One Locus with Multiple Alleles
As of August 1, 1998, 421 LDL receptor mutations had been identified in subjects with FH. Of these 421 mutations, 68 are large insertions, deletions, or rearrangements that are detectable by genomic blotting (http://www.ucl.ac.uk/fh/). In most populations, large structural rearrangements comprise 2 to 10 percent of the LDL receptor mutations.201– 203 The remaining 353 mutant alleles include bp substitutions, small deletions, and small insertions; these are compiled online at two Web sites: http://www.ucl.ac.uk 204,205 and http://www.umd.necker.fr.206,207 The locations of each of the 313 point mutations and small deletions/insertions relative to the exon/intron structure of the LDL receptor gene are shown in Fig. 120-22.
Location of 353 point mutations and small deletions/insertions (<25 bp) in the LDL receptor gene in individuals with FH (compiled on August 1, 1998). Exons are shown as vertical boxes and introns as the lines connecting them. The map is drawn approximately to scale.
Classification of Mutations
LDL receptor mutations can be divided into five classes based on their phenotypic effects on the protein (Fig. 120-23). Class 1 mutations fail to produce immunoprecipitable protein (null alleles). Class 2 mutations encode proteins that are blocked, either completely (Class 2A) or partially (Class 2B), in transport between the ER and the Golgi apparatus (transport-defective alleles). Class 3 mutations encode proteins that are synthesized and transported to the cell surface, but fail to bind LDL normally (binding-defective alleles). Class 4 mutations encode proteins that move to the cell surface and bind LDL normally, but are unable to cluster in clathrin-coated pits and thus do not internalize LDL (internalization-defective alleles). Class 5 mutations encode receptors that bind and internalize ligand in coated pits, but fail to discharge the ligand in the endosome and fail to recycle to the cell surface (recycling-defective alleles). These classifications are somewhat arbitrary since many LDL receptor alleles fall into more than a single class. For example, some transport-defective alleles are also recycling- and/or binding-defective.
Classification of LDL receptor mutations based on abnormal function of the mutant protein. These mutations disrupt the receptor's synthesis in the ER, transport to the Golgi complex, binding of apoprotein ligands, clustering in coated pits, and recycling in endosomes. Each class is heterogenous at the DNA level. (From Hobbs et al. 140 Used by permission of Annual Review of Genetics.)
For most LDL receptor mutations, no biosynthetic or functional studies of the receptor protein have been performed. The seventh edition of this textbook (see pages 1997 to 2001)208 contain tables listing the 127 LDL receptor mutations that have been characterized in sufficient detail (i.e., analysis of cellular LDL uptake, receptor biosynthesis, receptor transport, and gene structure) to allow classification according to the above scheme. These mutant alleles include 109 mutations identified in cultured fibroblasts obtained from skin biopsy specimens of 157 unrelated FH homozygotes and 13 FH heterozygotes that were collected over a 20-year period (1972 to 1992) in our laboratory. These cell strains are designated the Dallas collection.198,208 In each cell line, LDL receptor biosynthetic and functional activity studies were performed, the promoter and coding regions of the receptor gene were screened for sequence variations, and the size and quantity of the receptor mRNAs were determined. Thus, the functional effects of the molecular defects in the LDL receptor gene can be deduced for many alleles in the Dallas collection, but not for most of the other LDL receptor mutations in the current online database.
Class 1 Mutations: Null Alleles.
Class 1 fibroblasts exhibit less than 2 percent of the normal amount of high affinity binding, uptake, and degradation of 125I-LDL, which is the lowest amount that can be reliably detected. In these cells, LDL does not suppress HMG-CoA reductase activity or cholesterol synthesis, nor does it activate cholesteryl ester formation.3 Fig. 120-24 shows the striking biochemical differences in these LDL-mediated processes in fibroblasts derived from a normal subject and a homozygote with two null alleles. Ultrastructural studies using ferritin-labeled LDL have confirmed the absence of LDL receptors in these receptor-negative cells.131
Actions attributable to the LDL receptor in fibroblasts from a normal subject (•) and from a homozygote with the receptor-negative form of FH (Δ) incubated with varying concentrations of 125I-LDL or unlabeled LDL at 37°C for 5 h. Assays were performed in growing cells in monolayers.391 All data are normalized to 1 mg of total cell protein. The units for each assay are as follows: Binding, μg of 125I-LDL bound to cell surface; internalization, μg of 125I-LDL contained within the cell; hydrolysis of Apo B-100, μg of 125I-LDL degraded to 125I-monoiodotyrosine per hour; hydrolysis of cholesteryl esters, nM of [3 H]cholesterol formed per hour from the hydrolysis of LDL labeled with [3 H]cholesteryl linoleate; cholesterol synthesis, nM of [14C]acetate incorporated into [14C]cholesterol per hour by intact cells; cholesterol esterification, nM of [14C]oleate incorporated into cholesteryl [14C]oleate per hour by intact cells. (From M. S. Brown and J. L. Goldstein, Proc Natl Acad Sci USA 76:3330, 1979. Used by permission.)
Twenty-seven of 157 FH homozygote fibroblast strains in the Dallas collection (17 percent) produce no immunoprecipitatable LDL receptor protein, and thus have two null alleles of the Class 1 type209 (and unpublished data). In the Dallas collection, a significantly higher percentage of the Class I FH homozygotes are true homozygotes as compared with the overall sample—85 versus 41 percent—which is consistent with the relative rarity of the null alleles in the general population.
Multiple molecular defects in the LDL receptor gene produce null alleles. These include deletions that remove the LDL receptor promoter and thus produce no mRNA (example: FH French Canadian-1 210) as well as nonsense, frameshift, and splicing mutations.
Class 2 Mutations: Transport-Defective Alleles.
The rate of transport of the LDL receptor from the ER to the Golgi can be measured in pulse-chase experiments in which fibroblasts are incubated with [35 S]methionine, and the receptor is isolated by immunoprecipitation and SDS polyacrylamide gel electrophoresis.124 The normal receptor is synthesized in the ER as a partially glycosylated precursor that migrates with an apparent molecular mass of 120 kDa. When the receptor reaches the Golgi, the N- and O-linked sugars are processed so that the receptor migrates with an apparent mass of 160 kDa.124,211 In normal fibroblasts, this transport and processing occur within 60 min after synthesis. In Class 2 fibroblasts, transport to the Golgi is markedly delayed or abolished, and the apparent molecular mass of the receptor remains at 120 kDa.124,142,211 Eventually, the receptor is degraded without ever having reached the cell surface.212 Two subsets of Class 2 mutations are recognized.140 The more common Class 2B alleles are “leaky” in that a variable portion of the newly synthesized receptor is transported at a reduced rate to the Golgi and then to the cell surface. Class 2A alleles produce a protein that fails to be transported out of the ER so there is no detectable 160-kDa mature form of the receptor.
Class 2A and 2B alleles contain missense mutations or short in-frame deletions that partially or completely disrupt folding of the receptor. The failure of surface transport suggests that cells have a fail-safe mechanism that detects misfolded proteins and prevents their movement from the ER to the Golgi.154 A similar conclusion has been reached from the study of other cellular and viral proteins that follow the same transport pathway.213
A total of 85 (54 percent) of the 157 FH homozygotes in the Dallas collection have at least one Class 2 allele, making it the most common phenotype in the sample. Most Class 2 defects cluster in the exons that encode the ligand-binding domain, and most of the remainder are in the EGF precursor homology domain (see Fig. 120-22). The mutations are not evenly distributed among the ligand-binding repeats. Repeat 5 (the last repeat encoded by exon 4) contains more mutations than any other repeat. This suggests that repeat 5 occupies a crucial structural position so that any alteration in its sequence interferes with folding. The preferential involvement of this repeat may also reflect selection bias. In vitro mutagenesis studies showed that repeat 5 is the only repeat in which missense mutations reduce the binding of lipoproteins containing apo E as well as apo B-100.144 Thus, Class 2B receptors with missense mutations in repeat 5 that reach the cell surface should interfere with plasma clearance of IDL as well as LDL and produce a more severe phenotype than do mutations in other repeats (discussed below).
