|Home | About | Journals | Submit | Contact Us | Français|
5-Lipoxygenase (5-LO) catalyzes the rate-limiting step of the biosynthesis of proinflammatory leukotrienes from arachidonic acid (AA) and has been associated with atherosclerosis in animal models and humans. We previously reported that variants of a 5-LO promoter repeat polymorphism were associated with carotid atherosclerosis in humans, an effect that was exacerbated by high dietary AA but mitigated by high dietary n–3 fatty acids.
We sought to confirm these initial observations with a more clinically relevant phenotype such as myocardial infarction (MI).
The 5-LO polymorphism was genotyped in 1885 Costa Rican case-control pairs and tested for association with MI. Functional experiments were carried out to determine whether the associated alleles had differences in mRNA expression.
The frequency of variant genotype groups did not differ significantly between cases and controls. However, a significant gene × diet interaction was observed, in which, relative to the common 5 repeat allele, the 3 and 4 alleles were associated with a higher MI risk in the high (≥0.25 g/d) dietary AA group (odds ratio: 1.31; 95% CI: 1.07, 1.61) and with a lower risk in the low (<0. 25 g/d)AA group (0.77; 0.63, 0.94) (P for interaction = 0.015). Using allele-specific quantitation, the short alleles had expression approximately twice that of the 5 allele (P < 0.0001).
The 3 and 4 variants lead to higher 5-LO expression and provide additional evidence that these alleles are associated with greater risks of atherosclerosis and MI in the context of a high-AA diet.
The 5-lipoxygenase (5-LO) pathway, which generates leukotrienes from arachidonic acid (AA), has garnered a great deal of attention in the past few years for its potential role in cardiovascular disease (CVD)–related traits. This stems from a series of biochemical, genetic, and pharmacologic studies reported since 2002, which collectively have provided strong evidence for the proatherogenic role of leukotrienes, as reviewed by several groups of authors (1–3). For example, our group previously showed that 5-LO deficiency in mice protects against aortic lesion formation but leads to other metabolic disturbances (4–6). Other groups have reported the involvement of other 5-LO pathway genes in atherosclerosis traits as well, such as the leukotriene (LT) B4 (LTB4) receptors and 5-LO–activating protein (FLAP) (7–11).
Human studies have also provided evidence consistent with the notion that the 5-LO/LT pathway participates in atherosclerotic processes. Two studies have shown that 5-LO, FLAP, and LTA4 hydrolase (LTA4H) are abundantly expressed in arterial walls of CVD patients and that 5-LO has markedly greater expression in advanced lesions and localizes to inflammatory cells, such as macrophages, dendritic cells, and neutrophilic granulocytes (12, 13). Thus, 5-LO pathway genes and leukotrienes could potentially contribute to lesion progression in humans, from the initial stages of progression to the development of complex plaques that are prone to rupture and that are the cause of MI. As part of our genetic studies, we previously reported that persons carrying variant alleles of a 5-LO promoter polymorphism consisting of tandem Sp1-binding sites have significantly greater carotid atherosclerosis (14). An interesting extension of these analyses was that high dietary AA, the primary 5-LO substrate that leads to proinflammatory leukotrienes, exacerbated the atherogenic effect of the variant alleles, whereas high dietary n–3 fatty acids, such as docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA), that are also substrates for 5-LO but that generate noninflammatory leukotrienes, blunted this effect (14). To confirm these initial genetic and gene × diet interactions with a more clinically relevant phenotype such as MI, we genotyped the 5-LO promoter polymorphism in samples from a large study of Costa Ricans and carried out functional analyses of the at-risk alleles.
Cases for this study were adult patients who were survivors of a first acute MI as diagnosed by a cardiologist at any of the recruiting hospitals in the Central Valley of Costa Rica. A study cardiologist confirmed all cases according to the World Health Organization criteria for MI, which requires typical symptoms plus either elevation in cardiac enzyme concentrations or diagnostic changes in the electrocardiogram. Enrollment was carried out in the step-down unit of the recruiting hospitals, and cases were ineligible for participation if they died during hospitalization, were >75 y old on the day of their first MI, were physically or mentally unable to complete the questionnaire, or had a previous hospital admission related to CVD. For each case, one population-based control subject, matched for age (±5 y), sex, and county of residence, was recruited. The controls were randomly selected by using data from the National Census and Statistics Bureau of Costa Rica. Control subjects were ineligible if they had ever had an MI or if they were physically or mentally unable to complete the questionnaires. The catchment area consisted of 34 counties in the Central Valley of Costa Rica. Participation was 98% for cases and 88% for controls.
