| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Department of Biochemistry and Molecular Biology, Louisiana State University Health Sciences Center, New Orleans; 2Department of Biology, Xavier University, New Orleans, Louisiana; 3Departments of Pharmacology and Cancer Biology, Medicine, and Biochemistry, Sarah W. Stedman Nutrition and Metabolism Center and Duke University Medical Center, Durham, North Carolina; and 4Department of Biochemistry, Molecular Biology, and Biophysics, University of Minnesota, Minneapolis, Minnesota
Submitted 20 September 2006 ; accepted in final form 9 November 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
glucose-6-phosphatase; promoter; carbohydrate response element; carbohydrate response element-binding protein; adenosine 3',5'-cyclic monophosphate response element
-cells, since the human G-6-Pase gene promoter confers a glucose response to a reporter gene in a rat insulinoma cell line (41) and since G-6-Pase mRNA abundance in islets increases with hyperglycemia in diabetic rats (23). The transcriptional glucose response for several other glucose-responsive gene promoters, such as L-pyruvate kinase, fatty acid synthase (FAS), S14, and promoter PI of acetyl-coenzyme A carboxylase (ACCpI), is dependent on a carbohydrate response element (ChoRE) that consists of two motifs related to the E-box motif (5'-CACGTG-3') separated by five base pairs (20, 27, 35, 39, 43). Glucose-dependent transcriptional activation from a ChoRE is reported to be mediated by the transcription factor carbohydrate response element-binding protein (ChREBP) together with the dimerization partner Mlx (16, 17, 29, 44, 53, 54).
Because the regulation of the G-6-Pase gene plays an important role in glucose homeostasis, we sought to map the cis-regulatory sequences that confer glucose responsiveness to the rat G-6-Pase gene promoter using glucose-responsive cell lines. In HL1C hepatoma cells, the G-6-Pase gene promoter was moderately induced by glucose and required coexpression of glucokinase. The promoter was highly glucose responsive in INS-1-derived 832/13 rat insulinoma cells. Furthermore, the endogenous G-6-Pase mRNA was robustly upregulated by glucose in 832/13 cells as was G-6-Pase mRNA in rat pancreatic islets. We took advantage of the vigorous response in 832/13 cells to identify glucose-responsive elements, which were then tested in HL1C cells. We found that the rat G-6-Pase gene promoter possessed two glucose-responsive regions, a proximal glucose-responsive region (–230/–112 with respect to the transcription start site) that contains a hepatocyte nuclear factor 1 (HNF-1) binding site and two cAMP response elements (CREs) as key elements and a distal glucose-responsive region containing a typical ChoRE sequence (–3702/–3686). The transcription factor ChREBP was recruited to the G-6-Pase gene promoter in a glucose-dependent manner to the distal glucose-responsive region, whereas transcription factors of the cAMP response element-binding protein (CREB) and CAAT/enhancer-binding protein (C/EBP) families were required for the glucose response from the proximal glucose-responsive region. In HL1C cells, the glucose response required the proximal, but not the distal region. We conclude that the rat G-6-Pase gene promoter contains several distinct cis-regulatory elements involved in its glucose responsiveness.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
-mercaptoethanol and further supplemented with 11 mM glucose. Pancreatic islets were harvested from male Wister rats weighing 
250 g using the Liberase R1 enzyme (Roche Diagnostics, Indianapolis, IN) according to the guidelines of the manufacturer under a protocol approved by the Duke University Institutional Animal Care and Use Committee. Approximately 100 primary rat islets per condition were cultured for 24 h in 2 ml of RPMI 1640 medium containing 10% FCS plus 2 or 20 mM glucose. Primary hepatocytes were isolated using the collagenase perfusion method as previously described (9), under a protocol approved by the Louisiana State University Health Sciences Center, New Orleans Institutional Animal Care and Use Committee. Plasmids. Expression plasmids for dominant-negative A-CREB and A-C/EBP proteins as well as the empty expression plasmid pRc/CMV500 (36) were kindly provided by Dr. Charles Vinson, National Cancer Institute (Bethesda, MD). cDNA was generated from RNA isolated from primary hepatocytes from male Wistar rats as previously described (9). The 58/1575 fragment of the glucokinase cDNA (numbering according to Ref. 4), including the entire coding region, was amplified by PCR using PfuUltra High-Fidelity DNA Polymerase from Stratagene (La Jolla, CA). The sequences of the forward and reverse primers were 5'-ggaattccgcctcaggagtcaggaacat-3' and 5'- cgggatcccgagcatttgtggtgtgtggag-3', respectively. The amplified fragment was cloned into pcDNA3.1/Zeo(–) from Invitrogen (Carlsbad, CA) between EcoRI and BamHI sites to generate the glucokinase expression plasmid pcDNA3.1/Zeo(–)GK. The cloned GK sequence is identical to the 58/1575 sequence published previously (4), except for a G instead of A at position 1466. The 286/685 sequence of rat G-6-Pase cDNA was likewise amplified from cDNA (numbering according to Ref. 22) prepared from primary hepatocytes. Forward and reverse primers had sequences 5'-gtgggtcctggacactgact-3' and 5'-gaaaccaaacaggaagaaggtg-3', respectively. Single deoxyadenosines were added to the 3' ends of the PCR product by treatment with 20 units of Taq DNA polymerase (Invitrogen) and 0.2 mM dATP at 70°C