Identification of a Glucose Response Element in the Promoter of the Rat Glucagon Receptor Gene*

We cloned the 5′ upstream region of the rat glucagon receptor gene, demonstrating that the 5′ noncoding domain of the glucagon receptor mRNA contained two untranslated exons of 131 and 166 nucleotides (nt), respectively, separated by two introns of 0.6 and 3.2 kilobase pairs. We also observed an alternative splicing involving the 166-base pair exon. Cloning of up to 2 kilobase pairs of the newly identified genomic domain and transfection of various constructs driving a reporter gene, in pancreatic islet cell line INS-1, uncovered a strong glucose regulation of the promoter activity of plasmids containing up to nucleotide −868, or more, upstream from the transcriptional start point. This promoter activity displayed threshold-like behavior, with low activity of the promoter below 5 mm glucose, and maximal activation as of 10 mm glucose. This glucose regulation was mapped to a highly palindromic 19-nucleotide region between nt −545 and −527. Indeed, deletion or mutation of this sequence abolished the glucose regulation. This domain contained two palindromic “E-boxes” CACGTG and CAGCTG separated by 3 nt, a feature similar to the “L4 box” found in the pyruvate kinase L gene promoter. This is the first description of a G protein-coupled receptor gene promoter regulated by glucose.

We cloned the 5 upstream region of the rat glucagon receptor gene, demonstrating that the 5 noncoding domain of the glucagon receptor mRNA contained two untranslated exons of 131 and 166 nucleotides (nt), respectively, separated by two introns of 0.6 and 3.2 kilobase pairs. We also observed an alternative splicing involving the 166-base pair exon. Cloning of up to 2 kilobase pairs of the newly identified genomic domain and transfection of various constructs driving a reporter gene, in pancreatic islet cell line INS-1, uncovered a strong glucose regulation of the promoter activity of plasmids containing up to nucleotide ؊868, or more, upstream from the transcriptional start point. This promoter activity displayed threshold-like behavior, with low activity of the promoter below 5 mM glucose, and maximal activation as of 10 mM glucose. This glucose regulation was mapped to a highly palindromic 19-nucleotide region between nt ؊545 and ؊527. Indeed, deletion or mutation of this sequence abolished the glucose regulation. This domain contained two palindromic "E-boxes" CACGTG and CAGCTG separated by 3 nt, a feature similar to the "L4 box" found in the pyruvate kinase L gene promoter. This is the first description of a G protein-coupled receptor gene promoter regulated by glucose.
The primary physiological role of glucagon, together with insulin, is maintenance of normal glycemia. The liver has a central role in handling absorbed nutrients and in the regulation of hepatic glycogen disposability; this requires high density of glucagon receptors in the liver (for review see Refs. 1 and 2). The glucagon receptor mRNA has also been detected at variable levels in other tissues such as heart, kidney, adrenal gland, and adipose tissue (3)(4)(5), as well as in pancreatic islets, especially in B cells (6,7). The expression of the glucagon receptor mRNA is stimulated by glucose and inhibited by cyclic AMP, both in liver (8) and in cultured endocrine cells (9). Consequently, the promoter of the glucagon receptor gene could contain regulatory elements for these factors. It has been recently suggested by Burcelin et al. (10) that the glucose regulation of the glucagon receptor mRNA level differs from the glucose regulation of other genes such as Glut2, and might be mediated by triose metabolites, suggesting the existence of new enhancer sequences. To address this question, a prerequisite is the precise knowledge of the complete 5Ј mRNA sequence. For the glucagon receptor mRNA, there was an ambiguity; the glucagon receptor cDNA has been first cloned (from rat) simultaneously by us (11,12) and by Jelinek et al. (13). Our sequence (GenBank accession number L04796) was almost identical to that of Jelinek (GenBank accession number M96674) except for the 5Ј end; the first 25 nt 1 in our sequence (corresponding to positions Ϫ105 to Ϫ80 from the ATG codon) differed from the sequence published by Jelinek (13). The coding domain of the rat glucagon receptor gene is highly fragmented; it contains 12 introns with uneven splicing maturation (14) that may produce cloning artifacts resulting in the observed differences described for the 5Ј sequences. Therefore, we first identified the correct organization of the 5Ј end of the rat glucagon receptor mRNA. Thereafter, we cloned a fragment of genomic DNA located upstream from the transcription start point, and identified this fragment as a glucose activable promoter region using reporter gene studies. Finally, we localized a L4-like box sequence as the central motif of the glucose-induced gene stimulation.

MATERIALS AND METHODS
General Methods-Rat liver and heart fragments were homogenized in 4 M guanidine isothiocyanate, and total RNA was purified on CsCl gradient (15); poly(A) RNA was purified on oligo(dT)-cellulose (Invitrogen).
