Abstract
Age-dependent reproductive decline has become a significant global health concern as the average maternal age at first birth increases. Fertility loss associated with reproductive aging is driven in part by alterations to ovarian composition and function, dysregulation of folliculogenesis, and increased inflammatory signaling. Our understanding of the molecular changes underlying ovarian aging has been expanded by single-cell and spatial transcriptomic studies, which identified infiltration of immune cells as a feature of ovarian aging. However, the function of these age-associated immune cells and their potential contributions to the inflammaging phenotype remain unclear. In this study, we integrate single-cell and spatial transcriptomics to define changes in the composition and intercellular signaling in the aging mouse ovary. We identify specific macrophage and T cell subpopulations that increase with age and are key sources of pro-inflammatory signaling in old ovaries. Further, we predict bidirectional signaling between these pro-inflammatory cells and granulosa cell populations that may impair follicular growth and development while promoting immune cell recruitment. These findings provide insights into the mechanisms that drive ovarian inflammaging.
Similar content being viewed by others
Introduction
Female reproductive aging is a mounting concern in modern society. Global trends in delayed childbearing have underscored the increased risks of conception and pregnancy at advanced reproductive age, including miscarriage, birth defects, fetal chromosome abnormalities, preeclampsia, and pre-term birth1,2,3. Furthermore, perimenopause and menopause associated hormonal dysregulation have far-reaching impacts on female health, compromising cardiovascular, skeletal, cognitive, and immune function4,5. Despite the influence the female reproductive system has on general health and the health of offspring, our understanding of female reproductive biology has historically lagged behind other organ systems6. Moreover, the cellular heterogeneity and cyclic and dynamic nature of the ovary has made this particular female reproductive organ challenging to characterize at the cellular and molecular level.
The primary role of the ovary is to house and, upon reproductive maturity, facilitate the cyclic maturation of female germ cells, called oocytes7. Each oocyte is encapsulated by granulosa cells (GCs) in a highly organized structure called an ovarian follicle. The follicle is considered the functional unit of the ovary and serves as a supportive niche for the oocyte during its growth period7. The ovary’s ability to mediate follicular growth, atresia, ovulation, and luteinization is in part dependent on cooperative signaling between the follicle and immune cells of the surrounding stroma8. GCs support oocyte growth and development by supplying essential nutrients and metabolites. GCs can also secrete pro-inflammatory cytokines to regulate immune cell recruitment and activity9. During folliculogenesis macrophages localize to the theca layer of follicles and promote vascularization and GC proliferation by secreting growth factors, such as macrophage migration inhibitory factor (MIF)10. In the event of follicular atresia, macrophages invade past the theca layer and basal membrane to phagocytose apoptotic GCs. Similarly, macrophages play a prominent role in corpus luteum (CL) formation and regression, as they work in conjunction with leukocytes to promote luteal cell steroidogenesis and eventual CL clearing11.
With age, the constant cycles of tissue remodeling contribute to a dysregulated ovarian environment with extensive changes in tissue morphology, composition, and ultimately, function12,13. Significant efforts have been made to characterize the molecular underpinnings of these phenomena, largely through transcriptomic techniques. A central observation from many of these studies is significant changes in the aging immune milieu. These changes include expansion of lymphocyte and macrophage populations, changes in macrophage polarization, and emergence of age-specific sub-types of immune cells with pro-inflammatory expression profiles14,15,16,17,18,19. It has been hypothesized that these immune cells play a central role in the inflammatory, pro-fibrotic phenotype of the aging ovary, often referred to as inflammaging. However, the mechanisms by which they exert this potential impact remain unclear due in part to a limited understanding of their interactions with other cell types in the ovary. Although the specific sources and targets of proinflammatory signaling remain incompletely understood, inflammatory signaling pathways, such as transforming growth factor-β (TGFβ) and nuclear factor kappa B (NFκB), have been associated with age-dependent changes that may negatively impact oocyte viability and fertility via direct and indirect mechanisms20,21,22. For example, recent studies have shown that aging ovaries exhibit a loss of hyaluronan, which can activate pro-fibrotic inflammatory pathways and increase ovarian stiffness, ultimately hindering follicular expansion and ovulation16,21,23,24,. Additionally, changes in immune populations may also contribute to impairment of CL regression and clearance that is dependent on macrophage function24. Understanding how signaling between the follicular unit and immune populations change with age could provide novel insight to how this inflammation-induced dysregulation arises.
In this study, we used single-cell and spatial transcriptomics to define the immune-related cellular composition and signaling changes that occur with age in the murine ovary. In agreement with studies in other organisms and at earlier aging time-points, we identify dramatic alterations in immune cell populations that occur by fourteen months of age12,14,24,25. Notably, we identify multiple pro-inflammatory macrophage and lymphocyte populations with specific expression profiles that distinguish late-stage aging from early-stage aging in the ovary. Through in-depth analysis of cell-cell communication pathways, we provide evidence that immune populations within the aging ovary increase their inflammatory signal output via pathways that have been previously implicated in inflammaging, such as tumor necrosis factor (TNF), C-X-C motif chemokine ligand (CXCL), and TGFβ. Furthermore, we define how these age-altered immune populations interact with stromal fibroblasts and GCs and vice-versa, providing novel insight into how the age-related changes in ovarian composition impact cellular behavior, cyclic coordination, and, ultimately, ovarian function.
Results
Mapping ovarian aging with single-cell RNA-seq in the murine ovary
The mammalian ovary is a complex, dynamic organ that is subject to constant wound induction and healing due to the cyclic nature of follicular maturation and ovulation, as well as clearance of atretic follicles and corpus lutea (Fig. 1A). With age, the ovary demonstrates hallmark alterations in tissue composition, cyclicity, and appearance, including reduced follicle number, persistence of late-stage CLs, and enriched immune cell infiltration26. To characterize how changes at the intra- and intercellular level contribute to these macro-level phenotypes, we generated single-cell transcriptomes from dissociated whole ovaries isolated from reproductively young (3-month) and reproductively old (14-month) mice. After pre-processing to correct for ambient RNA and remove predicted doublets (see Materials and Methods), we obtained an integrated dataset containing 29,676 total cells (15,390 cells from old ovaries, 13,837 cells from young ovaries) (Fig. 1B, Supplementary Fig. 1A). Cells from young ovaries expressed an average of 4420 (±2172) genes and 27,972 (±34,239) transcripts per cell, and those from old ovaries expressed an average of 3364 (±1875) genes and 16,832 ( ± 19,072) transcripts per cell, indicating comparable datasets.
A Schematic of the ovary, highlighting the structural components and cell types identified and characterized in this study and their locations within the tissue. B UMAP plot of integrated young and old ovary single-cell datasets. C Violin plots of canonical cell-type marker genes used to classify clusters to cell types. D Changes in the proportion of cells in clusters from panel B between the young and old datasets. Error bars indicate the 95% confidence interval for the magnitude of the difference between ages as determined by bootstrapping. Dashed lines indicate absolute log2-fold changes greater than 0.58, and false discovery rate less than 0.05. Gray points indicate no significant difference between ages.
Dimension reduction using manifold approximation and projection (UMAP) and clustering of the integrated datasets identified twenty-two clusters (Fig. 1B). Based on the expression patterns of canonical marker genes, we classified each cluster into one of nine cell groups: granulosa (Fst)27,28, theca (Aldh1a1)28, stromal fibroblasts (Col1a1)12,27,28, epithelial (Krt18)27,28, endothelial (Cldn5)29, corpus luteum (CL) (Sfrp4)28, myeloid (Lyz2)28, non-B lymphocytes (NBLs, Cd3g)12,28, and B-cells (Cd79a)12,28 (Fig. 1B-C, Supplementary Fig. 1B). Roughly half of the total cells from each dataset were GCs (Old = 39%; Young = 54%), while the remaining groups ranged from 1% to 14% of total cells (Supplementary Fig. 1D, Supplementary Table 1). Notably, large cell types, such as oocytes and multinucleated giant cells (MNGCs), are absent from this dataset due to the 30 µm recovery size maximum.
We examined changes in ovarian cellular composition with age (Fig. 1D). One cluster, Epithelial.3 (C21), was uniquely identified in old ovaries. The remaining clusters were present in both age groups and had strong transcriptional signatures (Supplementary Table 1). Consistent with previous reports, we observed significant increases in immune cell populations in the old ovaries, including three out of four myeloid clusters and all lymphoid clusters (B cells and NBLs) (Fig. 1D)14,17,19. We also observed higher numbers of epithelial cells. However, as the increases in epithelial cells were dominated by elevated cell numbers in cluster 9 (Epithelial.1) from a single old sample, these changes were not further explored. Conversely, old ovaries exhibited reductions in a subset of granulosa and theca cells, likely reflecting the reduction in follicle number and ovulation with age30,31. The ~15% reduction in granulosa cells observed in old ovaries is consistent with other studies in mouse, primate, and humans12,25,32. Similar results were obtained with permutation testing using the nine cell type groupings (Supplementary Fig. 1D).
Generating a spatial atlas of the aging murine ovary
To provide additional support for cell type annotations and explore changes in ovarian composition and structure with age, we performed spatial transcriptomics analysis of young and old ovaries using the Slide-seq (v2) platform33 at near single-cell (10 μm) spatial resolution. Two representative datasets that best retained underlying tissue structure were selected for visualizations throughout. After filtering to remove beads that were not underlying the tissue, we obtained datasets containing 17,195 (Young) and 7532 (Old) beads, with a mean of 219 (±214, Young) and 215 (±220, Old) genes detected per bead (Fig. 2A–F). The reduced number of beads retained in the old ovary dataset was reproducible across both sections and may be due to impaired mRNA capture caused by altered tissue composition (e.g. fibrosis).
