Genetic Localization to Chromosome 1p32 of the Third Locus for Familial Hypercholesterolemia in a Utah Kindred
Abstract—Clinical familial hypercholesterolemia has been shown to result from mutations in 2 genes, the low density lipoprotein (LDL) receptor on chromosome 19 and apolipoprotein B on chromosome 2. However, we have recently described a Utah pedigree in which linkage to both genes was clearly excluded. A multipoint linkage analysis of 583 markers genotyped on 31 (18 affected) members of this pedigree was undertaken to localize a genetic region that may harbor a third gene that could result in clinical familial hypercholesterolemia. A multipoint log of the odds score of 6.8 was obtained for markers on 1p32. Haplotype carriers and affected status are completely concordant (18/18 persons). The phenotype is also expressed in young children (ages 4 and 9). Specific recombinant individuals in the pedigree restrict the region of linkage to an ≈17 cM interval between polymorphic markers D1S2130 and D1S1596. This region appears to overlap the region found linked to severe hypercholesterolemia in French and Spanish families. The identification of the gene in this region may provide important pathophysiological insights into new mechanisms that may lead to highly elevated LDL cholesterol and other associated dyslipidemic phenotypes.
- Received January 6, 2000.
- Accepted February 18, 2000.
Clinical familial hypercholesterolemia (FH) is defined as a very high (twice normal) LDL cholesterol level with clear bimodality within families. Total cholesterol levels are usually >7.0 mmol/L in adults, and elevated levels already appear at birth. Tendon xanthomas are diagnostic for this condition but are often absent, especially at young ages. Two genetic loci have been implicated in this clinical disorder: the LDL receptor gene (FH)1 and the apolipoprotein B gene (familial defective apo B, or FDB).2 There is some evidence that LDL cholesterol levels show some variation among affected persons in FH and FDB families.3 4 5 However, one is usually not able to distinguish clinically between FH and FDB,6 although there may be a trend toward lower LDL cholesterol levels in FDB patients.4
There are >600 mutations described for the LDL receptor.7 Despite this large number, mutations have not been identified in up to 30% of all persons with FH. In our recent study, we were able to find specific mutations in only 19 of 47 probands with clinical FH.8 Linkage to the LDL receptor was found in an additional 4 of 5 families large enough to test for linkage. Linkage analysis in 1 pedigree (kindred 1173) clearly excluded linkage to either the LDL receptor or the apo B gene.
FH and FDB are currently underdiagnosed throughout the world, and many of those who have been diagnosed with either of these 2 disorders are not adequately treated.9 10 The statin drugs have greatly increased physicians’ ability to control high levels of LDL cholesterol, yet not all persons respond equally to treatment. The existence of another locus that causes elevated LDL cholesterol similar to FH and FDB may elucidate other disease mechanisms for which new, effective drugs could be developed. Such drugs could not only help to target the defects of this new locus, but also might be used as therapy for other genetic cholesterol disorders such as FH, FDB, familial combined hyperlipidemia, and polygenic hypercholesterolemia. To investigate the possibility of another locus for elevated LDL cholesterol (referred to as FH38 ), we performed a genome-wide linkage analysis on kindred 1173.
Pedigree Ascertainment and Description
Kindred 1173 is extensively described in a previous publication.8 Lipid values and other measured phenotypes are provided in that article and are summarized here in Table 1⇓. In brief, this kindred was selected because the proband had a total cholesterol level >7.75 mmol/L (>300 mg/dL). No mutations in the LDL receptor were found in the proband of this family, and linkage analysis excluded linkage of LDL cholesterol to both the LDL receptor and apo B genes. The pedigree contains 38 individuals, 31 of whom were sampled. The 7 unsampled persons are deceased, but information about them was required to connect the pedigree together. All unaffected pedigree members had LDL cholesterol levels <3.88 mmol/L (<150 mg/dL) with no history of lipid-lowering medication use or elevated lipids. Fifteen of the 18 affected pedigree members had LDL cholesterol levels >5.17 mmol/L (>200 mg/dL). The remaining 3 affected persons were on statin medications at the time of cholesterol measurement and had historical total cholesterol levels >11.6 mmol/L (>450 mg/dL), and 2 of these also had tendon xanthomas. Further rationales for these classifications are provided in Reference 11 , which details specific algorithms, validated by DNA testing, for minimizing the false-positive and false-negative classification of FH.
