1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Susceptibility to develop nonalcoholic fatty liver disease (NAFLD) has genetic bases, but the associated variants are uncertain. The aim of the present study was to identify genetic variants that could help to prognose and further understand the genetics and development of NAFLD. Allele frequencies of 3,072 single-nucleotide polymorphisms (SNPs) in 92 genes were characterized in 69 NAFLD patients and 217 healthy individuals. The markers that showed significant allele-frequency differences in the pilot groups were subsequently studied in 451 NAFLD patients and 304 healthy controls. Besides this, 4,414 type 2 diabetes mellitus (T2DM) cases and 4,567 controls were genotyped. Liver expression of the associated gene was measured and the effect of its potential role was studied by silencing the gene in vitro. Whole genome expression, oxidative stress (OS), and the consequences of oleic acid (OA)-enriched medium on lipid accumulation in siSLC2A1-THLE2 cells were studied by gene-expression analysis, dihydroethidium staining, BODIPY, and quantification of intracellular triglyceride content, respectively. Several SNPs of SLC2A1 (solute carrier family 2 [facilitated glucose transporter] member 1) showed association with NAFLD, but not with T2DM, being the haplotype containing the minor allele of SLC2A1 sequence related to the susceptibility to develop NAFLD. Gene-expression analysis demonstrated a significant down-regulation of SLC2A1 in NAFLD livers. Enrichment functional analyses of transcriptome profiles drove us to demonstrate that in vitro silencing of SLC2A1 induces an increased OS activity and a higher lipid accumulation under OA treatment. Conclusions: Genetic variants of SLC2A1 are associated with NAFLD, and in vitro down-regulation of this gene promotes lipid accumulation. Moreover, the oxidative response detected in siSLC2A1-THLE2 cells corroborated the antioxidant properties previously related to this gene and linked the most representative clinical characteristics of NAFLD patients: oxidative injury and increased lipid storage. (HEPATOLOGY 2013)

Nonalcoholic fatty liver disease (NAFLD) has become one of the major causes of chronic liver disease worldwide1 as a result of the increased prevalence of obesity and type 2 diabetes mellitus (T2DM). The term NAFLD is used to describe a wide spectrum of fatty liver changes ranging from simple steatosis to nonalcoholic steatohepatitis (NASH).2 Although simple steatosis usually follows a benign course, NASH is prone to progress to hepatic fibrosis, cirrhosis, and/or hepatocellular carcinoma, leading to increased morbidity and mortality. It has been described that liver failure is the third foremost cause of death among people with NAFLD.1 The molecular mechanisms associated with the development and progression of this disease needs to be further explored. In this context, early identification of patients predisposed to develop NAFLD would be helpful to set up procedures that modify patients' lifestyle to avoid the progression to severe liver disease and/or complications.3, 4

During the last decade, genetic association studies have proved to be useful for identifying biomarkers for susceptibility to complex multifactorial diseases, including NAFLD,5, 6 but, in most of the cases, there is a lack of functional testing of the identified genes. Therefore, the aim of the present study was to identify genetic variants that are associated with, and directly influence the development of, NAFLD. For this purpose, a candidate gene-association study was performed, followed by functional analyses of the associated gene.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information


Collection of each cohort was approved by the relevant institutional ethics committees, and all subjects gave informed consent. All biopsies from NAFLD patients were revised and interpreted by a single pathologist at each hospital, always under the criteria defined by Brunt et al.,7 and the scoring system followed was the one described by Kleiner et al.8

Pilot Study.

Among all the samples obtained from the collaborative hospitals (Principe de Asturias Hospital, Madrid, Spain; Clinic Hospital, Barcelona, Spain; and Galdakao Hospital, Galdakao, Spain) only biopsy-proven NAFLD samples that were nondiabetic were chosen (69 cases). DNA was isolated from filter-paper blood and amplified with the REPLIg Mini Kit (Qiagen, Madrid, Spain). DNA from 217 control individuals was obtained from the DNA bank of the BIOEF Foundation (http://www.biobancovasco. org/).

Validation Phase.