In the Dallas collection, of the 32 Class 2 mutations within the ligand-binding domain, 29 involve amino acid residues that are highly conserved between multiple animal species.156 A more recent analysis of point mutations at the LDL receptor locus revealed that 74 percent of the mutations in the ligand-binding domain involved highly conserved amino acid residues that tended to be located in the 3′ half of the repeat.207
Insight into the mechanism by which sequence variations in the ligand-binding domain lead to a Class 2 phenotype has emerged from the crystal structure of repeat 5.21 Amino acid substitutions of any of the six residues that coordinate the binding of Ca2+ in the repeat interfere with the folding of the protein (Fig. 120-19). Short in-frame deletions or insertions of part of a repeat, such as occurs in FH Lithuania (G197del),140,214 change the spacing between highly conserved cysteine residues, thus interfering with disulfide bond formation. This is in contrast to the deletion of an entire cysteine-rich repeat, such as the FH Paris-1 (del exon 5)215 where repeat 6 is deleted, which does not impair its transport to the cell surface.
Most of the remaining Class 2 alleles bear mutations in the EGF precursor homology domain, and many of these map to one of the three growth factor repeats (A, B, and C). All missense mutations in the C repeat in the Dallas collection produce a Class 2 phenotype. The FH Lebanese (C660X) allele is a nonsense mutation in the third of six cysteine residues in repeat C.216 The mutant gene produces a reduced amount of receptor protein with an apparent molecular mass of 100 kDa that lacks the O-linked sugar, membrane spanning, and cytoplasmic domains. The truncated protein is apparently misfolded, and it fails to reach the Golgi.212,216 In vitro mutagenesis studies showed that the observed lack of transport of the mutant receptor out of the ER cannot be attributed simply to mispairing of the first two cysteine residues; when the first two cysteine residues were mutated to serines, the receptor protein was still sequestered in the ER.212 The remainder of Class 2 mutations in the EGF precursor domain are located in the region between the B and C growth factor repeats, which contains the six YWTD repeats that form the compact β-propeller structure.
Approximately 20 percent of the Class 2 mutations in the Dallas collection produce a complete block in transport of the receptor protein out of the ER (Class 2A), and most of these mutations are located within the EGF precursor homology domain. This region of the human receptor protein is 84 percent identical to the same region in the Xenopus receptor156 and is the second most highly conserved sequence of all domains in the receptor protein. Considered together, these findings suggest that this domain has a compact structure that is easily disrupted if only a single amino acid is changed.
In addition to FH, several other inborn errors of metabolism are caused by mutations that produce transport-defective proteins.140,217 These mutations involve genes for membrane proteins, such as leukocyte adhesion protein, sucrase-isomaltase, the insulin receptor, the cystic fibrosis transmembrane conductance regulator (CFTR), and rhodopsin. They also involve secretory proteins, including α1-antitrypsin, fibrillin, pro-α1(I) collagen, and pro-α2(I) collagen. Together with the observations on the LDL receptor, these findings suggest that a transport-defective phenotype may frequently underlie mutations involving cell-surface or secreted proteins. This high frequency is attributable to the ability of the cell to detect small regions of misfolding within a protein and thereby to interdict its movement to the cell surface.154,213
Class 3 Mutations: Binding Defective Alleles.
Many of the mutant alleles at the LDL receptor locus produce proteins that exhibit more than one type of functional defect. For example, in many Class 2B alleles, the receptor that reaches the cell surface is also defective in ligand binding. The failure of transport precludes study of the ligand-binding capacity, and thus the overall frequency of Class 3 alleles is difficult to estimate. Some mutant alleles produce receptors with defective ligand binding but normal intracellular transport, and thus are designated Class 3. Most of these alleles contain in-frame rearrangements in the cysteine-rich repeats of the ligand-binding domain or in the adjacent growth factor repeats of the EGF precursor domain.
Deletions of individual ligand binding repeats have different effects on the ability of the receptor to bind LDL and β-VLDL. For example, in FH Paris-1 (del exon 5) repeat 6 is deleted in its entirety, and the mutant LDL receptor binds β-VLDL with normal affinity but fails to bind LDL.215
In vitro mutagenesis studies have shown that replacement of a single conserved amino acid in a ligand-binding repeat produces the same functional abnormality as does deletion of the entire repeat.144 Thus, when the isoleucine adjacent to the third cysteine residue in each of the seven repeats was individually changed to an aspartic acid, the resultant receptors had the same binding characteristics as did receptors lacking each of those repeats.144 FH Puerto Rico (S156L)218 and FH French Canadian-3 (E207K),219 two slowly processed Class 2B alleles, have replacements in the highly conserved DxSDE sequence in repeats 4 and 5, respectively. The FH Puerto Rico (S156L) receptor fails to bind LDL, but binds β-VLDL with near-normal affinity.218 Substitution of a lysine for glutamic acid in the DxSDE sequence of repeat 5 as in the FH French Canadian-3 mutation results in reduced amounts of both β-VLDL and LDL binding.219 These results match the results obtained when repeats 4 and 5, respectively, were deleted in vitro144 and are consistent with the suggestion that certain missense mutations have deleterious effects on only a single repeat.
Normal binding of LDL, but not β-VLDL, requires at least one cysteine-rich repeat in the EGF precursor homology domain.148,220 If the A repeat or the A plus B repeats are deleted (as in FH Leuven-1 and FH Cape Town-2) (del exons 7 and 8), β-VLDL binding is maintained, but LDL binding is markedly reduced.220,221 Deletion of repeat B alone has no effect on binding of either ligand.220 If the entire EGF precursor domain is deleted, as in FH Osaka-2 (del exons 7–14),222 the receptor is unable to bind LDL but continues to bind β-VLDL.148 If such a truncated receptor is denatured in SDS, subjected to electrophoresis under nonreducing conditions, and transferred to nitrocellulose, it recovers its ability to bind LDL.148 These observations suggest that the EGF precursor homology domain plays a role in allowing the ligand binding domain to gain access to LDL on the cell surface in addition to its role in receptor recycling (discussed below).
Insertions within the ligand-binding domain of the LDL receptor can also produce a Class 3 phenotype. The best-characterized insertion, FH St. Louis (dup exons 2–8), encodes a receptor protein that is 50 kDa larger than normal (apparent molecular mass of 210 versus 160 kDa on SDS gels).223 The elongated protein results from a 14-kb duplication of exons 2 to 8 (discussed below) and contains 18 contiguous cysteine-rich repeats instead of the usual 9. Seven of these duplicated repeats are derived from the ligand-binding domain, and two come from the EGF precursor homology region. Remarkably, the elongated receptor is transported normally to the cell surface where it binds reduced amounts of LDL223 and β-VLDL (unpublished data). Thereupon, it undergoes efficient internalization and recycling.223 Thus, the only effect of the extensive duplication is a reduction in ligand binding.
Class 4 Mutations: Internalization-Defective Alleles.
