Lower female survival from an opportunistic infection reveals progesterone-driven sex bias in trained immunity

Abstract

Immune responses differ between females and males, although such sex-based variance is incompletely understood. Observing that bacteremia of the opportunistic pathogen Burkholderia gladioli caused many more deaths of female than male mice bearing genetic deficiencies in adaptive immunity, we determined that this was associated with sex bias in the innate immune memory response called trained immunity. Female attenuation of trained immunity varies with estrous cycle stage and correlates with serum progesterone, a hormone that decreases glycolytic capacity and recall cytokine secretion induced by antigen non-specific stimuli. Progesterone receptor antagonism rescues female trained immune responses and survival from controlled B. gladioli infection to magnitudes similar to those of males. These data demonstrate progesterone-dependent sex bias in trained immunity where attenuation of female responses is associated with survival outcomes from opportunistic infection.

Keywords: Burkholderia gladioli; CP: Immunology; glycolytic capacity; innate immune memory; oppotunistic infection; progesterone; progesterone receptor; sex as a biological variable; sex bias; trained immunity.

Figure 1.. Trained immunity imparts an enhanced survival advantage to male mice infected with B. gladioli:

(A) Experimental set-up. Female (F) and male (M) mice of Rag1−/− genotype (RAG1 KO) were injected retro-orbitally with either PBS (no training) or heat-killed C. albicans (exogenous training). After 1 week, mice were injected with live B. gladioli bacteria. Created with BioRender. (B) Survival curves (left) comparing control-untrained female vs. control-untrained male Rag1−/− mice after B. gladioli infection and comparison of mean survival days (right). (C) Survival curves (left) comparing exog-trained female vs. exog-trained male Rag1^−/− mice after B. gladioli infection, and comparison of mean survival days (right). (D) Survival mean change (increase) in days provided by immune training as estimated by assessing the survival difference at matched ordered ranks as described in Methods. Survival curves were analyzed by Kaplan-Meier two-sided log-rank statistics with p values obtained as described in Methods. Log rank p value 0.0238 has standard deviation of 1.53 × 10⁻⁴, and log rank p value of 0.0158 has standard deviation of 1.91 × 10⁻⁴. Mean survival (days) and survival mean change were compared with two-sided Mann-Whitney U test of means. Each dot represents a mouse, n = 5 for each group, with mean ± SD shown. p values < 0.05 are significant. See also Figures S1 and S2.

Figure 2.. Training is attenuated in mouse BM-MNCs when exposed to female serum:

(A) Tissue culture experimental timeline. Created with BioRender. (B) Supernatant levels of IL-6 and CXCL2 after LPS or poly(I:C) restimulation for 24 h. Cells used are female and male Rag1^−/− BM-MNCs that had been β-glucan-trained with sex-matched serum. (C) IL-6 and CXCL2 levels after restimulation of female and male cells trained with opposite sex serum. (D) Comparison of IL-6 and CXCL2 levels from female cells trained with matched versus opposite sex serum. (E) Comparison of IL-6 and CXCL2 levels from male cells trained with matched versus opposite sex serum. Means were compared with two-sided Mann-Whitney U test with Holm-Šidák correction for multiple hypothesis testing. Each dot plotted represents a mouse, n = 5 for each group with mean ± SD shown. p values <0.05 are significant. See also Figures S2 and S3.

Figure 3.. Serum from mouse estrous cycle stage influences attenuation of training of BM-MNCs:

(A) E2 (top) and P4 (bottom) concentrations in mouse serum measured from each stage of the estrous cycle, compared against male serum levels. (B) Concentrations of IL-6 (left) and CXCL2 (right) released in supernatant from female Rag1^−/− BM-MNCs after LPS restimulation, when cells had been β-glucan-trained in serum from each stage of the estrous cycle, compared with levels from male cells trained in male serum. (C) IL-6 (left) and CXCL2 (right) concentrations in supernatant after poly(I:C) restimulation from cells trained as in (B). Data were analyzed using Kruskal-Wallis one-way ANOVA with Dunn’s multiple comparisons tests. Each dot plotted represents a mouse, n = 4 per group in (A), and n = 5 per each group (B, C) with mean ± SD shown. p values <0.05 are significant.

Figure 4.. Attenuation of trained immunity is controlled by P4 and the nuclear PR:

(A) After LPS restimulation, supernatant levels of IL-6 (left) and CXCL2 (right) from female Rag1^−/− BM-MNCs that had been β-glucan-trained in female serum plus vehicle control (peach) or ER antagonist, PR antagonist, or both antagonists, compared with male cells trained in male serum (blue). (B) Similar to (A) after poly(I:C) restimulation. (C) After LPS restimulation, IL-6 (left) and CXCL2 (right) supernatant levels from female (n = 5) and male (n = 5) Rag1^−/− BM-MNCs trained in csFBS media with vehicle or P4 alone, PR antagonist alone, or P4 + PR antagonist combined. (D) Similar to (C) after poly(I:C) restimulation. (E) After LPS restimulation, IL-6 (left) and CXCL2 (right) supernatant levels from male BM-MNCs trained in male serum with vehicle or PR antagonist. (F) Similar to (E) after poly(I:C) restimulation. (G) Extracellular acidification rate (ECAR) of female Rag1^−/− BM-MNCs that had been trained with β-glucan plus vehicle control, or plus P4 (P4, 20 ng/mL) alone or combined with PR antagonist (PR ant), compared against untrained cells (left), and analyzed for basal glycolysis (middle) and maximum glycolytic capacity (right). (H) Survival curves (left) comparing Rag1^−/− females exogenously trained in the presence of vehicle control versus PR antagonist before B. gladioli infection, and comparison of mean survival days (right). In vitro cytokine data were analyzed using Kruskal-Wallis one-way ANOVA with Dunn’s tests post hoc. Glycolysis was assessed using parametric one-way ANOVA. Survival curves were analyzed by Kaplan-Meier one-sided log-rank statistics with p values obtained as described in Methods. Log-rank p value 0.0356 has standard deviation of 2.18 × 10⁻⁴. Mean survival (days) was compared using a one-sided Mann-Whitney U test of means. Each dot represents a mouse, n = 5 per sex per group, except for glycolysis data, where n = 3 mice per group with triplicate measurements at multiple time points with mean ± SD shown. p values < 0.05 are significant. See also Figure S4.