Biochemical Journal

Research article

Diet-induced epigenetic regulation in vivo of the intestinal fructose transporter Glut5 during development of rat small intestine

Takuji Suzuki, Veronique Douard, Kazuki Mochizuki, Toshinao Goda, Ronaldo P. Ferraris

Abstract

Metabolic complications arising from excessive fructose consumption are increasing dramatically even in young children, but little is known about ontogenetic mechanisms regulating Glut5 [glucose transporter 5; encoded by the Slc2a5 (solute carrier family 2 member 5) gene]. Glut5 expression is low postnatally and does not increase, unless luminal fructose and systemic glucocorticoids are present, until ≥14 days of age, suggesting substrate-inducible age- and hormone-sensitive regulation. In the present study, we perfused intestines of 10- and 20-day-old rats with either fructose or glucose then analysed the binding of Pol II (RNA polymerase II) and GR (glucocorticoid receptor), as well as acetylation of histones H3 and H4 by chromatin immunoprecipitation. Abundance of Glut5 mRNA increased only with fructose perfusion and age, a pattern that matched that of Pol II binding and histone H3 acetylation to the Glut5 promoter. Although many regions of the Glut5 promoter respond to developmental signals, fewer regions perceive dietary signals. Age- but not fructose-dependent expression of Sglt1 [sodium-dependent glucose co-transporter 1 encoded by the Slc5a1(solute carrier family 5 member 1) gene] also correlated with Pol II binding and histone H3 acetylation. In contrast, G6Pase (glucose-6-phosphatase; encoded by the G6pc gene) expression, which decreases with age and increases with fructose, is associated only with age-dependent changes in histone H4 acetylation. Induction of Glut5 during ontogenetic development appears to be specifically mediated by GR translocation to the nucleus and subsequent binding to the Glut5 promoter, whereas the glucocorticoid-independent regulation of Sglt1 by age was not associated with any GR binding to the Sglt1 promoter.

  • developmental programming
  • epigenetic regulation
  • fructose
  • glucose transporter 5 (Glut5)
  • histone acetylation
  • small intestine
  • transport

INTRODUCTION

Modern-day humans have developed an insatiable appetite for fructose, the sweetest of all natural sugars. The worldwide per capita consumption of corn syrup containing high fructose has increased dramatically in the last century, and this increase has been linked to the equally sharp rise in the prevalence of diabetes, the metabolic syndrome, fatty liver disease and childhood obesity [1,2]. Dietary fructose induces in adult mammals the expression and activity of several small intestinal genes, including its own transporter Glut5 [glucose transporter 5; encoded by the Slc2a5 (solute carrier family 2 member 5) gene] [3,4], thereby increasing fructose absorption rates and exacerbating the effects of fructose in diabetes [5]. When fructose is perfused into the small intestine of adults, Glut5 mRNA levels and fructose transport rates increase within 2–4 h [6]. Fructose-dependent Glut5 transcription is sensitive to actinomycin D and to cycloheximide [6], suggesting that fructose-induced Glut5 expression in vivo is regulated at the transcriptional and translational level.

In neonates of mammals, regulation of Glut5 involves complex interactions between age-related and diet-related signals. During the suckling period of rats (~0–14 days of age), Glut5 gene expression and transport activity are low [7] and, unlike in adults, induction by its substrate fructose is not possible [8]. However, Glut5 mRNA expression and activity can be stimulated by fructose consumption or perfusion into the intestinal lumen during weaning [3,8]. Hence regulation of Glut5 by its substrate undergoes a suckling-to-weaning transition and is therefore developmentally modulated. This developmental programming is understandable because rodents undergo during weaning a drastic switch from a high-fat fructose-free diet to a high-carbohydrate diet probably containing fructose [9]. In contrast, expression and activity of intestinal Sglt1 [sodium-dependent glucose co-transporter 1; encoded by the gene Slc5a1 (solute carrier family 5 member 1)] increase with age but is not regulated by dietary sugars in neonates [6,8], whereas those of the gluconeogenic enzyme G6Pase (glucose-6-phosphatase; encoded by the G6pc gene) increase with dietary fructose but is not regulated by age [10].

Abrupt changes in gene expression which occur frequently in differentiating cells such as the developing small intestine are accompanied with a major chromatin structural change that is triggered by modifications of the histone tail such as acetylation, methylation and phosphorylation [11,12]. These modifications are known as epigenetic regulation. In particular, acetylation of histones H3 and H4 is the most extensively studied because it is associated with transactivation of several genes. We have identified potential signalling pathways and the nuclear receptor involved in modulating Glut5 development [13,14], but there has been no study on the epigenetic regulation of this gene. Indeed, our previous studies demonstrated that the induction of Si (sucrase-isomaltase; encoded by the Si gene), an intestinal disaccharidase participating in the final digestion of starch hydrolysates and sucrose [15], is regulated by modifications of histone H3 during the differentiation of enterocytes along the crypt–villus axis of adult rat intestine [16]. Moreover, induction in adult mouse intestine of the Si gene by a diet rich in carbohydrate is associated with acetylation of histones H3 and H4 [17]. We therefore hypothesized in the present study that the characteristic induction of the Glut5 gene by its substrate fructose is associated with histone modifications during weaning, but not during suckling, and we used as controls Sglt1 and Gp6ase, constitutive genes displaying markedly different patterns of developmental expression.

