Edited by: Hans-Uwe Simon
X-chromosome Forkhead Box P3 polymorphisms associate with atopy in girls in three Dutch birth cohorts
Version of Record online: 21 DEC 2009
© 2009 John Wiley & Sons A/S
Volume 65, Issue 7, pages 865–874, July 2010
How to Cite
Bottema, R. W. B., Kerkhof, M., Reijmerink, N. E., Koppelman, G. H., Thijs, C., Stelma, F. F., Smit, H. A., Brunekreef, B., Van Schayck, C. P. and Postma, D. S. (2010), X-chromosome Forkhead Box P3 polymorphisms associate with atopy in girls in three Dutch birth cohorts. Allergy, 65: 865–874. doi: 10.1111/j.1398-9995.2009.02291.x
- Issue online: 1 JUN 2010
- Version of Record online: 21 DEC 2009
- Accepted for publication 10 November 2009
To cite this article: Bottema RWB, Kerkhof M, Reijmerink NE, Koppelman GH, Thijs C, Stelma FF, Smit HA, Brunekreef B, van Schayck CP, Postma DS. X-chromosome Forkhead Box P3 polymorphisms associate with atopy in girls in three Dutch birth cohorts. Allergy 2010; 65: 865–874.
Background: The Forkhead Box P3 (FOXP3) gene, located on the X-chromosome, encodes a transcription factor that directs T cells toward a regulatory phenotype. Regulatory T cells may suppress development of atopy. We evaluated whether single-nucleotide polymorphisms (SNPs) of FOXP3 are associated with atopy development in childhood.
Methods: Seven SNPs in FOXP3 were genotyped in 3062 children (51% boys) participating in the Allergenic study, which consists of three Dutch birth cohorts (PIAMA, PREVASC and KOALA). Association of FOXP3 SNPs with total serum IgE and sensitisation (presence of specific serum IgE to egg, milk, and indoor, i.e. house-dust mite, cat, and/or dog allergens) was investigated at ages 1, 2, 4, and 8. Analysis of variance and logistic regression were performed, stratified for gender.
Results: Our most consistent finding was observed for sensitisation to egg and indoor allergens. In girls, five FOXP3 SNPs (rs5906761, rs2294021, rs2294019, rs6609857 and rs3761548) were significantly associated with sensitisation to egg at ages 1 and 2 and with sensitisation to indoor allergens at age 2 (P < 0.05), but not at 4 and 8, a finding that was observed across the three cohorts. Rs5906761 and rs2294021 were associated with remission of sensitisation to food allergens in boys, as tested in the PIAMA cohort.
Conclusion: This is the first study showing across three cohorts that X-chromosomal FOXP3 genotypes may contribute to development of sensitisation against egg and indoor allergens in girls in early childhood. In addition, an association with remission of sensitisation to food allergens existed in boys only.
analysis of variance
forkhead box P3
immune dysregulation polyendocrinopathy enteropathy X-linked syndrome
regulatory T cell
Atopic diseases are complex in origin, i.e. development of disease results from interplay between genetic variants and environmental factors. Candidate genes for atopy have been investigated to understand the pathogenesis of atopy and to potentially open ways for novel preventive strategies and treatments.
CD4+ CD25+ regulatory T cells (Treg) play a key role in balancing immune responses to maintain or acquire tolerance against allergens (1, 2). Therefore, insufficient numbers or a compromised function of Tregs may contribute to the development and persistence of atopy and asthma (1–4). The development and function of Tregs is programmed by a transcription factor Forkhead Box P3 (FOXP3) (5). Genetic mutations in the FOXP3 gene cause immune dysregulation, poly-endocrinopathy, enteropathy, X-linked syndrome (IPEX), which is associated with high serum IgE levels, atopic dermatitis, eosinophilia (6), and food allergy (7, 8). This suggests that genetic variation in the FOXP3 gene may have an important influence on atopy development. The FOXP3 chromosomal location Xp11.23 has been linked to asthma and atopy (9), which further supports a possible role of FOXP3 in atopic disease.