For the validation stage, DNA from 451 patients provided by INSERM (Nice, France), INSERM (Paris, France), and Hospital Universitario Santa Cristina (Madrid, Spain) was tested (Table 1). All NAFLD cases were biopsy proven and nondiabetic. In addition, DNA from 24 biopsy-proven control individuals was supplied by the Hospital Universitario Santa Cristina, and DNA from other 279 control individuals was obtained from the DNA bank of the BIOEF Foundation, taking into account the following inclusion criteria: (1) absence of insulin resistance syndrome (i.e., no traces of hyperglycemia, hyperlipidemia, hypertension, or obesity); (2) healthy liver activity tested by measuring the levels of transaminases (aspartate aminotransferase [AST] and alanine aminotransferase [ALT]), gamma- glutamyl transferase (GGT), and total bilirubin concentration; and (3) body mass index (BMI) ≤30 kg/m2.

Table 1. Clinical Characteristics of Validation-Phase Participants (Cases and Controls)
Cases (n total = 451)MinimumMaximumAverageMedianStandard Deviation
  1. BMI, body mass index; AST/SGOT, aspartate aminotransferase/serum glutamic oxaloacetic transaminase; ALT/SGPT, alanine transaminase/serum glutamic pyruvate transaminase; gGT, gamma glutamyl transferase; HDL, high density lipoprotein; N, Number of individuals considered in each group; NA, not available.

BMI, kg/m222.7671.144.7544.498.44
AST/SGOT, IU/L119925.292212.2
ALT/SGPT, IU/L831934.972727.33
Glycated hemoglobin, %4.411.165.61.28
Glucose, mmol/L3.7320.156.445.612.5
Insulin, mIU/L212818.0714.114.03
GGT, IU/L935647.463444.55
Bilirubin, μmol/L3448.5174.51
Total cholesterol, mmol/L2.610.525.275.21.14
HDL cholesterol, mmol/L0.232.631.291.260.33
TGs, mmol/L0.5119.51.771.51.36
Controls (n total = 304)MinimumMaximumAverageMedianStandard Deviation
BMI, kg/m214.935.7524.8124.482.7
AST/SGOT, IU/L74721.11215.17
ALT/SGPT, IU/L66418.44177.14
Glycated hemoglobin, %4.265.0550.24
Glucose, mmol/L4.
Insulin, mIU/L2.211.26.467.42.86
GGT, IU/L35314.79136.25
Bilirubin, μmol/L2.5718.
Total cholesterol, mmol/L2.46.574.444.50.63
HDL cholesterol, mmol/LN/AN/AN/AN/AN/A
TGs, mmol/L0.967.992.252.120.82

T2DM Study.

The T2DM case-control study was carried out using Danish glucose-tolerant subjects from the Inter99 study9 (n = 4,567) and from the Steno Diabetes Center (SDC) (n = 730). Danish T2DM cases were obtained from the SDC (n = 1,695), the ADDITION study10 (n = 1,609), and from the Inter99 study (n = 380).

Liver Biopsies.

DNA and RNA were extracted from liver biopsies using the AllPrep DNA/RNA mini kit (Qiagen) from NAFLD patients and control individuals supplied by the same institutions that provided the DNAs for the validation phase.

Selection of Genes and Single-Nucleotide Polymorphisms.

According to previous studies concerning the transcriptomics, proteomics, and metabolomics of NAFLD,11-13 92 genes (Supporting Table 1) were selected. Additionally, genes encoding proteins involved in the methionine cycle were also included.14, 15


For the pilot association study, 3,072 single-nucleotide polymorphisms (SNPs) (Supporting Table 1) were selected (as described in the Supporting Materials) and genotyped in 69 NAFLD cases and 217 controls DNA samples using the GoldenGate technology developed by Illumina, Inc. (San Diego, CA). Validation of the five most significantly associated SNPs was carried out using predesigned TaqMan Assays (Applied Biosystems, Foster City, CA), following the manufacturer's instructions.

Variants used in the T2DM association study were genotyped by Kaspar (KBiosciences, Hoddesdon, UK).

Association Study.

