DNA sequence variation in human apolipoprotein C4 gene and its
effect on plasma lipid profile
M. Ilyas Kamboh
a,*, Christopher E. Aston
a, Richard F. Hamman
baDepartment of Human Genetics,Graduate School of Public Health,Uni6ersity of Pittsburgh,130 DeSoto Street,Pittsburgh, PA 15261, USA
bDepartment of Pre6enti6e Medicine and Biometrics,Uni6ersity of Colorado Health Science Center,Den6er,CO, USA
Received 18 March 1999; received in revised form 13 September 1999; accepted 3 November 1999
Abstract
Human apolipoprotein C-IV (apoC-IV, protein; APOC4, gene) is a new member of the APO E/C1/C2 gene cluster. In transgenic mice, human apoC-IV is predominantly associated with very low-density lipoprotein (VLDL) and thus may play an important role in lipid metabolism. To our knowledge, the extent and nature ofAPOC4 genetic variation and its role in lipid metabolism are unknown. In this study we have assessed the presence of genetic variation in all three exons ofAPOC4 and their flanking intronic sequence by SSCP and DNA sequencing. A total of five point mutations were observed, including two in the non-coding part of exon 1 (A609G and G620A), two in exon 2 (codons 36 and 52) and one in exon 3 (codon 96). The three mutations in exons 2 and 3 predict amino acid substitutions, Leu36Pro, Gly52Asp, and Leu96Arg. The frequencies of the variant alleles were: 0.010 for 609G, 0.039 for 620A, 0.502 for Pro36, 0.003 for Asp52 and 0.357 for Arg96. Significant pairwise linkage disequilibrium was observed between five of the tenAPOC4 pairs, including nt. 620/codon 36, nt. 620/codon 96, codon 36/codon 52, codon 36/codon 96 and codon 52/codon 96. A general linear model analysis reveled a significant association of the Leu36Pro and the Leu96Arg polymorphisms with triglyceride levels in women. This is consistent with the proposed role of apoC-IV in triglyceride metabolism. The characterization of APOC4 genetic variation may lead to the identification of a specific role of apoC-IV in lipid metabolism or in other physiologic pathways. © 2000 Elsevier Science Ireland Ltd. All rights reserved.
Keywords:ApoC-IV; Genetic variation; Association; Lipid; Linkage disequilibrium
www.elsevier.com/locate/atherosclerosis
1. Introduction
Recently, a new member of the apolipoprotein (apo, protein; APO, gene) gene family, designated APOC4, has been identified in the APO E/C1/C2 gene cluster on human chromosome 19q 13.2 [1]. TheAPOC4 gene is linked to theAPOC2 gene, with its 3%terminus lying only 555 bp upstream of theAPOC2 gene (see Fig. 1). The APOC4 gene consists of three exons and two introns and the total gene sequence encompasses about 4.2 kb region [1] (Fig. 1). TheAPOC4 cDNA sequence predicts a protein of 127 amino acids with 25 residues
in the signal peptide. Human apoC-IV is expressed only in liver [1,2], but it is not detected in normal plasma [3]. Although the function of apoC-IV is not clear at this stage, there is evidence that it may play a role in lipid metabolism because (1) like other apolipoproteins, it has two amphiphatic a-helical domains, which can
in-teract with lipid, and (2) the human apoC-IV in trans-genic mice is associated with very low density lipoprotein (VLDL) with expression of apoC-IV in transgenic mice causing hypertriglyceridemia [3].
To our knowledge, no genetic variation has been reported in the APOC4 gene. As common genetic variations in genes coding for apolipoproteins have been shown to affect plasma lipoprotein-lipid levels (reviewed in [4,5]), identification and examination of genetic variation in the APOC4 gene in relation to plasma lipid profile may help to understand its role in lipid metabolism. The aims of this study were: (1) to
The findings of this paper were presented at the 72nd Scientific Sessions of the American Heart Association in Atlanta, GA, USA, on November 7, 1999.
* Corresponding author. Tel.:+1-412-624-3066; fax:+ 1-412-383-7844.
