Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Aug 1.
Published in final edited form as: J Allergy Clin Immunol. 2016 Dec 21;140(2):534–542. doi: 10.1016/j.jaci.2016.10.041

Epigenome-wide Analysis Links SMAD3 Methylation at Birth to Asthma in Children of Asthmatic Mothers

Avery DeVries 1,2, Gabriela Wlasiuk 1, Susan J Miller 3, Anthony Bosco 4, Debra A Stern 1, I Carla Lohman 1, Janet Rothers 5, Anya C Jones 4, Jessie Nicodemus-Johnson 6, Monica M Vasquez 1, John A Curtin 7, Angela Simpson 7, Adnan Custovic 7, Daniel J Jackson 8, James E Gern 8, Robert F Lemanske Jr 8, Stefano Guerra 1,9, Anne L Wright 1,10, Carole Ober 6, Marilyn Halonen 1,11,12,13, Donata Vercelli 1,2,12,13
PMCID: PMC5479765  NIHMSID: NIHMS838362  PMID: 28011059

Abstract

Background

The timing and mechanisms of asthma inception remain imprecisely defined. Although epigenetic mechanisms likely contribute to asthma pathogenesis, little is known about their role in asthma inception.

Objective

To assess whether the trajectory to asthma begins already at birth and epigenetic mechanisms, specifically DNA methylation, contribute to asthma inception.

Methods

We used Methylated CpG Island Recovery Assay (MIRA)-chip to survey DNA methylation in cord blood mononuclear cells (CBMC) from 36 children (18 non-asthmatic, 18 asthmatic by age 9) from the Infant Immune Study (IIS), an unselected birth cohort closely monitored for asthma for a decade. SMAD3 methylation in IIS (n=60) and in two replication cohorts (The Manchester Asthma and Allergy Study, n=30, and the Childhood Origins of ASThma Study, n=28) was analyzed by bisulfite sequencing or Illumina 450K arrays. CBMC-derived IL-1β was measured by ELISA.

Results

Neonatal immune cells harbored 589 differentially methylated regions (DMRs) that distinguished IIS children who did and did not develop asthma by age 9. In all three cohorts, methylation in SMAD3, the most connected node within the network of asthma-associated DMRs, was selectively increased in asthmatic children of asthmatic mothers and was associated with childhood asthma risk. Moreover, SMAD3 methylation in IIS neonates with maternal asthma was strongly and positively associated with neonatal production of IL-1β, an innate inflammatory mediator.

Conclusions

The trajectory to childhood asthma begins at birth and involves epigenetic modifications in immunoregulatory and pro-inflammatory pathways. Maternal asthma influences epigenetic mechanisms that contribute to the inception of this trajectory.

Keywords: epigenetics, DNA methylation, childhood asthma, SMAD3

Graphical Abstract

graphic file with name nihms838362u1.jpg

INTRODUCTION

Asthma is the most prevalent chronic disease of childhood (1). Epidemiological evidence suggests that the disease often begins during the pre-school years even when chronic symptoms appear much later in life (2). However, firm criteria to pinpoint how early a child’s trajectory to asthma truly begins are currently lacking. The mechanisms underlying asthma inception also remain largely unknown. Subtle modifications of both innate and adaptive immune responses accompany and often precede the diagnosis of childhood asthma (3, 4), consistent with the notion that immune and respiratory alterations at an early window of susceptibility converge to place the child on a path to the disease. Genome-wide association studies (GWAS) have identified multiple genetic variants that influence asthma susceptibility (5) but have accounted for only a modest proportion of the total phenotypic variance, providing a compelling rationale for seeking additional risk factors for asthma. In this context, epigenetic mechanisms are especially worth investigating because environmental and developmental influences are essential for asthma pathogenesis (1), and epigenetic processes ensure the timed unfolding of developmental programs and plastic responses to environmental cues, including those delivered in utero by the maternal milieu (6).

Little is known about the role of epigenetic mechanisms in childhood asthma (7). A recent epigenome-wide study compared DNA methylation patterns in peripheral blood mononuclear cells from 6–12 year old inner-city children with persistent atopic asthma and healthy controls, and found that several immune genes involved in T cell maturation, Th2 immunity and oxidative stress were hypomethylated in asthmatic children (8). While these results are novel, their significance remains unclear because it is difficult to determine whether epigenetic alterations concurrent with asthma are a cause or a consequence of the disease. Moreover, a cross-sectional study cannot provide insights into the timing and mechanisms of asthma inception. A recent candidate gene study pointed to an association between IL2 promoter methylation at birth and asthma exacerbations during childhood (9), but relevant pathways were not further interrogated.

In an attempt to define when the trajectory to asthma begins and which pathways are involved, we performed an epigenome-wide search for DNA methylation signatures associated with childhood asthma in cord blood mononuclear cells (CBMC) from children enrolled in the Infant Immune Study (IIS). In this unselected birth cohort, the development of asthma and immune responses was monitored at multiple times from birth to age 9 (4, 10). We reasoned that the detection at birth of differentially methylated regions (DMRs) associated with asthma during childhood would both support a perinatal origin of the disease and highlight epigenetic mechanisms potentially contributing to asthma inception. We show herein that DNA methylation signatures associated with asthma during childhood were indeed present in neonatal blood immune cells, and clustered in immunoregulatory and pro-inflammatory pathways. Moreover, hypermethylation of the SMAD3 promoter was selectively detected in asthmatic children of asthmatic mothers and was associated with risk of childhood asthma in the IIS population and in two comparable birth cohorts.

METHODS

Study design and participants

The IIS unselected birth cohort includes 482 children and was designed to assess patterns of immune maturation in early life and their impact on asthma risk (4, 10). At enrollment, parents completed a respiratory health history questionnaire, a cord blood sample was obtained, and their child’s health was followed prospectively. Childhood asthma was defined as physician-diagnosed, with symptoms or medication use for asthma in the past year reported at least once on the age 2, 3, 5 or 9-year questionnaires.

Figure 1 shows an overview of our study design. We performed a nested case-control, epigenome-wide study of DNA methylation in CBMC (2–8×106 cells per sample) from a discovery population of 36 children from the IIS. This population was randomly selected among available samples but balanced for asthma status in the child (18 non-asthmatic, 18 asthmatic by age 9) and an effort was made to balance cases and controls by maternal asthma (Table E1). Additional targeted analysis of DNA methylation was performed in the 31 samples available from the IIS discovery population and in samples from 29 additional IIS children of comparable characteristics (Table E2). The distribution of cases and controls is provided in Figure 1. Availability of adequate samples was the only inclusion criterion besides asthma status of the child and her/his mother. Overall, the study population did not differ from the rest of the IIS population except for a greater proportion of asthma and maternal asthma, as per study design (Table E1).

Figure 1. Overview of study design.

Figure 1

Open in a new tab

IIS: Infant Immune Study, MAAS: Manchester Asthma and Allergy Study, COAST: Child Origins of ASThma study. NN: Non-asthmatic child with a Non-asthmatic mother, NA: Non-asthmatic child with an Asthmatic mother, AN: Asthmatic child with a Non-asthmatic mother, AA: Asthmatic child with an Asthmatic mother.