Data obtained by GoldenGate Assay were decoded and corrected in Genome Studio (2008; Illumina). TaqMan genotyping data obtained at the CFX-96 (Bio-Rad, Hercules, CA) were visualized and analyzed in a preliminary way using Bio-Rad CFX Manager v1.5 software.

Obtained genotypes and allele frequencies of NAFLD versus control individuals were compared using PLINK16 software v. 1.06. The analysis was done using an allelic test of single- and multimarker association, including all individuals. Data-filtering criteria were minor allele frequency (MAF) ≥0.01 and Hardy-Weinberg equilibrium (HWE) ≥0.001. Calculation of r2 and linkage disequilibrium (LD)-block estimation17 were analyzed in Haploview v. 4.118 (MAF ≥0.01 and HWE ≥0.001).

Genotyping concordance between genotyping experimental procedures was estimated with PLINK.16

Expression of SLC2A1 in Liver Biopsies.

Total RNA extracts were isolated from liver biopsies of control individuals and NAFLD patients by the DNA/RNA mini kit (Qiagen). Retrotranscription (RT) and quantitative polymerase chain reactions (qPCRs) were performed as described in the Supporting Materials. The ratio (Cthousekeeping/Cttarget gene) was calculated as described by Giulietti et al.19

Gene Silencing in THLE2 Cells.

The THLE2 human liver cell line20 was purchased from ATCC (The Global Bioresource Center) and grown in bronchial epithelial cell basal medium, plus a growth factor BulletKit (Lonza, Basel, Switzerland).

THLE2 cells were transfected as described elsewhere21 with 100 nM of short interfering RNA (siRNA) constructs that target SLC2A1 (Hs_SLC2A1_2 FlexiTube siRNA, SI00089264; Qiagen) and a negative control siRNA (AllStars negative control siRNA, 1022076; Qiagen) using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) twice every 24 hours during 48 hours. Silencing was checked by qPCR and western blotting, as described in the Supporting Materials.

Whole Genome Expression Profile of Silenced Hepatocytes.

Total RNA was extracted from triplicated siSLC2A1 and control THLE2 cell cultures using the RNeasy Mini Kit (Qiagen). Whole genome expression analysis was conducted using HumanHT12-v3 Expression BeadChips (Illumina). Complementary RNA synthesis was done with the Illumina TotalPrep RNA Amplification Kit (Ambion, Austin, TX), and subsequent amplification, labeling, and hybridization was performed according to the Illumina protocols.

Microarray Gene-Expression Data Analysis.

Raw expression data were log2 transformed and quantile normalized using the lumi package.22 To identify genes differentially expressed between SLC2A1-silenced and control cells, we employed the limma package.22

To understand the biological processes involved, we used two different enrichment functional analysis tools: gene ontology (GO)23 analysis, using FatiScan,24 (as detailed in the Supporting Materials), and Ingenuity (IPA; Ingenuity Systems, Inc., Redwood City, CA). For the Ingenuity analysis, only genes with a fold change >1.5 or <−1.5 (P < 0.05) were considered.

Determination of OS in SLC2A1-Silenced THLE2 Cells.

The quantity of intracellular reactive oxygen species (ROS) production was determined using dihydroethidium (DHE; Molecular Probes/Invitrogen, Carlsbad, CA),25 and fluorescent intensity was measured by fluorescent microscopy (Axio Imager D1; Carl Zeiss AG, Oberkochen, Germany). Images were analyzed by Axio Imager D1 software. Immunofluorescence (IF) intensity of approximately 10 cells from random fields was quantified for each sample using ImageJ software (National Institutes of Health, Bethesda, MD) and expressed as relative IF intensity. To minimize variations in measurements, all specimens were immunolabeled at the same time and under identical conditions.

Oleic Acid Treatment in SLC2A1-Silenced THLE2 Cells.

To induce triglyceride (TG) accumulation, cell-culture medium was supplemented with 1 mM of oleic acid (OA) bound to fatty-acid–free bovine serum albumin (BSA) (2:1 molar ratio). Controls were supplemented with BSA alone. THLE2 cells were kept for 16 hours in a 5% CO2/95% air incubator at 37°C. After treatments, the coverslips were washed with phosphate-buffered saline (PBS), and cells were fixed with 3.7% formaldehyde. Cells were permeabilized with 0.1% saponine and blocked with 10% fetal bovine serum in PBS. Cells were incubated for 1 hour with BODIPY 493/503 (1:200 dilution; Invitrogen) in blocking solution.

