{"paper_id":"9af650ba-4d95-4c24-b77e-a031d5117e3e","body_text":"Wu et al. Cell Discovery            (2022) 8:95 Cell Discoveryhttps://doi.org/10.1038/s41421-022-00438-7 www.nature.com/celldisc\nARTICLE Open Access\nSFRP4+ stromal cell subpopulation with\nIGF1 signaling in human endometrial regeneration\nBingbing Wu1,2,3,4,Y uL i1,2,3,N a n f a n gN i e1,2,3, Xilin Shen 2,5,W e iJ i a n g2, Yanshan Liu1,2,3,L i nG o n g2,3, Chengrui An2,3,\nKun Zhao2,3, Xudong Yao6, Chunhui Yuan2,3,J i n g h u iH u7,W e iZ h a o7, Jianhua Qian7 and XiaoHui Zou 1,2,3✉\nAbstract\nOur understanding of full-thickness endometrial regeneration after injury is limited by an incomplete molecular\ncharacterization of the cell populations responsible for the organ functions. To help fill this knowledge gap, we\ncharacterized 10,551 cells of full-thickness normal human uterine from two menstrual phases (proliferative and\nsecretory phase) using unbiased single cell RNA-sequencing. We dissected cell heterogeneity of main cell types\n(epithelial, stromal, endothelial, and immune cells) of the full thickness uterine tissues, cell population architectures of\nhuman uterus cells across the menstrual cycle. We identi fied an SFRP4\n+ stromal cell subpopulation that was highly\nenriched in the regenerative stage of the human endometria during the menstrual cycle, and the SFRP4 + stromal cells\ncould signi ficantly enhance the proliferation of human endometrial epithelial organoid in vitro, and promote the\nregeneration of endometrial epithelial glands and full-thickness endometrial injury through IGF1 signaling pathway\nin vivo. Our cell atlas of full-thickness uterine tissues revealed the cellular heterogeneities, cell population architectures,\nand their cell –cell communications during the monthly regeneration of the human endometria, which provide insight\ninto the biology of human endometrial regeneration and the development of regenerative medicine treatments\nagainst endometrial damage and intrauterine adhesion.\nIntroduction\nThe human uterine organ, especially the full-thickness\nendometria, is essential for fertilization and embryonic\ndevelopment. Human endometria, which mainly comprise\nendometrial epithelial and stromal cells, exhibit remark-\nable plasticity and undergo repeated injury and regen-\neration\n1,2. The highly dynamic properties of repeated\ninjury and scar-less repair during the menstrual cycle\nmake it an ideal model to study tissue regeneration\n3. Full-\nthickness injury or dysfunction of the human endometria\ncauses intrauterine adhesion, miscarriage, and uterine\nfactor infertility. The development of new regenerative\ntechnologies against intrauterine adhesion, miscarriage,\nand infertility diseases is hindered by our incomplete\nunderstanding of the molecular characterization of the\ncell populations responsible for scar-less endometrial\nregeneration during the menstrual cycle.\nThe tissue microenvironment is indispensable during\ntissue development\n4, homeostasis 5, regeneration, and\ndisease progression 6. Single cell analysis has been\nincreasingly utilized to dissect cell heterogeneity and\nstudy dynamic cell population architectures and their\nregulation during biological processes such as develop-\nment, tissue homeostasis, and pathology\n4–6. Organoid\ntechnology has been increasingly utilized to study cell-cell\ninteractions within the tissue microenvironment\n7,8. Thus,\nin this study, we dissected cell heterogeneity of main cell\ntypes of full-thickness uterine tissues, identi fied an\nSFRP4+ stromal cell subpopulation that are enriched in\nthe regenerative stage of the endometria during the\nmenstrual cycle as potential regenerative cell populations,\nand determined that the SFRP4\n+ stromal cells could\n© The Author(s) 2022\nOpen AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution andreproduction\nin any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a lin kt ot h eC r e a t i v eCommons license,and indicate if\nchanges were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicatedotherwise in a credit line to the material. If\nmaterial is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain\npermission directly from the copyright holder. To view a copy of this license, visithttp://creativecommons.org/licenses/by/4.0/.\nCorrespondence: XiaoHui Zou ( zouxiaohui@zju.edu.cn)\n1Clinical Research Center, the First Af filiated Hospital, School of Medicine,\nZhejiang University, Hangzhou, Zhejiang, China\n2Dr. Li Dak Sum & Yip Yio Chin Center for Stem Cell and Regeneration\nMedicine, Zhejiang University, Hangzhou, Zhejiang, China\nFull list of author information is available at the end of the article\nThese authors contributed equally: Bingbing Wu, Yu Li, Nanfang Nie, Xilin Shen\n1234567890():,;1234567890():,;\n1234567890():,;\n1234567890():,;\n\nsignificantly enhance the proliferation of human endo-\nmetrial epithelial organoid in vitro. In addition, this study\nfound that promotion of the regeneration of endometrial\nepithelial glands and full thickness endometrial injury\noccurred through the insulin-like growth factor 1 (IGF1)\nsignaling pathway in vivo. Our cell atlas of full-thickness\nuterine tissues revealed cellular heterogeneities, cell\npopulation architectures, and their communication dur-\ning the monthly regeneration of human endometria,\nwhich provides insight into the biology of human endo-\nmetrial regeneration and the development of regenerative\nmedicine treatments against endometrial damage and\nintrauterine adhesion.\nResults\nSingle cell RNA-seq pro filing and unbiased clustering of\ncells from full thickness human uterine tissues\nFirst, we used droplet-based single-cell RNA-seq (10×\nGenomics Chromium system) to pro file single-cell sus-\npensions from seven full-thickness normal human uterine\ntissues from two menstrual phases (Supplementary Table\nS1) (3 from the proliferative-NP and 4 from secretory\nphase-NS) (Fig. 1a). We then used the Cell Ranger Pipe-\nline (10× Genomics) to map the raw sequencing data. We\nisolated and pro filed 10,942 individual cells from human\nuterine tissue. We then filtered the data based on the\nnumber of counts (nCount_RNA < 60,000), features\n(nFeature_RNA > 500), and mitochondrial counts (per-\ncent. mt < 10) of each cell (Supplementary Fig. S1a –d),\nafter computational quality control 10,551 individual cells\nwere left, and the FindIntegrationAnchors function from\nthe Seurat package\n9 was used to integrate the tran-\nscriptomes of the filtered cells from the two groups (NP &\nNS). We then selected the highly variable feature genes\nfrom 2000 feature genes using the FindVariableFeatures\nfunction from the Seurat package\n9 as visualized in the\nElbow plot (Supplementary Fig. S1e). According to the\nvariance of each principal component (PC), we selected\ngenes in PC 1 –16 to perform the downstream graph-\nbased clustering of the filtered cells and partitioned all the\ncells into six main clusters, donor ID and secretory/pro-\nliferative origin phases of full-thickness uterus, which\nwere visualized using Uniform Manifold Approximation\nand Projection (UMAP) (Fig. 1b). The potential doublet of\nthe single-cell data was detected using DoubleFinder\n10\n(Supplementary Fig. S1f); each cluster possesses a unique\nset of marker genes (Fig. 1c) and gene ontology (Fig. 1d;\nSupplementary Tables S2, S3). As labeled by speci fically\nexpressed marker genes and gene ontologies, we named\nthe main clusters as endometrial epithelia, stroma,\nendothelia, smooth muscle, and immune cells in the\nhuman uterus.\nMyometrial smooth muscle showed elevated levels of\nDES and CNN1 (Fig. 1e). Speci fic genes expressed in\nmyometrial smooth muscle cells were enriched in Gene\nontology (GO) terms of regulation of smooth muscle\ncontraction and myo fibril assembly (Fig. 1d; Supplemen-\ntary Table S3). Endometrial stromal cells expressed ele-\nvated levels of COL1A1 and ECM1 (Fig. 1e). Specificg e n e s\nexpressed in stromal cell populations were enriched in GO\nterms of collagen metabolic process and extracellular\nstructure organization (Fig. 1d; Supplementary Table S3).