Apolipoprotein E (APOE) is a class of apolipoprotein found in the chylomicron and Intermediate-density lipoprotein (IDLs) that is essential for the normal catabolism of triglyceride-rich lipoprotein constituents.[4] In peripheral tissues, APOE is primarily produced by the liver and macrophages, and mediates cholesterol metabolism in an isoform-dependent manner. In the central nervous system, APOE is mainly produced by astrocytes, and transports cholesterol to neurons via APOE receptors, which are members of the low density lipoprotein receptor gene family.[5] APOE is the principal cholesterol carrier in the brain.[6] This protein is involved in Alzheimer’s disease and cardiovascular disease.[7]


The gene, APOE, is mapped to chromosome 19 in a cluster with Apolipoprotein C1 and the Apolipoprotein C2. The APOE gene consists of four exons and three introns, totaling 3597 base pairs. APOE is transcriptionally activated by the liver X receptor (an important regulator of cholesterol, fatty acid, and glucose homeostasis) and peroxisome proliferator-activated receptor γ, nuclear receptors that form heterodimers with Retinoid X receptors.[8] In melanocytic cells APOE gene expression may be regulated by MITF.[9]
APOE is 299 amino acids long and contains multiple amphipathic α-helices. According to crystallography studies, a hinge region connects the N- and C-terminal regions of the protein. The N-terminal region (residues 1–167) forms an anti-parallel four-helix bundle such that the non-polar sides face inside the protein. Meanwhile, the C-terminal domain (residues 206-299) contains three α-helices which form a large exposed hydrophobic surface and interact with those in the N-terminal helix bundle domain through hydrogen bonds and salt-bridges. The C-terminal region also contains a low density lipoprotein receptor (LDLR)-binding site.[10]


APOE is polymorphic,[11][12] with three major alleles: ApoE2 (cys112, cys158), ApoE3 (cys112, arg158), and ApoE4 (arg112, arg158).[7][13][14] Although these allelic forms differ from each other by only one or two amino acids at positions 112 and 158,[15][16][17] these differences alter APOE structure and function. These have physiological consequences:
E2 (rs7412-T, rs429358-T) has an allele frequency of approximately 7 percent.[18] This variant of the apoprotein binds poorly to cell surface receptors while E3 and E4 bind well.[19] E2 is associated with both increased and decreased risk for atherosclerosis. Individuals with an E2/E2 combination may clear dietary fat slowly and be at greater risk for early vascular disease and the genetic disorder type III hyperlipoproteinemia—94.4% of such patients are E2/E2, while only ∼2% of E2/E2 develop the disease, so other environmental and genetic factors are likely to be involved (such as cholesterol in the diet and age).[20][21][22] E2 has also been implicated in Parkinson’s disease,[23] but this finding was not replicated in a larger population association study.[24]
E3 (rs7412-C, rs429358-T) has an allele frequency of approximately 79 percent.[18] It is considered the “neutral” Apo E genotype.
E4 (rs7412-C, rs429358-C) has an allele frequency of approximately 14 percent.[18] E4 has been implicated in atherosclerosis,[25] Alzheimer’s disease,[26][27] impaired cognitive function,[28][29] reduced hippocampal volume,[29] HIV,[30] faster disease progression in multiple sclerosis,[31][32] unfavorable outcome after traumatic brain injury,[33] ischemic cerebrovascular disease,[34] sleep apnea,[35][36] accelerated telomere shortening [37] and reduced neurite outgrowth.[38] A notable advantage of the E4 allele (relative to E2 and E3) is a positive association with higher levels of vitamin D, which may help explain its prevalence despite its seeming complicity in various diseases or disorders.

