SEARCH

SEARCH BY CITATION

Keywords:

  • Asthma;
  • DNA methylation;
  • farm exposure;
  • IgE;
  • pyrosequencing

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Aknowledgments
  7. Authors' contributions
  8. Conflict of interest
  9. References
  10. Supporting Information

Background

Genetic susceptibility and environmental influences are important contributors to the development of asthma and atopic diseases. Epigenetic mechanisms may facilitate gene by environment interactions in these diseases.

Methods

We studied the rural birth cohort PASTURE (Protection against allergy: study in rural environments) to investigate (a) whether epigenetic patterns in asthma candidate genes are influenced by farm exposure in general, (b) change over the first years of life, and (c) whether these changes may contribute to the development of asthma. DNA was extracted from cord blood and whole blood collected at the age of 4.5 years in 46 samples per time point. DNA methylation in 23 regions in ten candidate genes (ORMDL1, ORMDL2, ORMDL3, CHI3L1, RAD50, IL13, IL4, STAT6, FOXP3, and RUNX3) was assessed by pyrosequencing, and differences between strata were analyzed by nonparametric Wilcoxon–Mann–Whitney tests.

Results

In cord blood, regions in ORMDL1 and STAT6 were hypomethylated in DNA from farmers' as compared to nonfarmers' children, while regions in RAD50 and IL13 were hypermethylated (lowest P-value (STAT6) = 0.001). Changes in methylation over time occurred in 15 gene regions (lowest P-value (IL13) = 1.57*10−8). Interestingly, these differences clustered in the genes highly associated with asthma (ORMDL family) and IgE regulation (RAD50, IL13, and IL4), but not in the T-regulatory genes (FOXP3, RUNX3).

Conclusions

In this first pilot study, DNA methylation patterns change significantly in early childhood in specific asthma- and allergy-related genes in peripheral blood cells, and early exposure to farm environment seems to influence methylation patterns in distinct genes.

Abbreviations
COPD

chronic obstructive pulmonary disease

GWAS

genomewide association study

PASTURE

Protection against allergy: study in rural environments

PBMC

peripheral blood mononuclear cells

SNP

single nucleotide polymorphism

Treg

regulatory T cells

TSS

transcription start site

Genetic predisposition and environmental influences contribute to asthma development [1]. More than 150 genes have been associated with asthma and related phenotypes [2, 3]. While environmental factors, such as cigarette smoke exposure and air pollution, aggravate the disease [4], living on a farm and the associated microbial exposure protect children from asthma and atopy [5, 6]. This protection is most effective when exposure takes place in the first 3 years of life [7, 8], but its nature and mechanism are still unclear.

The protective effect of farm environment exposure on allergy and asthma development may be mediated by epigenetic mechanisms that can regulate accessibility of genes for transcription by secondary, not mutation-based modification of DNA and DNA-associated proteins. It has previously been demonstrated in mice and humans that the environment influences epigenetic patterns [9, 10]. While epigenetic effects contribute to T-cell differentiation into Th2 and regulatory T cells (Treg) [11-13], specific studies on how epigenetics may affect early asthma development, influencing gene by environment interactions, are missing.

In this first pilot study, we studied the rural birth cohort PASTURE (Protection against allergy: study in rural environments) to investigate (a) whether epigenetic patterns in asthma candidate genes are influenced by farm exposure, (b) change over the first years of life, and (c) whether these changes may contribute to the development of asthma. Therefore, DNA methylation status of ten asthma and allergy candidate genes was analyzed in cord blood, and DNA extracted from whole blood at the age of 4.5 years in children living on farms and a rural control group not living on farms as well as in asthmatics and nonasthmatics from the same population. We selected three groups of genes for this explorative analysis: (i) genes that were associated with asthma in the first genome-wide association studies (GWAS) on asthma and IgE, (ii) genes involved in Th2 development; and (iii) genes involved in T-regulatory effects.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Aknowledgments
  7. Authors' contributions
  8. Conflict of interest
  9. References
  10. Supporting Information

Study population

The PASTURE population is a birth cohort from five European countries (Austria, Finland, France, Germany, and Switzerland) [14]. The study was specifically designed to investigate which environmental exposures and genetic and immunological mechanisms contribute to the protective farm effect over the first years of life. For that purpose, about 1000 pregnant women, living in rural areas were recruited during their last trimester of pregnancy. Approximately half of these women were living on or running a farm with livestock. Their children are referred to as ‘farmers’ children'.

For details of the selection process, please see the supplementary methods.