Representative H&E-stained tissue sections adjacent to those used for Slide-seq V2 analysis are shown for young (A) and old (D) ovaries. Slide-seq data are shown in x/y space for young (B) and old (E) datasets, with each point representing a sequenced bead (10 um diameter). Beads are colored according to their cluster identity shown in UMAP space (C, F) as determined using the Slide-seq expression data, illustrating the spatial distribution of cell types within tissue. G, H Visualization of the spatial datasets with points colored according to the tissue regions determined using Vesalius and cell identity assigned manually using structural features in the adjacent tissue slice. I Representative spatially enriched genes from pre-antral, early antral, late antral follicles, and the interstitium are highlighted in spatial feature plots. J DAB and hematoxylin staining for CD3ε in young (top) and old (bottom) ovary sections are shown. Insets highlight CD3ε+ staining in CLs (young) and interstitium near MNGCs (old). K Box plot depicts percent CD3ε+ area in CD3ε and IgG probed young and old ovary histology sections. Significance (p = 0.0091) as determined by unpaired t-test (young, n = 5, old, n = 4).
For each age, beads were clustered and visualized in UMAP space and in x/y tissue space, in which distinct underlying tissue structures could be identified (Fig. 2B, C, E, F). In young ovaries, we identified ten clusters, eight of which could be classified based on marker gene expression profiles and spatial distribution patterns (Supplementary Fig. 2A, B). This included a cluster (C7) that was spatially enriched in small preantral follicles and had high expression of classical markers of oocytes (Nlrp14, C86187/Sirena1, Ldhb, Zp2, Zp3)27,34,35. In the old ovary we identified nine clusters, seven of which could be classified based on expression profiles (Supplementary Fig. 2C, D). This included a small cluster (C7) expressing similar preantral/small antral GC and oocyte markers (including Ldhb and Kctd1436). The reduced number of early follicle GC/oocyte beads in the old Slide-seq data agrees with the observed decrease in the number of oocytes and early-stage follicles in aged ovaries15,37,38,39. Consistent with a previous study, we also noted that the theca cell cluster was not just localized around follicles, but throughout interstitial regions of the ovary as well40. This could be due to the interstitial origin of theca cells, as they have been observed to differentiate from interstitial fibroblasts41.
To better define cell type composition and perform spatial analysis of gene expression, we used the Vesalius package42, a computational tool that can organize spatial transcriptomic data into anatomically based regions. With Vesalius, we segmented our spatial datasets into tissue regions and assigned beads to corresponding histological structures (Fig. 2G, H). Significantly, these spatial analyses allowed us to globally assess whether putative cell type markers identified in our single-cell dataset were spatially enriched within specific tissue regions. An example of this approach is shown for the CL in Supplementary Fig. 2E, F, where we measured the spatial correlation for each gene detected in our Slide-seq dataset with beads that were located within the CL (beads in the CL.1 or CL.2 tissue regions, Fig. 2G). We observed a weak, positive correlation between a gene’s spatial enrichment in the CL and scores returned by statistical tests used to identify cluster marker genes from single-cell datasets (in this case, ROC or Wilcoxon tests implemented in Seurat). However, for both testing methods, we observed that some genes with high test values (such as Fdx1, Cyp11a1) had low spatial enrichment within the CL. Marker genes with improved spatial enrichment in the tissue regions of interest could be identified using spatial information provided by Vesalius - either by performing marker gene testing using the beads in the region or by identifying genes with high spatial correlation with the region (Supplementary Fig. 2E, bottom). High confidence marker genes were similarly identified throughout the tissue (Fig. 2I). The top ten spatially enriched genes for each tissue region in young and old ovaries are provided in Supplementary Data 1.
Although we observed clear spatial distribution patterns for macrophages in both ages, we could not determine the spatial distribution of the lymphoid cells in either the Slide-seq dataset. To provide spatial context to the observed expansion of NBLs in the old ovary, we performed immunohistochemistry for the T-cell marker CD3ε43. In the young ovary, CD3ε + cells were sparsely localized throughout the ovary with some enrichment in corpus lutea (Fig. 2J top, inset). In old ovaries, we observed a significant increase in CD3ε + cells (Fig. 2K). Additionally, we observed that CD3ε + cells were often concentrated in regions of the interstitial compartment near MNGCs (Fig. 2Jbottom, inset), in agreement with observed interactions between MNGCs and T-cells in aged ovaries44.
Taken together, our spatial transcriptomics analyses dataset captured age-based changes in tissue structure and cell-type distribution within the ovary that would not have been possible to identify with single-cell sequencing alone. Furthermore, the integration of spatial information with our single-cell datasets allowed for high-throughput identification and validation of high-confidence cell-type marker genes that would have otherwise required low-throughput methods, such as RNA-FISH.
Aging-associated expansion of immune cells
In agreement with earlier reports, we observed that the lymphocytes exhibited the most significant change in number with age, with over a two-fold increase in the proportion of total ovarian cells (Figs. 1D, S1E)12,14. Subcluster analysis on the NBL clusters identified nine subclusters with distinct marker genes (Fig. 3A, B). Subclusters NBL.1 and NBL.5 represented CD7+ and CD7- NK cells, respectively, both of which were slightly enriched in young ovaries (Fig. 3C). In contrast, subclusters NBL.6, 7, and 8 were significantly enriched in old ovaries. Subclusters NBL.7 and NBL.8 represented CD8/CD4 double-negative T cells (DNT), which have previously been reported in the ovary using flow-cytometry14. While the prior cytometry dataset from 10–11-month-old mice identified the CD8/CD4 DNT cells to be TCR γδ-negative, we observed evidence for the expression of both gamma and delta constant domain genes (Trgc1 and Trdc, respectively) and TCR γ variable genes Trgv2 and Trgv4, in a subset of the DNT cells (NBL.8). These cells had negligible expression of TCR alpha chain (Trac), Cd27 and interferon γ (Infg), supporting their classification as γδ-DNT17 cells, a recently identified T cell population characterized by their phenotypic flexibility, memory-associated activity, and migratory tendencies45,46 (Fig. 3B). A second subset of DNTs (NBL.7) had an expression profile more closely matching alpha/beta DNTs (Fig. 3B)47. Prior studies similarly detected DNT cells in the ovary14 and found the gδ-DNT17 population to increase by nine months of age12. These observations strongly suggest that with age, there is expansion of multiple DNT cell types in the ovary.
A UMAP plot of non-B cell lymphocyte (NBL) subcluster analysis with points colored by subcluster identity and labeled by subcluster cell type. B Dot plot of markers used to classify lymphoid subcluster cell types is shown. C Changes in the proportion of the NBL subclusters with age are shown, as determined using permutation testing. Error bars indicate the 95 percent confidence interval for the magnitude of the difference between ages as determined by bootstrapping. Dashed lines indicate absolute log2-fold changes greater than 0.58, and false discovery rate less than 0.05. Gray points indicate no significant difference between ages. D Bar plot highlights the relative signaling for all signaling pathways identified as significantly altered in NBL cells (>20% change in strength, p-value <= 0.01) between young and old datasets. E Heatmaps show the relative contribution of each subtype to the total pathway signaling, shown for incoming (top) and outgoing (bottom) signaling in both age groups. Colors represent relative contribution for each cell type (values scaled row-wise with row max = 1). Dendrograms represent the results of hierarchical clustering of both rows and columns. F Bar plots show the top ten signaling pathways between granulosa cells and stromal fibroblasts to NBLs in young (top) and old (bottom) datasets. Pathways with significant (>20%, p-value < 0.01) changes in strength between young and old are highlighted in blue. G Bar plots indicate the total strength of TNF signaling in young and old datasets (top) and the fraction of outgoing TNF signaling from old cell types (bottom). H Bar plots show the fraction of total B cell signaling sent to each cell group type in young and old datasets (left) and sources of total signaling being received by B cells in young and old datasets (right). I Bar plot demonstrates signaling pathways from GCs to B cells altered with age. J Chord plots demonstrate ADGRE signaling from NBL subtypes to B cells (depicted by lines connecting cell types) and strength of participation (depicted through thickness of lines) in young (left) and old (right) datasets. K Chord plots demonstrate MHC-I signaling from B cells to NBL subtypes (depicted by lines connecting cell types) and strength of signaling (depicted through thickness of lines) in young (left) and old (right) datasets.
The third subcluster identified to increase with age was NBL.6. These cells were Gata3high, Il7r/CD127 + , Thy1/Cd90+, Klrg1+, Ncr1/NKp46-, supporting their classification as innate lymphocyte ILC2 cells48,49, a cytokine-secreting cell type that is responsible for facilitating tissue repair by stimulating collagen deposition in fibroblasts, and fibrosis when dysregulated (Fig. 3B)50. The last subcluster that increased with age was NBL.0. This population is characterized as Cd8+/Cd4- T cells with high expression of granzyme K (Gzmk) and has not been previously reported in the ovary (Fig. 3B). Its expression profile is similar to the T cell (Cd8+/Cd4- TteK) population recently identified as a significant source of proinflammatory cytokines in rheumatoid arthritis51. Taken together, our data suggest that with age, the ovary experiences a reduction in NK cells and an expansion of multiple lymphocyte populations known for innate immune response and production of proinflammatory cytokines, including ILC2, T teK, and γδ-DNT17 T cells.
With age, we observed increased NBL signaling through pro-inflammatory signaling pathways, such as Annexin and TNF, with distinct patterns of signaling being sent and received by each lymphocyte type (Fig. 3D–E). Generally, NBLs had strong outgoing signaling to GCs and immune cells, with decreased signaling to GCs and increased signaling to immune cells with age (Fig. S3A). The major sources of signaling received by NBLs in both ages were GCs and stromal fibroblasts (Fig. S3B). This included signaling through pathways related to cell recruitment, attachment, and survival, including Collagen, JAM (Junction Adhesion Molecules), ADGRE (Adhesion G-Protein Coupled Receptors), and thrombospondin (THBS) (Fig. 3F). In old ovaries these signals were maintained, while signaling through the immune-activating MHC-I pathway increased (Fig. 3F). We noted that the age-enriched γδ-DNT17 T cells were a major source of proinflammatory signaling in the old ovary. For example, total ovarian tumor TNF signaling increased nearly 20-fold with age, with ~35% of total TNF signaling being sent from γδ-DNT17 cells (Fig. 3G). The majority of TNF signaling occurred through the TNF receptor 1 (Tnfrsf1a) (Fig. S3C), which is ubiquitously expressed and associated with pro-inflammatory and apoptotic responses. However, we observed evidence of TNF signaling through the hematopoietic-specific TNF receptor 2 (Tnfsf1b) for both myeloid and NBL cells, which has been associated with increasing cytokine production, enhancing cellular survival, and promoting proliferation52.
