Exendin-4 Promotes Liver Cell Proliferation and Enhances the PDX-1-induced Liver to Pancreas Transdifferentiation Process

Abstract

Over the last few years, evidence has accumulated revealing the unexpected potential of committed mammalian cells to convert to a different phenotype via a process called transdifferentiation or adult cell reprogramming. These findings may have major practical implications because this process may facilitate the generation of functional autologous tissues that can be used for replacing malfunctioning organs. An instructive role for transcription factors in diverting the developmental fate of cells in adult tissues has been demonstrated when adult human liver cells were induced to transdifferentiate to the pancreatic endocrine lineage upon ectopic expression of the pancreatic master regulator PDX-1 (pancreatic and duodenal homeobox gene 1). Since organogenesis and lineage commitment are affected also by developmental signals generated in response to environmental triggers, we have now analyzed whether the hormone GLP-1 (glucogen-like peptide-1) documented to play a role in pancreatic beta cell differentiation, maturation, and survival, can also increase the efficiency of liver to pancreas transdifferentiation. We demonstrate that the GLP-1R agonist, exendin-4, significantly improves the efficiency of PDX-1-mediated transdifferentiation. Exendin-4 affects the transdifferentiation process at two distinct steps; it increases the proliferation of liver cells predisposed to transdifferentiated in response to PDX-1 and promotes the maturation of transdifferentiated cells along the pancreatic lineage. Liver cell reprogramming toward the pancreatic beta cell lineage has been suggested as a strategy for functional replacement of the ablated insulin-producing cells in diabetics. Understanding the cellular and molecular basis of the transdifferentiation process will allow us to increase the efficiency of the reprogramming process and optimize its therapeutic merit.

INTRODUCTION

Transdifferentiation or adult cells reprogramming is a promising regenerative medicine approach, because it allows the conversion of one type of adult cell into another type of cell. Reprogramming liver cells toward the beta cell lineage and function has been extensively studied both in vivo and in vitro (1–9). An important instructive role for pancreatic transcription factors in dictating the process has been suggested (1–9). We have previously demonstrated that adult human liver cells can be propagated in vitro and induced to transdifferentiate along the pancreatic endocrine lineage using ectopic expression of PDX-1 (pancreatic and duodenal homeobox gene 1) (3). Transdifferentiated human liver cells acquire many beta cell-like characteristics, such as insulin production, processing, and storage as well as the ability to secrete the hormone in a glucose-regulated manner. Insulin-producing human liver cells were functional for prolonged periods of time and ameliorated hyperglycemia when implanted under the renal capsule of diabetic immunodeficient mice (3). Additional pancreatic transcription factors have been demonstrated to participate in endowing liver cells with pancreatic characteristics (10–13).

Soluble factors, proven instrumental in pancreatic beta cell differentiation, may also play roles in adult cell reprogramming in an as yet unknown manner (3, 10, 14). One of the major soluble factors suggested to affect beta cell differentiation, function, and survival in the pancreas is GLP-1 (glucagon-like peptide-1). GLP-1 is a glucose regulatory hormone with insulinotropic actions. GLP-1 is encoded in the proglucagon gene and is mainly produced in enteroendocrine L cells of the gut. It is secreted into the blood stream in response to nutrient intake (15, 16). In the pancreas, GLP-1 increases insulin gene expression, prohormone production, and its glucose-regulated insulin secretion. Moreover, GLP-1 stimulates the expression of many beta cell-specific genes, including PDX-1, glucokinase, and GLUT2, and induces beta cell proliferation, neogenesis, and survival (15, 16). Numerous studies suggested that GLP-1 promotes beta cell differentiation of pancreatic precursor cells and embryonic stem cells (17–19). Therefore, we sought to analyze this soluble effect on the transdifferentiation process of liver to pancreas.

Although several studies have shown that GLP-1 not only directly binds to liver membranes but also affects liver metabolism, no evidence for GLP-1 receptor expression in liver was found (20–22). Pancreatic GLP-1R (GLP-1 receptor) belongs to the B-class of the G-protein-coupled receptor superfamily (23, 24). Upon binding to its receptor, GLP-1 activates adenylate cyclase and cAMP production. Downstream effectors of cAMP include protein kinase A and the Epac family. Protein kinase A increases CREB⁴ transcriptional activity and up-regulates Irs2 expression (23, 24).

Along with the documented reports on GLP-1 effects on liver, we report here that the GLP-1 receptor agonist, exendin-4, augments the PDX-1-induced activation of the beta cell lineage and functions in the liver. Exendin-4 activates G-protein-coupled receptor signaling in liver cells and possesses a dual role in the PDX-1-induced liver to pancreas transdifferentiation. 1) It facilitates liver cell proliferation and increases the number of liver cells predisposed to undergo the reprogramming process; 2) it increases transdifferentiated cell maturation along the beta cell lineage and function. Taken together, exendin-4 is a permissive soluble factor that augments the therapeutic merit of the PDX-1-induced developmental redirection between liver and pancreas. Efficient adult cell reprogramming allows the potential generation of autologous surrogate beta cells for the treatment of diabetes, overcoming both the shortage in tissues from cadaveric donors and the need for life-long immune suppression.

EXPERIMENTAL PROCEDURES

Human Liver Cells

Adult human liver tissues were obtained from five different liver specimens from children 4–10 years old and eight individuals over 40 years old. Liver tissues were used with approval from the Committee on Clinical Investigations (the institutional review board).

