Maternal vitamin D depletion alters DNA methylation at imprinted loci in multiple generations

Abstract

Background: Environmental perturbation of epigenetic mechanisms is linked to a growing number of diseases. Characterizing the role environmental factors play in modifying the epigenome is important for disease etiology. Vitamin D is an essential nutrient affecting brain, bone, heart, immune and reproductive health. Vitamin D insufficiency is a global issue, and the role in maternal and child health remains under investigation.

Methods: We used Collaborative Cross (CC) inbred mice to characterize the effect of maternal vitamin D depletion on offspring phenotypic and epigenetic outcomes at imprinted domains (H19/Igf2, Snrpn, Dlk1/Gtl2, and Grb10) in the soma (liver) and germline (sperm). We assessed outcomes in two generations of offspring to determine heritability. We used reciprocal crosses between lines CC001/Unc and CC011/Unc to investigate parent of origin effects.

Results: Maternal vitamin D deficiency led to altered body weight and DNA methylation in two generations of offspring. Loci assayed in adult liver and sperm were mostly hypomethylated, but changes were few and small in effect size (<7 % difference on average). There was no change in total expression of genes adjacent to methylation changes in neonatal liver. Methylation changes were cell type specific such that changes at IG-DMR were present in sperm but not in liver. Some methylation changes were distinct between generations such that methylation changes at the H19ICR in second-generation liver were not present in first-generation sperm or liver. Interestingly, some diet-dependent changes in body weight and methylation were seemingly influenced by parent of origin such that reciprocal crosses exhibited inverse effects.

Conclusions: These findings demonstrate that maternal vitamin D status plays a role in determining DNA methylation state in the germline and soma. Detection of methylation changes in the unexposed second-generation demonstrates that maternal vitamin D depletion can have long-term effects on the epigenome of subsequent generations. Differences in vitamin D-dependent epigenetic state between cell types and generations indicate perturbation of the epigenetic landscape rather than a targeted, locus-specific effect. While the biological importance of these subtle changes remains unclear, they warrant an investigation of epigenome-wide effects of maternal vitamin D depletion.

Keywords: Collaborative Cross; Epigenetic inheritance; Imprinting; Maternal diet; Parent of origin; Vitamin D.

Fig. 1

Experimental crosses and dietary treatment scheme. a Dietary treatments are indicated in middle block: CON (AIN-93G), LVD (lacking vitamin D), standard chow (Teklad 8604). Arrows above indicate treatment at developmental timepoints. Shaded regions below dietary treatment scheme indicate treatment windows and time of harvest (dashed vertical line). b Illustration of crosses used in paternal transmission of genomes exposed to vitamin D depletion (highlighted region depicts direct exposure), developmental timepoints and cell types tested (parentheses). c Maternal (G₀ dam) plasma 25(OH)D levels (N = 5, 3, 5, 6, in order from left to right). X-axis indicates dam strain. G ₀ parental generation 0, G ₁ first generation of offspring, G ₂ second generation of offspring, PND postnatal day. Different letters on bars represent statistically significant groups with p value <0.05. Panel above the graph lists all statistically significant (p value <0.05) comparisons determined by regression analysis for overall diet-dependent effects (diet) and diet × strain interactive effects (diet × strain)

Fig. 2

Maternal (G₀) vitamin D deficiency is linked to (G₁) offspring developmental outcomes. a G₁ male 8 weeks body weight (N = 24, 24, 31, 25). b G₁ male 8 weeks percent lean (left y-axis, open bars) and fat mass (right x-axis, closed bars) (N = 3, 10, 5, 6). c Combined testes weight (N = 24, 17, 29, 17, upper panel) and testes weight corrected for body weight (N = 22, 17, 29, 14, bottom panel). d Total sperm count (N = 25, 17, 29, 15). Each dot represents an individual sample. Error bars represent standard error of the mean. Asterisks (*) or (***) indicate p value <0.05 or 0.005 determined by t test within each cross; panel above each graph lists all statistically significant (p value <0.05) comparisons determined by regression analysis for overall diet-dependent effects (diet), diet-independent parent of origin effects (PO), and diet-dependent parent of origin effects (diet × PO). n.s. not significant

Fig. 3

Grandmaternal (G₀) vitamin D deficiency is linked to G₂ offspring developmental outcomes. a G₂ male and female PND4, PND9 (female N = 40, 35, 23, 21 and male N = 49, 24, 31, 24, respectively) and male 8 weeks (N = 20, 10, 13, 11) body weight. b G₂ male 8 weeks body composition (N = 23, 10, 13, 11). c Combined testes weight (upper panel) and testes weight corrected for body weight (bottom panel) (N = 21, 10, 13, 11). d Total sperm count (N = 23, 10, 13, 11). Each dot represents an individual sample. Error bars represent standard error of the mean. Asterisks (*) or (***) indicate p value <0.05 or 0.005 determined by t test within each cross; panel above each graph lists all statistically significant (p value <0.05) comparisons determined by regression analysis for overall diet dependent effects (diet), diet-independent grandparent of origin effects (GPO), and diet-dependent grandparent of origin effects (diet × GPO). n.s. not significant

