Activation of IRS-2—Mediated Signal Transduction by IGF-1, but not TGF-α or EGF, Augments Pancreatic β-Cell Proliferation

The [methyl-³H]thymidine (25 Ci/mmol) was from Amersham (Piscataway, NJ). Antibodies to phosphorylated Erk1/2 were purchased from Promega (Madison, WI); antibodies to PKB total, phosphorylated (Ser⁴⁷³ and Thr³⁰⁸) PKB and phosphorylated GSK-3 were purchased from New England Biolabs (Beverly, MA); antibodies to GSK-3 total and EGFR were purchased from Santa Cruz (Santa Cruz, CA); and antibody to the p85 subunit of PI3K was purchased from Upstate (Lake Placid, NY). The total Erk antiserum was a gift from Dr. M. Cobb (University of Texas Southwestern Medical Center, Dallas, TX), and the p70^S6K and IRS-1 and -2 antisera and cDNA were gifts from Drs. M. White and M. Myers (Joslin Diabetes Center, Boston, MA). The activity assay for PKB was purchased from New England Biolabs, human recombinant IGF-1 from Gropep (Adelaide, Australia), and murine EGF from Sigma (St. Louis, MO). Purified human recombinant TGF-α was prepared by Stem Cell Pharmaceuticals (Seattle, WA). Protein kinase/phosphatase inhibitors were purchased from Calbiochem (La Jolla, CA). All other biochemicals were purchased from Sigma or Fisher Scientific (Pittsburgh, PA).

The glucose-sensitive pancreatic β-cell line INS-1 (37) was used in these experiments. INS-1 cells were maintained in RPMI 1640 medium containing 2 mmol/l l-glutamine, 1 mmol/l sodium pyruvate, 50 μmol/l β-mercaptoethanol, 100 units/ml penicillin, 100 μg/ml streptomycin, 10% FCS, and 11.2 mmol/l glucose and incubated at 37°C, 5% CO₂ as described (37). Under starvation conditions (before beginning experimental treatment), cells were incubated in RPMI 1640 medium containing 0.1% fatty acid-free BSA, 100 units/ml penicillin, 100 μg/ml streptomycin, and 0.5 mmol/l glucose and incubated as before. In all experiments, INS-1 cells utilized for controls were incubated with starvation media during treatment periods. Cells were subcultured at 80% confluence.

IRS-1 and -2 recombinant adenoviruses were prepared as previously described (38). Essentially, the cDNA encoding human IRS-1 or mouse IRS-2 was introduced into the shuttle vector pACCMVS, and homologous recombination was carried out in HEK293 cells that contain the transcription factor, E1A, which is required for viral replication. The generated recombinant adenovirus was amplified and purified from lysed 293 cells and utilized for infection of β-cells (38). Purified virus (5 × 10⁹ pfu/10⁶ cells) was added to INS-1 cells and incubated 2 h. Cells were washed and fresh media was added. After a 24-h incubation, cells were trypsinized and plated onto 96-well plates for [³H]thymidine incorporation assay.

Incorporation of [³H]thymidine was used as an indicator of DNA synthesis and INS-1 cell proliferation and was carried out as described previously (10,11). Briefly, INS-1 cells were cultured on 96-well plates (1 × 10⁴ cells/well), starved for 24 h, and treated with various growth factors with or without glucose (0.5–15 mmol/l) for 24 h. For the last 4 h of treatment, 1 μCi [³H]thymidine/well was added. The [³H]thymidine specifically incorporated into the INS-1 cell DNA was trapped on glass fiber filters and counted by liquid scintillation counting.

INS-1 cells were cultured on 6-well plates to 80% confluence. Cells were starved (RPMI 1640 without serum and only 0.5 mmol/l glucose) for 24 h before treatment. Quiescent cells were treated with growth factors in the absence or presence of increasing glucose concentrations for the indicated time periods. Cells were washed once with ice-cold PBS and lysed in buffer containing 50 mmol/l HEPES (pH 7.5), 1% Triton X-100, 2 mmol/l sodium vanadate, 50 mmol/l sodium fluoride, 10 mmol/l sodium pyrophosphate, 4 mmol/l EDTA, 10 μmol/l leupeptin, 10 μg/ml aprotinin, and 100 μmol/l phenylmethylsulfonylfluoride.

Immunoblot analysis of mitogenic signal transduction protein expression and protein tyrosine phosphorylation was performed as described previously, using horseradish peroxidase-based chemiluminescence reaction as a secondary detection method (39,40).