The isolation of human liver cells was performed as described (3, 5). The cells were cultured in Dulbecco's minimal essential medium (1 g/liter of glucose) supplemented with 10% fetal calf serum, exendin-4 (5 nm; Sigma), epidermal growth factor (20 ng/ml; Cytolab, Ltd.), and nicotinamide (10 mm; Sigma) and kept at 37 °C in a humidified atmosphere of 5% CO₂ and 95% air. The optimal concentration for exendin-4 treatment was determined as the concentration (0–20 nm) that induced the maximal C-peptide secretion from PDX-1 plus nicotinamide and epidermal growth factor (SF)-treated liver cells (data not shown).

Viral Infection

Liver cells were infected with recombinant adenoviruses at a multiplicity of infection of 1000 for 5 days. The adenoviruses used in this study were as follows: Ad-CMV-PDX-1, Ad-RIP-GFP (3, 5), Ad-CMV-GFP (Clontech), and Ad-GLUP-RED. Ad-GLUP-RED was prepared by insertion of 350 nucleotides of the 5′-flanking region of the rat glucagon promoter (generous gift from A. Manin) in the pACCMV.pLpA plasmid to drive ds-red2-n1 gene expression (KpnI/NotI, Invitrogen). The viral particles were generated by standard protocol (25). The specificity of the construct was analyzed in mouse glucagonoma-derived α-TC cells and rat insulinoma-derived RIN cells. Red fluorescence, indicating glucagon promoter activation, was identified only in α-TC and not in RIN cell lines.

cAMP Assay

Cells and medium were analyzed for cAMP level. Cells were seeded in 24-well plates. Following treatments, plates were snap-frozen in liquid nitrogen and stored at −70 °C until the analyses. Samples were diluted in 50 mm sodium acetate, pH 6.2, acetylated, and incubated overnight at −20 °C with ¹²⁵I-labeled cyclic AMP (PerkinElmer Life Sciences) and anti-cAMP antibody. The following day, 1% bovine serum albumin and ethanol were added, samples were centrifuged, and cAMP was measured in precipitate using an automatic γ counter (PerkinElmer Life Sciences).

RNA Isolation and Real-time Reverse Transcription-PCR

Total RNA was isolated, and complementary DNA was prepared and amplified as described (3, 5). The TaqMan fluorogenic probes and the Assay-On-Demand (Applied Biosystems, Foster City, CA) used in this study are as follows: human β-actin, Hs99999903_m1; human insulin, Hs00355773_m1; human glucagon, Hs00174967_m1; human somatostatin, Hs00356144_ m1; human PDX-1, Hs00426216_m1; human GLUT2, Hs00165775_m1; human PC2 (prohormone convertase 2), Hs00159922_m1; human PC1/3 (prohormone convertase 1/3), Hs00175619_m1; human SCG2 (secretogranin 2), Hs00185761_m1; human SGNE1 (secretory granule neuroendocrine 1), Hs00161638_m1; human PAX4, Hs00173014_m1; human PAX6, Hs00242217_m1; human NEUROD1, Hs00159598_m1; human NKX6.1, Hs00232355_m1; human NKX2.2, Hs00159616_m1; human ISL-1, Hs00158126_m1; human BRAIN4, HS00264887_s1; human ARX, Hs00417686_ m1; human MAFA, Hs01651425_s1; human glucokinase Hs01564555_m1.

Insulin and C-peptide Detection

Insulin and C-peptide secretion were measured by static incubations of primary cultures of adult liver cells 3–5 days after the initial exposure to the viral treatment, as described (3, 5). The glucose-regulated insulin and C-peptide secretion were measured at 17.5 mm glucose, which was determined by dose-dependent analyses to maximally induce insulin secretion from transdifferentiated liver cells, without having adverse effects on the function of treated cells (3, 5). Insulin and C-peptide secretion were detected by a radioimmunoassay using a human insulin radioimmunoassay kit (DPC, Los Angeles, CA) and human C-peptide radioimmunoassay kit (Linco Research, St. Charles, MO; <4% cross-reactivity to human proinsulin). The secretion was normalized to the total cellular protein measured by a Bio-Rad protein assay kit.

Immunofluorescence

Human liver cells treated with ad-CMV-PDX-1 for 5 days were plated on glass coverslides in 6-well culture plates. 3–4 days later, the cells were fixed and stained as described (3, 5). For BrdUrd labeling; Cells were incubated with cell proliferation labeling reagent (1:1000; Amersham Biosciences) for 6 h; after washing, the cells were cultured in growth media for further growth or fixed.

The antibodies used in this study were anti-insulin (1:100; Dako), anti-BrdUrd (1:200; Serotec, Raleigh, NC), anti-Ki67 (1:100; Abcam, Cambridge, UK), anti-PAX4 (1:100; R&D Systems, Minneapolis, MN), anti-MAFA (1:250; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), anti-PDX-1 (1:1000; a generous gift from C. Wright, Vanderbilt University, Nashville, TN), and anti-NKX6.1 (1:6000; a generous gift from C. B. Newgard, Duke University, Durham, NC). Secondary anti-rat IgG rhodamine-conjugated antibody (1:100), anti-guinea pig IgG cyanine (cy2)-conjugated antibody (1:200), anti-guinea pig IgG indocarbocyanine (cy3)-conjugated antibody (1:200), anti-rabbit IgG cyanine (cy2)-conjugated antibody (1:200), anti-rabbit IgG indocarbocyanine (cy3)-conjugated antibody (1:200), anti-goat IgG cyanine (cy2)-conjugated antibody (1:200), and anti-goat IgG indocarbocyanine (cy3)-conjugated antibody (1:200) were all from Jackson ImmunoResearch. Finally, the cells were stained with 4′,6-diamidino-2-phenylindole (Sigma). The slides were analyzed using a fluorescent microscope (Provis, Olympus).