\nEndothelial cells expressed elevated levels of VWF and\nCLDN5 (Fig. 1e). Speci fic genes expressed in endothelial\ncells showed enriched GO terms of endothelial cell mor-\nphogenesis and MHC protein complex assembly (Fig. 1d;\nSupplementary Table S3). Endometrial epithelial cells\nexpressed high levels of KRT8 and EPCAM (Fig. 1e). Spe-\ncific genes expressed in epithelial populations showed\nenriched GO terms of epithelial cell proliferation and\nepithelial cell differentiation (Fig. 1d; Supplementary Table\nS3). Vascular smooth muscle showed high levels of ADIRF\nand CRIP1 (Fig. 1e). Speci fic genes expressed in muscle\nc e l l ss h o w e de n r i c h e dG Ot e r m so fr e p r o d u c t i v es y s t e m\ndevelopment and muscle structure development (Fig. 1d;\nSupplementary Table S3). Endometrial immune cells\nexpressed high levels of PTPRC and CD68 (Fig. 1e). Spe-\ncific genes expressed in immune cell populations showed\nenriched GO terms of MHC protein complex assembly and\nimmune system processes (Fig. 1d; Supplementary Table\nS3). Immunohistochemistry (IHC) images from the\nHuman Protein Atlas (HPA) ( http://www.proteinatlas.org/\n)\n11 further validated the expression of speci ficm a r k e r so f\nthe five main cell populations as follows: ECM1 and\nCOL1A1 were expressed in the uterine stromal cells, DES\nwas expressed in the uterine myometrial cells, ADIRF was\nexpressed in the uterine vascular muscle cells, VWF and\nCLDN5 were expressed in the uterine endothelial cells,\nEPCAM and KRT8 were expressed in the uterine epithelial\ncells, and PTPRC and CD68 were expressed in the uterine\nimmune cells (Fig. 1f).\nEach main cluster can be further clustered into sub-\npopulations. In this study, we found 20 distinct sub-\npopulations in total from six main groups of full-thickness\nuterine tissues using single-cell technology. The uterine\nepithelial cells could be further clustered into five sub-\npopulations, each of which possesses a unique set of genes\nand gene ontology (Supplementary Fig. S2a, Tables S4, S5).\nWe named the epithelial subpopulations as antigen-\npresenting epithelia, EMT epithelial, secretory epithelia,\nproliferative epithelial, and ciliated epithelia (Supplemen-\ntary Fig. S2a). Uterine endothelial cells could be clustered\ninto two subpopulations, including in flammatory endothe-\nlial and secretory endothelial cells (Supplementary Fig. S2b,\nTables S8, S9). The uterine vascular smooth muscle cells\nwere clustered into four subpopulations: ADIRF\n+ vascular,\nsecretory vascular, in flammatory vascular, and DES +\nvascular smooth muscle cells (Supplementary Fig. S2c,\nWu et al. Cell Discovery            (2022) 8:95 Page 2 of 16\n\nFig. 1 (See legend on next page.)\nWu et al. Cell Discovery            (2022) 8:95 Page 3 of 16\n\nTables S10, S11). The immune cells of the human uterus\nclustered into macrophages and NK cells (Supplementary\nFig. S2d, Tables S12, S13). Uterine myometrial cells were\nfurther clustered into three subpopulations: MFAP5\n+\nmyometrial, DCN+ myometrial, and secretory myometrial\nmuscle cells (Supplementary Fig. S2e, Tables S14, S15).\nCell population architectures across the menstrual cycle\nidentified SFRP4 + stroma cells as potential regenerative\nendometrial cell populations\nUterine endometrial stromal cells clustered into four\nsubpopulations (Fig. 2a). As labeled by speci fically\nexpressed marker genes and gene ontologies (Fig. 2b, c;\nSupplementary Tables S6, S7), we named the uterus\nstromal subpopulations as secretory stroma, SFRP4\n+\nstroma, DCN + stroma, and in flammatory stroma. As\nshown by the feature plot, secretory stroma expressed\nelevated levels of SCGB1D2, SFRP4\n+ stromal cells\nexpressed high levels of SFRP4, DCN + stroma expressed\nelevated levels of DCN, and in flammatory stroma\nexpressed high levels of IL6 (Fig. 2d). Speci fic genes\nexpressed in stromal subpopulations enriched GO terms\nof regulation of protein metabolic process and regulation\nof immune response in secretory stroma, retinoic acid\nbiosynthetic process, and developmental process in\nSFRP4\n+ stroma, extracellular matrix organization, and\nregulation of cell migration in DCN + stroma, and\nresponse to cytokine and in flammatory response in\ninflammatory stroma (Fig. 2c).\nThe uterine endometria would undergo regeneration\nand differentiation under the in fluence of hormones\nduring the menstrual cycle. Although there have been\nsome relevant studies based on bulk tissues\n12, some recent\nsingle-cell studies on human endometria have failed to\nreveal the underlying mechanisms of regeneration and\ndifferentiation of human endometria at the single-cell\nlevel. Therefore, we then investigated the effect of the\nmenstrual cycle on the cellular and molecular dynamics of\nthe endometrial cell population architecture and further\nrevealed their regeneration and differentiation hierarchies.\nTo reconstruct the temporal dynamics of all cell\npopulations during the menstrual cycle, we first calculated\nthe relative proportion of all cell populations by decon-\nvolution analysis, using marker genes of each cell popu-\nlation generated in our cell atlas on a published dataset\n(GSE4888) on transcriptional pro filing of bulk human\nendometrium\n12 (Supplementary Fig. S3). There were\nmainly four patterns of endometrial cell population\narchitectures from proliferative through the early, mid-\nsecretory to late secretory phase of the menstrual cycle\n(Fig. 2f) and are as follows: Pattern 1, the SFRP4\n+ stromal\ncell populations dominated the first pattern; the propor-\ntion increased mainly in the proliferative phase and\ndecreased in the other phases of the endometria; Pattern\n2, the proliferative epithelial, ciliated epithelial, and\nantigen-presenting epithelial populations dominated the\nsecond pattern that increased from the proliferative phase\nto the early secretory phase and decreased afterwards;\nPattern 3, in flammatory stroma, EMT epithelial and\nsecretory epithelial cell populations of the third pattern\ndominated in the mid-secretory phase of the menstrual\ncycle; Pattern 4, the rest of the cell populations mainly\ndominated the fourth pattern that increased only in the\nlate secretory phase of the menstrual cycle, which was\nconsistent with previous results that NK cell subsets were\nreported to be abundant in the late secretory phase of the\nendometria that rebuild and maintain appropriate local\nmicroenvironment for pregnancy\n13. Monocytes/macro-\nphages are responsible for the breakdown and are asso-\nciated with repair and remodeling\n14.\nAs the dynamics of the cell population architectures\nshown above were highly correlated with the menstrual\ncycle, we investigated the potential cell populations\nresponsible for endometrial regeneration. The SFRP4\n+\nendometrial stromal cell population was shown to dom-\ninate the first pattern, mainly in the proliferative phase,\n(see figure on previous page)\nFig. 1 Single cell RNA-seq pro filing and unbiased clustering of cells from full thickness human uterine tissues. a Workflow shows sample\nprocessing, enzymatic digestion and drop-seq based single cell RNA-seq. b UMAP plot of 6 main clusters, donor ID and secretory/proliferative origin\nphases of full-thickness uterus by single cell RNA-seq (scRNA-seq). c Heatmap shows speci fically expressed gene signature of the 6 main clusters of\nfull-thickness uterus. d Gene ontology (GO) analysis of the speci fically expressed gene signature of the 6 main clusters of full-thickness uterus. e Violin\nplot showed speci fic marker genes from gene signatures of each cell cluster (DES, CNN1 for myometrial smooth muscle cells; COL1A1, ECM1 for\nstroma cells; KRT8, EPCAM for epithelial cell; CLDN5, VWF for endothelial cells; ADIRF, CRIP1 for vascular smooth muscle cells; PTPRC, CD68 for\nimmune cells). f Immunohistochemistry staining from the HPA further validated the expression of speci fic markers of the 6 main clusters: COL1A1\n(https://www.proteinatlas.org/ENSG00000108821-COL1A1/tissue/endometrium#img), ECM1 ( https://www.proteinatlas.org/ENSG00000143369-ECM1/\ntissue/endometrium#img) was expressed in the uterus stromal cells, DES ( https://www.proteinatlas.org/ENSG00000175084-DES/tissue/\nendometrium#img) was expressed in the uterine myometrial smooth muscle cells, ADIRF ( https://www.proteinatlas.org/ENSG00000148671-ADIRF/\ntissue/endometrium#img) was expressed in the uterine vascular smooth muscle cells, CLDN5 ( https://www.proteinatlas.org/ENSG00000184113-\nCLDN5/tissue/endometrium#img), VWF ( https://www.proteinatlas.org/ENSG00000110799-VWF/tissue/endometrium#img) were expressed in the\nuterus endothelial cells, EPCAM ( https://www.proteinatlas.org/ENSG00000119888-EPCAM/tissue/endometrium#img), KRT8 ( https://\nwww.proteinatlas.org/ENSG00000170421-KRT8/tissue/endometrium#img) were expressed in the uterus epithelial cells,, PTPRC ( https://\nwww.proteinatlas.org/ENSG00000081237-PTPRC/tissue/endometrium#img), CD68 ( https://www.proteinatlas.org/ENSG00000129226-CD68/tissue/\nendometrium#img) were expressed in the uterus immune cells. Scale bar, 25 µm.