Glucocorticoids are involved in differentiation and maturation of small intestinal cells [18,19]. In vivo approaches to determining mechanisms underlying the unique interaction between age-associated and sugar-related signals have shown that age-related changes in blood corticosterone levels may control the ability of the dietary signal to up-regulate Glut5 [13,20] because, during rat development, the glucocorticoid hormone increases markedly during the suckling to weaning transition at approx. 14 days of age [21]. Therefore we next hypothesized that Glut5 regulation by age involves signalling via the GR [glucocorticoid receptor; encoded by the Nr3c1 (nuclear receptor subfamily 3 group C member 1) gene]. Since it is already expressed at high levels before the endogenous corticosterone surge, we predicted that Sglt1 regulation by age does not involve the GR. This is the first study investigating hormonally mediated epigenetic regulation of intestinal genes in vivo.

MATERIALS AND METHODS

Animals

All the procedures conducted in the present study were approved by the Institutional Animal Care and Use Committee, University of Medicine and Dentistry of New Jersey, New Jersey Medical School. Pregnant female Sprague–Dawley rats purchased from Taconic were housed in the research animal facility under a 12-h light/12-h dark photoperiod in a temperature-controlled room (22–24 °C). Dams were fed a commercial diet (Purina Mills) ad libitum. After birth, rat pups were kept with their dams; age at birth was considered as day 0. At 10 (representing the suckling stage) or 20 (representing the weaning stage) days old, rat pups were removed randomly from their dams and used for perfusion.

Intestinal perfusion

The rat intestinal perfusion procedure was conducted following a method described previously [6]. Rat pups (not starved) were initially anaesthetized [0.2–0.4 ml/100 g of body weight, intraperitoneally, of a mixture of ketamine (20 mg/ml) and xylazine (2.5 mg/ml)] and kept under continuous anaesthesia for 4 h during the entire perfusion. Then, after opening the abdominal cavity, the intestine was exposed and a small incision was made first at 5 cm distal to the ligament of Treitz and then at 10 cm proximal from the ileocecal valve, and a catheter was inserted into the lumen via the first incision. After the contents were flushed, the small intestine was continuously perfused with sugar solution (100 mM fructose or glucose in Ringer's solution) at a rate of 30 ml/h at 37 °C using a peristaltic pump. Composition of the perfusion solution was as follows: 78 mM NaCl, 4.7 mM KCl, 2.5 mM CaCl2·5H2O, 1.2 mM MgSO4, 19 mM NaHCO3, 2.2 mM KH2PO4 and 100 mM glucose or fructose (pH 7.4) (300 mOsm). After perfusion, tissues for real-time PCR and Western blot analysis were frozen at −80 °C. Tissues for evaluating histone modification were processed according to the ChIP (chromatin immunoprecipitation) assay protocol described in detail below.

Immunohistochemistry

Rat pup jejunum was fixed in fresh 4% paraformaldehyde in PBS (pH 7.35) overnight at room temperature (24 °C) and then embedded in paraffin. Heat-induced antigen retrieval in a programmed pressure cooker (Retriever; Electron Microscopy Services) was performed on 5-μm-thick sections in 0.01 M sodium citrate buffer (pH 6.0). The sections were rinsed in PBS and blocked in 1% normal goat serum for 1 h at room temperature. The sections were then incubated in a rabbit anti-(rat Glut5) antibody (Chemicon International), diluted 1:500 in 1% BSA in PBS, or in a mouse anti-(rat GR) (US Biological), diluted 1:500 in 1% BSA in PBS, for 16 h at 4 °C then washed in PBS. The secondary antibody [Cy3 (indocarbocyanine)-labelled goat anti-(rabbit IgG), diluted 1:100] was applied for 1 h at room temperature. To examine translocation of GR from the cytoplasm to nucleus, immunostaining for GR as described was performed with the addition of propidium iodide staining of nuclei. As controls for non-specific staining, blocked sections were incubated with normal rabbit serum at 4 °C and stained with secondary antibody. Stained sections were mounted in Vectashield (Vector Laboratories) and examined at ×40 magnification with a water-immersion lens on a laser-scanning confocal microscope (Radiance 2100; Bio-Rad). All images with the same fluorescence were obtained with the same settings of the microscope. Non-specific staining with secondary antibodies was consistently negligible.

Western blot analysis

Briefly, supernatants were prepared, following methods described previously [14], from 100 mg of intestinal mucosal scrapes using T-Per Tissue Protein Extraction Reagent (Thermo Scientific) and analysed for protein concentration using the Bradford method. Intestinal proteins (50 μg) were separated using a pre-cast 12% Tris/HCl gel (Bio-Rad), transferred to a membrane and analysed with the anti-(rat Glut5) antibody (Chemicon International) diluted 1:1000. All membranes were stripped and reprobed with β-actin antibody (Chemicon International).

mRNA extraction, DNase treatment and RT (reverse transcription) reaction

Total RNA was extracted from 100 mg of scraped mucosa using 1 ml of TRIzol® reagent (Invitrogen). The total RNA concentration was determined by spectrophotometry (DU640; Beckman Coulter) and the quality was analysed using electrophoresis on a 1% agarose gel with ethidium bromide staining. To hydrolyse contaminating DNA in the RNA preparations, 20 μg of RNA was combined with 1 μl of RQ1 Rnase-free DNase I (Promega) and 10 μl of 10×DNase buffer in a final volume of 100 μl. After DNase treatment, RNA concentration and quality were analysed as described previously [14]. The cDNA was generated from 2.5 μg of DNase-treated RNA using SuperScript III RNase H-Reverse Transcriptase and oligo(dT)20 (Invitrogen) in a total volume of 20 μl.