Studies with the electron microscope show that the Class 4 receptors are distributed diffusely over the surface of the cell and are not concentrated in coated pits.142,224 They, therefore, cannot carry bound LDL into the cell. These rare mutations are of considerable historical interest because they provided the earliest evidence that cell-surface receptors must cluster in clathrin-coated pits in order to carry ligands into cells.3,224 Six internalization-defective alleles were characterized through cloning and confirmed by functional studies. All have mutations that alter the 50-amino acid cytoplasmic domain.152,225– 228 The mutations were subclassified into two groups, based on whether the mutation involves the cytoplasmic domain alone (Class 4A) or the cytoplasmic domain together with the adjacent membrane-spanning region (Class 4B).140
Two internalization-defective mutations, FH Bahrain (W792X) and FH Paris-3 (N796fs),225 have premature stop codons that leave the membrane-spanning region intact, but leave the receptor with only the first 2 or 6 of the normal 50 amino acids in the cytoplasmic tail, respectively. Another Class 4 mutation, the J.D. allele (FH Bari allele; Y807C), has a single base pair change that substitutes a cysteine for a tyrosine at amino acid residue 807 in the cytoplasmic domain.152 The latter observation stimulated a series of in vitro mutagenesis experiments, which revealed that position 807 must be occupied by an aromatic amino acid (tyrosine, phenylalanine, or tryptophan) for internalization to occur normally.229
Tyr807 is part of a tetrameric sequence NPVY (Asn-Pro-Val-Tyr) that is conserved in LDL receptors from six species.230 A variant of this sequence, NPxY (where x can be any amino acid), is present in one or more copies in the cytoplasmic tails of other members of the LDL receptor gene family, including LRP,231 megalin/gp330 protein (Heymann nephritis antigen),232 the VLDL receptor,233 and the chicken vitellogenin receptor234 (discussed below). Extensive in vitro mutagenesis experiments confirmed the importance of the NPxY sequence as a signal for directing the LDL receptor to coated pits.230 An NPxY sequence is also present in the cytoplasmic domains of at least 13 other cell-surface receptors, including several with tyrosine kinase activity (EGF, c-erb-B/neu, insulin, insulin-like growth factor 1), the β-subunits of the integrin receptors, and the amyloid precursor protein.230
Sequences other than NPxY are responsible for localization of other receptors in coated pits. In the transferrin receptor and the mannose 6-phosphate receptor, the signal for coated-pit clustering involves sequences of four to six amino acids that contain at least one aromatic residue, usually a tyrosine, but lack the other elements of the NPxY signal.235 Computer modeling236 and MRI studies237 suggest that the common feature in all of these sequences is the ability to form a tight turn that orients the aromatic residue in a fixed configuration. Presumably, the coated-pit assembly proteins recognize the aromatic residue within this tight turn, and they thereby move the receptor to a coated pit.238
The second subclass of internalization-defective alleles (Class 4B) produces truncated receptors that lack the membrane-spanning domain as well as the cytoplasmic tail. Most of these molecules are secreted from the cell, but approximately 10 percent remain adherent to the cell membrane where they bind LDL but do not internalize it, thus giving rise to the internalization-defective phenotype. Three deletion mutations— FH Rochester,226 FH Osaka-1,227 and FH Helsinki 228 —have this phenotype. All of the deletions extend from intron 15 to the noncoding region of exon 18, but each has a different end point. In each case, the truncated intron 15 is not removed by splicing. Translation of the mRNA continues into intron 15, and this produces an abnormal C-terminal sequence that contains 55 novel amino acids.227 This sequence includes a cluster of 14 hydrophobic amino acids that may constitute a pseudotransmembrane domain that anchors some of the receptors to the cell membrane.227 Inasmuch as this novel sequence lacks the NPxY signal, it is unable to direct the receptor to coated pits, hence the lack of internalization.
Class 5 Mutations: Recycling-Defective Alleles.
In addition to its role in facilitating the binding of LDL, the EGF precursor homology domain mediates the acid-dependent dissociation of receptor and ligand in the endosome, an event that is essential for receptor recycling.148 This function was discovered when the EGF precursor homology domain was deleted by in vitro mutagenesis of an expressible LDL receptor cDNA. The truncated LDL receptor bound and internalized β-VLDL, but it failed to release the ligand in the acidic environment of the endosome. The receptor was then degraded, apparently because it was unable to return to the surface in an unoccupied state.148 A naturally occurring mutation, FH Osaka-2 (del exons 7–14), with the same deletion and phenotype was described by Miyake et al.222 (discussed above).
Deletion of the A and B growth factor repeats produces a receptor with the same ligand-binding properties as FH Osaka-2 (that is, ability to bind β-VLDL but not LDL).220 van der Westhuyzen et al.221 demonstrated that such a mutant receptor, FH Cape Town-2 (del exons 7 and 8) is more rapidly degraded even in the absence of ligand binding, which suggests that the A and B repeats play a role in receptor recycling.
A total of 22 percent of the fibroblast strains in the Dallas collection have a Class 5 phenotype. The relative number of mutant alleles with this phenotype has been underestimated in prior analyses, because Class 5 mutations can produce a phenotype that superficially resembles Class 3 mutations in terms of deficient LDL binding. The Dallas collection contains 11 Class 5 alleles that are caused by missense mutations, all of which occur in the EGF precursor homology domain. Springer149 noted a disproportionately high number of mutations that are clustered in the region of the β-propeller domain that is predicted to lie in close physical proximity with EGF repeats B and C.
To date, only five naturally occurring mutations (four single-bp changes and one 3-bp deletion) have been mapped to the transcriptional regulatory elements of the LDL receptor gene (Fig. 120-20).198,239– 241 FH Pedi-2 has a 3-bp deletion, CTT(-92), involving the final C of the consensus sequence of repeat 1, which is adjacent to an Sp1 binding site. The mutation was found in a compound heterozygote whose other LDL receptor allele has a frameshift mutation in exon 2 (FH Pedi-1; E37fs).198 Functional assays performed on cultured fibroblasts from this homozygote revealed that the amount of LDL binding, uptake, and degradation was ~10 percent of normal (unpublished data). The other four mutant alleles, FH Columbia-2 [C(-42)G], C(-43)T, FH Albuquerque [C(-44)T], and C(-45)T, each contain single-bp substitutions in repeat 3.198,240,241 These mutations (demarcated by shaded boxes in Fig. 120-20) are located in the most conserved sequence of the Sp1 recognition site (CCGCCC). Changes at these sites are expected to reduce markedly the binding of Sp1, as was documented for one of these mutations that was found in two heterozygotes, one from Finland and the other from Denmark.239,240 In the Finnish study, a fragment from the LDL receptor promoter containing the C(-43)T nucleotide substitution was transfected into cultured cells. Transcription of a reporter gene was reduced by 95 percent when compared with the wild-type promoter.240 In the Danish study, the amount of endogenous mRNA generated from the C(-43)T mutant allele was less than 5 percent of that produced from the wild-type allele.239
The relative paucity of mutations within the promoter region of the gene is probably attributed to the small number of base pairs that are crucial for promoter function (<50).
General Characteristics of LDL Receptor Gene Mutations
Distribution of Point Mutations.
