Common Genomic Variation in the APOC3 Promoter Associated With Variation in Plasma Lipoproteins
Abstract We hypothesized that common genomic variation that affected the expression and/or function of the products of the APOC3, APOE, FABP2, and PON1 genes would be associated with variation in biochemical phenotypes in a previously unstudied human sample. We determined genotypes of functional genomic variants of APOC3, APOE, FABP2, and PON1 in 509 adult aboriginal Canadians from an isolated community in Northern Ontario. We tested for genotype associations with plasma lipoprotein traits. We found that (1) common variation at nucleotide −455 of the APOC3 promoter was associated with variation in plasma triglycerides (P=.006) and (2) common variation of APOE determining plasma isoforms of apo E was associated with variation in plasma apo B (P=.009). Analysis of subjects classed by APOC3 markers showed that homozygosity for presence of a C at nucleotide −455 and a T at nucleotide −482 was associated with significantly increased plasma triglycerides in both men and women. Furthermore, this allele was approximately twice as frequent in subjects within the highest quartile of plasma triglycerides as in subjects within the lowest quartile. Since the DNA variation detected by the APOC3 markers affects in vitro expression of the gene product, it is possible that the marker itself caused the associations. However, the associations could also have resulted from linkage disequilibrium with other functional variants in APOC3 or the closely linked APOA1 and/or APOA4 genes.
Presented in part at the 69th Scientific Sessions of the American Heart Association, New Orleans, La, November 10-13, 1996.
- Received December 18, 1996.
- Accepted April 30, 1997.
There have been numerous reported associations between atherosclerosis-related phenotypes and DNA markers of candidate genes that are involved in various intermediary pathways, such as lipoprotein metabolism.1 2 However, alleles of very few genes appear to be consistently related to intermediate phenotypes across diverse populations.2 Part of the inconsistency may relate to the fact that most DNA markers studied so far have not been demonstrated to have a functional impact on the structure or expression of the gene product. Thus, most associations have been attributed to linkage disequilibrium, with putative functional changes elsewhere at the genetic locus. Since this finding may vary between populations, factors such as admixture can result in false conclusions about genetic associations. One strategy to reduce such confounding in association studies may be to select DNA markers that are proven, by various assays, to directly mark a functional change in the gene of interest.
DNA markers of APOC3 have been consistently associated with variation in plasma TG in diverse study samples.3 4 5 Experiments in transgenic mice indicated that overexpression of human apo CIII caused profound hypertriglyceridemia.6 Common genomic variants of the promoter sequence of the APOC3 gene were found to be in linkage disequilibrium with each other and with the APOC3 Sst I marker that was most commonly used in genetic association studies over the last decade.7 Two APOC3 promoter sequence variants were in the vicinity of an insulin-response element.7 Recent expression studies showed that one common APOC3 promoter DNA sequence, namely a T at nt −455, was associated with a 40% to 50% insulin-mediated downregulation of APOC3 gene expression, albeit over a 108-fold range of insulin concentrations.8 In contrast, the other common DNA sequence, namely a C at nt −455, was associated with the loss of insulin-mediated downregulation of APOC3 gene expression.8 A DNA marker for the APOC3 C-containing insulin-response element may therefore mark a functional DNA change that could be more biologically relevant than the commonly studied Sst I polymorphism.
Another example of a marker that is likely to directly cause a genetic association is the APOE restriction isotype underlying the apo E protein polymorphism.9 10 The E4 allele of the APOE is associated with higher mean plasma concentrations of LDL cholesterol and apo B, while the E2 allele is associated with lower mean plasma concentrations of these analytes, than the E3 allele in almost all human populations.9 10 These associations almost certainly result from structural changes in apo E that affect lipoprotein metabolism, and the genotyping method directly assays the molecular basis of the association. This genomic variation underlies structural variation in apo E, which is felt to underlie the consistent association with metabolic phenotypes.1 2 9 10 Another example is the DNA marker system that detects the T54 allele of FABP2, which has been associated with diabetes-related phenotypes in Pima Indians11 and Mexican-Americans.12 This genomic variation underlies variation of in vitro protein function, which is felt to underlie the association with the metabolic phenotypes.11 Finally, a DNA marker system detects the R192 allele of PON1, which has been associated with increased paraoxonase activity,13 an atherogenic lipoprotein profile,14 and increased risk of coronary heart disease.15 16 This PON1 genomic variation is possibly the basis for the phenotype-genotype associations.