Cells were treated as above but grown on 15-cm plates. Lysate (1.5 ml) containing ∼2.0 mg total protein was immunoprecipitated with 5 μl antibody against either p85 or EGFR for 16 h at 4°C. The immune complexes were collected with 50 μl protein A-Sepharose for 1 h at 4°C, washed three times with lysis buffer, and resuspended in 40 μl of 2× SDS gel loading buffer. Samples were loaded onto a 7.5% SDS-PAGE gel, and coimmunoprecipitated proteins were detected via immunoblot analysis.

Cells were treated as above. Harvested protein (30 μg) was added to 100 μl PKB lysis buffer containing 20 μl anti-PKB conjugated to agarose beads (New England Biolabs) and incubated on a rocker at 4°C for 2 h. PKB activity was measured as described in the manufacturer’s directions for the PKB/Akt Kinase Assay Kit (New England Biolabs).

Protein assay was performed using the bicinchoninic acid method (Pierce, Rockford, IL). Data are presented as means ± SE. Statistically significant differences between groups were analyzed using Student’s t test, where P < 0.05 was considered statistically significant.

Experiments were conducted to define the signaling kinase cascades activated by TGF-α or EGF in INS-1 cells. In these experiments, the kinetics of activation were examined and compared with those signaling responses initiated by glucose and IGF-1. In preliminary experiments, it was found that over a concentration range between 0.1 and 100 nmol/l TGF-α, EGF, or IGF-1, maximum Erk1/2 phosphorylation was attained between 1 and 10 nmol/l for any of these growth factors in INS-1 cells incubated for 10 min at 15 mmol/l glucose (data not shown). This is comparable to that previously found for IGF-1-induced [³H]thymidine incorporation in INS-1 cells (11). In all of the following experiments, a 10-nmol/l concentration of growth factor was used in order to elicit a marked response. The time course of Erk1/2 phosphorylation was examined after 0, 2, 5, 10, 20 and 40 min in INS-1 cells treated with either basal (3 mmol/l) or stimulatory (15 mmol/l) glucose concentrations (Fig. 1A) alone or in the additional presence of 10 nmol/l IGF-1 (Fig. 1B), 10 nmol/l TGF-α (Fig. 1C), or 10 nmol/l EGF (Fig. 1D). Quantitative analyses of Western blots are shown in Figs. 1A–D, and data are expressed as the fold-increase above “time zero.” Glucose (15 mmol/l), in the absence of growth factors, significantly stimulated Erk1/2 phosphorylation ≥10-fold by 5 min (P ≤ 0.01) (Fig. 1A). The effect of glucose was greatest at 5 min, and a ≥5-fold effect was still readily apparent at 40 min (P ≤ 0.05, Fig. 1A). IGF-1 potentiated Erk1/2 phosphorylation at 15 mmol/l glucose, which also peaked after 5 min (15-fold; P ≤ 0.01) and was maintained as long as 40 min (≥10-fold; P ≤ 0.01, Fig 1B). In contrast to the magnitude of response observed with glucose and IGF-1, TGF-α and EGF induced a much more marked increase in Erk1/2 phosphorylation/activation in INS-1 cells. Erk1/2 activation reached a maximum at 5 min, but was >40-fold for TGF-α and >25-fold for EGF (P ≤ 0.01, Figs. 1C and D). Also, unlike that induced by glucose or IGF-1, phosphorylation of Erk1/2 induced by TGF-α and EGF was transient and diminished after 20 min. The increase in phosphorylation, which is still present with TGF-α/EGF plus 15 mmol/l glucose at 40 min, is equivalent to that produced by 15 mmol/l glucose in the absence of growth factor. Unlike IGF-1, TGF-α/EGF-induced Erk1/2 phosphorylation was independent of glucose concentration. In these experiments, total Erk1/2 remained unchanged under any of the conditions (Figs. 1A–D).

