Transcriptional activation of the proglucagon gene by lithium and beta-catenin in intestinal endocrine L cells.

The proglucagon gene encodes several peptide hormones that regulate blood glucose homeostasis, growth of the small intestine, and satiety. Among them, glucagon-like peptide 1 (GLP-1) lowers blood glucose levels in patients with diabetes and inhibits eating and drinking in fasted rats. Although proglucagon transcription and GLP-1 synthesis were shown to be activated by forskolin and other protein kinase A (PKA) activators, deleting or mutating the cAMP-response element (CRE) only moderately attenuates the proglucagon gene promoter in response to PKA activation. Therefore, PKA may activate proglucagon transcription via a mechanism independent of the CRE motif. Recently, PKA was shown to phosphorylate and inactivate GSK-3beta, a key mediator in the Wnt signaling pathway. We show here that lithium, an inhibitor of GSK-3beta, activates proglucagon gene transcription and stimulates GLP-1 synthesis in an intestinal endocrine L cell line, GLUTag. The activation was also observed in primary fetal rat intestinal cell (FRIC) cultures, but not in a pancreatic A cell line. Co-transfection of beta-catenin, a downstream effector of GSK-3beta activities, activated the proglucagon gene promoter without a CRE. Furthermore, forskolin and 8-Br-cAMP phosphorylated GSK-3beta at serine 9 in intestinal proglucagon-producing cells, and both lithium and forskolin induced the accumulation of free beta-catenin in these cell lines. These observations indicate that the proglucagon gene is among the targets of the Wnt signaling pathway.

(GLP-1) 1 and glucagon-like peptide 2 (GLP-2), however, are produced in the endocrine L cells and in selected endocrine neurons in the brain (1,2). GLP-1 is known as an insulinotropic hormone. It possesses potent effects to stimulate glucose-dependent insulin secretion, insulin gene expression, and B cell cAMP formation. Other effects of GLP-1 include inhibition of glucagon release and gastric emptying and, possibly, enhancement of peripheral insulin sensitivity (1,2). In addition, intracerebroventricular administration of GLP-1 was found to powerfully inhibit eating and drinking and alter body weight in fasted rats (3,4). Furthermore, a recent study shows that peripheral GLP-1 also plays a role in regulating macronutrient selection and food intake (5). GLP-2 was initially identified as a growth factor for the small intestine (6). Recent observations indicate that GLP-2 and its receptor also possess an overlapping function with GLP-1 in regulating gastric emptying and controlling satiety (7,8).
Numerous studies conducted in the past 15 years have identified more than a dozen transcription factors and signaling molecules that play roles in driving proglucagon gene expression (9 -31). However, none of the identified transcription factors or signaling molecules has been found to be tissue type-or cell type-specific.
Protein kinase A (PKA) is able to stimulate proglucagon gene transcription and glucagon or GLP-1 synthesis in both primary and transformed intestinal endocrine cells as well as in rat primary pancreatic islet cultures (11)(12)(13)(14)17). However, deleting or mutating the cAMP-response element (CRE) in the proglucagon gene promoter only partially attenuates its response to PKA activation in the small intestinal proglucagon-producing cell line STC-1 (13,17). This indicates that PKA may up-regulate proglucagon gene expression via a yet to be identified signaling pathway. Recently, PKA was found to phosphorylate and inactivate the serine/threonine kinase glycogen synthase kinase 3␤ (GSK-3␤) in several cell lineages (32,33). These observations prompted us to ask whether this newly identified function of PKA is engaged in proglucagon gene transcription. GSK-3␤ is a major mediator of the Wnt signaling pathway (34 -36). In normal epithelial cells, adenomatous polyposis coli together with GSK-3␤ and Axin bind to and phosphorylate ␤-catenin (␤-cat), targeting ␤-cat for proteasomal mediated degradation. In embryonic cells, Wnt signals inactivate GSK-3␤, resulting in free ␤-cat accumulation. Free ␤-cat then forms a bipartite transcription factor with a T cell factor (TCF)/lymphoid enhancer factor (LEF), namely cat/TCF, activating the Wnt responsive or cat/TCF target genes (36). Another notable inhibitor of GSK-3␤ is lithium, which mimics the function of the Wnt signals in embryonic cells (37,38).