Reverse transcription was performed using Superscript II™ (Life Technologies, Inc.). Anchored PCR of the 5Ј end was performed after single-strand ligation of phosphorylated oligonucleotide to cDNA using the 5Ј RACE kit (CLONTECH). Subcloning of PCR product was carried out by TA ligation to the pCRII plasmid (Invitrogen).
Sequencing of plasmids purified on Qiagen™ columns was performed with the Sequenase or Thermosequenase kit (U.S. Biochemical Corp; Amersham). Polymerase chain reaction was performed as indicated in the legend of figures. Analysis of amplified DNA fragments on agarose gel, transferred to nitrocellulose filters and hybridized to 5Ј-32 P-labeled oligodeoxynucleotide was, performed as detailed by Sambrook et al. (15).
Genomic Cloning and Construction of Reporter Gene Plasmids-Rat genomic DNA, partially digested with Sau3AI (15), was cloned in XhoIdigested and partially filled -FixII bacteriophage DNA (Stratagene), and packaged as recommended by the manufacturer. The resulting primary genomic library was screened at 40,000 plaque-forming units/ 145-mm plate. We initially screened the -FixII library with 32 P-labeled oligonucleotides complementary to nt ϩ147 to ϩ128 and to nt ϩ 772 to ϩ739 of the glucagon receptor CDS. One positive clone (P 10-C) was isolated. First, a 3.5-kb XbaI/PstI hybridizing fragment was subcloned into pUC18. Second, a 1-kb XbaI/XbaI fragment (located upstream from of the XbaI/PstI fragment) was subcloned in pZErO™-1 (Invitrogen). The XbaI/PstI fragment contained a 2.5-kb sequence corresponding to the 5Ј end of the coding domain of the glucagon receptor gene (Ref. 14; GenBank accession number L31574), and, in addition, a 1-kb sequence upstream from the previously described initiation codon. The 5Ј end extended sequence is listed in GenBank under accession number U63021.
In a second screening we used 32 P-labeled oligonucleotides J.for and S.for (described below). One positive -FixII clone (JS 3-A) was isolated, and a 1.3-kb BamHI/BamHI hybridizing fragment was subcloned first in pZErO™-1 (Invitrogen) and then in the BamHI site of pBlCat6 vector (clone B/B).
An oligonucleotide probe based on the 5Јend of this fragment was used as a probe to subclone a 2-kb PstI/SacI fragment in pBluescript SKϩ resulting in clone "P/S." The two subcloned fragments contain a 130-bp overlap with a unique XbaI site. We ligated a HindIII/XbaI fragment of clone P/S with a XbaI/XbaI fragment of the clone B/B in the pBlCat6 vector digested with HindIII/XbaI. The sequence of this constructed plasmid (called clone P/ctr/B) was deposited in GenBank under accession number U63022.
In the first transfection experiments, we used clones B/B and P/ctr/B. These plasmids gave only a low reporter gene activity in transfected cells (see below); these plasmids bear approximately 1 kb of sequence situated downstream from the mRNA start point. This intervening sequence may prevent the efficient transcription of the reporter gene. We therefore prepared shortened plasmids by deletion of the sequence located immediately after the end of the first untranslated exon, using a unique ApaI site.
The clone B/B was digested by ApaI and XhoI, blunted and ligated to give clone "pCB." The clone P/ctr/B was digested by ApaI and NotI, blunted and ligated to give the clone "pCC". The latter clone contained a XhoI site located 1 kb upstream from its 3Јend (ex-ApaI site), and a second XhoI site in the polylinker, so that XhoI digestion excised a 1-kb insert, which was subcloned in XhoI site of pBlCat6 vector in both its normal and reverse orientation (clones pCX and pCX i ).
The clone pCC was digested by BamHI, and a 0.5-kb fragment was removed; the plasmid was then recirculated by ligation. The first BamHI site is in position Ϫ287, and the second BamHI site is in the polylinker of pBlCat6, so BamHI digestion removed from the plasmid pCC, fragment an equivalent to the pCB plasmid insert. The remaining promoter domain (Ϫ2 kb to Ϫ287) was then excised by HindIII/BamHI and subcloned in the HindIII/BamHI sites of the pBlCat5 plasmid, i.e. upstream from the HSV thymidine kinase basal promoter. The resulting plasmid was called pTkCd.
Mutations-Mutations were introduced using the QuickChange sitedirected mutagenesis kit from Stratagene according to manufacturer's instruction. The principal is based on Pfu amplification of the whole plasmid and digestion of nonmutated DNA using DpnI digestion. DpnI does not digest unmethylated, Pfu-amplified DNA.
We used as forward primer GCACGTGTGACAGTTGCAATCTTTC for the preparation of pTkCd-m1 clone, and CCACACACGGTGGTCGT-GTGAGTGCTGCAATCTTTC as forward primer for the preparation of pTkCd-m2 clone. Clone pTkCd-m1 carries a single nucleotide change. Clone pTkCd-m2 contains a double mutation of the two first nucleotides of each palindromic E-box (change of four nucleotides).
