Genetic Epidemiology 7:199-210 (1990)
Genetic Epidemiology of Differences in Low-Density Lipoprotein (LDL) Cholesterol Concentration: Possible Involvement of Variation at the Apolipoprotein B Gene Locus in LDL Kinetics R.S. Houlston, P.R. Turner, B. Lewis, and S.E. Humphries Division of Chemical Pathology and Metabolic Disorders, United Medical and Dental Schools, St. Thomas’ Hospital (R.S.H., P,R. T., B.t.),and Charing Cross Sunley Research Centre, Hammersmith (R.S.H., S.E.H.), London, England Circulating levels of low-density lipoprotein (LDL) vary considerably within and between populations, paralleled by differing coronary heart disease (CHD) mortality rates. We have previously shown that variation in the apolipoprotein (apo) B gene as associated with certain restriction fragment length polymorphisms (RFLPs) influences the metabolism of LDL in the U.K. population. To investigate a possible genetic contribution to variation in LDL levels in differing populations we have extended this original study. RFLPs of the apo B gene were determined in samples of individuals from the United Kingdom, Finland, Italy, Spain, and Africa. Significant associations of LDL fractional catabolic rate with the apo B EcoRI and XbaI RFLP genotypes were detected only in the two North European populations. In the African population sample, the XbaI RFLP displayed a significant association with LDL apo B synthesis. The data suggest that variation in the apo B gene influences the metabolism of LDL and that it is different in individuals of different ethnic background. Key words: restrictionfragment length polymorphisms (RFLPs), apo B
Received for publication January 17, 1989; revised January 24, 1990. P.R. Turner’s present address is Universitad de Barcelona, Facultad de Medicina, Chant Lloreng 21,43201 Revs, Spain. Address reprint requests to R.S. Houlston, Charing Cross Sunley Research Centre, Lurgan Avenue, Hammersmith, London W6 8LW, England.
0 1990 Wiley-Liss, Inc.
200
Houlston et al.
INTRODUCTION
The metabolic determinants of circulating levels of low-density lipoprotein (LDL) are of interest in that LDL is causally related to the development of coronary heart disease (CHD) [Lewis 1983; Lipid Research Clinics Program, 19841. Concentrations of LDL vary considerably within populations, and distributions are strikingly different between different ethnic groups [Lewis et al., 1974; Keys, 1980; Neaton et al., 19841. It has been calculated that approximately 50% of the variability in normal plasma cholesterol levels in the general population is attributable to genetic differences among individuals [Sing and Orr, 1978; Moll et al., 1979; Rao et al., 19791. Variation at the apolipoprotein (apo) E gene locus has been calculated to contribute 4% of the phenotypic variation in total cholesterol levels in the general population [Boerwinkle and Sing, 19861. Studies on both apo E [Weintraub et al., 19871 and the LDL receptor [Weight et al., 19821 suggest that there is a significant genetic component to the metabolic fate of dietary cholesterol. The level of circulating cholesterol is thus the result of a complex interplay between environment and genetic factors. Apolipoprotein B is the major protein component of LDL and has a central role in the metabolism of this lipoprotein as the ligand for its cellular uptake through the liver B/E receptors and subsequent degradation; approximately 70% of LDL is catabolized by this route [Sparks and Sparks, 1985; Brown and Goldstein, 19831. Epidemiological studies have implicated elevated levels of apo B in patients with premature atherosclerosis [Sniderman et al., 19801. Case control studies have shown that levels of apo B discriminate well between patients with and without overt CHD or severe atherosclerosis [Whayne et al., 1981; Durrington et al., 19861. The heritibility of apo B is high [Berg, 1983; Hamsten et al., 19861, and evidence for a major gene determining levels of apo B has been reported with use of complex segregation analysis [Hasstedt et al., 1987; Pairetz et al., 19881. With the cloning of the human apo B gene [Lusis et al., 1985; Knott et al., 1986; Carlsson et al., 1985; Shoulders et al., 19851, a number of common restriction fragment length polymorphisms (RFLPs) have been identified [Shoulderset al., 1985; Priestly et al., 1985;Barni et al., 19861. Using such RFLPs, examination of how variation at this locus influences lipoprotein and apo levels has been made. The XbaI RFLP has been shown