Mechanisms of action of islet neogenesis-associated protein: comparison of the full-length recombinant protein and a bioactive peptide

Abstract

Islet neogenesis-associated protein (INGAP) was discovered in the partially duct-obstructed hamster pancreas as a factor inducing formation of new duct-associated islets. A bioactive portion of INGAP, INGAP^104–118 peptide (INGAP-P), has been shown to have neogenic and insulin-potentiating activity in numerous studies, including recent phase 2 clinical trials that demonstrated improved glucose homeostasis in both type 1 and type 2 diabetic patients. Aiming to improve INGAP-P efficacy and to understand its mechanism of action, we cloned the full-length protein (rINGAP) and compared the signaling events induced by the protein and the peptide in RIN-m5F cells that respond to INGAP with an increase in proliferation. Here, we show that, although both rINGAP and INGAP-P signal via the Ras/Raf/ERK pathway, rINGAP is at least 100 times more efficient on a molar basis than INGAP-P. For either ligand, ERK1/2 activation appears to be pertussis toxin sensitive, suggesting involvement of a G protein-coupled receptor(s). However, there are clear differences between the peptide and the protein in interactions with the cell surface and in the downstream signaling. We demonstrate that fluorescent-labeled rINGAP is characterized by clustering on the membrane and by slow internalization (≤5 h), whereas INGAP-P does not cluster and is internalized within minutes. Signaling by rINGAP appears to involve Src, in contrast to INGAP-P, which appears to activate Akt in addition to the Ras/Raf/ERK1/2 pathway. Thus our data suggest that interactions of INGAP with the cell surface are important to consider for further development of INGAP as a pharmacotherapy for diabetes.

regeneration of β-cells in diabetic patients is an important goal of diabetes research. In recent years, there has been increasing interest in the development of new strategies to induce β-cell regeneration and new islet formation in situ (3, 28, 40). Therefore, identification of bioactive molecules with the capacity to stimulate expansion of the remaining β-cell mass or with islet neogenic activity is crucial for harnessing the regenerative potential of the native pancreas.

Islet neogenesis-associated protein (INGAP) is a 16.8-kDa protein originally identified in a crude extract from a partially obstructed hamster pancreas (42). INGAP is expressed in the pancreas and duodenum (22, 39, 41) and has been shown to induce islet neogenesis in several species (43, 44). Structurally, INGAP is a member of the Reg family of secreted C-type lectins that comprises more than 25 members and is classified into four subfamilies based on the primary sequence (33, 58). INGAP belongs to the large Reg3 subfamily that has been identified predominantly in gastrointestinal tissues (pancreas, stomach, and liver) in rat, mouse, hamster, and humans (20, 39, 53, 56). Despite the ubiquity of Reg proteins, not much is known about their functions or the mechanisms of action. Although there is a consensus on the role of Reg1 as a β-cell mitogen (33, 52, 53, 56), much less is known about the functions of the Reg3 family. Expression of several Reg proteins, including INGAP, is influenced by a number of inflammatory cytokines (1, 19, 20, 36) or bacterial infections (32), thus implicating these proteins in the acute-phase response. Consistent with being C-type lectins, two members of the Reg3 family, α and γ, have been shown to bind carbohydrate ligands such as mannan and peptidoglycans and to have direct antimicrobial effects (10, 11). Human Reg3α has also been shown to bind lactose and extracellular matrix proteins and to induce adhesion of hepatocytes in vitro (12, 13). Although it remains to be determined whether other members of the family are able to act similarly, a number of studies suggest that Regs may bind specific cell surface receptors (perhaps in addition to or as an alternative to binding carbohydrates) and activate multiple signaling pathways (6, 23, 26, 50). To date, the only known Reg receptor (EXTL3) that specifically binds Reg1 is a transmembrane glycosyltransferase homologous to the multiple exostoses-like gene family (26). It is not known whether this receptor binds INGAP or other Reg proteins or whether a different receptor (if any) is involved in the mechanism of INGAP action.

One argument in favor of the receptor hypothesis is that the biological activity of INGAP appears to be mediated by a 15-amino acid fragment of the protein (AA 104–118), namely INGAP peptide (INGAP-P), which consists of a highly conserved IGLHDP amino acid motif and a unique sequence SHGTLPNGS not found in the other members of the Reg family (39). Synthetic INGAP peptide has been demonstrated to be as effective as the protein in inducing new islet formation and reversing streptozotocin-induced diabetes in hamsters and mice (43, 44), and therefore, it is a possible ligand for the receptor. Biological effects of a synthetic INGAP-P have been studied extensively both in vitro and in vivo in our laboratory and by others. To date, the most compelling data show that INGAP-P 1) induces in vitro regeneration of functional human islets from dedifferentiated, islet-derived, duct-like structures (25), 2) dose-dependently stimulates expansion of β-cell mass in rodents, dogs, and cynomolgus monkeys (29, 38, 43), and 3) increases insulin secretion and β-cell size and upregulates the expression of several genes related to β-cell function in rat neonatal islets in vitro (4, 5, 7). These important results were followed by clinical trials to investigate its efficacy and safety in humans, in which INGAP-P was found to have a signal effect, with an improvement of glucose homeostasis confirmed by A1C reduction at 90 days in patients with type 2 diabetes and by a significant increase in C-peptide secretion in patients with type 1 diabetes (18, 54).

