Conflict of interest The authors declare no conflict of interests.
Funding sources The epigenetic investigation was supported by National Institutes of Health, USA (R01-AI091905). The 18-year assessment of the 1989 Isle of Wight birth cohort was funded by grants from the National Institutes of Health, USA (R01 HL082925) and National Eczema Society/British Dermatological Nursing Group Research Awards.
Background Loss-of-function variants within the filaggrin gene (FLG) are associated with a dysfunctional skin barrier that contributes to the development of eczema. Epigenetic modifications, such as DNA methylation, are genetic regulatory mechanisms that modulate gene expression without changing the DNA sequence.
Objectives To investigate whether genetic variants and adjacent differential DNA methylation within the FLG gene synergistically act on the development of eczema.
Methods A subsample (n = 245, only females aged 18 years) of the Isle of Wight birth cohort participants (n = 1456) had available information for FLG variants R501X, 2282del4 and S3247X and DNA methylation levels for 10 CpG sites within the FLG gene. Log-binomial regression was used to estimate the risk ratios (RRs) of eczema associated with FLG variants at different methylation levels.
Results The period prevalence of eczema was 15.2% at age 18 years and 9.0% of participants were carriers (heterozygous) of FLG variants. Of the 10 CpG sites spanning the genomic region of FLG, methylation levels of CpG site ‘cg07548383’ showed a significant interaction with FLG sequence variants on the risk for eczema. At 86% methylation level, filaggrin haploinsufficient individuals had a 5.48-fold increased risk of eczema when compared to those with wild type FLG genotype (P-value = 0.0008).
Conclusions Our novel results indicated that the association between FLG loss-of-function variants and eczema is modulated by DNA methylation. Simultaneously assessing the joint effect of genetic and epigenetic factors within the FLG gene further highlights the importance of this genomic region for eczema manifestation.
Eczema is an inflammatory skin disorder that affects up to 25% of children in the United Kingdom.1,2 The major hallmarks of eczema are dry and itchy skin, immunoglobulin-E mediated sensitization in response to allergens and disrupted epidermal barrier function.1,3 To date, loss-of-function variants within the filaggrin gene (FLG) are the strongest risk factors for eczema development.2,4FLG encodes the essential filament aggregating protein for the formation of a functional skin barrier that prevents transepidermal water loss and penetration of environmental substances.2,3 Filaggrin haploinsufficiency, i.e. reduction in filaggrin protein expression in heterozygous individuals, is associated with approximately threefold increased risk for eczema.2,4
The limited explanation of the high heritability (up to 86%) and clustering of eczema within families provided by candidate-gene and genome-wide association studies suggest that genetic regulatory factors, other than DNA sequence variants, may account for the unexplained genetic effect.4,5 Epigenetic regulatory mechanisms are mitotically heritable and can alter gene expression without changing the DNA sequence.6 DNA methylation, widely studied in epidemiological investigations due to practical and biological reasons,7,8 along with histone modifications are important epigenetic marks that work hand-in-hand on influencing disease expression.9 We hypothesized that differential DNA methylation produces a higher risk for eczema among those with both FLG loss-of-function variants and high levels of DNA methylation in adjacent CpG sites.
Materials and methods
Study population and characteristics
A whole-population birth cohort was established on the Isle of Wight in 1989 to prospectively study the natural history of allergic conditions. After exclusion of adoptions, perinatal deaths and refusal, 1456 children (95%) were enrolled with follow-up assessments conducted at 1, 2, 4, 10 and 18 years. In this analysis, we focused on information collected at 18 years of age from 245 girls who were randomly selected for screening of epigenetic associations with allergic disorders for a mother-infant study. In all assessments, eczema was defined as chronic or chronically relapsing, itchy dermatitis lasting more than 6 weeks with characteristic morphology and distribution,10 following Hanifin and Rajka criteria.11
FLG genotyping and DNA methylation
Individuals carrying the minor allele for at least one of the FLG variants R501X, 2282del4, or S3247X were classified as having filaggrin haploinsufficiency. Detailed information on genotyping is provided by Ziyab et al.12 For measuring methylation levels, DNA was extracted from whole blood collected at 18 years of age,13 and bisulfite-treated for cytosine to thymine conversion using the EZ 96-DNA methylation kit (Zymo Research, Irvine, CA, USA). Genome-wide DNA methylation was assessed using the Illumina Infinium HumanMethylation450 BeadChip (Illumina, Inc., San Diego, CA, USA), which interrogates >484 000 CpG sites associated with approximately 24 000 genes. Arrays were processed using a standard protocol.14 The BeadChips were scanned using a BeadStation, and the methylation level (beta value) was calculated for each queried CpG locus using the Methylation Module of BeadStudio software (Illumina, Inc., San Diego, CA, USA).
After cleaning of the DNA methylation data, beta (β) values presented as the proportion of methylated (M) over methylated (M) and unmethylated (U) sites (β = M/[c + M + U], with c being constant to prevent dividing by zero) were used to estimate the effect of DNA methylation.15 The methylation levels of 10 CpG sites spanning the genomic region of the FLG were analysed. Wilcoxon Mann–Whitney tests were conducted to test for differences in methylation levels between yes/no strata of eczema and FLG variants. Statistical modelling was applied to assess statistical interaction on a multiplicative scale between FLG variants and methylation levels on the risk for eczema. Log-binomial regression models were used to estimate the risk ratios (RRs). GLIMMIX procedure in SAS 9.2 (SAS, Gary, NC, USA) was applied to determine the RRs of eczema associated with FLG variants at different methylation levels. To control for falsepositive results, we applied the false discovery rate (FDR) method to estimate adjusted P-values.16
The period prevalence of eczema was found to be 15.2% (37/244) in the subgroup with available methylation data, which was not significantly different (P-value = 0.703) from the period prevalence among girls in the original cohort 16.3% (107/656) at age 18 years. The combined proportion of carriers (heterozygous) of FLG variants R501X, 2282del4 and S3247X was 9.0% (23/234) and 9.3% (54/581) in the analysed and original cohort, respectively, showing no significant difference (P-value = 0.982). Of the 10 analysed CpG sites (Table 1), only methylation levels of one site (cg07548383) showed a marginal differential distribution across filaggrin haploinsufficiency present/absent groups, which was not statistically significant after adjustment for FDR. The majority of individuals (66.1%) had DNA methylation levels >83% for cg07548383. At DNA methylation >83%, the prevalence of eczema was 12.7% among individuals with wild type FLG and 33.3% among carriers of FLG variants.