for 20 min. The treated PCR product was ligated into the pGEM-T Easy vector (Promega) to generate the plasmid 286/685 pGEM-T. The cloned 286/685 sequence differs from that reported previously (22) by a G instead of T at position 381. The plasmid –4712/+122 pBSSK was constructed by inserting the –4712/+122 promoter sequence of the rat G-6-Pase gene in PstI and SalI sites of the multicloning site of pBluescript SK(+). Rat G-6-Pase gene promoter fragments –1642/+64, –729/+64, –600/+64, –400/+64, –350/+64, –300/+64, –250/+64, –230/+64, –226/+64, –200/+64, and –100/+64 were amplified by PCR from –4712/+122 pBSSK using primer sequences designed to flank the promoter sequences by an upstream NheI site and a downstream XhoI site. The promoter fragments were inserted in the NheI and XhoI restriction sites of pGL3-Basic vector from Promega (Madison, WI) to generate plasmids called X pGL3-Basic, where X denotes the inserted promoter fragment sequence. The plasmid –4078/+64 pGL3-Basic was made by a three-way ligation between the –4078/–1642 promoter fragment from –4712/+122 pBSSK generated by XbaI and HindIII digestion, the –1642/+64 promoter fragment from –1642/+64 pGL3-Basic generated by HindIII and XhoI digestion, and the vector backbone from –1642/+64 pGL3-Basic generated by XhoI and NheI digestion. The human G-6-Pase gene promoter fragment –963/+67 was amplified from DNA isolated from Jurkat cells using Tri Reagent from Molecular Research Center (Cincinnati, OH) according to the guidelines of the manufacturer and cloned in the NheI and XhoI sites of pGL3-Basic to generate the plasmid h-963/+67 pGL3Basic. Site-directed mutagenesis was performed with the QuikChange Site-Directed Mutagenesis Kit from Stratagene according to instructions from the manufacturer. One and two copies of the rat G-6-Pase gene promoter sequence –230/–208, one and two copies of the rat G-6-Pase gene promoter sequence –3706/–3682, a scrambled nucleotide sequence of two copies of gene promoter sequence –3706/–3682, two copies of the rat ACCpI sequence –126/–102 (numbering as in Ref. 35), two copies of the mouse G-6-Pase gene promoter sequence –2943/–2919, and two copies of the human G-6-Pase gene promoter sequence –2479/–2455 were cloned into vectors pGL3-Promoter (Promega) and pTA-Luc (Clontech Laboratories) between the SacI and NheI sites using oligonucleotides of a defined sequence. Multiple copies of the rat G-6-Pase gene promoter sequences –167/–156 or –143/–132 were inserted in a similar fashion between the NheI and XhoI restriction sites. The rat G-6-Pase gene promoter sequences –230/–187, –230/–162, –230/–137, and –230/–112 and the human G-6-Pase promoter sequence –230/–107 were inserted between the NheI and XhoI sites of pTA-Luc. The fidelity of G-6-Pase gene promoter inserts and of site-directed mutations were confirmed by DNA sequencing.
Transfection.
832/13 cells and HL1C cells were transfected in 12-well tissue culture plates. 832/13 cells were seeded at 1 x 106 cells/well in maintenance medium 1 day before transfection. Transfections were done with 1 µg Firefly Luciferase reporter plasmid, 0.25 µg phRL-TK plasmid (Promega) as a control for transfection efficiency, and 5 µl Lipofectamine 2000 (Invitrogen) in 0.5 ml DMEM. After 2 h, an additional 0.5 ml of DME was added along with glucose to provide a final concentration of either 2 or 20 mM. HL1C cells were seeded in maintenance medium 1 day before transfection at a concentration to provide 7 x 105 cells/well at the time of transfection. Transfections were done with polylysine-coated adenovirus (3) purchased from Baylor College of Medicine (Houston, TX). The transfection mixture for one well contained 400 ng firefly luciferase reporter plasmid, with or without 10 ng pcDNA3.1/Zeo(–)GK or pcDNA3.1/Zeo(–) plasmid, 100 ng phRL-TK plasmid, 7 x 107 virus particles, and 0.663 µg polylysine in HEPES-buffered saline, pH 7.3. An aliquot of 51.7 µl of the transfection mixture was added to wells containing cells covered with 0.5 ml DMEM. After incubation for 2 h at 37°C, an additional 0.5 ml DMEM was added with glucose to obtain a final concentration of 2 or 20 mM. After the cells were cultured for an additional 20 h, they were lysed, and firefly and Renilla luciferase activity was measured with the Dual-Luciferase Reporter Assay System (Promega) in a TD-20/20 luminometer (Turner Designs, Sunnyvale, CA). Relative light units (RLU) were calculated as a ratio of firefly luciferase activity and Renilla luciferase activity. Transfection experiments were designed as generalized randomized complete block designs with blocks being different transfection experiments, and with each treatment (a transfection mixture plus a certain glucose concentration) given to the experimental units, which were individual wells, with duplicate or triplicate treatments applied for each experiment. For each set of transfection experiments, at least two different sets of plasmid preparations were used. Transfection experiments were analyzed by ANOVA of a generalized randomized complete block design (24) assuming the blocks had random effects. Because the SDs of RLU measurements tended to be proportional with the means, the data were logarithm-transformed before ANOVA. Following the ANOVA, contrasts of reporter activity at 2 mM vs. at 20 mM glucose and contrasts comparing the fold expression (20 vs. 2 mM glucose) for different luciferase reporters were tested by the least-significant difference procedure. Means ± SE were calculated and retransformed back to linear scale for data presentation.