In this study, using cultivated intestinal endocrine cell lines as well as primary intestinal cells in culture, we examined whether Wnt signaling pathway/molecules mediate the effect of PKA in activating proglucagon gene transcription and GLP-1 synthesis.
Plasmids-Construction of the wild type and mutant rat proglucagon/luciferase (LUC) reporter gene plasmids have been described previously (15,17,39). ␤-cat expression plasmids were provided by Dr. Eric Fearon (40). The TCF-TK-LUC reporter gene construct (TOPFLASH) was provided by Dr. Burt Vogelstein (41). The G2-TK-LUC fusion gene was generated by inserting one copy of the rat proglucagon gene G2 enhancer-like element (9) into the TK-LUC plasmid. The DNA sequence for the top strain of G2 is AGGCACAAGAGTAAATAAAAAGTTTC-CGGGCCTCTG (9). It contains a potential binding site (AAGTTTC) for the bipartite transcription factor cat/TCF (35).
Cell Culture, Transfection, and LUC Reporter Gene Analysis-The intestinal GLUTag, STC-1, and the pancreatic InR1-G9 cell lines were grown and maintained in Dulbecco's modified Eagle's medium supplemented with appropriate serum (39). To examine the effects of lithium and forskolin on proglucagon mRNA expression, GLP-1 synthesis, and secretion, cells were grown in the medium containing the appropriate serum overnight. 6 h prior to the experiment, serum-containing medium was withdrawn, and serum-free medium was added. For the LUC reporter gene analysis, the intestinal endocrine cell lines GLUTag and STC-1 were transfected using LipofectAMINE (Invitrogen) per the manufacturer's instructions, whereas the InR1-G9 cell line was transfected by a method of calcium precipitation (42).
Fetal rat intestinal cell (FRIC) cultures were prepared using 19 -21 days-of-gestation fetal Wistar rats (Charles River Canada, Saint Constant, Quebec, Canada), as described in detail previously (11,31). In FRIC cultures, the endocrine L cells account for ϳ1% of the cell numbers.
RNA Extraction and Northern Blot Analysis-The methods used for RNA extraction and Northern blot analysis were described previously (39).
Cell Fractionation and Immunoblotting-The anti ␤-catenin antibody was purchased from BD Biosciences, and the anti-phospho-GSK-3␣ and -3␤ antibodies were from New England Biolabs (32). Cytoplasmic and membrane fractions were prepared based the protocol by Shimizu et al. (43). The methods used for immunoblotting and for examining the PKA phosphorylation of GSK-3 were described previously (32).

RESULTS
Lithium Activates Proglucagon mRNA Expression-Lithium is a notable inhibitor of GSK-3␤. We hypothesized that if PKA activates proglucagon gene transcription via inactivating GSK-3␤, lithium may also activate proglucagon mRNA expression. We examined this hypothesis in the mouse large intestinal GLUTag (13,(45)(46)(47) and the mouse small intestinal STC-1 (14) cell lines, as well as in the primary (FRIC) rat intestinal cultures. After incubation with 10 mM LiCl (48) for 4 h, both cell lines demonstrated ϳ3-fold increased proglucagon mRNA ex-pression ( Fig. 1). In the GLUTag cells, the activation was still observable after 12 h (3.9-fold). After a 24-hour incubation in two experiments, we still observed an enhanced proglucagon mRNA expression. Results from one experiment are presented in Fig. 1A. In two other experiments, however, proglucagon mRNA expression returned to untreated levels by 24 h (data not shown). In the STC-1 cell line, activated proglucagon gene expression was substantial during the whole experimental procedure (Fig. 1B). Fig. 1A also shows that proglucagon mRNA expression in the GLUTag cells is activated by the PKA acti- Total RNA was extracted for Northern blot analyses using cDNA probes for rat proglucagon (Glu) or tubulin (T). 10 g of RNA was loaded for each sample. For the cultivated cell lines, the membrane was exposed to x-ray film 8 -12 h. For the FRIC cultures, the membrane was exposed for 8 days. After the densitometric analyses, the effects of each chemical on proglucagon expression were calculated as the -fold change versus the untreated cells normalized against the tubulin mRNA.