Glucose and IGF-1 have been shown to stimulate PI3K activity in INS-1 cells (17). PKB is activated upon binding to PI3K-generated lipid products in the plasma membrane via its pleckstrin homology domain and subsequent phosphorylation of Thr³⁰⁸ by 3-phosphoinositide-dependent kinase (PDK-1), followed by a second phosphorylation at Ser⁴⁷³, which may be attributed to autophosphorylation (41,42). The time course of phosphorylation of PKB was evaluated, utilizing Thr³⁰⁸ and Ser⁴⁷³ phosphospecific antibodies, in INS-1 cells incubated 0, 2, 5, 10, 20 and 40 min with 3 and 15 mmol/l glucose alone or in the additional presence of 10 nmol/l IGF-1, 10 nmol/l TGF-α, or 10 nmol/l EGF (Fig. 2). Glucose alone did not significantly increase PKB phosphorylation as observed previously (18), and appeared slightly lower than that at time zero at either glucose concentration (Fig. 2). Phosphorylation of PKB at both the Ser⁴⁷³ and Thr³⁰⁸ sites induced by IGF-1 occurred rapidly (≤2 min) and was maintained ≥40 min (Fig. 2). The magnitude of PKB phosphorylation induced by IGF-1 was unchanged with time and was independent of glucose concentration (Fig. 2). TGF-α and EGF also induced a marked increase in PKB phosphorylation at both the Ser⁴⁷³ and Thr³⁰⁸ sites. However, this response to TGF-α and EGF was transient, peaking between 2 and 5 min but back to basal levels by 20 min (Fig. 2). In these experiments, total PKB remained unchanged under all conditions (Fig. 2).

PKB activity was simultaneously examined in PKB immunoprecipitates of INS-1 cells and correlated with the phosphorylation response observed (Figs. 2 and 3). Glucose did not significantly affect PKB activity between 2 and 40 min of treatment (Fig. 3A). However, glucose did modestly increase PKB activity (twofold) after 2 h (Fig. 3A). IGF-1 increased PKB activity (Fig. 3B) in parallel with PKB phosphorylation state (Fig. 2). The IGF-1-induced increase was twofold (P ≤ 0.05) between 2 and 40 min of treatment and further increased (four- to fivefold, P ≤ 0.05), independently of glucose, after 2 h (Fig. 3B). PKB activity was increased nearly eightfold in the presence of TGF-α and EGF irrespective of glucose concentration after 10 min of treatment (P ≤ 0.01, Figs. 3C and D). However, as was observed with PKB phosphorylation induced by TGF-α and EGF (Fig. 2), PKB activity diminished rapidly within 20 min and was negligible after 2 h, with only the slight twofold glucose-induced increase in PKB activity still present (Figs. 3C and D).

GSK-3 is a constitutively active kinase that is inactivated in response to insulin and IGF-1 by phosphorylation, as a downstream target of PKB (43). The time course of GSK-3 phosphorylation was examined after 0, 2, 5, 10, 20 and 40 min in INS-1 cells treated with either basal (3 mmol/l) or stimulatory (15 mmol/l) glucose concentrations (Fig. 4A) alone or in the additional presence of 10 nmol/l IGF-1 (Fig. 4B), 10 nmol/l TGF-α (Fig. 4C), or 10 nmol/l EGF (Fig. 4D). Quantitative analyses of Western blots are shown in Figs. 4A–D, and data are expressed as the fold-increase above time zero. Glucose, in the absence of growth factors, did not significantly increase GSK-3 phosphorylation between 2 and 40 min, as previously found (18) (Fig. 4A). IGF-1 rapidly increased levels of phosphorylated GSK-3 (threefold, P ≤ 0.05), and this response was also maintained at 40 min (Fig. 4B, P ≤ 0.05). IGF-1-induced GSK-3 phosphorylation correlated with IGF-1-induced PKB phosphorylation and was independent of glucose concentration (Figs. 2 and 4). The phosphorylation of GSK-3 was markedly increased by TGF-α and EGF, peaking at 10 min of treatment (≥10-fold, P ≤ 0.05; Figs. 4C and D). As for TGF-α- and EGF-induced PKB phosphorylation (Figs. 3C and D), TGF-α- and EGF-stimulated GSK-3 phosphorylation quickly fell to control levels by 20 min of treatment (Figs. 4C and D).

p70^S6K activation has been proposed to lie downstream of PKB and mTOR (44). However, glucose- and amino acid-induced p70^S6K activation has been shown to be independent of PKB stimulation, involving an alternative route mediated directly via mTOR activation (19,45). p70^S6K is phosphorylated at multiple sites, which can be determined from analysis of its decrease in electrophoretic migration in a SDS-polyacrylamide gel (44,46). Phosphorylation of p70^S6K was evaluated in INS-1 cells at 5, 40, and 120 min incubated with 3, 6, and 15 mmol/l glucose alone or in the additional presence of 10 nmol/l IGF-1, 10 nmol/l TGF-α, or 10 nmol/l EGF (Fig. 5). Control cells were treated with starvation media containing 0.5 mmol/l glucose for the indicated time periods. As demonstrated previously, glucose increased the phosphorylation state of p70^S6K, which was further increased by the addition of IGF-1 (Fig. 5) (10). However, this increase was delayed in onset, so that a significant increase in phosphorylation of p70^S6K was not observed until after 40 min, but thereafter was maintained ≥2 h (Fig. 5). In contrast to IGF-1, TGF-α and EGF did not enhance glucose-dependent phosphorylation of p70^S6K (Fig. 5) and had no effect on the phosphorylation of p70^S6K in the absence of glucose.