Fluorescent and differential interference contrast (DIC) images of fixed cells were acquired with an Olympus Fluoview FV500 confocal microscope (SGIker, UPV/EHU; Olympus Optical Co. Europa, Hamburg, Germany). To image BODIPY 493/503 staining, cells excited at 488 nm with an argon laser and fluorescence emission between 505 and 525 nm were captured. Hoechst-stained nuclei were excited with a laser diode at 405 nm and fluorescence emission was acquired with a 430-460-nm band-pass filter. Sequential acquisition of each fluorochrome was performed to avoid overlapping of fluorescent emission spectra.

Amount of protein was measured by the bicinconinic acid method, following manufacturer (Pierce Chemical Company, Rockford, IL) instructions, using BSA as the standard. Two percent sodium dodecyl sulfate was used to avoid overestimation in the highly lipidic samples.26 TG measurements were performed using 100 μg of cellular protein by a commercial kit (Menarine, Pomezia, Italy), following manufacturer instructions.


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

SNPs Located Within SLC2A1 Showed Association With NAFLD.

Pilot phase

A total of 1,536 SNPs, successfully genotyped in 69 cases and 217 controls, were considered for further analysis according to the described filtering criteria. Population stratification was null, based on the genotyped samples (see Supporting Materials). Eleven SNPs were shown to be significantly associated (P < 10−3) with NAFLD for the single-marker allelic test (Table 2A). These significant SNPs are located in the following genes: cytochrome P450, family 2, subfamily E, polypeptide 1 (CYP2E1), solute carrier family 2 (facilitated glucose transporter) member 1 (SLC2A1), 5-methyltetrahydrofolate-homocysteine methyltransferase (MTR), and serine/threonine kinase 11 (STK11). Seven of the eleven associated SNPs are located in SLC2A1, and all of them are in high LD (see Supporting Materials [Haplotype association analyses]). ,  

Table 2. Association Results in the Pilot and Validation Genotyping Studies A. Most significant (P < 10−3) associated SNPs in the pilot study
SNPChrGeneA1F_AF_UA2Chi-SquareP ValuePosition Within Gene
  1. Single-marker allelic associations were performed using standard chi-square tests within PLINK software.16

  2. Abbreviations: Chr, chromosome; A1, minor allele; A2, major allele; F_A, frequency of the minor allele in the affected population (cases); F_U, frequency of the minor allele in the unaffected population (controls).

rs2896938710CYP2E1A0.060T27.521.56E-07   Exon-9
rs10448791SMG7 (flanking SLC2A1)G0.630.41C17.432.98E-05   3'UTR
rs17708101SLC2A1A0.280.13G16.45.13E-05   Intron
rs8418561SLC2A1A0.270.12C15.398.73E-05   Intron
rs37542551MTRA0.260.44G13.592.28E-04   Intron
rs725903319STK11G0.330.52C12.933.24E-04   Intron
rs8418581SLC2A1A0.230.11C12.075.12E-04   Intron
rs46581SLC2A1G0.290.15C11.716.23E-04   3'UTR
rs8418481SLC2A1A0.280.15G11.576.69E-04   Intron
rs37542231SLC2A1A0.270.15T11.417.29E-04   Intron
rs22296821SLC2A1A0.270.14G119.12E-04   Exon-6
Table  . B. Validation of the Most Significant (P < 10−3) Associated SNPs in the Pilot Study (i)
GeneCHRSNPBPA1F_AF_UA2Chi-SquareP ValueOR95% CI inf95% CI sup
  1. Single-marker allelic associations were performed using standard chi-square tests within PLINK software.16

  2. Abbreviations: A1, minor allele; A2, major allele; F_A, frequency of the minor allele in the affected population (NAFLD cases); F_U, frequency of the minor allele in the unaffected population (controls); NA, not available.

