A New Susceptibility Locus for Myocardial Infarction, Hypertension, Type 2 Diabetes Mellitus, and Dyslipidemia on Chromosome 12q24

We examined the role of hepatic nuclear factor-1 alpha (HNF1a) gene polymorphism on coronary artery disease (CAD) traits in 4631 Saudi angiographed individuals (2419 CAD versus 2212 controls) using TaqMan assay on ABI Prism 7900HT sequence detection system. Following adjustment for confounders, the rs2259820_CC (1.19 (1.01–1.42); P = 0.041), rs2464196_TT (1.19 (1.00–1.40); P = 0.045), and rs2259816_T (1.13 (1.01–1.26); P = 0.031) were associated with MI. The rs2259820_T (1.14 (1.03–1.26); P = 0.011) and rs2464196_C (1.12 (1.02–1.24); P = 0.024) were associated with type 2 diabetes mellitus (T2DM), while the rs2393791_T (1.14 (1.01–1.28); P = 0.032), rs7310409_G (1.16 (1.03–1.30); P = 0.013), and rs2464196_AG+GG (1.25 (1.05–1.49); P = 0.012) were implicated in hypertension. Hypertriglyceridemia was linked to the rs2393791_T (1.14 (1.02–1.27); P = 0.018), rs7310409_G (1.12 (1.01–1.25); P = 0.031), rs1169310_G (1.15 (1.04–1.28); P = 0.010), and rs1169313_CT+TT (1.24 (1.06–1.45); P = 0.008) and high low density lipoprotein-cholesterol levels were associated with rs2259820_T (1.23 (1.07–1.41); P = 0.004), rs2464196_T (1.22 (1.06–1.39); P = 0.004), and rs2259816_T (1.18 (1.02–1.36); P = 0.023). A 7-mer haplotype CATATAC (χ 2 = 7.50; P = 0.0062), constructed from the studied SNPs, was associated with MI, and CATATA implicated in T2DM (χ 2 = 3.94; P = 0.047). Hypertriglyceridemia was linked to TGCGGG (χ 2 = 4.26; P = 0.039), and obesity to ACGGGT (χ 2 = 5.04; P = 0.025). Our results suggest that the HNF1a is a common susceptibility gene for MI, T2DM, hypertension, and dyslipidemia.


Background
The hepatocyte nuclear factor 1a gene (HNF1a, also known as transcription factor-1) on chromosome 12q24 encodes a transcription factor (TF) that binds to promoters of a variety of genes expressed predominantly in the liver [1,2]. Different sections of the HNF1a gene are transcribed to specific HNF1a isoforms, which might serve different yet unidentified functions [3][4][5][6]. This TF regulates several target genes involved in lipoprotein metabolism, including apolipoproteins, cholesterol synthesizing enzymes, and bile acid transporters, as well as glucose-stimulated insulin secretion [7,8] (Figure 1). Hence, alterations in the encoding gene are likely to lead to disorders in the various metabolic pathways associated with liver function. Currently, mutations in the HNF1a gene are known to be the most common cause of maturity onset diabetes of the young (MODY), a severe dominantly inherited form of nonketotic diabetes mellitus that is characterized by pancreatic beta-cell dysfunction [9][10][11][12][13][14][15]. This disease is a result of a primary defect in insulin secretion and usually develops at childhood, adolescence, or young adulthood. However, while MODY is predominantly inherited, its penetrance and expression may vary in presence or absence of a family history of diabetes [16]. Moreover, there appears to be a substantial heterogeneity in the etiology of the disease, even if the source of the disease may be the same [16][17][18]. Besides, HNF1a mutations have also been implicated in diabetic nephropathy related to type 1 diabetes [19] and noninsulin-dependent, type 2 diabetes mellitus (T2DM) [20][21][22][23], albeit not as pronounced as in MODY disorders [24][25][26]. Apart from diabetic disorders, single nucleotide polymorphisms (SNPs) in the HNF1a have also been linked to changes in plasma concentrations of the C-reactive protein (CRP) [27][28][29][30][31][32][33][34][35], arguably a powerful risk marker for cardiovascular disease. The gene has similarly been implicated in coronary artery disease (CAD)/myocardial infarction (MI) [30,36] as well as its risk traits, including dyslipidemic disorders [30,37], whereby an increased risk for CAD was also observed in T2DM patients harbouring HNF1a mutations [34]. However, these findings have either not been replicable in other studies or partly refuted by other investigators or meta-analysis studies [38][39][40][41]. Hence, the role of the HNF1A in the onset and development of CAD is yet to be clearly defined. Besides, the fact that HNF1A polymorphism has been linked with susceptibility to both CAD and its important risk traits points to some pleiotropic actions of the gene on disease pathways leading to atherosclerosis. In a preliminary linkage study in a heterozygous familial hypercholesterolemia setting, we established a link for early onset CAD to the locus of the HNF1a gene on chromosome 12q24 and suggested a role for this gene in this disease process. In the present study, therefore, we elected to test this notion further and to evaluate the potential role of the gene variants at this locus in CAD risk traits and its manifestation in a large cohort of angiographed Saudi individuals.