Atopy manifests itself serologically as elevated IgE levels and presence of specific IgE to common environmental allergens. The prevalence of elevated total IgE and specific IgE differs between boys and girls, boys showing higher prevalence than girls (10, 11). This may be attributed to gender-specific exposures, e.g. sex hormones, but may also result from involvement of genes from either sex chromosome. As FOXP3 is located on the X-chromosome, it may be involved in gender differences observed in atopy.
Recently, the presence of Tregs was suggested to be important not only in development of atopy, but also in development of tolerance to food allergens. Children who have outgrown food allergy were found to have more Tregs compared to children with persisting food allergy (12). Thus, FOXP3 may be important in the development of tolerance to food allergens and remission of sensitisation.
We aimed to test three hypotheses: (i) FOXP3 polymorphisms associate with the prevalence of elevated total serum IgE and specific IgE; (ii) because boys have only one X-chromosome, the associations will be sex specific; (iii) FOXP3 polymorphisms associate with remission of sensitisation to food allergens. There are no reports of common FOXP3 polymorphisms with impact on gene or protein function; therefore, we selected haplotype tagging single-nucleotide polymorphisms (SNPs). These were genotyped in 3062 children participating in our genetic cohort study [Allergenic (13)], composed of three well-characterized Dutch birth cohorts. We assessed earlier mentioned hypotheses at ages 1 through 8.
The Allergenic study includes three prospective Dutch birth cohorts of similar design, i.e. PIAMA (14), PREVASC (15, 16), and KOALA (17) [see (13) for a summarized description of the cohorts]. Genetic studies were approved by local medical ethics committees of participating institutes. All parents provided written informed consent.
Parents completed yearly distributed questionnaires derived from ISAAC (18) about allergic symptoms in the child. Information about general health, indoor environments like environmental tobacco smoke and pet exposure, socioeconomic characteristics, and demographic factors was also obtained by these questionnaires.
Total and specific IgE levels were determined in capillary or venous blood collected at ages 1, 4, and 8 in PIAMA, ages 1, 2, and 4 in PREVASC, and ages 1 and 2 in KOALA. Total IgE levels were measured by radioimmunoassay as described previously (19–21) and expressed as international units per milliliter (1 IU representing 2.4 ng of IgE). Specific IgE levels to house-dust mite (Dermatophagoides pteronyssinus), cat (Fel d1), dog (Can f1), egg, and milk were measured by Radio Allergo Sorbent Test. All analyses were performed by Sanquin Research (central laboratory of the blood banks, Amsterdam, the Netherlands).
The following atopic phenotypes were studied: total serum IgE, sensitisation to indoor aeroallergens (house-dust mite, cat, and/or dog), and food sensitisation to egg and milk allergens. Sensitisation was defined as presence of serum-specific IgE to the respective allergens using a cutoff value of 0.35 IU/ml.
Remission of sensitisation to food allergens
We defined children with remission of sensitisation to food allergens as children with sensitisation to egg and/or milk allergens at ages 1 and/or 4, but not at age 8. These children were compared to children with persistent sensitisation to food allergens, i.e. specific IgE to egg and/or milk allergens at ages 1 and/or 4 and at 8.
Haplotype tagging SNPs were selected from HapMap database (22). The HapMap data are based on genotype data from 30 mother–father–child trios from the Centre d'Etude du Polymorphisme Humain (CEPH) collection, which are US Utah residents with ancestry from northern and western Europe. These genotype data cover variation in the human genome with an average r2 of 0.95 (23). The tag SNPs rs3761548 and rs3761549 were selected by importing the FOXP3 genotype data of the CEPH population into haploview (version 3.2, 13 April 2005) on the following terms: minor allele frequency ≥0.1, Hardy–Weinberg equilibrium >0.05, and r² > 0.8. Additionally, FOXP3 SNPs with a reported minor allele frequency ≥0.1 in Caucasian populations were selected from database SNPper (http://snpper.chip.org, October 2004).