Trained personnel visited all study participants at their homes for data collection. Fieldworkers collected anthropometric measurements while study subjects were wearing light clothing and no shoes. All measurements were performed in duplicate, and the average was used for analyses. Sociodemographic characteristics, medical history, and lifestyle habits were collected by using a general questionnaire. Dietary intake was collected by using a food-frequency questionnaire (FFQ) that was developed and validated specifically to assess fatty acid intake among the Costa Rican population (15). The validity coefficient for the assessment of AA with the use of the FFQ was high (0.53), a finding consistent with the performance of the FFQ for other fatty acids (15), and the correlation coefficient between the FFQ and seven 24-h recalls was 0.62. Biological samples were always collected in the morning after an overnight fast. After a 12–14-h fast, blood samples (20 mL) were drawn in tubes containing 0.1% EDTA and immediately stored at 4 °C. Within 36 h, the samples were centrifuged at 2500 rpm for 20 min at 4 °C to isolate and aliquot plasma and white blood cells. The samples were then sealed and stored under N2 at −80 °C until analysis. For the present analyses, genotype information, complete data on all descriptive variables, and potential confounders were available from 1885 case-control pairs.
All subjects gave written informed consent on documents approved by the Human Subjects Committee of the Harvard School of Public Health and the University of Costa Rica. Approval for the present study was also obtained from the Institutional Review Board of the Keck School of Medicine, University of Southern California.
Genotyping of the 5-LO promoter polymorphism was performed by using previously described methods (14), without knowledge of case-control status. Genotyping of the single-nucleotide polymorphisms (SNPs) in exons 1 and 2 was carried out through allelic discrimination assays using the TaqMan system [Applied Biosystems Inc (ABI), Foster City, CA (16)]. Genotyping assays were custom-designed with the use of the Assays-by-Design service (ABI).
Total RNA was isolated from 123 buffy coats of controls with selected genotypes by using RNeasy kits (Qiagen Inc, Valencia, CA); it was then reverse-transcribed by using cDNA Archive kits (ABI). Real-time transcript abundance for 5-LO and β-actin were determined in triplicate by using previously developed assays (ABI). The amount of 5-LORNAwas normalized to β-actin RNA, and the replicates were averaged to determine RNA abundance in each sample relative to a reference sample. Wilcoxon’s rank-sum test was used to determine differences between genotype groups; P < 0.05 was considered statistically significant.
Allele-specific expression of the 3 and 4 promoter alleles relative to the 5 allele was determined by using pyrosequencing technology (17), which provides a sensitive method for quantitating allelic intensities in SNP genotyping reactions. These experiments were performed in subjects who were heterozygous for the 45 and 35 genotypes by using silent coding substitutions in exons 1 (C21T) and 2 (G270A), which have been reported to be in linkage disequilibrium with the 4 and 3 promoter alleles, respectively (18). Briefly, polymerase chain reaction (PCR) was used to amplify amplicons encompassing each SNP from cDNA and genomic DNA (as control reactions) of 3A/5G and 4T/5C heterozygotes by using the following primers—exon 1 (C21T): P-forward 5′CGCCATGCCCTCCTACAC3′ and P-reverse 5′CCACGAGGCTGAGGT AGATG3′; exon 2 (G270A): P-forward 5′CGCAAGTACTGGCTGAATGA3′ and P-reverse 5′CCTCAGGACAACCTCGACAT3′. PCR conditions were as follows: 0.4 µmol each primer/L, 1.5 mmol MgCl2/L, 200 µmol dNTP/L (Invitrogen, Carlsbad, CA), and 0.5U Platinum Taq polymerase (Invitrogen) with 35 cycles at 94 °C for 30s, 55 °C for 30s, and 72 °C for 45s. The PCR products were then subjected to pyrosequencing reactions according to the manufacturer’s protocol by using the following detection primers—exon 1 (C21T): 5′CCCTCCTACACGGTC3′; exon 2 (G270A): 5′CGATGTAGTCCCCGT3′. For genomic DNA, the signal ratio from the 2 alleles is expected to be 1, whereas a ratio different from 1 in the cDNA (RNA) reactions indicates differential transcript abundance. Each experiment was carried out in quadruplicate, and differences between the ratio obtained from RNA and that obtained from DNA were determined by using paired t tests; P < 0.05 was considered statistically significant.