\nWu et al. Cell Discovery            (2022) 8:95 Page 4 of 16\n\nFig. 2 Cell population architectures across the menstrual cycle identi fied SFRP4+ stroma cells as potential regenerative endometrial cell\npopulations. a UMAP plot of four uterus stroma cell sub-populations, donor ID and secretory/proliferative origin phases using Seurat. b Heatmap\nshows differential expressed gene signature of each sub-cluster from stromal cells. c Gene ontology (GO) analysis of the speci fically expressed gene\nsignature of each sub-population from stroma cells. d Featureplot depicted speci fic markers for each stroma sub-population ( SCGB1D2 for secretory\nstroma cells; SFRP4 for SFRP4+ stroma cells; DCN for DCN+ stroma cells; IL6 for inflammatory stroma cells). e Cell proportions of endometrial stromal in\nthe proliferative and secretory phase of the endometria. f Heatmap showed relative proportional score of each subpopulation in endometria during\nthe menstrual cycle from proliferative, early-secretory, mid-secretory to late-secretory phase of human endometria, the black arrow highlighted the\nproliferative epithelia and SFRP4 + stroma with high relative proportional score in proliferative phase. g Immunofluorescence staining of SFRP4 +\nstroma in different phase of menstrual cycle. Scale bar, 10 µm.\nWu et al. Cell Discovery            (2022) 8:95 Page 5 of 16\n\nFig. 3 (See legend on next page.)\nWu et al. Cell Discovery            (2022) 8:95 Page 6 of 16\n\nand we validated the results by immuno fluorescence\nstaining of SFRP4 + stroma in different phases of the\nmenstrual cycle. SFRP4 + stromal cells were highly enri-\nched in samples from the proliferative phase of the\nendometria compared with those from the secretory\nphase of the endometria (Fig. 2e, g). Thus, our results are\nconsistent with the deconvolution analysis that identi fied\nSFRP4\n+ stromal cells as potential regenerative endo-\nmetrial cell populations. To further con firm the above\nresults, we used external single-cell data 15 to verify\nSFRP4+ stroma as proliferative phase-speci fic endometrial\ncell populations, which depicted speci fic markers of\nSFRP4 for SFRP4 + stromal cells in the external single cell\ndata (Supplementary Fig. S4a, b). Different proportions of\nstromal subsets in distinct phases of the endometrium\nduring the menstrual cycle showed that SFRP4\n+ stromal\ncells were speci fically highly enriched in the proliferative\nphase of the endometria, accounting for more than 90% of\nthe stromal cells from the proliferative phase of the\nendometria (Supplementary Fig. S4c). As SFRP4\n+ stroma\nwas speci fically enriched in proliferative endometria, we\ninferred that SFRP4 + stroma is a potential regenerative\nendometrial cell population.\nNext, we examined the spatial distribution of SFRP4 +\nstromal cells. IHC staining images from the HPA showed\nthat the SFRP4\n+ staining stromal cells were equally dis-\ntributed in both the functional endometrial layer (Sup-\nplementary Fig. S5a), close to the uterine cavity, and the\nbasal endometrial layer (Supplementary Fig. S5b), close to\nthe myometrium, which was also consistent with the\nresults that SFRP4\n+ stromal cells were in the proliferative\nphase of the endometrium. In addition, more than 90% of\nthe stromal cells from the proliferative phase of the\nendometria were SFRP4 + stromal cells.\nConnectivity analysis revealing signals from SFRP4 +\nstroma cells promote endometrial epithelial organoid\nproliferation\nThe tissue microenvironment is indispensable for tissue\nhomeostasis and regeneration 5. Different cell populations\nin tissues are surrounded by each other, and communica-\ntion among cells regulates and balances cell populations to\nachieve proper regeneration\n4,5. Thus, we reconstructed the\nintrauterine connectivity map among the cell populations\nusing CellPhoneDB16. Finally, we obtained 400 signi ficant\nconnections of 463 and 485 ligand —receptor pairs among\n20 cell subpopulations from both the proliferative and\nsecretory phases of the human uterus, respectively (Fig. 3a;\nSupplementary Fig. S6a –c, Tables S16, S17). As shown in\nthe heatmaps, there were dramatic differences in terms of\nthe total number of receptor-ligand interactions from any\nof the two subpopulations from the microenvironment of\nthe full-thickness proliferative (Fig. 3a) and secretory\n(Supplementary Fig. S6a) human uterus.\nProliferation is the main stage during the regeneration\nprocess in the proliferative phase of endometria\n17. Thus,\nwe analyzed the potential regulatory cellular micro-\nenvironment of proliferative epithelia using the con-\nnectivity map. According to the unique temporal\ndistribution of proliferative epithelial and SFRP4\n+ stromal\ncell populations (Fig. 2f). We selected cell populations\nthat showed a similar temporal distribution to that of the\nproliferative epithelia of human endometria as the\nmicroenvironment of the proliferative epithelia in unique.\nThe connections of ligands from each cell population of\nthe regenerative microenvironment to receptors from the\nproliferative epithelia showed that ligands (WNTs, FGFs,\nIGF1, and MDK) from the SFRP4\n+ stromal cell popula-\ntion regulated the proliferative epithelia in the pro-\nliferative phase of human endometria (Fig. 3b). We then\nvalidated the temporal expression patterns of the ligands\nsurrounding the proliferative epithelia from the predicted\nconnectivity map using a published dataset (GSE4888) for\ntranscriptional pro filing of the bulk human endome-\ntrium\n12. We selected ligands with unique expression\npatterns that were highly correlated with the temporal\ndynamics of the proliferative epithelial cell population\n(Fig. 2f). As shown in the heatmaps, similar to the pattern\nof the proliferative epithelia, ligands of WNTs, FGFs,\nIGF1, and GDFs superfamily members were highly\nexpressed in the proliferative phase of human endometria\n(Fig. 3c).\nTo further validate the proliferative effect of the SFRP4\n+\nendometrial stromal cell population on endometrial epi-\nthelia, we first cultured stromal cells from endometria in\nthe proliferative phase of the menstrual cycle. As shown in\n(see figure on previous page)\nFig. 3 Connectivity analysis revealing signals from SFRP4 + stroma cells promote endometrial epithelial organoid proliferation in vitro.\na Heatmap shows the total numbers of receptor-ligand interactions from any of the two sub-populations from the microenvironment of full-\nthickness human uterus in proliferative phase. b Clustering of the interactions of ligands (uniquely expressed in proliferative endometria) from each\ncell sub-population of the proliferative endometria to receptors from the proliferative epithelial cell sub-population. c Heatmap shows expression of\nligands uniquely expressed in proliferative phase of the human endometria derived from another transcriptional dataset of the human endometria\nduring the menstrual cycle (GSE4888). d immunofluorescence staining of co-expression with SFRP4 and CD10 in cultured human endometrial\nstromal cells, scale bar, 100 μm. e Percentage of SFRP4 and CD10 double positive cells in each high-power field. n = 6. f organoids culturing of\nendometrial epithelia cells with or without SFRP4 + stroma co-cultured, scale bar, 200 μm. g quantification of endometrial epithelial organoids at\ndifferent time points of co-cuture, * p < 0.05; *** p < 0.001. h diameter of endometrial epithelial organoids at different time points of co-cuture,\n***p < 0.001.\nWu et al. Cell Discovery            (2022) 8:95 Page 7 of 16\n\nFig. 4 (See legend on next page.)\nWu et al. Cell Discovery            (2022) 8:95 Page 8 of 16\n\nthe immuno fluorescence staining with SFRP4 and CD10\n(Fig. 3d), approximately 97.28% of the cultured endo-\nmetrial stromal cells were SFRP4 and CD10 double-\npositive cells (Fig. 3e), which was consistent with the\nsingle-cell data showing that most of the stromal cells in\nthe proliferative phase of the endometria were SFRP4\n+\nendometrial stromal cells (Supplementary Figs. S4c, S5).\nNext, we co-cultured cultured SFRP4\n+ endometrial stro-\nmal cells with in vitro endometrial epithelial organoids\nand found that SFRP4\n+ endometrial stromal cells pro-\nmoted the proliferation of endometrial epithelial orga-\nnoids after 3, 7, and 14 days of co-culture (Fig. 3f).