An NPC1L1 gene promoter variant is associated with autosomal dominant hypercholesterolemia

Nutrition, Metabolism & Cardiovascular Diseases (2010) 20, 236e242 available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/nmcd

An NPC1L1 gene promoter variant is associated with autosomal dominant hypercholesterolemia ´.-L. Garcı´a-Otı´n b, S. Pampı´n d, M. Cofa ´n c, B. Martı´n a, M. Solanas-Barca b, A E. Ros c, J.-C. Rodrı´guez-Rey d, M. Pocovı´ a, F. Civeira b,* a

Departamento de Bioquı´mica y Biologı´a Molecular y Celular, Universidad de Zaragoza, and Ciber Enfermedades Raras (CIBERER), Instituto de Salud Carlos III (ISCIII), Spain b Laboratorio de Investigacio´n Molecular, Hospital Universitario Miguel Servet, IþCS, Zaragoza, Spain c Unitat de Lı´pids, Servei d’Endocrinologia i Nutricio´, Institut d’Investigacions Biome`diques August Pi i Sunyer, Hospital Clı´nic, Barcelona and Ciber Fisiopatologı´a de la Obesidad y Nutricio´n (CIBEROBN), ISCIII, Spain d Departamento de Biologı´a Molecular, Facultad de Medicina, Universidad de Cantabria, Santander, Spain Received 25 November 2008; received in revised form 19 March 2009; accepted 20 March 2009

KEYWORDS Familial hypercholesterolemia; Polymorphisms; Cholesterol intestinal absorption

Abstract Background and aims: A substantial number of subjects with autosomal dominant hypercholesterolemia (ADH) do not have LDL receptor (LDLR) or apolipoprotein B (APOB) mutations. Some ADH subjects appear to hyperabsorb sterols from the intestine, thus we hypothesized that they could have variants of the NiemannePick C1-Like 1 gene (NPC1L1). NPC1L1 encodes a crucial protein involved in intestinal sterol absorption. Methods and results: Four NPC1L1 variants (133A>G, 18C>A, 1679C>G, 28650A>G) were analyzed in 271 (155 women and 116 men) ADH bearers without mutations in LDLR or APOB aged 30e70 years and 274 (180 women and 94 men) control subjects aged 25e65 years. The AC haplotype determined by the 133A>G and 18C>A variants was underrepresented in ADH subjects compared to controls (p Z 0.01). In the ADH group, cholesterol absorption/ synthesis markers were significantly lower in AC homozygotes that in all others haplotypes. Electrophoretic mobility shift assay (EMSA) results revealed that the 133A-specific oligonucleotide produced a retarded band stronger than the 133G allele. Luciferase activity with NPC1L1 133G variant was 2.5-fold higher than with the 133A variant.

Abbreviations: ADH, autosomal dominant hypercholesterolemia; LDLR, LDL receptor gene; APOB, apolipoprotein B gene; NPC1L1, NiemannPick C1-Like 1; EMSA, electrophoretic mobility shift assay; CVD, cardiovascular disease; Apo, apolipoprotein; FH, familial hypercholesterolemia; FDB, familial defective apolipoprotein B; LDL, low-density lipoprotein; HDL, high-density lipoprotein; SNP, single nucleotide polymorphism; DNA, deoxyribonucleic acid; bp, base pair; PCR, polymerase chain reaction; IR, interquartile range. * Corresponding author. Laboratorio de Investigacio ´n Molecular, Hospital Universitario Miguel Servet, Avda. Isabel La Cato ´lica, 1-3, 50009 Zaragoza, Spain. Tel.: þ34 976765500; fax: þ34 976369985. E-mail address: [email protected] (F. Civeira). 0939-4753/$ - see front matter ª 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.numecd.2009.03.023

NPC1L1 gene and dominant hypercholesterolemia

237

Conclusion: The 133A>G polymorphism exerts a significant effect on NPC1L1 promoter activity. NPC1L1 promoter variants might explain in part the hypercholesterolemic phenotype of some subjects with nonLDLR/nonAPOB ADH. ª 2009 Elsevier B.V. All rights reserved.