We wished to determine whether the above genomic variants in APOC3, APOE, FABP2, and PON1 would be associated with variation in plasma lipoproteins in a native Canadian population. We specifically chose the DNA markers on the basis of the evidence that each resulted in a functional change and was thus not simply a marker. We were particularly interested to determine the effectiveness of this marker selection rationale to detect associations in a human sample that had never been previously studied.
The community of Sandy Lake, Ontario, is located 2000 km northwest of Toronto, in the subarctic boreal forest of central Canada. This isolated community is accessible only by air during most of the year. The ancestors of the contemporary residents of this region lived a nomadic, hunting-gathering subsistence typical of other Algonkian-speaking peoples of the northeastern subarctic. Since the development of the reservation and residential school systems, the lifestyle has changed radically from physically active to sedentary. The primary food source has changed from wildlife with roots and berries to processed high fat foods.
Seven hundred twenty-eight members of this community aged 10 years and above were enrolled, representing a 72% community participation rate. Female participants were not screened during pregnancy. Subjects who were pregnant at the time of recruitment had all determinations deferred until 3 months postpartum. Each measurement was performed twice, and the average was used in the analysis. Assessments included a questionnaire to assess medical history. Body mass index (BMI) was defined as weight/height2 (kg/m2).
Plasma samples were obtained with informed consent. For the current study, exclusion criteria included age below 18 years and an inadequate blood sample available for all biochemical and/or genetic determinations. The project was approved by the University of Toronto Ethics Review Committee.
Biochemical and Genetic Analyses
Blood for lipoprotein analyses was centrifuged at 2000 rpm for 30 minutes and the plasma was stored at −70°C. Plasma concentrations of lipids, lipoproteins, and apolipoproteins were determined as described.14 17 18 Established procedures were used to extract leukocyte DNA and to determine genotypes of FABP2 codon 54,9 PON1 codon 192,13 and APOE exon 4.19
New gene amplification reactions were devised to perform genotypes of the APOC3 promoter sequence. The primers C3-pro-5′ (5′-GTGAGAGCTCAGCCCCTGTAA-3′) and C3-pro-3′ (5′-TTTCACACTGGAAATTTCAGG-3′) were used in a gene amplification program with annealing temperature 60°C to amplify a 194-bp fragment of the APOC3 promoter that contained the insulin-response element. Digestion of a fragment that derived from the sequence containing C at −482 and T at −455 with endonuclease Msp I resulted in three fragments with sizes 148, 30, and 16 bp. Digestion of a fragment that derived from this sequence with endonuclease Fok I resulted in two fragments with sizes 122 and 72 bp. Digestion of a fragment that derived from an allele with the substitution of T for C at position −482 nt with endonuclease Msp I resulted in two fragments with sizes 164 and 30 bp. Digestion of a fragment that derived from an allele with the substitution of C for T at position −455 nt with endonuclease Fok I resulted in a single fragment with size 194 bp. All genotyping reactions were run with known controls.
The significance of deviations of observed genotype frequencies from those predicted by the Hardy-Weinberg equation were evaluated with χ2 tests. Linkage disequilibrium between the two APOC3 promoter markers was estimated using Hill and Robertson’s20 correlation coefficient of gene frequencies. SAS (Version 6.08) was used for all statistical comparisons involving biochemical variables.21 Quantitative variables were transformed and subjected to analysis of normality as described.18 All quantitative traits had significantly nonnormal distributions in this sample. Transformation using the log for each variable resulted in a distribution that was not significantly different from normal in the case of all variables but TG, in which case no transformation produced a distribution that was more normal than log transformation (data not shown).
