Atherosclerosis and Lipoproteins |
From the Institute for Medical Biology and Human Genetics (H.G.K., A.L., M.H., G.U.), University of Innsbruck, Innsbruck, Austria, and the Department of Medicine (F.J.R.), University of Witwatersrand, Johannesburg, South Africa.
Correspondence to Univ Prof Dr H.G. Kraft, Institute for Med. Biology and Human Genetics, Schöpfstr 41, A-6020 Innsbruck, Austria. E-mail Hans-Georg.Kraft{at}uibk.ac.at
| Abstract |
|---|
|
|
|---|
Key Words: lipoprotein(a) apolipoprotein(a) familial hypercholesterolemia homozygous familial hypercholesterolemia low density lipoprotein receptors
| Introduction |
|---|
|
|
|---|
Unlike all other lipoproteins, Lp(a) levels show an extreme
interindividual variability and are largely controlled by variation in
1 major gene, which is the structural gene for
apo(a).11 12 The apo(a) gene is highly polymorphic.
One of the polymorphisms, the kringle IV (K-IV) polymorphism,
which gives rise to the size polymorphism of the apo(a) protein, is
responsible for a large part of the variation in Lp(a) plasma levels.
In white populations,
50% of the variation in the Lp(a)
concentration is explained by the number of K-IV repeats in the apo(a)
allele.11 The category of this effect is the same in
every population studied to date, but the size of the effect differs
between populations. The mechanism of this association was elucidated
in in vitro cell studies that showed longer retention times for larger
isoforms in the endoplasmic reticulum.13
However, isoforms with an identical number of K-IV repeats show a considerable divergence of Lp(a) levels, which is explained by sequence variation at or close to the apo(a) gene locus. Variation at the apo(a) gene locus explains from 70% to >90% of the variation in Lp(a) concentration.14 Hence, every study that intends to find additional factors/genes that affect Lp(a) levels has to consider and stratify for the strong impact of the apo(a) gene. One way to achieve this is by using sibpair analysis in which Lp(a) levels are compared in sibpairs who are identical by descent (IBD) for their apo(a) alleles. Recently, we have performed such an analysis for studying the influence of heterozygous FH on Lp(a) levels and found a significant influence of the disease state.8 In the present study, we expand this approach by analyzing families in which both parents are heterozygous for FH and which consequently also embody FH homozygous children. This scheme should allow us to determine whether there is a gene-dosage effect of the FH causing LDL-R mutations on the quantitative Lp(a) trait, thus providing further evidence for the effect of an LDL-R mutation on plasma levels of the atherogenic Lp(a).
| Methods |
|---|
|
|
|---|
Screening for Mutations in the LDL-R Gene
Genomic DNA was extracted from whole blood collected in
EDTA-containing tubes according to standard
techniques.15 16 DNA screening for 3 founder-related
Afrikaner mutations, D206E (Afrik 1), V408 mol/L (Afrik 2), and D154N
(Afrik 3), was performed in a single reaction by a multiplex
amplification refractory mutation systempolymerase chain reaction, as
previously described.17 The DNA samples were also
screened for mutation P664L (FH-Gujerat) previously identified in South
African Indians.18 After screening for familial defective
apoB, subjects negative for these mutations underwent a more extensive
search by heteroduplex and/or single-strand conformational
polymorphism analysis.17
Lp(a)/Apo(a) Analysis
Samples were shipped on dry ice to Innsbruck for Lp(a) and
apo(a) analysis. Lp(a) levels were determined by
ELISA,19 and the assignment of the plasma level to the 2
alleles of 1 individual was performed after SDSagarose gel
electrophoresis and immunoblotting as
described.8
The number of K-IV repeats in the apo(a) alleles was determined by pulsed-field gel electrophoresis and Southern blotting.20 For the latter, genomic DNA was prepared as DNA-containing agarose plugs, which were subjected to a digestion with KpnI. The fragmented DNA was size-separated in a pulsed-field gel electrophoresis CHEF-mapper (Bio-Rad) and then transferred to a nylon membrane by alkaline blotting. The visualization of K-IVcontaining fragments was performed with a DIG-labeled (Boehringer-Mannheim) cDNA probe11 followed by chemiluminescence detection. The size determination of the individual alleles was accomplished by comparison with secondary standards.
