We performed a genetic association study between asthma and genes involved in the vitamin D pathway. SNPs located in five genes, including IL10, CYP24A1, IL1RL1, CYP2R1 and CD86, showed modest association with asthma or atopy in an asthma family study derived from the Saguenay_Lac-Saint-Jean population. Exploring gene-gene interactions in this pathway revealed significant two-gene models that refined risk evaluation for asthma. In an attempt to replicate these findings, we evaluated the significant genes in two additional Canadian studies and one Australian study. Again modest associations were observed for the IL10, CYP24A1 and CD86 genes for asthma or atopy. In general, SNPs showing association were different between studies or the orientation of the risk allele was reversed. Similarly, the significant two-gene models found in the original study were not replicated in the two additional Canadian studies and the Australian study. However, a significant two-gene model involving the IL10 and VDR gene was replicated in the American study.
Overall, no genes or gene-gene interactions in the vitamin D pathway were consistently associated with asthma or atopy across all populations. This can be explained by multiple factors as discussed previously [3, 42]. Briefly, asthma was defined differently between populations, including patient self-reported, physician's diagnosis, or a combination of clinical characteristics and objective threshold from methacholine challenges. In addition, the individual studies are either early or late-onset asthma. More specific for the current study, many environmental factors such as sun exposure and dietary vitamin D intake may differ between studies and explained some of the controversies and flip-flop observations. Considering the differences among populations, our replication effort may be more properly labeled as a descriptive study.
A number of SNPs in the IL10 gene surrounding the promoter region and extending up to intron 1 were significantly associated with asthma in the SLSJ collection. All these SNPs were in tight LD and most of the genetic diversity at this locus was captured by only two haplotypes. Based on functional studies [45–50], the three-SNP promoter haplotypes formed by rs180096, rs1800871, and rs1800872 have been referred as the high and low IL10-producing haplotypes. IL10 is known as an anti-inflammatory cytokine . Accordingly, it makes biological sense that a low IL10-producing haplotype would be associated with asthma, which is consistent with most of the literature [39, 52–54]. Surprisingly, none of the promoter polymorphisms in IL10 were associated with asthma or atopy in the SAGE and CAPPS populations with the exception of rs1800896, which was associated with asthma in CAPPS. However, the orientation of the risk allele was reversed compared to what was observed in the SLSJ study and the vast majority of the scientific literature. In contrast, SNPs located in the 3' end of the IL10 gene (rs3024498 and rs3024492) were associated with asthma in the replicate samples. The functional consequences of these SNPs are unknown but they were previously associated with serum levels of calcidiol (25-OH-D3), the prehormone form of vitamin D , and with longitudinal decline rate of lung function .
Other genes in the vitamin D pathway tested in the current study have been previously associated with asthma or asthma-related phenotypes. In the German Asthma Family Study, Wjst et al.  identified SNPs in the IL10 gene but also in the GC, CYP2R1 and CYP24A1 genes that were significantly associated with asthma or IgE levels. Interestingly, SNPs in these genes were also associated with serum levels of calcidiol (25-OH-D3) and calcitriol [1α,25(OH)2D3]. In the SLSJ population, three SNPs in the CYP24A1 gene showed modest evidence of association with asthma or atopy (p < 0.05). In addition, the only CYP24A1 SNP (rs2248359) genotyped in common between the German and SLSJ population showed a consistent result with an increase risk of asthma associated with the C allele. SNPs in the CYP2R1 gene, encoding an enzyme implicated in earlier steps of the vitamin D metabolic pathway, were also associated with asthma in both the German and the SLSJ studies. Only one SNP in the CD86 and CD28 genes reached significance in the SLSJ population. The CD28 and CD86 interaction represents an important costimulatory signaling pathway for CD4 positive T cell activation and Th2 cytokine production . The association between atopy and the CD86 promoter polymorphism rs2715267 was relatively strong in the SLSJ collection (p = 0.004). Different SNPs in the CD86 gene were also associated with asthma in the CAPPS and BHS studies. Thus, evidence is building to implicate CD86 as a susceptibility gene for asthma, but independent replications in larger population samples are required. In contrast, the overall results for CD28 are more consistent with the previous studies showing the lack of association with asthma [58–60]. Many SNPs in the IL1RL1 gene were associated with asthma in the SLSJ study. This receptor is required for the development of an effective Th2 response [61, 62] and mediates the biological effects of IL-33 . A functional promoter polymorphism (rs6543116) in the IL1RL1 gene was previously associated with atopic dermatitis and high total IgE levels in the sera from the same patients . More recently, a genome-wide association scan for blood eosinophil count identified a strong association with an intronic SNP (rs1420101) in this gene . The same SNP was then associated with asthma in a collection of ten different populations (7,996 cases and 44,890 controls). This SNP was not genotyped in the current study, but rs950880 can be used as a surrogate (r2 = 0.96 based on the European HapMap data). rs950880 was not associated with asthma or atopy in any of the four populations evaluated in the current study. However, a trend (p ≤ 0.10) in the same direction as the eosinophil/asthma study  was observed for asthma and atopy in the SLSJ study as well as for atopy in the BHS. Finally, the SLSJ results do not support previous observations with the GC and IL8 genes [55, 66, 67].