Association Experiments.
Once the SNPs of interest were identified, genotyping was achieved by TaqMan chemistry using the Applied Biosystems real-time Prism 7900HT Sequence Detection System (ABI Inc., CA, USA). Primers and the TaqMan fluorogenic probes bearing a suitable reporter dye on the 5 -end and a quencher dye on the 3 -end were designed using the Primer Express software V2.0 (ABI Inc., Foster City, CA, USA) and procured from Applied Biosystems (ABI, Warrington, UK). One probe (for allele 1) was labeled with VIC dye and the other (for allele 2) with FAM dye at the 5 -end, and serial dilutions were run to determine the optimal working concentration. For each reaction, a 25 L reaction was prepared by mixing 5 L containing 50 ng DNA, 12.5 L of 2x Universal mix (Eurogentec, Liege Science Park, Seraing, Belgium), 1.25 L of 20x probe assay mix, and 6.25 L DNase-free distilled water. Three no-template controls were included in each plate for normalization of emission signal. The thermal amplification profile for the first cycle occurred at 50 ∘ C for 2 min and 95 ∘ C for 10 min followed by 40 cycles of 94 ∘ C for 15 sec and 60 ∘ C for 30 sec. The plates were then scanned for FRET signal using the 7900HT sequence detection system and data analyzed using SDS 2.0 software (ABI, Foster City, CA, USA).  expressed as mean ± SEM. Associations with a two-tailed P value <0.05 were considered statistically significant.

Results
The results of whole genome scan experiments using the Affymetrix Gene Chip 250 Sty1 mapping array pointed to a number of genomic loci as potential risk for HFH and early onset of CAD in a study involving a family of 11 members harbouring HFH. These loci included that of the HNF1a gene on chromosome 12q24 (Figure 2). The ensuring sequencing of the gene in the family members and 200 other individuals from the general population revealed several informative SNPs (with frequencies of >0.1) that were of potential interest, from which seven were selected for the association studies in a population of 4631 candidates. These SNPs included (1) rs2393791 C>T, (2) rs7310409 A>G, (3) rs2259820 C>T (p.Leu459Leu), (4) rs2464196 T>C, (p.Ser487Asn), (5) rs2259816 G>T, (6) rs1169310 A>G, and (7) rs1169313 C>T numbered sequentially by their chromosomal positions ( Figure 3). We first performed the MWU test on the data, which demonstrated significant associations of the various SNPs with different disease traits (HNF1a Supplementary Data 2). We then subjected the data to binary logistical regression analysis for the respective conditions, whereby the coexisting disease traits and other covariates were treated as confounders, and Bonferroni adjustment for age and sex was performed on the data (HNF1a Supplementary Data 3). To begin with, the initial univariate analysis suggested that six of these variants rs7310409 A>G ( = 0.041), rs2259820 C>T ( = 0.011), rs2464196 C>T ( = 0.016), rs2259816 G>T (0.007), rs1169310 A>G ( = 0.016), and rs1169313 C>T ( = 0.023) conferred risk for MI (3044 cases versus 1587 controls). However, following adjustment for possible confounding effects of all other risk covariates, only the rs2259820 CC (odds ratio (95% confidence interval) = 1.19 (1.01-1.41); = 0.013), rs2464196 TT (1.19 (1.00-1.40); = 0.042), and rs2259816 T (1.13 (1.01-1.26); = 0.031) retained their significant association with the disease, while the relationships for the other three variants turned weaker or diminished (Table 2). Notably, the tests for the association of these variants with the classical CAD/MI risk traits also implicated the rs2259820 T (1.14 (1.03-1.26); = 0.011) and rs2464196 C = 0.023) conferring a risk for harbouring high low-density lipoprotein-cholesterol (hLDLC) levels. None of the studied SNPs was associated with gender or history of CAD.
