| Human Molecular Genetics | Pages |
Significant impact of the +93 C/T polymorphism in the apolipoprotein(a) gene on Lp(a) concentrations in Africans but not in Caucasians: confounding effect of linkage disequilibrium
Introduction
Results
C/T allele frequencies
Allelic associations of C/T with apo(a) K-IV repeats and PNRs
Effect of the C/T polymorphism on Lp(a) concentrations in Caucasians
Effect of the C/T polymorphism on Lp(a) concentrations in Africans
Discussion
Materials And Methods
Primer sequences
Statistical procedures
Acknowledgements
References
 |
Significant impact of the +93 C/T polymorphism in the apolipoprotein(a) gene on Lp(a) concentrations in Africans but not in Caucasians: confounding effect of linkage disequilibrium
Lipoprotein(a) [Lp(a)] is a quantitative genetic trait in human plasma associated with atherothrombotic disease. The major determinant of Lp(a) concentration is the apolipoprotein(a) [apo(a)] gene locus. Variation in the number of kringle IV repeats (K-IV VNTR) in apo(a) has a direct effect on Lp(a) concentrations but explains only a fraction of the large intra- and inter-population variance in Lp(a) levels. Effects on Lp(a) of other intragenic polymorphisms including a pentanucleotide repeat (PNRP) in the promoter likely reflect allelic associations with as yet unidentified sequence variation in the apo(a) gene. We have studied a candidate C[rarr]T transition in two European and two African populations. This polymorphism in the 5[prime] region of the apo(a) gene creates an ATG start codon thereby reducing apo(a) translation in vitro by 60%. All samples were also analyzed for the K-IV VNTR and the PNRP to stratify for their effects and to consider allelic associations. Consistent with the in vitro effect the C[rarr]T transition was associated with a significant reduction in Lp(a) levels in both African populations (P < 0.0056). In Caucasians, however, the effect wasnot significant. This was explained by linkage disequilibrium of the +93 T with apo(a) alleles of intermediate length (K-24-K-34) and with nine PNRs. In Europeans these alleles are associated with low Lp(a) which makes any potential effect of the +93 T undetectable in the total sample. From our results we conclude (i) that the +93 C/T polymorphism is the second known intragenic apo(a) polymorphism which affects Lp(a) levels directly in vivo; (ii) that allelic associations may mask the effect of a mutation; and (iii) that heterogeneity of an effect of a mutation across populations does not disprove causality.
INTRODUCTION
High plasma levels of lipoprotein(a) [Lp(a)] are a major susceptibility factor for atherothrombotic disease (coronary heart disease and stroke) in Caucasians and Asians (1). Lp(a) is a complex particle consisting of a low density lipoprotein (LDL) to which the plasminogen-related apolipoprotein(a) [apo(a)] is covalently bound by a single disulfide bridge (2-4). Although Lp(a) levels are considered remarkably stable within an individual they vary >1000-fold between individuals and their distribution also varies considerably between different populations (5,6). Twin and sibpair analyses (7-9) in Caucasians have revealed that the variation of Lp(a) plasma concentration is under genetic control and almost completely determined by variation at the apo(a) gene locus which explains >90% of the variability in Lp(a) concentrations (7,8,10). Several polymorphisms have been identified within the apo(a) gene which affect Lp(a) plasma levels. A strong and direct influence has been attributed to the size polymorphism of the apo(a) glycoprotein (11) which is determined by a variable number of 5.6 kb repeat units in the apo(a) gene containing exons coding for protein motifs called kringle IV type 2 [K-IV VNTR] (12). The number of K-IV type 2 repeats is negatively associated with Lp(a) plasma concentration and this effect has been detected in every population studied to date (5,13,14). Together with in vivo turnover studies of apo(a) (15) and of apo(a) metabolism in baboon primary hepatocytes (16) and Hep G2 cells transfected with recombinant apo(a) (17), this suggests that the number of K-IV repeats in the apo(a) gene directly affects synthesis of apo(a) by the liver (18). Depending on the population sample the K-IV VNTR does, however, explain only 30-70% of the total variability in Lp(a) concentrations and apo(a) alleles with identical K-IV-2 repeat number may cosegregate with dramatically different Lp(a) concentrations (19). Hence there must be further variation in the apo(a) gene.