Selection of genes and regions of interest

We studied genes that either were major findings in genomewide association studies on asthma and regulation of total IgE levels at the time the study was initiated (CHI3L1 [15], ORMDL3 [16], RAD50 [17], and STAT6 [17]) or were considered pivotal in T-cell differentiation and previously reported to be affected by epigenetic mechanisms in humans or mice (RAD50 [18, 19], IL4 [20, 21], IL13 [22, 23], STAT6 [24], FOXP3 [25, 26], and RUNX3 [27]). In addition, the study was extended to two further members of the ORMDL family (ORMDL1 and ORMDL2). Further detailed description of the investigated amplification products (= amplicons) is provided in the supplementary methods.

Methylation assays

Genomic DNA was processed by serial pyrosequencing [28] to quantitatively analyze bisulfite-treated DNA as described earlier [29] and in detail in the supplementary methods.

Statistical analyses

Statistical analysis was performed using the R software [30]. Differences in DNA methylation between the groups were assessed by nonparametric Wilcoxon–Mann–Whitney tests. Group-specific changes of overall DNA methylation levels were assessed by a methylation index (z-score) as previously proposed for exposure analysis [31], calculated separately for cord blood and 4.5-year samples.

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Aknowledgments
  7. Authors' contributions
  8. Conflict of interest
  9. References
  10. Supporting Information

We explored DNA methylation patterns in 23 regions in ten genes potentially related to asthma, T-cell differentiation into Th2 and Treg subtypes. Amplicons as shown in Figure 1 and Table S1 were analyzed by pyrosequencing. In most genes, a region spanning the transcription start site (TSS) was investigated, and in some cases, (ORMDL3, CHI3L1, RAD50, IL4, IL13, and STAT6) additional regions with known asthma-related SNPs (ORMDL3, CHI3L1, RAD50, IL13 and STAT6) or areas that were previously reported to undergo changes in methylation were analyzed as well (RAD50 and IL4). Overall, the DNA methylation status of 149 CpG sites was analyzed.

image

Figure 1. Position of amplification products. Investigated amplicons are shown in context of the gene structure. Exons are marked in orange, and the amplicons are shown in dark red.

Download figure to PowerPoint

Effect of farm exposure on methylation status

DNA extracted from cord blood from farmers' and nonfarmers' children was compared, and results showed significant differences in DNA methylation levels in four genes (Table 1, left panel), thus exceeding the stochastic expectations based on chance (23*0.05 = 1.15 random-positive results expected for this comparison). The regions around the TSS of ORMDL1 (ORMDL1_a) and a CpG island in intron 1 of STAT6 (STAT6_b) showed overall low methylation levels (< 5%) in both groups, but were significantly less methylated in farmers' children. In contrast, the locus studied in intron 24 of RAD50 harboring a hypersensitive site [32] (RAD50_b) was strongly methylated in general (Approximately 70%) and hypermethylated (3% difference) in the farm group. IL13_d spanning the asthma-associated SNP rs1800925 and displaying an average methylation degree of approximately 50% was also statistically significantly hypermethylated in the farm group.

Table 1. Cord blood methylation values (mean and standard deviations in%) comparing farming and asthma status
Genes and Amplicons(A) Farm children (n = 21)(B) Nonfarm children (n = 24)Difference between groups A:B (P-value)(C) Early asthmatics (n = 21)(D) Nonasthmatics (n = 24)Difference between groups C:D (P-value)
  1. Differential DNA methylation was assessed between farmers' and nonfarmers' children as well as between asthmatics and nonasthmatics.

  2. The P-value of the Wilcoxon–Mann–Whitney test is shown behind the respective two columns of the comparison. Bold indicates significant P-values.

