Association Analysis of Allelic Variants of USF1 in Coronary Atherosclerosis
Objective— USF1 regulates the transcription of more than 40 cardiovascular related genes and is well established as a gene associated with familial combined hyperlipidemia, a condition increasing the risk for coronary heart disease. No detailed data, however, exists on the impact of this gene to the critical outcome at the tissue level: different types of atherosclerotic lesions.
Methods and Results— We analyzed the USF1 in 2 autopsy series of altogether 700 middle-aged men (the Helsinki Sudden Death Study) with quantitative morphometric measurements of coronary atherosclerosis. SNP rs2516839, tagging common USF1 haplotypes, associated with the presence of several types of atherosclerotic lesions, particularly with the proportion of advanced atherosclerotic plaques (P=0.02) and area of calcified lesions (P<0.001) of the coronary arteries. Importantly, carriers of risk alleles of rs2516839 also showed a 2-fold risk for sudden cardiac death (genotype TT versus CC; OR 2.10, 95% CI 1.17 to 3.75, P=0.04). The risk effect of rs2516839 was present also in aorta samples of the men.
Conclusions— Our findings in this unique study sample suggest that USF1 contributes to atherosclerosis, the pathological arterial wall phenotype resulting in coronary heart disease and in its most dramatic consequence—sudden cardiac death.
In addition to established risk factors such as elevated total cholesterol levels, hypertension, smoking, and obesity, the progression of coronary heart disease (CHD) is modified by a still unknown composition of risk alleles of genes interacting with lifestyle risks. We initially identified USF1 as the first major gene associating with familial combined hyperlipidemia (FCHL),1 a phenotype characterized by elevated levels of serum total cholesterol, or triglycerides, or both. Later, several other studies also implied an association between the USF1 gene and severe common aberrations in lipid and glucose metabolism.2–12 A feature common to these aberrations is their contribution to the risk for premature CHD and myocardial infarction (MI). We have recently shown that allelic variants of the upstream transcription factor 1 (USF1) gene affect the prospective risk for cardiovascular disease (CVD) in two large population-based studies.13 Thus the USF1 gene is a promising candidate to contribute to the complex genetic background of CHD. The gene is biologically highly relevant, it encodes a ubiquitously expressed transcriptional regulator of several genes that are functionally important for the lipid accumulation, inflammation, and thrombotic complications of the coronary plaque.14–17 However, limited data so far exists on the association of the USF1 gene and the critical biological outcome: the extent and severity of coronary atherosclerosis. To study whether allelic variants of the USF1 relate to the progression of coronary atherosclerosis, we used an autopsy series of middle-aged Finnish men with detailed quantitative measurements of various types of atherosclerotic lesions in coronary arteries. Our data suggests that specific USF1 alleles contribute to coronary atherosclerosis.
The Prospective Autopsy Series of Middle-Aged Men
The Helsinki Sudden Death Study (HSDS) comprised 2 prospective series of altogether 700 men, aged 33 to 70 (mean age 53; Table 1), who underwent a medicolegal autopsy.18,19 The 2 series were collected at 10-year intervals; the first series (n=400) during 1981 to 1982 and the second series (n=300) during 1991 to 1992.
Protocols of 2 international studies were used to define atherosclerosis.20,21 The areas of different types of lesions were expressed in percentages by dividing the lesion area by the total area of the artery sample and multiplying it by 100%.
To comprehensively analyze the allelic diversity of the USF1 gene locus, htSNPs rs10908821, rs2774276, rs2073658, rs2516839, rs1556259, and rs2774279 (Figure 1) were first genotyped in the 1991 to 1992 autopsy series. SNPs rs2073658, rs2516839, and rs2774279 capturing the 2 most common haplotypes of USF1 (Figure 1) were genotyped in the older 1981 to 1982 series.
The linkage disequilibrium analysis and estimation of haplotype frequencies were performed for the HSDS 1991 to 1992 series with the publicly available Haploview software version 3.2.22
The distributions of the continuous atherosclerosis variables were highly skewed (Table 1). We therefore modeled the risk of atherosclerosis with age, BMI, and series-adjusted ordinal regression model (cumulative logit model) where quartiles of the atherosclerotic variable were the ordinal outcome (supplemental Table I, available online at http://atvb.ahajournals.org).
