Decoding functional and developmental trajectories of tissue-resident uterine dendritic cells through integrative omics.

Conceptualization, A.S., L.D., and G.M.; methodology, A.S., N.A., E.K., L.Y., J.Y., S.C., Q.L., M.R.G., J.F., and G.M.; investigation, A.S., N.A., E.K., L.Y., J.Y., S.C., Q.L., M.R.G., J.F., and G.M.; writing—original draft, A.S.; writing—review & editing, A.S., J.D., L.D., M.R.G., and G.M.; funding acquisition, G.M.; resources, L.D. and G.M.; supervision, G.M. and L.D.

To characterize the uDC population present in the human endometrium throughout the menstrual cycle, we performed single-cell RNA sequencing (scRNA-seq) analysis and determined the transcriptomic landscape of uDCs using LYZ+ and IRF8+ markers. A total of 2,229 DCs were extracted from uterine samples across different menstrual cycle stages: proliferative (early, mid, and late); secretory (early, mid, and late); menstrual; and early pregnancy (decidua basalis and decidua parietalis; Figures 1 A and 1B). Figure 1 Comprehensive analysis of uterine dendritic cell (uDC) subsets and their gene expression profiles (A) Schematic representation of the sample collection process and subsequent single-cell RNA sequencing (scRNA-seq), CITE-seq, and multiplex immunohistochemistry (IHC) workflow. Uterine tissue samples were enzymatically digested to obtain single-cell suspensions, which were then sorted via FACS to isolate CD45 + immune cells. The isolated cells underwent scRNA-seq analysis using the involving MARS-seq2.0 protocol. Subsequent bioinformatic analysis enabled the identification and characterization of distinct dendritic cell subpopulations. The findings were further validated through wet-lab techniques, including flow cytometry and IHC, to confirm the presence of these dendritic cell subpopulations. (B) UMAP plot showing all immune cells in gray and cells expressing LYZ and IRF8 in blue. Small subsets (indicated in boxes) were identified as dendritic cells, which was selected and used for downstream analysis. (C) UMAP plot visualizing the clustering of uDCs based on their gene expression profiles. Each point represents a single cell, colored according to the identified cluster (0–6), illustrating the distinct transcriptional landscapes within the uDC population. (D) Heatmap showing the expression of top 10 marker genes (shown on y axis) across different uDC clusters (shown on x axis). Each row represents a gene, and each column corresponds to a single cell, organized by cluster identity. Yellow color denotes upregulated genes, whereas pink denotes expression of the downregulated genes. (E) Spatial distribution for key markers ( LYZ , IRF8 , and CD14 ) across the uDC clusters shown on UMAP. (F) Expression level for key marker genes ( LYZ , IRF8 , and CD1 4) was quantified using violin plots. X -axis shows the clusters and y axis shows the expression level of the genes indicated on top of each plot. Comprehensive analysis of uterine dendritic cell (uDC) subsets and their gene expression profiles (A) Schematic representation of the sample collection process and subsequent single-cell RNA sequencing (scRNA-seq), CITE-seq, and multiplex immunohistochemistry (IHC) workflow. Uterine tissue samples were enzymatically digested to obtain single-cell suspensions, which were then sorted via FACS to isolate CD45 + immune cells. The isolated cells underwent scRNA-seq analysis using the involving MARS-seq2.0 protocol. Subsequent bioinformatic analysis enabled the identification and characterization of distinct dendritic cell subpopulations. The findings were further validated through wet-lab techniques, including flow cytometry and IHC, to confirm the presence of these dendritic cell subpopulations. (B) UMAP plot showing all immune cells in gray and cells expressing LYZ and IRF8 in blue. Small subsets (indicated in boxes) were identified as dendritic cells, which was selected and used for downstream analysis. (C) UMAP plot visualizing the clustering of uDCs based on their gene expression profiles. Each point represents a single cell, colored according to the identified cluster (0–6), illustrating the distinct transcriptional landscapes within the uDC population. (D) Heatmap showing the expression of top 10 marker genes (shown on y axis) across different uDC clusters (shown on x axis). Each row represents a gene, and each column corresponds to a single cell, organized by cluster identity. Yellow color denotes upregulated genes, whereas pink denotes expression of the downregulated genes. (E) Spatial distribution for key markers ( LYZ , IRF8 , and CD14 ) across the uDC clusters shown on UMAP. (F) Expression level for key marker genes ( LYZ , IRF8 , and CD1 4) was quantified using violin plots. X -axis shows the clusters and y axis shows the expression level of the genes indicated on top of each plot. The analysis of LYZ +/ IRF8+ DCs revealed a rich diversity within the uDC population, identifying seven distinct clusters depicted in a heatmap, with the top 10 genes for each cluster ( Figure 1 D). The heatmap illustrates distinct gene expression profiles for clusters 1 through 5, while clusters 0 and 6 demonstrate overlapping gene expression patterns, suggesting a shared transcriptional signature between these two subtypes of uDCs ( Figure 1 D). To better identify the different subpopulations of uDC, we used Uniform Manifold Approximation and Projection (UMAP) analysis revealing seven distinct subtypes of uDCs ( Figure 1 C). This unbiased clustering analysis did not rely on known DC markers. UMAP highlighted the complexity and heterogeneity within the uDC population, indicating multiple distinct transcriptomic programs within what was previously considered a singular cell type. We confirmed the identified populations as DCs, by assessing LYZ and IRF8 expression on the UMAP. Interestingly, all the clusters, except for cluster 1, were positive for LYZ ( Figure 1 E) while cluster 1 was positive for IRF8 . Only clusters 2 and 3 were double-positive for LYZ + /IRF8 + ( Figure 1 E). Violin plots illustrate the expression level of these genes ( Figure 1 F), and none of the clusters showed CD14 + cells (monocyte/macrophage marker), confirming the specificity of the approach for classical DCs ( Figure 1 F). The existing dogma is that DCs are recruited from the peripheral blood to the uterus. 21 From the cluster analysis, we observed that cluster 1 is the most diverse cluster in terms of gene signature and markers expression ( LYZ − /IRF8 + ; Figure 1 D); therefore, we hypothesized that cluster 1 might have been recruited from peripheral blood while the rest of the clusters are tissue-resident DCs. To test this hypothesis, we compared the newly identified clusters of uDCs ( Figure 2 A) to DC subpopulations present in peripheral blood mononuclear cells (PBMCs) using the publicly available scRNA-seq data ( Figure 2 B). 22 Our UMAP analysis delineated eight distinct clusters among the PBMC DCs ( Figure 2 B). We then performed a meta-analysis for the transcriptome of peripheral-blood-recruited uDCs (PB-uDCs) and peripheral blood DC clusters and found greatest gene overlap between PB-uDC and cluster 3 of the peripheral blood DCs ( Figures 2 A and 2B), which has been described as pDCs. 22 We compared the top 50 genes from cluster 1 uDCs to cluster 3 peripheral blood DCs and found 22 genes to be shared ( Figure 2 C), while similar number of genes are distinct for each cell type. These findings suggest that upon migrating into the uterus, cluster 1 undergoes differentiation acquiring characteristics specific to uDCs. Based on these findings, we referred cluster 1 as PB-uDC. Figure 2 Comparative analysis of peripheral blood mononuclear cell (PBMC) DCs and uterine DCs (A) UMAP visualization of DC from the uterine tissues showing seven clusters and cluster 1 identified as PB-uDCs. (B) UMAP visualization of DC from the PBMC dataset showing eight clusters, with cluster 3 highlighted, which was identified as plasmacytoid DCs. (C) Heatmap showing number of genes shared from comparing the top 50 markers from clusters 0–7 from PBMC and clusters 0–6 from uDCs. Cluster 1 from uDC has the highest number of shared genes with cluster 3 of PBMC DCs highlighted in black, and the list of 22 shared genes is shown on the side. The number of genes shared with each cluster of uDCs with each cluster of PB-DCs are shown in the box. (D) UMAP showing expressions of marker genes such as GZMB , JCHAIN , and IRF8 within cluster 1. Violin plots below show the quantification of expression levels of these marker genes in different clusters. GZMB and JCHAIN were specifically expressed in cluster 1, whereas IRF8 was found in clusters 1, 2, 3, and 4. (E) Circos plots showing the biological processes for the differentially expressed genes in the PB-uDC cluster along with the top 10 genes associated with each pathway. Red color indicates the genes were upregulated; blue color indicates that the genes were downregulated. (F) UMAP showing co-expression of the marker genes of the cluster— GZMB , JCHAIN , and IRF8. Yellow color indicates genes are co-expressed. (G) UMAP showing the expression of the marker gene GZMB across 10 phases of the menstrual cycle, from the menstrual stage through to decidualization (Decp and Decb), indicating dynamic changes in gene expression. Alongside is shown a line graph combined with a scatterplot overlay depicting the expression level of GZMB marker gene (1st y axis) and number of cells expressing the gene (2nd y axis) across different stages of the menstrual cycle ( x axis). Each point represents the expression level of a gene in a particular cluster, while the line graph emphasizes the overall expression trend, aiding in the identification of patterns or anomalies in gene expression related to specific menstrual phases. Comparative analysis of peripheral blood mononuclear cell (PBMC) DCs and uterine DCs (A) UMAP visualization of DC from the uterine tissues showing seven clusters and cluster 1 identified as PB-uDCs. (B) UMAP visualization of DC from the PBMC dataset showing eight clusters, with cluster 3 highlighted, which was identified as plasmacytoid DCs. (C) Heatmap showing number of genes shared from comparing the top 50 markers from clusters 0–7 from PBMC and clusters 0–6 from uDCs. Cluster 1 from uDC has the highest number of shared genes with cluster 3 of PBMC DCs highlighted in black, and the list of 22 shared genes is shown on the side. The number of genes shared with each cluster of uDCs with each cluster of PB-DCs are shown in the box. (D) UMAP showing expressions of marker genes such as GZMB , JCHAIN , and IRF8 within cluster 1. Violin plots below show the quantification of expression levels of these marker genes in different clusters. GZMB and JCHAIN were specifically expressed in cluster 1, whereas IRF8 was found in clusters 1, 2, 3, and 4. (E) Circos plots showing the biological processes for the differentially expressed genes in the PB-uDC cluster along with the top 10 genes associated with each pathway. Red color indicates the genes were upregulated; blue color indicates that the genes were downregulated. (F) UMAP showing co-expression of the marker genes of the cluster— GZMB , JCHAIN , and IRF8. Yellow color indicates genes are co-expressed. (G) UMAP showing the expression of the marker gene GZMB across 10 phases of the menstrual cycle, from the menstrual stage through to decidualization (Decp and Decb), indicating dynamic changes in gene expression. Alongside is shown a line graph combined with a scatterplot overlay depicting the expression level of GZMB marker gene (1st y axis) and number of cells expressing the gene (2nd y axis) across different stages of the menstrual cycle ( x axis). Each point represents the expression level of a gene in a particular cluster, while the line graph emphasizes the overall expression trend, aiding in the identification of patterns or anomalies in gene expression related to specific menstrual phases. Next, we defined the functional