Article Text

The protective effect of farm animal exposure on childhood allergy is modified by NPSR1 polymorphisms
1. S Bruce1,2,
2. F Nyberg3,4,
3. E Melén4,5,
4. A James6,7,
5. V Pulkkinen2,
6. C Orsmark-Pietras1,
7. A Bergström4,
8. B Dahlén6,7,
9. M Wickman4,8,
10. E von Mutius9,
11. G Doekes10,
12. R Lauener11,
13. J Riedler12,
14. W Eder9,13,
15. M van Hage14,
16. G Pershagen4,15,
17. A Scheynius16,
18. J Kere1,2,17
1. 1
Department of Biosciences and Nutrition, Karolinska Institutet, Stockholm, Sweden
2. 2
Department of Medical Genetics, University of Helsinki, Helsinki, Finland
3. 3
Astra Zeneca R&D Mölndal, Mölndal, Sweden
4. 4
Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden
5. 5
Astrid Lindgren Children’s Hospital, Karolinska University Hospital Solna, Sweden
6. 6
Department of Medicine, Division of Respiratory Medicine and Allergy, Karolinska University Hospital, Huddinge, Sweden
7. 7
Centre for Allergy Research, Karolinska Institutet, Stockholm, Sweden
8. 8
Sachs Children’s Hospital, Stockholm, Sweden
9. 9
University Children’s Hospital, Ludwig Maximilians University Munich, Munich, Germany
10. 10
Institute for Risk Assessment Sciences, Utrecht University, Utrecht, The Netherlands
11. 11
Children’s Hospital, Zurich University, Zurich, Switzerland
12. 12
Children’s Hospital Schwarzach, Schwarzach, Austria
13. 13
Children’s Hospital, Paracelsus Medical University, Salzburg, Austria
14. 14
Clinical Immunology and Allergy Unit, Department of Medicine Solna, Karolinska Institutet and University Hospital, Solna, Sweden
15. 15
Department of Occupational and Environmental Health, Stockholm County Council, Stockholm, Sweden
16. 16
Clinical Allergy Research Unit, Department of Medicine Solna, Karolinska Institutet and University Hospital, Solna, Sweden
17. 17
Clinical Research Centre, Karolinska University Hospital, Huddinge, Sweden
1. S Bruce, Department of Biosciences and Nutrition, Hälsovägen 7-9, 14157 Huddinge, Sweden; sara.bruce{at}biosci.ki.se

## Abstract

Background: Little is known about the asthma candidate gene neuropeptide S receptor 1 (NPSR1) in relation to environmental exposures, but recent evidences suggest its role as an effect modifier.

Objectives: To explore the interaction between NPSR1 polymorphisms and environmental exposures related to farming lifestyle and to study the in vitro effects of lipopolysaccharide (LPS) stimulation on NPSR1 expression levels.

Methods: We studied 3113 children from PARSIFAL, a European cross-sectional study on environmental/lifestyle factors and childhood allergy, partly focused on children brought up on a farm. Information on exposures and outcomes was primarily obtained from parental questionnaires. Seven tagging polymorphisms were analysed in a conserved haplotype block of NPSR1. Multivariate logistic regression was used to evaluate a multiplicative model of interaction. NPSR1 protein and messenger RNA (mRNA) levels in monocytes were measured after LPS stimulation by fluorescence activated cell sorting (FACS) and quantitative real-time polymerase chain reaction (PCR).

Results: A strong interaction was seen between current regular contact to farm animals and several NPSR1 polymorphisms, particularly rs323922 and rs324377 (p<0.005), with respect to allergic symptoms. Considering the timing of initiation of such current regular farm animal contact, significant interactions with these and two additional polymorphisms (SNP546333, rs740347) were revealed. In response to LPS, NPSR1-A protein levels in monocytes were upregulated (p = 0.002), as were NPSR1-A mRNA levels (p = 0.02).

Conclusions: The effect of farm animal contact on the development of allergic symptoms in children is modified by NPSR1 genetic background.