Statistical Analysis
Only nonparametric statistics were used because of
the positively skewed distribution of Lp(a) levels. To compare median
Lp(a) levels between groups, the Wilcoxon Mann-Whitney test was
used, and pairwise comparisons were performed with the Wilcoxon
matched-pairs signed rank test. A 2-tailed significance of
P<0.05 was considered to be statistically significant; for
multiple comparisons, the Bonferroni correction was applied. Apo(a)
K-IV allele frequencies were compared by
2
analysis after binning the alleles in the following way:
for group 1, the number of K-IV repeats was 15 to 20; for group 2, 21
to 25; for group 3, 26 to 30; for group 4, 31 to 35; and for group 5,
>35.
By this procedure an underrepresentation of individual cells was avoided. Other forms of binning have also been performed [small versus large apo(a) alleles and 5 groups containing equal numbers of apo(a) alleles], which did not change the outcome.
Univariate ANOVA calculations were performed to search for factors affecting Lp(a) levels. All calculations were performed by using the SPSS (version 8.0 for Windows) program.
| Results |
|---|
|
|
|---|
|
The mean and median Lp(a) levels in the homozygous versus the
heterozygous family members were 49.9 and 36.6 mg/dL versus 29.9 and
14.4 mg/dL, respectively (Table 1
). This difference was significant
(P=0.004, Wilcoxon Mann-Whitney test). Because this
difference could have been caused by a difference in the apo(a)
allele frequencies between the 2 groups, we compared the K-IV
allele frequencies between the FH homozygous and FH heterozygous
individuals. A
2 analysis of binned
apo(a) alleles by size (Table 2
)
showed that the distribution of apo(a) K-IV alleles was
indistinguishable between the 2 groups (
2
0.64, df 4, P>0.95). Also, the frequency of
"null" alleles was not different between FH homozygotes
(13.5%) and FH heterozygotes (11.5%).
|
|
Because most of the families had only 1 child, it was not possible to
compare Lp(a) levels in sibpairs IBD for apo(a) alleles but
nonidentical regarding their FH state. Instead, we compared Lp(a)
levels associated with alleles IBD, which were present either
in an FH homozygote or FH heterozygote. This was accomplished by
scanning of the apo(a) immunoblot from the apo(a)
heterozygotes, which allowed assignment of the respective fraction of
the total plasma Lp(a) concentration to each allele [allelic Lp(a)
levels]. Forty such IBD allele pairs could be drawn from the
sample. Again, higher average Lp(a) levels were found in association
with IBD alleles if these were present in FH homozygous
individuals (mean 22.6 mg/dL, median 12.55 mg/dL) as when present
in FH heterozygous individuals (mean 12.7 mg/dL, median 5.8 mg/dL).
Also, this difference was statistically significant
(P=0.005, Wilcoxon matched-pairs signed rank test).
On average, allelic Lp(a) levels were higher by 9.91 mg/dL in FH
homozygotes than in FH heterozygotes. This difference between FH
homozygotes and heterozygotes was present over the whole range of
K-IV alleles and hence was not affected by the number of K-IV
repeats [P=0.613 for the Spearman correlation coefficient
between K-IV repeat number and
Lp(a), Figure 2
].
|
Because of the structures of the families, which were ascertained
through FH homozygous children and in which both parents were FH
heterozygotes, we were unable to recruit a sufficient number of adult
nonaffected blood relatives. To allow a comparison with FH unaffected
individuals from the same South African population, we combined the
data from the present study with those from our previous study in
families with heterozygous FH.8 The nonaffected
individuals from this previous family study consisted of nonaffected
relatives (n=57) and spouses (n=43). From these 100 independent
individuals, we were able to determine 190 allelic Lp(a) values. The
box plot in Figure 4
shows allelic Lp(a)
levels for unaffected individuals (n=190), heterozygous FH individuals
(n=280), and homozygous FH individuals (n=52). Median allelic Lp(a)
levels rose from 4.95 mg/dL in unaffected individuals to 8.0 mg/dL in
heterozygous FH individuals to 13.2 mg/dL in homozygous FH individuals.