Quantitative real-time RT–PCR

Primers for mRNA expression were designed using primer3 software (http://frodo.wi.mit.edu/primer3/input.htm), and were purchased from Integrated DNA Technologies. Quantitative PCR analyses (MX 3000P; Stratagene) for candidate responsive genes were performed using the Brilliant SYBR® Green QPCR Master Mix with Rox (Bio-Rad), according to the method described by Douard et al. [13]. For each primer set (Supplementary Table S1 at http://www.BiochemJ.org/bj/435/bj4350043add.htm), i.e. Glut5, Sglt1 and G6Pase, as described in our previous study [13], the efficiency of the PCR (linear equation: y=slope+intercept) was measured in triplicate on serial dilutions of the same cDNA sample (pool of reverse-transcribed RNA samples). Real-time PCR efficiencies (E) for each reaction were then calculated using the following equation: E=[10(1/slope)]−1. Melting curve analysis also was performed for each gene to check the specificity and identity of the RT–PCR products. The relative expression ratio of a target gene was calculated as follows: Embedded Image

This calculation is based on real-time PCR efficiencies (E) of target genes and the cycle threshold (Ct) difference (Δ) of an unknown sample compared with a control [ΔCt(control–sample)]. In our experiment, the control is the 10-day-old pups perfused with glucose solution. Because the expression level of EF1α [elongation factor 1α; encoded by the Eef1a1 (eukaryotic translation elongation factor 1α1) gene] is not affected by glucose perfusion and fructose perfusion or by age of pups, the target gene expression was normalized to EF1α.

ChIP assay

Perfused small intestinal tissues were opened longitudinally on a glass plate then the mucosa was scraped and separated from underlyling muscle layers by a glass microscope slide. Following a method described previously [16], the scraped mucosa was incubated with fixative (1% formaldehyde, 4.5 mM Hepes, pH 8.0, 9 mM NaCl, 0.09 mM EDTA and 0.04 mM EGTA in PBS) for 30 min at 37 °C. The fixation reaction was terminated by the addition of glycine to a final concentration of 150 mM. After being washed in FACS solution (2% bovine serum and 0.05% NaN3 in PBS), the samples were sonicated in SDS lysis buffer [16 mM Tris/HCl, pH 8.0, 10 mM EDTA, 1% SDS and proteinase inhibitor tablets (Complete Mini, EDTA-free; Roche)] to yield DNA fragments ranging from 200 to 500 bp in size as confirmed by electrophoresis using a 1% agarose gel. The protein concentration (BCA™ Protein Assay Kit; Thermo Scientific) of all samples was adjusted to 2 μg/μl after sonication. The sample including 200 μg of protein was diluted 1:10 in ChIP dilution buffer [50 mM Tris/HCl, pH 8.0, 167 mM NaCl, 1.1% Triton X-100, 0.11% sodium deoxycholate and a protease inhibitor cocktail tablet (one tablet/10 ml solution)]. Approx. one-fifth of the total sample volume was used as the input fraction. The remaining solutions were incubated with 2 μg each of specific antibody, i.e. anti-(acetylated histone H3) (Millipore), anti-(acetylated histone H4) (Millipore), anti-Pol II (RNA polymerase II) (Santa Cruz Biotechnology) or rabbit IgG as a control. The protein–DNA–antibody complexes were immunoprecipitated by Protein G–Sepharose™ 4 Fast Flow (GE Healthcare Biosciences) containing 100 μg/ml salmon sperm DNA and 1% BSA in RIPA buffer (50 mM Tris/HCl, pH 8.0, 1 mM EDTA, 1% Triton X-100, 0.1% SDS and 0.1% sodium deoxycholate) with 150 mM NaCl. Protein G–Sepharose was washed twice in RIPA buffer/150 mM NaCl, five times in RIPA buffer/500 mM NaCl, twice in LiCl buffer (10 mM Tris/HCl, pH 8.0, 0.25 M LiCl, 1 mM EDTA, 0.5% Nonidet P40 and 0.5% sodium deoxycholate) and twice in TE buffer (10 mM Tris/HCl, pH 8.0, and 1 mM EDTA). Cross-linking reversal and DNA clean-up steps involved treatments in 400 μl of ChIP direct elution buffer (10 mM Tris/HCl, pH 8.0, 5 mM EDTA, 0.5% SDS and 300 mM NaCl) at 65 °C overnight and then in RNase A as well as proteinase K respectively. Input DNA and immunoprecipitated DNA were extracted by phenol/chloroform. Extracted DNA was then subjected to real-time PCR using primers corresponding to the indicated sites in the promoter regions of target genes. Primers for the ChIP assay (see Supplementary Table S2 at http://www.BiochemJ.org/bj/435/bj4350043add.htm) were designed using primer3 software as described above. The size of the amplified RT–PCR product was confirmed by 2% agarose gel electrophoresis (Supplementary Figure S1 at http://www.BiochemJ.org/bj/435/bj4350043add.htm). Only a single PCR product was obtained for each primer set, indicating that the primers were specific for the targeted Glut5, Sglt1 or G6Pase promoter region concerned. All ChIP signals were calculated by the 2(−ΔΔCt) method [22] and were normalized to the corresponding input signals as follows: Embedded Image where IP is immunoprecipitation. Non-specific antibody binding to protein–DNA fragments as indicated by the percentage input using IgG was negligible (Supplementary Figure S2 at http://www.BiochemJ.org/bj/435/bj4350043add.htm).