Lithium Activates the Rat Proglucagon Gene Promoter-The effect of lithium on proglucagon mRNA levels could be a result of enhanced transcription or reduced degradation or both. We therefore examined whether lithium activates the proglucagon gene promoter. Fig. 2A shows that, after being transfected into the GLUTag cells, the expression of the Ϫ1.1-kb and the Ϫ472-bp promoter constructs was activated ϳ5-fold by forskolin. The evolutionarily conserved CRE is located between Ϫ291 and Ϫ298 bp (9). When the wild type Ϫ302-bp promoter was examined, ϳ6-fold activation by forskolin was observed. However, even with the CRE-mutated [Ϫ302(M)] or deleted (Ϫ290) reporter gene constructs, forskolin still generated ϳ3-3.5-fold activation. This is consistent with previous studies on the STC-1 cell line (13,17). LiCl treatment also activated all the five promoter constructs examined by 2-3-fold, including those that carry a mutated [Ϫ302(M)] or deleted (Ϫ290) CRE. Lithium was also found to activate the proglucagon gene promoter constructs when transfected into the STC-1 cell line (Fig. 2B).
The stimulatory effect of LiCl on the proglucagon promoter was found to be dose-dependent (Fig. 2B) and was observable within 4 to 12 h after the treatment (data not shown).
To determine whether lithium also activates proglucagon gene expression in the pancreatic A cells, we conducted the above analyses against the hamster pancreatic A cell line InR1-G9. No appreciable activation on either proglucagon gene promoter or endogenous proglucagon mRNA expression was observed by LiCl treatment in this cell line (data not shown).
Lithium Stimulates GLP-1 Synthesis-We next examined the effect of lithium on GLP-1 synthesis and secretion in the GLUTag cell line and FRIC cultures. We demonstrated previously that GLP-1 synthesis and secretion in those cells could be activated by forskolin or other PKA activators (11, 14, 44 -47). Incubating the GLUTag cells with 10 mM LiCl for 4 h did not affect either the synthesis or secretion of GLP-1 (data not shown). When the incubation time was extended to 8 h, a 1.7-fold increase in the GLP-1 content of cell was observed (p Ͻ 0.01, Fig. 3A). In the same period, no change in GLP-1 secretion was observed as determined by an RIA of the cell free medium (Fig. 3B). Similarly, FRIC cultures were treated without or ␤ϪCat Transfection Activates the Proglucagon Gene Promoter without a Functional CRE-Lithium inactivates GSK-3␤ and induces free ␤-cat accumulation in other cell lineages (48). If lithium indeed stimulates proglucagon transcription via inactivation of GSK-3␤, overexpressing ␤-cat should mimic the effect of lithium treatment. A reporter gene construct, namely TOPFLASH, has been widely utilized to examine the cat-TCF activity in colon cancer and other cell lines (41, 48 -50). In this plasmid, the expression of a LUC reporter gene is driven by a minimum TK promoter fused with three copies of the TCF binding site. As shown in Fig. 5, this fusion gene is dose dependently activated by LiCl but not by KCl when transfected into the GLUTag cells.
We then directly examined the effect of the wild type and the constitutively active mutant ␤-cat, S33Y, on the expression of the proglucagon gene promoter. The S33Y mutant ␤-cat is resistant to degradation because it cannot be phosphorylated by GSK-3␤ (41). Co-transfection with the wild type ␤-cat cDNA stimulated all of the proglucagon gene promoter constructs examined, except for the Ϫ1.1-kb promoter construct, by 1.7-2.0-fold (Fig. 6). Compared with the wild type ␤-cat, the S33Y mutant ␤-cat activated all five constructs more effectively, varying from 2.5-to 4-fold (Fig. 6). In addition, the S33Y mutant ␤-cat was able to activate the Ϫ1.1 kb proglucagon gene promoter construct. The activation was also independent of the CRE motif on the proglucagon gene promoter (Fig. 6).