In addition to expanded T cell populations, we observed that B cells significantly increased with age (Fig. 1D, S1D). This was accompanied by increased expression of multiple factors implicated in B cell recruitment, maintenance, and development, such as Klf253, and Irf2bp254, and the immunoglobulin genes IgM and IgKC (Fig. S3D). Gene set enrichment analysis identified signatures consistent with increased B cell activation in old ovaries, including upregulation of cell adhesion molecules and factors associated with leukocyte activation and migration and antibody processing and presentation (Fig. S3E).
Changes in B cell proportion and expression profiles predicted changes in B cell signaling (Fig. S3F). In both ages, GCs were the major signaling partner for B cells (Fig. 3H). The top GC to B cell signaling pathways included MIF, amyloid precursor protein (APP), and THBS, while B cell signaling to GCs was exclusively through Cyclophilin A (CypA) (Fig. 3I). We also observed increased signaling between B cells and NBLs with age. Two notable signaling pathways that increased in strength were the ADGRE pathway and the Major Histocompatibility Complex Class I (MHC-I) signaling pathway (Fig. 3J, K). Both of these pathways play key roles in the pro-inflammatory activation of T-cells and NK-cells, which is consistent with increased inflammatory signaling in the ovary with age55,56,57. Taken together, our data identified significant signaling interactions between lymphocytes and follicular GCs, with old ovaries having increased signaling related to immune cell recruitment and activation, as well as proinflammatory signaling into the ovarian stroma.
Aging-associated expansion of proinflammatory macrophage populations and localization
Macrophages are the most abundant leukocytes in the ovary and play a pivotal role in ovarian physiology and pathology. To characterize how macrophage populations are affected with age, we performed subcluster analysis on the original myeloid clusters in our dataset (Fig. 1B, clusters 6, 14, 19 and 20), identifying thirteen subclusters (Fig. 4A, B, Supplementary Fig. 4A). We compared subcluster expression profiles with mouse single cell atlases28,58,59 and markers for ovarian macrophages (Fcgr1, Csf1r)60. Our subclusters contained ten putative macrophage clusters (subclusters M.0-5, M.8, M.10-12), with variable expression of the M1-like marker Mrc1 (Fig. 4C). These putative macrophages included both previously reported populations of ovarian tissue resident macrophages (Adgrehi and Adgreint TRM)60, as well as a third intermediate TRM (M.4). The remaining three subclusters included both conventional (M.6) and activated (M.7) dendritic cells, and eosinophils (M.9). Four subclusters were enriched in old ovaries (M.2, M.3, M.9 and M.12), while three subclusters were enriched in young ovaries (M.1, M.4 and M.10) (Fig. 4C).
A UMAP plot of subcluster analysis for the original myeloid clusters (6,14, 19, and 20) from Fig. 1B. B Dot plot shows expression of myeloid markers used for cell-type classification. C Changes in the proportion of the myeloid subclusters with age are shown, as determined using permutation testing. Error bars indicate the 95 percent confidence interval for the magnitude of the difference between ages as determined by bootstrapping. Dashed lines indicate absolute log2-fold changes greater than 0.58, and false discovery rate less than 0.05. Gray points indicate no significant difference between ages. D Bar plot highlights the relative signaling for all signaling pathways identified as significantly altered in myeloid cells (>20% change in strength, p-value <= 0.01) between young and old datasets. E Heatmaps show the relative contribution of each myeloid subtype to the total pathway signaling, shown for incoming and outgoing signaling in both age groups. Colors represent relative contribution for each cell type (values scaled row-wise with row max = 1). Dendrograms represent the results of hierarchical clustering of both rows and columns. F Bar plot shows the top signaling pathways from stromal fibroblasts to myeloid cells in young and old datasets. Chord plots show collagen signaling from stromal fibroblasts to myeloid cells (G), TGFβ signaling from myeloid cells to stromal fibroblasts (H), MHC-I signaling from myeloid cells to non-B cell lymphoid (NBL) cells (I), CXCL signaling from myeloid cells to NBLs cells (J), and CCL signaling from NBLs to myeloid cells in young (top) and old (bottom) datasets. Chord plot signaling is depicted by lines connecting subtypes, and the strength of signaling is depicted through the thickness of lines in young (top) and old (bottom) datasets. Source cell types are bold black, and target cell types are bold green.
The largest change in proportion was observed for subcluster M.3, rising from 4% of total myeloid cells in young ovaries to 19% in old (Supplementary Fig. 4B). This cluster was enriched for multiple markers associated with an immunosuppressive, tumor associated macrophage (TAM) phenotype (Arg1, Fn1, Gatm, Tgm2)49,61 and markers of fibrogenic scar-associated macrophages (SAMs) (Spp1, Cd9, Trem2 and Cd63)62. Due to its enrichment in old ovaries, we labeled this cluster aging-SAMs (aSAM). Subcluster M.5 also expressed SAM markers, but this subcluster had no age-based enrichment and exhibited comparatively lower expression of TAM marker genes and monocyte-derived proinflammatory mediators (Srgn and Tgfbi)63,64, thus this cluster was labeled constitutively-present SAMs (cSAMs). A second subcluster, Cd300e+ macrophages (M.12), was exclusively present in old ovaries and made up 2% of the total myeloid population (Supplementary Fig. 4B). This subcluster expressed markers that are characteristic of a pro-inflammatory myeloid phenotype, including genes such as Fcrl5, a marker recently identified in pro-inflammatory macrophages that emerge in non-alcoholic steatohepatitis (NASH)65, and Cd300e, an immune activating surface receptor that promotes pro-inflammatory cytokine production66,67,68. Interestingly, a small population of this cell type was exclusively present at 9 months in a prior ovarian aging study12, providing additional evidence that Cd300e+ macrophages are an aging-specific immune cell population. The eosinophil cluster (M.9) is characterized by high expression of S100a8 and S100a9, genes encoding calprotectin, part of the innate immune response69. In other tissues, calprotectin helps recruit other immune cells69; in the aging ovary, the increase in calprotectin-positive eosinophils may expand the immune cell milieu.
In old ovaries, we observed increased signaling through pathways associated with myeloid cell recruitment, macrophage polarization, and pro-inflammatory cytokine production (Fig. 4D). Notably, aSAMs adopt an overall pattern of signaling similar to the Adgrehi TRMs in old ovaries, while the Cd300e+ cells show a more distinct signaling profile (Fig. 4E). GCs and stromal fibroblasts were the strongest interactors with myeloid cells in both age groups (Supplementary Fig. 4C, D). Specific interactions between myeloid cells and stromal fibroblasts that increased with age included pathways related to macrophage recruitment, adhesion, and polarization, such as Collagen, MIF, and MK signaling pathways (Fig. 4F, G). Additionally, both the aSAM and TRM populations showed increased TGFβ (via macrophage-derived Tgfb1) signaling to the stromal fibroblasts (Fig. 4H), which is a hallmark inducer of tissue fibrosis70.
Aging also induced changes in signaling between the myeloid and NBL cell types, with different cell types showing distinct alterations in signaling pathways and partners. For example, MHC-I signaling increased from aSAMs and TRM populations, specifically into the T-teK lymphocytes (Fig. 4I). In contrast, pro-inflammatory Cxcl16-Cxcr6 (CXCL) signaling increased to multiple NBL cell types, including both αβ-DNT and γδ-DNT17, and T-teK cells, preferentially originating from aSAMs and specifically the Adgrehi TRMs (Fig. 4J). Signaling from NBLs to myeloid cells also changed with age, including increases in macrophage-activating Ccr5 signaling via the Ccl5 ligand from T-teK and NK cells (Fig. 4K).
Taken together, we show that Cd300e+ macrophages12 and aSAMs are signature macrophage populations in the aging ovary whose major signaling partners include GCs and stromal fibroblasts. Furthermore, our data suggest that aSAMs and TRMs in old ovaries receive significant signaling through recruitment pathways from stromal fibroblasts and send pro-fibrotic signals to the stroma through pathways including TGFβ. Additionally, we observe extensive crosstalk between aging-induced myeloid and lymphocyte populations through pathways that influence cell recruitment, inflammatory cytokine production, and macrophage polarization. All together this yields a picture of the aging ovary as an organ that has undergone a shift from an environment supporting cycles of follicle maturation, ovulation, and luteinization to an environment characterized by immune cells and inflammation.
Aging impacts granulosa cell function
With age, the number and type of follicles contained within the ovaries are reduced. However, our understanding of the impact of aging on GCs—the predominant cell type within the follicle—and their ability to support the oocyte, interact with other follicles, and coordinate with the surrounding stromal environment remains incomplete. Our integrated dataset contained six putative GC-cell subclusters (Fig. 5A) with distinct expression patterns for reported GC subtype marker genes, including markers of atretic GCs (Ghr, Cald, Pik3lp1)27, late antral GCs (Inhba, Nppc, Mro)27, mitotic GCs (Top2a, Ube2c)12,27, preantral GCs (Gatm, Kctd14)12,27, and small antral GCs (Amh)27 (Supplementary Fig. 5A). Multiple lines of evidence were used to assign GC types to our subclusters. First, examination of gene expression profiles along a pseudotime trajectory analysis of the GC subclusters agreed with known changes in GCs during follicle development (Supplementary Fig. 5B, C). GC type assignment was further confirmed by assessing the distinct spatial location of each GC type in the young ovary and compositional analysis of young ovary tissue regions after cell type transfer from the single cell reference dataset (Supplementary Fig. 5D, E). Based on these results, we were able to designate our GC subtypes as preantral, mitotic, small antral, cumulus, late antral/mural, and atretic types (Fig. 5B). Of these, late antral/mural GCs were enriched in old ovaries, while small antral GCs were enriched in young (Fig. 5B).