By studying the loci that contribute to human longevity, we aim to identify mechanisms that contribute to healthy aging. To identify such loci, we performed a genome-wide association study (GWAS) comparing 403 unrelated nonagenarians from long-living families included in the Leiden Longevity Study (LLS) and 1670 younger population controls. The strongest candidate SNPs from this GWAS have been analyzed in a meta-analysis of nonagenarian cases from the Rotterdam Study, Leiden 85-plus study, and Danish 1905 cohort. Only one of the 62 prioritized SNPs from the GWAS analysis (P < 1 × 10−4) showed genome-wide significance with survival into old age in the meta-analysis of 4149 nonagenarian cases and 7582 younger controls [OR = 0.71 (95% CI 0.65–0.77), P = 3.39 × 10−17]. This SNP, rs2075650, is located in TOMM40at chromosome 19q13.32 close to the apolipoprotein E (APOE) gene. Although there was only moderate linkage disequilibrium between rs2075650 and the ApoE ε4 defining SNP rs429358, we could not find an APOE-independent effect of rs2075650 on longevity, either in cross-sectional or in longitudinal analyses. As expected, rs429358 associated with metabolic phenotypes in the offspring of the nonagenarian cases from the LLS and their partners. In addition, we observed a novel association between this locus and serum levels of IGF-1 in women (P = 0.005). In conclusion, the major locus determining familial longevity up to high age as detected by GWAS was marked by rs2075650, which tags the deleterious effects of the ApoE ε4 allele. No other major longevity locus was found.


Worldwide human populations have shown an increase in mean life expectancy in the past two centuries (Oeppen & Vaupel, 2002). This is mainly because of environmental factors such as improved hygiene, nutrition, and health care. The large variation in healthy lifespan among the elderly has prompted research into the determinants of aging and lifespan regulation. The genetic contribution to human lifespan variation was estimated at 25–30% in twin studies (Gudmundsson et al., 2000; Skytthe et al., 2003; Hjelmborg et al., 2006). The most prominent genetic influence is observed in families in which the capacity to attain a long lifespan clusters (Perls et al., 2000; Schoenmaker et al., 2006). Exceptional longevity can be reached with a low degree of age-related disability (Christensen et al., 2008; Terry et al., 2008), raising the question whether protective mechanisms against disease exist in long-lived subjects.

In most experimentally modified animal model systems, single-gene mutations in many different genes have major life extension effects (Fontana et al., 2010; Kenyon, 2010). However, natural human and animal longevity is presumed to be a complex trait (Finch & Tanzi, 1997). In humans, both candidate gene and genome-wide genetic association approaches have been applied in an attempt to identify longevity loci. The frequency of genetic variants has been typically compared between nonagenarian cases and young controls, revealing loci at which genetic variants may contribute to a higher or lower probability of survival into old age. The initial candidate gene studies aimed at finding human longevity genes were dominated by contradictory results (Christensen et al., 2006). The more consistent evidence obtained by repeated observation in independent cohort studies for association with longevity has so far only been observed for three loci, the apolipoprotein E (APOE) locus (Schachter et al., 1994; Christensen et al., 2006), the FOXO3A locus (Willcox et al., 2008; Flachsbart et al., 2009; Pawlikowska et al., 2009; Soerensen et al., 2010), and the AKT1 locus (Pawlikowska et al., 2009). Thus, despite the expectation that longevity would be influenced by many genetic variants with small effect sizes, the effect of variants has consistently been shown in only three genes.

Hypothesis-free genome-wide approaches have also been undertaken. Genome-wide linkage scans reported evidence for linkage with longevity on chromosome 4q25 (Puca et al., 2001), 3p24-22, 9q31-34, and 12q24 (Boyden & Kunkel, 2010). However, the evidence for these loci is still very weak as the results, obtained in centenarians and their families, could not be replicated in nonagenarian sibling pairs (Beekman et al., 2006) or have yet to be tested in other studies. A meta GWAS of survival to 90 years or older in 1836 cases and 1955 controls did not find any significant genome-wide associations (Newman et al., 2010). Thus far, hypothesis-free approaches have not identified any loci involved in longevity.