Since the GR may regulate Glut5 transcription [13,21], we identified potential GREs (GR response elements) in the Glut5 and Sglt1 promoters, using ALGGEN PROMO (http://alggen.lsi.upc.es/cgi-bin/promo_v3/promo/promoinit.cgi?dirDB=TF_8.3) as follows. There are a total of 14 predicted GREs in the Glut5 promoter between 0 and 2000 bp upstream from the transcription start site: −11 to −20, −131 to −135, −615 to −619, −676 to −680, −771 to −775, −1004 to −1008, −1012 to −1016, −1031 to −1035, −1072 to −1076, −1186 to −1190, −1447 to −1451, −1538 to −1542, −1579 to −15783 and −1898 to −1902. A total of 20 predicted GREs are in the Sglt1 promoter: −12 to −16, −79 to −83, −125 to −129, −133 to −137, −139 to −143, −162 to −166, −342 to −346, −396 to −400, −618 to −622, −629 to −634, −668 to −672, −734 to −738, −815 to −822, −909 to −917, −952 to −956, −1656 to −1660, −1679 to −1684, −1688 to −1692, −1737 to −1741 and −1746 to −1750. We examined potential GR binding to the following regions of Glut5 and Sglt1: 0 to −200, −600 to −800, −1000 to −1200, −1400 to −1600 and −1800 to −2000 bp.

Statistical analyses

Values are presented as means±S.E.M. and were analysed using a two-way ANOVA to determine the significant differences in ChIP signals as a function of age and sugar in selected promoter regions of each target gene (see Table 1). If a two-way ANOVA indicated a significant difference, a one-way ANOVA was used to determine the significant differences in relative mRNA abundance and in ChIP signals of glucose-perfused 10- and 20-day-old rats and fructose-perfused 10- and 20-day-old rats in each target promoter region. The one-way ANOVA was followed by Tukey's multiple range tests to determine differences among the means (P<0.05), and results are depicted in the Figures by superscript letters.

View this table:
Table 1 P values based on two-way ANOVA indicating the statistical significance of effects of age and fructose perfusion on Pol II recruitment, histone acetylation and GR binding to promoter regions of Glut5, Sglt1 and G6Pase

Analysis by two-way ANOVA, with age and diet as treatments. *P<0.05 and **P<0.01. ×, not analysed statistically because GR binding was at or below background; −, ChIP analysis was not done.

RESULTS

Localization of Glut5 in jejunum of rat pups

Glut5 immunoreactive protein was localized mainly in the apical membrane of enterocytes lining the villi of both fructose- and glucose-perfused small intestine of 20-day-old rats (Figure 1, arrows). The expression of Glut5 was much higher in the fructose-perfused (Figure 1A) than in the glucose-perfused (Figure 1B) small intestine of 20-day-old rats. In contrast, Glut5 was virtually absent in the villi of 10-day-old rat (Figure 1C). These diet- and development-related increases in Glut5 expression were verified by Western blot analysis of mucosal homogenates (Figure 1D), confirming results from our previous work [13,14]. Dramatic increases in Glut5 levels shown in Figure 1(D) suggest that the diffuse staining in Figure 1(A) may reflect increased levels of immunoreactive Glut5 in the cytosol, prior to Glut5 insertion into the apical membrane.

Figure 1 Immunolocalization and expression of Glut5 in intestinal villi of rats

The intestinal villi from (A) 20-day-old pups perfused with fructose, (B) 20-day-old pups perfused with glucose and (C) 10-day-old pups perfused with fructose are shown. Arrows depict the brush border location of Glut5. In (C), because there was virtually no image at the confocal settings used for (A) and (B), the gain was increased 10-fold to generate the image shown [13]. Levels of immunoreactive Glut5 in the intestines of 10-day-old pups perfused with glucose (results not shown) were similar to those perfused with fructose. Representative photographs of several sections from two rats per treatment are shown. Yellow scale bar, 20 μm. (D) Effect of age and perfusion solution on levels of Glut5 protein as analysed by Western blotting of intestinal mucosal homogenates from 20-day-old pups perfused with fructose (20 d HF) or glucose (20 d HG) and 10-day-old pups perfused with fructose (10 d HF).

Modification of Glut5 promoter regions

Using two-way ANOVA, Glut5 mRNA expression increased with age (P<0.001) and with fructose perfusion (P<0.05), with the interaction being significant (P<0.05), suggesting that the effect of fructose depends on age. Mean expression was 50-fold higher in 20-day-old compared with 10-day-old rats. Glut5 mRNA expression levels were not different between glucose-perfused and fructose-perfused 10-day-old rats as determined by one-way ANOVA. However, Glut5 expression was significantly (P<0.05) increased 3.0-fold in fructose-perfused compared with glucose-perfused 20-day-old rats (Figure 2A).