Blood samples were collected at the University of Utah Cardiovascular Genetics Clinic in the morning after an overnight 12- to 16-hour fast. Measured LDL cholesterol was obtained by using a microscale procedure on a Roche FARA II automated analyzer, as described by Wu et al12 in our laboratory, which participates in the Centers for Disease Control and Prevention Standardization Program. For those persons who were currently on lipid-lowering medications, the highest historically measured or calculated LDL cholesterol value was used to define affected status for analysis. Three persons currently on statin therapy did not have available historical measured lipid levels and were classified as being affected. DNA was extracted from frozen buffy coats by using the Lipid Research Clinics guidelines.13 Written, informed consent was obtained from all subjects, and the project was approved by the University of Utah Institutional Review Board.
We genotyped the 31 family members for 583 anonymous autosomal markers with an average spacing of 5.7 cM for the genome search. Most of the markers are trinucleotide or tetranucleotide repeats and are described in public databases such as Marshfield (http://www.marshmed.org/genetics), Genethon (http://www.genethon.fr), and the Human Genome Database (GDB; http://gdbwww.gdb.org). Polymerase chain reaction products were analyzed at Myriad Genetics, Inc, on an ABI 377 fluorescent sequencing machines with the use of standard methods. Allele numbers were internally generated and do not refer to allele numbers defined in public databases. Inheritance of all alleles was verified by using the pedcheck program.14 In addition, 2 markers were added at the apo B locus and 3 markers at the LDL receptor locus to verify absence of linkage to these 2 genes. The genetic map used for all analyses and tables was generated internally by Myriad Genetics, Inc, using the cri-map program15 on 3916 meioses and closely corresponds to the Marshfield map. However, the Marshfield map is used in Figure 4⇓ to permit comparisons of recombinant boundaries between our study and another study.
Linkage analysis was performed after coding LDL cholesterol as a dichotomous phenotype as described above.11 There was clear bimodality of LDL cholesterol in the pedigree, so all persons could be easily assigned as affected or unaffected.8 The 7 members of the pedigree who were not sampled were coded as unknowns. For all 2-point linkage analyses, 2 general parametric models (dominant and recessive) were run by using the fastlink16 17 modifications of the mlink program.18 19 For the parametric analysis, we used a gene frequency of 0.001 for the dominant model and of 0.0447 for the recessive model. These gene frequency estimates yield a prevalence of affected persons of 0.2% for both models if the gene is fully penetrant. This is the estimated population frequency of FH and is the maximum expected frequency for this new locus. Power for linkage generally increases as the gene frequency decreases, thereby making this assumption conservative. Penetrances for a single liability class were set at 0.0 for the unaffected genotype (no sporadic cases) and 0.95 for the affected genotype (5% have the gene but do not express the phenotype). Even though transmission of the phenotype clearly appears to follow a dominant mode of inheritance in this pedigree, a recessive model was also used to help rule out other genetic regions that might contribute in a recessive manner to LDL cholesterol levels in the family.
Multipoint linkage analysis was performed with Myriad Genetics’ mclink software program.20 mclink generates multipoint marker haplotypes by using a blocked Gibbs sampling approach; convergence was verified with multiple runs. Generated haplotypes were compared with and agreed with visually assigned haplotypes. Multipoint plots of the best linkage region are produced showing maximum log of the odds (LOD) scores calculated at each marker across the chromosome.
Table 1⇑ shows the mean age and lipid levels for the affected and unaffected pedigree members. In addition, it shows means for these variables from 198 persons with FH who are known to have mutations in the LDL receptor.8 There is no significant difference in age among the 3 groups, although persons in the pedigree were 9 years younger. LDL and HDL cholesterol values were not different between the affected pedigree members and persons with FH, whereas triglycerides were significantly lower in the affected pedigree members compared with FH patients (P<0.001). Affected subjects were not overweight, with a mean body mass index of 22.7 kg/m2. There were no significant differences in triglycerides between the affected and unaffected pedigree members. HDL levels were significantly lower in the affected than the unaffected pedigree members (P<0.01).
Figure 1⇓ shows the distribution of the 2-point LOD scores for the dominant and recessive models for the 583 markers. The 2-point LOD scores presented are the maximum values across all recombination fractions. Table 2⇓ shows all markers under either the dominant or recessive models that had a 2-point LOD score >1.0. Only 16 markers (2.7%) had LOD scores >1.0 for the dominant model, with 8 of those markers occurring in the same region of chromosome 1. The maximum LOD score of 5.7 occurred for marker D1S2134. Three other chromosome 1 markers were all >3.0. Chromosome 17 was the only other chromosome with >1 marker with a LOD score >1.0. Other than chromosomes 1 and 17, no other chromosome had a marker with a LOD score >2.0.