A UMAP plot of granulosa cell (GC) subcluster analysis from combined single cell datasets is shown. B Changes in the proportion of the GC subclusters with age are shown, as determined using permutation testing. Error bars indicate the 95 percent confidence interval for the magnitude of the difference between ages as determined by bootstrapping. Dashed lines indicate absolute log2-fold changes greater than 0.58, and false discovery rate less than 0.05. Gray points indicate no significant difference between ages. C Spatial distribution of cell subtypes within a single late antral-stage follicle (highlighted in navy, top left panel). Bar plot (top right) shows the fraction of each subtype within the selected follicle determined by cell type transfer from the reference dataset. The location of each subtype (bottom left), as well as the distribution of distances for each cell to the center of the follicle (bottom right) is highlighted for each subtype. D Bar plot highlights the relative signaling for all pathways identified as significantly altered in GCs (>20% change in strength, p-value <= 0.01) between young and old datasets. E Heatmaps show the relative contribution of each GC subtype for each identified differential signaling pathway, shown for incoming and outgoing signaling in both age groups. Colors represent relative contribution for each cell type (values scaled row-wise with row max = 1). Dendrograms represent the results of hierarchical clustering of both rows and columns. F Bar plots show the identity of cell types receiving signaling from GCs, split by age. G Bar plots show the identity of cell types sending signals to GCs, split by age. H Bar plots depict the strength of total signaling from GC subtypes to myeloid cells, split by age. I Bar plot highlighst the strength of total signaling sent from late antral/mural GCs to myeloid subtypes, split by age. J Chord plots show SEMA3 (top) and Desmosterol (bottom) signaling from GC subtypes to myeloid subtypes. Chord plot signaling is depicted by lines connecting subtypes, and the strength of signaling is depicted through the thickness of lines in the young (left) and old (right) datasets. Major source cell types are bold/black, and target cell types are bold/green.
These assignments were largely in agreement with the recent characterization of GCs in mice (Supplementary Fig. 5F)12. However, our spatial information allowed for improved resolution of distinct subtypes within the follicle that had not previously been reported. For example, in antral follicles, mitotic and cumulus GCs were preferentially located near the center of the follicle, while antral/mural and atretic GCs localized more towards the follicle periphery (Fig. 5C). Although mitotic and cumulus cells shared a similar distribution in the follicle, the cumulus GCs had a unique signature of elevated expression of multiple glycolytic genes, in agreement with the reported role for cumulus cell glycolysis to support oocytes71. Similar patterns were observed in follicles undergoing atresia, with atretic GCs also being peripherally localized (Fig. 5C, Supplementary Fig. 5G). Gene set enrichment analysis of genes differentially expressed by GCs with age suggested that aging impacts each GC subtype differently (Supplementary Fig. 5H).
To better understand how the spatial distribution and age-related changes in GC subtype expression may impact their function, we examined GC cell-cell interactions to identify signaling pathways altered with age (Fig. 5D, E). Many of the outgoing pathways that increased with age were sent by atretic and late antral/mural GCs located at the periphery of the follicle and targeted immune cell populations likely situated within the theca layer and surrounding stroma (MHC-I, CXCL, MIF, ADGRE). Meanwhile, pathways that were reduced with age were affiliated with early follicle development, lipid metabolism, and androgen signaling (Anti-Müllerian hormone (AMH), Insulin-like growth factor (IGF), Bone morphogenic protein (BMP), Estradiol, Testosterone).
Consistent with previous studies, the majority of GC signaling was between GC populations in both age groups, suggesting they largely engage in autocrine and/or homotypic paracrine signaling (Fig. 5F, G)72,73. However, as discussed above, we also observed many signaling relationships with immune populations. For example, signaling from GCs to myeloid cells increased with age, with myeloid cells receiving ~20% of total GC signaling output (Fig. 5F). This was driven mostly by increased signaling from the late antral/mural GCs. In old ovaries, late antral/mural GCs sent signals primarily to TRMs and aSAMs (Fig. 5I). Interestingly, pathways with the largest relative increase in signaling from GCs to myeloid cells included multiple pathways implicated in the formation of lipid-laden macrophages, such as Semaphorins3 (SEMA3) and Desmosterol (Supplementary Fig. 5I, Fig. 5J).
Collectively, we observed that age significantly impacts GCs, leading to a reduction in expression of signaling pathways important for early-stage follicle maturation, such as BMP and AMH. Furthermore, we detect an increase in signaling from late antral and atretic GCs through pathways that recruit immune cells and may promote the formation of lipid-laden MNGCs. These novel findings provide compelling evidence that GCs are the main ovarian cell type that signals to and recruits immune cells.
Discussion
In this study, we leverage single-cell and spatial transcriptomics to investigate how age-related changes in ovarian cell composition and gene expression impact cell signaling and intercellular cooperativity, providing insight to how molecular-level changes impact ovarian physiology. This integrated analysis extends our understanding of how ovarian composition changes with age and provides significant insight into specific cell types and pathways that lead to the emergence of an inflammatory environment, often termed inflammaging. We find that immune cell populations infiltrate the ovary and predict that their signaling promotes and sustains inflammation. Our results are consistent with previous single-cell ovarian aging studies showing an expansion of immune cell populations in the ovary with age, while also unveiling additional populations at the 14-month timepoint that include macrophages (aSAMs) and lymphocytes (T-teKs, and DNT subgroups). Importantly, our cell-cell communication analyses defined putative signaling partners for both newly and previously identified immune populations. We identify GCs as one of the main ovarian cell types that engage with immune cells and the stroma, and report how these signaling relationships are affected with age and potentially contribute to the physiological dysregulation that is characteristic of ovarian aging.
Of the identified immune clusters, lymphoid populations were the most profoundly impacted with age. An age-dependent increase was observed in B-cells and four T-cell subtypes. The four T-cell subtypes, Cd8 + /Cd4- TteK (sub-cluster NBL.0), γδ-DNT17 (sub-cluster NBL.8), αβ-DNT17 (sub-cluster NBL.7), and ILC2 (sub-cluster 7), are affiliated with chronic inflammatory disease states and may be a source of the proinflammatory cytokine signatures in the aging ovary51,74. A previous analysis of ovarian aging in C57B6/J mice identified a similar Gata3 + ILC2 population in the ovary12. However, this population was not significantly increased at 9 months, suggesting that the expansion of the Gata3high ILC2 lymphocytes occurs at later stages of aging. These changes in lymphoid populations with age were accompanied by significant alterations in signaling output as well. Notably, two of the pathways most significantly enriched with age were Annexin and TNF – pathways that, when overexpressed, can promote T-cell activation and increase the production of other T-cell activating transcription factors75,76. For example, out-going TNF signaling strength from T-cells increased 20-fold by 14 months of age. While basal levels of TNF cytokine signaling within the ovary are critical for guiding folliculogenesis, steroidogenesis, and ovulation, elevated levels can disrupt ovarian homeostasis. Previous studies revealed that increased TNF secretion contributes to ovarian aging phenotypes, including ovulatory failure, increased atresia, and abnormal steroidogenesis. However, the primary sources and targets of this signaling have remained undefined77,78,79. Our study is the first to define a potential source of this age-related TNF signaling. Our analysis reveals that not only does the ovarian lymphocyte population increase with age, but it is accompanied by a significant increase in TNF and annexin signaling. Additionally, we defined an age-dependent increase in B-cell signaling via pathways responsible for immune cell recruitment, such as MIF80, APP80, and THBS81 consistent with the observed increase in a variety of immune cells.
Though macrophages as a whole did not significantly increase with age, we did observe distinct changes in multiple myeloid subtypes. This included increases in three of the ten putative macrophage sub-clusters - Cd300e+ macrophages (subcluster M.12), aSAMs (subcluster M.2) and Adgreint TRMs (M.3). These observations were complemented by an increase in pro-inflammatory signaling via pathways, such as MCH-I, CXCL, and CCL, that promote myeloid cell recruitment, macrophage polarization, and pro-inflammatory cytokine production. Stromal fibroblasts, NBLs, and GCs served as the predominant signaling partners of macrophages in both age groups. Though macrophage signaling partners remained consistent, underlying subtypes and the type and strength of pathway signaling were altered with age. Notably, there was an increase in collagen and TGFβ signaling activity between macrophages (aSAMs and TRMs) and stromal fibroblasts. Both of these pathways are major stimulants of collagen and fibronectin deposition as well as inhibitors of extracellular matrix-degrading enzymes such as matrix metalloproteinase82. Such activity promotes a stiff, profibrotic environment that can hinder ovulation83. Additionally, we observed a significant increase in inflammatory signaling activity between myeloid and lymphoid populations. The primary players in such pathways consistently included aSAMs, γδ-DNT17 cells, T-teK cells, and Adgrehi TRMs, which are all subtypes that increased in number with age. Previous studies have reported conflicting findings regarding age-related changes in macrophage population size, with some noting increases31,83, while others report decreases14,19 or no significant change12,84,85. However, despite these differences, reports consistently observe macrophages shifting towards a more activated state with age19,83,84. In the context of these other studies, our findings suggest that while there are no significant changes in the overall macrophage abundance with age, there is an expansion of subtypes that are associated with pro-inflammatory signaling. Taken together, our work in combination with previous studies suggests the existence of a positive feedback loop of inflammatory signaling and immune cell recruitment that continues to increase with age, contributing to the chronic inflammatory phenotype of the aging ovary.
Our investigation identified GCs as a major source of signaling to other GCs as well as to myeloid, B cell and T cell lineages. We observed striking changes in GC signaling partners and pathways with age. GC interactions with populations vital to follicle maturation and embryo maintenance-- such as theca, epi/endothelium, and CL cells -- diminish with age. Loss of communication with these cell types occurs concordantly with a decline in GC pathways related to follicle development, lipid metabolism, and androgen signaling pathways (AMH, IGF, BMP, Estradiol, Testosterone). We demonstrated previously that the oocyte exhibits an age-dependent increase in expression of key secreted factors, including Bmp15 and Gdf9, which regulate GC function and bidirectional communication86. This upregulation of oocyte-secreted factors may compensate for poorly functioning GCs. These findings suggest that the cell-type interactions and signaling pathways that facilitate follicle maturation and support oocytes become dysregulated with age, likely compromising oocyte quality. Moreover, the immune cell activation and recruitment pathway signaling strengthens between late antral/mural and atretic GCs and immune cells. The increase in aSAM signaling with late antral/mural and atretic GCs via SEMA3 and Desmosterol pathways was particularly notable as such pathways promote the formation of lipid-laden macrophages, such as MNGCs, that significantly increase in the ovary with age87. Furthermore, such observations of dysregulated immune cell interactions with follicles could provide context for the increase of trapped follicles and long-lived CLs that fail to be cleared by immune cells in old ovaries24. Taken together, these findings support an aging paradigm in which GCs transition from an oocyte-focused to an immune-focused signaling program, serving as an active player in ovarian inflammaging and dysfunction.
