Mutation in the ARH Gene and a Chromosome 13q Locus Influence Cholesterol Levels in a New Form of Digenic-Recessive Familial Hypercholesterolemia
We studied a Syrian family with 3 children who had low-density lipoprotein cholesterol (LDL) concentrations of 13.3, 12.2, and 8.6 mmol/L, respectively. Three other siblings and the parents all had LDL values <4.52 mmol/L, suggesting an autosomal-recessive mode of inheritance. The extended pedigree had 66 additional persons with normal LDL values. A genome-wide scan in the core family with 427 markers showed support for linkage on both chromosomes 1 and 13. Markers on chromosome 1 revealed a 3.07 multipoint LOD score between 1p36.1-p35, an 18-cM interval. Surprisingly, we also found linkage to 13q22-q32, a 14-cM interval, with a 3.08 LOD score. We had identified this locus earlier as containing a gene strongly influencing LDL in another Arab family with autosomal-dominant familial hypercholesterolemia and in normal dizygotic twins. We found evidence for an interaction between these loci. We next genotyped our twin panel and confirmed linkage of the 1p36.1-p35 locus to LDL (P<0.002) in this normal population. Elucidation of ARH, the LDL receptor adaptor protein at chromosome 1p35, caused us to sequence that gene. We first identified the genomic structure of ARH gene and then sequenced the gene in our family. We found an intron 1 acceptor splice-site mutation. This mutation was not found in any other family members, in 31 nonrelated Syrian persons, or in 30 Germans. Our results underscore the importance of ARH on chromosome 1 and the chromosome 13q locus to LDL, not only in families with unusual illnesses, but also to the general population.
Elevated plasma low-density lipoprotein cholesterol (LDL) concentrations, an important risk factor for atherosclerosis, are influenced by genetic factors. Elucidation of monogenic lipid disturbances has contributed greatly to our understanding of these disorders specifically and to lipid metabolism in general.1–4⇓⇓⇓ Ciccarese et al5 described families from Sardinia in whom hypercholesterolemia is inherited in an autosomal-recessive fashion (ARH). They studied 5 families and were successful in mapping an ARH locus to chromosome 15q25-q26, which they termed ARH-1. Eden et al6 and this study independently show that the autosomal-recessive familial hypercholesterolemia gene maps to chromosome 1p36.1-p35 (ARH). Recently, Garcia et al7 succeeded in cloning the gene residing at the chromosome 1p35 locus. The gene codes for a putative LDL receptor adaptor protein. In an earlier study, we found linkage of unexpectedly low LDL values in a family with autosomal-dominant familial hypercholesterolemia (FH) to a locus on chromosome 13q.8 We now report novel findings in an Arab kindred with ARH. We describe linkage to both the chromosome 1p35 and chromosome 13q loci, as well as evidence for an interaction between these loci. We identified the responsible mutation in the gene on 1p35 encoding for the LDL receptor adaptor protein and described the intron-exon structure of the gene. Finally, we show evidence of linkage in normal dizygotic twin children and their parents between LDL levels and the chromosome 1p35 locus. Our findings indicate that this gene may be of relevance to the general population and also underscore the importance of the chromosome 13q locus.
Materials and Methods
We studied a Druze family that resides in southern Syria. The Druze are a distinct ethnic group defined by descent and endogamy.9 They are concentrated in well-defined areas of the middle East, primarily in Lebanon and Syria. The family consists of a core pedigree that came to our attention because 3 children developed typical features of hypercholesterolemia, including giant tendon xanthomas. Seventy-two members of the family were recruited to participate in the study after informed consent was obtained from each individual or from a legal guardian. Total cholesterol, high-density lipoprotein cholesterol (HDL), and triglycerides were measured by an automated method after a 12-hour fast. The LDL values were calculated by means of the Friedewald formula.10 Genotyping was done with the ABI PRISM genotyping system (Applied Biosystems) as described earlier.8 All markers were amplified under a common set of PCR conditions by use of the True Allele PCR mix (PE Biosystems) containing AmpliTaq Gold DNA Polymerase and a final MgCl2 concentration of 2.5 mmol/L, with a 56, 58, and 60°C annealing temperatures. Electrophoresis and detection were done on an ABI 377 equipped with Genescan 3.1 software (Applied Biosystems). Genotyping was performed with Genotyper 2.0 software (Applied Biosystems). Genotypes were exported as a text file for subsequent linkage analysis. To test the hypothesis that an interaction takes place between chromosome 1 and 13 loci, GENEHUNTER-TWOLOCUS software was used according to Strauch et al.11 The program allowed us to perform parametric and nonparametric linkage analysis with 2 trait loci. In the parametric context, specific 2-locus models of disease may be tested.