We used CTTTGCTCCCACACACGGTGCAATCTTTCACCCAGGA-GCC as forward primer to produce the pCX⌬ clone. This clone is characterized by a 15-nt deletion located between Ϫ543 to Ϫ529, i.e. deletion of the fragment containing the two E-boxes of the palindromic G-box (L4-like box). Reverse primers were complementary to the above described forward primers. Mutated clones were sequenced as described above.
Transfection Studies-Insulin-secreting cell line INS-1 was kindly provided by Prof. Wollheim (University Medical Center, Geneva, Switzerland). These cells were cultured in modified RPMI 1640 as described previously (16). Cell culture material and media were from Life Technologies, Inc. Transfections were usually performed on cells 4 -5 days after seeding in six-well plates, in order to achieve roughly 60% cell confluence. The day before transfection, the usual cell medium was replaced with by RPMI 1640 medium with a low glucose concentration (5 mM instead of 11 mM). The cationic liposome Dosper (Boehringer Mannheim) was used according to the manufacturer's instructions. In standard transfections, we used 1.5 g of plasmid DNA and 10 l of Dosper reagent/35-mm diameter well. After 5 h, the transfection medium was replaced by RPMI 1640 medium, with variable glucose concentrations.
In several experiments, the transfection yield was estimated by cotransfection with 0.5 g of pcDNA3.1/his/lacZ (Invitrogen). Galactosidase activity was quantified on 2% of the cell extract using o-nitrophenyl-␤-D-galactopyranoside (Invitrogen) according to the manufacturer's recommended procedure. Measurement of galactosidase activity was highly reproducible, with intra-assay variation of less than 25%.
Chloramphenicol acetyltransferase (CAT) activity of transfected cells was assayed by acylation of [ 14 C]chloramphenicol (ICN) using n-butyryl-CoA (Promega) as indicated in the Promega protocol. An aliquot of the cell extract from individual wells was used for CAT assays and TLC analysis. Autoradiographic densities (measure of "volume" of absorption: integrated optical densities of the spot surface) were estimated using "Viber-Lourmat" image analysis apparatus. In control experiments we found a linear correlation between the amount of CAT in the assay and the "volume" of absorption measured in the range of 0 -125 milliunits. Activity of our samples from transfected cells never exceeded 100 milliunits of CAT.

RESULTS
Amplification of cDNA Using Primers Based on the Conflictual Sequences-Analysis of the 5Ј end sequence of the glucagon receptor mRNA was obtained by amplification of rat liver and heart cDNA using two forward primers J.for and S.for, based on the conflictual sequences reported in the literature. A forward primer based on an unambiguous sequence in the 5Ј end was used as a control. Fig. 1A shows the amplified products electrophoretic pattern. The amplification specificity was confirmed by Southern blotting, using internal oligodeoxynucleotides as probes (data not shown).
The major products were subcloned and sequenced. Obtained sequences, confirmed later by sequencing products of anchored PCR and by sequencing cloned genomic DNA (see below), demonstrated that glucagon receptor mRNAs contain two untranslated exons, and that the second one could be spliced out. We called the first 5Ј Anchored PCR of cDNA 5Ј Ends (5Ј RACE)-To identify the mRNA transcription start site, and estimate the relative abundance of these polymorphous glucagon receptor mRNA populations, we performed 5Ј end-anchored PCR.
We ligated liver and heart single-stranded cDNA to a phosphorylated anchor primer, and then used three different reverse primers to amplify this anchored cDNA. The procedure is detailed in the legend of Fig. 2. Hybridization using the internal labeled oligonucleotide probe (R-20) allowed us to visualize a major band with a smear for the liver and two separate bands for the heart (Fig. 2). The same pattern was observed in the three tested conditions. Shorter exposure clearly indicated that most of the DNA amplified from liver had the same mobility as the smaller fragment amplified from the heart. The smaller band corresponded to approximately 350 bp with R147, and about 160 bp with R-20 reverse primer. The size of the larger fragment amplified from the heart was in good agreement with the results expected in the presence of an additional 166-bp unspliced 5Ј exon. These results indicated that the average size of the 5Ј untranslated end of the glucagon receptor mRNA was approximately 180 bp or 180 ϩ 166 bp.
We subcloned the products of the second step of the anchored PCR into the pCRII vector, and hybridized recombinant colonies with 5Ј-labeled J.for and S.for oligonucleotides. 90% of clones originating from the liver hybridized with "J.for" only, and did not contain the US exon (confirmed by sequencing). In contrast, among the 55 colonies originating from the heart, 21 hybridized with the "J.for" probe only, 31 (representing the longer form comprising the 166-bp US exon) hybridized with both "J.for" and "S.for" probes, and 3 colonies hybridized with "S.for" only and contained truncated US exon sequences. This result suggested that all transcripts originated from a single promoter located upstream UJ exon, rather that from two separate promoters found upstream from each untranslated exon. The reason for a tissue-specific use of the second US exon remains unclear.
