Clinical characteristics of CPA patients with or without pregnancy showed in Table 1. No significant difference in age, BMI, SBP, DBP, potassium and tumor size was found between the non-pregnant and pregnant group (P > 0.05). The pregnant patients with CPA had higher serum cortisol, E2, Progesterone, and HCG concentration, while having lower serum FSH and LH concentrations (P < 0.01). Furthermore, compared with the non-pregnant patients, the duration is shorter, and the growth rate of the tumor is faster in pregnant patients with CPA (P < 0.05). The CT imaging of CPA was shown in Fig. 1.
Morphology is a critical and fundamental step in the diagnosis of adrenocortical adenoma. The histology results revealed that the adenoma cells were enlarged, and cytoplasmic lipids were increased, with lipid-rich foamy cytoplasm (Fig. 2A). Ki-67 staining was used to assess tumor cell proliferation activity. It showed that the percentage of Ki-67 positive cells are higher in the pregnant group than the non-pregnant group (8% vs 5.5%, P < 0.05) (Fig. 2B, C).
Comparative analysis of the genome-wide DNA methylation status between the adjacent tissue and adenoma tissue of the non-pregnant patients with CPA
To determine whether our present study was consistent with previous research in DNA methylation of CPA. Global DNA methylation was measured by Illumina's Infinium human methylation 850 K BeadChip. The differential methylation positions (DMPs) between adjacent and adenoma tissues were shown on the volcano plot (Δβ ≤ − 0.14 or Δβ ≥ 0.14, P < 0.05). Our results found that 10,053 DMP sites differed between adenomas and adjacent tissues, containing 2229 hypermethylated and 7824 hypomethylated sites in non-pregnant patients with CPA (Fig. 3A). A heat map was used to show the gene methylation level signature (Fig. 3B). According to previous studies, we found five hypermethylated genes and twenty-four hypomethylated genes differed between adenomas and adjacent tissues in our results that had been reported previously and we listed the studies in Additional file 1.
To identify the genome-wide DMPs difference between adenomas and adjacent tissues of non-pregnant adrenocortical adenomas groups, the GO and KEGG pathway enrichment was analyzed. For GO term analysis, we chose 7824 DMPs (containing 2925 genes) with hypomethylation in the adenoma tissue than the adjacent tissue. We just show the date that involved in molecular function. We found that adenoma tissues had the same hypo-methylated driven genes involved in adrenocortical adenoma development signaling, such as calcium ion binding, 3′,5′-cyclic-AMP phosphodiesterase activity, SH3/SH2 adaptor activity (Fig. 3C and Additional file 2). To determine the potential function differences between the two groups, KEGG pathways were performed on the hypo-methylated genes in adenoma tissues, which showed that Calcium signaling pathway, Cell adhesion molecules (CAMs), cAMP signaling pathway, MAPK signaling pathway, PI3K-Akt signaling pathway, WNT signaling and positive regulation of antigen receptor-mediated signaling pathway (Fig. 3D and Additional file 3). These pathways play important roles in the development of adrenocortical adenomas.
We then used the STRING-database software and Cytoscape software to analyze the top 1800 genes. It displays the number of nodes at 1203 and the number of edges showing 5807, then we analyze the functional enrichment in the network. It was considered that the top 20 genes with high degree of connectivity as the hub genes of PTPRC, SHH, CAMK2A (calcium ion binding), NOTCH1, ANXA1(transcription factor activity), NRXN1 (receptor activity), ITGB2(cell adhesion molecule binding), GNG7, GNG4 (signal transducer activity), FYN (phosphatidylinositol-4,5-bisphosphate 3-kinase activity), GRIN2B (extracellular-glutamate-gated ion channel activity), TYROBP (receptor binding), IGF1 (growth factor activity), TLR8V (drug binding), ITGB1 (cell adhesion molecule binding), CDH2 (beta-catenin binding), CCND1 (transcription factor binding), SYK (protein tyrosine kinase activity), PAX6 (HMG box domain binding), ADCY2 (phosphorus-oxygen lyase activity) in term of molecular function(Fig. 3E and Additional file 4).