We further evaluated the possible impact of the haplotypes at this locus on the various disease traits. The linkage disequilibrium plot for the studied SNPs is given in Figure 4. We used the most common 7-mer haplotype CATATAC (frequency = 0.505) constructed from the SNPs as a baseline for comparing their relationships with the disease traits (Table 3). Our results revealed that the baseline 7-mer haplotype CATATAC ( 2 = 7.50; = 0.0062) conferred a risk for MI (HNF1a Supplementary Data 5). Notably, its 6-mer (1-6) CATATA ( 2 = 7.48; = 0.0062) and 5-mer (1-5) CATAT ( 2 = 7.68; = 0.0058) derivatives were equally implicated in the disease. Further analysis pointed to the G>T change at this locus as explaining the difference between being causative and protective, as demonstrated by the fact that the 7-mer CATAGAC ( 2 = 6.06; = 0.014) was equally protective. Interestingly, the haplotype CATATAC ( 2 = 3.94; = 0.047) and its 4-mer derivative CATA was also implicated, albeit less significantly so in T2DM, pointing to MI and T2DM sharing common causative genomic sequences at this locus. No haplotype was positively associated with either hypercholesterolemia (hChol) or the harbouring low levels of high density lipoprotein-cholesterol (lHDLC). However, hypertriglyceridemia (hTG) was linked to TGCGGG ( 2 = 4.26; = 0.039) and its 5-mer TGCGG ( 2 = 4.61; = 0.032) derivative, while obesity was associated with

Discussion
The present study evaluated the relationships of gene variants on chromosomal 12q24 with CAD/MI and its various risk traits. We described the association for at least three of the studied SNPs, rs2259820, rs2464196, and rs2259816, and a weak link for two others, with the risk for MI. These variants reside in the chromosomal region that harbours the HNF1a gene, possibly pointing directly to this gene as the potential culprit. Currently, there appears to be stealth of information on the role of the HNF1a gene or its genomic locus in CAD/MI, in general. The available literature is somewhat conflicting, with some investigators implicating this locus in the disease [30,36] and others failing to establish such a relationship [38][39][40][41]. For example, although five of the variants included in the present study, rs2293791, rs7310409, rs2464196, rs2259816, and rs1169310, have been recently linked with changes in CRP [28,[32][33][34], a marker for cardiovascular disease, their direct involvement in CAD manifestation remains disputable. Thus, while some studies have implicated variants, such as the rs7310409 in acute coronary syndrome [40], rs2259816 in CAD [34], and rs2464196 in subclinical coronary atherosclerosis [30], others reported only a weak or no link at all for rs2259816 with CAD in the presence of a decrease in C-reactive protein levels [28]. The observation of associations for these variants with MI in the present study clearly indicates that further studies are warranted to ascertain globally the role of this genomic locus in atherosclerosis.