Quantitative RT-PCR. Total RNA was isolated from 832/13 cells and primary hepatocytes using TRI Reagent and from pancreatic islets using the RNeasy Micro Kit (Qiagen, Valencia, CA), and DNase treatment to eliminate genomic contamination. Quantities of rat G-6-Pase mRNA and 18S rRNA were assayed by real-time quantitative RT-PCR on an ABI Prism 7700 Sequence Detection System (Applied Biosystems, Foster City, CA) using the One-Step RT-PCR kit (Applied Biosystems) by a procedure essentially similar to a previously described procedure (38). The Taqman probe for detecting G-6-Pase mRNA contained 6-carboxyfluorescein at the 5' end, and the quencher 6-carboxy-N,N,N',N'-tetramethylrhodamine at the 3' end. The probe spanned the splice site between exon 3 and exon 4. The sequences of the forward primer, the reverse primer, and the Taqman probe were 5'-tcgctatctttcgtggaaagaaa-3', 5'-cccagtatcccaaccacaaga-3', and 5'-ttcaagcaccggaatccatacgttgact-3', respectively. 18S rRNA values were determined relative to a total RNA standard from 832/13 cells, the RNA concentration of which was determined by optical density at 260 nm. The G-6-Pase mRNA concentrations were determined relative to a specific G-6-Pase RNA standard that was made by in vitro transcription with the SpeI-digested plasmid 286/685 pGEM-T used as the template. In vitro transcription and gel purification of the transcript were done as previously described (38). The RNA concentration of the G-6-Pase RNA standard was determined by the optical density at 260 nm.
ChREBP antibodies. Polyclonal rabbit antibodies were raised against a synthetic peptide (residues 431–442 of rat ChREBP) that was conjugated to keyhole limpet hemocyanin (Sigma) via the sulfhydryl moiety of a Cys residue added at the end of the ChREBP sequence. Anti-ChREBP antibodies were affinity purified by column chromatography on a peptide-SulfoLink Plus column (Pierce) as previously described (13).
Chromatin immunoprecipitation assays.
832/13 cells, 90% confluent, were pretreated overnight in medium containing 5 mM glucose and then were treated with medium containing either 2 or 20 mM glucose for 6 h before exposure to 1% formaldehyde for 10 min at room temperature. Glycine was added to a final concentration of 0.125 M, and, after 5 min, the samples were washed two times and harvested in cold PBS with protease inhibitors (catalog no. 11836170001; Roche Diagnostics). Cells were collected by centrifugation for 4 min at 2,000 g and suspended in lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris, pH 8.1). Chromatin immunoprecipitation (ChIP) assays were performed following the Upstate Biotechnology ChIP assay kit protocol (catalog no. 17–295), with slight modifications. Briefly, the cell lysate was sonicated with glass beads to yield 100–1,000 bp genomic DNA fragments. The lysate (2 ml) was precleared with 80 µl of a 50% slurry of protein G-agarose that contained 32 µg sonicated salmon sperm DNA, 80 µg BSA, 160 µg recombinant protein G-agarose suspended in 10 mM Tris·HCl, pH 8.0, 1 mM EDTA, and 0.05% sodium azide for 30 min at 4°C with agitation. After centrifugation at 1,000 g for 2 min, aliquots of the supernatant were incubated with an antibody directed against ChREBP or with normal rabbit IgG (sc-2027; Santa Cruz Biotechnologies, Santa Cruz, CA) overnight with agitation at 4°C. Immunocomplexes were recovered by incubation with a 50% slurry of salmon sperm DNA/protein G-agarose, in the buffer described above, for 1 h at 4°C. The beads were washed 5 min each with low-salt (catalog no. 20–153), high-salt (catalog no. 20–154), LiCl buffer (catalog no. 20–155), and 10 mM Tris, pH 8.0, 0.5 mM EDTA buffer. The chromatin complexes were eluted by adding freshly prepared elution buffer (1% SDS, 0.1 M NaHCO3) with rotation at room temperature for 30 min. The cross-linking was reversed by adding NaCl to a final concentration of 500 mM and heating at 65°C for 4 h. After incubation with 20 mg proteinase K for 1 h, the DNA was purified using a Qiagen PCR cleanup column, and target genes were quantified by real-time PCR with SyberGreen (iTaq SyberGreen Supermix with ROX, catalog no. 170–8853; Bio-Rad) using the purified DNA as template. Standard curves were constructed using twofold serial dilutions of the unbound DNA extracted from the 2 mM glucose IgG treatment (2.5 µl) as a reference