In conclusion, this investigation uses single-cell and spatial transcriptomic technologies to capture the molecular biology of the aging ovary. The integration of spatial information with single-cell transcriptomics improves the identification of marker genes for each cell type. We corroborated previous findings in the field, especially regarding changes in immune cell populations, as well as extended our knowledge of how aging impacts cell organization and gene expression. We identify changes in intercellular signaling and signaling strength within and between different cell populations, providing a novel window into how cellular communication networks critical to the coordination of ovarian function are affected with age. Furthermore, we shed light on how altered cell populations and changes in cell signaling lead to a sustained state of ovarian inflammation.
Methods
Mouse colony
At the Stowers Institute for Medical Research (SIMR; Kansas City, MO, USA), all mouse experimental protocols were evaluated and approved by the Institutional Animal Care and Use Committee of the Stowers Institute for Medical Research. Mice (Mus musculus, C57BL6/J) were housed and monitored in a barrier facility with constant light cycle, humidity, and temperature, and offered food and water ad libitum. Core colony mouse strains, including C57BL6/J, of the Lab Animal Service Facility (LASF) at SIMR are routinely refreshed with mice from JAX laboratory every eighth generation to avoid genetic drift. All JAX mice are properly quarantined before being integrated into colonies. The aged mice were sourced from core colonies of the LASF at SIMR or the JAX laboratory (strain #000664) and aged in-house. The relevant SIMR IACUC protocol number is #2025-196, and SIMR’s Animal Welfare Assurance number is D16-00658 (Legacy A4304-01). Tissue samples from this colony were used for single-cell sequencing and spatial transcriptomic experiments.
At Northwestern University (Chicago, IL, USA), mice (Mus musculus, C57BL6/J) were maintained in accordance with the National Institutes of Health’s guidelines and housed in Northwestern University’s Center for Comparative Medicine barrier facility under constant light (12 h light/12 h dark), humidity, and temperature control, and offered food and water ad libitum. All animal experiments were approved by Northwestern University’s Institutional Animal Care and Use Committee. The relevant Northwestern IACUC protocol number is IS00018332, and Northwestern’s Animal Welfare Assurance number is A3283-01. Tissue samples from this colony were used for DAB and hematoxylin immunohistochemistry experiments.
All experiments complied with all relevant ethical regulations for animal use.
Ovarian tissue dissociation
Our protocol for tissue dissociation was modified from the mouse and rat adult brain dissociation kit (130-107-677; Miltenyi Biotec). Each biological replicate consisted of a single ovary from a single animal. For reproducibility, a total of two biological replicates were collected from 2-month-old mice for the young cohort, and two biological replicates were collected from >14-month-old mice for the old cohort. Ovaries were briefly washed in cold D-PBS. Ovaries were physically dissected on a 35 mm polystyrene petri dish with 3 mL Enzyme mix 1, subsequently mixed with 45 mL Enzyme mix 2 in a gentleMACs C tube (#130-093-237; Miltenyi Biotec) and then placed on the gentleMACS Octo Dissociator (#130-095-937; Miltenyi Biotec) running Program 37C_ABDK_01 for 30 minutes. The dissociated tissues and medium were applied to a Falcon® 40 µm cell strainer (#352340; Corning Incorporated) on a 50 mL centrifuge tube and washed with 10 mL of D-PBS. The total solution was applied to another 40 µm cell strainer on a 50 mL centrifuge tube to remove any undissociated debris, tissues, or cell masses larger than 40 µm. The cell suspension was transferred to a 15 mL centrifuge tube and centrifuged at 300 x g for 10 minutes at 4°C with full acceleration and full brake. The cell pellet was resuspended in 90 µL Debris Removal Solution and 310 µL D-PBS. The cell suspension was then gently overlayed with 400 µL D-PBS and centrifuged at 4°C and 3000 × g for 10 minutes, forming three-phases of separation. The top two phases were completely aspirated and discarded. D-PBS was added to the cell suspension to a final volume of 1.5 mL and subsequently centrifuged at 4°C and 1000 × g for 10 minutes. The supernatant was aspirated, and the cell pellet was resuspended in 90 µL cold D-PBS/0.04% BSA.
Next, to remove the red blood cells, 10 µL mouse Anti-Ter-119 MicroBeads (#130-049-901; Miltenyi Biotec) were mixed with the cell suspension and incubated for 15 minutes in the dark at 4°C. The cell suspension was washed with 1 mL cold D-PBS/BSA and centrifuged at 300 × g for 5 minutes. The cell pellet was resuspended with 500 µL cold D-PBS/BSA and placed on a magnetic stand. The supernatant was collected three times via washes with 500 µL cold D-PBS/BSA twice on the magnetic stand. The resulting 1.5 mL of cell suspension was centrifuged at 300 × g for 5 minutes and resuspended in 100 µL cold 1× Red Blood Cell Removal Solution for 10 minutes at 4°C. The cell suspension was diluted by 1 mL cold D-PBS/BSA, centrifuged at 4°C and 300 × g for 10 minutes, and the cells were resuspended by 100 µL D-PBS/BSA. The cell viability was examined using the Orange/Propidium Iodide assay on a Nexcelom Cellometer Auto T4. Samples with >500 cells/µL and >80% viability were used for cDNA library construction.
10X Genomic single-cell RNA sequencing
The cell suspension was loaded on a Chromium Single Cell Controller (10X Genomics; CA), based on live cell concentration. cDNA libraries were prepared using the Chromium Next GEM Single Cell 3’ GEM Chip G single Cell Kit (#1000120; 10X Genomics), the Chromium Next GEM Single Cell 3’ GEM, Library & Gel Bead Kit v3.1 (#1000121; 10X Genomics), the Chromium i7 Multiplex Kit (#120262; 10X Genomics) and the SPRIselect Reagent (#B23318, Beckman Coutler, Indianapolis, IN) with 12 cycles of PCR amplification to yield 150–500 ng of input cDNA according to the manufacturer’s instructions. Quality control was completed via the Qubit Fluorometer and the Agilent 2100 HS Bioanalyzer. Libraries were pooled in groups at equal molar concentration and sequenced on a NovaSeq S1 flowcell on the Illumina NovaSeq 6000 instrument RTA v3.4.4, recipe v1.7.0 or v1.7.5 using v1.5 chemistry with the following paired read lengths: 28 bp Read 1, 8 bp I7 Index, and 91 bp Read 2. With 12,000 to 16,000 cells per sample, libraries were sequenced to a depth necessary to achieve 53,000 to 82,000 mean reads per cell - around 900 million to 1 billion reads each -1456-2508 median genes per cell. The Cell Ranger package was used in the analysis. Raw reads were demultiplexed into Fastq format using cellranger mkfastq. Alignment, filtering, barcode counting, and UMI counting were done using Cell Ranger’s count pipeline.
Single-cell RNA-seq analysis
Output from CellRanger was further processed using SoupX88 (v. 1.6.2) to detect contamination from cell-free RNAs and generate background-corrected expression profiles. Corrected data sets were used to create Seurat objects using Seurat89 v. 4.3.0. Individual data sets were filtered to remove cells with fewer than 500 detected features, 1000 counts, and more than 20% mitochondrial counts. These cutoffs were selected based on visual inspection of QC plots for each metric and data set. Filtered data sets were used as input for DoubletFinder90 (v. 2.0.3) to identify and remove cells predicted to be doublets, transformed using SCtransform91 v2, and integrated to generate a final data set containing 31,632 cells. Clustering and dimensionality reduction were performed using the standard Seurat workflow (Seurat commands RunPCA, RunUMAP, FindNeighbors(dims = 1:30), and FindClusters). Clustering resolution parameters were selected after visual inspection of clustering tree plots generated with clustree (v. 0.5.0) for resolutions between 0.1 and 1.0. Two clusters that were only identified in one sample were removed prior to downstream analysis. Marker genes for each cluster were identified using the FindAllMarkers() function. Visualizations were generated using Seurat (v 4.3.0), SCpubr (v. 2.0.1), scCustomize (v. 1.1.3), EnhancedVolcano (v. 1.16.0), ggplot2 (v. 3.5.4), and UpSetR (v. 1.4.0).
To identify differentially expressed genes for B cells and for GCs, raw counts were extracted for the corresponding clusters from the integrated Seurat object and “pseudobulked” by summing within individual datasets. Genes were filtered to select only those that were expressed in at least 25 percent of cells (within the cluster) in at least one dataset. Differentially expressed genes were identified using quasi-likelihood generalized linear modeling implemented with edgeR (v.3.40.2).
Changes in composition between ages were detected using scProportionTest (v. 0.0.0.9). Pseudotime trajectory analysis of GCs was performed using monocle392 (v. 1.3.4). GO terms enriched along the GCs pseudotime trajectory were identified by GO enrichment analysis for the top fifty genes positively and negatively correlated with pseudotime, implemented using Metascape93 with all genes expressed by GCs as the background gene list. GSEA analysis was performed using WebGestaltR package94 (v. 0.4.6) with the non-redundant Gene Ontology (Biological Process, Molecular Function), KEGG, and Wikipathway databases and a minimum of five genes per gene set.
Cell-cell communication analysis was performed and visualized using CellChat95 (v. 2.0.1). Relative information flow bar plots were generated from the output of CellChat rankNet() function, with filtering to display only pathways that were statistically significantly different (p-value > 0.01) and had at least a 20% change in signaling between ages. Heatmap visualizations were generated from the output of CellChat netAnalysis_signalingRole_heatmap(), by filtering to keep only cell types of interest, and scaling the data across rows to show the relative contributions to incoming/outgoing signaling per pathway for each cell type.
