Increasing GLP-1–Induced β-Cell Proliferation by Silencing the Negative Regulators of Signaling cAMP Response Element Modulator-α and DUSP14

βTC-Tet cells were cultured as described previously (22). Growth arrest was induced by addition of 1 μg/ml tetracycline for 5–14 days. MIN6 cells (gift of Pr. Romano Regazzi, University of Lausanne, Lausanne, Switzerland) were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 25 mmol/l glucose supplemented with 15% fetal bovine serum (FBS) and 5 μmol/l 2-mercaptoethanol.

For microarray experiments, growth-arrested βTC-Tet cells were incubated for 2 h in medium containing 1% FBS. After 2 h, medium was changed and the incubations continued for 0 min, 45 min, 3 h, and 7 h in the presence of 100 nmol/l GLP-1 (7-36)-amide (Bachem, Bubendorf, Switzerland).

For analysis of signaling pathways, growth-arrested βTC-Tet cells were incubated for 2 h in medium containing 1% FBS, medium was then changed and the incubations continued with combinations of 100 nmol/l GLP-1 (7-36)-amide, 10 μmol/l H89, 50 μmol/l PD98059, 50 nmol/l wortmannin, 10 μmol/l forskolin, 50 μmol/l Epac activator 8-(4-chlorophenylthio)-2′-O-methyl-cAMP, 1 μmol/l nimodipine, or 5 μmol/l S(−)-Bay K8644. H89, wortmannin, PD98059, and nimodipine were added to the cells 30 min before GLP-1.

Total RNA was isolated using the RNeasy kit (Qiagen), and 1-μg aliquots of total RNA were amplified with the MessageAmp aRNA kit (Ambion). RNA integrity and concentration were assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies).

For microarray experiments, cDNAs were synthesized by reverse transcription of 5 μg amplified RNA and labeled by an indirect procedure with Cy5 (GLP-1–treated samples) or Cy3 (control samples) as described previously (23). The labeled probes were hybridized to cDNA microarrays (mouse 17K cDNA microarrays from the DNA Array Facility of the University of Lausanne, Lausanne, Switzerland; http://www.unil.ch/dafl). Scanning, image analysis, data transformation into log₂ intensity ratio (Cy5 to Cy3), and normalization with a print tip locally weighted linear regression (Lowess) method were performed as described previously (23). A total of 9,200 spots passed the quality control filtering. Gene selection was performed with univariate F tests. P values were adjusted for multiple testing using the false discovery rate method proposed by Benjamini and Hochberg (24). Using a P value of 0.05 resulted in the identification of 341 regulated genes, which were clustered into five groups using partitioning around medoids as clustering method with correlation distance. The whole microarray analysis was performed with the R software for statistical computing available at the Comprehensive R Archive Network (25).

Pancreatic islets were isolated from male Wistar rats (Janvier, Le Genest-St-Isle, France) and trypsinized, and β-cells were purified to >95% homogeneity by fluorescence-activated cell sorter (FACS) (FACStar-Plus; Becton Dickinson, Sunnyvale, CA) as described previously (26,27). β-Cells were seeded at a density of 3 × 10⁵ cells/10 ml DMEM containing 11.2 mmol/l glucose, 10% FBS, and 110 μg/ml sodium pyruvate. Cells were incubated in suspension overnight and then resuspended at a density of 4–5 × 10⁵ cells/ml, and aliquots of 50 μl were seeded as droplets on dishes coated with 804G-extracellular matrix (26) and further incubated for 48 h. Cells were then incubated for 2 h in medium containing 1% FBS and then exposed or not to 100 nmol/l GLP-1 (7-36)-amide for 45 min and 3 h. RNA isolation and amplification were performed as described above.

cDNAs were synthesized from 2.5 μg total RNA using the SuperscriptII reverse transcriptase (Invitrogen) primed with 50 pmol random hexamers. Amplification was performed in a 20-μl volume containing 1× QuantiTect SYBR Green PCR Master Mix (Qiagen) with cDNA template from 250 ng total RNA and 10 pmol specific primers (supplemental Table S2, which is available in an online appendix at http://dx.doi.org/10.2337/db07-1414). Data were normalized to the cyclophilin or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene. Quantitative RT-PCR was performed using the LightCycler system (Roche Diagnostics).