The recessive model resulted in 14 markers with 2-point LOD scores >1.0 (2.4%), with 6 of them occurring on chromosome 3, though grouped in 2 different unlinked locations (Table 2⇑). None of the LOD scores were >2.2. Table 2⇑ also shows the multipoint LOD scores from mclink. Chromosome 1 had a highly significant LOD score of 6.8 (0.0 recombination) at markers D1S2134 and D1S1661. Only chromosomes 1 and 17 show multipoint LOD scores >2.0, with chromosome 17 having a maximum multipoint LOD score of 2.6. Table 2⇑ also shows how the LOD scores changed from the 2-point to multipoint analyses. Under the multipoint dominant model, all of the chromosome 1 LOD scores increased. The largest LOD score for chromosome 17 decreased from 2.72 to 2.58, although the information from this marker helped to increase the LOD scores for surrounding markers. The multipoint LOD score for the chromosome 22 marker was also increased. However, this marker was the most telomeric marker tested, and the multipoint LOD score for the marker next to it was 0, suggesting that there is no linkage in this region. The multipoint LOD scores for the markers on chromosomes 2, 14, and 18 decreased.
The recessive multipoint LOD scores were all <1.0 except for chromosomes 1, 3, and 8. The chromosome 1 LOD score was 1.58 and overlapped with the region found for the dominant model. The multipoint LOD scores for chromosomes 3 and 8 were 2.46 and 1.33, respectively, with only chromosome 3 showing both suggestive linkage and surrounding marker support for linkage. Two-point LOD scores at markers near the LDL receptor locus (D19S865, D19S1165, and D19S221) were all < −2.0 at θ=0.05. The 2 markers near the apo B gene (D2S220 and D2S305) also had negative LOD scores (−1.9 and −1.7 at θ=0.01 and both <−4.0 at θ=0), excluding linkage.
Figure 2⇓ shows the shape of the significant multipoint linkage tracing for the dominant model for chromosome 1. The 1-LOD support region is from 66 to 78 cM, a 12 cM region. Figure 3⇓ shows the pedigree members with their estimated haplotypes and recombinants. Recombinants defining the minimal genetic region occur in individuals 7 and 50 between markers D1S2130 and D1S1596, resulting in an ≈17-cM region (Figure 4⇓). All persons with the disease haplotype (18/18) expressed the abnormal phenotype, and all phenotypically affected persons (18/18) had the disease haplotype. Two affected pedigree members were children, ages 4 and 9, when lipids were first measured.
This study shows that elevated LDL cholesterol levels in a Utah kindred, indicative of clinical FH, are linked to a region on chromosome 1p32. We have previously shown that this pedigree is clearly not linked to the LDL receptor or apo B gene,8 and additional markers near each gene in this study confirm the absence of linkage. A gene within a 17-cM region on chromosome 1 appears to be the major determinant of the observed elevated LDL cholesterol. There may be additional genes with minor effects on LDL cholesterol levels on chromosomes 3 and 17 in this pedigree. The LDL cholesterol levels of this pedigree are similar to those of FH pedigrees with known LDL receptor mutations, and penetrance is complete even at young ages. Triglyceride levels are significantly lower than in FH pedigrees, but mean age and body mass index are also lower. There were no differences in the frequency of tendon xanthomas. Therefore, genotyping to determine linkage or the presence of mutations appears to be necessary to differentiate the 3 FH-like disorders.
The linkage to chromosome 1p32 overlaps with a region recently found linked to high LDL cholesterol levels in a large French family and other smaller Spanish families.21 The European families had a maximum heterogeneity LOD score of 2.88 for all clinical FH-like pedigrees (27% estimated to be linked), and the LOD score for the single large pedigree was 3.13. The large size of the Utah pedigree greatly increases the power to detect linkage, while the high LOD score of 6.76 validates the previously published lower French LOD score.
Our linkage region is defined by 7 recombination events within the pedigree, giving a maximum potential region between D1S2130 and D1S1596. The addition of markers in this region could reduce its size substantially, since D1S1661 and D1S405 are not informative in recombinant meioses. It should be noted that none of our genotyped markers were the same as those used in the French study. The integration of the markers and recombinants from both studies is shown in Figure 4⇑, which uses the Marshfield map order and distances, rather than the Myriad or French sample map distances. Varret et al21 found 4 recombinants, 2 in affected persons and 2 in unaffected persons. Using unaffected persons for recombinant boundaries necessitates the assumption of nearly complete penetrance. One of their pedigree members was a haplotype carrier but not affected, while another was affected but not a haplotype carrier, suggesting that penetrance in their pedigree is incomplete. Nevertheless, using only the 2 affected persons delimits nearly the same recombinant region.
