing and cell counting, the isolated cells were seeded into 60 mm plastic dishes and cultured in DMEM with 20% FBS. The cultures were maintained at 37 °C in a humidified atmosphere of 5% CO 2 in air, and the medium was changed every second day. Subsequent experiments were conducted using cells between their third and sixth passages. The hUSMCs were identified by immunofluorescence staining of α-smooth muscle actin (α-SMA). To isolate hUFIBs, the tissue samples were enzymatically dissociated overnight at 37 °C using a digestion solution composed of 1 mL of DMEM/F12K containing the following components: 0.2% collagenase type IV (Solarbio, China), 0.2 mg/mL elastase (Roche, Basel, Switzerland), 0.5 mg/mL trypsin inhibitor, 2.0 mg/mL BSA, and 0.1 mg/mL DNaseI (Sigma). The enzymatic digestion was terminated by adding 2 mL of DMEM/F12K. The cell suspension was then centrifuged for 5 min at 300 ×  g and 4 °C. The resulting pellet was resuspended in DMEM/F12K. Subsequently, the cell suspension was passed through a 40-μm metal mesh once, and the cells were seeded into 60 mm plastic dishes. hUFIBs were confirmed to be positive for vimentin antibody. The endometrial tissue was dissected into 2-5 mm fragments and digested with 0.25% trypsin at room temperature for 1 h. Following removal of the undigested tissue, the remaining material was incubated in a solution containing 1 mg/mL collagenase type I for 2 h. The reaction was stopped by the addition of DMEM/F12K supplemented with 10% FBS and 1% penicillin-streptomycin. Subsequently, the digested cell suspension was centrifuged at 200 ×  g for 5 min and seeded at the density of 1 × 10 5 cells/cm 2 in 6-well plates and incubated at 37 °C in a humidified atmosphere with 5% CO 2 . The culture medium was refreshed every 2-3 days, and the cells were passaged at a ratio of 1:3-1:4. The isolated hUECs mainly consist of endometrial epithelial cells (EECs) and endometrial stromal cells (ESCs). hUECs were confirmed through immunofluorescence staining of CK7 for EECs and vimentin for ESCs. The levels of sFRP2 in hUSMCs, hUFIBs, and hUECs from the normal and US groups were determined by Western blot (WB) and qPCR. Briefly, the cells were seeded in 60 mm plastic dishes and grown for 3 days. RIPA or Trizol was added to extract protein or RNA. The detailed procedures for WB and qPCR were as described above. Additionally, hUSMCs, hUFIBs, and hUECs from the normal and US groups were seeded in 12-well plates overnight, fixed in 4% paraformaldehyde, and stained with sFRP2 primary antibody and FITC-conjugated secondary antibody for immunofluorescence analysis. Further experiments were conducted for hUFIBs. The levels of sFRP2, VEGFA, α-SMA, fibrinogen, Col1A1, and Col3A1 in hUFIBs from the normal and US groups were determined by WB and qPCR as described above. Furthermore, hUFIBs from the normal and US groups were seeded in 12-well plates at a density of 2 × 10 5 cells/well, fixed in 4% paraformaldehyde for 20 min, washed with PBS three times, incubated with 0.5% Triton-100 at room temperature for 20 min, and blocked with 5% BSA for 20 min. Antibodies against sFRP2, α-SMA, Col1A1, Col3A1, or vimentin were added. Then fluorescent-labeled secondary antibodies were added following the instructions of the Mix Fluorescence Immunohistochemical Mouse/Rabbit Kit (AiFang Biological, China). Nuclei were counter-stained with DAPI for 5 min after PBS washing. Images were captured using confocal microscopy (Carl Zeiss AG, LSM900) with laser intensity at 0.2%, master gain at 200 V, and digital gain at 1.0. The images were processed with ZEN software (Zeiss Germany). To evaluate collagen contraction, 4 × 10 4 hUFIBs (400 μL) from the normal and US groups were mixed with 200 μL of rat tail type I collagen matrix (3 mg/mL, Solarbio, China), which was then poured into a well of a 24-well plate. After solidification at 37 °C for 30 min, the cells were cultured in DMEM supplemented with 10% FBS. The diameter changes of the collagen gels were captured using a Canoscan 9000 F at 0 and 24 h. The measurements (in pixels) were analyzed using ImageJ, and the gel area at 24 h was reported as a percentage relative to the gel area at the 0 h time point. For migration assay, hUFIBs (1 × 10 5 cells per well) from both the normal and US groups were added to the upper chamber of each Transwell (8 μm), while fresh medium was added to the bottom compartment. After 3 days of culture at 37 °C, the cells that crossed through the chamber membrane were fixed and stained with 0.5% crystal violet before microscopic quantification (EVOS FL Auto Imaging System, Invitrogen, U.S.A.). All animal experiments were approved by the Animal Ethics Committee at Chongqing Medical University (No. 2018-109; Chongqing, China). Female C57BL/6 J mice (5-6 weeks, 15-20 g) were obtained from the Laboratory Animal Center of Chongqing Medical University and were housed at 25 °C with a 12-hour light-dark cycle and controlled humidity (30%-70%). The uterine surgery was performed as describe previously 9 . Briefly, mice were anesthetized with isoflurane, and a low abdominal incision was made to expose the uterine horns. A longitudinal incision (1 cm) at the midportion of the uterine horn was made using ophthalmic scissors. The incision penetrated through the myometrium and endometrium on the surgical side. It was closed with 8-0 absorbable sutures, and the ends of the incision edge were labeled with 8-0 nonabsorbable sutures. The surgery was performed on the left uterine horn. The uterus was placed back into the abdominal cavity, and the abdominal skin was closed using 3-0 nonabsorbable sutures. The control animals underwent low abdominal incisions but no uterine incisions. After 30 days post-surgery, the mice were sacrificed, and uterine tissues (1 cm in length) from the scarring site (Model) or the control group (Control) were collected for further experiments. Three samples from uterine scar sites of animals in the control and model groups were used for transcriptome analysis. For sampling at day 30 post-uterine surgery, the uterine samples at the scarring sites were obtained after removal of the remained blood. RNA extraction, quantification, sequencing, and data analysis were conducted following procedures similar to those used for the human myometrium. The RNA-seq data have been deposited in the NCBI GEO database under accession number GSE293910 . For in vivo sFRP2 knockdown (sFRP2KD) or overexpression (sFRP2OE), lentiviruses containing specific short hairpin RNA (shRNA) or the sFRP2-overexpressing fragment for mice were designed and synthesized by Hanbio (Shanghai, China). The negative control lentivectors for overexpression or the scrambled shRNA contained in the lentivirus used as negative controls for knockdown were also provided by Hanbio. These modified lentivectors for mice (10 6 units/per) combined with polybrene (100 ng) were separately transfected into the uterine horn of mice using a syringe. The uterine tissues were collected from mice at days 0, 3, 5, 7, and 14 post-transfection to evaluate the expression level of sFRP2 using WB and qPCR. In separate studies, uterine tissues were collected from mice with sFRP2OE induced by lentivirus for 7 days or from the control mice (i.e., sFRP2OE and Control) for various analyses. Furthermore, sFRP2OE mice and control mice were underwent uterine scar surgery (i.e., sFRP2OE+Model and Model). After 30 days, uterine tissues were collected from mice. Similarly, uterine tissues were obtained from mice with sFRP2KD induced by shRNA for 7 days or from the scrambled shRNA group, which were also subjected to uterine scar surgery (i.e., sFRP2KD+Model and Model). The uterine scar tissues from each group (Model and Control, sFRP2OE and Control, sFRP2OE+Model and Model, sFRP2KD+Model and Model) were lysed in cold RIPA buffer for WB analysis or Trizol for qPCR. WB analysis was conducted to determine the protein expression levels of sFRP2, VEGFA, α-SMA, fibrinogen, Col1A1, and Col3A1. The mRNA levels of sFRP2 (forward, AAACCAAGAATGAGGACGACAAC; reverse, TTGGTGTCTCTGTTGATGTACGT), VEGFA (forward, AAACGAACGTACTTGCAGATGTG; reverse, TCTTCCTTCATGTCAGGCTTTCT), α-SMA (forward, CCAGCTATGTGTGAAGAGGAAGA; reverse, TTGGTGATGATGCCGTGTTCTAT), fibrinogen (forward, GACGTTGCAGAGCTATCCATTTC; reverse, AGTGAATGAGTTGGCGGTGATAT), Col1A1 (forward, CCCGAGGTATGCTTGATCTGTAT; reverse, TCCCTCGACTCCTACATCTTCTG), Col3A1 (forward, CACCCTTCTTCATCCCACTCTTA; reverse, GTGGCTCCTCATCACAGATTATG), and GAPDH (forward, GAAGGTCGGTGTGAACGGAT; reverse, CCCATTTGATGTTAGCGGGAT) were quantified by qPCR. Tissues collected from the surgical site of the mouse uterus in each group (Model and Control, sFRP2OE and Control, sFRP2OE+Model and Model, sFRP2KD+Model and Model) were fixed in 4% paraformaldehyde or OCT compound for paraffin sections (5 μm) and frozen sections (7 μm), respectively. The paraffin sections were subjected to H&E staining, Masson’s trichrome staining, or immunohistochemistry staining of sFRP2, VEGFA, and α-SMA. The frozen sections were used for immunofluorescence staining. In this case, FITC-labeled secondary antibody stained with sFRP2 primary antibody and Cy3-labeled secondary antibody stained with primary antibodies of fibrinogen, Col1A1, and Col3A1 were developed to detect their expression and co-distribution. Fluorescence images were acquired by confocal microscopy with laser intensity at 0.2%, master gain at 300 V, and digital gain at 1.0. The mice in the sFRP2OE group and its control group (sFRP2OE and Control) were sacrificed at day 7 after lentivirus intervention. The mice in other groups (Model and Control, sFRP2OE+Model and Model, sFRP2KD+Model and Model) were sacrificed at day 30 after establishing uterine surgical scars. Uterine tissues (1 cm in length) from the scarring site or its control group were mixed with type II collagenase (0.1%) to digest the tissue. For culture experiments, mUFIBs were plated in DMEM with high glucose and supplemented with 1% penicillin and streptomycin, 20% FBS, and kept at 37 °C with 5% CO 2 . Fibroblasts were expanded to passages 3-5 for experimentation. The mUFIBs were identified by vimentin staining. mUFIBs isolated from each group (Model and Control, sFRP2OE and Control, sFRP2OE+Model and Model, sFRP2KD+Model and Model) were cultured in 60 mm plastic plates and lysed in cold RIPA buffer or Trizol for WB and qPCR analyses to detect the expression levels of sFRP2, VEGFA, α-SMA, fibrinogen, Col1A1, and Col3A1. For immunofluorescence analysis by confocal microscopy imaging (laser intensity at 0.2%, master gain at 200 V, and digital gain at 1.0), mUFIBs isolated from different groups were cultured in 12-well plates at a density of 2 × 10 5 cells/well, which were subjected to staining of sFRP2, α-SMA, Col1A1, Col3A1, and vimentin using the Mix Fluorescence immunohistochemical mouse/rabbit kit (AiFang Biological, China). mUFIBs were isolated from different groups (Model versus Control, sFRP2OE versus Control, sFRP2OE+Model versus Model, sFRP2KD+Model versus Model). For the collagen gel contraction and migration assay, the cell density was 4 × 10 4 mUFIBs/well and 1 × 10 5 mUFIBs/well, respectively. The methods for the collagen contraction assay and migration assay were described as above. Lentiviruses for sFRP2 knockdown (sFRP2KD) or overexpression (sFRP2OE) in hUFIBs were designed and synthesized by Hanbio (Shanghai, China). hUFIBs from the normal group were seeded into 6-well plates at a density of 1 × 10 6 cells per well and transfected with these modified lentivectors at a multiplicity of infection (MOI) of 4 in the presence of 6 μg/mL polybrene (Hanbio, China). After transfection for 24 h, the lentiviruses were removed and replaced with a supplemented medium containing 2.5 µg/mL puromycin. After 72 h of culture with puromycin, fluorescence was examined under a fluorescence microscope to determine the transfection efficiency of hUFIBs. For sFRP2 overexpression studies, a negative control lentivirus was used as the control group (Control), and PBS was used as a wild-type (WT) group. For sFRP2 knockdown studies, a scramble shRNA contained in lentivirus was used as a negative control (Control) and PBS was used as a WT group. In sFRP2 knockdown studies, each group was treated with recombinant human TGF-β (10 ng/mL, Proteintech) for 3 days to induce fibroblast activation and scarring after knocking down sFRP2. hUFIBs in different groups (WT, Control, and sFRP2OE; WT + TGF-β, Control+TGF-β, and sFRP2KD+TGF-β) were cultured in 60 mm plastic plates for WB and qPCR analyses. In addition, hUFIBs (2 × 10 5 cells/well) in different groups were cultured in 12-well plates for immunofluorescence analysis of sFRP2, fibrinogen, Col1A1, or Col3A1. Moreover, hUFIBs at 4 × 10 4 cells/well were used for collagen contraction assay, while hUFIBs at 1 × 10 5 cells/well were employed for migration assay. The procedures described above were used for WB, qPCR, immunofluorescence, collagen contraction, and migration studies. To assess the paracrine role of sFRP2 in mediating fibroblast scarring, the conditioned media (CM) from hUFIBs (2 × 10 6 cells) in the WT, control, and sFRP2OE, as well as WT, control, and sFRP2KD groups were collected. Their sFRP2 concentrations were quantified by ELISA according to the manufacture’s instruction (Jiangsu Meimian Industrial Co., Ltd). Meanwhile, hUFIBs from the normal group (10 6 cells/well) were treated with CM from the WT, sFRP2OE, and sFRP2KD groups, respectively. In a separate experiment, the CM from the sFRP2KD group was supplemented with rhsFRP2 (10 ng/mL), which was used to treat normal hUFIBs. After 3 days of incubation, fibrosis markers (fibrinogen, Col1A1, Col3A1) were evaluated by qPCR. Moreover, hUFIBs (10 5 cells) in different CM-treated or rhsFRP2-treated groups were seeded onto coverslips and stained with fibrinogen, Col1A1, and Col3A1 using the Mix Fluorescence immunohistochemical mouse/rabbit kit (AiFang Biological, China). Cells were imaged by confocal microscopy with laser intensity at 0.2%, master gain at 200 V, and digital gain at 1.0. To determine the expression levels of canonical pathway markers (β-catenin, Myc, cyclin D1) and non-canonical pathway markers (PKC, RhoA, JNK) in the human myometrial tissue (Normal and US) and hUFIBs in different groups (Normal and US; WT, Control, and sFRP2OE; WT + TGF-β, Control+TGF-β, sFRP2KD+TGF-β). Tissues and cells were lysed in Trizol and reversed to cDNA to detect the mRNA expression by qPCR analyses. The primers of β-catenin (forward, TTTGCCCAGCAAATCATGCG; reverse, TTGTGAACATCCCGAGCGAG), Myc (forward, CCCTCCACTCGGAAGGACTA; reverse, CGTTGTGTGTTCGCCTCTTG), cyclin D1 (forward, GAGGCGGATGAGAACAAGCA; reverse, GCAGTCCGGGTCACACTTG), PKC (forward, GGGTCACTGCTCTATGGACTTAT; reverse, TTCATCAGCAACCTCAGCCTTTA), RhoA (forward, CTTCGGAATGATGAGCACACAAG; reverse, GCCAATCCTGTTTGCCATATCTC), and JNK (forward, GCACCCTGAAGATCCTTGACTTT; reverse, ATGCAACCCACTGACCAGATATC) were used. The mRNA expression of WNT16, WNT5A, and WNT9A in human myometrial tissues (Normal and US) and mouse uteruses (Control and model) were measured by qPCR. Then, the expression profiles of WNT5A in human myometrial tissues, hUFIBs, mouse uterine tissues, and mUFIBs from different groups (Model and Control; sFRP2OE and Control; sFRP2OE+Model and Model; sFRP2KD+Model and Model) were measured by WB and qPCR analyses. The primers for human WNT16 (forward, AGTATGGCATGTGGTTCAGCA; reverse, GCGGCAGTCTACTGACATCAA), WNT5A (forward, GCCAAGGAGTTCGTGGACGC; reverse, CAGCACATGAGCTCGCAGCC), and WNT9A (forward, CCACCGTGAGAAGAACTGC; reverse, GCCTGCACTCCACATAGCA), as well as for mouse WNT16 (forward, CAGGGCAACTGGATGTGGTT; reverse, CTCGTGTCGGAACTGGCTTC), WNT5A (forward, CTGGCTCCTGTAGCCTCAAG; reverse, GCCGCGCTATCATACTTCTC), and WNT9A (forward, ATGCTGGATGGGTCCCTTCT; reverse, ACTGCCTGTTAGCCCGAAGTA) were used. Frozen sections of human myometrial tissues and mouse uterine tissues in different groups were stained with primary antibodies against sFRP2 and WNT5A to show their colocalization. hUFIBs (normal and US) and mUFIBs (Model and Control; sFRP2OE and Control; sFRP2OE+Model and Model; sFRP2KD+Model and Model) from each group were stained with WNT5A to analyze its expression profiles. Also, sFRP2OE hUFIBs or TGF-β-induced sFRP2KD hUFIBs were lysed with RIPA or Trizol to detect the expression of WNT5A. These cells from different groups were also stained with antibodies against sFRP2, WNT5A, and fibrinogen for immunofluorescence analyses using the Mix Fluorescence immunohistochemical mouse/rabbit kit. Fluorescence images were acquired following similar procedures as described above. For fluorescence colocalization analysis, multichannel images (sFRP2/WNT5A or sFRP2/WNT5A/fibrinogen) were individually threshold-processed, and colocalization was quantitatively determined as the pixel overlap between fluorescence channels using ImageJ software. In separate studies, hUFIBs (2 × 10 6 cells) from the WT, Control, and sFRP2OE groups were seeded in 60 mm plates. In a parallel experiment, hUFIBs in each group (WT, Control, and sFRP2OE) were treated with heparinases (0.1 U/mL, Sigma-Aldrich) for 12 h, followed by removal of heparinases. After 3 days of incubation, the CM from each group was collected after centrifugation at 300 × g to discard the cell debris. The concentrations of secreted WNT5A in the CM of each group (WT, Control, sFRP2OE; WT+Heparinase, Control+Heparinase, sFRP2OE+Heparinase) were quantified by ELISA (Jiangsu Meimian Industrial Co., Ltd). Additionally, hUFIBs in each group were collected and lysed in Trizol to detect the mRNA expression of PKC, RhoA, JNK by qPCR. Moreover, hUFIBs (2 × 10 6 ) from the normal group were cultured with recombinant human WNT5A protein (100 ng/mL) for 24 h and then seeded into 6-well plates for analysis of fibrinogen, Col1A1, and Col3A1 expressions by WB. Alternatively, they were seeded into 12-well plates to determine the expression of fibrinogen through immunofluorescence by confocal microscopy (laser intensity at 0.2%, master gain voltage at 200 V, and digital gain at 1.0). The control group was treated with the same volume of PBS. To examine the sFRP2-WNT5A interaction, normal hUFIBs were co-transfected with HA-tagged WNT5A (HA-WNT5A) and FLAG-tagged sFRP2 (FLAG-sFRP2) expression plasmids using Lipofectamine 3000 (Invitrogen). The plasmids were constructed by cloning full-length cDNA of WNT5A into pcDNA3.1-HA (HA-WNT5A) or full-length cDNA of sFRP2 into pcDNA3.1-FLAG vectors (FLAG-sFRP2, Tsingke Biotechnology Co., Ltd). Briefly, hUFIBs were seeded in 6-well plates at 5 × 10 4 cells/well and transfected at 70% confluence with either 2 μg of individual plasmids or both plasmids combined. Normal hUFIBs treated with cell culture medium alone were used as a control. Following 48 h of incubation, cells were collected after lysed in ice-cold lysis buffer for 30 min. After centrifugation at 12,000 × g for 15 min at 4 °C, 20 μL aliquots of supernatant were reserved as input controls for protein quantification. For immunoprecipitation assay, 500 μg of total protein from cell lysis in each group (Control, HA-WNT5A-transfected cells, FLAG-sFRP2-transfected cells, or HA-WNT5A/FLAG-sFRP2-cotransfected cells) was incubated with 20 μL of anti-HA magnetic beads (MedChemExpress, HY-K0237) or anti-FLAG magnetic beads (MedChemExpress, HY-K0207) at 4 °C with gentle rotation overnight. Beads were washed 3 times with ice cold RIPA buffer and eluted with 50 μL of 4× SDS loading buffer by boiling at 95 °C for 10 min. The immunoprecipitated proteins were separated by 10% SDS-PAGE and transferred to PVDF membranes. Membranes were blocked with 5% non-fat dry milk in TBST for 1 h and probed with primary antibodies against HA (ABclonal, 1:1000) or FLAG (Proteintech, 1:1000) overnight at 4 °C, followed by incubation with HRP-conjugated secondary antibody for 1 h at room temperature. Protein bands were visualized and detected with a chemiluminescent imaging system (Bio-Rad). hUFIBs isolated form human myometrial tissues, mUFIBs isolated from different mouse uterine tissues (Model and Control; sFRP2OE and Control; sFRP2OE+Model and Model; sFRP2KD+Model and Model), hUFIBs with sFRP2OE or sFRP2KD (WT, Control, and sFRP2OE; WT + TGF-β, Control+TGF-β, and sFRP2KD+TGF-β) were seeded at 2 × 10 5 cells in a 24-well plate and incubated overnight, followed by incubation with 2.5 μM Fluo-4 AM (a fluorescent probe for Ca 2+ , Beyotime, China) for 30 min. In addition, hUFIBs (2 × 10 6 cells/well) from different groups (WT, Control, and sFRP2OE; WT + TGF-β, Control+TGF-β, and sFRP2KD+TGF-β) were harvested to detect the expression of FZD5 by qPCR (forward, CACCAACGCTCTGAGACATAAGA; reverse, TATGGTGGCATGTGTCTGTAGTC). For immunofluorescence analysis, hUFIBs (normal and US) or mUFIBs (Model and Control; sFRP2OE and Control; sFRP2OE+Model and Model; sFRP2KD+Model and Model) were washed 3 times and fixed with 4% paraformaldehyde for 10 min after incubation with Fluo-4 AM. Also, hUFIBs with sFRP2OE or sFRP2KD (WT, Control, and sFRP2OE; WT + TGF-β, Control+TGF-β, and sFRP2KD+TGF-β) were stained with sFRP2 primary antibody and subsequently stained with fibrinogen primary antibody or FZD5 primary antibody (ABclonal, 1:500) after staining with Fluo-4 AM. Nuclei were counter-stained with DAPI for 5 min after PBS washing and observed by confocal microscopy (laser intensity at 0.2%, master gain voltage at 200 V, and digital gain at 1.0). Colocalization analysis of sFRP2, FZD5, and Ca 2+ in fluorescence images were analyzed using ImageJ 1.8.0. For flow cytometric analysis, UFIBs in each group were seeded in a 6-well plate overnight, stained with Fluo-4 AM (2.5 μM) for 30 min, and resuspended of 200 μL of PBS after pancreatic enzyme digestion, which were then quantified via flow cytometry (FACS AriaIII Flowmeter, BD Biosciences, U.S.A.). In a separate study, 2 × 10 6 cells of PBS-treated hUFIBs (WT), hUFIBs treated with scrambled shRNA (Control), and hUFIBs with sFRP2KD were stained with Fluo-4 AM (2.5 μM) for 30 min and collected using 200 μL of PBS. For each sample, baseline calcium levels were recorded for 50 seconds, followed by the addition of 10 ng/mL of TGF-β. The cells were continuously recorded for another 190 seconds to observe changes in the fluorescence intensity corresponding to the Ca 2+ influx. hUFIBs were treated with either CaCl 2 at 6 mM to induce Ca 2+ influx or PBS as the control. After 8 h of incubation, the solution in each well was removed, and the cells from different groups were collected for the detection of fibrinogen, Col1A1, and Col3A1 levels by WB. In addition, the fibrinogen expression was examined by immunofluorescence analysis via confocal microscopy imaging (laser intensity at 0.2%, master gain voltage at 200 V, and digital gain at 1.0). hUFIBs isolated from the normal human myometrium of uterine tissues and mUFIBs isolated from the normal mouse uterine tissues were separately seeded into a 24-well plate. Upon reaching 40-60% confluence, UFIBs were transfected with 40 pmol siRNA against sFRP2 (Genepharma Biotechnology, Shanghai, China), which is a pool of 3 target-specific siRNAs designed to knockdown sFRP2 gene expression. For hUFIBs, three sequences, i.e., 5′-CAUCAACCGAGAUACCAAA-3′ (sisFRP2-h64231), 5′-GGUAUGUGAAGCCUGCAAA-3′ (sisFRP2-h64232), and 5′-GCACCUGUGAGGAGAUGAA-3′ (sisFRP2-h64233) were used. For mUFIBs, the utilized three sequences are as follows: 5′-GCAUCGAGUACCAGAACAUTT-3′ (sisFRP2-374), 5′-GACAACGACAUCAUGGAAATT-3′ (sisFRP2-763), and 5′-GCAAGACCAUUUACAAGCUTT-3′ (sisFRP2-872). The negative control siRNA for humans and mice contains a scrambled sequence that does not specifically degrade any known cellular mRNA. PBS-treated UFIBs were used in the corresponding WT groups. After 72 h of transfection using Lipofectamine 3000 (Invitrogen, U.S.A.) following the manufacturer’s protocol, hUFIBs and mUFIBs in each group were harvested for WB and qPCR analyses. sisFRP2-h64231 for humans and sisFRP2-872 for mice were finally used for following therapeutic studies, by loading into ionizable lipid nanoparticles to afford LNP/sisFRP2. In an ethanol phase with a total volume of 112.5 µL, the ionizable lipid (DLin-MC3-DMA) was combined with DSPC, cholesterol, and DMG-PEG at a molar ratio of 50:10:38.5:1.5. A separate aqueous phase was prepared with siRNA (sisFRP2-h64231 or sisFRP2-872) in 10 mM citrate buffer (pH 3) to a total volume of 337.5 µL. Using a syringe pump, the ethanol and aqueous phases were combined to form LNP/siRNA via chaotic mixing using a microfluidic device designed with herringbone features as previously described 68 . LNPs were dialyzed against PBS for 2 h, sterilized using a 0.22 µm filter, and stored at 4 °C for further experiments. Empty LNPs were synthesized without siRNA. The condensation efficiency of LNP/sisFRP2 at different weight ratios was assessed using agarose gel electrophoresis (AGE), in combination with staining by 6× loading buffer containing GelRed (30 mM EDTA, 0.05% bromophenol blue, 0.05% xylene cyanol, and 36% glycerol). Naked siRNA solution (40 ng/μL) served as the negative control. Samples, including naked siRNA, empty LNPs, and LNP/sisFRP2, were loaded onto a 3.0% agarose gel in 1× TAE running buffer and subjected to electrophoresis at 120 V for 40 min. The siRNA bands were visualized under UV illumination using the Tanon 2500 Gel Image System. LNP/sisFRP2 formulated with the DLin-MC3-DMA:siRNA weight ratio of 15:1 was selected for further experiments. The size and size distribution of LNPs and LNP/sisFRP2 in aqueous solutions were measured using a Malvern Zetasizer NanoZS instrument (Malvern, U.K.). All measurements were conducted in triplicate at room temperature. Additionally, LNPs and LNP/sisFRP2 were negatively stained and observed by transmission electron microscopy (TEM) at 200 kV. The encapsulation efficiency of sisFRP2 in LNP/sisFRP2 was determined by Quant-iT-RiboGreen assay (Thermo Fisher Scientific, U.S.A.). Specifically, the LNP/sisFRP2 sample was diluted to approximately 2 ng/μL in two microcentrifuge tubes containing either 1× tris-EDTA (TE) buffer or TE buffer supplemented with 0.1% (v/v) Triton X-100 and incubated for 20 min to achieve lysis of LNPs. Subsequently, LNP/sisFRP2 in TE buffer, LNP/sisFRP2 in Triton X-100, and siRNA standards were plated in triplicate in black 96-well plates, into which the RiboGreen fluorescent detection reagent was added according to the manufacturer’s instructions. After 5 min, the fluorescence intensity was measured using a plate reader (Tecan), with an excitation wavelength of 490 nm and an emission wavelength of 520 nm. The RNA content was estimated by comparison to a standard curve. The encapsulation efficiency was calculated as (Cb-Ca)/Cb×100, where Ca is the content of RNA in TE buffer (free/unencapsulated RNA) and Cb is the content of RNA in Triton X-100 (total RNA). hUFIBs collected from the normal human myometrial tissue were used to examine biological effects of LNP/sisFRP2-h. In brief, hUFIBs were treated with either empty LNPs (LNP) or various doses of LNPs containing sisFRP2-h64231 (LNP/sisFRP2-h) at 50 and 150 nM of sisFRP2-h64231 for 3 days. Cells were then activated with 10 ng/mL TGF-β for additional 3 days to induce scarring and evaluate the effects of various LNPs. hUFIBs stimulated with TGF-β for 3 days without treatment with LNP or LNP/sisFRP2-h served as the model group. Cells in the control group were treated with PBS instead of TGF-β. In a separate study, hUFIBs in different groups (Control, Model, LNP, LNP/sisFRP2-h at 50 nM, LNP/sisFRP2-h at 150 nM) were cultured in 60 mm plastic plates and collected to detect the levels of sFRP2, WNT5A, VEGFA, α-SMA, fibrinogen, Col1A1, and Col3A1 expressions through WB and qPCR analyses. In addition, hUFIBs (2 × 10 5 cells/well) in each group were cultured in 12-well plates for immunofluorescence analysis to visualize the expression and co-distribution of sFRP2 with fibrinogen, Col1A1, and Col3A1 (imaging parameters: laser intensity at 0.2%, master gain voltage at 200 V, and digital gain at 1.0). Furthermore, hUFIBs in each group (4 × 10 4 cells/well) were utilized for collagen contraction assay to evaluate scar formation. Transwell assay (1 × 10 5 cells/well) was conducted for hUFIBs in each group to assess cell migration ability. For flow cytometric analysis, hUFIBs in different groups (Control, Model, LNP, LNP/sisFRP2-h at 50 nM, LNP/sisFRP2-h at 150 nM) were seeded in a 6-well plate overnight. They were then stained with Fluo-4 AM (2.5 μM) for 30 min, followed by resuspension in 200 μL of PBS after pancreatic enzyme digestion. The cells were quantified using flow cytometry to assess the levels of intracellular Ca 2+ . To estimate the increased Ca 2+ levels in each group, 2 × 10 6 cells of hUFIBs from each group were stained with Fluo-4 AM (2.5 μM) for 30 min and collected in 200 μL of PBS. Baseline calcium levels were recorded for 50 s, followed by continuous recording for another 190 s after treating with 10 ng/mL TGF-β. The flow cytometry data were analyzed using FlowJo software. To evaluate prophylactic efficacy of LNP/sisFRP2, three doses (0.125, 0.25, and 0.5 mg/kg sisFRP2) were administered by directly applying saline containing LNP/sisFRP2 to uterine scar sites of female mice. Control mice underwent a low abdominal incision without uterine injury and received saline