characteristics of this cluster. So, we extracted all differentially expressed p -value-significant genes ( p < 0.05), and log fold change of at least 0.6, for the PB-uDC subtype and used iPathway analysis. The biological processes that were found include immune response, cell communication, signaling, cell activation, locomotion, leukocyte activation, response to stress, localization, motility, signal transduction, and defense response, with the genes associated with each process ( Figure 2 E). To determine specific markers and characteristics of PB-uDCs, we identified DC-genes GZMB and JCHAIN from the bubble plot ( Figure S1 ). We observed that GZMB and JCHAIN are only expressed in PB-uDCs ( Figure 2 D), and co-expression of GZMB and JCHAIN is only detected in PB-uDCs ( Figure 2 F). IRF8 is detected in PB-uDCs as well as in clusters 2 and 3; however, the co-expression of GZMB and IRF8 is found only in PB-uDCs ( Figure 2 F). Violin plots confirmed that JCHAIN and GZMB were specific to PB-uDCs (cluster 1), whereas IRF8 was found in clusters 1, 2, 3, and 4 as well ( Figure 2 F). Finally, we evaluated the presence of PB-uDCs during different stages of the menstrual cycle. PB-uDCs were nearly undetectable in the endometrial samples collected at menstruation ( Figure 2 G). Their numbers increase at early proliferative, peak at mid-proliferative, and decrease afterward ( Figure 2 G). PB-uDCs are not detected in early decidua parietalis nor basalis ( Figure 2 G). This finding supports our hypothesis that cluster 1 represents a subset of uDCs recruited from peripheral blood in contrast with the other clusters that might have differentiated from local precursor cells. While cluster 1 (PB-uDCs) showed major separation in the UMAP, the rest of the clusters, although different, grouped closer together ( Figure 1 C). Consequently, we hypothesized that the remaining clusters may represent tissue-resident uDCs and have their origin from precursor cells located within the endometrium ( Figure 3 A). To test this hypothesis, we sought to ascertain a cluster with characteristics of progenitor cells by looking at the expression of KLF4 , NANOG , AXL , CX3CR1 , CD1C , and CLEC10A . Figure 3 Identification and characterization of progenitor cells on the UMAP (A) UMAP visualization uDCs with cluster 0 being identified as progenitor uDCs (P-uDCs). (B) UMAP plot depicts the expression of key marker genes for stem cells, including KLF4 and NANOG , as well as known marker genes for DC progenitors, such as AXL , CLEC10A , CX3CR1 , and CD1C . In the plot, all cells are represented in gray, while cells expressing the genes listed at the top of the UMAP are highlighted in purple. (C) UMAP highlighting co-expression patterns of DC progenitor marker genes. This panel illustrates cells co-expressing progenitor marker genes AXL and CX3CR1 , as well as cells co-expressing AXL and CD1C , within the UMAP. (D) Violin plots showing quantification for stem cell marker genes and DC progenitor marker genes. y axis shows the expression level, and x axis shows the clusters. (E) Lineage tracing using predictive slingshot indicates cluster 0 is progenitor from which all other clusters originate. The lines originate from cluster 0 and go through cluster 3 to cluster 2; another line originating from cluster 0 goes directly to clusters 4 and 5. (F) Circos plots showing the biological processes found for the genes differentially expressed in the P-uDC cluster along with the top 10 genes associated with each pathway. The genes are color coded red for upregulation and blue for downregulation. (G) UMAP showing the expression of the marker gene CLEC10A across 10 phases of the menstrual cycle, from the menstrual stage through to decidualization (Decp and Decb), indicating dynamic changes in gene expression. Alongside is a line graph combined with a scatterplot overlay depicting the expression level of CLEC10A marker genes (1st y axis) and number of cells expressing the gene (2nd y axis) across different stages of the menstrual cycle ( x axis). Each point represents the expression level of a gene in a particular cluster, while the line graph emphasizes the overall expression trend. Identification and characterization of progenitor cells on the UMAP (A) UMAP visualization uDCs with cluster 0 being identified as progenitor uDCs (P-uDCs). (B) UMAP plot depicts the expression of key marker genes for stem cells, including KLF4 and NANOG , as well as known marker genes for DC progenitors, such as AXL , CLEC10A , CX3CR1 , and CD1C . In the plot, all cells are represented in gray, while cells expressing the genes listed at the top of the UMAP are highlighted in purple. (C) UMAP highlighting co-expression patterns of DC progenitor marker genes. This panel illustrates cells co-expressing progenitor marker genes AXL and CX3CR1 , as well as cells co-expressing AXL and CD1C , within the UMAP. (D) Violin plots showing quantification for stem cell marker genes and DC progenitor marker genes. y axis shows the expression level, and x axis shows the clusters. (E) Lineage tracing using predictive slingshot indicates cluster 0 is progenitor from which all other clusters originate. The lines originate from cluster 0 and go through cluster 3 to cluster 2; another line originating from cluster 0 goes directly to clusters 4 and 5. (F) Circos plots showing the biological processes found for the genes differentially expressed in the P-uDC cluster along with the top 10 genes associated with each pathway. The genes are color coded red for upregulation and blue for downregulation. (G) UMAP showing the expression of the marker gene CLEC10A across 10 phases of the menstrual cycle, from the menstrual stage through to decidualization (Decp and Decb), indicating dynamic changes in gene expression. Alongside is a line graph combined with a scatterplot overlay depicting the expression level of CLEC10A marker genes (1st y axis) and number of cells expressing the gene (2nd y axis) across different stages of the menstrual cycle ( x axis). Each point represents the expression level of a gene in a particular cluster, while the line graph emphasizes the overall expression trend. CX3CR1 and CLEC10A were identified as specific markers for cluster 0 by bubble plot ( Figure S1 ). KLF4 and AXL were found expressed in all the clusters but PB-uDCs (cluster 1), confirming a common origin for the rest of the clusters ( Figure 3 B). CX3CR1 , CLEC10A, and CD1C were localized only in clusters 0 and 3 but not in clusters 4, 5, and 2, signifying that either cluster 0 or cluster 3 is identified as the tissue P-uDCs ( Figure 3 B). We observed that CX3CR1 is mainly expressed in cluster 0 and few cells also expressing CX3CR1 in cluster 3 ( Figure 3 B). CLEC10A and CD1C are mostly expressed in cluster 0 but also in a subgroup of clusters 3 ( Figure 3 B). AXL is expressed on all the clusters but cluster 1, suggesting AXL as a common marker for tissue-resident uDCs ( Figure 3 B). AXL was found to be co-expressed with CX3CR1 and CD1C in cluster 0 and slightly in cluster 3 ( Figure 3 C). Violin plots validate that CX3CR1 , CLEC10A , and CD1C were specifically found mainly in cluster 0 ( Figure 3 D). NANOG was not found expressed in any of the clusters ( Figures 3 B and 3D). To determine which of the two clusters may represent the tissue P-uDCs, we utilized the Slingshot algorithm for lineage tracing and observed a developmental trajectory from cluster 0 through cluster 3 to cluster 2 and to clusters 5, 6, and 4 ( Figure 3 E). This path supports the hypothesis that cluster 0 indeed serves as a progenitor state giving rise to other differentiated DC populations. Functional characteristics of this progenitor cluster were analyzed by examining the 784 differentially expressed genes (DEGs) that were statistically significant ( p value <0.05), and log fold change of at least 0.6, using iPathway analysis. 23 , 24 , 25 The functional characteristics of this cluster, depicted by the gene regulatory networks and signaling pathways, are associated with regulation of inflammation and tolerance, involving complement and type I and II interferon ( Figure 3 F). Related to the process of tolerance (allograft rejection), we identified IL-1b , FCGR2b , IGSF6 , CD1D , IL-18 , and CTSH , highly expressed in this cluster. These genes have been identified as immune regulatory factors modulating inflammation, immune cell activation, and antigen presentation. 26 , 27 , 28 , 29 , 30 Using CLEC10A as a specific marker for P-uDCs, we evaluated their presence during the different stages of the menstrual cycle. P-uDC numbers increase during the mid-proliferative and late secretory phases of the menstrual cycle ( Figure 3 G). These findings define cluster 0 as the potential tissue-resident P-uDCs. Since we observed multiple similarities between cluster 0 (P-uDCs) and cluster 3, we hypothesized that cluster 3 may represent a transitional stage of DC differentiation ( Figure 4 A). To test this, we analyzed specific genes for this cluster using bubble plot and identified PCLAF , TOP2A , MKI67 , TYMS , and TK1 ( Figure S1 ). Interestingly, these genes are associated with cell proliferation 31 , 32 , 33 , 34 and were found to be expressed only in cluster 3 ( Figure 4 B). The co-expression and expression levels across different cells within the cluster were quantified and visualized using violin plots, demonstrating significant expression of these proliferative markers ( Figures 4 C and 4D). Figure 4 Identification and characterization of proliferating/transitional uDCs (T-uDCs) (A) UMAP visualization uDCs with cluster 3 being identified as T-uDCs. (B) UMAP plot depicts the expression of key marker genes for proliferating cells, including PCLAF , TOP2A , PCNA , BIRC5 , MCM2 , and MKI67 . In the plot, all cells are represented in gray, while cells expressing the genes listed at the top of the UMAP are highlighted in purple. (C) Highlighting co-expression patterns of proliferative marker genes. This panel illustrates cells co-expressing progenitor marker genes PCLAF and TOP2A , as well as cells co-expressing MCM2 and MKI67 , within the UMAP. (D) Violin plots for proliferation marker genes are shown. y axis shows the expression level, and x axis shows the clusters. (E) Circos plots showing the biological processes for the T-uDC cluster along with the genes associated with each pathway. The genes are color coded red for upregulation and blue for downregulation. Processes are mostly cell-cycle-related such as DNA metabolic process, chromosome organization, DNA repair, recombination, replication, and packaging. (F) UMAP showing the expression of the marker gene PCLAF across 10 phases of the menstrual cycle, from the menstrual stage through to decidualization (Decp and Decb), indicating dynamic changes in gene expression. Below is shown a line graph combined with a scatterplot overlay depicting the expression level of PCLAF (1st y axis) and number of cells expressing the gene (2nd y axis) across different stages of the menstrual cycle ( x axis). Each point represents the expression level of a gene in a particular cluster, while the line graph emphasizes the overall expression trend. Identification and characterization of proliferating/transitional uDCs (T-uDCs) (A) UMAP visualization uDCs with cluster 3 being identified as T-uDCs. (B) UMAP plot depicts the expression of key marker genes for proliferating cells, including PCLAF , TOP2A , PCNA , BIRC5 , MCM2 , and MKI67 . In the plot, all cells are represented in gray, while cells expressing the genes listed at the top of the UMAP are highlighted in purple. (C) Highlighting co-expression patterns of proliferative marker genes. This panel illustrates cells co-expressing progenitor marker genes PCLAF and TOP2A , as well as cells co-expressing MCM2 and MKI67 , within the UMAP. (D) Violin plots for proliferation marker genes are shown. y axis shows the expression level, and x axis shows the clusters. (E) Circos plots showing the biological