Analysis of the distribution of 353 point mutations and small deletions/insertions in the LDL receptor gene reveals a higher-than-expected frequency of mutations in exon 4 (Fig. 120-22). One-fifth of all the mutations (78 of 353) are located in exon 4, which is larger than all the other exons and encodes ligand-binding repeats 3, 4, and 5. Within exon 4, the region that encodes the fifth ligand-binding repeat is particularly enriched in mutations. As mentioned previously, the overrepresentation of mutations within this region may be due to selection bias because these mutations tend to be associated with more severe disease.242 Mutations in exon 15 (clustered O-linked sugar domain) are underrepresented in the sample (Fig. 120-22). This region is the least-conserved domain of the receptor and, not surprisingly, contains only one nonsense and one splicing mutation. Deletion of this domain has no effect on LDL receptor function in fibroblasts118 and is associated with only mild hypercholesterolemia in humans (FH Espoo; del exon 15).243
The actual nucleotide sequence of a given repeat may also predispose to mutations. Three different duplications ranging in size from 8 to 21 bp involve the same 8 bp near the C-terminus of ligand-binding repeat 5.244 All three insertions are flanked by the sequence GAC, which may promote slipped strand mispairing.245 Day et al.204 found other hot spots for mutations at sites with short tandemly repeated sequences in exons 6, 11, and 12 of the LDL receptor gene.
Mutations Involving Alu Repeats.
Many deletions and insertions in the LDL receptor gene arise because of recombination between Alu-type elements, the predominant middle repetitive DNA sequences in mammals. The human genome contains about 910,000 Alu repeats246 that are distributed throughout all chromosomes. Most are located in intergenic regions and introns,247 but some are present in mature mRNAs, as in the 3′ untranslated region of the LDL receptor. Each Alu sequence of approximately 300 bp is composed of two tandem repeats designated the left and right arms. Alu repeats can be transcribed by RNA polymerase III, which recognizes sequence blocks in each arm that act as a bipartite promoter, resembling those of the 5S ribosomal RNA genes and tRNA genes.247 The large number of Alu repeats in the human genome and the possible instability stemming from their transcription have led to the idea that these sequences might serve as sites for genome rearrangements.248 This hypothesis has received its strongest support from the almost invariant finding of Alu sequences at the junctions of large rearrangements in mutant LDL receptor genes.215,222,223,226,227,249,250
Of the 11 deletions that have been sequenced and characterized from genomic DNA, 10 possess an Alu repeat at one or both mutation end points. The locations of 10 of these mutations are shown in Fig. 62-22 on p. 2001 of the seventh edition of this book.208 The other mutation, a 10-kb deletion (FH Konazawa) that removes exons 2 and 3251 is not shown. Two of the Alu-mediated deletions, FH Rochester (del exons 16–18) 226 and FH Osaka-2 (del exons 7–14) 227 resulted from recombination between two Alu sequences oriented in opposite directions. In one of these (FH Rochester) inverted repeats in the region surrounding the deletion joint may have formed an intrastrand double-stem loop structure that facilitated the recombination.226 More commonly, the recombination occurs between two Alu repeats in the same orientation, as in FH Paris-1 (del exon 5),215 FH London-1 (del exons 13 and 14), 249 and FH Aarhus-3 (del exon 5).250 It is presumed that homology between the Alu sequences led to a mispairing of chromosome 19 chromatids during meiosis, followed by an unequal crossover that deletes the intervening sequence. The same type of Alu-mediated recombination can result in the reciprocal event, namely a duplication of exons as found in FH St. Louis (dup exons 2–8) 223 (discussed above and below). Single Alu sequences can also instigate gene rearrangements as has been found at the recombination site in at least nine human genes,140 including one in the LDL receptor gene FH Potenza (del exons 13–15).252
An important question is whether the preponderance of Alu-mediated recombinations in the LDL receptor gene is caused by an unusually large number of Alu sequences in this gene. To date, approximately 25 percent of the 45-kb human LDL receptor gene has been sequenced, and 28 Alu repeats have been identified, indicating an average distribution of one Alu sequence per 1.6 kb of gene,140 which is twofold more frequent than the average region of the genome. In the human genome, one Alu repeat occurs on average every 3 kb (3 × 109 bp/900,000 Alu repeats). This twofold increase may be significant. It parallels the findings with globin genes. More Alu-mediated recombination events have been described in the α-globin locus, where the distribution of Alu repeats is 1 per 2 kb,253 than in the β-globin locus, where the frequency is 1 per 10 kb.254 Thus, it seems likely that the number of rearrangements in a region of DNA may rise in geometric fashion with the number of Alu sequences, each of which may act as a hot spot for recombination.255
Numerous LDL receptor mutations have been identified in two unrelated individuals with different ethnic backgrounds. Receptor haplotype analysis197a can be used to determine if the two individuals inherited the same mutant allele from a common ancestor. Unless there has been recombination within the LDL receptor locus, a mutation will remain associated with the haplotype of the allele on which the mutation originally arose. For example, the FH Afrikaner-1 mutation was also found in an individual from Maine. In FH Afrikaner-1 (D206E) and FH Maine (D206E), haplotype analysis with 10 RFLP sites showed an identical restriction pattern except for the 5′-most site.140 It is likely that these two mutant alleles descended from the same chromosome, and subsequently a recombinational event occurred, accounting for the single RFLP difference between the two alleles.
If identical mutations arose independently, they should be found on chromosomes with different haplotypes. For example, FH Gujerat (P664L) was identified in an Indian from Zambia as well as in an Englishman. The haplotypes of the two alleles differed at five of eight RFLP sites examined.256 Multiple recurrent mutations at the LDL receptor have been described, some of which were reviewed in the seventh edition of this textbook (see pages 1997–2000).208
Cytosine (C) to thymidine (T) transitions at a cytosine-guanine (CpG) dinucleotide are a frequent cause of mutations in the human genome.257 In most genetic disorders ~32 percent of the disease-causing mutations involve CpG dimers,245 but at the LDL receptor locus only 20 percent of the point mutations are C-to-T substitutions.207
Spontaneous De Novo Mutations.
The marked allelic genetic heterogeneity at the LDL receptor locus suggests a high mutation rate in this gene. Thus, it is surprising that only three FH heterozygotes have been identified with an LDL receptor mutation that is not found in either of their biological parents (proved by paternity testing).258– 260
Two Mutations in Same LDL Receptor Allele.
Jensen et al.261 described an LDL receptor allele that carried both a missense mutation in exon 11 (N543H) and a 9-bp deletion in exon 17 (del 778–780). The relative role of each mutation was determined by recreating each mutation separately in a recombinant LDL receptor expression construct. Both recombinant mutant proteins had only slightly reduced LDL receptor function when expressed in transfected cultured cells. In contrast, when both mutations were placed in the same expression construct, LDL receptor function was reduced by 75 percent. Eleven examples of double mutations in the same receptor allele have been reported and are listed at http://www.umd.necker.fr.207 Double mutations have also been identified in numerous other disease-causing genes.262 For most genetic disorders, molecular screening is terminated as soon as one disease-causing mutation in an allele is identified. It is possible that some of the discrepancies observed between genotype and phenotype in genetic diseases, including FH, result from undetected double mutations.