Because stimulatory glucose and IGF-1 appeared to maintain activation of certain signaling molecules >40 min and the activation of p70^S6K and PKB was delayed in onset in response to glucose, experiments examining phosphorylation/activation of signaling proteins were extended to 2 h. After 2 h of treatment, Erk1/2 phosphorylation remained increased at stimulatory glucose concentrations independently of growth factor presence (Fig. 6). The effect of IGF-1 on Erk1/2 phosphorylation was glucose dependent and maintained for 2 h (Fig. 6). IGF-1-stimulated PKB phosphorylation (at both Ser⁴⁷³ and Thr³⁰⁸) was also maintained up to 2 h, and as observed previously (18), was independent of glucose concentration (Fig. 6). In contrast to short-term incubation (<1 h), 15 mmol/l glucose slightly increased phosphorylation of PKB at Ser⁴⁷³ and Thr³⁰⁸ (Fig. 6), which correlated with increased activity of PKB (Fig. 3A). Likewise, GSK-3 phosphorylation followed that of PKB phosphorylation/activity, including a slight increase after 2 h of exposure to 15 mmol/l glucose (Fig. 6). We have also found that the glucose-dependent IGF-1-induced phosphorylation of Erk1/2 and p70^S6K, as well as glucose-independent IGF-1-induced phosphorylation of PKB and GSK-3, was maintained up to 24 h (data not shown). TGF-α and EGF did not affect Erk1/2, PKB, or GSK-3 phosphorylation at 2 h (Fig. 6), characteristic of their transient activity observed in β-cells. Only an effect attributed to 15 mmol/l glucose could be observed after 2 h in the presence of TGF-α or EGF.

[³H]thymidine incorporation was measured in quiescent INS-1 cells treated 24 h with 3 and 15 mmol/l glucose alone or in the additional presence of 10 nmol/l IGF-1, 10 nmol/l TGF-α, or 10 nmol/l EGF (Fig. 7). Control cells were treated with starvation media containing 0.5 mmol/l glucose and data are expressed as the fold-increase above control. TGF-α/EGF did not significantly enhance β-cell proliferation (Fig. 7). Moreover, neither TGF-α nor EGF potentiated the proliferative response occurring with 15 mmol/l glucose (∼20-fold over basal, P < 0.05) (Fig. 7). However, in contrast to TGF-α/EGF, IGF-1 in the presence of high glucose concentrations (15 mmol/l) markedly increased INS-1 cell proliferation (>60-fold, P < 0.05) (Fig. 7) as previously observed (10,11). The effect of IGF-1 on β-cell proliferation was glucose-dependent (10,11). In experiments in which IGF-1 was added to TGF-α and EGF, there was no synergistic effect observed on β-cell proliferation at permissive glucose concentrations, unlike that found with the combination of IGF-1 and growth hormone (data not shown) (10).

The differences in duration between TGF-α- and EGF-stimulated signaling events and those induced by IGF-1 and glucose might have been due to differential activation of early signal elements immediately downstream of the growth factor receptor. IRS-1 and -2 are cytosolic docking proteins, observed to bind to the activated, ligand-bound IGF-1R when it is stimulated (47) and important in mitogenic responses for β-cell proliferation (3). It was examined whether TGF-α- and EGF-induced Erk1/2 and PKB signal transduction occurred via upstream activation of IRS-1 and -2 in β-cells. INS-1 cells were incubated for 5 min with 3 and 15 mmol/l glucose alone or in the additional presence of 10 nmol/l IGF-1, 10 nmol/l TGF-α, or 10 nmol/l EGF (Fig. 8). The p85 subunit of PI3K was immunoprecipitated from the cell lysates and association with IRS-1 and -2 were measured by subsequent immunoblotting of these p85 immunoprecipitates. IGF-1 increased the association of IRS-1 and -2 with PI3K within 5 min, which was further enhanced at a stimulatory 15 mmol/l glucose concentration (Fig. 8). In parallel experiments in which the EGFR was immunoprecipitated instead of the p85 subunit of PI3K, no association of IRS-1 and -2 was detected during stimulation with TGF-α and EGF (data not shown).

