Hum Genet (1992) 88:463-470

9 Springer-Verlag1992

Association of apolipoprotein B gene variants with plasma apoB and low density lipoprotein (LDL) cholesterol levels Samir S. Deeb 1'2, 3, R. Alan Failor 1, B. Greg Brown 1, John D. Brunzell 1, John J. Albers 1, Arno G. Motulsky 1'2' 3, and Ellen Wijsman 1 1Department of Medicine, 2Department of Genetics, and 3Center for Inherited Diseases, Universityof Washington, Seattle, WA 98195, USA Received April 3, 1991 / Revised September 8, 1991

Summary. The contribution of the variants of the apolipoprotein (apo) B locus to the total variance in plasma apoB and cholesterol levels was examined in four independent populations, two that were composed of normal controls (n = 77 and 85) and two with coronary heart disease (n = 115 and 159). A correlation between genotype at the apoB-XbaI locus and apoB levels was observed. The effects of the (+; presence of restriction site) and ( - ) alleles were to increase or decrease the apoB and cholesterol levels by approximately 3.5 mg/dl, respectively. None of the 274 individuals in the coronary heart disease (CHD) groups was found to be a carrier of the apoB allele Arg3500-->Gln, previously shown to be associated with an apoB protein defective in binding to the low density lipoprotein receptor (LDL-R). No DNA sequence variants were found in the region encoding amino acid residues 3129-3532 within the putative LDL-R binding domain among 35 individuals with apoB levels above the 94th percentile (141 mg/dl).

Introduction Epidemiologic studies have established strong relationships between plasma levels of cholesterol-carrying lipoprotein particles and the development of atherosclerosis (Miller and Miller 1975; Fager et al. 1981; Maciejko et al. 1983). Elevated levels of low density lipoprotein (LDL) and depressed levels of high density lipoprotein (HDL) cholesterol are generally associated with an increased risk for premature atherosclerosis and coronary heart disease (Avogaro et al. 1978; Brunzell et al. 1984; DHHS 1980). An elevated level of lipoprotein (a) has been established as an independent risk factor for this disease (Berg 1983; Dahlen et al. 1986). It is generally accepted that both genetic and environmental factors contribute to variation in plasma lipoprotein levels and

Offprint requests to: S. Deeb, Schoolof Medicine, RG-25, University of Washington, Seattle, WA 98195, USA

the development of atherosclerosis, Genetic factors may include allelic variation at gene loci encoding the various classes of apolipoproteins. Restriction site polymorphisms (RFLPs) at or near some of the gene loci of these apolipoproteins have been used to test for association between particular alleles and lipid levels or coronary heart disease (CHD) in various populations. Various reviews have discussed the rationale of this approach and summarized the findings (Hegele and Breslow 1987; Humphries 1988; Lusis 1988; Cooper and Clayton 1988). Results of population association studies using restriction fragment length polymorphism (RFLP) markers at the apolipoprotein A-I/C-III/A-IV cluster of genes on chromosome 11 suggest that mutation(s) at this locus may be involved in the development of type IV/type V hypertriglyceridemia (Rees et al. 1983, 1985, 1986; Kessling et al. 1985; Henderson et al. 1987; Humphries 1988). Association between RFLP markers at this locus and CHD in men under the age of 60 is less clear (Deeb et al. 1986a; Frossard et al. 1986). However, in a Scottish study of 246 men aged 30-59 years who had first degree relatives with CHD, a significant association was observed between CHD and rare alleles of four RFLPs (XmnI, PstI, MspI, and SstI) at this locus (Price et al. 1989). This association was not significant when 713 men without a family history of CHD were included in the analysis. Recently, DNA markers at this locus gave evidence for linkage to the gene for familial combined hyperlipidemia (FCHL) in a subset of families with this condition (Wojciechowski et al. 1991). Another locus that has been the target of such studies is that of the apolipoprotein B (apoB) gene on chromosome 2. ApoB-100, the protein moiety of LDL, is the ligand responsible for clearance of plasma LDL cholesterol by the LDL receptor-mediated pathway. Results of population association studies at this locus have been contradictory. In a Boston-based study, a significant association was observed between apoB RFLP markers (detected with Xbal, EcoRI, and MspI) and of myocardial infarction, but not between the RFLP markers and lipid parameters (Hegele et al. 1986). In a study from London, certain haplotypes defined by RFLP