The results of targeted DNA methylation analysis performed in the IIS were replicated in CBMC from 30 children with asthmatic mothers from the Manchester Asthma and Allergy Study (MAAS) and 28 children from the Childhood Origins of ASThma (COAST) study (Figure 1). The characteristics of the MAAS and COAST study populations are described in Table E2. The MAAS unselected birth cohort includes 1085 children monitored for asthma and allergy from birth to age 11 years. An asthma diagnosis required at least one of the following criteria reported on age 5 or 8 questionnaires: 1) physician diagnosis of asthma, 2) the use of asthma medications during the previous 12 months. Controls required a negative report on both of these criteria and no report of wheezing in the previous 12 months (11). The COAST birth cohort enrolled 289 neonates at risk for asthma and allergy, i.e., having at least one parent with asthma and/or allergies. Asthma was diagnosed at age 6 years based on the documented presence of one or more of the following characteristics in the previous year: 1) physician diagnosis of asthma, 2) use of physician-prescribed albuterol for coughing or wheezing episodes, 3) use of a daily controller medication, 4) step-up plan including use of albuterol or short-term use of inhaled corticosteroids during illness, and 5) use of prednisone for asthma exacerbations (12). All three studies were approved by the appropriate Institutional Review Boards. Informed consent was obtained from the parents of all research participants.

DNA methylation profiling and DMR identification

DNA methylation was profiled by Methylated CpG Island Recovery Assay (MIRA)-chip (Roche-NimbleGen), as detailed in the Online Repository, Methods section, and Figure E1. DNA methylation microarray data from this publication were submitted to the NCBI Gene Expression Omnibus (GEO) database and assigned the identifier GSE85228.

Regions that were differentially methylated between asthmatics and non-asthmatics (DMRs) were identified using a Probe Sliding Window-ANOVA [Roche-NimbleGen (13)] that detects maximal inter-group differences in signal intensities relative to a user-defined threshold. Probability scores (P-values) are then assigned to each probe on the array using a repeated measure ANOVA model. In this study, DMRs were defined by region length ≥300 bp, magnitude (between-group mean log2 ratio difference) ≥0.2, significance threshold (adjusted using the Benjamini-Hochberg False Discovery Rate method to account for multiple testing) = 0.01, using a sliding window size = 750 bp (Table E3). A positive magnitude difference indicated hypermethylation in asthmatics relative to non-asthmatics. DMRs containing ≤5 CpG sites within a 750 bp window centered on the DMR [a configuration likely to result in sub-optimal capture of methylated DNA (14)] were excluded from further analyses.

Results of this analysis were technically validated by bisulfite sequencing (Online Repository, Methods section and Table E4). Microarray-derived estimates of DNA methylation intensity strongly correlated with DNA methylation levels measured by bisulfite sequencing over the entire DNA methylation range [Spearman correlation coefficient (ρ)=0.48, P=1.2×10−13] and at intermediate DNA methylation levels (8–92%, P=0.006; Figure E2). SMAD3 methylation estimates from bisulfite sequencing and microarrays were also highly correlated (ρ=0.46, P=0.009).

Functional DMR annotation is fully described in the Online Repository, Methods section. In brief, DMRs were annotated to genes based on the closest transcription start site. For pathway analysis, multiple genes were allowed to be associated with a single DMR (RefSeq genes +/− 5 kb).

DNA methylation analysis in the COAST cohort was performed using the Infinium HumanMethylation450 BeadChip array (Illumina) (Online Repository, Methods section).

Network analysis

A molecular interaction network was constructed by uploading the list of DMR-containing genes into the Ingenuity Pathway Analysis software (Qiagen) and using all available interaction data in the Ingenuity Systems Knowledge Base. Genes with no interactions were removed from the analysis to maximize the signal-to-noise ratio. The gene interaction network was then interrogated to detect enrichment for biological functions, canonical pathways and upstream regulators (defined as an enrichment for known targets of a given gene in a given gene list) using Ingenuity Pathway Analysis tools.

Cytokine measurements

IL-1β concentrations were measured by ELISA (Quantikine, R&D Systems) in the supernatants of LPS-stimulated IIS CBMC (n=57) also tested for SMAD3 promoter methylation.

Statistical analyses

DMRs were detected using a Probe Sliding Window-ANOVA which uses a repeated measure ANOVA model for the probes in each sliding window (13). Fisher’s Exact Test and one sample test of proportions were used to compare proportions for categorical variables, and Student’s t-test or Wilcoxon two-sample test were used to compare mean levels for continuous variables. Variables with skewed distributions were log-transformed prior to assessment by t-test. Two-sided P-values less than 0.05 were considered significant. Spearman correlation was used to test for association between median DNA methylation intensity (microarray) and percentage DNA methylation (bisulfite sequencing). When analyzing the genomic locations of DMRs and the co-localization of DMRs and DNase I hypersensitive sites, permutations were performed by randomly selecting 589 independent probes (the number of DMRs) and calculating the χ2 statistic for each sampling. We recorded the number of times (out of 50,000) that the permuted χ2 statistic was larger than the observed χ2 statistic and divided by the number of permutations to get the empirical P-value. Linear regression was used to test for an interaction between child asthma and maternal asthma on SMAD3 methylation. Pearson’s χ2 test was used to identify significant associations between child and parental characteristics at birth and asthma during childhood. Meta-analysis of the association between SMAD3 CpG7 methylation and childhood asthma risk in neonates born to asthmatic mothers was performed using estimates from each study cohort to compute the combined estimate of risk.

RESULTS

Cord blood cells from IIS children harbor asthma-associated DMRs

Our multi-step study of the contribution of epigenetic mechanisms to the development of asthma in early life included: 1) a discovery phase in which an epigenome-wide approach was used to identify candidate regions that were differentially methylated in cord blood samples from 36 IIS children (18 non-asthmatic, 18 asthmatic by age 9); 2) a targeted analysis phase in which DNA methylation of the most compelling candidate region was measured in CBMC from 29 IIS neonates who did, and 31 IIS neonates who did not develop asthma by age 9; and 3) a final phase in which the results of the targeted analysis performed in IIS were replicated in two comparable birth cohorts, MAAS and COAST (Figure 1).

Relying on the longitudinal design of the IIS birth cohort, we initially searched for neonatal epigenetic signatures of childhood asthma by profiling DNA methylation in CBMC from a discovery cohort of 36 children (Figure 1, Figure E1 and Table E1). Five hundred and eighty-nine independent regions were differentially methylated in asthmatic and non-asthmatic children (Table E3). Among these, 199 were hypermethylated and 390 were hypomethylated at birth in children who became asthmatic.