The effect of adenoviral-mediated overexpression of IRS-1 and -2 on β-cell proliferation was investigated. Firstly, INS-1 cells were infected with increasing doses of adenovirus (5–100 × 10⁸ pfu/10⁶ cells) containing either IRS-1 or -2, and levels of IRS-1 and -2 were evaluated by immunoblot analysis (Fig. 9A). A dose of 50 × 10⁸ pfu, a dose in which both proteins were increased >10-fold, was chosen for the experiments examining their effects on [³H]thymidine incorporation. An adenovirus that expressed GFP (AdV-GFP) was used as a control for these experiments and also indicated a >90% gene transfer efficiency mediated by the recombinant adenoviral vector as previously described (38). [³H]thymidine incorporation was measured as described above in quiescent, infected INS-1 cells treated with 3 and 15 mmol/l glucose alone or in the additional presence of 10 nmol/l IGF-1, 10 nmol/l TGF-α, or 10 nmol/l EGF (Fig. 9B). Control cells were treated with starvation media containing 0.5 mmol/l glucose, and data are expressed as the fold-increase above control. In INS-1 cells overexpressing IRS-2, glucose-dependent proliferation was >60-fold (P < 0.05) and glucose-IGF-1-dependent proliferation was ∼120-fold (P < 0.05) (Fig. 9B). Thus, IRS-2 overexpression enhanced glucose- and IGF-1-induced β-cell proliferation twofold above that of AdV-GFP-infected cells (Fig. 9B). Unlike IRS-2, IRS-1 did not significantly increase proliferation rates over that of control infected INS-1 cells stimulated with glucose with or without IGF-1 (Fig. 9B). In INS-1 cells overexpressing either IRS-1 or -2 and treated with TGF-α and EGF, no change in proliferation rate was observed above that already induced by glucose alone.

IGF-1 and glucose are mitogenic to pancreatic β-cells (9–11). Glucose-dependent IGF-1 signal transduction via IRS-mediated stimulation of PI3K activity has been shown to be essential for eliciting a mitogenic response in β-cells (10,11). Additionally, Erk1/2 activation and p70^S6K stimulation are known to facilitate the mitogenic responses of IGF-1 and glucose (10,11,18). Similar to IGF-1, TGF-α and EGF activated similar signaling pathways, although their magnitude of activation, duration of activity, and lack of glucose-dependency differed from that of IGF-1 in β-cells. A consequence of these differences was that TGF-α and EGF did not enhance INS-1 cell proliferation. In this side by side comparative study, the importance of duration, not magnitude, of the growth factor-induced signaling response appeared to be a significant factor for inducing β-cell mitogenesis. TGF-α and EGF induced marked signaling responses, including increased Erk1/2 phosphorylation and PKB phosphosphorylation and activity, which led to increased GSK-3 phosphorylation but did not progress to p70^S6K activation. These responses were much greater in magnitude than that induced by IGF-1, but were transient, lasting no longer than 20 min. Although, IGF-1 produced a comparatively modest increase in Erk1/2 and PI3K/PKB/GSK-3 signaling pathways compared with TGF-α/EGF, the responses were maintained ≥2 h and even up to 24 h. The activation of p70 ^S6K required >30 min of stimulation by glucose and IGF-1; therefore, the lack of a mitogenic response associated with TGF-α/EGF treatment is likely the result of the transient nature of the signal.