input. The protein content of the different treatment groups was normalized using the Pierce BCA protein assay before immunoprecipitation, and the quantity of the amplicons were expressed as a percentage of the total reference input. The primer sequences used to amplify the distal glucose-responsive G-6-Pase gene promoter region by PCR were 5'-gcatcagccctgtgtgaata-3' and 5'-gagttgagggcaaacagagc-3'. The PCR primers for the proximal glucose-responsive G-6-Pase gene promoter region were 5'-aggaccaggaaggaggtcac-3' and 5'-gccctgatctttggactcaa-3'. The ChIP experiments were analyzed as simple split-plot designs by ANOVA according to Ref. 24, with the two groups of formaldehyde-treated, sheared chromatin collected from cells grown at either 2 or 20 mM glucose representing whole plots and with the split plots being the two aliquots of the chromatin preparations that were precipitated with the antibody directed against ChREBP or with the control IgG. Separate ANOVAs were calculated for each target that was amplified by PCR. Because the SDs of target concentrations tended to be proportional to the means, the data were logarithm-transformed before the ANOVA. After ANOVA, group means were compared using the least-significant difference procedure. Inferences of the target concentration at 2 vs. 20 mM glucose for a particular antibody were made after calculation of the approximate degrees of freedom by Satterthwaite's procedure. Means ± SE were calculated and retransformed back to linear scale for presentation of the data.
Computer analysis of promoter regions. Repetitive elements were determined using the web-based RepeatMasker program (A. F. A. Smit, R. Hubley, and P. Green; RepeatMasker Open-3.0. 1996–2004 http://www.repeatmasker.org). Sequence alignment of promoter regions from rat, mouse, and human were done with the web-based lalign program (37) from the GeneStream server (xylian.igh.cnrs.fr). Sequences of the G-6-Pase gene promoter regions from mouse and human were obtained from GenBank accession no. NT_039521 and accession no. NT_010755.15, respectively.
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
HL1C cells lack expression of glucokinase, but, when glucokinase is introduced via an adenovirus, glucose increases endogenous rat G-6-Pase mRNA abundance in these cells (9). We found that the glucose response of a G-6-Pase gene promoter-reporter construct (–729/+64 pGL3-Basic) was dependent on cotransfection with the glucokinase expression vector pcDNA3.1/Zeo(–)GK (Fig. 2A). As shown in Fig. 2B, promoter fragments with 5' ends ranging from –400 to –4078 supported greater luciferase activity, both in low (2 mM) and high (20 mM) glucose concentrations, than did shorter promoter fragments or the empty reporter construct. This shows that the –400/–200 promoter fragment contains elements required for basal G-6-Pase gene promoter activity in HL1C cells. Furthermore, there was a modest glucose response for the rat –400/+64 G-6-Pase gene promoter fragment and longer promoter fragments, and none for the shorter fragments, suggesting that elements required for the glucose response are located between –400 and –200.
|
The human G-6-Pase gene promoter (h-963/+67), which is homologous to the rat –1642/+64 fragment, was not clearly glucose responsive in HL1C cells (Fig. 2B). However, the same human promoter fragment conferred a glucose response in 832/13 cells, although to a lesser degree than the rat G-6-Pase gene promoter (Fig. 2C). This observation confirms previous results from INS-1 cells (41).
The concentration of endogenous G-6-Pase mRNA is rapidly and potently induced (>100-fold after 24 h) by 20 mM glucose in 832/13 insulinoma cells (Fig. 3A). The upregulation occurred principally between 5 and 14 mM glucose, consistent with the S0.5 of glucokinase (Ref. 34 and Fig. 3B). Induction of G-6-Pase mRNA by glucose also occurs in rat pancreatic islets, as shown in Fig. 3C, with the concentration of G-6-Pase mRNA being 12-fold higher in 20 mM glucose than in 2 mM glucose. It is important to note that the abundance of G-6-Pase mRNA from islets treated with 20 mM glucose was still 80-fold lower than in primary hepatocytes freshly isolated from ad libitum-fed rats (data not shown).
|
An HNF-1 binding site is required, but not sufficient, for the complete glucose response of the G-6-Pase gene promoter.
