Integrated analysis of single-cell and bulk transcriptomic data reveals altered cellular composition and predictive cell types in endometriosis

Integrated analysis of single-cell and bulk transcriptomic data reveals altered cellular composition and predictive cell types in endometriosis | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Integrated analysis of single-cell and bulk transcriptomic data reveals altered cellular composition and predictive cell types in endometriosis Meihong Chen, Yan Shen, Guanqun Jiang, Liqun Wang, Qi Chen This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5277508/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The early diagnosis of endometriosis is delayed and the clinical treatment is difficult, which causes severe economic burden to the patients. A comprehensive study of the cell classification and composition of endometriosis, is essential for early diagnosis and pathogenesis. This study utilized single-cell data as a reference to accurately deconvolute cell types within bulk transcriptomics data. Five main cell types in endometriosis (epithelial cells, mesenchymal cells, endothelial cells, lymphocytes, and myeloid cells) revealed varying degrees of change compared to healthy controls. Notably, epithelial cells significantly decreased while mesenchymal cells increased. However, MUC5B epithelial cells, showed an increasing trend, consistent with the increasing trend of dStromal-late mesenchymal cells and eM2 macrophages.Pathway enrichment analysis revealed the EMT signalling pathway dominated in endometriosis. Additionally, a random forest method based on cell types could successfully distinguish endometriosis patients from healthy control, which illustrates the potential value of cell types in early diagnosis. Biological sciences/Biological techniques/Bioinformatics Health sciences/Diseases/Reproductive disorders/Urogenital reproductive disorders endometriosis CIBERSORTx scRNA-seq Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. Introduction Endometriosis is a common and challenging gynaecological condition characterized primarily by dysmenorrhea, chronic pelvic pain, infertility, etc 1,2 . The rising prevalence of endometriosis, along with high recurrence rates and the challenges of clinical management, significantly impacts women's quality of life and increases the burden on healthcare resources 3 . Moreover, patients with endometriosis frequently experience delayed diagnosis, with an average of 6.7 years and up to 4-11 years elapsing from symptom onset to pathological histological diagnosis following laparoscopic surgery 4,5 . Single-cell RNA sequencing (scRNA-seq) uncover detailed insights into microenvironment heterogeneity, functional differentiation and cellular interactions 6 . However, the cost-intensive and limited access to high-quality specimens hinder the widespread application of single-cell analysis. In contrast, using deconvolution methods to estimate cell populations in bulk transcriptomic data effectively addresses the disadvantages of single-cell data, providing a faster and more cost-effective approach for early disease research 7 . In this study, deconvolution results of bulk transcriptomics data display dramatic changes in cell types between ectopic endometriosis and normal endometrial tissue. Through pathway analysis, our data detail the molecular functions of each cell types and how they may promote the progression of endometriosis. Furthermore, the successful construction of the predictive model highlights the unique importance of cell types in the clinical application. 2. Results 2.1 Overall experimental design and bulk microarray database analysis Following overall design (Figure 1A), a total of five microarray datasets (GSE11691 8 , GSE7305 9 , GSE12768 10 , GSE25628 11 , and GSE5981 12 ) and one single-cell dataset (GSE179640 13 ) were included in this study. The analysis comprised 201 microarray samples, consisting of 96 healthy control samples and 105 endometriosis samples. After removing batch effects between different datasets (Sup Figure 1A), PCA analysis revealed that the healthy control and endometriosis groups formed two distinct clusters (Figure 1B), indicating differing molecular characteristics between the two groups. Subsequent differentially expressed gene analysis identified 115 significantly upregulated and 677 significantly downregulated genes (adjusted p-value 0.5, Figure 1C). Consistent with our previous research, the expression level of FXYD1 was significantly increased in endometriosis (log2Fold Change > 1) 14 . ELN as a key component of the extracellular matrix, was also upregulated in patients with endometriosis, potentially participating in the regulatory networks during the secretory phase 15 . Pathway analysis indicated a marked activation of the EMT pathway in endometriosis patients. Additionally, we observed an enrichment of several other pathways in these patients, including myogenesis, TNFA signalling, estrogen response, extracellular matrix dynamics, and immune effector processes (Figure 1D). Conversely, the downregulated differential genes were enriched in pathways such as E2F targets, G2M checkpoint, and MYC targets (Figure 1E). This further elucidated the distinctive features between patients and healthy individuals, aligning with the results of previous research 16 . 2.2 Deconvolution analysis revealed single-cell populations changes in endometriosis The cell composition between the endometriosis and healthy control groups varied dramatically. Using single-cell data for deconvolution of bulk data effectively provided cell proportions across numerous samples. Eight endometriosis samples from the single-cell database GSE179640 were selected and 24,438 cells passed quality control (Sup Figure 2A). Ultimately, we identified five major cell types: epithelial cells, mesenchymal cells, endothelial cells, lymphoid cells, and myeloid cells (Figure 2A & Sup Figure 2B) and 52 distinct cell subtypes (Figure 2B & Sup Figure 2C, Sup Table 1). Through the signature construction and bulk dissection (Sup Figure 3), the proportion of epithelial and endothelial cells was significantly decreased in endometriosis compared to the healthy control group (p-value = 1.4E-4), while the proportions of mesenchymal, myeloid and lymphoid exhibited varying degrees of increase (Figure 2C). Of note, most epithelial cell subtypes (SOX9_I, SOX9_f_II, Gla_s, and Cil) exhibited a notable decline in proportion within the endometriosis samples (Figure 3A). In contrast, the proportions of MUC5B, Lum, and KRT cells displayed a significant rise. Among the various mesenchymal cell subtypes, all demonstrated substantial upregulation except for the eSt cycling subtype, which exhibited a marked decrease in abundance. The dSt_early subtype showed a slight decrease (p-value = 0.33), while the dSt_m and dSt_l subtypes displayed significant increases (p-value = 3.8E-2 and p-value = 8.4E-5, respectively). Importantly, the dSt_l subtype became the most prevalent subtype among mesenchymal cells (Figure 3B). Although the overall proportions of immune and endothelial cells were relatively low, a significant grow in the proportions of eM2 and mast subtype cells was observed (Figure 3C-E). These results indicate that the composition of the cell population in endometriosis has changed significantly, reflecting disease-related tissue remodelling and alterations in the immune microenvironment. 2.3 Biological functions of various cell subtypes To further understand the functional roles of these significantly altered cell subtypes in endometriosis, we analysed the pivot cell markers within each subtype. The MUC5B cell subtype was distinguished by high expression of MUC5B, LTF, TFF3, and SAA1, while showing low expression of MT family genes (MT1H, MT1G, and MT1M) (Figure 2B & Figure 4A). Hallmark pathway analysis revealed that the genes prominently expressed in the MUC5B subtype were significantly enriched in epithelial mesenchymal transition (EMT), estrogen response, coagulation, KRAS signalling up, and interferon gamma response pathways (Figure 4B). The dStromal late mesenchymal cell subtype was characterized by evaluated expression of LEFTY2, ACTA2, EGR1, FOS, CXCL8, and CXCL2, and low expression of MMP7 and SCGB1D2 (Figure 2B & Figure 4C). This subtype exhibited the highest expression levels in pathways such as TNFA signalling via NFKB, hypoxia, EMT, MAPK signalling pathway, focal adhesion, and inflammatory response (Figure 4D). The eM2 cell subtype, identified as tissue-resident macrophages, showed high expression of markers such as FOLR2 and LYVE1 (Figure 4E). Enrichment analysis using Hallmark, KEGG, and GO pathways revealed significant enrichment in complement, KRAS signalling up, lysosome, and positive regulation of immune response (Figure 4F). 2.4 Integration of single-cell and bulk microarray reveals intersection genes and pathways We performed an overlap analysis between the marker genes of the three cell subtypes and the DEGs from bulk microarray data. For the MUC5B epithelial cell subtype, 21 out of 1,205 marker genes overlapped with the 114 upregulated DEGs from the microarray, including key genes such as S100A9, TFF3, and TLE2 (Figure 5A & Figure 5B). In the case of the dStromal late cell, 28 out of 779 marker genes were found to intersect with the 114 upregulated DEGs from the microarray, highlighting significant genes such as ACTA2, FOSB and EGR1 (Figure 5C & Figure 5D). For the myeloid eM2 subtype, 12 out of 612 marker genes displayed overlap with the 114 upregulated DEGs from the microarray, with notable genes like EPHX1, TYROBP and SOD3 (Figure 5E & Figure 5F). Subsequent pathway enrichment analysis of these intersecting genes revealed that the predominant signalling pathways included EMT, P53 pathway, positive regulation of cell migration, inflammatory response, and complement and coagulation cascades (Figure 5G & Sup figure 4). 2.5 Establishment of an early prediction model for endometriosis At first, we systematically examined published predictive markers of endometriosis, including downregulated genes such as BAX, FAS, and ESR1 17-19 , and upregulated genes like ESR2, PPARG, and ACTA2 19-21 (Figure 6A). Consistent with previous studies, these genes exhibited notable expression abnormalities in endometriosis. Regarding of the distinct cell composition and unique molecular characteristics, we hypothesis predictive model based on cell percentages could accurately distinguish endometriosis from healthy controls. Using 70% of the samples for training, our random forest model successfully identified 29 out of 31 endometriosis cases and 22 out of 28 healthy controls in the testing dataset (Table 1), yielding an overall AUC value of 0.932 (Figure 6B). Notably, MUC5B epithelial cells had the highest contribution to the model. Five of the top ten contributing cell types were mesenchymal subtypes, including HOX, ePV_1a, eSt_c, Fib, and dSt_l (Table 2), demonstrating the importance of mesenchymal cells in distinguishing endometriosis. In addition, myeloid cells (mast cells and eM2 macrophages) were also pivotal to the model performance. Table 1 Prediction result of the random forest model in the testing dataset Table 2 Top 15 most important cell types in the prediction model FALSE TRUE MeanDecreaseAccuracy MeanDecreaseGini MUC 6.349389 10.18686741 11.19534639 3.509903431 HOX 7.005972 5.959809913 9.309629485 3.265530055 Mas 8.554349 3.329296972 8.282943336 3.582387852 ePV_1a 5.174236 6.449464636 7.897985966 2.91620666 eSt_c 6.13349 6.755029796 7.876833257 1.506470136 Fib 6.517639 3.510529826 7.091543151 2.099887649 KRT 3.913174 6.113008499 6.987720918 2.35424138 dSt_l 1.851104 8.189318136 6.722879285 2.21917295 Art 6.928125 1.614410864 6.541262998 2.433950724 eM2 5.656597 3.722970618 6.181275007 2.509147755 Gla_s 5.956862 1.099191016 5.149002227 1.683667328 SOX9_f_I 2.680308 4.268303211 5.031055466 1.247652545 pDC 2.945008 4.405847704 4.648788358 2.243083172 Pla_B -0.41174 5.913050786 4.41702008 1.720396078 Mon 0.720371 4.855466586 3.683643304 2.517322473 3. Discussion Although endometriosis is a benign disease, it has malignant behaviours such as proliferation, distant