E-mail address:[email protected] (M.I. Kamboh).
search for genetic polymorphisms in the coding region and intron-exon boundaries of theAPOC4 gene, (2) to evaluate the extent of linkage disequilibrium between new APOC4 polymorphisms, as well as between
APOC4 polymorphisms and the well known APOE
polymorphism in this gene cluster, and (3) to evaluate the role of this new variation in affecting plasma lipid levels in the general population.
2. Methods
2.1. PCR amplification, SSCP, DNA sequencing and restriction analyses
Exons 1 and 2 and their intron-exon boundaries were amplified in one fragment each, while exon 3 and its intron-exon sequence was amplified in two overlapping fragments (Fig. 1). Table 1 presents the primer se-quences, PCR conditions and size for each amplified fragment. Each amplification was performed using a 50
ml reaction volume containing 20 mmol/l Tris – HCl (pH
8.4), 50 mmol/l KCl, 1.0 mmol/l MgCl2, 75 mmol/l of
each dNTP (Pharmacia), 0.2mmol/l of each primer, and
1 unit ofTaqDNA Polymerase (Life Technologies). An Omn-E from Hybaid was used for PCR, which con-sisted of an initial denaturation at 95°C for 5 min and then 35 cycles of 45 s of denaturation at 95°C, 45 s of annealing at the appropriate temperature (Table 1) and 45 s of extension at 72°C.
Mutation detection was performed using cold single-strand conformation polymorphism (SSCP) analysis.
Two microlitres of the amplified DNA was diluted in 8
ml of a stop solution consisting of 95% formamide, 20
mM EDTA, 0.05% xylene cyanol, 0.05% bromophenol blue followed by addition of 2 ml of a denaturing
solution of 0.5 M NaOH and 10 mM EDTA. The solution was heated to 42°C for 5 min and then 12 ml
was loaded into a pre-cast 4 – 20% TBE gel (Novex). Vertical electrophoresis was performed at a constant 20 mA per gel for 2 h at 28 – 32°C for the exons 1 and 2 fragments and at 10 – 17°C for the exon 3 fragment, using a Novex Thermoflow gel electrophoresis appara-tus and a Polyscience Digital Temperature Controller. The gels were stained with Sybr Green II RNA Gel Stain (Molecular Probes Inc.) using a 4:50 dilution with 1X TBE for 20 min and viewed with an Eagle Eye II (Stratagene) camera system.
Amplified DNA from subjects exhibiting different SSCP patterns were then sequenced using an Open-Gene™ Automated DNA Sequencing System (Visible Genetics Inc., Toronto, Canada), Thermo Sequenase™ Cycle Sequencing Core Kit (Amersham Life Science) and primers labeled with Cy5.5 from Visible Genetic Incorporated. Nucleotide substitutions were identified by comparing our DNA sequencing data with the pub-lished sequence of the APOC4 gene [1]. Restriction enzymes for four of the five mutations sites were iden-tified (Table 1), which were subsequently used for rou-tine screening. Twenty one microlitres of the amplified products were digested with appropriate restriction en-zymes and resolved on 4 – 20% polycrylamide gel for the exon 1 variation and on 3% agarose gel for the exon 2 and 3 polymorphisms.
Table 1
Primer sequence, PCR conditions and restriction enzymes used to detectAPOC4 polymorphisms
Nucleotide Annealing Restriction fragment size
Primers (5%–3%) Size (bp) Restriction
Exon
sequence numbera temperature (°C) enzyme (bp)b
EIF AGG GCC CTG CTC 558–577 67
1 241 MnlI wt 75,43,36,31,30,10,8,5,3
TGT GCA GC
E1R ACG TGG CTA CAG 798–776 mut 75,51,36,31,30,10,5,3
AGC CAC AGG AC
2 E2F GGG ATT CTA GGG 3060–3082 63 227 A6aI wt 227
TCC CAG CCT AC
E2R AGT CCC CTC CCA 3286–3267 mut 149,78
CCA CCC AG
BsmFI wt 227 mut 139,88
E3aF CCT CCA CTG TGA 3471–3494 60
3 248 HhaI wt 248
TGT CCT CTC TCC
E3aR GTA GCT GGG ACT 3718–3697 mut 150,98
ACA GGC ACC C
E3bF TT CAT AAA AGC 3669–3694
3 60 236 no enzyme
CAG GTG TGG TTG TG
E3bR TGG GGT GGG TTT 3904–3884 TTT TCT GGC
aNucleotide sequence of each primer is based upon Allan et al [1]. bwt, wild type, mut, mutant type.