Cell extracts were prepared as described previously (28). For ERK1/2 detection, cell lysates were prepared in 1% Triton, 10 mmol/l β-glycerophosphate, 50 mmol/l Tris, pH 7.5, 1 mmol/l EDTA, 1 mmol/l EGTA, 1 mmol/l sodium orthovanadate, 50 mmol/l NaF, 1 mmol/l phenylmethylsulfonyl fluoride, 4 μg/ml leupeptin, 2 μg/ml pepstatin, and 2 μg/ml aprotinin; incubated on ice for 30 min; and centrifuged for 10 min at 13,000 rpm, 4°C. Proteins were quantitated using the BCA Protein Assay Reagent kit (Pierce), separated on 10–12% SDS-PAGE, and transferred onto nitrocellulose membranes by electroblotting. Membranes were incubated with the following primary antibodies: CREM-1 (Santa Cruz Biotechnology, Santa Cruz, CA), DUSP14 (gift of Dr. Francesc Marti, University of Michigan Medical School, Ann Arbor, MI), RGS2 (Santa Cruz Biotechnology), c-myc (Roche Diagnostics), phospho-ERK1/2, and ERK1/2 (Cell Signaling Technology). The primary antibodies were revealed with anti-rabbit or anti-mouse IgG secondary antibodies (Amersham Biosciences) coupled to horseradish peroxidase using the ECL Western Blotting Detection kit (Amersham Biosciences).

The small heterogenous RNAs (shRNAs) for CREMα and DUSP14 were subcloned in the pSUPER vector (29). The target sequences can be found in supplemental Table S3. Selection of the shRNA was after cotransfection in HEK-293T cells (Lipofectamine2000; Invitrogen) of each pSUPER-shRNA plasmid with either pSV-CREMα (gift of Prof. Paolo Sassone-Corsi, University of California, Irvine, CA) or pcDNA3-DUSP14 (DUSP14 amplified by RT-PCR with the following primers: sense, 5′-CATGGGATCCACCATGGAGCAAAAGCTCATTTCTGAAG-AGGACTTGATGAGCTCCAGAGGTCACAG-3′; antisense, 5′-CATGCTCGAGCTAAAT-CCCCCAATAAGGCATC-3′, c-myc-tagged) and assessment of CREMα and DUSP14 expression by Western blotting.

10⁷ βTC-Tet or MIN6 cells were electroporated with 30 μg DNA and seeded in a 24-well plate. After 2 days, medium was changed to DMEM containing 0.5% FBS and 3 mmol/l glucose. After 12 h, medium was changed to contain either 3 or 8 mmol/l glucose, with or without 10 nmol/l exendin-4 (Bachem), and incubations were continued for 24 h. [³H]methylthymidine (0.25 μCi/well; Amersham Biosciences) was added for the last 12 h. Cells were then washed twice with ice-cold PBS and lysed on ice for 20 min with 10% trichloroacetic acid, and then 0.2 mol/l NaOH/1% SDS was added. After 10 min at room temperature, 0.2 mol/l HCl was added, and the solubilized material was transferred to scintillant. Radioactivity was determined by a liquid scintillation counter.

The shRNA sequences were cloned into the pLVTHM vector (gift of Prof. Didier Trono, Ecole Polytechnique Fédérale de Lausanne, Lausanne, Switzerland) (30). The H1-RNA promoter cassette in pLVTHM was excised by EcoRI/ClaI digestion and replaced by the EcoRI/ClaI-cut H1-shRNA cassette from the pSUPER-shRNA constructs to generate pLVTHM-shRNA vectors. The dominant-negative form of human DUSP14 (C111S; gift of Dr. Francesc Marti, University of Michigan Medical School, Ann Arbor, MI) (31) was c-myc tagged and cloned into the lentiviral Trip vector (28). Lentiviruses were prepared by transient transfection of HEK-293T cells (28). Virus titer was determined by p24 ELISA (PerkinElmer Life Sciences). βTC-Tet and MIN6 cells were transduced with a multiplicity of infection of 10–20. For pLVTHM-transduced cells, selection of infected cell populations was performed by FACS using the green fluorescent protein (GFP) fluorescence (FACStar-Plus; Becton Dickinson). Populations of cells transduced with the Trip vector were selected with 800 μg/ml G418 for 1 week and 400 μg/ml for an additional week.

