New DNA Methylation Array Opens a Window onto the Human Imprintome

Scientists have built and validated the first custom Infinium DNA methylation array designed specifically to profile the human imprintome, the collection of imprint control regions that govern whether genes inherited from the mother or the father are switched on. The tool, described in the open-access journal Epigenetics Communications, fills a conspicuous gap in epigenetic research: until now, the standard commercial methylation arrays used by thousands of laboratories worldwide captured only a small fraction of these biologically crucial regions, leaving much of the imprintome invisible to large-scale studies of human disease and environmental exposure.

Imprint control regions, or ICRs, are stretches of DNA carrying methylation marks that are laid down before gastrulation, very early in embryonic development, and then faithfully copied through every subsequent cell division. Because one parental allele is methylated while the other remains unmethylated, imprinted genes are expressed from a single copy, making them exquisitely sensitive gene dosage regulators. When these marks are disturbed, the consequences can be severe: imprinting abnormalities underlie Prader-Willi and Angelman syndromes, Silver-Russell and Kagami-Ogata syndromes, a range of cancers, neurological disorders, and chronic diseases linked to environmental contaminants. Because the marks are stable across tissues and across the life course, they also act as long-term molecular records of early-life exposures that questionnaires and conventional exposure assessments cannot recover.

The motivation for the new array stems from a 2022 study in which Dereje Jima and colleagues used whole-genome bisulfite sequencing, or WGBS, to map 1,488 candidate differentially methylated ICRs across the human genome. Before that work, only 24 human ICRs had been well characterized. WGBS remains the reference method because it can, in principle, read every CpG site in the genome, including all 22,157 CpG positions that fall within the newly catalogued ICRs. But it is expensive, demands high sequencing coverage, requires substantial computational infrastructure, and is impractical for the large sample numbers that epidemiological studies of chronic disease require. Existing targeted alternatives are similarly limited: pyrosequencing handles only short single sequences, pooled amplicon sequencing works only for small preselected target sets, and hybridization capture approaches such as targeted methyl-seq demand specialized instrumentation and expertise.

Commercial arrays offered no better solution. The Infinium Methylation EPIC v1 BeadChip covers just 1.4 percent of the CpG sites in the candidate ICRs, representing only 156 of the 1,488 regions with an average of two probes each. The newer EPIC v2 array improves on that, reaching 6.8 percent of the sites and 548 ICRs, but still leaves most of the imprintome unmeasured. To close this gap, the research team, led by Natalia Carreras-Gallo and Ryan Smith of TruDiagnostic together with Cathrine Hoyo, David Skaar, Randy Jirtle and colleagues at North Carolina State University and other institutions, extracted the 22,157 ICR-mapped CpG sites from the hg38 reference genome and worked with Illumina to design probes against them, applying the company’s validated quality thresholds for sequence, GC content and annealing temperature.

The resulting Human Imprintome array carries 22,819 probes in total, including 704 control probes and 22,115 CG probes. Of the CG probes, 10,438 target unique CpG sites, and 9,757 of those align successfully within the candidate ICR catalogue, covering 1,088 of the 1,488 regions, or 73.1 percent, with an average of nine probes per region and a maximum of 171. The design also retains 10,364 background-normalization probes and 1,313 multimapping probes, which were kept because they represent many ICRs but flagged with zeroed coordinates to prevent misassignment. Coverage remains sparse for ICRs on chromosomes 1 through 5 and chromosome 21, where repetitive sequence produces the most multimapping probes, and roughly 400 ICRs currently lack representation altogether, a limitation the team says it is already working to fix in a next version.

Validation proceeded along two fronts. First, the researchers processed matched samples on both the Human Imprintome array and the EPIC v2 array, using whole blood samples from an Alzheimer’s disease cohort recruited at the Duke Memory Clinic. Across 1,703 overlapping probe sites, Pearson correlation coefficients between the two platforms reached 0.788, 0.811 and 0.789 in the three samples tested, with a mean R-squared of 0.796. Notably, the beta-value distributions told a mechanistically satisfying story: while the full probe set of the Imprintome array showed peaks at methylation values of zero, one half and one, the subset of probes mapped to ICRs collapsed to a single dominant peak at one half, exactly the signature expected of parent-of-origin methylation in which one allele is fully methylated and the other fully unmethylated.

