Identifying CpG sites associated with eczema via random forest screening of epigenome-scale DNA methylation

A comparison of feature selection methodologies and learning algorithms in the development of a DNA methylation-based telomere length estimator

BACKGROUND

The field of epigenomics holds great promise in understanding and treating disease with advances in machine learning (ML) and artificial intelligence being vitally important in this pursuit. Increasingly, research now utilises DNA methylation measures at cytosine-guanine dinucleotides (CpG) to detect disease and estimate biological traits such as aging. Given the challenge of high dimensionality of DNA methylation data, feature-selection techniques are commonly employed to reduce dimensionality and identify the most important subset of features. In this study, our aim was to test and compare a range of feature-selection methods and ML algorithms in the development of a novel DNA methylation-based telomere length (TL) estimator. We utilised both nested cross-validation and two independent test sets for the comparisons.

RESULTS

We found that principal component analysis in advance of elastic net regression led to the overall best performing estimator when evaluated using a nested cross-validation analysis and two independent test cohorts. This approach achieved a correlation between estimated and actual TL of 0.295 (83.4% CI [0.201, 0.384]) on the EXTEND test data set. Contrastingly, the baseline model of elastic net regression with no prior feature reduction stage performed less well in general-suggesting a prior feature-selection stage may have important utility. A previously developed TL estimator, DNAmTL, achieved a correlation of 0.216 (83.4% CI [0.118, 0.310]) on the EXTEND data. Additionally, we observed that different DNA methylation-based TL estimators, which have few common CpGs, are associated with many of the same biological entities.

CONCLUSIONS

The variance in performance across tested approaches shows that estimators are sensitive to data set heterogeneity and the development of an optimal DNA methylation-based estimator should benefit from the robust methodological approach used in this study. Moreover, our methodology which utilises a range of feature-selection approaches and ML algorithms could be applied to other biological markers and disease phenotypes, to examine their relationship with DNA methylation and predictive value.

Epigenetic regulation of melanogenesis

Melanogenesis is a complex process in which melanin is synthesized in melanocytes and transported to keratinocytes, which involves multiple genes and signaling pathways. Epigenetics refers to the potential genetic changes that affect gene expression without involving changes in the original sequence of DNA nucleotides. DNA methylation regulates the expression of key genes such as tyrosinase (TYR), tyrosinase-related protein 1 (TYRP1), dopachrome tautomerase (DCT) and microphthalmia-associated transcription factor (MITF), as well as paracrine factors such as stem cell factor (SCF) and endothelin-1 (ET-1) in melanogenesis. Potential DNA methylation sites are present in the genes of melanogenesis-related signaling pathways such as "Wnt", "PI3K/Akt/CREB" and "MAPK". H3K27 acetylation is abundant in melanogenesis-related genes. Both the upstream activation and downstream regulation of MITF depend on histone acetyltransferase CBP/p300, and pH-induced H3K27 acetylation may be the amplifying mechanism of MITF's effect. HDAC1 and HDAC10 catalyze histone deacetylation of melanogenesis-related gene promoters. Chromatin remodelers SWI/SNF complex and ISWI complex use the energy of ATP hydrolysis to rearrange nucleosomes, while their active subunits BRG1, BRM and BPTF, act as activators and cofactors of MITF. MicroRNAs (miRNAs) can directly target a large number of melanogenesis-related genes, while long noncoding RNAs (lncRNAs) and circular RNAs (circRNAs) regulate melanogenesis in a variety of ways. Interactions exist among the epigenetic mechanisms of melanogenesis. For example, the methyl CpG binding domain protein 2 (MeCP2) links DNA methylation, histone deacetylation, and histone methylation. Epigenetic-based therapy provides novel opportunities for treating dermatoses that are caused by pigmentation disturbances. This review summarizes the epigenetic regulation mechanisms of melanogenesis, and examines the pathogenesis and treatment of epigenetics in pigmentation disorders.

Maternal atopy and offspring epigenome-wide methylation signature

The increase in the prevalence of allergic diseases is believed to partially depend on environmental changes. DNA methylation is a major epigenetic mechanism, which is known to respond to environmental factors. A number of studies have revealed that patterns of DNA methylation may potentially predict allergic diseases.Here, we examined how maternal atopy is associated with methylation patterns in the cord blood of neonates.We conducted an epigenome-wide association study in a cohort of 96 mother-child pairs. Pregnant women aged not more than 35 years old, not currently smoking or exposed to environmental tobacco smoke, who did not report obesity before conception were considered eligible. They were further tested for atopy. Converted DNA from cord blood was analysed using Infinium MethylationEPIC; for statistical analysis, RnBeads software was applied. Gestational age and sex were included as covariates in the final analysis.83 DM sites were associated with maternal atopy. Within the top DM sites, there were CpG sites which mapped to genes SCD, ITM2C, NT5C3A and NPEPL1. Regional analysis revealed 25 tiling regions, 4 genes, 3 CpG islands and 5 gene promoters, (including PIGCP1, ADAM3A, ZSCAN12P1) associated with maternal atopy. Gene content analysis revealed pointwise enrichments in pathways related to purine-containing compound metabolism, the G1/S transition of the mitotic cell cycle, stem cell division and cellular glucose homoeostasis.These findings suggest that maternal atopy provides a unique intrauterine environment that may constitute the first environment in which exposure is associated with methylation patterns in newborn.

