Self-Organizing Ovarian Somatic Organoids Preserve Cellular Heterogeneity and Reveal Cellular Contributions to Ovarian Aging

Abstract

Ovarian somatic cells are essential for reproductive function, but no existing ex vivo models recapitulate the cellular heterogeneity or interactions within this compartment. We engineered an ovarian somatic organoid model by culturing a stroma-enriched fraction of mouse ovaries in scaffold-free agarose micromolds. Self-organized ovarian somatic organoids maintained diverse cell populations, produced extracellular matrix, and secreted hormones. Organoids generated from reproductively old mice exhibited reduced aggregation and growth compared to young counterparts, as well as differences in cellular composition. Interestingly, matrix fibroblasts from old mice demonstrated upregulation of pathways associated with the actin cytoskeleton and downregulation of cell adhesion pathways, indicative of increased cellular stiffness that may impair organoid aggregation. Cellular morphology, which is regulated by the cytoskeleton, significantly changed with age and in response to actin modulation. Moreover, actin modulation altered organoid aggregation efficiency. Overall, ovarian somatic organoids have advanced knowledge of cellular contributions to ovarian aging.

Keywords: actin; cellular stiffness; organoid; ovary; reproductive aging; stroma.

FIGURE 1

Ovarian somatic organoids self‐assemble into solid structures containing key ovarian cell populations with distinct spatial organization. (A) Schematic of workflow to isolate primary mouse ovarian somatic cells and generate organoids and images of an empty agarose micromold (i) and a scan of an agarose micromold following organoid generation (ii). (B) Representative transmitted light images of murine ovarian somatic organoids cultured in IntestiCult at Days 1, 3, and 5 of culture. N = 4 micromolds. Representative image of H&E‐stained murine organoid section following 5 days of culture in IntestiCult. Scale bars for transmitted light images = 400 μm and scale bars for H&E image = 20 μm. (C) Representative images of murine ovarian somatic organoids at Days 1, 3, and 5 of additional culture following removal from micromolds after 3 initial days of culture. Scale bars = 200 μm. Quantification of average organoid area over 5 additional days of culture following removal from micromolds. Error bars represent the standard error of the mean (SEM). p < 0.05 for a versus b and p < 0.0001 for all other comparisons by one‐way ANOVA. N = 28 organoids. (D) Representative images of Ki67 (brown) and cleaved caspase‐3 (CC3, brown) IHC staining of mouse ovarian tissue sections and murine ovarian somatic organoid sections. Nuclei were detected by hematoxylin staining (blue‐purple). Scale bars = 20 μm. Insets are thresholded images highlighting positive pixels. Quantification of the percent Ki67 and CC3 positive area in regions of interest in mouse ovaries (N = 3 ovaries) and ovarian somatic organoids after 5 days of culture (N = 40–44). Error bars represent the standard error of the mean (SEM). (E) Representative images of IHC labelling of murine ovarian organoid sections after 6 days of culture using antibodies against Vimentin (brown), F4/80 (brown), FOXL2 (brown), and 3β‐HSD (green). For chromogenic staining nuclei were detected by hematoxylin (blue‐purple), and for fluorescent staining nuclei were detected by DAPI (blue). Arrows show F4/80‐positive macrophages enriched at the perimeter of organoids (black) and 3β‐HSD‐positive steroidogenic cells consistently depleted from the core of organoids (white). Insets are thresholded images highlighting positive pixels. Scale bars = 20 μm. N = 39–49 organoids. (F) Representative scan of H&E‐stained rhesus macaque ovarian tissue. Dashed line indicates separation of cortex (C) from medulla (M). Scale bar = 200 μm. (G) Representative images of ovarian somatic organoids generated from rhesus macaque ovarian cortex (top) or medulla (bottom) at Days 1, 3, and 5 of culture. Representative images of H&E‐stained rhesus macaque organoid sections following 5 days in culture. Scale bars for transmitted light images = 400 μm and scale bars for H&E images = 100 μm. N = 2 biological replicates for ovarian somatic cell isolation and ovarian somatic organoid generation from rhesus macaque ovarian tissue, 1–2 micromolds per biological replicate. (H) Representative images of IHC labelling of rhesus macaque medullary organoid sections after 6 days of culture using antibodies against Ki67 (brown), cleaved caspase‐3 (CC3, brown), Vimentin (brown), CD68 (brown), and 3β‐HSD (brown). Nuclei were detected by hematoxylin staining (blue‐purple). Insets are thresholded images highlighting positive pixels. Scale bars = 20 μm. N ≥ 30 organoids.