Fig. 4

Maternal vitamin D deficiency perturbs DNA methylation at imprinted regions in adult G1 soma (liver) and germline (sperm). a Schematic representation of imprinted regions assessed for DNA methylation changes. White, gray, and black lollipops indicate unperturbed methylation states from fully hypomethylated to partially and fully hypermethylated alleles, respectively. Location of lollipop and location of arrows indicates parental methylation or expression from maternal alleles (above gene) and paternal allele (below gene); b–c Bar graphs depict average DNA methylation across samples in the respective treatment group assayed in adult G1 liver (b) and sperm (c) for each sample analyzed (in order from left to right N = 10, 11, 10, 9 for all loci except Grb10 N = 10, 10, 9, 8 and N = 10, 10, 10, 9 liver and sperm, respectively). For each sample, the average methylation for all CpGs assayed across the locus is represented as a dot within the bar graph (see the “Methods” section). DMR differentially methylated region, Cbs CTCF binding site, ICR imprinted control region. Median values are indicated and linked by a horizontal line between box and whisker plots. Asterisks (*) or (***) indicate p value <0.05 or 0.005 determined by t test within each cross; panel above each graph lists all statistically significant (p value <0.05) comparisons determined by regression analysis for overall diet-dependent effects (diet), diet-independent parent of origin effects (PO), and diet-dependent parent of origin effects (diet × PO). n.s. not significant

Fig. 5

Grandmaternal vitamin D deficiency perturbs DNA methylation at imprinted regions in adult G₂ male soma (liver) and germline (sperm). a, b Bar graphs depict average DNA methylation across samples in the respective treatment group assayed in adult G₂ liver (a) and sperm (b) for each sample analyzed (in order from left to right, N = 10, 8, 10, 10 for all loci except Grb10 N = 10, 7, 9, 9). For each sample, the average methylation for all CpGs assayed across the locus is represented as a dot within the bar graph (see the “Methods” section). DMR differentially methylated region, Cbs CTCF binding site, ICR imprinted control region. Median values are indicated and linked by a horizontal line between box and whisker plots. Asterisks (*) or (**) indicate p value <0.05 or 0.01 determined by t test within each cross; panel above each graph lists all statistically significant (p value <0.05) comparisons determined by regression analysis for overall diet-dependent effects (diet), diet-independent grandparent of origin effects (GPO), and diet-dependent grandparent of origin effects (diet × GPO). n.s. not significant

Fig. 6

G₂ male neonatal DNA methylation and gene expression patterns. a Bar graphs depict average DNA methylation across samples in the respective treatment group assayed in neonatal G₂ liver (in order from left to right, N = 10 for all loci except H19PP N = 10, 9, 10, 10). For each sample, the average methylation for all CpGs assayed across the locus is represented as a dot within the bar graph (see the “Methods” section). b, c mRNA abundance relative to housekeeping genes Rplp0 and Gapdh (N = 10 for all genes). DMR differentially methylated region, ICR imprinted control region. Median values are indicated and linked by a horizontal line between box and whisker plots. Asterisks (*) indicates p value <0.05 determined by t test within each cross; panel above each graph lists all statistically significant (p value <0.05) comparisons determined by regression analysis for overall diet-dependent effects (diet), diet-independent grandparent of origin effects (GPO), and diet-dependent grandparent of origin effects (diet × GPO). n.s. not significant

Fig. 7

Diet-dependent DNA methylation profile in liver and sperm of G₁ and G₂. a Heat maps of difference in methylation (value = median of LVD methylation—median of CON methylation) detected in comparing LVD vs. CON groups. Negative values indicate hypomethylation in LVD group and positive values indicate hypermethylation in LVD group. Values represented by shades of color indicated below each panel. b Principal component analysis (PCA) plots for cross 1 and cross 2 was performed for each cell type, generation, and developmental timepoint using difference in methylation at all 7 loci (Igf2DMR1, H19Cbs2, H19Cbs4, IG-DMR, SnrpnICR, and Grb10DMR) between medians of LVD and CON groups. Principal component 1(PC1) and PC2 are the principal components that explain the majority of the variation and are represented on the x-axis and y-axis, respectively. Percent variation explained by each component is listed in parentheses on each axis