ANOVA was performed using the general linear models procedure to determine the sources of variation for fasting plasma total cholesterol, LDL cholesterol, HDL cholesterol, TG, apo B, and apo A-I, with F tests computed from the type III sums of squares.21 This form of sums of squares is applicable to unbalanced study designs and reports the effect of an independent variable after adjusting for all other variables included in the model. Independent covariates for each ANOVA were sex, age, and the log of BMI. Also included as independent class variables for ANOVA were four genotypes: FABP2 codon 54, PON1 codon 192, APOC3 position −455 nt, and APOE restriction isotype. For any significant associations, we also tested for a genotype×sex interaction in the ANOVA. For significant genotype-phenotype associations, a second ANOVA was performed in each sex separately, using genotype, age, and log BMI as independent variables. In each sex separately, we also performed pairwise comparisons using Bonferroni adjusted t tests to determine differences between genotypic means.
Biochemical Phenotypes in Whole Sample (Table 1⇓)
Out of a possible 534 subjects aged 18 years and older, sufficient DNA and phenotypic information was obtained for analysis from 509 subjects, of whom 289 (57%) were women. The mean±SD for lipid, lipoprotein, and apolipoprotein traits are shown in Table 1⇑. These values are very comparable to those reported in a reference Canadian population of 22 000 subjects.22 Eighteen men and 19 women were taking medication for hypertension; almost all of these took angiotensin-converting enzyme inhibitors. Twenty-three percent of men and 25% of women were current smokers.
Allele and Genotype Frequencies (Table 2⇓)
The frequencies of the alleles of the four genotype systems used in the subsequent ANOVA are shown in Table 2⇑. Some allele frequencies were markedly different from those reported in other populations. Specifically, the allele frequency of the PON1 Q192 was 0.223, which was very low compared with the frequency of ≈0.65 reported in white study samples.13 14 15 16 Also, the frequencies of both the E4 and E2 alleles of APOE were much lower than other populations studied, including Greenland Inuit.23 Observed genotype frequencies did not deviate from those predicted by the Hardy-Weinberg equation (Table 1⇑, all P>.20). Virtually complete linkage disequilibrium was observed between the APOC3 genotypes at positions −482 nt and −455 nt (r=.964, P<.0001), with T at position −482 nt occurring almost exclusively on alleles that contained C at position −455 nt. The frequencies for the four APOC3 promoter haplotypes −482C/−455T, −482C/−455C, −482T/−455C and −482T/−455C were, respectively, 0.544, 0.001, 0.010, and 0.438. The association analysis was performed using the APOC3 genotype at position −455 nt, but the associations with the APOC3 genotype at position −482 nt and APOC3 promoter haplotype were equally significant (data not shown).
Phenotype-Genotype Associations (Table 3⇓)
The results of the ANOVA are shown in Table 3⇑. Since we performed six ANOVAs, we set P<.01 as the nominal level of statistical significance. Sex, age, and log BMI were each highly significantly associated with variation in each plasma lipoprotein trait (Table 3⇑). APOC3 genotype was significantly associated with variation in log TG (P=.0060) and had a borderline association with variation in log apo B (P=.013). APOE genotype was significantly associated with variation in log apo B (P=.0087). At a nominal P<.01, PON1 genotype was not significantly associated with any biochemical variable. The same results were observed when multiple ANOVAs were performed post hoc using each genotype as the sole independent genetic variable and correcting for multiple comparisons (data not shown). We included sex×genotype interaction terms in two separate ANOVAs done to determine whether the association between APOC3 genotype and log TG and the association between APOE genotype and log apo B were gender specific. Both interaction terms were not significant at a nominal P<.01: the probability values for the sex×APOE genotype and sex×APOC3 genotype terms were 0.44 and 0.024, respectively.
Pairwise Comparisons (Table 4⇓)
The basis for the two significant phenotype-genotype associations detected by the above multivariate ANOVA was then examined with pairwise comparisons. The general linear models least-squares means for the log TG in subjects classified by genotypes of APOC3 and for log apo B in subjects classified by genotypes of APOE are shown in Table 4⇑.