Lp(a) levels associated with individual apo(a) alleles are thus
almost twice as high in homozygous FH compared with heterozygous FH and
3 times as high when homozygous FH individuals are compared with
unaffected individuals. This difference is highly significant
(P<0.001). In addition to the average Lp(a) levels, the
variance also increased with the number of mutated LDL-R alleles in
a subject (Figure 4
). The influence of the K-IV size
polymorphism on allelic Lp(a) levels in the 3 genotypic groups is
illustrated in Figure 3
, and the values are given in Table 3
. Apo(a) alleles were again binned
into 5 groups for this presentation. In every size group,
there is a gradual increase in average allelic Lp(a) levels from
non-FH, to heterozygous FH, and further to homozygous FH. The
intragroup difference was significant for 3 groups (15 to 20, 31 to 35,
and >35 K-IV repeats). In the 2 intermediate size groups (21 to 25 and
26 to 30 K-IV repeats), the significance level was not reached. The
largest relative increase in average allelic Lp(a) levels (298%) was
seen in the largest apo(a) size group (>35 K-IV repeats), and the
largest absolute increase (34.5 mg/dL) was in the smallest group (15 to
20 K-IV repeats).
|
|
|
An additional method, univariate ANOVA, was applied to determine the factors that contribute to Lp(a) levels. This showed that Lp(a) levels in the sample of 20 families were explained significantly by 2 factors: the number of K-IV repeats in the smaller apo(a) allele (explaining 46% of the variation) and the number of mutated LDL-R alleles in a given individual (explaining 9.7% of the variation). Variables that were tested but gave no significant contribution included age, sex, K-IV repeat number of the larger apo(a) allele, and additional polymorphisms in the apo(a) gene (pentanucleotide repeat polymorphism,21 Met/Thr,22 and +93 C/T polymorphism23 ).
| Discussion |
|---|
|
|
|---|
Most previous studies have analyzed the situation in FH heterozygotes. Case-control studies that compared Lp(a) levels in heterozygotes and controls uniformly found higher Lp(a) in FH,6 25 but these findings changed when FH families were studied. Results from some studies were in line with those from the case-control studies,2 26 but others were not.3 27 Furthermore, some authors have concluded that LDL-R mutations that result in heterozygous FH do not affect Lp(a) levels. However, none of these studies controlled rigorously for the effect of the apo(a) gene locus by apo(a) genotyping. We have recently performed a large study of South African and French-Canadian families with heterozygous FH.8 Using apo(a) typing and a sibpair approach, we observed a significant effect of the FH status on Lp(a) levels, which were significantly higher in the apo(a) identical FH heterozygous siblings.
In the present study, we have extended our investigation to FH homozygotes. This not only confirmed our conclusions from the analysis of FH heterozygotes but also demonstrated a significant gene-dosage effect. Not only were Lp(a) concentrations higher in the combined FH homozygotes from the 20 families analyzed in the present study than in almost any other sample of the white population [only patients with nephrotic syndrome have been reported to have higher Lp(a)28 ], but average Lp(a) concentrations were also significantly higher in the FH homozygotes (mean 49.9 mg/dL) than in the FH heterozygotes (mean 29.9 mg/dL) from the same families.
It is noteworthy that average Lp(a) levels in South African FH
heterozygotes from our previous study8 (mean 35.4 mg/dL)
were not different (P>0.05) from what we found in the
present study in an independent sample from the same population
(Afrikaners from South Africa). This is important because it allowed us
to use Lp(a) levels in the FH heterozygotes as the common standard
against which Lp(a) levels in FH homozygotes and in non-FH blood
relatives could be compared (in our previous study, there were no FH
homozygotes included, whereas in the present study, there were no
nonaffected family members included, with both situations being due to
the ascertainment and resultant structure of the analyzed
families). When the data from our previous and the present study
were combined, we obtained a very large data set, which showed a
stepwise and significant increase from unaffected to heterozygous and
homozygous family members (Table 1
, Figures 3
and 4
).
As previously mentioned, because of the small family size, it was not possible in the present study to compare sibpairs IBD for apo(a) alleles. Instead, we compared Lp(a) levels associated with alleles IBD between groups, a strategy we also had used in our earlier study.8
There are some previous reports on Lp(a) levels in FH homozygotes, but
the numbers of patients were low, no stratification for apo(a) gene
effects was performed, and no comparison with FH heterozygotes or
nonaffected family members was made. Guo et al29
analyzed Lp(a) and apo(a) phenotypes in 8 homozygous FH
subjects and compared them with 40 normolipidemic subjects. Their
homozygous FH individuals had exactly the same elevated mean Lp(a)
level (50 mg/dL) as in the present study. They did not detect an
association of the Lp(a) level with the size of the apo(a) isoforms,
which is not surprising in view of the very low resolution of their
phenotyping method (only 3 different isoforms were detected in the 8
homozygous FH subjects). If only intermediate-sized apo(a) alleles
(21 to 35 K-IV repeats) had been detected in the present study, we
also would have seen no influence of apo(a) allele size on Lp(a)
levels (Figure 3
), but in the apo(a) heterozygous FH
individuals, Guo et al also found a higher Lp(a) level associated with
the smaller isoform.