## Statistics from Altmetric.com

Epidemiological studies have shown that children living on farms have a lower prevalence of IgE mediated allergic diseases.15 The protective effects have specifically been correlated to contact with farm animals and farm milk consumption,510 with exposure to endotoxins or moulds as possible molecular explanations.7 1113 The timing is important, with prenatal and early life exposures showing particularly strong protective effects.6 14 Interestingly, several innate immunity related receptors, which directly interact with bacterial components, have been reported to have heterogeneous genetic effects suggestive of gene–environment interactions.1517

The neuropeptide S (NPS) receptor 1 (NPSR1; earlier named GPR154 or GPRA) was positionally cloned as an asthma candidate gene in families from Finland and Canada, and has been replicated for both childhood and adult populations of diverse geographic origin.1824 Neuropeptide S (NPS), the ligand, shows overlapping expression patterns with NPSR1 in several tissue types, including the epithelium of human colon and bronchi.25 26 NPSR1 is further expressed by macrophages and stimulation with NPS induces macrophage phagocytosis and chemotaxis.27 Stimulation of peripheral blood mononuclear cells (PBMC) by lipopolysaccharide (LPS), a potential molecular proxy for farm animal exposure, has been shown to increase mRNA levels of NPSR1.27 When the downstream target genes of NPSR1 signalling were studied, genes with ontology terms relating to immune response and chemotaxis were found to be overrepresented.28 Interestingly, recent genetic studies of NPSR1 have presented results suggesting gene–environment interactions.23 24

PARSIFAL (Prevention of Allergy Risk factors for Sensitisation In children related to Farming and Anthroposophic Lifestyle) is a European cross-sectional study designed to investigate the role of different lifestyles and environmental exposures in four groups: farm children, children from Steiner schools (who often have an anthroposophic lifestyle), and two corresponding reference groups.4 The large sample size and wealth of information on environmental exposures makes this a unique dataset for investigating gene–environment interactions. We have previously demonstrated that NPSR1 haplotypes are moderately associated with asthma and atopic sensitisation in this sub-sample of PARSIFAL.19 Here we explore potential effect modification by NPSR1 on environmental exposures related to the farming lifestyle. Since protective effects of the farming environment have been connected with endotoxins,7 12 13 we also investigated the effects of LPS stimulation on NPSR1 protein levels and mRNA expression levels in monocytes.

## METHODS

### Study design

This work is based on a sub-sample of children from the cross-sectional PARSIFAL study intended for gene–environment analyses. Recruitment and patient characteristics are described in detail elsewhere.4 In brief, the study included 14 893 children between 5–13 years old from four groups: farm children, Steiner school children, and two appropriate reference groups (see below), in five European countries (Austria, Germany, Holland, Sweden and Switzerland). Ethical approval for the study, including genetic analyses, was obtained in each country. The parents completed a detailed questionnaire on their child’s allergic diseases, infectious history and environmental exposures.4 6 2931 The present analysis included 3113 children with complete questionnaire data, adequate DNA material, and consent to genetic analyses. Information on non-participation and comparability between the whole PARSIFAL dataset and the genetic sub-sample can be found elsewhere.4 19

### Criteria for study group inclusion

Farm children included those children who currently were living on a farm and whose parents were running the farm. Their reference group was recruited from children in the same area but who did not meet the inclusion criteria for farm children. The Steiner group was recruited among children attending Steiner schools, whose parents often have an anthroposophic lifestyle. Their reference group was recruited from other schools in similar suburban/rural areas. For further information see Alfvén et al.4

### Health end point definitions and data on environmental exposures

Information on environmental exposures and health end points were reported by the parents of the children, except atopic sensitisation (see below). Children who had ever been diagnosed with asthma (by a physician), or with obstructive bronchitis more than once, were classified as having asthma, while children who reported sneezing, runny or stuffy nose in combination with itchy eyes in the last 12 months without having a cold at the same time were considered to have rhinoconjunctivitis.4 Two common cut-off values for atopic sensitisation were used: allergen-specific IgE level ⩾0.35 kU/l and ⩾3.5 kU/l in either Phadiatop (a mix of common inhalant allergens) and/or fx5 (a mix of common food allergens; Phadia AB, Uppsala, Sweden) analyses. To increase the power of the interaction analyses, the term “allergic symptoms” was broadly defined as asthma and/or rhinoconjunctivitis based on the definitions above. The health endpoint definitions used in this study have been amply documented and used in several PARSIFAL reports.4 9 10 14 19

