The Apolipoprotein E Polymorphism rs7412 Associates with Body Fatness Independently of Plasma Lipids in Middle Aged Men

Background The apolipoprotein E (APOE) gene is polymorphic, encoding one of 3 common alleles (ε2, ε3, ε4) produced from combinations of 2 non-synonymous SNPs (rs429358 and rs7412). APOE plays an important role controlling plasma lipids but its association with adipocyte functionality and body fatness remains to be determined. Methods We analyzed fasting plasma lipids and genotyped the two main APOE-SNPs (rs429358 and rs7412), both located in the fourth exon of the APOE, in 4660 Caucasian middle-aged men free of cardiovascular disease. Results The rs7412 SNP, which determines the APOE2 isoform, was significantly associated with Body Mass Index (BMI) and Waist Girth (WG) in a multivariate model accounting for age, smoking status and plasma lipids. BMI and WG were highest in TT homozygotes and lowest in CC homozygotes. This effect was independent of the rs429358 SNP, which failed to show any association with the BMI and WG variables. The odds ratio of being obese (BMI>30) for individuals carrying the APOε2 allele, present in 14% of the cohort and defined by the rs7412 SNP, was also significant in this multivariate model, with an OR of 1.27 (95% CI: 1.01–1.59). Conclusions This study provides an evidence of a lipid-independent association between the APOE SNP rs7412 and body fatness surrogates, BMI and WG, in a large cohort of middle-aged males.


Introduction
Obesity is a multifaceted disorder that increases the risk of developing cardiovascular disease and diabetes and is closely linked to disturbances in lipid and glucose metabolism [1]. Apolipoprotein E (APOE) binds to lipoprotein particles and mediates their catabolism, playing a crucial role in lipid regulation. The rs429358 and rs7412 single nucleotide polymorphisms (SNP) give rise to the e2, e3, and e4 haplotypes of the human APOE gene.
Although mainly produced by the liver, APOE is also synthesized and secreted by adipose tissues [2]. Despite the pivotal role of APOE on lipid metabolism and adipocyte activity [3][4][5], genome-wide association studies (GWAS) have failed to detect an association of APOE genotype with human adiposity. (Reviewed in [6]). This is partially explained by i) the heterogeneity of the studied populations, ii) the low presence of subjects with the less common e2 and e4 haplotypes, and iii) the confounding impact that plasma lipids have on obesity: body fatness is positively related to circulating cholesterol and triglycerides [7] and, in turn, circulating cholesterol and triglycerides are modulated in a APOE isoform-dependent manner [8]. To overcome these caveats, we studied a large cohort of 4,660 Spanish middle-aged men free of Cardiovascular Disease to investigate whether APOE-SNPs, rs429358 and rs7412, are associated with body fatness measures, body mass index and waist girth, over and above the effect on plasma lipids.

Study participants
The Aragon Workers Health Study (AWHS) is a longitudinal study of the workers of the General Motors Spain automobile factory in Figueruelas (Zaragoza, Spain) [9]. Workers were excluded from the cohort if they have clinically overt CVD. The study was approved by the local Institutional Review Board (Comité É tico de Investigación Clínica de Aragón, CEICA) and participants provided written informed consent.

Anthropometry and plasma biochemistry
Height and weight were determined using a standard steel strip stadiometer (SECA-440. SECA, Hamburg, Germany) and a digital electronic scale (SECA-778. SECA), respectively. Waist circumference was measured between iliac crest and the lower rib margin with flexible anthropometric tape (GulicK II -67019. Country Technologies, Gays Mills, WI. USA). Fasting serum triglycerides, total cholesterol and HDL-cholesterol were assayed by spectrophotometry (Chemical Analyzer ILAB 650, Instrumentation Laboratory).
Genotyping DNA from buffy coat was isolated using the FlexiGene DNA AGF3000 kit (Qiagen, Valencia, CA, USA) on an AutoGenFlex 3000 workstation (Autogen, Holliston, MA. USA) and genotyping was carried out in the Genetics Unit-Parque Cientifico de Madrid (Madrid. Spain). Briefly, samples were spotted onto 384 plates using a Beckman BioMek 2000 automated liquid handler (Beckman High Wycombe, UK) and diluted in a mix consisting in TaqMan Genotyping MasterMix (Applied Biosystems, Foster City, California)) and a mixture of pre-made TaqMan SNP genotyping assays; C_3084793_20 (rs429358) and C_904973_10 (rs7412) (Applied Biosystems). qPCR reactions were made in a HT7900 Fast Real-Time PCR System (Applied Biosystems). SDS 2.4 software (Applied Biosystems) was used for genotype calling.
Detailed description of the AWHS cohort has been previously provided [9]. Information regarding plasma lipid biochemistry and smoking status of the participants is detailed in Table 1. In a crude regression model, higher BMI and WG were inversely associated with smoking and plasma HDL and directly correlated with circulating cholesterol, TG, being an ex-smoker and age (p, 10 25 for all variables, in both fatness measures). A significant association was also observed between both the APOE polymorphism rs7412 as well as the rs429358 with circulating plasma cholesterol and triglycerides ( Table 2).
We next explored the contribution of each SNP to body fatness beyond their effects on plasma lipids. Multiple regression analyses considering SNP rs7412 or rs429358 as independent variables and either BMI or WG as dependent variables were used. After adjustment for variables that affect body weight (i.e. smoking status, plasma lipids, and age) the rs7412 genotype was significantly associated with BMI (P = 0.018) as well as WG (P = 0.048). Thus, individuals carrying the T/T genotype displayed both higher BMI and larger WG followed by C/T and C/C (Table 3). On the contrary, the same regression model did not reveal significant associations with either BMI or WG and the rs429358 SNP. We additionally tested for an interaction (epistasis) between rs7412 and rs429358 in our regression models and found P-values for the interaction term of 0.331 for BMI and 0.229 for WG. This precluded a possible epistasis between those SNPs and suggested that rs7412 acts independently on body fatness.
Multiple regression analyses involving classic APOE genotypes as independent variables and either BMI or WG as dependent variables revealed a significant effect of haplogenotypes on BMI after adjustment for age, lipids and smoking (P = 0.017) (Figure 1). A posteriori contrasts found a significant difference between e2e2 and non-e2 carriers (e3e3, e3e4 and e4e4, P = 0.038, P = 0.018 and P = 0.024, respectively), with the e2e2 genotype associated with higher BMI. It should be noted that the e2e2 genotype includes the T/T genotype for rs7412, which was associated to higher BMI values, as previously stated. Using this model, the odds ratio of being obese (BMI.30) for individuals bearing a e2 allele (13.4% of the cohort) was 1.27 (1.01-1.59, p = 0.035), when compared with those bearing no copies of this allele.

