Introduction

Cardiovascular disease is a worldwide problem and hypercholesterolemia contributes to cardiovascular diseases by promoting atherosclerosis.1 Cholesterol homeostasis in the human body is tightly regulated through dietary cholesterol absorption, biliary secretion and de novo synthesis. There are variations in serum cholesterol levels and inter-individual variations in cholesterol absorption rates might be a possible reason. It has been reported that cholesterol absorption is influenced by several factors, such as dietary composition, age and genetic factors.

Recently, the Niemann–Pick type C1-like 1 protein (NPC1L1) was identified as an intestinal cholesterol transporter.2, 3, 4, 5, 6, 7, 8 It is also the molecule target for the new cholesterol-lowering agent, ezetimibe.9 Human NPC1L1 is expressed mainly in the intestine and liver and is responsible for transporting cholesterol from intestinal lumen to enterocytes. It is now believed that intestinal cholesterol absorption is regulated by two transporters, NPC1L1 and ABCG5/8.10, 11 ABCG5/8 is a heterodimer transporter and may be responsible for selective reverse cholesterol transport from the enterocytes to the intestinal lumen. Several genetic variations in ABCG5/8 have been identified to cause sitosterolemia, a rare disorder characterized by increased serum sitosterol levels.12, 13, 14, 15 However, there are a limited number of studies that focus on the effects of NPC1L1 genetic variation. Therefore, the aim of this study was to identify genetic variations of NPC1L1 and to investigate whether these variations are associated with variations in serum cholesterol levels.

Materials and methods

Study subjects

A total of 224 consecutive individuals undergoing health check-ups at the National Taiwan University hospital were included in the study. Serum levels of total cholesterol, triglyceride, low-density lipoprotein (LDL)-cholesterol and high-density lipoprotein-cholesterol were determined independently. The demographic data and serum cholesterol profiles are shown in Table 1.

Table 1 Demographic data and serum lipid levels of the study subjects (n=224)
Full size table

Detection of genetic polymorphisms and genotyping

We screened the promoter and coding regions of the NPC1L1 gene for genetic polymorphisms by direct DNA sequencing of genomic DNA from 50 individuals. Total cellular DNA was isolated from peripheral blood lymphocytes using a PureGene genomic DNA purification kit (Gentra Systems Inc., Valencia, CA, USA). The promoter (from −971 to +1 bp) and coding regions of the NPC1L1 gene were amplified using PCR. The base sequence of the primer pairs is shown in Table 2. PCR was performed with an initial denaturation step of 95 °C for 5 min, followed by 35 cycles of denaturation at 95 °C for 30 s, annealing at 55 °C for 30 s and extension at 72 °C for 30 s. A final elongation step was performed at 72 °C for 10 min. DNA sequencing was performed using an automated sequence analyzer. Further genotyping was performed on the 224 individuals using PCR and direct DNA sequencing.

Table 2 Base sequence of the primer pairs used in PCR
Full size table

Plasmid construction

The promoter region between −1405 and +52 bp of the human NPC1L1 gene was amplified using human genomic DNA as the template. KpnI and BglII restriction sites were introduced into the forward and reverse primers, respectively, to facilitate subsequent cloning work. The forward primer was: 5′-CAGGTACCGGCTCATTCCCCATTCACCGACTGC (underline indicates KpnI cutting site) and the reverse primer was: 5′-GTAGATCTAGCGCAGGAGCAGGGCCCACAGCAG (underline indicates BglII cutting site). The PCR product was then digested with KpnI and BglII enzymes and inserted into a luciferase gene control vector, pGL3 basic vector (Promega, Madison, WI, USA). All constructs were confirmed by DNA sequence.

Cell culture and transient transfection

HepG2 cells were cultured in Dulbecco's modified Eagle's medium containing 2 mM L-glutamine, 1.5 g l–1 sodium bicarbonate, 100 μM non-essential amino acids, 1.0 mM sodium pyruvate and supplemented with 10% fetal bovine serum at 37 °C with 5% CO2.