Table Recently a pentanucleotide repeat polymorphism (PNRP) located 1.4 kb upstream of the apo(a) signal sequence within the promoter region of the apo(a) gene has been identified which also affects Lp(a) levels (20-22). Further, some intronic SSCP polymorphisms are also associated with Lp(a) concentrations (19). The effect of all these variations is not direct (20) and likely reflects linkage disequilibria of these polymorphisms with unidentified mutations in apo(a) which directly affect apo(a) synthesis and thereby Lp(a) concentration (21). As one candidate we have analyzed a C/T polymorphism in the 5[prime] untranslated region of the apo(a) gene. The C[rarr]T transition at position +93 of the transcription start site creates a new translation start codon and was found to reduce apo(a) translation in vitro by 60% (22).To test if this variation has an impact on Lp(a) plasma levels in the population we determined the C/T polymorphism in two Caucasian and two African populations which had also been characterized for the K-IV VNTR and the promoter PNRP. This enabled us to estimate the C/T effect on Lp(a) levels after stratification for these other variables and under consideration of possible allelic associations.
Population (n)
Frequencies
P-value
Genotypes
Alleles
(HWE)
C/C
C/T
T/T
C
T
Variance
Austrians (133)
102
30
1
0.880
0.120
3.97 × 10-4
>0.7
Danes (96)
68
28
0
0.854
0.146
6.49 × 10-4
>0.2
Black South Africans (213)
176
37
0
0.913
0.087
1.87 × 10-4
>0.3
Khoi San (58)
44
12
2
0.862
0.138
1.03 × 10-3
>0.5
RESULTS
C/T allele frequencies
The C/T polymorphism was analyzed by an ARMS (23) method in two Caucasian and two African populations: Austrians (n = 133), Danes (n = 96), black South Africans (n = 213) and Khoi San (n = 58). The allele frequencies were estimated by gene counting and are shown in Table 1. The observed numbers of C/C and T/T homozygotes and C/T heterozygotes were in agreement with expectations assuming Hardy-Weinberg equilibrium. In three populations (Austrians, Danes and Khoi San) the C/T allele frequencies were comparable with the frequencies reported in the initial publication for white North Americans (22) whereas in black South Africans the frequency of the T allele was significantly lower (P < 0.01).
Figure
Allelic associations of C/T with apo(a) K-IV repeats and PNRs
The distributions of the K-IV VNTR and the PNRP alleles in C/C homozygotes and C/T heterozygotes from all four populations (Austrians and Danes were pooled as Caucasians) are illustrated in Figure 1A. The distributions are significantly different for both polymorphic groups in Caucasians (K-IV VNTR, P < 0.01; PNRP, P < 0.001) and black South Africans (K-IV, P < 0.001; PNRP, P < 0.001) but not in the Khoi San. From the data a direct estimation of the K-IV repeat and PNR frequencies on C alleles can be performed using the C/C homozygotes. Since only few T/T homozygotes were present in the four population samples no direct calculation of K-IV and PNR allele frequencies on T alleles was possible. To get an estimate of the K-IV repeat frequency distribution on T alleles the K-IV repeat frequency on C alleles present within the C/T heterozygotes was estimated from their distribution in the group of C/C homozygotes. Using this distribution and, where necessary, binning with the next neighbor, the C allele from each single C/T heterozygote was assigned to one K-IV repeat allele. This at the same time assigned the T allele to the other K-IV repeat in this individual. The resulting K-IV repeat frequencies on T and C alleles are shown in Figure 1B. The same procedure was also employed to estimate the PNR allele frequencies on C and T alleles (lower panels of Fig. 1B).
The graphs demonstrate that the T at +93 is under-represented on small apo(a) alleles (<23 K-IV repeats) in Caucasians. No T alleles were found on small apo(a) K-IV repeat alleles compared with 16.7% on the C alleles. This difference is highly significant ([chi]2 = 11.98; df = 1; P < 0.001). Most T alleles were associated with K-IV alleles of intermediate size (26-34 repeats; Fig. 1). A linkage disequilibrium between the C/T site and the K-IV VNTR existed also in the black Africans where T alleles were strongly associated with K-IV repeats 21, 27, 29 and 31.