\nQuantification of endometrial epithelial organoids at dif-\nferent time points of co-culture showed that SFRP4\n+\nendometrial stromal cells could signi ficantly increase the\nnumber and diameter of endometrial epithelial organoids\ncompared with those cultured alone (Fig. 3g, h).\nSFRP4+ stromal cells promote the proliferation of\nendometrial epithelial organoid in vitro through\nregenerative IGF1 signaling\nTo study the underlying mechanisms by which SFRP4 +\nstromal cells promote the proliferation of endometrial\nepithelial organoids, we aimed to identify the key signaling\nmolecules and pathways. The result showed that the\nSFRP4\n+ stromal cell population was shown to promote\nthe proliferation of endometrial epithelia, most likely\nthrough the secretion of ligands (Fig. 3b).\nFirstly, we showed differentially expressed genes\nbetween SFRP4\n+ stromal cells and other stromal cell\nsubsets, among which 15 secreted ligands were sig-\nnificantly highly expressed in SFRP4 + stromal cells (Fig.\n4a). We further veri fied that the 15 secreted ligands were\nspecifically highly expressed in the proliferative phase in a\npublic dataset (GEO series: GSE4888) 12 (Fig. 4b). The\nVenn diagram showed that IGF1 was the only overlap\n(Fig. 4c) between the 15 secreted ligands (dark gray) sig-\nnificantly highly expressed in SFRP4\n+ stroma (light gray)\nand the interaction of ligands from SFRP4 + stroma with\nreceptors from the proliferative epithelial subset (purple)\n(Fig. 3b). Thus, IGF1 is potentially the key regenerative\nsignaling molecule secreted by SFRP4\n+ endometrial\nstromal cells that promotes the proliferation of endo-\nmetrial epithelia.\nNext, we studied the effect of IGF1 on endometrial epi-\nthelia. The 3D organoid culture showed that\nIGF1 supplementation could also promote organoid for-\nmation (Supplementary Fig. S7a), with a significant increase\nin the number (Supplementary Fig. S7b) and diameter\n(Supplementary Fig. S7c) of endometrial epithelial orga-\nnoids compared to those without IGF1 supplementation,\nwhich was consistent with the effect of the SFRP4\n+ stroma\non endometrial epithelial organoids (Fig.3f ) .W ea l s of o u n d\nthat IGF1 supplementation promoted the migration of\nendometrial epithelia in the 2D endometrial epithelial cul-\nture system (Supplementary Fig. S7d–f).\nNext, we investigated whether SFRP4\n+ stroma pro-\nmotes endometrial epithelial organoid formation through\nIGF1 signaling. We used siRNA to knock down IGF1\nexpression in the SFRP4\n+ stromal cells, three different\nsiRNAs were designed according to different regions of\nthe human IGF1 mRNA sequence (IGF1-si-1, IGF1-si-2,\nIGF1-si-3). qPCR was conducted to validate the ef ficiency\nof each siRNA 24 h after each of the three siRNAs against\nIGF1 were transfected into the SFRP4\n+ stromal cells. The\nresults showed that both IGF1-si-1 and IGF1-si-3 could\nsignificantly knock down the expression of IGF1 in the\nSFRP4\n+ stromal cells, and IGF1-si-3 showed the best\nperformance, therefore we chose IGF1-si-3 for the rest of\nthe experiments (Supplementary Fig. S8). After IGF1-si-3\nwas transfected into SFRP4\n+ stromal cells, the stromal\ncells were then co-cultured with endometrial epithelial\norganoids. The results showed that IGF1 knockdown\n(IGF1si-stroma) in SFRP4\n+ stromal cells signi ficantly\nhindered the formation of epithelial organoids (Fig. 4d),\nwith a signi ficant decrease in the number (Fig. 4e) and\ndiameter (Fig. 4f) of organoids formed compared to those\nthat were co-cultured with normal SFRP4 + stromal cells.\nWe also con firmed that SFRP4 + stroma promotes endo-\nmetrial epithelial organoid formation through\nIGF1 signaling using four IGF1 signaling pathway inhi-\nbitors (s1012, s1034, s1091, and s1093). The results\n(see figure on previous page)\nFig. 4 SFRP4 + stromal cells promote the proliferation of endometrial epithelial organoid in vitro through regenerative IGF1 signaling.\na Heatmap showed differentially expressed genes between SFRP4 + stromal cells and other stromal cell subsets. Among them, 15 secreted factors\nwere signi ficantly highly expressed in SFRP4 + stroma. b the transcriptional dataset (series: GSE4888) veri fied that the 15 secretory factors were\nspecifically highly expressed in proliferative phase. c Venn diagram of genes signi ficantly highly expressed in SFRP4 + stroma (light gray) compared to\nsecretory factors (dark gray), and interaction clusters of ligands from SFRP4 + stroma to receptors from the proliferative epithelial subset (purple).\nd organoids culturing of endometrial epithelia cells cocultured with SFRP4 + stroma from normal or IGF1 knockdown (IGF1si-stroma) cells, scale bar,\n200 μm. e quantification of endometrial epithelial organoids at different time points of co-cuture, ** p < 0.01; f diameter of endometrial epithelial\norganoids at different time points of co-cuture, * p < 0.05. g organoids culturing of endometrial epithelia cells with or without IGF1 inhibitors when\ncocultured with SFRP4 + stroma, 4 IGF1 inhibitors (s1012, s1034, s1091, s1093) are from Selleck.cn, scale bar, 200 μm. h quantification of endometrial\nepithelial organoids in different groups of co-culture, ** p < 0.01; ***p < 0.001. i diameter of endometrial epithelial organoids in different groups of co-\ncuture, ** p < 0.01; *** p < 0.001.\nWu et al. Cell Discovery            (2022) 8:95 Page 9 of 16\n\nshowed that supplementation of any of the four inhibitors\ninto the organoid culture system co-cultured with\nSFRP4\n+ stromal cells abolished the enhancement effects\nof SFRP4 + stromal cells on endometrial epithelial orga-\nnoids (Fig. 4g), with a signi ficant decrease in the number\n(Fig. 4h) and diameter (Fig. 4i) of organoids formed\ncompared to those cultured without the IGF1 signaling\npathway inhibitor. In this section, we identi fied the key\nIGF1 signaling molecules secreted by SFRP4\n+ stromal\ncells that promote the proliferation of endometrial epi-\nthelial organoids in vitro.\nWe also conducted the effects of IGF1 and knockdown\nor inhibition of IGF1 on stromal cell proliferation (Sup-\nplementary Fig. S8b, c). As the results showed that\nknockdown of IGF1 mRNA using IGF1 siRNA did not\naffect cell proliferation of stroma cells (Supplementary\nFig. S8b), while complete inhibition of IGF1 signaling\nusing receptor inhibitors of IGF1 signaling could inhibit\nthe cell proliferation of stroma cells compared with the\ncontrol (Supplementary Fig. S8c). The probable explana-\ntion is as follows: though knockdown of IGF1 mRNA\nusing IGF1 siRNA signi ficantly decrease the expression of\nIGF1, IGF1 siRNA could not completely shut-down the\nexpression of IGF1, there might be still basal constitutive\nexpression of IGF1 even after knockdown of IGF1 mRNA,\nwhile inhibition of IGF1 signaling using receptor inhibi-\ntors of IGF1 signaling could completely inhibit the\nIGF1 signaling, which explained the decrease in cell\nproliferation of stromal cells.\nSFRP4+ stromal cells promote the regeneration of\nendometrial epithelial glands and full thickness\nendometrial injury through IGF1 signaling pathway in vivo\nNext, we studied the regenerative potential of SFRP4 +\nendometrial stromal cells in an in vivo full-thickness\nendometrial injury model along with the effect of the key\nIGF1 signal molecule secreted from SFRP4 + stromal cells.\nA rat model of full-thickness endometrial injury was con-\nstructed (Supplementary Fig. S9). SFRP4\n+ endometrial\nstromal cells were mixed with gelatin methacryloyl\n(GelMA) hydrogel\n18 and transplanted into the injury site of\nthe rat full-thickness endometrial injury model.\nThe GelMA hydrogel was gelated under UV irradiation for\n15 s (Supplementary Fig. S10a). Comparing the 1H-NMR\nspectra (Nuclear magnetic resonance spectrometer,\nAVANCE III, Switzerland) of gelatin and GelMA, new sig-\nnals of the acrylic protons of methacrylic functions and\nmethyl function can be observed atδ = 5.3 ppm,δ = 5.5 ppm\nand at δ = 1.8 ppm. Therefore, we concluded that metha-\ncrylate (MA) was successfully grafted onto gelatin (Supple-\nmentary Fig. S10b). The microstructures of the GelMA\nhydrogels were observed by scanning electron microscopy\n(SEM). The SEM images revealed uniform porous micro-\nstructures throughout all samples (Supplementary Fig. S10c).