metastasis, and invasive behaviours. Patients primarily experience symptoms such as dysmenorrhea, chronic pelvic pain, and infertility 22 , and may even be at risk for malignant tumors, including ovarian cancer 23 . At present, there are various theories regarding the pathogenesis of endometriosis: transvascular reflux theory, body cavity epidermal metaphysiology, lymphatic and vein disseminate theory, and genetic immune theory, etc 24-26 . However, none of these theories can explicitly explain the occurrence of endometriosis. The development of endometriosis is not caused by a single factor 27 , but is influenced by multiple factors, such as the body's immune status 28 , inflammatory response 29 , angiogenesis 30 , and local hormone levels 31 . Therefore, exploring cell composition and subtype characteristics is crucial for studying the pathogenesis of endometriosis. To better understand the differences in cell subtype proportions in endometriosis, we integrated bulk microarray data related to endometriosis from public databases. Using the latest single-cell atlas of endometriosis and the deconvolution software, we revealed variations in five major cell types (epithelial cells, mesenchymal cells, endothelial cells, lymphatic cells, and myeloid cells). Although the proportion of epithelial cells significantly decreased, the proportion of MUC5B cells substantially increased. Notably, these cells exhibited higher expression of TFF3, which has been confirmed to be associated with inflammation 32 . Additionally, MUC5B-specific genes such as S100A9 and TIMP1 were also remarkably overexpressed at the bulk level. S100A9, a member of the S100 protein family, regulated cell migration, promoted the production of inflammatory factors and mediated immune responses 33 . Studies have found that S100A9 was highly expressed in endometriosis and may be involved in the recurrence mechanism of ovarian endometriomas 34 . The study found that mesenchymal cells was profoundly overexpressed in ectopic endometriosis, with the dStromal late subtype particularly elevated. These cells were involved in inflammatory response, cell adhesion, and angiogenesis. Key genes, including EGR1, bound to SNAY2 promoters to inhibit E-cadherin and promote metastasis. CXCL8 played a crucial role by binding to CXCR1 and activating the PTEN/AKT pathway, thereby promoting proliferation and inhibiting apoptosis in endometriosis cells 35 . Additionally, ACTA2, also known as alpha-smooth muscle actin (α-SMA), served as a marker for myofibroblasts associated with fibrosis in endometriosis 36 . Extensive research has demonstrated that α-SMA was significantly upregulated in endometriosis. Multiple factors contributed to its increased expression, which ultimately leaded to the development of fibrosis in endometriosis 37,38 . Macrophages were broadly classified into two main phenotypes: eM1 and eM2 macrophages. Recent studies have demonstrated that serum from women in endometriosis has the capacity to polarize macrophages towards both eM1 and eM2 phenotypes 39 . Similarly, our study also found elevated percentage of eM2 in ectopic endometrial tissues. Furthermore, we have observed that the physiological function of these eM2 cells positively regulated immune responses in the context of endometriosis. The presence of the EMT signalling pathway in all enrichment analyses revealed its predominant role in endometriosis. Numerous researchers have found that factors such as hypoxia, estrogen stimulation, and WNT4 may trigger EMT in endometriosis 40,41 . In the MUC5B epithelial cell subtype, S100A9 had been found to activate AKT1, leading to EMT in invasive pituitary adenomas 42 . However, the applicability of this mechanism to endometriosis required further investigation. The local inflammatory microenvironment was a hallmark characteristic of endometriosis, sustained by the synergistic activation of hormones and immune factors in ectopic endometrial tissue 43 . In this context, complement activation emerged as a crucial initiator of inflammatory cascade reactions. Core complement genes, including C3, CFH and CLU, were integral to the complement activation process. This activation modulated macrophages and mast cells, leading to the production of various inflammatory mediators and the recruitment of inflammatory cells, thereby amplifying the inflammatory response 44 . The upregulation of genes such as ACTA2, MYH9 and MYLK may indicate that ectopic lesions underwent repeated cycles of tissue damage and repaired due to recurrent bleeding and inflammation. These processes, facilitated by EMT and fibroblast-to-myofibroblast trans-differentiation, resulted in cellular contraction, excessive activation of cell migration, smooth muscle metaplasia, and fibrosis 45 . At present, the diagnostic markers of endometriosis were mainly based on serum, urine, peritoneal fluid 46 . Even when tissue samples were applied in diagnostic modelling, the focus was primarily on gene expression data rather than cellular composition. Our study showed that early diagnostic models based on cell types can be used to predict disease states of endometriosis successfully. In our model, epithelial cells, mesenchymal cells, and macrophages emerged as critical components. Notably, MUC5B epithelial cells displayed the largest contribution to diagnosing endometriosis. MUC5B cells, considered epithelial progenitor cells, played essential roles in tissue repair, regeneration, and cancer development. Tan successfully captured the tissue separation of MUC5B cells, but it remained unclear how this cell subgroup promoted endometriosis 13 . Mesenchymal cells accounted for the largest proportion of the top ten contributors, among which dStromal late cells were one of the key cells. Interestingly, these mesenchymal cells did not exhibit somatic mutations but display specific epigenetic abnormalities in the expression of key transcription factors 47 . Moreover, too many macrophages stimulated cytotoxic T helper cells to release inflammatory cytokines, leading to a endometritis environment that promoted endometriosis 48 , eM2 macrophages were found to be elevated in type III-IV endometriosis 49 , We also found that eM2 was the main factor in the early diagnosis of endometriosis. Additionally, several studies had reported an increased number of mast cells in endometriosis. Estrogen and stem cell factors provided the ideal microenvironment for their recruitment and differentiation. In turn, mast cells may release pro-inflammatory mediators that contribute to chronic pelvic pain and the progression of endometriosis 50 . 4. Methods 4.1 Collection and Preprocessing of public bulk transcriptomics datasets We conducted a comprehensive search in the Gene Expression Omnibus (GEO) database using the keyword "endometriosis" and release date before 29/02/2024 to collect bulk transcriptomics datasets. Seven datasets (GSE11691, GSE7305, GSE12768, GSE201912, GSE168902, GSE25628, and GSE51981) were identified. We then filtered the RNA-seq dataset GSE168902 to ensure consistency and comparability during data integration. Additionally, GSE201912 was excluded due to a minimal number of overlapping genes with the other datasets. For datasets generated from the Affymetrix platform, raw CEL files were downloaded and normalised using the rma function from the affy (v1.66.0) or oligo (v3.11) packages. For the GSE12768 dataset generated from the Cochin platform, we obtained the normalised data using the getGEO function from the GEOquery package. Probe IDs from the microarray were converted to gene symbols based on the corresponding GPL annotation files provided in GEO. Probes corresponding to multiple gene symbols were discarded. In contrast, genes corresponding to multiple probes was taken the maximum expression values. After normalizing each dataset individually, we integrated them into a merged dataset based on gene symbols. We employed the ComBat empirical Bayes batch correction algorithm from the sva package to remove batch effects between different datasets. Finally, we performed PCA analysis using factoextra to reduce the dimensionality of the molecular information from each sample for visualization. 4.2 Collection and preprocessing of scRNA-seq raw data The processed matrix data of Magda’s endometriosis atlas was downloaded from https://www.reproductivecellatlas.org/endometrium_reference.html. The single-cell RNA sequencing dataset for endometriosis (GSE179640) was downloaded from the GEO database and processed using the scanpy package 51 . Low-quality cells were filtered out following Magda's description 52 . The reference mapping tool scvi was used to integrate GSE179640 to the endometriosis atlas and automatically label the cell types annotation 53 . Additionally, normalization, dimensionality reduction, and clustering were performed with default parameter of scanpy. The clustering results were manually annotated for cell types using marker genes from the endometriosis atlas 4.3 Identification of differentially expressed genes and significant cell markers For the bulk transcriptomics dataset after batch effect removal, we constructed a design matrix comparing endometriotic tissue versus healthy tissue and performed differential gene analysis using the Limma package. Genes with an absolute log fold change (LogFC) > 0.5 and adjusted p-values < 0.05 were considered differentially expressed. For significant cell markers in the single-cell dataset, we used the FindAllMarkers function from the Seurat package to compare different cell subtypes within the same major cell type. The parameters were set to logfc.threshold = 0 and min.pct = 0.1. For the MUC5B, dStromal_late, and eM2 subtypes, hundreds of significant cell markers were identified using adjusted p-values 1, 0.5, and 0.1, respectively. 4.4 Pathways analysis Differentially expressed genes and cell markers were uploaded to the Metascape website for pathway analysis using the following parameters: a minimum of 3 overlapping genes, p < 0.05, and a minimum enrichment factor of 1.5 54 . The analysis included databases such as GO-BP, GO-CC, GO-MF, HALLMARK, and KEGG. Only pathways with adjusted p-values below 0.05 were considered significantly enriched. 4.5 CIBERSORTx deconvolution analysis We first randomly selected 1,000 cells from each cell type in GSE179640(or all available cells if fewer than 1,000) to construct a raw expression matrix. Total-count normalization was applied to standardize each cell to a library size of 10,000 reads. The normalized expression matrix was then uploaded to the CIBERSORTx cloud platform. Subsequently, we utilized the “Create Signature Matrix” feature with default parameters to build the single-cell signature matrix. The batch-corrected microarray expression matrix was also uploaded to the CIBERSORTx website. Finally, “Impute Cell Fractions” function was applied to estimate the proportions of different cell types in each bulk sample. We employed CIBERSORTx “Batch Correction Mode” to account for technical differences between the bulk and single-cell platforms. 4.6 Differentially expressed cell types We visualized the CIBERSORTx analysis results using the ggviolin function from the ggpubr package. The Wilcoxon signed-rank test was performed to compare the proportions of the same cell types between the healthy group and the endometriosis group. Cells with a p-value less than 0.05 were considered differentially expressed. 