LXR pathway drives hormonal response intensity in polycystic ovary syndrome

Abstract

Gonadotropin injections used to stimulate oocyte production during assisted reproductive technology (ART) procedures are associated with the risk of an abnormal response in predisposed patients suffering polycystic ovary syndrome (PCOS). Liver X receptors (LXR) pathway has been identified as key regulators during this process. This study explores the integration of the hormonal signals, cellular networks and molecular mechanisms linking sterol signaling with inflammation and immune infiltration. Pharmacological activation of LXR in a wild-type context protects against gonadotropin hyperstimulation mirroring the effect observed in LXR-deficient mice. Ovarian stimulation leads to immune cell infiltration orchestrated by granulosa cells in absence of LXR, resulting in an altered granulosa cell response to gonadotropin and enhanced inflammation. LXR controls inflammasome activity by regulating Thioredoxin Interacting Protein (TXNIP) gene expression in mural granulosa cells, thereby modulating IL1β production. This immune cell infiltration persists throughout ovulation in PCOS patients and is observed in cumulus oocytes complexes, highlighting the pivotal role of LXR path in regulating inflammatory processes during hormonal stimulation in ART procedures.

Keywords: Granulosa Cells; Hormonal Stimulation; Inflammasome; Liver X Receptors; Polycystic Ovary Syndrome.

Figure 1. LXRs activation prevents ovary from gonadotrophins hyperstimulation.

(A) Stimulation protocols: mice receive a single PMSG IP injection (7.5IU) and hCG IP injection (5IU) 46 h later (Stim) n = 14, or two IP injection (20IU) at hour 0 and 24 following a single hCG IP injection (10IU) (Hyperstimulation ++) n = 18 and a similar protocol with additionnal GW3965 (20 µg/mL) treatments 24 h before starting the protocol and together with following PMSG/hCG IP injection (Hyperstimulation ++ GW3965) n = 22. (B) Macroscopic representative pictures (upper panel) and HE-staining (bottom panel) of mouse ovaries after 96 post-hormonal protocols. Arrows indicated hemorrhagic cysts (Scale bars = 1 mm). (C) Hemorrhagic cysts have been counted in each group. (D) Macroscopic representative pictures (upper panel) and HE-staining (bottom panel) of wild type, LXR DKO, and TG-AMH-Lxrβ mouse ovaries following hormonal stimulation. Arrows indicated hemorrhagic cysts (Scale bars = 1 mm). (E) Expression of granulosa-specific markers: Cyp19a1, Fshr, and Inha were analyzed by RTqPCR from wild type n = 6, LXR DKO n = 6, and TG-AMH-Lxrβ n = 6 ovaries. Gene expressions were normalized using 36b4 gene expression. (F) Total number of oocytes retrieved in oviduct after ovulation following stimulation protocol from wild type n = 13, LXR DKO n = 8 and TG-AMH-Lxrβ n = 7 mice. In (E), boxes extend from the 25th to 75th percentile, the middle line shows the median, whiskers extend to the most extreme data. In (C) and (F), averages values ± SD are represented. Significance determined in (C) and (F) by Ordinary one-way ANOVA and in (E) by Mann and Whitney test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 (exact P values for these statistical comparisons are shown in Appendix Table S1). Source data are available online for this figure.

Figure 2. LXRs ablation is associated with immune cell infiltration within the ovary.

(A) Kinetic chart analysis of phenotype occurrence. (B) Hemorrhagic cysts quantification 40 h, 50 h, and 66 h post-PMSG injection observed in wild type, LXR DKO, and TG-AMH-Lxrβ ovaries (n for each comparison are shown in Appendix Table S1). (C) Principal component analysis of RNAseq dataset 40 h post-PMSG stimulation. (D) Venn diagram comparing wild type versus LXR DKO and LXR DKO versus TG-AMH-Lxrβ. (E) Gene ontology analysis using Cluster Profiler obtained from wild type versus LXR DKO (upper panel) and LXR DKO versus TG-AMH-Lxrβ (bottom panel) comparisons. (F) Immunodetection of CD45 (pan-immune cell marker) in green performed on wild type, LXR DKO, and TG-AMH-Lxrβ ovaries (Scale bars = 100 µm). (G) Quantification of CD45 staining performed on wild type n = 10, LXR DKO n = 10 and TG-AMH-Lxrβ n = 10 ovaries. (H) Analysis of infiltrated cell composition using CIBERSORTx platform conducted on RNAseq data. (I, J) Protocol for bone marrow transplantation following hormonal ovarian stimulation. After 1 month of bone marrow transplant, circulating B lymphocytes were analyzed 48 prior hormonal stimulation using C57BL6J, NSG and transplanted-NSG for each genotype donor. (K, L) Immunodetection of CD45 and quantification staining. C57BL/6 donor-specific MHC-II staining. NSG-Wild type n = 12, NSG-LXR DKO n = 14, NSG-TG-AMH-Lxrβ n = 13 and NSG n = 8 mice have been used. White arrows indicate infiltrated cells (Scale bars = 100 µm). Averages values ± SD are represented. Significance determined in (B) by Ordinary one-way ANOVA, in (G) by Mann and Whitney test and in (L) by Kolmogorov–Smirnov test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 (exact P values for these statistical comparisons are shown in Appendix Table S1). Source data are available online for this figure.

