The role of gene–ambient air pollution interactions in paediatric asthma

Types of AAP

AAP is a broad term that is typically subdivided either by the origin of the air pollution or by the specific bioactive compound that is measured.

In origin-based AAP classifications the effects of a gas mixture emitted by a specific source are evaluated. Examples are ‘traffic-related air pollution’ (TRAP), or ‘secondhand smoke’ also known as ‘environmental tobacco smoke’. The difficulty with this approach is that these categories encompass several bioactive components. The specific components of these mixtures are not necessarily consistent or reproducible. Conversely, an inherent advantage is that it captures possible interaction effects between different components. Studies assessing TRAP typically use some surrogate marker such as reported truck traffic on the street of residence or measured nitrogen dioxide (NO₂) [30, 31]. Unfortunately, these surrogate markers come with their own limitations and confounders [32].

Studies looking directly at specific bioactive compounds have generally measured benzo[a]pyrene (B[a]P), carbon monoxide (CO), lead, NO₂, ozone (O₃), particulate matter (PM) and/or sulphur dioxide (SO₂) [33–35]. While this approach allows for more exact and reproducible measurements, distinguishing association from causation remains complex. For example, when Slaughter et al. [36] found a significant association between CO and asthma symptom severity, this was hypothesized to be due to the effect of nonmeasured combustion byproducts.

Compounds of particular interest in paediatric asthma research have been NO₂, O₃ and PM.

NO₂ is a combustion byproduct and was initially viewed as mostly an indoor air pollutant [34]. However, since up to 80% of ambient NO₂ is caused by traffic, it has been used as a proxy for TRAP [9, 30]. A limitation of this approach is that while ambient NO₂ is mostly caused traffic, its specificity will be decreased in areas with other sources of exhaust fumes as it is a byproduct of all types of exhaust [32]. Increased NO₂ levels as measured outside children's homes have been associated with increased new-onset asthma [30].

O₃ is formed by an interaction between ultraviolet light and precursor compounds. It is most prevalent in areas with stagnant weather patterns and prolonged sunlight [37]. Several O₃ precursors are found in industrial and vehicle exhaust [9, 37]. Increased O₃ has been associated with increased asthma medication usage and morbidity [38–41]. Beyond this, higher mean 4-year O₃ concentrations have also been associated with increased asthma diagnosis in children engaged in sports [42].

PM is a heterogeneous group that is typically subdivided based on the size of particulates. PM₁₀ and PM_2.5 refer to particulates with an aerodynamic diameter below 10 and 2.5 μm, respectively [8, 34]. This group includes fine solids, dust, aerosols, soot, ash and pollen [37]. There is significant temporal and spatial variation in the content of PM [34]. PM can encompass compounds of specific interest such as polyaromatic hydrocarbons including B[a]P and diesel soot or diesel exhaust particulate (DEP) [37, 43]. PM is sometimes subdivided by pH and acidic PM was specifically associated with decreased lung function in children [44]. Both PM₁₀ and PM_2.5 exposure have been associated with increased asthma symptom frequency and decreased asthma control [8].

Link between AAP and asthma

The link between AAP exposure and asthma exacerbations is well established and has been extensively reviewed [8, 9]. This association remained significant in a meta-analysis [10].

Whether AAP causes asthma, however, has been a topic of discussion [45]. While an epidemiological link has been evident for years, it is only recently that evidence has supported a causal link. This link was reviewed by both Burbank et al. [11] and Hehua et al. [12] and was further supported in a recent systemic review by Khreis et al. [13].