### Genotyping

Detailed information on the genotyping methodology, polymorphism selection and quality assessment is given elsewhere.19 In brief, seven tagging polymorphisms reported in the Laitinen et al study18 were chosen in the conserved haplotype block of NPSR1 and genotyped on a MALDI-TOF (matrix associated laser desorption ionisation time of flight) platform. The selected polymorphisms are rs323917, rs323922, rs324377, SNP546333, rs324384, rs324396, and rs740347.

### Analysis of NPSR1 protein expression after LPS stimulation

Peripheral blood from 10 healthy volunteers was drawn into EDTA tubes and 200 μl was transferred to siliconised polypropylene FACS tubes, to minimise monocyte adherence. Blood samples were mixed with 800 μl RPMI-1640 medium containing l-glutamine (0.29 mg/ml), penicillin (100 U/ml) and streptomycin (100 μg/ml) (Gibco Lifetechnologies, Paisley, UK) and incubated for 16 h at 37°C, with or without 10 ng/ml LPS (Sigma-Aldrich, St Louis, Missouri, USA). For staining of intracellular receptor epitopes, red blood cells were lysed and leucocytes were fixed in 4% paraformaldehyde and permeabilised in 0.5% saponin. NPSR1-A and -B rabbit antibodies and pre-immune serum negative controls were raised and characterised as previously described18 26 27 and used with a secondary PE labelled goat-anti-rabbit IgG antibody (Southern Biotech, Alabama, USA). Samples were analysed on a FACSCalibur flow cytometer with CellQuest software (Becton Dickinson, San Jose, California, USA). Cell populations were detected using laser side and forward scatter and the monocyte population was further confirmed using a CD14-fluorescein isothiocyanate (FITC) antibody (Beckton Dickinson). On average 2500 gated cells were analysed and results are shown as median fluorescence intensity (MFI) values with the pre-immune serum MFI subtracted. The study was approved by the ethical review board at Karolinska Institutet.

### Analysis of NPSR1 mRNA expression after LPS stimulation

Human monocytes were obtained from 10 healthy blood donors as previously described.27 Briefly, peripheral blood cells were isolated by density gradient centrifugations using Ficoll-Paque Plus (Amersham Biosciences, Uppsala, Sweden). PBMCs were seeded onto cell culture bottles and incubated in complete RPMI 1640 medium supplemented with 0.29 mg/ml l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin (Gibco Lifetechnologies) and 10% fetal bovine serum (Sera-Lab, Sussex, UK) for 10–20 min. Adherent monocytes were harvested by scraping, rested over night in complete RPMI medium at a density of 1.5×106 cells/ml, and cultured for 6 h and 24 h with or without 10 ng/ml LPS (Sigma) at 37°C. Total RNA was isolated with RNeasy Mini Kit (Qiagen, Hamburg, Germany) from 3–9×106 cells. cDNA was synthesised with TaqMan reverse transcription reagents (Applied Biosystems, Foster City, California, USA) and oligo dT primers. Quantitative real-time polymerase chain reaction (PCR) was performed with ABI PRISM 7500 Sequence Detection System applying SybrGreen chemistry (Applied Biosystems). The primer sequences for human NPSR1-A were 5′-CCCCCTCATCTACTGTGT CTTCA-3′ (forward) and 5′TCTCTCCCGGAACGTCA TTCT-3′ (reverse). The primer sequences for NPSR1-B and elongation factor 1 α (EF-1α) have been documented previously.27 The relative gene expression differences were calculated with the comparative ΔΔCT method using EF-1α as a control.32 The study was approved by the ethical review board at University of Helsinki.

### Statistical analysis

Multivariate logistic regression was used for analyses and thus departure from a multiplicative model of interaction on the odds ratio (OR) scale was investigated. The statistical significance of effect modification was tested by a likelihood ratio (LR) test for the improvement of goodness of fit for the model when including the interaction term. All confidence intervals were calculated at the 95% level, p = 0.05.