Genomic DNA was extracted from buccal swabs or blood using standard methods (24). DNA was amplified by using REPLI-g UltraFast technology (Qiagen™; Benelux BV, Venlo, The Netherlands). Genotyping was performed by Competitive Allele-Specific PCR using KASPar™ genotyping chemistry, performed under contract by K-Biosciences (Herts, UK).
Because FOXP3 is located on the X-chromosome, analyses were stratified for gender. Analyses were performed in pooled data and in the three cohorts separately. Total serum IgE levels were logarithm transformed to approximate a normal distribution. Associations of FOXP3 genotypes with serum IgE levels and sensitisation at ages 1, 2, 4, and 8, and with the development of tolerance to food allergens at age 8, were assessed by analysis of variance (anova) and logistic regression analysis as appropriate. The associations of FOXP3 genotypes with sensitisation best fitted a logistic regression model with the number of alleles as a continuous variable; odds ratios (ORs) were presented per minor allele and adjusted for duration of breastfeeding, the presence of pets, cigarette smoke exposure, and older siblings at home and day care attendance in the first year of life. We present random-effects estimates of pooled effect sizes that assume the effect of FOXP3 on sensitisation and total IgE to vary between the three different cohorts, allowing a cohort-specific intercept and SNP effect term to be estimated. The functional effect of FOXP3 genotypes and the direction of its influence on atopy or tolerance development are unknown; therefore, all tests were two sided and considered significant at a P-value <0.05. Calculations were performed using spss 14.0 statistical software. Analyses of the pooled data were conducted using R version 2.9.0.
Haplotype frequencies were estimated by the EM-algorithm. The difference in frequency distributions between sensitized children and children who were not sensitized was evaluated by a log-likelihood ratio test, which was adjusted for hemizygosity in boys (in house software).
Non-Caucasian children (5.7%) were excluded from analyses to avoid spurious effects resulting from population stratification. Characteristics and a summary of the atopic phenotypes of the participants are presented in Table 1.
|Total serum IgE||n||Geom mean (interq range)¶||n||Geom mean (interq range)¶||n||Geom mean (interq range)¶||P†|
|1 year||171||6 (2–13)||114||7 (3–13)||318||5 (2–10)||0.007|
|2 years||–||–||175||9 (3–21)||313||10 (3–31)||0.78|
|4 years||304||31 (10–85)||103||16 (6–130)||–||–||<0.001|
|8 years||347||60 (22–200)||–||–||–||–||–|
|1 year||172||9 (2–29)||103||10 (4–29)||334||8 (3–17)||0.22|
|2 years||–||–||168||15 (5–36)||343||15 (5–48)||0.90|
|4 years||362||39 (14–104)||93||22 (10–64)||–||–||0.001|
|8 years||356||72 (26–225)||–||–||–||–||–|
|Sensitisation**||n||Prevalence (%)††||n||Prevalence (%)††||n||Prevalence (%)††||P†|
Single-nucleotide polymorphism selection and genotyping
The SNPs selected for FOXP3, data source, allele frequencies, and linkage disequilibrium are presented in Table 2 and Fig. 1. In girls, all FOXP3 genotypes were in Hardy–Weinberg equilibrium (chi-square test; P ≥ 0.01). Quality of genotype data was guaranteed by standards of K-Biosciences. Sixteen samples were genotyped in both genomic and amplified DNA and revealed <1% genotyping error. Allele frequencies did not significantly differ between the cohorts, neither between boys and girls (chi-square test; P > 0.05) and were similar to those previously reported in Caucasians (http://snpper.chip.org/bio/find-gene). Eight children with missing information on gender and 42 boys with heterozygosity were excluded from further analyses. Heterozygous genotypes in boys can be ascribed to <1% genotyping errors (113 of 21.434 genotypes). Because of linkage disequilibrium, association test results of rs4824747 and rs6609857 were very similar to the results of rs3761549 and rs2294019 (r2 = 0.97 and 0.99, respectively), and these are therefore not presented.