Western Blot Analyses

Cells were harvested and lysed for total protein extraction in radioimmunoprecipitation assay buffer (50 mm Tris-HCl, pH 7.4, 150 mm NaCl, 1% Nonidet P-40, 0.25% sodium deoxycholate, 1 mm EDTA, 1 mm NaF) together with a protease inhibitor mixture (Sigma). A total of 50 μg of protein extracts were electrophoretically separated on 10% polyacrylamide gels and blotted onto nitrocellulose membrane (Schleicher & Schuell). The membranes were incubated with primary antibodies, followed by horseradish peroxidase-conjugated goat anti-rabbit or goat anti-mouse secondary antibodies (Amersham Biosciences). Separated band intensities were quantified using ImageJ software.

Antibodies used in this study were anti-p53 (1:500), anti-poly(ADP-ribose) polymerase (1:1000), anti-cyclin D1 (1:200) (Santa Cruz Biotechnology, Inc.), anti-phosphoprotein kinase B (1:1000; Cell Signaling Technology, Danvers, MA), anti-phospho-ERK1/2 (1:10,000; Sigma), anti-phosphoprotein kinase Cζ (1:1000; Santa Cruz Biotechnology, Inc.), anti-phospho-CREB (1:1000; Cell Signaling Technology, Danvers, MA), and anti-β-actin (1:5000; Sigma).

Flow Cytometry

Five days after Ad-RIP-GFP or Ad-GluP-RED infection, cells (5 × 10⁵) were collected, washed twice in phosphate-buffered saline (5 min, 1000 × g), and resuspended in 400 μl of phosphate-buffered saline. Flow cytometry was performed (FACSCalibur, BD Biosciences) using the CellQuest program.

3-(4,5-Dimethylthiazol-2-yl)-2,5- diphenyltetrazolium Bromide Viability Assay

A total of 5 ×10³ cells/well were seeded in 96-well plates and cultured in the appropriate culture media, and 5 days later cells were cultured for 2 h with 500 μg/ml 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide reagent (Sigma). The medium was aspirated, and the cells were dissolved in dimethyl sulfoxide. Absorbance of the formazan product was measured by an enzyme-linked immunosorbent assay reader.

Statistical Analyses

Statistical analyses were performed with a two-sample Student t test assuming unequal variances.

RESULTS

Exendin-4 Activates G-protein-coupled Receptor Signaling in Liver Cells in Vitro

GLP-1 and its receptor's agonist, exendin-4, have a well established role in beta cell function (15, 16). GLP-1 also has known effects on the function of liver cells and even specifically binds to the liver cell membranes. Despite the fact that we and others failed to find evidence for this receptor expression in the liver either at gene expression or protein levels (data not shown) (20–22), we sought to examine whether exendin-4 affects the transdifferentiation process of liver cells to insulin-producing cells. In the pancreas, GLP-1 activates a specific G-protein-coupled receptor, which activates a well characterized intracellular pathway. Therefore, we first analyzed whether exendin-4 activates in liver a signal transduction pathway that resembles that induced by GLP-1 in the pancreatic beta cells. The effect of exendin-4 on liver cells was analyzed by the administration of exendin-4 (5 nm) to adult human liver cells in vitro. Indeed, exendin-4 administration caused an increase in the intracellular levels of cAMP within 1 min, which turned significant within 5 min (Fig. 1a), suggesting that a G-protein-coupled receptor is activated upon exendin-4 treatment in the liver cells. Furthermore, exendin-4 treatment activated in liver cells a signaling pathway that is characteristic of the GLP-1R signaling pathway (23, 24), including increased phosphorylation of CREB, protein kinase B, protein kinase Cζ, and ERK1/2 within 1–5 min (Fig. 1b). These data suggest the capacity of exendin-4 to activate a signal transduction pathway in liver cells, which resembles that in pancreatic beta cells.

FIGURE 1.

GLP-1R agonist, exendin-4, activates G-protein-coupled receptor and signal transduction pathway in liver cells in vitro. Adult human liver cells were cultured with exendin-4 (5 nm) for the indicated times (0–60 min). a, intracellular cAMP levels were measured by radioimmunoassay. Results represent mean ± S.D. n = 8 in two different experiments. b, cell lysates were immunoblotted using anti-phospho-CREB (pCREB), anti-phosphoprotein kinase B (pPKB), anti-phospho-ERK1/2 (pERK1/2), and anti-phosphoprotein kinase Cζ (pPKCζ). Anti-β-actin serves as protein load control. Representative results are shown (n = 3). *, p < 0.05.

Exendin-4 Enhances PDX-1-induced Activation of Beta Cell Phenotype in Liver Cells

Ectopic expression of PDX-1 in adult human liver cells resulted in the activation of pancreatic lineage and function (3, 5). Moreover, we reported that soluble factors, such as SF, promote the transdifferentiation process induced by PDX-1 (3). To evaluate the effect of exendin-4 on the transdifferentiation process, we analyzed the ability of exendin-4 to activate ectopic insulin and glucagon promoters driving the expression of green fluorescent protein or red fluorescent protein, respectively (Ad-RIP-GFP, recombinant adenoviruses expressing green fluorescent protein under the control of rat insulin-1 promoter; Ad-GLUP-RED, recombinant adenoviruses expressing red fluorescent protein under the control of rat glucagon promoter).