Figure 2 Effect of age and sugar on Glut5 mRNA expression, and on Pol II binding and acetylation of histones in the Glut5 promoter of neonatal rat small intestine

(A) mRNA expression of Glut5. Total RNA was extracted from glucose- or fructose-perfused small intestine of 10-day or 20-day-old rats (the primers used are listed in Supplementary Table S1 at http://www.BiochemJ.org/bj/435/bj4350043add.htm). The mRNA abundance was measured by real-time RT–PCR using EF1α as a reference gene. Values are means±S.E.M. (n=3–5). Letters indicate significant differences as determined by one-way ANOVA. Differences among the means were detected by Tukey–Kramer's multiple range test at P<0.05. (BD) ChIP assays were performed using anti-Pol II (B), anti-(acetylated histone H3) (C) and anti-(acetylated histone H4) (D) antibodies in the small intestine of 10- or 20-day-old rats perfused with either glucose (HG) or fructose (HF). ChIP signals were detected by real-time RT–PCR using the primers for the designated promoter regions of Glut5 (Supplementary Table S2 at http://www.BiochemJ.org/bj/435/bj4350043add.htm). The ChIP signals were normalized to input signals. Values are means±S.E.M. (n=3), with each n analysed by ChIP two or three times. The region shaded in grey in (BD) indicates non-specific ChIP signals obtained using IgG instead of antibodies. The letters indicate significant differences (within a specific region) that were detected by one-way ANOVA followed by Tukey–Kramer's multiple range test at P<0.05.

Pol II binding to the promoter regions of the Glut5 gene increased 4-fold (from −1200 to −1000 bp), 7-fold (from −800 to −600 bp) and 15-fold (from −200 to 0 bp) in 20-day-old compared with 10-day-old rats, hence the effect of age on Pol II binding was indeed highly significant (Table 1). For 20-day-old rats, fructose also tended to increase Pol II binding in each region examined, but was statistically significant only between −200 and 0 bp (near the transcription start site), where coincidentally there was significant interaction between age and sugar effects (Table 1). In the region between −200 to 0 bp, Pol II binding was significantly increased 1.3-fold by fructose perfusion in 20-day-old rats (Figure 2B). The pattern of Pol II binding to the promoter region of the Glut5 gene between −200 to 0 bp correlated well with Glut5 mRNA expression (Figure 2A).

We then investigated whether the acetylation of histone H3 and H4 on promoter regions of the Glut5 gene changed with development or sugar perfusion. Using two-way ANOVA, the effect of age on acetylation of histone H3 was highly significant for each Glut5 promoter region examined (Table 1). Fructose tended to enhance histone H3 acetylation in each promoter region examined, and H3 acetylation was statistically significant from −1600 to −1000 bp (Table 1). Acetylation levels of histone H3 increased 2–4-fold with age and 1.5-fold with fructose perfusion (Figure 2C). There were no significant effects of age and sugar on acetylation of histone H4 in the entire promoter region of Glut5 (Figure 2D and Table 1).

Modification of Sglt1 promoter regions

Using two-way ANOVA, Sglt1 mRNA abundance increased with age (P<0.005), but did not change with sugar perfusion (P= 0.16) (Figure 3A). There was no statistical interaction between age and sugar perfusion (P= 0.16). Mean Sglt1 mRNA expression increased with age by approx. 2-fold. There was a tendency for glucose to enhance Sglt1 mRNA abundance in 20-day-old rats.

Figure 3 Effect of age and sugar on Sglt1 mRNA expression, and on Pol II binding and acetylation of histones in the Sglt1 promoter of neonatal rat small intestine

(A) Sglt1 mRNA abundance was measured by real-time RT–PCR using EF1α as a reference gene. Values are means±S.E.M. (n=3–5). Letters indicate significant differences as determined by one-way ANOVA. (BD) ChIP assays were performed using anti-Pol II (B), anti-(acetylated histone H3) (C) and anti-(acetylated histone H4) (D) antibodies in the small intestine of 10- or 20-day-old rats perfused with glucose (HG) or fructose (HF). ChIP signals were detected by real-time RT–PCR using the primers for promoter regions of Sglt1 (Supplementary Table S2 http://www.BiochemJ.org/bj/435/bj4350043add.htm). These signals were normalized to input signals. Values are means±S.E.M. (n=3), with each n analysed two or three times. The region shaded in grey in (BD) indicates ChIP signals obtained using IgG. The letters indicate significant differences (within a specific region) detected by one-way ANOVA followed by Tukey–Kramer's multiple range test at P<0.05.

The effect of age on Pol II binding in representative Sglt1 promoter regions was highly significant between −1800 to 0 bp (P= 0.0004–0.002 by two-way ANOVA; Table 1). However, it seems clear that significant Pol II binding occurs in 20-day-old rats only, as Pol II binding in 10-day-old rats was not above background levels (Figure 3B). Fructose did not alter Pol II binding in the Sglt1 promoter of 20-day-old rats.

Analysis of patterns of histone H3 acetylation showed some similarity with Pol II binding. Using two-way ANOVA, acetylation levels of histone H3 were significantly increased by approx. 2-fold in 20-day-old (Figure 3C) compared with 10-day-old rats in Sglt1 promoter regions between −1800 and −800 bp (Table 1). There was no effect of fructose perfusion on H3 acetylation. There were also no significant effects of age and sugar on acetylation of histone H4 in each Sglt1 promoter region analysed (Figure 3D and Table 1).