As data highlighting the role of AAP in both asthma severity and incidence became available, differences in AAP sensitivity between ethnic groups became apparent. In 2014, Wendt et al. [46] noted a significantly higher effect of increased O₃ levels on black patients as opposed to white patients with asthma. Recently, Native American ancestry was also shown to interact with AAP and influence response to bronchodilator therapy [47]. This increased sensitivity likely contributes to the high asthma burden of disease in these populations. While differences in environmental exposures likely play a role, a genetic modification of the effect of air pollution was suggested as far back as 2004 [37]. Of note, environmental and genetic factors may also work synergistically. For example, as we will describe further in the article, many genetic variants associated with oxidative stress are related to AAP sensitivity. Socio–economic factors leading to a diet poor in antioxidants could logically further predispose at-risk populations. Beyond cohorts based on ethnicity, differences in AAP sensitivity in cohorts based on ancestral atopic disease status have been described in France [48]. This observation seems to support the notion that the observed differences are in part due to genetic differences and not solely due to differences in environmental factors secondary to socio–economic status.

Genes controlling antioxidant defences

Oxidative stress is defined as an imbalance between oxidants and antioxidant systems with excessive oxidant availability [49]. AAP is known to induce oxidants and increase oxidative stress. This is thought to be one of the major pathways by which AAP influences asthma [50, 51].

The specific oxidising systems in the lung and the potential role of air pollution in shifting the oxidative balance has been extensively reviewed elsewhere [52]. Here, we will focus on how genetic variants that weaken the response to oxidative stress seem to increase the sensitivity to AAP in asthma. These appear to be localised along one of five antioxidant pathways.

First, the glutathione pathway is likely the best studied in the context of AAP sensitivity. Glutathione-S-transferase (GST) enzymes conjugate reactive oxygen species with reduced glutathione to facilitate their elimination [53, 54]. The cytosolic GST enzymes are subdivided into α, µ, π, σ and θ classes based on their isoform [54]. Regarding the effect of AAP, the θ and μ members of the GST family, encoded by GST Pi 1 (GSTP1) and GST Mu 1 (GSTM1), respectively, have been of particular interest [54]. This is both due to their relative importance in the lung, with GSTP1 estimated to provide up to 90% of pulmonary GST activity, and the relatively high frequency of GSTP1 or GSTM1 null genotypes [55].

The GSTM1 null genotype was shown to be associated with increased markers of oxidative stress after ozone exposure and facilitates DEP-induced allergic inflammation in adults [56, 57]. Specifically in paediatric asthma, Romieu et al. [58] showed patients with this genotype experienced significant decreases in forced mid-expiratory flow at 25–75% of forced vital capacity when exposed to high O₃ concentrations whereas others did not.

The results of specific variants within these genes have been more heterogeneous. Up to 40% of the population has two isoleucines as opposed to valine in GSTP1 codon 105 (genotype GSTP1 I/I) [57]. The GSTP1 I/I genotype is associated with DEP-induced allergic inflammation in adults and increased paediatric asthma [57, 59, 60]. Conversely, NO_x exposure led to allergic sensitisation only in patients with genotype GSTP1 I/V or V/V but not in those with GSTP1 I/I [55]. Furthermore, there was a negative association between asthma, O₃ and PM_2.5 exposure for GSTP1 I/I homozygotes as opposed to a positive association for those with a GSTP1 I/V or V/V genotype [61]. Finally, in high PM₁₀ exposure communities, GSTP1 I/V or V/V was a risk factor for asthma [60]. We currently lack a clear explanation for these seemingly contradictory findings related to these specific GSTP1 variants.

Two possible explanations for this observed heterogeneity have been described. On one side, we know that the components of PM have a wide spatio–temporal variation. Since these variants change the catalytic properties of the GST enzymes, it is possible that the effect of the variant is dependent on the specific PM components to which a patient is exposed [62]. On the other, gene–gene interactions may complicate the observed gene–environment relationship. For example, variants within the nicotinamide adenine dinucleotide plus hydrogen (NAD(P)H) quinone oxidoreductase 1 (NQO1) gene modulate the asthma risk associated with a GSTM1 null genotype [63, 64]. This gene encodes NAD(P)H dehydrogenase 1, a phase 2 detoxification reaction protein that helps protect against damage from oxidative stress.