Forkhead Box P3 and total serum IgE
In pooled data, FOXP3 SNPs were not significantly associated with serum IgE level at any age, neither in girls nor in boys. In the KOALA cohort separately, rs3761549 was associated with IgE level at age 1 in girls (P = 0.03); girls homozygous for the minor allele had a significantly higher IgE level (geometric mean 19.6 IU/ml) compared to heterozygous girls (geometric mean 5.2 IU/ml), and girls homozygous for the wild-type alleles had the lowest IgE levels (geometric mean 4.6 IU/ml). These findings were not observed in other age groups, neither in other cohorts.
Forkhead Box P3 and sensitisation
Ages 1 and 2
Table 3 shows results of association analyses of FOXP3 SNPs and sensitisation at age 1 in girls (no associations were observed in boys, data not shown); Table 4 shows results at age 2 in girls and boys. We observed associations for multiple FOXP3 SNPs, and the associations existed with various allergens at different ages. FOXP3 SNPs were associated with sensitisation to egg at age 1 and 2 and with sensitisation to indoor allergens at age 2. The minor allele of rs3761548 associated with a decreased risk of sensitisation, whereas all other minor alleles associated with an increased risk. The associations of FOXP3 SNPs with prevalence of sensitisation were different between girls and boys. In contrast to our expectations, the associations were more prominent in girls. For instance, sensitisation to indoor allergens at age 2 significantly associated with FOXP3 SNPs rs3761548 and rs2294019 in girls (Table 4A), but not in boys (Table 4B). Sensitisation to egg allergens at age 1 significantly associated with FOXP3 SNP in girls, but no significant associations were observed in boys at this age (data not shown). Similarly, associations of FOXP3 SNPs were observed for egg sensitisation at age 2 in girls, but only with rs3761549 in boys from the KOALA cohort. Associations were observed in pooled data, and significant associations or trends with the same direction of effects were found in separate cohorts. For instance, Table 4 shows an OR of 4.7 (P = 0.001), a significant association in girls between rs3761549 and egg sensitisation in the total cohort, an association that was in the same direction in both PREVASC (OR 5.4, P = 0.08, n = 85) and KOALA (OR 4.7, P = 0.005, n = 288). Associations between SNPs in FOXP3 and sensitisation were comparable when using different cutoff levels for specific IgE, i.e. 0.35 and 0.7 IU/ml (data not shown).
|Girls SNP||Egg sensitisation|
|KOALA (n = 291)||PREVASC (n = 71)||PIAMA (n = 157)||Pooled (n = 519)|
|OR 95% CI||P†||OR 95% CI||P†||OR 95% CI||P†||OR 95% CI||P‡|
|rs5906761||0.7 0.29–1.50||0.318||15.8 1.0–247.7||0.049||4.1 1.1–15.4||0.037||1.4 0.7–2.5||0.316|
|rs3761548||0.6 0.2–1.6||0.309||0.1 0.0–1.9||0.132||0.44 0.1–1.7||0.226||0.5 0.3–1.0||0.049|
|rs3761549||4.9 1.8–13.4||0.002||2.1 0.2–28.2||0.567||–§||1.5 0.5–4.3||0.446|