Asthma-associated DMRs were distributed across all chromosomes but were non-randomly distributed with respect to genome location (χ2= 19.49, df = 7, P=0.007), with an overrepresentation at intergenic regions (standardized residual: 2.91) and an underrepresentation at transcription start sites (standardized residual: −2.23, with absolute values > 1.96 significantly contributing to the overall χ2 test statistics) (Figure E3). Of the DMRs with the most significant methylation differences, some mapped to biologically plausible genes such as ATG9A (involved in autophagy and innate immune responses (15); −log10 adjusted P-value 5.569), OR6K6 (an olfactory receptor expressed in sputum during asthma exacerbations (16); −log10 adjusted P-value 4.408), and GALNT2 (an N-acetylgalactosaminyltransferase carrying polymorphisms associated with lung function (17); −log10 adjusted P-value 4.895). Moreover, several asthma-associated DMRs mapped to loci (+/− 5kb from transcription start sites) identified in GWAS for asthma (RORA, SMAD3) and asthma-related traits such as atopic dermatitis (FLG), allergic rhinitis (TMEM232), lung function (ANK1, DLEU7, SNRPN, RORA, CFDP1) and airflow obstruction (SPATA13) (http://www.ebi.ac.uk/gwas). Functional annotation using Ingenuity Pathway Analysis linked DMR-associated genes to biological processes including immune function and immune and lung development (Table E5). We also mapped the locations of neonatal asthma-associated DMRs to DNase I hypersensitive sites, regions of increased chromatin accessibility typically endowed with regulatory activity (18). Overall, 72 asthma-associated DMRs mapped to DNase I hypersensitive sites (29 monocyte-specific, 15 T cell-specific, 28 in both cell types: Table E6), which represented a significant enrichment (Table E7, permutation P-value < 2×10−5 for both cell types). The co-localization of asthma-associated differential DNA methylation and enhanced chromatin accessibility supports the potential biological significance of asthma-associated DMRs.

Asthma-associated DMRs cluster in regulatory and pro-inflammatory gene networks

To elucidate the functional implications of our findings and gain a pathway- based view of asthma-associated methylome alterations at birth, Ingenuity Pathway Analysis tools were used to construct a molecular interaction network of DMR- associated genes on the basis of prior knowledge of the physical and functional connections between the molecules encoded by those genes. The network included 146 genes organized around several nodes, the most connected of which was the transcription factor SMAD3 (Figure 2). When the network was interrogated to detect enrichment for known targets of upstream regulators, IL1B, an innate pro-inflammatory cytokine overexpressed in asthmatics (19–21), emerged as the top regulator of a gene subset that included RORA and RELB, transcription factors essential for innate and adaptive responses, UBD, an innate immunity gene regulated by the NF-kB inhibitor A20/TNFAIP3 (22), and the asthma-associated neurotrophin BDNF (19) (overlap P=2.26×10−5, Table 1).

Figure 2. Molecular interaction network of asthma-associated DMRs.

Figure 2

Open in a new tab

The network was constructed using all available interaction data in the Ingenuity Systems Knowledge Base. The most highly connected gene (SMAD3, 17 connections) is highlighted in black. Targets of IL1B, the top upstream regulator of network genes, are highlighted in gray. Types of interactive molecules are defined in the Legend.

Table 1.

Upstream regulators of genes containing asthma-associated DMRs

Upstream Regulator Molecule Type Overlap P* Target molecules in dataset
IL1B cytokine 2.26E-05 ACAN, BDNF, CFTR, COL10A1, CXCL10, CYP7A1, FGFR3, GNAS, LDHA, NCOA2, NEUROD1, PFKP, RELB, RORA, SCX, SNCA, UBD
HSPG2 enzyme 8.60E-05 ACAN, COL10A1, FGFR3
BDNF growth factor 1.43E-04 BDNF, DUSP6, NRXN1, PRKCZ, PTBP1, RPL35A, SLC17A8, TMEM45A, TTC3
SNCA enzyme 1.56E-04 BDNF, CFL1, DNM1, SNCA, SYN3
PRMT5 enzyme 3.51E-04 CXCL10, CYP7A1, FASN
Open in a new tab
*

Enrichment for known targets of a given gene in a given gene list. The Table shows the top hits provided by this analysis.

The SMAD3 promoter is significantly hypermethylated in asthmatic children of asthmatic mothers

SMAD3 is not only the most connected node in the network of asthma-associated DMRs (Figure 2), but is also a well-replicated asthma-associated gene from GWAS (23–25). Moreover, SMAD3 acts as a master regulator of TGF-β signaling, thereby controlling the differentiation of Treg and Th17 cells that play critical and opposite roles in asthma (1, 26). Finally, the SMAD3 DMR lies within the distal promoter of the gene, a location that provides ample opportunity for DNA methylation to influence SMAD3 gene expression (Figure E4). Therefore, subsequent analyses of the nexus between neonatal DNA methylation and trajectory to childhood asthma specifically targeted SMAD3. We used bisulfite sequencing to precisely quantify methylation at the SMAD3 promoter DMR (321 bp, 8 consecutive CpG sites) in a total of 60 IIS neonates (31 non-asthmatics, 29 asthmatics: Figure 1). Mean SMAD3 methylation levels at birth were 41.3% (95% CI 35.5–47.5) in non-asthmatics and 47.1% (95% CI 40.6–53.1) in asthmatics, a difference that did not reach statistical significance (P=0.2 by Wilcoxon two-sample test: Figure E5). Because these results may have reflected heterogeneity within the study population, we next examined the entire IIS population for associations between childhood asthma and potential risk factors measurable at birth in the child (sex, ethnicity, mode of delivery, total cord IgE, 17q21 rs8076131 genotype) and parents (maternal asthma, maternal allergy, paternal asthma, paternal allergy, maternal smoking during pregnancy). As shown in Table E8, maternal asthma exhibited a distinctive association with childhood asthma in IIS (P=0.003 by χ2 test). When the relation between SMAD3 methylation at birth and childhood asthma was examined separately in children with and without maternal asthma, SMAD3 methylation was found to be significantly increased in asthmatic compared to non-asthmatic children of asthmatic mothers (P=0.005 by Wilcoxon two-sample test). In contrast, asthmatic and non-asthmatic children of non-asthmatic mothers did not differ in their SMAD3 methylation levels [P for interaction (by linear regression) = 0.001; Figure 3A].

Figure 3. Effects of maternal asthma n the association between neonatal SMAD3 methylation and childhood asthma in IIS (panel A), MAAS (panel B) and COAST (panel C).

Figure 3

Open in a new tab

SMAD3 methylation In IIS and MAAS was assessed by bisulfite sequencing and expressed as mean percent DNA methylation across 8 consecutive CpG sites in the SMAD3 DMR. In COAST, percent SMAD3 methylation at cg02486855, the seventh CpG in the SMAD3 DMR, was assessed by the Infinium HumanMethylation450 BeadChip array. N: non-asthmatic, A: asthmatic. P-values by Wilcoxon two-sample test. The results presented in panel A were unaffected after adjusting for ethnicity.

Although associations between SMAD3 variants [rs17228058 (25), rs744910 (23), rs17294280 (24)] and asthma have been reported in GWAS, asthma-related SMAD3 methylation differences were unlikely to be influenced by SMAD3 genotype. Indeed, sequencing identified no polymorphisms within the SMAD3 DMR (data not shown). Moreover, asthma-associated SMAD3 variants are located at least 80 kb away from the SMAD3 DMR, whereas the relationship between DNA methylation and genetic variation appears to decay rapidly beyond 5 kbs (27).

