Glucagon-Like Peptide 1 Inhibits the Sirtuin Deacetylase SirT1 to Stimulate Pancreatic β-Cell Mass Expansion

Human GLP-1 fragment 7–36 amide, exendin-4, nicotinamide, and resveratrol were obtained from Sigma (St. Louis, MO). RPMI-1640 medium, FCS, and other culture media were purchased from Invitrogen (Burlington, ON, Canada). Anti-FKHR antibody was purchased from Cell Signaling (Beverly, MA). Antiacetyl-lysine and anti-SirT1 antibodies were obtained from Millipore (Bedford, MA). Anti–guinea-pig insulin was purchased from Sigma.

INS832/13 cells (24) were grown in RPMI-1640 medium supplemented with 10 mmol/L HEPES, 10% heat-inactivated FCS, 2 mmol/L l-glutamine, 1 mmol/L sodium pyruvate, 50 μmol/L β-mercaptoethanol, 100 IU/mL penicillin, and 100 μg/mL streptomycin at 37°C in a humidified 5% CO₂ atmosphere. Cells at 70% confluence were washed with phosphate-buffered saline and preincubated in serum-free RPMI-1640 medium supplemented with 3 mmol/L glucose and 0.1% BSA (Sigma) for at least 4 h before treatment. This condition mimics calorie restriction and was shown to activate SirT1.

Rat islets were isolated from male Wistar rats (250 g) by collagenase digestion. Islets were subsequently purified over a Histopaque gradient and handpicked under a microscope.

Human islets were isolated from organ donors at the Department of Surgery, Montreal General Hospital, McGill University Health Center, Montreal, Quebec, Canada (three separate donors were received). Human ethics approval was obtained through the McGill University Health Center ethics committee. Donors were between ages 42 and 65, and none had a history of diabetes or metabolic disorder. Islets were isolated by digestion with Liberase CI (Boehringer Mannheim, Indianapolis, IN) followed by purification over a Ficoll gradient using a CORE 2991 Cell Processor (COBE BCT, Denver, CO). After isolation, islets were rested overnight in complete CMRL medium.

Proteins were extracted and quantified by BCA assay (Roche, Indianapolis, IN). FoxO1 protein complexes were immunoprecipitated from 1 mg of total protein extracts and subjected to electrophoresis on 8 or 10% polyacrylamide gels.

Cells were fixed to isolate intact chromatin and sheered in 200-base pair fragments, and SirT1 DNA/protein complexes were immunoprecipitated using an anti-SirT1 antibody (Millipore). Bound DNA was amplified by PCR using sequence-specific primers designed for amplification of the forkhead binding site region of the Foxa2 promoter (16).

Cells were treated for 15 min, and SirT1 activation was measured using a commercially available fluorometric assay (Biomol, Plymouth Meeting, PA). In brief, the kit uses a SirT1 substrate consisting of amino acids 379–382 of human p53 (Arg-His-Lys-Lys[Ac]). The assay's fluorescence signal is generated in proportion to the amount of deacetylation of Lys-382, a known in vivo target of SirT1. Cell extracts were incubated for 10 min at 37°C with a 25 μmol/L concentration of the fluorogenic SirT1 substrate. Reactions were stopped by the addition of 1 mmol/L nicotinamide, and the fluorescence resonance energy transfer signal was determined.

NAD⁺, NADH, and ATP levels were measured in cell extracts using commercial kits (NAD/NADH quantification kit and ATP fluorometric kit from Biovision, Mountain View, CA).

Nicotinamide phosphoribosyltransferase (Nampt) activity was measured by the conversion of ¹⁴C-labeled nicotinamide to [¹⁴C]nicotinamide mononucleotide as described previously (25). In brief, cell lysates (10 μg total protein) were incubated with ATP, 5-phosphoribosyl-1-pyrophosphate, and [¹⁴C]nicotinamide for 30 min; filtered through glass fiber filters; and washed with acetone. Filters were air-dried and transferred into scintillation vials for quantification.

Cells were transfected with either control green fluorescent protein (GFP), wild-type SirT1, or a catalytically inactive SirT1 mutant (26) using Lipofectamine 2000 (Invitrogen) as described previously (16). Transfection efficiency was 70%. The following day, proliferation of transfected cells was evaluated using BrdU incorporation assay (Roche). In brief, cells were treated for 24 h in the absence or presence of 1 nmol/L GLP-1, and BrdU was added to the culture medium for the last 60 min of the incubation period.