ORMDL3_a12.69 (1.05)12.90 (1.26)0.89712.56 (1.05)13.04 (1.24)0.435
ORMDL3_b83.42 (2.01)82.48 (2.56)0.10682.51 (2.19)83.31 (2.47)0.363
ORMDL3_c86.22 (3.14)87.27 (2.36)0.38387.90 (2.09)85.83 (2.94) 0.032
ORMDL3_d85.68 (2.67)85.54 (3.05)0.93586.00 (3.21)85.17 (2.39)0.579
ORMDL1_a2.87 (0.33)3.19 (0.37) 0.007 2.97 (0.31)3.10 (0.43)0.255
ORMDL2_a6.12 (1.89)5.41 (1.12)0.3395.95 (1.86)5.61 (1.30)0.779
CHI3L1_a33.44 (5.06)32.70 (4.75)0.81933.07 (3.86)33.02 (5.66)0.933
CHI3L1_b25.81 (6.99)26.06 (3.38)0.25825.33 (3.81)26.53 (6.87)0.988
RAD50_a4.07 (1.01)4.28 (1.43)0.8024.20 (0.99)4.16 (1.43)0.660
RAD50_b71.76 (2.00)68.96 (3.30) 0.005 70.14 (3.12)70.54 (3.05)0.734
RAD50_c87.72 (0.96)87.04 (1.33)0.07687.32 (1.01)87.39 (1.42)0.549
IL13_a18.69 (2.48)18.77 (2.67)0.96318.99 (2.62)18.49 (2.51)0.659
IL13_b83.89 (1.79)83.74 (2.36)0.80184.13 (1.92)83.51 (2.22)0.478
IL13_c7.33 (0.74)7.10 (1.39)0.5877.43 (1.03)6.95 (1.22)0.473
IL13_d53.78 (1.67)52.42 (1.99) 0.030 53.48 (1.86)52.85 (1.99)0.403
CNS187.99 (2.42)87.26 (3.18)0.45687.74 (2.25)87.45 (3.41)0.990
IL4_a68.67 (5.04)67.09 (7.03)0.47866.30 (7.45)69.46 (3.92)0.243
IL4_b64.07 (4.56)63.55 (4.54)0.79763.71 (4.02)63.85 (5.06)0.917
STAT6_a5.36 (1.25)6.64 (2.62)0.1376.32 (2.10)5.79 (2.26)0.244
STAT6_b3.79 (0.34)4.23 (0.47) 0.001 4.04 (0.42)3.99 (0.51)0.753
STAT6_c91.29 (1.23)90.57 (1.89)0.49991.04 (1.92)90.72 (1.33)0.215
FOXP3_a94.94 (0.96)94.84 (0.99)0.82794.70 (0.99)95.07 (0.93)0.319
RUNX3_a2.40 (0.32)2.48 (0.47)0.4412.33 (0.45)2.54 (0.34)0.133

When analyzing the 4.5-year samples (Table S2), only one significant finding (thus not exceeding random stochastic expectations) was observed. Not the region in intron 1 (STAT6_b), but the closely neighbored transcription start site of STAT6 (STAT6_a) was found to be less methylated in farmers' children aged 4.5 years. ORMDL1_a, RAD50_b, and IL13_d did not show differences in methylation status between farmers' and nonfarmers' children at that age.

Methylation status and early asthma

No differences in methylation of the investigated genes (exceeding those expected by chance) were observed when children without asthma were compared with those with early asthma symptoms (Table 1, right panel). In DNA from cord blood, a region in intron 1 of ORDML3 (ORMDL_c) located close to the asthma-related SNP rs8076131 was hypermethylated in early asthmatics compared with nonasthmatics (87.9% > 85.83%; P = 0.032). A second partly overlapping amplicon (ORMDL3_d) was also slightly more methylated in early asthmatics than in nonasthmatics, but this finding did not reach statistical significance. The TSS of ORMDL2 (ORMDL2_a), which showed overall low methylation levels (< 10%), was hypomethylated in early asthmatics (Table S2).

Effects of farming on methylation and asthma development

Next, we investigated DNA methylation differences between asthmatics and nonasthmatics after stratification for farm exposure. Within the children growing up on farms, no differences in methylation were observed in relation to the early asthma status (Table 2, left panel). Then we compared methylation in nonasthmatic farmers' children to asthmatic nonfarmers' children as these groups represent the extremes in exposure and disease status. If farming exposure influences methylation status of asthma candidate genes, which in turn might affect asthma development, differences between these two groups should exist. Indeed, in cord blood, loci in intron 1 of ORMDL3 (ORMDL3_c) and STAT6 (STAT6_b) showed statistically significant hypermethylation in asthmatics not being exposed to a farm environment compared with the healthy farmers' children (Table 2, right panel) but no differences were detected at age 4.5 (Table S3).

Table 2. Cord blood methylation values (mean and standard deviations in%) comparing asthma status within substrata
Genes and AmpliconsFarm childrenNonfarm children
(A) Early asthmatics (n = 9)(B) Nonasthmatics (n = 12)Difference between groups A:B (P-value)Difference between groups B:C (P-value)(C) Early asthmatics (n = 12)(D) Nonasthmatics (n = 12)
  1. Differential DNA methylation was assessed between asthmatics and nonasthmatics within farmers' children and between nonasthmatic farmers' children and asthmatic nonfarmers' children. The P-values of the respective Wilcoxon–Mann–Whitney tests are shown in the middle of the table. Bold indicates significant P-values.