FIGURE 2

Organoids exhibit age‐dependent differences in aggregation, growth, and function. (A) Representative images of ovarian somatic organoids generated from reproductively young (6–12 weeks) or reproductively old (10–14 months) mice at days 1, 3, and 5 of culture. Scale bars = 400 μm. N = 4 micromolds per age. (B, C) Quantification of the average number of aggregates per microwell (B) and average area per organoid (C) over 5 days of culture for organoids generated from reproductively young and old mice. Error bars represent the standard error of the mean (SEM). *p < 0.05, **p < 0.01, and ****p < 0.0001 by two‐way ANOVA. N = 4 micromolds per age. (D, E) Representative images of Ki67 (D, brown) and cleaved caspase‐3 (CC3, E, brown) IHC staining of young and old ovarian somatic organoid sections. Nuclei were detected by hematoxylin staining (blue‐purple). Scale bars = 20 μm. Insets are thresholded images highlighting positive pixels. Quantification of the percent Ki67 (D) and CC3 (E) positive area in young (N = 42–46) and old (N = 44–57) ovarian somatic organoids after 5 days of culture. Error bars represent the standard error of the mean (SEM). *p < 0.05 and ****p < 0.0001 by Welch's t‐tests. (F) Representative images of hyaluronan binding protein (HABP, green) assays performed with young and old ovarian somatic organoid sections. Nuclei (blue) were detected with DAPI. Scale bars = 20 μm. Organoids are outlined by dashed lines. Quantification of the percent HABP‐positive area in young (N = 43) and old (N = 53) ovarian somatic organoids after 5 days of culture. Error bars represent the standard error of the mean (SEM). (G) Representative images of Picrosirius Red (PSR)‐stained young and old ovarian somatic organoid sections. Scale bars = 20 μm. Insets are thresholded images highlighting positive pixels. Quantification of the percent PSR‐positive area in young (N = 43) and old (N = 53) ovarian somatic organoids after 5 days of culture. Error bars represent the standard error of the mean (SEM). (H, I) Gene expression of Col1a1 (H) and Acta2 (I) in ovarian somatic organoids measured by RT‐qPCR following 48‐h treatment with 10 ng/mL TGF‐β or control. Gene expression for TGF‐β‐treated organoids was graphed as fold‐change over young control. *p < 0.05 by Welch's t‐tests after performing Shapiro–Wilk tests to confirm normality. N = 5 trials. (J, K) Estradiol (J) and progesterone (K) secretion measured in conditioned media of young and old ovarian somatic organoids by ELISAs at Days 1, 3, and 5 of culture. *p < 0.05 and **p < 0.01 by Kruskal–Wallis tests followed by Dunn's multiple comparisons tests to compare between time points for each age group separately. N = 9 micromolds over 3 trials.