The consensus binding site for cat/TCF has been suggested to be WWGTTTC (35). We localized several potential cat/TCF binding sites on the proglucagon gene promoter. One such binding site is downstream of Ϫ290 bp and is a part of the G2 enhancer-like element (9). This site is conserved among the proglucagon genes of humans and rodents (9,22). We inserted one copy of the G2 element in front of the TK promoter in the TK-LUC fusion gene plasmid, and the new fusion gene was named G2-TK-LUC (Fig. 7). LiCl was found to dose dependently activate G2-TK-LUC by up to 3-fold when transfected into the GLUTag (Fig. 7) and STC-1 (data not shown) cells. Furthermore, G2-TK-LUC was activated by forskolin by ϳ5fold (Fig. 7).
PKA Phosphorylates GSK-3␤, and Both LiCl and PKA Activators Induce Free ␤-Cat Accumulation-Recent studies have shown that PKA phosphorylates and inactivates GSK-3␤ in the HEK293 and NIH 3T3 cell lines (32). In addition, PKA was also found to phosphorylate and inactivate GSK-3␤ in neuronal cells, and GSK-3␤ inactivation is linked to the inhibition of neuronal cell apoptosis (33). To further examine our hypothesis that Wnt pathway mediates the activation of proglucagon gene transcription by PKA, we examined GSK-3 phosphorylation and inactivation by PKA in the intestinal endocrine cell lines. As shown in Fig. 8, treating the GLUTag cells with forskolin or 8-Br-cAMP induced the phosphorylation of GSK-3␣ at serine 21 and GSK-3␤ at serine 9. In contrast, treating the GLUTag cells with 8-Br-cGMP did not change the phosphorylation status of GSK-3␣ or GSK-3␤ (Fig. 8).
We then asked whether lithium treatment or PKA activation would lead to free ␤-cat accumulation in the two intestinal proglucagon-producing cell lines. The amount of ␤-cat in cell membrane and cell cytosol was examined by Western blot analysis following a cellular fractionation procedure (43). In four separate experiments we observed free ␤-cat accumulation in response to either LiCl or forskolin treatment, with activation levels varying from 2.5 to 10-fold. Data from one experiment are shown in Fig. 9. In this experiment, incubating the GLUTag cells with 10 mM LiCl for 2 h led to an ϳ3-fold increase in cytosolic ␤-cat, whereas the change in ␤-cat in the membrane portion was minimal. Similarly, treating the GLUTag cells with forskolin for 2 h led to a 2.9-fold increase in cytosolic ␤-cat accumulation, whereas the change of ␤-cat in the membrane portion was minimal (Fig. 9A). We also found that treatment of the GLUTag cells with 10 nM insulin for 12 h generated an appreciable effect on free ␤-cat accumulation (2.9-fold), whereas the effect at 2 or 4 h was minimal (Fig. 9A). Similar results regarding free ␤-cat accumulation in response to these three reagents were obtained in examining the STC-1 cell line (Fig. 9B). DISCUSSION In mammals, a single proglucagon gene encodes three major peptide hormones expressed in three different tissues (1). These hormones play critical roles in blood glucose homeostasis, the growth of small intestines, and satiety (1)(2)(3)(4)(5)(6)(7)(8). Uniquely, different hormones encoded by the same proglucagon gene may possess opposite roles, such as glucagon versus GLP-1 in blood glucose homeostasis, or overlapping roles, such as GLP-1 and GLP-2 in mediating satiety (7,8). Therefore, it is desirable and interesting to explore the molecular mechanisms that underlie proglucagon gene transcription and the biosynthesis of each individual hormone in a cell-specific manner. Although previous studies have identified numerous transcription factors and signals that may up-or down-regulate proglucagon gene transcription, none of the identified factors specifically regulate proglucagon transcription in pancreatic A cells only or in intestinal endocrine cells only (9 -26, 51-53). Our observations indicated that lithium (as an inhibitory agent of GSK-3␤) and cat/TCF selectively up-regulates proglucagon gene transcription and GLP-1 synthesis in the endocrine cells of the gut but not in a pancreatic A cell line. It will be interesting to explore the molecular mechanisms underlying this cell type-specific transactivation event.