processes for the T-uDC cluster along with the genes associated with each pathway. The genes are color coded red for upregulation and blue for downregulation. Processes are mostly cell-cycle-related such as DNA metabolic process, chromosome organization, DNA repair, recombination, replication, and packaging. (F) UMAP showing the expression of the marker gene PCLAF across 10 phases of the menstrual cycle, from the menstrual stage through to decidualization (Decp and Decb), indicating dynamic changes in gene expression. Below is shown a line graph combined with a scatterplot overlay depicting the expression level of PCLAF (1st y axis) and number of cells expressing the gene (2nd y axis) across different stages of the menstrual cycle ( x axis). Each point represents the expression level of a gene in a particular cluster, while the line graph emphasizes the overall expression trend. To understand the function of the cluster, we extracted the 1,000 DEGs that were statistically significant ( p value <0.05) and had a log fold change of at least 0.6. The functional analysis of this cluster was characterized by an abundance of cell-cycle-associated pathways, including E2F targets, MYC targets, and G2/M checkpoints ( Figure 4 E). This enrichment in pathways critical for cell proliferation confirms the transitional and proliferative nature of the cells in cluster 3. The pathway analysis thus serves as a validating cross-reference, substantiating the functional insights derived from the UMAP analysis. Using PCLAF as a marker for cluster 3, we determined the temporal dynamics to understand how the cluster changes over different stages of the menstrual cycle. Cluster 3 appeared at the mid-proliferative, peaked at the late secretory, and declined rapidly afterward ( Figure 4 F). Very few cells from this cluster are observed in decidual samples ( Figure 4 F). Based on these findings, we define cluster 3 as transitional uDCs (T-uDCs). Having identified the progenitor and transitional uDC, we investigated whether these two clusters could generate conventional uDC (C-uDC). Cluster 2 was identified as a candidate for C-uDCs based on shared expression of LYZ with P-uDC and T-uDC and the shared expression of IRF8 with PB-uDC ( Figure 5 B). This suggests that these two clusters 0 and 3 (P-uDCs and T-uDCs) and cluster 2 have a tissue resident origin, while PB-uDCs and cluster 2 share some maturation characteristics. Figure 5 Identification and characterization of conventional uDCs (C-uDCs) on the UMAP (A) UMAP visualization uDCs with cluster 2 being identified as C-uDCs. (B) UMAP projection displays the distribution of all analyzed cells (depicted in gray) alongside the subpopulation expressing cDC markers CLEC9A , XCR1 , LYZ , and IRF8 (highlighted in blue). (C) Illustrates the subset of cells that co-express CLEC9A and XCR1 , indicative of cDC identity (top panel). Middle panel depicts a distinct group of cells characterized by the co-expression of both progenitor ( AXL ) and cDC markers ( CLEC9A ), representing a potential DC subset arising from the resident progenitor DCs. Bottom panel shows co-expression of LYZ and IRF8 in the cluster 2, both the markers used for isolation of uDCs from all immune cells. (D) Violin plots quantifying the expression level and cluster specificity for the C-uDC marker genes. (E) Circos plots showing the biological processes for the C-uDC cluster along with the genes associated with each pathway. (F) UMAP showing the expression of the marker gene XCR1 across 10 phases of the menstrual cycle, from the menstrual stage through to decidualization (Decp and Decb), indicating dynamic changes in gene expression. Alongside is shown a line graph combined with a scatterplot overlay depicting the expression level of XCR1 (1st y axis) and number of cells expressing the gene (2nd y axis) across different stages of the menstrual cycle ( x axis). Each point represents the expression level of a gene in a particular cluster, while the line graph emphasizes the overall expression trend, and the expression peaks in late-secretory (WOI) period. Identification and characterization of conventional uDCs (C-uDCs) on the UMAP (A) UMAP visualization uDCs with cluster 2 being identified as C-uDCs. (B) UMAP projection displays the distribution of all analyzed cells (depicted in gray) alongside the subpopulation expressing cDC markers CLEC9A , XCR1 , LYZ , and IRF8 (highlighted in blue). (C) Illustrates the subset of cells that co-express CLEC9A and XCR1 , indicative of cDC identity (top panel). Middle panel depicts a distinct group of cells characterized by the co-expression of both progenitor ( AXL ) and cDC markers ( CLEC9A ), representing a potential DC subset arising from the resident progenitor DCs. Bottom panel shows co-expression of LYZ and IRF8 in the cluster 2, both the markers used for isolation of uDCs from all immune cells. (D) Violin plots quantifying the expression level and cluster specificity for the C-uDC marker genes. (E) Circos plots showing the biological processes for the C-uDC cluster along with the genes associated with each pathway. (F) UMAP showing the expression of the marker gene XCR1 across 10 phases of the menstrual cycle, from the menstrual stage through to decidualization (Decp and Decb), indicating dynamic changes in gene expression. Alongside is shown a line graph combined with a scatterplot overlay depicting the expression level of XCR1 (1st y axis) and number of cells expressing the gene (2nd y axis) across different stages of the menstrual cycle ( x axis). Each point represents the expression level of a gene in a particular cluster, while the line graph emphasizes the overall expression trend, and the expression peaks in late-secretory (WOI) period. Gene Ontology (GO) analysis of differentially expressed genes that were significant ( p < 0.05), and had a log fold change of at least 0.6, revealed some shared biological pathways between cluster 2 and PB-uDCs, indicating functional similarities ( Figures 5 E and 2 D). Specific markers for cluster 2, identified from the bubble plot ( Figure S1 ), included CLEC9A and XCR1 , distinguishing it from PB-uDCs ( Figures S1 and 5 B). The co-expression of CLEC9A and XCR1 further confirmed that cluster 2 represents C-uDCs ( Figure 5 C). Violin plots quantified the gene expression across different clusters, highlighting their significant expression in cluster 2 ( Figure 5 D). These findings support the identification of cluster 2 as C-uDCs. Temporal analysis of cluster 2 during the menstrual cycle revealed that C-uDCs were detected during early to mid-proliferative phase, peaking in the late secretory stage, which coincides with the window of implantation ( Figure 5 F), suggesting a potential role during embryo implantation. A major function of DCs is antigen presentation to T cells in the lymph node to induce either tolerance or T cell activation. 35 We hypothesized that one of the identified clusters within the tissue-resident uDCs could reveal a transcriptome signature indicative of migratory/antigen presentation potential. To test our hypothesis, we focused on cluster 4 ( Figure 6 A) and examined the marker genes. We identified LAMP3 , CCR7 , BIRC3 , and FSCN1 specifically expressed in cluster 4 ( Figures 6 B and S1 ). LAMP3 is involved in the lysosomal function and membrane trafficking and is a marker for mature DCs, particularly those that migrate to lymph nodes. 36 CCR7 is crucial for the migration of DCs to lymph nodes, guiding them from peripheral tissues to the lymphoid organs. 10 , 36 BIRC3 , also known as cIAP2 , influences the survival of DCs, ensuring that they live long enough to effectively migrate and present antigens. 36 , 37 In migratory DCs, FSCN1 facilitates the formation of DC projections (dendrites), aiding in their motility and interaction with T cells in the lymph nodes. 38 The specific expression of these markers in cluster 4 was quantified using violin plots ( Figure 6 C), and their co-expression is shown ( Figure 6 D). Figure 6 Identification and characterization of migratory uDCs (M-uDCs) in the human endometrium (A) UMAP visualization of uDCs, with cluster 4 identified as M-uDCs. (B) UMAP displaying the distribution of all analyzed cells (gray) and the sub populations expressing the migratory DC markers LAMP3 , CCR7 , BIRC3 , and FSCN1 (highlighted in blue). (C) Violin plots quantifying the expression levels and cluster specificity for the M-uDCs marker genes LAMP3 , CCR7 , BIRC3 , and FSCN1 . (D) Illustrates the subset of cells co-expressing LAMP3 with CCR7 and BIRC3, indicative of migratory identity. Co-expression analysis is shown with color thresholds to highlight the specific gene interactions. (E) Circos plots showing the biological processes for the M-uDC cluster along with the genes associated with each pathway. (F) UMAP showing the expression of the marker gene LAMP3 across 10 phases of the menstrual cycle, from the menstrual stage through to decidualization (Decp and Decb). Below is a line graph combined with scatterplot overlay depicting the expression levels of LAMP3 (1st y axis) and the number of cells expressing the gene (2nd y axis) across different stages of the menstrual cycle ( x axis). Each point represents the expression level of LAMP3 in a particular cluster, while the line graph emphasizes the overall expression trend, with expression peaks during the mid-proliferative to late secretory phases with some presence in early decidual tissues. Identification and characterization of migratory uDCs (M-uDCs) in the human endometrium (A) UMAP visualization of uDCs, with cluster 4 identified as M-uDCs. (B) UMAP displaying the distribution of all analyzed cells (gray) and the sub populations expressing the migratory DC markers LAMP3 , CCR7 , BIRC3 , and FSCN1 (highlighted in blue). (C) Violin plots quantifying the expression levels and cluster specificity for the M-uDCs marker genes LAMP3 , CCR7 , BIRC3 , and FSCN1 . (D) Illustrates the subset of cells co-expressing LAMP3 with CCR7 and BIRC3, indicative of migratory identity. Co-expression analysis is shown with color thresholds to highlight the specific gene interactions. (E) Circos plots showing the biological processes for the M-uDC cluster along with the genes associated with each pathway. (F) UMAP showing the expression of the marker gene LAMP3 across 10 phases of the menstrual cycle, from the menstrual stage through to decidualization (Decp and Decb). Below is a line graph combined with scatterplot overlay depicting the expression levels of LAMP3 (1st y axis) and the number of cells expressing the gene (2nd y axis) across different stages of the menstrual cycle ( x axis). Each point represents the expression level of LAMP3 in a particular cluster, while the line graph emphasizes the overall expression trend, with expression peaks during the mid-proliferative to late secretory phases with some presence in early decidual tissues. To understand the biological function of this cluster, we extracted all significant DEGs ( p < 0.05 and log fold change of at least 0.6) and conducted a pathway analysis using iPathway analysis. The main pathways expressed in cluster 4 are related to immune system process, immune response, cytokine production, regulation of cytokine production, and signal transduction ( Figure 6 E). The top genes differentially expressed in this cluster include chemokines ( CCL19 , CCL22 , and CXCL9 ), cytokines ( IL4l1 ), and chemokine receptors ( IL7R ). Transcriptional factors such as RELB are highly expressed in this cluster and are known to be essential for the regulation of T reg differentiation. 