Taken together, these data support the potential of INGAP as a candidate agent for the treatment of diabetes that possesses both islet-neogenic and insulinotropic activities. However, the relatively short plasma half-life of INGAP-P and the need for administration in a high dose calls for improvement of the drug profile (18). In this context, characterization of the mechanism of action becomes an important priority for further research.

For this reason, we have recently cloned a full-length recombinant protein [recombinant INGAP (rINGAP)] that is much more stable than the peptide (≥5 days in cell culture) and is 6-His-tagged (2). To provide a detailed, side-by-side comparison of the biological effects of INGAP-P and the full-length protein, we chose a well-characterized β-cell line, RIN-m5F (14), that responds to INGAP-P by an increase in proliferation (34). Our data show that rINGAP is 100 times more potent on a molar basis than INGAP-P and that, although they both signal via activation of the Ras/Raf/ERK pathway, the upstream signaling events may differ between the protein and peptide. Using confocal microscopy, we found several differences between the protein and peptide in cell binding and internalization. We show that fluorescent-labeled rINGAP forms patches on the cell surface in a fashion consistent with receptor binding and clustering, whereas INGAP-P rapidly internalizes. INGAP-induced activation of ERK1/2 is reduced significantly by pertussis toxin (for both protein and peptide), thus suggesting that, despite differences in cell binding, both INGAPs act via a G protein-coupled receptor (GPCR).

MATERIALS AND METHODS

rINGAP and INGAP-P.

INGAP-P, a 15-amino acid fragment of INGAP protein (AA 104–118; MW 1,501.6), was synthesized and HPLC-purified at the Sheldon Biotechnology Centre (McGill University, Montreal, QC, Canada). A full-length recombinant rINGAP containing COOH-terminal 6-His tag (MW 17.6 kDa) was cloned from hamster pancreatic tissue by directional cloning of the PCR product generated with Superscript III RT and Platinum Pfx DNA Polymerase (Invitrogen) into pcDNA3.1D/V5-His-TOPO expression vector (Invitrogen). This construct was used for recloning into a lentiviral vector and expression in H293 cells (described in detail in Ref. 2). Purification of rINGAP was carried out using Cobalt resin [BD Talon (BD Biosciences) or Fractogel EMD Chelate (M) (Merck)], as described (2).

Cell culture.

RIN-m5F cells (passage 18) were purchased from ATCC and maintained at 37°C/5% CO₂ in RPMI-1640 medium (Invitrogen) containing 25 mM glucose, 10% FBS (Montreal Biotech), and antibiotics/antimycotics (Invitrogen). The following experiments were carried out on cells from passages 25–31. Cells were plated in 60-mm TC dishes (1 × 10⁶ cells/dish) and allowed to grow for 24–48 h, followed by the serum withdrawal for 24 h prior to the treatment. INGAP-P, rINGAP, epidermal growth factor (EGF; 10 ng/ml; Sigma), or exendin-4 (Ex-4; 10 nM) (Bachem) were administered in serum-free medium for the times indicated. For the analysis of signaling pathways, cells were pretreated for 40 min with the following inhibitors (all from Calbiochem; concentrations were calculated to exceed IC₅₀ 10–20 times or as suggested by the manufacturer): 100 nM wortmannin [phosphatidylinositol 3-kinase (PI3K) pathway], 10 μM PD-98059 (MEK), 100 nM AG1478 [EGF receptor (EGFR)], 100 nM PP2 (Src), 100 nM Raf kinase inhibitor 1 (c-Raf), 1 μM H-89 (PKA), 100 nM PKA inhibitor 14–22 amide (PKi), 250 μM SQ-22536 (adenylate cyclase), 1 μM bisindolylmaleimide I (Bis; PKC), and 100 nM SB-203580 (p38). To inhibit GPCR signaling, cells were pretreated with 100 ng/ml pertussis toxin (Ptx; Enzo Life Sciences, Plymouth Meeting, PA) for 24 h.

Assessment of cell proliferation by bromodeoxyuridine immunostaining.

Cells plated in eight-well or four-well chamber slides (5 × 10⁴ or 1 × 10⁵ cells/well) were treated with INGAP, EGF, or Ex-4 for 24 h, as described above, and 50 μM bromodeoxyuridine (BrdU) was added during the last 3 h of treatment. Cells were washed with PBS and fixed in methanol for 10 min at −20°C. Immunostaining for BrdU was carried out using mouse anti-BrdU antibody (Roche), following the manufacturer's protocol. This was followed by detection with secondary horseradish peroxidase (HRP)-conjugated antibody (broad spectrum; Histostain-Plus) and AEC chromogen (both from Zymed Laboratories). Slides were counterstained with hematoxylin. BrdU-positive and -negative nuclei were counted (total of 200/well), and the percentage of BrdU-positive nuclei was calculated.

Western blot analysis.