In a few studies, such as the Ashkenazi Jewish Centenarian Study and the Leiden Longevity Study (LLS), different generations of long-lived families are being investigated for parameters and pathways contributing to the longevity phenotype (Atzmon et al., 2004; Schoenmaker et al., 2006). The survival benefit of the LLS families is marked by a 30% decreased mortality risk in the survival analysis of three generations, i.e., the parents of the probands in this study (nonagenarian sibling pairs), their unselected additional siblings, and their offspring (Schoenmaker et al., 2006). As compared to their partners, the offspring of nonagenarians siblings have a lower prevalence of type 2 diabetes, myocardial infarction and hypertension (Westendorp et al., 2009), a beneficial glucose, lipid, and thyroid metabolism, and a preservation of insulin sensitivity with age (Rozing et al., 2009, 2010a,b; Vaarhorst et al., 2011; Wijsman et al., 2011). Hence, in middle age, these families display beneficial metabolic profiles.

Because the longevity phenotype is inherited in the LLS families, they offer a route to identify genetic variants that influence human longevity. Previously, we tested whether the absence of GWAS-identified alleles promoting common diseases might explain their familial longevity (Beekman et al., 2010). Longevity was not easily explained by the absence of disease-susceptibility alleles. More likely therefore, the genome of the long-lived harbors longevity-promoting alleles. To identify such loci, we performed a GWAS comparing nonagenarian siblings from the LLS and younger population controls. We subsequently investigated emerging candidate SNPs in nonagenarian cases from the Rotterdam Study, the Leiden 85-plus study, and the Danish 1905 cohort.



A GWAS was performed in nonagenarian participants from the LLS and middle-aged controls from the Rotterdam Study (RS). Genotype data for 516,721 SNPs that passed quality control thresholds were analyzed in a comparison of 403 unrelated nonagenarians (94 years on average) and 1670 controls (58 years on average). A flow chart of the consecutive analysis steps is depicted in Fig. 1, and a description of the population samples investigated in the GWAS and subsequent replication studies is given in Table 1. Results of the association analysis of stage 1 are depicted in Fig. S1. None of the SNPs reached genome-wide significance (P < 5 × 10−8).

Table 1

Characteristics of the genotyped samples used for analysis
Fig. 1

Flow chart of experimental work.

Replication studies

We prioritized the SNPs that had the most significant association with survival into old age according to the analysis of stage 1 (P < 1 × 10−4, Table S1). For 58 of the 62 selected SNPs, successful genotyping was obtained in the replication cohorts. In stage 2, these 58 SNPs were tested for association comparing 960 RS replication cases (mean age of 93 years), 1208 Leiden 85-plus replication cases (mean age of 92 years), and 1578 Danish replication cases (mean age of 93 years) with appropriate middle-aged population controls (Table 1). Meta-analysis for the 58 SNPs, comprising a total of 4149 nonagenarian cases and 7582 younger controls (from the LLS GWAS, RS replication, Leiden 85-plus replication, and Danish replication studies), was performed.

Rs2075650 on chromosome 19 was the only SNP that was associated with survival into old age at the genome-wide significance level (P = 3.39 × 10−17) (Table S2A). The minor allele was underrepresented among the older cases as compared to middle-aged controls, hence associated with the decreased probability of carriers surviving into old age corresponding to an odds ratio (OR) below unity [OR = 0.71 (95% CI 0.65-0.77)]. This effect is observed in both sexes (Table S2B, C). The remaining 57 SNPs did not show genome-wide significant effects on longevity either in men or women (Table S2B for men and S2C for women). The association of rs2075650 with survival did show some heterogeneity across the four studies (P = 0.0495), which is mainly because of the RS.

rs2075650 and the APOE ε2/ε3/ε4 polymorphism

Rs2075650 is located in the TOMM40 gene, next to the APOE gene (Fig. S2). APOE was previously associated with longevity (Schachter et al., 1994; Christensen et al., 2006). The ApoE protein has three isoforms (ApoE ε2, ApoE ε3, and ApoE ε4) which are defined by two SNPs, rs7412 (Arg136Cys; ε2) and rs429358 (Cys112Arg; ε4). A meta-analysis of rs7412 and rs429358, in the LLS GWAS study, the Leiden 85-plus replication study, and the Danish replication study samples (3189 cases and 5757 controls), showed a significant association of rs429358 with longevity [OR = 0.62 (95% CI 0.56–0.68), P = 1.33 × 10−23], which was comparable to rs2075650 [OR = 0.67 (95% CI 0.61–0.74), P = 9.15 × 10−17]. Rs7412 also showed an association with longevity, with a higher prevalence of the minor allele in nonagenarians [OR = 1.31 (95% CI 1.17–1.46), P = 1.35 × 10−6].