The dissimilarities in the kinetics and amplitude of activation of signaling pathways associated with TGF-α/EGF or IGF-1 treatment may be the result of differences in protein complex formation of signaling molecules immediately downstream of their receptors, the EGFR and IGF-1R, respectively. Activation of the IGF-1R intrinsic tyrosine kinase activity leads to tyrosine phosphorylation of IRS-1 and/or IRS-2, which via docking of the p85 regulatory subunit of PI3K, causes activation of PI3K and initiation of downstream signaling events (47). We found IGF-1 induced a significant increase in the association of IRS-1 and IRS-2 with p85 in INS-1 cells. This association, though present at 3 mmol/l glucose, was substantially greater in the presence of stimulatory glucose (15 mmol/l), underlying the glucose-dependent aspect of IGF-1-induced mitogenesis in β-cells (11). In contrast, neither TGF-α nor EGF was able to promote p85 recruitment to IRS-1 or -2, indicating that the EGFR does not induce signal transduction via IRS irrespective of glucose concentration. EGFR binding directly to p85, however, is inducible by TGF-α and EGF in INS-1 cells (data not shown). Dissimilar to IGF-1R activation, TGF-α/EGF-induced activation of the EGFR intrinsic tyrosine kinase leads to EGFR autophosphorylation and the recruitment of the p85 subunit of PI3K directly to the EGFR, but might also involve alternative adapter molecules to members of the IRS family (29,30,48). In general, it has been unclear whether EGFR ligands can induce proliferation of differentiated β-cells, perhaps with the exception of betacellulin (34), although they are capable of causing mitogenesis within pancreatic tumors (25,26,31–33) and promote β-cell neogenesis (35). The differences in mitogenic responses of pancreatic cells to these growth factors have been attributed to the presence of certain ErbB receptor subtypes and receptor interactions with certain adapter molecules (22,29,30,34–36). It is possible that the lack of a prolonged signaling response by TGF-α or EGF in INS-1 cells might be due to the absence of a particular ErbB receptor subtype and/or associated adapter molecule. However, other possibilities, such as a more rapid TGF-α/EGF receptor desensitization, perhaps involving decreased TGF-α/EGF receptor recycling and/or efficient phosphotyrosine phosphatase-mediated inactivation, should also be considered; to establish such studies, however, further experimentation is required. Notwithstanding, the data revealed in this study indicate that TGF-α/EGF do not induce a mitogenic response in INS-1 cells (a model of relatively well-differentiated mature β-cells [37]), despite a marked activation of Erk1/2 and PI3K signaling pathways.

The inability of TGF-α/EGF to signal via IRS, unlike IGF-1, could also contribute to the failure to maintain activation of signaling pathways leading to mitogenesis in the β-cell. Indeed, the importance of IRS-2 in particular for initiating signaling events leading to increased β-cell mitogenesis was emphasized in this study in adenoviral-mediated IRS-2 overexpressing INS-1 cells in which glucose- and IGF-1-induced β-cell proliferation was more than doubled, whereas TGF-α/EGF had no effect. Intriguingly, adenoviral-mediated IRS-1 overexpression did not significantly increase glucose or IGF-1-induced β-cell mitogenesis. These data provided further evidence suggesting a specific key role of IRS-2 in the promotion of β-cell growth (3) and are consistent with increased expression of IRS-2 within proliferating β-cells (49). Moreover, these observations are also inversely correlated with the marked reduction in β-cell mass found in IRS-2—deficient mice, but not in IRS-1-deficient mice (14,15,50,51). Therefore, in this study the data suggest that maintained activation of PI3K and Erk1/2 mitogenic signal transduction pathways required for β-cell proliferation is at least in part maintained in β-cells because of growth factor signaling that occurs specifically via IRS-2. However, other factors that associate with IRS-2 (rev. in ref. 52) could also contribute to a stable signaling complex and may also play a role in maintaining a continuous IRS-2-mediated signaling response that leads to increased β-cell proliferation.

In summary, this study further demonstrated the importance of glucose-dependent activation of Erk1/2 and p70^S6K signaling elements, but not necessarily PKB, to commit a β-cell to a mitogenic event. Moreover, sustained activation of these mitogenic signal transduction pathways appears to be required to drive β-cell proliferation. Such a prolonged activation could be a consequence of specific signaling via IRS-2. Growth factors, such as TGF-α and EGF, which only induce a transient activation (<20 min) of PI3K and Erk1/2 signal transduction pathways, do not signal via IRS-2 or initiate a stable complex formation capable of signaling to Erk1/2 or PI3K in INS-1 cells, thus failing to activate p70^S6K and to induce β-cell proliferation. Therefore, if human β-cells are to be propagated in vitro to increase the number of cells for eventual β-cell replacement therapy of type 1 diabetes, it will be important to use growth factors that maintain activation of signal transduction pathways (particularly those involving IRS-2, Erk1/2, and p70^S6K) in cell culturing strategies. Likewise, if the therapeutic goal of maintaining a compensatory increase in β-cell mass to adapt to peripheral insulin resistance is to be considered a viable pharmaceutical strategy to treat type 2 diabetes, such an approach must consider inducing a sustained activation of β-cell mitogenic signal transduction pathway, particularly those mediated via IRS-2.

This work was supported by National Institutes of Health Grant DK 55267 (to C.J.R.) and by the Juvenile Diabetes Foundation International (to C.J.R.). We are grateful to Stem Cell Pharmaceuticals for provision of TGF-α. We thank Cynthia Jacobs for the preparation of this manuscript.