The evaluated main determinants were focused on the children’s lifetime or prenatal exposures, related to the farm environment. Throughout the manuscript, “current” refers to exposures that occurred during the last 12 months and “regular” refers to exposures that occurred at least once a week. The investigated exposures were: “ever regular farm animal contact”, “current regular farm animal contact”, “current regular visits to stable or barn”, “mother’s farm animal contact during pregnancy”, “mother worked regularly in stable or barn during pregnancy”, “ever consumption of farm milk”. For “current regular farm animal contact”, timing of the initiation of such regular exposure was further investigated (current regular contact with initiation in first year of life/after first year but more than a year ago/last 12 months), with no current contact as the reference category. The seven genotyped NPSR1 polymorphisms were evaluated as potential effect modifiers. The polymorphisms were included as variables with two genotype groups with reference “non-risk” genotype(s), determined from a main effect logistic regression analysis of each polymorphism separately (supplemental table 1). Two polymorphisms (rs323917 and rs740347) had such low minor allele frequencies that the rare minor allele homozygotes for analytical reasons were pooled with the heterozygotes and a dominant model was used. For the other five polymorphisms, a recessive model was found to estimate the strongest main risk effect.

Table 1 Prevalences of allergic phenotypes and farm related exposures among the PARSIFAL children, by study group

The following potential confounding factors were evaluated: country (Austria/Germany/Holland/Sweden/Switzerland), age (5–6/7–8/9/10–11/12–13 years), sex (male/female), heredity (mother’s or father’s reported asthma and/or rhinoconjunctivitis), parental education (low/medium/high), maternal smoking during pregnancy (yes/no), current smoking at home (yes/no), older siblings (no/one/two/three or more older siblings), and study group (farm/farm reference/Steiner school/Steiner reference). Relevant confounders were identified starting from the full model, with relevant confounding defined using the customary criterion of approximately 10% change in the investigated effect estimates. All analyses were carried out using the statistical software STATA 9.0 (College Station, Texas, USA).

Where evidence for interaction was detected on the genotype level, haplotype by environment analyses were performed in R using the haplo.glm algorithm in the haplo.stats package,33 which includes an expectation maximisation (EM) step to estimate probabilities for haplotypes. Evidence for an overall haplotype–environment interaction was obtained through an LR test for the improvement of goodness of fit for the model when including versus excluding the interaction term.

The global significance for the interaction between NPSR1 and timing of current regular contact with farm animals in determining risk of allergic symptoms was tested by permutation. The affection status was permuted 10 000 times, each followed by seven multivariate logistic regressions as described above, after which the number of nominally significant interactions was recorded (randomised χ2 tests). The proportion of permutation tests where the number of nominally significant LR tests exceeded or was equal to the original number of significant tests provided the global p value.

Paired Wilcoxon signed rank tests were used to analyse the effect of LPS stimulation on NPSR1 protein and mRNA levels.

## RESULTS

### Descriptive analyses and main effects

The prevalence of farm related exposures and allergic diseases in the four study groups are presented in table 1. As expected, the exposures were most common among farm children, but a substantial proportion of children in the other groups was also exposed. Of the investigated potential confounder variables, only gender, heredity and study group were identified as confounders of the investigated associations and interactions and were thus included in all analyses described below. In an initial analysis, protective main effects of the farm related exposures on allergic symptoms and atopic sensitisation (most pronounced for ⩾3.5 kU/l) were observed in this sub-sample of PARSIFAL (table 2). Overall genetic associations of NPSR1 haplotypes have previously been described in this population.19 In the gene–environment interaction analyses, our focus was on the haplotype tagging polymorphisms for a number of reasons. First, because of the ease of interpretation when using binary genetic coding—that is, two levels of genetic exposure and two levels of environmental exposures, which also increases the power of the analysis in that the cross-classification cell counts increase. Second, choosing a baseline in the regression becomes intuitive for single nucleotide polymorphisms (SNPs) where one allele or genotypic combination defines risk in relation to another, which is not true for haplotypes that are generally numerous and show complex association patterns. Third, since these polymorphisms were chosen as tags, they are actually fewer than the total number of haplotypes observed, reducing the risk of type I error. Minor allele frequencies and crude ORs for the genotypic model selected for each polymorphism for use in the interaction analyses are presented in supplemental table 1.

