Widespread domain-like perturbations of DNA methylation in whole blood of Down syndrome neonates

Abstract

Introduction: Down syndrome (DS) is the most frequent genetic cause of intellectual disability. Despite the fact that more than 50 years have passed since the discovery of its genetic aberrations, the exact pathogenesis of the DS phenotype has remained largely unexplained. It was recently hypothesized that the DS pathogenesis involves complex (epi)genetic, molecular and cellular determinants. To date, many reports have addressed epigenetic aberrations associated with DS at different developmental stages/ages and tissue types, but to our best knowledge not in DS newborns. This study aimed to investigate genome-wide methylation patterns in DS newborns compared to non-trisomic newborns.

Method: We analyzed blood samples obtained from ten newborns with DS and five age-matched non-trisomic newborns. Epigenetic profiles were obtained from extracted DNA using the Illumina Infinium 450K array. Since aberrant blood cell distribution is known to be present in DS, we applied two distinct models: with and without correction for estimated blood cell distribution.

Results: Differentially methylated position (DMP) analysis of the uncorrected model detected 19525 significant hits (51,2% hypomethylated). In the corrected model, we found 121953 significant DMPs (49,8% hypomethylated). Independent of the used model we observed a chromosome 21 dosage effect. Moreover, we detected 46 and 145 differentially methylated regions in the uncorrected and corrected model respectively, both showing hypomethylation overrepresentation. Replication analyses of DMPs and DMRs found by Bacalini et al. (2015) showed a large overlap.

Conclusion: In this study, we found methylation profile differences between DS newborns and controls reflecting a systematically affected epigenetic profile. The observed chromosome 21 dosage effect suggests the involvement of affected essential regulatory factors/regions or altered expression of chromatin modeling enzymes located on chromosome 21. Additional research is necessary to substantiate these hypotheses.

Fig 1. Estimated blood cell distribution.

Relative estimated distribution of CD8+ and CD4+ T cells, natural killer cells, B cells, monocytes, and granulocytes among down syndrome (DS) and controls.

Fig 2. Manhattan plots.

(A) Manhattan plot DMP association analysis (probes associated with SNP, MAF>0,05, and probes located on X and Y chromosomes were excluded from analysis), corrected for gender but not for blood cell distribution. X-axis represents ranked chromosomes, Y-axis represents–log10 (P-value). Redline indicates significance level (Bonferroni; P < 5 10⁻⁷). (B) Manhattan plot DMP association analysis (probes associated with SNP, MAF>0,05, and probes located on X and Y chromosomes were excluded from analysis), corrected for gender and blood cell distribution. X-axis represents ranked chromosomes, Y-axis represents–log10 (P-value). Red line indicates significance level (Bonferroni; P < 5 10⁻⁷).

Fig 3. Volcano plots.

(A) Volcano plot DMP association analysis (X and Y chromosomes were excluded from analysis), corrected for gender but not for blood cell distribution. X-axis represents (effect size) adjusted mean delta difference (δ based on adjusted coefficients), Y-axis represents–log10(q-value). Horizontal line indicates significance level (P < 5 10⁻⁷), vertical lines indicate δ_abs > 0,5. (B) Volcano plot DMP association analysis (X and Y chromosomes were excluded from analysis), corrected for gender and blood cell distribution. X-axis represents (effect size) adjusted mean delta difference (δ based on adjusted coefficients), Y-axis represents–log10(q-value). Horizontal line indicates significance level (P < 5 10⁻⁷), vertical lines indicate δ_abs > 0,1.

Fig 4. Distribution of hypo -and hypermethylation over the autosomes in DS vs. controls.

(A) Relative to the total number of probes, percentages of hypo- and hypermethylated probes. Hypomethylated δ < 0, hypermethylated δ > 0. DMP association analysis corrected for gender but not for blood cell distribution. (B) Relative to the total number of probes, percentage of hypo- and hypermethylated probes. Hypomethylated δ < 0, hypermethylated δ > 0. DMP association analysis corrected for gender and blood cell distribution. (C) Relative to the total number of genome-wide significant (q<0,05) probes, percentage of hypo- and hypermethylated probes. Hypomethylated δ < 0, hypermethylated δ > 0. DMP association analysis corrected for gender but not for blood cell distribution. (D) Relative to the total number of genome-wide significant (q<0,05) probes, percentage of hypo- and hypermethylated probes. Hypomethylated δ < 0, hypermethylated δ > 0. DMP association analysis corrected for gender and blood cell distribution.

Fig 5. Chromosome 21 Manhattan plots.

(A) Manhattan plot chromosome 21 only, DMP association analysis, corrected for gender. X-axis represents ranked position chromosome 21, Y-axis represents q-value DMPs. Red line indicates significance level (q < 0,05). (B) Manhattan plot chromosome 21 only, DMP association analysis, corrected for gender and blood cell distribution. X-axis represents ranked position chromosome 21, Y-axis represents q-value DMPs. Red line indicates significance level (q < 0,05).

Fig 6. Regional plots of replicated DMRs.

CpGs within DMRs are annotated between colored vertical lines. (A) RUNX1, overlapping (4 of 5 probes) DMRs detected in the uncorrected and corrected model are highlighted separately, (B) NAV1, DMR detected in the uncorrected model, (C) HOXA5 and (D) HOXA6, for both genes, DMRs detected in the corrected model. X-axis represents the chromosomal position and Y-axis represents methylation index (Beta value). Tracks: CpG: localization of CpG sites (UCSC, HG19), RefGen: Reference gene, blocks represents exonic regions, arrow indicates direction of transcription (UCSC, HG19), CpG island: blocks represent the localization of CpG islands (> 300bp, GC content > 50%, ratio observed/expected CpG > 0,6). K562 H3k27me3: histone 3 lysine 27 3 methylation mark in K562 (chronic myelogenous leukemia human cell line, gain of the mark is associated with decreased activity of gene). K562H3k27ac: histone 3 lysine 27 acetylation mark in K562 (chronic myelogenous leukemia human cell line, an increased level of this mark is associated with increased gene activity, i.e., active enhancer mark).