Correlation Networks Provide New Insights into the Architecture of Testicular Steroid Pathways in Pigs

Abstract

Steroid metabolism is a fundamental process in the porcine testis to provide testosterone but also estrogens and androstenone, which are essential for the physiology of the boar. This study concerns boars at an early stage of puberty. Using a RT-qPCR approach, we showed that the transcriptional activities of several genes providing key enzymes involved in this metabolism (such as CYP11A1) are correlated. Surprisingly, HSD17B3, a key gene for testosterone production, was absent from this group. An additional weighted gene co-expression network analysis was performed on two large sets of mRNA-seq to identify co-expression modules. Of these modules, two containing either CYP11A1 or HSD17B3 were further analyzed. This comprehensive correlation meta-analysis identified a group of 85 genes with CYP11A1 as hub gene, but did not allow the characterization of a robust correlation network around HSD17B3. As the CYP11A1-group includes most of the genes involved in steroid synthesis pathways (including LHCGR encoding for the LH receptor), it may control the synthesis of most of the testicular steroids. The independent expression of HSD17B3 probably allows part of the production of testosterone to escape this control. This CYP11A1-group contained also INSL3 and AGT genes encoding a peptide hormone and an angiotensin peptide precursor, respectively.

Keywords: AKR1C; CYP11A1; HSD17B3; WGCNA; androstenone; co-expression network; estrogens; porcine testis; steroids synthesis; testosterone.

Figure A1

Landscape of genes involved in the pathway of sex steroids production in the porcine testis. The initial step to produce all sex steroids in the testes is the transformation of cholesterol in pregnenolone. In all mammals, the transformation of pregnenolone in androstenedione opens the common pathway to estrogens and androgens. Pregnenolone is also the starting product in the androstenone pathway. Epiandrosterone, 19-nor-testosterone, testosterone and androstenedione are androgens. 17β-estradiol and estrone are estrogens. Androstenone is a pheromone. The pathways of the biosynthesis of epiandrosterone and 19-nor-testosterone are not completely known.

Figure 1

Correlations detected by the analysis of RT-qPCR results. These RT-qPCRs were performed on 33 genes, but only correlations concerning 18 genes and with p < 0.01 are reported. The correlations between the testicular expression of genes and plasma concentrations of testosterone and 17β-estradiol are presented at the bottom of this diagram.

Figure 2

Part of HCA of mRNAseq-P (A) and from mRNAseq-PLW (B). Among more than 9000 genes, an HCA was built by retaining the top 100 genes with the highest variance (see Supplementary Figure S2). We selected for this figure the part of these HCA including some genes already identified as highly correlated in this study. A green arrow indicated the genes present in the group of genes correlated to CYP11A1 and characterized by RT-qPCR (see Figure 1).

Figure 3

Presentation of CYP11A1-cluster-P162. An HCA was built with data (mRNAseq-P) observed for the 469 genes included in WGCNA-CYP11A1. The analysis of this clustering allowed the characterization of a group of 162 genes linked closely to CYP11A1 (CYP11A1-cluster-P162). For the composition of sub-clusters, we have prioritized genes present in the group around CYP11A1 characterized by RT-qPCR (see Figure 1).

Figure 4

Presentation of CYP11A1-cluster-PLW89. An HCA was built with data (mRNAseq-PLW) observed for the 469 genes included in the correlation module previously defined by WGCNA analysis around CYP11A1. The analysis of this clustering allowed the selection of a group of 307 genes linked by correlations. In this figure, only the sub-cluster (89 genes) including CYP11A1 is shown. For the composition of sub-clusters, we have prioritized genes present in the group around CYP11A1 characterized by RT-qPCR (see Figure 1).

Figure 5

Analysis of genes likely to belong to CYP11A1-group. (A) The HCA analyses of HCAs led to defining two clusters around CYP11A1: CYP11A1-cluster-P162 and CYP11A1-cluster-PLW379 from WGCNA-CYP11A1. The comparison of the genes’ content of CYP11A1-cluster-P162 and CYP11A1-cluster-PLW379 allowed us to define CYP11A1-group_159 and CYP11A1-group_382 including 159 and 382 genes, respectively. (B) The comparison of genes’ contents of CYP11A1-group_159 and CYP11A1-group_382 and the two lists of genes highly correlated to CYP11A1 (CYP11A1/list-P68 and CYP11A1/list-PLW105) led to characterizing of 45 genes constituting the core of CYP11A1-group. We suggest considering only the 45 + (13 + 27) genes as the correlation group around CYP11A1.

Figure 6

Presentation of CYP11A1-group (85 genes). The 45 genes presented in the shaded box constitute the core group. Their expression are highly correlated to that of CYP11A1 in both datasets (mRNAseq-P and -PLW). The other 40 genes are strongly correlated to CYP11A1 in only one dataset. All genes cited in Appendix A are indicated in blue. Genes that are indicated with a grey-blue background are genes cited in this study. * In mRNAseq analysis, the RSEM pipeline attributed to this gene counts of all AKR1C genes that have different known specificities. Nevertheless, RT-qPCR results (Figure 1) showed that the gene involved in testicular steroid synthesis is AKR1C4. ** In mRNAseq analysis, the RSEM counts attributed to this gene include counts of the three CYP19A genes.

Figure 7

Characterization of the correlation group around HSD17B3. The analysis of HCAs led to defining HSD17B3-cluster-P14, HSD17B3-cluster-PLW63 and HSD17B3-cluster-PLW267 from WGCNA-HSD17B3. The list of genes strongly correlated to HSD17B3 in each mRNA-seq was established: HSD17B3/list-PLW388 and HSD17B3/list-P28. The comparison between these five elements was not favorable to the characterization of a large correlation group around HSD13B3.

Figure 8

Possible regulation network of genes from the CYP11A1-group. Transcription factors (in blue) that may contribute to the regulation of genes (in black) from the CYP11A1-group were identified using human orthologous genes. The core (highlighted in a blue-green background) of this network included four genes and three TFs.