FACS analyses revealed that exendin-4 treatment per se does not activate insulin or glucagon promoters in the liver cells (Fig. 2). However, exendin-4 distinctly affects the insulin and the glucagon promoters both activated by PDX-1 (Fig. 2). Exendin-4 increases the number of insulin promoter-activating cells (GFP positive cells) (Fig. 2a) and the intensity of this promoter activity (mean fluorescence; Fig. 2b). On the other hand, exendin-4 has only a marginal effect on the PDX-1-induced glucagon promoter activation (Fig. 2, c and d). Moreover, exendin-4 further enhanced the insulin promoter activity over that of PDX-1 and SF treatment alone (Fig. 2b). These data suggest that exendin-4 and SF may promote the effect of PDX-1 on the insulin promoter activity, via pathways that are partially distinct.

FIGURE 2.

Exendin-4 increases the number of insulin-expressing cells and insulin gene expression in PDX-1-treated liver cells. Adult human liver cells were co-infected by Ad-CMV-PDX-1 (multiplicity of infection 1000) and Ad-RIP-GFP (multiplicity of infection 1000; a and b) or Ad-GLUP-RED (multiplicity of infection 1000; b and c) supplemented with SF and/or exendin-4 (Ex-4; 5 nm). Five days later, the cells were analyzed by FACS for percentage of fluorescent positive cells (a and c) and mean fluorescence intensity (b and d). Cells infected only by Ad-RIP-GFP or Ad-GLUP-RED served as controls. Results are presented mean ± S.D.; n ≥ 9 in three different experiments. *, p < 0.05; **, p < 0.1; significance represents the differences between Ad-CMV-PDX-1 and the other treatments. #, p < 0.05; significance represents the differences between Ad-CMV-PDX-1 and control cells. e, quantitative real-time reverse transcription-PCR analyses for insulin, glucagon, and somatostatin gene expression levels. The results are normalized to β-actin gene expression within the same cDNA sample and are presented as the relative levels of the mean ± S.D. versus cells treated with Ad-CMV-PDX-1 and SF. (Results are presented as the relative levels compared with Ad-CMV-PDX-1 and SF, because the expression of the pancreatic hormones examined was not detected in Ad-CMV-GFP- and exendin-4-treated cells). n ≥ 6 in three independent experiments. *, p < 0.05.

In correlation with the activation of the ectopic promoters, exendin-4 promotes the activation of insulin gene expression more than 150-fold compared with PDX-1 and SF treatment alone, whereas glucagon gene expression increases only by 2.5-fold (Fig. 2e) as measured by quantitative real-time PCR. Taken together, these data suggest that exendin-4 may predominantly promote beta cell differentiation in PDX-1- and SF-treated liver cells.

Exendin-4 Promotes Beta Cell-like Maturation of Transdifferentiated Liver Cells

Functional pancreatic beta cells are characterized by their ability to secrete mature, processed insulin in response to elevated glucose concentrations. The ability to express, produce, process, and secrete insulin requires the activation of numerous genes within the same cell. Therefore, we analyzed the effect of exendin-4 on the expression of factors instrumental to beta cell-specific function.

In addition to the 150-fold increase in the insulin gene expression (Fig. 2), exendin-4 enhanced the PDX-1 and SF effect on the activation of PC2, PC1/3, SCG2, and SGNE1 (Fig. 3, a and b) expression, which are engaged in prohormone processing and granule assembly, respectively. The increase in the expression of pancreatic marker is partially explained by the capacity of exendin-4 to increase expression of pancreatic transcription factor genes, such as endogenous PDX-1, NEUROD1, ISL-1, NKX2.2, and modestly MAFA (Fig. 3, c and d). The expression of NKX6.1 and PAX4 genes was activated in cells treated with PDX-1 and SF; however, exendin-4 did not further increase the expression of these genes (Fig. 3d). Along with the modest increase in glucagon gene expression (Fig. 2), the expression of alpha cell-specific transcription factors, such as ARX, BRAIN4, and PAX6, was not affected by exendin-4 treatment of PDX-1 and SF-treated cells (data not shown). The robust activation of insulin but not other pancreatic hormones, taken together with the profile of pancreatic transcription factor gene expression induced by exendin-4, further strengthens the notion that exendin-4 may preferentially promote beta cell lineage activation without affecting alpha cell differentiation of PDX-1-treated liver cells.

FIGURE 3.

Exendin-4 promotes beta cell maturation in PDX-1-induced transdifferentiated liver cells. Shown is quantitative reverse transcription-PCR analysis of SGNE1 and SCG2 (a); PC2, PC1/3, and GLUT2 (b); endogenous hPDX-1, NEUROD1, and isl1 (c); and NKX6.1, PAX4, NKX2.2, and MAFA (d) gene expression levels. The results are normalized to β-actin gene expression within the same cDNA sample and are presented as the relative levels of the mean ± S.D. versus cells treated with the control virus Ad-CMV-GFP. n ≥ 10 in three independent experiments. Insulin (e) and C-peptide (f) secretion were measured by incubation for 15 min at 2 and 17.5 mm glucose in KRB. n ≥ 10 in three independent experiments. *, p < 0.05; significance represents the differences between Ad-CMV-PDX-1 and SF and Ad-CMV-PDX-1, SF, and exendin-4 treatments. The significance of Ad-CMV-PDX-1 and SF effect in comparison with Ad-CMV-GFP treatment was already published (3).