Table  . (ii)
LocusHaplotypeF_AF_UChi-SquareDFP ValueSNPs
  1. Sliding window (of two adjacent SNPs when available) allelic associations were calculated using standard chi-square tests within PLINK software.16 (i): single allelic association; (ii): sliding window allelic association.

  2. Abbreviations: Haplotype, allelic combination of the two adjacent studied SNPs; F_A, frequency of each haplotype in the affected population (NAFLD cases); F_U, frequency of each haplotype in the unaffected population (controls).

Validation phase

Genotyping reproducibility was tested by comparing the genotypes of the pilot study samples obtained by GoldenGate (Illumina) and TaqMan (Applied Biosystems) technologies. The resulting concordance rate was 0.98.

A genotyping confirmation study was carried out with 451 biopsy-proven NAFLD patients and 304 controls (Table 1). The two SNPs located in SLC2A1 (Supporting Fig. 1, rs4658 and rs841856) increased their significance of association to 8.08 × 10−6 (odds ratio [OR], 1.76; 95% confidence interval [CI]: 1.25-2.48) and 5.93 × 10−5 (OR, 1.83; 95% CI: 1.32-2.53), respectively. However, the SNPs within MTR, CYP2E1, and STK11 did not maintain their significance when increasing the number of genotyped samples (P > 10−3). Additionally, multimarker analysis (sliding windows of 2 nearby SNPs, PLINK16; Table 2B) of rs4658 and rs841856 showed that the ancestral variant (C+G alleles) was less present in the affected population (P = 2.38 × 10−7). Therefore, as expected, the minor alleles (G+T) were more frequent in the case population (P = 1.15 × 10−3).

The association of rs4658 and rs841856 genotypes in NAFLD cases and controls was tested for interactions with their clinical characteristics (Table 1). However, no significant relationships were observed (Kruskal-Wallis's chi-square test = the value for each comparison, P > 0.05; Supporting Fig. 2).

Negative Association of SNPs rs4658 and rs841856 With T2DM.

To see whether the NAFLD-associated SNPs were related to T2DM, we genotyped rs4658 and rs841856 in a T2DM case-control cohort, and both of them were determined to be independent (rs4658: ncases = 3,482, ncontrols = 4,738; OR, 0.99 [95% CI: 0.89-1.10], P = 0.82; rs841856: ncases = 3,462, ncontrols = 4,820; OR, 1.03 [95% CI: 0.93-1.15], P = 0.59).

SLC2A1 Expression Is Decreased in NAFLD Liver Biopsies.

RT-qPCR for the SLC2A1 and two housekeeping genes (glyceraldehyde-3-phosphate dehydrogenase [GAPDH] and acidic ribosomal phosphoprotein [ARP]) was carried out using RNA extracted from liver biopsies of 39 NAFLD patients and 30 controls. The ratio (Cthousekeeping/Cttarget gene), based on the media threshold cycle values obtained for each group of samples (NAFLD and controls), showed that SLC2A1 is down-regulated in NAFLD liver biopsies (P = 2.7382 × 10−6; Fig. 1). Further analysis indicated that homozygous carriers of rs4658 G-allele have lower expression of SLC2A1 messenger RNA (mRNA) in liver (P = 0.0422; Table 3).

thumbnail image

Figure 1. Differential expression of SLC2A1 in liver biopsies. Y-axis indicates the FC of SLC2A1 mRNA expression detected in control and NAFLD liver biopsies relative to the characterized housekeeping genes (GAPDH and ARP). Values of relative FC and their standard deviation (SD) are detailed in the table, where “n” is the number of biopsies considered in each group and a t test was performed.

Download figure to PowerPoint

Table 3. rs841856 and rs4658 Genotypes Related to SLC2A1 Expression
SNP_IDGenotypeFC_SLC2A1 Mean ± SEMNComparisonsP Value
  1. Left tables: mean values of FC in expression of SLC2A1 calculated for each genotyped biopsy group (N: number of samples within each group). Right tables: P value of pairwise comparisons of FC mean values for each genotype of SLC2A1 SNPs obtained by Welch's two-sample t test.