The association between SMAD3 promoter hypermethylation and childhood asthma replicates in two independent birth cohorts

The association between neonatal SMAD3 hypermethylation and childhood asthma found in IIS neonates born to asthmatic mothers was assessed for replication in two birth cohorts, MAAS and COAST (Table E2). Bisulfite sequencing of the SMAD3 DMR in 30 MAAS children born to asthmatic mothers revealed significantly higher DNA methylation in the asthmatic group (P=0.049 by Wilcoxon two-sample test; Figure 3B). Comparable results were obtained by assessing SMAD3 methylation in 28 COAST neonates using the Illumina Infinium HumanMethylation450 BeadChip array. Figure 3C shows that methylation at cg02486855, the seventh CpG in the SMAD3 DMR and the only SMAD3 DMR CpG interrogated on the Illumina platform, did not significantly differ between asthmatic and non-asthmatic children of non-asthmatic mothers (P=0.4 by Wilcoxon two-sample test). In contrast, significant cg02486855 hypermethylation was detected in neonates born to asthmatic mothers who developed asthma during childhood, compared to neonates who did not (P=0.04 by Wilcoxon two-sample test; Figure 3C). The data generated by bisulfite sequencing and on the Illumina platform were comparable because when percent methylation values for SMAD3 CpG7 were extracted from the bisulfite sequencing data for IIS and MAAS, again we found significant hypermethylation in asthmatic compared to non-asthmatic children of asthmatic mothers in IIS (P=0.004 by Wilcoxon two-sample test; Figure E6), and a difference approaching significance in MAAS (P=0.09 by Wilcoxon two-sample test; Figure E6).

Cord blood contains a mixture of cell types with potentially distinct DNA methylation profiles. The availability of COAST DNA methylation data generated on the Illumina450 BeadChip array allowed us to estimate cord blood cell proportions using an algorithm recently developed that integrates Illumina DNA methylation data with information from a cord blood reference panel (28). Figure E7 shows that cell proportions thus estimated did not significantly differ between asthmatic and non-asthmatic children, regardless of maternal asthma history. Most importantly, the SMAD3 methylation differences detected between asthmatic and non-asthmatic children of asthmatic mothers remained significant (P=0.04 by Wilcoxon two-sample test) even after adjusting for CBMC composition (Figure E8).

Finally, we asked whether the SMAD3 methylation levels measured at birth in children of asthmatic mothers are associated with risk for childhood asthma in our three cohorts. A meta-analysis revealed that for each 10% increase in SMAD3 CpG7 methylation there is nearly a two-fold increased risk of childhood asthma (meta-analysis Odds Ratio=1.95, [95%CI: 1.23, 3.10],P=0.005; heterogeneity P=0.5).

Maternal asthma modifies the relation between neonatal SMAD3 methylation and IL-1ββ β producing capacity

SMAD3 knock-out mice exhibit increased expression of Il1b (29), which is also the top upstream regulator of genes containing asthma-associated DMRs in this study and a critical pro-inflammatory mediator in human asthma (19–21). Therefore, we relied again on the IIS to explore the relationship between neonatal LPS-induced IL-1β production, SMAD3 promoter methylation and childhood asthma. The distributions of SMAD3 methylation and log IL-1β production in non-asthmatic and asthmatic children without and with a maternal history of asthma were compared by dividing mean percentage SMAD3 methylation and log IL-1β secretion at the median, thereby creating four groups (low/low, low/high, high/low and high/high). No distribution differences were observed among children of non-asthmatic mothers (P=0.79 by Fisher’s sExact Test; Figure 4A). In contrast, the children of asthmatic mothers who developed asthma were almost entirely found among those with high SMAD3 promoter methylation and high IL-1β secretion (P=0.009 by Fisher’s Exact Test; Figure 4B). Moreover, asthmatic children of asthmatic mothers had higher LPS-induced IL-1β than non-asthmatic children (P=0.03 by Student’s t test), whereas comparable IL-1β levels were measured in children without a maternal history of asthma (P=0.65 by Student’s t test; Table 2). These results suggest that a strong relationship exists between neonatal SMAD3 methylation, production of the innate cytokine IL-1β, and childhood asthma. Furthermore, our data suggest that this relationship is powerfully influenced by maternal asthma.

Figure 4. Effects of maternal asthma on the relation between neonatal SMAD3 methylation and IL-1β protein production.

Figure 4

Open in a new tab

The distributions of SMAD3 methylation and log IL-1β producing capacity in non-asthmatic and asthmatic children without and with a maternal history of asthma were compared by dividing mean percentage SMAD3 methylation and log IL-1β secretion at the median, thereby creating four groups (low/low, low/high, high/low and high/high). Distributions for asthmatics and non-asthmatics were compared across all four quadrants (panel A) or focusing the analysis on the high/high and low/low quadrants (panel B) (P by Fisher’s Exact Test). N: non-asthmatic, A: asthmatic.

Table 2.

LPS-stimulated CBMC-production of IL-1β in IIS children with or without a history of maternal asthma.

Children with Asthmatic Mothers Children with Non-asthmatic Mothers
Asthmatic* (n=7) Non-asthmatic (n=12) Asthmatic* (n=19) Non-asthmatic (n=16)
Geometric Mean (ng/ml) 16.5 6.0 6.7 7.7
95% CI (ng/ml) 6.9–39.4 4.8–7.6 4.1–10.9 4.7–12.5
P† 0.03 0.65
Open in a new tab
*

Physician-diagnosed with symptoms or medication use for asthma in the past year reported at least once on the age 2, 3, 5 or 9-year questionnaires

†

by Student's t-test after log10 transformation.

DISCUSSION

Although asthma is the most common chronic complex disease of childhood (1), the timing and mechanisms of its inception remain largely unknown. This gap in knowledge severely hinders efforts aimed at preventing this disease, which no therapeutic regimen can currently cure. Our study, the first epigenome-wide search for asthma-associated methylome signatures at birth, sheds new light on these critical but still open questions. The finding that almost 600 genomic regions were differentially methylated at birth between children who did and did not develop asthma later in life strongly suggests that the trajectory to asthma begins at birth if not prenatally and involves epigenetic mechanisms. As importantly, data from three birth cohorts showed that SMAD3 promoter hypermethylation was associated with childhood asthma selectively in neonates with a maternal history of asthma - one of the strongest, and mechanistically one of the most elusive, risk factors for asthma in the child (30, 31). These findings are especially noteworthy in the context of the current dearth of characteristics measurable at birth that effectively predict asthma during childhood. Of note, the link between epigenetic SMAD3 dysregulation at birth and childhood asthma appeared to be selective for neonates born to asthmatic mothers, a finding that suggests that heterogeneity is deeply embedded in the pathogenesis of, and the trajectory to, childhood asthma (1, 32). More generally, to our knowledge this is the first time that a neonatal epigenetic characteristic linked to asthma during childhood is robust enough to replicate across three independent birth cohorts.

