Our study examined whether exposure to SHS both alone and in combination with AAP was associated with increased methylation and decreased expression of IFN-γ in Teffs and of Foxp3 in Tregs. Our assessment of SHS exposure, based on parent questionnaires, was dichotomous: subjects were determined to have either been exposed to SHS or not. The validity of questionnaires asking about parental smoking and smoking in the home to determine pediatric exposure to SHS has been demonstrated previously, with questionnaire-determined SHS exposure confirmed by cotinine analysis
[1, 34, 35]. To confirm SHS exposure in some subjects, we performed urinary cotinine analysis on a subset of subjects from our Fresno cohort, and found that there was a significantly higher level of urinary cotinine in our questionnaire-determined SHS vs. nSHS subjects.
In both cohorts, we found that subjects exposed to SHS showed a significantly greater percentage of methylation of CpG sites in the IFN-γ promoter in their T effector cells when compared to the Teffs of subjects not exposed to SHS (Figure
2A). Consequently, expression of IFN-γ was significantly decreased in the Teffs of SHS subjects in comparison with the Teffs of nSHS subjects in both cohorts (Figure
2B). Interestingly, while there was a significant increase in methylation and decrease in expression of IFN-γ in Teffs associated with AAP exposure (that is, between the FSHS and SSHS groups and between the FnSHS and SnSHS groups), we saw no significant difference in the levels of methylation associated with AAP exposure alone (FnSHS) vs. SHS exposure alone (SSHS), though we did see a significant difference between expression levels of IFN-γ in both groups, with decreased expression in Teffs from FnSHS subjects vs. Teffs from SSHS subjects. When we performed linear regression analysis to determine the association between frequency of methylation and expression levels of IFN-γ, our results were inconsistent within the individual groups (FSHS, FnSHS, SSHS, SnSHS; data not shown), but we found a significant negative association between methylation and expression of IFN-γ in Teffs in both the SHS and the nSHS groups (Figures
2C,D) that was also seen when data from all subject sizes were analyzed (Figure
2E), with increases in R
values also seen as population size increased. The disparity at the smaller group level therefore seems to be due to statistics and sample size rather than to any true difference between populations, as increasing the sample size led to more consistent and significant data. We can therefore conclude that there is in fact a negative association between CpG methylation of the IFN-γ promoter regions and expression of IFN-γ in Teffs, a result that matches those of previous studies
[20–22]. Furthermore, as IFN-γ is the major Th1 cytokine
, our finding of increased methylation and decreased expression of IFN-γ in Teffs of subjects exposed to SHS and subjects exposed to AAP, as well as subjects exposed to both, suggests that SHS and AAP impair normal Th1 function both independently and in conjunction with each other. However, as we did not assess Teff or Th1 function, we cannot verify this.
Similarly, we found that subjects exposed to SHS had a significantly greater percentage of methylation of CpG sites in the promoter and intron regions of Foxp3 in their regulatory T cells when compared to Foxp3 in the Tregs of nSHS subjects, and that expression of Foxp3 in the Tregs of SHS subjects was therefore significantly decreased. The same was true of subjects exposed to AAP vs. subjects not exposed to AAP (Figure
4A,B). Linear regression analysis showed a significant negative association between methylation of the regulatory regions of Foxp3 and expression of the gene within both the SHS population (Figure
4C) and the nSHS population (Figure
4D), as well as within the entire population (Figure
4E), despite inconsistent associations in the individual groups; as with IFN-γ, these inconsistencies were likely due to decreased sample size having decreased statistical power. Stable expression of Foxp3 is necessary for normal Treg function
; our results therefore suggest that exposure to SHS or AAP alone, and to both SHS and AAP, may impair Treg function via methylation and consequent decreased expression of Foxp3. While we did not assess Treg function, a previous study has shown that increased methylation of Foxp3 leads to decreased Treg activity, supporting this conclusion
. The significantly increased IgE concentrations in SHS subjects vs. nSHS subjects also provide positive evidence for decreased Treg function, as decreased Treg function leads to reduced immunosuppression of Th2 cells, leading to increased levels of the Th2 cytokines IL-4 and IL-13, which induce increased B cell production of IgE
. However, as this association was only seen in FSHS subjects vs. nSHS subjects, it would appear that changes in IgE are more influenced by the combination of AAP and SHS exposure rather than AAP or SHS alone.
As expected, we found that both AAP alone and SHS alone were associated with significant increases in methylation and decreases in expression of both IFN-γ in Teffs and Foxp3 in Tregs (Figures
4A,B). Of greater interest is the interaction between these exposures and their combined and relative effects upon the methylation and expression of these two genes. Our finding that there is no significant difference between the levels of methylation of IFN-γ in Teffs and the levels of methylation and expression of Foxp3 in Tregs in subjects exposed to SHS alone vs. subjects exposed to AAP alone suggests that neither is more significantly associated with methylation than the other. However, when looking at calculated effect sizes, SHS exposure alone seems to have a greater association with methylation and transcription of both genes than AAP exposure alone; in every measurement except that of IFN-γ expression levels, the SHS effect size is at least twice as large as that of the AAP effect size. (It should, however, be noted that because our SHS-only sample size is very small, any conclusions we draw are more indicative of trends.) Additionally, when looking at relative contributions of SHS and AAP to variance of the means via two-way ANOVA, while significant effects on the total variance were associated with both exposures, the degree of the effect differed, with SHS contributing more to variation of methylation of both genes and AAP contributing more to variation of expression of both genes. The implications of these findings are as yet unclear and require more data regarding timing and relative degree of AAP and SHS exposures in relation to each other.