Following treatments, cells were placed on ice, washed with PBS, and solubilized in lysis buffer (Cell Signaling Technology, Beverly, MA) containing 2.5 mM Na₄P₂O₇, 1 mM Na₃VO₄, and Complete protease inhibitor cocktail tablet (Roche). Equal amounts of protein [20–50 μg, measured with DC Protein assay (Bio-Rad)] were resolved by 10% SDS-PAGE, followed by transfer onto Nitrocellulose membrane (Bio-Rad) at 250 mA for 90 min, and analyzed with different antibodies. Anti-ERK1/2 (MAPK 44/42) and anti-phospho-ERK1/2 (Thr²⁰²/Tyr²⁰⁴), anti-p38 MAPK and anti-phospho-p38 MAPK (Thr¹⁸⁰/Tyr¹⁸²), anti-C-Raf and anti-phospho-C-Raf (Ser³³⁸), anti-phospho-(pan)-PKC (γ-Thr⁵¹⁴), and rabbit polyclonal antibodies were purchased from Cell Signaling Technology. Following primary antibody incubation, blots were washed and then incubated in a secondary anti-mouse or anti-rabbit HRP-conjugated antibody (Cell Signaling Technology), washed, and developed using the ECL system (GE Healthcare). To analyze expression of several proteins on the same blot, membranes were first incubated with phosphoantibodies, followed by stripping (0.2 M glycine, 0.1% SDS, and 0.05% Tween-20, pH 2.2) prior to probing with corresponding non-phospho-primary antibodies.

Ras activation and phospho-Akt (Ser⁴⁷³) ELISA.

Ras-GTP and Akt activation in INGAP-stimulated RIN-m5F cells were analyzed using Ras activation and phospho-Akt (Ser⁴⁷³) ELISA kits purchased from Millipore; 1 × 10⁶ cells were plated in 60-mm plates for 48 h, followed by a 24-h starvation in serum-free medium. Cells were treated with the growth factors at 37°C for the times indicated. Plates were then placed on ice and washed with ice-cold PBS prior to cell lysis in 150 μl of Mg+ lysis buffer containing a cocktail of protease inhibitors (New England Biolabs). Ten microliters of cell lysates was used for Ras-GTP ELISA, and the readings were normalized by the amounts of protein (DC protein assay; Bio-Rad).

PKC and PKA kinase activity assay.

Cell lysates prepared as for Western blots and containing 5–10 μg/sample of crude protein were assayed by ELISA using PKA and PKC kinase activity assay kits purchased from Assay Design (Ann Arbor, MI) according to the supplied protocol and normalized to total protein.

Visualization of fluorescent rINGAP and INGAP-P.

One hundred micrograms of rINGAP was labeled with DyLight-488 or DyLight-594 (ThermoScientific) as specified in the instructions. INGAP-P was labeled with either 5-carboxyfluorescein (5-FAM) or FITC during the synthesis at the Sheldon Biotechnology Centre (McGill University) or Canpeptide (Ponte Clair, QC, Canada). Fluorescent rINGAP (50 nM) or INGAP-P (8.35–16.7 μM) was added to RIN-m5F cells grown in glass chamber slides (Beckton-Dickinson or Lab-Tek) for various intervals, followed by washing with PBS and fixation in 4% paraformaldehyde. Slides were mounted using VectaShield medium (Vector) or Prolong Gold (Invitrogen) with DAPI for counterstaining of nuclei and examined under confocal microscope Zeiss LSM 510 or Olympus FV10i. For live confocal imaging, cells were grown in Nunc chambered coverglass slides (ThermoScientific). Nuclei were stained with 0.01% DAPI prior to incubation with INGAP, followed by washing. Live imaging was carried out at 37°C and 5% CO₂.

To study INGAP internalization, cholera toxin subunit B (CTB; Alexa fluor 594, 5 μg/ml) transferrin (25 μg/ml, Texas Red) and LysoTracker Red DND99 (50 nM; all from Invitrogen) were used in comigration assays with DyLight-488 rINGAP and FAM-INGAP-P. To assess INGAP colocalization with early endosomes, fixed cells were permeabilized with 0.1% Triton × 100 for 10 min, blocked in 5% goat serum, and probed with anti-early endosome antigen 1 (EEA1) rabbit primary antibody (1:200; Abcam) overnight at 4°C, followed by the secondary donkey anti-rabbit DyLight594-conjugated antibody (1:500) for 1 h at room temperature. Cells incubated with DyLight-594-labeled rINGAP were probed with anti-clathrin and anti-caveolin rabbit antibodies (Abcam), followed by FITC-labeled goat anti-rabbit secondary antibody (Abcam). Nuclei were counterstained with DAPI included in the mounting medium (Prolong Gold; Invitrogen). The following inhibitors of endocytosis were used: dansylcadaverine (100–300 μM), filipin (1 μg/ml), cytochalasin D (25 μg/ml) (all from Sigma), and wortmannin (100 nM; Calbiochem).

Statistical analysis.

Experiments were repeated at least three times. Results are expressed as means ± SE. Since most of our data were normalized to control values prior to calculation of fold change, we could not assume that the data follow Gaussian distributions. Therefore, for statistical analyis, we employed the nonparametric Kruskal-Wallis test, followed by Wilcoxon rank-sum test for pairwise comparison, using the SPSS software (IBM). A P value of <0.05 was considered significant.

RESULTS

INGAP-P and rINGAP dose-dependently increase proliferation of RIN-m5F cells, but with different molar efficiencies.