Islets from male C57BL/6 mice were isolated by collagenase digestion (32) and Histopaque density gradient purification. They were suspended in RPMI-1640 containing 11.1 mmol/l glucose supplemented with 10% FBS and 2 mmol/l glutamax. Immediately after isolation, batches of 25 or 50 islets were infected with the lentiviruses at a multiplicity of infection of 40 for 3 h in a total volume of 300 μl. Islets were then seeded in plates coated with extracellular matrix derived from bovine corneal endothelial cells (Novamed, Jerusalem, Israel) (33) or kept in suspension for 7 days. For proliferation measurements, islets were then treated with 50 nmol/l exendin-4 for 48 h and incubated for the last 24 h with either 10 μmol/l 5′-bromo-2′-deoxyuridine (BrdU) or 0.5 μCi/well [³H]methylthymidine. Incorporated radioactivity was measured as described above. For BrdU detection, cells were washed with PBS, fixed in ethanol, and incubated with an anti-BrdU antibody (BrdU labeling and detection kit I; Roche Diagnostics) and an anti-GLUT2 antibody (34) for 1 h at 37°C. After washing, cells were stained with Cy3-conjugated antibodies (Jackson Immunoresearch) and FITC-conjugated antibodies (Calbiochem) for 30 min at 37°C. Cells were mounted with Vectashield containing DAPI (Reactolab, Servion, Switzerland) and examined under a fluorescence microscope. For each condition, 2–5,000 GLUT2-positive cells were counted. In the basal state, 1–2% of GLUT2-positive cells were BrdU positive.

Results are given as means ± SEM. Statistical analyses were performed using Student's t tests or a χ² test (for proportion comparison) followed by post hoc pairwise comparisons with P values corrected by Holm method.

To identify genes regulated by GLP-1 in β-cells, we performed a transcriptomic analysis of growth-arrested βTC-Tet cells (22,35) treated with GLP-1 for 45 min, 3 h, and 7 h or left untreated. Using an unsupervised analytical method, the regulated genes were grouped into five different clusters according to their kinetics of up- or downregulation (supplemental Fig. S1). The complete list of genes whose expression was significantly regulated is presented in supplemental Table S1.

On analysis of the regulated genes, we identified four genes that were rapidly and strongly upregulated and that may act as negative regulators of GLP-1 signaling in β-cells: RGS2, which reduces cAMP production in βTC cells (36); the ATF/CREB family members CREM and ICERI, which inhibit CREB transcriptional activity (37); and DUSP14, which inactivates the ERK1/2 signaling pathway (31). RGS2 was increased by eightfold after 45 min of GLP-1 treatment. CREM and ICERI were maximally induced by 24- and 30-fold, respectively, after 3 h, and DUSP14 exhibited a threefold increase in expression after 3–7 h of treatment. Quantitative RT-PCR analysis confirmed the expression profiles observed in the microarray experiments (Fig. 1).

At the protein level, increased expression of RGS2 was confirmed after 45 min of GLP-1 treatment (Fig. 2). CREMα was the major CREM isoform present in βTC-Tet cells, and its expression was upregulated by GLP-1 at 3 h (Fig. 2). ICERI and ICERIγ were hardly detectable in nonstimulated βTC-Tet cells, but their levels were strongly increased at 3 and 6 h of treatment (Fig. 2). DUSP14 expression was increased after 6 h of GLP-1 treatment (Fig. 2). These data indicate that GLP-1 rapidly and strongly induced expression of negative regulators of its own signaling pathway.