Table 2 Main environmental effects of farm related exposures on allergic symptoms and atopic sensitisation in all PARSIFAL children

### Interaction between farm animal exposure and NPSR1 polymorphisms

First, we assessed whether the protective effects of farm related exposures was influenced by NPSR1 polymorphisms. Strong effect modification for allergic symptoms was seen for regular farm animal contact (table 3). The strongest effect modification was seen for current regular farm animal contact, especially by rs323922 and rs324377 (p value for interaction  = 0.001 and 0.002). The protective effects of the other variables against development of allergic symptoms were not dependent on genotype (supplemental table 2). In a separate analysis for asthma and rhinoconjunctivitis, rs324396 seemed to differ somewhat between the phenotypes. However, the overall trends were similar and the combined variable for allergic symptoms was consistently showing more significant effects (table 3). No gene–environment interactions were observed for any of the main environmental determinants when atopic sensitisation was used as outcome (neither ⩾0.35 kU/l nor ⩾3.5 kU/l, data not shown).

Table 3 Effect modification of regular farm animal contact on allergic symptoms† by NPSR1 genotypes, among all PARSIFAL children

When investigating the effect of timing of initiation of current regular farm animal contact on allergic symptoms, significant effect modification were seen with rs323922, rs324377, SNP546333, and rs740347 (tables 4 and for asthma and rhinoconjunctivitis separately). If the current regular farm animal contact was initiated during the first year of life, there was a tendency towards a protective effect regardless of genotype, while if the contact was initiated after early infancy, the environmental effects differed substantially depending on NPSR1 background (fig 1). To investigate further the consistency of the interaction, a stratified analysis was performed for farm children and for all non-farm children in the study (Steiner school children combined with the two reference groups) separately, and a similar interaction trend was observed in both strata (table 4). The power was low for the separate analyses of asthma and rhinoconjunctivitis, resulting in broader confidence intervals and lack of cases for some combinations of exposure and genotype (table 5).

Figure 1 The effect of current regular farm animal contact on allergic symptoms is dependent on NPSR1 polymorphism in the PARSIFAL (Prevention of Allergy Risk factors for Sensitisation In children related to Farming and Anthroposophic Lifestyle) children. Odds ratios (OR) and 95% confidence intervals (CI) for the effect of timing of current regular farm animal contact in strata of rs324377 are shown. The two reference groups (no current regular exposure to farm animals, stratified by genotype), are indicated at OR = 1. The following group includes those with current regular farm exposure that was initiated during the first year of life, the next group after the first year until recently, and the last group initiated their current regular farm animal contact recently (last 12 months). All groups are compared to the reference group (that has no current farm animal contact) in their genotype stratum.
Table 4 The effect of initiation of current regular farm animal contact on allergic symptoms* is modified by NPSR1 genotypes, among all PARSIFAL children and in farmer and non-farmer subgroups
Table 5 The effect of initiation of current regular farm animal contact on asthma and rhinoconjunctivitis is modified by NPSR1 genotypes, among all PARSIFAL children

To get a composite view of the interaction on the haplotype level, the overall haplotype by environment interaction significance was calculated for ever regular farm animal contact (p = 0.06), current regular farm animal contact (p = 0.14), and timing of initiation of current regular farm animal contact (p = 0.01). Furthermore, the permuted p value for an overall interaction between NPSR1 and timing of initiation of current regular farm animal contact was 0.008 (10 000 permutations), thus indicating a significant global gene–environment interaction.

### NPSR1 expression after LPS stimulation

Since the protective effects of the farming environment have been attributed to high levels of endotoxins, for example, we hypothesised that the observed interaction between farm animal contact and NPSR1 might be linked to the innate immune response. Therefore, the direct in vitro effects of LPS stimulation on NPSR1 protein and mRNA levels were investigated. CD14+ PBMCs from 10 blood donors showed a definite pattern of upregulation of the NPSR1-A isoform after 16 h of LPS stimulation (p = 0.002, supplemental fig 1 A, B and D), as measured by FACS analysis, while we did not observe an upregulation for the B isoform (supplemental fig 1 A, C, and E). In the parallel mRNA investigations, where monocytes from 10 blood donors were stimulated with LPS, the NPSR1-A isoform was found to show a modest upregulation trend at 6 h (p = 0.02), while the B isoform response showed a similar but not significant trend (supplemental fig 1 F and G). The NPRS1 mRNA levels after 24 h of LPS stimulation showed a large variation for both isoforms and no trends could be found (data not shown).