  2. Abbreviation: SEM, standard error of the mean.

rs841856GG1.0253 ± 0.085441GG versus GT0.9540
 GT1.0173 ± 0.107719GG versus TT0.9944
 TT1.0226 ± 0.30092GT versus TT0.9890
rs4658CC1.0322 ± 0.092737CC versus CG0.7612
 CG1.0718 ± 0.090722CC versus GG0.0517
 GG0.5462 ± 0.14773CG versus GG0.0422

Effect of Silencing SLC2A1 in THLE2 Cells.

To study the consequence of SLC2A1 diminution, THLE2 cells were transfected with siRNA SLC2A1 that successfully knocked down the expression of the targeted gene, as shown in gene- and protein-expression levels by qPCR and western blotting, respectively (Fig. 2). Then, the whole genome expression profile of siSLC2A1-THLE2 cells was characterized with HumanHT12-v3 Expression BeadChips (Illumina). Triplicates of RNA extracted from siSLC2A1- and from siControl-THLE2 cells were studied. Raw data can be found in the ArrayExpress Browser at under ID: E-MEXP-3310, and differential expression results are shown in Supporting Table 3.

thumbnail image

Figure 2. SLC2A1 is down-regulated after SLC2A1 siRNA transfection. (A) Total RNA was extracted from triplicates of silenced THLE-2 cell cultures, and SLC2A1 mRNA expression was analyzed by RT-qPCR. Differences between siControl and siSLC2A1 were statistically significant: *P < 0.05 (t test). Subsequently, the same mRNA was hybridized to the Human HT-12 Expression BeadChips (Illumina, Inc., San Diego, CA). (B) After siRNA transfection, cells were harvested for western blotting analysis to monitor the efficiency of silencing by comparing protein levels of Glut1.

Download figure to PowerPoint

The first principal component, reflecting the major pattern of variability, was used to generate a gene ranking according to the response of SLC2A1 silencing, and enrichment functional analysis revealed that a good proportion of functional categories, which were enriched with regulated genes in siSLC2A1 cells, were related to lipid metabolism, transport and storage, apoptosis, and OS. Among the top overexpressed genes in siSLC2A1-THLE2 cells, the regulation (see Supporting Table 3; adjusted P < 0.001) of FOSB (fold change [FC]: +19.95), BHLHB2 (FC: +5.78), EGR1 (FC: +4.90), SGK (FC: +3.37), JUNB (FC: +3.24), Serpine3 (FC: +3.13), LDLR (FC: +3.12), MT1E (FC: +3.11), CTGF (FC: +3.07), and LIF (FC: +3.05) is remarkable, because their coding proteins are closely related to NAFLD pathogenesis and deeply associated with lipid metabolism. In addition, the genetic networks regulated by interferon (IFN), including both IFN-alpha/beta-inducible genes (IFI27 [FC: −7.45], OAS2 [FC: −7.22], IFI44L [FC: −6.42], and IFITM1 [FC: −5.25]), were repressed (adjusted P < 0.001) in cells after SLC2A1 silencing.

Complementary IPA (Ingenuity Systems) assessment revealed that the differentially expressed genes in siSLC2A1 cells were involved in apoptosis, hepatic steatosis, and cell death of hepatocytes (Table 4).

Table 4. Most Significant (P < 10−2) Regulated Pathways in siSLC2A1-THLE2 Cells After Gene-Set Enrichment Analysis
GO IDGO TermAdjusted P Value
GO:0044255Cellular lipid metabolic process1.69 × 10−3
GO:0045444Fat cell differentiation2.46 × 10−3
GO:0006979Response to OS3.41 × 10−3
GO:0006917Induction of apoptosis6.57 × 10−3
GO:0006869Lipid transport8.16 × 10−3
GO:0019915Lipid storage1.38 × 10−2

Consequences of SLC2A1 Silencing on Cell Metabolism.

