Androgens Modulate the Immune Profile in a Mouse Model of Polycystic Ovary Syndrome

Abstract

Polycystic ovary syndrome (PCOS) is associated with a low-grade inflammation, but it is unknown how hyperandrogenism, the hallmark of PCOS, affects the immune system. Using a PCOS-like mouse model, it is demonstrated that hyperandrogenism affects immune cell populations in reproductive, metabolic, and immunological tissues differently in a site-specific manner. Co-treatment with an androgen receptor antagonist prevents most of these alterations, demonstrating that these effects are mediated through androgen receptor activation. Dihydrotestosterone (DHT)-exposed mice displayed a drastically reduced eosinophil population in the uterus and visceral adipose tissue (VAT). A higher frequency of natural killer (NK) cells and elevated levels of IFN-γ and TNF-α are seen in uteri of androgen-exposed mice, while NK cells in VAT and spleen displayed a higher expression level of CD69, a marker of activation or tissue residency. Distinct alterations of macrophages in the uterus, ovaries, and VAT are also found in DHT-exposed mice and can potentially be linked to PCOS-like traits of the model. Indeed, androgen-exposed mice are insulin-resistant, albeit unaltered fat mass. Collectively, it is demonstrated that hyperandrogenism causes tissue-specific alterations of immune cells in reproductive organs and VAT, which can have considerable implications on tissue function and contribute to the reduced fertility and metabolic comorbidities associated with PCOS.

Keywords: NK cells; Polycystic ovary syndrome; eosinophils; hyperandrogenism; immunology; insulin resistance.

Figure 1

Immune populations in blood and secondary lymphoid organs are differently affected by androgen exposure. a) Experimental design. b) Frequency of neutrophils in blood, expressed as a percent of CD45⁺ immune cells (n = 11 Control, 9 DHT, 9 DHT + Flutamide). c) Frequency of eosinophils in blood (n = 12, 11, 8). d) Frequency of monocytes in blood (n = 11, 9, 9). e) Neutrophil count in blood (n = 5, 10, 6). f) CD45+ immune cell count in blood (n = 5, 10, 7). g) Frequency of CD3⁺ T cells in blood (n = 8, 10, 9). h) Frequency of CD4⁺ T helper cells in blood (n = 8, 10, 9). i) Frequency of CD8⁺ cytotoxic T cells in blood (n = 8, 10, 9). j) Thymus weight (n = 12, 11, 8). k) Frequency of NK cells in blood (n = 8, 10, 9). (l) CD69 expression on NK cells in blood (n = 8, 10, 9). m) Frequency of NK cells in spleen (n = 12, 11, 8). n) CD69 expression on NK cells in spleen (n = 7, 10, 8). o) Plasma levels of IL‐18 (n = 8, 8, 8). p) Plasma levels of resistin (n = 11, 11, 9). Data are presented as means ± SD. n indicates the number of biologically independent samples examined. Statistical analysis was assessed by one‐way ANOVA with Dunnett's multiple comparison b,e–j,m–o) or by Kruskal‐Wallis with Dunn's multiple comparison c,d,k,l,p), and significant differences were indicated with p values. Source data are provided as a Source Data File.

Figure 2

Uterine eosinophil and NK cell populations are markedly altered by androgen exposure in PCOS‐like mice. a) Frequency of eosinophils among immune cells in uterus (n = 11 Control, 11 DHT, 7 DHT + Flutamide). b) Representative plots of CD45⁺ immune cells, gated on live single cells. c) Frequency of NK cells in uterus (n = 12, 11, 7). d) CD69 expression on NK cells in uterus (n = 12, 11, 7). e) Frequency of macrophages in uterus (n = 11, 11, 7). f) MHC‐II expression on macrophages (n = 11, 11, 7). g) CD11c expression on macrophages (n = 11, 11, 7). h) Eotaxin (CCL11) levels in uterus (n = 10, 9, 7). i) IL‐5 levels in uterus (n = 10, 9, 7). j) IFN‐γ levels in uterus (n = 10, 9, 7). k) TNF‐α levels in uterus (n = 10, 9, 7). l) CCL2 levels in uterus (n = 10, 9, 7). m) IL‐18 levels in uterus (n = 10, 9, 7). Data are presented as means ± SD. n indicates the number of biologically independent samples examined. Statistical analysis was assessed by Kruskal‐Wallis with Dunn's multiple comparison a,g,d,m), one‐way ANOVA with Dunnett's multiple comparison c,e,f) or mixed‐effects ANOVA with Bonferroni's multiple comparison test h–l) and significant differences were indicated with p values. Source data are provided as a Source Data File.