Mice were maintained on a mixed background (129/sv and C57BL/6J). Wild-type and SirBACO transgenic mice (27) aged 6–12 months were treated with daily intraperitoneal injections of exendin-4 (10 nmol/kg) or saline for 7 days (n = 5 each). Mice were killed, and pancreas sections were processed for insulin and Ki67 immunohistochemistry to assess β-cell proliferation. Cross-sectional islet area was measured using Image Pro Plus software (Media Cybernetics, Silver Spring, MD). Results are presented as relative β-cell areas (i.e., percent of insulin + areas relative to total pancreas areas). For each animal, at least four sections spaced 80 μm apart were studied.

Data are presented as means ± SEM. Statistical analyses were performed with SPSS (SPSS Inc., Chicago, IL) using ANOVA.

We have previously shown that inhibition of the forkhead transcription factor FoxO1 mediates the pleiotropic action of the glucoincretin hormone GLP-1 (10). FoxO1 is subject to various posttranslational modifications, including Akt-dependent phosphorylation and SirT1-mediated deacetylation. Here, we sought to test the hypothesis that GLP-1 inhibits SirT1 to enhance FoxO1 acetylation and to stimulate β-cell mass expansion.

We first evaluated FoxO1 acetylation and binding to SirT1 by Western blot in INS832/13 cells. We resorted to the use of a pancreatic β-cell line because of the high amount of biological material required to perform the assays. INS cells were incubated in serum-free medium, a condition known to activate SirT1, in the presence or absence of GLP-1 and the SirT1 activators H₂O₂ or resveratrol. Figure 1A and B shows that GLP-1 induced FoxO1 acetylation, an action prevented by resveratrol but not by H₂O₂. Coimmunoprecipitation experiments revealed that GLP-1 inhibited the physical association between FoxO1 and SirT1 (Fig. 1A). However, this action was inhibited by both H₂O₂ and resveratrol. The action of GLP-1 on FoxO1 acetylation was transient, with a maximal effect observed between 15 and 45 min. FoxO1 acetylation returned to basal value after ∼180 min (Fig. 1C). Taken together, these results suggest that GLP-1 inhibits SirT1-mediated FoxO1 deacetylation.

Chromatin immunoprecipitation assay was performed to assess whether SirT1 is recruited to the promoter region of FoxO1 target genes. Figure 1D shows that SirT1 was physically associated to a region of the Foxa2 promoter encompassing the forkhead binding site (10). Moreover, SirT1 binding to the Foxa2 promoter was reduced upon GLP-1 treatment. This result indicates that SirT1 binding to the Foxa2 promoter mirrors that of FoxO1, as we have previously demonstrated (10). Altogether, these findings indicate that SirT1 interacts with FoxO1 and is recruited to the promoter of FoxO1 target genes in a GLP-1–inhibitable manner. These data suggest a role for SirT1 in the transcriptional response to the glucoincretin hormone in β-cells. We next performed subcellular fractionation to address whether GLP-1 provokes SirT1 nuclear exclusion, as it does for FoxO1 (Fig. 1E). Our results show that SirT1 is predominantly nuclear in INS cells and that its subcellular localization was not altered by GLP-1 treatment. Conversely, GLP-1 induced cytoplasmic translocation of FoxO1, as we have published previously (10).