To identify the transcription start point, we subcloned the products of the second anchored PCR step in the pCRII vector, and sequenced the 5Ј end of 30 clones originating from both liver and heart. All the inserts yielded the same sequence, but starting at various points ranging from 73 to 131 nt upstream from the 3Ј end of exon UJ (see Fig. 3). These sizes were in good agreement with the average size of the 5Ј RACE amplified fragments shown in Fig. 2.
A supplement verification of the 5Ј mRNA length was performed by two additional independent anchored PCR. Primers were located on exon UJ, in order to favor the longest 5Ј transcripts. In a population of 20 clones, we did not find a sequence exceeding the point indicated in Fig. 3B by "start exon UJ," strongly suggesting that we had indeed identified the actual mRNA 5Јend.
Screening of Rat Genomic Library and Genomic Organization-We constructed and screened a bacteriophage -FixII rat genomic DNA library. The first genomic clone obtained contained the complete glucagon receptor coding domain, and 1.5-kb sequence upstream from the ATG initiation codon. The genomic DNA sequence upstream from position Ϫ79 (GenBank accession number U63021) diverged from both published cDNA 5Ј end sequences and represented thus obviously an intron. This sequence showed 83% identity (over 992 nt upstream from the ATG) with the reported putative mouse glucagon receptor gene promoter (GenBank accession number L38612) (17).
In the second screening, we obtained DNA fragments located up to 5 kb upstream from the initiation codon (GenBank accession number U63022). The genomic organization of the whole glucagon receptor gene deduced from these data is illustrated in Fig. 3A. The comparison of the rat glucagon receptor genomic sequence with published cDNA sequences confirmed the existence of two exons in the 5Ј untranslated domain. Even our longest 5Ј-cDNA clones sequences were identical to the genomic DNA sequence. These clones indicate the putative Heart or liver cDNA fragments were amplified using two alternative forward primers fitting conflicting glucagon receptor 5Ј end sequences; S.for was based on sequence nt Ϫ101 to Ϫ82 of our sequence (L04796), and J.for was based on sequence Ϫ105 to Ϫ86 of Jelinek's sequence (M96674). We used a forward primer G.for based on nt Ϫ15 to ϩ3 (located downstream from the position where the glucagon receptor sequences diverged) as a control. The same reverse primer R147 (corresponding to nt ϩ128 to ϩ147 of glucagon receptor CDS) was used with the three forward primers. This reverse primer is located downstream of the 100-bp intron present in the genomic DNA at position nt ϩ63. After 30 cycles of PCR (94, 60, and 72°C, 1 min. each) the reaction products were analyzed on 1.2% agarose gel. The intensity of cDNA fragments amplification was lower in heart as compared with liver, as expected since the glucagon receptor mRNA content is 7 times lower in heart than in the liver (3). The specificity of the amplification was confirmed by Southern blotting, using internal oligodeoxynucleotides as probe (data not shown).Panel B, schema of structure of the 5Ј end of the glucagon receptor mRNA deduced from the PCR results. The S.for primer, based on our 5Јend sequence, amplified a single 250-bp fragment as expected (246 bp) for the mRNA corresponding to the sequence L04796 (3). Likewise the G.for primers also yield only the expected 189-bp DNA fragment. Minor fragments, 100 bp longer, originated from immature mRNA containing the 100-bp intron I (14). Primer J.for, based on the 5Ј end published by Jelinek, amplified two fragments of 250 and 400 bp in both heart and liver. The larger fragment predominated in the heart. The 250-bp fragment corresponded to the published sequence, M96674 (252 bp). The 400-bp fragment possessed an additional 166-bp exon anchored at position nt Ϫ79. The 3Ј end of this additional exon corresponded to our sequence L04796. transcription start site (Fig. 3B). The experimentally determined size of 5Ј domain of cDNA (Fig. 2) is in good agreement with this position of the start point.
The sequence neighboring the putative start site was in relative agreement with the consensus sequence of promoters lacking TATA boxes. The genomic sequence upstream from this putative start site was highly (G ϩ C)-rich and could correspond to the basal gene promoter. Moreover, we found at position Ϫ527 to Ϫ 545 upstream from the transcription start point, highly palindromic sequence of 19 nucleotides containing a canonical E-box "CACGTG", followed by three nucleotides "tga" and another palindromic E-box "CAGCTG" (Fig. 3B). This feature is very similar to the "L4 box" found in pyruvate kinase L or Spot14 promoter, in which it is thought to be the glucose response element (GIRE or ChoRE; see "Discussion"). We call this nucleotide stretch the "G-box".