Genetic variants in PVRL2-TOMM40-APOE region are associated with human longevity in a Han Chinese population.


Human longevity results from a number of factors, including genetic background, favorable environmental, social factors and chance. In this study, we aimed to elucidate the association of human longevity with genetic variations in several major candidate genes in a Han Chinese population.


A case-control association study of 1015 long-lived individuals (aged 90 years or older) and 1725 younger controls (30-70 years old) was undertaken. Rs2075650 in TOMM40 was firstly genotyped using the ABI SNaPshot method in an initial cohort consisted of 597 unrelated long-lived individuals and 1275 younger controls enrolled from Sichuan. Secondly, eighteen tag single-nucleotide polymorphisms (SNPs) in the PVRL2-TOMM40-APOE locus were genotyped for extensive study in the same cohort. Finally, 5 associated SNPs were genotyped in a replication cohort including 418 older individuals and 450 younger controls. The genotype and allele frequencies were evaluated using the χ2 tests. The linkage disequilibrium (LD) block structure was examined using the program Haploview.


The case-control study of rs2075650 in TOMM40 showed significant difference in allele frequencies between cases and controls (P = 0.006) in an initial study. Of the 18 SNPs genotyped, rs405509 in APOE and another three SNPs (rs12978931, rs519825 and rs395908) in the PVRL2 gene also showed significant association with human longevity in extensive study in the same cohort. Rs2075650 in TOMM40, rs405509 in APOE and rs519825 in PVRL2 showed a significant association with human longevity in a replication cohort.


These results suggested that PVRL2, TOMM40 and APOE might be associated with human longevity. However, further research is needed to identify the causal variants and determine which of these genes are involved in the progress of human longevity.


Association between apolipoprotein E polymorphism and Alzheimer disease in Tehran, Iran.


Epsilon 4 allele of apolipoprotein E (APOE-epsilon4) is a major risk factor for Alzheimer’s disease (AD). The association of APOE allele frequencies with AD remains unknown in developing countries. We examined the frequency of APOE alleles in 105 patients with AD and 129 cognitively normal subjects of similar age and sex (control group), in Tehran, Iran.

The APOE-epsilon4 allele frequency was significantly higher in the AD subjects than in the control group (21% versus 6.2%, p < 0.001). In addition, the OR for APOE-epsilon4 heterozygous and homozygous subjects were 3.2 (p = 0.001) and 12.75 (p = 0.01), respectively. The OR was not uniform across age groups.

The AD subjects carrying one or two APOE-epsilon4 allele showed earlier age-at-onset (p < 0.001).

These data suggest that the APOE-epsilon4 allele increase the risk for AD in Tehran population in a dose and age-dependent manner.

Although the APOE-epsilon2 allele frequency was lower in the AD subjects than in the control group (0.95% versus 2.7%, p = 0.15), APOE-epsilon2 was not associated with the onset of AD in Tehran’s population. The OR for epsilon2 allele in AD subjects was 0.34 (p = 0.21). The genotype frequencies for epsilon3, epsilon4, and epsilon2 alleles in control subjects were 91.2, 6.1, and 2.7%, respectively. These values were similar to that reported for Turkish, Greece, Japanese, Spanish, and Moroccan populations, but they were significantly different from the reported values for the other ethnic populations.

This observation emphasizes the importance of geographical location and ethnical background of the subjects in the study of APOE genotypes and their association with AD.


Join 25,000 people in helping redefine health with health concierge, healthy aging and precision medicine.