39 In summary, this gene signature further suggests a mature, antigen-processing uDC phenotype. Temporal analysis of cluster 4 revealed these cells are present in the mid-proliferative phase and decrease at the late-late secretory phase. Interestingly, these cells are detected in the decidua of early pregnancies ( Figure 6 F). Accordingly, we define cluster 4 as migratory uDCs (M-uDCs). Clusters 5 and 6 follow a different path of differentiation compared to clusters 3 and 4 ( Figure 3 E), suggesting shared biological functions. To identify specific markers for cluster 5, we used the bubble plot ( Figure S1 ) and identified IL22RA2 , CDH17 , CD207 , and SYT2 genes as potential markers. IL22RA2 , SYT2 , and CDH17 are found expressed only in cluster 5, while CD207 is shared with clusters 0 and 6, indicating CD207 as a marker for this line of differentiation ( Figure S2 ). The presence of IL22RA2 , CDH17 , CD207 , and SYT2 in cluster 5 suggests an antigen-presenting function, as these genes are crucial for antigen processing and presentation. IL22RA2 encodes a receptor for interleukin-22 ( IL-22 ), involved in inflammatory responses and tissue repair, influencing DC maturation and function. 40 CDH17 is crucial for DCs’ migration to lymph nodes and interaction with T cells. 41 , 42 UMAP analysis showed specific expression of these genes in cluster 5 ( Figure S2 B) and corroborated with quantified expression levels using violin plots ( Figure S2 C). Co-expression showed CDH17 + CD207 and CDH17+ IL22RA2 specificity to cluster 5 ( Figure S2 D). iPathway analysis (for significant DEGs [ p < 0.05] and log fold change of at least 0.6) revealed “response to stimulus” as significant ( Figure S2 E). Temporal analysis showed increased cell numbers in early proliferative, peaking in early-secretory phases, and declining through the late-secretory phases, and non-existent in the Decb stage ( Figure S2 F). These characteristics define cluster 5 as antigen-presenting uterine DCs (A-uDCs). Cluster 6 shares several markers with cluster 5, including HLA-DBQ2 , involved in recognizing and presenting common self-antigens to induce tolerance ( Figures S3 B and S3C), further supporting the premise that these clusters may play a role on tolerance to paternal antigens during conception. The temporal distribution of these cells through the menstrual cycle has similar pattern as cluster 5 ( Figure S3 F). NONO (non-POU domain-containing Octamer-binding protein) and EIF5A are expressed in all uDC clusters ( Figure S3 C). NONO is involved in regulating immune-related gene expression and synthesis of pro-inflammatory cytokines. 43 EIF5A plays a critical role in DCs’ maturation. Thus, NONO and EIF5A , along with LYZ and IRF8 , can be used as general markers for uDC. The cyclic hormonal nature of the human endometrium influences the nature and number of immune cells present within the female reproductive tract. 44 To determine the impact of hormone-regulated changes on the global presence of DCs and their specific clusters, we employed UMAP analysis for detailed visualization. This analysis revealed a broad distribution of cells in each menstrual stage; with varying numbers and types throughout the menstrual cycle ( Figure 7 A). During menstruation, very few cells were localized in clusters 0, 1, 2, and 4. In the proliferative phase, there is an early increase in cluster 0 (P-uDCs) and the recruitment of peripheral DCs (cluster 1, PB-DCs). The number of PB-DCs is high in the mid-proliferative phase and decreases afterward, becoming minimal in the rest of the cycle ( Figure 7 ). Clusters 3 (T-uDCs) and 2 (C-uDCs) are present mainly during the secretory phase, peaking in the late secretory phase. If pregnancy occurs (Decp and Decb), cluster 2 (C-uDCs) will remain present in the decidua; however, in the absence of pregnancy, their number will decrease and disappear during menstruation. Clusters 5 and 6 are mainly present during the early secretory phase. Cluster 4 is detected in the mid-proliferative phase and remains present until the late secretory phase ( Figure 7 A). Figure 7 Sequential transition from progenitor to differentiated states in uDCs maturation (A) UMAP visualization of cells from each stage of the menstrual cycle. Gray represents all cells, while blue highlights the cells coming from the specific stage indicated on the top of the UMAP. (B) Histogram showing the distribution of cells from different menstrual cycle stages across each cluster of the UMAP. The y axis represents the percentage of cells, and x axis indicates the menstrual cycle stage. The different colors in the histogram bar represent different clusters that are indicated on the right. (C) Area plot showing cumulative percentage of cells ( y axis) expressing key genes for P-uCDs, T-uDCs, and C-uDCs shown on x axis. AXL has a marked presence in each stage. In proliferative stages, uDCs express AXL along with CD1C and CX3CR1 genes, which are lost by the cells in later stages of differentiation. The cells gain proliferative markers such as PCNA and MKI67 in their differentiating stages and then lose these proliferative markers to gain the CLEC9A and XCR1 C-uDCs markers. Sequential transition from progenitor to differentiated states in uDCs maturation (A) UMAP visualization of cells from each stage of the menstrual cycle. Gray represents all cells, while blue highlights the cells coming from the specific stage indicated on the top of the UMAP. (B) Histogram showing the distribution of cells from different menstrual cycle stages across each cluster of the UMAP. The y axis represents the percentage of cells, and x axis indicates the menstrual cycle stage. The different colors in the histogram bar represent different clusters that are indicated on the right. (C) Area plot showing cumulative percentage of cells ( y axis) expressing key genes for P-uCDs, T-uDCs, and C-uDCs shown on x axis. AXL has a marked presence in each stage. In proliferative stages, uDCs express AXL along with CD1C and CX3CR1 genes, which are lost by the cells in later stages of differentiation. The cells gain proliferative markers such as PCNA and MKI67 in their differentiating stages and then lose these proliferative markers to gain the CLEC9A and XCR1 C-uDCs markers. The percentage of cells in each cluster varied in relation to the days of the menstrual cycle. Cluster 0 is present throughout the menstrual cycle, with its percentage increasing during the proliferative and early secretory phases. Cells differentiated from cluster 0 appear in the proliferative phase and increase during the secretory and early pregnancy phases. The percentage of cells in cluster 1 increase during the proliferative phase, decreasing afterward. The percentage of cells from cluster 4, 5, and 6 increase only during mid-secretory and are present during early pregnancy in the decidua basalis but not in parietalis ( Figure 7 B). This finding suggests that sub-cellular heterogeneity of DCs within the endometrium remains remarkably dynamic throughout the menstrual cycle, challenging previous assumptions about presence of a single type of DC in the human endometrium. Our next objective was to map the developmental trajectory of tissue-resident uDCs. Using the Slingshot algorithm, we identified a developmental trajectory of uDCs starting from cluster 0, progressing through cluster 3, and culminating in cluster 2 ( Figure 3 E). We use AXL+ as markers of cluster differentiation since it was expressed in all three clusters (0, 3, and 2). High percentages of AXL+/CX3CR1+/CD1A+ are found in cluster 0, and their percentage decreases as they progress through clusters 3 and 2 ( Figure 7 C). These progenitor cells transition to a proliferative state in cluster 3, indicated by high percentages of AXL+/MKI76+/PCNA+ , decreasing as they progress into cluster 2. As the percentage of cells expressing MKI67+/PCNA+ decrease, we observe an increase in the percentage of CLEC9A+/XCR1+ cells found in cluster 2 ( Figure 7 C). This trajectory highlights a well-defined pathway of tissue-resident uDCs maturation that is consistent across the menstrual cycle, reinforcing the notion of a complex yet orderly DC developmental process. To quantitatively assess tissue residents uDCs (clusters 0 and 2–6) and blood-recruited uDCs (cluster 1), we performed average gene expression across clusters based on their specific gene signature. Analysis of average gene expression across clusters confirmed these patterns, clearly distinguishing PB-uDCs (cluster 1, CCR7, SELL, ITGA4, GZMB, and JCHAIN) from tissue-resident subsets, clusters 0 and 2–6 (high expression of tissue-residency markers including CX3CR1, CLEC10A, CD1C, and AXL; Figure S8 A). Notably, PB-uDCs accounted for only 21.2% of the total DC population, while the remaining 78.8% comprised tissue-resident uDCs ( Figure S8 B). Inflammation plays a crucial role in uterine homeostasis, but there are different types of inflammatory signatures. We aimed to identify the specific inflammatory responses each cluster contributes to the endometrial milieu. Detailed investigation into the genes driving these functions 45 , 46 revealed that clusters 0, 1, 2, and 4 exhibit inflammatory response hallmark functions. We extracted genes associated with inflammatory response pathways from these clusters and generated a heatmap to visualize the gene expression patterns ( Figure S4 ). Our heatmap analysis revealed distinct inflammatory signatures for each cluster. We then analyzed the functions of most-highly upregulated genes in each cluster compared to other clusters (indicated by arrows; Table S1 ). The inflammatory genes specifically enriched in cluster 0 are primarily associated with immune regulation of the adaptive and innate immune responses. Cluster 2 was most prominent during the late-secretory phase, also known as the window of implantation, as observed from the UMAP. Analyzing the specific genes identified in cluster 2’s inflammatory signature, we found pro-inflammatory mediators distinctly expressed, including XCR1 , SOD1 , BCL6 , and HDAC9 . These genes are associated with the process of implantation, and their regulation, such as BCL6 , is indicative of pregnancy complications like miscarriage. Based on these findings, we propose that cluster 2 plays a pivotal role in mediating implantation and early pregnancy ( Figure S4 and Table S1 ). Our next objective was to analyze the communication/signaling between various subtypes of the uDCs by dissecting the ligand-receptor signaling networks between the various subtypes of uDCs (CellChat; Figure S5 ). 