2.2. PCR amplification, subcloning into TA cloning
6ector and haploid DNA sequencing
Double heterozygotes for the amino acid polymor-phisms at codons 36, 52 and 96 were chosen to deter-mine the phase of chromosomes unequivocally. These samples were PCR amplified using the primer pair E2F and E3bR to obtain a 845 bp fragment encompassing the three polymorphic sites. PCR was performed using the following cycling conditions on a PE9600 (Perkin Elmer) thermal cycler: denaturation at 95°C for 5 min and then 35 cycles of 35 s of denaturation at 95°C, 35 s of annealing at 63°C and 35 s of extension at 72°C. A final cycle of extension was carried out for 5 min at 72°C. The PCR products were subcloned using TOPO™ TA Cloning kit (Invitrogen). Briefly, the kit takes advantage of the non-template-dependent termi-nal transferase activity ofTaq polymerase, which adds a single deoxyadenosine (A) to the 3% end of PCR products. The kit supplies a linearized vector with a single overhanging 3% deoxythymidine (T) residue that allows efficient ligation of the PCR product into the vector. Ligation of the PCR product to the vector was performed at room temperature. The pCR-II-TOPO-PCR product vector was then transformed into TOP10 competent cells. The transformed cells were plated and allowed to grow overnight at 37°C. Positive clones were selected based on their white or light blue color, placed in growth medium and incubated overnight. Plasmid
DNA was isolated and used for DNA sequencing as described above. The -40 M13 forward (23-mer) primer was used to sequence the codon 36 and 52 sites and the codon 96 site was sequenced using the E3aR primer.
2.3. Subjects
A total of 592 non-Hispanic whites from the San Luis Valley, CO, were screened for APOC4 polymor-phisms and their effects on quantitative lipid traits. The subjects were unrelated and between the ages of 20 and 74 years and had a normal response to a standard 75-g oral glucose tolerance test (1985 World Health Organi-zation criteria). The study population and design have been described elsewhere [6,7].
2.4. Statistical analysis
Allele frequencies (p, q) at each site were calculated by allele counting. Heterozygosity (gene diversity) was computed asH( =1−(p2+q2). Hardy – Weinberg
subcloning and haploid DNA sequencing. Pairwise linkage disequilibrium (D) was calculated using the equationD=h−pq[8], where h is the frequency of the haplotype with the rare alleles at both sites, p is the frequency of the rare allele at one site and q at the second site. Significance of the linkage disequilibrium between two sites was determined byx2 analysis based upon differences between observed and expected num-bers of haplotypes [9]. A general linear model (GLM) analysis was used to determine the association of geno-types and other independent variables with plasma quantitative traits. Those lipid and apolipoprotein val-ues that were not normally distributed were trans-formed (logarithm or square root). The independent variables considered were the linear effect of age, cigarette smoking and body mass index (BMI). Signifi-cance of the genotypic effects was based on the F-test conditional on all other variables being in the model (i.e. the type 3 sum of squares as reported by SAS). A nominal significance level of aB10% was used in the discussion. The contribution of each polymorphism to
Fig. 3. Restriction digestion patterns for fourAPOC4 mutant sites in exons 1, 2 and 3. The enzyme used to cut each fragment is given at far right of each gel picture. Genotypes are given above each sample. M indicates the markers used and the sizes of corresponding bands of the marker are given in bp on the left-hand side.