Reporter Gene Study of the Promoter-We subcloned the putative promoter domain of the glucagon receptor gene in the pBlCat6 plasmid as illustrated in Fig. 4A. For the transfection studies, we deleted all sequences downstream from the first exon (UJ). Transfections were performed in INS-1 cells: a cell line established (16) and kindly provided by Prof. Wollheim. These cells contain the glucagon receptor mRNA and express functional glucagon receptors (18). Glucose regulated expression of the glucagon receptor in these cells was demonstrated by the adenylyl cyclase stimulation: the activity was higher (225 pmol of cAMP/min/mg of protein; 4.2-fold stimulation) using membranes from cells grown in the presence of 20 mM glucose, then with the membranes from cells grown in the presence of 5 mM glucose (100 pmol of cAMP/min/mg of protein; 2-fold stimulation). An INS-1 subclone that expresses low levels of the glucagon receptor and a RIN cell line that does not express the glucagon receptor were used as controls of transfection specificity (data not shown).
We studied the glucagon receptor promoter activity and its regulation by glucose in the INS-1 cells expressing greatly glucagon receptors. We constructed CAT plasmid variants con-taining various putative promoter sequences. The DNA fragments tested are described in Fig. 4. We observed that the transfection with the short pCB (Ϫ236 to ϩ156) plasmid gave the same, relatively large CAT activity in cells grown in medium containing either 5 or 20 mM glucose (Fig. 5).
In contrast, cells transfected by the longer plasmids pCX (Ϫ869 to ϩ156) and pCC (Ϫ2 kb to ϩ156) expressed a low CAT activity in the presence of glucose 5 mM and a spectacularly higher CAT activity in the presence of 20 mM glucose (Fig. 5). Transfection realized with plasmid pCX (Ϫ869 to ϩ156) and plasmid pCC (Ϫ2 kb to ϩ156), gave similar results, suggesting that clone pCX (Ϫ869 to ϩ156) contained the DNA fragment responsible for the glucose responsiveness. This enhancer should be located upstream from the plasmid pCB insert sequence (i.e. between position Ϫ236 to Ϫ869 upstream from the transcription start site). Results obtained with the short pCB plasmid shown that these regulatory elements inhibited gene transcription at low, 5 mM glucose and activated the transcription at high, 20 mM glucose concentrations.
The shortened plasmid pCP (Ϫ541 to ϩ156) gave a lower activity, with reduced glucose stimulation. The deleted plasmid pCD (pCX deleted between Ϫ267 and Ϫ550 nt) had a lower, nonregulated activity (Fig. 5). These results suggest that the glucose regulatory element is located close to the PmlI site and DraIII restriction sites, including nucleotides Ϫ541 to Ϫ550. This corresponds indeed to the position of the 19-nt palindromic sequence "G-box" indicated above (Fig. 3B).
We also tested the promoter activity of plasmids bearing inserts in the reversed orientation (Fig. 6). Plasmid pCXi (ϩ156 to Ϫ868) possessed a low activity that remained glucose regulated. Plasmid pCBi (ϩ156 to Ϫ236) possessed a low activity that was no regulated by glucose (Fig. 6). As the putative glucose enhancer sequence is highly palindromic, it is not surprising that glucose stimulation was also observed in reverse orientation of plasmid pCXi i . Absolute activity value of reversed plasmids was lower, which was expected, as in this orientation the normal basal promoter was also in the reverse orientation. This was confirmed by the low activity of the reversed plasmid pCBi containing only the putative basal promoter (i.e. GC-rich domain) (Fig. 6).
We removed the basal promoter (insert of the clone pCB) by BamHI digestion of the whole promoter domain (plasmid pCC). We subcloned this genomic fragment (Ϫ2 kb to Ϫ287) upstream from the HSV thymidine kinase basal promoter in the pBlCat5 plasmid. Cells transfected with this plasmid, called pTkCd, which contains the reporter gene driven by the recombinant glucagon receptor enhancer and the tk basal promoter, displayed much higher activities that pCX plasmid. The pTkCd plasmid retained the same level of sensitivity that the pCX plasmid to glucose activation (Fig. 7). Therefore, this plasmid will be used as a model for further studies of the glucose enhancer.
We realized a glucose activation dose response curve. Average results are shown in Fig. 8. Most of activity variation was observed between 5 and 10 mM glucose, with about half-maximal activity observed at an average of 7.5 mM glucose. These glucose concentrations are within physiological glycemia range. In addition, the glucose stimulation of the gene transcription suggests a threshold-like behavior. Indeed, analysis of individual results indicates that the majority of experimental points offer a low activity up to 5 mM glucose, and high activity as of 10 mM glucose. Three dose-response experiments were performed with pCX plasmid (pBlCat6 derivative) Fig. 8, and one, with similar relative results, with pTkCd plasmid (pBlCat5 derivative including HSV tk promoter).