The –400/–200 rat G-6-Pase gene promoter fragment contains two E-boxes, –357/–352 (CAATTG) and –316/–311 (CAGTTG), of which the former is preserved as an E-box in the human G-6-Pase gene promoter. The –400/–200 fragment also contains a sequence (–226/–212) homologous to mouse and human G-6-Pase gene promoter sequences that bind HNF-1
or HNF-1 (26, 45, 47). This HNF-1 binding site is important for the full promoter response to insulin, cAMP, and glucocorticoids, as well as for basal activity in various cell lines (7, 25, 45, 47, 52). We tested the involvement of these sites in the glucose response by deleting the E-boxes and mutating the HNF-1 binding site (–218/–216 ATT 
CGC). In addition, we attempted to narrow down the location of important cis-regulatory elements by testing G-6-Pase gene promoter fragments –350/+64, –300/+64, and –250/+64 for glucose responsiveness.
Figure 4A shows that deletion of the E-boxes had no affect on the glucose response. Consistent with these results, the full glucose response was retained in the –250/+64 fragment, which eliminated the E-boxes but retained the HNF-1 site. However, mutation of the HNF-1 binding site abolished the glucose response in 832/13 cells. In HL1C cells, mutation of the HNF-1 site also blocked glucose responsiveness, as well as basal activity, in the context of the –400/+64 G-6-Pase gene promoter construct (Fig. 4B). Thus the HNF-1 binding site is essential for the G-6-Pase glucose response.
|
Fig. 7A, and data not shown). Likewise, insertion of the –400/–200 fragment in either pGL3-Promoter or pTA-Luc did not confer glucose responsiveness (data not shown). We conclude that the HNF-1 site is necessary, but not sufficient, for the glucose response of the proximal part of the G-6-Pase gene promoter.
|
|
|
–11/+14 and –36/–12, lack the transcription start site and the TATA box, respectively. This suggests that interfering with the normal transcription initiation of the G-6-Pase gene promoter results with a reduced, albeit not abolished, glucose response. The –211/–187 construct lost the 3'-flanking sequence of the HNF-1 site, which may have caused reduced responsiveness. The deletions of the remaining three constructs with reduced or absent glucose responsiveness (–186/–162, –161/–137, and –136/–112) remove sequences that, in the homologous regions of the mouse and human G-6-Pase gene promoters, are important cis-regulatory sequences such as insulin-responsive sequences, CREs, and glucocorticoid response elements (7, 25, 26, 40, 46, 52). Especially intriguing was that the –186/–162 deletion cuts through an upstream CRE at –165/–158 (termed CRE1 in Ref. 47), the –161/–137 deletion cuts through CRE1 and a downstream CRE at –141/–134 (termed CRE2 in Ref. 47), and the –136/–112 deletion cuts through CRE2, suggesting that CREs are important for the glucose response.
We subsequently tested the effects of CRE1 and CRE2 on glucose responsiveness in the context of –230/+64 pGL3-Basic. This was done by mutagenesis of CRE1 (–162/–157 CGTAAA AAGCTT) and CRE2 (–139/–135 GCATC CTAGA) similar to that of Ref. 47. Mutation of CRE2 significantly reduced the glucose response, whereas mutation of CRE1 alone, or in combination with the CRE2 mutation, completely abolished it (Fig. 6A). CRE1 is thus absolutely required for a glucose response, and CRE2 is required for the full response, in the context of –230/+64 pGL3-Basic. We confirmed that these sites are also important for glucose responsiveness of the G-6-Pase gene promoter in HL1C cells. When the two CRE sites were mutated in the –230/+64 promoter construct, the glucose response was lost (Fig. 6B).
We screened for the smallest promoter fragment that includes the HNF-1 site that would confer glucose responsiveness to a heterologous promoter. When the –230/+64 G-6-Pase promoter fragment is inserted upstream of the minimal promoter of the pTA-Luc vector, the resulting luciferase activity increases after treatment with 20 mM glucose in 832/13 cells (Fig. 6C). No glucose-induced expression was seen with insertions of the –230/–187 or the –230/–162 fragments in pTA-Luc. By contrast, insertion of the –230/–137 fragment, which contains CRE1, but not CRE2, resulted in significant but reduced glucose induction compared with –230/+64. Insertion of the –230/–112 fragment, which contains CRE1, CRE2, and the HNF-1 site, was fully glucose responsive.
A combination of the HNF-1 binding site and the CREs is sufficient for a glucose response in a heterologous promoter context. We tested whether a combination of the HNF-1 binding site and the CREs from the rat G-6-Pase gene promoter would be sufficient for conferring a glucose response to the minimal promoter of pTA-Luc. Two copies of the HNF-1 binding site (–230/–208) were inserted between the SacI and NheI restriction sites. In addition, multiple copies of the CRE1 region (–167/–156) and/or CRE2 region (–143/–132) were inserted between the NheI and XhoI restriction sites. As shown in Fig. 7A, the combination of multiple HNF-1 binding sites and CREs promotes a glucose response in 832/13 cells. The construct with four CRE1 sequences promote a stronger glucose response and stronger basal expression than that with four CRE2 sequences. The strongest response was seen with the plasmid that contained two copies of the HNF-1 binding site and six copies of the CRE1 region. This plasmid also exhibited significant glucose-dependent Luciferase expression in HL1C cells (Fig. 7B). Finally, when four copies of the CRE 1 region were inserted in pTA-Luc, in the absence of an HNF-1 site, the resulting plasmid gave a glucose response in 832/13 cells (Fig. 7C). Thus multiple copies of a CRE sequence are sufficient for a glucose response in 832/13 cells.
