Genetic contributions to epigenetic-defined endotypes of allergic phenotypes in children

Abstract

Asthma is a common respiratory disease, with contributions from both genes and the environment and significant heterogeneity in underlying endotypes; yet, little is known about the relative contributions of each to these endotypes. To address this gap, we used nasal mucosal cell DNA methylation (DNAm) and gene expression and genotypes for 284 children in the Urban Environment and Childhood Asthma (URECA) birth cohort. Using an unbiased data-reduction approach and 37,256 CpGs on a custom-content Asthma&Allergy array, empirical Bayesian factorization was implemented to identify three DNAm signatures that were associated with phenotypes reflecting allergic diseases (allergic asthma and allergic rhinitis), allergic sensitization (atopy) (specific and total immunoglobulin E), and/or type 2 inflammation (eosinophil count and fractional exhaled nitric oxide [FeNO]). These associations were replicated in the Infant Susceptibility to Pulmonary Infections and Asthma (INSPIRE) and the Children's Respiratory Environment Workgroup (CREW) cohorts. The genes that were correlated with each signature in URECA reflected three cardinal endotypes of asthma: inhibited immune response to microbes, impaired epithelial barrier integrity, and activated type 2 immune pathways. To estimate the genetic contributions to these signatures, we used a common set of genotypes available in the three cohorts. The joint SNP heritability of each signature was 0.21 (p = 0.037), 0.26 (p = 1.7 × 10^-8), and 0.17 (p = 7.7 × 10^-6), respectively. The heritabilities of the DNAm signatures suggest that genetic variation contributes significantly to epigenetic signatures of allergic phenotypes and that susceptibility to the development of specific endotypes of asthma is present at birth and is poised to mediate individual epigenetic responses to early-life environments.

Keywords: DNA methylation; T2 inflammation; asthma and allergy endotypes; barrier function; viral response.

Figure 1

Overview of the study We focused our studies are children and young adults in three asthma cohorts. We defined 16 DNAm signatures in nasal mucosal cells from children in the URECA cohort, of which three signatures were associated with phenotypes reflecting allergic and T2 inflammation. These DNAm-phenotype associations were replicated in nasal mucosal cells and nasal lavage cells in two additional cohorts. Gene-expression studies in the same cells from URECA children revealed enrichments for distinct sets of genes reflecting key endotypes of asthma and allergic diseases. Heritability studies of the three signatures were performed in the combined sample, using the subset of meSNPs for the CpGs in the three signatures, eSNPs for genes correlated with any of the three signatures, and fine-mapped SNPs from an allergic rhinitis GWAS, revealing that all three DNAm signatures were significantly heritable.

Figure 2

Overlap and number of CpGs and genes associated with DNAm signatures 5, 8, and 16 (A) 7,761 out of 10,813 (71.8%) CpGs assigned to a signature (lfsr < 0.05) are unique to only one signature; 4.2% are shared by all three. (B) 4,694 out of 7,943 (59.1%) genes correlated (FDR < 0.05) with at least one of the signatures are unique to only one signature; 7.1% are shared by all three.

Figure 3

Significant upstream regulators of genes correlated with DNAm signatures 5, 8, and 16 (FDR < 0.05) (A) Interferon-gamma (IFNG) is an upstream regulator of genes correlated with DNAm signature 5 (mean log₂FC = −1.77; p = 3.99 × 10⁻⁵). GATA-binding protein 2 (GATA2) is downstream of IFNG and significantly activated in children with atopy. (B) Septin-2 (SEPTIN2) is an upstream regulator of genes correlated with DNAm signature 8 (mean log₂FC = −2.66; p = 4.11 × 10⁻¹⁴). Tectonic family member 2 (TCTN2) is downstream of SEPTIN2 and significantly inhibited in children with atopy. (C) Interleukin-13 (IL13) is an upstream regulator of genes correlated with DNAm signature 16 (mean log₂FC = 5.54; p = 3.54 × 10⁻⁴). Periostin (POSTN) is downstream of IL13 and significantly activated in children with atopy. Scatterplots show the correlation between expression of representative genes from each network (y axis) and the individual scores for each DNAm signature (x axis). The red arrow denotes the direction of the correlation with atopy in URECA subjects. The residuals shown in the scatterplots (y axes) are after regressing out site, sequencing batch, ancestry PCs 1–3, and percent squamous and percent ciliated epithelial cells as covariates. FDR-adjusted p values are shown. Red symbols, activated genes; blue symbols, inhibited genes; gray symbols, not differentially expressed.

Figure 4

Heritability of the DNAm signatures (A) QQ plots of GWAS p values (upper) and enrichment p values (lower) for each signature. Genomic inflation factors, or lambdas (λ), were calculated based on all SNPs (black points). (B) The contribution of each set of SNPs to the heritability of each signature. The total heritability of each signature was 0.21 (p = 0.037) for signature 5, 0.26 (p = 1.7 × 10⁻⁸) for signature 8, and 0.17 (p = 7.7 × 10⁻⁶) for signature 16. (C) The per-SNP contribution to the heritability of each signature.