All data were analyzed with SAS software (version 9.0; SAS Institute Inc, Cary, NC). The significance of differences in health characteristics and potential confounders was assessed by using McNemar’s test and paired t tests, if normally distributed, or Wilcoxon’s signed-rank test, if not normally distributed. Odds ratios (ORs) and 95% CIs were estimated from multiple conditional logistic regression models using a dominant model that included the 55 genotype as the reference group, carriers of “short” alleles with 3 and 4 Sp1 repeats (33,34,44,35,45,36,37, 46,48) in group 1, and carriers of “longer” alleles with ≥6 repeats (56,57,58,59,66,67) in group 2. Median dietary AA (0.25 g/d), tertiles of dietary AA, and tertiles of DHA and EPA (g/1000 kcal) were used to stratify subjects into low- and high-intake groups. Likelihood ratio tests were used to test for interactions; confounders such as income, cigarette smoking, dietary saturated fat, and total energy intake were included. Other variables tested— but not included in the final model because they did not affect the results—were physical activity, history of diabetes, history of hypertension, fiber intake, cholesterol intake, polyunsaturated fatty acid intake, aspirin use, and abdominal adiposity.
The clinical characteristics of the cases and controls used for this study are shown in Table 1. As expected, compared with the controls, the cases had higher frequencies of risk factors for MI, including abdominal adiposity, diabetes, and hypertension, and had higher intake of dietary cholesterol and various fatty acids (Table 1).
The GC-rich core promoter region of 5-LO contains a sequence of tandem repeats (GCGGG) that are binding sites for the transcription factor Sp1 and that potentially regulate transcriptional activation under inflammatory conditions. As our group (14) and other investigators (18) reported previously, the most common promoter allele in the white population consists of 5 repeats and has a frequency of ≈0.80, whereas alleles of 4, 3, and longer than 5 repeats (ie, 6, 7, 8) are less common with frequencies of ≈0.15, 0.03, and (collectively) 0.02, respectively. The frequencies of the various genotype groups in the Costa Ricans, which are comparable to those reported previously, are shown in Table 2.
Our group’s initial human studies showed that the variant alleles (ie, 3, 4, 6, 7, or 8 repeats) were associated with greater carotid atherosclerosis than was the 5 allele (14). Although our group was previously not able to assess the effect of the alleles with less than 5 repeats separately from that of the longer alleles, because of the small size of the cohort in that study, the substantially larger Costa Rican study did allow us to carry out this analysis in the present study. As shown in Table 3, there was no association between the variant 5-LO promoter alleles and risk of MIunder a dominant model. Adjustment for confounders such as smoking and dietary fat intake did not alter these findings (Table 3), and nor did further adjustments for physical activity, history of diabetes, history of hypertension, fiber intake, cholesterol intake, polyunsaturated fatty acid intake, aspirin use, and visceral adiposity (data not shown). A recessive model, in which persons who are homozygous for the 3 and 4 alleles were considered as a separate group (ie, 33, 34, and 44), also did not show an association between the 5-LO promoter repeats and MI (data not shown).