For transfection studies, HepG2 cells were plated onto 96-well plate (8 × 104 cells per well) before 24-h transfection.16 The cells were transiently co-transfected with 250 ng of pGL3 basic luciferase vector, pGL-762T or pGL-762C constructs, and 25 ng pRL-TK vector (Renilla luciferase-bearing vector) using jetPEI transfection reagent according to the manufacturer’s instructions. In all transfections, pRL-TK vector (Renilla luciferase-bearing vector—Promega) served as an internal control to normalize the efficiency of transfection. After 24-h transfection, the cells were washed with phosphate-buffered saline solution and incubated for an additional 24 h in either fresh medium, 40 μM of lovastatin, 10 μg ml–1 or 20 μg ml–1 cholesterol (Sigma, St Louis, MO, USA), and 5 or 15 μM of ezetimibe (Schering-Plough/Merck, Kenilworth, NJ, USA).17 Finally, luciferase assay was performed using the Dual-GloLuciferase Reporter Assay System according to the manufacturer's instructions (Promega).

Statistical analysis

Data were expressed as mean±s.e.m. Parametric data between groups were compared using the Student's t-test. A P-value of less than 0.05 was considered statistically significant. Linkage disequilibrium and haplotype analyses were performed using Powermarker software. For comparing groups of cells treated with increasing concentrations of ezetimibe, linear regression was calculated by GraphPad Prism software. Hardy–Weinberg equilibrium was assessed by the χ2 test.

Results

Identification of NPC1L1 polymorphisms

To detect genetic polymorphisms of the NPC1L1 gene, genomic DNA from 50 individuals (100 chromosomes) were analyzed. In total, 11 single nucleotide polymorphisms (SNPs) were identified. Among these, five were in the NPC1L1 coding region, with two causing amino-acid changes (P282T and A1238V). For the remaining six SNPs, four were intronic SNPs, and two located in the promoter region. The allele frequencies of the SNPs are shown in Table 3. Among the 11 SNPs, g.−762T>C (ss119336593) and g.1679C>G (rs2072183) were common and the frequencies of the minor allele were 34 and 36%, respectively. In addition, our results revealed that the two common SNPs, g.−762T>C and g.1679C>G, were highly linked (D′ value=0.7459, P-value <0.00001) (Table 4).

Table 3 The NPC1L1 SNPs and their allele frequencies identified in 50 subjects
Full size table
Table 4 Allele frequencies, linkage disequilibrium and haplotype analysis for g.−762T>C and g.1679C>G
Full size table

Promoter assay for NPC1L1 promoter SNP

As to the two common SNPs, g.−762T>C was in the promoter region and g.1679C>G was synonymous, we therefore performed luciferase assay to assess whether g.−762T>C contributed to variations in the promoter activity of the NPC1L1 gene. We cloned two promoter constructs (pGL-762T and pGL-762C), and transient transfection into HepG2 cells was performed. In luciferase assays, both of the promoter constructs (pGL-762T and pGL-762C) showed a significant increase in promoter activity compared with the pGL3 basic vector without NPC1L1 promoter region (3.1-fold, P-value=0.0278 and 10.8-fold, P-value=0.0001, respectively) (Figure 1). In addition, the pGL-762C construct showed significantly higher promoter activity than that of pGL-762T construct (3.5-fold, P-value=0.00157), suggesting that the minor C allele may contribute to increase NPC1L1 transcript production (Figure 1).

Figure 1
figure 1

Effect of exogenous cholesterol or lovastatin on the Niemann–Pick type C1-like 1 (NPC1L1) promoter activity. pGL-762T and pGL-762C indicate −762T and −762C promoter constructs, respectively. pGL3 basic represents a negative control vector, which does not have a NPC1L1 promoter sequence. HepG2 cells were transiently transfected with the promoter constructs. pGL-762C had a significantly higher promoter activity than pGL-762T (P<0.005). Supplementing with cholesterol (10 or 20 μg ml–1) decreased NPC1L1 promoter activity, while adding lovastatin (40 μM) increased promoter activity. Data are means±s.e.m. (n=∼3–6).

Full size image

Effect of added cholesterol, lovastatin and ezetimibe on the promoter activity of NPC1L1

We further investigated whether the −762T and −762C differed in their response to cholesterol, lovastatin (a cholesterol synthesis inhibitor) as well as ezetimibe (an intestinal cholesterol absorption inhibitor). When HepG2 cells were incubated with cholesterol at 10 μg ml–1, the promoter activity of constructs, pGL-762T and pGL-762C, decreased slightly (0.95- and 0.82-fold, respectively). Furthermore, when cells were treated with higher cholesterol contents (20 μg ml–1), the promoter activity of both pGL-762T and pGL-762C constructs declined significantly (0.78- and 0.65-fold, respectively). This result suggested that the inhibitory effect of cholesterol was dose dependent. In contrast, when lovastatin was added, the promoter activity showed a significant increase (2.17-fold for pGL-762T and 1.76-fold for pGL-762C), demonstrating that the promoter activity of NPC1L1 was regulated by cholesterol level. However, when cells were treated with either lovastatin (40 μM) or cholesterol (20 μg ml–1), the promoter activity of pGL-762C construct was still significantly higher than that of pGL-762T (2.8-fold, P-value=0.04545, n=5–6 for treatment lovastatin; 2.9-fold, P-value=0.00003, n=4–6 for treatment cholesterol) (Figure 1).

