The FOS/AP-1 Regulates Metabolic Changes and Cholesterol Synthesis in Human Periovulatory Granulosa Cells

Abstract

FOS, a subunit of the activator protein-1 (AP-1) transcription factor, has been implicated in various cellular changes. In the human ovary, the expression of FOS and its heterodimeric binding partners JUN, JUNB, and JUND increases in periovulatory follicles. However, the specific role of the FOS/AP-1 remains elusive. The present study determined the regulatory mechanisms driving the expression of FOS and its partners and functions of FOS using primary human granulosa/lutein cells (hGLCs). Human chorionic gonadotropin (hCG) induced a biphasic increase in the expression of FOS, peaking at 1 to 3 hours and 12 hours. The levels of JUN proteins were also increased by hCG, with varying expression patterns. Coimmunoprecipitation analyses revealed that FOS is present as heterodimers with all JUN proteins. hCG immediately activated protein kinase A and p42/44MAPK signaling pathways, and inhibitors for these pathways abolished hCG-induced increases in the levels of FOS, JUN, and JUNB. To identify the genes regulated by FOS, high-throughput RNA sequencing was performed using hGLC treated with hCG ± T-5224 (FOS inhibitor). Sequencing data analysis revealed that FOS inhibition affects the expression of numerous genes, including a cluster of genes involved in the periovulatory process such as matrix remodeling, prostaglandin synthesis, glycolysis, and cholesterol biosynthesis. Quantitative PCR analysis verified hCG-induced, T-5224-regulated expression of a selection of genes involved in these processes. Consistently, hCG-induced increases in metabolic activities and cholesterol levels were suppressed by T-5224. This study unveiled potential downstream target genes of and a role for the FOS/AP-1 complex in metabolic changes and cholesterol biosynthesis in granulosa/lutein cells of human periovulatory follicles.

Keywords: FOS; JUN proteins; granulosa cells; human; ovulation.

Figure 1.

The effect of hCG on the expression of FOS and Jun family members. Human granulosa/lutein cells (hGLCs) obtained from women undergoing a standardized IVF procedure were acclimated for 6 days and then cultured without (control) or with hCG (1 IU/mL) for 0, 1, 3, 6, 12, or 24 hours. (A) The levels of mRNA for FOS and Jun family members were measured by qPCR and normalized to the levels of RNA18S5 in each sample (n ≥ 4 independent experiments). (B) Representative Western blot images detecting FOS, JUN, JUNB, and JUND proteins (intact and phosphorylated forms). The levels of ACTB in each lane were used as a loading control. The band intensities for each protein were measured by ImageJ and normalized to the intensity of ACTB in the corresponding sample. The experiments were repeated at least three times with independent samples. (C) Nuclear extracts were immunoprecipitated with a phosphorylated FOS antibody (3 µg/mL) or normal rabbit IgG (3 µg/mL), and protein-antibody complexes were separately detected by Western blot analyses using antibodies for JUN, JUNB, or JUND. Bars with no common superscripts within each treatment group and (*) between treatments are significantly different (P < 0.05). **P = 0.052.

Figure 2.

The effect of hCG on intracellular signaling pathways. Primary hGLCs were cultured with or without hCG (1 IU/mL) for 0, 0.5, 1, 2, or 3 hours. (A) Western blot analyses were performed to measure the levels of protein for phosphorylated (p)CREB, pERK1/2, pAKT, and p-p38MAPK. The levels of ACTB protein were used as a loading control. (B) The band intensities for each protein were measured by ImageJ and normalized to the intensity of ACTB in the corresponding sample. The experiments were repeated at least 3 times with independent samples. Bars with no common superscripts within each treatment group and (*) between treatments are significantly different (P < 0.05). **P = 0.084; ***P = 0.07.

Figure 3.