DNA Methylation at Birth is Associated with Childhood Serum Immunoglobulin E Levels

Immunoglobulin E (IgE) is known to play an important role in allergic diseases. Epigenetic traits acquired due to modification of deoxyribonucleic acid (DNA) methylation (DNAm) in early life may have phenotypic consequences through their role in transcriptional regulation with relevance to the developmental origins of diseases including allergy. However, epigenome-scale studies on the longitudinal association of cord blood DNAm with IgE over time are lacking. Our study aimed to examine the association of DNAm at birth with childhood serum IgE levels during early life. Genome-scale DNAm and total serum IgE measured at birth, 5, 8, and 11 years of children in the Taiwan Maternal and Infant Cohort Study were included in the study in the discovery stage. Linear mixed models were implemented to assess the association between cord blood DNAm at ~310K 5′-cytosine-phosphate-guanine-3′ (CpG) sites with repeated IgE measurements, adjusting for cord blood IgE. Identified statistically significant CpGs (at a false discovery rate, FDR, of 0.05) were further tested in an independent replication cohort, the Isle of Wight (IoW) birth cohort. We mapped replicated CpGs to genes and conducted gene ontology analysis using ToppFun to identify significantly enriched pathways and biological processes of the genes. Cord blood DNAm of 273 CpG sites were significantly (FDR = 0.05) associated with IgE levels longitudinally. Among the identified CpGs available in both cohorts (184 CpGs), 92 CpGs (50%) were replicated in the IoW in terms of consistency in direction of associations between DNA methylation and IgE levels later in life, and 16 of the 92 CpGs showed statistically significant associations (P < .05). Gene ontology analysis identified 4 pathways (FDR = 0.05). The identified 16 CpG sites had the potential to serve as epigenetic markers associated with later IgE production, beneficial to allergic disease prevention and intervention.

Sarcoptic mange severity is associated with reduced genomic variation and evidence of selection in Yellowstone National Park wolves (Canis lupus)

Population genetic theory posits that molecular variation buffers against disease risk. Although this "monoculture effect" is well supported in agricultural settings, its applicability to wildlife populations remains in question. In the present study, we examined the genomics underlying individual-level disease severity and population-level consequences of sarcoptic mange infection in a wild population of canids. Using gray wolves (Canis lupus) reintroduced to Yellowstone National Park (YNP) as our focal system, we leveraged 25 years of observational data and biobanked blood and tissue to genotype 76,859 loci in over 400 wolves. At the individual level, we reported an inverse relationship between host genomic variation and infection severity. We additionally identified 410 loci significantly associated with mange severity, with annotations related to inflammation, immunity, and skin barrier integrity and disorders. We contextualized results within environmental, demographic, and behavioral variables, and confirmed that genetic variation was predictive of infection severity. At the population level, we reported decreased genome-wide variation since the initial gray wolf reintroduction event and identified evidence of selection acting against alleles associated with mange infection severity. We concluded that genomic variation plays an important role in disease severity in YNP wolves. This role scales from individual to population levels, and includes patterns of genome-wide variation in support of the monoculture effect and specific loci associated with the complex mange phenotype. Results yielded system-specific insights, while also highlighting the relevance of genomic analyses to wildlife disease ecology, evolution, and conservation.

Gaussian Bayesian network comparisons with graph ordering unknown

A Bayesian approach is proposed that unifies Gaussian Bayesian network constructions and comparisons between two networks (identical or differential) for data with graph ordering unknown. When sampling graph ordering, to escape from local maximums, an adjusted single queue equi-energy algorithm is applied. The conditional posterior probability mass function for network differentiation is derived and its asymptotic proposition is theoretically assessed. Simulations are used to demonstrate the approach and compare with existing methods. Based on epigenetic data at a set of DNA methylation sites (CpG sites), the proposed approach is further examined on its ability to detect network differentiations. Findings from theoretical assessment, simulations, and real data applications support the efficacy and efficiency of the proposed method for network comparisons.

Shared DNA methylation signatures in childhood allergy: The MeDALL study

BACKGROUND

Differential DNA methylation associated with allergy might provide novel insights into the shared or unique etiology of asthma, rhinitis, and eczema.

OBJECTIVE

We sought to identify DNA methylation profiles associated with childhood allergy.