47 The signaling received (incoming) by the seven subtypes of uDCs were classified into mainly five patterns ( Figure 8 A) and include IL16 , GRN , ANNEXIN , CCL , CXCL , VISFATIN , and CD137 ( Figure 8 A). The signaling sent (outgoing) from the seven subtypes of uDCs were also classified into five major patterns and include MIF , GAS , COMPLEMENT , ANNEXIN , CCL , CXCL , BTLA , VISFATIN , CD137 , TNF , and ncWNT ( Figure 8 B). Figure 8 Receptor-ligand interaction between different subtypes of the uDCs (A) Visualization of intercellular communication within the uDC network, illustrated through river plots that delineate both incoming and outgoing signaling patterns. Figure highlights the incoming communication pathways to target cells, mapping how various signals are received by different uDC subtypes from their counterparts. (B) Complements this by showing the outgoing communication from these cells, detailing the types of signals transmitted outward and their respective targets across the cellular network. (C and D) graphically represent the expansion, migration, homing, and differentiation stages of uDCs, showcasing specific receptor-ligand pairs such as CD74/CXCR4 and CCL4/CCR5 that facilitate these processes. The diagrams provide a concise view of how distinct signaling mechanisms contribute to the spatial and functional organization of uDCs. (E) Details the interactions within the PB-uDC cluster, emphasizing the survival and migration pathways mediated by key molecules like GAS6 and AXL . Receptor-ligand interaction between different subtypes of the uDCs (A) Visualization of intercellular communication within the uDC network, illustrated through river plots that delineate both incoming and outgoing signaling patterns. Figure highlights the incoming communication pathways to target cells, mapping how various signals are received by different uDC subtypes from their counterparts. (B) Complements this by showing the outgoing communication from these cells, detailing the types of signals transmitted outward and their respective targets across the cellular network. (C and D) graphically represent the expansion, migration, homing, and differentiation stages of uDCs, showcasing specific receptor-ligand pairs such as CD74/CXCR4 and CCL4/CCR5 that facilitate these processes. The diagrams provide a concise view of how distinct signaling mechanisms contribute to the spatial and functional organization of uDCs. (E) Details the interactions within the PB-uDC cluster, emphasizing the survival and migration pathways mediated by key molecules like GAS6 and AXL . First, we analyzed the communication pattern of LGALS9-CD44 , which is essential for the homing of hematopoietic stem cells to their bone marrow niches, and it was found specifically on the P-uDCs; suggesting of autocrine signaling. The second pattern is represented by CCL4/CCR4 , which identify only one cluster expressing CCL4 (P-uDCs) and three clusters of uDCs expressing CCR5 , T-uDCs, PB-uDCs, and C-uDCs. This pathway is essential for the differentiation of DCs as well as in the recruitment of immune cells to sites of infection or inflammation and reveals a communication between P-uDCs and T-uDCs, PB-uDCs, and C-uDCs ( Figure 8 C). The third pattern is represented by MIF/CD74-CXCR4 and links T-uDCs (in cluster 3) expressing MIF , with C-uDCs, PB-uDCs, and P-uDCs expressing the MIF receptor CD74-CXCR4 . MIF/CD74-CXCR4 interaction is associated with cell migration and promotes antigen presentation ( Figure 8 D). The fourth pattern involves GAS6-AXL pathway linking the PB-uDCs, expressing GAS6 and all the tissue-resident uDC subtypes expressing AXL . 48 , 49 This pattern demonstrates a direct and distinct communication between the resident and recruited DCs ( Figure 8 E). To validate our RNA-seq data, we determined protein expression of markers definitory of DCs and individual subsets. First, we reanalyzed our CITE-seq performed on endometrial single-cell suspensions to link surface protein expression with key genes identified by our scRNA-seq analysis 50 ( Figure 9 Ai). Using antibody sequencing information, we identified the DC population by selecting CD45 + , lineage-negative cells that expressed CD11c and HLA-DR. These cells formed multiple clusters, all of which displayed surface expression of canonical markers for DCs (CD11c high , HLA-DR high , CD83) and tissue residency (CD69; Figure 9 A). Consistent with the transcriptomic data, CITE-seq also revealed multiple clusters of uDCs, including a clearly separated cluster 1 corresponding to PB-uDCs, recruited from peripheral blood, and the rest of the cluster corresponding to tissue residents uDCs ( Figure 9 Aii). Figure 9 Multimodal validation and characterization of uterine dendritic cells (uDCs) in human endometrium (A) CITE-seq analysis of DCs in the human endometrium. (i) UMAP showing all cell types identified in the endometrial CITE-seq dataset. (ii) DCs were isolated using antibody-derived tags (CD11c + HLA-DR + CD83 + ), forming distinct clusters. (iii and iv) Violin plots showing surface protein expression of canonical DC markers (CD11c, HLA-DP, HLA-DQ, and HLA-DR) across identified clusters. (B) UMAP feature plots depicting the expression of specific genes: IRF8, LYZ, GZMB, CD1C, CLEC9A, LAMP3, CLEC10A, and CX3CR1. (C) Flow cytometry plots in (C) show the expression of CX3CR1 on P-uDCs. (D) C-uDCs displaying expression of XCR1. (E) M-uDCs displaying the expression of CCR7, identifying their surface marker profile. (F) Illustrates the expression of CD11C and granzyme B in PB-uDCs, indicating the proportion of cells with cytotoxic potential. (G) Depicts the expression of CCR5 in combination with CD1C, CX3CR1, and CD103 on uDCs. (H) shows the expression of CXCR4 in combination with CD1C, CX3CR1, and CD103, indicating the presence and distribution of this chemokine receptor on uDCs. Multimodal validation and characterization of uterine dendritic cells (uDCs) in human endometrium (A) CITE-seq analysis of DCs in the human endometrium. (i) UMAP showing all cell types identified in the endometrial CITE-seq dataset. (ii) DCs were isolated using antibody-derived tags (CD11c + HLA-DR + CD83 + ), forming distinct clusters. (iii and iv) Violin plots showing surface protein expression of canonical DC markers (CD11c, HLA-DP, HLA-DQ, and HLA-DR) across identified clusters. (B) UMAP feature plots depicting the expression of specific genes: IRF8, LYZ, GZMB, CD1C, CLEC9A, LAMP3, CLEC10A, and CX3CR1. (C) Flow cytometry plots in (C) show the expression of CX3CR1 on P-uDCs. (D) C-uDCs displaying expression of XCR1. (E) M-uDCs displaying the expression of CCR7, identifying their surface marker profile. (F) Illustrates the expression of CD11C and granzyme B in PB-uDCs, indicating the proportion of cells with cytotoxic potential. (G) Depicts the expression of CCR5 in combination with CD1C, CX3CR1, and CD103 on uDCs. (H) shows the expression of CXCR4 in combination with CD1C, CX3CR1, and CD103, indicating the presence and distribution of this chemokine receptor on uDCs. The violin plots ( Figure 9 Aiii and iv) further quantified the expression of DC markers across CITE-seq clusters, confirming that all identified populations displayed hallmark DC markers. We then analyzed the expression of IRF8 and LYZ, the two markers used for the identification of the DCs in the scRNA-seq analysis. Expression pattern of IRF8 and LYZ in the CITE-seq UMAP was consistent with those observed in our scRNA-seq data ( Figure 9 B). As previously described in the scRNA-seq analysis, we identified PB-uDCs (IRF8 and GZB), transitional- and progenitor-uDCs (CX3CR1, CD1C, and CLEC10A), C-uDCs (CLEC9A), and a small number of M-uDCs (LAMP3). To further validate the presence of DCs and their subtypes in the human endometrium, we performed spectral flow cytometry on endometrial samples from two donors, using a stringent, multiparametric gating strategy to isolate canonical human DCs while excluding T cells, NK cells, and monocyte/macrophage populations. We began by gating all cells using a forward scatter area (FSC-A) versus side scatter area (SSC-A) plot, with 73.4% of events passing this initial gate ( Figure S6 A). Next, singlets were selected using FSC-height versus FSC-area to exclude doublets and aggregates (93.4% of the previously gated population; Figure S6 B). Live cells were then identified using a Live/Dead Blue viability dye versus FSC-A plot, with 59.4% of cells gated as viable ( Figure S6 C). From this population, CD45 + leukocytes were gated (26.7%; Figure S6 D), followed by exclusion of CD3 + T cells based on a CD19 versus CD3 plot (67% were CD3–; Figure S6 E). DCs were then isolated by selecting CD11c+ cells and excluding CD56 + NK cells (24.8%; Figure S6 F), and finally, CD14–HLA-DR+ myeloid cells were gated to exclude macrophages while retaining classical DCs (32.1%; Figure S6 G). This sequential gating approach resulted in the enrichment of ∼1% of the original cell population, representing a purified population of bona fide DCs. From this population, we assessed expression of key subtype markers identified from our transcriptomic analysis. First, we looked for the presence of CX3CR1 and CD103, markers of progenitor and C-uDCs. As shown in Figures 9 C and 9D, flow cytometry analysis revealed a distinct CX3CR1 + uDC corresponding to progenitor DC identified in the single-cell analysis and CD103 + /CX3CR1 − corresponding to C-uDCs. 51 Next, we looked for M-uDCs based on expression of CCR7; similar as described in the scRNA-seq, we observed a small population of M-uDCs identified by the expression of CD11C + /CCR7 + (9D). Finally, we looked for the presence of GZMB + cells that we defined as the uDCs recruited from peripheral blood (PB-uDCs). Flow cytometry analysis of GZMB + cells revealed specific CD11C + /GZMB + DCs that do not share any other tissue-resident uDC markers ( Figure 9 F). To validate the uDC interactions data described in Figure 8 E, we examined the protein expression of key receptors, CCR5 and CXCR4. CXCR4 was found to be expressed in cells corresponding to clusters 0, 2, 3, and 5, supporting the predicted interaction with CD74 and its ligand MIF. Additionally, CCR5 expression was detected on CX3CR1 + and CD103+ cells, which correspond to the progenitor (P-uDC), transitional (T-uDC), and C-uDC subtypes, respectively ( Figures 9 G and 9H). These findings confirm the presence of receptor-ligand pairs predicted computationally and reinforce the functional relevance of inter-subtype signaling pathways identified in our transcriptomic dataset. To further validate the existence of these subtypes of uDCs, we use multiplex immunohistochemistry (IHC) in endometrial samples collected at mid-secretory phase, a period when we observe the presence of majority of the clusters ( Figure 7 B). We designed a panel of antibodies to serve as specific marker for each subtype based on the identified genes from the scRNA-seq analysis. For peripheral-blood-recruited uDCs, we used IRF8 and GZMB , for P-uDCs CLEC10A , for C-uDCs XCR1 , and for T-uDCs TOP2A . LYZ and IRF8 were used as general markers for DC, CD14 for macrophages, and DAPI for nuclear staining. As shown in Figure 10 A, we identify CD14 − /IRF8 + /GZMB + /LYZ − cells corresponding to peripheral blood recruited DCs. Similarly, we detected cells characterized as CD14 − /CLEC10A + /LYZ + /TOP2A − corresponding to P-uDCs ( Figure 10 B), cDCs as CD14 − /LYZ + /IRF8 + /XCR1 + ( Figure 10 C), and T-uDCs as CD14 − /LYZ + /CLEC10A − /TOP2A + ( Figure 10 D). Figure 10 Validation of uterine dendritic cell (uDC) subtypes in the mid-secretory phase of the endometrium by multiplex IHC (A) PB-uDCs: IHC staining for various markers such as DAPI (blue) for nuclei, CD14 (cyan), GZMB (red), IRF8 (yellow), and LYZ (green) is shown. Cells stained for IRF8 and GZMB , lacking CD14 and LYZ , indicating PB-uDCs. (B) P-uDCs: cells stained for various markers such as DAPI (blue) for nuclei, CD14 (cyan), CLEC10A (red), TOP2A (yellow), and LYZ (green) are shown. Cells stained for CLEC10A and LYZ , lacking CD14 and TOP2A, indicating P-uDCs. (C) C-uDCs: IHC staining for marker genes such as DAPI (blue) for nuclei, CD14 (cyan), XCR1 (yellow), IRF8 (red), and LYZ (green) are shown. Cells stained for XCR1, IRF8 , and LYZ , lacking CD14 , indicating C-uDCs. (D) T-uDCs: IHC staining for marker genes such as DAPI (blue) for nuclei, CD14 (cyan), CLEC10A (red), TOP2A (yellow), and LYZ (green) are shown. Cells stained for TOP2A and LYZ , lacking CD14 and CLEC10A, indicating transitional uDCs. Scale bars, 50 μm. Validation of uterine dendritic cell (uDC) subtypes in the mid-secretory phase of the endometrium by multiplex IHC (A) PB-uDCs: IHC staining for various markers such as DAPI (blue) for nuclei, CD14 (cyan), GZMB (red), IRF8 (yellow), and LYZ (green) is shown. Cells stained for IRF8 and GZMB , lacking CD14 and LYZ , indicating PB-uDCs. (B) P-uDCs: cells stained for various markers such as DAPI (blue) for nuclei, CD14 (cyan), CLEC10A (red), TOP2A (yellow), and LYZ (green) are shown. Cells stained for CLEC10A and LYZ , lacking CD14 and TOP2A, indicating P-uDCs. (C) C-uDCs: IHC staining for marker genes such as DAPI (blue) for nuclei, CD14 (cyan), XCR1 (yellow), IRF8 (red), and LYZ (green) are shown. Cells stained for XCR1, IRF8 , and LYZ , lacking CD14 , indicating C-uDCs. (D) T-uDCs: IHC staining for marker genes such as DAPI (blue) for nuclei, CD14 (cyan), CLEC10A (red), TOP2A (yellow), and LYZ (green) are shown. Cells stained for TOP2A and LYZ , lacking CD14 and CLEC10A, indicating transitional uDCs. Scale bars, 50 μm. These findings confirm the presence of distinct subtypes of uDCs present in the mid-secretory phase of the endometrium, corroborating the insights from scRNA-seq analysis.