Although pancreatic ductal cells have been understood to be a particular target of INGAP (38, 42), a number of studies, including the results of clinical trials, suggest that β-cells are also responsive to INGAP stimulation in a number of ways, including potentiation of glucose-stimulated insulin secretion and upregulation of the corresponding genes, as well as increase in cell viability and proliferation (1, 4, 7, 18, 34, 49, 57). There was no significant effect on insulin expression in our experiments on RIN-m5F cells, but we observed that both INGAP-P and rINGAP dose-dependently induced BrdU incorporation after 24 h (Fig. 1A). The effect of rINGAP appeared to plateau at 1 nM (×1.64 increase), and so this dose was used throughout the study, whereas INGAP-P was used at 835 nM (×1.72 increase in BrdU). Similar increases in proliferation were observed for EGF (10 ng/ml) and Ex-4 (10 nM), which were used as positive controls (Fig. 1A) (17).

Fig. 1.

Effect of islet neogenesis-associated protein (INGAP) on proliferation in RIN-m5F cells. A: INGAP increases bromodeoxyuridine (BrdU) incorporation. RIN-m5F cells were treated with indicated amounts of INGAP peptide (INGAP-P) or recombinant INGAP (rINGAP) for 24 h in chamber slides. Exendin-4 (Ex-4) and epidermal growth factor (EGF) were used as positive controls. Each dose was compared with the control, and the data were analyzed as a ratio of BrdU(+) cells (%) in INGAP-treated to untreated control. Shown are means ± SE of at least 3 independent experiments (*P < 0.05 and †P < 0.01 compared with untreated control; Kruskal-Wallis and Wilcoxon rank-sum test). B: INGAP induces phosphorylation of ERK1/2 in RIN-m5F. RIN-m5F cells (1 × 10⁶) plated in 60-mm tissue culture plates were treated with INGAP for the times indicated. Blots (30 μg of protein) were probed with anti-phospho (p)-ERK1/2 (Thr²⁰²/Tyr²⁰⁴) antibody, followed by stripping/reprobing with anti-total ERK1/2 antibody (Cell Signaling Technology), and quantified by densitometry using Image J software. The results of quantification are presented in Fig. 7C.

The increase in BrdU incorporation was consistent with a rapid temporal activation of ERK1/2, which was observed between 1 and 15 min after the addition of either rINGAP or INGAP-P (Fig. 1B; quantification is shown in Fig. 7C). These results show that both protein and peptide act in a similar manner but with different molar efficiencies (≥100-fold). Therefore, we next investigated whether INGAP protein and INGAP-P interact differently with the cell surface and/or activate different signaling pathways.

Fig. 7.

Involvement of Ras/Raf/ERK1/2 activation in signaling events induced by INGAP-P and rINGAP. Cell lysates collected in the time course experiments with rINGAP or INGAP-P were used for multiple analyses by ELISA [Ras-GTP and Akt (Ser⁴⁷³); Millipore] and by Western blot (p-ERK1/2, p-c-Raf). The data are presented as a fold change over the 0-min time point, which equals 1 and is shown as a dotted line in all charts. A: Ras activation was measured by Ras-GTP ELISA, as described in materials and methods (*P < 0.05; †P < 0.01). B: c-Raf phosphorylation, measured by Western blot/densitometry (Image J) as a ratio of p-c-Raf to total c-Raf. The changes were not found to be statistically significant. C: quantification of relative ERK1/2 phosphorylation measured by Western blot/densitometry (Image J) as a ratio of p-ERK1/2 to total ERK1/2 (*P < 0.05 compared with time 0). D: changes in Akt phosphorylation in RIN-m5F cells treated with INGAP-P and rINGAP. E: phosphorylation of ERK1/2 and Akt induced by EGF and Ex-4, quantified as in C and D (*P < 0.05 for p-ERK1/2; §P < 0.05 for p-Akt).

Interaction of rINGAP with the cell surface is characterized by clustering and slow internalization.

Using DyLight 488-labeled rINGAP, we observed binding to the cell surface of RIN-m5F cells within minutes of exposure (Fig. 2). Bound rINGAP forms small clusters and patches on the cell surface resembling the cross-linking of membrane multiprotein complexes described for other ligands (15, 31, 37, 45, 55). This is different from a homogenous staining exhibited by CTB (Alexa fluor 594) and transferrin (Texas Red) (both from Invitrogen), which were used as positive markers for caveolin- and clathrin-mediated endocytosis. rINGAP binding is observed both at 37°C and on ice (Fig. 2, A–D), which is suggestive of a high-affinity receptor.

Fig. 2.

Binding of rINGAP is characterized by capping on the cell surface at 37°C or on ice and by slow internalization. A and B: cells were prechilled on ice for 15 min and incubated with DyLight488-rINGAP and cholera toxin subunit B (CTB; Alexa fluor 594, 5 μg/ml; Invitrogen) (A) or transferrin (25 μg/ml, Texas Red; Invitrogen) (B). C and D: 1-h incubation with cholera toxin subunit B (CTB) and transferrin, respectively, at 37°C. E and F: cells were incubated for 5 or 24 h with labeled rINGAP and costained with 50 nM LysoTracker Red DND99 (LT; Invitrogen) for the last hour. G and H: chase experiment. Cells were incubated with rINGAP for 1 h, followed by washing and a chase period of 5 (G) or 24 h (H) without the presence of labeled INGAP. LT (50 nM) was added 1 h prior to fixation in 4% PFA. I: negative control. Bars, 20 μm. J and K: cells were incubated with DyLight-594-labeled rINGAP for 1 (J) or 15 min (K), fixed, and probed with anti-clathrin and anti-caveolin rabbit antibodies, followed by FITC-labeled goat anti-rabbit secondary antibody. L: no primary antibody control to J and K. Nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI) and included in the mounting medium (Prolong Gold; Invitrogen). Images were taken with Olympus FV10i (A–I) and Zeiss LSM-510 (J–L) confocal microscopes.