Modification of G6Pase promoter regions

Using two-way ANOVA, the expression of the gluconeogenic enzyme G6Pase decreased markedly with age (P<0.01), but increased significantly (P<0.05) with fructose perfusion (Figure 4A). Mean G6Pase expression decreased 4-fold with age. Fructose perfusion increased G6Pase mRNA expression in 10- as well as in 20-day-old rats.

Figure 4 Effect of age and sugar on G6Pase mRNA expression, and on Pol II binding and acetylation of histones in the G6Pase promoter of neonatal rat small intestine

(A) mRNA expression of G6Pase was measured by real-time RT–PCR using EF1α as a reference gene. Values are means±S.E.M. (n=3–5). The letters indicate significant differences by one-way ANOVA followed by Tukey–Kramer's multiple range test at P<0.05. (BD) ChIP assays were performed using anti-Pol II (B), anti-(acetylated histone H3) (C) and anti-(acetylated histone H4) (D) antibodies in the small intestine of 10- or 20-day-old rats perfused with glucose (HG) or fructose (HF). ChIP signals were detected by real-time RT–PCR using the primers for promoter regions of G6Pase (Supplementary Table S2 http://www.BiochemJ.org/bj/435/bj4350043add.htm). The ChIP signals were normalized to input signals. Values are means±S.E.M. (n=3). The region shaded in grey in (BD) indicates the ChIP signals of IgG. The letters indicate significant differences (within a specific region) detected by one-way ANOVA followed by Tukey–Kramer's multiple range test at P<0.05.

Despite the marked effect of age and fructose perfusion on steady-state levels of G6Pase mRNA, age and fructose perfusion had no effect on Pol II binding to the G6Pase promoter (Figure 4B). This is consistent with previous work showing that the fructose-induced increase in G6Pase mRNA levels was not inhibited by the Pol II inhibitor actinomycin D [13]. Moreover, histone H3 acetylation did not change with age or sugar in any region (Figure 4C).

There was a significant effect of age but not of sugar perfusion on histone H4 acetylation (Table 1). There is a tendency for histone H4 acetylation to decrease with age throughout the entire promoter region of G6Pase (Figure 4D). A two-way ANOVA showed that histone H4 acetylation decreased significantly with age from −1900 to −600 bp and between −350 and −150 bp (Table 1). Thus the age-related decrease in histone H4 acetylation parallels changes in G6Pase mRNA abundance.

Translocation of the GR during fructose perfusion

GRs translocate from the cytoplasm to the nucleus when ligands such as corticosterone bind to the GR. We have shown previously that plasma glucocorticoid levels in rats were initially low during the suckling stage then increased markedly during weaning [21]. In the present study, we show that GRs translocated from the cytoplasm to the nucleus in the villus of 20-day-old but not 10-day-old rats (Figure 5). The GR translocated to the nuclear region in both fructose- (Figures 5A–5C) and glucose- (Figures 5D–5F) perfused intestines. There seemed to be no perfusionrelated difference in levels of GR in 20- as well as in 10-day-old rats. The GR did not translocate to the nuclear region in 10-day-old rats perfused with fructose (Figures 5G–5I) or with glucose (Figures 5J–5L). However, if we injected 10-day-old rats with the glucocorticoid analogue dexamethasone, the GR would translocate to the nucleus followed by increases in Glut5 expression [13].

Figure 5 Translocation of the GR in rat small intestine

The intestines of 20- (AF) and 10- (GL) day-old pups were perfused with fructose (HF) or glucose (HG) for 4 h. Inset: close-up of villus regions. Yellow scale bar, 20 μm. In 20-day-old pups (A and D), arrows depict GRs having translocated to spaces occupied by the nuclei (B and E). When the images are merged, the arrows now show yellow/green where the nuclei are found, indicating movement of GRs to the nuclear region (C and F). The GRs are localized primarily in the cytosol of 10-day-old pups perfused with fructose (G) or glucose (J). Note the arrow indicating an absence of GR immunostaining in areas where nuclei are located (H and K). When images are merged, the arrows indicate primarily nuclear staining with propidium iodide (red) (compare I and L with C and F). These images are typical of two pups from each treatment.

Binding of the GR on Glut5 and Sglt1 promoter regions

We examined representative regions of the Glut5 promoter that may include GREs. Age tended to increase GR binding in all promoter regions (Figure 6). The patterns of GR binding to Glut5 seem to parallel GR translocation shown in Figure 5. A two-way ANOVA indicated highly significant age-sensitive GR binding from −2000 to −1000 bp and from −200 to 0 bp (Table 1) where several GREs are located. GR binding was not observed in the representative promoter regions of Sglt1 (results not shown), even though these regions had GREs (see the list in the Materials and methods section).

Figure 6 Binding of the GR to Glut5 promoter regions

ChIP assays were performed using an anti-GR antibody in fructose (HF)- or glucose (HG)-perfused small intestine of 10- or 20-day-old rats. ChIP signals were detected by real-time RT–PCR using the primers to detect binding in the promoter regions of Glut5. The ChIP signals were normalized to input signals. Values are means±S.E.M. (n=3), with each n being analysed at least twice. The letters indicate significant differences detected by one-way ANOVA followed by Tukey–Kramer's multiple range test at P<0.05. The region shaded in grey indicates non-specific ChIP signals obtained using IgG instead of antibodies. There is significant binding of the GR to the Glut5, but not to the Sglt1 (results not shown), promoter.