Figure 3. LXR DKO ovaries showed an alteration of NF-κB signaling response.

(A) GSEA comparisons of wild type versus LXR DKO and LXR DKO versus TG-AMH-Lxrβ. Categories in red indicate NF-κB pathway deregulations. (B) Heatmap of NF-κB targets genes (Boston University list) using 40 h post-PMSG dataset. (C) Nfkbia, Irf1, Csf1, and Spi1 gene expression analyzed by RT-qPCR on wild type n = 7, LXR DKO n = 7, and TG-AMH-Lxrβ n = 5 ovaries. (D) HE-staining (upper panel) and p65 immunodetection in red (middle panel), magnification of granulosa compartment with p65 staining using fire scale and nucleus in green (bottom panel). Wild type and TG-AMH-Lxrβ ovaries showed a cytoplasmic staining compared to LXR DKO that exhibit a nucleus translocation of p65 in yellow (white arrows) (Scale bars = 100 µm). In (C), boxes extend from the 25th to 75th percentile, the middle line shows the median, whiskers extend to the most extreme data. In (C), averages values ± SD are represented. Significance determined in (C) by Mann and Whitney test. *P < 0.05, **P < 0.01 (exact P values for these statistical comparisons are shown in Appendix Table S1). Source data are available online for this figure.

Figure 4. LXR DKO mice reveals an impaired response to PMSG stimulation specific to Granulosa cells.

(A) Principal component analysis from granulosa cell collected on immature mice before and after PMSG stimulation for 48 h (Madogwe et al, 2020). (B) GSEA from PMSG-responders granulosa specific gene list using wild type versus LXR DKO (left panel) and TG-AMH-Lxrβ versus LXR DKO (right panel) comparison. (C) Venn diagram identifying leading-edge genes deregulated in both comparisons. (D) Heatmap of the 248 expression gene profiles identified as PMSG-responders of granulosa cells from wild type, LXR DKO, and TG-AMH-Lxrβ mouse ovaries. (E) Gene expression profiles of Txnip, Ezh1, and Ghr (Cluster 1) and Rnf128, Shisha6, and Ghitm (Cluster 2) in granulosa cells PMSG-stimulated 0 h n = 3 vs. PMSG-stimulated 48 h n = 3 compared to 40 h post-PMSG induced wild type n = 3, LXR DKO n = 3 and TG-AMH-Lxrβ n = 3 ovaries dataset.

Figure 5. Txnip expression is under the control of both PMSG and LXR activity.

(A) Txnip expression analysis by RT-qPCR in wild type n = 18, LXR DKO n = 14, and TG-AMH-Lxrβ n = 15 mouse ovaries. (B) Immunodetection of TXNIP in wild type, LXR DKO, and TG-AMH-Lxrβ ovaries in red (upper panel, scale bar = 1 mm; middle panel, scale bar = 100 µm), high magnification reveals that TXNIP is accumulated is the entire mural granulosa layer in LXR DKO compared to both wild type and TG-AMH-Lxrβ that harbor a centrifuge expression gradient pattern (black brackets: TXNIP intensity signal vs green brackets: granulosa layer). TXNIP detection did not harbor any difference in cumulus granulosa cells whatever the genotype (scale bar = 100 µm). (C) Gene expression analysis of TXNIP and CYP19A1 in KGN human granulosa cell line in response to DMSO n = 6 or forskolin n = 6 treatment. (D) Txnip expression in primary granulosa cell cultures from wild type, LXR DKO, and TG-AMH-Lxrβ mouse ovaries in response to forskolin treatment (n for each comparison are shown in Appendix Table S1). (E) Single-cell RNAseq box plots of both TXNIP and NR1H2 expression from granulosa cell originated from follicle at various stage of maturation. In (A), (C), and (D), boxes extend from the 25th to 75th percentile, the middle line shows the median, whiskers extend to the most extreme data. In (A), (C) and (D), averages values ± SD are represented. Significance determined in (A), (C), and (D), by Mann and Whitney test. *P < 0.05, **P < 0.01 (exact P values for these statistical comparisons are shown in Appendix Table S1). Source data are available online for this figure.