An expanded Materials and Methods section can be found in the online data supplement available at http://www.circresaha.org.
The core pedigree consisted of 2 second-cousin parents with normal LDL values who had 6 children, as shown in Table 1. Three children had LDL values consistent with FH and had tendon xanthomas. Three other children had normal LDL values. The members of the core family stem from an extended normocholesterolemic pedigree (online Figure 1 in the online data supplement available at http://www.circresaha.org). We first excluded the LDL receptor locus by direct sequencing and the apolipoprotein (apo) B locus by genotyping locus-linked microsatellite markers. We also genotyped apo E in the core family. All the members were homozygous for apo E3. We performed a total genome scan of the core pedigree (online Figure 2). The affected probands in the core pedigree were homozygous for the marker D1S199, whereas the nonaffected siblings were heterozygous. Two-point linkage analysis yielded a logarithm of the odds (LOD) score of 2.37 for this marker at recombination fraction (θ=0.00). We genotyped the markers D1S2667, D1S186, and 12 markers in between. These markers cover a wide region that contains the tumor necrosis factor receptor superfamily member 1B (TNFRSF1B). This gene has been implicated in familial-combined hyperlipidemia.12,13⇓ To exclude TNFRSF1B, we sequenced the gene and found no mutations.
The recombination event in the individual 5.14 at D1S513 established the proximal boundary of the region as shown by the haplotypes in Figure 1. An ancestral recombination event defined the distal boundary of the region at D1S2826. We then genotyped all the family members for the markers in the region. These marker data enabled us to estimate a more accurate allele frequency, independent of that available in the Genome Database. Table 2 shows the calculated 2-point LOD score in the core family depending on the observed allele frequency. The highest 2-point LOD score was 2.61 at the marker D1S2843 (θ=0.00).
Figure 2 (top panel) shows the calculated multipoint LOD scores for all genotyped markers in the region. We examined whether or not this locus contributes quantitatively (QTL) to cholesterol, LDL, HDL, and triglyceride levels. Heterozygous persons had total cholesterol, LDL, HDL, and triglyceride concentrations that were not significantly different from the 44 individuals with the wild-type haplotypes, consistent with an autosomal recessive mode of inheritance.
The genome scan yielded an additional LOD score of 2.63 at the markers D13S1267 and D13S1240. Genotyping more markers enabled us to identify a region of homozygosity on chromosome 13. The boundaries of this region are defined by the recombination events that had occurred in the individual 4.5 at the markers D13S1300 and D13S1266. Haplotypes in the core family are shown in Figure 3. The region thus extends over 13q22-q32, a 14-cM interval. We found this result remarkable in light of our earlier finding that a cholesterol-influencing gene resides at this locus on chromosome 13.8 We therefore genotyped chromosome 13 markers in the entire pedigree. We found one additional homozygous person (5.15) who has the same haplotype as the affected offspring. This individual is also a daughter of the proband’s father, but stems from a different mother. Table 2 gives the 2-point LOD scores. The multipoint analysis yielded a LOD score of 3.08 shown in Figure 2 (bottom panel). We tested this locus as a QTL in the family; however, the locus did not have a QTL effect.