For further details, please see supplemental materials, available online at http://atvb.ahajournals.org.
The cause of death for one third of the study subjects was sudden cardiac death (SCD) attributable to coronary heart disease (CHD) with or without myocardial infarction (MI; Table 1). In detailed computer-assisted morphometric measurement of different types of atherosclerotic plaques, over 90% of the study subjects had signs of early atherosclerosis and as many as 40% had areas of complicated atherosclerotic lesions in their coronary arteries (Table 1).
We genotyped a total of 6 SNPs covering the 5.7 kb USF1 gene and defining a total of 5 haplotypes in the Finnish population because of the high extent of linkage disequilibrium between the SNPs in this population (Figure 1). These haplotype-tagging SNPs facilitated the monitoring of most of the allelic diversity of USF1 in Finns.
Association of USF1 Variants With Atherosclerosis
Coronary and Aortic Atherosclerosis
In both autopsy series, collected with a 10-year interval, a significant risk effect of the T-allele of SNP rs2516839 was observed in ordinal regression analysis: Carriers of 2 risk alleles were 2.4 times as likely to have more severely calcified coronary arteries in the 1981 to 1982 series (OR 2.39, 95% CI 1.30 to 4.40) and 3.2 times as likely in the 1991 to 1992 series (OR 3.20, 95% CI 1.68 to 6.09) than noncarriers (Figure 2). Although the risk effect of rs2516839 was in general more evident in the later autopsy series (1991 to 1992), in both series the carriers of the risk allele (T) had greater odds of having larger advanced atherosclerotic lesion areas as the noncarriers both in their coronary arteries and in abdominal aorta.
In the combined data set of the 2 autopsy series, with increased power to detect significant risk effects of the USF1 alleles, we observed htSNP rs2516839 to significantly associate with areas of several types of advanced atherosclerotic lesions (Table 2). T-allele of the SNP correlated with a higher probability of more severe atherosclerosis. Conversely, carrying 2 C-alleles of this SNP seemed to protect from advanced atherosclerotic lesions. Although this protective effect was statistically significant in ordinal regression analysis for several atherosclerotic lesion types in the coronary arteries, and for the area of advanced (characterized by fibrous or complicated area) and calcified atherosclerotic lesions both in coronary arteries and abdominal aorta, the trend of this protective effect was detectable for all the lesion types studied (excluding coronary fatty streaks; Table 2). Further, although SNPs rs2073658 and rs2774279 both showed association with calcification of the coronary arteries, their risk effects were nonsignificant at the presence of the risk effect of rs2516839 (supplemental Table II).
Carriers of the risk allele (T) of rs2516839, the SNP associated with coronary and aortic atherosclerosis, also had a suggestive increased risk for having >50% coronary narrowing in more than one coronary artery (for details please see supplemental materials).
This risk-associated T allele of the SNP does not tag a single haplotype of USF1, but is present in the 2 most common (61%) haplotypes of USF1 (CTCTAG and CCCTAA, Figure 1). In our study sample, CTCTAG associated with the proportion of advanced lesion area out of the total lesion area and CCCTAA associated with the extent of coronary calcification. These associations were, however, weaker than those obtained from analysis of rs2516839 genotypes (for details please see supplemental materials).
USF1 Variants and Sudden Cardiac Death
In the combined data set the carriers of the rs2516839 risk (TT) genotype had a 2-fold risk for SCD when compared to that of the carriers of the protective genotype (TT versus CC; OR 2.10, 95% CI 1.17 to 3.75, P=0.01, P adjusted for multiple comparisons=0.04, and CT versus CC; OR 1.92, 95% CI 1.09 to 3.39, P=0.02, P adjusted for multiple comparisons=0.10). In the 1991 to 1992 series, increased risk for SCD was associated with the rs2516839 risk allele (T) containing genotypes (TT versus CC; OR 2.95, 95% CI 1.11 to 7.85, P=0.03, P adjusted for multiple comparisons=0.12, and CT versus CC; OR 2.94, 95% CI 1.12 to 7.72, P=0.03, P adjusted for multiple comparisons=0.25). The risk effect of SNP rs2516839 on SCD in the older 1981 to 1982 series failed to reach statistical significance, although we observed an OR suggesting a trend for risk (OR 1.71, 95% CI 0.82 to 3.58, P=0.15). In the older 1981 to 1982 series increased risk for SCD was, however, attributed to the closely linked AA-genotype of rs2774279 (OR 2.39, 95% CI 1.09 to 5.24, P=0.03, P adjusted for multiple comparisons=0.45, when compared to that of the GG genotype).