A further example of a gene–gene–environment interaction can be found when looking at variants within epoxide hydrolase 1 (EPHX1). This gene encodes the microsomal EPH, an important class of enzymes involved in detoxification. These variants impacted the risk of early persistent asthma only in children with a GSTM1 null genotype that lived within 75 m of a major road [65].

The second and third antioxidant pathways involved in AAP are the catalase (encoded by CAT) and myeloperoxidase (encoded by MPO) pathways. These enzymes catalyse the reduction of hydrogen peroxide (H₂O₂). Children with variants in both the promotor region of the CAT and MPO genes were found to be more likely to be absent from school due to respiratory illness when exposed to AAP [66].

The fourth pathway is the heme oxygenase-1 (HO-1 encoded by HMOX-1) pathway [67]. By catalysing heme, HO-1 creates both biliverdin and bilirubin which in turn act as scavengers of reactive oxygen species [67]. HMOX-1 is induced by oxidative stress and this inducibility is inversely correlated to the number of (GT)_n repeats. Non-Hispanic white children with fewer than 23 (GT)n repeats were found to be at a reduced risk for asthma. This effect was more pronounced for children living in a low-O₃ community. It was hypothesised that the difference in promotor activity may allow children to respond more optimally to temporary increases in oxidative stress but that this effect is less significant if the antioxidant defences are chronically stimulated by high background O₃ concentrations [68].

The fifth pathway is centred around thioredoxin. Reduced thioredoxin can catalyse the reduction of oxidised cellular proteins. It in turn is then reduced again by thioredoxin reductase. Deficiencies within this system have been associated with COPD, asthma and lung injury [69]. A recent genome wide interaction study linked variants within thioredoxin domain containing 8 (TXNDC8), a gene encoding one of the thioredoxin reductase enzymes, to sensitivity to NO₂ in Caucasians [70].

Aside from variants within the oxidative stress defences, two variants that expose the host to increased intrinsic oxidative stress also increase the sensitivity to AAP. First, a variant within cytochrome P450 2E1 (encoded by CYP2E1) increased air pollution sensitivity [35]. This enzyme plays a central role in detoxification reactions and is known to increase oxidative stress [71]. Second, a variant leading to increased expression of B4GALT5 significantly modified the effect of traffic related AAP on asthma in children [72]. This gene encodes the enzyme B4GALT5 involved in the synthesis of lactosylceramide. The lactosylceramide-centric signalling pathways increase oxidative stress when activated [73].

Closely related to oxidative stress is nitrosative stress. Nitrosative stress is defined as an overproduction of NO free radicals [74]. These can react with reactive oxygen species to form reactive nitrogen species (RNS). Both NO and RNS have been shown to play a significant role in allergic asthma [75]. Arginase, encoded by ARG1 and ARG2, competes with NO synthase and decreases the nitrosative stress. Variants within both ARG1 and ARG2 are associated with asthma but the relationship between variants within ARG1 and asthma was found to be modulated by ozone exposure [76].

There has been significant research interest in modulating oxidative stress in asthma. The results of both paediatric and adult studies were recently expertly reviewed by Sahiner et al. [77]. In brief, the current literature is heterogeneous with inconsistent results of anti-oxidant supplementation. While a study by Romieu et al. [78] analysed subjects by ozone exposure, to our knowledge, no studies have stratified patients by AAP exposure and genetic profile.

Genes controlling the immune response

Beyond variants in genes directly related to antioxidant defences, variants in genes controlling the response of the immune system have also been found to interact with AAP. Most of the variants described to date are closely associated with the nuclear factor κΒ (NF-κB) pathway. NF-κB is a family of nuclear transcription factors that functions as a central pro-inflammatory regulator in both the innate and adaptive immune system [79]. When activated, NF-κB upregulates the transcription of inflammatory proteins and can regulate cell differentiation and proliferation [79]. One of NF-κB's downstream effectors, the NOD-, LRR- and pyrin domain-containing protein 3 inflammasome, has been suggested as a possible target in severe asthma [80]. Mitochondrial damage, secretion of tumour necrosis factor α (TNF-α) and activation of Toll-like receptors (TLRs) can lead to activation of the NF-κB pathway.

