M2 Macrophages are Major Mediators of Germline Risk of Endometriosis and Explain Pleiotropy With Comorbid Traits

Abstract

Endometriosis is a common gynecologic condition that causes chronic, life-altering symptoms including pain and infertility. There is an urgent need for new non-hormonal targeted therapeutics to treat endometriosis, but until very recently, the cellular and molecular signatures of endometriotic lesions are undefined, hindering the development of clinical advances. Integrating inherited risk data from analyses of >45 0000 individuals with ≈35 0000 single-cell transcriptomes from 21 patients, M2-macrophages as candidate drivers of disease susceptibility are uncovered, and nominating IL1 signaling as a central hub impacted by germline genetic variation associated with endometriosis risk. Extensive functional follow-up confirmed these associations and revealed a pleiotropic role for this pathway in endometriosis. Population-scale expression quantitative trait locus analysis demonstrates that genetic variation controlling IL1A expression is associated with endometriosis risk variants. Manipulation of IL1 signaling in state-of-the-art in vitro decidualized endometrial organoids impacts epithelial differentiation, and in an in vivo endometriosis model, treatment with anakinra (an interleukin-1 receptor antagonist) results in a significant, dose-dependent reduction in spontaneous and evoked pain and dampened pro-angiogenic signaling. Together, these studies highlight non-diagnostic cell types as central to endometriosis susceptibility and support IL1 signaling as an important actionable pathway for this disease.

Keywords: GWAS; IL1A; IL1B; M2 macrophages; anakinra; angiogenesis; endometriosis; inflammation; organoids; single cell transcriptomics.

Figure 1

Endometriosis risk gene expression is enriched in M2 macrophages, dendritic cells and plasmablasts/plasma cells. A) UMAP representation of the major immune and structural cell types in endometriosis. The data set includes 362700 cells from 49 samples taken from 21 patients. Erythrocytes are excluded. B) UMAP representation of the immune subclusters present. C) Expression of the canonical markers that define the immune clusters. D–F) Distribution of the single cell disease relevance score in D) the UMAP space, E) per cell cluster, and F) stage. The Monte Carlo test is significant for M2 macrophages (p = 0.01), dendritic cells (p = 0.05), and endothelial cells (p = 0.04). Geary's C as a test statistic for heterogeneity within cell type is significant for macrophages (p = 0.03), dendritic cells (p = 0.03), and plasma cells/plasmablasts (p = 0.05). G) scDRS variance explained by different cell categories according to linear regression. UMAP, uniform manifold approximation and projection; NK, natural killer cells. Only the 200798 cells from endometrioma and peritoneal endometriosis are included in F). Each stage group is represented by 7 patients. In E) and F) upper and lower bounds of boxes represent upper and lower quartiles, with central lines at the median. Solid lines above and below the boxes denote the minimum and maximum data points, and black dots denote outliers. Blue squares highlight cell populations with significant heterogeneity in risk scores (p < 0.05).

Figure 2

Endometriosis risk is consistently associated with IL1A/IL1B gene expression. A) Spearman correlation between gene expression and scDRS. B) Top five enriched pathways among genes correlated with scDRS. C) scDRS heterogeneity relative to PROGENy pathway activity. Spearman correlation. D) Gene expression of candidate susceptibility genes at the 2q41.1 endometriosis risk locus in immune subsets. E) Expression quantitative trait locus analysis of differences in candidate risk gene expression associated with endometriosis risk variant genotype.

Figure 3

M2 macrophages play a central role in cell‐cell communications in endometriosis. A) Predicted interactions for every cell cluster. B,C) pathways affected by B) incoming and, C) outgoing signals for M2 macrophages. D–F) Ligand‐receptor pairs identified for D) TNF, E) Visfatin, F) CXCL. G) IL1 signaling expression in non‐immune cell clusters and subclusters. H) Expression of interleukin‐1 pathway genes in EnEpi and EnS. I) Expression of interleukin‐1 pathway genes in EnEpi and EnS of eutopic endometrium from patients with and without a confirmed diagnosis of endometriosis. Genes are sorted by expression rank. EnEpi, endometrial‐type epithelium; EnS, endometrial‐type stroma; NK, natural killer cells. In H and I genes are ordered from low to high expression.

Figure 4

Stimulation of interleukin‐1 signaling disrupts the organization of human endometrial organoid‐stromal co‐cultures. A) UMAP representation of single‐cell transcriptomes for 39568 endometrial epithelial cells from organoids. B) IL1 signaling for ciliated and non‐ciliated cells. T‐test p < 0.0001(****). C) Expression of selected markers from IL1 signaling pathway. D) Spearman correlation between estrogen activity and IL1 signaling (p = 0.017). E‐G) 3‐D Incucyte photos capturing the development of the organoid co‐cultures. E) Single cell seeding of epithelial and mesenchymal stem cells. Arrowhead denotes a representative single cell. F) Treatment with estradiol (E2) and cell proliferation. Arrowhead shows a representative proliferating organoid. G) Treatment with E2 in combination with progesterone (P4), stromal cell decidualization (arrowhead), and epithelial organoid expansion (arrow). H) Treatment with IL1B results in disordered epithelial differentiation (arrowhead).

Figure 5

Blocking IL1 signaling reduces endometriosis‐associated pain in a mouse model. Endometriosis‐like lesions were induced in C57BL/6J mice, allowed to grow for 4 weeks, and treated with anakinra at the indicated dose. Spontaneous pain was decreased after treatment as measured by A) Abdominal squashing and B) abdominal contortions. C) Evoked pain measured using von Frey fibers on the abdomen was also decreased in a dose‐dependent manner. D) Masson's trichrome staining of the resulting lesions, including representative microscopic images (left) and masks (right) indicating in red pixels scored as positive for collagen staining (blue). E) Lesion area measured and F) area stained positive for collagen. G) Angiogenesis dot‐blot array using extracts from vehicle and anakinra‐treated lesions. Highlighted growth factors showed at least two‐fold reduction in mean normalized signal. H) ELISA‐measured VEGF concentration in vehicle‐ and anakinra‐treated lesions. ^* p < 0.05, ^** p < 0.01, ^***** p < 0.001, ^**** p < 0.0001 by ANOVA (B,C,E,F) or Student's t‐test (A,H). Investigators were blinded to group allocation until all analyses were complete. n = 10 mice per group, except for the 10 mg kg⁻¹ group, where n = 9.

Figure 6

scDRS identifies cell types that may explain associations with traits that have shared germline genetics with endometriosis. scDRS scores for other traits that share risk genetics with endometriosis. A black border highlights cell types with significant associations with the trait. An asterisk indicates significant heterogeneity detected.