Our study highlighted SMAD3 and IL-1β as main players in the trajectory to childhood asthma. SMAD3, the most connected node in the network of asthma-associated DMRs, is known to be critical for the regulation of both the asthma-protective Treg and the asthma-promoting Th17 cell differentiation programs. Altered Treg and Th17 activities have been reported in childhood asthma (1, 26) and conversely, maternal exposure to asthma-protective environments such as farming has been shown to activate the Treg compartment (33) and influence the expression of Th17 markers (34). On the other hand, IL-1β, which emerged as the primary upstream regulator of genes harboring asthma-associated DMRs, is increasingly recognized as a key asthma mediator in both children (20) and adults, especially in neutrophilic asthma (19, 21). Our data emphasize the functional connection between SMAD3 and IL-1β. Indeed, in neonates who became asthmatic by age 9, SMAD3 promoter hypermethylation, an epigenetic configuration consistent with low SMAD3 expression, was strongly associated with high IL-1β production. This convergence is likely to destabilize the Treg program, enhance inflammation, and promote Th17 differentiation (35), ultimately favoring the development of asthma. While it is unclear whether these mechanisms operate pre- and/or perinatally, detection of the relationship between SMAD3 methylation and IL-1β production selectively among children of asthmatic mothers implies that the in utero environment is critical for directing the epigenetic trajectory towards childhood asthma.

Our results should be interpreted with caution because our discovery population was small, albeit longitudinally phenotyped for asthma in great detail, and environmental exposures were not comprehensively assessed. Moreover, DNA methylation did not distinguish between 5-methylcytosine and other cytosine modifications and was assessed in mixed rather than isolated cell populations. However, our data from the COAST population suggest that differential methylation by asthma status did not reflect asthma-associated differences in CBMC proportions. We also acknowledge that the DNA methylation differences we detected were not extreme, albeit more substantial than those recently reported in other studies (8, 36). This is a recurring theme in the literature (37), and systematic studies are needed to comprehensively assess how relatively modest DNA methylation differences modify disease trajectories. In general, the functional impact of such differences will likely depend on the regulatory properties of the locus in which they reside, and the extent to which additional epigenetic processes, such as post-translational histone modifications, influence those properties.

Gene expression analyses are often used to complement epigenetic studies, but samples for such analyses were not collected at birth in our study populations. On the other hand, samples for cytokine protein assessments were available and proved essential to integrate our epigenetic findings. Finally, our search for asthma-associated differential methylation returned almost 600 regions, only some of which mapped to genes involved in immunoregulation and inflammation. DMRs that reside in functionally interesting genes but lack a link to immune regulation or inflammation may also contribute to asthma pathogenesis, albeit through different mechanisms. With these caveats, we propose that in a proportion of children with childhood-onset asthma, a distinctive methylome is in place already at birth, particularly within innate immunoregulatory and pro-inflammatory pathways, and promotes a trajectory that may ultimately lead to clinical disease. Some of the epigenetic mechanisms that contribute to the inception of this trajectory are strongly influenced by the milieu associated with maternal asthma. A scenario in which epigenetic modifications at an early window of susceptibility promote a long-term developmental trajectory to asthma is consistent with the emerging paradigm that chronic non-communicable diseases have their origins in early life through an epigenetic calibration of set points for later responsiveness and function (38, 39).

Supplementary Material

1. Figure E1. Workflow of DNA methylation profiling.

Genomic DNA was fragmented by sonication and divided into two aliquots: one underwent methyl-binding protein methylation capture (MBDCap) and was enriched for methylated DNA, while the other was left untreated. The two fractions were then whole genome-amplified, fluorescently labeled [captured (enriched)=red, untreated (input)=green], mixed and hybridized to Human DNA Methylation 2.1M Deluxe Promoter Arrays (Roche NimbleGen). After scanning the arrays, the hybridization intensities for each probe were extracted separately for the red and green channels, and were quantile-normalized to remove technical noise. The difference between normalized intensities [log2 enriched – log2 input = log2 (enriched/input), referred to as the log2 ratio] was then calculated and used as a measure of methylation levels. Measurements for 16 genomic regions were validated by bisulfite sequencing. Regions differentially methylated in the asthmatic and non-asthmatic groups were detected using Roche-Nimblegen scripts that implement a probe sliding-window ANOVA. DMRs were functionally annotated using Homer and Ingenuity Pathway Analysis.

10. Table E1.

Characteristics of IIS study subjects.

11. Table E2.

Characteristics of non-asthmatic and asthmatic children in the IIS, MAAS and COAST study populations.

12. Table E3.

Complete list of annotated asthma-associated DMRs (n=589) in IIS children*

13. Table E4.

Primers used for bisulfite sequencing.

14. Table E5.

Functional annotation of asthma-associated DMRs in IIS children.

15. Table E6.

Overlap between asthma-associated DMRs in IIS children and DNase I HSS in monocytes and naïve T cells.

16. Table E7.

Asthma-associated DMRs in IIS children and DNase I HSS

17. Table E8.

Association between child and parental risk factors at birth and asthma during childhood in IIS children.

2. Figure E2. Correlation of bisulfite sequencing and DNA methylation microarray results.

Each dot corresponds to the median normalized log2 ratios of enriched/input sample across probes (for microarray data) or the median percentage DNA methylation across CpG sites (for bisulfite sequencing data) for each region in each subject. The correlation between median normalized log2 ratios from microarray analysis and median percent DNA methylation from bisulfite sequencing was assessed by measuring the Spearman correlation coefficient (ρ).

3. Figure E3. Distribution of asthma-associated DMRs by chromosome and genomic location.

A) Manhattan plot of asthma-associated DMRs. The x axis depicts the chromosomal location of each DMR. The y-axis shows the Benjamini Hochberg (BH)-adjusted P-values from the probe sliding window ANOVA. The minimum value of the y-axis was set at 2 because a significance threshold of 0.01 was imposed on the adjusted P-values. Genes associated with top hits are labeled. B) Genomic location of asthma-associated DMRs. The annotation is based on the center of the DMR, as defined by the Homer package. The genomic distribution of DMRs differed significantly from the distribution of probes across the entire array (χ2= 19.49, df = 7, P-value = 0.007). Fifty-thousand permutations were performed to assess whether this distribution differed from random expectations (permutation P-value = 0.008). Standardized residuals were computed for each category in order to assess which genomic locations contributed to the overall statistical significance. Absolute values > 1.96 significantly contribute to the overall χ2 test statistics. UTR: untranslated region, TSS: transcription start site (−1kb to +100 bp), TTS: transcription termination site (−100 bp to +1kb).

4. Figure E4. Regulatory potential of the SMAD3 DMR.

The SMAD3 DMR was mapped to the human genome (hg19/GRCh37) using University of California Santa Cruz (UCSC) Genome Browser tools. Tracks are specified on the left. UCSC genes are shown in blue. Nearby CpG islands (with identifier) are shown in green. DNase I hypersensitive sites (HSS) identified in CD14+ monocytes and CD4+ naïve T cells are shown in red and blue, respectively. Post-translational histone modifications (H3K27Ac, H3K4Me1, and H3K4Me3) identified in CD14+ monocytes, CD4+ naïve T cells, and peripheral blood mononuclear cells (PBMC) (Roadmap Epigenomics Project) mark poised/active regulatory regions (E7). The bottom track shows common SNPs (MAF > 10%) from the HapMap CEU population, none of which are associated with asthma. The genomic region that surrounds the SMAD3 DMR (+/− 100 bp) is highlighted in red. The figure covers 5 kb upstream and downstream of the SMAD3 DMR.

5. Figure E5. Association between SMAD3 DNA methylation at birth and childhood asthma in the IIS cohort.

SMAD3 methylation was assessed by bisulfite sequencing and expressed as mean percentage DNA methylation across 8 consecutive CpG sites in the SMAD3 DMR. N: non-asthmatic, A: asthmatic. P-value by Wilcoxon two-sample test.