In most populations of the world, the frequency of heterozygous FH is ~0.2 percent (discussed above). In certain populations, the frequency is much higher, owing to founder effects (Table 120-4). The highest frequency is found in three different populations of South Africa—the Ashkenazi Jews (1 in 67),103 the South African Indians (estimated at 1 in 100),263 and the Afrikaners, in whom the prevalence of FH (1 in 100) is fivefold higher than in the European population from which it originated.43
Table 120-4: Inbred Populations with Mutant LDL Receptor Alleles that Account for >15% of the Mutant Alleles in that Population |Favorite Table|Download (.pdf) Table 120-4: Inbred Populations with Mutant LDL Receptor Alleles that Account for >15% of the Mutant Alleles in that Population
|Inbred Population ||Mutation ||Percent of FH Heterozygotes with Mutation |
|Christian Lebanon || FH Lebanese (C660X) ||100 |
|South African: |
|Ashkenazi Jews || FH Lithuania (G197del) ||80 |
|Asian Indians || FH Gujerat (P664L) ||>15* |
|Afrikaners || FH Afrikaner-1 (D206E) ||60–70 |
| || FH Afrikaner-2 (V408M) ||20–30 |
|French Canada || FH French Canadian-1 (del 5′ flanking region-intron 1) ||60 |
| || FH French Canadian-4 (W66G) ||18 |
|Iceland || FH Iceland (IVS4+2T>C) ||60 |
|Finland || FH Helsinki (del exons 15–18) ||34 |
| || FH North Karelia (P288fs) ||34 |
|Sephardic Jews || FH Sephardic (D147H) ||>15* |
|Druze || FH Druze (Y167X) ||>15* |
|Ashkenazi Jews || FH Lithuania (G197del) ||35 |
|Norway || FH Elverum (IVS3+1G>A) ||28 |
|Greece || FH Genoa (D528G) ||23 |
| || FH Afrikaner-2 (V408M) ||15 |
|Spain || E10X ||20 |
|Belgium (Southern) || C122X ||16 |
|Denmark || FH French Canadian-4 (W66X) ||15 |
| || FH Cincinnati-5 (W23X) ||15 |
The high frequency of FH in the Ashkenazi Jews of South Africa is likely attributable to their origin from a genetically isolated population of ~40,000 Lithuanian Jews who emigrated to South Africa between 1880 and 1910. In support of this hypothesis, a mutation identified in a patient from New Jersey, FH Lithuania (del G197) 140 was present in 8 of 10 South African Ashkenazi Jews tested.214 This same mutation comprises 35 percent of mutant LDL receptor genes in the Ashkenazi Jews of Israel, 64 percent of whom are of Lithuanian origin.214
The Indians of South Africa are suspected of having a high frequency of FH because the same LDL receptor mutation (P664L) was identified in four unrelated Indians residing in South Africa.263,264 This mutation was previously identified in a Zambian subject.265 The Zambian subject and the four South African Indians are all Muslims originating from the province of Gujerat in India. Many Indians from this region emigrated to different parts of the world, so it is likely that this mutation will be identified in Indians with FH in other locales.
In the Afrikaners, three alleles, FH Afrikaner-1 (D206E), FH Afrikaner-2 (V408M), and FH Afrikaner-3 (D154N), comprise over 90 percent of the mutant LDL receptor genes responsible for FH.266, 266a The most common allele, FH Afrikaner-1, is found in 60 to 70 percent of the FH patients. FH Afrikaner-2 and FH Afrikaner-3 alleles are present in 20 to 30 percent and 5 to 10 percent, respectively. The high frequency of these three alleles can also be attributed to founder effects. The Afrikaners are descended from about 2000 original settlers—mostly from Holland, Germany, and France—who emigrated to the Cape of South Africa in the seventeenth and eighteenth centuries. The ancestry of an individual with the Afrikaner-2 mutation was traced to a small town in the Netherlands (Ankijk), which was the site of departure of some of the original emigrants to South Africa.267 The other common mutant allele, FH-Afrikaner-1, appears to have its origin in England rather than the Netherlands.267a In the nineteenth century, the Afrikaners moved into the interior of the country, the Transvaal, where they remained largely isolated from the surrounding populations. The fertility rate was high, and the population grew dramatically to its present size of approximately 3 million. In addition to FH, Afrikaners have a high frequency of several other genetic disorders, owing to founder effects.267
Another African country with a high incidence of FH is Tunisia. The incidence of FH in this North African country is ~1 in 165, and approximately 60 percent of the FH homozygotes are offspring of consanguineous marriages.267a
The frequency of FH heterozygotes in the French Canadian population is estimated to be 1 in 270 based on the number of FH homozygotes identified in Quebec Province in 1981.101 An even higher frequency (1 in 154) was found in the northeastern region of the province. The French Canadians have a high frequency of several other genetic disorders,269 all of which are attributed to founder effects. The 5.3 million modern French Canadians are descended from about 8000 French settlers who emigrated to Quebec Province from western France between 1608 and 1763, founding an agrarian population that has remained physically and socially isolated.269
French Canadian-1, a large deletion in the 5′ end of the LDL receptor gene that produces a Class 1 phenotype,210 was found in 59 percent of 130 French Canadian FH heterozygotes from Montreal.210,219 The same mutation was identified in only one of 42 FH heterozygotes who lived in the western part of France and in none of 30 Parisian FH heterozygotes.270 This finding supports the idea that the high frequency of this mutation in French Canadians is due to a founder effect. Screening for the French Canadian-1 mutation plus five other less common mutations (French Canadian-2 to -6) will detect ~80 percent of the mutant LDL receptor alleles in the French Canadian population.219,271 Two of these mutations, FH French Canadian-3 (E207K) and FH French Canadian-4 (W66G), were identified in FH homozygotes from France who are patients of Professor J. L. deGennes and Dr. F. Dairou of the Hopitaux de Paris.
A total of four mutations are known to be common causes of FH in the Middle East—FH Lebanese (C660X), FH Lithuania (del G197), FH Sephardic (D174H), and FH Druze (Y167X). The FH Sephardic and FH Druze mutations have been identified in unrelated families in Israel.272,273 The Druze are a relatively small Islamic sect that has remained remarkably isolated for ~1000 years,272 but the exact frequency of FH within the group has not been determined.
The Christian Lebanese population played an important role in delineation of the genetics of FH, owing to Khachadurian's pioneering clinical studies (discussed above). The Dallas collection contains eight FH homozygotes with a Christian Lebanese ancestry. All were found to be homozygous for the FH Lebanese (C660X) allele216 (and unpublished observations), which has not yet been identified in any other population. The exact frequency of the mutation in Lebanon has not been determined, but the high frequency of homozygosity suggests that it is common.97
In general, European and North American populations have a plethora of LDL receptor mutations, each of low frequency and typically limited to a single family. Exceptions include the Finnish population in which either of two mutant alleles, FH Helsinki (del exons 16–18) or FH North Karelia (frameshift 287), is present in two-thirds of individuals with heterozygous FH.274,275 The prevalence of each of these mutations varies in different geographic regions, FH Helsinki predominating in the Northern region and FH North Karelia in the Eastern region of Finland. Screening for these two mutations, plus two less frequent mutations, detects over 75 percent of the mutant alleles in Finland.276 Three mutations accounted for 43 percent of the defective LDL receptor alleles in 476 Norwegian FH heterozygotes.277 The population of Iceland is largely of Norwegian origin, but a splicing mutation found in 60 percent of Icelandic FH heterozygotes278 has not been identified in Norway. In Denmark, 42 percent of individuals with FH have one of three LDL receptor mutations: W23X (15 percent), W66G (15 percent), and W556S (12 percent).279, 279a
Clusters of unrelated FH patients sharing the same mutant allele have been identified in other regions of Europe. In Greece, 60 percent of 150 FH heterozygotes had one of six different mutations.280,281 Two mutations (E10X or del G518) were present in 30 percent of FH heterozygotes in the Aragon region of Spain.282 In southern Belgium, a nonsense mutation of exon 4 was found in 16 percent of individuals suspected of having FH.283 In two regions of England, Manchester284 and the southern part of the country,285 two different mutations each had a frequency of ~10 percent. In Japan, 30 percent of the FH heterozygotes had one of five different mutations, none of which had a frequency >15 percent.286
Correlation of Receptor Mutations and Clinical Expression
Clinical Variability among FH Homozygotes.