An even stronger and consistent linkage disequilibrium was detected between the C/T and the PNRPs in all populations except Khoi San. Whereas the majority of C alleles from Caucasians (73.9%) had eight (TTTTA) repeats, the T alleles were predominantly associated with nine repeats (64.4%). This was true for both Caucasian samples (data not shown). This difference was also highly significant ([chi]2 = 51.796; df = 3; P < 0.001).
An allelic association of the same type and magnitude was present in the black South Africans ([chi]2 = 30.37, df = 5, P < 0.001; see Fig. 1B) but not in the Khoi San.
In addition we tested for the allelic association using the EH (Estimating Haplotypes) program of the LINKAGE package (28). This maximum likelihood procedure yielded significant linkage disequilibria for all three loci at the apo(a) gene (PNRP, C/T and K-IV polymorphism; P < 0.001) as well as separately between C/T and PNRP (P < 0.001) and C/T and K-IV polymorphism (P < 0.01), respectively.
Effect of the C/T polymorphism on Lp(a) concentrations in Caucasians
The mean and median Lp(a) concentrations in Austrians, Danes and the pooled Caucasians are shown in Table 2 separately for the three genotypic groups. There was only one T/T homozygote in the Caucasian sample which was not considered for further analysis. The differences in Lp(a) concentrations between C/C homozygotes and C/T heterozygotes were not statistically significant for Danes (P = 0.843) and for the pooled sample (P = 0.122). It was of borderline significance in the Tyroleans (P = 0.0465). In both populations and in the pooled sample the T allele was associated with the lower mean Lp(a) plasma concentrations which would be expected under the hypothesis generated from the in vitro data. The small difference between the C/C and C/T groups was, however, almost entirely explained by the frequency distribution of the apo(a) K-IV alleles among the groups [see expected mean Lp(a) levels in Table 2]. Univariate ANOVA using Lp(a) concentration or log transformed Lp(a) concentration as dependent variable demonstrated that only the K-IV VNTR (P < 0.001) and the PNRP (P < 0.025) but not the C/T polymorphism had an effect on Lp(a) levels.
Figure
We further plotted log transformed Lp(a) levels against the sum of K-IV repeats in both apo(a) alleles of each subject separately for the three genotypic groups (Fig. 2, lower panel). If the T allele had an effect on Lp(a) levels a shift in the regression line of C/T heterozygotes versus C/C homozygotes might be expected. This was not observed. The regression lines were almost identical for the two groups. This analysis did not consider that in C/T heterozygotes with one short K-IV repeat and high Lp(a) concentration the C rather than the T is likely to be associatedwith the short allele (Fig. 1B) and high Lp(a) levels.
As demonstrated above, the C/T site the K-IV repeats and the PNRP are in allelic association, e.g. a T at +93 is associated with higher K-IV repeat copy numbers and PNR = 9 both of which are associated with low Lp(a) levels. Such allelic association might mask an effect of the C/T polymorphism on Lp(a) concentration. Thus no effect of the C/T polymorphism on the in vivo Lp(a) concentration could be demonstrated by different types of analysis in the Caucasian sample despite its considerable size.
Table 2.
Population
Austrians
Danes
Pooled Caucasians
Black Africans
Khoi San
Pooled Africans
Genetype
C/C
C/T
T/T
C/C
C/T
T/T
C/C
C/T
T/T
C/C
C/T
T/T
C/C
C/T
T/T
C/C
C/T
T/T
Lp(a) mean
18.3
11.7
0.5
16.9
13.7
-
17.8
13.8
0.5
25.9
15.2
-
38.5
22.9
23.9
28.2
17.1
23.9
SD
22.9
15.2
-
20.5
24.6
-
22.1
19.7
-
28.0
17.1
-
28.8
24.3
12.5
28.4
19.1
12.5
Lp(a) median
8.9
5.8
0.5
6.5
10.0
-
7.8
7.7
0.5
18.5
8.04
-
32.2
14.2
23.9
19.8
8.9
23.9
P-value
0.04651
0.8432
0.1221
0.00172
0.0331
0.00561
Expected mean3
18.1
13.1
3.0
17.0
12.5
-
18.1
13.2
3.0
27.1
26.3
-
33.5
38.1
42.8
28.4
29.2
42.8
Effect of the C/T polymorphism on Lp(a) concentrations in Africans
As expected, mean and median Lp(a) levels in the South African blacks (24.1 mg/dl, 16.3 mg/dl) and Khoi San (39.5 mg/dl, 32.2 mg/dl) were significantly higher than in Caucasians (16.7 mg/dl, 7.9 mg/dl) (5,29). As in the European populations Lp(a) levels were lower in C/T heterozygotes compared with C/C homozygotes but in both African groups this difference was larger and it was significant. In the Khoi San and in the South African blacks Lp(a) levels were almost twice as high in the C/C group as compared with the C/Ts (Table 2). In contrast to the Caucasians this was not caused by differences in K-IV VNTR allele frequencies between the two groups. Direct comparison of K-IV VNTR allele frequency distribution did not show any difference (black South Africans: P = 0.982, Khoi San: P = 0.376).