ORMDL3_a12.47 (0.92)12.90 (1.18)0.6670.72012.63 (1.20)13.17 (1.34)
ORMDL3_b83.73 (1.60)83.17 (2.33)0.6560.11781.59 (2.18)83.45 (2.70)
ORMDL3_c87.78 (2.64)85.22 (3.12)0.179 0.048 87.98 (1.81)86.50 (2.72)
ORMDL3_d86.31 (3.39)85.05 (1.66)0.4370.72385.77 (3.20)85.28 (2.99)
ORMDL1_a2.81 (0.27)2.91 (0.37)0.6190.3563.09 (0.29)3.31 (0.41)
ORMDL2_a6.79 (2.12)5.44 (1.44)0.1710.7005.00 (0.90)5.78 (1.21)
CHI3L1_a32.56 (3.98)34.03 (5.77)0.6790.97833.41 (3.91)31.93 (5.61)
CHI3L1_b24.25 (4.04)27.22 (8.86)0.6040.49726.41 (3.45)25.66 (3.48)
RAD50_a4.25 (0.93)3.93 (1.08)0.4560.5584.15 (1.09)4.43 (1.78)
RAD50_b71.38 (1.78)72.02 (2.18)0.6790.07069.16 (3.67)68.77 (3.08)
RAD50_c87.52 (0.73)87.90 (1.14)0.2700.10787.17 (1.19)86.88 (1.54)
IL13_a19.70 (2.82)17.93 (1.97)0.3830.71318.46 (2.45)19.11 (2.97)
IL13_b83.71 (1.96)84.03 (1.72)0.6020.44384.44 (1.91)82.89 (2.66)
IL13_c7.37 (0.73)7.28 (0.79)0.9320.8877.48 (1.24)6.65 (1.48)
IL13_d54.33 (1.50)53.41 (1.74)0.4270.38252.73 (1.89)52.11 (2.15)
CNS187.77 (2.15)88.19 (2.75)0.7200.62887.71 (2.41)86.78 (3.92)
IL4_a67.26 (5.85)69.82 (4.20)0.5030.27165.51 (8.76)69.02 (3.74)
IL4_b63.90 (3.81)64.20 (5.30)1.0000.86363.59 (4.29)63.49 (5.11)
STAT6_a5.58 (1.31)5.18 (1.22)0.5520.0656.87 (2.45)6.39 (2.90)
STAT6_b3.89 (0.34)3.71 (0.34)0.394 0.012 4.15 (0.46)4.32 (0.49)
STAT6_c91.32 (1.36)91.25 (1.18)1.0000.88790.81 (2.32)90.31 (1.35)
FOXP3_a94.60 (1.03)95.22 (0.85)0.2300.47894.78 (1.00)94.90 (1.04)
RUNX3_a2.39 (0.34)2.41 (0.31)0.8990.5262.29 (0.51)2.70 (0.32)

Group-specific changes of overall DNA methylation levels

We investigated whether overall methylation of the 23 selected gene regions was different between the analyzed groups (farmers, nonfarmers; asthmatics, and nonasthmatics). A z-score was used to normalize the data allowing comparison of all amplicons. For the 4.5-year samples, no differences were detected (Figs S1a and S1b). In cord blood however, farm children showed a trend toward an overall lower level of changes in DNA methylation at the analyzed loci compared with children not exposed to farm environment (P = 0.093). This provided a further evidence for environment-specific DNA methylation changes in the analyzed samples (Fig. S1c). No differences were observed comparing asthmatics and nonasthmatics in the cord blood samples (Fig. S1d). When correlation of DNA methylation between different amplicons and genes was analyzed, several amplicons related to Th2-regulatory genes showed significant correlation in DNA methylation. In the 4.5-year samples, several amplicons in genes of the ORMDL gene family showed ‘phased’ DNA methylation (Supplementary Figure 2).

Time trends in methylation between cord blood and whole blood at 4.5 years of age

Significant and large differences in DNA methylation of up to 7% (IL13_a; CpG island 5′ of the TSS of IL13) were found when methylation in cord blood was compared with whole blood at 4.5 years of age across strata (Table 3). The ORMDL3 locus showed decreased methylation over time in farmers' children as well as nonfarmers' children, in early asthmatics and nonasthmatics, while lower methylation was observed for ORMDL1 only in nonfarmers' children and nonasthmatics. On chromosome 5, RAD50 amplicons showed hypomethylation in a regulatory element (DNAse hypersensitive site) in intron 24 (RAD50_b) over time with the strongest effect observed in children exposed to the farm environment, whereas the closely located amplicon RAD50_c and the region around the RAD50 TSS were slightly more methylated over time. IL13 was hypermethylated over time with the smallest effect observed in asthmatics. In contrast, methylation increase in the IL4 locus was strongest in asthmatics and nonfarmers, respectively. Relatively small effects were observed for STAT6 with a slight decrease in methylation. CHI3L1, another gene suggested to relate to asthma in GWAS, as well as the amplicons tested in the T-regulatory genes FOXP3 and RUNX3 showed no significant difference in the time trend analyses.