The promoting effect of exendin-4 on the maturation of PDX-1-induced transdifferentiated liver cells along the beta cell lineage is further manifested by increased insulin and C-peptide secretion in response to glucose; exendin-4 increases both the total amount and the glucose stimulated insulin and C-peptide secretion at 17.5 mm by 80%. No significant levels of insulin or C-peptide secretion were detected at low glucose concentrations (2 mm; Fig. 3, e and f). Exendin-4 alone has no effect on pancreatic gene expression activation or insulin secretion in liver cells (Fig. 3). Taken together, our data suggest that exendin-4 increases PDX-1-induced transdifferentiated liver cell maturation preferentially along the beta cell lineage.

The effect of exendin-4 on promoting the activation of the pancreatic beta cell lineage and function cannot be explained by increasing transdifferentiated cell proliferation, because we could not detect an increase in insulin and Ki67 (Fig. 4, c–f) or BrdUrd co-localization (1% co-localization of insulin and BrdUrd; data not shown) when exendin-4 treatment was preceded by PDX-1 ectopic expression. Moreover, the same treatment did not increase the number of cells capable of activating the insulin promoter (Fig. 4a compared with Fig. 2a) but markedly increased glucose-regulated C-peptide secretion (Fig. 4b), further strengthening the notion that when applied together or after the molecular manipulation (Ad-CMV-PDX-1), exendin-4 promotes the developmental redirection process by possibly increasing transdifferentiated cells maturation along the pancreatic lineage.

FIGURE 4.

Exendin-4 promotes beta cell maturation of PDX-1 induced transdifferentiated liver cells without inducing cell proliferation. Adult human liver cells were treated with exendin-4 subsequent to viral infection with Ad-CMV-PDX-1, Ad-RIP-GFP, and SF. a, cells were analyzed by FACS for percentage of fluorescent positive cells. Results are presented as mean ± S.D.; n ≥ 6 in two different experiments. b, C-peptide secretion was measured by incubation for 15 min at 2 and 17.5 mm glucose in KRB. n ≥ 10 in four independent experiments. *, p < 0.05; significance represents the differences between Ad-CMV-PDX-1 and SF and Ad-CMV-PDX-1, SF, and exendin-4 (Ex-4) treatments. The significance of the Ad-CMV-PDX-1 and SF effect in comparison with Ad-CMV-GFP treatment was already published (3). Shown are double immunofluorescence analyses for 4′,6-diamidino-2-phenylindole (DAPI; blue) (c), Ki67 (green) (d), and insulin (red) (e) of cells treated with exendin-4 for 2 days subsequent to PDX-1 treatment as well as a merged image (f). The arrows indicate Ki67-positive cells. Original magnification was ×20.

Exendin-4 Promotes Liver Cell Proliferation

Because our above data demonstrate that exendin-4 and PDX-1 co-treatment increases the number of transdifferentiated cells (RIP-GFP-positive; Fig. 2a) without increasing the proliferation of these cells (Fig. 4), we sought to analyze whether exendin-4 affects liver cell proliferation prior to the PDX-1 treatment. We supplemented the growth media of liver cells by exendin-4 (5 nm) without treating the cells with PDX-1. Indeed, exendin-4 treatment reduced the doubling time of adult human liver cells from 4.55 ± 1.78 to 3.4 ± 1.3 days (p ≤ 0.02). The increased number of cells was also documented by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (Fig. 5a), and the accelerated rate of cells proliferation was associated by increases in cyclin D1 levels, decreases in tumor suppressor p53 levels, and changes in the cleavage of the apoptosis-associated protein poly(ADP-ribose) polymerase, as indicated by Western blot analyses (Fig. 5b). Taken together, these data suggest that exendin-4 not only promotes liver cell proliferation but may also affect the survival of these cells (23, 26).

FIGURE 5.

Exendin-4 promotes liver cell proliferation and survival. Adult human liver cells were supplemented with exendin-4. a, viability was assessed over 5 days using a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. Results represent mean ± S.D. n = 10 of two different experiments. *, p < 0.005. b, Western blot analyses were preformed 3 days after exendin-4 supplement for total p53, cyclin D1, and poly(ADP-ribose) polymerase (PARP) in control untreated cells (lane 1) and exendin-4-treated cells (lane 2). β-Actin serves as the protein load control. Experiments were repeated three times, band intensities were quantified, and S.D. value was calculated. #, p < 0.05.

Exendin-4 Increases the Number of Liver Cells That Undergo PDX-1-induced Reprogramming toward the Beta Cell Phenotype

To analyze whether the promoting role of exendin-4 on liver cell proliferation is relevant to its capacity to enhance the reprogramming process, we analyzed the effect of exendin-4 pretreatment on the transdifferentiation capacity of PDX-1. We hypothesized that if exendin-4 increases the number of insulin-positive cells (Fig. 2) without increasing transdifferentiated cell proliferation, it may increase the number of liver cells that are predisposed to undergo the process once treated by PDX-1.