|rs2294021||0.7 0.3–1.5||0.346||3.5 0.6–20.9||0.164||4.5 1.1–18.2||0.037||1.4 0.7–2.5||0.327|
|rs2294019||0.5 0.2–1.4||0.199||5.0 1.0–24.4||0.048||3.9 1.2–12.5||0.022||1.5 0.8–3.1||0.241|
|Girls SNP||Milk sensitisation|
|KOALA (n = 299)||PIAMA (n = 157)||Pooled (n = 456)|
|OR 95% CI||P†||OR 95% CI||P†||OR 95% CI||P‡|
|rs5906761||0.9 0.4–1.9||0.692||1.1 0.7–1.9||0.642||1.0 0.6–1.6||0.886|
|rs3761548||0.8 0.3–1.8||0.533||1.2 0.7–2.2||0.429||1.0 0.6–1.7||0.986|
|rs3761549||2.1 0.8–5.3||0.127||0.6 0.3–1.5||0.269||1.1 0.4–3.0||0.831|
|rs2294021||0.8 0.4–1.9||0.661||2.1 0.8–5.3||0.127||0.9 0.6–1.5||0.827|
|rs2294019||1.4 0.6–3.0||0.464||0.6 0.3–1.1||0.106||0.9 0.5–1.8||0.797|
|(A) Girls SNP||Egg||Milk||Indoor*|
|KOALA (n = 288)||PREVASC (n = 85)||Pooled (n = 373)||KOALA (n = 288)||KOALA (n = 260)||PREVASC (n = 149)||Pooled (n = 409)|
|OR 95% CI||P†||OR 95% CI||P||OR 95% CI||P‡||OR 95% CI||P†||OR 95% CI||P†||OR 95% CI||P†||OR 95% CI||P‡|
|rs5906761||1.2 0.5–3.1||0.716||0.8 0.2–2.4||0.645||0.9 0.4–1.9||0.803||1.0 0.6–1.7||0.873||1.6 0.8–2.9||0.152||1.2 0.5–2.9||0.751||1.5 0.9–2.4||0.117|
|rs3761548||0.3 0.1–0.9||0.039||0.5 0.1–2.1||0.350||0.4 0.2–0.9||0.030||1.0 0.6–1.6||0.926||0.4 0.2–0.7||0.005||1.3 0.5–3.5||0.566||0.5 0.3–0.9||0.024|
|rs3761549||4.7 1.6–13.9||0.005||5.4 0.8–35.3||0.080||4.7 1.9–11.6||0.001||1.2 0.6–2.6||0.588||1.7 0.8–3.8||0.177||0.3 0.0–2.4||0.257||1.2 0.5–2.6||0.660|
|rs2294021||1.2 0.5–3.0||0.746||1.1 0.4–3.4||0.873||1.0 0.5–2.1||0.909||1.0 0.6–1.6||0.881||1.8 0.9–3.3||0.077||1.2 0.5–3.1||0.687||1.6 0.9–2.6||0.058|
|rs2294019||1.2 0.4–3.2||0.751||1.5 0.5–5.1||0.502||1.3 0.6–2.7||0.484||0.7 0.4–1.3||0.249||2.2 1.1–4.0||0.017||1.2 0.4–3.4||0.694||1.9 1.1–3.1||0.015|
|(B) Boys SNP||Egg||Milk||Indoor*|
|KOALA (n = 320)||PREVASC (n = 88)||Pooled (n = 408)||KOALA (n = 319)||KOALA (n = 278)||PREVASC (n = 149)||Pooled (n = 427)|
|OR 95% CI||P†||OR 95% CI||P†||OR 95% CI||P‡||OR 95% CI||P†||OR 95% CI||P†||OR 95% CI||P†||OR 95% CI||P‡|
|rs5906761||0.8 0.5–1.3||0.400||1.0 0.3–2.8||0.943||0.9 0.6–1.4||0.685||1.3 0.9–1.7||0.134||1.0 0.6–1.5||0.835||0.8 0.5–1.4||0.401||0.9 0.6–1.2||0.505|
|rs3761548||0.8 0.5–1.3||0.325||1.3 0.4–3.6||0.674||0.8 0.5–1.2||0.332||0.8 0.6–1.1||0.176||1.0 0.6–1.5||0.908||1.1 0.7–1.9||0.651||1.1 0.8–1.5||0.750|
|rs3761549||2.1 1.2–3.7||0.012||–§||1.7 1.0–2.9||0.055||0.9 0.5–1.6||0.755||1.2 0.6–2.3||0.631||1.2 0.6–2.5||0.570||1.1 0.7–1.8||0.568|
|rs2294021||0.9 0.5–1.4||0.532||0.9 0.3–2.7||0.898||1.0 0.6–1.4||0.839||1.4 1.0–1.9||0.055||0.9 0.6–1.4||0.789||0.7 0.4–1.2||0.250||0.9 0.6–1.2||0.381|
|rs2294019||0.7 0.4–1.3||0.319||0.5 0.2–1.7||0.260||0.8 0.5–1.2||0.286||1.3 1.0–1.8||0.088||0.7 0.4–1.2||0.198||0.7 0.4–1.3||0.207||0.7 0.5–1.1||0.099|
Ages 4 and 8
At age 4, no significant associations were found in girls and boys (data not shown). At age 8 in girls, rs3761549 was associated with sensitisation to indoor allergens (P = 0.04, OR = 0.5, data not shown). In boys at age 8, rs5906761 associated with milk sensitisation (P = 0.04, OR 1.8, data not shown).