GRs interact with CBP [CREB (cAMP-response-elementbinding protein)-binding protein]) and C/EBPδ (CCAAT/enhancer-binding protein δ), which act as co-activators of transcription [23]. In the present study, we examined CREB and C/EBPδ binding to the Glut5 promoter region. CREB binding was much greater than IgG; however, a two-way ANOVA showed no significant effect of age and fructose perfusion on CREB binding in all of the Glut5 promoter regions examined (Supplementary Figure S3A at http://www.BiochemJ.org/bj/435/bj4350043add.htm). In contrast, there is a modest effect of age on C/EBPδ binding in the −800 to −600 bp region of the Glut5 promoter (Supplementary Figure S3B).

DISCUSSION

Ontogenetic development in vivo involves cross-talk among various organs and tissues, and is impossible to replicate in vitro. In all omnivorous mammals, fructose is an important source of calories after weaning, hence its transporter Glut5 is inducible at this stage of development, and a dietary and an endocrine signal seem essential for Glut5 enhancement: fructose [8] and corticosterone [8] respectively. In the present study, we show that epigenetic mechanisms involved in the developmental regulation in vivo of the inducible transporter Glut5 differs markedly from those of Sglt1 and G6Pase (Figure 7), constitutive enzymes essential for normal carbohydrate transport and metabolism in intestinal cells. We have then shown that both GR translocation to the nucleus and GR binding to the Glut5 promoter are dependent on age. Since we have shown previ-ously that inhibiting GR translocation prevents fructose-induced Glut5 transcription [13], our findings suggest that age-dependent GR translocation may precede GR binding to the Glut5 promoter followed by Glut5 transcription. Our novel findings also include fructose-induced changes in histone acetylation that are link-ed with age-induced changes associated with Glut5 development.

Figure 7 Schematic diagram summarizing the effect of age and sugar perfusion on acetylation of histones and binding of Pol II to the promoter regions of Glut5, Sglt1 and G6Pase

The distribution of histone H3 (A) and H4 (B) acetylation on each gene is shown. (C) The binding levels of Pol II on the promoter region of each gene is shown. The grey broken line indicates the effect of age, the solid black line indicates the effects of both age and fructose-perfusion, and the dotted line indicates no effect of age and fructose-perfusion. Age clearly influences epigenetic regulation of all three genes, with Sglt1 and Glut5 involving histone H3 and with G6Pase involving histone H4. Pol II binding in vivo increases with age in Glut5 and Sglt1, paralleling age-related increases in mRNA expression. Only Glut5 is responsive to fructose perfusion.

Glut5

The acetylation of histones H3 and H4 by cofactors with HAT (histone acetyltransferase) activity is observed in euchromatin regions where chromatin structure is opened so that transcriptional factors and co-regulators are able to bind to this region easily. Thereafter the transcription of mRNA is activated and maintained by general transcriptional complex including Pol II [24]. Age-related increases in acetylation of histone H3 were dramatic, seemed to involve the entire proximal promoter region of Glut5 and were associated with increases in Pol II binding in more downstream regions. In contrast, acetylation of histone H3 and Pol II binding in the Sglt1 promoter involved more upstream regions. Glut5 transcription is associated with specific histone proteins, since histone H4 acetylation did not change. Fructose perfusion resulted in relatively smaller (compared with age effects), but still striking increases in Pol II binding and histone H3 acetylation, indicating that an age-related signal (such as age-related increases in blood corticosterone levels) is the primary regulator of Glut5 development. Perhaps the modest fructose-responsive component complements the substantial age-related response, thereby explaining the dramatic induction of Glut5 during weaning. In contrast, histone modifications accompanying changes in Sglt1 gene expression were responsive only to age (Figure 7), and changes were smaller in magnitude compared with those of Glut5.

Ontogenetic development in vivo has been shown to involve histone acetylation or deacetylation to regulate gene expression. During metamorphosis from tadpole to frog, increases in thyroid hormone allow the thyroid receptor to activate gene expression in various organ systems, including the small intestine, through histone H4 acetylation [25]. In rats, acetylation of histones H3 and H4 regulates intestinal expression of a retinol-binding protein during the perinatal stage [26]. Thus remodelling of chromatin structure through the acetylation of histone proteins was involved in the expression of small intestinal genes and development of small intestine.

We have shown previously that the blood concentration of corticosterone increased markedly during development from suckling to weaning in rats [21], and triggers the signalling cascade that eventually results in age-sensitive GR translocation to the nuclei. GR expression itself is independent of age and sugar [13]. After chromatin remodelling, the GR then directly binds to Glut5 promoter regions (Figure 8). Our findings cannot determine the sequence of histone H3 acetylation and GR binding, only that both are involved in Glut5 transcription. Without GR binding, Glut5 transcription is not possible as what happens in 10-day-old rats with low circulating levels of corticosterone and in 10-dayold rats with high circulating levels of corticosterone but injected with the GR inhibitor RU486 [13]. The acetylation levels of histone H3 would then increase in the GR-bound regions of the Glut5 promoter and would be followed by Pol II binding. To explain the additional effect of fructose perfusion on age effects, we hypothesized that the additional binding of transcriptional factors which have HAT activity to the promoter region of Glut5 from −2000 to −1000 bp may be stimulated by fructose perfusion, thereby increasing Glut5 gene expression in 20-day-old rats (Figure 8). The identity of this novel factor(s) able to bind to the Glut5 promoter between −2000 and −1000 bp and to modify the acetylation of histone H3 in fructose-perfused small intestine of 20-day-old rats is the subject of ongoing studies. Since fructose metabolism is unregulated and leads to rapid and marked increases in glycolytic intermediates including acetyl-CoA [27], fructose metabolism itself may enhance histone acetylation in the nucleus, as acetyl-CoA serves as a substrate for HATs [28].