Therefore, we temporally separated the exendin-4 treatment from that of Ad-CMV-PDX-1 viral infection. Five days prior to the PDX-1 and SF treatment, cells were pretreated by exendin-4 (5 nm), which was supplemented to the growth media. As demonstrated in Fig. 6, when exendin-4 treatment preceded the transdifferentiation process (Ad-CMV-PDX-1 infection), both the number of cells capable of activating ectopic insulin promoter and the insulin secretion substantially increase. To analyze whether increased liver cell proliferation due to exendin-4 treatment is responsible for the increase in the number of insulin-producing cells activated by ectopic PDX-1 expression, we labeled liver cells with BrdUrd 2 days after exendin-4 treatment but 3 days before the molecular manipulation (Ad-CMV-PDX-1 infection). BrdUrd inclusion subsequent to exendin-4 treatment reflects the soluble factor's effect on liver cells proliferation. Moreover, BrdUrd and insulin co-localization may suggest a relationship between exendin-4-activated liver cell proliferation and the efficient induction of the pancreatic lineage by PDX-1. Indeed, our results demonstrate that pretreatment with exendin-4 increases the number of both BrdUrd-positive and insulin-positive cells in comparison with exendin-4-untreated control cells (Fig. 6). Importantly, as opposed to Ki67 expression, which reflects only the “present” state of proliferation, BrdUrd applied subsequent to exendin-4 but prior to PDX-1 treatments irreversibly “tags” cells that proliferate prior and possibly during the transdifferentiation process. If liver cell proliferation is relevant to the transdifferentiation process, it will be manifested by double immunostaining for both BrdUrd and insulin.

FIGURE 6.

Exendin-4 increases the number of liver cells predisposed to undergo the transdifferentiation process induced by PDX-1 by increasing the cells proliferation. Adult human liver cells were treated with exendin-4 prior to viral infection with Ad-CMV-PDX-1, Ad-RIP-GFP, and SF. a, FACS analysis for determining the percentage of fluorescent positive cells. Results are presented as mean ± S.D.; n ≥ 6 in two different experiments. b, C-peptide secretion was measured by incubation for 15 min at 2 and 17.5 mm glucose in KRB. n ≥ 10 in four independent experiments. *, p < 0.05; significance represents the differences between Ad-CMV-PDX-1 and SF and Ad-CMV-PDX-1, SF, and exendin-4 (Ex-4) treatments. The significance of Ad-CMV-PDX-1 and SF effect in comparison with Ad-CMV-GFP treatment was already published (3). Double immunofluorescence analyses are shown for 4′,6-diamidino-2-phenylindole (DAPI; blue) (c and g), insulin (green) (d and h), and BrdUrd (red) (e and i). c–f, control cells treated with PDX-1; g–j, cells treated with exendin-4 for 5 days prior to PDX-1 treatment. Original magnification was ×40. k, 800 cells were analyzed under the fluorescent microscope by staining for BrdUrd and insulin. The percentage of cells double-positive for BrdUrd and insulin was calculated for 50 cells positive for insulin staining.

Double immunofluorescence analyses revealed that in controls (PDX-1- and SF-treated), 6.6% of the cells are insulin-positive and 9% are BrdUrd-positive (Fig. 6). Pretreatment with exendin-4 increases the percentage of both insulin- and BrdUrd-positive cells to 10 and 14%, respectively (Fig. 6). Interestingly, although only 17% of insulin-positive cells were also positive to BrdUrd in control cells, 72% of the insulin-positive cells were labeled also by BrdUrd upon exendin-4 pretreatment.

The percentages of both BrdUrd and insulin-positive cells were raised by about 40% upon exendin-4 treatment. However, what is striking is the increase in the percentage of cells that were positively co-labeled by both insulin and BrdUrd, which increased from 17 to 72%, reflecting a 423% increase. These data may suggest that exendin-4 treatment increases the number of cells that are predisposed to undergo the developmental redirection process.

Taken together, our data suggest that exendin-4 has a dual role in the reprogramming process of liver to pancreas, as manifested in Fig. 7; when applied together with or after the PDX-1 treatment, exendin-4 promoted C-peptide secretion by 73%. However, when applied before the reprogramming process, the total number of cells increased, and the amount of processed insulin produced by an initial similar number of liver cells was increased by 217% compared with control liver cells treated only by PDX-1 and soluble factors. It seems that exendin-4 affects the transdifferentiation process 2-fold; it increases both the number and the maturation of transdifferentiated liver cells, depending on the stage of its administration.

FIGURE 7.

Schematic illustration of the experimental course. a, the experiments were temporally divided into two stages; the growth factors, molecular manipulation, and duration of each stage are shown. At day 10 of the experiments, C-peptide secretion was measured. b, quantification of C-peptide secretion levels is normalized to the total number of liver cells in the different treatments.

Exendin-4 Increases the Maturation of PDX-1-induced Transdifferentiated Liver Cells but Not the Proliferation of These Cells