On the basis of our group’s previous observations, we next determined whether risk of MI was mediated by an interaction between the 5-LO promoter alleles and dietary AA intake. Under a dominant model with the same genotype classifications as described above, this analysis showed a significant gene × diet interaction (P = 0.015), in which persons in the high dietary AA group carrying 1 or 2 copies of the shorter 3 and 4 repeats had a greater risk of MI (OR: 1.31; 95% CI: 1.07, 1.61) that did persons with 55 in the low-AA group (Table 4). In contrast, the shorter 3 and 4 repeats were protective in the context of low dietary AA (0.77; 0.63, 0.94). After adjustment for various confounders, the ORs for MI in both dietary AA groups remained essentially unchanged (Table 4). We next used the same analytic strategy with tertiles of dietary AA and observed a trend toward a lower MI risk in the lowest tertile (0.79; 0.60, 1.03) and a higher risk in the highest tertile (1.23; 0.94, 1.62) with the shorter5-LOrepeats. The OR for MI risk in the middle tertile was 0.94 (0.72, 1.24).
Although these latter results are not statistically significant (P for interaction = 0.15)—which we attribute to decreased power, because there were 35% fewer individuals in each genotype or diet group—they support the significant nutrigenetic association with median dietary AA concentrations. A recessive model in which persons who were homozygous for the 3 and 4 alleles were stratified by median dietary AA also yielded a marginally significant gene × diet interaction (P = 0.059) in the same direction as the interaction found under a dominant model, which remained unchanged after adjustment for confounders (P = 0.056). We also carried out similar analyses with the median and tertiles of dietary n–3 fatty acids (g/1000 kcal), such as EPA and DHA, which were based on our previous observations, but we did not detect any significant gene × diet interactions. For example, with the use of the same fully adjusted dominant model as was used for dietary AA, the ORs for MI risk in the lowest, middle, and highest tertiles of dietary DHA and EPA with the shorter 5-LO repeats were 0.97 (0.74, 1.27), 0.69 (0.52, 0.92), and 0.86 (0.65, 1.14), respectively.
To investigate functional differences between the 5-LO promoter alleles, we isolated total RNA from 123 buffy coat–derived leukocytes with different genotypes and carried out real-time quantitation of 5-LO concentrations. These analyses were performed only with RNA from controls to avoid any potential effects of medication use on 5-LO expression in the cases and focused on the shorter promoter alleles because a nutrigenetic association with MI was observed with these variants. As shown in Figure 1, there were no significant differences in 5-LO mRNA concentrations between persons who were homozygous or heterozygous for the 3 and 4 variants and those who were homozygous for the 5 repeat allele. 5-LO expression between genotype groups also did not differ significantly when stratified by dietary AA concentrations or when the 3 and 4 alleles were analyzed separately (data not shown).
As another method for evaluating the functional relevance of the associated variants, we carried out allele-specific expression using 2 silent substitution SNPs located in exon 1 (C21T and Thr6Thr) and exon 2 (G270A and Thr90Thr) that had previously been reported to be in linkage disequilibrium with the 4 and 3 repeat alleles, respectively (18). Because both SNPs are present in the coding sequence of the 5-LO mRNA, the transcript abundance from a promoter containing 3 or 4 repeats relative to a 5 allele promoter can be determined in persons who are heterozygous for the haplotype harboring the respective promoter variant and SNP. As shown in Figure 2A, the ratio obtained from RNA in 4T/5C heterozygotes was 1.7, which is indicative of a transcript abundance from the 4 allele that was nearly 2-fold that from the 5 allele. Similarly, the signal ratio from RNA of 3A/5G heterozygotes was ≈2.5 (Figure 2B), which is indicative of transcript abundance from the 3A allele that was >2-fold that from the 5G allele. As would be expected, control reactions using genomic DNA from the same heterozygous persons yielded allelic ratios of 1 for both SNPs (Figure 2A and B).
In the present study, we have provided evidence for a nutrigenetic association between functional 5-LO promoter alleles and MI in a large Costa Rican population. Other investigators have also reported associations of other 5-LO/LT pathway genes with CVD traits. For example, deCode Genetics reported an association between FLAP haplotypes and MI or stroke (20); this association has been replicated in some populations (20) but not others (21, 22). More recently, haplotypes of the LTA4H gene have been associated with MI in whites and African Americans; the effect was more pronounced in the latter group (23). Finally, Iovannisci et al (24) showed that LTC4 synthase variants increase coronary artery calcification and carotid atherosclerosis. It is important that these other human studies are also bolstered by functional data showing that the associated variants or haplotypes lead to increased leukotriene production. When this information is taken together with our present results, there is a growing body of biochemical and genetic evidence supporting the notion that genetically controlled increased 5-LO/LT pathway activity is an important mediator of CVD phenotypes in humans.