Fig. 2. DNA sequence flanking the mutation sites inAPOC4 exons 1, 2, and 3. Exon 1, (a) wild type sequence for positions 609 and 620, (b) mutant type sequence; Exon 2, (a) wild type sequence for position 3139, (b) mutant type sequence for position 3139, (c) wild type sequence for position 3187, (d) mutant type sequence for position 3187. Exon 3, (a) wild type sequence for position 3568, (b) mutant type for position 3568. The position of each variable site is indicated by an arrow and the number of the nucleotide.
variation in the dependent quantitative traits was esti-mated as described elsewhere [10]. All calculations were done in SAS (version 6.08).
3. Results
3.1. DNA sequence 6ariation
DNA sequencing of SSCP variant patterns seen in exon 2 identified two point mutations at nucleotide positions 3139 (TC) and 3187 (GA) (Fig. 2). The 3139 mutation affected codon 36 and predicts the re-placement of residue Leu (CTG) with Pro(CCG) in the mature apoC4 protein. Similarly, the nucleotide change at position 3187 affected codon 52 and would result in the substitution of Asp(GAC) for Gly (GGC). The missense mutations at codons 36 and 52 created restric-tion sites forA6aI andBsmFI, respectively (Fig. 3). The variant pattern of exon 3 corresponded to a missense mutation (TG) (Fig. 2) at codon 96 (nucleotide position 3568), which changed Leu(CTC) to Arg(CGC). This mutation also created a restriction site for HhaI (Fig. 3). Thus, of the 902 bp of sequence analyzed, five variable sites were identified with a frequency of about one mutation per 180 bp.
3.2. Population frequency of 6ariable sites
Following the identification of five variable sites in 50 individuals, all 592 subjects were screened to examine the population distribution of these mutations (Table 2). The frequencies of three mutant alleles at exon 1
(A609G, G620A) and exon 2 (Gly52Asp) were observed at a low frequency of about 1% each. However, vari-ants observed in exon 2 (Leu36Pro) and exon 3 (Leu96Arg) were common polymorphisms. The fre-quencies of the Leu36Pro alleles were almost identical (0.498/0.502). The frequencies of the Leu96Arg alleles were 0.643/0.357. The average heterozygosities for these two polymorphisms were 0.50 and 0.46, respectively, which is at or near the maximum heterozygosity attain-able by a diallelic system.
3.3. Linkage disequilibrium
The three amino acid polymorphisms in exons 2 and 3 are contained in a single 845 bp fragment that could be subcloned and the haploid DNA inserts could then be sequenced to determine unequivocally the phase of double heterozygotes. There was one double het-erozygote at the codon 36 (TC) and codon 52 (GA) sites, two double heterozygotes at the codon 52 (GA) and codon 96 (TG) sites and 205 double heterozygotes at the codons 36 and 96 sites. The 845 bp fragment was amplified from the three individuals double het-erozygotes at codons 36/52 and codons 52/96 and from six double heterozygote individuals at codons 36/96. Subcloning and sequencing of the double heterozygote individual at codons 36/52 revealed the presence of the TG and CA haplotypes. The two double heterozygotes at codons 52/96 showed the presence of two (GT, AG) of the four possible haplotypes (AT, AG, GT and GG), indicating a non-random association between the two sites. The six double heterozygotes sequenced at codons 36/96 demonstrated the presence of three haplotypes including, CG, CT and TT. The fourth possible haplo-type, TG, was not observed. A random combination of six alleles at codons 36 (TC), 52 (GA) and 96 (TG) would result into eight possible haplotypes: H1=TGT, H2=CAG, H3=CGT, H4=CGG, H5=TGG, H6=TAT, H7=TAG and H8=CAT. However, only the H1, H2, H3 and H4 haplotypes were observed in this study, which strongly indicate that the three amino acid polymorphisms are in linkage disequilibrium.
The maximum-likelihood pairwise linkage disequi-librium between all APOC4 polymorphisms and be-tween APOC4 and two APOE (codons 112 and 158) polymorphisms is presented in Table 3. We chose the
APOE gene because of its close proximity to the
APOC4 gene. Among theAPOC4 polymorphisms link-age disequilibrium was observed between five pairs: nt. 620/codon 36, nt. 620/codon 96, codon 36/codon 96, codon 36/codon 52 and codon 52/codon 96. None of the APOC4 polymorphisms showed linkage disequi-librium with either of the two APOE polymorphisms. However, the two APOE polymorphisms were in non-random association with each other in this sample population.