Haplotype analyses confirmed information from the SNP analyses not adding any new information and are therefore not presented.
Forkhead Box P3 and the presence of remission of sensitisation to food allergens
In boys, minor alleles of rs5906761 and rs2294021 were more frequently observed in children with persistent sensitisation to food allergens, compared to children who had remission of sensitisation to food allergens (51.1%vs 27.9%, P = 0.02 and 51.1%vs 25.6%, P = 0.01, Fig. 2). No significant associations were observed in girls. Similar results were obtained for milk allergens only (data not shown).
This is the first large population-based cohort study that evaluated whether X-chromosome FOXP3 SNPs are associated with development of serum IgE, sensitisation, and remission of sensitisation to food allergen. We observed gender-specific associations of FOXP3 SNPs with presence of serum-specific IgE to egg and indoor allergens, a sexual dimorphism that was most prominent at ages 1 and 2. The associations between FOXP3 SNPs and sensitisation were present in girls specifically. This finding was present in a large birth cohort of 3062 Dutch children that pooled data from three cohorts and showed results in the same direction across the three cohorts. In addition, we found that FOXP3 SNPs were associated with remission of sensitisation to food allergens in boys. This suggests that there exists a genetic regulation of the sexual dimorphism observed in atopy development in childhood.
Because we are the first to evaluate FOXP3 polymorphisms in relation to atopy development during childhood, we cannot directly compare our results to other studies. However, importance of the FOXP3 gene in early IgE development is strongly suggested by findings in FOXP3 mutant mice and children with IPEX syndrome, which both show early life elevated serum IgE levels (25, 26). One study in adults evaluated a FOXP3 microsatellite polymorphism with serum IgE levels in a small group of asthma patients and found no association (27). Although studying asthma patients, it is consistent with our study because we also did not find significant associations of the haplotype tagging SNPs in FOXP3 with total serum IgE in children 8 years of age.
The strengths of our study are its large sample size and prospective follow-up, which enabled us to evaluate the influence of FOXP3 polymorphisms at four cross-sectional time points during childhood from age 1 through 8. This resulted in replication of the pooled results in the three separately investigated cohorts, found at two cross-sectional time points, i.e. in children aged 1 and 2. Moreover, the prospective follow-up gave us the unique opportunity to investigate the role of FOXP3 in the remission of sensitisation to food allergens. Finally, selection of haplotype tagging SNPs has made it very unlikely that we have missed an important signal from this gene.
To appreciate our results, we should also consider some potential limitations. First, our study represents a selected population with a relatively high number of children with atopic parents compared to the general population. As a result of recruitment strategies, children with an allergic mother were overrepresented in the PIAMA cohort, as were children with a first-degree family member with asthma in the PREVASC cohort. Our results may thus not be fully representative of the general population, yet previous genetic analyses in our cohorts have shown that irrespective of cohort differences, genetic associations with atopy phenotypes exist in the same direction within the three cohorts (13).
Second, we have not corrected for multiple comparisons, and because we tested for associations of five SNPs and five phenotypes, one might argue that some of our results may result from type I statistical error. However, because of linkage disequilibrium between the evaluated SNPs and potential clustering of outcome variables, for example, elevated IgE levels and the presence of specific IgE, the statistical tests performed are not completely independent. Moreover, the associations were found with the same SNPs and with the same direction of effects in three cohorts at two different ages, which is convincing when considering that replication of genetic associations has been proven to be difficult in the previous years (28). Furthermore, our findings were robust to the use of different cutoff levels for specific IgE. Taken together, this argues against false-positive findings. The assessment of remission of sensitisation to food allergens could only be tested in the PIAMA cohort that performed a follow-up of 8 years. Therefore, these findings warrant replication in other birth cohorts.
