Transcriptome profiling in fast versus slow-growing rainbow trout across seasonal gradients

Abstract

Background: Circannual rhythms in vertebrates can influence a wide variety of physiological processes. Some notable examples include annual reproductive cycles and for poikilotherms, seasonal changes modulating growth. Increasing water temperature elevates growth rates in fishes, but increases in photoperiod regime can have similar influences even at constant temperature. Therefore, in order to understand the dynamics of growth in fish it is important to consider the background influence of photoperiod regime on gene expression differences. This study examined the influence of a declining photoperiod regime (winter solstice) compared to an increasing photoperiod regime (spring equinox) on white muscle transcriptome profiles in fast and slow-growing rainbow trout from a commercial aquaculture strain.

Results: Slow-growing fish could be characterized as possessing transcriptome profiles that conform in many respects to an endurance training regime in humans. They have elevated mitochondrial and cytosolic creatine kinase expression levels and appear to suppress mTOR-signaling as evidenced by elevated TSC2 expression, and they also have elevated p53 levels. Large fish display a physiological repertoire that may be consistent with strength/resistance physiology having elevated cytoskeletal gene component expression and glycogen metabolism cycling along with higher PI3K levels. In many respects small vs. large fish match eccentric vs. concentric muscle expression patterns, respectively. Lipid metabolic genes are also more elevated in larger fish, the most notable being the G0S2 switch gene. M and Z-line sarcomere remodelling appears to be more prevalent in large fish. Twenty-three out of 26 gene families with previously reported significant SNP-based growth differences were detected as having significant expression differences.

Conclusions: Larger fish display a broader array of genes showing higher expression, and their profiles are more similar to those observed in December lot fish (i.e., an accelerated growth period). Conversely, small fish display gene profiles more similar to seasonal growth decline phases (i.e., September lot fish). Overall, seasonal timing was coupled to greater differences in gene expression compared to differences associated with fish size.

Fig. 1

Experimental design depicting the selection of size-matched differences of a large and small fish selected from 3 different paternal half-sib families in each of two seasonal spawning lots (September and December). September fish were sampled during a declining growth phase (winter solstice) while December lot fish were sampled during an increasing growth phase (spring equinox) when they were approximately 15 months of age.

Fig. 2

Number of contigs (Y-axis) with differential gene expression at the nominal P ≤ 0.05 and FDR 0.05 level in large and small fish, and across seasonal groupings (December vs. September lots) out of 31,600 possible contigs assessed (edge R analysis)

Fig. 3

Multidimensional Scaling Plot (edgeR analysis) of gene expression profiles observed within December lot fish (blue polygon), September lot fish (red polygon), large fish (orange polygon), and small fish (purple polygon)

Fig. 4

Results from the REVIGO analysis analyzing significant differences in FDR corrected gene counts for biological process GOSlim categories. Large and Small fish differences are depicted in the two left-had panels, while differences between December vs. September fish are shown in the right-hand panels. Gene Ontology categories with significantly different higher gene counts between large vs. small fish and between December vs. September fish are indicated

Fig. 5

Neighbour-joining tree depicting the SNP allele genetic distances (1-CSSC values) of all 12 fish used in the analysis. September and December seasonal groupings are depicted as Sept. and Dec., respectively, while Large and Small fish are shown as L and S, respectively

Fig. 6

Pairwise Absolute Differences (PAD coefficient) expression levels based upon RPKM counts among all 12 fish sampled. RPKM values among 31,600 contigs, following edgeR filtering, are shown in Panel a, while Panel b depicts results from the most highly upregulated (RPKM ≥ 5) 18,756 contigs