Introduction The two most common monogenic disorders resulting in increased total cholesterol and low-density lipoprotein (LDL) cholesterol are familial hypercholesterolemia (FH) and familial defective apolipoprotein (Apo) B (FDB), which are caused by mutations in the genes encoding for the LDL receptor (LDLR) protein and Apo B, respectively [1]. A rare third form of autosomal dominant hypercholesterolemia (ADH), named FH3, is due to mutations in the gene encoding for proprotein convertase subitilisin/kexin 9 (PCSK9) [2]. Depending on the clinical diagnostic criteria used, mutations in these genes explain 60-80% of ADH cases in most populations [3]. Cholesterol homeostasis is maintained by balancing intestinal cholesterol absorption, endogenous cholesterol synthesis and biliary excretion of cholesterol and bile salts [4]. In humans, the LDL-cholesterol level relates to the efficiency of intestinal cholesterol absorption [5], and we recently reported that subjects with ADH and no mutations in LDLR or APOB were characterized by high intestinal cholesterol absorption, as assessed by elevated levels of circulating phytosterols [6]. Thus, we hypothesized that genetic variation at the NiemannePick C1-Like 1 (NPC1L1) gene could be the basis for some forms of nonLDLR/nonAPOB ADH. NPC1L1, a recently identified sterol intestinal transporter, promotes the passage of sterols across the brush border membrane of enterocytes, acting as a critical protein for intestinal cholesterol absorption [7,8]. NPC1L1 is predominantly expressed in the liver and the proximal intestine [7,9] and shares 51% amino acid homology with NiemannePick C1, a protein involved in intracellular cholesterol trafficking that is defective in NiemannePick type C storage disease [10]. Sequence variants in NPC1L1 are associated with variations in sterol absorption and LDLcholesterol in the general population [11]. It is unknown whether NPC1L1 gene variation is associated with ADH. Hence, the aim of this work was to investigate the association between ADH and NPC1L1 gene variants.

Methods Study subjects Subjects with a clinical diagnosis of ADH and healthy controls were studied. The ADH group was made up of 271 consecutive unrelated subjects aged 30e70 years attending four Spanish lipid clinics that were recruited during a large FH case-finding program and in whom thorough genetic testing uncovered no molecular defects in LDLR or APOB [6]. Secondary hyperlipidemia was ruled out in all subjects. A clinical diagnosis of ADH was made in subjects with offtreatment LDL-cholesterol levels above the age- and sexspecific 95th percentiles of a Spanish reference population

[12], triglyceride below 200 mg/dL, and familial vertical transmission with at least two first-degree relatives with hypercholesterolemia. Clinical data, history of cardiovascular disease (CVD), demographic and anthropometric measurements, and a physical examination in search of tendon xanthomas were obtained in each subject. The control group consisted of 274 healthy unrelated men and women volunteers aged 25e65 years, who underwent a medical examination as reported elsewhere [13]. Exclusion criteria were parental history of dyslipidemia and current acute illness or use of drugs that might influence glucose or lipid metabolism. A fasting blood sample was obtained from all subjects for lipid and genomic analyses. All participants provided informed consent to a protocol approved by the ethical committees from each institution.

Biochemical determinations Lipid profile To obtain a baseline lipid profile in asymptomatic subjects, overnight fasting blood was drawn after at least 4 weeks without hypolipidemic drug treatment in each lipid clinic participating in the study. None of the control subjects received cholesterol-lowering drugs. In patients with CVD, baseline lipid levels were obtained prior to the cardiovascular event from reviews of clinical histories. Total cholesterol, triglycerides and high-density lipoprotein (HDL) cholesterol were determined with standard enzymatic methods and LDL-cholesterol was estimated with the Friedewald equation. Noncholesterol sterols Noncholesterol sterols, including plant sterols and cholesterol precursors, and their ratios to cholesterol are accepted as surrogate markers for the efficiency of cholesterol absorption and cholesterol synthesis, respectively [14]. In a subsample of ADH subjects who were treatment-naı¨ve for cholesterol-lowering drugs and phytosterol-supplemented foods, serum noncholesterol sterol concentrations were determined by gas chromatography, as described [6]. Noncholesterol sterols are expressed as ratios to cholesterol (102  mmol/mol cholesterol). Inter-assay and intra-assay coefficients of variation for noncholesterol sterols were: 5 and 3.2% for lathosterol; 1.9 and 1.6% for campesterol; 2 and 1.8% for sitosterol; and 7.9 and 4.5% for lanosterol. All biochemical determinations were processed blinded with respect to genetic analysis and in random order.

