Integrated analysis of lncRNA and mRNA expression in rainbow trout families showing variation in muscle growth and fillet quality traits

Abstract

Muscle yield and quality traits are important for the aquaculture industry and consumers. Genetic selection for these traits is difficult because they are polygenic and result from multifactorial interactions. To study the genetic architecture of these traits, phenotypic characterization of whole body weight (WBW), muscle yield, fat content, shear force and whiteness were measured in ~500 fish representing 98 families from a growth-selected line. RNA-Seq was used to sequence the muscle transcriptome of different families exhibiting divergent phenotypes for each trait. We have identified 240 and 1,280 differentially expressed (DE) protein-coding genes and long noncoding RNAs (lncRNAs), respectively, in fish families exhibiting contrasting phenotypes. Expression of many DE lncRNAs (n = 229) was positively correlated with overlapping, neighboring or distantly located protein-coding genes (n = 1,030), resulting in 3,392 interactions. Three DE antisense lncRNAs were co-expressed with sense genes known to impact muscle quality traits. Forty-four DE lncRNAs had potential sponge functions to miRNAs that affect muscle quality traits. This study (1) defines muscle quality associated protein-coding and noncoding genes and (2) provides insight into non-coding RNAs involvement in regulating growth and fillet quality traits in rainbow trout.

Figure 1

Effect of WBW as a variable on muscle yield and other quality traits (fat content, shear force and whiteness index) variations. WBW showed moderate regression coefficient (R²) values of 0.56 and 0.50 with muscle yield and fat content, respectively. Fillet whiteness and shear force had low coefficient values of 0.18 and 0.01 with WBW, respectively.

Figure 2

Phenotypic variation in growth and muscle quality traits in 98 families (~5 fish each).

Figure 3

Venn diagram showing unique DE protein-coding genes (a) and DE lncRNAs (b) for each trait in addition to common genes between different traits. Number of DE genes for each trait is shown in (c) (FDR < 0.05 and fold change ≥ 2 or ≤−2).

Figure 4

Consistency between RNA-Seq and qPCR measurements of DE protein-coding genes (a) and DE lncRNAs (b) (R² = 0.89 & 0.81, respectively). Correlations between expression pattern of MYSS (c) and STC (d) with muscle yield and shear force, respectively.

Figure 5

(a–l) Orientation of lncRNAs relative to the overlapping protein-coding loci (on the left) and comparison of the expression patterns (TPM) of DE lncRNAs with the overlapping DE protein-coding genes across families showing significant phenotypic variations (on the right).

Figure 6

(a) Correlation between single DE lncRNA and co-expressed distant protein-coding genes in addition to their correlations with the phenotypes (WBW, muscle yield, and fat content). The color intensity of the nodes reflects the correlations between the genes and phenotypes. The color of the edges reflects the correlation between the protein-coding genes and DE lncRNA where red denotes 0.90 > R ≥ 0.85, purple color denotes 0.95 > R ≥ 0.90, and green color means R ≥ 0.95. (b) Correlation between two antisense DE lncRNAs and their co-expressed distant protein-coding genes that are significantly upregulated in fish families showing variations in WBW, muscle yield, and fat content and known to have an impact on the phenotypes. Edge colors reflect the strength of expression correlation between the DE lncRNAs and protein-coding genes.

Figure 7

Gene enrichment analysis of protein-coding genes co-expressed with DE lncRNAs. Enriched gene-sets are represented as red nodes connected according to their GO/KEGG pathway relations. Color intensity of the node represents the fold enrichment while node size represents number of genes in the gene-set. Enriched terms belonging to metabolic pathways, energy and growth-related mechanisms were predominant.