Pairwise t tests with Bonferroni correction showed that subjects of both sexes with the C/C genotype of APOC3 position −455 nt had significantly higher mean TG than the other two genotypic classes (Table 4⇑). Identical results were obtained using genotypes of APOC3 position −482 nt and haplotypes formed by genotypes of these two markers (data not shown).
Pairwise t tests with Bonferroni correction showed no differences in plasma apo B between the APOE genotypes for either sex, although the trends to lower apo B in subjects with E2, intermediate apo B in subjects homozygous for E3, and higher apo B in subjects with E4 were similar to those seen in other populations (Table 4⇑).
Subjects were divided into quartiles for untransformed plasma TG, the values for which ranged from 0.38 to 8.34 mmol/L. The cut points for the 25th and 75th percentiles were 1.05 and 1.95 mmol/L, respectively. The frequency of the C allele of APOC3 position −455 nt was significantly higher among subjects in the highest plasma TG quartile than in subjects in the lowest TG quartile (0.539 versus 0.388, P<.001). There were significantly more C/C homozygotes in the highest TG quartile than in the lowest TG quartile (28.7% versus 13.2%, P<.001). The results were identical when alleles of the genotype at APOC3 position −482 nt and the APOC3 haplotype constructed from the two genotypes were used (data not shown).
We found in this study of aboriginal Canadians from Sandy Lake a significant association between APOC3 genetic variation and variation in fasting plasma concentrations of TG. This association was due to TG-raising influence of the genomic variant containing C at position −455 nt of the APOC3 gene, with C/C homozygotes of both sexes having the highest mean plasma TG concentrations. However, similar associations were observed when both the genotype at position −482 nt of APOC3 and the APOC3 haplotype constructed from the two genotypes were studied. This suggests that the association of APOC3 markers with TG may be related to linkage disequilibrium between the markers and another actual functional variant at this locus, which could be within the APOC3 gene or even the neighboring APOA1 or APOA4 genes. Alternatively, the variants tested may truly affect APOC3 gene expression, which then has an impact at the clinical level.
In vitro, the variant containing C at position −455 nt and T at position −482 nt is resistant to insulin-mediated suppression of APOC3 gene transcription, which was maximally about 50% in the insulin-responsive variant containing T at position −455 nt and C at position −482 nt.8 While mean TG levels were within the physiological range in the Sandy Lake sample, they were higher by approximately 30% in homozygotes for the C at position −455 nt and T at position −482 nt than in homozygotes for T at position −455 nt. Furthermore, there were more than twice as many homozygotes for C at position −455 nt and T at position −482 nt in the highest quartile for plasma TG than in the lowest quartile. This suggests that the genomic variation of APOC3, which is associated with the loss of insulin-mediated downregulation of gene expression, might be associated with plasma lipoprotein variation. However, these findings are also consistent with linkage disequilibrium between the markers and another functional change at this locus. Our hope had been to demonstrate that these markers for altered in vitro expression of APOC3 would be associated unequivocally with biologically plausible phenotypes. However, our results do not exclude the possibility of linkage with other functionally important DNA changes in APOC3, or even APOA1 or APOA4.