We further tested whether the association of rs2516839 with SCD was independent of its association with coronary calcification. Fitting the calcification covariate in the model reduced the risk associated with rs2516839 from odds ratio 2.10 to 1.59 (genotype TT versus CC). Fitting other lesions types, however, led to more modest reduction in the risk associated with rs2516839 (supplemental Table III and supplemental results). These result suggested that part of the increased SCD risk associated with rs2516839 was attributable to the contribution of rs2516839 to the development of severe coronary artery disease characterized by coronary calcification.
Here we have studied the impact of allelic variants of USF1 gene on the biological outcome of disturbed lipid metabolism and vessel wall endothelial function; the atherosclerosis of coronary arteries. We observed an association of USF1 variants with coronary narrowing and several atherosclerotic lesion phenotypes, measured with computer-assisted planimetry, both in coronary arteries as well as in abdominal aorta. The risk effect of USF1 was evident in two distinct autopsy series, and we observed consistent and significant associations in the data after adjusting for multiple comparisons.
The risk effect of USF1 was more evident in the later autopsy series (1991 to 1992). There the proportion of study subjects affected with advanced atherosclerotic lesions was smaller than in the 1981 to 1982 series (Table 1). This decrease is likely to be explained by reduced clustering of lifestyle risk factors, which can increase the environmental “noise” and complicate genetic analyses in the 1981 to 1982 series compared to the 1991 to 1992 series. Nevertheless, indication of the risk effect of USF1 was present in both series.
Importantly, the atherosclerosis-associated USF1 variant significantly increased the risk for the most severe consequence of atherosclerosis: prehospital sudden cardiac death. SCD is nowadays the most significant contributor to mortality from CHD, in particular in middle age, where up to 80% of deaths attributable to CHD belong to this category.23–25 Previous studies have suggested a role for USF1 in the etiology of CHD; we initially identified USF1 as the first major gene associating with familial combined hyperlipidemia (FCHL) in families who were ascertained through a proband with premature CHD. Subsequently, we observed USF1 variants to contribute to the risk of incident cardiovascular disease among women.13 Although we observed an association between USF1 variants and CHD also at our present study, our results were not directly comparable with the previous findings as our study sample consisted of men only. Interestingly, in our study, a part of the increased risk for SCD associated with USF1 was attributable to the contribution of the gene to coronary calcification. The gene, however, also seemed to have a risk effect independent from the effect of atherosclerotic lesions.
In a recent population-based cohort study, Reiner and colleagues observed a USF1 variant (rs3737787, which strongly correlates with rs2073658 of our study) to associate with risk of coronary artery calcium.12 In line with this, the amount of coronary calcification was significantly associated with variation at the USF1 locus in our study, although rs2073658 was not the most significantly associated variant. While we observed USF1 to associate also with other types of atherosclerotic lesions, all of which were significantly correlated with the amount of calcification, the association with calcification remained most substantial. Some evidence already exists on how USF1 could participate in the build up of calcified lesions in the arteries; USF1 is involved in the regulation of osteopontin expression in arterial smooth muscle cells.26,27 Osteopontin is an inhibitor of vascular calcification,28 and as a transcription factor USF1 could regulate the expression of other important genes involved in similar processes as well.
Most of the studies on the USF1 gene have examined only a few allelic variants of the gene, instead of comprehensively using the genetic variation at the USF1 locus.2–4,8,9,11,29 Here, with knowledge of the linkage disequilibrium structure at the USF1 locus in our study population, we genotyped 6 htSNPs to capture the full allelic diversity of the USF1 gene in Finns, known to show wider linkage disequilibrium intervals between the SNPs.30 The SNP showing the strongest evidence of an association with atherosclerosis and SCD was rs2516839, tagging the 2 most common haplotypes of USF1.