A typical CRE element is located on both human and rat proglucagon gene promoters (9,17,22). Previous studies in the mouse small intestinal STC-1 cell line have shown that deleting or mutating this CRE motif generates only partial attenuation in response to PKA activation (13,17). We show here that this is also true for the GLUTag cell line. These observations further support our proposal that PKA stimulates proglucagon gene transcription not only via phosphorylation of the CREbinding proteins (CREB) but also through pathways that crosstalk with PKA.
The cross-talk between the Wnt signaling pathway and PKA or G protein-coupled receptors has been realized very recently. Meigs et al. (54) have shown that constitutively activated G␣ 12 and G␣ 13 interact with E-cadherin to cause the release of ␤-cat and the subsequent stimulation of cat/TCF-mediated transcription. The chimeric receptor of the ligand binding and transmembrane domains of the ␤2 adrenergic receptor and the cytoplasmic domains of Frizzled-1 can also stimulate the cat/TCF transcriptional activation through a mechanism that appears to involve signaling through G␣ q and/or G␣ o (55). Similarly, Fujino et al. (56) reported the phosphorylation of GSK-3␤ and stimulated TCF-mediated transcription by prostaglandin E 2 through the EP2 or EP4 receptor, likely via PKA and phosphatidylinositol 3-kinase, respectively. Furthermore, treating fibroblasts and neuronal cells with forskolin or 8-Br-cAMP may lead to enhanced phosphorylation and inactivation of GSK-3␤ (32,33). Very recently, Yusta et al. (57) reported that GLP-2 inhibits cell apoptosis via its G protein-coupled receptor in association with PKA-dependent inactivation of GSK-3␤. Here we extend these observations into the intestinal proglucagonproducing cell (Fig. 8). This, in combination with our observation that both lithium and forskolin activated the G2-TK fusion promoter (Fig. 7), supports our notion that PKA activates proglucagon gene transcription via cross-talk with the Wnt pathway. Previous studies have implicated G2 in regulating proglucagon gene expression via binding with members of the Foxa transcription factor family (20,(51)(52)(53). In addition, by examining the effect of membrane depolarization on proglucagon gene expression in pancreatic A cell lines, Furstenau et al. (23) identified a calcium-response element within this enhancer-like element (23). Further studies are needed to examine the binding of cat/TCF to G2 and determine how G2 is implicated in specifying cell type-specific regulation of proglucagon gene transcription in response to lithium and other signaling molecules.
Although phosphorylation and inactivation of GSK-3␤ by PKA, PKB/Akt, and PKC have been demonstrated in several cell lineages (32,33,(57)(58)(59)(60)(61)(62), whether this inactivation leads to free ␤-cat accumulation has not been previously reported. Ding et al. (58) have shown that a 2-hour insulin treatment, possibly through PKB activation, leads to phosphorylation of GSK-3␤ at FIG. 10. A diagram showing mechanisms for the regulation of proglucagon gene transcription by lithium and PKA. Lithium is able to inactivate GSK-3␤ and therefore induce the accumulation of free ␤-catenin, which will form a complex with a given TCF. The bipartite transcription factor cat-TCF then stimulates proglucagon mRNA transcription, possibly through interacting with the G2 enhancer like element (9,23). PKA is able to activate proglucagon gene transcription via the evolutionarily conserved CRE element on the proglucagon gene promoters (9,(11)(12)(13)(14)17). In addition, PKA may also activate proglucagon gene transcription by inactivating the GSK-3␤ and inducing the accumulation of free ␤-catenin. It is recognized however, that GSK-3␤ inactivation and free ␤-cat accumulation in response to PKA could be two separate events. Glu, proglucagon gene. A dotted line shows the CRE pathway documented previously (11)(12)(13)(14)17).

FIG. 9. Both LiCl and forskolin stimulate free ␤-cat accumulation in intestinal endocrine cell lines.