Genetic studies LDL receptor (LDLR) and Apo B (APOB) The deoxyribonucleic acid (DNA) of individuals with a clinical diagnosis of ADH was screened for LDLR and APOB mutations

238 following standard protocols with sensibility > 99% [15]. Only ADH subjects without a functional mutation in the LDLR or APOB genes were included in this study. NPC1L1 gene polymorphisms Four NPC1L1 gene variants (133A>G rs17655652, 18C>A rs41279633, 1679C>G rs2072183 and 28650A>G rs3187907) were selected because of a minor allele frequency > 0.1 in Caucasian populations and previously reported associations with LDL-cholesterol and/or responses to ezetimibe [11,16]. To analyze the two polymorphisms in the promoter (133A>G and 18C>A), a 365 base pair (bp) DNA fragment was amplified by polymerase chain reaction (PCR) on an Applied Biosystems GeneAmp 9700 system. The primers used to detect the polymorphisms were: forward primer: 50 GAC CCT AGC ACC TGC GTG ATG A-30 and reverse primer: 50 GTA ACG CTC GCC TGG TAC ACG G-30 . The polymorphisms 1679C>G and 28650A>G located in Exon 2 and 30 UTR of the NPC1L1 gene, respectively, were analyzed by two PCR reactions performed simultaneously on a Applied Biosystems GeneAmp 9700, followed by restriction enzyme analysis. For the detection of polymorphism 1679C>G, the primer pairs used were 50 -ACC TGC TCC TGC CAA GAC T-30 (forward) and 50 -GCA GAG GAT GAT GAT GAG GAC-30 (reverse). For the polymorphism 28650A>G: forward primer, 50 -ACC CAG GTG ACT GAA GGC TG-30 ; reverse primer, 50 -CAG GAT CTT CAT CTC CAC TGC-30 . Electrophoretic mobility shift assays (EMSA) Nuclear extracts were prepared from subconfluent Caco-2 cell cultures. The oligonucleotides used in the retardation studies were 50 -GGC AGG GCT CAG CCT CAT T-30 and 50 -GGC AGG GCT CGG CCT CAT T-30 . Nucleotides at the 133A>G mutation are in bold and italic. The preparation of nuclear extracts, labeling of oligonucleotides and electrophoresis were done as described [17]. Exposed EMSA were quantified using a phosphor imager BioRad-GS-363 (Multi-Analyst Software) and corrected according to the density of the probe. For supershift experiments, 1 mg of anti-RXRa (Santa Cruz Biotechnologies, Santa Cruz, CA) was added before the addition of the probe. Cell lines and culture conditions The human hepatocarcinoma cell line HepG2 (H-8065) and the human colon carcinoma cell line Caco-2 (HTB-37) were obtained from the American Type Culture Collection. Both cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with Glutamax-I, 4.500 mg/ L D-glucose, 110 mg/L sodium pyruvate, 100 U/mL penicillin, 100 mg/mL streptomycin-sulfate, 1% nonessential amino acids and fetal bovine serum (5% for Hep G2 and 10% for Caco-2 cells). The cells were cultured in 75 cm2 plastic flasks and incubated at 37  C at 5% CO2. Construction of the reporter plasmids A 1846 bp fragment extending from coordinates 1.796 to þ13 of the human NPC1L1 gene promoter (nucleotide positions are indicated with reference to the ATG start codon) including the polymorphism 133A>G were amplified from genomic DNA of homozygous individuals. The amplification was done with Accuprime Pfx DNA Polymerase

B. Martı´n et al. (Invitrogen, Carlsbad, CA, USA) with primers 50 -AAA AAA AAA AAA GCT AGC GGA CTC TAT CTC TCT GTG GTG C-30 and the reverse primer: 50 TTT TTT TTT TTT AAG CTT CCG GCC TCC GCC ACC CCA GGT C-30 which contain NheI and Hind III restriction sites (underlined). The amplified fragments were digested with these enzymes and cloned into the promoterless plasmid pGL3 Basic (Promega, Madison, WI, USA). The sequences of the plasmids were confirmed by automated capillary sequencing on a GE-Healthcare MegaBACE 500. Plasmid DNA from the constructs was prepared using the HiSpeed Plasmid Midi Kit (Qiagen, Hilden, Germany), according to the manufacturer’s instructions. Transient transfection and luciferase assay Confluent cells (10  106) were harvested and resuspended in 500 mL of PBS supplemented with 50 mL of MgCl2 containing 10 mg of the reporter plasmids. Two-hundred and fifty nanograms of the control plasmid pRL-TK (Promega), in which the luciferase gene Renilla is under the control of the TK promoter, were included as an internal control of transfection efficiency. Transfection was carried out by electroporation at 300 V and 975 mF. The luciferase activity was measured 48 h after transfection. Preparation of cell extracts and measurement of luciferase activity were done with the Dual Luciferase kit (Promega) following manufacturer’s instructions. All transfections were done four times and performed in triplicate with two different preparations of DNA.