In individuals with homozygous FH, the concentration of plasma LDL and the severity of coronary atherosclerosis vary, even among individuals who are homozygous for the same mutations, indicating that other genes play important roles. This variability is exemplified by two individuals, both of whom were homozygous for the French Canadian-1 allele, which abolishes production of receptor mRNA and protein.210 In one of the homozygotes, FH 49, symptoms of coronary atherosclerosis did not develop until age 17. Despite treatment with multiple lipid-lowering medications and ileal bypass, her cholesterol level remained between 600 and 1200 mg/dl. Nevertheless, she lived until age 33. Although she had longstanding angina pectoris, she did not die of a myocardial infarction, but of malignant melanoma.210 Another individual, FH 549, also homozygous for the French Canadian-1 allele, died suddenly at age 3 and at autopsy had severe three-vessel coronary atherosclerotic disease.48,210
Some of the clinical variability results from the nature of the particular LDL receptor mutation. FH homozygotes with one or two mutant receptor alleles whose products retain some LDL receptor function have a lower level of plasma cholesterol and have less aggressive coronary atherosclerosis than individuals with two totally nonfunctional receptor alleles (see Fig. 120-4). Moojani287 compared 10 French Canadians homozygous for a large deletion (del 5′ flanking region-intron 1; French Canadian-1) with an equal number of homozygotes with a missense mutation in exon 3 (W66F; French Canadian-4). Total plasma cholesterol levels in the FH homozygotes with the deletion were significantly higher than the levels in those with the missense mutation, with no overlap between the two groups (1068 versus 644 mg/dl). Homozygosity for the deletion was associated with earlier age of onset of coronary artery disease (12.7 versus 23.6 years).
In general, the severity of clinical manifestations of homozygous FH tends to correlate with the activity level of the LDL receptor, as assessed in cultured fibroblasts. An exception to this general trend is seen with the French Canadian-4 mutation. This missense mutation (W66F) is the most perplexing of all the mutations in the LDL receptor gene. The W66F mutant allele produces a receptor of normal size that is processed at the usual rate, but its functional activity in fibroblasts from patients homozygous for the mutation varies from 25 to 100 percent of controls.198,288 Yet, despite this variability in fibroblast assays, FH homozygotes with this mutation all share the clinical manifestations of classic homozygous FH. The lack of correlation between the activity of a mutant LDL receptor allele in fibroblasts and the clinical expression of the disease suggests a liver-specific derangement in the function of this particular mutant protein.
Clinical Variability among FH Heterozygotes.
The nature of the molecular defect in the LDL receptor appears to have less effect on the plasma level of LDL cholesterol in FH heterozygotes than in homozygotes. For example, the plasma levels of total cholesterol in subjects heterozygous for the French Canadian-1 and -4 alleles (324 versus 288 mg/dl) were much more similar than levels in the corresponding homozygotes (1068 versus 644 mg/dl).287 In the deletion group (French Canadian-1 allele), the plasma cholesterol level of heterozygotes was ~33 percent of that of the homozygotes, whereas in the missense group (French Canadian-4 allele) the level in the heterozygotes was ~50 percent of that of the homozygotes. The proportionally higher cholesterol level in FH heterozygotes with the missense mutation suggests that the mutant protein may act in a dominant-negative fashion to interfere with the synthesis or function of the wild-type receptor.
Individuals heterozygous for a mutant LDL receptor allele with no receptor activity tend to have higher plasma levels of LDL cholesterol than those having alleles that retain some residual activity. This is illustrated in the Afrikaner population where FH heterozygotes with the Afrikaner-2 mutation (V408M), which produces no receptor activity, have significantly higher plasma levels of LDL than those with the Afrikaner-1 mutation (D206E), which has ~20 percent of normal receptor activity (plasma LDL levels of 360 mg/dl versus 308 mg/dl).289
The clinical manifestations of heterozygous FH are also modified by environmental factors, especially diet. In Tunisia, many individuals heterozygous for an LDL receptor mutation had plasma levels of LDL cholesterol lower than the 95th percentile when compared to North American controls.268 Cardiovascular complications are distinctly uncommon, and no tendon xanthomas were found in 27 Tunisian FH heterozygotes (ages 32 to 80), all of whom were parents of FH homozygotes. It is likely that the low-fat Mediterranean diet consumed in this region of the world is responsible for the lower plasma levels of LDL.
The clinical expression of FH in China is markedly attenuated as compared to that of FH in Western countries.290 Xanthomas are absent, and symptomatic coronary artery disease is rare in Chinese FH heterozygotes. To determine if genetic differences between Chinese and Caucasians were responsible for the milder disease in the Chinese, Pimstone et al.291 compared the mean levels of plasma LDL cholesterol in 18 FH heterozygotes from mainland China and 16 from Canada. The Chinese heterozygotes had significantly lower mean LDL levels (174 versus 298 mg/dl). The lower plasma levels of LDL in FH subjects living in China reflects the much lower plasma levels of cholesterol in the general population of this country, owing to their low-fat diet.292
Variability Caused by Differential Ligand Receptor Interactions.
Among individuals with different LDL receptor mutations, one source of clinical variability stems from the relative effect of different mutations on the ability of the receptor to bind lipoproteins containing apo B-100 or apo E. Several mutations in the ligand binding and EGF precursor homology domains of the receptor abolish binding of LDL, without impairing binding of apo E-containing lipoproteins, such as β-VLDL (discussed above).
Normal LDL metabolism begins with the triglyceride-rich particle, VLDL, which is secreted by the liver. In capillaries, the triglycerides of VLDL are removed by lipoprotein lipase, forming a denser, more cholesterol-rich, apo E-containing particle termed “IDL.” Approximately 50 percent of this IDL is cleared by hepatic LDL receptors, which recognize apo E with high affinity.293 Even when receptors are normal, some IDL remains in the plasma, where it is converted into LDL.293
In FH homozygotes with a mutation that blocks binding of both IDL and LDL, the synthesis of LDL is increased because the precursor IDL cannot be cleared efficiently by the liver.293 The degradation of LDL is decreased, owing to the defective receptors. As a result of this combined overproduction and undercatabolism, LDL rises to high levels.3,293 In contrast, if the mutant LDL receptor retains the ability to bind IDL, but not LDL, the production of LDL may be normal. The major abnormality is a delayed removal of LDL from plasma, and the overall effect is a more moderate elevation in plasma LDL.
Experimentally, it is difficult to obtain sufficient plasma IDL to perform binding studies with human cells. As a substitute for IDL, we use β-VLDL,144,220 which is a mixture of lipoproteins of d <1.006 g/ml obtained from the plasma of cholesterol-fed rabbits.120,141 β-VLDL contains particles that are equivalent to IDL, as well as chylomicron remnants, which also bind to LDL receptors by virtue of the apo E content.
Clinical studies of a few FH individuals who have LDL receptors that retain the ability to bind β-VLDL suggest that these mutations are indeed associated with relatively mild disease. The proband of the O. family294 is homozygous for the FH Denver-2 (D283N) allele, which substitutes asparagine for the first aspartic acid residue of the DxSDE sequence in repeat 7 of the ligand binding domain.140 The mutation produces a Class 2B receptor that reaches the cell surface slowly and binds β-VLDL, but not LDL294 (and unpublished observations). LDL turnover studies performed in the proband and her obligate heterozygous parents demonstrated a reduced fractional catabolic rate of the injected radiolabeled LDL, but no increase in LDL production.295 This is in marked contrast to the results with other FH homozygotes and heterozygotes in whom LDL production is increased. Both heterozygous parents of FH Denver-2 have only a modest elevation in plasma cholesterol296 and lack symptomatic coronary atherosclerosis, as do the other heterozygotes in the family.