treatment, while model mice were subjected to surgically induced uterine scars, followed by saline application. All mice were euthanized at day 30 post-intervention, and uterine scar tissues were collected for analysis. One portion of the tissue was homogenized in Trizol for qPCR quantification of sFRP2, fibrinogen, Col1A1, and Col3A1 mRNA expression. The remaining tissue was fixed in 4% formalin for histological assessment via H&E staining and immunohistochemical detection of sFRP2. Following uterine scar surgery as previously described, mice were randomly assigned to different groups. The model group received saline application to the uterine scar site. The LNP group was treated with saline containing empty LNPs applied to the scar site, while the LNP/sisFRP2 group received saline containing LNP/sisFRP2 (0.5 mg/kg sisFRP2) applied to the scar site. Mice in the control group underwent abdominal incision without uterine injury and received saline treatment. After uterine surgery and intervention for 30 days, female mice in each group (Control, Model, LNP, LNP/sisFRP2) were mated to male C57BL/6 J mice (6-8 weeks, 20-25 g). The day on which a virginal plug appeared was recorded as embryonic day 0.5 (E0.5). Some pregnant mice were sacrificed at embryonic day 4 (E4) to obtain uterine tissues for sectioning and staining with Ki67 antibody (1:500) to assess the status of endometrial receptivity. Embryos with the enclosed uteruses from different groups were collected at E7.5 and subjected to H&E staining to visualize the implantation site. The angle (θ) between the long axis of the embryo (uterine-embryonic axis) and the mesometrial-antimesometrial (M-AM) axis was measured, and embryos with an aberrant uterine-embryonic axis (>10° angle relative to the M-AM axis) were considered to have disorientation of embryonic implantation. The uteruses from different groups were also weighted at E7.5. At E12.5, abdominal ultrasound was performed on mice from each group to observe the status of the embryos. Embryos in each group were also collected for gross morphology analysis to determine absorption rates. In addition, abdominal ultrasound was used to monitor fetal growth on E17.5. Fetal and placental tissues at E17.5 were collected to evaluate fetal development (i.e., the litter size, weight, and crown-rump length (CRL) and placental development (i.e., the weight, spongiotrophoblast layer/labyrinth ratio S/L, and maximum invasion distance) in each group. In addition, the pregnancy rate, the number of pups per litter (including the absorbed, live, and dead fetuses), and the morphology of fetuses in each group at birth were detected to analyze pregnancy outcomes. To evaluate potential immune responses to LNP/sisFRP2 treatment, the uteruses of female C57BL/6 mice (6 weeks old) were surgically exposed and treated with either LNP/sisFRP2 (0.5 mg/kg sisFRP2 in saline) or saline alone, applied topically to the normal uterine surface before abdominal closure. Plasma samples were collected at 24 h and day 30 post-treatment to quantify TNF-α and IL-6 levels via ELISA (Jiangsu Meimian Industrial Co., Ltd). Uterine and splenic tissues from each group were weighed to calculate organ index values and subsequently fixed in 4% formalin for histological analysis. Immunohistochemistry analysis was performed to assess F4/80-positive macrophages in uterine sections and CD3-positive lymphocytes in both uterine and splenic sections, while neutrophil infiltration was evaluated by immunofluorescence staining of uterine cryosections with MPO antibody. Statistical analysis was conducted using SPSS 23.0 software. Flow cytometry data were analyzed using Flow Jo 10.0. The Mauchly’s test of sphericity was employed to assess the heterogeneity of variances. Data in accordance with normal distribution were compared by two-tailed unpaired t -test for two groups or one-way ANOVA for more than 2 groups. Statistical significance was defined as p  < 0.05. Further information on research design is available in the  Nature Portfolio Reporting Summary linked to this article.

To explore pathological changes associated with US, RNA-sequencing (RNA-seq) was conducted for the uterine myometrial tissues excised from pregnant women with US, using the normal uterine myometrium from pregnant women without US as a control. Detailed clinical characteristics of all normal and US-affected myometrial samples are provided in Supplementary Table  1 . The heatmap and volcano map showed differentially expressed genes (DEGs) in myometrial tissues between the normal and US groups (Fig.  1 a, b), with 333 genes were upregulated and 263 genes were downregulated in the US group. Given the established paradigm of fibrogenesis initiation through localized overexpression of fibrosis-associated genes at scar sites, we strategically prioritized the analysis of upregulated transcripts. Our screening identified 108 genes exhibiting significant enrichment in scar formation-associated Gene Ontology (GO) pathways (Fig.  1c , highlighted in blue). Notably, sFRP2, a canonical Wnt signaling modulator with well-documented roles in fibrotic activation across diverse organ systems including the liver, lung, and heart 15 , 16 , 18 , demonstrated exceptional differential expression profiles. Remarkably, sFRP2 manifested as the most statistically robust DEG in scar-related pathways (fold change = 8.5, p-value = 2.2 × 10 −9 ). This compelling association, coupled with the persistent knowledge gap in sFRP2’s involvement in US pathogenesis, prompted systematic investigation of its potential mechanistic contributions to this disorder. Fig. 1 RNA-seq analysis, quantification of scar formation-related biomarkers, and histological examination of human myometrial tissues without US (normal) or with US. a ,  b  Heatmap ( a ) and volcano plots ( b ) illustrate DEGs identified through RNA-seq analysis of human myometrial tissues from normal and US pregnant women ( n  = 3 samples). c GO enrichment analysis showing up-regulated GO terms in the US group compared to the normal group ( n  = 3 samples). Blue, scar formation-related pathways; Red, Wnt-related pathways; Green, Ca 2+ -related pathways. d ,  e  Representative WB bands ( d ) and quantification ( e ) of sFRP2, VEGFA, and α-SMA in human myometrial tissues from the normal and US groups. f ,  g Relative mRNA levels ( f ) and immunohistochemistry analysis ( g , n  = 6 samples) of sFRP2, VEGFA, and α-SMA in human myometrial tissues from the normal and US groups. The red arrows indicate neovascularization. h Human myometrial tissue sections stained with Masson’s trichrome or H&E ( n  = 6 samples). i ,  j  WB analysis of protein levels ( i ) and qPCR quantification of mRNA levels ( j ) of fibrinogen, Col1A1, and Col3A1 in human myometrial tissues. k – m Immunofluorescence analysis of the spatial co-distribution of sFRP2 with fibrinogen ( k ) Col1A1 ( l ) and Col3A1 ( m ) in human myometrial tissues of the US group ( n  = 6 samples). Box plots in e , f , j illustrate data distributions ( n  = 6 samples), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests. Source data are provided as a Source Data file. a ,  b  Heatmap ( a ) and volcano plots ( b ) illustrate DEGs identified through RNA-seq analysis of human myometrial tissues from normal and US pregnant women ( n  = 3 samples). c GO enrichment analysis showing up-regulated GO terms in the US group compared to the normal group ( n  = 3 samples). Blue, scar formation-related pathways; Red, Wnt-related pathways; Green, Ca 2+ -related pathways. d ,  e  Representative WB bands ( d ) and quantification ( e ) of sFRP2, VEGFA, and α-SMA in human myometrial tissues from the normal and US groups. f ,  g Relative mRNA levels ( f ) and immunohistochemistry analysis ( g , n  = 6 samples) of sFRP2, VEGFA, and α-SMA in human myometrial tissues from the normal and US groups. The red arrows indicate neovascularization. h Human myometrial tissue sections stained with Masson’s trichrome or H&E ( n  = 6 samples). i ,  j  WB analysis of protein levels ( i ) and qPCR quantification of mRNA levels ( j ) of fibrinogen, Col1A1, and Col3A1 in human myometrial tissues. k – m Immunofluorescence analysis of the spatial co-distribution of sFRP2 with fibrinogen ( k ) Col1A1 ( l ) and Col3A1 ( m ) in human myometrial tissues of the US group ( n  = 6 samples). Box plots in e , f , j illustrate data distributions ( n  = 6 samples), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests. Source data are provided as a Source Data file. Analyses by Western blot (WB), quantitative real-time PCR (qPCR), and immunohistochemistry further confirmed significantly higher mRNA and protein levels of sFRP2 in the scarred myometrium (Fig.  1d–g ). It has been well demonstrated that excessive angiogenesis can facilitate scar tissue formation by supplying essential nutrients and oxygen 19 . Accordingly, we also analyzed uterine expression profiles of vascular endothelial growth factor A (VEGFA), one of the most important positive regulators of angiogenesis. It was found that the protein and mRNA levels of VEGFA significantly increased in the US group, along with the increased sFRP2 expression. Meanwhile, in line with the absence of the muscular layer in scar tissues 20 , we detected decreased α-smooth muscle actin (α-SMA, typically expressed in muscle tissues), in the scarred myometrium. Examination of Masson’s trichrome-stained histological sections indicated extensive deposition of collagen fibers at the scar site (Fig.  1h ), while hematoxylin and eosin (H&E) staining showed irregularly arranged and severely impaired muscular tissues. This suggested that overproduced fibrosis and deposited extracellular matrix (ECM) occupy the defected muscular layer. Further WB and qPCR assays revealed increased levels of markers associated with fibrosis and ECM deposition, including fibrinogen, collagen type I alpha 1 chain (Col1A1), and collagen type III alpha 1 chain (Col3A1) in the myometrial tissue of the US group (Fig.  1 i, j and Supplementary Fig.  1 ). Additionally, immunofluorescence staining showed a notable spatial co-distribution of overexpressed sFRP2 with increased expressions of fibrinogen, Col1A1, and Col3A1 in the scarred myometrium (Fig.  1k–m ). These results suggest that the elevated expression of sFRP2 in the myometrium is notably associated with scar development by simultaneously activating fibrosis and angiogenesis. Of note, angiogenesis can create a conducive environment for ECM deposition, thereby compensating for the deficiency in the muscular layer of scar tissues. To further explore the US pathogenesis, we isolated three major types of cells from the scarred human myometrium (Fig.  2a ), including uterine smooth muscle cells (hUSMCs), uterine fibroblasts (hUFIBs), and endometrial cells (hUECs). Among these cell types, hUFIBs from the US group exhibited the most significant increase in sFRP2 expression (Fig.  2b–d and Supplementary Fig.  2a ). Moreover, considering the importance of fibroblasts as the predominant connective tissue cells during the scar formation 21 , we focused on further exploration with hUFIBs. Consistent with the findings from the myometrial tissue, levels of sFRP2, VEGFA, fibrinogen, Col1A1, and Col3A1 were elevated in hUFIBs from the US group (Fig.  2e–h and Supplementary Fig.  2 b, c). Interestingly, unlike the myometrial tissue, a significant increase in α-SMA was observed in hUFIBs of the US group (Fig.  2 e, f and Supplementary Fig.  2b ). Since α-SMA acts as a hallmark of mature myofibroblasts 22 , this implied that the increased sFRP2 expression is strongly associated with the fibroblast-to-myofibroblast transition (FMT) in hUFIBs from the US group. This pathologically compensatory response is generally triggered to address the muscular layer deficiency in scar tissues 23 . Similarly, immunofluorescence staining of hUFIBs revealed increased expressions of sFRP2, α-SMA, Col3A1, and Col1A1 in the US group (Fig.  2i, j ). Furthermore, we detected spatial co-distribution of α-SMA, Col3A1, and Col1A1 with increased sFRP2, indicating that elevated sFRP2 is closely related to the myofibroblast transition of hUFIBs and ECM deposition. Fig. 2 Expression of sFRP2 and biomarkers associated with the scar formation in various cells isolated from the normal and scarred uteruses. a A diagram depicting isolation of hUSMCs, hUFIBs, and hUECs from human uterine tissues. b ,  c Representative WB bands ( b ) and quantitative analysis ( c ) of sFRP2 in hUSMCs, hUFIBs, and hUECs sourced from both normal and US cohorts. d mRNA levels of sFRP2 in hUSMCs, hUFIBs, and hUECs. e WB analysis of protein expression profiles of sFRP2, VEGFA, and α-SMA in hUFIBs. f Quantified mRNA expression levels of sFRP2, VEGFA, and α-SMA in hUFIBs. g ,  h  WB analysis ( g ) and qPCR quantification ( h ) of fibrinogen, Col1A1, and Col3A1 in hUFIBs. i , j  Immunofluorescence analysis of the co-distribution of sFRP2 with α-SMA ( i ), Col3A1, or Col1A1 ( j ) in hUFIBs ( n  = 6 samples). k Microscopic images of migrated hUFIBs. l Digital photos show collagen gel contractility assay of hUFIBs, with white circles marking the collagen periphery after 24 h of culture. Data in bar graphs ( c , d ) are presented as means ± standard deviation ( n  = 6 samples). Box plots in f , h illustrate data distributions ( n  = 6 samples), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests for two groups and one-way ANOVA with post hoc LSD tests for multiple groups. Source data are provided as a Source Data file. a A diagram depicting isolation of hUSMCs, hUFIBs, and hUECs from human uterine tissues. b ,  c Representative WB bands ( b ) and quantitative analysis ( c ) of sFRP2 in hUSMCs, hUFIBs, and hUECs sourced from both normal and US cohorts. d mRNA levels of sFRP2 in hUSMCs, hUFIBs, and hUECs. e WB analysis of protein expression profiles of sFRP2, VEGFA, and α-SMA in hUFIBs. f Quantified mRNA expression levels of sFRP2, VEGFA, and α-SMA in hUFIBs. g ,  h  WB analysis ( g ) and qPCR quantification ( h ) of fibrinogen, Col1A1, and Col3A1 in hUFIBs. i , j  Immunofluorescence analysis of the co-distribution of sFRP2 with α-SMA ( i ), Col3A1, or Col1A1 ( j ) in hUFIBs ( n  = 6 samples). k Microscopic images of migrated hUFIBs. l Digital photos show collagen gel contractility assay of hUFIBs, with white circles marking the collagen periphery after 24 h of culture. Data in bar graphs ( c , d ) are presented as means ± standard deviation ( n  = 6 samples). Box plots in f , h illustrate data distributions ( n  = 6 samples), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests for two groups and one-way ANOVA with post hoc LSD tests for