Figure 8 Schematic model of Glut5 regulation in small intestinal cells of 20-day-old pups

Age-related increases in corticosterone levels lead to binding with and dimerization of the GR in the cytosol, which then translocate to the nuclei and simultaneously stimulate modest levels of histone H3 acetylation. This step is required for Glut5 regulation as fructose is unable to stimulate Glut5 in 10-day-old rats without the GR in the nuclei. Injecting 10-day-old rats with the glucocorticoid analogue dexamethasone relieves the age-related inhibition of Glut5 by fructose [13]. Perfusion with fructose leads to recruitment of unknown transcription factors (TF) which have HAT activity, thereby increasing the magnitude of age-related increases in histone H3 acetylation in the Glut5 promoter.

Sglt1

Epigenetic regulation of Sglt1 is also age-dependent but is not affected by sugar perfusion and does not involve the GR. Sglt1 is only modestly regulated by diet in 20-day-old rodents, probably because there is little variation in composition of milk [6], their main source of nutrition. However, in adult rodents consuming a more varied diet, Sglt1 activity and transporter number [29] increase with glucose consumption. This increase in Sglt1 gene expression by high carbohydrate feeding in adult mice is accompanied with the induction of histone H3 acetylation [30], but there are no reports that have evaluated the induction of Sglt1 gene by epigenetic regulation during small intestinal development. Thus, unlike Glut5, where there is a sharp transition from scarcity in suckling to substrate-regulated abundance in weaning stage, there is a gradual transition of Sglt1 to substrate-dependent regulation in adults. In rodent evolution, it is possible that programmed age-dependent increases in Sglt1 expression may be designed to meet hard-wired increases in glucose consumption, whereas Glut5 expression increases only after weaning when the opportunity to consume fructose arises.

Transcription of the Sglt1 gene is regulated by HNF-1 (hepatocyte nuclear factor-1) [31,32], which plays a crucial role in the maturation of small intestinal epithelial cells and in the induction of small intestinal genes Si and Lph (lactase-phlorizin hydrolase) [33]. HNF-1α and Cdx-2 interact to modulate induction of the Si gene during the suckling–weaning transition in mice. HNF-1α, however, is not required for histone acetylation of the Lph gene in vivo [33].

G6Pase

G6Pase is expressed in both the intestine and liver and plays a pivotal rule in hepatic gluconeogenesis [10]. However, the function of G6Pase in the small intestine is not known. In neonatal rats, G6Pase expression is quite robust and is stimulated by fructose [34]. Intestinal G6Pase mRNA expression is also strongly induced in diabetic or starved rats and becomes normal upon insulin treatment or refeeding respectively [35]. There have been no studies regarding the ontogenetic and dietary regulation of G6Pase. Negative regulation of intestinal G6Pase with age is specifically associated with decreased histone H4 acetylation, as histone H3 did not change. It is interesting to note that negative regulation of hepatic G6Pase by IL-6 (interleukin-6) is also associated with decreases in histone H4 acetylation [36] in vivo.

Although fructose increases mRNA abundance of both Glut5 and G6Pase in vivo, binding of Pol II and acetylation of histones H4 and H3 were not changed by sugar perfusion in the G6Pase promoter (Figure 7), unlike that in the Glut5 promoter. This suggests marked differences in mechanisms regulating G6Pase and Glut5. Our previous work indicated that increased expression of Glut5, but not G6Pase, is prevented by the Pol II inhibitor actinomycin D [36]. Thus the remarkable difference in epigenetic regulation between Glut5 and G6Pase may be due to fructose-induced de novo transcription of Glut5 and to fructose-enhanced mRNA stability of G6Pase.

AUTHOR CONTRIBUTION

Takuji Suzuki and Veronique Douard helped design the study, performed the experiments and analysed the data, as well as writing the initial drafts of the manuscript. Kazuki Mochizuki helped design the study, conducted the experiments, analysed the data and wrote the manuscript. Toshinao Goda helped in staff co-ordination, data analysis and manuscript revisions. Ronaldo Ferraris co-ordinated the study and designed the experiments, helped in data analysis and wrote the final versions (initial submission and final revision) of the manuscript.

FUNDING

This work was supported by the National Science Foundation [grant number IOS-0722365] and the National Institutes of Health [grant number RDK075617] (to R.P.F.). T.S. was supported by a Grant-in-Aid for Japan Society for the Promotion of Science Researcher Fellows for Young Scientists [grant number 20-11635].

Abbreviations: C/EBPδ, CCAAT/enhancer-binding protein δ; ChIP, chromatin immunoprecipitation; CREB, cAMP-response-element-binding protein; EF1α, elongation factor 1α; Glut5, glucose transporter 5; GR, glucocorticoid receptor; GRE, GR response element; G6Pase, glucose-6-phosphatase; HAT, histone acetyltransferase; HNF-1, hepatocyte nuclear factor-1; Lph, lactase-phlorizin hydrolase; Pol II, RNA polymerase II; RT, reverse transcription; Sglt1, sodium-dependent glucose co-transporter 1; Si, sucrase-isomaltase

References

View Abstract