In this study, all samples were obtained from the international maternity and child health hospital of Medical school, Shanghai Jiaotong University, Shanghai, China. The detailed patients and sample information can be found in Table S1.Informed consent from patients were obtained before this work. All procedures involved were approved by the Institutional Review Board of international maternity and child health hospital.

All endometrial tissue samples were processed independently using an identical protocol. In brief, the tissues were minced into small fragments, each measuring less than 1 mm in diameter. These fragments underwent enzymatic digestion with collagenase I (1 mg/ml; Thermo Fisher) and deoxyribonuclease I (0.2 mg/ml; Thermo Fisher) for 50 minutes on a rotator at 37 °C. The resulting digested mixture was filtered through a 70-μm cell strainer to isolate dissociated cells. Following this, the supernatant was discarded after centrifugation at 300g for 10 minutes, and the pellet was resuspended in 1 mL of red blood cell lysis buffer (Thermo Fisher) and incubated at 4 °C for 10 minutes to eliminate intact red blood cells. To further purify the single-cell suspension, a Dead Cell Removal Kit (Miltenyi) was employed to eliminate dead cells. The resultant pellet was then resuspended in 5 mL of calcium- and magnesium-free PBS containing 0.04% weight/volume BSA. Finally, 10 µl of this suspension was counted using a hemocytometer under an inverted microscope, with Trypan blue utilized to distinguish live cells. After confirming the quality of the single-cell suspensions, three samples from each group were selected for 10x Genomic sequencing, given the high costs associated with single-cell RNA sequencing.

Six samples were processed using the 10x Genomics Chromium platform , as detailed in Table S1. The freshly prepared single-cell suspension was diluted to a concentration of 700-1200 cells per μl, following the guidelines provided in the 10x Genomics Chromium Next GEM Single Cell 3’ Reagent Kits v3.1 (No. 1000268) for computational analysis and library construction. The resulting library was sequenced on the Illumina Nova 6000 PE150 platform.

Furthermore, eight samples were examined with the BD Rhapsody™ platform, as outlined in Table S1. In accordance with the manufacturer’s guidelines, the single-cell suspension was modified to the correct volume for processing and capture using the BD Rhapsody™ Enhanced Cartridge Reagent Kit (Cat. No. 664887) along with the BD Rhapsody™ Cartridge Kit (Cat. No. 633733). Reverse transcription was carried out utilizing the BD Rhapsody™ cDNA kit (Cat. No. 633773), and the BD Rhapsody™ WTA Amplification Kit (Cat. No. 633801) was employed for constructing the DNA library. High-throughput sequencing was performed in PE-150 mode.

The FASTQ files from 10xGenomics Chromium platform were initially analyzed using the 10xGenomics Chromium pipeline, employing the default parameters from the Cell Ranger single-cell software suite (v6.1.2) (10xGenomics) (accessible at https://www.10xgenomics.com/support/software/cell-ranger/latest). Following this, the FASTQ files from the BD Rhapsody™ platform were processed via the BD Rhapsody WTA analysis pipeline, utilizing the GRCh38 human genome as a reference (Seven Bridges, Charlestown, Mass). The resulting data were subsequently evaluated using SeqGeq 1.8.0 (BD Biosciences).

The single-cell RNA-seq data were analyzed using the Seurat R package (version 4.0.4). To exclude doublets, the R package DoubletFinder was employed. After conducting quality control, a total of 103,539 cells were retained for subsequent bioinformatics analyses. To mitigate batch effects between samples and experiments, integration was performed using the fastMNN method. Dimensionality reduction was achieved with the “RunUMAP” function, allowing visualization through Uniform Manifold Approximation and Projection (UMAP). For subgroup clustering, cells of various types were separately extracted and clustered based on their first 22 clusters, using a resolution of 0.5. The Wilcoxon rank-sum test, executed with the “FindAllMarkers” function, identified markers for each cluster. Only genes with |avg_log2FC| > 0.25 and p_val_adj < 0.05 were considered significant marker genes. The marker genes for each cluster are detailed in Supplementary Table S2.

To identify the genes responsible for endometrial polyps, we compare the expression profiles between 5 endometrial polyps and 5 eutopic endometrium. Adapter trimming and quality control were automated using Trim Galore (version 0.6.7). For genomic alignment, Hisat2 (version 2.1.0) was utilized to map sequences to the human reference genome (GRCh38). The read counts for each gene were calculated using featureCounts (version 1.6.3). Differentially expressed genes between the samples were determined using DEGseq2, applying thresholds of |Log2-fold change| > 1 and a p-value < 0.05.

To explore enhanced cell-cell communication between mast cells and other cells, we utilized CellPhoneDB, a comprehensive tool designed to identify and visualize cell-type interactions based on signaling characteristics. we employed the METHOD 3: Retrieval of Differentially Expressed Interactions available in CellPhoneDB version 3¹⁶ , which leverages differentially expressed genes to extract and highlight enhanced cell communication.

To investigate the role of WT1 in mast cell proliferation, we performed siRNA-mediated knockdown of WT1 in the HMC-1 mast cell line. The experiments were conducted in triplicate, with three samples undergoing WT1 knockdown and three control samples treated with a non-targeting siRNA. RNA sequencing was carried out to assess changes in gene expression 24 hours post-transfection. Following transfection, cell counts were measured at 0, 24, and 48 hours using CounterSTAR (ALIT Biotech, China).

To identify differentially expressed genes (DEGs), we utilized the “FindMarkers” function in Seurat. The log fold change (Log2FC) and the adjusted p-values for each DEG were evaluated through ROC analysis. Only genes that met the criteria of |avg_logFC| > 0.25 and FDR (false discovery rate) indicated by p_val_adj < 0.05 were classified as DEGs.

The transcriptional regulatory network was examined using the GENIE3 (version 1.10.0) and RcisTarget (version 1.8.0) R packages as part of the SCENIC workflow (version 1.1.2.1), applying the default settings (28). Reference transcription factors (TFs) corresponding to hg19 were obtained via RcisTarget. The gene expression data from 31,503 cells were normalized using Seurat and served as input.Initially, co-expression modules were identified by analyzing the relationships between transcription factors and potential target genes based on the gene expression data through GENIE3. Following this, cis-regulatory motif enrichment analysis was conducted on all potential target genes within each co-expression module using RcisTarget. Only those target genes that were enriched with the specific motifs of their corresponding TFs were designated as direct target genes. Each transcription factor, along with its direct target genes, was categorized as a regulon. Finally, the gene regulatory interactions between fibroblast and epithelial cells were inferred using the Seurat FindAllMarkers function.

It is was performed following a previous study¹⁷ , For pathway enrichment analysis and visualization of differentially expressed genes (DEGs), we followed a comprehensive protocol outlined by Reimand et al18. and utilized Metascape¹⁹ . The initial step involved using g:Profiler to conduct pathway enrichment analysis on the DEG list. Subsequently, the results were visualized through EnrichmentMap (version 3.5.0) within Cytoscape (version 3.10.1). To facilitate the navigation and interpretation of the pathway enrichment outcomes, we employed AutoAnnotate (version 1.3.2) in Cytoscape. Metascape serves as an online platform that offers extensive resources for the annotation and analysis of gene lists.

Fresh human specimens were fixed in a 4% paraformaldehyde solution for 24 hours, subsequently embedded in paraffin, and sectioned into 5 μm slices. These sections underwent deparaffinization in xylene and rehydration in ethanol. Antigen retrieval was conducted using Tris-EDTA (pH 9.0) (Biosharp, Anhui, China) in a microwave. For the subsequent procedures, a staining kit from Absin (Shanghai, China) was utilized.In brief, the slides were treated with 3% H2O2 to inhibit endogenous peroxidase activity, followed by blocking with 5% BSA. The sections were then incubated overnight at 4 °C with primary antibodies targeting CD63 (1:200) and TPSB2 (1:200). This was followed by incubation with a secondary antibody. Staining was carried out using DAB and hematoxylin. Observations and imaging of the slides were performed with a microscope (Leica, Germany).

Paraffin sections with a thickness of 7 μm were prepared. Following dewaxing and rehydration, heat-mediated antigen retrieval was conducted using a citrate-based solution (pH 6.0). The sections were subsequently permeabilized in PBS supplemented with 0.4% Triton X-100 for 30 minutes at room temperature. After washing with PBS, the samples were then blocked with 5% bovine serum albumin (BSA) at room temperature for 1 hour. Following this, they were incubated overnight at 4 with primary antibodies against trytase (1:100, Abcam, ab8978, Cambridge, UK),. After washing with PBS, the samples were treated with 555-conjugated secondary antibodies (Invitrogen, California, A-31572, USA) for 1 hour in the dark at room temperature, then stained with 4,6-diamidino-2-phenylindole (DAPI). Immunofluorescence imaging was carried out using a confocal microscope (Leica, Germany).

Cell distribution between the two groups was analyzed using unpaired two-tailed Wilcoxon rank-sum tests. For comparing gene expression or gene signatures between the two cell groups, unpaired two-tailed Student’s t-tests were employed. All statistical analyses and visualizations were conducted using R. The specific statistical tests applied in the figures are detailed in the figure legends, with statistical significance defined as p < 0.05.