Slide-Seq
The Slide-seq pucks used in this study were generated and sequenced at the Broad Institute (Cambridge, MA) by Dr. Fei Chen’s group according to the methods and supplementary information provided in a previous study33. The spatial barcode sequencing file of the pucks was generated via monobase ligation chemistry and provided by the Chen lab to be used for later analysis. The pucks were received at the Stowers Institute for Medical Research on small glass coverslips in 1.7 mL LoBind tubes at room temperature and were stored in the dark at 4oC.
Ovaries were briefly washed in cold-PBS and individually submerged in optimal cutting temperature (OCT) compound. Tissues were moved to a cryo-mold with fresh OCT and oriented for sagittal tissue sectioning. Tissue blocks were then flash frozen at −70°C. After freezing, tissue blocks were acclimated to −13°C in a Cryostat NX70 (Thermo Fisher Scientific) for 30 minutes prior to sectioning. Tissue blocks were then mounted on a cutting block with OCT and sectioned at a 5° cutting angle with 10-μm section thickness. Regions of interest (multiple oocytes on the same plain) were identified prior to tissue capture. Four histological cryo-sections from 2 female mice for each age cohort were individually placed on the slide-seq puck slides. This sample size was determined based on puck availability. Pucks with adhered tissue were then removed from the cryostat and placed into a 1.5-ml Eppendorf tube containing 200 µL hybridization buffer (6X SSC, 2 unit/µL Lucigen RNase inhibitor) for 30 minutes at room temperature to facilitate the binding of mRNA to the spatially barcoded beads of the puck. The puck platform was sensitive to the integrity and makeup of the old tissue, which resulted in lower mRNA binding to the barcoded beads.
First-strand cDNA synthesis was performed in RT solution (115 µL water, 40 µL 5X Maxima RT buffer, 20 µL 10 mM dNTPs, 5 µL RNase Inhibitor, 10 µL 50 uM Template Switch Oligo, 10 µL Maxima H-RTase) for 30 minutes at room temperature, followed by 1.5 hours at 52°C. After reverse transcription, the tissue was removed by adding 200 µL 2X Tissue Digestion Buffer (100 mM Tris pH 8.0, 200 mM NaCl, 2% SDS, 5 mM EDTA, 32 unit/µL Proteinase K) and incubating at 37°C for 30 minutes. Following this incubation, 200 µL wash buffer (10 mM Tris pH 8.0, 1 mM EDTA, 0.01% Tween-20) was then added and the mixture pipetted to remove the beads from the coverslip. A series of 3 washes was performed on the beads using 200 µL wash buffer for the first two washes and 200 µL 10 mM Tris-HCl (pH 8.0) for the last wash, pelleting for 3 minutes at 3000 RCF, removing the supernatant, and resuspending beads. Samples were then treated with exonuclease I solution (170 µL water, 20 µL ExoI buffer, 10 µL ExoI) for 50 minutes at 37°C. A series of 2 washes was performed on the beads using 200 µL wash buffer and pelleting for 3 minutes at 3000 RCF, removing the supernatant and resuspending. Samples were then resuspended in 0.1 N NaOH and incubated for 5 minutes at room temperature, followed by a series of 2 washes in 200 µL wash buffer. The beads were resuspended in 200 µL TE, pelleted, the supernatant removed, resuspended in Second Strand Mix (133 µL water, 40 µL 5X Maxima RT buffer, 20 µL 10 mM dNTPs, 2 µL 1 mM dN-SMRT oligo, 5 µL Klenow Enzyme) and incubated for 1 hour at 37°C. The beads were washed 3 times with 200 µL wash buffer, resuspended with 200 µL water, and then pelleted for 3 minutes at 3000 RCF. The final resuspension of the tissue digestion was in 200 uL of 10 mM Tris-HCl (pH 8.0). The final resuspension prior to Second Strand Synthesis was in TE. The 200 uL cDNA PCR master mix consisted of 88 µL H2O, 100 µL Terra PCR Direct Buffer, 4 µL Terra Polymerase (Takara, 639270), 4 µL 100 uM Truseq PCR handle primer, and 4 µL 100 µM SMART PCR primer. The final cDNA purification was eluted in 20 µL of H2O. Subsequent library preparation was performed using the Nextera XT kit (Illumina, FC-131-1096), starting with 600 pg of cDNA using a specific P5-Truseq PCR hybrid oligo in place of the Nextera XT i5 adapter, according to manufacturer’s instructions. The PCR product was purified twice with 0.6X AMPure XP beads (Beckman Coulter, A63881), resuspended in 20 µL water, and checked for quality and quantity using the Bioanalyzer (Agilent) and Qubit Fluorometer (Thermo Fisher Scientific). Libraries were pooled and sequenced as 42-bp pair-end reads on four Illumina NextSeq 500 flow cells on the NextSeq 500 instrument using NextSeq Control Software 2.2.0.4 with the following paired read length: 42 bp Read1, 8 bp i7 Index and 42 bp Read2. Following sequencing, Illumina Primary Analysis version NextSeq RTA 2.4.11 and cellranger mkfastq (4.0.0) were run to demultiplex reads for all libraries and generate FASTQ files. Alignment, filtering, barcode, and unique molecular identifier (UMI) counts were done using cellranger count (4.0.0). The barcode-gene matrices were filtered with >=10 unique molecular identifier (UMI) counts using the unique barcode list. Each puck received approximately 400,000 to 500,000 spatial beads, corresponding to 175 genes per bead.
Slide-Seq analysis
Syrah96 was downloaded from GitHub (https://github.com/0x644BE25/Syrah) (version 480d2da, which was private at the time, used with “vs2” bead oligo version and maxLinkDist=5). After running Syrah, DigitalExpression from the dropseq (2.3.0) pipeline was used to generate digital expression matrices from the Syrah bam files (arguments MIN_NUM_TRANSCRIPTS_PER_CELL = 10 and READ_MQ = 10). These matrices were used to generate a Seurat object with SCTransform v2, followed by UMAP, clustering, and marker identification for each puck. Two pucks that best retained the underlying tissue structure were selected for further analysis and visualizations. For these pucks, tissue segmentation was performed using the R package Vesalius42 (v. 1.0.1). Stray beads were identified based on examination of the Vesalius territories and number of features per bead and were manually removed using the R package tidygate (v. 0.4.9). An hematoxylin and eosin (H&E)-stained image of the adjacent tissue section was used to assist in the identification of tissue boundaries for removal of stray beads and for the assignment of tissue region to structural and morphological histology features. Genes that were spatially enriched in tissue regions were identified either by using Seurat FindMarkers() for the beads in the region of interest, or by computing the Pearson correlation coefficient between a binary vector of the beads’ presence or absence in the tissue region and a vector for the expression data for each gene. Assignment of beads to a cell type was performed using robust cell type decomposition (spacexr, v. 2.2.1), using the respective single-cell datasets (young or old) to build a reference for each SlideSeq dataset.
For the analysis of the spatial distribution of GCs within the follicle, bead positions were transformed into micron space. To do this, the raw puck files containing all beads were filtered to remove stray beads using a k-nearest neighbors approach using the dbscan::kNNdist() function with k = 10. Beads were filtered to remove beads below the 98.5th percentile based on manual inspection of various filtering thresholds. Once stray beads were removed, coordinates for each bead (in micron space) were calculated by taking the range of values in the X and Y dimensions and rescaling them to achieve a diameter of 3000 microns (based on an assumed 3 mm diameter puck). Spatial distributions of cell types within the follicle were determined using the cell type assigned by RCTD and calculating the distance of each bead to the centroid of the beads corresponding to the Vesalius tissue regions for the follicle of interest.
DAB and hematoxylin immunohistochemistry
Five and four ovaries, respectively, were collected from 6-12 week or 15 month-old C57BL/6 mice, fixed in Modified Davidson’s fixative (Electron Microscopy Sciences, Hatfield, PA) overnight at 4°C, then washed in 70% ethanol and processed in an automated tissue processor (Leica Biosystems, Deer Park, IL). Cohort size was selected to allow for robust statistical analysis. Ovaries were embedded in paraffin and sectioned at 5 μm thickness. All mice were sampled while in diestrus to account for potential effects of ovarian cyclicity.
Tissue sections were deparaffinized in Citrosolv, rehydrated in graded ethanol dilutions, and then washed in deionized water. Antigen retrieval was carried out by incubation in 1X Reveal Decloaker (Biocare Medical, Concord, CA) in a steamer for 30 minutes. Endogenous peroxide activity was blocked by incubation with 3% hydrogen peroxide, and endogenous avidin and biotin were blocked using an avidin/biotin blocking kit (Vector Laboratories). Slides were blocked in 3% bovine serum albumin (BSA, Sigma-Aldrich) and 10% goat serum (Vector Laboratories) for 1 hour, prior to overnight incubation with a primary antibody for CD3ε (1:400; Cell Signaling Technology (Cat##78588), Danvers, MA) diluted in 3% BSA at 4°C. Protein concentration-matched non-immune rabbit IgG was used for negative controls. Sections were washed in TBS with 0.1% Tween 20 and incubated with a biotinylated anti-rabbit (1:200; Vector Laboratories (Cat# BA-1000) secondary antibody for 1 hour at room temperature. Signal amplification was performed with a Vectastain Elite ABC Kit (Vector Laboratories), and detection was performed with a 3,3’- diaminobenzidine (DAB) with the DAB Peroxidase (HRP) Substrate Kit (Vector Laboratories). Tissue sections were counterstained with Harris hematoxylin, dehydrated, and mounted with Cytoseal XYL. Sections were imaged with an EVOS M7000 (Thermo Fisher Scientific) at 20X.
CD3ε+ staining quantification and analysis
Images of DAB and hematoxylin-stained sections were assigned a de-identified name to allow for blinded analysis and then opened using ImageJ software. H DAB color deconvolution was performed on both CD3ε-probed sections and adjacent IgG-probed control sections. For each image, the color 2 channel (corresponding to DAB) was selected, thresholded at 6-188, and then converted to a binary mask.
A region of interest (ROI) was manually drawn around the ovary selected for analysis. Within the ROI, particles were analyzed using a particle size filter set to five to infinity (units based on pixel area). The total area of the analyzed particles (representing CD3ε+ staining) was summed and normalized to the total ROI area to calculate the CD3ε+ signal per unit area of the ovary.