Statistical analyses Allele and genotype frequencies were calculated for each single nucleotide polymorphism (SNP) using the chi-square test. HardyeWeinberg equilibrium was tested with the chisquare test. Haploview software package v. 3.32 was used to calculate linkage disequilibrium statistics and estimate haplotype frequencies using the ExpectationeMaximization algorithm [18]. PHASE v.2.1 was used with the case-control setting option to estimate the global haplotype frequencies in the study groups, to perform a permutation (1000 permutations were set up) signification test for differences in haplotype frequencies between groups, and to obtain the most likely pairs of haplotypes for each individual [19]. Association analysis between the genotype and various lipid variables was performed using unpaired t-tests. Comparisons of quantitative variables without a normal distribution were made with nonparametric statistics by using the ManneWhitney test. The between-group relative luciferase activity was compared by using t-tests. Statistical analyses were performed using the Statistical Package for the Social Sciences software, version 12.0 (SPSS Inc Chicago, IL). A p value < 0.05 was considered statistically significant for all analyses.

Results Subject characteristics ADH group was composed of 271 subjects (155 women and 116 men), median age 53 years (interquartile range (IR) 42e61). Control groups included 274 subjects (180 women and 94

NPC1L1 gene and dominant hypercholesterolemia

239

men), median age 34 years (IR 26e48). The frequency of tendon xanthomas and cardiovascular events in the ADH group was 26 (9.6%) and 33 (12.3%) respectively. As expected, the ADH group had significantly higher total cholesterol (333.0  42.0 mg/dL versus 190.9  38.5 mg/dL), LDLcholesterol (252.5  39.4 mg/dL versus 121.0  34.3 mg/dL) and triglycerides (63 IR 45e90 mg/dL versus 110 IR 88e145 mg/dL) than the control group. No between-group differences in HDL-cholesterol were observed (median 54 mg/ dL in both groups). Genotype analyses The genotype distribution of NPC1L1 polymorphisms (133A>G, 18C>A, 1679C>G and 28650A>G) in 271 ADH and 274 controls is presented in Table 1. The genotype frequencies for all SNPs analyzed in cases and controls were in HardyeWeinberg equilibrium. The ADH group had a significantly higher (p Z 0.027) minor allele frequency than the control group for the 133A>G polymorphism (0.29 vs. 0.25). The genotype distribution of 18C>A, 1679C>G and 28650A>G NPC1L1 polymorphisms was similar in ADH cases and controls. No significant associations among NPC1L1 polymorphisms and lipid levels were observed in any group. Haplotype analyses We constructed haplotypes with all four NPC1L1 SNPs. Although 16 different haplotypes were found, six of them accounted for 98.8% of all haplotypes identified. The estimated frequencies of the most common (>2%) haplotypes are shown in Table 2. The PHASE v2.1 permutation test for significant differences in haplotype frequencies in ADH and control groups gave a p value of 0.07, indicating a strong tendency for nonrandom distribution between ADH and control subjects, thus suggesting an association. The three haplotypes consisting of common alleles for the polymorphisms 133A>G and 18C>A (AeC) were more frequent in controls than in ADH subjects. When considering the haplotypes with these two promoter SNPs, the

Table 1 Genotype distributions of NPC1L1 gene polymorphisms in ADH and controlsa SNP

Controls

ADH

p

Genotype distribution

133A>G rs17655652 AA AG GG 18C>A rs41279633 CC CA AA 1679C>G rs2072183 CC CG GG 28650A>G rs3187907 AA AG GG a

n

(%)

n

(%)