Our present data also implicated both the rs2259820 and rs2464196 in T2DM, reaffirming a role for this genomic region in diabetic disease pathways. Interestingly, the rs2464196 and two other variants linked to MI were also implicated in HTN, pointing to a possible common link for the three cardiovascular diseases. To our knowledge, this is the first report linking the rs2259820 to T2DM. This variant resides on exon 7 of the HNF1a gene, suggesting that this genic region is important for the manifestation of this disease. However, as in the case of CAD, a number of investigations addressing the role of the HNF1a gene polymorphism with respect to diabetic disease conditions have yielded somewhat inconsistent results. Thus, for example, among the variants included in the present study, the rs2259816 G>T has been associated with diabetic nephropathy in type 1 diabetes patients but not with T2DM [19] or other cardiovascular risk Disease Markers 7 traits including dyslipidemia, hypertension, or obesity [40]. On the other hand, a study by Qi et al. (2011) [46] suggested that the combined risk of acquiring CAD in diabetic patients increased significantly in the presence of the rs2259816 and other risk loci related to various genes. As discussed above, in the present study, this variant was linked to MI but only weakly associated with T2DM, adding to the inconsistency in the literature about the impact of the HNF1a polymorphism on disease manifestation, in general. As demonstrated in Figure 1, HNF1a regulates several genes involved in lipoprotein metabolism, such as apolipoproteins, cholesterol synthesizing enzymes, and bile acid transporters as well as glucose-stimulated insulin secretion [1,7]. Hence, we deemed it necessary to evaluate its potential influence on lipid profiles in our study population. Our data indicates that the variants exhibited even stronger relationships with hTG and the harbouring of hLDLC levels than with the diseases discussed above. The findings of an association with dyslipidemia are in agreement with previous studies implicating some of the HNF1a gene variants in dyslipidemia, T2DM [38], hTG, and high low-density lipoprotein-cholesterol (hLDLC) levels [37]. Specifically, the rs2258287 has been linked to acquiring hLDLC levels [37], and a study in the Oji-Cree population described an association of the G319S with "hypertriglycemic waist" contributing to greater risk of T2DM [22]. Besides, mice null for HNF1a have also shown altered plasma cholesterol levels [7]. Put together, our findings furnish unequivocal support to the notion that the chromosomal region encompassing the HNF1a gene is a susceptibility locus for MI, HTN, T2DM, and dyslipidemia, the important causes for atherosclerosis. This scenario also points to possible complex interactions between diabetes and dyslipidemia in atherosclerosis disease pathways. Similar interaction has been suggested recently between HNFIa gene, hyperglycemia, cardiovascular risk, and T2DM [39].  The table shows selected haplotypes associated with disease. The most frequent 7-mer haplotype CATATAC (0.505) was employed as the baseline to determine the relative effects of the other haplotypes. The studied SNPs are (1) rs2393791, (2) rs7310409, (3) rs2259820, (4) rs2464196, (5) rs2259816, (6) rs1169310, and (7) rs1169313 arranged sequentially by their chromosomal positions, and blocks represent the range of variants constituting the respective haplotypes. * < 0.01; * * < 0.005 by 2 test.
Even more important is the fact that, while the large majority of HNF1a variants associated with disease to date are coding, in the present study, it was primarily noncoding SNPs that were implicated. These include the rs2259816 located on intron 7 of the gene, which was associated with MI but was protective against T2DM. These observations seem to point to the likelihood that other entities related to the function of this locus might offer an explanation for these observations. Hence, we thought it worthwhile to explore further the probability of haplotyping this region as being more informative than individual SNPs in establishing the impact of the changes in the HNF1a gene locus on disease manifestation. Indeed, our study revealed several stretches of sequences that were associated with MI, T2DM, and changes in triglyceride and LDL-cholesterol levels. Notably, the difference between the causative 7-mer CATATAC and protective CATAGAC for MI was the genomic position of the rs2259816, which displayed the most significant association with the disease. Even more interesting was the observation of an association of the former haplotype with T2DM, demonstrating a link for the two diseases at haplotype level. Thus, MI and T2DM share not only common causative variants but also a common genomic region at the locus of the HNF1a gene. However, hTG was also associated with a number of other haplotypes that differed, however, from those implicated in MI and T2DM, suggestive of other independent entities as the possible underlying causes of these disorders. This could occur in various fashions, including the likelihood of the changes influencing gene expression as cis-acting regulators of nearby genes. For potential mechanisms involving SNPs in Disease Markers 9 the 3-untranslated region (3 -UTR), it can be speculated that such genic changes are likely to interfere with actions of gene regulatory elements, such as the microRNAs, thereby influencing mRNA maturation processes, for example. However, this notion needs to be verified further.
We conclude that the HNF1a locus constitutes a risk factor for MI, T2DM HTN, and dyslipidemia. Furthermore, since the majority of implicated variants were noncoding, we speculate that some other entities at this genomic locus, rather than the HNF1A gene per se, are likely to be the primary contributors to the disease pathways of atherosclerosis.