Variability Caused by an LDL-Lowering Gene.
Studies of one FH family, the P. family, have provided evidence that some of the variability in clinical expression of FH may be attributable to a dominant LDL-lowering allele that suppresses the effect of receptor mutations.218 The mutant LDL receptor allele in the P. family has a missense mutation that substitutes leucine for serine in the DxSDE sequence of the fourth ligand-binding repeat.218 The mutant receptor binds β-VLDL, but not LDL. This allele, FH-Puerto Rico (S156L), is present in homozygous form in 1 individual and in heterozygous form in 18 relatives. LDL turnover studies performed in the homozygote as well as in four heterozygotes from the P. family disclosed a phenotype similar to that seen in the O. family (discussed above). The fractional catabolic rate for LDL was decreased without an increase in the production rate.297
One-third of the FH heterozygotes in the P. family, including the proband's mother, have concentrations of plasma LDL below the 90th percentile of the population despite their LDL receptor mutation. The remaining 12 heterozygotes have shown LDL levels above the 95th percentile. Genetic analysis suggests the presence of a gene segregating in an autosomal dominant fashion whose effect is to lower the plasma LDL level. This putative suppressor gene is not linked to the LDL receptor locus itself or to the genes for apo B-100, apo E, apo C-I, apo C-II, apo C-III, hepatic lipase, lipoprotein lipase, LRP, VLDL receptor, 7α-hydroxylase, HMG-CoA reductase, SREBP-1, SREBP-2, SCAP, and the ileal bile acid transporter, as determined by linkage analysis218 (and unpublished observations). Evidence for an LDL-lowering allele has also been found in a French Canadian family with FH due to a deletion of exons 2 and 3 of the LDL receptor gene.298 Two of the eight carriers of this mutation in this family have a normal LDL cholesterol level.
Production of Half-Normal Number of LDL Receptors in Heterozygotes
In normal fibroblasts, the synthesis of the LDL receptor is regulated by feedback transcriptional suppression.112,115,160 The number of receptors declines about tenfold when cellular cholesterol stores are increased by incubation of cells with a usable exogenous source of cholesterol, such as LDL or cholesterol dissolved in ethanol.115 The number of receptors increases again when the exogenous cholesterol is removed and cellular cholesterol levels fall. In the steady state, the number of receptors is adjusted to allow just enough LDL uptake to provide sufficient cholesterol for cell growth and to balance cholesterol losses.112,115
Under conditions that induce a maximal rate of receptor synthesis (i.e., vigorous growth in the absence of an exogenous source of cholesterol), FH heterozygote cells in tissue culture produce about half as many functional receptors as do normal cells.299 More important, when grown in the presence of increasing amounts of exogenous cholesterol, the heterozygote and normal cells suppress their LDL receptor activities in parallel. Over a ten- to twentyfold range of LDL receptor activities, at any given concentration of intracellular cholesterol the heterozygote cells express about half as many receptors as do normal cells.299 This relationship is evident even in the range of LDL receptor levels in which the heterozygote cells clearly have the capacity to produce as many active LDL receptors as do the normal cells—that is, when the appropriate number of receptors is less than half the maximal number.299 The lack of compensation by the normal receptor allele is due to the heterozygote cells, at all levels of exogenous cholesterol, expressing a slightly higher activity of HMG-CoA reductase, which compensates for their cellular cholesterol deficit by synthesis rather than receptor-mediated uptake.
These findings indicate that in the heterozygous cells the regulatory mechanism dictates that the normal allele produces only the amount of gene product that it would normally produce at a given level of cellular cholesterol. The failure of the regulatory mechanism to stimulate the normal allele at the LDL receptor locus to produce twice its normal amount of gene product leaves the heterozygote cells with a persistent 50 percent deficiency in LDL receptors under all conditions of cell growth. Pharmacologic interventions can perturb this cellular regulatory mechanism in vivo (discussed below).
LDL Receptor Mutations in Animals
A rabbit model for FH, the Watanabe-heritable hyperlipidemic (WHHL) rabbit, was discovered in the late 1970s by Yosio Watanabe, a veterinarian at Kobe University in Japan (Fig. 120-25).300 Homozygous WHHL rabbits are markedly hypercholesterolemic from the time of birth, and severe atherosclerosis develops by 2 to 3 years of age.293 The mutation is a 12-bp in-frame deletion that removes 4 amino acids from the fourth ligand-binding repeat of the LDL receptor.154 The mutant protein has a Class 2B/3 phenotype and reaches the cell surface at a markedly reduced rate.294 The few receptors that reach the surface retain the ability to bind β-VLDL, but not LDL. WHHL rabbits are extremely useful in dissecting the role of the LDL receptor in vivo (discussed below)293 and in testing strategies for treatment of FH using gene therapy.301
Photograph of Dr. Yoshio Watanabe, Kobe University, Japan, together with one of his WHHL rabbits. (Courtesy of Dr. Toru Kita.)
A second animal model for FH was described by Scanu et al.302 in a pedigree of rhesus monkeys with hypercholesterolemia inherited in an autosomal dominant pattern. Fibroblasts from the hypercholesterolemic monkeys showed a half-normal level of LDL receptor activity, protein, and mRNA. The responsible mutation, a point mutation that produces a stop codon at amino acid 284, was identified by selectively amplifying and sequencing each exon of the monkey LDL receptor gene with oligonucleotide primers homologous to the human sequences.303
The technique of homologous recombination in cultured embryonic stem cells was used to produce a line of mice that lack functional LDL receptor genes (LDLR–/– mice).304 Homozygous male and female LDLR–/– mice are viable and fertile. Total plasma cholesterol levels are twofold higher than those of wild-type littermates, owing to a seven- to ninefold increase in IDL and LDL without a significant change in HDL. Plasma triglyceride levels are normal. The half-lives for intravenously administered 125I-VLDL and 125I-LDL are prolonged by 30 times and 2.5 times, respectively, but the clearance of 125I-HDL is normal in the LDLR–/– mice. Unlike wild-type mice, LDLR–/– mice responded to moderate amounts of dietary cholesterol (0.2% cholesterol/10% coconut oil) with a major increase in the cholesterol content of IDL and LDL particles.
When the LDLR–/– mice were fed a high-cholesterol diet (1.25 percent cholesterol/7.5 percent cocoa butter/7.5 percent casein/0.5 percent cholic acid), the total plasma cholesterol rose from 246 to >1500 mg/dl, whereas wild-type littermate mice fed the same diet had cholesterol levels <160 mg/dl.305 After 7 months on this diet, the LDLR–/– mice developed massive xanthomatous infiltration of the skin and subcutaneous tissue (Fig. 120-26A-D ). The aorta and coronary ostia exhibited gross atheromata, and the aortic and coronary ostia were thickened by cholesterol-laden macrophages (Fig. 120-26E,F ).
Xanthomas and aortic arch atherosclerosis in LDL receptor knockout mice (left panels) as compared with xanthoma-free and atherosclerosis-free wild-type mice (right panels). Mice of both genotypes were fed a 1.25 percent cholesterol diet for 8 to 9 months. The LDLR−/− mice show xanthomatous involvement of ears and eyelids (xanthelasmas) (B), tendon xanthomas of the hind paw (D), and extensive aortic plaque formation visible through the thin aortic wall (F). The total cholesterol level was 173 to 264 mg/dl in the wild-type mice and 1893 to 2360 mg/dl in the LDLR−/− mice. (Data from Ishibashi et al. 305 Used by permission of Journal of Clinical Investigation.)