6. Figure E6. SMAD3 CpG7 methylation in IIS and MAAS.

SMAD3 methylation was assessed by bisulfite sequencing and expressed as percent methylation at SMAD3 CpG7 in neonates from the IIS and MAAS cohorts. N: non-asthmatic, A: asthmatic. P-values by Wilcoxon two-sample test.

7. Figure E7. Estimation of CBMC composition in the COAST cohort.

Proportions of CBMC cell types were estimated using DNA methylation data generated on the Illumina HumanMethylation450 BeadChip and a cord blood reference panel (E11). N: non-asthmatic, A: asthmatic. P-values by Wilcoxon two-sample test.

8. Figure E8. Effects of maternal asthma on the association between neonatal SMAD3 methylation and childhood asthma in COAST after adjusting for CBMC composition.

Residual methylation values for SMAD3 (cg02486855, Illumina HumanMethylation450 BeadChip) were calculated by adjusting for CBMC composition using linear regression. N: non-asthmatic, A: asthmatic. P-values by Wilcoxon two-sample test.

9

KEY MESSAGES.

  • The trajectory to childhood asthma begins at birth and involves epigenetic modifications in innate pro-inflammatory and immunoregulatory pathways.

  • Neonatal immune cells collected at birth harbor differentially methylated regions that distinguish children who will and will not develop asthma by age 9.

  • In three independent birth cohorts, DNA methylation at the SMAD3 promoter was selectively increased in asthmatic children of asthmatic mothers and was associated with risk of childhood asthma.

Acknowledgments

The investigators would like to thank the children and their parents for their continued support and enthusiasm. We greatly appreciate the commitment they have given to the project. We would also like to acknowledge the hard work and dedication of the study team (post-doctoral scientists, research fellows, nurses, physiologists, technicians and clerical staff).

FUNDING: This work was supported by RC1HL100800 and ES006694 (to DV), R01AI042268 (to ALW) and P01HL070831 (to RFL). AB is supported by a BrightSpark Foundation McCusker Fellowship. This report includes independent research supported by National Institute for Health Research Respiratory and Allergy Clinical Research Facility at University Hospital of South Manchester NHS Foundation Trust. The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health. MAAS was supported by the Asthma UK Grants No 301 (1995–1998), No 362 (1998–2001), No 01/012 (2001–2004), No 04/014 (2004–2007), The BMA James Trust, The Moulton Charitable Foundation (2004-current), The Medical Research Council (MRC) Grant G0601361. The authors would like to acknowledge the North West Lung Centre Charity for supporting this project.