159 95 20 196 73 5 155 107 11 189 78 6

58 35 7 71 27 2 57 39 4 69 29 2

131 125 16 177 86 9 146 109 17 182 86 4

48 46 6 65 32 3 54 40 6 67 32 1

most abundant (AC) was significantly (p Z 0.011) more frequent in the control group than in ADH subjects (Table 3). No differences in total and LDL-cholesterol were found among haplotypes. Cholesterol absorption and synthesis markers Table 4 shows the serum noncholesterol sterol to cholesterol ratios by promoter haplotypes in a subgroup of 79 treatment-naı¨ve ADH subjects. No between-gender differences in sterol ratios were observed. The lathosterol ratio was significantly (p Z 0.03) higher and the lanosterol ratio was nonsignificantly (p Z 0.08) higher in AC homozygotes than in the other haplotypes. The ratio of serum campesterol þ sitosterol (plant sterols) to serum lanosterol þ lathosterol (cholesterol precursors), which reflects the balance between cholesterol absorption and synthesis, was nearly significantly (p Z 0.057) lower in AC homozygotes than in the other haplotypes. Functional study of the L133A>G polymorphism To determine if the 133A>G polymorphism influenced promoter strength, we performed reporter gene constructions containing either the 133A or 133G allele upstream of the luciferase reporter gene of pGL3Basic and performed transient transfections in both Caco-2 and HepG2 cell lines. The promoterless vector pGL3Basic was transfected as a negative control. As expected, the cells transfected with the negative control had negligible luciferase activity. Luciferase activity observed with pGL3 (133G) was 2.36fold higher than that seen with the construct pGL3 (133A) in the Caco-2 cell line (p Z 0.012) and 2.82-fold higher in the HepG2 cell line (p Z 0.028), suggesting that the A to G nucleotide change results in increased NPC1L1 promoter activity. To investigate the changes in proteineDNA binding in the vicinity of the 133A>G NPC1L1 polymorphism, we carried out EMSAs using nuclear extracts prepared from Caco-2 cell cultures and the oligonuclotides described. Fig. 1 is representative of any of the three experiments carried out. Caco-2 cell nuclear extracts were able to retard either the A- or the G-allele oligonucleotides, but the retarded bands produced with the A- specific oligonucleotide were always more intense (ratio of 1.76  0.17). The analysis of the sequence with MatInspector [20] predicted the existence of a binding site for the nuclear

0.027

Table 2 Haplotype distribution of NPC1L1 polymorphisms in ADH and in controlsa

0.205

Haplotype

Controls Allele number

Allele number

ACCA GCCA AAGG ACGA AAGA ACCG

129 63 25 21 16 13

111 74 32 20 18 10

0.456

0.631

ADH, autosomal dominant hypercholesterolemia; SNP, single nucleotide polymorphism. p refers to chi-square test.

ADH

p

0.039 0.128 0.162 0.761 0.585 0.338

a ADH, autosomal dominant hypercholesterolemia. p values refer to chi-square tests between specific haplotype and all other haplotypes.

240

B. Martı´n et al.

Table 3 Haplotype distributions of NPC1L1 promoter polymorphisms in ADH and in controlsa Haplotype

AC GC AA GA

Controls

ADH

p

n

(%)

n

(%)

329 136 83 0

60.0 24.8 15.2 0.0

285 155 102 2

52.4 28.5 18.7 0.4

0.011 0.170 0.112 0.476

a

ADH, autosomal dominant hypercholesterolemia. p values refer to chi-square tests between specific haplotype and all other haplotypes.

receptor RXR. In the supershift assay with an RXRa-specific antibody, the intensity of the A-allele band was markedly reduced, in contrast with the G-specific oligonucleotide band (Fig. 1), suggesting that the SNP alters the proteinbinding specificity of the DNA sequence.

Discussion The major finding of this study is that genetic variation in NPC1L1 is associated with ADH without functional mutations in the two major loci responsible for this phenotype, LDLR and APOB. Our results expand the previous knowledge about the contribution of NPC1L1 to the interindividual variation of serum sterols in normolipidemic populations [11,21], and also in the lipid-lowering response to the drug ezetimibe [16]. NPC1L1 plays an important role in the intestinal absorption of sterols, including plant sterols and cholesterol, thus it is conceivable that some forms of ADH might be caused by genetic variants at the NPC1L1 gene locus that result in enhanced sterol absorption. Recently, we showed that increased intestinal cholesterol absorption, as measured by plasma phytosterol levels, is frequent in nonLDLR/nonAPOB ADH subjects [6]. In the present work, NPC1L1 gene variation was associated with the balance between intestinal absorption and hepatic synthesis, as determined by the ratio of campesterol þ sistosterol (two major plant sterols coming entirely from the diet) to lathosterol þ lanosterol (two noncholesterol sterols that are intermediate products of cholesterol synthesis).