Using pharmacological agents, we analyzed the signaling pathways involved in the induction of RGS2, CREM, ICERI, and DUSP14 by GLP-1. We showed that forskolin induced the expression of each gene (supplemental Fig. S2) and that the PKA inhibitor H89 prevented their induction (supplemental Fig. S3). The cAMP-binding protein Epac2 was not involved because its specific activator 8-(4-chlorophenylthio)-2′-O-methyl-cAMP had no effect (Supplemental Fig. S4). There was no evidence for the involvement of the ERK1/2 and PI 3-kinase signaling pathways because the MEK1/2 inhibitor PD98059 or the PI 3-kinase inhibitor wortmannin did not block induction of these genes (Supplemental Fig. S3). Interestingly, however, Ca²⁺ influx through l-type voltage-dependent Ca²⁺ channels was required for full induction of RGS2, CREM, and ICERI by GLP-1, as shown by the partial inhibitory effect of nimodipine (supplemental Fig. S5), but Ca²⁺ influx was not sufficient for induction of either gene as observed in Bay K8644–treated cells (data not shown). Thus, induction of RGS2, CREM, ICERI, and DUSP14 depends on the cAMP/PKA pathway and is potentiated by Ca²⁺ influx.

To evaluate whether RGS2, CREM, ICERI, and DUSP14 were similarly regulated in primary β-cells, we seeded cell sorter–purified primary rat β-cells (26,27) on extracellular matrix-coated dishes and incubated them in the presence or absence of 100 nmol/l GLP-1 for 45 min or 3 h. RGS2 mRNA expression was upregulated by 26-fold at 45 min of GLP-1 treatment (Fig. 3); CREM and ICERI transcripts were induced by 30- and 24-fold, respectively, after 3 h (Fig. 3); and DUSP14 expression was upregulated by threefold after 3 h (Fig. 3). These results demonstrate that GLP-1 induced a pattern of RGS2, CREM, ICERI, and DUSP14 gene expression that was very similar in primary islet β-cells and the βTC-Tet cell line.

The effect of CREMα and DUSP14 on β-cell proliferation was first tested by reducing their expression in the βTC-Tet cells. Different shRNAs were subcloned in the pSUPER vector and tested for their capacity to reduce the expression of CREMα and DUSP14 in transiently transfected HEK-293T cells, as assessed by quantitative RT-PCR (data not shown) and Western blot analysis. We identified shRNAs (shCREM-2 and shDUSP-2) that suppressed CREMα or DUSP14 expression by >90% (Fig. 4A and D).

These pSUPER-shRNAs were then electroporated into βTC-Tet cells. After a 2-day culture period, cells were exposed or not to exendin-4 for 24 h, [³H]thymidine was added for the last 12 h to assess proliferation, and RNA was extracted to evaluate the reduction of CREMα and DUSP14 expression. The electroporation conditions led to transduction of ∼60% of the cells and an observed reduction of CREMα and DUSP14 mRNA expression at the end of the experiment of ∼30–40% (Fig. 4B and E, respectively). The rate of βTC-Tet cell proliferation was increased by 1.4-fold by reducing CREMα expression (Fig. 4C), and the same increase in proliferation was seen when cells were kept at 3 or 8 mmol/l glucose. Similarly, reduction of DUSP14 expression by shDUSP-2 increased [³H]thymidine incorporation by 1.5- and 1.6-fold at 3 and 8 mmol/l glucose, respectively (Fig. 4F). In contrast to what has been reported with the INS-1 cells (10), exendin-4 did not increase proliferation in βTC-Tet cells.

To further support the involvement of CREMα and DUSP14 in β-cell proliferation, we generated stable cell lines by lentiviral-mediated gene transduction for expression of shCREM-2 or a dominant-negative form of DUSP14 (DUSP14-DN) (31). Pools of transduced cells were selected by cell sorter based on GFP fluorescence (CREMα) or using G418 (DUSP14-DN). Cell replication was assessed by cell counting. Expression of shCREM-2 led to a reduction of CREMα mRNA expression and an increased rate of cell proliferation compared with control cells (Fig. 5A and B). Cells stably expressing DUSP14-DN had increased ERK1/2 phosphorylation and showed an increased rate of proliferation compared with control cells expressing LacZ (Fig. 5C and D).