464 alleles at the apoB locus were observed to be associated with either obesity, plasma cholesterol, or coronary heart disease (Rajput-Williams et al. 1988). Frequencies of alleles of an apoB minisatellite were found to be significantly different in a population of Austrian males with severe coronary heart disease as c o m p a r e d with normal controls (Friedl et al. 1990). The association with heart disease was not confirmed by several other studies ( D e e b et al. 1986a; A b u r a t a n i et al. 1987; Ferns and Dalton 1986; Young et al. 1987). The apoB XbaI polymorphism (a silent C-to-T transition in codon 2488 located less than I kb 5' upstream of the proposed ligand domains) was observed to be associated with plasma cholesterol and apoB levels in a small (56) sample of Norwegian students (Berg 1986), and with total and L D L cholesterol levels in apparently healthy individuals f r o m Finland (Alto-Setala et al. 1988) and the United Kingdom (Talmud et al. 1987). Allelic variation within the proposed L D L receptor binding domain (Knoff et al. 1986; Yang et al. 1986; Milne et al. 1989) could lead to diminished binding to the L D L receptor and consequently to elevated plasma levels of the apoprotein and of L D L particles. A relatively u n c o m m o n mutant apoB-100 allele associated with defective L D L particles and hypercholesterolemia has been described (Soria et al. 1989). The disorder, referred to as familial defective apolipoprotein B-100, results from a mutation in codon 3500 substituting glutamine for arginine in close proximity to the p r o p o s e d ligand domain (Innerarity et al. 1987). Using the monoclonal antibody MB-19 that recognizes two variants of apoB, Gavish et al. (1989) demonstrated that these two apoB variants were unequally represented in plasma of individuals heterozygous for the two alleles. This allele-specific inequality was relatively c o m m o n , inherited in a codominant manner, and linked to l h e apoB gene locus, implying the presence of relatively c o m m o n apoB gene variants in the population that influence plasma apoB and L D L levels. Thus mutations in the apoB gene might account for a significant fraction of the genetic variance in apoB and L D L levels. In this study we assessed the significance of genetic variation at the apoB locus by analysis of a D N A variant in a population study, and by sequencing of a region of the apoB gene p r e s u m e d to code for the L D L receptor binding domain. In our earlier study ( D e e b et al. 1986a) we were unable to detect significant differences in the frequency of R F L P alleles at any of the apolipoprotein loci between C H D and control populations. However, the apoB-XbaI ( + ) allele a p p e a r e d to be associated with elevated plasma cholesterol levels in these populations (Deeb et al. 1986a). Since this was the only significant association observed out of a n u m b e r of similar comparisons and since the association was only significant at the 5% level, the biological significance of this correlation was in question. To test the replicability of this finding we retested only this m a r k e r in two new populations, one with coronary heart disease and another of "normal" individuals. The C H D population was included since variation in plasma apoB and cholesterol levels were higher in such patients than in normals, making it potentially more

likely that we could observe genotypic effects if present. It was also of interest to determine if the genotypic effects existed in both normal and C H D individuals. The results confirm the previously reported association ( D e e b et al. 1986a) between XbaI alleles of the apoB gene and plasma cholesterol levels. The other aim of this study was to examine the nucleotide sequence of a segment of the apoB gene, located within the putative L D L receptor binding domain, for variants that result in amino acid substitutions that may be associated with elevated plasma apoB levels. None were found within this region.

Materials and methods

DNA preparation DNA was prepared from venous blood leukocytes by lysis in proteinase K - sodium dodecyl sulfate followed by phenol extraction and ethanol precipitation either manually or on an Applied Biosystems (Foster City, Calif.) model 340-A nucleic acid extractor, according to the manufacturer's instructions.

Southern blot analysis Approximately 10 gg DNA was digested with a restriction endonuclease (20 units) and subjected to electrophoresis on 1.0% gels. After alkali denaturation (0.5 N NaOH, 1.5 M NaC1, 60 min) and neutralization (l.0 M Tris HC1, pH 7.5, 1.5 M NaC1, 60 rain), DNA was transferred onto nitrocellulose (Schleicher and Schuell, Keene, N.H.) as described earlier (Maniatis et al. 1982). The DNA probes and conditions used for hybridization and washing of blots were described earlier (Deeb et al. 1986a).