Second, the team compared the array against WGBS, the current gold standard, using sixteen autopsy brain specimens from the Duke/UNC Alzheimer’s Disease Research Center brain bank, comprising equal numbers of Alzheimer’s cases and controls among non-Hispanic Black and non-Hispanic White donors. Samples were pooled within each of four groups to boost sequencing coverage, which ranged from 10X to 15X with bisulfite conversion above 97 percent. After filtering CpG sites with fewer than ten sequencing reads, correlations between array and sequencing measurements ranged from 0.532 to 0.657, with a mean R-squared of 0.569. The authors attribute the lower agreement relative to the EPIC v2 comparison to the modest WGBS coverage and to fundamental differences between array-based and sequencing-based methylation measurement, rather than to any flaw in the array itself.

Reliability testing reinforced the platform’s credentials. Eight umbilical cord blood samples from the Newborn Epigenetics Study, a prospective birth cohort of 2,681 pregnant women recruited in Durham, North Carolina between 2005 and 2011, and seventeen whole blood samples from the Alzheimer’s cohort were each run twice on the Imprintome array. Intraclass correlation coefficients between replicate runs ranged from 0.799 to 0.945, with a mean of 0.868, indicating that the array delivers highly reproducible measurements, a prerequisite for biomarker development and for multi-center studies where batch effects and platform drift can otherwise swamp true biological signals.

The practical advantages of the new tool extend beyond coverage. Because it is built on the familiar Infinium BeadChip platform, it slots into existing laboratory workflows, from bisulfite conversion of 200 nanograms of DNA through hybridization on the iScan imaging system, and into established bioinformatic pipelines, including the sesame R package, which the authors used for preprocessing and which supports custom array manifests. Arrays are faster, cheaper and more automatable than next-generation sequencing for the targeted question of ICR methylation, and by concentrating probes on regions of interest, a custom array avoids paying for genome-wide content irrelevant to the research question. The team has also released comprehensive annotations of every probe and its ICR assignments, and plans to make data and resources available through the humanicr.org website.

The broader promise is considerable. With environmental and lifestyle factors estimated to account for 70 to 90 percent of the burden of common chronic diseases, yet with disease often emerging decades after exposure, stable early-established epigenetic marks are among the few measurable archives of developmental history the human body keeps. A reliable, scalable readout of the imprintome could accelerate the discovery of ICRs associated with cancer, neurological disease, imprinting syndromes and environmentally driven illness, sharpen causal inference in exposure epidemiology, and guide therapeutic strategies aimed at growth regulation and developmental disorders. For a field that until recently could name only two dozen human imprint control regions, the arrival of a dedicated, validated array marks a decisive step toward making the full imprintome a routine object of study rather than a specialized pursuit.

Subject of Research: Development and validation of a custom Infinium DNA methylation array for profiling human imprint control regions

Article Title: Creation and validation of the first infinium DNA methylation array for the human imprintome

Article References: Carreras-Gallo, N., Dwaraka, V. B., Jima, D. D., Skaar, D. A., Mendez, T. L., Planchart, A., Zhou, W., Jirtle, R. L., Smith, R., & Hoyo, C. (2024). Creation and validation of the first infinium DNA methylation array for the human imprintome. Epigenetics Communications, 4(1), Article 5. https://doi.org/10.1186/s43682-024-00028-6

Image Credits: AI Generated

DOI: 10.1186/s43682-024-00028-6

Keywords: genomic imprinting, imprintome, imprint control regions, DNA methylation, Infinium array, methylation array, epigenetics, whole-genome bisulfite sequencing, EPIC v2, custom BeadChip, environmental epigenetics, Alzheimer's disease