Table 2
Genotype, allele frequencies and heterozygosities (H() of APOC4 polymorphisms
Polymorphic position (nt. or amino Genotype % Allele acid change)/restriction enzyme
Exon 1: nucleotide 609 (AG) AA=580 (98.0) A=0.990
Exon 2: nucleotide 3139 (codon 36) TT=136 (23.6)
(TC) (LeuPro)/A6aI TC=302 (52.4) T=0.498 Exon 2: nucleotide 3187 (codon 52) GG=572 (99.3)
GA=4 (0.7)
Exon 3: nucleotide 3568 (codon 96)
(TG) (LeuArg)/HhaI TG=279 (48.3) G=0.357 GG=67 (11.6) H( =0.459 Total=578
Table 3
Pairwise linkage disequilibrium (D) between fiveAPOC4 and twoAPOEvariable sites
APOC4/codon 36 APOC4/codon 52 APOC4/codon 96 APOE/codon 112 APOE/codon 158 APOC4/nt. 620
−0.002 (NSa) 0.0 (NSa)
APOC4/nt.609 0.001 (NSa) 0.005 (NSa) 0.009 (NSa) 0.009 (NSa) APOC4/nt. 620 −0.005 (P=0.04) 0.0 (NSa) −0.014 (P=0.007) −0.006 (NSa) −0.0001 (NSa)
−0.002 (P=0.046)
APOC4/codon 36 −0.080 (P=0.00001) −0.007 (NSa) −0.004 (NSa)
APOC4/codon 52 −0.001 (P=0.007) 0.0 (NSa) −0.0004 (NSa)
APOC4/codon 96 −0.004 (NSa) 0.005 (NSa)
−0.007 APOE/codon 112
(P=0.024)
aNS, non-significant.
3.4. Association with quantitati6e traits
A general linear model (GLM) was used to analyze the association of the five APOC4 polymorphisms simultaneously with the two APOE polymorphisms. Since this analysis takes into account multiple compari-sons, the significant levels do not need to be adjusted. The APOE polymorphisms of codon 112 (E3/E4) and codon 158 (E3/E2) were included in the analysis be-cause of their known effects on plasma cholesterol levels. A summary of the P-values from the GLM is given in Table 4. APOC4/codon 36 and codon 96 polymorphisms showed a significant association with triglyceride levels in women. The Leu36 and Arg96 alleles were associated with elevated triglyceride levels (Table 5). The codon 36 and codon 96 polymorphisms explained 2.1 and 1.5% of the variation in triglyceride levels, respectively in women. The codon 36 polymor-phism also showed significant association with apoA-I levels in women. The APOC4/nt. 620 variable site was associated with HDL-cholesterol levels in men and Lp(a) levels in women (Table 5).
Previously we have described the association of the
APOE polymorphism with plasma cholesterol levels in this population [11]. In that analysis, variation in the
APOE gene (codons 112 and 158) was treated as one locus with three alleles (APOE*2, APOE*3 and
APOE*4). However, in this paper we have analyzed the contribution of the codons 112 and 158 polymorphisms separately. The APOE*4 (codon 112) and APOE*2 (codon 158) polymorphisms showed the expected asso-ciations with cholesterol and LDL-cholesterol levels (Table 4). TheAPOEcodon 112 (codon 158) polymor-phism explained 0.9% (4.2%) and 1.1% (4.5%) of the variation in cholesterol, and LDL-cholesterol, respec-tively. The greatest effect of the APOE polymorphism was on apoB levels where it accounted for 1.5% (codon 112) and 6.5% (codon 158) of the apoB variation. To investigate the possibility whether an interaction exists between the APOE and APOC4 polymorphisms in affecting cholesterol levels, a pairwise interaction was
examined between all polymorphisms, but no signifi-cant interaction was found (data not shown).
