One-carbon metabolism nutrients impact the interplay between DNA methylation and gene expression in liver, enhancing protein synthesis in Atlantic salmon

Abstract

Supplementation of one-carbon (1C) metabolism micronutrients, which include B-vitamins and methionine, is essential for the healthy growth and development of Atlantic salmon (Salmo salar). However, the recent shift towards non-fish meal diets in salmon aquaculture has led to the need for reassessments of recommended micronutrient levels. Despite the importance of 1C metabolism in growth performance and various cellular regulations, the molecular mechanisms affected by these dietary alterations are less understood. To investigate the molecular effect of 1C nutrients, we analysed gene expression and DNA methylation using two types of omics data: RNA sequencing (RNA-seq) and reduced-representation bisulphite sequencing (RRBS). We collected liver samples at the end of a feeding trial that lasted 220 days through the smoltification stage, where fish were fed three different levels of four key 1C nutrients: methionine, vitamin B6, B9, and B12. Our results indicate that the dosage of 1C nutrients significantly impacts genetic and epigenetic regulations in the liver of Atlantic salmon, particularly in biological pathways related to protein synthesis. The interplay between DNA methylation and gene expression in these pathways may play an important role in the mechanisms underlying growth performance affected by 1C metabolism.

Keywords: B-vitamins; DNA methylation; Nutrition; SAM; methionine; omics; one-carbon metabolism; smoltification.

Figure 1.

Overview of experimental design and nutrient levels in the feeding trial with growth performance. (a) a schematic diagram depicting different stages, pallet sizes, diets, and sampling points throughout the trial. (b) diagram showing targeted nutrients (blue) along with SAM and SAH (green) in the 1C metabolism that includes folate and methionine cycles. (c) barplots showing levels of four 1C nutrients (B6, folate, B12, and methionine) analysed for both 3 mm and 4 mm pellets for three diet groups (ctrl, 1C+, and 1C++). (d) line plot showing the growth rates in body weights (g) for three diet groups along with p-values calculated by ANOVA.

Figure 2.

Clustering analysis of gene expression differences among three diets. (a) a PCA plot displaying the clusters of three diet groups - ctrl (blue, semi-transparent), 1C+ (yellow, semi-transparent), and 1C++ (red, semi-transparent) – using 27 RNA-seq samples including nine samples from each group. Top 1000 high variance genes were used as input data. (b) a dot plot showing the DBSCAN result in a PCA format with two identified clusters, DEG C1 (blue) and DEG C2 (red), on the pooled set of DEGs generated from three pair-wise comparisons (1C+ vs ctrl, 1C++ vs ctrl, 1C++ vs 1C+). (c) line plots showing normalized read counts as expression levels for three diet groups with the total averages for DEG C1 (blue) and DEG C2 (red) clusters.

Figure 3.

Global and regional DNA methylation landscapes along with SAM/SAH ratio. (a) box plots showing SAM and SAH levels (nmol/g) with SAM/SAH ratio for three diet groups. P-values are calculated by ANOVA, and red stars on the top indicate statistical significance from the base mean by t-test. Dotted horizontal lines indicate base means. (b) a t-SNE plot displaying the clusters of three diet groups - ctrl (blue, semi-transparent), 1C+ (yellow, semi-transparent), and 1C++ (red, semi-transparent) – using 27 RRBS samples. (c) a ridge density plot showing average methylation rates of 157,201 CpG sites for three diet groups. (d) a schematic diagram showing the definition of three different genomic regions – gene body (GB, green), promoter (P, blue), and flanking regions (flank, red) along with intergenic regions (IGRs). (e) ridge density plots showing average methylation rates for three diet groups in four different genetic regions. (f) a running average line plot showing average DNA methylation rates of all RRBS samples within 5000 bp up- and down-stream around TSSs. The running average lines were calculated separately for five different genetic regions – RS (red), P (dark yellow and green), and GB (blue and pink).

Figure 4.

Distributions and counts of DMCs identified by three pair-wise comparisons. (a) violin plots showing distributions and counts of non-DMCs, hypo DMCs, and hyper DMRs in four different genomic regions for three comparisons. The x-axis represents the percentage differences of methylation rates. (b) stacked barplots showing the number of DMCs per gene, 1 DMC (red), 2 DMCs (green), and greater than or equal to 3 DMCs (blue), in three different genomic regions for three comparisons. The x-axis represents the proportions of DMC counts in percentage.

Figure 5.

DNA methylation landscapes in five gene expression groups and six different genomic regions. (a) a density plot showing the distribution of gene expression calculated from normalized read counts from RNA-seq samples. Red vertical lines represent threshold values to define five different gene groups. (b) a barplot showing the count of genes in the five gene expression groups - none, very low, low, med, and high – defined by different gene expression levels. The ‘none’ group contains genes without any expression. (c) ridge density plots displaying distributions of methylation rates for five different gene groups in six genetic regions.