4.7 Diagnostic model construction The collected bulk microarray samples were randomly divided into training and testing sets in a 7:3 ratio using the caret package. A classification model was developed using the randomForest package with the proportions of various cell subtypes as input features and disease status as the prediction target. The number of trees was set to 1,000 for model construction. The model's performance was evaluated based on accuracy and the area under the ROC curve (AUC) of the testing dataset. Declarations Data availability Data and materials from this study are available from the corresponding author on reasonable request. Acknowledgements This research was funded by National Natural Science Foundation of China (81960276). Authors’contributions Conceptualisation: M.C., L.W. and Q.C.; Methodology: M.C., Y.S. and G.J.; Formal analysis: M.C. and G.J.; Writing: M.C., Y.S. and Q.C.; Supervision: L.W. and Q.C. Competing interests The authors declare that they have no competing interests Consent to publish Not applicable. Correspondence and requests for materials should be addressed to Q.C. or L.W. References Zondervan, K. T., Becker, C. M. & Missmer, S. A. Endometriosis. The New England journal of medicine , 382 (2020). Saunders, P. T. K. & Horne, A. W. Endometriosis: Etiology, pathobiology, and therapeutic prospects. Cell 184 , 2807-2824 (2021). Missmer, S. A. et al. Impact of endometriosis on women's life decisions and goal attainment: a cross-sectional survey of members of an online patient community. BMJ open 12 , e052765 (2022). Agarwal, S. K., Chapron, C., Giudice, L. C., Laufer, M. R. & Taylor, H. S. Clinical diagnosis of endometriosis: a call to action. American Journal of Obstetrics and Gynecology 220 (2019). Anastasiu, C. V., Moga, M. A., Neculau, A. E., Blan, A. & Chicea, L. M. Biomarkers for the Noninvasive Diagnosis of Endometriosis: State of the Art and Future Perspectives. International Journal of Molecular Sciences 21 , 1750 (2020). Cheng, C., Chen, W., Jin, H. & Chen, X. A Review of Single-Cell RNA-Seq Annotation, Integration, and Cell-Cell Communication. Cells 12 (2023). https://doi.org:10.3390/cells12151970 Nguyen, H., Nguyen, H., Tran, D., Draghici, S. & Nguyen, T. Fourteen years of cellular deconvolution: methodology, applications, technical evaluation and outstanding challenges. Nucleic Acids Res 52 , 4761-4783 (2024). https://doi.org:10.1093/nar/gkae267 Hull, M. L. et al. Endometrial-peritoneal interactions during endometriotic lesion establishment. Am J Pathol 173 , 700-715 (2008). https://doi.org:10.2353/ajpath.2008.071128 Hever, A. et al. Human endometriosis is associated with plasma cells and overexpression of B lymphocyte stimulator. Proc Natl Acad Sci U S A 104 , 12451-12456 (2007). https://doi.org:10.1073/pnas.0703451104 Borghese, B. et al. Gene expression profile for ectopic versus eutopic endometrium provides new insights into endometriosis oncogenic potential. Mol Endocrinol 22 , 2557-2562 (2008). https://doi.org:10.1210/me.2008-0322 Crispi, S. et al. Transcriptional profiling of endometriosis tissues identifies genes related to organogenesis defects. J Cell Physiol 228 , 1927-1934 (2013). https://doi.org:10.1002/jcp.24358 Tamaresis, J. S. et al. Molecular classification of endometriosis and disease stage using high-dimensional genomic data. Endocrinology 155 , 4986-4999 (2014). https://doi.org:10.1210/en.2014-1490 Tan, Y. et al. Single-cell analysis of endometriosis reveals a coordinated transcriptional programme driving immunotolerance and angiogenesis across eutopic and ectopic tissues. Nat Cell Biol 24 , 1306-1318 (2022). https://doi.org:10.1038/s41556-022-00961-5 Chen, M. et al. Bioinformatic analysis reveals the importance of epithelial-mesenchymal transition in the development of endometriosis. Sci Rep 10 , 8442 (2020). https://doi.org:10.1038/s41598-020-65606-9 Aghajanova, L. & Giudice, L. C. Molecular evidence for differences in endometrium in severe versus mild endometriosis. Reprod Sci 18 , 229-251 (2011). https://doi.org:10.1177/1933719110386241 Pei, F. L. et al. Construction and evaluation of endometriosis diagnostic prediction model and immune infiltration based on efferocytosis-related genes. Front Mol Biosci 10 , 1298457 (2023). https://doi.org:10.3389/fmolb.2023.1298457 Panzan, M. Q. et al. Evaluation of FAS and caspase-3 in the endometrial tissue of patients with idiopathic infertility and recurrent pregnancy loss. Eur J Obstet Gynecol Reprod Biol 167 , 47-52 (2013). https://doi.org:10.1016/j.ejogrb.2012.10.021 Juhasz-Böss, I. et al. Endometrial expression of estrogen receptor β and its splice variants in patients with and without endometriosis. Arch Gynecol Obstet 284 , 885-891 (2011). https://doi.org:10.1007/s00404-010-1768-7 Yilmaz, B. D. & Bulun, S. E. Endometriosis and nuclear receptors. Hum Reprod Update 25 , 473-485 (2019). https://doi.org:10.1093/humupd/dmz005 McKinnon, B., Bersinger, N. A., Huber, A. W., Kuhn, A. & Mueller, M. D. PPAR-gamma expression in peritoneal endometriotic lesions correlates with pain experienced by patients. Fertil Steril 93 , 293-296 (2010). https://doi.org:10.1016/j.fertnstert.2009.07.980 Sanchez, A. M. et al. The endometriotic tissue lining the internal surface of endometrioma: hormonal, genetic, epigenetic status, and gene expression profile. Reprod Sci 22 , 391-401 (2015). https://doi.org:10.1177/1933719114529374 Ahn, S. H., Singh, V. & Tayade, C. Biomarkers in endometriosis: challenges and opportunities. Fertil Steril 107 , 523-532 (2017). https://doi.org:10.1016/j.fertnstert.2017.01.009 Van Le, L. et al. Discussion: 'Ovarian epithelial carcinoma with pelvic endometriosis,' by Wang et al. Am J Obstet Gynecol 208 , e1-2 (2013). https://doi.org:10.1016/j.ajog.2013.03.014 Laganà, A. S. et al. The Pathogenesis of Endometriosis: Molecular and Cell Biology Insights. Int J Mol Sci 20 (2019). https://doi.org:10.3390/ijms20225615 Zondervan, K. T., Becker, C. M. & Missmer, S. A. Endometriosis. N Engl J Med 382 , 1244-1256 (2020). https://doi.org:10.1056/NEJMra1810764 Taylor, R. N., Kane, M. A. & Sidell, N. Pathogenesis of Endometriosis: Roles of Retinoids and Inflammatory Pathways. Semin Reprod Med 33 , 246-256 (2015). https://doi.org:10.1055/s-0035-1554920 Borghese, B., Zondervan, K. T., Abrao, M. S., Chapron, C. & Vaiman, D. Recent insights on the genetics and epigenetics of endometriosis. Clin Genet 91 , 254-264 (2017). https://doi.org:10.1111/cge.12897 Zhang, T., De Carolis, C., Man, G. C. W. & Wang, C. C. The link between immunity, autoimmunity and endometriosis: a literature update. Autoimmun Rev 17 , 945-955 (2018). https://doi.org:10.1016/j.autrev.2018.03.017 Liang, B. et al. Efficacy, safety and recurrence of new progestins and selective progesterone receptor modulator for the treatment of endometriosis: a comparison study in mice. Reprod Biol Endocrinol 16 , 32 (2018). https://doi.org:10.1186/s12958-018-0347-9 Laschke, M. W. & Menger, M. D. Anti-angiogenic treatment strategies for the therapy of endometriosis. Hum Reprod Update 18 , 682-702 (2012). https://doi.org:10.1093/humupd/dms026 Shao, R., Cao, S., Wang, X., Feng, Y. & Billig, H. The elusive and controversial roles of estrogen and progesterone receptors in human endometriosis. Am J Transl Res 6 , 104-113 (2014). Henze, D. et al. Endometriosis Leads to an Increased Trefoil Factor 3 Concentration in the Peritoneal Cavity but Does Not Alter Systemic Levels. Reprod Sci 24 , 258-267 (2017). https://doi.org:10.1177/1933719116653676 Cluzeau, T. et al. Pro-inflammatory proteins S100A9 and tumor necrosis factor-α suppress erythropoietin elaboration in myelodysplastic syndromes. Haematologica 102 , 2015-2020 (2017). https://doi.org:10.3324/haematol.2016.158857 Zhu, G. et al. Identification of differentially expressed proteins associated with recurrence in ovarian endometriotic cysts. Syst Biol Reprod Med 66 , 59-69 (2020). https://doi.org:10.1080/19396368.2019.1688425 Li, M. Q. et al. CXCL8 enhances proliferation and growth and reduces apoptosis in endometrial stromal cells in an autocrine manner via a CXCR1-triggered PTEN/AKT signal pathway. Hum Reprod 27 , 2107-2116 (2012). https://doi.org:10.1093/humrep/des132 Wang, Y. et al. EGR1 induces EMT in pancreatic cancer via a P300/SNAI2 pathway. J Transl Med 21 , 201 (2023). https://doi.org:10.1186/s12967-023-04043-4 Yang, G., Deng, Y., Cao, G. & Liu, C. Galectin-3 promotes fibrosis in ovarian endometriosis. PeerJ 12 , e16922 (2024). https://doi.org:10.7717/peerj.16922 Gong, Y. et al. Lysine acetyltransferase 14 mediates TGF-β-induced fibrosis in ovarian endometrioma via co-operation with serum response factor. J Transl Med 22 , 561 (2024). https://doi.org:10.1186/s12967-024-05243-2 Nie, M. F. et al. Serum and Ectopic Endometrium from Women with Endometriosis Modulate Macrophage M1/M2 Polarization via the Smad2/Smad3 Pathway. J Immunol Res 2018 , 6285813 (2018). https://doi.org:10.1155/2018/6285813 Lih Yuan, T. et al. Oestrogen-induced epithelial-mesenchymal transition (EMT) in endometriosis: Aetiology of vaginal agenesis in Mayer-Rokitansky-Küster-Hauser (MRKH) syndrome. Front Physiol 13 , 937988 (2022). https://doi.org:10.3389/fphys.2022.937988 Yang, Y. M. & Yang, W. X. Epithelial-to-mesenchymal transition in the development of endometriosis. Oncotarget 8 , 41679-41689 (2017). https://doi.org:10.18632/oncotarget.16472 Huang, N. et al. Intracellular and extracellular S100A9 trigger epithelial-mesenchymal transition and promote the invasive phenotype of pituitary adenoma through activation of AKT1. Aging (Albany NY) 12 , 23114-23128 (2020). https://doi.org:10.18632/aging.104072 Patel, B. G. et al. Pathogenesis of endometriosis: Interaction between Endocrine and inflammatory pathways. Best Pract Res Clin Obstet Gynaecol 50 , 50-60 (2018). https://doi.org:10.1016/j.bpobgyn.2018.01.006 Agostinis, C. et al. Immunological Basis of the Endometriosis: The Complement System as a Potential Therapeutic Target. Front Immunol 11 , 599117 (2020). https://doi.org:10.3389/fimmu.2020.599117 Zhang, Q., Duan, J., Olson, M., Fazleabas, A. & Guo, S. W. Cellular Changes Consistent With Epithelial-Mesenchymal Transition and Fibroblast-to-Myofibroblast Transdifferentiation in the Progression of Experimental Endometriosis in Baboons. Reprod Sci 23 , 1409-1421 (2016). https://doi.org:10.1177/1933719116641763 Pant, A., Moar, K., T, K. A. & Maurya, P. K. Biomarkers of endometriosis. Clin Chim Acta 549 , 117563 (2023). https://doi.org:10.1016/j.cca.2023.117563 Anglesio, M. S. et al. Cancer-Associated Mutations in Endometriosis without Cancer. N Engl J Med 376 , 1835-1848 (2017). https://doi.org:10.1056/NEJMoa1614814 Vallvé-Juanico, J., Houshdaran, S. & Giudice, L. C. The endometrial immune environment of women with endometriosis. Hum Reprod Update 25 , 564-591 (2019). https://doi.org:10.1093/humupd/dmz018 Poli-Neto, O. B., Meola, J., Rosa, E. S. J. C. & Tiezzi, D. Transcriptome meta-analysis reveals differences of immune profile between eutopic endometrium from stage I-II and III-IV endometriosis independently of hormonal milieu. Sci Rep 10 , 313 (2020). https://doi.org:10.1038/s41598-019-57207-y McCallion, A. et al. Estrogen mediates inflammatory role of mast cells in endometriosis pathophysiology. Front Immunol 13 , 961599 (2022). https://doi.org:10.3389/fimmu.2022.961599 Wolf, F. A., Angerer, P. & Theis, F. J. SCANPY: large-scale single-cell gene expression data analysis. Genome Biol 19 , 15 (2018). https://doi.org:10.1186/s13059-017-1382-0 Marečková, M. et al. An integrated single-cell reference atlas of the human endometrium. Nat Genet 56 , 1925-1937 (2024). https://doi.org:10.1038/s41588-024-01873-w Lopez, R., Regier, J., Cole, M. B., Jordan, M. I. & Yosef, N. Deep generative modeling for single-cell transcriptomics. Nat Methods 15 , 1053-1058 (2018). https://doi.org:10.1038/s41592-018-0229-2 Zhou, Y. et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun 10 , 1523 (2019). https://doi.org:10.1038/s41467-019-09234-6 Additional Declarations No competing interests reported. Supplementary Files supplementaryfigure1.jpg Additional file1: Supplementary Figure 1. Boxplot of merging endometriosis sample from 5 datasets before and after removing batch effects supplementaryfigure2.jpg Additional file2: Supplementary Figure 2. (A). The violin plots representing the number of genes by counts, total count and mitochondrial percentage in single cell data before and after removing low quality cells. (B) The UMAP plots showing distribution of mature markers in major cell types. (C). Dot plot showing expressed percentage and abundance of mature markers in endothelial cells, lymphoid cells, and myeloid cells, respectively. supplementaryfigure3.jpg Additional file3: Supplementary Figure 3. Heatmap of single-cell signature matrix from CIBERSORTx result supplementaryfigure4.jpg Additional file4: Supplementary Figure 4. Consistent result between bulk transcriptomics and single-cell analysis. Venn diagrams illustrating the overlap of enriched pathways between MUC5B (A), dStromal_late (B), and eM2 (C) subtypes and those identified in bulk microarray data. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5277508","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":375690222,"identity":"24f8ffbe-b04f-4982-93cc-c649d7d7cf3b","order_by":0,"name":"Meihong Chen","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAq0lEQVRIiWNgGAWjYDACCQY2BoYKGzk29vYDpGg5k2bMx3MmgQQtjG2HE+dJOBgQp4N/dnfaA6CW9DYJhgSGHxXbiLDkztntBgzn0nPbpBsPMPacuU1Yi4FE7jYJhjLr3DaZAwnMjG1Ea2FjTmeTSDAgRUubcwLxWiRugLScSTNsAwbyQaL8wj8DpKXCRl6+vf3ggx8VRGgBAeY/UMYB4tSPglEwCkbBKCAIACNnOAcfn5+vAAAAAElFTkSuQmCC","orcid":"","institution":"Second Affiliated Hospital of Nanchang University","correspondingAuthor":true,"prefix":"","firstName":"Meihong","middleName":"","lastName":"Chen","suffix":""},{"id":375690223,"identity":"4b45fcad-134e-4d86-9fbc-3a0c7b4a7163","order_by":1,"name":"Yan Shen","email":"","orcid":"","institution":"china","correspondingAuthor":false,"prefix":"","firstName":"Yan","middleName":"","lastName":"Shen","suffix":""},{"id":375690224,"identity":"c08621bb-9a31-4502-864a-1fe214b4c376","order_by":2,"name":"Guanqun Jiang","email":"","orcid":"","institution":"University of Southampton","correspondingAuthor":false,"prefix":"","firstName":"Guanqun","middleName":"","lastName":"Jiang","suffix":""},{"id":375690225,"identity":"c5a9e524-62a3-42d5-97e7-c922391e5f5f","order_by":3,"name":"Liqun Wang","email":"","orcid":"","institution":"china","correspondingAuthor":false,"prefix":"","firstName":"Liqun","middleName":"","lastName":"Wang","suffix":""},{"id":375690226,"identity":"d87f0580-f886-4932-bf99-2ede9be8b5a7","order_by":4,"name":"Qi Chen","email":"","orcid":"","institution":"Second Affiliated Hospital of Nanchang University","correspondingAuthor":false,"prefix":"","firstName":"Qi","middleName":"","lastName":"Chen","suffix":""}],"badges":[],"createdAt":"2024-10-16 16:38:22","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5277508/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5277508/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":69443010,"identity":"ee1a631b-31c1-4751-930d-68f244515e4e","added_by":"auto","created_at":"2024-11-20 11:31:48","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":2707765,"visible":true,"origin":"","legend":"\u003cp\u003eDistinct molecular characteristic between ectopic endometriosis and healthy control. (A) Workflow showing the study design (details provided in the Methods). (B) Principal component analysis (PCA) showing microarray patients with disease status. (C) Volcano plot comparing significantly (limma package; p value \u0026lt;0.05) changed genes between the healthy control and ectopic endometriosis. Grey dotted line indicates the threshold for p-value = 0.05 and absolute value of log2 Fold change less than 0.5. Blue and red points represented down-regulated and up-regulated differentially expressed genes respectively. Hallmark, KEGG and GO enrichment analysis of up-regulated (D) and down-regulated (E) differentially expressed genes. Hallmark pathways are indicated in light beige, orange indicates KEGG pathways, red shows molecular function, bule shows biological process and slate gray shows cellular component.\u003c/p\u003e","description":"","filename":"Figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5277508/v1/bc4d5a1d1666a772ca99e4f1.jpg"},{"id":69441968,"identity":"28edfa49-628d-487d-82cd-38408a077804","added_by":"auto","created_at":"2024-11-20 11:23:49","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":10154000,"visible":true,"origin":"","legend":"\u003cp\u003eDeconvolution analysis revealed different cell population between ectopic endometriosis and healthy control. (A) Uniform manifold approximation and projection (UMAP) dimension reduction plot of endometrial tissue from GSE179640. Different colours indicate different cell-type. (B) Dot plot showing expressed percentage and abundance of mature markers in epithelial and mesenchymal cell types, respectively. (C) Violin plot showing 5 major cell compositions in healthy control (red) and ectopic endometriosis (blue) groups. Bonferroni adjusted p values indicated\u003c/p\u003e","description":"","filename":"Figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5277508/v1/9463c47087318e21529b760d.jpg"},{"id":69440791,"identity":"bcedb5be-39bc-4f81-8bab-1550cad843c8","added_by":"auto","created_at":"2024-11-20 11:15:48","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":421842,"visible":true,"origin":"","legend":"\u003cp\u003eViolin plot showing minor cell compositions of epithelial (A), mesenchymal (B), lymphoid (C), myeloid (D) and endothelial (E) cell in healthy control (red) and ectopic endometriosis (blue) groups. Bonferroni adjusted p values indicated.\u003c/p\u003e","description":"","filename":"figure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5277508/v1/c81a811685cfe9c9e91a9f00.jpg"},{"id":69440800,"identity":"7d7e0a75-6682-4b01-b199-a6a23ffe8af3","added_by":"auto","created_at":"2024-11-20 11:15:49","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1738578,"visible":true,"origin":"","legend":"\u003cp\u003eMechanistic exploration of cell types with noteble changes in proportions. Volcano plot comparing significantly marker genes of MUC5B (A), dStromal_late (C) and eM2 (E), respectively. Grey dotted line indicates the threshold for p-value = 0.05 and absolute value log2 fold change of less than 1 for MUC5B, 0.5 for dStromal_late, and less than 0.1 for eM2, individually. Blue and red points represented down-regulated and up-regulated differentially expressed genes, respectively. Hallmark, KEGG and GO enrichment analysis of MUC5B (B), dStromal_late (D) and eM2 (F) significantly marker genes. Hallmark pathways are indicated in light beige, orange indicates KEGG pathways, red shows molecular function, bule shows biological process and slate gray shows cellular component.\u003c/p\u003e","description":"","filename":"Figure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5277508/v1/37810a5c30c5455dc48a129e.jpg"},{"id":69440792,"identity":"f9eb35e5-3ae6-409e-b291-0ab79d89014a","added_by":"auto","created_at":"2024-11-20 11:15:48","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":265443,"visible":true,"origin":"","legend":"\u003cp\u003eConsistent result between bulk transcriptomics and single-cell analysis. Venn diagrams illustrating the overlap of MUC5B (A), dStromal_late (C) and eM2 (E) marker genes and DEGs from bulk microarray. Violin plots displaying the expression levels of overlapping genes for MUC5B (B), dStromal_late (D), and eM2 (F) in control and ectopic endometriosis samples. (G) Violin plots displaying the activation score of enrichment pathway in control and ectopic endometriosis samples. Bonferroni adjusted p values indicated.\u003c/p\u003e","description":"","filename":"Figure5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5277508/v1/0be15bfb13b10524c5765d9d.jpg"},{"id":69440801,"identity":"19879b3a-34b4-407c-8a0c-0d7acfbc6cdf","added_by":"auto","created_at":"2024-11-20 11:15:49","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":220463,"visible":true,"origin":"","legend":"\u003cp\u003eExploring the predictive value of cell types in ectopic endometriosis using Machine Learning. (A) Violin plots displaying the expression levels of mature predictive genes in control and ectopic endometriosis samples. Bonferroni adjusted p values indicated. (B) AUROC (area under the receiver-operating characteristic curve) analysis in the testing cohort. The random forest model achieved an AUROC score of 0.932\u003c/p\u003e","description":"","filename":"figure6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5277508/v1/dc4644c33750248de512f135.jpg"},{"id":79540852,"identity":"6f9a51e4-282c-47be-b001-29f02be72410","added_by":"auto","created_at":"2025-03-31 03:31:52","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":16194481,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5277508/v1/caef866d-b570-4302-81df-1ba7f7469804.pdf"},{"id":69440796,"identity":"b9d252d4-e724-4595-82f2-758f666288df","added_by":"auto","created_at":"2024-11-20 11:15:48","extension":"jpg","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":194164,"visible":true,"origin":"","legend":"\u003cp\u003eAdditional file1: Supplementary Figure 1. Boxplot of merging endometriosis sample from 5 datasets before and after removing batch effects\u003c/p\u003e","description":"","filename":"supplementaryfigure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5277508/v1/d0859462319c228c3e8cb23b.jpg"},{"id":69441967,"identity":"509111f6-3f65-4b03-bcb0-7189f9833929","added_by":"auto","created_at":"2024-11-20 11:23:48","extension":"jpg","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":1805880,"visible":true,"origin":"","legend":"\u003cp\u003eAdditional file2: Supplementary Figure 2. (A). The violin plots representing the number of genes by counts, total count and mitochondrial percentage in single cell data before and after removing low quality cells. (B) The UMAP plots showing distribution of mature markers in major cell types. (C). Dot plot showing expressed percentage and abundance of mature markers in endothelial cells, lymphoid cells, and myeloid cells, respectively.\u003c/p\u003e","description":"","filename":"supplementaryfigure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5277508/v1/66820023785997a14dd601a9.jpg"},{"id":69440797,"identity":"e80fd506-d57e-461c-8066-2936a749e68f","added_by":"auto","created_at":"2024-11-20 11:15:49","extension":"jpg","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":86689,"visible":true,"origin":"","legend":"\u003cp\u003eAdditional file3: Supplementary Figure 3. Heatmap of single-cell signature matrix from CIBERSORTx result\u003c/p\u003e","description":"","filename":"supplementaryfigure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5277508/v1/04a33899cf7f3a44fe5190e8.jpg"},{"id":69440799,"identity":"2d564899-fdfe-4f3e-bbd8-bab658638cce","added_by":"auto","created_at":"2024-11-20 11:15:49","extension":"jpg","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":372491,"visible":true,"origin":"","legend":"\u003cp\u003eAdditional file4: Supplementary Figure 4. Consistent result between bulk transcriptomics and single-cell analysis. Venn diagrams illustrating the overlap of enriched pathways between MUC5B (A), dStromal_late (B), and eM2 (C) subtypes and those identified in bulk microarray data.\u003c/p\u003e","description":"","filename":"supplementaryfigure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5277508/v1/cde05ce2a852ffc22297e578.jpg"}],"financialInterests":"No competing interests reported.","formattedTitle":"Integrated analysis of single-cell and bulk transcriptomic data reveals altered cellular composition and predictive cell types in endometriosis","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eEndometriosis is a common and challenging gynaecological condition characterized primarily by dysmenorrhea, chronic pelvic pain, infertility, etc\u003csup\u003e1,2\u003c/sup\u003e. The rising prevalence of endometriosis, along with high recurrence rates and the challenges of clinical management, significantly impacts women\u0026apos;s quality of life and increases the burden on healthcare resources\u003csup\u003e3\u003c/sup\u003e. Moreover, patients with endometriosis frequently experience delayed diagnosis, with an average of 6.7 years and up to 4-11 years elapsing from symptom onset to pathological histological diagnosis following laparoscopic surgery\u003csup\u003e4,5\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eSingle-cell RNA sequencing (scRNA-seq) uncover detailed insights into microenvironment heterogeneity, functional differentiation and cellular interactions\u003csup\u003e6\u003c/sup\u003e. However, the cost-intensive and limited access to high-quality specimens hinder the widespread application of single-cell analysis. In contrast, using deconvolution methods to estimate cell populations in bulk transcriptomic data effectively addresses the disadvantages of single-cell data, providing a faster and more cost-effective approach for early disease research\u003csup\u003e7\u003c/sup\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn this study, deconvolution results of bulk transcriptomics data display dramatic changes in cell types between ectopic endometriosis and normal endometrial tissue. Through pathway analysis, our data detail the molecular functions of each cell types and how they may promote the progression of endometriosis. Furthermore, the successful construction of the predictive model highlights the unique importance of cell types in the clinical application.