Figure 3

Ovarian macrophage populations are decreased by androgen exposure in PCOS‐like mice. a) Frequency of macrophages in ovaries, expressed as percent of CD45⁺ immune cells (n = 9 Control, 8 DHT, 6 DHT + Flutamide). b) Frequency of NK cells in ovaries (n = 8, 10, 9). c) CD69 expression on NK cells in ovaries (n = 8, 10, 9). d) Frequency of CD4⁺ T helper cells in ovaries (n = 8, 10, 9). e) Frequency of CD8⁺ cytotoxic T cells in ovaries (n = 8, 10, 9). Data are presented as means ± SD. n indicates the number of biologically independent samples examined. Statistical analysis was assessed by one‐way ANOVA with Dunnett's multiple comparison a–c), or Kruskal‐Wallis with Dunn's multiple comparison d,e), and significant differences were indicated with p values. Source data are provided as a Source Data File.

Figure 4

The peripubertal DHT‐induced mouse model is a non‐obese but insulin‐resistant model of PCOS. a) Experimental design. b) Fat mass (n = 12 Control, 11 DHT, 9 DHT + Flutamide). c) Insulin levels at baseline and 15 min following glucose administration (n = 12, 12, 10). d) Blood glucose levels during oGTT (n = 12, 12, 10). e) HOMA‐IR, calculated from fasted glucose and insulin levels (n = 12, 12, 10). f) Glycosylated hemoglobin levels (HbA1c) (n = 34, 31, 21). Data are presented as means ± SD. n indicates the number of biologically independent samples examined. Statistical analysis was assessed by two‐way ANOVA with Bonferroni's multiple comparison test b–d,f), or Kruskal‐Wallis with Dunn's multiple comparison e) and significant differences were indicated with p values. Source data are provided as a Source Data File.

Figure 5

DHT‐exposed PCOS‐like mice display an aberrant immune profile in VAT albeit unaltered fat mass. a) Frequency of macrophages among immune cells in VAT (n = 8 Control, 10 DHT, 7 DHT + Flutamide). b) Representative plots of subpopulations of macrophages (CD45⁺CD11b⁺SSC^low/midF4/80⁺Ly6G^mid) based on the expression of CD11b and CD11c and frequency CD11b^highCD11c⁺. c) MHC‐II expression on CD11b^highCD11c⁺ macrophages (n = 8, 10, 7). d) MHC‐II expression on CD11b^midCD11c⁻ macrophages (n = 8, 10, 7). e) Frequency of eosinophils in VAT (n = 8, 10, 7). f) CD69 expression on NK cells in VAT (n = 10, 11, 8). g) Frequency of NK cells in VAT (n = 10, 11, 8). h) IL‐5 levels in VAT (n = 16, 14, 15). i) Eotaxin (CCL11) levels in VAT (n = 10, 9, 7). j) IL‐4 levels in VAT (n = 16, 14, 15). k) IL‐13 levels in VAT (n = 10, 9, 7). l) IFN‐γ levels in VAT (n = 16, 16, 15). m) TNF‐α levels in VAT (n = 16, 16, 15). n) IL‐18 levels in VAT (n = 9, 9, 8). o) Resistin levels in VAT (n = 11, 12, 10). p) IL‐10 levels in VAT (n = 16, 14, 15). q) IL‐2 levels in VAT (n = 16, 16, 15). Data are presented as means ± SD. n indicates the number of biologically independent samples examined. Statistical analysis was assessed by one‐way ANOVA with Dunnett's multiple comparison a–f, n,o), Kruskal‐Wallis with Dunn's multiple comparison g), or mixed‐effects ANOVA with Bonferroni's multiple comparison test h–m,p,q) and significant differences were indicated with p values. Source data are provided as a Source Data File.