The effect of blocking either the PKA or the MAPK signaling pathway on the expression of protein for FOS and Jun family members. Primary hGLCs were cultured without or with (A) hCG (1 IU/mL) ± H89 (an inhibitor of the PKA signaling pathway) or (B) hCG (1 IU/mL) ± SCH772984 (an inhibitor of the MAPK signaling pathway) for 1 hour. Representative Western blot images showed phosphorylated (p)CREB, pERK1/2, FOS, JUN, and JUNB proteins. ACTB in each lane was used as a loading control. Densitometric analyses were performed by ImageJ to measure the band intensities for each protein. Protein band intensities were normalized to the intensity of ACTB in the corresponding sample. The experiments were repeated at least 3 times with independent samples. Bars with no common superscripts are significantly different (P < 0.05).

Figure 4.

RNA-seq analyses using hGLCs treated with hCG or hCG+T-5224. High-throughput RNA-seq analysis was performed using hGLCs obtained from 4 patients (the numbers 1-4 represent individual patient). The cells were cultured with hCG (I IU/mL, H) or hCG+T-5224 (20 μM, TH) for 12 hours. (A) Principal component analysis (PCA) clustering was conducted after normalization to determine whether there are overall treatment differences or batch effects. (B) Mean-difference (MD) plot displays gene-wise log2-fold changes (log_FCs) against average expression values together with a plot of sample expression. Dots in red, gray, and blue mean genes that show increases, no changes, or decreases in expression values by the addition of T-5224, respectively. Arrows point to an array of genes that are increased or decreased by T-5224. (C) Heatmap indicates the group difference of differentially expressed genes (DEGs) between H- and TH-treated hGLCs and the repeatability within each group. Thresholds used in MD plot and heatmap were FDR < 0.01 and |log_FC| > 1 to select DEGs in between groups.

Figure 5.

Validation of RNA-seq data by qPCR analysis. Most highly downregulated genes (9 genes) were selected from the list of DEGs by hCG+T-5224 treatment compared with hCG in hGLCs. The cells were cultured without (control) or with hCG (1 IU/mL) ± T-5224 (20 µM) for 12 hours. qPCR analysis was performed to measure the levels of mRNA for 9 selected DEGs. The levels of mRNA for each gene were normalized to the levels of RNA18S5 in each sample (n = 5-6 independent batches). Bars with no common superscripts are significantly different (P < 0.05).

Figure 6.

The effect of T-5224 on the expression of genes involved in glycolysis and cholesterol de novo biosynthesis. (A) A schematic diagram depicts selected DEGs involved in glycolysis (SLC2A1, PFKFB3, PKFKB4, and LDHA) and cholesterol de novo biosynthesis (ACSS1, ACSS2, HMGCS1, MVK, MVD, FDPS, SQLE, LSS, CYP51A1, EBP, and DHCR24). (B) The levels of 10 selected DEGs depicted were measured by qPCR analysis and normalized to the levels of RNA18S5 in each sample (n = 5-6 independent batches). Bars with no common superscripts are significantly different (P < 0.05).

Figure 7.

The effect of T-5224 on granulosa cell metabolic changes. Primary hGLCs were cultured without (control) or with hCG (1 IU/mL) ± T-5224 (20 µM) for 24 hours. (A) Cell metabolic activity was assessed by recording the absorbance at 490 nm using MTS assays. The experiments were repeated in 10 independent batches which had 4 replicates per batch. OD, optical density. (B) Seahorse XF glycolysis stress tests were performed to measure glycolytic function in cells by directly measuring the extracellular acidification rate (ECAR). The experiments were repeated in 4 independent batches with at least 4 replicates per batches. (C) Cholesterol levels in hGLC lysates were measured by cholesterol fluorescence assay. The fluorescence was read using excitation wavelength at 535 nm and emission wavelength at 590 nm. The experiments were repeated in 4 independent batches. (D) The concentration of progesterone was measured in condition media. The experiments were repeated in 4 independent patient samples. (E) Numbers of cells per treatment were assessed by in situ counting nuclei fluorescently labeled with Hoechst 33 342 in 4 independent patient samples, each with 4 replicates. Bars with no common superscripts were significantly different (P < 0.05).