ABBREVIATIONS

CBMC

cord blood mononuclear cells

COAST

Childhood Origins of ASThma

DMR

differentially methylated region

GWAS

genome-wide association study

IIS

Infant Immune Study

MAAS

Manchester Asthma and Allergy Study

Footnotes

AUTHOR CONTRIBUTIONS: ADV and GW measured and analyzed DNA methylation. SJM, AB, DAS, JR, JNJ, ACJ and MMV performed statistical and network analyses. ICL generated ELISA data. JAC, AS and AC designed and provided samples from the MAAS study. DJJ, JEG, and RFL designed and provided data from the COAST study. ALW and MH designed and provided samples from the IIS study. CO, SG, and MH contributed to data interpretation. DV designed the study, oversaw data analyses and interpretation, and wrote the paper with ADV. All authors reviewed the final manuscript.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  • 1.Martinez FD, Vercelli D. Asthma. Lancet. 2013;382(9901):1360–72. doi: 10.1016/S0140-6736(13)61536-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Stern DA, Morgan WJ, Halonen M, Wright AL, Martinez FD. Wheezing and bronchial hyper-responsiveness in early childhood as predictors of newly diagnosed asthma in early adulthood: a longitudinal birth-cohort study. Lancet. 2008;372(9643):1058–64. doi: 10.1016/S0140-6736(08)61447-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Halonen M, Lohman IC, Stern DA, Ellis WL, Rothers J, Wright AL. Perinatal TNF-alpha production influenced by maternal pregnancy weight gain, predicts childhood asthma. Am J Respir Crit Care Med. 2013;188(1):35–41. doi: 10.1164/rccm.201207-1265OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Rothers J, Halonen M, Stern DA, Lohman IC, Mobley S, Spangenberg A, et al. Adaptive cytokine production in early life differentially predicts total IgE levels and asthma through age 5 years. J Allergy Clin Immunol. 2011;128(2):397–402. doi: 10.1016/j.jaci.2011.04.044. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Meyers DA, Bleecker ER, Holloway JW, Holgate ST. Asthma genetics and personalised medicine. The lancet Respiratory medicine. 2014;2(5):405–15. doi: 10.1016/S2213-2600(14)70012-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Barua S, Junaid MA. Lifestyle, pregnancy and epigenetic effects. Epigenomics. 2015;7(1):85–102. doi: 10.2217/epi.14.71. [DOI] [PubMed] [Google Scholar]
  • 7.DeVries A, Vercelli D. Early predictors of asthma and allergy in children: The role of epigenetics. Current opinion in allergy and clinical immunology. 2015;15(5):435–9. doi: 10.1097/ACI.0000000000000201. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Yang IV, Pedersen BS, Liu A, O'Connor GT, Teach SJ, Kattan M, et al. DNA methylation and childhood asthma in the inner city. J Allergy Clin Immunol. 2015;136(1):69–80. doi: 10.1016/j.jaci.2015.01.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Curtin JA, Simpson A, Belgrave D, Semic-Jusufagic A, Custovic A, Martinez FD. Methylation of IL-2 promoter at birth alters the risk of asthma exacerbations during childhood. Clinical and experimental allergy : journal of the British Society for Allergy and Clinical Immunology. 2013;43(3):304–11. doi: 10.1111/cea.12046. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Oddy WH, Halonen M, Martinez FD, Lohman IC, Stern DA, Kurzius-Spencer M, et al. TGF-β in human milk is associated with wheeze in infancy. J Allergy Clin Immunol. 2003;112(4):723–8. doi: 10.1016/s0091-6749(03)01941-9. [DOI] [PubMed] [Google Scholar]
  • 11.Deliu M, Belgrave D, Simpson A, Murray CS, Kerry G, Custovic A. Impact of rhinitis on asthma severity in school-age children. Allergy. 2014;69(11):1515–21. doi: 10.1111/all.12467. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Calişkan M, Bochkov YA, Kreiner-M⊘ller E, B⊘nnelykke K, Stein MM, Du G, et al. Rhinovirus wheezing illness and genetic risk of childhood-onset asthma. N Engl J Med. 2013;368(15):1398–407. doi: 10.1056/NEJMoa1211592. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.van Breda SG, Claessen SM, Lo K, van Herwijnen M, Brauers KJ, Lisanti S, et al. Epigenetic mechanisms underlying arsenic-associated lung carcinogenesis. Archives of toxicology. 2015;89(11):1959–69. doi: 10.1007/s00204-014-1351-2. [DOI] [PubMed] [Google Scholar]
  • 14.Yegnasubramanian S, Lin X, Haffner MC, DeMarzo AM, Nelson WG. Combination of methylated-DNA precipitation and methylation-sensitive restriction enzymes (COMPARE-MS) for the rapid, sensitive and quantitative detection of DNA methylation. Nucleic Acids Res. 2006;34(3):e19. doi: 10.1093/nar/gnj022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Saitoh T, Fujita N, Hayashi T, Takahara K, Satoha T, Leea H, et al. Atg9a controls dsDNA-driven dynamic translocation of STING and the innate immune response. Proc Natl Acad Sci U S A. 2009;106(49):20842–6. doi: 10.1073/pnas.0911267106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Bosco A, Ehteshami S, Stern DA, Martinez FD. Decreased activation of inflammatory networks during acute asthma exacerbations is associated with chronic airflow obstruction. Mucosal Immunol. 2010;3(4):399–409. doi: 10.1038/mi.2010.13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Pendergrass SA, Brown-Gentry K, Dudek S, Frase A, Torstenson ES, Goodloe R, et al. Phenome-wide association study (PheWAS) for detection of pleiotropy within the Population Architecture using Genomics and Epidemiology (PAGE) Network. PLoS genetics. 2013;9(1):e1003087. doi: 10.1371/journal.pgen.1003087. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Maurano MT, Humbert R, Rynes E, Thurman RE, Haugen E, Wang H, et al. Systematic localization of common disease-associated variation in regulatory DNA. Science. 2012;337(6099):1190–5. doi: 10.1126/science.1222794. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Hastie AT, Moore WC, Meyers DA, Vestal PL, Li H, Peters SP, et al. Analyses of asthma severity phenotypes and inflammatory proteins in subjects stratified by sputum granulocytes. The Journal of allergy and clinical immunology. 2010;125(5):1028–36. e13. doi: 10.1016/j.jaci.2010.02.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Raedler D, Ballenberger N, Klucker E, Bock A, Otto R, Prazeres da Costa O, et al. Identification of novel immune phenotypes for allergic and nonallergic childhood asthma. J Allergy Clin Immunol. 2014 doi: 10.1016/j.jaci.2014.07.046. [DOI] [PubMed] [Google Scholar]
  • 21.Simpson JL, Phipps S, Baines KJ, Oreo KM, Gunawardhana L, Gibson PG. Elevated expression of the NLRP3 inflammasome in neutrophilic asthma. Eur Respir J. 2014;43(4):1067–76. doi: 10.1183/09031936.00105013. [DOI] [PubMed] [Google Scholar]
  • 22.Damrauer SM, Studer P, da Silva CG, Longo CR, Ramsey HE, Csizmadia E, et al. A20 modulates lipid metabolism and energy production to promote liver regeneration. PLoS One. 2011;6(3):e17715. doi: 10.1371/journal.pone.0017715. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Moffatt MF, Gut IG, Demenais F, Strachan DP, Bouzigon E, Heath S, et al. A large-scale, consortium-based genomewide association study of asthma. N Engl J Med. 2010;363(13):1211–21. doi: 10.1056/NEJMoa0906312. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Ferreira MA, Matheson MC, Tang CS, Granell R, Ang W, Hui J, et al. Genome-wide association analysis identifies 11 risk variants associated with the asthma with hay fever phenotype. J Allergy Clin Immunol. 2014;133(6):1564–71. doi: 10.1016/j.jaci.2013.10.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Hinds DA, McMahon G, Kiefer AK, Do CB, Eriksson N, Evans DM, et al. A genome-wide association meta-analysis of self-reported allergy identifies shared and allergy-specific susceptibility loci. Nat Genet. 2013;45(8):907–11. doi: 10.1038/ng.2686. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Wenzel SE. Asthma phenotypes: the evolution from clinical to molecular approaches. Nat Med. 2012;18(5):716–25. doi: 10.1038/nm.2678. [DOI] [PubMed] [Google Scholar]
  • 27.Bell JT, Pai AA, Pickrell JK, Gaffney DJ, Pique-Regi R, Degner JF, et al. DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome biology. 2011;12(1):R10. doi: 10.1186/gb-2011-12-1-r10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Bakulski KM, Feinberg JI, Andrews SV, Yang J, Brown S, SLM, et al. DNA methylation of cord blood cell types: Applications for mixed cell birth studies. Epigenetics. 2016;11(5):354–62. doi: 10.1080/15592294.2016.1161875. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Anthoni M, Fyhrquist-Vanni N, Wolff H, Alenius H, Lauerma A. Transforming growth factor-beta/Smad3 signalling regulates inflammatory responses in a murine model of contact hypersensitivity. The British journal of dermatology. 2008;159(3):546–54. doi: 10.1111/j.1365-2133.2008.08696.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Lim RH, Kobzik L, Dahl M. Risk for asthma in offspring of asthmatic mothers versus fathers: a meta-analysis. PLoS One. 2010;5(4):e10134. doi: 10.1371/journal.pone.0010134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Paaso EM, Jaakkola MS, Rantala AK, Hugg TT, Jaakkola JJ. Allergic diseases and asthma in the family predict the persistence and onset-age of asthma: a prospective cohort study. Respiratory research. 2014;15:152. doi: 10.1186/s12931-014-0152-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Moore WC, Bleecker ER. Asthma heterogeneity and severity-why is comprehensive phenotyping important? The Lancet Respiratory medicine. 2014;2(1):10–1. doi: 10.1016/S2213-2600(13)70288-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Schaub B, Liu J, Hoppler S, Schleich I, Huehn J, Olek S, et al. Maternal farm exposure modulates neonatal immune mechanisms through regulatory T cells. J Allergy Clin Immunol. 2009;123(4):774–82. doi: 10.1016/j.jaci.2009.01.056. [DOI] [PubMed] [Google Scholar]
  • 34.Lluis A, Ballenberger N, Illi S, Schieck M, Kabesch M, Illig T, et al. Regulation of TH17 markers early in life through maternal farm exposure. J Allergy Clin Immunol. 2014;133(3):864–71. doi: 10.1016/j.jaci.2013.09.030. [DOI] [PubMed] [Google Scholar]
  • 35.Basu R, Whitley SK, Bhaumik S, Zindl CL, Schoeb TR, Benveniste EN, et al. IL-1 signaling modulates activation of STAT transcription factors to antagonize retinoic acid signaling and control the TH17 cell-iTreg cell balance. Nat Immunol. 2015;16(3):286–95. doi: 10.1038/ni.3099. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Nicodemus-Johnson J, Naughton KA, Sudi J, Hogarth K, Naurekas ET, Nicolae DL, et al. Genome-wide Methylation Study Identifies an IL-13 Induced Epigenetic Signature in Asthmatic Airways. Am J Respir Crit Care Med. 2015 doi: 10.1164/rccm.201506-1243OC. in press. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.DeVries A, Vercelli D. Epigenetics in allergic diseases. Curr Opin Pediatr. 2015;27(6):719–23. doi: 10.1097/MOP.0000000000000285. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Wang G, Walker SO, Hong X, Bartell TR, Wang X. Epigenetics and early life origins of chronic noncommunicable diseases. The Journal of adolescent health : official publication of the Society for Adolescent Medicine. 2013;52(2 Suppl 2):S14–21. doi: 10.1016/j.jadohealth.2012.04.019. [DOI] [PubMed] [Google Scholar]
  • 39.McDade TW. Early environments and the ecology of inflammation. Proc Natl Acad Sci U S A. 2012;109(Suppl 2):17281–8. doi: 10.1073/pnas.1202244109. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1. Figure E1. Workflow of DNA methylation profiling.

