The role of ALOX5AP, LTA4H and LTB4R polymorphisms in determining baseline lung function and COPD susceptibility in UK smokers

Background We have previously shown evidence that polymorphisms within genes controlling leukotriene B4 (LTB4) production (ALOX5AP and LTA4H) are associated with asthma susceptibility in children. Evidence also suggests a potential role of LTB4 in COPD disease mechanisms including recruitment of neutrophils to the lung. The aim of the current study was to see if these SNPs and those spanning the receptor genes for LTB4 (LTB4R1 and LTB4R2) influence baseline lung function and COPD susceptibility/severity in smokers. Methods Eight ALOX5AP, six LTA4H and six LTB4R single nucleotide polymorphisms (SNPs) were genotyped in a UK Smoking Cohort (n = 992). Association with baseline lung function (FEV1 and FEV1/FVC ratio) was determined by linear regression. Logistic regression was used to compare smoking controls (n = 176) with spirometry-defined COPD cases (n = 599) and to more severe COPD cases (GOLD stage 3 and 4, n = 389). Results No association with ALOX5AP, LTA4H or LTB4R survived correction for multiple testing. However, we showed modest association with LTA4H rs1978331C (intron 11) with increased FEV1 (p = 0.029) and with increased FEV1/FVC ratio (p = 0.020). Conclusions These data suggest that polymorphisms spanning ALOX5AP, LTA4H and the LTB4R locus are not major determinants of baseline lung function in smokers, but provide tentative evidence for LTA4H rs1978331C (intron 11) in determining baseline FEV1 and FEV1/FVC ratio in Caucasian Smokers in addition to our previously identified role in asthma susceptibility.


Background
Chronic obstructive pulmonary disease (COPD) is a complex respiratory disease with genetic and environmental contributors to pathophysiology [1,2]. Evidence suggests the dihydroxy leukotriene, leukotriene B 4 (LTB 4 ), plays a role in this disease as its production is elevated in the airways of COPD subjects [3,4]. The altered inflammatory response of the airways of COPD sufferers is a result of lymphocytes and neutrophils, suggested in part to be the result of cigarette smoke inhalation [5]. Importantly, LTB 4 has been shown to have chemotactic activity recruiting inflammatory cells to the lung [6,7]. LTB 4 is implicated in the neutrophillic inflammation of COPD and has been suggested as the major chemotactic agent in more severe forms of this disease [8]. It has been established that the cysteinyl leukotrienes (CysLTs; LTC 4 , LTD 4 and LTE 4 ) play a significant role in bronchoconstriction and airway inflammation in asthma [9] but their role in COPD is less clear.
Studies have suggested that polymorphisms spanning leukotriene pathway genes may be important in determining leukotriene production and susceptibility to allergic disorders, such as asthma [10]. LTB 4 and the CysLTs are produced from arachidonic acid in a multi-enzyme pathway called the 5-lipoxygenase (5-LO) pathway. Single nucleotide polymorphisms (SNPs) in two 5-LO pathway genes; 5-lipoxygenase activating protein (ALOX5AP) and leukotriene A 4 hydrolase (LTA4H) have shown an association with LTB 4 overproduction from ionomycinstimulated neutrophils and with myocardial infarction (MI) susceptibility [11,12]. 5-lipoxygenase activating protein (FLAP) with 5-LO is involved in the synthesis of LTA 4 which can be conjugated with glutathione by LTC 4 synthase to form LTC 4 and subsequent CysLTs or converted to LTB 4 by the enzyme LTA 4 hydrolase (LTA 4 H) [13]. FLAP is involved in the production of all leukotrienes; however LTA 4 H is specifically involved in LTB 4 production.
A recent study has suggested that LTA 4 H contains an aminopeptidase activity as well as having a role in LTB 4 production [14]. This aminopeptidase activity cleaves the neutrophil chemoattractant proline-glycine-proline (PGP), a COPD biomarker, responsible for the influx of neutrophils into the lung -contributing to chronic inflammation. Cigarette smoke was found to inhibit this aminopeptidase activity thereby leading to accumulation of PGP and neutrophils [14]. This dual role could have important consequences for the design of therapeutics targeting LTA 4 H.
We have recently reported evidence that SNPs spanning ALOX5AP and LTA4H are asthma susceptibility markers [15]. SG13S114A, SG13S89A and SG13S41G (ALOX5AP) and rs1978331C (LTA4H) were associated with asthma and asthma related phenotypes (atopy, FEV 1 , bronchial hyperresponsiveness) in a family based association study using 341 asthma families with two affected siblings [15]. Several haplotype associations were also observed [15]. To date, no study has investigated the role of these SNPs with respect to COPD or baseline lung function in smokers.
Smoking is associated with decline in lung function and is a major risk factor for the development of COPD; we therefore investigated the role of ALOX5AP, LTA4H and LTB4R SNPs in smokers.
The aim of the current study was to determine whether polymorphisms spanning ALOX5AP, LTA4H and the LTB4R locus influence baseline lung function (FEV 1 and FEV 1 /FVC ratio) in smokers and whether they contribute to susceptibility to develop COPD or a more severe form of COPD in smokers. We genotyped twenty SNPs spanning these three loci in a cohort recruited for COPD or smoking history (n = 992 subjects) and completed a series of association analyses.