to be associated with variation in serum cholesterol and triglyceride levels [Law et al., 1986; Berg, 1986; Talmud et al., 1987; Aalto-Setala et al., 19881 and to account for between 4% and 10%of the total phenotypic variance in cholesterol levels in individuals from the general population [Talmud et al., 1987, 19891. Additionally this RFLP has been suggested to represent an independent risk factor for ischaemic heart disease without influence on circulating levels of apo B or LDL [Hegele et al., 19861. Recently, we have published two studies on the relationship between LDL kinetics and the XbaI RFLP, suggesting that the effect of the variation associated with this RFLP is mediated primarily through variation in the apo B protein itself affecting the catabolic rate of LDL [Houlston et al., 1988; Demant et al., 19881. In the Scottish study, individuals possessing the X2 (presence of cutting site) allele had the lowest LDL fractional catabolic rate (FCR), providing an attractive explanation for the higher levels of cholesterol documented to be associated with the X2 allele in previous population studies [Law et al., 1986; Berg, 1986; Talmud et al., 1987; Aalto-Setala et al., 19881. We have now extended these studies to compare the potential contribution of
Apolipoprotein B Gene and LDL Kinetics
201
variation in the apo B gene to LDL kinetics in five areas, the United Kingdom, Finland, Italy, Spain and Africa. MATERIALS AND METHODS Subjects
Subjects studied (n = 14-23 in each of the five groups) were men aged 35-49 years chosen to represent the three deciles of cholesterol at each participating centre [International Collaborative Study Group, 19861.Criteria for exclusion were body mass index less than 20 or more than 27, recent illness, therapeutic diets or lipid-lowering drugs, and physical signs of severe familial hyperlipidaemia. Kinetics and Cell Analysis
The FCR of LDL was determined for each individual using autologous 1251-labelled LDL and the urine plasma ratio method [Turner et al., 1984; Janus et al., 19801. LDL synthetic rate was calculated from the product of the FCR and the pool size; the pool size was estimated from the plasma concentration of LDL apo B and the body weight [Janus et al., 19801. The rate of degradation of LDL was determined for each individual using autologous '251-labelledLDL and underpressed lymphocytes. Polymorphism Studies
Isolation of DNA, digestion, Southern blotting and hybridisation analysis were as previously described [Talmud et al., 19871. Enzymes and probes used were, for apo B, EcoRI, pAB3 cDNA; XbaI, pAB3.5 genomic fragment [Barni et al., 1986; Talmud et al., 19871; for Apo AI/CIII/AIV; XmnI, 2.2 kb PstI fragment genomic fragment of the apo A1 gene [Kessling et al., 19851, PUvII, 1.O kb PuvII fragment of apo A1 [Kessling et al., 19881; for the LDL receptor, PvuII and NcoI, 1.9 kb BamHI fragment of the LDL receptor (a gift of Dr. D. Russel) [Taylor et al., 19891. Statistical Analysis
A one-way analysis of variance was performed to test the null hypothesis that phenotypic variation was not associated with genetic variation at each of the loci investigated. The F statistic was used to determine the level of significance of differences between genotypes. We considered statistical significance to be at the 0.05 level. The percentage variance of the lipid traits attributable to any RFLP was obtained from the multiple polynomial equation. Differences in allele frequencies of the two apo B RFLPs between populations were assessed by gene counting and x2 analysis. All statistical manipulations were carried out using the MINITAB statistical program (State College, PA). RESULTS
Within each population sample, LDL cholesterol varied directly with the production rate of LDL apo B, and there was an inverse relationship between LDL cholesterol and LDL apo B-FCR. Correlations between the three kinetic parameters and LDL concentrations are shown in Table I. Mean levels of plasma LDL were significantly different and were ranked in the order United Kingdom > Finland > Italy > Spain >
202
Houlston et al.
TABLE I. LDL Cholesterol Values, LDL Apo B Production Rate, and FCR and LDL Degradation by Mononuclear Cells in Samples From Five Different Populations (Means f SD): Correlations of LDL ConcentrationsWith LDL Kinetic Parameters Given in Brackets
Population
N
LDL cholesterol (mmol/liter)
United Kingdom
22
4.31 k 1.23
Finland
22
3.7 -+ 1.09
Italy
23
3.65
?
1.30
Spain
14
3.26
k
1.09
Africa
15
2.26
?