Beyond their individual effects, AAP and SHS also appear to have a greater association with altered methylation and expression of Foxp3 and IFN-γ in Tregs and Teffs when combined, as can be seen by the increased methylation and decreased expression of these genes in the FSHS subjects (AAP + SHS) vs. the FnSHS subjects (AAP alone) and the SSHS subjects (SHS alone), as well as the increased effect size when compared with the SnSHS group. Given that AAP and SHS both share certain components (including PAH, which we know to be associated with increased methylation and decreased expression of Foxp3 in Tregs
[4, 37]), this finding seems logical. The details of this interaction, however, are unclear. Looking at effect sizes for AAP and SHS on methylation and expression of IFN-γ suggests these the two exposures are associated with these changes in an additive fashion, as the sum of the individual effect sizes of AAP and SHS exposures alone is close to the effect size of AAP + SHS for both IFN-γ methylation and transcription. However, two-way ANOVA found significant interaction between AAP and SHS on variation of mean transcription levels of IFN-γ, which would suggest a synergic association. Foxp3 effect sizes also suggest a synergic association rather than a perfectly additive one, but do not present a clear picture: the effect size of AAP + SHS is nearly twice as large as the sum of the individual effect sizes of AAP and SHS on Foxp3 transcription, while the effect size of AAP + SHS on Foxp3 methylation is nearly half the sum of the individual effect sizes of the exposures. Additionally, while two-way ANOVA showed a significant interaction between AAP and SHS on the variance of Foxp3 methylation levels, suggesting a synergic association, the interaction between the two exposures was not found to be significant with relation to Foxp3 transcription, which would suggest an additive association.
In part, our findings conflict with previous findings that acute SHS exposure contributes more significantly to clinical symptoms of asthma than does AAP exposure
, though it should be noted that in the Rabinovitch et al. study, there was some overlap between measures, as the study looked at PM2.5 as the measure of AAP, which can also be found in SHS. However, our study only assessed if subjects had been exposed to SHS or not, without looking at the time of most recent exposure, so a direct comparison between our study and the Rabinovitch et al. study is not appropriate (indeed, our cotinine assay results would suggest that a number of our SHS subjects had not experienced acute SHS exposure in relation to sample collection time). Additionally, we have not yet looked at any functional assays or clinical assessments to determine if the differences in methylation and expression of our two genes of interest are borne out in the immunophenotypes of the subjects. At the same time, our findings are intriguing, showing that the combination of AAP and SHS is more significantly associated with methylation and expression of IFN-γ in Teffs and Foxp3 in Tregs than independent exposure to these compounds, and thus impairs normal immunosuppressive functions to a greater degree than would be seen with either AAP or SHS alone.
Our findings, to our knowledge, are novel in that they show a significant relationship between SHS and AAP and hypermethylation and therefore decreased expression of IFN-γ and Foxp3, two genes crucial to the suppression of the asthmatic and allergic responses. SHS exposure is associated with increased comorbidity of asthma and allergic rhinitis, but the mechanisms by which these occur are still unclear. Several studies have implicated SHS and AAP in epigenetic changes to the human genome, but the list of specific genes impacted is incomplete
[11, 13, 16, 17]. We chose to study IFN-γ and Foxp3 both because of their significance to immunosuppression and because previous studies have associated exposure to ambient air pollution and/or allergens with hypermethylation of these genes
[4, 23, 28, 38]. The results of our study therefore imply that at least one way in which SHS and AAP may contribute to asthma and allergies is via epigenetic modifications that disrupt the normal function of genes crucial to Teffs and Tregs.
Our findings are supported by evidence that exposure to cigarette smoke or AAP is associated with decreased numbers and function of Teffs and Tregs
[4, 39–43]. Decreased presence and activity of Teffs and Tregs have also been implicated in the prevalence and severity of asthma, with various studies finding skewing of Th2/Th1 ratios in favor of Th2 activity and of Th17/Treg ratios in favor of Th17 activity in patients with moderate to severe asthma
[44–47]. We can thus infer that increased exposure to SHS and AAP leads to hypermethylation of IFN-γ in Teffs and Foxp3 in Tregs, leading to decreased expression of these genes, causing impaired function of Teffs and Tregs that therefore leads to increased severity and prevalence of asthma and allergies. This is supported by our finding of elevated levels of IgE in FSHS-exposed subjects vs. FnSHS subjects. However, since this difference between IgE levels in SHS and nSHS subjects was not seen in the Stanford cohort, IgE and the allergic response appears to be more associated with AAP exposure than with SHS exposure (two-way ANOVA, however, showed a significant SHS effect of 6.57% (P < 0.05) vs. a significant AAP effect of 3.53% (P < 0.05)).