We used the Quick Change site-directed mutagenesis kit for FIG. 2. Southern blot of anchored PCR of the 5 end the of glucagon receptor cDNA (5 RACE) from liver and heart. The liver and heart mRNA were reverse transcribed using the R426 reverse primer (complementary to nt ϩ426 to 408) of the glucagon receptor CDS. The reaction product was purified and ligated to a phosphorylated anchor primer as recommended by the manufacturer (CLONTECH 5Ј RACE kit). In the first amplification step, the ligated cDNA was amplified using the PCR anchor primer (CLONTECH) together with R356 (based on nt ϩ356 to ϩ339, samples 1-4) or R147 (based on nt ϩ147 to ϩ128, samples 5 and 6) reverse primers. The product of the first PCR was diluted 1000 times and reamplified using the same PCR anchor primer, and reverse primers R147 (lanes 1 and 2) or R-20 (based on nt Ϫ20 to Ϫ39, lanes [3][4][5][6]. The product of the second PCR was separated on 1.2% agarose gel, transferred to nitrocellulose membrane, hybridized using R-20-labeled oligonucleotide as a probe, and autoradiographed. the preparation of several mutated clones. First, we deleted the G-box from the pCX clone, yielding the pCX⌬ clone that lacks a 15-nucleotide domain that includes the two E-boxes. This deletion produced a total suppression of the glucose stimulation of the CAT activity in the transfected cells (Fig. 9A). In addition, two mutations were introduced in the G-box present in the pTkCd plasmid. In the first clone pTkCd-m1, we mutated the second E-box CAGCTG to CAGTTG. In the second mutated clone pTkCd-m2, we changed CA of the two E-boxes to TG. Transfections of INS-1 cells with the mutated plasmids showed that, whereas the C 3 T point mutation of pTkCd-m1 did not impair glucose stimulation of the reporter gene expression, the CA 3 TG mutations of pTkCd-m2 almost totally suppressed glucose activation of the reporter gene. These results demon-strate the key role of the G-box in the glucose activation of the rat glucagon receptor gene expression. DISCUSSION We have cloned the 5Ј upstream domain of the glucagon receptor gene, we demonstrated that it contains a functional glucose-regulated promoter, and we located its glucose regulatory element. This is the first time that a G protein-coupled receptor gene promoter regulated by glucose has been cloned.
Discussion of Gene Organization-The glucagon receptor belongs to the type B G protein-coupled receptor family (GPCRs; see GPCR data base (Ref. 19)). Several genes of this receptor family have been described, and share a similar coding domain, i.e. exon/intron organization pattern (14, 20 -27). This pattern FIG. 4. Construction of plasmids used for transfection studies. The ligation of a HindIII/XbaI fragment of clone P/S and a XbaI/XbaI fragment of clone B/B with the pBlCat6 vector digested by HindIII/XbaI yielded clone P/ctr/B. P/ctr/B was digested by ApaI and NotI, then blunted and ligated to obtain clone pCC. This clone contained a XhoI site 1 kb upstream from its 3Ј end (i.e. ApaI site). A second XhoI site is located in the polylinker, so that XhoI digestion excised a 1-kb insert that was subcloned in the XhoI site of the pBlCat6 vector to obtain clone pCX. DraIII cuts at position Ϫ550 and Ϫ267 nt. DraIII-digested plasmid was blunted with T4 DNA polymerase and ligated to obtain clone pCD.

FIG. 5. CAT activity of INS-1 cell extracts transfected with plasmids containing the glucagon receptor promoter.
The left panel summarizes the pBlCat6 plasmids used, with the various rat glucagon receptor gene 5Ј sequences: clone pCC (Ϫ2 kb to ϩ156); pCX (Ϫ869 to ϩ156) and the short clone pCB (Ϫ236 to ϩ156) were constructed as indicated in Fig. 4. The pCP clone (Ϫ541 to ϩ156) was pCX shortened with EcoRV (cutting in polylinker) and PmlI (cutting in the first E-box, see Fig. 3B). The pCD plasmid (Ϫ869 to Ϫ550, Ϫ267 to ϩ156) was prepared by internal deletion (between Ϫ267 and Ϫ550 nt) of pCX with the DraIII enzyme. The right panel shows the results of transfection experiments. The CAT activity of transfected cells was assayed by acylation of [ 14 C]chloramphenicol by n-butyryl-CoA, after cell culture in media containing 5 mM (hatched bars) or 20 mM (open bars) glucose. The spot densities were calculated in each experiment as percentage of activity produced by pCX plasmid. The mean of three to five separate experiments performed in duplicates Ϯ S.E. is represented.
is distinct from the exon/intron organization of the other GPCR families.
Two types of 5Ј untranslated domains, with or without introns, have been described in this B type receptor family. In other words, the gene promoter could be located either immediately upstream of the ATG initiation site, or at a more distal position, such as the promoter of the porcine calcitonin receptor gene located 20 kb from the ATG initiation site and separated from this site by two 5Ј untranslated exons and two introns (21). The rat PTH receptor possesses two promoters, located upstream from the first or the third untranslated exon respectively, with different tissue specificities (22,23). An alternative splicing of at least four untranslated exons was described for the human PACAP receptor (24).
In contrast, the gene promoter is close to the ATG initiation site in both the human and the rat VPAC1 receptor genes (formerly VIP 1 receptor) (25,26), as well as in the rat GLP-1 receptor gene (27).