On a regular chow diet, the LDLR–/– mice have LDL cholesterol levels that are much lower than those in FH homozygotes,305 primarily because mouse liver, in contrast to human liver, produces apo B-48 in addition to apo B-100 (Chap. 115). Lipoproteins containing apo B-48 can be removed from plasma by alternate receptors, especially LRP (discussed below). To circumvent this problem, and to create a more exact model of the human disease, LDLR–/– mice have been crossed with mice expressing a human apo B-100 transgene306 and also with mice that produce only mouse apo B-100 (and no apo B-48) in their livers.307 These two lines of mice manifest plasma levels of LDL cholesterol that are as high as in FH homozygotes and develop extensive complex atherosclerotic lesions, even when the mice consume a regular low-fat chow diet.
The LDLR–/– mice offer a new model for the study of environmental and genetic factors that modify the process of atherosclerosis and xanthomatosis. They also provide a useful model to explore somatic-cell gene therapy targeted at the liver. For example, intravenous injection of a recombinant replication-defective adenovirus encoding the human LDL receptor driven by the cytomegalovirus promoter reduced the elevated plasma IDL/LDL level to normal, restored expression of LDL receptor protein in the liver, and increased the clearance of 125I-VLDL.304
A strain of spontaneously hypercholesterolemic pigs was found to have a missense mutation (R94C) in the third ligand repeat of the LDL receptor gene.157, 157a Biosynthetic studies of the mutant receptor have not been performed, but this allele is likely to have a Class 2 phenotype because insertion of an additional cysteine residue into the repeat is predicted to interfere with its folding. The inheritance pattern of the hypercholesterolemia associated with this mutation appears to be autosomal recessive; pigs heterozygous for the mutation are normolipidemic.
The LDL receptor gene is a mosaic of exons shared with several other genes.139,140,147 These shared sequences fall into three types. The first type is the 40-amino acid cysteine-rich repeat of the LDL-binding domain, which is found in complement factors (C7, C8α, C8β, C9). The second type of shared sequence is the growth factor repeats of the EGF homology region, which are found in plasma proteases of the blood-clotting system (factor IX, factor X, protein C, thrombospondin), cell-surface receptors (TGF-α precursor, ELAM-1, GMP-140, thrombomodulin), adhesive glycoproteins (tenascin, cartilage proteoglycan core protein), and developmental proteins (Notch, lin-12, delta).140 The third type of shared sequence, the six 40-amino acid YWTD repeats that form a β-propeller structure,149 is interdigitated between clusters of the growth factor repeats. These six YWTD repeats are found in similar locations in the LDL receptor and the EGF precursor.139,140,147 In general, all three types of shared sequences are encoded by discrete exons. These findings support Gilbert's exon-shuffling hypothesis308 in which exons encoding functional protein domains are reused during evolution.
As discussed above, many genes of diverse function possess one or two of the shared types of sequences that are found in the LDL receptor. Only one family of genes shares sequences of all three types. This family is designated the LDL receptor gene family (Fig. 120-27), and all seven known members are cell-surface proteins that bind ligands and transport them into cells by receptor-mediated endocytosis in coated pits. In addition to these three sets of shared exon-encoded sequences, all of these proteins, except the insect receptor, have at least one copy of the NPxY internalization sequence in the cytoplasmic domain (see Fig. 120-27).
Similarities in shared exons and transmembrane orientation in seven members of the LDL receptor gene family. The mutant human LDL receptor (shown at extreme right) refers to FH St. Louis (dup exons 2–8). The evolutionary significance of this duplication mutation is described in the text. Naturally occurring mutations that disrupt receptor function are denoted in parentheses: FH, familial hypercholesterolemia in mammals; R/O, restrictor ovulator in chickens; and yolkless in Drosophila melanogaster.
Five members of the LDL receptor gene family—the LDL receptor, LRP, the VLDL receptor, the neuro apo E/LR8B receptor, and the chicken vitellogenin receptor—are known to bind and internalize plasma lipoproteins. LRP differs structurally from the LDL receptor primarily because of size. It contains 31 complement-type ligand-binding repeats organized in 4 clusters and 22 growth factor repeats that are separated by 8 YWTD-containing β-propellers. LRP differs functionally from the LDL receptor in that it only binds lipoproteins that are enriched in vitro with excess apo E.309 LRP is able to bind and internalize at least 20 known ligands, ranging from lipoprotein particles (chylomicron remnants) and protease inhibitors (α-2-macroglobulin) to plasminogen activators.310 Each ligand probably binds to a unique combination of ligand-binding repeats in LRP.
The VLDL receptor differs functionally from the LDL receptor in that it binds apo E-containing lipoproteins (without the need for enrichment with exogenous apo E), but not LDL.233 This receptor and the LDL receptor show strikingly different tissue distributions in vivo. The LDL-receptor mRNA is expressed primarily in adrenal cortex and liver with virtually no expression in heart, skeletal muscle, and adipose tissue.3,154 The VLDL-receptor mRNA is expressed at highest levels in the heart with moderately high expression in skeletal muscle, adipose tissue, and brain, but only in trace amounts in the liver.233 Although this pattern of expression supports the hypothesis that the VLDL receptor may be involved in uptake of triglycerides from VLDL,233 a knockout of its gene in mice revealed no abnormality in plasma-lipoprotein levels on normal, high-carbohydrate, or high-fat diets.310a The sole abnormality detected was a modest decrease in body weight and adipose tissue mass.
The chicken vitellogenin receptor, expressed exclusively in the developing oocyte, binds and internalizes two yolk proteins: VLDL and vitellogenin. It does not bind LDL.237 The chicken vitellogenin receptor transports 1.5 g of protein into the chicken oocyte each day, as the single cell grows to the size of a grade AA jumbo egg over seven days. A mutation in this receptor causes the female-sterile “restricted ovulator” phenotype, in which chicken oocytes do not grow to normal size.311 A mutation in the Drosophila homologue of the vitellogenin receptor also produces a female-sterile phenotype, yolkless.312
Megalin/gp330 is a large protein that resembles LRP.232 It resides in coated pits on the tubular (apical) surface of epithelial cells in the renal glomerulus and proximal tubule. It is a target for an autoantibody that circulates in rats with Heymann-type autoimmune nephritis.313 The function of megalin/gp330 is unknown, but it binds most, if not all, of the ligands that are recognized by LRP, including apo E-enriched lipoproteins.310,314
The genomic rearrangements that underlie exon shuffling in the LDL receptor gene family may be continuing at the present time owing to the recombinational activity of the Alu elements (discussed above). Figure 120-27 shows the structure of a mutant LDL receptor that is produced by the FH St. Louis (dup exons 2–8) allele and compares it with the other members of the LDL receptor gene family. As discussed above, this mutant gene has undergone an Alu-mediated duplication of 14 kb, which reiterates exons 2 to 8.223 Although the resulting protein has two complete ligand-binding domains, it binds LDL and β-VLDL poorly. In contrast to the earlier duplications that gave rise to the seven repeats in the ligand-binding domain of the wild-type LDL receptor, this new rearrangement is deleterious, and so is expected to disappear through natural selection. It is possible to envision another duplication that might improve or expand the function of the LDL receptor, especially given that different combinations of cysteine-rich repeats mediate binding of different apoprotein ligands to the LDL receptor.144