Subjects and baseline characteristics
Subjects were recruited from five UK centres for smoking history and/or COPD diagnosis (n = 992) [16]. Subjects collected from Nottingham (n = 537) were Caucasian, > 40 years and had > 10 pack-year smoking history. Subjects collected from other UK centres (n = 455) were recruited for physician and spirometry defined COPD, Caucasian, > 40 years, smokers with > 10 pack-year history. The combined subjects (n = 992) recruited for smoking history or COPD diagnosis was stratified into 'healthy' smokers (smoking controls) (n = 176, post-bronchodilator (BD) salbutamol FEV 1 > 80% predicted and postBD FEV 1 /FVC > 0.7) and COPD cases (n = 599, postBD FEV 1 < 80% predicted and postBD FEV 1 /FVC ratio < 0.7). Subjects not meeting these criteria (or with missing data) were excluded from the case control analyses (n = 217). To investigate whether SNPs determined severity of COPD in the smokers we compared smoking controls with postBD spirometry, i.e. the GOLD classifications [17]. Ethical approval was obtained from the relevant ethics committees (Nottingham, Sheffield, Manchester, Leicester and Oxford) and informed consent from all subjects was obtained.

Selection of SNPs and genotyping
Twenty SNPs were genotyped across ALOX5AP (eight), LTA4H (six) and LTB4R (six) ( Figure 1A-C). SNPs spanning ALOX5AP and LTA4H have previously been shown to tag haplotypes associated with myocardial infarction and LTB 4 production [11,12] and with asthma susceptibility in our recent study [15]. Six LTB4R SNPs were chosen for their ability to tag linkage disequilibrium (LD) blocks or for inferred function, once the region had been sequenced in Caucasian individuals (22 SNPs validated in 35 Caucasian subjects, data not shown). Genotyping was completed by Kbioscience (Hertfordshire, UK) using KASPar technology. Hardy-Weinberg equilibrium was assessed in all subjects using Haploview software [18].

Statistical analyses
Linear regression (SPSS V15, SPSS Inc., Chicago, IL) was used to determine the contribution of each SNP to baseline FEV 1 (litres) or FEV 1 /FVC ratio using the additive model (e.g. AA vs. AC vs. CC) including age, gender, height and smoking pack-years as covariates. The COPD susceptibility analyses were completed using logistic regression in the additive model in two ways. Firstly, the smoking controls (n = 176) vs. all COPD subjects (n = 599) and then the smoking controls (n = 176) vs. GOLD 3 and 4 subjects (n = 389). Both analyses included age, gender and pack-years as covariates (Table  1). Based on the 80 tests completed (analysis of 20 SNPs with 4 outcomes), a conservative Bonferroni correction suggested a p < 10 -4 when reporting results as significant. With respect to power (based on lowest and highest minor allele frequency), there was between 77-99% power to detect a 50 ml difference in FEV 1 and between 58-99% power to detect a 5% difference in FEV 1 /FVC ratio. Analyses of COPD susceptibility were relatively underpowered, with between 28-91% power with an odds ratio of 1.5 and 68-99% power with an odds ratio of 2.0. All analyses considered an error rate of 5%.