To evaluate a possible additive effect of reducing CREMα and DUSP14 on cell proliferation, we determined the proliferative rate of the cells stably expressing shCREM-2 after electroporation of shDUSP-2. This induced a further 1.5-fold increase in proliferation at either 3 or 8 mmol/l glucose (Fig. 5E). Likewise, cells stably expressing DUSP14-DN and electroporated with shCREM-2 showed an 1.3-fold increase in cell replication compared with control cells at either 3 or 8 mmol/l glucose (Fig. 5F).

Together, these results indicate a role for both the CREB and ERK1/2 signaling pathways in controlling cell replication and that these pathways may control proliferation additively. Because proliferation is not increased on exendin-4 treatment, the CREB and ERK1/2 pathways may be constitutively active in these cells. This may result from the constitutive activity of the GLP-1 receptor and the ectopic expression of the preproglucagon gene, which may lead to production of GLP-1 or glucagon and autocrine activation of their receptors (38).

To evaluate whether CREMα and DUSP14 were also involved in the regulation of proliferation in primary β-cells, we infected mouse islet cells with lentiviruses expressing shCREM-2, shDUSP-2, or DUSP14-DN or with control vectors. Islets were then seeded on extracellular matrix–coated dishes and kept for 7 days in culture to allow monolayer formation and full expression of either the shRNAs or DUSP14-DN. In preliminary experiments, using GFP expression from the pLVTHM vector, we established that with a multiplicity of infection of 40, >80% of the cells were infected. Monolayer cultures were then exposed or not to exendin-4 for 48 h. β-Cell proliferation was assessed by immunodetection of BrdU incorporated in GLUT2-positive cells during the last 24 h of culture (Fig. 6A).

Expression of shCREM-2 led to a ∼75% reduction of CREMα mRNA (Fig. 6B) but did not modify the basal rate of cell proliferation (Fig. 6C). In contrast to the β-cell lines, exendin-4 treatment markedly increased by 2.5-fold cell proliferation. The reduction of CREMα expression induced a further 1.3-fold increase in BrdU incorporation. Expression of shDUSP-2 reduced DUSP14 mRNA expression by ∼50% (Fig. 6D). This did not change basal BrdU incorporation but increased the exendin-4–stimulated proliferation by 1.5-fold (Fig. 6E). Similarly, expression of DUSP14-DN (Fig. 6F) increased exendin-4–induced BrdU incorporation by 1.3-fold (Fig. 6G). Figure 6H shows that proliferation measured by [³H]thymidine was also increased by exendin-4 and that reducing DUSP14 expression by transduction of shDUSP-2 further augmented the proliferative effect of exendin-4.

Overall, these results show that the proliferation-inducing effect of exendin-4 can be enhanced by reducing the expression of CREMα or DUSP14 or by overexpressing a dominant-negative form of DUSP14.

A major goal of current research in β-cell biology is to understand the mechanisms that stimulate or limit β-cell proliferation and how these mechanisms can be exploited to increase β-cell mass to improve treatment of type 1 and type 2 diabetes. Here, using a transcriptomic approach, we found that several negative regulators of the cAMP/PKA/CREB and MAPK/ERK1/2 pathways are rapidly and strongly upregulated by GLP-1 treatment of β-cells. Importantly, we show that reducing or preventing the normal expression or activity of these gene products leads to increased β-cell line proliferation and increased induction by GLP-1 of primary β-cell proliferation.

Microarray analysis of transcripts expressed in growth-arrested, well-differentiated βTC-Tet cells (22,35) exposed to GLP-1 revealed five sets of genes with distinct time-dependent patterns of expression. Three sets showed increased expression with highest transcript levels reached at 45 min, 3 h, or 7 h; two sets showed reduced expression levels reaching nadir at 3 or 7 h. Here, we focused on the very early events, i.e., genes with maximal increased expression at 45 min or 3 h, because we reasoned that these may be critical for conditioning the long-term response of the cells to GLP-1 treatment, and we selected four genes that negatively regulate the signaling pathways activated by this hormone. RGS2 is a GTPase-activating protein that inactivates Gα (39) and also directly inhibits the activity of some adenylyl cyclases (40). CREMα and ICERI are repressors of cAMP-induced, CREB-dependent transcription (37). It has been reported that expression of ICERI and ICERIγ is markedly increased in diabetes (41) and that transgenic overexpression of ICERIγ in pancreatic β-cells causes early diabetes and reduced β-cell proliferation (42), suggesting a role of the cAMP/PKA/CREB pathway in β-cell proliferation. Finally, DUSP14 is a MAPK phosphatase, which inhibits the activity of ERK1/2 (31). We showed that the induction of these genes by GLP-1 was dependent on the activation of the cAMP/PKA/CREB pathway and was potentiated by Ca²⁺ influx through voltage-dependent Ca²⁺ channels. Importantly, these genes were similarly regulated in βTC-Tet cells and in cell sorter–purified primary β-cells.