Table 4 Serum noncholesterol sterol ratios in 79 ADH subjects according to the NPC1L1 promoter polymorphisms Genotype Campesterol/cholesterol

AC/AC (n Z 20)

190 (131e286) Sitosterol/cholesterol 204  96 Lanosterol/cholesterol 11.3  8.4 Lathosterol/cholesterol 161  77 Absorption/synthesis ratioa 2.16 (1.70e2.91)

All others (n Z 59)

p

210 (152e279) 208  100 8.9  4.0 130  46 2.99 (1.85e4.39)

0.596 0.870 0.082 0.031 0.057

Values are mean  SD or median (interquartile range). Units are 102 mmol/mol cholesterol. a Absorption/synthesis ratio Z (campesterol þ sitosterol)/ (lanosterol þ lathosterol).

Figure 1 The 133A>G SNP results in a differential binding of Caco2 proteins. The four lanes on the left are retardations of labeled A-allele-specific nucleotide. The next four correspond to retardations of labeled G-allele-specific nucleotide. Both A and G alleles were incubated with (E) and without (NE) nuclear extract and a 100 excess of cold nucleotide (either A-allele-specific, G-specific). Supershift assays were carried out with 2 mg of anti-RXR. Retarded band is indicated with a black arrow.

Interestingly, the ADH group showed a lower frequency of the 133A and 18C NPC1L1 allele in association with an increase in cholesterol synthesis markers ra‘ther than a decrease in intestinal absorption markers. Intestinal sterol absorption and hepatic cholesterol biosynthesis are highly and inversely correlated and closely regulated [22] thus it is difficult to establish what the primary effect is from the measurements of plasma sterol levels. Because NPC1L1 is highly expressed in humans in both the apical surface of absorptive intestinal cells and the liver [7], where it localizes to the canalicular membrane [23] and promotes the retention of biliary cholesterol by liver cells [24], we cannot exclude that its genetic variation differentially influences cholesterol translocation from the intestinal lumen to the enterocytes and from bile to the hepatocytes. The higher frequency of the rare 133G NPC1L1 allele among individuals with high plasma cholesterol is consistent with the notion that this gene variant might increase NPC1L1 function. This was confirmed by reporter gene assays which showed that the promoter corresponding to the G-allele produces a stronger protein signal than the one having the A variant (2.3-fold and 2.8fold in Caco-2 and HepG2 cell lines, respectively). Also, the EMSAs showed that the A>G transition decreased the ability of the sequence to bind proteins by an average of

NPC1L1 gene and dominant hypercholesterolemia 1.76  0.17-fold. The fact that the G-variant promoter is stronger while binding less protein suggests that the binding protein might act as a transcriptional repressor. The supershift experiments indicated that RXRa was part of the proteineDNA complexes and that its presence was more important in the complex formed with the A-allele oligonucleotide. The retinoid receptors are emerging as major players in many cellular processes. RXRa is a promiscuous member of this family able to form heterodimers with many other nuclear receptors. As a result it can act as an activator or a repressor by recruiting coactivators or interacting with corepressors, respectively [25]. Therefore, the more intense binding of RXRa might well explain the decrease in promoter activity corresponding to the A-allele. A second important finding of our study is that the two polymorphisms associated with nonLDLR/nonAPOB ADH are located in the promoter region of the NPC1L1 gene. Growing evidence shows that functionally relevant polymorphisms in various promoters alter both transcriptional activity and affinities of existing proteineDNA interactions, thus influencing disease expression in humans [26]. To date, a large number of genetic variants located in the promoter gene regions have been identified and associated with abnormal phenotypes including monogenic diseases, such as those in LDLR causing FH [17,27] or in the cystic fibrosis transmembrane conductance regulator (CFTR) gene in cystic fibrosis [28]. More often, promoter mutations show milder effects than mutations in the coding region and are usually associated with incomplete dominance, an observation that supports the influence of other genetic or environmental factors in determining the clinical phenotype [26]. This is the case in most cases of nonLDLR/nonAPOB ADH subjects. We recently studied a large sample of ADH subjects from Spain and showed that those without functional mutations in LDLR or APOB have lower total and LDL-cholesterol (237  49 mg/ dL versus 302  69 mg/dL, p < 0.001), incidence of tendon xanthomas, family history of premature CVD and presence of hypercholesterolemia from childhood than subjects with FH or FDB [29], clearly suggesting a lower penetrance, higher expression variability and milder phenotype than molecularly defined ADH. A limitation of our work is that we have not studied nuclear families to demonstrate cosegregation of the NPC1L1 gene variants with hypercholesterolemia or sterol absorption/cholesterol synthesis markers. This also impedes assessing whether the NPC1L1-associated hypercholesterolemia phenotype is the result of combinations of common and rare alleles rather than a monogenic disorder. We could not detect any NPC1L1 haplotype unique to nonLDLR/nonAPOB ADH subjects, thus the first model of complex trait genes probably fits better with some forms of genetic hypercholesterolemia, as it occurs with hyperlipoprotein(a) [30]. In summary, the results of our study in a large cohort of subjects with personal and family history of high LDLcholesterol in whom LDLR and APOB genetic testing failed to detect any functional mutation suggest that NPC1L1 promoter variants might explain, at least in part, the hypercholesterolemic phenotype. Our data also demonstrate that the 133A>G NPC1L1 polymorphism exerts a significant effect on intestinal and hepatic cell promoter activity that probably explains part of the cholesterol absorption/synthesis rate variability in these subjects.