GLUTag (A) and STC-1 (B) cells were grown in the absence or presence of 10 nM insulin (Ins), 10 mM LiCl, or 10 M forskolin (For) plus 10 M IBMX for the indicated periods of time. Cytosolic and membrane ␤-cat were examined by Western blot analysis as described under "Experimental Procedures." The same membranes were stripped followed by hybridization with an anti-␤-actin antibody (loading control). After the densitometric analyses of the photograms, the effect of each chemical on ␤-cat appearance in both cytosol and membrane fractions were calculated as the -fold change versus the untreated cells normalized against ␤-actin. serine 9 and inactivation of GSK-3␤ enzymatic activity in a number of epithelial cell lines. However, free ␤-cat levels were not altered in their study. In contrast, both Wnt and lithium induced free ␤-cat accumulation but did not affect the phosphorylation status of GSK-3␤ (58). Based on these observations, Ding et al. (58) hypothesized that insulin and the Wnt signals regulate GSK-3␤ through different mechanisms and therefore lead to a distinct downstream event and that the phosphorylation of GSK-3␤ at serine 9 may not be sufficient to induce free ␤-cat accumulation. In our study, we also found that treating the intestinal endocrine cell lines with 10 nM insulin for 2 h generated no effect on free ␤-cat accumulation. However, the effect began to be observable after 4 h and was enhanced after 12 h (Fig. 9). More importantly, we demonstrated that, in the intestinal proglucagon producing cells, forskolin treatment led to enhanced free ␤-cat accumulation within 2 h. We speculate that, in these particular endocrine cell lines, phosphorylation of GSK-3␤ at serine 9 by PKA may be sufficient to inactivate its ability to facilitate ␤-cat degradation. However, at this stage, we cannot eliminate the possibility that phosphorylation of GSK-3␤ by PKA and the accumulation of free ␤-cat in response to PKA are two independent events. Nevertheless, our results clearly indicate that PKA activates proglucagon gene transcription in the endocrine L cells at least in part through accumulation of free ␤-cat. A model illustrating our current understanding of the mechanisms underlying proglucagon gene activation by PKA and lithium is shown in Fig. 10.
Our results also provide a potential molecular mechanism for the similar effect of lithium and GLP-1 to suppress food and water intake (3,4,(63)(64)(65)(66)(67)(68)(69)(70). Scientists have observed for a number of years that peripheral administration of lithium in rats causes a spectrum of effects, including reduced food/water intake, decreased salt ingestion after sodium depletion, and induced robust conditioned taste aversions (63)(64)(65)(66)(67)(68)(69). These effects can be mimicked by central (intracerebroventricular) administration of GLP-1 (3). More importantly, the effects provoked by lithium can be blocked by pre-treating the animals with GLP-1 receptor antagonists (67,68). Taken together, it is reasonable to speculate that lithium mediates satiety, at least in part, through up-regulation of the brain GLP-1 pathway. Unfortunately, there is currently no brain proglucagon-producing cell line to examine this hypothesis. It will therefore be necessary to develop other methodologies to examine whether brain proglucagon mRNA transcription and GLP-1 synthesis are modulated by lithium and other signaling molecules in the Wnt signaling pathway.
We have reported previously that PKA activators regulate both the synthesis and secretion of GLP-1 (11)(12)(13)(14). Other molecules, such as PKC activators, may stimulate GLP-1 secretion but not its synthesis (14). In this study, although we obtained evidence that lithium significantly enhanced GLP-1 synthesis, it had no significant effect on GLP-1 secretion. One may speculate that in the in vivo setting, Wnt signaling molecules may cross-talk with other pathways that are implicated in GLP-1 secretion. Alternatively, lithium may simultaneously activate both the Wnt signaling pathway for GLP-1 synthesis and an as yet to be determined pathway that is engaged in stimulating GLP-1 secretion. Furthermore, lithium may alter the response of the intestinal L cells to known secretagogues, including nutrients, as well as several neuro/endocrine gut hormones (31). We have recently observed a similar synergistic effect of fatty acids on glucose-dependent insulinotropic peptide (71) and of leptin on gastrin-releasing peptide-stimulated GLP-1 secretion (72). These hypotheses deserve further examination in vivo.
In summary, this study indicated that proglucagon is among the downstream target genes of cat-TCF or the Wnt signaling pathway. Our results provide a novel mechanism by which PKA up-regulates proglucagon gene transcription in the intestinal endocrine L cells and a potential explanation as to how lithium controls food/water intake. Additional studies are required to examine proglucagon gene transcription and GLP-1 synthesis in the brain in response to lithium and other Wnt signaling molecules.