\u003c/p\u003e"},{"header":"2. Results","content":"\u003cp\u003e2.1 Overall experimental design and bulk microarray database analysis\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFollowing overall design (Figure 1A), a total of five microarray datasets (GSE11691\u003csup\u003e8\u003c/sup\u003e, GSE7305\u003csup\u003e9\u003c/sup\u003e, GSE12768\u003csup\u003e10\u003c/sup\u003e, GSE25628\u003csup\u003e11\u003c/sup\u003e, and GSE5981\u003csup\u003e12\u003c/sup\u003e) and one single-cell dataset (GSE179640\u003csup\u003e13\u003c/sup\u003e) were included in this study. The analysis comprised 201 microarray samples, consisting of 96 healthy control samples and 105 endometriosis samples. After removing batch effects between different datasets (Sup Figure 1A), PCA analysis revealed that the healthy control and endometriosis groups formed two distinct clusters (Figure 1B), indicating differing molecular characteristics between the two groups.\u003c/p\u003e\n\u003cp\u003eSubsequent differentially expressed gene analysis identified 115 significantly upregulated and 677 significantly downregulated genes (adjusted p-value \u0026lt; 0.05, absolute value of log2Fold Change \u0026gt; 0.5, Figure 1C). Consistent with our previous research, the expression level of FXYD1 was significantly increased in endometriosis (log2Fold Change \u0026gt; 1)\u003csup\u003e14\u003c/sup\u003e. ELN as a key component of the extracellular matrix, was also upregulated in patients with endometriosis, potentially participating in the regulatory networks during the secretory phase\u003csup\u003e15\u003c/sup\u003e. Pathway analysis indicated a marked activation of the EMT pathway in endometriosis patients. Additionally, we observed an enrichment of several other pathways in these patients, including myogenesis, TNFA signalling, estrogen response, extracellular matrix dynamics, and immune effector processes (Figure 1D). Conversely, the downregulated differential genes were enriched in pathways such as E2F targets, G2M checkpoint, and MYC targets (Figure 1E). This further elucidated the distinctive features between patients and healthy individuals, aligning with the results of previous research\u003csup\u003e16\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e2.2 Deconvolution analysis revealed single-cell populations changes in endometriosis\u003c/p\u003e\n\u003cp\u003eThe cell composition between the endometriosis and healthy control groups varied dramatically. Using single-cell data for deconvolution of bulk data effectively provided cell proportions across numerous samples. Eight endometriosis samples from the single-cell database GSE179640 were selected and 24,438 cells passed quality control (Sup Figure 2A). Ultimately, we identified five major cell types: epithelial cells, mesenchymal cells, endothelial cells, lymphoid cells, and myeloid cells (Figure 2A \u0026amp; Sup Figure 2B) and 52 distinct cell subtypes (Figure 2B \u0026amp; Sup Figure 2C, Sup Table 1). Through the signature construction and bulk dissection (Sup Figure 3), the proportion of epithelial and endothelial cells was significantly decreased in endometriosis compared to the healthy control group (p-value = 1.4E-4), while the proportions of mesenchymal, myeloid and lymphoid exhibited varying degrees of increase (Figure 2C).\u003c/p\u003e\n\u003cp\u003eOf note, most epithelial cell subtypes (SOX9_I, SOX9_f_II, Gla_s, and Cil) exhibited a notable decline in proportion within the endometriosis samples (Figure 3A). In contrast, the proportions of MUC5B, Lum, and KRT cells displayed a significant rise. Among the various mesenchymal cell subtypes, all demonstrated substantial upregulation except for the eSt cycling subtype, which exhibited a marked decrease in abundance. The dSt_early subtype showed a slight decrease (p-value = 0.33), while the dSt_m and dSt_l subtypes displayed significant increases (p-value = 3.8E-2 and p-value = 8.4E-5, respectively). Importantly, the dSt_l subtype became the most prevalent subtype among mesenchymal cells (Figure 3B). Although the overall proportions of immune and endothelial cells were relatively low, a significant grow in the proportions of eM2 and mast subtype cells was observed (Figure 3C-E). These results indicate that the composition of the cell population in endometriosis has changed significantly, reflecting disease-related tissue remodelling and alterations in the immune microenvironment.\u003c/p\u003e\n\u003cp\u003e2.3 Biological functions of various cell subtypes\u003c/p\u003e\n\u003cp\u003eTo further understand the functional roles of these significantly altered cell subtypes in endometriosis, we analysed the pivot cell markers within each subtype. The MUC5B cell subtype was distinguished by high expression of MUC5B, LTF, TFF3, and SAA1, while showing low expression of MT family genes (MT1H, MT1G, and MT1M) (Figure 2B \u0026amp; Figure 4A). Hallmark pathway analysis revealed that the genes prominently expressed in the MUC5B subtype were significantly enriched in epithelial mesenchymal transition (EMT), estrogen response, coagulation, KRAS signalling up, and interferon gamma response pathways (Figure 4B). The dStromal late mesenchymal cell subtype was characterized by evaluated expression of LEFTY2, ACTA2, EGR1, FOS, CXCL8, and CXCL2, and low expression of MMP7 and SCGB1D2 (Figure 2B \u0026amp; Figure 4C). This subtype exhibited the highest expression levels in pathways such as TNFA signalling via NFKB, hypoxia, EMT, MAPK signalling pathway, focal adhesion, and inflammatory response (Figure 4D). The eM2 cell subtype, identified as tissue-resident macrophages, showed high expression of markers such as FOLR2 and LYVE1 (Figure 4E). Enrichment analysis using Hallmark, KEGG, and GO pathways revealed significant enrichment in complement, KRAS signalling\u0026nbsp;up, lysosome, and positive regulation of immune response (Figure 4F).\u003c/p\u003e\n\u003cp\u003e2.4 Integration of single-cell and bulk microarray reveals intersection genes and pathways\u003c/p\u003e\n\u003cp\u003eWe performed an overlap analysis between the marker genes of the three cell subtypes and the DEGs from bulk microarray data. For the MUC5B epithelial cell subtype, 21 out of 1,205 marker genes overlapped with the 114 upregulated DEGs from the microarray, including key genes such as S100A9, TFF3, and TLE2 (Figure 5A \u0026amp; Figure 5B). In the case of the dStromal late cell, 28 out of 779 marker genes were found to intersect with the 114 upregulated DEGs from the microarray, highlighting significant genes such as ACTA2, FOSB and EGR1 (Figure 5C \u0026amp; Figure 5D). For the myeloid eM2 subtype, 12 out of 612 marker genes displayed overlap with the 114 upregulated DEGs from the microarray, with notable genes like EPHX1, TYROBP and SOD3 (Figure 5E \u0026amp; Figure 5F).\u0026nbsp;Subsequent pathway enrichment analysis of these intersecting genes revealed that the predominant signalling pathways included EMT, P53 pathway, positive regulation of cell migration, inflammatory response, and complement and coagulation cascades (Figure 5G\u0026nbsp;\u0026amp; Sup figure 4).\u003c/p\u003e\n\u003cp\u003e2.5 Establishment of an early prediction model for endometriosis\u003c/p\u003e\n\u003cp\u003eAt first, we systematically examined published predictive markers of endometriosis, including downregulated genes such as BAX, FAS, and ESR1\u003csup\u003e17-19\u003c/sup\u003e, and upregulated genes like ESR2, PPARG, and ACTA2\u003csup\u003e19-21\u003c/sup\u003e (Figure 6A). Consistent with previous studies, these genes exhibited notable expression abnormalities in endometriosis.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eRegarding of the distinct cell composition and unique molecular characteristics, we hypothesis predictive model based on cell percentages could accurately distinguish endometriosis from healthy controls. Using 70% of the samples for training, our random forest model successfully identified 29 out of 31 endometriosis cases and 22 out of 28 healthy controls in the testing dataset (Table 1), yielding an overall AUC value of 0.932 (Figure 6B). Notably, MUC5B epithelial cells had the highest contribution to the model. Five of the top ten contributing cell types were mesenchymal subtypes, including HOX, ePV_1a, eSt_c, Fib, and dSt_l (Table 2), demonstrating the importance of mesenchymal cells in distinguishing endometriosis. In addition, myeloid cells (mast cells and eM2 macrophages) were also pivotal to the model performance.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ePrediction result of the random forest model in the testing dataset\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003cimg src=\"data:image/png;base64,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\" width=\"927\" height=\"366\"\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eTop 15 most important cell types in the prediction model\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" align=\"\" width=\"528\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eFALSE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003eTRUE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003eMeanDecreaseAccuracy\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003eMeanDecreaseGini\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eMUC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e6.349389\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e10.18686741\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e11.19534639\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e3.509903431\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eHOX\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e7.005972\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e5.959809913\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e9.309629485\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e3.265530055\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eMas\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e8.554349\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e3.329296972\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e8.282943336\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e3.582387852\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eePV_1a\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e5.174236\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e6.449464636\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e7.897985966\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e2.91620666\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eeSt_c\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e6.13349\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e6.755029796\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e7.876833257\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e1.506470136\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eFib\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e6.517639\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e3.510529826\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e7.091543151\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e2.099887649\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eKRT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e3.913174\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e6.113008499\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e6.987720918\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e2.35424138\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003edSt_l\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e1.851104\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e8.189318136\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e6.722879285\