The upstream G-6-Pase gene promoter region contains a ChoRE.
We tested whether the –3702/–3686 rat G-6-Pase gene promoter sequence, CATATG-CTGAG-CACATG, consisting of two E-boxes separated by five base pairs, functions as a ChoRE in 832/13 cells. Deletion of the –3702/–3686 sequence from –4078/+64 pGL3-Basic resulted in a significant decrease in glucose responsiveness (Fig. 8A), showing its involvement in the total response. By comparison, deletion of the adjacent –3639/–3620 sequence, which is completely conserved between the rat and human promoters, had no significant effect on glucose responsiveness (data not shown). Mutation of the HNF-1 binding site (–218/–216 ATT CGC) in –4078/+64 pGL3-Basic led to a drastic reduction in luciferase expression, but the residual activity was still significantly upregulated by glucose (Fig. 8A). This residual responsiveness was further decreased when the –3702/–3686 fragment was deleted in addition to the HNF-1 site mutation (Fig. 8A, bottom).
|
We did not observe a glucose effect of the G-6-Pase ChoRE in HL1C cells in transient transfection experiments, nor did we observe a response from the potent ACCpI ChoRE (data not shown). This suggests that transcriptional enhancement from ChoREs is muted in HL1C cells under these experimental conditions.
Glucose responsiveness of the human G-6-Pase gene promoter. The proximal glucose-responsive region (–230/–112) and the distal ChoRE (–3702/–3686) of the rat G-6-Pase gene promoter occur in regions that have homology to the human G-6-Pase gene promoter. We investigated whether the homologous regions of the human promoter are involved in a glucose response.
We first tested whether the HNF-1 binding site and the two CREs are involved in the glucose response of the human G-6-Pase gene promoter. In the context of the human –963/+67 G-6-Pase gene promoter fragment, we mutated CRE1 (–157/–152 CGTAAA AAGCTT) and CRE2 (–134/–130 GCATC CTAGA). In addition, two mutations of the HNF-1 site were introduced. Mutation mA (–226/–221 AGTTAA GGCCGC) is similar to that of Ref. 26, and mutation mB (–218/–216 ATT CGC) is similar to the HNF-1 site mutation of the rat G-6-Pase gene promoter. The results in Fig. 9A show that, although the mutated fragments all conferred a modest glucose response, the degree of upregulation was significantly reduced when compared with the wild-type promoter sequence. This indicates that the HNF-1 binding site, CRE1, and CRE2 are all required for the full glucose response of the human G-6-Pase gene promoter. We further confirmed that the human –230/–107 promoter fragment, which is homologous to the rat –230/–112 promoter fragment, is capable of conferring glucose responsiveness to the minimal promoter of the pTA-Luc vector (Fig. 9B).
|
ChREBP binds to the distal promoter region in a glucose-dependent manner. ChREBP has recently been described as the transcription factor that is responsible for glucose-activated transcription from a ChoRE (16, 17, 29, 44, 53, 54). We tested whether ChREBP binds to the rat G-6-Pase gene promoter in 832/13 cells in situ using a quantitative ChIP assay that amplified the –3793/–3580 or –274/–63 promoter fragments after immunoprecipitation of chromatin with an antibody directed against ChREBP. Whereas the downstream G-6-Pase gene promoter region did not appear to bind ChREBP, there was significant glucose-dependent binding of ChREBP to the upstream G-6-Pase gene promoter containing the ChoRE (Fig. 10).