Conflicting data have been published for the 5Ј end of glucagon receptor gene: Buggy et al. reported the sequence of a putative human glucagon receptor promoter, separated from the initiation codon by an undescribed 5-kb intron (28). In contrast, Burcelin et al. (17) reported the sequence of a putative mouse glucagon receptor promoter adjacent to the initiation codon. However, no experimental data demonstrating the promoter activity were provided. Our results suggest that sequences upstream from the ATG of the mouse glucagon receptor belong to an intron (85% homology with our rat intronic sequence). A comparison of our experimentally active rat promoter with the sequence of the putative human glucagon receptor promoter indicates that the best significant homology (65% over 120 nt) is located in the vicinity of the mRNA start point. The size of the human 5Ј untranslated domain, estimated by primer extension assay, was larger (475 nt) than in rat (376 nt). The complete 5Ј sequence of human gene has not yet been reported (28), but comparison of the rat and the human glucagon receptor cDNAs shows a high (82%) homology in the coding domain (29,30).
Herein we provide evidence for a polymorphism of the 5Ј untranslated domain of the rat glucagon receptor mRNA. Indeed, two introns of 0.6 and 3.2 kb place the transcription start site (of exon UJ) at more than 4 kb from the ATG initiation codon (Fig. 3B). In the 5Ј untranslated domain, a 166-bp exon (US) was present between the two introns. This exon seems to FIG. 6. CAT activity of INS-1 cell extract transfected with plasmids containing the glucagon receptor promoter in both orientations. Left panel, schematic representation of the pBlCat6 plasmids with inserts in the normal orientation (plasmids pCX (Ϫ869 to ϩ156) and pCB (Ϫ236 to ϩ156)), and in the reverse orientation (plasmid pCX i (ϩ156 to Ϫ868) and pCB i (ϩ156 to Ϫ236)). Right panel, results of transfection experiments performed as in Fig. 5  be spliced in the mature mRNA from liver and less frequently in the heart, but this requires further investigation.
The mRNA 5Ј end polymorphism could be explained a priori either by the existence of two alternative promoters or by an alternative splicing. Two alternative promoters have indeed been described for the PTH receptor (23) as well as for several genes involved in the glucose metabolism (31,32). Our experimental results show that the majority of the mRNA molecules containing the US exon also contains the UJ exon, suggesting a unique transcriptional initiation domain. Nevertheless, we cannot rule out the existence of an alternative promoter, active in other tissues expressing the glucagon receptor, or active only during specific life periods.
The size of the products synthesized by anchored PCR at the 5Ј end of the rat cDNA (Fig. 2) yielded an average size of 180-nt untranslated mRNA (without exon US) and of 370 nt (with exon US) (Fig. 2). This size is comparable to that obtained by sequencing, supporting that most transcription start at either 210 or 376 nt (without or with exon US) upstream from the ATG. However, other (downstream) putative start positions are also observed. This situation is similar for the PTH receptor (22) or GLP-1 receptor (27) genes, in which multiple start points are described. A cDNA with multiple start points is generally observed for genes without TATA boxes but with GC-rich domain in the basal promoter, which is the case for all genes described to date for this receptor family.
However, we cannot rule out a gene organization similar to the PTH receptor gene; this would imply existence of a third and very short third 5Ј untranslated exon together with the existence of two basal promoters, located, respectively, upstream from the first and second exon (23). Indeed, the sequence TCTGGAAAGTTTGCAGG, located 9 nucleotides downstream of the 5Ј-most mRNA experimental start point of, is, according to the GeneID program (Ref. 33, geneind@darwin. bu.edu), a potential acceptor splicing site. Thus, few nucleotides upstream may represent the third 5Јexon, present on a still unsequenced genomic fragment. In this hypothesis, the genomic sequence upstream from the start point would have a basal promoter activity (becoming a second basal promoter as for the PTH receptor gene; Ref. 23), and the G-box would be located on the first intron. Such a location of a glucose regulatory element in the first intron has been described for the fatty acid synthase gene (see below).
Discussion of Glucose Regulation-Several gene promoters are up-regulated by increasing of extracellular glucose concentration. Glucose activation of two genes expressed in the liver, pyruvate kinase L (PKL) and Spot14 (a lipogenesis-associated protein) have been extensively studied. It appears that glucose stimulates the transcription of these two genes via similar motifs present in the promoter. These carbohydrate response elements were called GIRE (34) or ChoRE (35) for the PKL or Spot14 genes, respectively. The core motif in these elements is the so called E-box, constituted by the palindromic sequence CACGTG, the consensus recognition sequence of transcription factors belonging to the c-Myc family. Glucose regulation of PKL gene uses two imperfect E-boxes, separated by 5 nt, forming a perfect palindrome coined L4 box: CACGGGn 5 CCCGTG (36). A similar feature is required for glucose activation of the Spot14 gene: CACGTGn 5 CCCGTG (37). The spacing distance of 5 nt between these two E-boxes was shown to be essential for its activity (37). It is noteworthy that the L4 box found in the pyruvate kinase L gene is active in INS-1 cell line, whereas neither the proinsulin I nor the glucokinase mRNA levels are increased under the same conditions (38).