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Gil Mor ( [email protected] ). This study did not generate new unique reagents. • Single-cell RNA-seq data have been deposited at GEO: GSE288249 and are publicly available as of the date of publication. • Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. • This paper does not report original code. Single-cell RNA-seq data have been deposited at GEO: GSE288249 and are publicly available as of the date of publication. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. This paper does not report original code.

This study provides an in-depth characterization of bona fide DCs in the human endometrium and reveals the existence of distinct subsets with differential functions that are tightly regulated by the menstrual cycle. Employing scRNA-seq on uterine tissues collected across different menstrual phases and during early pregnancy, and validated with CITE-seq, flow cytometry, and IHC, we show that, in addition to DCs recruited from peripheral blood, the majority of uDCs differentiate from resident P-uDCs acquiring distinct characteristics. The presence of resident P-uDCs suggest a self-renewing source essential for the preparation of the endometrium for embryo implantation and successful pregnancy. The human endometrium undergoes complete tissue remodeling every menstrual cycle, involving multiple cell types, including epithelial, stroma, and immune cells. 5 , 52 , 53 , 54 The regenerative capacity of the endometrium is attributed to stem/progenitor cells, which resides in both the epithelium and stromal cell components of the basalis layer. 55 , 56 Traditionally, the origin of immune cells has been attributed to active recruitment from the peripheral blood. However, this concept has been challenged with recent data that suggest the existence of immune progenitor cells within the human endometrium. 57 The findings described in this study show that, although there is a recruited component of the uDC population, there is also a distinct function-specific tissue-resident uDC component. uDCs represent a small percentage of the endometrial cellular component but play multiple roles including tissue renewal, antigen presentation, and preparation of the endometrium for successful embryo implantation and placentation. 46 , 58 , 59 They contribute to fetal-maternal tolerance by regulating T cell activation, paternal antigen presentation, maintaining fetal-maternal tolerance, and creating a distinct inflammatory environment for embryo implantation. 60 However, it was unclear how these multiple and diverse functions are performed by the two classical classifications of DCs: cDCs and pDCs. Therefore, we tested the hypothesis that within the uDC population, subclusters with specific functions must exist. Using single-cell transcriptomic analysis of the LYZ+/IRF8+ DCs, we identified seven uDC subtypes, including cells with characteristics of resident progenitor cells and peripheral-blood-recruited DCs. We chose these two transcription factors because IRF8 is expressed by DCs that originate from hematopoietic progenitors in the bone marrow, and LYZ is a marker for circulating progenitors of cDCs. 17 , 22 , 61 Interestingly, we observed that all resident uDCs exhibit AXL expression. Notably, cluster 1, which we identified as being recruited from peripheral blood, lacks this characteristic. Therefore, we propose that AXL expression could serve as a potential marker for distinguishing endometrial resident DCs from those derived from peripheral blood. Cluster 1, identified as PB-uDC, exhibits a distinct gene expression profile partially resembling that of peripheral blood pDCs, with a 44% similarity to the known gene signatures described in the literature. 22 , 62 This includes high IRF8 expression and involvement in specific functional pathways such as allograft rejection, tumor necrosis factor alpha (TNF-α) signaling, cell motility, movement, and locomotion. These findings provide additional evidence that these cells are actively migrating and being recruited. However, the remaining 56% difference highlights the context-dependent nature of DC marker expression, influenced by the specific tissue microenvironment. Since no previous studies have described this subset, we propose renaming it as peripheral-blood-recruited uDCs (PB-uDCs), reflecting their origin from peripheral blood and the distinct modifications in their gene expression profiles acquired within the tissue environment. We observe that PB-uDCs are recruited from peripheral blood during the proliferative phase, which aligns with previous studies showing increased numbers of recruited DCs during the proliferative phase, indicative of a hormonal influence on the recruitment of peripheral uDCs. 50 , 63 , 64 , 65 Having identified clusters of tissue-resident uDCs, we then sought to distinguish between progenitors and differentiated uDCs. Using AXL as the marker for tissue residents uDCs, we pinpointed cluster 0, as the tissue P-uDCs. This cluster is characterized by the expression of markers such as KLF4, CX3CR1, CLEC10A , and CD1C and exhibit a trajectory of differentiation, transitioning through cluster 3 (transitional/proliferative uDCs, T-uDCs) and culminating in either cluster 2 (C-uDCs) or a second trajectory to clusters 4, 5, and 6. We identified cluster 3 as a transitional type of uDCs (T-uDCs) showing high expression of PCLAF , TOP2A , MKI67 , TYMS , and TK1 , genes associated with a proliferative state. These cells appear during the proliferative phase of the menstrual cycle, suggesting a potential role on the regeneration of the endometrium following menstruation. To our knowledge, no previous studies have described this specific transitional uDC subset or its potential role in endometrial regeneration, highlighting a distinct developmental aspect of DC biology in the context of the menstrual cycle. Cluster 2 revealed characteristics of C-uDCs marked by CLEC9A and XCR1 expression, which are consistent with the classical DC1 subtype described in the literature. 62 However, other well-known markers of DC1, such as CD141, which is typically highly expressed, were not found in this cluster within the endometrium. This suggests a tissue-specific adaptation of gene expression in these cells. Based on this distinct marker profile and their specialized role in the endometrium, we propose renaming this cluster to reflect its tissue-specific characteristics. This cluster aligns with cDC functions, particularly antigen presentation and the expression of a distinct set of inflammatory mediators, necessary for the preparation of the endometrium for embryo implantation. Indeed, in animal studies and in human samples we demonstrated a specific inflammatory signature necessary for the preparation of the epithelium of the lumen to become receptive to the embryo. 20 , 66 This inflammatory signature, responsible for the removal of the mucin-16 layer, expression of adhesion molecules, osteopontin ( SPP1 ) and CD44 , and increase in vascularity, was mainly determined by DCs. 20 , 66 In this study, we identified C-uDCs as the specific uDC cluster responsible for establishing this distinct inflammatory signature required for the success of embryo implantation. Furthermore, the number of C-uDCs increase during the late secretory phase, coinciding with the window of implantation. Clusters 4, 5 and 6 were identified as immune modulatory uDCs. Cluster 4 showed characteristics of M-uDCs, depicted by the expression of LAMP3, CCR7, BIRC3 , and FSCN1 , and underscores the dynamic nature of M-uDCs in immune surveillance and response to foreign antigens within the endometrium. Chemokine receptor such as CCR7 , enables the migration of uDCs to the lymph nodes. 10 These cells are prominent during the mid-proliferative to late secretory phases and are detected in early pregnancy decidual tissues, indicating their potential role in maternal-fetal tolerance. Cluster 5 most closely resembled the classical DC2 (cDC2) subtype described in the literature, which is associated with antigen presentation and immune regulation. 62 However, due to tissue-specific adaptations, some of the expected cDC2 markers might exhibit differential expression in the endometrium. Clusters 5 and 6, which share several markers with cluster 4, appear to be involved in antigen presentation and immune regulation. They are characterized by specific markers such as IL22RA2 , CDH17 , CD207 , and SYT2. Although a number of recent studies have utilized scRNA-seq to map the spatial and temporal dynamics of the endometrium, 67 none have focused specifically on immune cells, including DCs. 68 , 69 , 70 Among the limited studies mentioning DCs, functional subsets remain poorly characterized, often discussed only in the context of diseases like endometriosis. 71 This underscores a significant gap in the literature, which our study addresses by characterizing DC subtypes and their functional roles across different phases of the menstrual cycle in healthy human endometrium. Early studies, such as those by Rodriguez-Garcia’s group, 58 , 72 suggested the presence of distinct subsets of DCs based on the detection, by flow cytometry, of CD11c, CD11b, CD14, CD1c, and CD103 in the female reproductive tract and their role in the protection against viral infections, such as HIV. 50 , 59 None of those studies looked at the different phases of the menstrual cycle. An important finding of this study is that the presence of the different uDC clusters varies according to the changes of the human endometrium during the different phases of the menstrual cycle, which is in line with previous studies describing changes in the uDC population according to the phase of the menstrual cycle and the layer of the endometrium. 63 , 65 Previous studies were limited to the use of few markers, such as CD1a and CD83 for immunocytochemistry or flow cytometry, therefore, constraining the possibility to identify all the different sub-populations identified in this study. We and others demonstrated that endometrial biopsies lead to an increase in pro-inflammatory cytokine levels, which was associated with successful pregnancy outcomes. 20 , 73 This outcome was related to the activation of DCs. 66 Indeed, depletion of DCs in mice, prior to implantation, is linked to implantation failure. 17 , 74 Additional studies in different species have further strengthened the concept that a specific inflammatory signature is necessary for the induction of molecular and functional changes of endometrial cells, including epithelial and stromal during the process of embryo implantation. 75 , 76 By extracting and analyzing genes associated with inflammatory pathways, we identified that clusters 0, 1, 2, and 4 exhibit inflammatory responses as hallmark functions. As indicated above, cluster 2 stood out as particularly prominent during the late-secretory phase, coinciding with the window of implantation. The specific inflammatory mediators distinctly expressed in this cluster, including XCR1 , 77 SOD1 , 78 BCL6 , 79 and HDAC9 , 80 are critical for the implantation process. Furthermore, this provides insights into better understanding of the pathophysiology and the cellular origin associated with pregnancy complications such as implantation failure and miscarriage characterized by dysregulation of some of these gens, such as BCL6 . BCL6 is crucial for the survival and differentiation of stem/progenitor cells within the endometrium and plays a role in the differentiation, migration, and invasion of trophoblastic cells. 77 , 78 , 79 , 80 PB-uDCs (clusters 1) and M-uDCs (cluster 4) also contribute to the inflammatory landscape, each with distinct gene expression profiles. PB-uDCs' inflammatory genes are involved in pathways that likely support the recruitment and activation of peripheral immune cells, whereas M-uDCs' inflammatory signature suggests a role in antigen presentation and migration, essential for effective immune surveillance and response. These findings collectively underscore the importance of diverse inflammatory responses among uDC subtypes in creating a balanced and dynamic endometrial environment, crucial for successful implantation and pregnancy maintenance. While this study presents a detailed characterization of multiple DC subtypes in the human endometrium, there are still multiple questions to be addressed. Key inferences regarding developmental trajectories and functional specialization are based on transcriptomic and computational analyses. Ongoing studies in our lab aim to evaluate the function of these identified subtypes. Future research is necessary for lineage tracing and determination of the signaling pathways associated with the regulation of uDCs differentiation and function. Functional validation of progenitor capacity, lineage relationships, and the specific contributions of uDC subsets to implantation is critically needed. In conclusion, our findings fill a critical gap in understanding the origin, molecular characteristics, and roles of uDCs in the human endometrium. Our results reveal a dynamic and heterogeneous uDC landscape in the human endometrium, with distinct subpopulations playing specialized roles in immune regulation, antigen presentation, and preparation for implantation. Using the single-cell analysis in samples from the different phases of the menstrual cycle and validated by CITE-seq and immunocytochemistry, we were able to provide a comprehensive mapping of the different uDC clusters with association of their potential function and time within the endometrium. These findings open venues for further research into the therapeutic potential of targeting specific uDC subsets in reproductive health and disease. A limitation of our study is the variability introduced by collecting samples from multiple patients, as obtaining repeated samples from the same individuals would be neither ethical nor practical. It is noteworthy that all published studies on endometrial sampling across the menstrual cycle follow similar designs, and, to our knowledge, no study has achieved a sample size as large as ours. Endometrial biopsy is an invasive and painful procedure and performing it multiple times poses significant risks to patients, including severe pain and discomfort, increased health risks, potential long-term damage to the endometrial lining, and the risk of developing Asherman syndrome, a condition with lasting health implications. It is important to note that our donor cohort was stringently selected to include only healthy women with no known fertility issues or gynecological disorders. While this enhances the biological clarity of immune dynamics in a physiologically normal endometrium, it may limit the generalizability of our findings to individuals with subfertility or pathological conditions such as endometriosis or PCOS, where DC composition and behavior could differ significantly. Future studies, including broader patient populations, will be necessary to fully capture immunological diversity across reproductive health contexts.