Further, when we calculated by linear regression expected Lp(a) levels derived from the K-IV VNTR allele frequencies the expected Lp(a) was practically identical with the observed value in the C/C homozygotes whereas in the C/T group the expected value was higher than the observed and corresponded to the expected and observed values in C/C homozygotes. The two T/T homozygotes in this group also showed a much lower Lp(a) concentration than expected from their number of K-IV repeats. Together this demonstrates that the observed differences in Lp(a) concentration are true effects of the C/T site and not caused by frequency differences of K-IV or PNR alleles.
In a next step we plotted the concentration of Lp(a) in an individual against the sum of the K-IV repeats from both alleles separately for C/C and T/T homozygotes and C/T heterozygotes.
The graph in Figure 2 illustrates that C/T heterozygotes on average had lower Lp(a) concentrations than C/C homozygotes with a similar K-IV repeat number. The number of T/T homozygotes was too low to compare the regression lines.
DISCUSSION
Since high plasma Lp(a) levels are considered as an independent genetic risk factor for atherothrombotic disease their genetic regulation has been the focus of numerous studies. Turnover studies have revealed that differences in Lp(a) concentrations are determined almost exclusively by the synthetic rate rather than by differences in catabolism (30,31). Therefore one might speculate that factors which influence apo(a)'s translation might also affect Lp(a) plasma levels. The presence of a T at the polymorphic C/T +93 site in the 5[prime] untranslated region of the apo(a) gene (22) creates a new start codon which is followed by a stop codon soon thereafter. In vitro studies have shown that the presence of the false start codon at +93 leads to a 60% reduction of apo(a) translation (22). We here have analyzed whether the +93 C/T polymorphism has an effect on plasma Lp(a) levels in vivo by performing an epidemiological study including two Caucasian (Tyroleans from Austria, Danes) and two African populations (South African blacks, Khoi San). Because of the known strong causal effect of the apo(a) K-IV VNTR which is present in all populations, and the weak association of a 5[prime] PNRP with Lp(a) in Caucasians (20), these polymorphisms were also considered in our analysis. This turned out to be essential for the interpretation of our findings.
In both African ethnic groups, the Khoi San and the black South Africans, Lp(a) concentrations were significantly lower in C/T heterozygotes as compared with C/C homozygotes (Table 2). The reduction in C/T heterozygotes from both groups was close to that expected from the in vitro data (22). Assuming that in C/C homozygotes as a group both alleles on average contribute equally to Lp(a) concentration the mean C-allele-associated Lp(a) concentration is [sim]14 mg/dl in the pooled African population (Table 2). If mutation to a T at +93 results in a 60% reduction on one allele the average Lp(a) concentration in the C/T group is expected to be [sim]19 mg/dl which is not far from the 17.1 mg/dl observed. This rough calculation assumes that there are no allelic associations of the C/T polymorphism with other intragenic variations affecting Lp(a) concentrations e.g. the K-IV VNTR and 5[prime] PNRP. This was true for the Khoi San (Fig. 1) but not for the other groups.
The reduced Lp(a) levels in the C/T heterozygotes might be spurious reflecting random differences in K-IV alleles between groups or even a systematic difference due to allelic association. Lp(a) concentrations were, however, still significantly lower in African C/T heterozygotes when the K-IV VNTR was considered. When we calculated Lp(a) levels in the C/C and C/T groups from the respective frequencies of K-IV VNTR alleles by linear regression analysis two results emerged which clearly demonstrated that the reduced Lp(a) levels in African C/T subjects are caused by the presence of the T and not by differences in K-IV VNTR alleles. First, observed and expected Lp(a) values did not differ significantly in C/C homozygotes (P = 0.727). Second, expected mean Lp(a) did not differ significantly between C/C and C/T groups (P = 0.515). Thus the significantly lower Lp(a) observed in the C/T genotype is not explained by a different K-IV VNTR allele frequency of this group. Direct testing also demonstrated that K-IV VNTR frequencies are not significantly different between the African C/C and C/T groups (P = 0.999).