FIGURE 3

Heterogenous cell populations are maintained in ovarian somatic organoids irrespective of age, over 6 days in culture. (A) Uniform manifold approximation and projection (UMAP) plot of primary mouse ovarian somatic cells from reproductively young (6–12 weeks) and reproductively old (10–14 months) mice following plating in 2D, prior to organoid generation (monolayer) and following dissociation of ovarian somatic organoids at Days 1 and 6 of culture where scRNAseq was performed for two replicates per group and each biological replicate included cells pooled from at least five mice. Unbiased clustering revealed 11 distinct cell populations. (B, C) Quantification of the percentage of cells in each cluster for monolayer (Mono), Day 1 organoids (D1), and Day 6 (D6) organoids for each age. Error bars represent the standard error of the mean (SEM). (C) shows percentage of cells in low abundance clusters (0‐Endothelial, 3‐Granulosa 2, and 4‐Immune). Cluster color coding as in A. Statistical significance of differences in the percentage of cells in each cluster was not determined because scRNAseq was performed for two replicates per group. (D, E) UMAP plots showing cells colored based on (D) Ki67 expression (negative and positive) and (E) cell cycle stage (G1, G2M, S). (F) Chord diagrams showing cell–cell communication in Day 1 and 6 organoids for each age created using CellChat. The color of the chords indicate the cell cluster sending the signal and the size of the chords are proportional to signal strength. Statistical significance of differences in cell interactions was not determined because scRNAseq was performed on two replicates per group. (G–L) Representative images of vimentin (G, brown), alpha‐smooth muscle Actin (⍺‐SMA, H, brown), desmin (I, brown), F4/80 (J, brown), FOXL2 (K, brown), and 3β‐HSD (L, green) IHC staining of young and old ovarian somatic organoid sections at Days 1 and 6 of culture. For chromogenic staining nuclei were detected by hematoxylin (blue‐purple), and for fluorescent staining nuclei were detected by DAPI (blue). Scale bars = 20 μm. Insets are thresholded images highlighting positive pixels. Quantification of the positive area for each marker in young and old ovarian somatic organoids at Days 1 and 6 of culture (Young Day 1: N = 16–41, Old Day 1: N = 34–50, Young Day 6: N = 15–29, Old Day 6: N = 37–59). Error bars represent the standard error of the mean (SEM). *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 by Kruskal–Wallis tests followed by Dunn's multiple comparisons tests. Statistical significance only for Young Day 1 versus Old Day 1, Young Day 6 versus Old Day 6, Young Day 1 versus Young Day 6, and Old Day 1 versus Old Day 6 is shown on graphs. All other statistically significant comparisons are listed in Table S2.

FIGURE 4

Actin pathways, which regulate cell morphology, are upregulated in ovarian mesenchyme 1 cells with age. (A) GO analysis of differentially expressed genes in the Mesenchyme 1 cluster with age for monolayer cells. Pathways upregulated and downregulated with age are labeled and displayed in manually grouped categories. scRNAseq was performed for two replicates per age and each biological replicate included cells pooled from at least five mice. (B) Raincloud plots showing expression of genes driving upregulated Actin‐related pathways in the Mesenchyme 1 cells from young and old mice. (C) Schematic of the morphological analysis pipeline. Over 89,000 cells spanning both ages (young and old) and all treatments (vehicle, LatA, and JASP) collected from 8 independent biological replicates were utilized for morphological analysis in all subsequent panels in Figure 4. (D) Representative images of ovarian somatic cells from reproductively young (6–12 weeks) and reproductively old (10–14 months) mice. F‐Actin (phalloidin, yellow), vimentin (green), and DAPI (blue) were detected by immunocytochemistry. Scale bars = 200 μm. (E) Box and whisker plots showing quantification of nuclear area, cellular area, and nuclear to cellular area ratio for ovarian somatic cells from young and old mice. ***p < 0.001 by t‐tests. (F) 2D UMAP visualization based on 106 morphological parameters. 8 k‐means clusters are layered on top of the UMAP space with representative cellular (outer, green) and nuclear (inner, blue), and raw image morphology for each cluster. Panels below show distribution of high (red) and low (gray) nuclear area (N: Area), cellular area (C: Area), vimentin expression, and F‐Actin expression displayed within the UMAP space. High and low represent standard scaling of expression across the dataset. (G) Violin plots of the nuclear area, cellular area, vimentin, and F‐Actin content for each of the 8 k‐means clusters. (H) Representative images of an ovarian somatic cells within each morphological cluster. DAPI (blue), F‐Actin (phalloidin, green), and vimentin (red) were detected by immunocytochemistry. Brightness of fluorescent staining was adjusted equivalently for all images to highlight localization. Scale bar = 50 μm.