Reproductive medicine has made significant progress, yet the underlying mechanisms that control embryo implantation and pregnancy remain poorly understood. One critical component of this process is the human endometrium, the dynamic inner lining of the uterus that undergoes dramatic changes throughout the menstrual cycle and, in the absence of pregnancy, completely remodels every month. The menstrual cycle 1 averages 28 days and is divided into two major phases by the event of ovulation (day 14): proliferative phase, before ovulation, 2 and the secretory phase. During the proliferative phase, the endometrium regenerates, while the secretory phase involves differentiation and preparation for pregnancy. During the secretory phase, the endometrium enters a narrow period of cellular and molecular receptivity, ideal for embryo implantation, known as the window of implantation (days 19–23 of the menstrual cycle). Without pregnancy, the endometrium sheds, initiating a new cycle of regeneration. 3 The source of the different cell types associated with each menstrual remodeling remains poorly understood. Immune cells, including B cells, T cells, macrophages, natural killer (NK) cells, and dendritic cells (DCs), are major components of the human endometrium. 4 , 5 Some immune cells form lymphoid aggregates, with a B cell core surrounded by CD8 + T cells and encircled by macrophages. 6 Recent studies show the expansion of CD8 + tissue-resident memory (TRM) cells, group 1 innate lymphoid cells (ILC1), and NK cells, during subsequent pregnancies, indicating their stable persistence in the basal layer between pregnancies. 7 , 8 , 9 The origin of other immune cells such as DCs is unknown. DCs are specialized antigen-presenting cells crucial for capturing, processing, and presenting antigens to T cells, thus initiating and regulating immune responses. 10 DCs include conventional DCs (cDCs), plasmacytoid DCs (pDCs), and monocyte-derived DCs (moDCs), 11 each with distinct functions and localizations. DCs are crucial for activating naive T cells, 12 , 13 initiating adaptive immune responses and maintaining tolerance by presenting self-antigens in a non-inflammatory context, preventing autoimmunity and maintaining tissue homeostasis. 14 In the human endometrium, DCs balance tolerance to a semi allogeneic fetus and defense against pathogens during pregnancy. 15 , 16 In addition, we and others have shown that DCs and macrophages create an inflammatory gradient affecting the mucin layer on the epithelium, which allows the apposition and adhesion of the blastocyst to the epithelium and promotes implantation. 17 , 18 We previously revealed that depleting uterine DCs (uDCs) severely impairs implantation and leads to embryo resorption. 17 This impairment is related not to immune tolerance but to successful decidualization. 19 , 20 These findings indicate that, besides their immune functions, uDCs also have a trophic role in regulating embryo implantation. Despite their recognized importance, the origin, molecular characteristics, and precise roles of uDCs in the endometrium remain unknown. The complex endometrial milieu, cellular heterogeneity, small percentage compared to other immune cells, and dynamic molecular signaling pose significant challenges in understanding the specific contributions of uDC subtypes. We hypothesized that the diverse functions of DCs in the endometrium are fulfilled by different subsets of DCs, with renewal provided by tissue resident progenitor cells. Specifically, we aimed to (1) map the transcriptional landscape of uDCs at high resolution across different stages of the menstrual cycle and early pregnancy to uncover time-specific changes in their abundance and function and (2) investigate whether certain uDC subtypes arise from local progenitor populations, thus revealing tissue-specific mechanisms of maintenance and differentiation. To test this hypothesis, we performed single-cell analysis of uDCs throughout the menstrual cycle and early pregnancy. Our data reveal the existence of multiple subtypes of DCs in the human endometrium. More importantly, we show the presence of resident progenitor uDCs (P-uDCs) within the endometrium, suggesting a self-renewing source of these pivotal cells. These findings challenge the notion that DCs are solely recruited from peripheral blood, offering fresh perspectives on the immunological underpinnings of endometrial receptivity and blastocyst implantation.

The authors declare that they have no competing interests.

REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Anti-LYZ Sino Biological Cat# 110207-M095; Anti-CLEC10A Abcam Cat#: EPR27400-166; RRID: AB_315086 Anti-Topoisomerase IIα Cell Signaling Technology (CST) CAt #12286S, RRID: AB_2797871 Anti-CD14 Sino Biological Cat#: 10073-R001 Anti-IRF8 Cell Signaling Technology (CST) Cat#: 83413S Anti-XCR1 Cell Signaling Technology (CST) Cat#: 44665S; RRID: AB_2799269 Anti-Granzyme B Sino Biological Cat#: 10345-R002 Biological samples Human endometrial biopsies Shenzhen Zhongshan Obstetrics & Gynecology Hospital Deposited data GEO (scRNA-seq comparison dataset GSE288249 Software and algorithms Seurat v5.1.0; RRID: SCR_016341 MetaCell R Version 2024.04.0 + 735 Slingshot – CellChat – HISAT2 v0.1.6 HALO Image Analysis Software v3.3, Indica Labs Other CITE-seq antibodies (custom panel) Parthasarathy et al., 50 2023 Flow cytometry panels Rodriguez-Garcia et al., 58 2017 This study was approved by the Institutional Review Board (IRB) of Shenzhen Zhongshan Obstetrics & Gynecology Hospital (formerly Shenzhen Zhongshan Urology Hospital), Shenzhen, China (IRB protocol number: SZZSECHU-2020010). Written informed consent was obtained from all participants prior to enrollment. Endometrial biopsies were collected from healthy, HIV-negative, reproductive-age women between 18 and 40 years of age. All participants had a BMI between 18 and 25 kg/m 2 and regular menstrual cycles (3–5 days every 25–35 days) and were not receiving hormonal stimulation. All donors tested negative for HIV, HBV, HCV, and syphilis. Exclusion criteria included: (1) Two or more prior spontaneous abortions, (2) Three or more biochemical pregnancies, (3) Known uterine pathologies (e.g., endometritis, endometriosis, polyps, adenomyosis, polycystic ovary syndrome, intrauterine adhesions), (4) Current or recent bacterial, fungal, or viral infection, and (5) Diagnosed autoimmune or thyroid disorders. Two or more prior spontaneous abortions, Three or more biochemical pregnancies, Known uterine pathologies (e.g., endometritis, endometriosis, polyps, adenomyosis, polycystic ovary syndrome, intrauterine adhesions), Current or recent bacterial, fungal, or viral infection, and Diagnosed autoimmune or thyroid disorders. No additional information regarding ancestry, race, or ethnicity was collected. Endometrial biopsies were collected using a Pipelle catheter inserted transcervically into the uterine cavity to sample tissue from the stratum functionale. Tissue fragments were soft and measured several millimeters in thickness. On average, each biopsy yielded approximately 110,000 cells, which were used for single-cell suspensions and CD45 + immune cell enrichment. Samples were allocated to experimental groups based on menstrual cycle stage, determined by self-reported cycle tracking. Staging and sample sizes were as follows: (1) Menstrual (days 1–3): n = 3 (2) Early proliferative (days 4–7): n = 8 (3) Mid proliferative (days 8–11): n = 8 (4) Late proliferative (days 12–14): n = 5 (5) Early secretory (days 15–18): n = 7 (6) Mid secretory (days 19–22): n = 8 (7) Late secretory (days 23–25): n = 13 (8) Late-late secretory (days 26–28): n = 3 (9) Early pregnancy (gestational weeks 5–8): n = 5 Menstrual (days 1–3): n = 3 Early proliferative (days 4–7): n = 8 Mid proliferative (days 8–11): n = 8 Late proliferative (days 12–14): n = 5 Early secretory (days 15–18): n = 7 Mid secretory (days 19–22): n = 8 Late secretory (days 23–25): n = 13 Late-late secretory (days 26–28): n = 3 Early pregnancy (gestational weeks 5–8): n = 5 All samples were from female participants, and no male subjects were included in the study. As the study exclusively focused on the female reproductive tract, sex was not a variable, but was intrinsic to the biological system under investigation. Therefore, sex-based comparisons were not applicable, and gender identity data were not collected. Endometrial tissues were digested in 2 mL RPMI-1640 medium containing 2 mg/mL collagenase IV (17104-019, Invitrogen) and 0.0125 mg/mL DNase I (D4527, Sigma) for 20 min at 37°C. The digested tissue was passed through a 70 μM cell strainer to obtain a single-cell suspension. Red blood cells were lysed using RBC lysis buffer for 10 min at room temperature, followed by washing with PBS. Single-cell suspensions were subjected to FACS using a BD FACSAria Fusion sorter. CD45 + cells (Biolegend, 304006) were sorted into 384-well plates (Eppendorf) for downstream single-cell RNA sequencing (scRNAseq). Sorted cells were processed for transcriptome library preparation following the MARS-seq2.0 protocol. 