TNF-α is a pro-inflammatory cytokine that plays a role in the innate immune response [81]. It is primarily released by macrophages and leads to phosphorylation of NF-κB. Dysregulation of this pathway has been implicated in asthma and is of particular interest in refractory asthma [81]. With regards to the interaction between AAP and asthma, patients with variants associated with the TNF-α gene were found to be protected from asthma if they grew up in a low O₃ exposure community [82]. The authors speculated that this variant may dampen the inflammatory response to lower doses of oxidative stress [82]. The TLR-2 and TLR-4 receptors bind to either Gram-positive bacteria, mycoplasma, yeast and spirochetes or Gram-negative bacteria, respectively. Variants within these genes were shown to significantly modulate the risk for doctor-diagnosed asthma in children exposed to PM_2.5 [83].

As noted above, NF-κB also plays a central role in T-cell activation and differentiation [79]. Paradoxically, besides leading to T-cell activation, NF-κB activation also stimulates differentiation of regulatory T-cells (Treg). These Tregs play a central role in controlling the immune response and preventing chronic inflammation [79]. This dual action is of specific interest within asthma as excessive activation of type 2 helper T-cells (Th2 cells) and insufficient downregulation by Treg cells is thought to play a central role in allergic asthma [84]. Unsurprisingly, variants within genes that affect the balance between Treg and Th2 cells have been associated with increased AAP sensitivity. Cytotoxic T-lymphocyte-associated protein 4 (CTLA4, encoded by CTLA4) has a Treg-cell mediated inhibitory effect on T-cells [85]. Variants within CTLA4 leading to decreased CTLA4 activity seemed associated with increased ambient B[a]P asthma risk [35]. Signal transducer and activator of transcription 4 (encoded by STAT4) is another protein with a central role in T-cell differentiation [35]. Variants within STAT4 have been shown to increase the odds for asthma in patients exposed to high levels of AAP [35].

Finally, Ierodiakonou et al. [70] identified variants within EPHA3 in their NO₂ genome-wide interaction study. This gene encodes an Eph receptor and member of the family of tyrosine kinases. This receptor is known to play a role in cell–cell interactions and has been studied mostly in an oncogenic context [86]. To date, its potential role in asthma remains unclear.

It is worth noting that many of the proteins associated with these variants have been targeted either in asthma or other inflammatory processes. For example, due to interesting experimental and animal model results, TNF-α blockade has been an area of extensive research in paediatric asthma [87]. Unfortunately, the results of clinical studies have been mixed. Given the data available from genetic studies, one wonders if these mixed results are partially due to different exposures and genetic susceptibilities in the populations studied.

Other implicated variants

Variants in genes with a role in airway development and repair and bronchial responsiveness have also been associated with AAP sensitivity. Children homozygous for threonine as opposed to cysteine in the 508 position of TGF-B1 were found to be at increased risk for asthma if they resided close to a major roadway. This variant is known to increase transforming growth factor B1 (TGF-B1) expression [88]. The transforming growth factor superfamily proteins serve a wide array of functions including regulating cell growth, proliferation and apoptosis [89]. These proteins are secreted in an inactive form and rely on external stimuli, such as oxidative stress, for their activation [89]. Specifically in asthma, TGF-B1 increases smooth muscle proliferation and airway remodelling. Furthermore, TGF-B1 helps to regulate the balance between T-cell populations and is required to maintain peripheral Treg cells [89].

Finally, variants associated with ADCY2 were shown to significantly interact with NO₂ on asthma risk [72]. ADCY2 encodes adenylyl cyclase type 2, a receptor expressed on airway smooth muscle cells [90]. Activity of the adenylyl cyclases is upregulated by stimulation of the B-adrenergic receptor and causes smooth muscle relaxation [90].