In view of these clear results in Africans, the findings in Caucasians at first view were surprising and in apparent conflict with the results in Africans. No significant effect of the C/T polymorphism was observed in either of the two Caucasian population samples nor in the pooled Caucasian sample (n = 229) when appropriately analyzed. Although it is hard to envisage that a 60% reduction of apo(a) translation in vitro cannot be detected in vivo in Caucasians as opposed to Africans there are arguments to explain such a scenario.
The initial analysis of the C/T polymorphism without consideration of the K-IV VNTR (and the PNRP) had suggested a mild effect on Lp(a) levels in particular in Tyroleans as predicted from the in vitro data (22). C/T heterozygotes had lower average Lp(a) concentrations. The impression from this analysis is, however, misleading. As discussed above two potential sources of bias exist if the K-IV VNTR and PNR polymorphisms are not considered. First, there may be chance differences in K-IV or PNR allele frequencies between the C/T polymorphic groups which could either generate a false effect or mask a true effect. However, even with different kinds of analyses in which frequency differences were considered no measurable influence on Lp(a) could be detected for the T in position +93. The slight difference in observed Lp(a) concentrations between the C/C and C/T groups (Table 2) was entirely explained by differences in the K-IV allele frequencies among groups. Measured Lp(a) concentrations in the groups were not significantly different from those expected according to their K-IV VNTR frequencies from regression analysis (Table 2).
Second, there might exist allelic associations of the C/T site with the K-IV VNTR and/or the 5[prime] PNR. An allelic association of +93 T with intermediate/large K-IV repeats and a PNR 9 was indeed observed here. Such alleles are associated with low Lp(a) in Caucasians. On one hand this association might falsely suggest a lowering effect of the +93 T on Lp(a) levels and even result in an overestimation of the +93 effect, if present. Our analysis suggests that the slightly lower observed and expected mean in Caucasian C/T heterozygotes is explained by the higher frequency of intermediate/large K-IV repeats and possibly the PNR 9 in this group which is not a result of chance but rather of the allelic association of +93 T with intermediate K-IV repeats and with the PNR 9. This is consistent with the finding that Lp(a) is lower in the C/T Caucasian heterozygotes but that there is no difference between observed and expected values in this group. On the other hand this association might also mask a true biological effect of the C/T polymorphism. The allelic association of the +93 T with intermediate/large K-IV repeats suggests that it affects the expression of a fraction of alleles which produce very low concentrations of Lp(a) anyhow. Such an effect may not be detectable in the whole sample. The +93 T alleles represent only 13% of all alleles in the sample. They are present only in heterozygotes and in association with intermediate/high copy numbers of K-IV repeats. They were absent from small K-IV alleles associated with high Lp(a). In such a situation even an effect which might be large for the individual allele may be small for the total sample and hence undetectable. This explains why no effect of the C/T polymorphism on the in vivo Lp(a) concentration can be demonstrated in the analyzed sample of Caucasians despite its considerable size. A further hint that the effect of the C/T polymorphism is present but masked in the Caucasians is provided by the fact that the only T/T homozygote in this group had a several fold lower Lp(a) concentration than expected.
Why, then, is a strong +93 T effect present in black Africans in which +93 T is also in linkage disequilibrium with the K-IV VNTR and PNR 9 and the frequency of the +93 T allele is even lower than in Caucasians? The reason is likely threefold. First, K-IV VNTR alleles of a length of 21 and 27-31 repeats are not associated with low Lp(a) in Africans; secondly, the PNR is not associated with Lp(a) levels in black Africans; and thirdly Lp(a) levels are significantly higher in African populations than in Caucasians. Together this explains why the allelic associations, though present, do not mask the effect of the C/T polymorphism in black Africans.
Together, three different situations were present in the analyzed groups with respect to the effect of the C/T polymorphism on Lp(a) levels: (i) in Khoi San there was a strong effect and no allelic association; (ii) in black Africans there was also a strong effect despite allelic association of C/T with other polymorphisms; and (iii) there was no effect in Caucasians probably due to allelic association.