Our recombinants move their telomeric boundary (D1S472) toward the centromere, whereas their centromeric boundary (D1S211) is within our region. In our laboratory, the D1S211 marker is difficult to score and is no longer used. We are hesitant to use this marker to define a recombinant boundary. Using the D1S197 boundary instead of the D1S211 boundary yields a 5.7-cM region from D1S2130 to D1S197 compared with a 1.4-cM region if the D1S211 boundary is valid. Note that the order in the Marshfield map for markers D1S472 and D1S255, which are 1 cM apart, is switched from the order estimated directly from the family data in the study by Varret et al. This switch has the effect of increasing their reported linked interval by a small amount over that reported in their study, but after recombinants from the 2 studies are combined, it has no effect.
There are no obvious candidate genes in the minimal linkage region, and the region appears to be a gene-poor region. Therefore, to validate further the recombinant boundaries and to narrow the linkage region so that cloning of the responsible gene may proceed, we are currently adding more markers to the region and expanding the family to find additional recombinants. Additional clinically defined FH families are also being studied to determine whether any might be linked to the FH3 locus. We are also investigating whether other phenotypes might be linked to this region, suggesting a pleiotropic effect of the gene.
Our findings, then, substantiate and extend previous work8 21 that identified a third genetic locus responsible for FH in humans. Other known FH loci are represented by the LDL receptor and by the ligand for it (apo B). The genetic lesion that underlies FH3 may be another component of the same pathway, or it may represent a novel mechanism that contributes to human cholesterol metabolism. Questions of whether a substantial fraction of clinically defined FH families are linked to FH3 and whether other non-FH phenotypes are associated with FH3 can be addressed in future work.
This study was funded by Novartis Pharmaceuticals Corp; Myriad Genetics, Inc; and National Institutes of Health grants HL21088 and HL47561 (to S.C.H. and P.N.H.). We gratefully acknowledge valuable technical assistance from A. Bandley and R. Robertson. Family studies were coordinated by S. Frogley. Database support was provided by M. Francis, J. Fraser, and R. Gress. This work is dedicated to the memory of Roger R. Williams, whose enthusiasm and encouragement helped make it possible.
Goldstein JL, Schrott HG, Hazzard WR, Bierman EL, Motulsky AG. Hyperlipidemia in coronary heart disease, II: Genetic analysis of lipid levels in 176 families and delineation of a new inherited disorder, combined hyperlipidemia. J Clin Invest. 1973;52:1544–1568.
Innerarity TL, Weisgraber KH, Arnold KS, Mahley RW, Krauss RM, Vega GL, Grundy SM. Familial defective apolipoprotein B-100: low density lipoproteins with abnormal receptor binding. Proc Natl Acad Sci U S A. 1987;84:6919–6923.
Gallagher JJ, Myant NB. Variable expression of the mutation in familial defective apolipoprotein B-100. Arterioscler Thromb. 1993;13:973–976.
Hansen PS, Defesche JC, Kastelein JJ, Gerdes LU, Fraza L, Gerdes C, Tato F, Jensen HK, Jensen LG, Klausen IC, Faergeman O, Schuster H. Phenotypic variation in patients heterozygous for familial defective apolipoprotein B (FDB) in three European countries. Arterioscler Thromb Vasc Biol. 1997;17:741–747.
Wilson DJ, Gahan M, Haddad L, Heath K, Whittall RA, Williams RR, Humphries SE, Day IN. A World Wide Web site for low-density lipoprotein receptor gene mutations in familial hypercholesterolemia: sequence-based, tabular, and direct submission data handling. Am J Cardiol. 1998;8:1509–1511.
Haddad L, Day INM, Hunt S, Williams RR, Humphries SE, Hopkins PN. Evidence for a third genetic locus causing familial hypercholesterolemia: a non-LDLr, non-apoB kindred. J Lipid Res. 1999;40:1113–1122.
World Health Organization. Familial Hypercholesterolaemia (FH). Report of a WHO Consultation. Geneva, Switzerland: World Health Organization; 1998.
Wu LL, Warnick GR, Wu JT, Williams RR, Lalouel JM. A rapid micro-scale procedure for determination of the total lipid profile. Clin Chem. 1989;35:1486–1491.
Lander ES, Green P. Construction of multilocus genetic linkage maps in humans. Proc Natl Acad Sci U S A. 1987;84:2363–2367.
Cottingham RW Jr, Idury RM, Schaffer AA. Faster sequential genetic linkage computations. Am J Hum Genet. 1993;53(1):252–63.
Lathrop GM, Lalouel JM, Julier C, Ott J. Strategies for multilocus linkage analysis in humans. Proc Natl Acad Sci U S A. 1984;81:3443–3446.
Thomas A, Gutin A, Abkevich V, Bansal A. Multilocus linkage analysis by blocked Gibbs sampling. Stat Comput. In press.
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.