Identification of common regulators of genes in co-expression networks affecting muscle and meat properties

Abstract

Understanding the genetic contributions behind skeletal muscle composition and metabolism is of great interest in medicine and agriculture. Attempts to dissect these complex traits combine genome-wide genotyping, expression data analyses and network analyses. Weighted gene co-expression network analysis (WGCNA) groups genes into modules based on patterns of co-expression, which can be linked to phenotypes by correlation analysis of trait values and the module eigengenes, i.e. the first principal component of a given module. Network hub genes and regulators of the genes in the modules are likely to play an important role in the emergence of respective traits. In order to detect common regulators of genes in modules showing association with meat quality traits, we identified eQTL for each of these genes, including the highly connected hub genes. Additionally, the module eigengene values were used for association analyses in order to derive a joint eQTL for the respective module. Thereby major sites of orchestrated regulation of genes within trait-associated modules were detected as hotspots of eQTL of many genes of a module and of its eigengene. These sites harbor likely common regulators of genes in the modules. We exemplarily showed the consistent impact of candidate common regulators on the expression of members of respective modules by RNAi knockdown experiments. In fact, Cxcr7 was identified and validated as a regulator of genes in a module, which is involved in the function of defense response in muscle cells. Zfp36l2 was confirmed as a regulator of genes of a module related to cell death or apoptosis pathways. The integration of eQTL in module networks enabled to interpret the differentially-regulated genes from a systems perspective. By integrating genome-wide genomic and transcriptomic data, employing co-expression and eQTL analyses, the study revealed likely regulators that are involved in the fine-tuning and synchronization of genes with trait-associated expression.

Fig 1. Outline of the systems biology approach to dissect the molecular networks of complex traits like meat quality.

The transcript data were integrated with SNP genotype to map expression QTLs (eQTL) of all present transcripts as revealed by microarray analysis. Previously obtained co-expression networks which were correlated with meat quality were analyzed ([15]; white elements) in more depth to reveal the highly connected hub genes in the co-expression modules. Subsequently transcripts and corresponding eQTL of trait-associated modules were addressed. In order to reveal major sites of orchestrated regulation of these genes, eQTL for individual genes including the hub genes in the modules were detected and sites, where several eQTL of multiple genes existed, were considered as hotspots of eQTL harboring likely common regulators of genes in the modules. Moreover, the eigengene of each module was used for association analysis to identify common regulators of genes in module. Genes within the major sites of regulation were listed that were found to be correlated with the expression of genes in the respective module. Finally, the functional link of these potential regulators was exemplarily validated by siRNA knockdown.

Fig 2. Networks of the top 150 correlated genes, including hub genes, in modules orange, dark-turquoise, tan, red, and black.

Nodes of darker color within a module are the top 10 hub genes.

Fig 3. ExpressionQTL identified with the expression values of genes in the modules or the eigengene value of the respective module.

Black and red dashed lines are a genome-wide significance threshold corresponding to negative log₁₀ (NegLog10)>4 and >5, respectively. a-b, Manhattan plots of genome-wide association analyses of (a) transcript abundance of genes and (b) eigengene value for module dark-turquoise. c-d, Manhattan plots of genome-wide association analyses of (c) transcript abundance of genes and (d) eigengene value for module orange.

Fig 4. Knockdown of regulator genes by RNA interference reveals regulation within modules.

siRNAs were designed to target regulator genes in modules dark-turquoise and orange, and transfected into murine C2C12 muscle cells in vitro. Relative mRNA expression was measured by qPCR 48 hours after transfection. Expression was normalized to Hrrt1 and Ppia internal controls. a) Zfp36l2 as a partial regulator of genes in module dark-turquoise. Expression of Zfp36l2 was significantly reduced relative to its expression in control cells at 48 hours post-transfection of siRNA. Expression of other genes (Fam134b, Irs2, Ndel1, Nr4a3, Ppargc1a, Crem, Sdc4) in module dark-turquoise was significantly reduced; these genes are functionally enriched in apoptosis and cell death. b) Cxcr7 as a regulator of genes in module orange. siRNA targeting significantly reduced the levels of Cxcr7 relative to control at 48 hours post-transfection. Expression of most top hub genes in module orange, like Egr1, Zfp36, Fos, Klf4, Ankrd1, Otud1, Adamts1, Gadd45b, Ier5, Tiparp, and Jun, was also reduced. The data represent mean±SEM (n = 4–6 independent experiments). * indicated significant level at p<0.05 and ** indicated significant level at p<0.01.