Although the first signs of internalization are observed after 15 min, the protein appears to remain clustered on the cell surface for several hours (Fig. 2, C–E), unlike transferrin and CTB, which internalize within 1 h (Fig. 2, C and D). After a 5-h incubation, most of the fluorescent label is seen inside of cells (Fig. 2E) and is colocalized partially with the lysosomal marker LysoTracker red. After 24 h, all labeled rINGAP appears to internalize and associate with lysosomes, albeit partially, but shows no further binding to the cell surface (Fig. 2F).

Interestingly, in the chase experiments, when cells were exposed to DyLight488 rINGAP for only 1 h, followed by washing and culture without rINGAP for 5 or 24 h, the amount of internalized rINGAP was not significantly lower than after continuous incubation (Fig. 2, G and H). This, together with the aforementioned observation, suggests that the available INGAP receptor pool is all ligand bound within 1 h and that the receptor turnover time probably exceeds 24 h.

The lack of comigration between rINGAP and CTB or transferrin suggests that rINGAP is not internalized via either a clathrin- or caveolin-mediated pathway. This is in line with the results of immunostaining for clathrin and caveolin, which showed no colocalization with rINGAP (Fig. 2, J and K). Because of a long internalization time of rINGAP, usage of specific inhibitors of clathrin-mediated (chlorpromazin and dansylcadaverin) or caveolin-mediated endocytosis (filipin and β-methylcyclodextrin), as well as dynasore (dynamin inhibitor), was not practical due to their fast-developing cytotoxicity. We believe that the comigration experiments, as well as the lack of colocalization, are sufficient to rule out clathrin- and caveolin-mediated endocytosis as a main internalization route for rINGAP. Besides, we found that rINGAP internalization is inhibited by wortmannin (inhibitor of fluid-phase pinocytosis and PI3K) and by cytochalasin D (inhibitor of actin polymerization) (Fig. 3, C and D), which is suggestive of macropinocytosis as a major mechanism for rINGAP endocytosis (24). Further comigration experiments with dextran and treatments with other inhibitors, e.g., amiloride (48), could be used to verify this possibility. Of note, no colocalization with the early endosomal marker EEA1 has been observed for rINGAP (Fig. 4I), which may further support the macropinocytosis route of internalization.

Fig. 3.

Binding and internalization of fluorescently labeled rINGAP are inhibited partially by 100 nM wortmannin and cytochalasin D, suggestive of macropinocytosis. RIN-m5F cells plated in chamber slides were incubated for 5 h with 50 nM DyLight 488-rINGAP without inhibitors (A) and in the presence of cytochalasin D (25 μg/ml; B) or wortmannin (100 nM; C). D: negative control. Nuclei were counterstained with DAPI included in the mounting medium (Prolong Gold; Invitrogen). Images were taken with an Olympus FV10i confocal microscope.

Fig. 4.

5-Carboxyfluorescein (FAM)-labeled INGAP-P is internalized rapidly into the cytoplasm of RIN-m5F cells. Cells grown in chamber slides were treated with FAM-labeled INGAP-P for the times indicated and fixed with 4% PFA. A, B, D, and F: cells were stained with Lysotracker for 1 h as in Fig. 2. C and E: fixed cells were costained for early endosome antigen 1 (EEA1) and visualized with secondary donkey anti-rabbit DyLight594-conjugated antibody. Nuclei were counterstained with DAPI included in the mounting medium (Prolong Gold). Images were taken with an Olympus FV10i confocal microscope. G: a continuous incubation with FAM-INGAP for 24 h. H: chase experiment. Cells were incubated with FAM-INGAP for 1 h, followed by washing and a chase period of 24 h, without the presence of labeled INGAP. LysoTracker Red DND99 (50 nM) was added 1 h prior to fixation in 4% PFA. I: no colocalization with EEA1 was observed for rINGAP (2 h). Nuclei were counterstained with DAPI included in the mounting medium (Prolong Gold). Images were taken with an Olympus FV10i confocal microscope.

INGAP-P is rapidly internalized into the cytoplasm without clustering on the cell surface.

In contrast to rINGAP, we did not observe any binding of FAM-labeled INGAP-P to the membrane. However, the labeled peptide was visible in the cytoplasm of RIN-m5F cells after 5 min of incubation, reaching a plateau after 30 min (Fig. 4). As seen in Fig. 4C, INGAP-P appears to colocalize with early endosomes after 30 min and then gradually migrates into the lysosomal compartment, colocalizing with LysoTracker red (Fig. 4, D and F).