We had observed in our earlier study that the chromosome 13 locus was linked to total and LDL cholesterol in a panel of dizygotic twins and their parents from Berlin.8 We therefore also genotyped this group for the markers D1S552, D1S199, and D1S2843. This analysis yielded linkage of total and LDL cholesterol to the chromosome 1 locus. The multipoint LOD score for total cholesterol was 1.82 (P=0.0019) and 1.86 (P=0.0017) for LDL cholesterol.
In the single-locus analysis, the FH trait appeared to be recessive at both the chromosome 1 and 13 loci. All 3 affected children in the core family (individuals 5.9, 5.10, and 5.11) were homozygous by descent for a few markers at both regions. We regarded this remarkable result as a clear incentive to perform a linkage analysis with 2-trait loci. We used GENHUNTER-TWOLOCUS to test the hypothesis that only a simultaneous homozygous-mutant genotype at the trait loci on both chromosome 1 and 13 leads to the disease (“recessive and recessive”), a multiplicative 2-locus model.14 In addition to the nonparametric analysis with 2 disease loci, we performed a parametric 2-locus analysis under the assumption of this multiplicative model. We also analyzed the data under a heterogeneity model and under an additive model assuming a recessive action of both loci. The models are shown in Table 3. Because 2-locus analysis is computationally much more intensive than an analysis with only 1 disease locus, we were only able to include 4 of the 6 children of the core family. We first chose to include the 3 affected children and 1 unaffected child (5.13) in the GENEHUNTER-TWOLOCUS analysis. The nonparametric 2-locus analysis yielded a maximum nonparametric linkage (NPL) score of 8.64 (P=3.8×10−6). The P value reflects a test of the null hypothesis that both disease loci are unlinked to their corresponding marker maps. The parametric analysis yielded a maximum 2-locus LOD score of 5.41 under the multiplicative model. The heterogeneity and additive models yielded maximum LOD scores of 2.82 and 2.75, respectively. For nonparametric analysis, as well as parametric analysis under all 3 models, the maximum score occurred directly at the markers D1S2843 and D13S1267, the same position for the maximal single-locus LOD score. The results remained the same when we replaced individual 5.13 by another unaffected child; the maximal change in the LOD-score was less than 1%. To compare the results of the parametric 2-trait locus analysis to the single-locus results, we next recalculated the single-locus LOD scores for the pedigree without the individuals 5.12 and 5.14. The maximum LOD scores were 2.80 for chromosome 1 and 2.83 for chromosome 13. Hence, the 2-locus LOD score for the multiplicative model was superior to both single-locus LOD scores.
The superiority of the multiplicative model, compared with the other models, could be because of the fact that 3 of the 4 children were affected. To test this possibility, we repeated the 2-trait locus analysis using all 3 unaffected children, but only 1 affected child 5.9. The resulting LOD scores were 3.06 for the multiplicative model, 1.86 for the heterogeneity model, and 1.66 for the additive model. Again, the LOD scores did not change by more than 1% when another affected child was selected for the analysis.
We identified the genomic structure of ARH as shown in Table 4. First, we scanned the bacterial artificial chromosome (BAC) library with primers from the cDNA of ARH and found that the 121-O3 clone harbored the ARH gene. ARH-specific primers, designed based on the ARH cDNA sequence, were used for sequencing the cloned genomic DNA (BAC clone) to determine the positions of the introns and the intron-exon boundaries given in Table 4. Intron-exon boundaries were defined by comparing the genomic sequence with the cDNA sequence. ARH consists of 9 exons. The exon size ranges from 35 (exon 8) to 2066 bp (exon 9, the last exon). Exon 1 contains 21 bp of untranslated sequence followed by the initiation codon ATG. All the exon-intron boundary sequences conform to the consensus splice donor (GT) and acceptor (AG) sites, with the exception of the splice donor site of intron 8, which is GG instead of GT.