Table 1. Characteristics and descriptive statistics of CpG methylation sites within the filaggrin gene
Location of CpG*
*Refers to the location of the CpG site relevant to the genomic regions of the filaggrin gene: ‘Body’– intron or exons; ‘TSS1500’ and ‘TSS200’– 1500 or 200 base pairs before the transcription starting site; ‘5′ UTR’– contains elements for controlling gene expression – it begins at the transcription start site and ends one nucleotide before the start codon of the coding region.
IQR, interquartile range.
The interaction term ‘FLG variants × CpG site (cg07548383) methylation’ showed statistical significance (P-value = 0.02; Table 2). The RRs of eczema associated with FLG variants increased as the methylation levels increased (Fig. 1). For instance, compared with methylation levels ≤0.84, at methylation levels of 0.85 and 0.86 the RRs of eczema among those with FLG variants when compared with those without FLG variants were 2.61 (P-value = 0.023) and 5.48 (P-value = 0.0008) respectively. The observed increased risk of eczema (RR = 5.48) due to joint effect of both risk factors remained statistically significant after correcting of multiple testing (adjusted P-value = 0.008).
Table 2. Statistical interaction between filaggrin variants and DNA methylation of cg07548383 on the risk for eczema at age 18 years
The CpG site ‘cg07548383’ was modeled as continuous covariate.
FLG variants × cg07548383
This is the first study to determine the role of both genetic and epigenetic factors within the genomic region of FLG gene on the risk for eczema. Previously, we showed an association between FLG variants and eczema at the age of 18 years (RR = 1.6; P-value = 0.031).12 Our working hypothesis for this analysis was that genetic variants and adjacent differential DNA methylation within the genomic region of FLG (Fig. 2) act jointly on increasing the risk for eczema. In this report, we show that at 86% methylation level, FLG variants carriers have a 5.48-fold increased risk for eczema. Our results indicate that the risk of eczema among filaggrin haploinsufficient study subjects is modulated by DNA methylation that occurred in the ‘body’ (intragenic region) of the FLG gene.
Our observation of the role of an intragenic CpG site is further supported by the emerging evidence, which suggests that intragenic DNA methylation may have a major role in regulating gene expression and phenotype variation.8,17 Hence, a change in the research paradigm from mainly focusing on DNA methylation at the promoter regions to incorporating intragenic regions is rapidly emerging and scientifically reasoned. However, the mechanistic role of intragenic DNA methylation is not clear yet. We assume that the effect of DNA methylation depends on genetic variants and has an effect of its own only if there are few genetic variations. In the case of haploinsufficiency of FLG, we assume that DNA methylation has no independent effect, but higher intragenic methylation may result in additional masking and thus increasing the effect of the haploinsufficiency.
A limitation is that our results can only be generalized to women. However, this is due to the focus of the study, namely to assess epigenetic inheritance from mother to offspring, and not related to a disease-specific selection. Another limitation is that the 95% CI (2.04–14.72) bounding the RR at the highest methylation level (86%) is wide, which is due to the limited number of individuals (n = 29) with 86% methylation level. Evidence of selection bias is absent since prevalence of eczema and FLG variants is comparable between analysed and original cohort. Multiple testing was a concern as we tested the joint effect of differential DNA methylation of 10 CpG sites and FLG variants separately. Nevertheless, the observed increased risk remained statistically significant after penalizing its P-value for FDR. Regarding reliability and specificity, a recent report demonstrated that the Infinium HumanMethylation450 array, which was used to obtained DNA methylation profiles in this study, had strong reproducibility and high validity.18 Using interaction between continuous DNA methylation levels and FLG variants to explain the occurrence of eczema, we showed a dose-response relationship and avoided an attempt to identify a threshold as our sample size is too small to determine a threshold. A major strength of our study is having information on genetic variants that are adjacent to methylation sites, which allowed us to determine whether those two factors interact. The CpG site cg07548383 is in close proximity and within the same exon-3 to the most prevalent FLG variants (R501X and 2282del4) (Fig. 2).
The extent to which DNA methylation measured in blood relate to other tissues and whether can be used as a biomarker for phenotype variation is unclear and an area of current scientific dispute.7,8,19 Furthermore, we were not able to assess whether the FLG expression is modulated by differential DNA methylation in the current study as data on gene expression or proteins were not collected yet. In summary, the novel finding of a complex interplay between genetic and epigenetic factors is of high importance and needs to be explored in future studies. Therefore, considering both genetic and epigenetic factors that regulate the formation of the epidermal barrier will lead to an improved assessment of genetic risks.
We would like to acknowledge the help of all the staff at The David Hide Asthma and Allergy Research Centre in undertaking the 18 year and previous assessments of 1989 Isle of Wight birth cohort. We would also like to acknowledge the help of the participants and their families who have helped us with this project over the last two decades.