First, ROS induction was assessed using DHE, revealing a significant rise of superoxide production in siSLC2A1-THLE2 cells (P < 0.05; Fig. 3).

thumbnail image

Figure 3. SLC2A1-silenced THLE2 cells increase steady-state levels of ROS, as detected using ROS-sensitive dye (DHE). (A) THLE2-silenced cells (triplicates) were stained using DHE (Invitrogen, Carlsbad, CA), following the manufacturer's instructions; fluorescent images were taken by using an Axio imager D1 microscope (Carl Zeiss AG, Oberkochen, Germany). (B) IF intensity of approximately 10 cells from random fields for each sample (triplicates) was calculated using the ImageJ software (National Institutes of Health, Bethesda, MD) and is expressed as arbitrary units (*P < 0.05).

Download figure to PowerPoint

Furthermore, we determined the effect of OA, the most abundant lipid in the diet, serum, and liver tissue,27, 28 that promotes the formation of TG-enriched lipid droplets, in siSLC2A1- and control-THLE2-cells. OA addition to culture medium did not have any effect in the regulation of SLC2A1 expression (differential expression: P = 0.073). Microscopy studies using Bodipy493/503, a fluorescent dye for neutral lipids, confirmed that siSLC2A1-THLE2 cells significantly increased the presence of lipid droplets when OA was added to culture medium (P = 0.018; Fig. 4A,C). Moreover, cell morphology was evaluated by Nomarski/DIC interferential light microscopy. THLE2 cells analysis, at high magnification using DIC interference contrast, was useful to identify more precisely the size and number of lipid droplets that greatly increased after SLC2A1 silencing (Fig. 4B).

thumbnail image

Figure 4. OA treatment induces steatosis in siSLC2A1-THLE2 cells. (A) Fluorescent images of double staining THLE2-silenced cells. Lipid droplets were stained with BODIPY 493/503 (green), and nucleus was stained with with DAPI (blue). Note that after 1 mM OA treatment, siSLC2A1-silenced cells present higher lipid droplets levels, also shown. (B) By scanning microscope (Normaski). (C) Graphic representation of TG quantification using the Menarine (Pomezia, Italy) commercial kit. Eight independent experiments were performed in triplicates. Differences between siControl and siSLC2A1-transfected cells were statistically significant (*P < 0.05).

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Association studies have proved very helpful in identifying some NAFLD biomarkers, but revealed several restrictions resulting from the limited phenotypic characterization of case and control individuals, because surrogate markers, such as ALT and AST levels, for the diagnosis of NAFLD have been used.1 Given that, in the present project, only biopsy-proven NAFLD cases were analyzed, and not only one, but several SNPs located within the SLC2A1 gene showed association with NAFLD. This fact reinforces the association of this gene, because the frequencies of the minor alleles of most of SLC2A1 studied SNPs exhibited identical frequencies in the case cohort.

Although SLC2A1 has been previously related to T2DM,29 the variants (rs4658 and rs841856) associated with NAFLD in this study did not show any relation with this trait in the studied cohort. In addition, the genotypes of these SNPs have never been linked to obesity30 and did not present any bias for BMI in the present report. Altogether, these results suggest that minor alleles along the whole SLC2A1 sequence are related to the susceptibility to develop NAFLD.

mRNA levels of SLC2A1 were lower in the liver biopsies that harbored the minor allele for rs4658 in homozygosis (NAFLD samples). Although a bigger cohort should be tested, this result indicates that the presence of the minor allele of rs4658 is directly related with a down-regulation of the expression of this gene. In silico function analysis of associated SNPs showed that, apart from the SNPs that cause changes in the amino-acidic sequence of GLUT1 protein, variation in SNPs, such as rs2229682, rs841856, and rs841858, could trigger a mismatched binding site for transcription factors and/or microRNAs, depending on the allele that is present in the cells (Supporting Table 2). These mismatches could work as potential regulators of the expression of SLC2A1, although more experiments with specific gene-sequence combinations are required to elucidate these events.