Pairwise comparisons did not disclose a significant association between APOE genetic variation and plasma apo B, although the trend in the relationship was consistent with that seen in other populations.9 10 The relationship between APOE genetic variation and plasma lipoproteins that is seen in many populations9 10 was also absent in other aboriginal samples.23 The difficulty in detecting the APOE association with plasma lipoproteins in native samples may relate to the unique distribution of alleles, which can result in small numbers of subjects with specific genotypes. For example, in Sandy Lake, the frequencies of APOE E4 and E2 alleles were 0.105 and 0.005, respectively. These frequencies are considerably lower than those reported in white populations.9 10 These frequencies are also markedly different from those reported in Greenland Inuit and American Indians, both of which have about twice as high a prevalence of the E4 allele and a similarly low prevalence of the E2 allele.10 23
We also found that the low activity variant of paraoxonase, encoded by PON1 Q192, had a very low frequency in Sandy Lake and tended to be associated with lower plasma HDL cholesterol. Human paraoxonase is located on HDL24 and can retard the accumulation of lipid peroxidation products in LDL in vitro.25 Both the plasma concentration of paraoxonase and an individual’s PON1 genotype are associated with variation in plasma lipid and lipoprotein concentrations.14 26 Biochemical studies suggested that the low activity variant of paraoxonase had a very low prevalence in native Canadians, compared with other populations.27 The high activity variant of paraoxonase, encoded by PON1 R192, has been associated with increased susceptibility to coronary heart disease.15 16 Somewhat paradoxically, the PON1 R192 allele was also associated with higher plasma HDL cholesterol,14 just as we found in Sandy Lake. But in a study of Hutterites, the PON1 Q192 allele was associated with a lower ratio of LDL:HDL cholesterol.14 These disparities may be due to interpopulational differences in linkage disequilibrium with functional genetic determinants at this locus.
The T54 variant of FABP2 has increased affinity for long-chain fatty acids in vitro compared with the A54 variant.11 While we found no significant association between FABP2 genotype and any biochemical variable in our multivariate model, preliminary analyses indicate that FABP2 genotype may be associated with variation in traits such as percent body fat and BMI (data not shown). This possibility requires further study, as does the possibility that variation in FABP2 is related to variation in traits related to carbohydrate metabolism, such as plasma glucose and insulin.
The differences in PON1 and APOE allele frequencies in Sandy Lake compared with other samples might have resulted from founder effects involving the ancestors of the contemporary community. Archeological studies suggested that hunter-gatherers inhabited the Sandy Lake region 6000 years ago.28 The current inhabitants of the Sandy Lake region lived a hunter-gatherer subsistence until about 70 years ago. The present community is largely descended from one clan who established the present reservation. Alternatively, selection pressure might have produced the present allele frequencies.
In summary, we have observed a significant association between variation in the insulin-response element of the APOC3 promoter and variation in plasma TG. Specifically, the presence of a C at nucleotide −455 and a T at nucleotide −482 of the APOC3 promoter is associated with higher plasma TG. This modest association might be due to a direct effect of the genetic variation or to linkage disequilibrium with another functional change at the APOA1/APOC3/APOA4 locus. Genetic variation at this gene cluster may be associated with a predisposition to markedly abnormal phenotypes, such as hypertriglyceridemia, when secondary genetic or environmental factors are present. Understanding the background of genetic predisposition to abnormal phenotypes may be important in native populations, which may develop an increased prevalence of metabolic diseases as their lifestyles change.29 30
Selected Abbreviations and Acronyms
|BMI||=||body mass index|
|FABP||=||intestinal fatty acid binding protein|
This study was supported in part by operating grants from the National Institutes of Health (No. 91-DK-01), the Ontario Ministry of Health (No. 04307), the Heart and Stroke Foundation of Ontario (No. T2978), the Medical Research Council of Canada (No. MA-13430), and the St Michael’s Hospital Foundation. Dr Hegele is a Career Investigator of the Heart and Stroke Foundation of Ontario. We thank the chiefs and council of the community of Sandy Lake, the Sandy Lake community surveyors, the Sandy Lake nurses, the staff of the University of Toronto Sioux Lookout program, and the Department of Clinical Epidemiology of the Samuel Lunenfeld Research Institute.
Breslow JL. Genetic basis of lipoprotein disorders. J Clin Invest. 1989;84:373-380.
Mehrabian M, Lusis AJ. Genetic markers for atherosclerosis and related risk factors. In: Molecular Genetics of Coronary Artery Disease: Candidate Genes and Processes in Atherosclerosis. Lusis AJ, Rotter JI, Sparkes RS, eds. Basel, Switzerland: Karger; 1992:363-418.
Rees A, Stocks J, Sharpe CR, Vella MA, Shoulders CC, Katz J, Jowett NI, Barelle FE, Galton DJ. Deoxyribonucleic acid polymorphism in the apolipoprotein A-I-C-III gene cluster: association with hypertriglyceridemia. J Clin Invest. 1985;76:1090-1095.