To further analyze the potential additive roles of exendin-4 in increasing the efficiency of PDX-1-induced transdifferentiation, we pretreated liver cells by exendin-4 (as in Fig. 7B) and added the soluble factor also together with PDX-1 (as in Fig. 7E). Fig. 7F demonstrates a further increase in glucose-regulated C-peptide secretion under this treatment, which is about 2-fold increased compared with the combined effect of PDX-1 and exendin-4 (Fig. 7E) and about 3-fold increased compared with PDX-1 and soluble factors (Fig. 7A). We analyzed whether this treatment is associated with an increase in adult beta cell markers at the molecular level. Expression of all of the adult beta cell markers was increased in cells pretreated by exendin-4 prior to the molecular manipulation compared with cells treated as in Fig. 7E, possibly reflecting the activation of the transdifferentiation in the increased number of predisposed liver cells (Fig. 6K). However, when exendin-4-pretreated cells were treated also by PDX-1 and exendin-4, whereas the expression of insulin and glucokinase genes expression further increased (Fig. 8a), the expression of PC1/3 and SGNE decreased (Fig. 8a). It is not yet clear whether the decrease is related to the role of exendin-4 in increasing liver cell proliferation. Regardless, the level of expression of these factors was similar to that of cells co-treated with PDX-1 and exendin-4. Co-localization of PDX-1, insulin, and additional beta cell-specific transcription factors was analyzed by immunofluorescence. Interestingly, exendin-4 pretreatment combined with PDX-1 and exendin-4 co-treatment increased the number of insulin- and beta cell-specific transcription factor-positive cells to about 20–30% of the cells in culture, depending on the pancreatic marker analyzed (Fig. 8b), compared with only 10% insulin-positive cells upon exendin-4 pretreatment followed by PDX-1 alone (Fig. 6k). Importantly, all of the insulin-positive cells co-localize with the ectopic PDX-1 and the other induced pancreatic transcription factors, such as NKX6.1. For an as yet unknown reason, PAX4 was expressed only in 50% of insulin-positive cells, whereas MAFA exhibited a leaky expression in numerous liver cells (Fig. 8b). To analyze whether the 2–3-fold increase in the number of insulin-positive cells (Figs. 7F and 8b, compared with Fig. 6k) is caused by transdifferentiated cell proliferation, we labeled the exendin-4-pretreated cells 5 days after the PDX-1 and exendin-4 co-treatment by BrdUrd or Ki67 and analyzed the co-localization of these proliferation markers with insulin. Only 2 and 1% of the insulin-positive cells, respectively, were positive also to the proliferation markers, which at this time point reflected the effect of exendin-4 on transdifferentiated cell proliferation. These data suggest that the substantial increase in the number of insulin-positive cells is not attributable to transdifferentiated cell proliferation but rather to the maturation of these cells. The increase in the number of insulin-positive cells in pre- and co-treatment with exendin-4 compared with pretreatment alone most likely reflects increased insulin expression (Fig. 8a) and production (Figs. 7F and 8b), which may allow a more efficient detection of the hormone by immunofluorescence (Fig. 8b). Taken together, our data demonstrate a significant role for exendin-4 in increasing the efficiency of the PDX-1-induced transdifferentiation of liver to pancreatic beta cells, by both increasing predisposed liver cell proliferation and transdifferentiated cell maturation along the beta cell lineage and function.

FIGURE 8.

The combined effect of exendin-4 treatment on PDX-1 induced transdifferentiation of liver to endocrine pancreas. Adult human liver cells were treated with 5 nm exendin-4 (Ex-4) prior to or during the viral infection with Ad-CMV-PDX-1 and SF. a, quantitative reverse transcription-PCR analysis of insulin, GLUT2, glucokinase (GK), SGNE, and PC1/3 gene expression levels. The results are normalized to β-actin gene expression within the same cDNA sample and are presented as the relative levels of the mean ± S.D. compared with cells treated with PDX-1, SF, and exendin-4. n ≥ 4 in two independent experiments. *, p < 0.05. b, double immunofluorescence analyses of PDX-1 (green) and insulin (red) (I), PDX-1 (green) and PAX4 (red) (II), PDX-1 (green) and NKX6.1 (red) (III), MAFA (green) and insulin (red) (IV), PAX4 (green) and insulin (red) (V), and NKX6.1 (green) and insulin (red) (VI) of cells treated with exendin-4 for 5 days before and 5 days subsequent to PDX-1 + SF treatment. *, merge figures of control virus-treated cells (treated with exendin-4 for 5 days prior and subsequent to Ad-β-galactosidase + SF + exendin-4 treatment). The nuclei were stained using 4′,6-diamidino-2-phenylindole (DAPI; blue). Original magnification was ×20.

DISCUSSION

Our study demonstrates a dual role for GLP-1R agonist, exendin-4, in promoting the PDX-1-induced activation of the pancreatic lineage and function in adult human liver cells. The roles of exendin-4 in promoting PDX-1-induced reprogramming of liver to pancreas are manifested by its capacity to increase insulin gene expression 150-fold compared with PDX-1 and SF treatment as well as insulin production and its glucose-regulated secretion (Fig. 2). Exendin-4 specifically augments the beta cell-like phenotype without significantly affecting glucagon or somatostatin gene expression (Fig. 2). The promoting effects that exendin-4 has on the maturation of transdifferentiated liver cells can be explained by its capacity to increase beta cell-specific pancreatic transcription factor expression (Figs. 3 and 8b). By supplementing exendin-4 at distinct time points during the PDX-1-induced developmental switch, we found that it plays a dual role in the process. When introduced to liver cells prior to PDX-1, it increases the proliferation of liver cells capable of undergoing the developmental switch. However, when applied on cells subsequent to PDX-1, it increases transdifferentiated cell maturation without increasing the proliferation of these cells (Figs. 4–8). The relevance of liver cell proliferation induced by exendin-4 to the efficiency of the PDX-1-induced transdifferentiation process is best exemplified by the fact that 72% of insulin-positive cells were proliferating prior to or during the cell fate switch event (Fig. 6). A role for exendin-4 on transdifferentiated cell maturation is best exemplified in Fig. 8b, in which 20–30% of the cells in culture are positive to insulin, which co-localizes with numerous beta cell-specific transcription factors. The fact that only 2% of insulin-positive cells in Fig. 8b were also positive to BrdUrd that was introduced subsequent to PDX-1 treatment suggests that at this time point, exendin-4 promotes the transdifferentiated cells maturation along the beta cell lineage but not the proliferation of these cells. These data may suggest that 10% insulin-positive cells upon exendin-4 pretreatment followed by PDX-1 treatment alone (Fig. 6k) may reflect only transdifferentiated cells producing relatively high levels of insulin. Additional liver cells in culture that acquired only restricted beta cell characteristics and marginal insulin production capacity may remain undetected in the immunofluorescence assay in Fig. 6k. Once exendin-4 is added also together with the molecular manipulation, it increases the insulin production to immunofluorescence-detectable levels.