Table 3. Time trends in DNA methylation from cord blood and whole blood at 4.5 years of age
Genes and AmpliconsFarm childrenNonfarm childrenEarly asthmaticsNonasthmatics
Mean difference P-value Mean difference P-value Mean difference P-value Mean difference P-value
  1. Mean differences between methylation values of cord blood and 4.5-year samples (4.5-year values are subtracted from the cord blood values) of one stratum are given with the P-value of the Wilcoxon–Mann–Whitney test analyzing the difference between the four groups of asthmatics, nonasthmatics, farmers' and nonfarmers' children at birth and at 4.5 years of age. Bold indicates significant P-values.

ORMDL3_a1.51 0.0012 1.82 0.0002 1.59 0.0001 1.75 0.0010
ORMDL3_b2.27 0.0102 0.260.8651.290.0851.190.276
ORMDL3_c3.73 0.0078 4.68 9.37*10−6 5.03 3.52*10−5 3.57 0.0019
ORMDL3_d4.58 0.0003 4.60 0.0005 4.68 0.0013 4.50 0.0001
ORMDL1_a0.110.1480.34 0.0022 0.160.1170.30 0.0089
ORMDL2_a−0.100.536−0.590.3020.320.907−0.940.067
CHI3L1_a−2.090.221−0.800.728−0.810.421−2.030.370
CHI3L1_b−2.090.0610.280.646−0.380.967−1.220.134
RAD50_a−0.600.103−0.120.759−0.230.554−0.460.212
RAD50_b5.61 0.0002 3.38 0.0065 4.23 0.0025 4.76 0.0002
RAD50_c−0.010.979−0.86 0.0271 −0.66 0.0315 −0.270.734
IL13_a−6.29 1.75*10−5 −6.09 2.36*10−6 −4.89 0.0009 −7.28 1.57*10−8
IL13_b−1.010.120−1.030.173−0.200.639−1.78 0.0095
IL13_c−0.77 0.0091 −0.78 0.0334 −0.380.091−1.26 0.0016
IL13_d3.31 0.0002 1.92 0.0151 2.32 0.0053 3.08 0.0030
CNS1−0.710.496−1.480.106−0.940.319−1.320.169
IL4_a−2.050.353−5.21 0.0065 −5.40 0.0164 −2.070.112
IL4_b0.370.757−1.840.285−1.670.452−0.090.493
STAT6_a0.96 0.0160 1.6 0.0388 1.57 0.0090 1.050.0776
STAT6_b0.34 0.0157 0.53 0.0002 0.32 0.0087 0.54 0.0011
STAT6_c0.330.986−0.99 0.0068 −0.510.343−0.330.096
FOXP3_a0.730.0560.170.3990.310.3360.540.074
RUNX3_a0.120.3760.120.2290.060.5000.170.173

Replication of time trends in methylation

For replication, we selected the regions showing the most stable difference in methylation. The methylation status of three regions that showed consistently significant differences of more than two percent points in the time trend analysis (Table 3) were analyzed in additional 30 samples (from Austria and Switzerland). Again, the CpG island 5′ of the TSS of IL13 (IL13_a) showed an increase in methylation over all groups, which was very similar to the increase observed in the initial sample. Results were also comparable with the initial set for the intronic region ORMDL3_d except for early asthmatics where an increase in methylation was observed (Table S4). Results were replicated for the hypersensitive site in intron 24 of RAD50 (RAD50_b) harboring SNP rs2240032 which has been shown to be related to IgE levels [17]. Again, only the group of early asthmatics in the additional sample set did not show the same decline in methylation over time.

When the replication data set and the initial data set were combined (Table S5), the methylation in the CpG island of IL13 (IL13_a) decreased significantly for all four strata (P < 0.001). ORMDL3 and RAD50 showed a clear increase in methylation from cord blood to the 4.5-year samples. This increase was significant or at least close to the nominal significance level of P = 0.05 for all four strata except for the nonfarmers in intron1 of ORMDL3 (ORMDL3_d).

To investigate whether methylation changes over time are associated with asthma development, we analyzed those children from the combined sample for which DNA was available at both time points and compared the individual changes according to disease status. The intronic region of ORMDL3 (ORMDL3_d) and the CpG island 5′ of the TSS of IL13 (IL13_a) showed a significant decrease in methylation over time in asthmatic farmers' children in contrast to an increasing methylation in asthmatic nonfarmers' children (Tables 4 and 5).