Polymerase chain reaction (PCR) amplification and direct sequencing An 813-bp fragment of exon 26 was amplified using genomic DNA as template and the following two oligodeoxynucleotides as primers: (9542) 1. 5' CAGGCTTGAAGGAATTCTTG 3' (10355) 2. 5' AGCCACTGACACTTCCATA 3' Thirty cycles of 1 min at 94~ 1 min at 55~ and 1 min at 72~ using the Perkin-Elmer/Cetus Gene Amp kit and DNA thermal cycler were used to amplify the desired fragment. PCR products were purified from unincorporated dNTPs and nonspecific DNA products by electrophoresis in a low-melting point agarose gel (SeaPlaque agarose, FMC BioProducts, Me.). The isolated gel slice containing the PCR product was then extracted twice with phenol to remove the agarose, and DNA was then precipitated with ethanol and 0.3 M NaOAc. Reagents of the Sequenase kit (U.S. Biochemical Co., Ohio) were used for sequencing reactions according to the following procedure: (1) About 1 pmol of the double-strand PCR product was mixed with 5 pmol of sequencing primer (either one of the PCR primers) in 10 gl of sequencing buffer (40 mM Tris HC1, pH 7.5, 20 mM MgC1, and 50 mM NaC1). (2) The template-primer mix was heated in a boiling water bath for 5 min followed by quick chilling on ice. After a quick spin to bring down the condensation, the template-primer mix was added to 5 gl of labeling mix (20 mM DTT, 0.6 gM each of dGTP, dATP, dTTP, dCTP, 10 gCi of 35S dATP (NEN Research Products, specific activity > 1000 Ci/mmol) and 2 units of Sequenase. The labeling

465 reaction was carried out at room temperature for 2 rain. Termination reaction was then done by adding 3.5 lal of labeling mix to 2.5 gl of each of the four termination mixes (80 gM each of dGTP, dATP, dTTP, dCTP plus 8.0 ~tM of one of the four dideoxyribonucleotides), and incubated at 37~ for 5 rain. Then 3 gl of stop solution (95% formamide, 20 mM E D T A , 0.05% bromophenol blue, 0.05% xylene cyanol) was added to the mixture. The samples were heated to 75~ for 2 min immediately before loading on an 8% polyacrylamide/7 M urea gel for sequence analysis. At least 200 bases could be read after an overnight exposure.

Single-strand conformation polymorphism (SSCP) analysis A 1215-bp segment of exon 26 of the apoB gene was PCRamplified using genomic D N A as template and the following oligonucleotide primers: (9542) 1. 5' C A G G C T T G A A G G A A T T C T T G 3' (10756) 2. 5' T A T G C G T T G G A G T G T G G C T We used 32 cycles of I rain denaturation at 94~ 1 rain annealing at 54~ and 1.25 min elongation at 72~ The amplified D N A was labeled by incorporation of 32p-dCTP according to the procedure described by Cawthon et al. (1990). The 1215-bp fragment was digested with both NcoI and SstI to yield 259-, 278-, 372-, and 306-bp subfragments (given in the 5' to 3' order). The subfragmerits were then denatured by heating at 90~ for 3 rain and the single strands separated on a 4.5% nondenaturing polyacrylamide gel according to the procedure of Orita et al. (1989) as modified by Cawthon et al. (1990). The gel contained 10% glycerol and was run at room temperature using a fan for cooling.

Screening for the apoB 3500 mutation Genomic D N A samples were screened for the apoB 3500 mutant allele according to the procedure of Soria et al. (1989). Oligonucleotide primers were first used to PCR amplify a D N A fragment encompassing the mutation conducted in a Perkin-Elmer/ Cetus thermal cycler (Saiki et al. 1988). The amplified D N A was then dot-blotted onto nylon membranes (Hybond-N, Amersham) and hybridized with 3:p-labeled oligonucleotide probes specific to the wild-type and mutant alleles. D N A from a known heterozygote for the 3500 mutation (a kind gift of Dr. Brian McCarthy, Gladstone Foundation, San Francisco) was used as a control. PCR, hybridization, and washing conditions were exactly as described (Soria et al. 1989).

Populations The two original C H D and control populations (CHD-1, control-I) were collected for a patient-control study and were described previously (Deeb et al. 1986a). The two new population samples (CHD-2 and control-2) used in this study were Caucasoids from the Seattle area and collected specifically for this study, which aims at assessing the contribution of the apoB locus to plasma apoB and cholesterol levels. The CHD group was included since it displays greater variability in these phenotypes, making it more likely the genotypic effects could be detected. The C H D groups were composed of males less than 65 years of age who were consecutive patients at the Cardiac Catheterization Laboratory. They were all angiocardiographically proven to have 50% or more stenosis in at least one coronary artery (NV50 > 1). Examinations were performed at the Cardiac Catheterization Laboratory, University of Washington. The control population was mainly composed of ostensibly healthy medical students at the University of Washington and unrelated healthy and normolipidemic spouses from family studies for lipid abnormalities.