4. Discussion
This study was designed to search for DNA variation and to determine its role in lipid metabolism in the
APOC4 gene, which is a new member of the APO E/C1/C2 gene cluster. All three exons ofAPOC4 (614 bp) and their flanking intronic sequence (288 bp) were subjected to SSCP followed by DNA sequence. A total of five single nucleotide substitutions were identified, giving an overall frequency of one mutation per 180 bp. All variable sites were dialellic and genotype distribu-tions in Hardy – Weinberg equilibrium. Two mutadistribu-tions were observed in the non-coding part of exon 1 (nt. 609 and nt. 620), two in exon 2 (codons 36 and 52) and one in exon 3 (codon 96). All three mutations in the coding region would lead to amino acids substitutions. Among the five variable sites, only two (codon 36 and codon 96) exist at appreciable frequency, with respective het-erozygosity values of about 50 and 46%. The codon 52 variable site was extremely rare with B1% carrier frequency. The carrier frequencies of the exon 1 muta-tions were 2 and 8%. However, these frequencies are from a non-Hispanic white population and they may vary in other ethnic groups.
I
.
Kamboh
et
al
.
/
Atherosclerosis
152
(2000)
193
–
201
199
Table 4
P-values* from the general linear model analysis indicating association of genotypes and other covariates with plasma lipid and apolipoprotein levelsa
Covariate Men Women
ApoB Lp[a] Cholesterol HDL-C
Cholesterol HDL-C LDL-C TG ApoA-I LDL-C TG ApoA-I ApoB Lp[a]
0.33 0.59 0.0001 0.0001 0.0001 0.0001
0.21 0.24 0.48 0.015 0.48
0.74 0.28
0.85 Age
0.0001 0.39 0.75 0.96 0.44 0.0001 0.16 0.0001 0.40 0.53 0.79
Ln BMI 0.38 0.005 0.42
0.65 0.44 0.20 0.03 0.27 0.10 0.30
0.84 0.31
0.43 0.56
Smoking 0.91 0.44 0.25
0.64
0.84 0.33 0.92 0.19 0.23 0.46 0.63 0.60 0.43 0.82 0.21 0.25 0.40
APOC4/nt. 609
0.35 0.86 0.90 0.53 0.50 0.997 0.28 0.46 0.36 0.16 0.05
APOC4/nt. 620 0.74 0.06 0.79
0.95 0.72 0.27 0.36 0.71 0.03 0.05
0.41 0.37
0.40 0.91 0.80
APOC4/Codon 36 0.46 0.25
0.55
0.74 0.48 0.96 0.86 0.79 0.49 0.56 0.75 0.56 0.69 0.73 0.92 0.48
APOC4/Codon 52
0.89 0.47 0.15 0.31 0.38 0.08 0.25 0.11 0.35
APOC4/Codon 96 0.39 0.64 0.35 0.92 0.55
0.01 0.58 0.03 0.17 0.06 0.49 0.06
0.12 0.58
0.92 0.44
APOE/Codon 112 0.14 0.15 0.10
0.003
APOE/Codon 158 0.02 0.81 0.007 0.37 0.64 0.71 0.0001 0.01 0.0001 0.40 0.28 0.0009 0.23
Table 5
Adjusted mean (9S.E.) lipid and apolipoprotein levels among APOC4 genotypes based upon significant associations in Table 4
APOC4/nt. 620 HDL-C(mg/dl)in Lp[a](mg/dl)in women
A general linear model analysis identified a gender-specific significant association ofAPOC4/codon 36 and 96 polymorphisms with triglyceride levels in women, but not in men. The observed association of the
APOC4 polymorphisms with triglyceride levels appears to be consistent with its postulated role in triglyceride metabolism. Like apoC-I, apoC-II and apoC-III [12 – 15], the expression of human apoC-IV in transgenic mice results in a hypertriglyceridemic phenotype [3]. ApoC-IV in transgenic mice is associated mainly with VLDL [3] and therefore it is expected to play a role in triglyceride metabolism. However, the contribution of
APOC4 genetic variation in affecting plasma triglyce-ride levels is small (:2%). This paradox may be ex-plained by the fact that apoC-IV is absent in human plasma [3] and thus, evaluating its role in lipid metabolism may not be easy. The modest association of
APOC4 polymorphisms with triglycerides may also be explained if the APOC4 polymorphisms are in linkage disequilibrium with functional loci in the nearby
APOC2 gene. It is also possible that apoC-IV partici-pates in other metabolic pathways unrelated to lipid metabolism. The identification of structural variation in theAPOC4 gene may facilitate the determination of its functional role in future studies.