The data presented here also have some general implications. In association studies heterogeneity of the effect of a mutation across populations is generally considered as an indication that the mutation is not causally involved in the generation of the effect. Furthermore, presence of identical effects in evolutionarily distant populations is taken as indirect evidence for a causal relationship. The mutation in apo(a) analyzed here contradicts these paradigms. A mutation which has a demonstrated biological effect in vitro had a large impact on the phenotype [Lp(a) levels] in vivo in one group but not in another. We demonstrate that masking by linkage disequilibrium is the likely reason.
Second, we demonstrate that strong linkage disequilibria may be conserved over a relatively long evolutionary distance. The allelic association of +93 T with PNR 9 which are 1.4 kb apart in populations as diverse as Danes and South African blacks is remarkable and suggests that this haplotype is old, predating the split between Africans and Caucasians. It further suggests that the 5[prime] PNRP in apo(a) is remarkably stable.
MATERIALS AND METHODS
EDTA blood was obtained from unrelated healthy Caucasian individuals from the province of Tyrol, Austria (n = 133), from Aarhus, Denmark (n = 96), and from unrelated black subjects (n = 213) and unrelated Khoi San (n = 58) from South Africa. These samples have been described previously (5). Genomic DNA was prepared as DNA containing agarose plugs exactly as described (5) and stored in 0.5 M EDTA (pH = 8.0) at 4°C. The +93 C/T polymorphism was analyzed by an ARMS (amplification refractory mutation system) (23) based method. A plug containing [sim]3.5 µg genomic DNA was washed twice in TE buffer and melted in 50 µl H2O. Two PCR reactions were made with each of these DNA samples mixing 0.5 µl of this mixture with 40 pmol of primer CTUP, 4 pmol of primer CTLO and 12.5 pmol of either C- or T-specific primer. One unit of Taq polymerase (Dynazyme) was added after a hot start (98°C for 3 min). Apo(a) specific primers were chosen using the published sequence (22) and employing the program OLIGO (Medprobe, Norway).
Primer sequences
Upper: CTUP 5[prime]-AAA GGC AAT GTG GAG CAG CTG AG
Lower: CTLO 5[prime]-TGA ATT GCA CAT AAA GCC ATG GC
C-specific: 5[prime]-GTC CCA ATC CCA GGA CG
T-specific: 5[prime]-GTC CCA ATC CCA GGA CA
The PCR program consisted of two steps: 10 cycles with an annealing temperature of 65°C were followed by 20 cycles with an annealing temperature of 55°C. In the first step only the larger primers CTUP and CTLO annealed and generated an apo(a)- specific product of 395 bp. In the second step, depending on the sequence present in the sample, an amplification product of 225 bp was generated employing the smaller C- or T-specific primer as reverse primer together with CTUP as forward primer. The generation and the size of the fragments were analyzed in a 2% agarose gel after staining with ethidium bromide. The amplification of the 395 bp product which had to be present in each sample served as an internal control in the assay. The amplification of the 225 bp product with the C- (T-) specific primer indicated the presence of a C(T) at position +93 in the genomic DNA. Homozygotes yielded PCR products only in one of the two tests whereas heterozygotes gave products in both trials. The assay was validated by sequencing the PCR fragments of one homozygote (C/C) and one heterozygote (C/T). In the heterozygote the sequencing reaction was performed individually for the two alleles which had been separated in a pulsed field gel after digestion of genomic DNA with KspI (24). For all other alleles we used the statistical method of estimating the phase of the C/T polymorphs with the other markers as described below because physically isolating all alleles from pulsed field gels would have been too cumbersome. Lp(a) concentration, the apo(a) K-IV VNTR and the PNRP have been determined previously for all subjects and have been published elsewhere (20).
Statistical procedures
Mean and median Lp(a) levels between C/C homozygotes, C/T heterozygotes and T/T homozygotes were first compared directly using the non-parametric Kruskal-Wallis One-Way ANOVA. Since the K-IV VNTR and the PNRP together explain a large fraction of the variation of Lp(a) plasma concentration it was, however, necessary to stratify for the effect of these variables when estimating the effect of the C/T polymorphism on Lp(a) levels. To compensate for the influence of the K-IV polymorphism we calculated expected mean Lp(a) levels using the K-IV allele frequencies as described (25). Additionally, univariate analyses of variance (ANOVA) (26) were performed using the sum of the K-IV repeats in the two alleles, the sum of the pentanucleotide repeats, and the C/T polymorphism as variables and Lp(a) concentration or log transformed Lp(a) concentration as dependent variables. In a further analysis the log transformed Lp(a) concentrations were plotted against the sum of K-IV repeat numbers in both alleles and the linear regression was calculated separately for the C/C and C/T phenotypic groups.