Clinical characteristics and genotyping
Subject characteristics for the smoking controls (n = 176) and COPD subjects (n = 599) and the entire cohort (n = 992) are shown ( Table 1). Comparison of the smoking controls (n = 176) and COPD sufferers (n = 599) show differences in baseline lung function (percent predicted FEV 1 96.03% compared to 40.31%) as anticipated (p < 0.0001). These subjects also showed differences for age, gender and pack-years so these variables were included as covariates in analyses. Genotyping completion rates were > 96% for all twenty SNPs genotyped and did not show deviation from Hardy-Weinberg equilibrium (p > 0.05). Minor allele frequencies for ALOX5AP and LTA4H SNPs were similar to those observed in our previous study [15].  represents the LD plot in the smokers (n = 992) for each loci respectively..* represents SNPs that define the haplotype HapA, #which define HapB and^which define the haplotype (HapK) previously associated with MI [11,12]. Polymorphisms spanning ALOX5AP, LTA4H and LTB4R are not associated with baseline FEV 1

and FEV 1 /FVC in smokers
To assess whether SNPs in ALOX5AP, LTA4H and LTB4R influence baseline lung function in smokers we determined their role in baseline FEV 1 and FEV 1 /FVC in the entire population (n = 992) using linear regression in the additive model (Additional File 1 Table S1). The FEV 1 analyses did not identify any significant association with ALOX5AP, LTA4H or LTB4R SNPs. LTA4H rs1978331C (intron 11) (p = 0.029, mean FEV 1 values: TT 1.468 ± 0.039L, TC 1.599 ± 0.034L and CC 1.594 ± 0.057L) and rs2660899 (5'UTR) (p = 0.024; GG 1.580 ± 0.030L, GT 1.504 ± 0.044L and TT 1.192 ± 0.158L) were associated with increased FEV 1 , although this did not meet Bonferroni correction. Analysis with FEV 1 /FVC ratio again did not show any significant association. However, the same LTA4H rs1978331C (intron 11) showed p = 0.020 with the same direction of effect, increased FEV 1 /FVC ratio (mean ratio ranging from 53.8% in the major homozygotes to 57.4% in the minor homozygotes).
Polymorphisms spanning ALOX5AP, LTA4H and LTB4R do not determine COPD susceptibility To determine whether SNPs spanning ALOX5AP, LTA4H and LTB4R act as determinants of COPD susceptibility in smokers, we completed case-control association analyses comparing the smoking controls (n = 176) with the COPD subjects (n = 599), defined by postbronchodilator spirometry ( Table 2). No significant associations were observed. This analysis was corrected for age, gender and pack-years as these traits were significantly different between the study groups (Table 1).