\nTo detect the cytocompatibility of the GelMA hydrogel,\nwe encapsulated and cultured fibroblast L929 in the\nhydrogel. According to the results of cell live/death\nstaining, several dead cells (red color) were detected on the\nfirst day but disappeared on the seventh day (Supple-\nmentary Fig. S10d). The results showed that cell viability\nwas 70% on the first day and 100% on the third and\nseventh days, indicating that the cells encapsulated in the\nhydrogel were active after long-term culture (Supple-\nmentary Fig. S10e). We also collected GelMA hydrogel\nextract at 24 h for cell culture. The results showed that\nthere was no signi ficant difference in the cell proliferation\nrate between cells cultured in hydrogel extract medium\nand normal medium (Supplementary Fig. S10f). These\nresults indicate that the GelMA hydrogel had good\ncytocompatibility.\nThe IGF1 signaling pathway inhibitor (s1091) was pre-\nviously proven safe after in vivo administration and\nalready used in a phase 3 study\n19, therefore, s1091 was\nalso mixed into the hydrogel together with SFRP4 +\nendometrial stromal cells to study the involvement of the\nkey IGF1 signaling pathway during the regeneration of\nfull-thickness endometrial injury after SFRP4\n+ stromal\ncell transplantation in vivo.\nThe histology results showed that transplantation of the\nSFRP4+ stroma cells into the injury site promoted the\nregeneration of the rat endometria (Fig. 5a), with a thicker\nendometrial layer formed compared with that of the\ninjury group (Fig. 5b). IHC staining of the endometrial\nepithelial glands using an anti-FOXA2 antibody showed\nthat SFRP4\n+ stroma cells could promote new gland for-\nmation during the regeneration of the damaged endo-\nmetria (Fig. 5c, d). The IGF1 signaling pathway inhibitor\n(s1091) abolished the regeneration of the SFRP4\n+ stroma\ncells both in full-thickness endometria and in the for-\nmation of new endometrial epithelial glands (Fig. 5b, d),\nwhich was consistent with the in vitro results that SFRP4 +\nstromal cells promote the formation of human endo-\nmetrial epithelial organoids. In this section, we show that\nSFRP4\n+ stromal cells can promote the regeneration of\nendometrial epithelial glands and full-thickness endo-\nmetrial injury through the IGF1 signaling pathway in vivo.\nDiscussion and conclusion\nOur understanding of full-thickness endometrial\nregeneration after injury is limited by the incomplete\nmolecular characterization of the cell populations\nresponsible for organ functions. Cell heterogeneity, cell\npopulation architecture, and regulation of complex tissues\nand organs are vital for tissue development, homeostasis,\nregeneration, and pathology\n4–6, however, the precise\nmolecular mechanisms remained unknown until the\nbroad applications of single-cell RNA-seq. In this study,\nwe reconstructed the cell subpopulation architectures and\nWu et al. Cell Discovery            (2022) 8:95 Page 10 of 16\n\ncommunications map of full-thickness human uterine tis-\nsues from two menstrual phases (proliferative and secre-\ntory phases) using unbiased single-cell RNA-sequencing.\nOur study further sheds light on an SFRP4\n+ endometrial\nstromal cell subpopulation that is highly enriched in the\nregenerative stage of the endometria during the menstrual\ncycle. SFRP4\n+ stromal cells signi ficantly enhanced the\nproliferation of endometrial epithelial organoids in vitro\nand promoted the regeneration of endometrial epithelial\nglands and full-thickness endometrial injury through the\nIGF1 signaling pathway in vivo. Our cell atlas of full-\nthickness uterine tissues revealed the cellular and mole-\ncular mechanisms regulating the monthly regeneration of\nhuman endometria, which provides insights into the biol-\nogy of human endometrial regeneration and the develop-\nment of regenerative medicine treatments against\nendometrial damage and intrauterine adhesion.\nMost previous studies on uterine biology were based on\nbulk uterus/endometrium tissue transcriptomic analysis\n20\nor a comparison between the different regions of the\ntissue\n21. With advances in technology and development\nanalysis pipelines, studies have reached the single-cell\nlevel\n22–25. To the best of our knowledge, all cell popula-\ntions throughout the menstrual cycle were included in our\nstudy, which provided the most detailed and dynamic cell\npopulations of the uterine tissue to date in comparison\nwith previous bulk, single-cell studies on endometrial\ntissue or in vitro endometrial organoids\n15,24,26,27.\nSpecifically, in comparison with two recent single cell\nstudy on endometria during the menstrual cycle 15,25, our\nstudy provided more detailed and dynamic cell popula-\ntions of the uterus tissue across the menstrual cycles, with\na total of 20 functional distinct sub-populations identi fied\nwhich could be further grouped into 6 main clusters\nnamed as endometrial epithelia, stroma, endothelia,\nsmooth muscle, and immune cells of the full-thickness\nuterus tissues by using the single-cell technology, and\nreconstructed the spatiotemporal cell population archi-\ntectures of the full-thickness human uterus tissues during\nthe menstrual cycle. In Wang et al. paper\n15, six cell types\nwere identi fied: stromal fibroblast, endothelium, macro-\nphage, lymphocyte, ciliated epithelium and unciliated\nepithelium, which was consistent with main clusters in\nour study. Their main findings failed to provide the sub-\npopulation of the main endometrial cell, which did not\nfully exploit the advantages of the single-cell technology.\nThus, the results from our study provided more detailed\nand heterogeneous cell populations of the uterus tissue\nacross the menstrual cycles, and more abundant insight\nand resolution to the biology of the human endometria\nregeneration and differentiation.\nAnd, in comparison with another single cell study on\ntemporal and spatial dynamics of human endometria, we\nmainly fully characterized the stromal cell populations,\nwhile their study was mainly on the endometrial epithelial\ncells, with the focus on the molecular mechanisms on the\nFig. 5 SFRP4 + stromal cells promote the regeneration of endometrial epithelial glands and full thickness endometrial injury through\nIGF1 signaling pathway in vivo. a H&E staining of rat uterus 1 week after operation in endometrial injury group (injury), material only group\n(Gelma), SFRP4+ stromal cell therapy group (Gelma + stroma) and cell compound IGF1 inhibitor group (Gelma + stroma + s1091). b comparison of\nendometrial thickness in different operation groups, n = 5, * p < 0.05; *** p < 0.001. c Immunohistochemical staining of FOXA2 showed endometrial\nglands 1 weeks after after operation in injury group, Gelma group, Gelma + stroma group and Gelma + stroma + s1091 group. d quantification of\nendometrial glands in in different operation groups, n = 5. scale bar, 500 μm.\nWu et al. Cell Discovery            (2022) 8:95 Page 11 of 16\n\ndifferentiation of the epithelia towards secretory and\nciliated lineages 25, while our study was mainly focus on\nthe regeneration of the endometria, which may provide\ncomplementary evidences to the whole picture of the\nhuman endometrial biology. In their study, 14 clusters of\ncells were identi fied, which could be grouped into five\nmain cellular categories: (1) immune (lymphoid and\nmyeloid); (2) epithelial (SOX9\n+, lumenal, glandular and\nciliated); (3) endothelial (arterial and venous); (4) sup-\nporting—perivascular cells (PV STEAP4 and PV MYH11),\nsmooth muscle cells and fibroblasts expressing C7\n(fibroblasts C7); and (5) stromal –nondecidualized endo-\nmetrial (eS) and decidualized endometrial (dS). The main\ndifferences between our clusters and clusters in Garcia-\nAlonso et al.'s paper\n25 were the epithelial and stromal cell\nclusters:\nIn our study, there were five epithelial clusters: Anti-\ngen_presenting, EMT, Secretory, proliferative, ciliated.