Genetic regulators of mineral amount in Nelore cattle muscle predicted by a new co-expression and regulatory impact factor approach

Abstract

Mineral contents in bovine muscle can affect meat quality, growth, health, and reproductive traits. To better understand the genetic basis of this phenotype in Nelore (Bos indicus) cattle, we analysed genome-wide mRNA and miRNA expression data from 114 muscle samples. The analysis implemented a new application for two complementary algorithms: the partial correlation and information theory (PCIT) and the regulatory impact factor (RIF), in which we included the estimated genomic breeding values (GEBVs) for the phenotypes additionally to the expression levels, originally proposed for these methods. We used PCIT to determine putative regulatory relationships based on significant associations between gene expression and GEBVs for each mineral amount. Then, RIF was adopted to determine the regulatory impact of genes and miRNAs expression over the GEBVs for the mineral amounts. We also investigated over-represented pathways, as well as pieces of evidences from previous studies carried in the same population and in the literature, to determine regulatory genes for the mineral amounts. For example, NOX1 expression level was positively correlated to Zinc and has been described as Zinc-regulated in humans. Based on our approach, we were able to identify genes, miRNAs and pathways not yet described as underlying mineral amount. The results support the hypothesis that extracellular matrix interactions are the core regulator of mineral amount in muscle cells. Putative regulators described here add information to this hypothesis, expanding the knowledge on molecular relationships between gene expression and minerals.

Figure 1

Correlation network among genes and miRNAs with expression values correlated to at least one mineral. This network shows all the significant correlations among genes and miRNAs in the PCIT general and PCIT miRNA analysis. (A) Complete network, (B) Network with just the correlations regarding the genes and miRNAs with expression values correlated to more than one mineral. It is the internal circle of the complete network with more details, (C) Correlations among the mineral’s GEBVs.

Figure 2

Representation of the contrasting samples considering the genomic estimated breeding values of all 10 minerals together, based on the PCA score. Orange circles represent the samples with the highest scores (positive contrast) and the green circles represent the samples with the lowest scores (negative contrast).

Figure 3

Co-expression networks among genes and miRNAs being part of enriched pathways (DEGs and correlated to a mineral), hubs, TFs, miRNAs or presenting a significant RIF regarding nine of the minerals in study. (A) Mg, (B) Fe, (C) Ca, (D) Se, (E) K, (F) Na, (G) Cu, (H) P, (I) S. Red lines represent the correlations with a significant RIF gene or miRNA.

Figure 4

Co-expression network containing DEGs for Zn, genes or miRNAs with expression values that are correlated to these DEGs and are also a hub or a significant RIF for Zn, ora miRNA correlated to Zn. Their functional attributes are presented in different colors or shapes. Red lines represent the correlations with a significant RIF gene or miRNA. This network is presented in separate for the others in Fig. 3 because there are no DEGs for Zn in the network taking part of enriched pathways.

Figure 5

Flowchart representing the steps of the methodology.