We next sought to directly test whether GLP-1 affects SirT1 activity using an in vitro deacetylase assay. Figure 2A shows that GLP-1 blunted SirT1 activity to the same extent as the SirT1 inhibitor nicotinamide. Conversely, SirT1 activity was increased in resveratrol-treated cells. Because SirT1 activity is dependent on NAD⁺, we measured changes in the NAD⁺-to-NADH ratio in response to GLP-1 treatment in INS cells and isolated human islets. Figure 2B shows that GLP-1 decreased the NAD⁺-to-NADH ratio by 30% in INS cells, an action mimicked by the addition of 10% serum. Conversely, resveratrol increased the NAD⁺-to-NADH ratio by 60%. Both GLP-1 and serum also elicited a reduction in the NAD⁺-to-NADH ratio in human islets (Fig. 2C), whereas the effect of resveratrol (∼25%) did not reach statistical significance. The reduction in NAD⁺ availability provides a possible mechanism by which GLP-1 could acutely regulate SirT1 activity. To directly test this possibility, we supplemented the assay buffer with 450 μmol/L NAD⁺ (Fig. 2D). Supplementation of NAD⁺ suppressed GLP-1 as well as serum inhibition of SirT1 activity. In an effort to unravel the mechanism by which GLP-1 could alter the NAD⁺-to-NADH ratio, we investigated its action on the activity of Nampt, the rate-limiting enzyme for the biosynthesis of NAD, and on the cellular ATP content. GLP-1 decreases the activity of Nampt (Fig. 2E), thereby revealing a novel mechanism by which GLP-1 could decrease the NAD⁺-to-NADH ratio. FK-866 (10 nmol/L), a specific Nampt pharmacological inhibitor, was used as a positive control. Conversely, GLP-1 did not significantly change the cellular ATP content (Fig. 2F), consistent with a recent publication by Peyot et al. (28). It is interesting that the addition of serum did not alter Nampt activity but significantly increased the cellular ATP concentration. This suggests that serum and GLP-1 could inhibit SirT1 via distinct mechanisms.

We next measured SirT1 expression by quantitative PCR after overnight incubation in serum-free medium in the absence or presence of GLP-1 (Fig. 3A). Our data show that glucose and serum deprivation, a condition mimicking calorie restriction, increased SirT1 expression by fivefold compared with complete growth medium in INS cells. GLP-1 treatment completely abolished the rise in SirT1 expression. We next sought to confirm these observations in isolated rat islets (Fig. 3B). Both GLP-1 and serum decreased SirT1 expression in rat islets as well, although the detected changes were more modest. Thus, SirT1 expression was repressed by 30% in the presence of GLP-1 and by 40% in response to serum. We assessed SirT1 protein levels by Western blot to test whether alterations in SirT1 expression were translated into changes in protein levels. Figure 3C and D shows that GLP-1 decreased SirT1 protein levels by 70% in INS cells, an effect mimicked by serum.

To test whether SirT1 inactivation mediates GLP-1 action on β-cell proliferation, we sought to perform SirT1 gain of function both in vitro and in vivo. INS832/13 cells were either transfected with wild-type SirT1, a catalytically inactive SirT1 mutant, or with control GFP, and β-cell replication was evaluated by BrdU incorporation. Transfection of the cells with SirT1 led to a fivefold increase in protein levels as compared with GFP (Fig. 4A). Overexpression of wild-type SirT1 inhibited GLP-1–induced β-cell proliferation but failed to significantly affect the effect of serum (Fig. 4B). Overexpression of a catalytically inactive SirT1 mutant did not significantly alter β-cell proliferation, indicating that the effects of SirT1 on β-cell proliferation are dependent on its deacetylase activity.

To demonstrate the implication of SirT1 in GLP-1 action in vivo, we determined whether the effects of GLP-1 on β-cell mass were curtailed in transgenic mice with increased dosage of SirT1 (SirBACO mice) as previously described (27). Thus, mice were administered with either saline or the long-acting GLP-1 analog exendin-4 (10 nmol/kg) daily for 7 days. Cross-sectional β-cell areas were estimated by morphometry of pancreatic sections stained for insulin (Fig. 5A), and β-cell proliferation was evaluated by Ki67 staining (Fig. 5B). Exendin-4 administration increased the relative β-cell area by ∼2.5-fold and stimulated β-cell replication by ∼50% in wild-type animals. However, GLP-1 effects on both β-cell area and β-cell replication were abolished in SirBACO mice. Taken together, our results indicate that SirT1 gain of function prevents the mitogenic action of GLP-1 and identify SirT1 as a negative regulator of β-cell mass expansion.