Our data do not provide significant evidence that SHS or AAP and impaired Teff and Treg function are necessarily causative in the development of asthma or allergic rhinitis. Complete clinical data is needed to examine the link between SHS and AAP and these diseases.
In addition to examining the independent associations of AAP and SHS with the methylation and expression of our candidate genes, we looked to see if we could begin to determine the relative impacts of the two exposures on mechanisms of the immune system. A previous study conducted on patients from Denver looked at the relative impact of the two exposures on clinical symptoms, and demonstrated that recent SHS exposure decreased the impact of AAP on asthmatic exacerbations while simultaneously increasing asthmatic exacerbations
. Similarly, we found that subjects exposed to SHS in addition to AAP had increased methylation and decreased expression of Foxp3 in their Tregs and of IFN-γ in their Teffs when compared with the Tregs and Teffs of subjects exposed to AAP alone, suggesting that SHS is associated with aggravation of the effects of exposure to AAP.
At the same time, we did not find a significant difference in IFN-γ methylation in Teffs or in Foxp3 methylation and expression in Tregs between subjects exposed to AAP alone and subjects exposed to SHS alone, suggesting that the SHS + AAP association is a combinatory one. Further study is needed to examine functional consequences of the epigenetic modifications of IFN-γ and Foxp3 on Teffs and Tregs, respectively, as well as to see if these relative associations hold true for other genes. SHS and AAP are not the only factors known to influence asthma, nor are IFN-γ and Foxp3 the only genes relevant to the development and exacerbation of the disease
[2, 13, 48–50]; other factors, both environmental and genetic, are currently under investigation. We have, to date, examined other indoor environmental exposures such as pollen, dust mites, endotoxin, and mold, but found the greatest differences associated with SHS exposure or lack thereof. Going forward, we will continue to assess these other exposures.
Moreover, SHS and AAP are not the only factors potentially impacting DNA methylation. Previous studies have demonstrated that certain genes are either hyper- or hypomethylated with age
[51–54], but no study to date of which we know has assessed DNA modification of Foxp3 or IFN-γ with respect to age in different T cell subsets. When we performed linear regression of age vs. methylation for both genes, we found no association. Gender has also been implicated in differences in methylation
[15, 54, 55], but we found no significant difference in methylation of either gene in any group associated with gender, and when controlling for gender, obtained no difference in our findings. These results suggest that any contributions of age and gender to the methylation of Foxp3 and IFN-γ in Tregs and Teffs, respectively, are relatively small in relation to the impacts of SHS and/or AAP. Future studies should, however, continue to monitor age and gender.
One further confounding factor is suggested by the significantly increased total IgE seen in FSHS patients vs. nSHS patients. Increased IgE levels are frequently associated with increased allergic severity
, which is in turn often associated with increased methylation and decreased expression of Foxp3 in Tregs and of IFN-γ in Teffs
[21, 25]. Additionally, studies have found an association between SHS exposure and increased IgE levels
. This suggests that SHS might affect methylation by increasing allergic severity. Overall allergic severity for a given subject, however, is difficult to determine
[58, 59]; we did perform skin tests and administer questionnaires concerning allergic rhinitis, and found that subjects with a total IgE >25 kU/L reported symptoms of allergic rhinitis. Given our small sample size, however, we can only infer at this point that total IgE levels may be indicative of clinical atopy. Additionally, given the measures we examined, we did not see any significant difference in allergic severity between SHS and nSHS subjects; the most significant difference between the populations was their SHS exposure status, suggesting that SHS influences methylation directly rather than via an intermediate. It should also be noted that, while increased total IgE is frequently correlated with increased allergic severity, this does not always hold true
Further limitations of our study include our small sample size, as well as the lack of temporality, given that this is a cross-sectional analysis.
Future studies should aim to correlate our findings of genetic differences with phenotypic differences. Teff and Treg functional assays should be performed to determine if the observed changes in methylation and expression of IFN-γ and Foxp3, respectively, impair the normal behavior of these cells. Immunophenotype analysis to evaluate the relative numbers and activity levels of Teffs and Tregs between SHS and nSHS subjects (with and without AAP) from this study should also be performed. Moreover, data from more subjects should be added to our analysis to increase the potential significance of our findings, and data should be collected at multiple time points to examine the temporality of these methylation associations, if any exist. Along these lines, it will also be important to gain a more nuanced picture of the association between these exposures and methylation of these genes, looking at such factors as timing of exposure, duration and extent of exposure, prenatal vs. postnatal exposure (particularly with respect to SHS), and whether this methylation association persists after the exposure has been removed from the environment. Additionally, clinical allergy and asthma profiles and their potential relation to SHS exposure, both with and without AAP, should be examined. To better understand the mechanisms underlying expression of both IFN-γ and Foxp3, methylation at individual CpG sites should also be analyzed. Finally, it is necessary to study other genes important in allergies and asthma that could be epigenetically modified, with repeated measurements performed for these genes as well as IFN-γ and Foxp3 to assess the durability of these changes over time.