Finally, the contribution of APOE genetic variation in affecting plasma cholesterol levels is well known. The
APOE polymorphism (codons 112 and 158) explains about 4 – 18% of the variation in cholesterol levels in different population groups (reviewed in [16]). As the twoAPOE polymorphisms are in close proximity, they are treated as one locus with three alleles in genetic analyses. However, such analyses cannot differentiate the individual locus contribution to plasma cholesterol variation. As has been done conventionally, we previ-ously reported the association of the triallelic APOE
polymorphism with cholesterol levels in this population [11]. In this paper we have treated the codon 112 and 158 polymorphisms separately to examine their individ-ual effects. Our data show that the contribution of the
APOE/codon 158 variable site is greater than the
APOE/codon 112 variable site. This is consistent with the only published report that also reached a similar conclusion [17]. In 964 Mexican Americans from Starr County, TX, Hanis et al. [17], have shown that the codon 158 position mediates most of the observed effects associated with the triallelic APOE polymor-phism. Their data, along with ours, indicate that the two APOE loci can be treated as independent markers despite their close proximity. Our observation that the
APOE codon 158 contributes more to the variation in plasma cholesterol levels also substantiates the original observations that the cholesterol lowering effect of the
APOE*2 allele (codon 158) is two to three times the cholesterol raising effect of the APOE*4 allele (codon 112) [18].
feasible to subclone the fragment and determine un-equivocally the phase of double heterozygotes on hap-loid DNA. We used this strategy on the three amino acid polymorphic sites in theAPOC4 gene (codons 36, 52 and 96). This strategy allowed us to detect significant linkage disequilibrium between codons 36/52 and codons 52/96, which otherwise was non-significant when maximum-likelihood estimates were used. An-other factor that may influence the maximum-likeli-hood estimates of linkage disequilibrium is the relative allele frequencies at variable sites. It is noteworthy that the three variable sites (nt. 620, codon 36 and codon 96) that exhibited linkage disequilibrium using maximum-likelihood estimates are more common than the other two sites that did not show significant linkage disequi-librium. This indicates that the detection of linkage disequilibrium by using maximum-likelihood estimates largely depends on relative allele frequencies. A similar observation has also been noted in the APOA1 –C3 gene cluster where significant linkage disequilibrium was present only among those variable sites that had appreciable frequencies [8]. Another possibility is that some of these are ancient mutations and sufficient time has been elapsed for the linkage disequilibrium to de-cay. This possibility could be examined by screening other racial groups to determine if these mutations arose before the split of major human races. The two
APOE polymorphic sites were also in linkage disequi-librium, but not with any of theAPOC4 polymorphic sites. The absence of linkage disequilibrium between the
Acknowledgements
This study was supported by an NIH grants HL 44672, DK 30747 and CRC RR 00051. We acknowl-edge the technical assistance of Michael Carlin, Wei Si and Purnima Desai; the analytical assistance of Andrea McAllister, and the clerical assistance of Noel Eisel.
References
[1] Allan CM, Walker D, Segrest JP, Taylor JM. Identification and characterization of a new human gene (APOC4) in the apolipo-protein E, C-I, and C-II gene locus. Genomics 1995;28:291 – 300. [2] Allan CM, Taylor S, Taylor JM. Two hepatic enhancers, HCR.1 and HCR.2, coordinate the liver expression of the entire human apolipoprotein E/C-I/C-IV/C-II gene cluster. J Biol Chem 1997;272:29113 – 9.
[3] Allan CM, Taylor JM. Expression of a novel human apolipo-protein (apoC-IV) causes hypertriglyceridemia in transgenic mice. J Lipid Res 1996;37:1510 – 8.