multiple groups. Source data are provided as a Source Data file. Myofibroblast activation at the injury site is typically characterized by high migratory ability and strong contractility 24 , 25 . Transwell assay showed high migratory potential of hUFIBs from the US group (Fig.  2k and Supplementary Fig.  2d ). Further, the contractile activity of hUFIBs was assessed by the collagen gel contraction assay. hUFIBs from the US group more significantly reduced the collagen gel area, as compared to those of the normal group (Fig.  2l and Supplementary Fig.  2e ), indicating their strong contractility. These findings demonstrated that hUFIBs from the US group exhibit an escalated transformation towards myofibroblasts. Collectively, our results support the notion that the increased expression of sFRP2 in hUFIBs is associated with the enhanced FMT, excessive ECM deposition, and fibrosis activation in the pathogenesis of US. To unravel the underlying mechanisms linking sFRP2 to uterine scar formation, a mouse model of US was established (Fig.  3a ). A longitudinal incision of 1 cm in length was made at the midportion of the left uterine horn of female mice, which penetrated through the myometrium and endometrium on the surgical side. The control group underwent a low abdominal incision without any uterine incision 9 . It is worth noting that neither the uterine surgery group (referred to as the model group) nor the control group experienced any postsurgical mortality or infections. At day 30 post-surgery, uteruses were collected for evaluation. RNA-seq revealed an increased sFRP2 expression (fold change = 5.2, p-value = 2.0 × 10 −8 ) in the scarred uterus of mice (Fig.  3b, c ). Differential expression profiling identified 361 upregulated and 242 downregulated genes, with 89 of the upregulated genes functionally associated with scar formation pathways. Both GO enrichment analysis (scar-related pathways highlighted in blue, Fig.  3d ) and gene set enrichment analysis (GSEA) (Supplementary Fig.  3 ) indicated significant enrichment of fibrotic pathways in model mice. Furthermore, GSEA revealed conserved upregulation of angiogenesis-related pathways in both human and murine transcriptomic datasets, reinforcing the translational relevance of our findings. Various validation results consistently demonstrated increased levels of sFRP2 and VEGFA, along with decreased α-SMA expression in the uteruses of the model group (Fig.  3e–i and Supplementary Fig.  4a ). Histological analysis after Masson’s trichrome and H&E staining revealed increased collagen deposition and disrupted muscle arrangement in the scarred mouse uteruses (Fig.  3j, k ). Correspondingly, we found elevated expressions of fibrinogen, Col1A1, and Col3A1 in the model group (Supplementary Fig.  4b–d ). Confocal imaging further confirmed co-distributions of increased sFRP2 with elevated levels of fibrinogen, Col1A1, and Col3A1 in the model group (Supplementary Fig.  4e–g ). Consequently, this mouse model of uterine surgical scarring thoroughly recapitulates the pathogenesis of US in the human myometrium. Fig. 3 RNA-seq analysis, quantification of scar formation-related biomarkers, and histological examination of uteruses from mice with uterine surgical scarring, as well as analysis of scar transformation of mUFIBs. a A schematic diagram showing the protocol for establishing uterine scars in mice (model), with control animals undergoing a low abdominal incision without uterine incision (control). b ,  c  Heatmap ( b ) and volcano plots ( c ) illustrate RNA-seq data comparing mouse uterine tissues from the surgical scar group with the non-scar group ( n  = 3 mice). d GO analysis showcases the up-regulated GO terms in the model group versus the control group, with blue indicating scar-related pathways and green highlighting Ca 2+ -related pathways ( n  = 3 mice). e – i Analysis of sFRP2, VEGFA, and α-SMA in mouse uterine tissues by WB ( e ) qPCR ( f ) and immunohistochemistry ( g – i , n  = 6 mice). Red arrows indicate neovascularization. j ,  k  Microscopic images of mouse uterine tissue sections stained with Masson’s trichrome ( j , n  = 6 mice) or H&E ( k , n  = 6 mice). l A workflow depicts the isolation of mUFIBs. m WB analysis of the expression levels of fibrinogen, Col1A1, and Col3A1 in mUFIBs.   n , o  Immunofluorescence analysis of the co-distribution of sFRP2 with α-SMA ( n ), Col3A1, and Col1A1 ( o ) in mUFIBs ( n  = 6 mice). p Microscopic images of migrated mUFIBs. q Digital photos show the contractility performance of mUFIBs in collagen gel contractility assay. White circles depict the collagen boundary after 24 h of incubation. Box plots in f display data distribution ( n  = 6 mice), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests. Source data are provided as a Source Data file. a A schematic diagram showing the protocol for establishing uterine scars in mice (model), with control animals undergoing a low abdominal incision without uterine incision (control). b ,  c  Heatmap ( b ) and volcano plots ( c ) illustrate RNA-seq data comparing mouse uterine tissues from the surgical scar group with the non-scar group ( n  = 3 mice). d GO analysis showcases the up-regulated GO terms in the model group versus the control group, with blue indicating scar-related pathways and green highlighting Ca 2+ -related pathways ( n  = 3 mice). e – i Analysis of sFRP2, VEGFA, and α-SMA in mouse uterine tissues by WB ( e ) qPCR ( f ) and immunohistochemistry ( g – i , n  = 6 mice). Red arrows indicate neovascularization. j ,  k  Microscopic images of mouse uterine tissue sections stained with Masson’s trichrome ( j , n  = 6 mice) or H&E ( k , n  = 6 mice). l A workflow depicts the isolation of mUFIBs. m WB analysis of the expression levels of fibrinogen, Col1A1, and Col3A1 in mUFIBs.   n , o  Immunofluorescence analysis of the co-distribution of sFRP2 with α-SMA ( n ), Col3A1, and Col1A1 ( o ) in mUFIBs ( n  = 6 mice). p Microscopic images of migrated mUFIBs. q Digital photos show the contractility performance of mUFIBs in collagen gel contractility assay. White circles depict the collagen boundary after 24 h of incubation. Box plots in f display data distribution ( n  = 6 mice), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests. Source data are provided as a Source Data file. Then mouse uterine fibroblasts (mUFIBs) were extracted from the scar site of the US group or the same site of the control group (Fig.  3l ). Consistent with the findings in hUFIBs, mUFIBs from the model group exhibited increased expressions of sFRP2, VEGFA, α-SMA, fibrinogen, Col1A1, and Col3A1 (Fig.  3m–o and Supplementary Fig.  5a–e ). Transwell and collagen gel contraction assays further confirmed myofibroblast differentiation and activation of mUFIBs from the model group (Fig.  3p, q and Supplementary Fig.  5f, g ). Taken together, our findings from the mouse uterine surgical scarring model also demonstrate the critical role of sFRP2 in the formation of uterine scars, primarily through its involvement in FMT, ECM deposition, and fibrosis activation. These findings mirror the role played by sFRP2 in fibroblasts of the scarred human myometrium. To further investigate the potential role of sFRP2 in vivo, we administered lentiviral vectors encoding a sFRP2-expressing fragment into the uterine horns of female mice. The control group received negative control lentivectors. Uteruses were collected at days 3, 5, 7, and 14 post-transfection for analyses. We found that sFRP2 overexpression peaked significantly at the 7th day post-transfection, compared to that at days 3 and 5 (Supplementary Fig.  6 ). Moreover, the elevated expression of sFRP2 at day 14 showed no significant difference compared to the 7th day. Therefore, the 7th day post-intervention was selected for further assessment (Fig.  4a ). Fig. 4 Effects of sFRP2 overexpression in the mouse uterine tissue on scar formation-related biomarkers, histological structures of the uterus, and scar transformation of isolated mUFIBs. a A flowchart shows the induction of sFRP2 overexpression (sFRP2OE) in the uterine tissue of mice, with control animals receiving negative control lentivectors. b – g  WB detection ( b , c ), qPCR quantification ( d ), and immunohistochemistry analysis ( e – g , n  = 6 mice) of sFRP2, VEGFA, and α-SMA in the mouse uterine tissues after 7 days of sFPR2OE intervention. The red arrows in f indicate neovascularization. h , i Histological sections of mouse uterine tissues stained with Masson’s trichrome or H&E ( n  = 6 mice). j WB analysis of fibrinogen, Col1A1, and Col3A1 levels in uterine tissues of sFRP2OE mice and control mice. k A diagram details the isolation of mUFIBs from both control and sFRP2OE mice. l Microscopic images and quantification of migrated mUFIBs from control and sFRP2OE groups. m Digital images (left) and quantitative analysis (right) assess the contractility of mUFIBs, with white circles marking the collagen boundary after 24 h of incubation. Box plots in c , d , l , m illustrate data distributions ( n  = 6 mice), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests. Source data are provided as a Source Data file. a A flowchart shows the induction of sFRP2 overexpression (sFRP2OE) in the uterine tissue of mice, with control animals receiving negative control lentivectors. b – g  WB detection ( b , c ), qPCR quantification ( d ), and immunohistochemistry analysis ( e – g , n  = 6 mice) of sFRP2, VEGFA, and α-SMA in the mouse uterine tissues after 7 days of sFPR2OE intervention. The red arrows in f indicate neovascularization. h , i Histological sections of mouse uterine tissues stained with Masson’s trichrome or H&E ( n  = 6 mice). j WB analysis of fibrinogen, Col1A1, and Col3A1 levels in uterine tissues of sFRP2OE mice and control mice. k A diagram details the isolation of mUFIBs from both control and sFRP2OE mice. l Microscopic images and quantification of migrated mUFIBs from control and sFRP2OE groups. m Digital images (left) and quantitative analysis (right) assess the contractility of mUFIBs, with white circles marking the collagen boundary after 24 h of incubation. Box plots in c , d , l , m illustrate data distributions ( n  = 6 mice), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests. Source data are provided as a Source Data file. In contrast to the control group, sFRP2 expression was significantly increased in the sFRP2 overexpression (sFRP2OE) group (Fig.  4b–e ). Additionally, sFRP2OE led to enhanced mRNA and protein levels of VEGFA and α-SMA in the uteruses (Fig.  4b–d ). Different from decreased α-SMA expression in the uterus from mice with surgical scars due to the muscle layer defects, sFRP2OE in the normal uterus only promoted the FMT, thus leading to an increase in α-SMA. Consistently, histological analyses also revealed increased VEGFA and α-SMA expressions as well as enhanced collagen synthesis and ECM deposition in the uteruses of sFRP2OE mice (Fig.  4f–i ). Furthermore, fibrinogen, Col1A1, and Col3A1 were upregulated post sFRP2OE (Fig.  4j and Supplementary Fig.  7a, b ). Importantly, immunofluorescence staining showed that elevated sFRP2 protein was predominantly co-distributed with fibrosis-related biomarkers in scar tissues (Supplementary Fig.  7c–e ). Further, mUFIBs were isolated from the sFRP2OE and control groups (Fig.  4k ). We observed increased sFRP2 in mUFIBs from the sFRP2OE group (Supplementary Fig.  8a–c ). In comparison to the control group, mUFIBs from the sFRP2OE group exhibited significantly higher levels of VEGFA, α-SMA, fibrinogen, Col1A1, and Col3A1 (Supplementary Fig.  8 ). Additionally, mUFIBs from the sFRP2OE group displayed notably increased migration capacity and stronger contractibility, compared with the control group (Fig.  4l, m ). These results collectively substantiate that sFRP2 overexpression in the normal uterine tissue can induce characteristic scar features, such as enhanced angiogenesis, fibroblast-myofibroblast differentiation, ECM deposition, and fibrotic activation, confirming its ability to initiate US formation independent of surgical injury. Based on these findings, we next investigated whether sFRP2 interacts synergistically with surgical injury to exacerbate US in female mice. We induced uterine surgical scarring in mice after lentiviral transfection-mediated sFRP2OE for 7 days. Subsequently, we collected uterine tissues at day 30 post-model establishment (sFRP2OE+Model). Comparative analysis with the Model group (treated with negative control lentiviruses plus uterine surgical scarring) revealed significantly elevated sFRP2 expression in the sFRP2OE + Model group (Supplementary Fig.  9a–d ). Furthermore, the sFRP2OE + Model group exhibited notably increased expressions of scar formation-associated biomarkers, including VEGFA, fibrinogen, Col1A1, and Col3A1, compared with the Model group (Supplementary Figs.  9 , 10 ). To further validate the upregulation of sFRP2 in fibroblasts, mUFIBs from the sFRP2OE+Model and Model groups were isolated. Different assessments affirmed significantly increased sFRP2 mRNA and protein levels in mUFIBs of the sFRP2OE+Model group, along with remarkably elevated expressions of VEGFA, α-SMA, fibrinogen, Col3A1, and Col1A1 (Supplementary Fig.  11 ). In addition, sFRP2 showed notable co-distribution with the examined biomarkers relevant to scar formation (Supplementary Fig.  12a, b ). Similarly, mUFIBs from the sFRP2OE + Model group exhibited considerably enhanced myofibroblast differentiation (Supplementary Fig.  12c–f ). These results demonstrate that sFRP2 serves dual roles in uterine scarring, functioning both as an initiator of fibrotic formation and as an amplifier of surgery-induced scar progression. To further validate the conclusion that sFRP2 acts as an initiator rather than a defensive signal in scar formation, we investigated the effects of sFRP2 knockdown (sFRP2KD) in vivo. A lentiviral vector containing specific short hairpin RNA (shRNA) targeting sFRP2 was injected into the uteruses of mice. At days 3, 5, 7, and 14 post-injection, uterine tissues were collected. According to the knockdown efficiency (Supplementary Fig.  13 ), we selected 7-day intervention for subsequent experiments, as it exhibited the optimal knockdown effect and corresponded to the intervention timing in the sFRP2OE experiment. After the sFRP2KD lentivirus was injected into the uteruses of mice for 7 days, the uterine scar-inducing surgery was performed. The control group was treated with scramble shRNA and the same model induction procedure. At day 30 after uterine scar establishment, all mice were sacrificed for analyses. Remarkably, we observed that sFRP2KD significantly