SNP rs2516839, located in an untranslated exon of USF1, did not consistently associate with serum triglyceride levels in Finnish familial combined hyperlipidemia families, although a suggestive association was observed in one of the analyses.1 The common allele (T) of the SNP, the risk allele in our data, associated with higher cholesterol and triglyceride levels also in Utah families ascertained for type 2 diabetes.7 In another study, our protective genotype (CC) of SNP rs2516839 showed suggestive association with lower risk of metabolic syndrome in Chinese hospital cases.6 However, in a population based data set from the Finns the C-allele of rs2516839 was associated with higher lipid values among study subjects with CVD.13 Unfortunately, we do not have data on the cholesterol levels of our study sample and thus could not directly relate atherosclerotic lesion areas to cholesterol levels, and the exceptional ascertainment of our study population prevents direct comparisons with previous studies. What emerges from this study is evidence of association of USF1 with quantitative atherosclerotic phenotypes of arterial wall, the tissue level biological end state resulting in CHD.
The previous studies underline the general influence of USF1 in regulating the expression of critical genes of lipid and glucose metabolism. As a transcription factor regulating the expression of more than 40 cardiovascular related genes29 involved in lipid metabolism, hemostasis, inflammation, and endothelial function, USF1 could contribute to the development of atherosclerosis and its complications through various pathways. Evidence of these contributions already exists for lipolysis,4,8 inflammation,1,12,29 and now for calcification from our study and Reiner and colleagues.12 Variation at the USF1 locus could thus contribute to the increased risk of CHD among FCHL patients by affecting lipid metabolism related pathways, which increase the levels of serum lipids. Another separate pathway leading to an increased risk of CHD could stem from the effect of USF1 on coronary calcification, and suggestive evidence for such separate effect already exists.12 This separate effect could involve inflammation pathways with markers such as CRP.14,31 USF1 variants could also contribute to SCD via factors affecting plaque stability. How exactly the effect of USF1 is related in these different pathways remains to be assessed with further studies, for instance with studies addressing the relationship between the USF1 gene variants and arterial transcript profiles.
In addition to the role of USF1 in various atherosclerosis pathways, its genetic analysis is possibly complicated by other genetic and environmental factors. The effect of USF1 could be cell type– or tissue-specific, involve interaction with other genes, and depend on different hormonal/environmental cues. Some evidence of sex- or age-specific effects of USF1 already exists.1,10–13 Further, the influence of genetic variation in USF1 could be more readily observed in subjects with certain pathophysiological condition, such as CHD, diabetes, or obesity. These gene-gene or gene-environmental interactions may explain some of the heterogeneity between previous USF1 studies.
The number of functional genetic variants at the USF1 locus remains also unsolved. The locus could harbor several rare or common variants with distinct effects. Some evidence of the functional role of USF1 variants on the transcript levels of downstream genes already exists from fat biopsy samples.1,29 Sequencing of the complete USF1 and neighboring region in large study samples and detailed functional studies would facilitate the identification of the variants of USF1 with functional relevance.
The variant showing the strongest evidence of an association with atherosclerosis and SCD in our study, rs2516839, is located within the 5′ untranslated region of the first exon of USF1. Although untranslated variants do not alter the amino acid sequence of a protein, they can for instance affect the transcription level of mRNA, splicing patterns, and other posttranscriptional modifications of mRNA, or stability or localization of mRNA in the cell. The effect of rs2516839 and other USF1 variants on such features should be assessed in future functional studies. Rs2516839 could also be a marker for another yet unrecognized functional marker or domain.