\nFour epithelial clusters were reported in Garcia-Alonso\net al ’s paper\n25 (SOX9+, lumenal, glandular and ciliated)\n(1) SOX9 populations (MMP7 +SOX9+); (2) ciliated cells\n(TPPP3+); (3) lumenal cells (PTGS1 +PAX2+); and (4)\nglandular cells (SCGB2A2 +). So, we map the expression\nof markers of the four epithelial clusters of theirs in our\nepithelial data (Supplementary Fig. S11a), and we find that\nmarker of SOX9 populations (MMP7\n+SOX9+) is highly\nexpressed in the EMT cluster in our study, which shows\nthat the SOX9 populations in their paper is actually EMT\ncluster in our study, marker of ciliated cells (TPPP3 +)i s\nalso highly expressed in CILIATED cluster in our study,\nwhich shows that the ciliated populations in their paper is\nthe same as the CILIATED cluster in our study, marker of\nglandular cells (SCGB2A2\n+) is highly expressed in\nSECRETORY cluster in our study, which shows that the\nglandular populations in their paper is actually SECRE-\nTORY cluster in our study, and marker of lumenal cells\n(PTGS1\n+PAX2+) is highly expressed in ANTI-\nGEN_PRESENTING cluster in our study, while the\nPROLIFERATIVE epithelial cluster in our study also\nexpress moderate level of the SOX9 populations marker\nMMP7. Thus, these results suggest that the epithelial\nclusters identified in our study were consistent and can be\nvalidated in the published single-cell datasets.\nIn case of stromal cell clusters, there were four epithelial\nclusters in our study: secretory stroma, SFRP4\n+ stroma,\nDCN+ stroma, and in flammatory stroma. As shown by\nthe feature plot, secretory stroma expressed high levels of\nSCGB1D2, SFRP4\n+ stroma cells expressed high levels of\nSFRP4, DCN + stroma expressed high levels of DCN,\ninflammatory stroma expressed high levels of IL6. There\nwere three fibroblast/stromal clusters in Garcia-Alonso\net al ’s paper 25: (1) Fibro C7 populations (C7 +); (2) non-\ndecidualized endometrial (eS) (MMP11 +, CRABP2 +) and\n(3) decidualized endometrial (dS) (CFD, IL15, FOXO1).\nSo, we map the expression of markers of the three\nfibroblast/stromal clusters in their paper in our stromal\ncells data (Supplementary Fig. S11b), and we find that\nmarker of C7 populations (C7) is highly expressed in the\nDCN cluster in our study, which shows that the C7\npopulations in their paper is actually DCN cluster in our\nstudy, marker of eS populations (MMP11, CARBP2) is\nhighly expressed in the SFRP4 cluster in our study, which\nshows that the eS populations in their paper is actually\nSFRP4 cluster in our study, marker of dS populations\n(CFD, IL15, FOXO1) is highly expressed in the SECRE-\nTORY and INFLAMMATORY clusters in our study,\nwhich shows that the dS populations in their paper could\nbe further divided into two clusters (SECRETORY and\nINFLAMMATORY) in our study. Thus, these results\nsuggest that the stromal clusters identi fied in our study\nwere also consistent and can be validated in the published\nsingle-cell datasets.\nPrevious studies have shown that endometrial stem/\nprogenitor cells and endometrial mesenchymal stem cells\nare responsible for the regeneration of human and mouse\nendometria\n2,28. As there was an increasing attention on\nfibroblasts heterogeneity and functions in this field,\nfibroblasts were suspected to participate in tissue health\nand diseases through physical or biochemical niches by\nproducing extracellular matrix or soluble signal mole-\ncules\n29. One of the populations being largely proposed as\na regenerative population in the endometrium is the\nstromal SUSD2 +, associated to perivascular areas 2,28,s o\nwe checked the SUSD2 + cells in our data (Supplementary\nFig. S11c, d), and found SUSD was highly enriched in the\nvascular cell population (ADIRF\n+ vascular cells) (Sup-\nplementary Fig. S11d), but there ’s no positive expression\nin any of the stroma cells in our data (Supplementary Fig.\nS11c) and con firms the perivascular nature of the\nSUSD2\n+ cells, and also indicated that these cells were\ndifferent from the regenerative SFRP4 + stroma cells\nfound in our study. In this study, we revealed hetero-\ngeneity in endometrial stromal cells and identi fied a\nregenerative SFRP4\n+ endometrial stromal cell sub-\npopulation that was speci fically enriched in the regen-\nerative stage of endometria during the menstrual cycle.\nSFRP4\n+ stromal cells signi ficantly enhanced the pro-\nliferation of endometrial epithelial organoids in vitro and\npromoted regeneration of endometrial epithelial glands\nand full-thickness endometrial injury in vivo. These\nresults highlight the crucial role of fibroblast-like stromal\ncells during menstrual endometrial repair and regenera-\ntion. SFRP4\n+ stromal cells also provide a novel cell source\nfor tissue engineering and regenerative medicine treat-\nment of endometrial injury, thin endometria, and\nintrauterine adhesions.\nA previous study showed that IGF1 is involved in the\nregulation of reproductive tissues (such as endometria)\nWu et al. Cell Discovery            (2022) 8:95 Page 12 of 16\n\nand functions in the endometrium under the control of\nhormone receptors in mice 30. In this study, we identi fied\nIGF1 as a key endogenous signaling molecule secreted by\nhuman SFRP4 + stromal cells during the regenerative\nphase of the menstrual cycle, which regenerates endo-\nmetrial epithelial organoid formation in vitro, gland for-\nmation in vivo and full-thickness endometrial\nregeneration. This identi fication provided a promising\nbioactive molecule for tissue engineering and regenerative\nmedicine treatment of endometrial injury, thin endome-\ntria, and intrauterine adhesion, as IGF1 was reported to be\nfunctional in the treatment of injury in peripheral Nerve\n31\nand cartilage reconstruction in Osteoarthritis 32. In addi-\ntion to its role in the endometrial epithelia, IGF1 was also\nshown to promote interleukin 10 (IL10) expression in\nbone marrow stem cells, which suggests additional func-\ntions of IGF1 in immune regulation\n33.\nMaterials and methods\nHuman uterus collection\nSeven full-thickness (including endometrium and\nmyometrium layer) normal human uterine samples (all\nwith normal menstrual cycle) from two menstrual phases\n(proliferative and secretory phase) were collected from the\nnormal part of uterus from hysterectomy due to leio-\nmyoma (Supplementary Table S1), after the whole uterus\ntissues were surgically removed from hysterectomy, the\nnormal part of the uterus full-thickness (endometrium\nand myometrium layer) visible to the naked eye were\nselected to avoid the leiomyoma site. Tissues of about\n1 cm in length, 1 cm in width and 2 cm in depth were\nimmediately cut with a scalpel and transported to the\nlaboratory in cell culture medium at a low temperature of\n4 °C for subsequent tissue digestion experiments. All tis-\nsues were all collected from the First Af filiated Hospital,\nSchool of Medicine, Zhejiang University. Approval for\nutilizing the patient samples in this study was obtained\nfrom Ethics Committee of the First Af filiated Hospital,\nSchool of Medicine, Zhejiang University (Approval\nReference Number: 2018-113). Patients taking any hor-\nmones were excluded from the study.\nSingle cell suspension preparation\nA single-cell suspension was prepared as described in\na previous study 34.B r i efly, full-thickness uterus tissue\nwas minced into small cubes wi th scissors, and digested\nin 20 –30 mL digestion enzyme mixtures containing\n1.25 U/mL Dispase II (Sigma, D4693)/0.4 mg/mL col-\nlagenase V (Sigma, C-9263) solution in RPMI 1640\nmedium (ThermoFisher Scienti fic, 21875-034) with\ngentle shaking at 37 °C for 20 –30 min, with the digested\ntissue supernatant neutralized by 10% FBS in RPMI\n1640 medium and replaced by new digestion enzyme\nmixtures every 20 –30 min. Digested cells were collected,\nand red blood cells were removed using red cell lysis\nbuffer (Beyotime Biotech, C3702). The stromal cells and\nsmooth muscles in the neutralized digested tissue\nsupernatant were collected by passing the digested\nsupernatant through 70 μm cell sieves (Corning). The\nepithelial cells were backwashed and further digested\nwith TrypLE (Thermo Fisher Scienti fic) at 37 °C for\n10 min. The digested supernatant was passed through\n70 μm cell sieves to obtain a single epithelial cell sus-\npension. Finally, as we used the different digestion\nmethods to get the stromal cells, myometrial muscles\nand epithelial cells separately, we need to combine the\nstromal cells, smooth muscles and epithelial cells to get\nthe uterus single cell suspension for single cell analysis,\nthe stromal cells, smooth muscles and epithelial cells\nwere combined at a ratio of 1:1:1 to get the uterus single\ncell suspension for further single cell analysis (Fig. 1a).\nSingle cell capture, pre-ampli fication and sequencing and\nBioinformatic analysis\nSingle-cell capture and pre-ampli fication were con-\nducted on a GemCode instrument (10× Genomics)\naccording to the manufacturer ’s ‘instructions (Chro-\nmium™ Single Cell 3 ’ Reagent Kit v2). The generated\nlibrary was sequenced using the Illumina X10 platform\nand the generated sequencing reads were aligned and\nanalyzed using the Cell Ranger Pipeline (10× Genomics).