## DISCUSSION

In this study we show how effects of an environment can differ depending on genetic background, with exposures to farm animals exerting protective or risk effects for allergic symptoms, depending on the inherited NPSR1 variant. Robust results on interaction were also presented for a variable describing the timing of initiation of current regular farm animal contact. It is clear that independent replication is needed; however, the temporal phenomenon observed here suggests that an environmental exposure can show a more homogeneous effect if introduced early in life, while different mechanisms may act later on to modify such effects by, for example, a person’s genetic setup. This was also supported by the finding that the protective effect of farm animal contact during pregnancy was not dependent on NPSR1 genotypes.

NPSR1 is expressed by macrophages,27 and thus has potential for direct or indirect interaction with environmental endotoxins. Since the protective effects of farm animal exposure have been connected with higher levels of endotoxins, for example, we investigated the in vitro effects of LPS and found that the NPSR1-A (but not B) protein levels were specifically upregulated by LPS in CD14+ PBMCs (monocytes). We further demonstrated that purified monocytes upregulate NPSR1-A (with a similar trend for NPSR1-B) mRNA levels after 6 h of LPS stimulation. Previously, the more heterogeneous cell population PBMCs have been shown to upregulate NPSR1 mRNA levels after 16 h of LPS stimulation for both isoforms.27

A limitation of this experiment is its small sample size, although it is larger than the previous study.27 It should also be noticed that several other environmental components typical for farm animal exposure might affect NPSR1 abundance in monocytes and macrophages, and would be worthy of study.3436 Further, the most relevant tissue or cell type for a functional study is not entirely clear. Here we focused on peripheral blood monocytes, since NPSR1 has been demonstrated to affect their function27 and because innate immunity in general has been of interest in previous studies with similar scope.14 15 Although more thorough studies are needed, all experimental results taken together suggest that NPSR1 levels are affected by LPS stimulation, and this can serve as a first molecular hypothesis for our epidemiological results.

The NPSR1 polymorphisms that were investigated in this study are located in the second intron of the gene, comprising a block of tight linkage disequilibrium that was originally identified as associated to asthma related traits.18 So far, no functional characteristics have been described for these polymorphisms, while many hypotheses can be proposed, including transcription factor binding, regulation of alternative splicing of the two isoforms A and B, and their potential linkage disequilibrium to a disease causing polymorphism. The heterogenic responses to LPS stimulation observed here may suggest that inherited NPSR1 variants have an effect. Our limited sample size makes this difficult to assess, but further subject and data collections are ongoing to address this important issue.

The large sample size of the PARSIFAL study and its wealth of information on environmental exposures make it suitable for assessment of gene–environment interactions. The subset of the study available for genetic analyses has earlier been concluded essentially representative.19 Further, we have previously noted that the frequencies of genetic variants across the countries were quite homogeneous, suggesting that population stratification is not expected.19 In the present analyses we investigated typical farm related environmental exposure in all the four study groups. The finding that timing of initiation of farm animal contact is important when studying NPSR1 modification was further analysed in farmer and non-farmer children, separately. It should be noted that when analysing only farmers, there are few individuals in the reference group (no current exposure), while there are fewer exposed for the non-farmers. Nevertheless, it is reassuring that both analyses present a similar interaction, despite the different general environments of the groups of children.