Restoration and maintenance of β-cell mass is a major goal of diabetes treatment (29–31). GLP-1, a potent antidiabetes medication (4), enhances glucose-stimulated insulin secretion and promotes β-cell mass expansion, at least in rodents. However, the molecular mechanism by which GLP-1 exerts its action on β-cell mass is not fully elucidated. We have previously demonstrated that the transcription factor FoxO1, an important regulator of β-cell mass (16–18), is a prominent transcriptional effector of GLP-1 action (10). Thus, GLP-1 provoked nuclear exclusion of FoxO1 to relieve a molecular brake on Pdx1 and Foxa2 expression. Also, the effect of exendin-4 on β-cell mass expansion was abolished in transgenic mice with β-cell–specific FoxO1 gain of function (10). Acetylation/deacetylation of FoxO1 regulates its transcriptional activity (19). On one hand, acetylation of FoxO1 by CBP/p300 reduces its DNA binding activity and increases its susceptibility to phosphorylation by Akt (20). On the other hand, acetylated FoxO1 binds to the promyelocytic leukemia–associated protein Pml and is targeted to specific nuclear subdomains (Pml nuclear bodies) (32). Upon calorie restriction or oxidative stress, FoxO1 is deacetylated by the NAD⁺-dependent sirtuin deacetylase SirT1, which renders FoxO1 immobile within the nuclear compartment and results in FoxO1 transcriptional activity (26). This prompted us to investigate the role of SirT1 in GLP-1 action.

Our findings show that GLP-1 increases FoxO1 acetylation via inhibition of the NAD+-dependent SirT1 deacetylase. Our study provides two possible mechanisms by which GLP-1 could regulate SirT1. First, we show that GLP-1 acutely decreases the cellular NAD⁺-to-NADH ratio, possibly via its hitherto uncharacterized action on Nampt. Second, GLP-1 downregulates SirT1 expression, thereby providing a long-term mechanism by which chronic or prolonged exposure to GLP-1 receptor agonists could also inhibit SirT1. The fact that GLP-1 failed to increase the intracellular ATP content seemingly refutes the hypothesis that GLP-1 could inhibit SirT1 via the stimulation of β-cell glucose metabolism. However, because our assay did not take into account compartmentalization of ATP production or breakdown, this possibility cannot be completely ruled out. We also demonstrate that the effects of GLP-1 and its long-acting analog exendin-4 on β-cell mass were blunted in a pancreatic β-cell line with increased dosage or SirT1 as well as in transgenic mice with SirT1 gain of function. Our results are consistent with a model in which GLP-1 inhibits SirT1-mediated FoxO1 deacetylation to relieve the constraint on β-cell proliferation and β-cell mass expansion (Fig. 6). Thus far, the action of GLP-1 agonists on β-cell mass has not been demonstrated in humans in vivo. Moreover, the reported increase in β-cell mass in mice has been shown to be restricted after the initial burst of proliferation in the postnatal period (33,34). However, a recent publication suggests that exendin-4 can stimulate β-cell replication in human islet grafts (35).

Previous studies using genetically engineered mice with SirT1 gain (36) or loss of function (37) indicate that SirT1 could enhance insulin secretion, an effect mediated via reduction of Ucp2 expression. These findings garnered much attention by identifying SirT1, a molecule already scrutinized for its role in longevity, as a potential target for diabetes treatment. However, the action of SirT1 on insulin secretion is difficult to conciliate with its role in calorie restriction (38,39). In addition, these effects were not reproduced in SirBACO mice (27). This discrepancy may be attributed to differences in the level and pattern of SirT1 expression between the various animal models. At first glance, our findings that the glucoincretin hormone GLP-1 inhibits SirT1 may seem conflicting with the action of SirT1 on insulin secretion. However, the short-term effects of GLP-1 on insulin secretion are not transcriptional in nature (40).

In conclusion, we show that GLP-1 suppresses SirT1 activity to stimulate β-cell mass expansion. Our study describes a novel mechanism of action for the glucoincretin hormone and identifies for the first time SirT1 as a negative regulator of pancreatic β-cell mass. Thus, SirT1 could link β-cell replication to the hormonal/nutritional status of the cells.

J.B. is a Canadian Diabetes Association scholar and a Fonds de la Recherche en Santé du Québec Junior Investigator. This work was supported by a research grant to J.B. from the National Sciences and Engineering Research Council.

No potential conflicts of interest relevant to this article were reported.

P.-O.B.-D. and L.V. conducted experiments and performed data analyses. N.K. and W.G. supplied SirBACO mice and reviewed the manuscript. J.B. supervised the project, conducted experiments, and wrote the manuscript.

The authors acknowledge Drs. Domenico Accili (Columbia University), Yves Deshaies (Université Laval), and Frédéric Picard (Université Laval) for numerous useful discussions.