Interestingly, many of the roles that exendin-4 plays in the process of PDX-1-induced liver to pancreas transdifferentiation resemble its roles in pancreatic beta cells. GLP-1R agonists increase beta cell mass by increasing neogenesis and inhibiting beta cell apoptosis (26). GLP-1R agonists accelerate maturation and differentiation of fetal porcine islets and improve the glucose-stimulated insulin secretion from pancreatic islets (27). Many of these characteristics were recapitulated also in transdifferentiated liver cells, including a substantial increase in the expression of early and beta cell-specific transcription factors (Fig. 3). However, although the GLP-1R agonist accelerates mature pancreatic beta cell proliferation, it does not significantly increase transdifferentiated liver cell proliferation.

It is not clear which intracellular signaling pathways mediate the GLP-1 signals in liver. As in rodents and humans, a single structurally identical GLP-1R has been identified and is expressed in a wide range of tissues, including insulin glucagon- and somatostatin-producing cells in the pancreas, but not in liver, fat, or muscle (20–22). However, numerous studies demonstrate that exendin-4 or GLP-1 affects tissues lacking the GLP-1R, among them liver, skeletal muscle, and fat (20–22). Moreover, a binding assay revealed an association of ¹²⁵I-GLP-1 with hepatocyte plasma membrane, and by cross-linking assays, a 63-kDa protein, similar to “classical” GLP-1R mass, was detected (20–22). Although it was suggested that transdifferentiated HepG2 human hepatoma cell line expresses the pancreatic GLP-1R (28), we could not detect GLP-1R sequences in our human liver cultures either before or after PDX-1 treatment (data not shown), suggesting that GLP-1R may not be included among the many other pancreatic genes, the expression of which is activated in the transdifferentiated liver cells.

Indeed, despite the fact that the “known” GLP-1R was not identified in liver, we demonstrate that treating human liver cells with exendin-4 activates a signaling cascade characteristic of G-protein-coupled receptors, which is initiated by prompt elevations in intracellular cAMP levels (Fig. 1). Further analyses are needed to identify the receptor that mediates the GLP-1R agonist effect in liver. However, our data are in agreement with previous data demonstrating GLP-1R agonist effects on liver cells in vitro (15, 16). As was demonstrated, exendin-4 treatment activated in liver cells a signal transduction pathway that includes the accumulation of cAMP and increased the phosphorylation of CREB, protein kinase B, protein kinase Cζ, and ERK1/2 within 1–5 min (Fig. 1). This pathway was suggested to mediate the GLP-1 signal transduction in beta cells (23, 24).

In pancreatic islets, GLP-1 represses glucagon gene expression in alpha cells. In contrast, in transdifferentiated liver cells, exendin-4 does not clearly affect alpha cell differentiation at all, but it shifts the PDX-1-induced reprogramming process preferentially toward insulin-producing cells. This fact is manifested by beta cell-like maturation, increased insulin production, and its glucose regulated secretion (Fig. 3). The preferential enhancement of beta cell differentiation and function is explained by specific increases in transcription factor expression characteristic of beta cells but not alpha cells (Figs. 2 and 3).

When PDX-1 treatment is applied on liver cells pretreated with exendin-4, the number of transdifferentiating cells (positively stained to insulin) increases by 36%. Interestingly, 72% of the insulin-positive cells were also positive to BrdUrd (Fig. 6). The fact that most of the transdifferentiating cells co-localize with the proliferation marker may suggest that either exendin-4 preferentially induces the proliferation of liver cells predisposed to undergo the PDX-1-activated transdifferentiation process or that the proliferation step by itself, which is associated with a decrease in p53 expression (Fig. 5), could promote the transdifferentiation process.

Increased proliferation may augment liver cell dedifferentiation and possibly the plasticity of these cells, thus increasing the capacity of these cells to acquire the new developmental fate. However, we ruled out the possibility that the exendin-4 treatment increased hepatic dedifferentiation, because it did not decrease adult hepatic marker expression or increase immature markers (5) (data not shown).

Increased cell proliferation could contribute in an additional mode to the increased rate of the developmental switch induced by PDX-1 in liver. Because the reprogramming process is considered a wide epigenetic event, which involves the activation of otherwise silent genetic information, cells that are at S phase or during mitosis contain a “loose” chromatin compaction, compared with non-dividing cells. This change in the chromatin compaction may allow a better accessibility of PDX-1 to the relevant pancreatic promoters, which can be more efficiently activated.

Downstream target gene accessibility is not the only limiting factor that controls the activation of otherwise silent pancreatic genetic information in the liver; additional factors, such as chromatin demethylation or acetylation, may also play important roles in the efficiency of the epigenetic process. Indeed, PDX-1 has been demonstrated to modulate chromatin structure by increasing histone deacetylase activity and decreasing chromatin methylation (29–31). Taken together, our data suggest a dual promoting effect of GLP-1R agonist in promoting the PDX-1-induced liver reprogramming preferentially toward the beta cell lineage and function, in a process that involves accelerated predisposed liver cell proliferation and transdifferentiated cell maturation along the beta cell lineage.

Developing means to efficiently induce adult cell reprogramming will allow the generation of autologous pancreatic beta cells to be used in cell replacement therapy for diabetic patients. In this process, the diabetic patient can serve also as the donor of his own therapeutic tissue, alleviating both the shortage in tissue availability from cadaver donors and the need for life-long suppression of the immune system.