Lipid analysis Plasma lipids and lipoproteins were determined on all patients and control groups at the Northwest Lipid Research Center (LRC) by standard technologies of the L R C program (Lipid Research Clinic 1974). ApoB was measured by radioimmunoassay (Albers et al. 1975).

Statistical analysis ApoB levels were adjusted for age and sex using the following equations: a = age, Y = trait to be adjusted (e.g., apoB, LDL); Y/ = observed trait level; YI = adjusted trait level. Y1 = apoB level (adjusted to a mean 98.694 mg/dl). Y2 = total cholesterol level (adjusted to mean of 198.113 mg/dl). Y3 = L D L cholesterol (calculated; adjusted to mean of 126.163 mg/dl). ApoB levels: For males > 17 years of age: Y~ = I/1 - (1.901654 x a - 0.000175454 x a 3 + 36.13429) + 98.694. For females > 17 years of age: YI = Y1 - (0.795234 x a + 63.669358) + 98.694. Total cholesterol levels: For males > 17 years of age: Y~ = Y2 - (2.735699 x a - 0.00025 x a 3 + 105.0329) + 198.113. For females > 17 years of age: Y) = Y2 - (1.291317 x a + 136.011329) + 198.113. LDL cholesterol levels: For males > 17 years of age: Y~ = Y3 - (2.268495 x a - 0.0002153 x a 3 + 55.80278) + 126.163. For females > 17 years of age: Y~ = I(3 - (0.853306 x a + 82.216177) + 126.163. Possible differences in allele frequency between patients and controls were tested by a Chi square test. Differences in mean apoB levels among genotypes in the pooled data were tested with oneway analysis of variance, under a random effect model. Differences in mean apoB levels among genotypes within populations were tested with Friedman's rank test (Neter et al. 1986) because Bartlett's test (Neter et al. 1986) indicated violation of the assumption of equality of variances in the various populations needed to perform the test with two-way analysis of variance. Differences in mean apoB level among genotypes were also compared with oneway analysis of variance after pooling the data from patients and controls. Average effects of alleles were estimated as described by Falconer (1981) under the assumption of Hardy-Weinberg equilibrium. Contribution of the apoB locus to overall variance of apol3 levels was calculated with methods suggested by Boerwinkle and Sing (1986). Linkage disequilibrium, D, between pairs of markers, was tested with a Chi square test. Patients and controls were analyzed separately because observed gene frequency differences in the two groups could give spurious evidence of linkage disequilibrium if the groups were pooled. When the null hypothesis that D = 0 for a pair of markers could not be rejected, sample sizes necessary to reject the null hypothesis with a power of at least 80% were computed (Thompson et al. 1988) to determine whether the actual sample sizes were likely to be sufficient to reject the null hypothesis if it is false.

Results

Linkage disequilibrium at the apoB locus m a r k e r s at t h e a p o B g e n e l o c u s , (PvuII, MS) previously shown to map to 2p23( D e e b et al. 1 9 8 6 b ) , w e r e u s e d in this s t u d y

Four RFLP

XbaI, EcoRI, 2p24

466 (Fig. 1). These m a r k e r s were used to estimate the contribution o f alleles of the a p o B gene to the variance in plasma a p o B and cholesterol levels in the population. K n o w i n g the linkage relationships a m o n g these markers

Pvu~E _~_ I

,

i

i

,

O

I

,

,

,

i

I

5

i

I

,

i

i

i

I

20

'

41 O

I

i

I

:

,

,

15

,

25 3500

RI

I

f

V///////A I '

i

.

10

Xb o I

'

I

_

,

l

|| ,

I

i

I

.30

J

i

i

L

I

:55

I

MS

IIII-. . . . . i

i

i

Chr. 2

I

45

50 kbp

Fig.1. Structure of the human apoB gene. Shown are the polymorphic sites used to determine linkage disequilibrium, The XbaI site was used in studies of association with plasma apoB and cholesterol levels. The minisatellite (MS) is polymorphic in the number of repeats. The putative LDL receptor binding region is indicated with a box. Exons and introns are indicated with thick and thin bars', respectively. Structure of the gene adapted from Blackhart et al. (1986) Table 1. Linkage disequilibrium (D') between markers at the apoB locus. * P < 0.001. MS, minisatellite RFLP pair