The same procedure was performed in IgG control images to quantify non-specific staining. The average non-specific signal per unit area detected in IgG control sections was subtracted from each CD3ε+ signal value to account for background staining and obtain corrected CD3ε+ signal values.
Statistics and reproducibility
The single-cell datasets included two replicates of each age. Differences in cell proportions between ages were determined using the ‘permutation_test‘ function in the scProportionTest R package, with the default number of permutations (n = 1000). Changes in proportion were deemed significant if the absolute log2 fold difference was greater than 0.58, and error bars in plots throughout represent the 95% confidence interval.
For Slide-seq, analysis was performed on a single tissue slice for each age.
For DAB and hematoxylin immunohistochemistry, we collected and analyzed 1-2 ovarian sections from 5 young mice and 4 old mice.
For test statistics for marker genes, either for beads in the specified tissue regions of the young mouse ovary Slide-seq V2 dataset, or clusters in the integrated single-cell dataset, were performed using the ‘FindMarkers‘ function in Seurat (v. 4.3.0.1) with the parameters ‘test.use = “roc”, only.pos = TRUE‘ for the ROC analysis, or ‘test.use = “wilcox”, only.pos=TRUE‘ for Wilcoxon testing.
Significant changes in signaling pathways between young and old single-cell datasets (n = 2 per age) were determined using the ‘rankNet‘ function implemented in the CellChat R package (v. 2.1.0). Significance was determined by performing paired Wilcoxon tests by specifying ‘do_stats=TRUE‘. The results from the rankNet function were further filtered to display pathways where p-value <= 0.01 and the relative change in signaling between ages was >= 20 percent.
Comparison of the area positive for CD3ε was performed using an unpaired t-test.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
Original data underlying this manuscript can be accessed from the Stowers Original Data Repository at https://www.stowers.org/research/publications/libpb-2605. RNA seq data is available via GEO GSE317144 and the Single Cell Portal accession SCP3321.
References
Zabak, S., Varma, A., Bansod, S. & Pohane, M. R. Exploring the complex landscape of delayed childbearing: factors, history, and long-term implications. Cureus 15, e46291 (2023).
Frejka, T. & Sardon, J.-P. First birth trends in developed countries. Demogr. Res. 15, 147–180 (2006).
Gladys, M. & Martinez, K. D. Fertility of Men and Women Aged 15–49 in the United States: National Survey of Family Growth, 2015–2019. Natl. Health Stat. Rep. 179, 1–22 (2023).
Whiteley, J., DiBonaventura, M., Wagner, J. S., Alvir, J. & Shah, S. The impact of menopausal symptoms on quality of life, productivity, and economic outcomes. J. Women’s. Health 22, 983–990 (2013).
El Khoudary, S. R. et al. Patterns of menstrual cycle length over the menopause transition are associated with subclinical atherosclerosis after menopause. Menopause 29, 8–15 (2021).
Merone, L., Tsey, K., Russell, D. & Nagle, C. Sex Inequalities in Medical Research: A Systematic Scoping Review of the Literature. Women’s. Health Rep. 3, 49–59 (2022).
Richards, J. S. & Pangas, S. A. The ovary: basic biology and clinical implications. J. Clin. Invest. 120, 963–972 (2010).
Kinnear, H. M. et al. The ovarian stroma as a new frontier. Reproduction 160, R25–R39 (2020).
Field, S. L., Dasgupta, T., Cummings, M. & Orsi, N. M. Cytokines in ovarian folliculogenesis, oocyte maturation and luteinisation. Mol. Reprod. Dev. 81, 284–314 (2014).
Matsuura, T. et al. Anti-macrophage inhibitory factor antibody inhibits PMSG-hCG-induced follicular growth and ovulation in mice. J. Assist. Reprod. Genet 19, 591–595 (2002).
Kirsch, T. M., Friedman, A. C., Vogel, R. L. & Flickinger, G. L. Macrophages in corpora lutea of mice: characterization and effects on steroid secretion. Biol. Reprod. 25, 629–638 (1981).
Isola, J. V. V. et al. A single-cell atlas of the aging mouse ovary. Nat. Aging 4, 145–162 (2024).
Winkler, I. et al. The cycling and aging mouse female reproductive tract at single-cell resolution. Cell 187, 981–998.e925 (2024).
Ben Yaakov, T., Wasserman, T., Aknin, E. & Savir, Y. Single-cell analysis of the aged ovarian immune system reveals a shift towards adaptive immunity and attenuated cell function. Elife 12, e74915 (2023).
Shen, L., Liu, J., Luo, A. & Wang, S. The stromal microenvironment and ovarian aging: mechanisms and therapeutic opportunities. J. Ovar. Res 16, 237 (2023).
Amargant, F. et al. Ovarian stiffness increases with age in the mammalian ovary and depends on collagen and hyaluronan matrices. Aging Cell 19, e13259 (2020).
Foley, K. G., Pritchard, M. T. & Duncan, F. E. Macrophage-derived multinucleated giant cells: hallmarks of the aging ovary. Reproduction 161, V5–V9 (2021).
Huang, Y. et al. Inflamm-aging: a new mechanism affecting premature ovarian insufficiency. J. Immunol. Res 2019, 8069898 (2019).
Zhang, Z., Schlamp, F., Huang, L., Clark, H. & Brayboy, L. Inflammaging is associated with shifted macrophage ontogeny and polarization in the aging mouse ovary. Reproduction 159, 325–337 (2020).
Machlin J. H. et al. Fibroinflammatory signatures increase with age in the human ovary and follicular fluid. Int. J. Mol. Sci. 22, 4902 (2021).
Rowley J. E. et al. Low molecular weight hyaluronan induces an inflammatory response in ovarian stromal cells and impairs gamete development in vitro. Int. J. Mol. Sci. 21, 1036 (2020).
Zhang, J. et al. A multi-omic single-cell landscape of the aging mouse ovary. Geroscience 47, 4485–4498 (2025).
Babayev, E. et al. Cumulus expansion is impaired with advanced reproductive age due to loss of matrix integrity and reduced hyaluronan. Aging Cell 22, e14004 (2023).
Mara, J. N. et al. Ovulation and ovarian wound healing are impaired with advanced reproductive age. Aging 12, 9686–9713 (2020).
Wang, S. et al. Single-cell transcriptomic atlas of primate ovarian aging. Cell 180, 585–600.e519 (2020).
Lan, T. C. T. et al. Aging disrupts spatiotemporal coordination in the cycling ovary. BioRxiv. (2024).
Morris M. E. et al. A single-cell atlas of the cycling murine ovary. Elife 11, e77227 (2022).
Han, X. et al. Mapping the Mouse Cell Atlas by Microwell-Seq. Cell 172, 1091–1107.e1017 (2018).
Walchli, T. et al. Single-cell atlas of the human brain vasculature across development, adulthood and disease. Nature 632, 603–613 (2024).
Wu, J. et al. Aging conundrum: A perspective for ovarian aging. Front Endocrinol. 13, 952471 (2022).
Lliberos, C. et al. Evaluation of inflammation and follicle depletion during ovarian ageing in mice. Sci. Rep. 11, 278 (2021).
Jin, C. et al. Molecular and genetic insights into human ovarian aging from single-nuclei multi-omics analyses. Nat. Aging 5, 275–290 (2025).
Stickels, R. R. et al. Highly sensitive spatial transcriptomics at near-cellular resolution with Slide-seqV2. Nat. Biotechnol. 39, 313–319 (2021).
Liang, L. F., Chamow, S. M. & Dean, J. Oocyte-specific expression of mouse Zp-2: developmental regulation of the zona pellucida genes. Mol. Cell Biol. 10, 1507–1515 (1990).
Roller, R. J., Kinloch, R. A., Hiraoka, B. Y., Li, S. S. & Wassarman, P. M. Gene expression during mammalian oogenesis and early embryogenesis: quantification of three messenger RNAs abundant in fully grown mouse oocytes. Development 106, 251–261 (1989).
Berkel, C. Hormonally-regulated KCTD14 Expression is higher in oocytes from antral follicles compared to those from preantral follicles in mice, with implications in fertility. Reprod. Sci. 32, 1374–1378 (2025).
Duncan F. E., Confino R., Pavone M. E. Female Reproductive Aging: From Consequences to Mechanisms, Markers, and Treatments, 2nd edn. Elsevier (2019).
Parkening, T. A., Collins, T. J. & Elder, F. F. Orthotopic ovarian transplantations in young and aged C57BL/6J mice. Biol. Reprod. 32, 989–997 (1985).
Broekmans, F. J., Soules, M. R. & Fauser, B. C. Ovarian aging: mechanisms and clinical consequences. Endocr. Rev. 30, 465–493 (2009).
Mantri, M., Zhang, H. H., Spanos, E., Ren, Y. A. & De Vlaminck, I. A spatiotemporal molecular atlas of the ovulating mouse ovary. Proc. Natl. Acad. Sci. USA 121, e2317418121 (2024).
Richards, J. S., Ren, Y. A., Candelaria, N., Adams, J. E. & Rajkovic, A. Ovarian follicular theca cell recruitment, differentiation, and impact on fertility: 2017 update. Endocr. Rev. 39, 1–20 (2018).
Martin, P. C. N., Kim, H., Lovkvist, C., Hong, B. W. & Won, K. J. Vesalius: high-resolution in silico anatomization of spatial transcriptomic data using image analysis. Mol. Syst. Biol. 18, e11080 (2022).
DeJarnette, J. B. et al. Specific requirement for CD3epsilon in T cell development. Proc. Natl. Acad. Sci. USA 95, 14909–14914 (1998).
Converse, A. et al. Multinucleated giant cells are hallmarks of ovarian aging with unique immune and degradation-associated molecular signatures. PLoS Biol. 23, e3003204 (2025).
Ribot, J. C. et al. CD27 is a thymic determinant of the balance between interferon-gamma- and interleukin 17-producing gammadelta T cell subsets. Nat. Immunol. 10, 427–436 (2009).
McKenzie, D. R., Comerford, I., Silva-Santos, B. & McColl, S. R. The emerging complexity of gammadeltaT17 cells. Front Immunol. 9, 796 (2018).