|
, C/EBP-, HNF-3, HNF-3
, and the forkhead transcription factor FKHR (26, 40, 52). We tested whether the glucose responsiveness of the proximal glucose-responsive region requires transcription factors related to CREB and C/EBP. Expression plasmids for dominant-negative proteins (A-CREB and A-C/EBP) were included in transfection experiments and compared with transfections with the empty expression plasmid pRc/CMV500. A-CREB is a dominant-negative of CREB and likely also of the closely related family members activating transcription factor-1 and cAMP-responsive element modulator (1), whereas A-C/EBP is a dominant-negative of C/EBP family members (21, 36). Figure 11, top, shows that the A-CREB expression plasmid and, to a lesser degree, the A-C/EBP expression plasmid diminish the glucose response of the –400/+64 fragment of the rat G-6-Pase promoter in a dose-dependent manner. A similar result was obtained for a reporter plasmid containing four copies of the G-6-Pase CRE1 region (–167/–156) upstream of the minimal promoter of pTA-Luc (data not shown). In contrast, Fig. 11, bottom, shows that the A-CREB and A-C/EBP expression plasmids had no effect on the glucose response on the pTA-Luc-derived plasmid containing two copies of the rat G-6-Pase ChoRE region (–3706/–3682). We conclude that the proximal glucose-responsive region of the rat G-6-Pase promoter, but not the distal ChoRE, requires transcription factors of the CREB and C/EBP families for full glucose responsiveness. These results further illustrate that the glucose responsiveness in 832/13 cells of the proximal and distal parts of the G-6-Pase promoter occur by disparate mechanisms.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
A distal glucose-responsive region is a typical ChoRE: the –3702/–3686 sequence consists of two E-boxes (CANNTG) separated by five base pairs, and its deletion from –4078/+64 pGL3-Basic reduced the glucose induction approximately threefold and conferred glucose responsiveness to heterologous promoters. The E-box half-site CACG has previously been described as being important for the function of a ChoRE (35). The rat G-6-Pase ChoRE does not contain a CACG motif. However, it is a near-perfect palindrome that contains the sequence 5'-CATGTG-3', which is also part of the ChoREs of the rat ACCpI and the rat FAS promoter (35, 39). The sequences in the mouse and the human G-6-Pase gene promoters that are homologous to the rat G-6-Pase ChoRE each have only one consensus E-Box. The mouse sequence, which retains the CATGTG motif, also functioned as a ChoRE. The human sequence, which does not, at most functioned as a very weak ChoRE. This further suggests that the CATGTG motif is important for the function of the G-6-Pase ChoRE.
The mechanism by which glucose-responsive transcriptional activation occurs from a ChoRE has lately been described (16, 17, 29, 44, 54). Increased glucose flux leads to dephosphorylation of the cytoplasmic ChREBP transcription factor, which allows nuclear localization and increased binding of ChREBP together with its dimerization partner Mlx to the ChoRE with ensuing activation of target genes. The G-6-Pase ChoRE might work by such a mechanism, since we have observed binding of ChREBP in 832/13 cells in situ to the distal glucose-responsive region of the promoter in a glucose-dependent manner. It has been reported that the mRNA and the transcription rate of ChREBP are upregulated by glucose in INS-1 cells (53). However, preliminary experiments indicated that the binding of ChREBP to the G-6-Pase gene promoter occurs rapidly, within 15 min, after exposure to 20 mM glucose (P. Zhang, and D. K. Scott, unpublished observations). Thus increased binding of ChREBP to the G-6-Pase ChoRE is not likely to be due simply to an increased concentration of ChREBP protein in these cells.
A proximal glucose-responsive region (–230/–112) provides a glucose response to a heterologous minimal promoter in 832/13 cells. There are no obvious ChoREs, nor any E-boxes in this region. Within this region, an HNF-1 binding site is necessary, but not sufficient, for glucose responsiveness. Therefore, it can be described as an accessory factor for the glucose response. The HNF-1 binding site plays a similar role for the response of the mouse or human G-6-Pase gene promoter to insulin, cAMP, and glucocorticoids (25, 45, 47).
Two CREs are surprisingly also required for the full glucose response. These CREs are completely conserved between mouse, rat, and human G-6-Pase gene promoters. In the context of the –200/+64 pGL3/Basic construct, the CREs are not sufficient for a response. The glucose response thus seems to involve a glucose-dependent interplay of transcription factors binding to the HNF-1 and CRE sites. Indeed, a combination of multiple CREs and HNF-1 binding sites from the rat G-6-Pase gene promoter is sufficient for conferring glucose responsiveness to the promoter of the pTA-Luc vector in both 832/13 and HL1C cells. Furthermore, multiple copies of the CRE1 region are sufficient for conferring a glucose response to the promoter of the pTA-Luc vector in 832/13 cells. We therefore suggest that the transcription factors activating the proximal G-6-Pase gene promoter in a glucose-dependent manner interact with the CREs of the promoter. The CRE1 site is a bona fide CRE, since it can confer responsiveness to cAMP and protein kinase A to a heterologous promoter and allow in vitro binding of the CREB transcription factor (40, 47). It may also be involved in glucocorticoid responsiveness and basal activity of the promoter (40), and it overlaps an insulin-responsive sequence and a glucocorticoid response element (7, 52). The CRE2 site overlaps binding motifs for HNF-3 and C/EBP, and gel shift assays and in vitro footprinting have revealed binding of CREB, C/EBP-, C/EBP-, HNF-3, HNF-3, and the forkhead transcription factor FKHR to the region of CRE2 (26, 52). Recently, C/EBP- and C/EBP- have been shown to be constitutively bound to the rat G-6-Pase gene promoter in rat FAO hepatoma cells in situ, whereas CREB binds after induction with forskolin (11). A potential candidate for mediating a glucose response through the CREs in insulinoma cells is CREB which is activated by glucose in INS-1 cells (51). Indeed, we found that CREB or a closely related family member, as well as a member of the C/EBP family of transcription factors, are required for the full glucose responsiveness.