We sequenced the ARH gene in the core family and found a mutation that lies in the acceptor site of intron 1. The mutation converts the original acceptor-splice site from G→C, a transversion mutation. Figure 4A shows the intron 1/exon 2 junction in a normal proband (top) and in an affected child (bottom). The affected offspring are all homozygous for the mutation, whereas the parents are heterozygous. Figure 4B shows the mutated nucleotide sequence of the coding and noncoding strands at the intron 1/exon 2 junction. We found a restriction enzyme PvuII to cut the PCR product of exon 2 at the mutated donor-acceptor site. We used this enzyme to screen the entire family by using the exon 2 PCR product. On recognition, the enzyme cuts the 316-bp PCR product of exon 2 producing 2 fragments, 256 and 60 bp, in the homozygous state and 3 fragments, 316, 256, and 60 bp, in the heterozygous state. Figure 4C shows the status of the core family members with regard to the ARH G transversion mutation. The parents (4.5 and 3.3) are heterozygous for the mutation. The 3 affected children (5.9, 5.10, and 5.11) are homozygous. In agreement with the reported haplotypes, 2 of unaffected children (5.13 and 5.14) are carriers and 1 (5.12) is a noncarrier. We screened the entire family and found 21 individuals who were heterozygous for the mutation. These individuals are the same who share the haplotype with the affected patients for the microsatellite markers genotyped in the region. Only the 3 affected offspring in the core family are homozygous for the mutation. We also screened 31 normal Syrian individuals for the mutation and another 30 normal unrelated German individuals. None had the mutation.
We also found 3 single nucleotide polymorphisms (SNPs) in ARH gene. One variant lies in exon 6 at bp 625 of ARH cDNA converting C→T and leads to an amino acid substitution from Pro to Ser. The second SNP lies in the intron 6 at bp 28 converting A→G. The third SNP lies in exon 7 at bp 675 of ARH cDNA converting A→G but without any change in the coding sequence. To explore a possible effect of the variant in exon 6 on the phenotype, we genotyped the whole family for these 3 SNPs. Five persons were homozygous for the variant in exon 6. We also genotyped our twin panel for the 3 SNPs. The data showed no significant association between these variants and LDL levels (P<0.1).
Our linkage data provide strong evidence that ARH is influenced by 2 loci in the family in our present study, namely 1p36.1-p35 and 13q22-q32. Our findings support the recent observations reported by Eden et al.6 Furthermore, they corroborate earlier findings from our group regarding chromosome 13 as a locus for a gene influencing cholesterol concentrations.8 We also excluded the recently described autosomal-dominant FH locus on chromosome 1 and the ARH-1 locus on chromosome 15 mapped by Varret et al3 and Ciccarese et al,5 respectively.
In addition to the chromosome 1p36.1-p35 locus, we also found compelling evidence for linkage to 13q22-q32. We do not believe that this finding is related to chance. In our earlier study, we identified this locus as the site for a gene that lowers LDL.8 We studied a family with autosomal-dominant familial hypercholesterolemia that had affected heterozygous and homozygous members with lower than expected LDL values. We corroborated this locus with a linkage analysis in dizygotic twins and their parents. Eden et al6 did not report any evidence of linkage to 13q22-q32 in their study. However, the family in our present study and the family we reported earlier are both unrelated Arab families8; their genetic background may be similar.
Garcia et al7 cloned the gene responsible for ARH residing at chromosome 1p35. Their exciting finding was a new protein that is responsible for LDL receptor malfunction. The gene they identified codes for a protein that has a phosphotyrosine binding domain PTD. This domain was found to interact with the cytoplasmic tail of the LDL receptor.15 Malfunction of this interaction could account for a disabled LDL receptor. ARH is expressed in many tissues, even in those that have only a very low LDL receptor expression. Conceivably ARH has other functions; however, similar to the families described by Garcia et al,7 our affected family members have no other discernible phenotypes.
As reported in the present study, we determined the exon-intron structure of ARH. All splice sites conform to the GT/AG rule, except for intron 8, where the donor splice site GT was converted to GG. However, there are a few exceptions to this rule.16–20⇓⇓⇓⇓ We identified the intron 1 acceptor-site mutation in the affected persons of the core family. The mutation replaces the G bp of the acceptor-splice site (AG) by C. We corroborated our findings by analysis of the other family members. We also examined 31 healthy Syrians and 30 Germans. All were wild-type for the mutation. A defect in intron 1 splice-donor site will either skip exon 2 on splicing or will retain intron 1 in the processed mRNA. Both cases contain stop codons that would result in a truncated protein missing the amino acids coded by exons 2 to 9. We also found 3 SNPs in exon 6, intron 6, and exon 7. The SNP in exon 6 causes an amino acid exchange Pro→Ser. We excluded this variant when we found other homozygous family members who had normal total cholesterol and LDL levels. Moreover, we examined these variants in our twin panel and found that they were not associated with total cholesterol and LDL levels.
The parametric and nonparametric 2-locus analysis was not only done to verify that ARH is indeed governed by 2 loci, but also to gain insight into how the 2 genes may act together. Due to the restrictions on the pedigree size with GENEHUNTER-TWOLOCUS, we were not able to include all 6 children in the analysis simultaneously. We performed the analysis with all 3 affected children and 1 unaffected child. A recessive-and-recessive multiplicative model gave the strongest results, irrespective of the unaffected child selected. The model reflects that only individuals with a homozygous mutant genotype at both loci will develop the trait. Here, the 2-locus LOD score was 5.41, which was almost twice as high as the result obtained for the heterogeneity and additive models. With the latter models, the disease may be caused by a homozygous-mutant genotype at only 1 locus.
It may be argued that the multiplicative 2-locus model gave the strongest results merely because 2 unaffected individuals were not included in the analysis. We therefore repeated the 2-locus calculations with all 3 unaffected children and 1 affected child. The resulting LOD score was 3.06 for the multiplicative model, 1.86 for the heterogeneity model, and 1.66 for the additive model. Hence, even in this setting the superiority of the multiplicative model remained over the other 2 models. These results support a multiplicative action of both loci, where a homozygous mutant genotype at both chromosome 1 and 13 loci is necessary to express the disease.
Our 2-locus multiplicative model suggests that a mutation at only 1 locus is not sufficient to develop the disease. However, our results suggest that the chromosome 1 mutation has a dramatic effect on the adaptor protein. Our analysis raises the possibility that 2 proteins coded by 2 separate genes at distant loci are able to perform the same biological function. A mutation at 1 locus may disable one of the proteins, whereas the second is still available for normal function. Such a model would require disruption of both genes, disabling both proteins to express the phenotype. If our hypothesis is correct, a mutation on chromosome 1 leading to a dramatically altered protein is still in accord with the multiplicative 2-locus trait model. Our linkage and interaction analysis strongly supports the hypothesis that 2 loci are involved in this form of recessive FH in our family.
We were interested to observe linkage for the 1p36.1-p35 locus to total and LDL cholesterol in normal dizygotic twins and their parents. We verified earlier findings that LDL, HDL, total cholesterol, triglycerides, and body mass index are all strongly influenced by genetic variance.21 In the Syrian family, the chromosome 1 locus did not appear as a QTL. The effect here is autosomal-recessive and qualitative in nature. In the twins, the locus appeared as a QTL, suggesting that other variations in the responsible gene are operative. We believe that these studies are unique because we not only mapped a digenic syndrome, but also suggested its relevance in a completely unrelated, normal population. Identification of ARH and its mutations are clearly important to our understanding of cholesterol metabolism and cardiovascular risk. Elucidating the gene at the chromosome 13q locus and the basis for the interaction will have a high priority.
Hussam Al-Kateb is supported by a fellowship from the Deutscher Akademischer Austauschdienst (DAAD). We thank Atakan Aydin for his help in solving technical problems, and Athanasios Vergopolous for sequencing the LDL receptor in the family.
Original received October 29, 2001; revision received April 1, 2002; accepted April 1, 2002.
- ↵Goldstein JL, Brown MS. Familial hypercholesterolemia: identification of a defect in the regulation of 3-hydroxy-3-methylglutaryl coenzyme A reductase activity associated with overproduction of cholesterol. Proc Natl Acad Sci U S A. 1973; 70: 2804–2808.
- ↵Varret M, Rabes JP, Saint-Jore B, Cenarro A, Marinoni JC, Civeira F, Devillers M, Krempf M, Coulon M, Thiart R, Kotze MJ, Schmidt H, Buzzi JC, Kostner GM, Bertolini S, Pocovi M, Rosa A, Farnier M, Martinez M, Junien C, Boileau C. A third major locus for autosomal dominant hypercholesterolemia maps to 1p34.1-p32. Am J Hum Genet. 1999; 64: 1378–1387.
- ↵Hunt SC, Hopkins PN, Bulka K, McDermott MT, Thorne TL, Wardell BB, Bowen BR, Ballinger DG, Skolnick MH, Samuels ME. Genetic localization to chromosome 1p32 of the third locus for familial hypercholesterolemia in a Utah kindred. Arterioscler Thromb Vasc Biol. 2000; 20: 1089–1093.
- ↵Garcia CK, Wilund K, Arca M, Zuliani G, Fellin R, Maioli M, Calandra S, Bertolini S, Cossu F, Grishin N, Barnes R, Cohen JC, Hobbs HH. Autosomal recessive hypercholesterolemia caused by mutations in a putative LDL receptor adaptor protein. Science. 2001; 292: 1394–1398.
- ↵Knoblauch H, Müller-Myhsok B, Busjahn A, Ben Avi L, Bähring S, Baron H, Heath SC, Regina Uhlmannm, Faulhaber H-D, Shpitzen S, Aydin A, Reshef A, Rosenthal M, Eliav O, Mühl A, Lowe A, Schurr D, Harats D, Jeschke E, Friedlander Y, Schuster H, Luft FC, Leitersdorf E. A cholesterol-lowering gene exists on chromosome 13q. Am J Hum Genet. 2000; 66: 157–166.
- ↵Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972; 18: 499–502.
- ↵Glenn CL, Wang WY, Benjafield AV, Morris BJ. Linkage and association of tumor necrosis factor receptor 2 locus with hypertension, hypercholesterolemia and plasma shed receptor. Hum Mol Genet. 2000; 9: 1943–1949.
- ↵Geurts JM, Janssen RG, van Greevenbroek MM, van Der Kallen CJ, Cantor RM, Bu Xd, Aouizerat BE, Allayee H, Rotter JI, de Bruin TW. Identification of TNFRSF1B as a novel modifier gene in familial combined hyperlipidemia. Hum Mol Genet. 2000; 9: 2067–2074.
- ↵Trommsdorff M, Borg JP, Margolis B, Herz J. Interaction of cytosolic adaptor proteins with neuronal apolipoprotein E receptors and the amyloid precursor protein. J Biol Chem. 1998; 273: 33556–33560.
- ↵Broderick TP, Schaff DA, Bertino AM, Dush MK, Tischfield JA, Stambrook PJ. Comparative anatomy of the human APRT gene and enzyme: nucleotide sequence divergence and conservation of a nonrandom CpG dinucleotide arrangement. Proc Natl Acad Sci U S A. 1987; 84: 3349–3353.
- ↵Schroeder HW Jr, Walter MA, Hofker MH, Ebens A, Willems van Dijk K, Liao LC, Cox DW, Milner EC, Perlmutter RM. Physical linkage of a human immunoglobulin heavy chain variable region gene segment to diversity and joining region elements. Proc Natl Acad Sci U S A. 1988; 85: 8196–8200.
- ↵Alexandre D, Chuchana P, Brockly F, Blancher A, Lefranc G, Lefranc MP. First genomic sequence of a human Ig variable lambda gene belonging to subgroup I: functional genes, pseudogenes and vestigial sequences are interspersed in the IGLV locus. Nucleic Acids Res. 1989; 17: 3975.
- ↵Knoblauch H, Busjahn A, Münter S, Nagy Z, Faulhaber H-D, Schuster H, Luft FC. Heritability analysis of lipids and three gene loci in twins link the macrophage scavenger receptor to high-density lipoprotein cholesterol concentrations. Arterioscler Thromb Vasc Biol. 1997; 17: 2054–2060.