We performed single-cell RNA sequencing to construct a global endometrial polyps (EPs) atlas, which profiled both ectopic endometrial and adjacent eutopic endometrium tissues (EU) from seven patients with endometrial polyps (Fig.1A). In sections stained with hematoxylin and eosin (H&E), we noted a twofold increase in the cross-sectional area of the glandular epithelium in the MP group compared to the EU group (Fig 1B), After rigorous quality control, a total of 103,539 cells were identified, including 52974 cells from eutopic endometrium and 50565 cells from the polyps, respectively. Average 2840 genes and 8914 unique molecular identifiers (UMIs) were obtained per cell, indicating sufficient coverage and transcripts representative(Fig. S1). Next, we performed unsupervised clustering analysis and genomic signature identification, 8 clusters were defined in the endometrium (Fig.1C and D), including epithelial cells (9.3 %), fibroblasts (67.1%), smooth muscle cells (SMCs) (2.5 %), endothelial cells (2.1 %), macrophages (3.8 %), NK cells (6.8 %), mast cells (1.1%), and CD8 T cells (7.2%). The number for each cell type in endometrial samples is shown in Table1. The correlation between these cell types confirms their identification (Fig. S2).

Next, we examined the changes in cell composition associated with endometrial polyps and discovered a reduction in the proportion of fibroblasts within these polyps, while the proportions of epithelial cells (p=0.049), mast cells (p=0.002), and macrophages (p=0.017) increased (Fig. 1E). To validate these changes in cell composition within the polyps, we performed immunohistochemistry staining using cell-specific markers. Indeed, we observed that the numbers of CD68-positive macrophages and TPSB2-positive mast cells in the polyps were approximately twice as high as those in the eutopic endometrium (Fig. 1F and G), consistent with previous reports , . Moreover, the levels of Ki67 in the glandular epithelium of the polyps were elevated, corroborating the increase in epithelial cells we had observed (Fig. 1H). Collectively, these observations suggest glandular proliferation and immune cell infiltration, including macrophages and mast cells, in the polyps, likely contributing to the proliferative and inflammatory alterations observed.

(C) Uniform manifold approximation and projection (UMAP) plot showing the eight cell types of endometrial polyps.

(D) Dot plot showing the expression of representative genes for each cell type.

(E) Bar graph showing cell proportions (%) of various cell types in EU (solid red) and EP (light red) groups, with significant differences indicated by p-values.

(F) Immunohistochemical staining for cell proliferation marker Ki67 in EU and EP. Scale bars, 60 μm and 5 μm (zoomed-in images). Left, representative images. Right, the proportion of Ki67 positive cells was quantiffed.

(G) Immunohistochemical staining for macrophage marker CD63 in EU and EP. Scale bars, 60 μm and 5 μm (zoomed-in images). Left, representative images. Right, the proportion of CD63 positive cells was quantified.

(H) Immunohistochemical staining for mast cell marker TPSB2 in EU and EP. Scale bars, 60 μm and 5 μm (zoomed-in images). Left, representative images. Right, the proportion of TPSB2 positive cells was quantiffed.

To simultaneously define different gene expressions at the global and cellular levels, bulk RNA-seq was performed parallel for endometrial polyps and eutopic endometrium (Fig.2A and 2B). 94 upregulated and 16 downregulated genes were detected in endometrial polyps (log2FoldChange >1 & pvalue < 0.05 Supplemental Table2). Next, the GO analysis was performed. The results showed that regulation of inflammatory response, negative regulation of cell activation, and cellular lipid catabolic process associated signal pathway were activated in endometrial polyps. However, the genes involved in the proteoglycan metabolic process, neuron migration, and gas homeostasis were downregulated (Fig.2C).

We assessed alterations in the expression of inflammation-associated genes within the cellular layers of endometrial polyps to explore particular anomalies in molecular expression across various cell types. Notably, CD163, which was found to be upregulated in the bulk RNA-seq analysis, showed increased expression specifically in macrophages, while its expression remained unchanged in other cell types. Similarly, CST7, also upregulated in the bulk RNA-seq analysis, exhibited increased expression in mast cells, with no significant changes observed in other cells. Conversely, S100A9 and LPL, both upregulated in the bulk RNA-seq analysis, were found to be downregulated in macrophages and smooth muscle cells (SMCs) but remained unchanged in most other cell types (Fig.2D and 2E). These differences reflect the cellular heterogeneity in gene expression changes. Collectively, our findings highlight the complex and cell type-specific molecular landscape within endometrial polyps.