Discussion
The human APOE locus (19q13.2) encodes a polymorphic protein of 299 amino acids with three haplotypes; e2, e3, and e4, produced from combinations of two rs429358 and rs7412 coding non-synonymous SNPs. These haplotypes give rise to three homozygous (e2e2, e3e3, e4e4) and three heterozygous genotypes (e2e3, e2e4, e3e4). Several studies have supported linkage to obesity phenotypes at chromosome 19q13, where APOE gene is located. Thus, as early as 2002, a genome scan based on German families with obese children described an association between a region very close to 19q13 and obesity in children and adolescents [10]. Subsequent imprinted linkage analyses confirmed this suggestive connection between BMI and the 19q13 locus [11,12].
In agreement with our results, APOE e2 carriers (ie, carriers of the minor allele of rs7412) were described to have higher plasma APOE concentration as compared with subjects with e3 and e4 genotype, being plasma APOE concentration directly correlated with BMI [13]. Likewise, the ARIC Study reported that the APOE isoforms are associated with increasing BMI in the order: e2.e3. e4 [14]. However, this relation is controversial with studies showing associations of the e2 allele with increased BMI [14][15][16][17] or the e4 allele with decreased BMI measures [18,19], while others showed the e4 genotype to be associated with obesity [20,21]. A recent study showed no association between APOE polymorphisms and body fat mass in obese children [22], but another study described lower BMI in 8-year-old children carrying APOE4 [23]. We hypothesize that these seemly conflicting results could be explained by the confounding effects of plasma lipids when the relationship between APOE and adiposity is investigated. In this vein, a recent study failed to show any significant effect of the interaction between APOE genotype and BMI on blood lipid levels [24]. However, the predictive power of the regression model for LDL-C improved when gene-BMI interaction and gene-BMI interaction plus gene-nutrient interaction were added [24]. The authors reasoned that sample size of the study was relatively small (,1000 individuals), perhaps affecting its ability and power to detect the effect of such interactions on lipid levels.  Associations between APOE isoforms and circulating lipid concentrations were described more than 30 years ago [25], and is also well known that these lipids are, in turn, associated with obesity [7]. Once lipid effects were normalized across genotypes we showed that the APOE-SNP rs7412 had a lipid-independent effect on body fatness. One may argue that controlling for lipids, which might be on a causal pathway between exposure (APOE-SNPs) and outcome (body fatness) could lead to an over-adjusted model. However, this would be the case if the only causal link between APOE and fatness was via plasma lipid modulation. In this regard, a number of biological pathways have been described in previous reports by us and others that suggest a lipidindependent role of APOE in the modulation of adipocyte function. These APOE isoform-dependent effects have been demonstrated in vivo as well as in vitro models [4,5,20,26]. It can also be argued that BMI might not completely capture obesity status and WG is a more adequate measure of abdominal fat and total fat. In this sense, the fact that APOE-effect acts not only upon BMI but WG also lends support to the existence of APOEdependent lipid-independent mechanisms regulating adipose tissue mass and thus obesity.
Our results also support the notion that, while circulating cholesterol and TGs are determined by both APOE SNPs acting synergistically, the rs7412 SNP affects BMI and WG independently of rs429358. This is of particular interest in regards to the 12.4% of the AWHS cohort individuals in whom a single SNP determination (i.e. the presence of a single T allele in the SNP rs7412) can inform about their obesity risk.
Our study has some inherent limitations. Due to its crosssectional design, we cannot draw conclusions about causal relations between APOE-SNPs and fatness measurements. We do know, however, that the relation is independent of APOE effects on plasma lipids and relies on the rs7412 SNP. Additionally, since we focused on middle-aged males we cannot generalize our results to all adult populations. The strength of this study is that the large and homogeneous sample size of the AWHS Study allows a valuable opportunity for in-depth analysis of APOE allele effects in Caucasian middle-aged men free of CVD. This demographic is specially interesting as it represents a growing population of individuals who have adopted the food and lifestyle patterns of developed countries and are at high risk of developing metabolic disease. Our findings warrant further investigation in women, as well as in more heterogeneous populations, and underscore the importance of APOE in human metabolism beyond its role on lipoprotein regulation.

Conclusions
Polymorphisms rs429358 and rs7412 in the human APOE gene are strongly associated with the variation of cholesterol and triglyceride levels in plasma. Further, once lipid variation is normalized across genotypes, the SNP rs7412 is also associated with body weight pointing towards the existence of a lipidindependent effect of APOE on body fatness.