Table 4. Changes between cord blood and 4.5-year methylation values (mean and standard deviations in%) in relation to farming and asthma status
Genes and Amplicons(A) Farm children (n = 30)(B) Nonfarm children (n = 26)Difference between groups A:B (P-value)(C) Early asthmatics (n = 24)(D) Non asthmatics (n = 32)Difference between groups C:D (P-value)
  1. Pooled analysis of German discovery samples and Austrian and Swiss samples from replication data set for which DNA from both time points was available. The means of the individual differences from cord blood and the 4.5-year samples (4.5-year values are subtracted from the cord blood values) are given with the respective P-value of the Wilcoxon–Mann–Whitney test on difference between farmers' and nonfarmers' children as well as between asthmatics and nonasthmatics. Bold indicates significant P-values.

ORMDL3_d5.33 (7.35)1.5 (10.95)0.03420.95 (7.08)6.03 (10.56)0.2615
RAD50_b2.83 (5.43)1.06 (3.92)0.2393−0.12 (5.22)3.85 (3.85) 0.0069
IL13_a−5.21 (5.66)−5.20 (7.05)0.9678−2.94 (4.26)−6.88 (7.21) 0.0107
Table 5. Changes between cord blood and 4.5-year methylation values (mean and standard deviations in%) regarding asthma in relation to farm status
Genes and AmpliconsFarm childrenNonfarm children
(A) Early asthmatics (n = 14)(B) Nonasthmatics (n = 16)Difference between groups A:B (P-value)Difference between groups B:C (P-value)(C) Early asthmatics (n = 10)(D) Nonasthmatics (n = 16)
  1. Pooled analysis of German discovery sample and Austrian and Swiss samples from replication data set for which DNA from both time points was available. The means of the individual differences from cord blood and the 4.5-year samples (4.5-year values are subtracted from the cord blood values) are given with the respective P-value of the Wilcoxon–Mann–Whitney test on difference between asthmatics and nonasthmatics within farmers' children and between nonasthmatic farmers' children and asthmatic nonfarmers' children. Bold indicates significant P-values.

ORMDL3_d3.78 (5.14)7.34 (9.43)0.5224 0.0155 −3.65 (7.67)4.93 (11.72)
RAD50_b1.13 (5.99)4.29 (4.63)0.2520 0.0006 −2.00 (3.31)3.28 (2.68)
IL13_a−1.45 (4.35)−8.22 (4.82) 0.0062 0.0789−4.25 (3.96)−5.85 (8.68)

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Aknowledgments
  7. Authors' contributions
  8. Conflict of interest
  9. References
  10. Supporting Information

With this pilot study, we provide further evidence that farm environment affects the epigenome, as it was very recently proposed for methylation of the CD14 gene promoter [33]. Potentially, gene regulatory mechanisms may play a role in the protection against asthma and allergy observed in children exposed to farm environment during early childhood (or even before). GWAS-derived asthma candidate genes and genes regulating Th2 differentiation showed changes in methylation. These changes were observed in children growing up on farms and their peer controls and in relation to the development of wheeze in the first 4 years of life. Distinct changes in methylation especially over time were observed in genes of the ORMDL family as well as in different genes related to IgE regulation and Th2 differentiation, especially RAD50 and IL13.

Clearly, the presented study is an explorative study based on a small sample size. However, the samples were selected from a much larger study population based on stringent selection criteria and are representative for our farming cohort based on demographic criteria. Quite uniquely, the individuals included in this study had DNA extracted for methylation analysis at birth from cord blood and at different time points later on during early childhood, which would in principle allow to discern if observed differences in DNA methylation precede the disease and might therefore be at least partly causative or ‘only’ disease associated. In addition, samples were selected based on the farm environment exposure to investigate epigenetic effects of this strong protective factor on asthma and allergy development in early life. Thus, the very specific longitudinal design of this methylation study counterbalances at least partly the limitation of small sample numbers available for analysis.

So far, very few studies investigated DNA methylation patterns in more than one amplicon and in more than a single gene in complex diseases. In contrast to cancer, it is more difficult to define target genes and target tissue for DNA methylation analysis in a disease like asthma where both lung-specific and general immunological mechanisms may play a role. Thus, in the absence of preexisting and reliable data, this study explored whether such analyses could be feasible for candidate genes related to asthma and allergy development in cells derived from peripheral blood. In other areas of research, recent studies have suggested that DNA from peripheral blood cells might be suitable to detect phenotype-specific DNA methylation changes, although they do not directly represent the tissue of action. It was shown that weight-associated DNA methylation patterns as well as DNA methylation that correlated with a dietary intervention program could be detected in PBMCs [34, 35].