81 Specifically, HISAT (version 0.1.6) with default parameters was used to map the reads to the human reference genome hg38, and reads with multiple mapping positions were excluded. Reads were associated with genes if they were mapped to an exon, using Homo sapiens Ensembl90 for reference. Reads were condensed into original molecules by counting the same unique molecular identifiers (UMI). The batches used exhibited low cross single-cell contamination (under 3%), confirmed through spurious UMI detection in empty wells. While formal power analysis is challenging for scRNA-seq studies due to the distinct nature of single-cell variability and sampling, our study was designed to achieve robust detection and characterization of rare cell types, specifically dendritic cells (DCs), which constitute less than 1% of the endometrial cell population. Our comprehensive single-cell sequencing approach, combined with deep sequencing depth and rigorous quality control, allowed us to reliably identify and analyze these rare DC populations. To ensure statistical relevance, we employed the following methodology: (1) Detection of Rare Cell Populations: The statistical power of our study was evidenced by the detection of rare uDC subpopulations within the CD45 + immune cell compartment. The combination of single-cell RNA-seq, CITE-seq, and stringent bioinformatics analysis allowed us to capture rare cell states effectively, demonstrating sufficient resolution to identify DCs reliably. (2) Gene Filtering and Normalization: Genes expressed in fewer than three cells were filtered out to reduce noise. We performed log-normalization on the data and identified highly variable features to focus on biologically meaningful variation. (3) Clustering and Differential Expression Analysis: Clustering was performed at a resolution of 0.5, which was empirically determined to capture distinct cell populations. Differential expression analysis used a minimum log fold change of 0.6 and an adjusted p-value threshold of 0.05, based on the Benjamini-Hochberg correction, to control for false discovery rates. Detection of Rare Cell Populations: The statistical power of our study was evidenced by the detection of rare uDC subpopulations within the CD45 + immune cell compartment. The combination of single-cell RNA-seq, CITE-seq, and stringent bioinformatics analysis allowed us to capture rare cell states effectively, demonstrating sufficient resolution to identify DCs reliably. Gene Filtering and Normalization: Genes expressed in fewer than three cells were filtered out to reduce noise. We performed log-normalization on the data and identified highly variable features to focus on biologically meaningful variation. Clustering and Differential Expression Analysis: Clustering was performed at a resolution of 0.5, which was empirically determined to capture distinct cell populations. Differential expression analysis used a minimum log fold change of 0.6 and an adjusted p-value threshold of 0.05, based on the Benjamini-Hochberg correction, to control for false discovery rates. Through these methods, we ensured that our analysis was statistically robust and capable of capturing rare and functionally relevant cell types. The count matrix is obtained from the scRNAseq data, containing gene expression levels for all CD45 + cells. The count matrix was pre-processed, which includes normalization and quality control, to remove low-quality cells and genes, ensuring robust downstream analysis. Next, the MetaCell analysis is initialized using the count matrix using R (Version 2024.04.0 + 735 (2024.04.0 + 735)). MetaCell clusters cells based on their gene expression profiles by constructing a k-nearest neighbor (k-NN) graph of cells using highly variable genes. MetaCell then partitions the k-NN graph into smaller, highly connected subgraphs called metacells. Following this, the metacells are examined for the expression of DC markers. Specifically, the focus is on cells expressing LYZ (Lysozyme) and IRF8 (Interferon Regulatory Factor 8), which are characteristic markers for DCs. Finally, cells from the identified metacells that show high expression of LYZ and IRF8 are extracted. These extracted cells are further analyzed to study specific dendritic cell subpopulations within the CD45 + cell population. To analyze DCs extracted from CD45 + cells, the count matrix and metadata containing the stages of the menstrual cycle were loaded into R. The data dimensions were checked to ensure correct loading. Essential R packages, including Seurat (5.1.0, RRID: SCR_016341 ; version and package identifier), tidyverse, ggpubr, Matrix, dplyr, and patchwork, were installed and loaded. A Seurat object was created using the counts data, metadata, and appropriate filtering criteria (min.cells = 3, min.features = 200). Quality control steps involved calculating the median UMI counts per cell (1719) and the median number of genes per cell (1065). The percentage of mitochondrial genes was calculated, although no mitochondrial genes were found in this dataset. Histograms of UMI counts and gene counts per cell were plotted to visualize data distribution. A violin plot of the QC columns was generated to assess the distribution of these metrics. Scatterplots comparing UMI counts to gene counts were created to detect potential outliers (correlation co-efficiency = 0.98). To ensure data quality, we filtered genes with fewer than three counts across all cells and set the clustering resolution at 0.5 to capture distinct cell populations. Differential expression analysis employed a minimum log fold change of 0.6 and an adjusted p -value threshold of 0.05. The count data were log-normalized and variable features were identified. The data were scaled and subjected to PCA. The PCA results were visualized, and the loadings were examined. The optimal number of principal components for further analysis was determined using an elbow plot. A nearest neighbor graph was constructed with the selected number of dimensions (12 dimensions were selected). Clustering was performed at a resolution of 0.5. UMAP embedding was carried out to visualize the clustering results in a two-dimensional space, which was then plotted. The steps include data preparation, quality control, normalization, feature selection, PCA, clustering, and visualization, ensuring a robust and reproducible analysis pipeline. Slingshot algorithm was used for lineage tracing (trajectory analysis), CellChat package was used for receptor-ligand analysis using default parameters, Bioconductor and Seurat packages were used for all other analyses. Tissue processing for validations with CITE-Seq and flow cytometry were performed as published previously 82 with the following procedure. Endometrial tissues were minced and enzymatically digested using the Tumor Dissociation Kit (Miltenyi Biotec) in combination with 0.01% DNase I (Worthington Biochemical), following mechanical dissociation on a gentleMACS Octo Dissociator. Single-cell suspensions were filtered through 100, 70, and 30 μm MACS SmartStrainers (Miltenyi Biotec) and washed with PBS +2% FBS. Red blood cells were removed using CD235a MicroBeads (Miltenyi Biotec). Viable cells were enriched by density gradient centrifugation and dead cells were excluded using the Dead Cell Removal Kit (Miltenyi Biotec). For immune cell enrichment, cells were gated on CD45 expression. Dendritic cells (DCs) were characterized using flow cytometry as CD45 + , CD3 − , CD19 − , CD56 − , HLA-DRhigh, CD11c+ cells, in accordance with the approach outlined by Rodriguez-Garcia et al. (2017) 58 as follows. For flow cytometry, cells were stained with a panel of fluorophore-conjugated antibodies including: CD45, CD3, CD19, CD56, HLA-DR, CD11c, CD14, and additional DC subset markers (see key resources table ). Dead cells were excluded using LIVE/DEAD Fixable Yellow or Zombie dye (BioLegend). Fluorescence-minus-one (FMO) controls were used to set gates. Samples were acquired on a MACSQuant 10 or Cytek Aurora flow cytometer and analyzed using FlowJo v10 or OMIQ. Sample preparation for single-cell antibody and RNA sequencing was carried out as outlined by Parthasarathy et al. (2023) 50 with the following procedure. For CITE-seq, ∼10,000 viable CD45 + cells per sample were stained with TotalSeq-A oligo-conjugated antibodies (BioLegend) targeting key surface proteins, incubated for 30 min at 4°C, and washed to remove unbound antibodies. Samples were processed using the BD Rhapsody Single-Cell Analysis System according to manufacturer instructions. RNA-capture beads with unique molecular identifiers (UMIs) were used to barcode transcripts, and libraries were generated for both mRNA and antibody-derived tags (ADTs). Sequencing was performed on a NovaSeq 6000 using 100 bp paired-end reads, aiming for 60,000 reads per cell for transcriptome libraries, ∼600 reads per cell for sample tags, and ∼850 reads per cell for ADTs. Raw FASTQ files were trimmed to 75 bp, aligned to the GRCh38 reference genome, and analyzed using Partek Flow v10.0. Cells with >30% mitochondrial reads, 4,000 genes were excluded. RNA and protein matrices were normalized separately. Weighted nearest-neighbor (WNN) analysis was performed to integrate protein and RNA signals. Downstream analysis of scRNA-seq data was conducted as detailed in section developmental trajectory from progenitor to differentiated states in uDCs . The endometrial biopsy was rinsed with PBS, fixed in 4% PFA at room temperature for 4–6 h, and subsequently dehydrated overnight. The paraffin-embedded tissue was sectioned into slides with a thickness of 4 μM and subjected to multiplex immunohistochemistry using the PANO 5-plex IHC kit (Panovue, 10144100100) and Bond Polymer Refine Detection (Leica, DS9800-CN) on the automatic dyeing machine (BOND RX, Leica). The primary antibodies were used: Anti-LYZ (Sino Biology, 110207-M095), Anti-CLEC10A (abcam, ab315086), Anti-Topoisomerase IIα (CST, 12286S), Anti-CD14 (Sino Biology, 10073-R001), Anti-IRF-8 (CST, 83413S), Anti-XCR1 (CST, 44665S), Anti-Granzyme B (Sino Biology, 10345-R002), followed by incubation with horseradish peroxidase-conjugated secondary antibodies and tyramine signal amplification. After labeling all target antigens, the nuclei were counterstained with DAPI (Panovue, Beijing, China). The multispectral images were acquired using an automated slide-scanner (SLIDEVIEW VS200, Olympus) and analyzed with HALO software (Version 3.3, Indica labs). Citing data sources: Publicly available dataset at https://doi.org/10.1126/science.aah4573 was used for comparison.

This work was supported in part by 10.13039/100000002 NIH grant NIAID 5R01AI145829-05 (G.M.) and 5P42ES030991 (GM Project 4) R01HD111146 (G.M.). We acknowledge the assistance of the Wayne State University Proteomics Core that is supported through NIH grants P30 ES020957 , P30 CA022453 , and S10 OD030484 . A.S. was in part supported by training grant T32GM142519 , NIH. General Program of the 10.13039/501100001809 National Natural Science Foundation of China ( 82371684 ), and the Shenzhen Medical Research Fund ( B2302006 ) (L.D.).