0.74
F P df
6.66
< 0.001 4/94
LDL-FCR (poolslday)
LDL production (mg/kg/day)
0.325 k 0.07 ( - 0.457***) 0.353 k 0.09 (-0.401) 0.406 rt 0.13 (-0.31) 0.423 2 0.15 ( -0.46) 0.361 k 0.10 (-0.10)
7.51 2 3.0 (0.685*) 7.54 2 2.4 (0.609*) 8.91 2 3.3 (0.56*) 6.87 2 2.4 (0.35) 6.33 f 2.7 (0.64**)
2.50 0.05 4/94
2.18
Genetic Epidemiology of Differences in Low-Density Lipoprotein (LDL) Cholesterol Concentration: Possible Involvement of Variation at the Apolipoprotein B Gene Locus in LDL Kinetics R.S. Houlston, P.R. Turner, B. Lewis, and S.E. Humphries Division of Chemical Pathology and Metabolic Disorders, United Medical and Dental Schools, St. Thomas’ Hospital (R.S.H., P,R. T., B.t.),and Charing Cross Sunley Research Centre, Hammersmith (R.S.H., S.E.H.), London, England Circulating levels of low-density lipoprotein (LDL) vary considerably within and between populations, paralleled by differing coronary heart disease (CHD) mortality rates. We have previously shown that variation in the apolipoprotein (apo) B gene as associated with certain restriction fragment length polymorphisms (RFLPs) influences the metabolism of LDL in the U.K. population. To investigate a possible genetic contribution to variation in LDL levels in differing populations we have extended this original study. RFLPs of the apo B gene were determined in samples of individuals from the United Kingdom, Finland, Italy, Spain, and Africa. Significant associations of LDL fractional catabolic rate with the apo B EcoRI and XbaI RFLP genotypes were detected only in the two North European populations. In the African population sample, the XbaI RFLP displayed a significant association with LDL apo B synthesis. The data suggest that variation in the apo B gene influences the metabolism of LDL and that it is different in individuals of different ethnic background. Key words: restrictionfragment length polymorphisms (RFLPs), apo B
Received for publication January 17, 1989; revised January 24, 1990. P.R. Turner’s present address is Universitad de Barcelona, Facultad de Medicina, Chant Lloreng 21,43201 Revs, Spain. Address reprint requests to R.S. Houlston, Charing Cross Sunley Research Centre, Lurgan Avenue, Hammersmith, London W6 8LW, England.
0 1990 Wiley-Liss, Inc.
200
Houlston et al.
INTRODUCTION
The metabolic determinants of circulating levels of low-density lipoprotein (LDL) are of interest in that LDL is causally related to the development of coronary heart disease (CHD) [Lewis 1983; Lipid Research Clinics Program, 19841. Concentrations of LDL vary considerably within populations, and distributions are strikingly different between different ethnic groups [Lewis et al., 1974; Keys, 1980; Neaton et al., 19841. It has been calculated that approximately 50% of the variability in normal plasma cholesterol levels in the general population is attributable to genetic differences among individuals [Sing and Orr, 1978; Moll et al., 1979; Rao et al., 19791. Variation at the apolipoprotein (apo) E gene locus has been calculated to contribute 4% of the phenotypic variation in total cholesterol levels in the general population [Boerwinkle and Sing, 19861. Studies on both apo E [Weintraub et al., 19871 and the LDL receptor [Weight et al., 19821 suggest that there is a significant genetic component to the metabolic fate of dietary cholesterol. The level of circulating cholesterol is thus the result of a complex interplay between environment and genetic factors. Apolipoprotein B is the major protein component of LDL and has a central role in the metabolism of this lipoprotein as the ligand for its cellular uptake through the liver B/E receptors and subsequent degradation; approximately 70% of LDL is catabolized by this route [Sparks and Sparks, 1985; Brown and Goldstein, 19831. Epidemiological studies have implicated elevated levels of apo B in patients with premature atherosclerosis [Sniderman et al., 19801. Case control studies have shown that levels of apo B discriminate well between patients with and without overt CHD or severe atherosclerosis [Whayne et al., 1981; Durrington et al., 19861. The heritibility of apo B is high [Berg, 1983; Hamsten et al., 19861, and evidence for a major gene determining levels of apo B has been reported with use of complex segregation analysis [Hasstedt et al., 1987; Pairetz et al., 19881. With the cloning of the human apo B gene [Lusis et al., 1985; Knott et al., 1986; Carlsson et al., 1985; Shoulders et al., 19851, a number of common restriction fragment length polymorphisms (RFLPs) have been identified [Shoulderset al., 1985; Priestly et al., 1985;Barni et al., 19861. Using such RFLPs, examination of how variation at this locus influences lipoprotein and apo levels has been made. The XbaI RFLP has been shown to be associated with