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e2.21917295\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eArt\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e6.928125\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e1.614410864\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e6.541262998\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e2.433950724\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eeM2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e5.656597\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e3.722970618\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e6.181275007\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e2.509147755\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eGla_s\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e5.956862\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e1.099191016\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e5.149002227\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e1.683667328\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eSOX9_f_I\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e2.680308\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e4.268303211\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e5.031055466\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e1.247652545\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003epDC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e2.945008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e4.405847704\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e4.648788358\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e2.243083172\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003ePla_B\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e-0.41174\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e5.913050786\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e4.41702008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e1.720396078\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003eMon\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 13.447%;\"\u003e\n \u003cp\u003e0.720371\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 19.1288%;\"\u003e\n \u003cp\u003e4.855466586\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 29.5455%;\"\u003e\n \u003cp\u003e3.683643304\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 24.4318%;\"\u003e\n \u003cp\u003e2.517322473\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n"},{"header":"3. Discussion","content":"\u003cp\u003eAlthough endometriosis is a benign disease, it has malignant behaviours such as proliferation, distant metastasis, and invasive behaviours. Patients primarily experience symptoms such as dysmenorrhea, chronic pelvic pain, and infertility\u0026nbsp;\u003csup\u003e22\u003c/sup\u003e, and may even be at risk for malignant tumors, including ovarian cancer\u003csup\u003e23\u003c/sup\u003e. At present, there are various theories regarding the pathogenesis of endometriosis: transvascular reflux theory, body cavity epidermal metaphysiology, lymphatic and vein disseminate theory, and genetic immune theory, etc\u003csup\u003e24-26\u003c/sup\u003e. However, none of these theories can explicitly explain the occurrence of endometriosis. The development of endometriosis is not caused by a single factor\u003csup\u003e27\u003c/sup\u003e, but is influenced by multiple factors, such as the body\u0026apos;s immune status\u003csup\u003e28\u003c/sup\u003e, \u0026nbsp;inflammatory response\u003csup\u003e29\u003c/sup\u003e, \u0026nbsp;angiogenesis\u003csup\u003e30\u003c/sup\u003e, and local hormone levels\u003csup\u003e31\u003c/sup\u003e. Therefore, exploring cell composition and subtype characteristics is crucial for studying the pathogenesis of endometriosis.\u003c/p\u003e\n\u003cp\u003eTo better understand the differences in cell subtype proportions in endometriosis, we integrated bulk microarray data related to endometriosis from public databases. Using the latest single-cell atlas of endometriosis and the deconvolution software, we revealed variations in five major cell types (epithelial cells, mesenchymal cells, endothelial cells, lymphatic cells, and myeloid cells). Although the proportion of epithelial cells significantly decreased, the proportion of MUC5B cells substantially increased. Notably, these cells exhibited higher expression of TFF3, which has been confirmed to be associated with inflammation\u003csup\u003e32\u003c/sup\u003e. Additionally, MUC5B-specific genes such as S100A9 and TIMP1 were also remarkably overexpressed at the bulk level. S100A9, a member of the S100 protein family, regulated cell migration, promoted the production of inflammatory factors and mediated immune responses\u0026nbsp;\u003csup\u003e33\u003c/sup\u003e. Studies have found that S100A9 was highly expressed in endometriosis and may be involved in the recurrence mechanism of ovarian endometriomas\u003csup\u003e34\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eThe study found that mesenchymal cells was profoundly overexpressed in ectopic endometriosis, with the dStromal late subtype particularly elevated. These cells were involved in inflammatory response, cell adhesion, and angiogenesis. Key genes, including EGR1, bound to SNAY2 promoters to inhibit E-cadherin and promote metastasis. CXCL8 played a crucial role by binding to CXCR1 and activating the PTEN/AKT pathway, thereby promoting proliferation and inhibiting apoptosis in endometriosis cells\u003csup\u003e35\u003c/sup\u003e. \u0026nbsp;Additionally, ACTA2, also known as alpha-smooth muscle actin (\u0026alpha;-SMA), served as a marker for myofibroblasts associated with fibrosis in endometriosis\u003csup\u003e36\u003c/sup\u003e. Extensive research has demonstrated that \u0026alpha;-SMA was significantly upregulated in endometriosis. Multiple factors contributed to its increased expression, which ultimately leaded to the development of fibrosis in endometriosis\u003csup\u003e37,38\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eMacrophages were broadly classified into two main phenotypes: eM1 and eM2 macrophages. Recent studies have demonstrated that serum from women in endometriosis has the capacity to polarize macrophages towards both eM1 and eM2 phenotypes\u003csup\u003e39\u003c/sup\u003e. Similarly, our study also found elevated percentage of eM2 in ectopic endometrial tissues. Furthermore, we have observed that the physiological function of these eM2 cells positively regulated immune responses in the context of endometriosis.\u003c/p\u003e\n\u003cp\u003eThe presence of the EMT signalling pathway in all enrichment analyses revealed its predominant role in endometriosis. Numerous researchers have found that factors such as hypoxia, estrogen stimulation, and WNT4 may trigger EMT in endometriosis\u003csup\u003e40,41\u003c/sup\u003e. In the MUC5B epithelial cell subtype, S100A9 had been found to activate AKT1, leading to EMT in invasive pituitary adenomas\u003csup\u003e42\u003c/sup\u003e. However, the applicability of this mechanism to endometriosis required further investigation. The local inflammatory microenvironment was a hallmark characteristic of endometriosis, sustained by the synergistic activation of hormones and immune factors in ectopic endometrial tissue\u003csup\u003e43\u003c/sup\u003e. In this context, complement activation emerged as a crucial initiator of inflammatory cascade reactions. Core complement genes, including C3, CFH and CLU, were integral to the complement activation process. This activation modulated macrophages and mast cells, leading to the production of various inflammatory mediators and the recruitment of inflammatory cells, thereby amplifying the inflammatory response\u003csup\u003e44\u003c/sup\u003e. The upregulation of genes such as ACTA2, MYH9 and MYLK may indicate that ectopic lesions underwent repeated cycles of tissue damage and repaired due to recurrent bleeding and inflammation. These processes, facilitated by EMT and fibroblast-to-myofibroblast trans-differentiation, resulted in cellular contraction, excessive activation of cell migration, smooth muscle metaplasia, and fibrosis\u003csup\u003e45\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eAt present, the diagnostic markers of endometriosis were mainly based on serum, urine, peritoneal fluid\u003csup\u003e46\u003c/sup\u003e. Even when tissue samples were applied in diagnostic modelling, the focus was primarily on gene expression data rather than cellular composition. Our study showed that early diagnostic models based on cell types can be used to predict disease states of endometriosis successfully. In our model, epithelial cells, mesenchymal cells, and macrophages emerged as critical components. Notably, MUC5B epithelial cells displayed the largest contribution to diagnosing endometriosis. MUC5B cells, considered epithelial progenitor cells, played essential roles in tissue repair, regeneration, and cancer development. Tan successfully captured the tissue separation of MUC5B cells, but it remained unclear how this cell subgroup promoted endometriosis\u003csup\u003e13\u003c/sup\u003e. Mesenchymal cells accounted for the largest proportion of the top ten contributors, among which dStromal late cells were one of the key cells. Interestingly, these mesenchymal cells did not exhibit somatic mutations but display specific epigenetic abnormalities in the expression of key transcription factors\u003csup\u003e47\u003c/sup\u003e. Moreover,\u0026nbsp;too many macrophages stimulated cytotoxic T helper cells to release inflammatory cytokines, leading to a endometritis environment that promoted endometriosis\u003csup\u003e48\u003c/sup\u003e, eM2 macrophages were found to be elevated in type III-IV endometriosis\u003csup\u003e49\u003c/sup\u003e, We also found that eM2 was the main factor in the early diagnosis of endometriosis. Additionally, several studies had reported an increased number of mast cells in endometriosis. Estrogen and stem cell factors provided the ideal microenvironment for their recruitment and differentiation. In turn, mast cells may release pro-inflammatory mediators that contribute to chronic pelvic pain and the progression of endometriosis\u003csup\u003e50\u003c/sup\u003e.\u003c/p\u003e"},{"header":"4. Methods","content":"\u003cp\u003e4.1 Collection and Preprocessing of public bulk transcriptomics datasets\u003c/p\u003e\n\u003cp\u003eWe conducted a comprehensive search in the Gene Expression Omnibus (GEO) database using the keyword \u0026quot;endometriosis\u0026quot; and release date before 29/02/2024 to collect bulk transcriptomics datasets. Seven datasets (GSE11691, GSE7305, GSE12768, GSE201912, GSE168902, GSE25628, and GSE51981) were identified. We then filtered the RNA-seq dataset GSE168902 to ensure consistency and comparability during data integration. Additionally, GSE201912 was excluded due to a minimal number of overlapping genes with the other datasets.\u003c/p\u003e\n\u003cp\u003eFor datasets generated from the Affymetrix platform, raw CEL files were downloaded and normalised using the rma function from the affy (v1.66.0) or oligo (v3.11) packages. For the GSE12768 dataset generated from the Cochin platform, we obtained the normalised data using the getGEO function from the GEOquery package. Probe IDs from the microarray were converted to gene symbols based on the corresponding GPL annotation files provided in GEO. Probes corresponding to multiple gene symbols were discarded. In contrast, genes corresponding to multiple probes was taken the maximum expression values.\u003c/p\u003e\n\u003cp\u003eAfter normalizing each dataset individually, we integrated them into a merged dataset based on gene symbols. We employed the ComBat empirical Bayes batch correction algorithm from the sva package to remove batch effects between different datasets. Finally, we performed PCA analysis using factoextra to reduce the dimensionality of the molecular information from each sample for visualization.