Our group previously reported that variant 5-LO promoter alleles were associated with a greater degree of carotid atherosclerosis, which was enhanced by high dietary AA (14). However, because of the small samples sizes, our group was unable to differentiate between the shorter and longer alleles in that study. Given the large Costa Rican study, this analysis was possible, and it showed that the shorter repeats exhibit a nutrigenetic association with MI. Thus, these results replicate our previous observations with a more clinically relevant endpoint and support the notion that the shorter alleles are associated with CVD traits. It interesting that, whereas the shorter repeats increased the risk of MI in the presence of high dietary AA, they had a protective effect in the presence of low dietary arachidonate, a phenomenon that had also been observed previously (14). It is possible that the shorter alleles have opposite effects because they are sensitive to substrate availability and thus could perhaps be down- or up-regulated on the basis of cellular AA concentrations.
Another interesting observation from our group’s earlier study was that the proatherogenic effect of the variant alleles was mitigated by high dietary intakes of DHA and EPA (14), which are also 5-LO substrates but which lead to noninflammatory leukotrienes. However, we did not observe a similar protective effect in the Costa Ricans with either the shorter or longer alleles. Such a negative finding may be due to the relatively low consumption of fish by the population in the Central Valley region of Costa Rica (25), from which this sample was collected. With respect to the longer repeats, the results suggest that these alleles are protective, but, because of their low frequency, a more definitive assessment of their effect will require larger sample sizes.
The prevailing evidence for how 5-LO and leukotrienes promote atherogenesis is through inflammatory-mediated mechanisms in the artery wall. For example, both our group (4) and other investigators (12, 13) previously reported that expression of the 5-LO, FLAP, or LTA4H gene (or all 3) is abundant in atherosclerotic lesions, that it co-localizes with macrophages that have infiltrated the arterial wall, and that it increases with both plaque progression and instability. On the basis of this evidence and the known chemotactic properties of leukotrienes, it can be presumed that the 5-LO pathway increases the risk of developing CVD through inflammatory processes initiated through the oxidation of AA. It is interesting that previous studies also suggested that n–3 fatty acids protect against sudden cardiac death and ventricular arrhythmias through other mechanisms that occur within myocytes (26) and that are distinct from lipoxygenase-catalyzed oxidation reactions. Thus, it is possible that 5-LO- and AA-derived leukotrienes promote atherosclerotic vascular disease through an inflammatory mechanism in leukocytes—which do express 5-LO—whereas the cardioprotective effects of n–3 fatty acids occur through the alteration of myocyte function—which, to our knowledge, does not express 5-LO or other leukotriene synthesis enzymes.
To assess functional differences between the shorter promoter repeats, we used a robust and internally controlled allele-specific expression assay that is much less susceptible to the confounding effects of other potential trans-acting regulators of 5-LO expression. These experiments showed that the 3 and 4 repeat alleles have significantly higher 5-LO mRNA transcript abundance than does the common 5 allele. Notably, such allelic imbalances were not observed in genomic DNA from the same heterozygous persons. Therefore, these functional differences are consistent with the genetic associations and gene-diet interactions for the 5-LO promoter alleles and CVD traits in the Costa Rican population and other populations (14). Such observations, however, could appear counterintuitive, because the shorter alleles have 1 or 2 fewer binding sites for the transcription factor Sp1, which one would predict may lead to less 5-LO expression and a more protective effect. For example, previous in vitro promoter constructs showed that variant alleles with either fewer or more than 5 repeats had less transcriptional activity than did the 5 allele (18). Subsequent cell transfection studies by other investigators also reported similar functional differences, whereby a positive linear relation was observed between the number of Sp1 repeats and promoter activity (27, 28). One potential explanation for this apparent paradox may be related to the more complex transcriptional regulation of 5-LO in vivo than may be seen in such in vitro experiments. As an example, Uhl et al (29) reported that the GC-rich Sp1 binding sites in the 5-LO promoter undergo methylation in certain cell types, which contributes to tissue specific expression of 5-LO. Such an effect would not be observed with cloned constructs in transfection studies and could potentially account for the differences between the in vitro studies and the present results.