Discussion
This was the first study to investigate polymorphisms spanning genes involved with LTB 4 production and activity with lung function and COPD susceptibility in smokers. A UK smoking cohort comprising n = 992 individuals with > 40 years and > 10 pack-years smoking history was used to determine whether SNPs in ALOX5AP, LTA4H and LTB4R influenced baseline lung function and susceptibility to develop COPD in smokers. LTB 4 has been shown to be important for the inflammation observed in COPD, with this mediator upregulated in COPD subjects [4]. We hypothesised that polymorphisms in these genes may influence susceptibility to develop airway obstruction in smokers that is in part driven by LTB 4 . We have found that polymorphisms spanning ALOX5AP, LTA4H and the LTB4R locus are not associated with lung function or COPD susceptibility in smokers as no SNP survived correction for multiple testing. However, we provide tentative evidence for association between LTA4H rs1978331C (intron 11) and lung function measures in these subjects. We have previously investigated the role of polymorphic variation in the genes of the 5-lipoxygenase pathway e.g. ALOX5, LTC4S, CYSLTR1 in asthma and allergy susceptibility [15,19,20] and as determinants of clinical responses to therapies targeting this pathway [21]. These studies provide accumulating evidence that polymorphic variation in these genes influence disease phenotypes in disorders where leukotrienes play a significant role [10], also confirmed with other non-respiratory diseases e.g. MI [11,12]. To date, no study has specifically looked at genetic determinants of leukotriene production/activity in smokers with or without COPD. While no association survived the Bonferroni correction, additive model analyses with rs1978331C (LTA4H, intron 11) showed a p = 0.029 with an increase in FEV 1 and p = 0.020 with FEV 1 /FVC ratio.
The mean FEV 1 and FEV 1 /FVC values for the TC heterozygotes and CC homozygotes were similar, but the presence of the minor C-allele for these genotype groups gave higher trait values when compared to the TT homozygotes, suggesting a dosage effect does not occur. These findings provide tentative evidence suggesting that variants in LTA4H may determine lung function in COPD.
We next sought to investigate whether polymorphisms spanning these genes determine susceptibility to develop COPD. Case-control association analyses were completed with 'healthy' control smokers and smokers with physician diagnosed COPD (including spirometry). No significant associations with polymorphisms spanning ALOX5AP, LTA4H and LTB4R were identified. We also completed another case-control analysis involving COPD sufferers at the severe end of the spectrum. GOLD groups 3 and 4 were chosen as this represented the most severe cases based on spirometry. Again no significant associations were observed.
There is an interesting link for LTA4H with COPD and asthma; in our group's previous study, we showed preliminary association with rs1978331C (LTA4H, intron 11) and asthma susceptibility in 341 families (protection, p = 0.036) [15]. A recent study has shown a similar effect in a different disease; heterozygosity at two LTA4H SNPs, one rs1978331 (intron 11), is significantly associated with protection from tuberculosis infection, lower mortality amongst patients with severe tuberculosis infection and protection from the development of severe leprosy disease [22]. These two studies show the same protective direction of association and provide further support for a functionally significant role of rs1978331 or (another SNP tagged by this) in determining LTA 4 H expression or activity.
Suggestive association with COPD (p = 0.02 to 0.05) with four LTA4H SNPs within the promoter region (these SNPs were not analysed in our current study) was reported by another group [23]. We have not identified any association with LTA4H SNPs located in the 5'UTR (rs2540482, rs2660845 and rs2540475) with lung function or COPD susceptibility in smokers. Interestingly, this group also tested different ALOX5AP SNPs to our current study and found no association with COPD [23]. These and our own data provide suggestive support for a role of Logistic regression was used to compare genotype frequencies between smoking controls (n = 176) and total COPD cases (n = 599) using the additive model with the covariates age, gender and pack-years. OR, odds ratio; 95% CI, 95% confidence interval. 0, 1 and 2 represent the number of major homozygote, heterozygote and minor homozygote genotype frequencies.
LTA4H SNPs in determining baseline lung function in smokers potentially suggesting a role for genetically determined LTB 4 in COPD (LTA4H converts LTA 4 to LTB 4 ). This may be a result of neutrophilic inflammation being important in COPD and severe COPD [24]. While this study did not show significant protective association with LTA4H rs1978331C with lung function in smokers, the same direction of effect was observed with asthma susceptibility in our previous study [15] and with the HapK (rs1978331(T/C), rs17677715(T/C), rs2540482(T/C), rs2660845(A/G) and rs2540475(C/T)) haplotype association that conferred a modest risk of myocardial infarction in Icelandic subjects for rs1978331T [12]. As previously mentioned, a protective effect has also been observed with tuberculosis infection [22]. LTA 4 H has a pro-inflammatory role generating LTB 4 through its epoxide-hydrolase activity (intracellular) and an anti-inflammatory role through its amino-peptidase activity to breakdown PGP, facilitating resolution (extracellular). Cigarette smoke selectively inhibits the ability of LTA 4 H to break-down PGP leading to neutrophil accumulation and contributing to COPD pathogenesis [14]. rs1978331 may affect the levels of transcription of the LTA4H gene. Decreased transcription could lead to decreased protein levels of LTA 4 H which may contribute to the protective physiological effect, through reduction in the formation of the inflammatory LTB 4 . However, this mechanism would lead to the accumulation of PGP and so neutrophillic inflammation, counteracting the situation. LTB 4 and PGP are both neutrophil chemoattractants [25,26]. rs1978331 may alter splicing efficiency of LTA4H. rs1978331 is in intron 11 of the gene and exon 10 and 11 of LTA4H contains the zinc-binding domain which is required for both the epoxide hydrolase and aminopeptidase activities [27]. The two functional sites are different but overlapping [28]. Altered splicing in this region could affect the ability of LTA 4 H to generate LTB 4 and/or degrade PGP. Presence of the C-allele may cause splicing events that reduce LTB 4 formation, but the aminopeptidase activity may remain functional, which could lead to less neutrophil chemotaxis and so less inflammation. Presence of the T-allele may cause splicing events that lead to increased LTB 4 production.