Because of the peculiar genetic regulation and large interindividual
variation of Lp(a) levels, studies intended to detect differences in
Lp(a) concentrations between groups must obey certain rules. Because
Lp(a) levels are determined to a very large degree by the apo(a) gene,
the gold standard is to compare Lp(a) levels between individuals with
both apo(a) alleles IBD. Because this was not possible within this
group, we have used the second best approach: comparing
allele-associated Lp(a) levels of IBD apo(a) alleles
present in either homozygous or heterozygous FH individuals (Figure 1
). In addition, it is necessary to investigate sufficiently
high numbers of individuals to reduce the influence of fluctuations
created by chance. The significant gene dosage effect would have been
missed if, by chance, only individuals with very low differences (<5
mg/dL) who represented 47.5% of all pairs had been
analyzed.
Rader et al30 performed a turnover study that used
radioactively labeled Lp(a) and LDL in 5 homozygous FH individuals, 4
heterozygous relatives, and 8 normolipidemic controls. They followed
the disappearance of radioactivity from the plasma and calculated the
fractional catabolic rate (FCR). They concluded that the catabolism of
Lp(a) was not significantly different in the homozygous FH individuals
compared with the heterozygous parents and controls. This is certainly
true for the homozygous FH patient 2 shown in Figure 2
of their study,
but reanalysis of the kinetic parameters in
Table 1
of Rader et al30 gave a different
result. The FCR of Lp(a) was significantly reduced in homozygous and
heterozygous FH individuals when compared with normal controls
(P=0.020). The effect of mutations in the LDL-R gene on FCR
of Lp(a) was not as pronounced as it was on the FCR of LDL, but it was
present! This new finding is also more compatible with a further
finding of that study. Rader et al showed that a fraction of the Lp(a)
appears in the LDL density range and that this newly Lp(a)-derived LDL,
also designated as Lp(a-), has the same reduced FCR in FH individuals
as does LDL. If this conversion of Lp(a) to LDL is a significant part
of Lp(a) catabolism and is not due to the labeling treatment, as
suggested earlier by Knight et al,31 then one should
also expect a reduced FCR of Lp(a) in FH subjects.
Together, our data clearly demonstrate that average Lp(a) concentrations in FH homozygotes are in a range above the 90th percentile of Lp(a) in healthy white populations and twice as high as in FH heterozygotes. This leaves little doubt that LDL-R mutations that result in FH with elevated LDL also result in hyperlipoprotein(a).
The challenging question now is what the mechanism underlying this effect may be. In vivo the LDL-R seems to play a minor role in the uptake of Lp(a), although this is still under debate. Possibly, the elevated levels of Lp(a) in FH are not a consequence of the LDL-R defect but result from poorly understood metabolic changes in FH. These might include the metabolism of fatty acids, which is known to affect lipoprotein synthesis32 or the so-called direct synthesis of LDL.33 Potentially, elucidation of this mechanism will shed light not only on the regulation of Lp(a) metabolic pathways but also on secondary metabolic disturbances in FH.
| Acknowledgments |
|---|
Received March 10, 1999; accepted August 20, 1999.
| References |
|---|
|
|
|---|
leu mutation in the LDL receptor gene.
J Clin Invest. 1991;88:483492.
This article has been cited by other articles:
![]() |
G. Miltiadous, V. Saougos, M. Cariolou, and M. S. Elisaf Plasma Lipoprotein(a) Levels and LDL-Cholesterol Lowering Response to Statin Therapy in Patients with Heterozygous Familial Hypercholesterolemia Ann. Clin. Lab. Sci., January 1, 2006; 36(3): 353 - 355. [Abstract] [Full Text] [PDF] |
||||
![]() |
S. Canizales-Quinteros, C. A. Aguilar-Salinas, E. Reyes-Rodriguez, L. Riba, M. Rodriguez-Torres, S. Ramirez-Jimenez, A. Huertas-Vazquez, V. Fragoso-Ontiveros, A. Zentella-Dehesa, J. L. Ventura-Gallegos, et al. Locus on Chromosome 6p Linked to Elevated HDL Cholesterol Serum Levels and to Protection Against Premature Atherosclerosis in a Kindred With Familial Hypercholesterolemia Circ. Res., March 21, 2003; 92(5): 569 - 576. [Abstract] [Full Text] [PDF] |
||||
![]() |
S. Gonbert, B. Saint-Jore, P. Giral, C. Doucet, J. Chapman, and J. Thillet Molecular Analysis of Apo(a) Fragmentation in Polygenic Hypercholesterolemia: Characterization of a New Plasma Fragment Pattern Arterioscler. Thromb. Vasc. Biol., August 1, 2001; 21(8): 1353 - 1358. [Abstract] [Full Text] [PDF] |
||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
ATVB Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 2000 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |