Autosomal Recessive Hypercholesterolemia Caused by Mutations in a Putative LDL Receptor Adaptor Protein

See allHide authors and affiliations

Science  18 May 2001:
Vol. 292, Issue 5520, pp. 1394-1398
DOI: 10.1126/science.1060458


Atherogenic low density lipoproteins are cleared from the circulation by hepatic low density lipoprotein receptors (LDLR). Two inherited forms of hypercholesterolemia result from loss of LDLR activity: autosomal dominant familial hypercholesterolemia (FH), caused by mutations in the LDLR gene, and autosomal recessive hypercholesterolemia (ARH), of unknown etiology. Here we map theARH locus to a ∼1-centimorgan interval on chromosome 1p35 and identify six mutations in a gene encoding a putative adaptor protein (ARH). ARH contains a phosphotyrosine binding (PTB) domain, which in other proteins binds NPXY motifs in the cytoplasmic tails of cell-surface receptors, including the LDLR. ARH appears to have a tissue-specific role in LDLR function, as it is required in liver but not in fibroblasts.

The liver is the major site of synthesis and clearance of cholesteryl ester–rich lipoproteins. More than 70% of circulating LDL is removed from the blood via hepatic LDLR-mediated endocytosis (1). In individuals with two mutant LDLR alleles (homozygous FH), the rate of clearance of LDL from the blood is decreased, resulting in hypercholesterolemia, xanthomatosis (deposition of cholesterol in skin and tendons) and premature coronary artery disease (CAD) (1). LDLR activity in cultured skin fibroblasts from FH homozygotes is also very low (1). A rare autosomal recessive form of hypercholesterolemia (ARH) that clinically resembles FH but is not due to mutations in LDLR has been described (2–9). These patients have markedly impaired hepatic LDLR function but normal or only modestly reduced LDLR function in cultured fibroblasts (35, 7, 8).

To elucidate the molecular basis of ARH we performed a whole-genome linkage study in four ARH families (Fig. 1A). Two were of Sardinian origin and had low LDL clearance rates in vivo (ARH1 and ARH2) (8), and two were of Lebanese origin (ARH3 and ARH4), including the original family described with this disorder (2). The probands of the four families were offspring of consanguineous unions and all families showed horizontal transmission of hypercholesterolemia. All affected family members were severely hypercholesterolemic and had very large xanthomas; some had premature CAD (Table 1). Plasma LDL levels tended to be lower and the onset of symptomatic CAD somewhat later in these probands than in FH homozygotes. LDLR function in cultured fibroblasts from affected family members were normal or only moderately reduced (3, 7,8), thus ruling out a diagnosis of homozygous FH.

Figure 1

Pedigrees (A), linkage analysis (B), and fine mapping (C) of the ARH gene. (A) The four pedigrees used for gene mapping (ARH1 to ARH4) are shown. ARH1 and ARH2 are Sardinians, and ARH3 and ARH4 are Lebanese. Fasting plasma total cholesterol levels (TC) are shown. (B) Distribution of lod scores in the linked region on chromosome 1. A total genome scan was performed initially in ARH1 and ARH2 and then additional markers were typed in all four families. The maximum lod score was 7.4 over a ∼1-cM region on chromosome 1. (C) Fine mapping within the linked region in ARH2. Genomic DNA was extracted from whole blood that had been collected from the deceased probands and stored at –20°C for more than 10 years, or from fresh leukocytes isolated from venous blood. The region of homozygosity shared by the affected individuals in this family is boxed. Squares, males; circles, females; double lines, consanguineous matings; filled squares, affected individuals.

Table 1

Molecular defects in ARH and clinical characteristics of probands in four families with ARH (Fig. 1). Genomic DNA was extracted from cultured fibroblasts or leukocytes. The coding regions of the gene were screened for sequence variation using SSCP and dideoxy sequencing. The nucleotides and amino acids were numbered from the A of the initiation codon (ATG). The age at the time of diagnosis is provided. The plasma cholesterol and LDL-cholesterol levels were measured by the referring physician. LDLR activity was assessed as described in the reference and is provided as a percentage of normal control fibroblasts studied simultaneously. Abbreviations: TC, fasting plasma total cholesterol; ref., reference; ins, insertion; F, female; CAD, symptomatic or documented coronary artery disease; AS, aortic murmur or stenosis; yr, years; M, male; ND, not done; NIDDM, non–insulin-dependent diabetes; Tx, treatment; MI, myocardial infarction; del, deletion; X, stop; amino acids: W, tryptophan; Q, glutamine; P, proline; H, histidine.

View this table:

Multipoint linkage analysis revealed significant linkage [logarithm of the odds ratio for linkage (lod) score, 7.4] to a 5.7-cM interval on 1p35, demarcated by the polymorphic loci D1S2864 and D1S2787 (Fig. 1B) (10). This interval overlaps to a chromosomal region on 1p35–p36 linked to ARH in two other families (11). We found no linkage to 15q25–q26, which was previously found to be associated with ARH in five Sardinian families (9), including ARH1 (Fig. 1). We refined the linked region to a ∼1-cM interval extending from D1S1152 to D1S2885 by identifying a region of homozygosity shared by all affected family members but not by a normocholesterolemic sibling in ARH2 (Fig. 1C) (12). The coding sequences of 13 genes that mapped to this interval were screened for sequence variation by PCR and the single strand conformation polymorphism (SSCP) technique (13). Two abnormally migrating bands were identified in the predicted coding sequences of a cDNA (DKFZp586D0624) in probands from ARH1 and ARH3.

The gene structure and predicted amino acid sequence of the protein encoded by this cDNA are shown in Fig. 2. The gene spans ∼25 kb and has nine exons and eight introns (Fig. 2A). The predicted amino acid sequence contains a 170–amino acid motif that shares considerable sequence similarity with the phosphotyrosine binding (PTB) domains of many adaptor proteins (14, 15) (Fig. 2B). PTB domains bind the consensus sequence NPXY, which is present in the cytoplasmic domains of several cell-surface receptors, including the epidermal growth factor receptor (16), the insulin receptor (17), nerve growth factor receptor (TrkA) (18), and the LDLR (19,20). The integrity of the NPXY sequence in the cytoplasmic tail of the LDLR is absolutely required for internalization (19,20), and the LDLR has been shown in vitro to bind other proteins containing PTB domains (21, 22).

Figure 2

Gene structure (A), predicted amino acid sequence (B), and location of mutations in ARH probands (C). (A) The cDNA for DKFZp586D0624 (GenBank Accession NumberAL117654) was prepared by PCR amplification of reverse-transcribed liver poly(A)+ mRNA. A P1 clone containing the entire gene (290N7, Incyte Genomics, Inc., Palo Alto) was used to amplify the introns and to sequence intron-exon boundaries.ARH is encoded by nine exons and spans ∼25 kb. Filled rectangles, exons; lines, introns. (B) The predicted amino acid sequences of human, mouse, and Xenopus ARH. Numbers to the right correspond to human sequence. The alignment of the inferred amino acid sequences displays 67% identity among the three proteins. The regions of amino acid identity are boxed. ARH has a highly conserved PTB domain at the amino terminus (indicated by blue shading, 89% identity). Alignment was constructed with PSI-BLAST (23). The boundaries of the PTB domain are according to Pfam 6.0 database [(35, 36); domain PF00640]. (C) Schematic representation of ARH showing the location of the mutations identified in this study.

Database searches (23) revealed orthologous proteins in mouse and Xenopus that share 89% sequence identity with the human protein in the PTB domain (Fig. 2B). Several regions in the COOH-terminal half of these protein are also highly conserved. These blocks do not appear to be shared with other proteins currently in the database. The closest paralogs of ARH are theDrosophila NUMB protein (24) and theCaenorhabditis elegans CED-6, an adaptor protein involved in cell engulfment (25). These proteins share 33% (52%) and 34% (60%) sequence identity (similarity) with the human protein, respectively.

The coding region of ARH was sequenced using genomic DNA from the affected family members of ARH1, ARH2, ARH3, and ARH4 (Fig. 1). Two different mutations that cause premature termination of translation were identified in the Sardinian families. The affected individuals in ARH1 were homozygous for a single base-pair insertion in exon 4 (Table 1) that introduces a premature termination codon at amino acid 170, truncating the protein in the terminal portion of the PTB domain. Affected individuals in ARH2 were homozygous for a nonsense mutation at codon 22. The ARH gene was sequenced in 10 other unrelated Sardinian probands. Four Sardinian patients were homozygous for the frameshift mutation in exon 4, and three were homozygous for the nonsense mutation in exon 1. The remaining three probands were compound heterozygotes for the two mutations. The finding that only two mutations account for ARH in these 12 apparently unrelated Sardinian probands probably reflects genetic drift, which has been observed for other diseases on the island (26, 27). There was significant overlap in the distribution of the mutation on the island; neither mutation was found in 50 normolipidemic Sardinians (28).

The four affected Lebanese siblings in ARH3 (Fig. 1) were homozygous for a nonsense mutation in codon 136, which stops translation in the terminal region of the PTB binding domain. Both ARH4 probands were homozygous for a missense mutation substituting a histidine for proline at amino acid 202, which is outside the PTB domain. The LDLR activity was normal in the ARH3 fibroblasts (3). Neither of the mutations found in ARH3 or ARH4 was present in 15 normolipidemic individuals from Lebanon, seven unrelated Lebanese FH homozygotes with a molecularly defined defect in the LDLR gene (29), or in 50 normocholesterolemic Caucasians. ARH probands from two other unrelated consanguineous families (ARH5 and ARH6) were homozygous for different frameshift mutations located in a string of seven guanine residues in exon 1 (Table 1). Both mutations are predicted to truncate the protein near the NH2-terminus (Fig. 2C).

We performed Northern blot analysis to assess the size and relative abundance of the ARH mRNA in cultured fibroblasts from the probands of the ARH1, ARH3, and ARH4 families (Fig. 3A). A 3.1-kb mRNA was detected in the control fibroblasts. In contrast to the LDLR mRNA, the levels of ARH mRNA were not affected by the addition of sterols to the medium. Only trace amounts of ARH mRNA were detected in the ARH1 fibroblasts, in which both ARH alleles contained the frameshift mutation in exon 4, and in the ARH3 fibroblasts, in which were homozygous for a nonsense mutation in exon 4. Normal levels of ARH mRNA were present in the ARH4 fibroblasts, which harbored a homozygous missense mutation. Studies are in progress to determine the functional effects of this mutation. ARH4 was the only family in which the parents had evidence of a possible defect in cholesterol metabolism; the father had a myocardial infarction at age 28. In one ARH family described by Norman et al. (6), both parents had moderately elevated plasma LDL-cholesterol levels. These observations raise the possibility that some ARH mutations result in codominant, rather than recessive hypercholesterolemia.

Figure 3

Expression of ARH and LDLR in fibroblasts (A) and human tissues (B). (A) Fibroblasts were plated at a density of 1.5 × 105 in 100-mm dishes and grown in 5% fetal calf serum (FCS). On day 4, the medium was changed to 10% human lipoprotein–poor serum (HLPPS) (minus sterols) or 10% FCS (plus sterols). Cells were harvested on day 5, and total RNA was isolated using RNA-STAT-60 (Tel-Test, Inc., Friendswood, Texas). Northern blotting was performed using 20 μg of RNA (37). The coding regions of ARH, LDLR, and GAPDH (as a loading control) were amplified by PCR, and the fragments were radiolabeled (Megaprime DNA Labeling System, Amersham Pharmacia Biotech, Piscataway, New Jersey) before incubation with the filter (1 × 106 cpm/ml). ARH3-P, father of proband for ARH-3. (B) Northern blot analysis of human tissues. Radiolabeled ARH, LDLR, and β-actin probes were incubated with the RNA blot (Clontech Laboratories, Inc., Palo Alto, California) in Rapid-Hyb buffer (Amersham) (1 × 106 cpm/ml).

Preliminary results using a mammalian two-hybrid system indicate that the PTB domain of ARH interacts with the cytoplasmic tail of the LDLR, and additional studies are under way to characterize the specificity and physiological significance of this interaction. PTB domains differ in their selectivity for NPXY sequences in different proteins, which allows for specificity in the biological response (30). For example, the Drosophila SHC (30) and mouse Disabled (21) adaptor proteins, bind to only a subset of NPXY sequences. ARH appears to be a close phenocopy of homozygous FH, which suggests that all clinical sequelae of ARH mutations are attributable to defective LDLR activity, and this in turn suggests that ARH binds specifically to the LDLR. However, although both ARH and LDLR appear to be nearly ubiquitously expressed (Fig. 3B), LDLR expression is relatively low in some of the same tissues that express high levels of ARH (kidney, placenta) (1), raising the possibility that this protein may be involved in other receptor pathways. None of the 16 probands examined in this study have other obvious shared phenotypes that would suggest defective signaling or functioning of NPXY-containing proteins, with the possible exception of NIDDM (non–insulin-dependent diabetes mellitus) (Table 1).

The defect in LDLR function in ARH appears to be not only receptor-specific, but also tissue-specific. We have been unable to identify a consistent defect in LDLR function (binding, uptake or internalization) in cultured fibroblasts from ARH patients. It is possible that another PTB domain protein compensates for the absence of ARH in cultured fibroblasts, or that adaptor molecules are not required for receptor-medicated endocytosis of LDL in fibroblasts and possibly other extrahepatic cells.

ARH may participate in a step in the itinerary of the LDLR pathway that is specific to polarized cells like hepatocytes (31). ARH may be required for trafficking of LDLR to the basolateral surface. In contrast to fibroblasts, LDLRs in hepatocytes do not cluster in coated pits; conceivably, ARH may target the LDLR to the coated pit after the receptor binds LDL (32). Alternatively, ARH may participate in the recycling of the LDLR from the lysosome to the basolateral cell surface after dissociating from LDL. Although the specific role of ARH in the functioning of the LDLR remains to be defined, the crucial role of this protein is revealed by the profound hypercholesterolemia that occurs in this disease.

  • * To whom correspondence should be addressed. Email: helen.hobbs{at}


View Abstract

Navigate This Article