The T-allele in HapK was functionally associated with Logistic regression was used to compare genotype frequencies between smoking controls (n = 176) and GOLD stage 3/4 cases (n = 389) using the additive model with the covariates age, gender and pack-years. OR, odds ratio; 95% CI, 95% confidence interval. 0, 1 and 2 represent the number of major homozygote, heterozygote and minor homozygote genotype frequencies. Black bold indicated (p = ≤ 0.05).
LTB 4 overproduction from ionmycin stimulated neutrophils in MI patients [12]. Other factors could complicate this potential mechanism, such as the lung environments in asthma and COPD and the presence/absence of cigarette smoke. This information could have important consequences for the design of any therapeutics inhibiting LTA 4 H. Reducing LTA 4 H activity will reduce LTB 4 production, but neutrophilic inflammation will persist as PGP will no longer be degraded. A more selective inhibitor strategy would be required to block LTB 4 production, but leave the aminopeptidase activity intact. This could take advantage of the different substrate specificities of the non-overlapping regions of the 'active site', small molecules which bind to this hydrophobic part of the site can alter the substrate preference of the aminopeptidase activity [28]. Consideration of both LTB 4 and PGP and consideration of the SNPs spanning LTA4H will be required when designing therapeutics. This is the first study investigating lung function in smokers and genetic variants specific to genes involved with LTB 4 production and activity. Overall we have identified that polymorphisms spanning ALOX5AP, LTA4H and LTB4R are not major determinants of lung function in smokers. However, these data highlight the potential importance of LTA4H polymorphisms in particular rs1978331C (LTA4H, intron 11). Although the rs1978331 association did not survive correction for multiple testing, the previous associations with; asthma/lung function [15], MI [12] and TB [22] suggest it may be a true association of modest effect size and this SNPs does influence LTA4H expression and/or activity. While no association survived the Bonferroni correction, additive model analyses with rs1978331C (LTA4H, intron 11) showed a p = 0.029 with an increase in FEV 1 and p = 0.020 with FEV 1 / FVC ratio. The mean FEV 1 and FEV 1 /FVC values for the TC heterozygotes and CC homozygotes were similar, but the presence of the minor C-allele for these genotype groups gave higher trait values when compared to the TT homozygotes, suggesting a dosage effect does not occur. For rs1978331 TT versus TC genotype groups there was a 131 ml difference in FEV 1 and for TT versus CC a 126 ml difference in FEV 1 was observed. The level of FEV 1 at a given time depends on 1) the maximum lung function obtained during development, and 2) the rate of decline of lung function with age. Lung function reaches a maximum by age 25-35 years [29]. In smokers the rate of decline in FEV 1 is accelerated and has been calculated to be~38.2 ml/year in males and 23.9 ml/year in females [29] therefore the differences observed between LTA4H rs1978331 genotypes can be considered clinically relevant and equate to > 3 years decline in FEV 1 . These findings therefore provide tentative evidence suggesting that variants in LTA4H may determine clinically relevant lung function levels in smokers.
It is important to acknowledge the limitations of our study. Other SNPs spanning these large genes could be important. There may also be another functional variant in linkage disequilibrium with rs1978331. We cannot exclude the contribution of polymorphisms spanning other 5-LO pathway genes e.g. ALOX5, although existing data did not support their inclusion [20,30,31]. The magnitude of effect of SNPs are modest but in line with the predicted relative risk attributed to a single SNP in a single gene in complex disorders. Finally, the number of individuals used in this study was modest and we have not completed extensive replication in multiple cohorts and so caution is required in the interpretation of our novel findings. To our knowledge these SNPs did not show association with lung function and/or COPD in recent GWAS studies. We have also completed a comprehensive look up of genes previously associated with lung function including LTA4H and ALOX5AP in 20,288 individuals from the general population (the SpiroMeta consortium) and did not identify these genes as containing major determinants of lung function in this large general population cohort [32].

Conclusions
In conclusion, these data did not confirm the hypothesis that polymorphisms in genes specific to LTB 4 production and activity are major determinants of baseline lung function in smokers and do not determine susceptibility to develop COPD. However, rs1978331 (LTA4H, intron 11) may have a modest effect on lung function parameters in smokers. Heterozygosity of this polymorphism has previously been correlated with LTB 4 production, asthma and TB. These findings may be important when considering potential approaches to target LTA4H in COPD.

Additional material
Additional file 1: Baseline lung function (FEV 1 and FEV 1 /FVC ratio) and ALOX5AP, LTA4H and LTB4R SNPs in the smokers (n = 992). This table shows the results of the association analysis between leukotriene pathway SNPs and baseline FEV 1 and FEV 1 /FVC using the additive model. Covariates included in the model were age, gender, height and pack years. Associations with p < 0.05 are shown in bold black.