attenuated scar formation in the uterus (Supplementary Figs.  14 , 15 ). Furthermore, mUFIBs isolated from the sFRP2KD group exhibited reduced angiogenesis, inhibited FMT, decreased ECM deposition, and alleviated fibrotic activation (Supplementary Figs.  16 , 17 ). These experiments collectively suggest that sFRP2 acts as a positive regulator of scar pathogenesis in the uterus. The above studies provided substantial evidence for the crucial role of sFRP2 in mice. To further confirm its effects in humans, hUFIBs were isolated from the normal myometrium. Lentivirus containing a human sFRP2 segment was transfected into hUFIBs for 3 days to overexpress sFRP2. A negative control lentivector was transfected into hUFIBs as the control group, while the wild-type (WT) group was treated with PBS. WB and qPCR results demonstrated that the upregulation of sFRP2 led to increased expressions of VEGFA, α-SMA, fibrinogen, Col1A1, and Col3A1 in hUFIBs (Fig.  5a–d and Supplementary Fig.  18 ). Immunofluorescence staining supported this finding, showing that the elevation of sFRP2 corresponded with increased levels of fibrinogen, Col1A1, and Col3A1 in the sFRP2OE group (Supplementary Fig.  19a–c ). Besides, notable co-distribution was observed between sFRP2 and these scar formation-related biomarkers. Moreover, hUFIBs in the sFRP2OE group showed significantly enhanced migration and contractile capacity (Fig.  5e, f and Supplementary Fig.  19d, e ). Fig. 5 sFRP2 overexpression or knockdown affects scar formation-related biomarkers and scar transformation in hUFIBs. a ,  b  WB ( a ) and qPCR ( b ) analyses of sFRP2, VEGFA, and α-SMA in hUFIBs with sFRP2 overexpression. Wild-type (WT) cells received PBS treatment, while the control group was treated with negative control lentivectors. c , d  WB analysis ( c ) and qPCR quantification ( d ) of fibrinogen, Col1A1, and Col3A1 in hUFIBs subjected to different treatments. e Microscopic images of migrated hUFIBs across WT, control, and sFRP2OE groups. f Digital photos show the contractility of hUFIBs in different groups, with white circles depict the collagen boundary after 24 h of incubation. g , h WB ( g ) and qPCR ( h ) analyses of the effects of sFRP2 knockdown and subsequent TGF-β stimulation for 3 days on the expression of sFRP2, VEGFA, and α-SMA in hUFIBs. i , j WB analysis ( i ) and qPCR quantification ( j ) of fibrinogen, Col1A1, and Col3A1 in hUFIBs undergoing sFRP2 knockdown and TGF-β treatment for 3 days. The WT group was treated with PBS and TGF-β, while the control group received scrambled shRNA and TGF-β. k – n  GSEA of k canonical and m non-canonical Wnt signaling pathways in human myometrial RNA-seq datasets, with corresponding qPCR validation of l canonical (β-catenin, Myc, cyclin D1) and n non-canonical (PKC, RhoA, JNK) Wnt pathway markers in hUFIBs from normal and US groups. Box plots in b , d , h , j , l , n illustrate data distributions ( n  = 6 samples), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests for two groups and one-way ANOVA with post hoc LSD tests for multiple groups. Source data are provided as a Source Data file. a ,  b  WB ( a ) and qPCR ( b ) analyses of sFRP2, VEGFA, and α-SMA in hUFIBs with sFRP2 overexpression. Wild-type (WT) cells received PBS treatment, while the control group was treated with negative control lentivectors. c , d  WB analysis ( c ) and qPCR quantification ( d ) of fibrinogen, Col1A1, and Col3A1 in hUFIBs subjected to different treatments. e Microscopic images of migrated hUFIBs across WT, control, and sFRP2OE groups. f Digital photos show the contractility of hUFIBs in different groups, with white circles depict the collagen boundary after 24 h of incubation. g , h WB ( g ) and qPCR ( h ) analyses of the effects of sFRP2 knockdown and subsequent TGF-β stimulation for 3 days on the expression of sFRP2, VEGFA, and α-SMA in hUFIBs. i , j WB analysis ( i ) and qPCR quantification ( j ) of fibrinogen, Col1A1, and Col3A1 in hUFIBs undergoing sFRP2 knockdown and TGF-β treatment for 3 days. The WT group was treated with PBS and TGF-β, while the control group received scrambled shRNA and TGF-β. k – n  GSEA of k canonical and m non-canonical Wnt signaling pathways in human myometrial RNA-seq datasets, with corresponding qPCR validation of l canonical (β-catenin, Myc, cyclin D1) and n non-canonical (PKC, RhoA, JNK) Wnt pathway markers in hUFIBs from normal and US groups. Box plots in b , d , h , j , l , n illustrate data distributions ( n  = 6 samples), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests for two groups and one-way ANOVA with post hoc LSD tests for multiple groups. Source data are provided as a Source Data file. As well demonstrated, TGF-β can potently stimulate the proliferation, activation, and myofibroblast differentiation of fibroblasts, thus leading to increased ECM deposition 26 . The upregulated TGF-β expression is responsible for the development of many fibrotic diseases 27 . To examine the effect of sFRP2KD on scar-like transformation of fibroblasts, human recombinant TGF-β was used to stimulate hUFIBs in vitro. In this case, lentivirus carrying shRNA against sFRP2 was transfected into hUFIBs for 3 days to realize sFRP2KD. hUFIBs transfected with scramble shRNA served as the control group, while hUFIBs in the WT group was treated with PBS. Subsequently, all groups were stimulated with TGF-β. We found that sFRP2KD significantly abrogated TGF-β-induced expression of VEGFA, α-SMA, fibrinogen, Col1A1, and Col3A1 in hUFIBs (Fig.  5g–j and Supplementary Fig.  20 ), as compared to the control and WT groups. Immunofluorescence staining of fibrinogen, Col1A1, and Col3A1 further confirmed that sFRP2KD effectively blocked TGF-β-induced fibroblast scarring (Supplementary Fig.  21 ). Consistently, the migration of fibroblasts was notably decelerated following sFRP2KD (Supplementary Fig.  22a ). In addition, TGF-β-treatedhUFIBs after sFRP2KD exhibited reduced contractility compared to the control and WT groups (Supplementary Fig.  22b ), indicating attenuated fibrotic activation of sFRP2KD hUFIBs, even under TGF-β stimulation. Collectively, our results demonstrated that overexpression of sFRP2 leads to excessive production of scar formation-related biomarkers and increased FMT in hUFIBs. By contrast, sFRP2 knockdown effectively reduced the expression of these biomarkers and attenuated the myofibroblast differentiation of TGF-β-stimulated hUFIBs. These findings underscore the pivotal role of sFRP2 as a key initiator in promoting fibroblast-mediated scarring. The above experiments confirmed that sFRP2 drives fibroblast-mediated scarring, contributing to US formation. As a secreted protein, sFRP2 has been shown to influence the activation of surrounding cells through paracrine signaling and sustain cell growth through autocrine signaling 28 , 29 . Consequently, we sought to determine whether sFRP2 affects fibroblast activation through autocrine or paracrine mechanisms. As aforementioned, sFRP2 overexpression (sFRP2OE) in hUFIBs directly induced fibrotic activation, whereas sFRP2 knockdown (sFRP2KD) suppressed TGF-β-induced upregulation of fibrosis markers. These findings suggest that endogenous sFRP2 autonomously regulates the fibrotic process in hUFIBs. Accordingly, sFRP2 promotes fibroblast scarring through an autocrine mechanism. Furthermore, previous studies have demonstrated that immune cell-derived sFRP2 can activate neighboring fibroblasts through paracrine signaling, triggering downstream pathways that promote fibroblast activation and ECM deposition, key features of fibrosis and tissue repair 30 . To examine the paracrine function of sFRP2 in hUFIBs, we collected the conditioned media (CM) from hUFIBs subjected to different treatments and quantified sFRP2 levels. Our analysis revealed that the CM from sFRP2OE hUFIBs contained significantly elevated sFRP2 concentrations compared to the WT and control groups, while sFRP2 secretion was markedly reduced in the sFRP2KD-derived CM (Supplementary Fig.  23a ). This result indicated that overexpression of sFRP2 can enhances its extracellular release, whereas sFRP2 knockdown effectively suppresses its secretion. Next, we treated naive hUFIBs with CM from WT, sFRP2OE, and sFRP2KD groups. Notably, CM from sFRP2OE hUFIBs significantly upregulated fibrotic markers including fibrinogen, Col1A1, and Col3A1, while this pro-fibrotic effect was markedly attenuated in hUFIBs treated with the sFRP2KD-derived CM. To further validate these findings, we supplemented sFRP2KD CM with recombinant human sFRP2 protein (rhsFRP2) and observed remarkable restoration of scar-like phenotypes in treated hUFIBs (Supplementary Fig.  23b, c ). These results provide compelling evidence that sFRP2 exerts paracrine effects on UFIBs mainly by activating neighboring fibroblasts. In summary, sFRP2 can promote US formation through dual mechanisms: paracrine activation of neighboring fibroblasts and autocrine amplification of fibroblast activation. These findings highlight sFRP2 as a pivotal regulator in US pathogenesis, coordinating both intercellular signaling to adjacent cells and intracellular pathways to drive fibrotic progression, thereby underscoring the importance of sFRP2 in US development. To gain a more profound insight into how sFRP2 regulates fibroblasts and promotes uterine scar formation, we identified enriched GO terms related to the Wnt signaling pathway (Fig.  1c , highlighted in red). As a Wnt-binding protein with a frizzled-like cysteine-rich domain, sFRP2 can regulate Wnt signaling 31 . Wnt signaling is known to operate through canonical and non-canonical pathways. Recent studies have implicated both pathways in fibrosis activation and scar formation 32 , 33 . To determine which specific Wnt pathway mediates sFRP2-induced US, we performed GSEA of human myometrium transcriptome data. Our results revealed significantly greater upregulation of non-canonical pathway components compared to canonical pathway elements (Fig.  5k, m ). Further supporting this observation, we examined mRNA expressions of key pathway markers, including canonical markers (β-catenin, MYC proto-oncogene (Myc), cyclin D1) and non-canonical markers [protein kinase C (PKC), Ras homolog family member A (RhoA), c-Jun N-terminal kinase (JNK)] in both human myometrial tissues and hUFIBs. While canonical markers were elevated in the myometrium but unchanged in hUFIBs (Fig.  5l and Supplementary Fig.  24a ), non-canonical pathway components showed significant upregulation in both tissues and hUFIBs (Fig.  5n and Supplementary Fig.  24b ). In addition, we analyzed expression of canonical (β-catenin, Myc, cyclin D1) and non-canonical (PKC, RhoA, JNK) Wnt pathway markers in both sFRP2OE and TGF-β-stimulated sFRP2KD groups. Of note, only non-canonical Wnt markers were upregulated in sFRP2OE hUFIBs (Supplementary Fig.  24c ), while their expression was reduced in sFRP2KD cells even under TGF-β stimulation (Supplementary Fig.  24d ). These results demonstrate that sFRP2 specifically activates the non-canonical Wnt pathway in UFIBs. Based on these findings, we focused subsequent investigations on the non-canonical Wnt pathway. Further transcriptome analysis revealed differential expression of four Wnt family genes in both human and mouse tissues, with WNT16, WNT5A, and WNT9A demonstrating upregulation and WNT2B exhibiting marked downregulation in both human myometrial and mouse uterine tissues (Fig.  6a ). Subsequent qPCR validation demonstrated that among these candidates, only WNT5A showed significant upregulation in both tissues (Fig.  6b ). WNT5A, a key player in the Wnt signaling pathway, can drive the FMT in pulmonary fibrosis and promote cardiac fibrosis 34 , 35 . Consequently, we hypothesize that sFRP2 can induce fibroblasts scarring by regulating the expression of non-canonical WNT5A, ultimately leading to uterine scar formation. Therefore, we investigated the role of WNT5A in the scarred myometrial tissue and in the process of sFRP2-induced scar-like transformation of UFIBs. We found an increased WNT5A level in the scarred human myometrium (Fig.  6c and Supplementary Fig.  25a ), showing an evident co-localization with sFRP2 (Fig.  6d ). Moreover, elevated levels of WNT5A were observed in isolated hUFIBs from the US group (Fig.  6 e, f and Supplementary Fig.  25b ). Further, the overexpression of sFRP2 in hUFIBs significantly increased the WNT5A expression (Fig.  6g and Supplementary Fig.  26a ), compared to the control and WT groups, thereby up-regulating the expression of fibrinogen (Supplementary Fig.  26b ), a typical fibrosis marker. To visualize spatial relationships, we quantified fluorescence intensity profiles of sFRP2, WNT5A, and fibrinogen along defined cell axes. Intensity plots indicated that while sFRP2/WNT5A frequently co-localized, both proteins showed adjacent but distinct distributions relative to fibrinogen deposits (Supplementary Fig.  26c ). Conversely, sFRP2KD significantly inhibited the expressions of WNT5A and fibrinogen in TGF-β-stimulated hUFIBs (Fig.  6h and Supplementary Fig.  26d–f ). Fig. 6 sFRP2 regulates WNT5A expressions and Ca 2+ influx in myometrial tissues and hUFIBs. a Venn diagram analysis of differentially expressed WNT family genes comparing human (blue) and mouse (red) transcriptomes. b qPCR analysis of the expression of WNT16, WNT5A, and WNT9A in human myometrial and mouse uterine scar tissues. c WB analysis of WNT5A levels in human myometrial tissues. d Immunofluorescence imaging and plot profile analysis delineate the co-localization of sFRP2 with WNT5A in the human myometrium with US ( n  = 6 samples). e ,  f  WB ( e ) and immunofluorescence ( f ) analyses of WNT5A in hUFIBs ( n  = 6 samples). g WB analysis of WNT5A in hUFIBs following sFRP2 overexpression (sFRP2OE), PBS-treated hUFIBs (WT), and negative control lentivector-treated hUFIBs (Control). h WB bands illustrate WNT5A levels in sFRP2-knockdown hUFIBs with TGF-β. WT cells received PBS and TGF-β, while control cells were treated with scramble shRNA and TGF-β. i WNT5A concentrations in conditioned media (CM) derived from WT, Control, and sFRP2OE groups, as well as corresponding heparinase-supplemented counterparts. j Immunoprecipitation (IP) assay indicates the interaction between FLAG-sFRP2 and HA-WNT5A in normal hUFIBs. k – m  WB ( k , l ) and immunofluorescence ( m , n  = 6 samples) analyses of fibrinogen, Col1A1, and Col3A1 in hUFIBs after WNT5A stimulation, using PBS as the control treatment. n , o  Immunofluorescence ( n , n  = 6 samples) and flow cytometric ( o ) analyses illustrate Ca 2+ influx (stained with Fluro-4 AM) in hUFIBs. p Flow cytometric quantification of