We further investigated the response of the two promoter constructs to ezetimibe. Although cells were incubated with 5 μM of ezetimibe, the promoter activity of pGL-762T and pGL-762C dropped by 0.79- and 0.89-fold, respectively, respectively). At a higher concentration of ezetimibe (15 μM), the promoter activity further declined (0.70- and 0.69-fold of basal level, respectively). These results demonstrated that the inhibitory effect of ezetimibe on the NPC1L1 promoter activity was dose dependent (R2=0.3683, P-value=0.0164 for pGL-762T and R2=0.3201, P-value=0.035 for pGL-762C by linear regression). When comparing the two promoter constructs, the promoter activity of pGL-762C construct was significantly higher than that of pGL-762T at both ezetimibe concentrations (Figure 2).

Figure 2
figure 2

Effect of ezetimibe on the Niemann–Pick type C1-like 1 (NPC1L1) promoter activity. HepG2 cells were transiently transfected with two promoter constructs, pGL-762T and pGL-762C. After 24 h, cells were incubated with ezetimibe for another 24 h. The promoter activities of both pGL-762T and pGL-762C were decreased by ezetimibe treatment. The promoter activity of pGL-762C remained significantly higher than that of pGL-762T. Data are means±s.e.m. (n=∼4–7).

Full size image

Association between NPC1L1 polymorphisms and serum cholesterol levels

For analysis of association between NPC1L1 SNP and cholesterol levels, only SNPs with a frequency of >5% were used. Hardy–Weinberg equilibrium was tested for g.−762T>C and g.1679C>G. Both did not significantly violate the Hardy–Weinberg equilibrium (P-value=0.0687 and 0.1834 in the χ2 test). Tables 5 and 6 show the results of association analysis of serum cholesterol levels and genotypes for SNP g.−762T>C and g.1679C>G, respectively. This demonstrated that the genotype of g.−762T>C was significantly associated with serum total cholesterol and LDP-cholesterol level in a recessive genetic model (P-value=0.0477 and 0.0494, n=224), but was not associated with triglyceride and high-density lipoprotein-cholesterol level. The minor allele (g.−762T>C) had a frequency of 34.1% (n=153/448), and individuals with homozygous −762C allele had 9.0% higher serum total cholesterol and 12.7% higher serum LDL-C, respectively (total cholesterol value=214.7±9.0 vs 196.9±2.6 mg dl–1, P-value=0.0477; LDL-cholesterol value=131.2±8.1 vs 116.4±2.2 mg dl–1, P-value=0.0494) (Figure 3).

Table 5 Association between serum cholesterol levels and g.−762T>C polymorphism
Full size table
Table 6 Association between serum cholesterol level and g.1679C>G polymorphism
Full size table
Figure 3
figure 3

Serum cholesterol levels in patients with different Niemann–Pick type C1-like 1 (NPC1L1) genotypes. In a recessive model, homozygous carriers of the g.−762T>C minor allele had significantly higher serum total cholesterol (by 9.0%) and low-density lipoprotein (LDL)-cholesterol (by 12.7%) levels. *P-value <0.05.

Full size image

If we divided the patients into normocholesterolemic (total cholesterol <240 mg dl–1) and hypercholesterolemic (total cholesterol <240 mg dl–1) groups, the genotype distribution of g.−762T>C differed significantly between the two groups (Table 7). On the other hand, the genotype distribution of g.1679C>G did not differ significantly between the two groups (Table 7). We also analyzed the genotype distributions of both polymorphisms by using 200 mg dl–1 as the cholesterol cutoff value. We found that there were no significant differences in the genotype distributions in both polymorphisms.