Tas S. Strong association of a single nucleotide substitution in the 3′-untranslated region of the apolipoprotein-C-III gene with common hypertriglyceridemia in Arabs. Clin Chem. 1989;35:256-259.
Zeng Q, Dammerman M, Takada Y, Matsunaga A, Breslow JL, Sasaki J. An apolipoprotein CIII marker associated with hypertriglyceridemia in Caucasians also confers increased risk in a west Japanese population. Hum Hered. 1994;95:371-375.
Ito Y, Azrolan N, O’Connell A, Walsh A, Breslow JL. Hypertriglyceridemia as a result of human apolipoprotein CIII gene expression in transgenic mice. Science. 1990;249:790-793.
Dammerman M, Sandkuijl LA, Halaas JL, Chung W, Breslow JL. An apolipoprotein CIII haplotype protective against hypertriglyceridemia is specified by promoter and 3′-untranslated region polymorphisms. Proc Natl Acad Sci U S A. 1993;90:4562-4566.
Li WW, Dammerman MM, Smith JD, Metzger J, Breslow JL, Leff T. Common genetic variation in the promoter of the human apo CIII gene abolishes regulation by insulin and may contribute to hypertriglyceridemia. J Clin Invest. 1995;96:2601-2605.
Baier LJ, Sacchettini JC, Knowler WC, Eads J, Paolisso G, Tataranni PA, Mochizuki H, Bennett PH, Bogardus C, Prochazka M. An amino acid substitution in the human intestinal fatty acid binding protein is associated with increased fatty acid binding, increased fat oxidation, and insulin resistance. J Clin Invest. 1995;95:1281-1287.
Mitchell BD, Kammerer CM, O’Connell P, Harrison CR, Manire M, Shipman P, Moyer MP, Stern MP, Frazier ML. Evidence for linkage of postchallenge insulin levels with intestinal fatty acid binding protein (FABP2) in Mexican-Americans. Diabetes. 1995;44:1046-1053.
Hegele RA, Brunt JH, Connelly PW. A polymorphism of the paraoxonase gene associated with variation in blood pressure in a genetic isolate. Arterioscler Thromb Vasc Biol. 1995;15:89-95.
Serrato M, Marrian AJ. A variant of the human paraoxonase/arylesterase (HUMPONA) gene is a risk factor for coronary artery disease. J Clin Invest. 1995;96:3005-3008.
Hegele RA, Evans AJ, Tu L, Ip G, Brunt JH, Connelly PW. A gene-gender interaction affecting lipoproteins in a genetic isolate. Arterioscler Thromb. 1994;14:671-678.
Hegele RA, Brunt JH, Connelly PW. Multiple genetic determinants of variation of plasma lipoproteins in a genetic isolate. Arterioscler Thromb Vasc Biol. 1995;15:861-871.
Hixson JE, Vernier DT. Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with HhaI. J Lipid Res. 1990;31:545-548.
SAS Institute Inc. SAS/STAT User’s Guide, Version 6. Cary, NC: SAS Institute Inc; 1987.
Connelly PW, MacLean DR, Horlick L, O’Connor B, Petrasovits A, Little JA. Plasma lipids and lipoproteins and the prevalence of risk for coronary heart disease in Canadian adults. Can Med Assoc J. 1992;146:1977-1987.
de Knijff P, Johansen L, Rosseneu M, Frants RR, Jespersen J, Havekes LM. Lipoprotein profile of a Greenland Inuit population: influence of anthropometric variables, apo E and A4 polymorphism and lifestyle. Arterioscler Thromb. 1992;12:1371-1379.
Kalow W. Interethnic variation of drug metabolism. Trends Pharmacol Sci. 1991;12;102-107.
Fiddler T, Stevens JR. Killing the Shamen. Toronto, Canada: Penumbra Press; 1985.
Trowell HC, Burkitt DP. Western diseases: their emergence and prevention. Cambridge, Mass: Harvard University Press; 1981.