Besides differences in the dynamics of cell binding and internalization, some other differences between the protein and peptide have been observed. For example, internalized INGAP-P appears to degrade faster, as shown in 24-h experiments with continuous and “chase” incubations (Fig. 4, G and H). Also, internalization of INGAP-P was inhibited on ice or by preincubation with the caveolae inhibitor filipin (Fig. 5, A and B), suggesting that this process might be mediated by caveolae/lipid raft endocytosis. The inhibitor of clathrin-dependent endocytosis dansylcadaverine did not have a significant effect (Fig. 5C). On the other hand, INGAP-P internalization is inhibited by a 15-min preincubation with cytochalasin D, resulting in formation of small clusters on the cell surface (Fig. 5D). This suggests that actin filaments are involved in the process of INGAP-P internalization. However, it is unlikely to be macropinocytosis, since wortmannin did not appear to have an inhibitory effect on this process (Fig. 5E).

Fig. 5.

Internalization of FAM-INGAP-P is inhibited on ice by filipin and cytochalasin D but not by wortmannin or dansylcadaverine. Cells grown in chamber slides were treated with FAM-labeled INGAP-P (16.7 μM) for 1 h under various conditions, i.e., on ice (A) and in the presence of 1 μg/ml filipin (B). C: dansylcadaverine (300 μM). D: cytochalasin D (25 μg/ml). E: wortmannin (100 nM). F: 1 μl of DMSO (same volume as for the inhibitors). Nuclei were counterstained with DAPI included in the mounting medium (Prolong Gold; Invitrogen) and imaged using either Zeiss LSM 510 (A and B) or Olympus FV10i (C–F).

To investigate whether rINGAP and INGAP-P act via the same receptor, both DyLight488-rINGAP and FAM-INGAP-P were used in competition experiments with ×20 molar excesses of the unlabeled protein and peptide. The results show that internalization of the protein is inhibited partially by cold protein and the peptide by cold peptide, but they do not appear to inhibit each other (Fig. 6), at least at the given concentration, which suggests that they likely do not bind the same receptor.

Fig. 6.

Molar excess competition assay for binding and internalization of fluorescently labeled rINGAP and INGAP-P. RIN-m5F cells plated in chamber slides were incubated with FAM-INGAP-P for 1 h (left) or with DyLight-488 rINGAP for 5 h (right). A and B: no inhibition. C and D: with 167 μM INGAP-P (10× molar excess). E and F: with 1 μM rINGAP (20× molar excess).

Signaling pathways leading to ERK1/2 phosphorylation by both rINGAP and INGAP-P involve Ras-Raf activation.

Activation of ERK1/2 may be mediated by a number of signaling cascades initiated at the cell membrane level by receptor tyrosine kinases (RTK) or by different classes of GPCRs. These signaling cascades include the PKC, PKA, PI3K, or Ras/Raf-dependent pathways (30, 35, 46). Since the nature of the INGAP receptor is unknown, we screened for both RTK and GPCR-initiated signaling events using phosphospecific antibodies and pharmacological inhibitors of all of the above-mentioned pathways. For comparison, we used EGF (10 ng/ml) and Ex-4 (10 nM), which was found to be mitogenic for RIN-m5F cells at the indicated concentrations (Fig. 1A). Because EGF signals through the classical RTK pathway and Ex-4 is an agonist of the G protein-coupled glucagon-like peptide-1 (GLP-1) receptor (17), such a comparison may provide important clues to how INGAP works.

Activation of low-molecular-weight Ras family GTPases is the first key event in the signaling through RTKs such as EGFR. However, it became apparent that the mechanisms of MAP kinase activation by GPCRs may also include Ras activation by cross-talk between GPCRs and RTKs, e.g., transactivation of EGFRs shown for several GPCR ligands, including GLP-1 (8, 30). In keeping with this notion, our results show a rapid Ras activation by both INGAP-P and rINGAP (Fig. 7A) that precedes phosphorylation of c-Raf (Fig. 7B) and ERK1/2, which peaks at 10 min (Fig. 7C).

Since INGAP-P has been shown previously to activate the PI3K/Akt signaling pathway (5, 25), and because this pathway can be involved in cell proliferation, we measured phospho-Akt (Ser⁴⁷³) in a time course experiment and observed a weak increase (not statistically significant) by INGAP-P at 30 min but not by rINGAP (Fig. 7D). In contrast, both EGF and Ex-4 induced a transient Akt activation at 1 min, which preceded that of ERK1/2 (Fig. 7E). This is in line with previous studies showing that GLP-1 and EGF-like ligands stimulate proliferation in β-cells via activation of the PI3K/Akt pathway (8, 9). Accordingly, more late activation of Akt (at 30 min) than ERK1/2 (10 min) by INGAP-P suggests that the PI3K signaling is not involved in ERK1/2 phosphorylation in RIN-m5F cells. The fact that Akt does not seem to be activated by rINGAP indicates that signaling events upstream of Ras/Raf/ERK activation may vary between INGAP-P and rINGAP. Of note, we did not observe significant activation of either p38 MAPK (Western blot), PKA (ELISA), or PKC (Western blot and ELISA) by either protein or peptide (data not shown).

Pharmacological inhibition of signaling pathways implicates GPCR in mitogenic effects of INGAP on RIN-m5F cells.

To investigate signaling events implicated in INGAP-induced proliferation, we employed specific pharmacological inhibitors of Raf (Raf inhibitor 1), PI3K (wortmannin), PKC (Bis), PKA (H-89, PKi), adenylate cyclase (SQ), Src (PP2), and EGFR (AG-1478). In addition, Ptx was used to examine whether INGAP actions were mediated by a GPCR. The effectiveness of these inhibitors was judged by ERK1/2 phosphorylation after 10 min of treatment with INGAP, EGF, or Ex-4 (Fig. 8).