Table 7 Genotype distributions of two common single nucleotide polymorphisms of the NPC1L1 gene in normocholesterolemic and hypercholesterolemic patients
Full size table

Discussion

The NPC1L1 gene encodes for the intestinal protein that is responsible for cholesterol absorption. This protein is also the target of the new cholesterol-lowering agent, ezetimibe. As diagnosis and treatment of hyperlipidemia are important clinical issues, there has been much research effort focusing on genetic polymorphisms of the NPC1L1 gene. It has been shown that genetic variations in NPC1L1 result in differences in response to ezetimibe. Hegele et al.18 reported that certain haplotypes (incorporating g.1679C>G, g.25286A>C and g.27621T>C) are associated with inter-individual variations in drug response. Simon et al.19 also demonstrated that patients carrying g.−18C>A had better response to ezetimibe therapy. However, these variations did not influence baseline cholesterol levels. Cohen et al.,20 on the other hand, demonstrated that rare non-synonymous SNPs of the NPC1L1 gene were associated with baseline plasma LDL-cholesterol levels. However, the incidences of these SNPs were extremely low. In this study, we demonstrated that the allele frequencies of the NPC1L1 polymorphisms in people of Chinese ethnicity were quite different from those reported in Western countries. For instance, g.−762T>C polymorphism was common in Chinese with a minor allele frequency of 34.1% in this, whereas its frequency is extremely low (<1%) in Caucasians as reported by Simon et al. On the other hand, SNP g.−18C>A is common in Western reports, but was absent in the 50 individuals of our study. For another common SNP, g.1679C>G (L272L), the minor allele frequency was similar (30–40%) among our study and previous reports.

Functional studies were performed for the promoter polymorphism (g.−762T>C) and we demonstrated that the minor allele contributed to a higher promoter activity. We also showed that the minor allele was associated with higher serum cholesterol levels. Theoretically, variations in NPC1L1 transcript amount might be related to variations in baseline cholesterol levels. In NPC1L1 knockout mice, intestinal cholesterol absorption decreased by 69%. In humans, Miettinen and Kesaniemi21 reported that lower cholesterol absorption was associated with lower serum levels of cholesterol. It is therefore reasonable that variations in promoter activity might be related to changes in serum cholesterol levels.

Our functional analysis of the NPC1L1g.−762T>C polymorphism suggested that homozygous carriers of the minor allele of g.−762T>C might have more NPC1L1 gene expression in enterocytes, leading to an increase in cholesterol absorption. We also demonstrated that NPC1L1 promoter activity was regulated by cholesterol concentration, suggesting that NPC1L1 promoter regulation was similar to other cholesterol metabolism genes, such as HMG-CoA reductase, LDL receptor and high-density lipoprotein receptor.22, 23, 24 Besides, our results were consistent with data reported by other group, NPC1L1 transcription activity was inhibited while cells were treated with ezetimibe.25 In addition, it has been reported that NPC1L1 promoter is regulated mainly by SREBP-2, a transcription factor-regulated cholesterol metabolism-related genes expression.17 Both putative sterol regulatory elements and Ying Yang-1 (YY1)-binding sites are present in the promoter region of NPC1L1, suggesting that these transcription factors may regulate the NPC1L1 gene expression.26 Alrefai et al.17 also demonstrate that there are two putative sterol regulatory elements (∼−657 to −648 and ∼−35– to −26 bp) within human NPC1L1 promoters. YY1 has been proved to be a negative regulator of SREBP-dependent genes.27, 28, 29 However, it is still unknown whether YY1 also participates in regulating NPC1L1 expression. According to a report by Davies et al., there are two putative YY1 conserved binding sites located in the promoter (∼−771 to −752 bp) and coding region (∼−10 to +11 bp) of the NPC1L1 gene. The consensus sequence of the YY1 binding site is 5′-CCAT-3′ or 5′-ACAT-3′. The g.−762T>C polymorphism is located in the YY1 conserved binding sites in the promoter region. The minor allele might cause a reduction of YY1's DNA-binding activity, reducing the transcription inhibition effect of YY1 and leading to an increase in NPC1L1 transcripts.

In summary, our studies showed that SNP g.−762T>C is significantly associated with serum total cholesterol and LDL-cholesterol levels. This was consistent with our finding that the minor allele promoter construct (g.−762T>C) showed significantly higher promoter activity than the major allele construct. These results suggest that homozygous carriers of the minor allele of g.−762T>C might have more NPC1L1 expression in enterocytes, and therefore leads to increased cholesterol absorption. However, it is still unclear whether g.−762T>C is associated with hypercholesterolemia-related diseases (for example, atherosclerosis), and this needs to be further investigated.