Allele frequencies were estimated by gene counting which are also the maximum likelihood estimates.The distribution of the K-IV alleles between C/C homozygotes and C/T heterozygotes was compared using a t-test since the K-IV alleles are normally distributed. The comparison of the PNR allele frequencies in the C/C and C/T groups was performed using a [chi]2 goodness of fit test (27). This test was also used to test if the C/T polymorphism was in Hardy-Weinberg equilibrium in the populations.
Linkage disequilibria between the C/T polymorphism and the K-IV VNTR and the PNRP in the apo(a) gene were calculated using the EH program (version 1.11) from the LINKAGE package (28). To compile this program on a PC the maximum number of alleles at the K-IV repeat locus had to be reduced to seven and this was achieved by binning. Apo(a) alleles containing 11-15 K-IV repeats were summarized in group 1, alleles comprising 16-20 repeats were in group 2 and so forth until group 7 which contained all alleles with >40 K-IV repeats. To illustrate the allelic associations of K-IV and PNR repeats with C and T alleles, we calculated the expected frequencies of K-IV and PNR alleles among the C alleles of C/T heterozygotes from the respective distribution in the C/C homozygotes. When the phase of the C alleles had thus been determined the remaining K-IV and PNR alleles were then assigned to T alleles. The obtained distributions are represented graphically (Fig. 1B). Except for the EH program, all calculations were performed using the SPSS package (Windows version 6.1.2).
ACKNOWLEDGEMENTS
The excellent technical work of S. Rauchenwald and S. Höfle is gratefully acknowledged. DNA from the Danish subjects was kindly provided by L. Lemming (Aarhus, Denmark), and blood from the African populations was a generous gift from R. Delport and H. Vermaak (Pretoria, South Africa). We thank M. Knapp (Bonn, Germany) for helpful discussions. This work was funded by grant P 11695-MED from the Austrian Science foundation (Fonds zur Förderung der wissenschaftlichen Forschung in Österreich) to G.U.
REFERENCES
*To whom correspondence should be addressed. Tel: +43 512 507 3451; Fax: +43 512 507 2861; Email: hans-georg.kraft@uibk.ac.at
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 6 Jan 1998
Copyright© Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Ober, A. S. Nord, E. E. Thompson, L. Pan, Z. Tan, D. Cusanovich, Y. Sun, R. Nicolae, C. Edelstein, D. H. Schneider, et al. Genome-wide association study of plasma lipoprotein(a) levels identifies multiple genes on chromosome 6q J. Lipid Res., May 1, 2009; 50(5): 798 - 806. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. M. Luke, J. P. Kane, D. M. Liu, C. M. Rowland, D. Shiffman, J. Cassano, J. J. Catanese, C. R. Pullinger, D. U. Leong, A. R. Arellano, et al. A Polymorphism in the Protease-Like Domain of Apolipoprotein(a) Is Associated With Severe Coronary Artery Disease Arterioscler Thromb Vasc Biol, September 1, 2007; 27(9): 2030 - 2036. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J-P Chretien, J Coresh, Y Berthier-Schaad, W H L Kao, N E Fink, M J Klag, S M Marcovina, F Giaculli, and M W Smith Three single-nucleotide polymorphisms in LPA account for most of the increase in lipoprotein(a) level elevation in African Americans compared with European Americans J. Med. Genet., December 1, 2006; 43(12): 917 - 923. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Pepe, G. Chimienti, G. M. Liuzzi, B. L. Lamanuzzi, M. Nardulli, F. Lolli, E. Angles-Cano, and S. Mata Lipoprotein(a) in the Cerebrospinal Fluid of Neurological Patients with Blood-Cerebrospinal Fluid Barrier Dysfunction Clin. Chem., November 1, 2006; 52(11): 2043 - 2048. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Berglund and R. Ramakrishnan Lipoprotein(a): An Elusive Cardiovascular Risk Factor Arterioscler Thromb Vasc Biol, December 1, 2004; 24(12): 2219 - 2226. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Matthes, P. Aguilar-Martinez, L. Pizzi-Bosman, R. Darbellay, L. Rubbia-Brandt, E. Giostra, M. Michel, T. Ganz, and P. Beris Severe hemochromatosis in a Portuguese family associated with a new mutation in the 5'-UTR of the HAMP gene Blood, October 1, 2004; 104(7): 2181 - 2183. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. M. Marcovina, M. L. Koschinsky, J. J. Albers, and S. Skarlatos Report of the National Heart, Lung, and Blood Institute Workshop on Lipoprotein(a) and Cardiovascular Disease: Recent Advances and Future Directions Clin. Chem., November 1, 2003; 49(11): 1785 - 1796. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Huby, V. Afzal, C. Doucet, R. M. Lawn, E. L. Gong, M. J. Chapman, J. Thillet, and E. M. Rubin Regulation of the Expression of the Apolipoprotein(a) Gene: Evidence for a Regulatory Role of the 5' Distal Apolipoprotein(a) Transcription Control Region Enhancer in Yeast Artificial Chromosome Transgenic Mice Arterioscler Thromb Vasc Biol, September 1, 2003; 23(9): 1633 - 1639. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. R. Holmer, C. Hengstenberg, H.-G. Kraft, B. Mayer, M. Poll, S. Kurzinger, M. Fischer, H. Lowel, G. Klein, G. A.J. Riegger, et al. Association of Polymorphisms of the Apolipoprotein(a) Gene With Lipoprotein(a) Levels and Myocardial Infarction Circulation, February 11, 2003; 107(5): 696 - 701. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Rubin, F. Paultre, C. H. Tuck, S. Holleran, R. G. Reed, T. A. Pearson, C. M. Thomas, R. Ramakrishnan, and L. Berglund Apolipoprotein [a] genotype influences isoform dominance pattern differently in African Americans and Caucasians J. Lipid Res., February 1, 2002; 43(2): 234 - 244. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Rubin, T. A. Pearson, R. G. Reed, and L. Berglund Fluorescence-based, Nonradioactive Method for Efficient Detection of the Pentanucleotide Repeat (TTTTA)n Polymorphism in the Apolipoprotein(a) Gene Clin. Chem., October 1, 2001; 47(10): 1758 - 1762. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Ogorelkova, H. G. Kraft, C. Ehnholm, and G. Utermann Single nucleotide polymorphisms in exons of the apo(a) kringles IV types 6 to 10 domain affect Lp(a) plasma concentrations and have different patterns in Africans and Caucasians Hum. Mol. Genet., April 1, 2001; 10(8): 815 - 824. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. G. Kraft, A. Lingenhel, F. J. Raal, M. Hohenegger, and G. Utermann Lipoprotein(a) in Homozygous Familial Hypercholesterolemia Arterioscler Thromb Vasc Biol, February 1, 2000; 20(2): 522 - 528. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. M. Marcovina, H. Kennedy, G. B. Bon, G. Cazzolato, C. Galli, E. Casiglia, M. Puato, and P. Pauletto Fish Intake, Independent of Apo(a) Size, Accounts for Lower Plasma Lipoprotein(a) Levels in Bantu Fishermen of Tanzania : The Lugalawa Study Arterioscler Thromb Vasc Biol, May 1, 1999; 19(5): 1250 - 1256. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Fontana, V. Mooser, P. Bovet, C. Shamlaye, B. Burnand, V. Lenain, S. M. Marcovina, W. Riesen, and R. Darioli Dose-Dependent Inverse Relationship Between Alcohol Consumption and Serum Lp(a) Levels in Black African Males Arterioscler Thromb Vasc Biol, April 1, 1999; 19(4): 1075 - 1082. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Huby, C. Dachet, R. M. Lawn, J. Wickings, M. J. Chapman, and J. Thillet Functional Analysis of the Chimpanzee and Human apo(a) Promoter Sequences. IDENTIFICATION OF SEQUENCE VARIATIONS RESPONSIBLE FOR ELEVATED TRANSCRIPTIONAL ACTIVITY IN CHIMPANZEE J. Biol. Chem., June 15, 2001; 276(25): 22209 - 22214. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||