Our data suggest that genes related to distinct aspects of immunology change methylation status over time in peripheral blood cells (cord blood versus whole blood). Interestingly, methylation differences did not affect all investigated genes equally but certain genes (ORMDL gene family and genes involved in Th2/Th1 differentiation) showed significant and constant methylation differences, while others did not (Chitinase and T-regulatory genes). This implies that there is no general change in methylation over time and provides some further support that the findings are not solely based on chance. Interestingly, a further study examining methylation time trends in children also reported significant changes in methylation status of a gene related to innate immunity (CD14) studied comparatively in whole blood at 2 and 10 years of age [36]. Even though our study is based on only a few genes and regions investigated within those genes the observed differences are striking and should be explored further.

ORMDL3 was the first asthma candidate gene identified by a GWAS [16].

A recent study reported an interaction of ORMDL3 (and partly for ORMDL2) with Th2-related genes IL4, IL13, and STAT6. In murine lung epithelial cells, expression of ORMDL3 could be induced by IL4 and IL13 in a STAT6-dependent manner [37]. Differential DNA methylation of the CpG island in the putative promoter region of IL13 (covered by IL13_a) seems to have an influence on the IL13 gene expression [38]. The IL13-RAD50 locus has been shown to be involved in the T-cell differentiation, and this differentiation seems to be modulated by epigenetic effects [11, 12, 22]. The hypersensitive site in RAD50 (analyzed by amplicon RAD50_b) has been identified as locus control region influencing the expression of the Th2-related cytokines IL4 and IL13 [19, 32]. The clustering of changes in gene specific methylation patterns over time in genes functionally relevant for asthma also suggests (but does not exclude) that differences in cell composition between cord blood and whole blood at 4.5 years of age are not likely to explain these differences sufficiently. This is in line with published results from a recent study where FACS analyses showed that the major cell fraction of CD4+ T cells in PBMCs do not vary significantly over time. Thus, the authors argue that differences in methylation observed for specific pathways cannot be explained by changes in cell composition alone [39].

Therefore, we fully acknowledge that the cell selection for epigenetic studies needs further evaluation and clarification. Unfortunately, no cell type–specific DNA methylation analysis is possible in our data set. There may be tissues and cell types that are more relevant and informative for methylation studies in asthma, but still it is not clear which tissues and cells these are. On the other hand, peripheral blood cells are easily accessible in children and patients, and our study may therefore be very useful in its explorative nature. However, effects restricted to small subpopulations of the blood cells such as Tregs will probably not be detected. This may be the case for FOXP3 for which in purified Treg cells differential methylation in relation to air pollution was shown previously [40]. However, the fact that differences in the DNA methylation status between groups are visible in this crude analysis is reassuring but should by no means imply that other cell sources would not be more suitable.

Overall, we performed a number of analyses and by averaging all CpGs within one amplicon and subsequently relating the observed number of significant results to the stochastically expected number of positive associations by chance, we addressed the issue of multiple testing [41], while at the same time, the exploratory character of this study was not reduced. For a limited number of findings, replication in a second data set was performed and the results confirmed our initial results. Our results are comparable with the data from the few studies published in this field.

The magnitude of the differences in methylation between the investigated strata was in general quite modest in our study when compared with differences observed in cancer studies [42, 43]. In some cases, the magnitude of effects was within or close to the error range of pyrosequencing. However, the results are very well in line with other studies in complex diseases. For CD14, average methylation at an intragenic CpG island was reported to change from 5.5% to 6.8% within 8 years [36]. Methylation of a CpG in the ALOX12 gene in children with persistent wheezing was 5.1% compared to 6.3% in children who never wheezed [44]. A further gene, PTGDR was identified to be hypomethylated in allergic patients by around 2% [45]. A farm environment was associated with an approximately 2% lower DNA methylation degree at the CD14 promoter in placental tissue [33]. This result would also support the observed trend in our study for an overall lower DNA methylation degree in various tissues of the child when exposed in utero to a farm environment. However, so far only a few genes have been analyzed for their association with the farm environment not allowing current to draw definite conclusions from the available data. In addition, effects may have been masked by other environmental factors influencing DNA methylation, for example, passive smoke exposure [31]. The functional relevance of such small differences in methylation and their effect on disease development need to be further investigated. It remains questionable whether they alone can explain the varying phenotypes or whether a cascade of several different epigenetic changes is required. Furthermore, it is intriguing that genes strongly associated with a disease due to DNA mutations also show disease-specific DNA methylation patterns. In a genome-wide methylation analysis, it has recently been shown that methylation patterns in SERPINA1 are strongly associated with COPD. SERPINA1 is the gene coding for alpha1-antitrypsin, and genetically determined alpha1-antitrypsin deficiency is the strongest genetic susceptibility factor for the development of COPD [46].