METHODS

Within the European Mechanisms of the Development of Allergy (MeDALL) consortium, we performed an epigenome-wide association study of whole blood DNA methylation by using a cross-sectional design. Allergy was defined as having symptoms from at least 1 allergic disease (asthma, rhinitis, or eczema) and positive serum-specific IgE to common aeroallergens. The discovery study included 219 case patients and 417 controls at age 4 years and 228 case patients and 593 controls at age 8 years from 3 birth cohorts, with replication analyses in 325 case patients and 1111 controls. We performed additional analyses on 21 replicated sites in 785 case patients and 2124 controls by allergic symptoms only from 8 cohorts, 3 of which were not previously included in analyses.

RESULTS

We identified 80 differentially methylated CpG sites that showed a 1% to 3% methylation difference in the discovery phase, of which 21 (including 5 novel CpG sites) passed genome-wide significance after meta-analysis. All 21 CpG sites were also significantly differentially methylated with allergic symptoms and shared between asthma, rhinitis, and eczema. The 21 CpG sites mapped to relevant genes, including ACOT7, LMAN3, and CLDN23. All 21 CpG sties were differently methylated in asthma in isolated eosinophils, and 10 were replicated in respiratory epithelium.

CONCLUSION

Reduced whole blood DNA methylation at 21 CpG sites was significantly associated with childhood allergy. The findings provide novel insights into the shared molecular mechanisms underlying asthma, rhinitis, and eczema.

Machine Learning-Based DNA Methylation Score for Fetal Exposure to Maternal Smoking: Development and Validation in Samples Collected from Adolescents and Adults

BACKGROUND

Fetal exposure to maternal smoking during pregnancy is associated with the development of noncommunicable diseases in the offspring. Maternal smoking may induce such long-term effects through persistent changes in the DNA methylome, which therefore hold the potential to be used as a biomarker of this early life exposure. With declining costs for measuring DNA methylation, we aimed to develop a DNA methylation score that can be used on adolescent DNA methylation data and thereby generate a score for in utero cigarette smoke exposure.

METHODS

We used machine learning methods to create a score reflecting exposure to maternal smoking during pregnancy. This score is based on peripheral blood measurements of DNA methylation (Illumina's Infinium HumanMethylation450K BeadChip). The score was developed and tested in the Raine Study with data from 995 white 17-y-old participants using 10-fold cross-validation. The score was further tested and validated in independent data from the Northern Finland Birth Cohort 1986 (NFBC1986) (16-y-olds) and 1966 (NFBC1966) (31-y-olds). Further, three previously proposed DNA methylation scores were applied for comparison. The final score was developed with 204 CpGs using elastic net regression.

RESULTS

Sensitivity and specificity values for the best performing previously developed classifier ("Reese Score") were 88% and 72% for Raine, 87% and 61% for NFBC1986 and 72% and 70% for NFBC1966, respectively; corresponding figures using the elastic net regression approach were 91% and 76% (Raine), 87% and 75% (NFBC1986), and 72% and 78% for NFBC1966.

CONCLUSION

We have developed a DNA methylation score for exposure to maternal smoking during pregnancy, outperforming the three previously developed scores. One possible application of the current score could be for model adjustment purposes or to assess its association with distal health outcomes where part of the effect can be attributed to maternal smoking. Further, it may provide a biomarker for fetal exposure to maternal smoking. https://doi.org/10.1289/EHP6076.

An analysis of the effect of mu-opioid receptor gene (OPRM1) promoter region DNA methylation on the response of naltrexone treatment of alcohol dependence

This study explored the effect of OPRM1 promoter region DNA methylation on the outcome of treatment with the opioid antagonist naltrexone (NTX) for alcohol dependence (AD). Ninety-three patients with DSM-IV AD [41 African Americans (AAs) and 52 European Americans (EAs)] received double-blind treatment with NTX or placebo for at least three months. Relapse to heavy drinking was assessed during the first 13 weeks of the trial. Peripheral blood methylation levels of 33 CpG units in the OPRM1 promoter region were quantified using Sequenom EpiTYPER technology. Bayesian logistic regression was used to analyze the effects of NTX treatment, CpG methylation, CpG methylation×NTX treatment, and age on AD relapse. The Random Forest machine learning algorithm was applied to select AD relapse predictors. No significant effect of individual OPRM1 promoter CpG units on AD relapse was observed in either AAs or EAs. Age was significantly associated with AD relapse in EAs, among whom older subjects had a lower relapse rate. Random forest analyses revealed that the prediction rate for AD relapse reached 66.0% with five top variables (age and four CpG units; ranked by their importance to AD relapse) in the prediction model. These findings suggest that methylation levels of individual OPRM1 promoter CpG units do not contribute significantly to inter-individual variation in NTX response. However, the age of subjects in combination with a cluster of specific OPRM1 promoter CpG units may affect NTX treatment outcome. Additional studies of OPRM1 DNA methylation changes during and after NTX treatment of AD are needed.