The difference of genome-wide DNA methylation profiles between pregnant and non-pregnant patients with CPA
To identify the genome-wide DNA methylation difference between pregnant and non-pregnant groups, hierarchical clustering was used to analysis the methylation patterns of DMPs. The top 5000 DMPs differences (p < 0.05) were chosen to draw the heatmap and scatter plot figures. The results showed that the two groups have distinct methylation patterns. The pregnant group contained higher hypermethylated DMPs (5043 vs 2229 sites) and lower hypomethylated (2477 vs 7824 sites) (Fig. 4A–D). We also extracted the relevant information from the genome-wide DNA methylation sites and examined the differences between the two groups of 5 kb upstream and downstream of the transcription start site (TSS). The TSS upstream and downstream methylation distribution results visually showed that the chromosomal hyper-methylation sites regions of the pregnant (Fig. 4F) are higher than those of non- pregnant group (Fig. 4E). To further investigate the chromosomal distribution of the DNA methylation status corresponding to the 5000 DMPs, we then determined the proportion of DMPs sites that were hypermethylated and hypomethylated, respectively, on pregnant and non-pregnant chromosomes (Fig. 4G). The results also show that the pregnant groups had more hypermethylated DMPs (67.94% vs 22.16%) and less hypomethylated DMPs (32.93% vs 77.84%). We calculated the proportion of DNA methylation status on each chromosome, the proportion of hypermethylated DMPs was higher on chromosomes 1 and X but lower on chromosome 2 (Fig. 4H).
Identifying the promoter DNA methylation region signature of CPA during pregnancy
The DNA promoter region is important for gene expression regulation. Hypomethylation is typically used as a positive gene expression regulator. We compared the DNA promoter region methylation status between the two groups (Fig. 5A, B). In general, the pregnant group had a lower DNA promoter region methylation percentage than non-pregnant group (33.25% vs 66.48%). Similar results with the genome-wide DNA methylation pattern, the DNA promoter region indicated that there were 1978 methylated probes in the pregnant group, among them 576 DMPs were completely hypomethylated and 1109 DMPs were completely hypermethylated (Fig. 5C).
The GO and KEGG pathway analysis were used to investigate the DNA promoter region epigenetic status of the pregnant group. The GO enrichment analysis showed that hypomethylated 488 genes and most genes were significantly enriched in tissue development, stem cell development, cell fate commitment, cell–cell adhesion, cell proliferation, cell development, stem cell proliferation, and cell–cell signaling (Fig. 5D and Additional file 5). We also conducted the KEGG Enrichment analysis on 488 genes. We discovered some genes enriched linked to tumor proliferation signaling pathway, including 13 genes that were associated with Wnt signaling pathway, 10 genes with Ras/MAPK signaling pathway, 4 genes with PI3K-Akt signaling pathway,2 genes with p53 signaling pathway, 2 genes with VEGF signaling pathway and 4 genes with Jak-STAT signaling pathway (Fig. 5E and Additional file 6).
We then used STRING-database and Cytoscape software to analyze the 488 genes. It displays 432 nodes and shows 91 edges, then we analyzed the functional enrichment in the network, focusing on the key proteins that regulate cell population proliferation, cell and cell adhesion, cell death, and mitotic cell cycle. It was considered that those top genes with high degree of connectivity as the hub genes of ACC: SOX2 (SRY (sex determining region Y)-box 2), PAX6 (Paired Box 6), POMC (Pro-opiomelanocortin), MMP9 (Matrix metallopeptidase 9), IGF2 (Insulin Like Growth Factor 2), AR (Androgen Receptor), ADCY2 (Adenylate Cyclase 2), HTR5A (5-Hydroxytryptamine Receptor 5A), NGF (Nerve growth factor), CASR (Calcium Sensing Receptor), BDKRB2 (Bradykinin Receptor B2), OPRK1 (Opioid Receptor Kappa 1), AKT3 (AKT Serine/Threonine Kinase 3), SGK1 (Serum/Glucocorticoid Regulated Kinase 1), TBX3 (T-Box Transcription Factor 3) (Fig. 5F and Additional file 7).
Identification and validation of the promoter region DNA methylation marker
The above selected four genes have been described as potential oncogenes in hypopharyngeal carcinoma , breast cancer , osteosarcoma , and adrenocortical adenomas . Bisulfite pyrosequencing was used to verify their promoter methylation profiles. The promoter region DNA average methylation was TBX3 (28.20% versus 20.61%), AKT3 (38.90% versus 28.90%), SOX2 (18.38% versus 16.14%), and SGK1 (21.33% versus 12.33%) in the two groups (Fig. 6A, D, G, J).
To determine whether the promoter hypomethylation was linked to the higher protein expression of the above four genes. we performed immunohistochemistry to examine the degree of protein expression and found that hypomethylation of the above four genes were linked to a higher protein staining level (Fig. 6B, E, H, K), Statistical analysis also showed that the pregnant group had higher levels of TBX3, AKT3, SOX2, and SGK1 than non-pregnant adenomas group, with a significant difference (P < 0.05 or P < 0.01) (Fig. 6C, F, I, L).