241

Conflict of interest None of the authors has any conflict of interest

Acknowledgments The authors thank Dr Juan Ferrando and Dr Jose ´ Puzo for facilitating subjects’ data and blood samples for this study. Beatriz Martı´n is the recipient of a fellowship from Fundacio ´n Cuenca Villoro. Grants were also received from the Spanish Ministry of Health FIS PI06/0365, PI06/1238, RETIC RECAVA and the Spanish Ministry of Education and Science (SAF2005-07042).

References [1] Goldstein JL, Hobbs H, Brown MS. Familial hypercholesterolemia. In: Scriver CR, Beaudet AL, Sly WS, Valle D, editors. The metabolic and molecular basis of inherited disease. 8th ed, vol. 2. New York: McGraw-Hill; 2001. p. 2863e913. [2] Abifadel M, Varret M, Rabes JP, Allard D, Ouguerram K, Devillers M, et al. Mutations in PCSK9 cause autosomal dominant hypercholesterolemia. Nat Genet 2003;34:154e6. [3] Fouchier SW, Kastelein JJ, Defesche JC. Update of the molecular basis of familial hypercholesterolemia in The Netherlands. Hum Mutat 2005;26:550e6. [4] Lammert F, Wang DO. New insights into the genetic regulation of intestinal cholesterol absorption. Gastroenterology 2005; 129:718e34. [5] Kesa ¨niemi YA, Miettinen TA. Cholesterol absorption efficiency regulates plasma cholesterol level in the Finnish population. Eur J Clin Invest 1987;17:391e5. [6] Garcı´a-Otı´n AL, Cofa ´n M, Junyent M, Recalde D, Cenarro A, Pocovı´ M, et al. Increased intestinal cholesterol absorption in autosomal dominant hypercholesterolemia and no mutations in the low-density lipoprotein receptor or apolipoprotein B genes. J Clin Endocrinol Metab 2007;92:3667e73. [7] Altmann SW, Davis Jr HR, Zhu LJ, Yao X, Hoos LM, Tetzloff G, et al. NiemannePick C1 Like 1 protein is critical for intestinal cholesterol absorption. Science 2004;303:1201e4. [8] Davis Jr HR, Zhu LJ, Hoos LM, Tetzloff G, Maguire M, Liu J, et al. NiemannePick C1 Like 1 (NPC1L1) is the intestinal phytosterol and cholesterol transporter and a key modulator of whole-body cholesterol homeostasis. J Biol Chem 2004;279: 33586e92. [9] Davies JP, Scott C, Oishi K, Liapis A, Ioannou YA. Inactivation of NPC1L1 causes multiple lipid transport defects and protects against diet-induced hypercholesterolemia. J Biol Chem 2005; 280:12710e20. [10] Davies JP, Levy B, Ioannou YA. Evidence for a NiemannePick C (NPC) gene family: identification and characterization of NPC1L1. Genomics 2000;65:137e45. [11] Cohen JC, Pertsemlidis A, Fahmi S, Esmail S, Vega GL, Grundy SM, et al. Multiple rare variants in NPC1L1 associated with reduced sterol absorption and plasma low-density lipoprotein levels. Proc Natl Acad Sci U S A 2006;103:1810e5. [12] Go ´mez-Gerique JA, Gutie ´rrez-Fuentes JA, Montoya MT, Porres A, Rueda A, Avellaneda A, et alDRECE study group. Lipid profile of the Spanish population: the DRECE (diet and risk of cardiovascular disease in Spain) study. Med Clin 1999; 113:730e5. [13] Merin ˜o-Ibarra E, Artieda M, Cenarro A, Goicoechea J, Calvo L, Guallar A, et al. Ultrasonography for the evaluation of visceral fat and the metabolic syndrome. Metabolism 2005;54:1230e5. [14] Miettinen TA, Tilvis RS, Kesa ¨niemi YA. Serum plant sterols and cholesterol precursors reflect cholesterol absorption and