Velikkakam, T., Gollob, K. J. & Dutra, W. O. Double-negative T cells: Setting the stage for disease control or progression. Immunology 165, 371–385 (2022).
Olguin-Martinez, E., Ruiz-Medina, B. E. & Licona-Limon, P. Tissue-specific molecular markers and heterogeneity in Type 2 innate lymphoid cells. Front. Immunol. 12, 757967 (2021).
Sadeghalvad, M., Khijakadze, D., Orangi, M. & Takei, F. Flow cytometric analysis of innate lymphoid cells: challenges and solutions. Front. Immunol. 14, 1198310 (2023).
Qin, M. et al. Tissue microenvironment induces tissue specificity of ILC2. Cell Death Discov. 10, 324 (2024).
Jonsson, A. H. et al. Granzyme K(+) CD8 T cells form a core population in inflamed human tissue. Sci. Transl. Med. 14, eabo0686 (2022).
Louisa, F. Alim 1 CK, Fernando Souza-Fonseca-Guimarães 3. Molecular mechanisms of tumour necrosis factor signalling via TNF receptor 1 and TNF receptor 2 in the tumour microenvironment. Curr. Opin. Immunol. 86, 102409 (2023).
Wittner, J. & Schuh, W. Kruppel-like factor 2: a central regulator of B cell differentiation and plasma cell homing. Front. Immunol. 14, 1172641 (2023).
Korholz, J. et al. Novel mutation and expanding phenotype in IRF2BP2 deficiency. Rheumatology 62, 1699–1705 (2023).
Tseng, W. Y., Stacey, M. & Lin, H. H. Role of Adhesion G Protein-Coupled Receptors in Immune Dysfunction and Disorder. Int. J. Mol. Sci. 24, 15141 (2023).
Kaur, S. & Roberts, D. D. Emerging functions of thrombospondin-1 in immunity. Semin Cell Dev. Biol. 155, 22–31 (2024).
Sun, S. & Wang, W. Mechanosensitive adhesion G protein-coupled receptors: Insights from health and disease. Genes Dis. 12, 101267 (2025).
Fei, L. et al. Systematic identification of cell-fate regulatory programs using a single-cell atlas of mouse development. Nat. Genet 54, 1051–1061 (2022).
Wang, R. et al. Construction of a cross-species cell landscape at single-cell level. Nucleic Acids Res. 51, 501–516 (2023).
Li, N. et al. Two distinct resident macrophage populations coexist in the ovary. Front. Immunol. 13, 1007711 (2022).
Ma, R. Y., Black, A. & Qian, B. Z. Macrophage diversity in cancer revisited in the era of single-cell omics. Trends Immunol. 43, 546–563 (2022).
Martinez, F. O. et al. Genetic programs expressed in resting and IL-4 alternatively activated mouse and human macrophages: similarities and differences. Blood 121, e57–e69 (2013).
Scuruchi, M. et al. Serglycin as part of IL-1beta induced inflammation in human chondrocytes. Arch. Biochem. Biophys. 669, 80–86 (2019).
Lecker, L. S. M. et al. TGFBI production by macrophages contributes to an immunosuppressive microenvironment in ovarian cancer. Cancer Res. 81, 5706–5719 (2021).
Shen, Z. et al. Expansion of macrophage and liver sinusoidal endothelial cell subpopulations during non-alcoholic steatohepatitis progression. iScience 26, 106572 (2023).
Coletta, S. et al. The immune receptor CD300e negatively regulates T cell activation by impairing the STAT1-dependent antigen presentation. Sci. Rep. 10, 16501 (2020).
Pagliari, M. et al. Helicobacter pylori affects the antigen presentation activity of macrophages modulating the expression of the immune receptor CD300E through miR-4270. Front. Immunol. 8, 1288 (2017).
Isobe, M. et al. The CD300e molecule in mice is an immune-activating receptor. J. Biol. Chem. 293, 3793–3805 (2018).
Carnazzo, V. et al. Calprotectin: two sides of the same coin. Rheumatology 63, 26–33 (2024).
Frangogiannis, N. Transforming growth factor-beta in tissue fibrosis. J. Exp. Med. 217, e20190103 (2020).
Sugiura, K., Pendola, F. L. & Eppig, J. J. Oocyte control of metabolic cooperativity between oocytes and companion granulosa cells: energy metabolism. Dev. Biol. 279, 20–30 (2005).
Hornick, J. E., Duncan, F. E., Shea, L. D. & Woodruff, T. K. Multiple follicle culture supports primary follicle growth through paracrine-acting signals. Reproduction 145, 19–32 (2013).
Zhou, H. et al. Synergy of Paracrine signaling during early-stage mouse ovarian follicle development in vitro. Cell Mol. Bioeng. 11, 435–450 (2018).
Akitsu, A. & Iwakura, Y. Interleukin-17-producing gammadelta T (gammadelta17) cells in inflammatory diseases. Immunology 155, 418–426 (2018).
D’Acquisto, F. et al. Annexin-1 modulates T-cell activation and differentiation. Blood 109, 1095–1102 (2007).
Hayden, M. S. & Ghosh, S. Regulation of NF-kappaB by TNF family cytokines. Semin. Immunol. 26, 253–266 (2014).
Richardson, S. J., Senikas, V. & Nelson, J. F. Follicular depletion during the menopausal transition: evidence for accelerated loss and ultimate exhaustion. J. Clin. Endocrinol. Metab. 65, 1231–1237 (1987).
Faddy, M. J., Gosden, R. G., Gougeon, A., Richardson, S. J. & Nelson, J. F. Accelerated disappearance of ovarian follicles in mid-life: implications for forecasting menopause. Hum. Reprod. 7, 1342–1346 (1992).
Gougeon, A., Ecochard, R. & Thalabard, J. C. Age-related changes of the population of human ovarian follicles: increase in the disappearance rate of non-growing and early-growing follicles in aging women. Biol. Reprod. 50, 653–663 (1994).
Grieb, G., Merk, M., Bernhagen, J. & Bucala, R. Macrophage migration inhibitory factor (MIF): a promising biomarker. Drug N. Perspect. 23, 257–264 (2010).
Sondag, C. M. & Combs, C. K. Amyloid precursor protein mediates proinflammatory activation of monocytic lineage cells. J. Biol. Chem. 279, 14456–14463 (2004).
Saitoh, O., Olson, E. N. & Periasamy, M. Muscle-specific RNA processing continues in the absence of myogenin expression. J. Biol. Chem. 265, 19381–19384 (1990).
Umehara, T. et al. Female reproductive life span is extended by targeted removal of fibrotic collagen from the mouse ovary. Sci. Adv. 8, eabn4564 (2022).
Isola, J. V. V. et al. Reproductive Ageing: Inflammation, immune cells, and cellular senescence in the aging ovary. Reproduction 168, R25–R39 (2024).
Briley, S. M. et al. Reproductive age-associated fibrosis in the stroma of the mammalian ovary. Reproduction 152, 245–260 (2016).
Duncan, F. E. et al. Age-associated dysregulation of protein metabolism in the mammalian oocyte. Aging Cell 16, 1381–1393 (2017).
Losslein, A. K. et al. Monocyte progenitors give rise to multinucleated giant cells. Nat. Commun. 12, 2027 (2021).
Young, M. D. & Behjati, S. SoupX removes ambient RNA contamination from droplet-based single-cell RNA sequencing data. Gigascience 9, giaa151 (2020).
Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184, 3573–3587.e3529 (2021).
McGinnis, C. S., Murrow, L. M. & Gartner, Z. J. DoubletFinder: Doublet Detection in Single-Cell RNA Sequencing Data Using Artificial Nearest Neighbors. Cell Syst. 8, 329–337.e324 (2019).
Hafemeister, C. & Satija, R. Normalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression. Genome Biol. 20, 296 (2019).
Cao, J. et al. The single-cell transcriptional landscape of mammalian organogenesis. Nature 566, 496–502 (2019).
Zhou, Y. et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat. Commun. 10, 1523 (2019).
Liao, Y., Wang, J., Jaehnig, E. J., Shi, Z. & Zhang, B. WebGestalt 2019: gene set analysis toolkit with revamped UIs and APIs. Nucleic Acids Res 47, W199–W205 (2019).
Jin, S. et al. Inference and analysis of cell-cell communication using CellChat. Nat. Commun. 12, 1088 (2021).
Brewster, C. E., Mann, F. G., Benham-Pyle, B. W. & Sánchez Alvarado A. Syrah, a Slide-seqV2 pipeline augmentation. BioRxiv. (2022).
Acknowledgements
We are grateful to the Stowers Institute for Medical Research (SIMR) Sequencing and Discovery Genomics Technology Center, specifically Anoja Perera and Amanda Lawlor, for their technical expertise. We also appreciate the computational support from the SIMR Computational Biology Technology Center, particularly Jay Unruh and Madelaine Gogol. Special thanks to Dr. Fei Chen’s group at the Broad Institute for providing the Slide-seq pucks used to capture our spatial transcriptomics data. We are also indebted to Carolyn Brewster, who wrote the Syrah pipeline, which we used to process our spatial transcriptomics data. Special thanks to the SIMR Rodent Department for their support. We thank Mark Miller for the contribution of original illustrations. Finally, we extend our gratitude to Prianka Hashim for many insightful discussions throughout this project. This work was supported in part by R01HD105752-01 from NICHD to FED and JLG, F31HD116553-01 from the NICHD to AMG, and the Stowers Institute for Medical Research.
Author information
Authors and Affiliations
Contributions
Anna Galligos and Joseph Varberg led the analysis of data, interpretation of data, and wrote the manuscript. Wei-Ting Yueh led data acquisition. Aubrey Converse and Seth Malloy contributed to data acquisition. Fatimah Aljubran contributed to data interpretation. Francesca Duncan and Jennifer Gerton supervised the work.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Communications Biology thanks Jose Victor Isola, Ulises Urzúa, and the other anonymous reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Christopher Hine and Mengtan Xing. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Galligos, A., Varberg, J.M., Yueh, WT. et al. Multicellular origins of murine ovarian inflammaging. Commun Biol 9, 593 (2026). https://doi.org/10.1038/s42003-026-09826-1
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s42003-026-09826-1