Fig. 8.

Effect of pharmacological inhibitors on ERK1/2 phosphorylation by INGAP, EGF, and Ex-4. RINm5F cells grown in 60-mm plates were pretreated for 30–40 min with the indicated inhibitors, except for pertussis toxin (Ptx; 24-h pretreatment). After a 10-min treatment with growth factors, cells were placed on ice and lysed; 30 μg of proteins was probed for p-ERK1/2. Data are shown as a ratio of p-ERK1/2 to total ERK1/2 relative to the “no inhibitors” group (equal to 1, shown as a dotted line). A: inhibitors of G-protein coupled receptor (Ptx, 100 ng/ml), adenylate cyclase [SQ-22536 (SQ); 250 μM], and PKA (PKi, 100 nM; H-89, 1 μM) or DMSO as a vehicle control. B: inhibitors of PKC [bisindolylmaleimide (Bis); 1 μM], phosphatidylinositol 3-kinase [wortmannin (Wm); 100 nM], Src (PP2; 100 nM), EGF receptor (EGFR; AG, 100 nM), MEK (PD), and c-Raf (Raf-1, 100 nM), (*P < 0.05 and §P < 0.01 compared with “no inhibitor” group for each growth factor treatment).

As shown in Fig. 8A, INGAP-P- and rINGAP-induced activation of ERK1/2 was inhibited by ∼40% after a 24-h exposure to Ptx but was not affected by AG-1478 (Fig. 8B). This suggests that INGAP likely signals through a GPCR but that this signaling does not involve the EGFR, as has been shown previously for GLP-1 (8). Ptx also inhibited early Ras activation induced by INGAP, EGF, or Ex-4, (Fig. 9), which further supports the idea that INGAP signals via a GPCR-Ras pathway. Consistent with the previous implication of Ras-Raf signaling, pretreatment with Raf kinase inhibitor 1 reduced ERK1/2 activation (Fig. 8B) by all growth factors tested. Interestingly, Src inhibitor PP2 was effective for rINGAP but not for INGAP-P (Fig. 8B). Further highlighting the differences in signaling between the protein and peptide, inhibition of PKC had a stimulatory effect on rINGAP-induced ERK1/2 activation, whereas no such effect was seen for INGAP-P (Fig. 8B).

Fig. 9.

Inhibition of G protein-coupled receptor signaling results in diminished Ras activation. RIN-m5F cells grown in 60-mm plates were pretreated with Ptx for 24 h prior to the addition of growth factors for 1, 3, 5, and 10 min. Control cells were treated with growth factors (no Ptx) for the same intervals. Cells were harvested in Mg+ lysis buffer and subjected to the Ras-GTP ELISA, as described in materials and methods. The results are shown as a ratio of Ptx treated/control per each time point and are means ± SE of at least 3 independent experiments (*P < 0.05 compared with control, which equals 1 and is shown as a dotted line).

For comparison, ERK1/2 activation by EGF was inhibited with AG-1478, wortmannin, PP2, H-89, Ptx, and Raf inhibitor 1, whereas only Raf inhibitor 1 and Ptx had an effect on Ex-4. As expected, PD-98059 was an effective inhibitor for all growth factors tested.

DISCUSSION

Based on the original studies of discovery and cloning of INGAP (39), it has been believed that peptide^104–118 (INGAP-P) is an active center of the protein, since it had essentially the same effects on target tissues. Since then, research has focused on the peptide as a more clinically relevant compound (28–32) that is now in phase II clinical trials (18). However, the peptide has a limited stability and has to be administered in high doses. In an effort to improve INGAP-P efficacy, we searched for clues in the full-length INGAP protein, which we cloned recently and which displayed a much greater stability and at least 100 times higher molar efficiency in inducing in vitro regeneration of functional human islets from dedifferentiated, islet-derived, duct-like structures (2). Likewise, significant differences in effective concentrations of INGAP-P and protein have been observed for both proliferative [ARIP cells, hamster ductal explants (39)] and differentiating effects (human pancreatic ductal epithelial cells; Assouline-Thomas B, unpublished observations). In the current study, we also demonstrate that rINGAP is more efficient on a molar basis (>100 times) than INGAP-P in stimulation of proliferation in RIN-m5F cells, although both ligands in principle activate the same Ras/Raf/Erk pathway.

The data suggest that the difference in molar efficiency is most likely based on how rINGAP and INGAP-P interact with the cell surface. rINGAP binds RIN-m5F cells in a manner consistent with the clustering of the ligand-receptor complexes, which may also include other membrane proteins such as integrins, proteoglycans, and various glycoconjugates and which may contribute to a higher intensity of the initiated signaling (47). Formation of such complexes could also explain delayed internalization of rINGAP that probably occurs by macropinocytosis, as indicated by the inhibition experiments with wortmannin and cytochalasin D. At present, it is unclear whether rINGAP internalization plays a role in the downstream signaling or whether it only serves the purpose of ligand-receptor degradation once the signaling cascade is initiated. Similar internalization kinetics have been described for hepatoma-derived growth factor, which binds its receptor, and heparan/sulphate proteoglycans (55). As shown for hepatoma-derived growth factor, receptor binding initiates downstream signaling, whereas the heparan/sulphate-based internalization, also by macropinocytosis, modulates the strength of the related signaling processes. Such a possibility for INGAP should be explored once INGAP binding partners are identified.