\u003c/p\u003e\n\u003cp\u003e4.2 Collection and preprocessing of scRNA-seq raw data\u003c/p\u003e\n\u003cp\u003eThe processed matrix data of Magda\u0026rsquo;s endometriosis atlas was downloaded from https://www.reproductivecellatlas.org/endometrium_reference.html. The single-cell RNA sequencing dataset for endometriosis (GSE179640) was downloaded from the GEO database and processed using the scanpy package\u003csup\u003e51\u003c/sup\u003e. Low-quality cells were filtered out following Magda\u0026apos;s description\u003csup\u003e52\u003c/sup\u003e. The reference mapping tool scvi was used to integrate GSE179640 to the endometriosis atlas and automatically label the cell types annotation\u003csup\u003e53\u003c/sup\u003e. Additionally, normalization, dimensionality reduction, and clustering were performed with default parameter of scanpy. The clustering results were manually annotated for cell types using marker genes from the endometriosis atlas\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e4.3 Identification of differentially expressed genes and significant cell markers\u003c/p\u003e\n\u003cp\u003eFor the bulk transcriptomics dataset after batch effect removal, we constructed a design matrix comparing endometriotic tissue versus healthy tissue and performed differential gene analysis using the Limma package. Genes with an absolute log fold change (LogFC) \u0026gt; 0.5 and adjusted p-values \u0026lt; 0.05 were considered differentially expressed.\u003c/p\u003e\n\u003cp\u003eFor significant cell markers in the single-cell dataset, we used the FindAllMarkers function from the Seurat package to compare different cell subtypes within the same major cell type. The parameters were set to logfc.threshold = 0 and min.pct = 0.1. For the MUC5B, dStromal_late, and eM2 subtypes, hundreds of significant cell markers were identified using adjusted p-values \u0026lt; 0.05 and thresholds of absolute LogFC \u0026gt; 1, 0.5, and 0.1, respectively.\u003c/p\u003e\n\u003cp\u003e4.4 Pathways analysis\u003c/p\u003e\n\u003cp\u003eDifferentially expressed genes and cell markers were uploaded to the Metascape website for pathway analysis using the following parameters: a minimum of 3 overlapping genes, p \u0026lt; 0.05, and a minimum enrichment factor of 1.5\u003csup\u003e54\u003c/sup\u003e. The analysis included databases such as GO-BP, GO-CC, GO-MF, HALLMARK, and KEGG. Only pathways with adjusted p-values below 0.05 were considered significantly enriched.\u003c/p\u003e\n\u003cp\u003e4.5 CIBERSORTx deconvolution analysis\u003c/p\u003e\n\u003cp\u003eWe first randomly selected 1,000 cells from each cell type in GSE179640(or all available cells if fewer than 1,000) to construct a raw expression matrix. Total-count normalization was applied to standardize each cell to a library size of 10,000 reads. The normalized expression matrix was then uploaded to the CIBERSORTx cloud platform. Subsequently, we utilized the \u0026ldquo;Create Signature Matrix\u0026rdquo; feature with default parameters to build the single-cell signature matrix. The batch-corrected microarray expression matrix was also uploaded to the CIBERSORTx website. Finally, \u0026ldquo;Impute Cell Fractions\u0026rdquo; function was applied to estimate the proportions of different cell types in each bulk sample. We employed CIBERSORTx \u0026ldquo;Batch Correction Mode\u0026rdquo; to account for technical differences between the bulk and single-cell platforms.\u003c/p\u003e\n\u003cp\u003e4.6 Differentially expressed cell types\u003c/p\u003e\n\u003cp\u003eWe visualized the CIBERSORTx analysis results using the ggviolin function from the ggpubr package. The Wilcoxon signed-rank test was performed to compare the proportions of the same cell types between the healthy group and the endometriosis group. Cells with a p-value less than 0.05 were considered differentially expressed.\u003c/p\u003e\n\u003cp\u003e4.7 Diagnostic model construction\u003c/p\u003e\n\u003cp\u003eThe collected bulk microarray samples were randomly divided into training and testing sets in a 7:3 ratio using the \u003cem\u003ecaret\u003c/em\u003e package. A classification model was developed using the \u003cem\u003erandomForest\u003c/em\u003e package with the proportions of various cell subtypes as input features and disease status as the prediction target. The number of trees was set to 1,000 for model construction. The model\u0026apos;s performance was evaluated based on accuracy and the area under the ROC curve (AUC) of the testing dataset.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData and materials from this study are available from the corresponding author on reasonable request.\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research was funded by National Natural Science Foundation of China (81960276).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo;contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualisation: M.C., L.W. and Q.C.; Methodology: M.C., Y.S. and G.J.; Formal analysis: M.C. and G.J.; Writing: M.C., Y.S. and Q.C.; Supervision: \u0026nbsp;L.W. and Q.C.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to publish\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCorrespondence\u0026nbsp;\u003c/strong\u003eand requests for materials should be addressed to Q.C. or L.W.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eZondervan, K. T., Becker, C. M. \u0026amp; Missmer, S. A. Endometriosis. \u003cem\u003eThe New England journal of medicine\u003c/em\u003e, 382 (2020). \u003c/li\u003e\n\u003cli\u003eSaunders, P. T. K. \u0026amp; Horne, A. W. Endometriosis: Etiology, pathobiology, and therapeutic prospects. \u003cem\u003eCell\u003c/em\u003e \u003cstrong\u003e184\u003c/strong\u003e, 2807-2824 (2021). \u003c/li\u003e\n\u003cli\u003eMissmer, S. A.\u003cem\u003e et al.\u003c/em\u003e Impact of endometriosis on women\u0026apos;s life decisions and goal attainment: a cross-sectional survey of members of an online patient community. \u003cem\u003eBMJ open\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, e052765 (2022). \u003c/li\u003e\n\u003cli\u003eAgarwal, S. K., Chapron, C., Giudice, L. C., Laufer, M. R. \u0026amp; Taylor, H. S. Clinical diagnosis of endometriosis: a call to action. \u003cem\u003eAmerican Journal of Obstetrics and Gynecology\u003c/em\u003e \u003cstrong\u003e220\u003c/strong\u003e (2019). \u003c/li\u003e\n\u003cli\u003eAnastasiu, C. V., Moga, M. A., Neculau, A. E., Blan, A. \u0026amp; Chicea, L. M. Biomarkers for the Noninvasive Diagnosis of Endometriosis: State of the Art and Future Perspectives. \u003cem\u003eInternational Journal of Molecular Sciences\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 1750 (2020). \u003c/li\u003e\n\u003cli\u003eCheng, C., Chen, W., Jin, H. \u0026amp; Chen, X. A Review of Single-Cell RNA-Seq Annotation, Integration, and Cell-Cell Communication. \u003cem\u003eCells\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e (2023). https://doi.org:10.3390/cells12151970\u003c/li\u003e\n\u003cli\u003eNguyen, H., Nguyen, H., Tran, D., Draghici, S. \u0026amp; Nguyen, T. Fourteen years of cellular deconvolution: methodology, applications, technical evaluation and outstanding challenges. \u003cem\u003eNucleic Acids Res\u003c/em\u003e \u003cstrong\u003e52\u003c/strong\u003e, 4761-4783 (2024). https://doi.org:10.1093/nar/gkae267\u003c/li\u003e\n\u003cli\u003eHull, M. L.\u003cem\u003e et al.\u003c/em\u003e Endometrial-peritoneal interactions during endometriotic lesion establishment. \u003cem\u003eAm J Pathol\u003c/em\u003e \u003cstrong\u003e173\u003c/strong\u003e, 700-715 (2008). https://doi.org:10.2353/ajpath.2008.071128\u003c/li\u003e\n\u003cli\u003eHever, A.\u003cem\u003e et al.\u003c/em\u003e Human endometriosis is associated with plasma cells and overexpression of B lymphocyte stimulator. \u003cem\u003eProc Natl Acad Sci U S A\u003c/em\u003e \u003cstrong\u003e104\u003c/strong\u003e, 12451-12456 (2007). https://doi.org:10.1073/pnas.0703451104\u003c/li\u003e\n\u003cli\u003eBorghese, B.\u003cem\u003e et al.\u003c/em\u003e Gene expression profile for ectopic versus eutopic endometrium provides new insights into endometriosis oncogenic potential. \u003cem\u003eMol Endocrinol\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 2557-2562 (2008). https://doi.org:10.1210/me.2008-0322\u003c/li\u003e\n\u003cli\u003eCrispi, S.\u003cem\u003e et al.\u003c/em\u003e Transcriptional profiling of endometriosis tissues identifies genes related to organogenesis defects. \u003cem\u003eJ Cell Physiol\u003c/em\u003e \u003cstrong\u003e228\u003c/strong\u003e, 1927-1934 (2013). https://doi.org:10.1002/jcp.24358\u003c/li\u003e\n\u003cli\u003eTamaresis, J. S.\u003cem\u003e et al.\u003c/em\u003e Molecular classification of endometriosis and disease stage using high-dimensional genomic data. \u003cem\u003eEndocrinology\u003c/em\u003e \u003cstrong\u003e155\u003c/strong\u003e, 4986-4999 (2014). https://doi.org:10.1210/en.2014-1490\u003c/li\u003e\n\u003cli\u003eTan, Y.\u003cem\u003e et al.\u003c/em\u003e Single-cell analysis of endometriosis reveals a coordinated transcriptional programme driving immunotolerance and angiogenesis across eutopic and ectopic tissues. \u003cem\u003eNat Cell Biol\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 1306-1318 (2022). https://doi.org:10.1038/s41556-022-00961-5\u003c/li\u003e\n\u003cli\u003eChen, M.\u003cem\u003e et al.\u003c/em\u003e Bioinformatic analysis reveals the importance of epithelial-mesenchymal transition in the development of endometriosis. \u003cem\u003eSci Rep\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 8442 (2020). https://doi.org:10.1038/s41598-020-65606-9\u003c/li\u003e\n\u003cli\u003eAghajanova, L. \u0026amp; Giudice, L. C. Molecular evidence for differences in endometrium in severe versus mild endometriosis. \u003cem\u003eReprod Sci\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 229-251 (2011). https://doi.org:10.1177/1933719110386241\u003c/li\u003e\n\u003cli\u003ePei, F. L.\u003cem\u003e et al.\u003c/em\u003e Construction and evaluation of endometriosis diagnostic prediction model and immune infiltration based on efferocytosis-related genes. \u003cem\u003eFront Mol Biosci\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 1298457 (2023). https://doi.org:10.3389/fmolb.2023.1298457\u003c/li\u003e\n\u003cli\u003ePanzan, M. Q.\u003cem\u003e et al.\u003c/em\u003e Evaluation of FAS and caspase-3 in the endometrial tissue of patients with idiopathic infertility and recurrent pregnancy loss. \u003cem\u003eEur J Obstet Gynecol Reprod Biol\u003c/em\u003e \u003cstrong\u003e167\u003c/strong\u003e, 47-52 (2013). https://doi.org:10.1016/j.ejogrb.2012.10.021\u003c/li\u003e\n\u003cli\u003eJuhasz-B\u0026ouml;ss, I.\u003cem\u003e et al.\u003c/em\u003e Endometrial expression of estrogen receptor \u0026beta; and its splice variants in patients with and without endometriosis. \u003cem\u003eArch Gynecol Obstet\u003c/em\u003e \u003cstrong\u003e284\u003c/strong\u003e, 885-891 (2011). https://doi.org:10.1007/s00404-010-1768-7\u003c/li\u003e\n\u003cli\u003eYilmaz, B. D. \u0026amp; Bulun, S. E. Endometriosis and nuclear receptors. \u003cem\u003eHum Reprod Update\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 473-485 (2019). https://doi.org:10.1093/humupd/dmz005\u003c/li\u003e\n\u003cli\u003eMcKinnon, B., Bersinger, N. A., Huber, A. W., Kuhn, A. \u0026amp; Mueller, M. D. PPAR-gamma expression in peritoneal endometriotic lesions correlates with pain experienced by patients. \u003cem\u003eFertil Steril\u003c/em\u003e \u003cstrong\u003e93\u003c/strong\u003e, 293-296 (2010). https://doi.org:10.1016/j.fertnstert.2009.07.980\u003c/li\u003e\n\u003cli\u003eSanchez, A. M.\u003cem\u003e et al.\u003c/em\u003e The endometriotic tissue lining the internal surface of endometrioma: hormonal, genetic, epigenetic status, and gene expression profile. \u003cem\u003eReprod Sci\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 391-401 (2015). https://doi.org:10.1177/1933719114529374\u003c/li\u003e\n\u003cli\u003eAhn, S. H., Singh, V. \u0026amp; Tayade, C. Biomarkers in endometriosis: challenges and opportunities. \u003cem\u003eFertil Steril\u003c/em\u003e \u003cstrong\u003e107\u003c/strong\u003e, 523-532 (2017). https://doi.org:10.1016/j.fertnstert.2017.01.009\u003c/li\u003e\n\u003cli\u003eVan Le, L.\u003cem\u003e et al.