[4] Breslow J. Apoliprotein genetic variation and human disease. Physiol Rev 1988;68:85 – 132.
[5] Humphries SE. DNA polymorphism of the apolipoprotein gene: their use in investigating the genetic component of hyperlipi-demia and atherosclerosis. Atherosclerosis 1988;72:89 – 108. [6] Hamman RF, Marshal JA, Baxter J, Kahn LR, Mayer EJ,
Dolean M, Murphy JR, Lezotte DC. Methods and prevalence of non-insulin-dependent diabetes mellitus in a biethnic Colorado population: the San Luis Valley Diabetes Study. Am J Epidemiol 1989;129:295 – 311.
[7] Rewers M, Shetterly SM, Hoag S, Marshall J, Hamman RF. Is the risk of coronary heart disease lower in Hispanics than in non-Hispanic whites? The San Luis Valley Diabetes Study. Eth-nicity Dis 1993;3:44 – 54.
[8] Thompson EA, Deeb S, Walker D, Motulsky AG. The detection of linkage disequilibrium between closely linked markers: RFLPs at the AI-CIII apolipoprotein genes. Am J Hum Genet 1988;42:113 – 24.
[9] Kamboh MI, Hamman RF, Ferrell RE. Two polymorphisms in the APOA-IV coding gene: their evolution and linkage disequi-librium. Genet Epidemiol 1992;9:305 – 15.
[10] Boerwinkle E, Sing CF. Bias of the contribution of single effects to the variance of a quantitative trait. Am J Hum Genet 1986;39:137 – 44.
[11] Kamboh MI, Aston CE, Hamman RF. The relationship of APOE polymorphism and cholesterol levels in normoglycemic and diabetic subjects in a biethnic population from the San Luis Valley, Colorado. Atherosclerosis 1995;112:145 – 59.
[12] Aalto-Seta¨la¨ K, Fisher EA, Chen X, Chajek-Shaul T, Hayek T, Zechner R, Walsh A, Ramakrishnan R, Ginsberg HN, Breslow JL. Mechanism of hypertriglyceridemia in human apolipoprotein (apo) CIII transgenic mice. Diminished very low-density lipo-protein fractional catabolic rate associated with increased apoC-III and reduced apoE on the particles. J Clin Invest 1992;90:1889 – 900.
[13] de Silva HV, Lauer SJ, Wang J, Simonet WS, Weisgraber KH, Mahley RW, Taylor JM. Overexpression of human apolipo-protein C-III in transgenic mice results in an accumulation of apolipoprotein B48 remnants that is corrected by excess apolipo-protein E. J Biol Chem 1994;269:2324 – 35.
[14] Shachter NS, Hayek T, Leff T, Smith JD, Rosenberg DW, Walsh A, Ramakrishnan R, Goldberg IJ, Ginsberg HN, Breslow JL. Overexpression of apolipoprotein CII causes hypertriglyce-ridemia in transgenic mice. J Clin Invest 1994;93:1683 – 90. [15] Simonet WS, Bucay N, Pitas RE, Lauer SJ, Taylor JM. Multiple
tissue-specific elements control the apolipoprotein E/C-I gene locus in transgenic mice. J Biol Chem 1991;266:8651 – 4. [16] Kamboh MI, Aston CE, Ferrell RE, Hamman RF. Impact of
the apolipoprotein E polymorphism in determining interindivid-ual variation in total cholesterol and low density lipoprotein cholesterol in Hispanics and non-Hispanic whites. Atherosclero-sis 1993;98:201 – 11.
[17] Hanis CL, Hewett-Emmett D, Douglas TC, Bertin JK, Schull WJ. Effects of the apolipoprotein E polymorphism on levels of lipids, lipoproteins, and apolipoproteins among Mexican-Ameri-cans in Starr County, Texas. Arterioscler Thromb 1991;11:362 – 70.
[18] Davignon J, Gregg RE, Sing CF. Apolipoprotein E polymor-phism and atherosclerosis. Arteriosclerosis 1988;8:1 – 21.