In conclusion, our study is a further step in deciphering epigenetic contributions to asthma development in childhood. In our analysis, the strongest effect was observed for time trends in methylation of specific asthma- and allergy-related genes. However, also farming seems to influence methylation in a number of the genes and gene families selected for investigation. It suggests the feasibility of such studies in asthma, and it highlights that a strong protective environmental factor, such as growing up on a farm, which is associated with disease development influences methylation and that timing of DNA collection for epigenetic analysis is crucial.

Aknowledgments

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Aknowledgments
  7. Authors' contributions
  8. Conflict of interest
  9. References
  10. Supporting Information

This work was supported by funding from the European Union, EU FP7 KBBE-2007-2-2-06 (EFRAIM), and the German ministry of education and research (BMBF) as part of the national genome research network (NGFN), with grant NGFN 01GS0810, and the German Research Council (DFG) by grant SFB587, project B16. S. Michel was also supported by a PINA fellowship.

Authors' contributions

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Aknowledgments
  7. Authors' contributions
  8. Conflict of interest
  9. References
  10. Supporting Information

Sven Michel contributed to study execution, performed the data analyses and drafted the first version of the manuscript; Florence Busato and Nicolas Mazaleyrat contributed to the study execution, Jon Genuneit was involved in the data analyses; Juha Pekkanen, Anne M. Karvonen, Maija-Riitta Hirvonen, Jean-Charles Dalphin, Josef Riedler, Charlotte Braun-Fahrländer, Roger Lauener and Juliane Weber contributed to data collection and study management; Erika von Mutius contributed to data collection and was responsible for the design of the PASTURE study; Michael Kabesch and Jörg Tost designed the study, contributed to the data collection and interpretation and wrote the final version of the manuscript.

Conflict of interest

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Aknowledgments
  7. Authors' contributions
  8. Conflict of interest
  9. References
  10. Supporting Information

All authors declare that they have no competing financial or personal interests.

The PASTURE study group (in alphabetical order by study center): Department of Environmental Health, National Institute for Health and Welfare, PO Box 95 FIN-70701 Kuopio, Finland (Anne Hyvärinen, Sami Remes, Pekka Tiittanen); Department of Environmental Science, Inhalation Toxicology Laboratory, University of Eastern Finland, PO Box 1627, Kuopio, Finland (Marjut Roponen); Université de Franche-Comté, Department of Respiratory Disease, University Hospital, 25000 Besancon, France (Marie-Laure Dalphin, Vincent Kaulek); Ulm University, Institute of Epidemiology and Medical Biometry, Helmholtzstraße 22, D-89081 Ulm, Germany (Gisela Büchele); LMU Munich, University Children's Hospital, Lindwurmstrasse 4, D-80337 Munich, Germany (Martin Depner, Markus Ege, Bianca Schaub); Department of Clinical Chemistry and Molecular Diagnostics, Philipps University of Marburg, Marburg, Germany (Petra Pfefferle, Harald Renz); Swiss Tropical and Public Health Institute, Socinstr. 57, PO Box, 4002 Basel, Switzerland (Sondhja Bitter, Georg Loss); University of Basel, Petersplatz 1, 4003 Basel, Switzerland (Sondhja Bitter, Georg Loss); Christine Kühne-Center for Allergy Research and Education, Hochgebirgklinik Davos, Herman-Burchard-Str. 1, CH.7265 Davos-Wolfgang (Remo Frei, Caroline Roduit); Utrecht University, Institute for Risk Assessment Sciences (IRAS), Division of Environmental Epidemiology, PO Box 80178, 3508TD, Utrecht, The Netherlands (Gert Doeckes).

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Aknowledgments
  7. Authors' contributions
  8. Conflict of interest
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Aknowledgments
  7. Authors' contributions
  8. Conflict of interest
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
all12097-sup-0001-TableS1-S6.docWord document286K

Table S1. Description of investigated regions

Table S2. Methylation values (mean and standard deviations in %) of 4.5 year samples comparing farming and asthma status.

Table S3. Methylation values (mean and standard deviations in %) of 4.5 year samples comparing asthma status within substrata.

Table S4. Replication set – time trends in DNA methylation from cord blood and whole blood at age 4.5 years

Table S5. Combined analyses – time trends in DNA methylation from cord blood and whole blood at age 4.5 years

Table S6. Characteristics of the population used for the initial and the replication set.

all12097-sup-0002-FigureS1.tifimage/tif1594KFigure S1. DNA methylation patterns in main strata.
all12097-sup-0003-FigureS2.tifimage/tif2006KFigure S2. Pairwise correlation across loci.
all12097-sup-0004-DataS1.docWord document247K 

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.