A third limitation to our study design may be the variable proportion of the cohorts that participated in each age group, e.g. at age 1, all cohorts contributed to the IgE measurements, whereas at age 8, only PIAMA could be evaluated. As a result, age groups may have been subject to variable selection effects. However, the association results at age 1 and 2 show similar direction of the effects in all three separate cohorts, and therefore do not likely reflect an effect of a selected cohort.
We chose to study the association of FOXP3 SNPs with sensitisation to food and indoor aeroallergens separately, because the prevalence of sensitisation to food allergens shows a clearly different age-related pattern compared to sensitisation to aeroallergens (29), and we hypothesized that allergen exposures and pathogenesis of sensitisation are different among these types of allergens. Furthermore, several studies indicate that early sensitisation to egg, not milk, is associated with increased sensitisation to aeroallergens, rhinitis, and asthma later in life (30, 31), suggesting that there may also be a distinct mechanism underlying sensitisation to different food allergens. Therefore, and because the prevalence of sensitisation to milk allergens was found to be significantly different between the cohorts, we evaluated FOXP3 in relation to sensitisation to egg and milk allergens separately.
Detection of genetic association in boys with remission of sensitisation to egg and/or milk supports our initial hypothesis; however, the main effects of FOXP3 polymorphisms on sensitisation were observed in females. This is in contrast to our expectations, because the X-linked IPEX syndrome expresses itself in hemizygous boys, but rarely in girls (6). The pathogenetic or methodological explanation of our observations is unknown, but can be speculated upon. First, Wildin et al. (6) have proposed several explanatory mechanisms for FOXP3 to be involved in susceptibility to immunological disease in females. They observed variable expression of disease in FOXP3 heterozygous female mice that were also heterozygous to a mutation of another X-chromosomal gene, γC. This suggests that genes on the alternative X-chromosome can interact with the FOXP3 locus through influence on X-inactivation thereby modifying disease outcome. Further explanations proposed by these authors (6) included other epigenetic mechanisms such as (i) sex-specific hormonal influences or (ii) imprinting, where allelic expression is dependent on the sex of the parent from whom the specific allele was inherited. Interestingly, it has recently been shown that Y-chromosome-linked polymorphisms may differentially regulate the expression of X-linked and autosomal genes (32). Thus hypothetically, FOXP3 gene expression may be differentially regulated in males and females through suppressive or activation mechanisms regulated by the Y-chromosome. In addition to these pathogenic mechanisms, an alternative explanation for not finding a comparable effect in boys could be less statistical power to detect an effect that increases with the number of alleles, because girls have twice the number of alleles as boys. Finally, the associations found can be spurious as a result of multiple testing, although we consider this least likely because the associations were observed in three birth cohorts, both at age 1 and 2 and at two specific IgE cutoff levels.
In conclusion, our results suggest that FOXP3 polymorphisms contribute to both development of sensitisation to egg and aeroallergens early in childhood and remission of sensitisation to food allergens in a gender-specific way. Our observation that FOXP3 may be important early in the development of atopy should stimulate future study on FOXP3 function and Tregs focussing on early pathogenic mechanisms of atopy and tolerance development. This should then encompass studies in both female and male animals and humans that not only evaluate genetic association, but also include epigenetic mechanisms and interaction with sex hormones involved in atopic sensitisation or tolerance development.
The authors thank the children and parents of the PIAMA, PREVASC, and KOALA study for their participation. In addition, we acknowledge the field workers, secretaries, and scientific collaborators dedicated to the PIAMA, PREVASC, and KOALA cohorts, Marcel Bruinenberg for his advice on DNA isolation, processing, and genotyping, and Ilja Nolte for her advice on haplotype analyses.
This study was financially supported by ZonMW grant number 912-03-031. G.H. Koppelman is supported by a Zon-Mw VENI grant, number 91656091.
- 24Molecular cloning, 3rd edn. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, 2001., .