\nThe raw count data of each single cell were deposited into\nthe public database of the Genome Sequence Archive for\nHuman (GSA-Human) under accession number\nHRA000928. Single-cell analysis was conducted using\nSeurat\n9. The potential doublet of single-cell data was\ndetected using DoubleFinder 10. A connectivity map was\nconstructed according to a previous ligand-receptor\ndataset\n35 using CellPhoneDB 16. GO analysis was con-\nducted using http://geneontology.org. Gene set variation\nanalysis (GSVA) was used to perform deconvolution\nanalysis\n36.\nHuman endometrial epithelial organoid co-cultured with\nSFRP4+ stroma\nSingle endometrial epithelial and stromal cell suspensions\nwere digested, as described in the single-cell suspension\npreparation section. SFRP4 stromal cells were digested from\nhuman uterine samples during the proliferative phase. Cell\nsuspensions were centrifuged and resuspended in ice-cold\nMatrigel (Corning, 536231) with 1 × 10\n4 epithelial cells and\n0.5 × 104 stroma per well. Drops of matrigel-cell suspension\n(20 μL) were plated into 48-well plates (Costar, 3548),\nallowed to set at 37 °C, and overlaid with 250 μLe n d o -\nmetrial organoid expansion medium (ExM) with or without\nIGF1 inhibitors. Endometrial organoid expansion medium\n(ExM) was obtained from a previous study 34, containing\nN2 supplement (Gibco), B27 supplement (Gibco), 50 ng/mL\nWu et al. Cell Discovery            (2022) 8:95 Page 13 of 16\n\nE G F( P e p r o t e c h ) ,1 0 0 n g / m LN o g g i n( P e p r o t e c h ) ,5 0 0 n g /\nmL R-spondin-1 (Peprotech), 100 ng/mL FGF10 (Pepro-\ntech), 50 ng/mL HGF (Peprotech), 500 nM ALK-4, -5, -7\ninhibitor (Selleck, A83-01), 10 nM nicotinamide (Sigma),\nand 1.25 mM N-acetyl-L-cysteine (Sigma) in advanced\nDMEM/F12 medium (Gibco, C11330500BT). IGF1 inhibi-\ntors included BMS-536924 (Selleck, s1012) at a final con-\ncentration of 2μM ,N V P - A E W 5 4 1( S e l l e c k ,s 1 0 3 4 )a tafinal\nconcentration of 2.5 μM, linsitinib (Selleck, s1091) at a final\nconcentration of 2 μM, GSK1904529A (Selleck, s1093) at a\nfinal concentration of 5 μM. The medium was changed\nevery 2–3 days. All replicates for each one experiment were\nperformed on organoids derived from a same individual\nbiopsy, and we repeated the same experiment for three\ntimes, and we also repeated the organoid experiments using\ncells from different donors (marked in Supplementary\nTable S1). After endometrial epithelial organoid culture, the\npictures were taken and saved. As shown in the figure\nbelow, the endometrial epithelial organoid spheres in the\npictures are automatically recognized and selected with the\ncount and measure object ’s function plug-in in Image-Pro\nPlus 6.0 (the red logo is the endometrial epithelial organoid\nautomatically recognized by the software), and the software\ncalculates the number of epithelial organoids and the dia-\nmeter of each epithelial organoid ball at the same time for\nfurther statistical analysis (Supplementary Fig. S8d).\nTransfection of siRNA\nIGF1 siRNA were purchased from RiboBio. siRNA\nreagent was dissolved in a stock solution at a concentra-\ntion of 20 μM. The transfection complex reagent,\nincluding 5 μL siRNA stock solution, 83 μL OPTI-MEM\n(Gibco), and 12 μL Lipo2000 (Invitrogen), was mixed and\nvortexed gently and incubated for 10 min at room tem-\nperature (20 °C). The transfection complex reagent was\nadded to a 6-well plate with 2.5 × 10\n5 stromal cells and\ngently mixed with a final concentration of 50 nM siRNA\nin each well. The transfection reagent was replaced with\nfresh culture medium after 4 h incubation in 37 °C incu-\nbator. Stromal cells transfected with siRNAs were used for\nfurther analysis.\nSynthesis of GelMA\nGelMA was fabricated as described previously 37. Type A\ngelatin (Sigma-Aldrich) was dissolved in PBS) at 50 °C to\nobtain a 10% w/v homogeneous solution. Then a 0.1 mL\nmethacrylic anhydride (MA) (Sigma-Aldrich) per gram of\ngelatin was added to the gelatin solution at a rate of\n0.5 mL/min, with continuous stirring. The mixture was\nallowed to react at 50 °C for 3 h. The GelMA solution was\ndialyzed against deionized water using 8 –14 kDa cutoff\ndialysis tubing (VWR Scienti fic USA) for 6 days at 50 °C\nto remove unreacted MA and any byproducts. The\nGelMA solution was frozen overnight at −80 °C, then\nlyophilized, and stored at −20 °C until further use. GelMA\nat a concentration of 10% was used to mix the cells in the\nin vivo study.\nAnimal experiment\nNatural-normal female SD rats (at the age of 8 weeks)\nwere kept in a speci fic pathogen-free air-conditioned\nroom and allowed free access to food and water at the\nAnimal Center of Zhejiang University of Medicine. All\nexperiments were approved by the Animal Experimental\nEthical Inspection of the First Af filiated Hospital, College\nof Medicine, Zhejiang University (2018-095). Ten rats\n(including 20 uterine horns) were randomly divided into\nfour groups ( five uterine horns in each group): endo-\nmetrial injury group (injury), material only group\n(GelMA), human SFRP4\n+ stromal cell therapy group\n(GelMA + stroma), and human SFRP4 + stromal cell\ncompound IGF1 inhibitor group (GelMA + stroma +\ns1091). No immunosuppressive drugs were used during\nthe animal experiments. Endometrial injury was induced\nas follows after the animal was anesthetized: a midline\nincision in the abdomen was made and the uterus was\nexposed. In the injury alone group, a 1 cm longitudinal\nincision was made on the opposite side of the mesome-\ntrium, with the endometrial layer exposed. The endo-\nmetrial layers, with a length of 1 cm and 0.5 cm in width,\nwere then torn off, and the smooth muscle layer remained\nintact (Supplementary Fig. S9). Finally, the injury sites\nwere marked with 6-0 non-absorbable silk sutures, and\nthe longitudinal incision wound was closed after endo-\nmetrial injury. In the repair group, after injury, 15 μL\nGelMA hydrogel with or without 2 × 10\n4 stromal cells and\ns1091 were added to the injured wound site of each\nuterus, then irradiated by UV for 15 s to gelling in situ.\nThe concentration of s1091 was 2 μM in the GelMA\nsolution. After the surgery, the abdominal cavity was\nwashed with 0.9% (w/v) normal saline. Then, the rectus\nabdominis, skin, and fascia were closed using sutures.\nHistology and immuno fluorescence staining\nFull-thickness normal human uterine tissues and rat\nuterine tissues were fixed in 4% (w/v) paraformaldehyde,\ndehydrated in an ethanol gradient, embedded in paraf fin\nand sectioned at a 10 μm thickness. Then, the 10 μm-thick\nparaffin sections were stained with hematoxylin and eosin.\nImmunostaining was carried out as follows: The 10 μm\nparaffin sections were rehydrated, antigen retrieved,\nrinsed three times with PBS, and treated with blocking\nsolution (1% BSA) for 1 h, prior to incubation with pri-\nmary antibodies at 4 °C overnight. The primary antibodies\nSFRP4 (Novus, NBP2-76870), CD10 (Abcam, ab34199),\nand FOXA2 (IHCeasy, KHC0140) were used. Secondary\nantibodies, goat anti-rabbit Alexa Fluor 546 (Invitrogen,\nA11035), donkey anti-mouse Alexa Fluor 488 (Invitrogen,\nWu et al. Cell Discovery            (2022) 8:95 Page 14 of 16\n\nA21202), and DAPI (Beyotime, China), were used to\nvisualize the respective primary antibodies and cell nuclei.\nAll procedures were performed according to the manu-\nfacturer’s instructions.\nStatistical analysis\nA quantitative comparison of the radius and number of\norganoids cultured in different conditioned media was\nconducted using ANOVA in PRISM 5.0, with all P values\nless than 0.05. considered statistically signi ficant. Migra-\ntion of endometrial epithelial cells in 2D culture media\nbetween groups of IGF1 supplement and the control was\ncompared using an unpaired t test in PRISM 5.0, with all\nP values less than 0.05, considered statistically signi ficant.\nQuantitative comparison of the thickness of regenerated\nendometria and number of glands formed among differ-\nent groups was conducted using ANOVA in PRISM\n(version 5.0), with all P values less than 0.05, considered\nstatistically signi ficant. The expression of IGF1 genes\namong groups transfected with different siRNAs against\nIGF1 were compared using ANOVA in PRISM (version\n5.0), with all P values less than 0.05, considered statisti-\ncally signi ficant.