We then tested the hypothesis that CREMα and DUSP14 limit the proliferative effect of GLP-1 by reducing or preventing their expression using shRNAs or by increasing the activity of ERK1/2 by expression of a dominant-negative form of DUSP14. In transiently or stably transfected β-cells, this led to increased proliferation as assessed by [³H]thymidine incorporation or cell counting. Furthermore, we demonstrated an additive effect of the cAMP/PKA/CREB and MAPK/ERK1/2 pathways in controlling proliferation because the proliferation of β-cells stably expressing shRNA-CREMα was further increased when electroporated with the shRNA-DUSP14 and the proliferation of β-cells expressing DUSP14-DN was increased by electroporation of the shRNA-CREMα. In these cell lines, however, we found that exendin-4 did not enhance proliferation, in contrast to studies with INS-1 cells (10).

In primary mouse islets, however, exendin-4 stimulated proliferation as assessed by BrdU or [³H]thymidine incorporation. Then, transducing shRNA-CREMα, shRNA-DUSP14, or DUSP14-DN increased the proliferative effect of exendin-4 but had no effect on the basal rate of proliferation.

Thus, the above observations suggest that both the cAMP/PKA/CREB and MAPK/ERK1/2 signaling pathways are involved in the control of β-cell proliferation. In βTC-Tet cells, GLP-1 did not increase the proliferation rate. This may be due to an ectopic expression of the preproglucagon gene, which leads to autocrine activation of the GLP-1 and glucagon receptors and permanent elevated expression of CREMα and DUSP14 (38). Thus, shRNA-mediated knockdown of these two genes or expression of a dominant-negative form of DUSP14 is sufficient to increase their proliferation.

In primary β-cells, transduction of shRNA to knockdown CREMα or DUSP14 or expression of DUSP14-DN increases replication only on exendin-4 treatment. Thus, in primary β-cells, the intracellular signaling pathways activated by GLP-1 and controlling proliferation can be further increased by preventing the expression of two of their negative regulators.

Together, our data demonstrate that β-cells have evolved a very tight control system to limit GLP-1–activated proliferation by inducing several negative regulators of its own signaling pathway. These genes are induced by the cAMP pathway and potentiated by influx of Ca²⁺, suggesting that their induction is more potent in mature β-cells than in their precursors because Ca²⁺ influx after GLP-1 treatment is a characteristic of well-differentiated β-cells (43). Thus, both the induction and the effect of these negative regulators of GLP-1–induced action may be more pronounced in mature β-cells than in precursor cells (Fig. 7). This may be required to prevent exaggerated mature β-cell proliferation and abnormally high insulin secretion. The expression of these negative regulators may, however, be reduced to increase GLP-1–dependent mature β-cell proliferation in diabetic patients. Recent studies have shown that replication of mature β-cells is a major mechanism for replacing lost β-cells with little, if any, contribution of precursor cell expansion (44–46). Increasing the mass of existing β-cells by interfering with the mechanisms we have described here could be used to improve GLP-1–based therapy of type 2 diabetes (47,48) so long as reducing the expression of these negative regulators could be finely tuned so as not to overcome the natural feedback loops that normally prevent excessive β-cell growth.

This work has received grants from the Swiss National Science Foundation (3100A0-113525), the Juvenile Diabetes Research Foundation (Program Project 7-2005-1158), and the European Union (Integrated Project Eurodia LSHM-CT-2006-518153, Framework Programme 6 of the European Community).

We thank the University of Lausanne DNA Array Facility for help with cDNA microarray experiments and Katharina Rickenbach for help in preparing cell sorter–purified β-cells.