Replicated methylation changes associated with eczema herpeticum and allergic response

Background

Although epigenetic mechanisms are important risk factors for allergic disease, few studies have evaluated DNA methylation differences associated with atopic dermatitis (AD), and none has focused on AD with eczema herpeticum (ADEH+). We will determine how methylation varies in AD individuals with/without EH and associated traits. We modeled differences in genome-wide DNA methylation in whole blood cells from 90 ADEH+, 83 ADEH−, and 84 non-atopic, healthy control subjects, replicating in 36 ADEH+, 53 ADEH−, and 55 non-atopic healthy control subjects. We adjusted for cell-type composition in our models and used genome-wide and candidate-gene approaches.

Results

We replicated one CpG which was significantly differentially methylated by severity, with suggestive replication at four others showing differential methylation by phenotype or severity. Not adjusting for eosinophil content, we identified 490 significantly differentially methylated CpGs (ADEH+ vs healthy controls, genome-wide). Many of these associated with severity measures, especially eosinophil count (431/490 sites).

Conclusions

We identified a CpG in IL4 associated with serum tIgE levels, supporting a role for Th2 immune mediating mechanisms in AD. Changes in eosinophil level, a measure of disease severity, are associated with methylation changes, providing a potential mechanism for phenotypic changes in immune response-related traits.

Electronic supplementary material

The online version of this article (10.1186/s13148-019-0714-1) contains supplementary material, which is available to authorized users.

Effect of total glucosides of paeony on the changes of IL-4 and ICAM-1 levels in eczema mouse model serum

Effect of total glucosides of paeony on the changes of IL-4 and ICAM-1 levels in eczema mouse model serum was investigated. A total of 38 KM mice of SPF grade were divided into 3 groups: the control group (n=10), the model group (n=15) and the treatment group (n=13). The pathological model of chronic eczema in mouse right ear was induced using dinitrochlorobenzene acetone solution. Two ears of mice in the control group and the left ear of mice in the model and treatment groups were smeared with acetone as control. The mice in the treatment group were treated by administration with total glucoside of paeony. The changes of IL-4 and ICAM-1 levels were measured using caudal vein blood collection. The mouse ear weight was measured and the relationship among IL-4 and ICAM-1 levels, ear thickness and treatment time was analyzed. Mouse ear thickness in the model group was higher than that in the treatment and control groups (P<0.05). The weight of the mouse right ear in the model and treatment groups was significantly higher than that of the left ear (P<0.05). Furthermore, The IL-4 and ICAM-1 levels of mice in the model group were higher than that in the treatment and control groups (P<0.05). The IL-4 and ICAM-1 levels of mice in the model and treatment groups increased compared to that before modeling (P<0.05). The IL-4 and ICAM-1 levels of mice were positively correlated with ear thickness in the model group (r=0.865, P=0.002; r=0.833, P=0.009). In addition, the IL-4 level of mice was positively correlated with the ICAM-1 level in the model group (r=0.812, P=0.014). Finally, IL-4 and ICAM-1 may be involved in the pathologic process of chronic eczema. Therefore, the study showed that the total glucosides of paeony may play a role in the treatment of chronic eczema by regulating the IL-4 and ICAM-1 levels.

Does epigenetics play a role in human asthma?

Asthma and other allergic diseases are among the most prevalent chronic non-communicable diseases of childhood. According to the World Health Organization, asthma affects >7.0 million children under 18 in the United States, with an economic burden that is estimated to exceed that of tuberculosis and HIV/AIDS combined. Despite much research, the natural history of asthma and its pathogenesis are still in many ways elusive. This review discusses our current understanding of the role epigenetic processes play in asthma pathogenesis, focusing on genome-wide, population-based studies.

Epigenetics in allergic diseases

PURPOSE OF REVIEW

Allergic diseases are among the most prevalent chronic diseases of childhood, affecting more than 7 million children in the United States. Epidemiological evidence supports the idea that the inception of allergic diseases is typically before the preschool years, even when chronic symptoms do not emerge until adulthood. The role of epigenetic mechanisms (particularly DNA methylation) in allergic disease is under active investigation because these mechanisms are known to be at the interface of gene regulation, environmental stimuli, and developmental processes, all of which are essential for the pathogenesis for asthma and allergy. This article specifically reviews genome-wide DNA methylation studies in allergic disease.

RECENT FINDINGS

Differential DNA methylation at specific regions appears to be associated with concurrent allergic disease. A few studies have identified methylation signatures predictive of disease.

SUMMARY

DNA methylation signatures have been shown to be associated with several allergic disease phenotypes, typically concurrently with disease. The few that have been found to precede diagnosis are especially interesting because they highlight an early trajectory to disease.