These glucose regulatory elements have different locations in the different promoters: located only at Ϫ168 to Ϫ144 nt from the start point of the pyruvate kinase L gene, whereas in the Spot14 gene it is found at nt Ϫ1448 to Ϫ1431. The gene of fatty acid synthase, another glucose-regulated gene, also possesses a sequence feature closely related to L4 box. However, this glucose regulatory element is located in the first intron (ϩ283 to ϩ303) (39).
We observed that the sequence motif tgCACGTGtgaCAGCT-Gca in the promoter domain of the glucagon receptor located at Ϫ545 to Ϫ527 is very similar to the motif of the regulatory elements described above. This highly palindromic (underlined), 19-nucleotide sequence contained two perfect, 6-nucleotide palindromes (uppercase). The first one represents a canonical E-box sequence. Results observed with different constructs obtained by with restriction digestion (Fig. 5) suggested that the glucose regulation is centered on this domain. We confirmed the essential role of this motif, the coined G-box, by the mutational studies (Fig. 9), where the deletion or mutation of this motif suppressed glucose stimulation.
The E-boxes found in our G-box motif are more palindromic than in the L4 box and in the L4-like glucose regulatory elements described previously. The palindromic nature of the E-boxes seems to be important for the glucose regulation; mutations that increase the palindromic nature of the second E-box of the L4 box in the Spot14 promoter strongly increase its glucose-stimulated enhancer activity (37).
As this enhancer is palindromic, it is not surprising that the glucose stimulation was also observed in the reverse orientation (Fig. 6). Absolute activity of the reversed plasmids is lower, which is expected, as in this orientation the normal basal promoter is also in the reverse orientation. The transcription start sites were not investigated for reverse orientation plasmids. Consequently, we did not determine which domain acts as a basal promoter in these plasmids. However, a GC-rich domain may act as a promoter in the reverse orientation (40), and a similar glucose regulatory element may act as enhancer downstream from the transcription start site (39).
Recently, USF-2 (upstream stimulatory factor 2) was shown to be directly involved in the glucose-induced stimulation of PKL and Spot14 gene transcription (41,42). USF-2 is a mem- ber of the c-Myc family of transcription factors characterized by helix-loop-helix/leucine zipper domains. However, another transcription factor may act on the glucagon receptor G-box including a 3-nt spacing; indeed, the 5 nt inserted between the two E-boxes are essential for the previously described L4 box activation (37).
It was clearly demonstrated that, in a first time, glucose has to be phosphorylated to glucose 6-phosphate before the stimulation of the described genes occurs (38,43). According to Doiron et al. (44), an intermediate of the nonoxidative branch of pentose phosphate pathway, the xylulose 5-phosphate, mediates the glucose signal to the transcriptional machinery. An involvement of a 5Ј-AMP-activated protein kinase has recently been described (45).
According to Burcelin et al. (10), the glucose stimulation of the hepatic glucagon receptor mRNA level, involves triose phosphate intermediates of the glucose metabolism. This again suggests that the L4-box and the G-box do not interact with the same factors.
The aldolase A, pyruvate kinase M2, and probably Glut2 genes are regulated by glucose via dephosphorylation of Sp1 (46). Burcelin et al. (10) recently demonstrated that, in mouse, the glucose-induced up-regulation of the Glut2 and glucagon receptor mRNA levels involve different mechanisms, supporting our data.
The present results have established for the first time the localization and the sequence of a highly active, glucose-regulated glucagon receptor promoter, involving a novel L4-like box. A better understanding of the regulation will require the study of the protein elements involved in the glucose regulation of the glucagon receptor, as well the study of the metabolites of glucose that are involved within the regulatory pathway. (Ϫ869 to ϩ156) plasmid, or the G-box deleted plasmid pCX⌬ prepared by QuickChange site-directed mutagenesis kit (Stratagene). Transfected cells were grown in either 5 mM glucose or 20 mM glucose, and the CAT activity of cells extracts was assayed as described under "Material and Methods." The mean Ϯ S.E. of three (for pCX⌬) or five (for pCX) independent experiments, performed in duplicate, is represented. Panel B, INS-1 cells were transfected by pTkCd plasmid (pBlCat5, tk basal promoter with fragment Ϫ2 kb to Ϫ287) and two mutants of the G-box prepared by QuickChange sitedirected mutagenesis kit (Stratagene) as described under "Materials and Methods." Mutations are indicated in bold and underlined. Transfected cells were grown in either 5 mM glucose or 20 mM glucose, and the CAT activity of cell extracts was assayed as described under "Material and Methods." The mean of two independent experiments, performed in duplicate, is represented.