It is reasonable to expect interactions among functionally related genes. In the SLSJ study, some two-gene models were more efficient at evaluating the risk of asthma compared with single SNP models. The models implicating functional variants in the VDR and IL10 genes are particularly interesting. We previously demonstrated that genetic variants in the VDR gene are associated with asthma . Since the latter publication , we have genotyped more SNPs in the VDR gene in the same population and many SNPs between intron 2 and exon 9 are associated with asthma (see Additional file 12). The mechanism explaining this association is still unclear, but allele-specific expression at this locus implicates the presence of genetic variants that influence VDR expression . In fact, we observed that the highly expressed haplotype was overtransmitted to asthmatic individuals, while the low expressed haplotype was undertransmitted . These VDR haplotypes can be tagged by the SNP rs1544410 with the G allele being associated with the high-risk/high-expressed haplotype and the A allele being associated with the low-risk/low-expressed haplotype. As mentioned above, the low IL10-producing haplotype was associated with asthma in the same population (SLSJ). These IL10 haplotypes can be tagged by the SNP rs1800871 with the T allele being associated with the high-risk/low-producing haplotype and the C allele being associated with the low-risk/high-production haplotype. By combining the information of both SNPs (IL10-rs1800871 and VDR-rs1544410) we obtained the two-gene model shown in Figure 5b and that was a significant (p = 0.00018) predictor of the risk of asthma. In this model, the risk of asthma increased additively with the number of G alleles (tagging the high-risk/high-expressed haplotype) at the VDR locus and with the number of T alleles (tagging the high-risk/low-producing haplotype) at the IL10 locus. Similarly, other intriguing interactions between plausible candidate genes were observed in the SLSJ population, including IL10 and IL1RL1 (Figure 5d) as well as IL10 and CD86 (Figure 5e). Although interesting, the large majority of these two-gene models were not replicated in the other populations. In the CAMP study, a two gene-model involving IL10 and VDR variants mirrored the observation made in the SLSJ sample. However, further work will be required to demonstrate the relevance of these two-gene models in asthma.
The cascade of enzymatic reactions that lead to the biosynthesis of 1α,25(OH)2D3 is complex and requires the participation of many gene products (Figure 1). It is conceivable that polymorphisms in any of several genes in this pathway could lead to differences in endogenous biosynthesis and bioavailability of 1α,25(OH)2D3. At a more downstream step in the pathway, genetic variants in the VDR itself or in genes implicated in the transcriptional machinery, can influence the sensitivity of the 1α,25(OH)2D3 stimulus. Further downstream are transcriptionally regulated genes, or vitamin responsive genes, that mediate the broad physiological actions of 1α,25(OH)2D3. Again, polymorphisms in these genes are prime candidates to influence the vitamin D response. However, at this level the task is rather challenging owing to the large number of genes regulated by 1α,25(OH)2D3 . An extra layer of complexity exists, knowing that the regulation of gene expression by 1α,25(OH)2D3 is highly cell-specific [69, 70]. Accordingly, a large number of genes can determine the overall response to the vitamin D pathway. With this in perspective, gene selection in the current study was not comprehensive. On the other hand, the results of the current study may have implications in a wider scope than the field of asthma and asthma-related phenotypes. SNPs associated with asthma in the vitamin D pathway genes may influence the overall vitamin D sensitivity and consequently influence many other diseases .