Allele frequency

Sample size

D' %

PvuII/XbaI PvuII/EcoRI PvuII/MS XbaI/EcoRI XbaI/MS EcoRI/MS

0.074/0.490 0.077/0.193 0.079/0.335 0.497/0.178 0.471/0.321 0.184/0.330

242 246 248 369 259 264

96* 34 81 96* 68* 84*

Table 2. Variation of apoB and cholesterol levels as a function of genotype at the apoBXbaI locus

Population

is i m p o r t a n t in the choice of appropriate m a r k e r s for association studies as well as for interpretation of the results. T h e extent of linkage disequilibrium, D, between pairwise combinations of the R F L P m a r k e r s was assessed as described u n d e r Materials and m e t h o d s . As shown in Table 1, highly significant positive linkage disequilibrium was o b s e r v e d b e t w e e n the following pairs: PvuII/XbaI, PvuII/MS, XbaI/EcoRI, XbaI/MS, and EcoRI/MS. These results show that the region extending f r o m PvuII to the minisatellite (MS), approximately 37 kb in length, d e m o n s t r a t e s a high degree of linkage disequilibrium. The low f r e q u e n c y of the PvuII R F L P most likely precluded detection of significant linkage disequilibrium, since a very large sample size would be required in cases of negative disequilibrium ( T h o m p s o n et al. 1988). Based on these results it can be expected that any D N A variants that influence apoB levels and are within this region would be in linkage disequilibrium with some or all the R F L P markers.

Frequency of apolipoprotein B alleles as a function of plasma lipid and lipoprotein concentrations Plasma levels of a p o B and cholesterol (total and L D L cholesterol) were significantly associated with the genotype at the apoB XbaI locus in both the C H D and control populations. Table 2 shows plasma levels of a p o B , L D L cholesterol, and total cholesterol as a function of g e n o t y p e of the apoB-XbaI site in b o t h the original and new populations studied. Presence of the 5.0-kb ( + ) allele was correlated with higher a p o B , L D L cholesterol, and total cholesterol levels in the population of n o r m a l and c o r o n a r y heart disease patients. Association of the ( + ) allele with higher a p o B levels was significant at the 0.015 level in the c o m b i n e d population. The average effects of the apoB-XbaI alleles on a p o B and cholesterol levels in the four populations are given

BXbal

Mean plasma level (mg/dl) _+ SD genotype ApoB Total cholesterol

LDL cholesterol

Control1

++ +-

110.3_+15.3 (18) a 106.5 _+17.7 (41) 96.6 4- 18.2 (18)

204.7_+26.7 (20) 205.2 _+31.1 (43) 187.7_+35.2 (22)

120.9+29.2 (20) 119.8 _+27.9 (43) 107.2 _+27.4 (22)

Control2

++ +--

126.7 -+ 38.6 (32) 113.8 _+28.7 (32) 108.7 _+35.4 (21)

217.3 _+45.3 (32) 203.7 _+37.0 (33) 206.0 _+48.8 (21)

140.0 + 40.2 (32) 128.3 + 33.0 (33) 130.8 + 41.8 (21)

CHD1

++ +--

128.8_+32.7 (30) 126.5_+33.5 (56) 128.5 _+39.2 (29)

216.5_+44.1 (30) 211.1_+41.3 (55) 213.0 + 52.1 (27)

145.0+45.3 (28) 137.7+40.9 (52) 137.0 _+49.0 (25)

CHD2

++ +--

124.3_+26.0 (29) 112.5 ___31.7 (75) 110.0-+37.3 (55)

214.0_+45.5 (29) 200.8 + 39.5 (75) 189.1-+45.3 (55)

136.2+40.9 (27) 126.0 _+37.9 (73) 114.5+35.5 (51)

Combined

++ +--

123.9 + 29.8 (109) 115.3 _+29.0 (204) 111.9-+35.2(121)

213.9 -+ 41.9 (111) 204.9 _4-37.8 (206) 196.8-+45.2(125)

136.8 -+ 39.7 (107) 128.1 + 35.7 (201) 120.7_+38.0(119)

a Numbers in parentheses are number of individuals in each genotypic class

467 Control 1 7 6 5

Control 2

CHD-1

CHD-2

Pooled

a(+)

a(+)

g3 ,,=,1 go

3-1

Association of apolipoprotein B gene variants with plasma apoB and low density lipoprotein (LDL) cholesterol levels.

The contribution of the variants of the apolipoprotein (apo) B locus to the total variance in plasma apoB and cholesterol levels was examined in four ...
923KB Sizes 0 Downloads 0 Views