Figure 6. Impaired inflammasome activity triggers hemorrhagic phenotype observed in LXR DKO.

(A) Heatmap of Nlrp3, Txnip, Pycard (Asc), Il1b, and Casp1 using 40 h post-PMSG dataset. (B) Nlrp3, Pycard (Asc), and Il1b expression analysis by RT-qPCR in wild type n = 17, LXR DKO n = 16, and TG-AMH-Lxrβ n = 15 mouse ovaries. (C) TXNIP, NLRP3, ASC, IL1B protein accumulation in wild type, LXR DKO, and TG-AMH-Lxrβ mouse ovaries. ACTIN was used as a loading control. (D) Correlation plots between NLRP3, ASC, and IL1B using RT-qPCR data from in wild type, LXR DKO, and TG-AMH-Lxrβ mouse ovaries. (E) Protocol for hormonal stimulation as described Fig. 1A with n = 19 or without MCC950 n = 20 co-treatment, an inflammasome inhibitor. MCC950 was injected every 24 h, by four injections starting 24 h prior first PMSG injection (7.5 UI) until hCG final injection (5IU). Hemorrhagic cysts have been quantify comparing DMSO, as a control, and MCC950 treatment. (F) HE-staining (left panel) and IL1B immunodetection (right panel) were performed using LXR DKO ovary (Scale bar = 1 mm). Magnification indicated IL1B staining, using fire scale, and nucleus in green. Strong staining has been observed (white arrows, white asterisk) both surrounding immune cells (ROI 1) as well as cystic follicle in the antrum (ROI 3) compared to mural granulosa cells of non-cystic follicle (ROI 2) (Scale bar = 100 µm. In (B), boxes extend from the 25th to 75th percentile, the middle line shows the median, whiskers extend to the most extreme data. In (E), averages values ± SD are represented. Significance determined in (B) and (E), by Mann and Whitney test. *P < 0.05, **P < 0.01 (exact P values for these statistical comparisons are shown in Appendix Table S1). Source data are available online for this figure.

Figure 7. Inflammatory signature in cumulus oophorus reveals an immune infiltration both in human and mouse models in a context of PCOS.

(A) Granulosa cells were collected after COC hyaluronidase treatment and process for subsequent RNAseq library and sequencing. (B) GSEA comparisons of human cohort PCOS versus Control as well as mouse models, LXR DKO versus wild type and LXR DKO versus TG-AMH-Lxrβ. Categories are red indicated upregulated and is blue downregulated pathways. Bold pathway is associated with immune signature. (C) Venn diagram comparing common leading-edge lists of genes between human and mouse datasets, this analysis leads to identify 96 DEGs that composed the HPS signature. (D) Heatmap corresponding to the HPS signature. (E) HPS signature significance calculation using GSVA between human cohort PCOS n = 5 versus Control n = 11 samples. (F) Gene ontology analysis using Cluster Profiler obtained from human cohort PCOS versus Control regarding the HPS signature. (G) Human adult ovary single-cell RNAseq analysis. CXCR4, PTRC, ITGB2, CYBB, PIK3R5, TNFRSF1B, FPR1, IL18RAP, and HCLS1 are plotted using tSNE clustering. Immune cell clusters have been indicated using red dashed lines. (H) CD45 immunodetection of immune cells in the cumulus-oocyte complex from wild type, LXR DKO, and TG-AMH-Lxrβ (white arrows, scale bar = 100 µm). (I) HPS signature analysis on single cell cluster of immune cells in a murine model of PCOS (Luo et al, 2024). In (E) and (I), average values ± SD are represented. Significance determined in (E), by Mann and Whitney test. *P < 0.05, ***P < 0.001, ****P < 0.0001 (exact P values for these statistical comparisons are shown in Appendix Table S1).  Source data are available online for this figure.