Given the high linkage disequilibrium at the USF1 locus and at its surroundings, we cannot exclude the potential contribution of other genetic variants neighboring USF1 to the association signal we observed. Future studies, preferably in populations with reduced linkage disequilibrium at the locus, should assess the relative impact of a large number of variants in an extended chromosomal region. Accordingly, Reiner and colleagues suggested in their USF1 study that “the true susceptibility effect on mortality may be attributable, at least in part, to neighboring polymorphisms in other genes.”12
To conclude, our data provide genetic evidence that specific USF1 alleles would contribute to coronary atherosclerosis. The significance of the role of USF1 in atherosclerotic process was evidenced by data from 2, similarly ascertained, independent autopsy series as well as by the observed association also with aortic lesions. The USF1 gene variant contributing to advanced atherosclerosis also associated with a 2-fold increased risk for SCD, which accounts nowadays for most of the mortality associated with coronary heart disease in the Western world.
We thank Mervi Alanne, Kirsi Auro, Pekka Ellonen, Minttu Jussila, Kaisa Silander, and Minna Suvela for their excellent technical assistance. Samuli Ripatti is thanked for his assistance in the data analysis. Seppo Tyynelä is acknowledged for morphometric measurements of the coronary and aortic artery samples and coronary casts.
Sources of Funding
This study has been supported by the Center of Excellence in Complex Disease Genetics of the Academy of Finland, the Jenny and Antti Wihuri Foundation, Aarne Koskelo Foundation, the Finnish Foundation for Cardiovascular Research, and GenomEUtwin supported by the European Commission under the program “Quality of Life and Management of the Living Resources” of 5th Framework Programme (no. QLG2-CT-2002-01254), and by the Emil Aaltonen Foundation, Medical Research Fund of Tampere University Hospital, the Pirkanmaa Regional Fund of the Finnish Cultural Foundation, and the Yrjö Jahnsson Foundation.
L.P. is a member of the board of directors of Orion Corporation.
Original received September 20, 2007; final version accepted January 31, 2008.
Pajukanta P, Lilja HE, Sinsheimer JS, Cantor RM, Lusis AJ, Gentile M, Duan XJ, Soro-Paavonen A, Naukkarinen J, Saarela J, Laakso M, Ehnholm C, Taskinen MR, Peltonen L. Familial combined hyperlipidemia is associated with upstream transcription factor 1 (USF1). Nat Genet. 2004; 36: 371–376.
Putt W, Palmen J, Nicaud V, Tregouet DA, Tahri-Daizadeh N, Flavell DM, Humphries SE, Talmud PJ. Variation in USF1 shows haplotype effects, gene: gene and gene: environment associations with glucose and lipid parameters in the European Atherosclerosis Research Study II. Hum Mol Genet. 2004; 13: 1587–1597.
Huertas-Vazquez A, Aguilar-Salinas C, Lusis AJ, Cantor RM, Canizales-Quinteros S, Lee JC, Mariana-Nunez L, Riba-Ramirez RM, Jokiaho A, Tusie-Luna T, Pajukanta P. Familial combined hyperlipidemia in Mexicans: association with upstream transcription factor 1 and linkage on chromosome 16q24.1. Arterioscler Thromb Vasc Biol. 2005; 25: 1985–1991.
Zeggini E, Damcott CM, Hanson RL, Karim MA, Rayner NW, Groves CJ, Baier LJ, Hale TC, Hattersley AT, Hitman GA, Hunt SE, Knowler WC, Mitchell BD, Ng MC, O’Connell JR, Pollin TI, Vaxillaire M, Walker M, Wang X, Whittaker P, Kunsun X, Jia W, Chan JC, Froguel P, Deloukas P, Shuldiner AR, Elbein SC, McCarthy MI. Variation within the gene encoding the upstream stimulatory factor 1 does not influence susceptibility to type 2 diabetes in samples from populations with replicated evidence of linkage to chromosome 1q. Diabetes. 2006; 55: 2541–2548.
van der Vleuten GM, Isaacs A, Hijmans A, van Duijn CM, Stalenhoef AF, de Graaf J. The involvement of upstream stimulatory factor 1 (USF1) in Dutch patients with familial combined hyperlipidemia. J Lipid Res. 2007; 48: 193–200.
Lee JC, Weissglas-Volkov D, Kyttala M, Sinsheimer JS, Jokiaho A, de Bruin TW, Lusis AJ, Brennan ML, van Greevenbroek MM, van der Kallen CJ, Hazen SL, Pajukanta P. USF1 contributes to high serum lipid levels in Dutch FCHL families and U.S. whites with coronary artery disease. Arterioscler Thromb Vasc Biol. 2007; 27: 2222–2227.