Genomic DNA was fragmented by sonication and divided into two aliquots: one underwent methyl-binding protein methylation capture (MBDCap) and was enriched for methylated DNA, while the other was left untreated. The two fractions were then whole genome-amplified, fluorescently labeled [captured (enriched)=red, untreated (input)=green], mixed and hybridized to Human DNA Methylation 2.1M Deluxe Promoter Arrays (Roche NimbleGen). After scanning the arrays, the hybridization intensities for each probe were extracted separately for the red and green channels, and were quantile-normalized to remove technical noise. The difference between normalized intensities [log2 enriched – log2 input = log2 (enriched/input), referred to as the log2 ratio] was then calculated and used as a measure of methylation levels. Measurements for 16 genomic regions were validated by bisulfite sequencing. Regions differentially methylated in the asthmatic and non-asthmatic groups were detected using Roche-Nimblegen scripts that implement a probe sliding-window ANOVA. DMRs were functionally annotated using Homer and Ingenuity Pathway Analysis.

NIHMS838362-supplement-1.tif (374.5KB, tif)
10. Table E1.

Characteristics of IIS study subjects.

NIHMS838362-supplement-10.xlsx (13.4KB, xlsx)
11. Table E2.

Characteristics of non-asthmatic and asthmatic children in the IIS, MAAS and COAST study populations.

NIHMS838362-supplement-11.xlsx (87.2KB, xlsx)
12. Table E3.

Complete list of annotated asthma-associated DMRs (n=589) in IIS children*

NIHMS838362-supplement-12.xlsx (12.3KB, xlsx)
13. Table E4.

Primers used for bisulfite sequencing.

NIHMS838362-supplement-13.xlsx (37.3KB, xlsx)
14. Table E5.

Functional annotation of asthma-associated DMRs in IIS children.

NIHMS838362-supplement-14.xlsx (17.8KB, xlsx)
15. Table E6.

Overlap between asthma-associated DMRs in IIS children and DNase I HSS in monocytes and naïve T cells.

NIHMS838362-supplement-15.xlsx (9.9KB, xlsx)
16. Table E7.

Asthma-associated DMRs in IIS children and DNase I HSS

NIHMS838362-supplement-16.xlsx (12.8KB, xlsx)
17. Table E8.

Association between child and parental risk factors at birth and asthma during childhood in IIS children.

NIHMS838362-supplement-17.docx (38.7KB, docx)
2. Figure E2. Correlation of bisulfite sequencing and DNA methylation microarray results.

Each dot corresponds to the median normalized log2 ratios of enriched/input sample across probes (for microarray data) or the median percentage DNA methylation across CpG sites (for bisulfite sequencing data) for each region in each subject. The correlation between median normalized log2 ratios from microarray analysis and median percent DNA methylation from bisulfite sequencing was assessed by measuring the Spearman correlation coefficient (ρ).

NIHMS838362-supplement-2.tif (103MB, tif)
3. Figure E3. Distribution of asthma-associated DMRs by chromosome and genomic location.

A) Manhattan plot of asthma-associated DMRs. The x axis depicts the chromosomal location of each DMR. The y-axis shows the Benjamini Hochberg (BH)-adjusted P-values from the probe sliding window ANOVA. The minimum value of the y-axis was set at 2 because a significance threshold of 0.01 was imposed on the adjusted P-values. Genes associated with top hits are labeled. B) Genomic location of asthma-associated DMRs. The annotation is based on the center of the DMR, as defined by the Homer package. The genomic distribution of DMRs differed significantly from the distribution of probes across the entire array (χ2= 19.49, df = 7, P-value = 0.007). Fifty-thousand permutations were performed to assess whether this distribution differed from random expectations (permutation P-value = 0.008). Standardized residuals were computed for each category in order to assess which genomic locations contributed to the overall statistical significance. Absolute values > 1.96 significantly contribute to the overall χ2 test statistics. UTR: untranslated region, TSS: transcription start site (−1kb to +100 bp), TTS: transcription termination site (−100 bp to +1kb).

NIHMS838362-supplement-3.tif (851.3KB, tif)
4. Figure E4. Regulatory potential of the SMAD3 DMR.

The SMAD3 DMR was mapped to the human genome (hg19/GRCh37) using University of California Santa Cruz (UCSC) Genome Browser tools. Tracks are specified on the left. UCSC genes are shown in blue. Nearby CpG islands (with identifier) are shown in green. DNase I hypersensitive sites (HSS) identified in CD14+ monocytes and CD4+ naïve T cells are shown in red and blue, respectively. Post-translational histone modifications (H3K27Ac, H3K4Me1, and H3K4Me3) identified in CD14+ monocytes, CD4+ naïve T cells, and peripheral blood mononuclear cells (PBMC) (Roadmap Epigenomics Project) mark poised/active regulatory regions (E7). The bottom track shows common SNPs (MAF > 10%) from the HapMap CEU population, none of which are associated with asthma. The genomic region that surrounds the SMAD3 DMR (+/− 100 bp) is highlighted in red. The figure covers 5 kb upstream and downstream of the SMAD3 DMR.

NIHMS838362-supplement-4.tif (1MB, tif)
5. Figure E5. Association between SMAD3 DNA methylation at birth and childhood asthma in the IIS cohort.

SMAD3 methylation was assessed by bisulfite sequencing and expressed as mean percentage DNA methylation across 8 consecutive CpG sites in the SMAD3 DMR. N: non-asthmatic, A: asthmatic. P-value by Wilcoxon two-sample test.

NIHMS838362-supplement-5.tif (11.4MB, tif)
6. Figure E6. SMAD3 CpG7 methylation in IIS and MAAS.

SMAD3 methylation was assessed by bisulfite sequencing and expressed as percent methylation at SMAD3 CpG7 in neonates from the IIS and MAAS cohorts. N: non-asthmatic, A: asthmatic. P-values by Wilcoxon two-sample test.

NIHMS838362-supplement-6.tif (22.9MB, tif)
7. Figure E7. Estimation of CBMC composition in the COAST cohort.

Proportions of CBMC cell types were estimated using DNA methylation data generated on the Illumina HumanMethylation450 BeadChip and a cord blood reference panel (E11). N: non-asthmatic, A: asthmatic. P-values by Wilcoxon two-sample test.

NIHMS838362-supplement-7.tif (74.4MB, tif)
8. Figure E8. Effects of maternal asthma on the association between neonatal SMAD3 methylation and childhood asthma in COAST after adjusting for CBMC composition.

Residual methylation values for SMAD3 (cg02486855, Illumina HumanMethylation450 BeadChip) were calculated by adjusting for CBMC composition using linear regression. N: non-asthmatic, A: asthmatic. P-values by Wilcoxon two-sample test.

NIHMS838362-supplement-8.tif (11.4MB, tif)
9
NIHMS838362-supplement-9.xlsx (11.9KB, xlsx)

RESOURCES