Ca 2+ influx in hUFIBs subjected to different treatments. q Immunofluorescence analysis of the relationship of sFRP2 with Ca 2+ and fibrinogen in sFRP2OE hUFIBs ( n  = 6 samples). r Dynamic tracing of Ca 2+ influx in hUFIBs of sFRP2KD, WT, and Control groups with TGF-β ( n  = 6 samples). Box plots in b , i , l , o , p illustrate data distributions ( n  = 6 samples), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests for two groups and one-way ANOVA with post hoc LSD tests for multiple groups. Source data are provided as a Source Data file. a Venn diagram analysis of differentially expressed WNT family genes comparing human (blue) and mouse (red) transcriptomes. b qPCR analysis of the expression of WNT16, WNT5A, and WNT9A in human myometrial and mouse uterine scar tissues. c WB analysis of WNT5A levels in human myometrial tissues. d Immunofluorescence imaging and plot profile analysis delineate the co-localization of sFRP2 with WNT5A in the human myometrium with US ( n  = 6 samples). e ,  f  WB ( e ) and immunofluorescence ( f ) analyses of WNT5A in hUFIBs ( n  = 6 samples). g WB analysis of WNT5A in hUFIBs following sFRP2 overexpression (sFRP2OE), PBS-treated hUFIBs (WT), and negative control lentivector-treated hUFIBs (Control). h WB bands illustrate WNT5A levels in sFRP2-knockdown hUFIBs with TGF-β. WT cells received PBS and TGF-β, while control cells were treated with scramble shRNA and TGF-β. i WNT5A concentrations in conditioned media (CM) derived from WT, Control, and sFRP2OE groups, as well as corresponding heparinase-supplemented counterparts. j Immunoprecipitation (IP) assay indicates the interaction between FLAG-sFRP2 and HA-WNT5A in normal hUFIBs. k – m  WB ( k , l ) and immunofluorescence ( m , n  = 6 samples) analyses of fibrinogen, Col1A1, and Col3A1 in hUFIBs after WNT5A stimulation, using PBS as the control treatment. n , o  Immunofluorescence ( n , n  = 6 samples) and flow cytometric ( o ) analyses illustrate Ca 2+ influx (stained with Fluro-4 AM) in hUFIBs. p Flow cytometric quantification of Ca 2+ influx in hUFIBs subjected to different treatments. q Immunofluorescence analysis of the relationship of sFRP2 with Ca 2+ and fibrinogen in sFRP2OE hUFIBs ( n  = 6 samples). r Dynamic tracing of Ca 2+ influx in hUFIBs of sFRP2KD, WT, and Control groups with TGF-β ( n  = 6 samples). Box plots in b , i , l , o , p illustrate data distributions ( n  = 6 samples), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by two-tailed unpaired t -tests for two groups and one-way ANOVA with post hoc LSD tests for multiple groups. Source data are provided as a Source Data file. Regarding the specific mechanism of sFRP2-mediated WNT5A regulation in US, recent studies have revealed that sFRP2 forms heterodimers with WNT5A, enhancing exosome-mediated WNT5A re-secretion. This process is facilitated through sFRP2 binding to WNT5A, which promotes its accumulation on cell-surface heparan sulfate proteoglycans (HSPGs) and subsequent endocytosis for exosomal WNT5A secretion 36 . To verify whether sFRP2 forms a heterodimer with WNT5A and accumulates on HSPGs to promote WNT5A secretion in hUFIBs, we initially quantified WNT5A secretion levels via ELISA in the WT, Control, and sFRP2OE cells. Parallel experiments with heparinase treatment were performed to assess whether heparinase can disrupt sFRP2-WNT5A heterodimer binding to HSPGs and consequently inhibit WNT5A secretion. ELISA analysis revealed significantly increased WNT5A secretion in sFRP2OE hUFIBs, which was effectively blocked by heparinase treatment (Fig.  6i ). Furthermore, qPCR quantification confirmed that heparinase-mediated inhibition of sFRP2-WNT5A dimer formation reduced WNT5A re-secretion, thereby suppressing activation of the non-canonical Wnt pathway (Supplementary Fig.  27 ). To further validate the physical interaction between sFRP2 and WNT5A, we constructed expression vectors by cloning synthetic cDNA fragments encoding full-length human-derived WNT5A (tagged with an HA epitope, HA-WNT5A) and sFRP2 (tagged with a FLAG epitope, FLAG-sFRP2) into the pcDNA3.1 vector. Subsequently, HA-tagged WNT5A and FLAG-tagged sFRP2 were transfected into normal hUFIBs either individually or in combination. Co-immunoprecipitation assay indicated FLAG-sFRP2 in HA pull-down complexes and HA-WNT5A in FLAG pull-down complexes (Fig.  6j ), demonstrating that sFRP2 directly binds WNT5A to form a heterodimer, thereby facilitating WNT5A re-secretion. Furthermore, after hUFIBs from the normal myometrium were treated with exogenous human WNT5A, fibrosis-related biomarkers, such as fibrinogen, Col1A1, and Col3A1, were significantly increased in hUFIBs (Fig.  6k–m ). Similarly, the mouse US tissue exhibited an elevated expression of WNT5A, co-localizing with sFRP2 (Supplementary Fig.  28a–d ). The increased expression of WNT5A was also detected in mUFIBs from the model group (Supplementary Fig.  28e–h ). In addition, sFRP2OE in the mouse uterus resulted in significantly increased WNT5A levels in uterine tissues and isolated mUFIBs (Supplementary Fig.  29 ), concomitant with enhanced sFRP2 expressions. Surgical induction of scarring in sFRP2OE mice (sFRP2OE+Model) caused further upregulation of WNT5A expressions in uterine tissues and mUFIBs, along with similar changes in sFRP2 levels (Supplementary Fig.  30 ). By contrast, sFRP2KD led to decreased WNT5A expressions in uterine tissues and mUFIBs (Supplementary Fig.  31 ). These findings collectively demonstrate elevated WNT5A expression in uterine tissues and UFIBs from both human and mice with US, likely mediated through sFRP2-WNT5A heterodimer formation, WNT5A recycling secretion, and subsequent activation of the non-canonical Wnt signaling pathway. Moreover, WNT5A alone was able to promote the fibrotic activation of hUFIBs, highlighting its potential role as a crucial intermediate regulator in sFRP2-initiated uterine scar formation. We also observed significant enrichment in pathways related to Ca 2+ in RNA-seq data from both human and mouse US tissues (Figs.  1c and 3 d, highlighted in green). As well documented, disruption of intracellular Ca 2+ homeostasis induces pro-fibrotic cellular responses 37 . In particular, increased intracellular Ca 2+ can upregulate fibrotic biomarkers in fibroblasts. Additionally, previous research has demonstrated sFRP2’s capacity to modulate Ca 2+ influx in immune cells 38 . Furthermore, emerging evidence indicates that sFRP2 can bind to frizzled receptor 5 (FZD5), triggering increased cytosolic Ca 2+ concentration and subsequent activation of downstream profibrotic pathways 31 , 39 . Therefore, we further examined whether sFRP2 can modulate Ca 2+ homeostasis through binding to FZD5 in fibroblasts and induce scar formation. An increased Ca 2+ influx was detected in hUFIBs from the US group (Fig.  6 n, o). We also validated the significantly increased Ca 2+ influx in hUFIBs with sFRP2OE, as compared to the control and WT groups (Fig.  6p ). Importantly, immunofluorescence analysis revealed that FZD5 co-localized with sFRP2 (Supplementary Fig.  32 a, b). Additionally, FZD5 expression, Ca 2+ influx, and fibrinogen expression were in tandem with the alteration of sFRP2 expression (Fig.  6q and Supplementary Fig.  32c ). Notably, TGF-β stimulation triggered FZD5 upregulation and extracellular Ca 2+ influx in normal fibroblasts. However, this TGF-β-mediated FZD5 overexpression and Ca 2+ mobilization were significantly attenuated in sFRP2-knockdown hUFIBs (Supplementary Figs.  32d–f and 33 a, b), with concomitant reductions in fibrinogen production (Supplementary Fig.  33c ). Moreover, dynamic Ca 2+ influx in hUFIBs was monitored by flow cytometry. Initially, a stable Ca 2+ influx baseline was established, followed by stimulation with TGF-β to induce a rapid increase in intracellular Ca 2+ levels. We found that the peak of intracellular Ca 2+ levels was significantly lower in the sFRP2KD group compared to the WT and control groups (Fig.  6r ). Also, a significantly increased Ca 2+ influx was found in mUFIBs isolated from mice with US, sFRP2OE mice, and sFRP2OE mice with US, relative to the corresponding controls (Supplementary Figs.  34 – 36 ). By contrast, the Ca 2+ influx was markedly attenuated in mUFIBs from sFRP2KD mice with US (Supplementary Fig.  37 ). These results demonstrate that sFRP2 can positively regulate the Ca 2+ influx in UFIBs. To examine the impact of Ca 2+ influx on fibrotic activation of hUFIBs without sFRP2 intervention, hUFIBs from the normal group were treated with CaCl 2 to stimulate Ca 2+ influx. We observed increased levels of fibrinogen, Col1A1, and Col3A1 in hUFIBs after Ca 2+ stimulation (Supplementary Fig.  38 ). This finding suggests that Ca 2+ influx, primarily mediated by sFRP2-FZD5 binding, is sufficient to promote hUFIBs-dependent fibrotic activation. Consequently, Ca 2+ signaling also serves as a key mediator in sFRP2-induced fibroblast fibrosis and scar formation. Integrating these mechanistic findings, we conclude that sFRP2 promotes the ECM deposition and FMT in UFIBs primarily through WNT5A-mediated activation of non-canonical Wnt signaling pathways and induction of Ca 2+ influx via binding to FZD5. These synergistic pathways collectively drive US pathogenesis. The above findings position sFRP2 as a therapeutic target for US. While TGF-β and VEGFA are established mediators of fibrosis 14 , 40 , our work reveals sFRP2 as a previously unrecognized upstream regulator that initiates fibroblast priming prior to downstream pathway activation. This mechanistic insight leads us to propose that sFRP2 inhibitors are effective therapies for US. As a proof of concept, siRNAs against human sFRP2 were initially designed. Among them, sisFRP2-h64231 exhibited the highest knockdown efficiency in hUFIBs (Supplementary Fig.  39 ). Then sisFRP2-h64231 was packaged into ionizable lipid nanoparticles (referred to as LNP/sisFRP2-h) (Fig.  7a ), since this type of nonviral vectors have been successfully used in several RNA therapies approved by U.S. Food and Drug Administration 41 – 44 . The lipid-to-siRNA weight ratio was selected based on agarose gel electrophoresis results (Supplementary Fig.  40 ), with a final ratio of 15:1. Both blank lipid nanoparticles (LNP) and LNP/sisFRP2-h exhibited spherical morphology, showing relatively narrow size distribution profiles and with the average diameter of 105 and 125 nm for LNP and LNP/sisFRP2-h64231, respectively (Fig.  7 b, c). The encapsulation efficiency of sisFRP2-h64231 was determined to be 91.4 ± 3.7%. Fig. 7 Lipid nanoparticles encapsulating sisFRP2 (LNP/sisFRP2-h) down-regulate sFRP2 in hUFIBs. a Schematic illustration of preparation of LNP/sisFRP2-h via microfluidic mixing. The ethanol solution containing DLin-MC3-DMA, DSPC, DMG-PEG, and cholesterol was rapidly mixed with an acidic aqueous solution containing sFRP2 siRNA (sisFRP2) in a microfluidic device to prepare LNP/sisFRP2-h. b , c  Transmission electron microscopy images (left) and size distribution profiles (right) of empty LNP ( b , n  = 6 samples) and LNP/sisFRP2-h ( c , n  = 6 samples). d A workflow showing treatment regimens. Control, hUFIBs treated with PBS alone; TGF-β, hUFIBs treated with PBS and TGF-β; LNP, hUFIBs treated with empty LNP and TGF-β; LNP/sisFRP2-h, hUFIBs treated with LNP/sisFRP2-h64231 at 50 or 150 nM of sisFRP2-h64231 and stimulated with TGF-β. e , f WB analysis of protein levels of sFRP2, WNT5A, VEGFA, and α-SMA in hUFIBs after different treatments. g mRNA levels of sFRP2, WNT5A, VEGFA, and α-SMA in hUFIBs. h – j  Immunofluorescence analysis of the co-distribution of sFRP2 with fibrinogen ( h ), Col1A1 ( i ), or Col3A1 ( j ) in hUFIBs ( n  = 6 samples). k Digital photos show the contractility of hUFIBs in different groups. White circles outline the collagen boundary after 24 h of incubation. l Microscopic images show hUFIBs migration subsequent to respective treatments. m Time-dependent Ca 2+ influx in hUFIBs post TGF-β stimulation ( n  = 6 samples). Box plots in g illustrate data distributions ( n  = 6 samples), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by one-way ANOVA with post hoc LSD tests. Source data are provided as a Source Data file. a Schematic illustration of preparation of LNP/sisFRP2-h via microfluidic mixing. The ethanol solution containing DLin-MC3-DMA, DSPC, DMG-PEG, and cholesterol was rapidly mixed with an acidic aqueous solution containing sFRP2 siRNA (sisFRP2) in a microfluidic device to prepare LNP/sisFRP2-h. b , c  Transmission electron microscopy images (left) and size distribution profiles (right) of empty LNP ( b , n  = 6 samples) and LNP/sisFRP2-h ( c , n  = 6 samples). d A workflow showing treatment regimens. Control, hUFIBs treated with PBS alone; TGF-β, hUFIBs treated with PBS and TGF-β; LNP, hUFIBs treated with empty LNP and TGF-β; LNP/sisFRP2-h, hUFIBs treated with LNP/sisFRP2-h64231 at 50 or 150 nM of sisFRP2-h64231 and stimulated with TGF-β. e , f WB analysis of protein levels of sFRP2, WNT5A, VEGFA, and α-SMA in hUFIBs after different treatments. g mRNA levels of sFRP2, WNT5A, VEGFA, and α-SMA in hUFIBs. h – j  Immunofluorescence analysis of the co-distribution of sFRP2 with fibrinogen ( h ), Col1A1 ( i ), or Col3A1 ( j ) in hUFIBs ( n  = 6 samples). k Digital photos show the contractility of hUFIBs in different groups. White circles outline the collagen boundary after 24 h of incubation. l Microscopic images show hUFIBs migration subsequent to respective treatments. m Time-dependent Ca 2+ influx in hUFIBs post TGF-β stimulation ( n  = 6 samples). Box plots in g illustrate data distributions ( n  = 6 samples), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by one-way ANOVA with post hoc LSD tests. Source data are provided as a Source Data file. We next assessed the efficacy of LNP/sisFRP2-h in inhibiting TGF-β-mediated scar-like transformation of hUFIBs. We treated hUFIBs isolated from the normal human myometrium with empty LNP or LNP/sisFRP2-h at 50 and 150 nM of sisFRP2-h64231 for 3 days, followed by stimulation with TGF-β. hUFIBs in the control group were treated with PBS, without TGF-β activation, while cells in the TGF-β group were treated with TGF-β alone (Fig.  7d ). While TGF-β treatment increased sFRP2 expression in hUFIBs, LNP/sisFRP2-h, particularly at 150 nM of sisFRP2-h64231, significantly downregulated expressions of sFRP2 and WNT5A as well as reduced the Ca 2+ influx in hUFIBs, leading to inhibited VEGFA excretion, decreased FMT, reduced collagen synthesis and deposition, attenuation of fibrosis activation, and declined dynamic Ca 2+ influx (Fig.  7e–m and Supplementary Figs.  41 – 43 ). These data demonstrate that LNP/sisFRP2-h can effectively