\u003c/em\u003e Discussion: \u0026apos;Ovarian epithelial carcinoma with pelvic endometriosis,\u0026apos; by Wang et al. \u003cem\u003eAm J Obstet Gynecol\u003c/em\u003e \u003cstrong\u003e208\u003c/strong\u003e, e1-2 (2013). https://doi.org:10.1016/j.ajog.2013.03.014\u003c/li\u003e\n\u003cli\u003eLagan\u0026agrave;, A. S.\u003cem\u003e et al.\u003c/em\u003e The Pathogenesis of Endometriosis: Molecular and Cell Biology Insights. \u003cem\u003eInt J Mol Sci\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e (2019). https://doi.org:10.3390/ijms20225615\u003c/li\u003e\n\u003cli\u003eZondervan, K. T., Becker, C. M. \u0026amp; Missmer, S. A. Endometriosis. \u003cem\u003eN Engl J Med\u003c/em\u003e \u003cstrong\u003e382\u003c/strong\u003e, 1244-1256 (2020). https://doi.org:10.1056/NEJMra1810764\u003c/li\u003e\n\u003cli\u003eTaylor, R. N., Kane, M. A. \u0026amp; Sidell, N. Pathogenesis of Endometriosis: Roles of Retinoids and Inflammatory Pathways. \u003cem\u003eSemin Reprod Med\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 246-256 (2015). https://doi.org:10.1055/s-0035-1554920\u003c/li\u003e\n\u003cli\u003eBorghese, B., Zondervan, K. T., Abrao, M. S., Chapron, C. \u0026amp; Vaiman, D. Recent insights on the genetics and epigenetics of endometriosis. \u003cem\u003eClin Genet\u003c/em\u003e \u003cstrong\u003e91\u003c/strong\u003e, 254-264 (2017). https://doi.org:10.1111/cge.12897\u003c/li\u003e\n\u003cli\u003eZhang, T., De Carolis, C., Man, G. C. W. \u0026amp; Wang, C. C. The link between immunity, autoimmunity and endometriosis: a literature update. \u003cem\u003eAutoimmun Rev\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, 945-955 (2018). https://doi.org:10.1016/j.autrev.2018.03.017\u003c/li\u003e\n\u003cli\u003eLiang, B.\u003cem\u003e et al.\u003c/em\u003e Efficacy, safety and recurrence of new progestins and selective progesterone receptor modulator for the treatment of endometriosis: a comparison study in mice. \u003cem\u003eReprod Biol Endocrinol\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 32 (2018). https://doi.org:10.1186/s12958-018-0347-9\u003c/li\u003e\n\u003cli\u003eLaschke, M. W. \u0026amp; Menger, M. D. Anti-angiogenic treatment strategies for the therapy of endometriosis. \u003cem\u003eHum Reprod Update\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 682-702 (2012). https://doi.org:10.1093/humupd/dms026\u003c/li\u003e\n\u003cli\u003eShao, R., Cao, S., Wang, X., Feng, Y. \u0026amp; Billig, H. The elusive and controversial roles of estrogen and progesterone receptors in human endometriosis. \u003cem\u003eAm J Transl Res\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 104-113 (2014). \u003c/li\u003e\n\u003cli\u003eHenze, D.\u003cem\u003e et al.\u003c/em\u003e Endometriosis Leads to an Increased Trefoil Factor 3 Concentration in the Peritoneal Cavity but Does Not Alter Systemic Levels. \u003cem\u003eReprod Sci\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 258-267 (2017). https://doi.org:10.1177/1933719116653676\u003c/li\u003e\n\u003cli\u003eCluzeau, T.\u003cem\u003e et al.\u003c/em\u003e Pro-inflammatory proteins S100A9 and tumor necrosis factor-\u0026alpha; suppress erythropoietin elaboration in myelodysplastic syndromes. \u003cem\u003eHaematologica\u003c/em\u003e \u003cstrong\u003e102\u003c/strong\u003e, 2015-2020 (2017). https://doi.org:10.3324/haematol.2016.158857\u003c/li\u003e\n\u003cli\u003eZhu, G.\u003cem\u003e et al.\u003c/em\u003e Identification of differentially expressed proteins associated with recurrence in ovarian endometriotic cysts. \u003cem\u003eSyst Biol Reprod Med\u003c/em\u003e \u003cstrong\u003e66\u003c/strong\u003e, 59-69 (2020). https://doi.org:10.1080/19396368.2019.1688425\u003c/li\u003e\n\u003cli\u003eLi, M. Q.\u003cem\u003e et al.\u003c/em\u003e CXCL8 enhances proliferation and growth and reduces apoptosis in endometrial stromal cells in an autocrine manner via a CXCR1-triggered PTEN/AKT signal pathway. \u003cem\u003eHum Reprod\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 2107-2116 (2012). https://doi.org:10.1093/humrep/des132\u003c/li\u003e\n\u003cli\u003eWang, Y.\u003cem\u003e et al.\u003c/em\u003e EGR1 induces EMT in pancreatic cancer via a P300/SNAI2 pathway. \u003cem\u003eJ Transl Med\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 201 (2023). https://doi.org:10.1186/s12967-023-04043-4\u003c/li\u003e\n\u003cli\u003eYang, G., Deng, Y., Cao, G. \u0026amp; Liu, C. Galectin-3 promotes fibrosis in ovarian endometriosis. \u003cem\u003ePeerJ\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, e16922 (2024). https://doi.org:10.7717/peerj.16922\u003c/li\u003e\n\u003cli\u003eGong, Y.\u003cem\u003e et al.\u003c/em\u003e Lysine acetyltransferase 14 mediates TGF-\u0026beta;-induced fibrosis in ovarian endometrioma via co-operation with serum response factor. \u003cem\u003eJ Transl Med\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 561 (2024). https://doi.org:10.1186/s12967-024-05243-2\u003c/li\u003e\n\u003cli\u003eNie, M. F.\u003cem\u003e et al.\u003c/em\u003e Serum and Ectopic Endometrium from Women with Endometriosis Modulate Macrophage M1/M2 Polarization via the Smad2/Smad3 Pathway. \u003cem\u003eJ Immunol Res\u003c/em\u003e \u003cstrong\u003e2018\u003c/strong\u003e, 6285813 (2018). https://doi.org:10.1155/2018/6285813\u003c/li\u003e\n\u003cli\u003eLih Yuan, T.\u003cem\u003e et al.\u003c/em\u003e Oestrogen-induced epithelial-mesenchymal transition (EMT) in endometriosis: Aetiology of vaginal agenesis in Mayer-Rokitansky-K\u0026uuml;ster-Hauser (MRKH) syndrome. \u003cem\u003eFront Physiol\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 937988 (2022). https://doi.org:10.3389/fphys.2022.937988\u003c/li\u003e\n\u003cli\u003eYang, Y. M. \u0026amp; Yang, W. X. Epithelial-to-mesenchymal transition in the development of endometriosis. \u003cem\u003eOncotarget\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 41679-41689 (2017). https://doi.org:10.18632/oncotarget.16472\u003c/li\u003e\n\u003cli\u003eHuang, N.\u003cem\u003e et al.\u003c/em\u003e Intracellular and extracellular S100A9 trigger epithelial-mesenchymal transition and promote the invasive phenotype of pituitary adenoma through activation of AKT1. \u003cem\u003eAging (Albany NY)\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 23114-23128 (2020). https://doi.org:10.18632/aging.104072\u003c/li\u003e\n\u003cli\u003ePatel, B. G.\u003cem\u003e et al.\u003c/em\u003e Pathogenesis of endometriosis: Interaction between Endocrine and inflammatory pathways. \u003cem\u003eBest Pract Res Clin Obstet Gynaecol\u003c/em\u003e \u003cstrong\u003e50\u003c/strong\u003e, 50-60 (2018). https://doi.org:10.1016/j.bpobgyn.2018.01.006\u003c/li\u003e\n\u003cli\u003eAgostinis, C.\u003cem\u003e et al.\u003c/em\u003e Immunological Basis of the Endometriosis: The Complement System as a Potential Therapeutic Target. \u003cem\u003eFront Immunol\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 599117 (2020). https://doi.org:10.3389/fimmu.2020.599117\u003c/li\u003e\n\u003cli\u003eZhang, Q., Duan, J., Olson, M., Fazleabas, A. \u0026amp; Guo, S. W. Cellular Changes Consistent With Epithelial-Mesenchymal Transition and Fibroblast-to-Myofibroblast Transdifferentiation in the Progression of Experimental Endometriosis in Baboons. \u003cem\u003eReprod Sci\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 1409-1421 (2016). https://doi.org:10.1177/1933719116641763\u003c/li\u003e\n\u003cli\u003ePant, A., Moar, K., T, K. A. \u0026amp; Maurya, P. K. Biomarkers of endometriosis. \u003cem\u003eClin Chim Acta\u003c/em\u003e \u003cstrong\u003e549\u003c/strong\u003e, 117563 (2023). https://doi.org:10.1016/j.cca.2023.117563\u003c/li\u003e\n\u003cli\u003eAnglesio, M. S.\u003cem\u003e et al.\u003c/em\u003e Cancer-Associated Mutations in Endometriosis without Cancer. \u003cem\u003eN Engl J Med\u003c/em\u003e \u003cstrong\u003e376\u003c/strong\u003e, 1835-1848 (2017). https://doi.org:10.1056/NEJMoa1614814\u003c/li\u003e\n\u003cli\u003eVallv\u0026eacute;-Juanico, J., Houshdaran, S. \u0026amp; Giudice, L. C. The endometrial immune environment of women with endometriosis. \u003cem\u003eHum Reprod Update\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 564-591 (2019). https://doi.org:10.1093/humupd/dmz018\u003c/li\u003e\n\u003cli\u003ePoli-Neto, O. B., Meola, J., Rosa, E. S. J. C. \u0026amp; Tiezzi, D. Transcriptome meta-analysis reveals differences of immune profile between eutopic endometrium from stage I-II and III-IV endometriosis independently of hormonal milieu. \u003cem\u003eSci Rep\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 313 (2020). https://doi.org:10.1038/s41598-019-57207-y\u003c/li\u003e\n\u003cli\u003eMcCallion, A.\u003cem\u003e et al.\u003c/em\u003e Estrogen mediates inflammatory role of mast cells in endometriosis pathophysiology. \u003cem\u003eFront Immunol\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 961599 (2022). https://doi.org:10.3389/fimmu.2022.961599\u003c/li\u003e\n\u003cli\u003eWolf, F. A., Angerer, P. \u0026amp; Theis, F. J. SCANPY: large-scale single-cell gene expression data analysis. \u003cem\u003eGenome Biol\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 15 (2018). https://doi.org:10.1186/s13059-017-1382-0\u003c/li\u003e\n\u003cli\u003eMarečkov\u0026aacute;, M.\u003cem\u003e et al.\u003c/em\u003e An integrated single-cell reference atlas of the human endometrium. \u003cem\u003eNat Genet\u003c/em\u003e \u003cstrong\u003e56\u003c/strong\u003e, 1925-1937 (2024). https://doi.org:10.1038/s41588-024-01873-w\u003c/li\u003e\n\u003cli\u003eLopez, R., Regier, J., Cole, M. B., Jordan, M. I. \u0026amp; Yosef, N. Deep generative modeling for single-cell transcriptomics. \u003cem\u003eNat Methods\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 1053-1058 (2018). https://doi.org:10.1038/s41592-018-0229-2\u003c/li\u003e\n\u003cli\u003eZhou, Y.\u003cem\u003e et al.\u003c/em\u003e Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 1523 (2019). https://doi.org:10.1038/s41467-019-09234-6\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"endometriosis, CIBERSORTx, scRNA-seq","lastPublishedDoi":"10.21203/rs.3.rs-5277508/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5277508/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"The early diagnosis of endometriosis is delayed and the clinical treatment is difficult, which causes severe economic burden to the patients. A comprehensive study of the cell classification and composition of endometriosis, is essential for early diagnosis and pathogenesis. This study utilized single-cell data as a reference to accurately deconvolute cell types within bulk transcriptomics data. Five main cell types in endometriosis (epithelial cells, mesenchymal cells, endothelial cells, lymphocytes, and myeloid cells) revealed varying degrees of change compared to healthy controls. Notably, epithelial cells significantly decreased while mesenchymal cells increased. However, MUC5B epithelial cells, showed an increasing trend, consistent with the increasing trend of dStromal-late mesenchymal cells and eM2 macrophages.Pathway enrichment analysis revealed the EMT signalling pathway dominated in endometriosis. Additionally, a random forest method based on cell types could successfully distinguish endometriosis patients from healthy control, which illustrates the potential value of cell types in early diagnosis.","manuscriptTitle":"Integrated analysis of single-cell and bulk transcriptomic data reveals altered cellular composition and predictive cell types in endometriosis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-11-20 11:15:43","doi":"10.21203/rs.3.rs-5277508/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"8124ab37-01ac-4e17-a6da-6cd39e73c056","owner":[],"postedDate":"November 20th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":39997201,"name":"Biological sciences/Biological techniques/Bioinformatics"},{"id":39997202,"name":"Health sciences/Diseases/Reproductive disorders/Urogenital reproductive disorders"}],"tags":[],"updatedAt":"2025-03-31T03:23:42+00:00","versionOfRecord":[],"versionCreatedAt":"2024-11-20 11:15:43","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5277508","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5277508","identity":"rs-5277508","version":["v1"]},"buildId":"k6vKHA0u1VdKjwwnw531e","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}