variation in serum cholesterol and triglyceride levels [Law et al., 1986; Berg, 1986; Talmud et al., 1987; Aalto-Setala et al., 19881 and to account for between 4% and 10%of the total phenotypic variance in cholesterol levels in individuals from the general population [Talmud et al., 1987, 19891. Additionally this RFLP has been suggested to represent an independent risk factor for ischaemic heart disease without influence on circulating levels of apo B or LDL [Hegele et al., 19861. Recently, we have published two studies on the relationship between LDL kinetics and the XbaI RFLP, suggesting that the effect of the variation associated with this RFLP is mediated primarily through variation in the apo B protein itself affecting the catabolic rate of LDL [Houlston et al., 1988; Demant et al., 19881. In the Scottish study, individuals possessing the X2 (presence of cutting site) allele had the lowest LDL fractional catabolic rate (FCR), providing an attractive explanation for the higher levels of cholesterol documented to be associated with the X2 allele in previous population studies [Law et al., 1986; Berg, 1986; Talmud et al., 1987; Aalto-Setala et al., 19881. We have now extended these studies to compare the potential contribution of
Apolipoprotein B Gene and LDL Kinetics
201
variation in the apo B gene to LDL kinetics in five areas, the United Kingdom, Finland, Italy, Spain and Africa. MATERIALS AND METHODS Subjects
Subjects studied (n = 14-23 in each of the five groups) were men aged 35-49 years chosen to represent the three deciles of cholesterol at each participating centre [International Collaborative Study Group, 19861.Criteria for exclusion were body mass index less than 20 or more than 27, recent illness, therapeutic diets or lipid-lowering drugs, and physical signs of severe familial hyperlipidaemia. Kinetics and Cell Analysis
The FCR of LDL was determined for each individual using autologous 1251-labelled LDL and the urine plasma ratio method [Turner et al., 1984; Janus et al., 19801. LDL synthetic rate was calculated from the product of the FCR and the pool size; the pool size was estimated from the plasma concentration of LDL apo B and the body weight [Janus et al., 19801. The rate of degradation of LDL was determined for each individual using autologous '251-labelledLDL and underpressed lymphocytes. Polymorphism Studies
Isolation of DNA, digestion, Southern blotting and hybridisation analysis were as previously described [Talmud et al., 19871. Enzymes and probes used were, for apo B, EcoRI, pAB3 cDNA; XbaI, pAB3.5 genomic fragment [Barni et al., 1986; Talmud et al., 19871; for Apo AI/CIII/AIV; XmnI, 2.2 kb PstI fragment genomic fragment of the apo A1 gene [Kessling et al., 19851, PUvII, 1.O kb PuvII fragment of apo A1 [Kessling et al., 19881; for the LDL receptor, PvuII and NcoI, 1.9 kb BamHI fragment of the LDL receptor (a gift of Dr. D. Russel) [Taylor et al., 19891. Statistical Analysis
A one-way analysis of variance was performed to test the null hypothesis that phenotypic variation was not associated with genetic variation at each of the loci investigated. The F statistic was used to determine the level of significance of differences between genotypes. We considered statistical significance to be at the 0.05 level. The percentage variance of the lipid traits attributable to any RFLP was obtained from the multiple polynomial equation. Differences in allele frequencies of the two apo B RFLPs between populations were assessed by gene counting and x2 analysis. All statistical manipulations were carried out using the MINITAB statistical program (State College, PA). RESULTS
Within each population sample, LDL cholesterol varied directly with the production rate of LDL apo B, and there was an inverse relationship between LDL cholesterol and LDL apo B-FCR. Correlations between the three kinetic parameters and LDL concentrations are shown in Table I. Mean levels of plasma LDL were significantly different and were ranked in the order United Kingdom > Finland > Italy > Spain >
202
Houlston et al.
TABLE I. LDL Cholesterol Values, LDL Apo B Production Rate, and FCR and LDL Degradation by Mononuclear Cells in Samples From Five Different Populations (Means f SD): Correlations of LDL ConcentrationsWith LDL Kinetic Parameters Given in Brackets
Population
N
LDL cholesterol (mmol/liter)
United Kingdom
22
4.31 k 1.23
Finland
22
3.7 -+ 1.09
Italy
23
3.65
?
1.30
Spain
14
3.26
k
1.09
Africa
15
2.26
?
0.74
F P df
6.66
< 0.001 4/94
LDL-FCR (poolslday)
LDL production (mg/kg/day)
0.325 k 0.07 ( - 0.457***) 0.353 k 0.09 (-0.401) 0.406 rt 0.13 (-0.31) 0.423 2 0.15 ( -0.46) 0.361 k 0.10 (-0.10)
7.51 2 3.0 (0.685*) 7.54 2 2.4 (0.609*) 8.91 2 3.3 (0.56*) 6.87 2 2.4 (0.35) 6.33 f 2.7 (0.64**)
2.50 0.05 4/94
2.18