prevent TGF-β-stimulated hUFIBs scarring in vitro. In view of preclinical therapeutic evaluations in mice, siRNAs targeting mouse sFRP2 were redesigned and screened through in vitro transfection experiments in mUFIBs (Supplementary Fig.  44 ). The ultimately selected sisFRP2-872 was then encapsulated into LNP at a lipid/siRNA weight ratio of 15:1, resulting in a final siRNA therapy defined as LNP/sisFRP2 (Supplementary Fig.  45a ). LNP/sisFRP2 also exhibited a nearly spherical shape (Fig.  8a ), with a mean diameter of 119 nm (Supplementary Fig.  45b ). The encapsulation efficiency of sisFRP2-872 was 90.3 ± 2.6%. Fig. 8 Dose-dependent therapeutic effects of LNP/sisFRP2 in a mouse model of US. a A representative TEM image of LNP/sisFRP2 ( n  = 6 samples). b A workflow depicts the treatment protocols. In these experiments, female mice were euthanized at day 30 post-uterine scarring and subsequent interventions. Control, mice subjected to a low abdominal incision without uterine incision; Model, mice with US induced by uterine scar surgery and receiving topical saline treatment; LNP/sisFRP2, US mice treated with LNP/sisFRP2 at 0.125, 0.25, and 0.5 mg/kg of sisFRP2. c Immunohistochemical analysis of sFRP2 in uterine tissues from different groups ( n  = 6 mice). d – g mRNA levels of sFRP2 ( d ) fibrinogen ( e ) Col1A1 ( f ) and Col3A1 ( g ) in uterine tissues from different groups. h H&E-stained uterine tissue sections for different groups ( n  = 6 mice). The black arrows indicate uterine glands. Box plots in d – g illustrate data distributions ( n  = 6 mice), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by one-way ANOVA with post hoc LSD tests. Source data are provided as a Source Data file. a A representative TEM image of LNP/sisFRP2 ( n  = 6 samples). b A workflow depicts the treatment protocols. In these experiments, female mice were euthanized at day 30 post-uterine scarring and subsequent interventions. Control, mice subjected to a low abdominal incision without uterine incision; Model, mice with US induced by uterine scar surgery and receiving topical saline treatment; LNP/sisFRP2, US mice treated with LNP/sisFRP2 at 0.125, 0.25, and 0.5 mg/kg of sisFRP2. c Immunohistochemical analysis of sFRP2 in uterine tissues from different groups ( n  = 6 mice). d – g mRNA levels of sFRP2 ( d ) fibrinogen ( e ) Col1A1 ( f ) and Col3A1 ( g ) in uterine tissues from different groups. h H&E-stained uterine tissue sections for different groups ( n  = 6 mice). The black arrows indicate uterine glands. Box plots in d – g illustrate data distributions ( n  = 6 mice), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by one-way ANOVA with post hoc LSD tests. Source data are provided as a Source Data file. According to previous studies, effective doses for intralesional siRNA-LNP administration typically ranged from 0.1 to 1 mg/kg siRNA, with 0.25 mg/kg generally demonstrating significant target gene knockdown 45 . For therapeutic evaluation of locally administered LNP/sisFRP2, we conducted dose-response studies via direct topical application to uterine scar lesions at 0.125, 0.25, and 0.5 mg/kg sisFRP2 (Fig.  8b ). Control groups included model mice receiving saline treatment and sham-operated mice undergoing abdominal incision without uterine trauma. At day 30 post-treatment, immunohistochemical analysis of uterine tissues revealed dose-dependent attenuation of sFRP2 protein expression following LNP/sisFRP2 administration (Fig.  8c ). qPCR analysis corroborated these findings, showing significant downregulation of sFRP2 mRNA levels in all LNP/sisFRP2 groups, with the maximal suppression observed at 0.5 mg/kg (Fig.  8d ). Consistently we observed substantial reductions in key fibrotic markers, including fibrinogen, Col1A1, and Col3A1, across all LNP/sisFRP2 treatment groups compared to the untreated model control (Fig.  8e–g ). As observed in the model group, US can damage the myometrium and reduce the number of uterine glands 9 . We found that the abnormal structures in the myometrium and endometrium were effectively normalized by LNP/sisFRP2, showing a notably improved muscular tissue layer in the myometrium and increased uterine glands in the endometrium (Fig.  8h ). Of note, the improved myometrial structure is well consistent with the reduced levels of sFRP2, fibrinogen, Col1A1, and Col3A1. These results demonstrate that local delivery of sisFRP2 using LNP can effectively prevent US and alleviate scarring-associated adverse effects on the myometrium and endometrium in mice. The endometrium serves as the primary interface dominating embryo-maternal interactions, offering an ideal environment for endometrial receptivity. US can disrupt the normal structure and function of the endometrium, thus affecting endometrial receptivity 46 . The above results substantiated that sisFPR2 therapies effectively restored the normal endometrial structure. To further examine whether sisFPR2 therapies can improve endometrial receptivity, female mice subjected to different treatments were mated with male mice before being sacrificed at a predetermined time (Fig.  9a ). Uterine tissues at embryonic day 4 (E4) were excised for evaluating endometrial receptivity at the pre-implantation stage. It is worth noting that endometrial receptivity is characterized by inhibited epithelial cell proliferation and promoted stromal cell proliferation in the endometrium 47 . Immunofluorescence staining of Ki67 (a cell proliferation marker) in the uterine tissue revealed increased proliferation of epithelial cells and reduced stromal cell proliferation in the endometrium of the model and LNP groups (Fig.  9b ). By contrast, LNP/sisFRP2 treatment notably improved these aberrations, thus enhancing uterine receptivity for decidualization. Fig. 9 Treatment with LNP/sisFRP2 improves uterus function and pregnancy outcomes. a A schematic illustrating treatment protocols. b Immunofluorescence imaging of Ki67 in the mouse uterus on E4 of gestation ( n  = 6 pregnant mice). c A sketch depicts the embryonic orientation within the uterus, highlighting the relative position of the section plane, with the deviation angle (θ) indicated by the red dashed line (the uterine-embryonic axis) against the black dashed line (the mesometrial-antimesometrial long (M-AM) axis). d H&E-stained sections of uterine tissues isolated on E7.5, with the uterine-embryonic axis and M-AM axis outlined by red and black dashed lines, respectively ( n  = 6 pregnant mice). e Uterus weights on E7.5 following different treatments. f Determination of the θ values post-implantation and the proportion of uterine-embryo disorientation on E7.5, categorizing disorientation as θ > 10°. g The upper panels show digital photographs of representative uteri on E12.5, with red arrows indicating scarring sites ( n  = 6 pregnant mice). The lower panels illustrate ultrasonographic images of the fetus and placenta at sites of uterine scarring on E12.5 ( n  = 6 pregnant mice), with white dashed lines delineating the embryo. h Assessment of the embryonic absorption rates at E12.5. i Ultrasound images (upper), gross morphology of fetuses (middle), and H&E-stained fetal sections (lower) at E17.5 ( n  = 6 pregnant mice), with the red line indicating the crown-rump length on ultrasound images. j Placental gross morphology (top) as well as whole slide (middle) and high-magnification views (lower) of H&E-stained placental sections ( n  = 6 pregnant mice), with black dashed lines highlighting the spongiotrophoblast layers, and red and yellow dashed lines denoting the labyrinth and max invasion distance, respectively. Box plots in e , f , h illustrate data distributions ( n  = 6 pregnant mice), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by one-way ANOVA with post hoc LSD tests. Source data are provided as a Source Data file. a A schematic illustrating treatment protocols. b Immunofluorescence imaging of Ki67 in the mouse uterus on E4 of gestation ( n  = 6 pregnant mice). c A sketch depicts the embryonic orientation within the uterus, highlighting the relative position of the section plane, with the deviation angle (θ) indicated by the red dashed line (the uterine-embryonic axis) against the black dashed line (the mesometrial-antimesometrial long (M-AM) axis). d H&E-stained sections of uterine tissues isolated on E7.5, with the uterine-embryonic axis and M-AM axis outlined by red and black dashed lines, respectively ( n  = 6 pregnant mice). e Uterus weights on E7.5 following different treatments. f Determination of the θ values post-implantation and the proportion of uterine-embryo disorientation on E7.5, categorizing disorientation as θ > 10°. g The upper panels show digital photographs of representative uteri on E12.5, with red arrows indicating scarring sites ( n  = 6 pregnant mice). The lower panels illustrate ultrasonographic images of the fetus and placenta at sites of uterine scarring on E12.5 ( n  = 6 pregnant mice), with white dashed lines delineating the embryo. h Assessment of the embryonic absorption rates at E12.5. i Ultrasound images (upper), gross morphology of fetuses (middle), and H&E-stained fetal sections (lower) at E17.5 ( n  = 6 pregnant mice), with the red line indicating the crown-rump length on ultrasound images. j Placental gross morphology (top) as well as whole slide (middle) and high-magnification views (lower) of H&E-stained placental sections ( n  = 6 pregnant mice), with black dashed lines highlighting the spongiotrophoblast layers, and red and yellow dashed lines denoting the labyrinth and max invasion distance, respectively. Box plots in e , f , h illustrate data distributions ( n  = 6 pregnant mice), with the central lines representing the mean, box boundaries indicating standard deviation, and whiskers showing the minimum and maximum values. The p values were calculated by one-way ANOVA with post hoc LSD tests. Source data are provided as a Source Data file. On the other hand, US and restricted uterine receptivity may cause abnormal embryo orientation and stagnant embryonic development at the post-implantation stage 11 . To further evaluate beneficial effects of sisFRP2 therapies, we examined embryo orientation at the implantation sites on E7.5 (Fig.  9c ). Histological analysis revealed significantly larger angles (> 10°) between the uterine-embryonic axis and the mesometrial-antimesometrial (M-AM) axis in the model and LNP groups (Fig.  9d ). The uterus weight on E7.5 showed an opposite changing profile, compared to the aberrant uterine-embryonic axis (Fig.  9e ). Treatment with LNP/sisFRP2 effectively restored the aberrant angles between the uterine-embryonic axis and the M-AM axis as well as reduced the number of disoriented embryos (Fig.  9f ). At E12.5, gross morphology observation and ultrasound imaging of embryo implantation sites in the scarred region showed obvious embryo absorption in the model and LNP groups (Fig.  9 g, h), agreeing with the previous finding that the aberrant uterine-embryonic axis may cause embryo resorption and loss 9 . In contrast, the LNP/sisFRP2 group showed significantly low resorption rates. Ultimately, examination of surviving embryos on E17.5 indicated that fetuses in the model and LNP groups exhibited slower development, accompanied with fetal hematoma and morphological abnormalities (Fig.  9i and Supplementary Fig.  46a ). Treatment with LNP/sisFRP2 efficaciously mitigated these adverse outcomes. The placenta, as a vital bridge between the mother and fetus, plays a crucial role in supporting fetal development in the womb by facilitating the exchange of nutrients, oxygen, waste products 48 . However, the presence of uterine scars can impair the placenta’s structure and function, such as disrupting the normal development of the placenta and affecting the blood flow to the placenta/embryo 49 , 50 . Histological analysis on E17.5 indicated an abnormally pale appearance of the placenta with scar tissues attached to the maternal side in the model and LNP groups (Fig.  9j ). Consistently, both the model and LNP groups showed significantly decreased placental weight, a notably enlarged spongiotrophoblast layer, and an extended trophoblast invasion distance (defined as the maximal invasion distance (Fig.  9j and Supplementary Fig.  46b ), indicating over-invasion of trophoblast cells into the decidua, accompanied by a reduced labyrinth layer. This further caused insufficient vascular branching in the labyrinthine layer of the placenta, thereby impairing embryo growth. Importantly, these placental abnormalities were reversed following LNP/sisFRP2 treatment. In a parallel experiment, we also examined typical pregnancy outcomes of mice subjected to different treatments. It was found that LNP/sisFRP2 intervention effectively enhanced the pregnancy rate and live fetus rate as well as reduced the numbers of fetuses with external deformities (Supplementary Table  2 ). These data demonstrated that adverse pregnancy outcomes resulting from US can be notably alleviated by sisFRP2 therapies. Taken together, these results indicate that LNP/sisFRP2 can effectively prevent abnormal implantation and development of embryos as well as mitigate adverse pregnancy outcomes associated with US. The clinically validated LNP delivery platform has demonstrated excellent safety profiles for nucleic acid delivery across multiple administration routes 51 , 52 . To evaluate potential immune responses following local uterine administration of LNP/sisFRP2, we conducted safety studies at both acute (24 h) and chronic (30 days) time points in female mice. Quantitative analysis of collected plasma samples indicated that levels of the proinflammatory cytokines TNF-α and IL-6 remained comparable between LNP/sisFRP2-treated animals and control mice at all examined time points, with no statistically significant differences detected in either the acute or chronic phase (Supplementary Fig.  47a ). Histopathological evaluation revealed comparable levels of neutrophil and macrophage infiltration in uterine tissues between control and LNP/sisFRP2 groups at both time points (Supplementary Fig.  47b, c ). Furthermore, we observed similar CD3 + lymphocyte populations in both uterine and splenic tissues across all groups at both examination periods (Supplementary Fig.  47d, e ). Consistently, organ index measurements of the uterus and spleen showed no significant intergroup variations, suggesting the absence of treatment-related edema (Supplementary Fig.  47f ). Collectively, these comprehensive safety assessments demonstrate that localized LNP/sisFRP2 treatment does not provoke measurable acute or chronic immune activation in the tested mouse model.