In contrast, INGAP-P seems to interact with the membrane in a short-lived transient fashion, translocating into the cytoplasm in a few minutes. The observed sensitivity of this process to filipin, an inhibitor of caveolae-based endocytosis, and to cytochalasin D, an inhibitor of actin assembly, but not to the inhibitors of clathrin-based endocytosis or to wortmannin implicates caveolae as the major route of internalization for INGAP-P.

In view of the observed differences, the main question is whether the peptide binds a specific receptor and whether this receptor is the same as for the protein. Although the lack of binding on ice may suggest an argument against the receptor for INGAP-P, a number of studies show that low temperature may affect the peptide ligand conformation, which may in turn affect the binding (27). Another argument for a receptor is that activation of the Ras/Raf/ERK1/2 cascade by both INGAP-P and rINGAP is blocked by Ptx, suggesting the involvement of a GPCR. As shown for PTH and PTH-related peptide, the same receptor may activate different signaling pathways when activated by related but not identical ligands (16, 21). Therefore, it is possible that INGAP-P, as an active center of rINGAP, binds the same receptor, but due to different duration and conformational changes produced by the receptor-ligand interactions, the downstream signaling responses vary. However, it is also possible that INGAP-P and rINGAP bind and activate different receptors, especially because no competition between the two has been found in our experiments.

The lack of knowledge about INGAP receptor(s) complicates delineation of signaling mechanisms upstream of the Ras/Raf/ERK cascade. The results of screening with pharmacological inhibitors for most common transducing molecules reported previously for INGAP-P signaling, such as PI3K, PKA, adenylate cyclase (25, 51), and PKC, show that these kinases are not involved in ERK1/2 activation in RIN-m5F cells induced by either rINGAP or by INGAP-P. To the contrary, we observed a stimulatory effect of Bis (PKC inhibitor) on rINGAP-induced p-ERK1/2. Together with sensitivity to Ptx, these data suggest that both ligands likely act via a G_i protein-coupled receptor(s) that does not activate PKC or adenylate cyclase (30). Thus the proposed signaling pathway for rINGAP looks like GPCR/Src/?/Ras/Raf/Erk, whereas the pathway for INGAP-P remains less defined and may be described as GPCR/?/ Ras/Raf/Erk (question marks signify possible missing links in the pathway that are yet to be identified).

This is the first study that compared INGAP protein and INGAP-P, using RIN-m5F cells as a model and focusing on the proliferative response and ERK1/2 activation. However, the effects of INGAP-P on this cell line have also been studied by others reporting an increase in proliferation as well as an upregulation of the muscarinic M₃ receptor after 72 h, which was mediated by activation of NF-κB (34). Of note, upregulation of the M₃ receptor by INGAP-P via NF-κB and the association with that increase in insulin secretion was also observed in MIN6 cells (34) and in neonatal rat islets (5), thus implicating NF-κB activation by INGAP-P as a common mechanism in insulin-secreting cells. Signaling events leading to NF-κB activation are yet to be investigated. It would be interesting to determine whether rINGAP also activates NF-κB and upregulates the M₃ receptor in β-cell lines and pancreatic islets and whether it does so with a higher molar efficiency than the peptide.

A number of signaling pathways activated by INGAP-P have been reported, with the PI3K/Akt pathway being the most common (5, 25). A rapid activation of ERK1/2 (MAPK3/1) by INGAP-P has also been shown in neonatal rat islets (5), and involvement of cAMP-PKA in the INGAP-P-induced neurite outgrowth has been reported (51). However, although our data do not indicate the involvement of PI3K/Akt in the proliferative action of INGAP (Fig. 8), we did observe weak Akt phosphorylation in cells treated with INGAP-P for 30 min, which comes later than the Ras/Raf/Erk activation. Therefore, it can be speculated that INGAP-P-induced PI3K/Akt signaling induces other responses in RIN-m5F cells, e.g., the NF-κB pathway, which should be investigated in future studies. Likewise, a broader assessment of signaling activated by rINGAP has to be undertaken after short-term and a long-term exposure. Inclusion of other cell types, such as ductal cells or pancreatic islets into these studies, is also needed to fully understand the potential and the mechanism of action of rINGAP and INGAP-P.

Taken together, our data show that both rINGAP and INGAP-P stimulate cell proliferation via the Ras/Raf/Erk pathway and suggest involvement of a G_i protein-coupled receptor(s). The major differences between the two are related to cell binding and internalization, which we believe are responsible for a much lower molar efficiency of the peptide. These data suggest that modifications of the peptide leading to a more rINGAP-like cell binding might be an interesting approach to improve INGAP-P efficacy as a therapy for diabetes.

GRANTS

This work was supported by the Canadian Institute of Health Research.

DISCLOSURES

The authors have no conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

M.P. and L.R. did the conception and design of the research; M.P., J.Z., M.M., J. Ding, J.M., and B.A.-T. performed the experiments; M.P., J. Daoud, J.Z., M.M., J. Ding, and J.M. analyzed the data; M.P. and J. Daoud interpreted the results of the experiments; M.P. prepared the figures; M.P. drafted the manuscript; M.P. and L.R. edited and revised the manuscript; L.R. approved the final version of the manuscript.