In HL1C cells, the G-6-Pase gene promoter is glucose responsive through the proximal glucose-responsive region. The CREs and the HNF-1 binding site are also required elements for the response in this cell line. However, there was no clear glucose response from the G-6-Pase ChoRE sequence, and the potent ChoRE of ACCpI also failed to confer glucose responsiveness in transient transfection experiments in this cell line. It is thus apparent that HL1C cells do not elicit a strong glucose-mediated response from traditional ChoREs, at least not under the employed experimental conditions. A recent report indicates that Mlx is required for the full glucose response of the G-6-Pase gene in primary rat hepatocytes (28). This observation provides evidence to suggest that the ChREBP·Mlx dimer acts through a ChoRE, with the distal ChoRE identified here as a likely cis-regulatory element.
To our knowledge, this is the first report of glucose-responsive regions in the rat G-6-Pase gene promoter, although the human G-6-Pase gene promoter has previously been described as glucose responsive. In the rat insulinoma cell line INS-1, luciferase reporter constructs with –1227/+57 and –161/+4 sequences of the human G-6-Pase gene promoter were reported to exhibit glucose responsiveness (41). Consistent with these results, we observed glucose activation from the –963/+67 fragment of the human G-6-Pase gene promoter in 832/13 cells and found that the HNF-1 binding site, the CRE1, and the CRE2 are required for full glucose responsiveness. The human G-6-Pase gene promoter was also found to be glucose responsive in the human enterocyte cell line Caco-2/TC7 and the human hepatoma cell line HepG2 (8), but different promoter regions seemed involved. In HepG2 cells, the –299/+57 region was sufficient for a glucose response, but in Caco-2/TC7 cells, elements between –299 and –1227 were required. A binding site for the transcription factor aryl receptor nuclear translocator (ARNT) was required for the full response, and a decrease in the concentration of ARNT protein by RNA interference diminished the glucose response (8). The ARNT binding site, as well as three other regions for putative transcription factor-binding sites between –299 and –1227 of the human promoter, are located in Alu elements with no clear homology in the –4078/+64 rat G-6-Pase gene promoter. Thus there appears to be both cell-line and species-specific differences with regard to the glucose-responsive regions of the G-6-Pase gene promoter.
Upregulation of G-6-Pase by glucose in hepatocytes counteracts the effects of insulin in the fed state. Metabolic control analysis of gluconeogenesis in rat hepatocytes indicated that G-6-Pase has a very low flux control coefficient (12). The induction of G-6-Pase by glucose may thus have a limited effect on the gluconeogenesis rate. On the other hand, an increase in the G-6-Pase activity causes a noticeable decrease in the concentration of glucose 6-phosphate and in the rates of glycolysis and glycogen synthesis in rat hepatocytes (2). The authors of the latter study hypothesized that the main regulatory function of G-6-Pase is to buffer the glucose 6-phosphate concentration and that the induction by glucose is a compensatory mechanism. This could limit the glycogen storage in the postprandial period, as has been previously suggested (5). Consistent with this view, among the metabolic changes of rats with a 1.6- to 3-fold increase in hepatic G-6-Pase activity, there was a significant decrease in hepatic glycogen content (50).
We observed that glucose induces G-6-Pase mRNA in rat islets. This glucose induction may explain the increased amounts of G-6-Pase expression in islets from diabetic and/or obese rodents (19, 23, 48). Glucose-mediated release of insulin from pancreatic -cells occurs through glucose metabolism and an increase in the ATP-to-ADP ratio (reviewed in Ref. 14). If G-6-Pase were sufficiently overexpressed, increased cycling between glucose and glucose 6-phosphate with concomitant hydrolysis of ATP as well as a decrease in glycolytic flux would be expected to result in decreased insulin secretion. This has been demonstrated in INS-1 cells overexpressing the G-6-Pase via recombinant adenovirus (49). It has also been reported that pancreatic islets from hyperglycemic ob/ob mice show increased glucose cycling (18). However, even after treatment with 20 mM glucose, the concentration of G-6-Pase mRNA in rat islets remains 80-fold lower than the concentration that we have observed in freshly isolated hepatocytes from ad libitum-fed male Wistar rats (data not shown). It is thus questionable whether the induction of G-6-Pase by glucose is sufficient to acutely affect insulin secretion or other aspects of -cell physiology.
It is remarkable that G-6-Pase mRNA is induced by glucose in several cell types of human and rodent origin and by several different mechanisms, suggesting an important physiological role for this glucose induction. From the present study, we conclude that the rat G-6-Pase gene promoter responds to glucose through multiple cis-regulatory elements.
|
GRANTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
Present addresses: P. Zhang, M. Charbonnet, and D. K. Scott, Div of Endocrinology, Univ of Pittsburgh Medical Center, Pittsburgh, PA; C. Doumen, Dept of Mathematics and Natural Sciences, Collin County Community College, Plano, TX; and J. W. Haycock, PhosphoSolutions, Denver, CO.
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 The nucleotide sequence of the –4078/–1643 of the rat G-6-Pase gene promoter has been deposited in the GenBank database under GenBank accession no. DQ100345. 
|
REFERENCES |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