\nAcknowledgements\nThis work was supported by the National Key R&D Program of China\n(2018YFC1105102), the National Natural Science Foundation of China\n(81871127, 82171616, 31870973), the Key Scienti fic and Technological\nInnovation Team of Zhejiang Province (2013TD11). Zhejiang Provincial Natural\nScience Foundation of China (LZ22H040001). The Science and Technology\nprogram of Jinhua Science and Technology Bureau (Grant No. 2021-3-001).\nAuthor details\n1Clinical Research Center, the First Af filiated Hospital, School of Medicine,\nZhejiang University, Hangzhou, Zhejiang, China. 2Dr. Li Dak Sum & Yip Yio Chin\nCenter for Stem Cell and Regeneration Medicine, Zhejiang University,\nHangzhou, Zhejiang, China.\n3Zhejiang Provincial Key Laboratory of Tissue\nEngineering and Regenerative Medicine, Hangzhou, Zhejiang, China.\n4International Institutes of Medicine, The 4th Af filiated Hospital of Zhejiang\nUniversity School of Medicine, Hangzhou, Zhejiang, China. 5Chu Kochen\nHonors College, Zhejiang University, Hangzhou, Zhejiang, China. 6Zhejiang\nUniversity-University of Edinburgh Institute, Hangzhou, Zhejiang, China.\n7Department of Gynecology, the First Af filiated Hospital, School of Medicine,\nZhejiang University, Hangzhou, Zhejiang, China\nAuthor contributions\nB.W.: acquisition of clinical sample, data, data analysis and interpretation,\nmanuscript writing; Y.Li: acquisition of sample, sample processing; X.S., W.J.,\nK.Z., C.A.: data analysis; L.G., X.Y.: manuscript preparation; Y.Liu, N.N.: processing\nof sample for immunostaining, scaffold characterization; W.Z., J.H., J.Q.\nacquisition of clinical sample; X.Z.: conception and design, manuscript writing.\nCompeting interests\nThe authors declare no competing interests.\nPublisher’s note\nSpringer Nature remains neutral with regard to jurisdictional claims in\npublished maps and institutional af filiations.\nSupplementary information The online version contains supplementary\nmaterial available at https://doi.org/10.1038/s41421-022-00438-7.\nReceived: 5 January 2022 Accepted: 17 June 2022\nReferences\n1. Ono, M. et al. Side population in human uterine myometrium displays phe-\nnotypic and functional characteristics of myometrial stem cells. Proc. Natl.\nAcad. Sci. USA 104, 18700–5( 2 0 0 7 ) .\n2. Gargett, C. E., Nguyen, H. P. T. & Ye, L. Endometrial regeneration and endo-\nmetrial stem/progenitor cells.Rev. Endocr. Metab. Dis. 13, 235–51 (2012).\n3 . M a y b i n ,J .A .&C r i t c h l e y ,H .O .D .M e n strual physiology: implications for\nendometrial pathology and beyond.Hum. Reprod. Update21,7 4 8–61 (2015).\n4. Camp, J. G. et al. Multilineage commun ication regulates human liver bud\ndevelopment from pluripotency.Nature 546, 533–8 (2017).\n5. Zepp, J. A. et al. Distinct mesenchymal lineages and niches promote epithelial\nself-renewal and myofibrogenesis in the lung. Cell 170,1 1 3 4–48 (2017).\n6. Puram, S. V. et al. Single-cell transcri ptomic analysis of primary and metastatic\ntumor ecosystems in head and neck cancer. Cell 171,1 6 1 1–1624.e24 (2017).\n7. Wan, A. C. A. Recapitulating cell-cell in teractions for organoid construction -\nare biomaterials dispensable?Trends Biotechnol.34,7 1 1–21 (2016).\n8. Bagley, J. A., Reumann, D., Bian, S., Levi-Strauss, J. & Knoblich, J. A. Fused\ncerebral organoids model interactions between brain regions.Nat. Methods\n14,7 4 3–51 (2017).\n9. Satija, R., Farrell, J. A., Gennert, D., Schi e r ,A .F .&R e g e v ,A .S p a t i a lr e c o n s t r u c t i o n\nof single-cell gene expression data.Nat. Biotechnol.33, 495–502 (2015).\n10. McGinnis, C. S., Murrow, L. M. & Gartner, Z. J. DoubletFinder: doublet detection\nin single-cell RNA sequencing data using artificial nearest neighbors.Cell Syst.\n8,3 2 9–337.e4 (2019).\n11. Uhlen, M. et al. Proteomics. tissue-based map of the human proteome.Science\n347, 1260419 (2015).\n12. Talbi, S. et al. Molecular phenotypi ng of human endometrium distinguishes\nmenstrual cycle phases and underly ing biological pro cesses in normo-\novulatory women.Endocrinology147,1 0 9 7–121 (2006).\n13. Fu, B. Q. et al. Natural killer cells promote fetal development through the\nsecretion of growth-promoting factors.Immunity 47,1 1 0 0–1113.e6 (2017).\n14. Cousins, F. L., Kirkwood, P. M., Saunders, P. T. K. & Gibson, D. A. Evidence for a\ndynamic role for mononuclear phagocytes during endometrial repair and\nremodelling.Sci. Rep. 6, 36748 (2016).\n15. Wang, W. et al. Single-cell transcri ptomic atlas of the human endometrium\nduring the menstrual cycle. Nat. Med. 26, 1644–53 (2020).\n16. Vento-Tormo, R. et al. Single-cell reconstruction of the early maternal-fetal\ninterface in humans. Nature 563,3 4 7–53 (2018).\n1 7 . B i l y k ,O . ,C o a t h a m ,M . ,J e w e r ,M .&P o stovit, L. M. Epithelial-to-mesenchymal\nt r a n s i t i o ni nt h ef e m a l er e p r o d u c t i v et r a c t :f r o mn o r m a lf u n c t i o n i n gt od i s e a s e\npathology.Front Oncol. 7, 145 (2017).\n18. Yue, K. et al. Synthesis, properties, and biomedical applications of gelatin\nmethacryloyl (GelMA) hydrogels.Biomaterials73,2 5 4–71 (2015).\n19. Fassnacht, M. et al. Linsitinib (OSI-906 ) versus placebo for patients with locally\nadvanced or metastatic adrenocortical carcinoma: a double-blind, rando-\nmised, phase 3 study. Lancet Oncol. 16,4 2 6–35 (2015).\n20. Diaz-Gimeno, P., Ruiz-A lonso, M., Blesa, D. & Simon, C. Transcriptomics of the\nhuman endometrium.I n tJ .D e v .B i o l .58,1 2 7–37 (2014).\n21. Evans, G. E. et al. In the secretory en dometria of women, luminal epithelia\nexhibit gene and protein expressions that differ from those of glandular\nepithelia.Fertil. Steril. 102,3 0 7–317.e7 (2014).\n22. Proserpio, V. & Lonnberg, T. Single-ce ll technologies are revolutionizing the\napproach to rare cells. Immunol. Cell Biol. 94,2 2 5–9( 2 0 1 6 ) .\n23. Wu, B. et al. Reconstructing lineage hierarchies of mouse uterus epithelial\ndevelopment using single-cell analysis.Stem Cell Rep. 9, 381–96 (2017).\n24. Krjutskov, K. et al. Single-cell transcriptome analysis of endometrial tissue.Hum.\nReprod. 31,8 4 4–53 (2016).\n25. Garcia-Alonso, L. et al. Mapping the temporal and spatial dynamics of the\nhuman endometrium in vivo and in vitro. Nat. Genet. 53,1 6 9 8–711 (2021).\n26. Koh, W. et al. Single cell transcriptomes derived from human cervical and\nuterine tissue during pregnancy.Adv. Biosyst. 3, e1800336 (2019).\n2 7 . F i t z g e r a l d ,H .C . ,D h a k a l ,P . ,B e h u r a ,S .K . ,S c h u s t ,D .J .&S p e n c e r ,T .E .S e l f -\nrenewing endometrial epithelial organoids of the human uterus. Proc. Natl\nAcad. Sci. USA\n116,2 3 1 3 2–42 (2019).\n2 8 . L v ,Q . ,W a n g ,L . ,L u o ,X .&C h e n ,X .A d u l ts t e mc e l l si ne n d o m e t r i a lr e g e n -\neration: Molecular insights and clinical applications. Mol. Reprod. Dev. 88,\n379–94 (2021).\nWu et al. Cell Discovery            (2022) 8:95 Page 15 of 16\n\n29. Plikus, M. V. et al. Fibroblasts: Origins, de finitions, and functions in health and\ndisease. Cell 184,3 8 5 2–72 (2021).\n30. Ogo, Y. et al. IGF-1 gene expression i s differentially regulated by estrogen\nreceptors alpha and beta in mouse endometrial stromal cells and ovarian\ngranulosa cells.J. Reprod. Dev. 60, 216–23 (2014).\n31. Slavin, B. R. et al. Insulin-like gro wth factor-1: a promising therapeutic\ntarget for peripheral nerve injury. Front Bioeng. Biotechnol. 9, 695850\n(2021).\n32. Hossain, M. A. et al. IGF-1 facilitates cartilage reconstruction by regulating PI3K/\nAKT, MAPK, and NF-kB signaling in rabbit osteoarthritis. J. Inflamm. Res. 14,\n3555–68 (2021).\n33. Wang, L. et al. Transplant of insulin -like growth factor-1 expressing bone\nmarrow stem cells improves functional regeneration of injured rat uterus by\nNF-kappaB pathway.J. Cell Mol. Med 22,2 8 1 5–25 (2018).\n34. Turco, M. Y. et al. Long-term, hormone-responsive organoid cultures of human\nendometrium in a chemically defined medium.Nat. Cell Biol.19,5 6 8–77 (2017).\n3 5 . R a m i l o w s k i ,J .A .e ta l .Ad r a f tn e t w o rk of ligand-receptor-mediated multi-\ncellular signalling in human.Nat. Commun. 7, 7866 (2016). (vol 6, 7866, 2015).\n36. Diaz-Mejia, J. J. et al. Evaluation of methods to assign cell type labels to cell clusters\nfrom single-cell RNA-sequencing data.F1000Res8, ISCB Comm J-296 (2019).\n37. Hong, Y. et al. A strongly adhesive hemostatic hydrogel for the repair of\narterial and heart bleeds. Nat. Commun. 10,2 0 6 0( 2 0 1 9 ) .\nWu et al. Cell Discovery            (2022) 8:95 Page 16 of 16","source_license":"CC0","license_restricted":false}