As part of our functional characterization of the 5-LO promoter alleles, we also compared total mRNA concentrations from persons carrying the shorter 3 and 4 repeats with those from persons who were homozygous for the 5 repeat allele. In contrast to the allele specific expression studies described above, these analyses did not show differences in 5-LO mRNA concentrations. It is possible that, in addition to cis-acting elements within its own promoter, 5-LO may be regulated by other trans-acting genetic factors, variations of which could complicate the quantitation of total 5-LO mRNA concentrations. Alternatively, isolated buffy coats are composed of a mixture of various leukocyte populations, each of which expresses 5-LO in different concentrations; for example, macrophages and granulocytes have higher expression than do other types of leukocytes (30). Therefore, 5-LO expression could vary between persons, regardless of genotype, depending on the cellular composition of buffy coats. A more definitive assessment of differences in total mRNA concentrations between variant 5-LO alleles will require additional experiments using specific and isolated cell types, such as macrophages, lymphocytes, or neutrophils (or all 3). Obtaining such cell populations, particularly with respect to genotype, would also allow cell-specific methylation analyses of the 5-LO core promoter region and the Sp1 repeats to be carried out, which, as discussed above, may also increase our understanding of the in vivo regulation of the 5-LO gene.
Despite the relatively large size of the Costa Rican study, we did not detect an association of the 5-LO promoter repeats with MI without using dietary AA in the analyses. A recent study also failed to detect an association with the 5-LO promoter polymorphisms and MI in Spanish patients; however, that study included ≈80% fewer subjects than our Costa Rican population (27). Such negative findings could be due to the complexity of the phenotype under study, because MI likely is under the genetic control of pathophysiologic processes that make coronary plaques unstable and susceptible to rupture and of those processes that contribute to lesion progression. As a result, the association of 5-LO with MI may be more difficult to detect than its effect on lesion size. In addition, other genetic or dietary differences (or both) between the Costa Rican and US populations may also confound a nutrigenetic association with 5-LO. For example, according to the United Nations Food and Agriculture Organization (Internet: www.fao.org), beef and eggs, which are the primary sources of dietary AA, are ≈3 (62.1 compared with 23.3 g/d for each person) and ≈1.5 (40.5 compared with 25.9 g/d for each person) times as available in the United States as in Costa Rica, respectively. Thus, one potential limitation of our study is that the association of 5-LO with atherosclerosis may be stronger and more readily detectable in a US population (14), in whom dietary AA is more abundant. Finally, it is noteworthy that the gene × diet interactions between the shorter 5-LO promoter repeats and MI risk in the Costa Ricans went in the opposite directions to those in the US population. Without knowledge of the dietary status of subjects, the association between these alleles and MI could be neutralized in an unstratified analysis that considered all persons together.
In conclusion, we showed that 5-LO promoter alleles with <5 Sp1-binding motifs are more susceptible to the effects of dietary AA on the risk of MI and are associated with higher 5-LO expression than are alleles with 5 repeats. These results are consistent with the notion that certain persons could benefit more from certain dietary modifications, such as a reduction in the intake of beef and eggs (major sources of AA), and they suggest that personalized dietary recommendations may become more effective through the incorporation of genetic information. More comprehensive analyses of the 5-LO gene using haplotype-based approaches in other populations will be required to further clarify the role of the 5-LO gene in MI and other CVD traits and to determine which alleles are the actual causal variants.
2Supported by grants no. HL079353 (to HA), HL30568 (to AJL), and HL060692 (to HC) from the National Institutes of Health; grant no. 0355031Y (to MM) from the American Heart Association; and Reserch Facilities Improvement Program grant no. C06 (RR10600-01, CA62528-01, and RR14514-01) from the National Center for Research Resources.
The authors’ responsibilities were as follows—HA and HC: designed the study, obtained funding, and wrote the manuscript; AB: carried out data analyses and contributed intellectual content to the manuscript; JH and HW: obtained data and contributed intellectual content to the manuscript; and MM and AJL: obtained funding and contributed intellectual content to the manuscript. None of the authors had a personal or financial conflict of interest.