242

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

B. Martı´n et al. synthesis in volunteers of a randomly selected male population. Am J Epidemiol 1990;131:20e31. Tejedor D, Castillo S, Mozas P, Jimenez E, Lopez M, Tejedor MT, et alSpanish FH Group. Reliable low-density DNA array based on allele-specific probes for detection of 118 mutations causing familial hypercholesterolemia. Clin Chem 2005;51:1137e44. Simon JS, Karnoub MC, Devlin DJ, Arreaza MG, Qiu P, Monks SA, et al. Sequence variation in NPC1L1 and association with improved LDL-cholesterol lowering in response to ezetimibe treatment. Genomics 2005;86:648e56. Mozas P, Galetto R, Albajar M, Ros E, Pocovi M, RodriguezRey JC. A mutation (49C>T) in the promoter of the low density lipoprotein receptor gene associated with familial hypercholesterolemia. J Lipid Res 2002;43:13e8. Excoffier L, Slatkin M. Maximum-likelihood estimation of molecular haplotype frequencies in a diploid population. Mol Biol Evol 1995;12:921e7. Stephens M, Scheet P. Accounting for decay of linkage disequilibrium in haplotype inference and missing data imputation. Am J Hum Genet 2005;76:449e62. Quandt K, Frech K, Karas H, Wingender E, Werner T. MatInd and MatInspector: new fast and versatile tools for detection of consensus matches in nucleotide sequence data. Nucleic Acids Res 1995;23:4878e84. Fahmi S, Yang C, Esmail S, Hobbs HH, Cohen JC. Functional characterization of genetic variants in NPC1L1 supports the sequencing extremes strategy to identify complex trait genes. Hum Mol Genet 2008;17:2101e7. Santosa S, Varady KA, AbuMweis S, Jones PJ. Physiological and therapeutic factors affecting cholesterol metabolism: does

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

a reciprocal relationship between cholesterol absorption and synthesis really exist? Life Sci 2007;80:505e14. Yu L, Bharadwaj S, Brown JM, Ma Y, Du W, Davis MA, et al. Cholesterol-regulated translocation of NPC1L1 to the cell surface facilitates free cholesterol uptake. J Biol Chem 2006; 281:6616e24. Temel RE, Tang W, Ma Y, Rudel LL, Willingham MC, Ioannou YA, et al. Hepatic NiemannePick C1-like 1 regulates biliary cholesterol concentration and is a target of ezetimibe. J Clin Invest 2007;117:1968e78. Weston AD, Blumberg B, Underhill TM. Active repression by unliganded retinoid receptors in development: less is sometimes more. J Cell Biol 2003;161:223e8. Johnson AD, Wang D, Sadee W. Polymorphisms affecting gene regulation and mRNA processing: broad implications for pharmacogenetics. Pharmacol Ther 2005;106:19e38. Smith AJ, Ahmed F, Nair D, Whittall R, Wang D, Taylor A, et al. A functional mutation in the LDLR promoter (139C>G) in a patient with familial hypercholesterolemia. Eur J Hum Genet 2007;15:1186e9. Rowntree R, Harris A. DNA polymorphisms in potential regulatory elements of the CFTR gene alter transcription factor binding. Hum Genet 2002;111:66e74. Civeira F, Ros E, Jarauta E, Plana N, Zambon D, Puzo J, et al. Comparison of genetic versus clinical diagnosis in familial hypercholesterolemia. Am J Cardiol 2008;102:1187e93. Merin ˜o-Ibarra E, Puzo J, Jarauta E, Cenarro A, Recalde D, Garcı´a-Otı´n AL, et al. Hyperlipoproteinaemia(a) is a common cause of autosomal dominant hypercholesterolaemia. J Inherit Metab Dis 2007;30:970e7.