Here we find markedly elevated sFRP2 expression in human myometrial tissues with US, which exhibited distinct pathological features including excessive angiogenesis, muscular layer defects, increased collagen deposition, and fibrotic activation. Further comparative analysis revealed significantly higher expression of sFRP2 in hUFIBs sourced from the US group, compared to other cell types. These US-derivedhUFIBs manifested prominent scar-like transformation characteristics, as evidenced by significantly elevated levels of VEGFA, α-SMA, fibrinogen, Col1A1, and Col3A1, along with high migratory ability, strong contractility, and myofibroblast activation. In mice with uterine surgical scarring, we confirmed the essential functions of sFRP2 in mediating scar transformation of mUFIBs to promote US. These abnormalities in UFIBs were named as fibroblast scarring, to emphasize their special changes. Furthermore, we hypothesize that sFRP2 positively modulates US, instead of feedback signaling. In support of this hypothesis, overexpression of sFRP2 in the uterus of normal mice led to scarring features in both the uterus and mUFIBs. Moreover, sFRP2OE aggravated uterine surgical scarring in mice, and mUFIBs from sFRP2OE mice with surgery-induced uterine scars showed additionally promoted fibroblast scarring. By contrast, knockdown of sFRP2 in the mouse uterus significantly inhibited US and fibroblast scarring. We also found that sFRP2OE induced scar-like transformation in hUFIBs, while sFRP2KD blocked TGF-β-induced fibrotic activation of hUFIBs. Moreover, Transwell assays revealed distinct patterns of cellular distribution across experimental groups, with particularly uneven cell distribution observed for UFIBs from US tissues, sFRP2OE UFIBs, and TGF-β-stimulated UFIBs. These heterogeneous migration patterns likely reflect pathologically relevant cell behaviors characteristic of fibroblast activation, including upregulated ECM components, altered cell-cell and cell-matrix interactions, and the formation of cellular aggregates, which are all hallmark features of activated fibroblasts in fibrotic microenvironments 53 , 54 . These findings collectively demonstrate that sFRP2 acts as a critical positive regulator in US. Previous studies indicate that upregulated sFRP2 can mediate cardiac fibrosis, while sFRP2 knockdown or therapy with sFRP2 antibody reduces fibrosis and improves cardiac function in various animal models 16 , 55 . However, exogenously administered sFRP2 at therapeutic doses inhibited fibrosis and improved cardiac function in a rat model of myocardial infarction 56 . Meanwhile, increasing evidence has demonstrated that sFRP2 exhibits opposite activities in different tissues and cell types 57 , 58 . Consequently, available findings reveal functional diversity and multiples roles of sFRP2 under different pathophysiological conditions, such as cardiac development, cardiovascular disease, metabolic diseases, and cancers 18 , 59 , 60 . Herein, our results highlight a pivotal role of sFRP2 in mediating US by activating UFIBs and promoting fibroblast-to-myofibroblast trans-differentiation. sFRP2 is a well-recognized Wnt regulator 61 . Aged skin fibroblasts showed increased excretion of sFRP2 and WNT5A, which augmented angiogenesis and drove an invasive melanoma phenotype by the non-canonical Wnt signaling pathway in tumor cells 62 . As a Wnt ligand, WNT5A may promote fibrosis in the lung and heart, by activating the non-canonical Wnt pathway 33 , 63 . A study on endometriosis patients indicated that overexpressed sFRP2 positively regulates endometriosis lesion growth via canonical Wnt signaling 64 . Moreover, sFRP2 was able to trigger WNT5A expression in tumors 61 . Our studies revealed elevated WNT5A levels in both scarred uterine tissues and isolated UFIBs from humans and mice, showing evident co-localization with sFRP2. Further experiments demonstrated that sFRP2OE increased WNT5A expression, while sFRP2KD decreased WNT5A. We also proved that sFRP2 can directly interact with WNT5A to form functional heterodimers, which subsequently promote WNT5A re-secretion and consequent activation of non-canonical Wnt signaling pathways. Meanwhile, Ca 2+ signaling is also critical for the scar formation 65 . sFRP2 can regulate Ca 2+ influx in mouse splenic B cells and mesangial cells 17 , 38 . We substantiated that sFRP2 has a major role in promoting Ca 2+ influx by binding to FZD5 in UFIBs. Notably, the activation of WNT5A or Ca 2+ signaling alone can induce fibroblast scarring. These findings suggest that sFRP2 promotes US, mainly by activating the non-canonical Wnt pathway and increasing Ca 2+ influx. Our integrated findings establish sFRP2 as the pathogenic initiator of US through its coordinated regulation of fibroblast activation, angiogenesis, and ECM remodeling. This central role is supported by its upregulation in human and mouse uterine scar tissues, causal effects in functional studies, and mechanistic links to both non-canonical Wnt signaling activation and Ca 2+ mobilization cascades. Accordingly, sFRP2 can serve as a therapeutic target to prevent or treat US. To explore translational potential of sFRP2-targeting interventions, we established an ionizable LNP platform for siRNA delivery and demonstrated that LNP/sisFRP2 can effectively reduce sFRP2 expression in hUFIBs, thus inhibiting fibroblast scarring. A single dose of LNP/sisFRP2 applied on the uterine scar of mice significantly restored the uterine structure. Correspondingly, LNP/sisFRP2 therapy increased endometrial receptivity, promoted normal embryo implantation and development, improved the placental structure to support embryo growth, and rescued adverse pregnancy outcomes. Compared with previously explored strategies, such as cell transplantation and growth factor treatments 66 , 67 , our sFRP2-based siRNA therapy owns multiple advantages, like high specificity, excellent efficacy, easy formulation, patient-friendly delivery, good safety, and promising translation potential. While our study has identified sFRP2 as a key therapeutic target, it is important to acknowledge that US is a polygenic disease involving dynamic crosstalk between TGF-β, VEGF, PDGF, and TNF-α signaling cascades 10 . However, therapeutic translation to reproductive-aged women necessitates rigorous risk evaluation, as systemic multi-target interventions could potentially compromise endometrial receptivity and embryo implantation. Notably, our localized LNP/sisFRP2 delivery system can minimize such reproductive risks. Moving forward, developing precisely controlled combination therapies (e.g., concurrent sFRP2 and TGF-β inhibition) will be crucial for achieving optimal therapeutic outcomes while preserving fertility in this vulnerable population. In conclusion, our findings demonstrate that sFRP2 excreted by fibroblasts can initiate the scar formation in the uterus via activating non-canonical Wnt signaling and increasing Ca 2+ influx in UFIBs. sFRP2 may function as a therapeutic target for the prophylactic treatment of US, while site-specific inhibition of sFRP2 via LNP-mediated siRNA delivery represents a promising and translational strategy for precision prevention of US and improving reproductive health of women with US. Our findings also provide valuable insights for the development of targeted therapeutic interventions for other diseases associated with excessive scarring or fibrotic conditions.

Uterine scarring (US), characterized by the formation of scar tissues on the uterus, can be caused by various factors, such as uterine surgeries and infections. In particular, cesarean sections are a common cause of US 1 , 2 , with the global increase in cesarean section rates contributing to a higher prevalence of US among women 3 . Other surgical interventions that can lead to US include myomectomy, dilatation and curettage, hysteroscopic resection, and uterine malformation correction 4 , 5 . Depending on the extent and location of US, its symptoms include pelvic pain and menstrual irregularities 6 . In addition, US is an increasing risk factor for many adverse pregnant consequences that threaten fetal-maternal health, such as infertility, placental abnormalities, ectopic pregnancy, uterine rupture, intrauterine growth restriction, recurrent miscarriages, preterm birth, postpartum hemorrhage, and even fetal death 5 , 7 – 9 . Currently, there are no pharmaceutical strategies available to effectively address the issue of collagen deposition in uterine scars. A deep understanding of the pathophysiological process and underlying cellular and molecular signaling pathways involved in US is essential for the development of effective therapies for this condition. A few available studies have indicated that uterine surgery can lead to the continuous deposition of collagen 10 , 11 , which in turn stimulates the proliferation and differentiation of fibroblasts in the uterus, ultimately resulting in the formation of uterine scars 10 . Moreover, the intricate and extensive interactions of specific growth factors and cytokines during abnormal wound healing play a critical role in the development of US 10 , 12 . For instance, the upregulation of transforming growth factor (TGF)-β, vascular endothelial growth factor (VEGF), and platelet-derived growth factor (PDGF) in fibroblasts can promote fibrosis and scarring 13 , 14 . Similarly, overexpression of tumor necrosis factor (TNF)-α and interleukin (IL)−6 also contributes to the scarring process in the uterus 10 . However, the detailed pathological features and key molecular factors responsible for scar implantation in the uterus remain largely unclear. Recently, emerging evidence has demonstrated that secreted frizzled-related protein 2 (sFRP2), a key member of the sFRP family, serves as a critical mediator of fibrotic progression in multiple tissues, including the cardiac, pulmonary, and renal systems 15 – 17 . However, the pathological significance of sFRP2 in US remains unclear. In this study, we found a significant upregulation of sFRP2 in human and mouse uterine tissues with US as well as in isolated uterine fibroblasts (UFIBs). Overexpression of sFRP2 in the normal uterus induced US features in female mice and exacerbated surgery-induced US in mice as well. Human and mouse UFIBs with overexpressed sFRP2 also displayed characteristics of fibrotic transformation. By contrast, knockdown of sFRP2 in the mouse uterus markedly suppressed uterine scar development and scarring features in UFIBs. Further analyses revealed that sFRP2 mediates scar development in the uterus by regulating non-canonical Wnt/Ca 2+ signaling. Accordingly, sFRP2 was explored as a potential therapeutic target for preventing US. We successfully screened and formulated effective siRNA targeting sFRP2 into lipid nanoparticles. In a mouse model of uterine surgical scarring, the siRNA/lipid nanoparticles demonstrated strong efficacy in reducing sFRP2 expression and preventing uterine scar formation, thereby alleviating adverse pregnancy outcomes caused by US.

Source Data Source Data