When effect modification between NPSR1 and farm related variables was analysed, we used both atopic sensitisation and allergic symptoms as outcome. However, it was only for allergic symptoms that we found evidence for effect modification. This could be due to chance, but of note is that some studies have suggested that different environmental exposures related to the farming lifestyle are responsible for protection against different types of allergic phenotypes.6 14 The definition of sensitisation by serology measurements may also be of importance. For completeness, we performed all analyses using two common cut-offs, ⩾0.35 kU/l and ⩾3.5 kU/l, but no genetic interaction effects were seen with either cut-off. We have also performed interaction analyses by stratifying for sensitisation (that is, allergic symptoms with or without detectable sensitisation) and there were trends for interaction in both groups, emphasising the independence of these two phenotypes (data not shown).

For allergic symptoms, we performed interaction analyses for asthma and rhinoconjunctivitis separately and also for a combined variable of the two. The incentive for the combination was to increase the number of cases, thus acquiring a higher statistical power to perform detailed gene–environment interaction analysis. Some of the observed small differences between the symptoms might be of biological interest, but we have to conclude that a larger, better powered study would be needed to make any such conclusions.

#### Electronic database information

“R” (http://www.r-project.org); haplo.stats package (http://mayoresearch.mayo.edu/mayo/research/schaid_lab/software.cfm)

We have recently used this population to evaluate the overall association between NPSR1 variants and childhood allergic disease.19 Here, we chose to consider how environmental effects could be modified by an inherited NPSR1 allele. This seemed intuitive, because the genetic setup is fixed at birth, while the environment is interchangeable. However, statistically the interaction effects could be viewed in both ways and the presented results implicitly mean that the genetic effects also differ depending on environment. Of interest is that the tag polymorphisms that showed association in our first study (rs324384, rs324396, distinguishing H1/H3 from rest) did not show significant interaction in this study. Further, when performing haplotype–environment analyses, the haplotypes showing effect modification (using H1 as reference) were H2 and H4, both of which did not show association in our previous analysis with the PARSIFAL children,19 but were associated with risk in the original NPSR1 identification study.18 These new findings emphasise the importance of considering gene–environment effects in association studies of complex diseases such as asthma and atopic sensitisation. Of note, a recent association study of NPSR1 presented flip-flop association patterns for different populations, possibly suggesting that different environments might cause the discrepant association patterns.24 Consideration of environmental exposures and simultaneously the timing of exposures may be more important than earlier perceived.

In conclusion, this study adds NPSR1 to the list of genes suggested to modify the protective effects of environmental exposures,1517 and further indicates that genetic effect modification of environmental exposures may be influenced by the timing of initiation of such exposures.

## Acknowledgments

The authors thank all children and parents in the PARSIFAL study, as well as fieldworkers in all centres and the PARSIFAL Genetics study group.

## Footnotes

• Competing interests: SB, EM, AJ, COP, AB, BD, MW, EvM, GD, RL, JR, WE, and MvH declare no competing interests; AS was appointed to a full-time professorship in Clinical Allergy Research by Karolinska Institutet of Stockholm, Sweden in November 1995; this was a donation given at one occasion for 15 million SEK from Kabi Pharmacia, Uppsala, Sweden and the annual proceeds cover the salary for AS and the salary for one administrative assistant/research assistant until retirement and AS has no obligation to the donor; JK and VP are co-inventors on patents related to the discovery of GPRA as an asthma susceptibility gene; GP received $38 000 in 2002 and$10 000 in 2003 from AstraZeneca as research grants for epidemiological registry studies of comorbidities in lung cancer patients; FN is employed by AstraZeneca (AZ), holds some shares and AZ also supports his academic part-time adjunct position at Karolinska Institutet.

• Funding: This study was supported by the Stockholm County Council, the Swedish Foundation for Health Care Science and Allergy Research, the Swedish Society of Medicine, Konsul Th. C. Bergh’s Foundation, the Swedish Environmental Protection Agency, the European Union QLRT 1999-01391, the Kuehne Foundation, the Swedish Foundation for Health Care Science and Allergy Research, the Swedish Asthma and Allergy Association’s Research Foundation, the Swedish Research Council, the Swedish Heart and Lung Foundation, Sigrid Jusélius Foundation, Päivikki and Sakari Sohlberg Foundation, the Academy of Finland and the Swiss National Research Foundation.

• Patient consent: Obtained

• ▸ Additional tables and figure are published online only at http://jmg.bmj.com/content/vol46/issue3

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.