Reiner AP, Carlson CS, Jenny NS, Durda JP, Siscovick DS, Nickerson DA, Tracy RP. USF1 gene variants, cardiovascular risk, and mortality in european-americans. analysis of two U.S. cohort studies. Arterioscler Thromb Vasc Biol. 2007; 27: 2736–2742.
Komulainen K, Alanne M, Auro K, Kilpikari R, Pajukanta P, Saarela J, Ellonen P, Salminen K, Kulathinal S, Kuulasmaa K, Silander K, Salomaa V, Perola M, Peltonen L. Risk alleles of USF1 gene predict cardiovascular disease of women in two prospective studies. PLoS Genet. 2006; 2: e69.
Szalai AJ, Wu J, Lange EM, McCrory MA, Langefeld CD, Williams A, Zakharkin SO, George V, Allison DB, Cooper GS, Xie F, Fan Z, Edberg JC, Kimberly RP. Single-nucleotide polymorphisms in the C-reactive protein (CRP) gene promoter that affect transcription factor binding, alter transcriptional activity, and associate with differences in baseline serum CRP level. J Mol Med. 2005; 83: 440–447.
Providence KM, White LA, Tang J, Gonclaves J, Staiano-Coico L, Higgins PJ. Epithelial monolayer wounding stimulates binding of USF-1 to an E-box motif in the plasminogen activator inhibitor type 1 gene. J Cell Sci. 2002; 115: 3767–3777.
Ilveskoski E, Perola M, Lehtimaki T, Laippala P, Savolainen V, Pajarinen J, Penttila A, Lalu KH, Mannikko A, Liesto KK, Koivula T, Karhunen PJ. Age-dependent association of apolipoprotein E genotype with coronary and aortic atherosclerosis in middle-aged men: an autopsy study. Circulation. 1999; 100: 608–613.
Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005; 21: 263–265.
Salomaa VV, Lundberg V, Agnarsson U, Radisauskas R, Kirchhoff M, Wilhelmsen L. Fatalities from myocardial infarction in Nordic countries and Lithuania. The MONICA Investigators. Eur Heart J. 1997; 18: 91–98.
Bidder M, Shao JS, Charlton-Kachigian N, Loewy AP, Semenkovich CF, Towler DA. Osteopontin transcription in aortic vascular smooth muscle cells is controlled by glucose-regulated upstream stimulatory factor and activator protein-1 activities. J Biol Chem. 2002; 277: 44485–44496.
Renault MA, Jalvy S, Potier M, Belloc I, Genot E, Dekker LV, Desgranges C, Gadeau AP. UTP induces osteopontin expression through a coordinate action of NFkappaB, activator protein-1, and upstream stimulatory factor in arterial smooth muscle cells. J Biol Chem. 2005; 280: 2708–2713.
Giachelli CM, Speer MY, Li X, Rajachar RM, Yang H. Regulation of vascular calcification: roles of phosphate and osteopontin. Circ Res. 2005; 96: 717–722.
Naukkarinen J, Gentile M, Soro-Paavonen A, Saarela J, Koistinen HA, Pajukanta P, Taskinen MR, Peltonen L. USF1 and dyslipidemias: converging evidence for a functional intronic variant. Hum Mol Genet. 2005; 14: 2595–2605.
Service S, DeYoung J, Karayiorgou M, Roos JL, Pretorious H, Bedoya G, Ospina J, Ruiz-Linares A, Macedo A, Palha JA, Heutink P, Aulchenko Y, Oostra B, van Duijn C, Jarvelin MR, Varilo T, Peddle L, Rahman P, Piras G, Monne M, Murray S, Galver L, Peltonen L, Sabatti C, Collins A, Freimer N. Magnitude and distribution of linkage disequilibrium in population isolates and implications for genome-wide association studies. Nat Genet. 2006; 38: 556–560.
Norja S, Nuutila L, Karhunen PJ, Goebeler S. C-reactive protein in vulnerable coronary plaques. J Clin Pathol. 2007; 60: 545–548.