FIGURE 5

Ovarian somatic cell morphology shifts with age and upon modulation of the Actin cytoskeleton. (A) Contour overlays on the UMAP space showing enriched morphology for ovarian somatic cells from young (purple) and old (red) mice. Over 89,000 cells spanning both ages (young and old) and all treatments (vehicle, LatA, and JASP) collected from 8 independent biological replicates were utilized for morphological analysis for this and all subsequent panels in Figure 5. (B) Heatmap showing the fractional abundance of ovarian somatic cells from young and old mice within each k‐means cluster. (C) Representative 20× and 60× images of ovarian somatic cells from young and old mice treated with vehicle control, Latrunculin A (LatA), and Jasplakinolide (JASP). F‐Actin (phalloidin, yellow), vimentin (green), and nuclei (DAPI, blue) were detected by immunocytochemistry. Arrows show distinct Actin enrichments. 20× Scale bars = 200 μm. 60× Scale bars = 50 μm. (D) Contour overlays on the UMAP space showing enriched morphology for ovarian somatic cells from young (purple) and old (red) mice treated with LatA. (E) Contour overlays on the UMAP space showing enriched morphology for ovarian somatic cells from young (purple) and old (red) mice treated with JASP. (F) Heatmap showing the fractional abundance of ovarian somatic cells from young and old mice with or without LatA or JASP treatment within each k‐means cluster. (G) Bubble plot showing correlation of morphological enrichment between experimental groups.

FIGURE 6

Modulating the Actin cytoskeleton alters the aggregation of ovarian somatic organoids. (A) Representative images of primary ovarian somatic cells following treatment with Latrunculin A (LatA) or vehicle control. F‐actin (phalloidin, yellow), vimentin (green), and nuclei (DAPI, blue) were detected by immunocytochemistry. Scale bars = 50 μm. Representative images of organoids generated from young and old primary ovarian somatic cells following treatment with Latrunculin A (LatA) or vehicle control at 1 and 9 h post seeding into agarose micromolds. Scale bars = 500 μm. Right panels for each timepoint are thresholded masks of cells in agarose micromolds. (B, C) Kernel‐smoothed distribution curves for histograms quantifying the frequency of different numbers of particles per well at 1 h (B) and 9 h (C) following seeding of primary ovarian somatic cells from young (green) and old (red) mice treated with a vehicle control or primary ovarian somatic cells from old mice treated with LatA (blue) into agarose micromolds. N = 1 micromolds per condition. Representative images for N = 2 additional micromolds per condition are shown in Figure S9A,B. (D) Representative images of primary ovarian somatic cells following treatment with Jasplakinolide (JASP) or vehicle control. F‐actin (phalloidin, yellow), vimentin (green), and nuclei (DAPI, blue) were detected by immunocytochemistry. Arrows show distinct actin enrichments. Scale bars = 50 μm. Representative images of organoids generated from young and old primary ovarian somatic cells following treatment with Jasplakinolide (JASP) or vehicle control at 1 h post seeding into agarose micromolds. Scale bars = 500 μm. Right panels for each timepoint are thresholded masks of cells in agarose micromolds. (E) Kernel‐smoothed distribution curves for histograms quantifying the frequency of different numbers of particles per well at 1 h following seeding of primary ovarian somatic cells from young (green) and old (red) mice treated with a vehicle control or primary ovarian somatic cells from young mice treated with JASP (purple) into agarose micromolds. N = 1 micromolds per condition. Representative images for N = 3 additional micromolds per condition are shown in Figure S9F–H.