The general transcriptional adjustments of the regulation of SLC2A1 expression shown in NAFLD liver biopsies were studied by silencing this gene in THLE2 cells with siRNA. Differential gene-expression functional analysis identified enriched GO categories related to NAFLD disease, such as lipid metabolism, apoptosis, and OS. Increased expression of FOSB, BHLHB2, EGR1, SGK, JUNB, Serpine3, LDLR, MT1E, CTGF, and LIF genes is directly related with the aforementioned pathways. Specifically, hepatic BHLHB2 has been shown to mediate the ligand-dependent liver X receptor signal to regulate the expression of genes related to fatty-acid synthesis31 and early growth response protein has been implicated in the control of lipid-biosynthetic genes.32 In addition, a progressive rise in Serpine3 level has been related to an increasing degree of steatosis.33 Furthermore, CTGF may link steatosis and fibrosis through increased leptin levels,34 and, finally, a new physiological role for TORC2, mediated by SGK, has been described in the regulation of Caenorhabditis elegans lipid accumulation and growth.35 Moreover, fatty liver has been reported to reduce biological response to IFN-alpha.36 These data are consistent with the down-regulated genes detected after silencing SLC2A1 in human hepatocytes being transcriptionally regulated by IFN-alpha and -beta. IFITM1 is one of the genes involved in the model to predict moderate to severe hepatic steatosis, NASH, and fibrosis in patients infected with chronic hepatitis C.37

In addition, there is consistent evidence that mitochondrial dysfunction has a central role in fatty liver disorders independently of its etiology. Mitochondria participate, at different levels, in NAFLD progression by impairing fatty liver homeostasis and inducing overproduction of free radicals (ROS) that, in turn, trigger lipid peroxidation (LPO) and cell death. OS enhances the release of LPO products and cytokines, which, together, trigger the liver lesions of NASH.2, 38-40 It has been described that Glut1 is the responsible of transporting dehydroascorbid acid into the mitochondria, where it is reduced to mitochondrial ascorbid acid (mitAA; reduced form of vitamin C).39 mitAA quenches ROS, protecting the mitochondria genome and membrane from oxidative damages. Hence, silencing of SLC2A1, and therefore of Glut1 translation, significantly increases ROS levels. Moreover, gene-profile experiments in siSLC2A1-THLE2 cells rendered an OS panel, which allows us to conclude that SLC2A1 actively contributes to regulate a proinflammatory environment at early stages of fatty liver disease.

According to the fact that SLC2A1 protein (GLUT1) is inserted in the plasma membrane of only the last two perivenular hepatocytes in the liver cell plate,41, 42 exactly in the same cells where steatosis starts in adult NAFLD,43 we propose that impairment in lipid metabolism in siSLC2A1 cells promotes the accumulation of TGs. The observed diminished amount of SLC2A1 in the liver of NAFLD patients and the effect of OA medium enrichment in siSLC2A1-THLE2 cells indicate either that the decrease of SLC2A1 mRNA levels induces TG accumulation or, vice versa, that a fat-rich environment activates some mechanism (e.g., glucose/fatty-acid cycle44, 45) that down-regulates the expression of SLC2A1 in hepatic cells. In conclusion, these observations indicate that hepatic SLC2A1 mediates lipid metabolism, providing new potential insights into NAFLD pathogenesis.46

Advanced high-throughput sequencing studies will be required to identify the precise SLC2A1 causative genetic variants in each ethnicity, as well as functional tests, to decipher the underlying mechanism that causes NAFLD susceptibility/development. Nevertheless, these data identified and demonstrated that SLC2A1 is implicated in lipid storage and inflammatory-damaging response, and both are critical features of NAFLD disease.


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

The authors thank all the patient and control individuals for their participation in this study and the technical and human support provided by Servicios Generales from the University of Basque Country UPV/EHU (SGIker) and La Unidad de Formación e Investigación, UFI 11/20, from the University of Basque Country UPV/EHU.


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

HEP_26052_sm_SuppFig2A.tif262KSupporting Information Figure 2A
HEP_26052_sm_SuppFig2B.tif261KSupporting Information Figure 2B
HEP_26052_sm_SuppFig2C.tif269KSupporting Information Figure 2C
HEP_26052_sm_SuppFig2D.tif265KSupporting Information Figure 2D
HEP_26052_sm_SuppInfo.doc442KSupporting Information
HEP_26052_sm_SuppTab1.xls322KSupporting Information Table 1
HEP_26052_sm_SuppTab3.xls6470KSupporting Information Table 3

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.