Selective Insulin Signaling through A and B Insulin Receptors Regulates Transcription of Insulin and Glucokinase Genes in Pancreatic β Cells

Molecular Cell, Vol. 7, 559–570, March, 2001, Copyright 2001 by Cell Press

Selective Insulin Signaling through A and B Insulin Receptors Regulates Transcription of Insulin and Glucokinase Genes in Pancreatic ␤ Cells Barbara Leibiger,*§ Ingo B. Leibiger,*§k Tilo Moede,* Sabine Kemper,* Rohit N. Kulkarni,† C. Ronald Kahn,† Lina Moitoso de Vargas,‡ and Per-Olof Berggren* * The Rolf Luft Center for Diabetes Research Department of Molecular Medicine Karolinska Institutet S-171 76 Stockholm Sweden † Research Division Joslin Diabetes Center and Department of Medicine Harvard Medical School Boston, Massachusetts 02215 ‡ Department of Medicine New England Medical Center and Tufts University School of Medicine Boston, Massachusetts 02111

Summary Insulin signaling is mediated by a complex network of diverging and converging pathways, with alternative proteins and isoforms at almost every step in the process. We show here that insulin activates the transcription of its own gene and that of the ␤ cell glucokinase gene (␤GK) by different mechanisms. Whereas insulin gene transcription is promoted by signaling through insulin receptor A type (Ex11⫺), PI3K class Ia, and p70s6k, insulin stimulates the ␤GK gene by signaling via insulin receptor B type (Ex11⫹), PI3K class II–like activity, and PKB (c-Akt). Our data provide evidence for selectivity in insulin action via the two isoforms of the insulin receptor, the molecular basis being preferential signaling through different PI3K and protein kinases. Introduction Understanding selectivity in signal transduction is one of the most challenging tasks in current cell biology. Over the years, insulin signaling has served as one of the model examples in hormone-induced signal transduction. Malfunction of insulin signaling, referred to as insulin resistance, is one of the major causes of type 2 diabetes mellitus (non-insulin-dependent diabetes mellitus), the most common metabolic disorder in man. Insulin has been shown to exhibit pleiotropic effects involving mitogenic and/or metabolic events. Moreover, the effect of insulin is tissue as well as development dependent. The fact that insulin may transduce its signal through a variety of pathways has been discussed in extensive detail (White and Kahn, 1994). The two major pathways described to date, which employ insulin rek To whom correspondence should be addressed (e-mail: ingo@

enk.ks.se). §These authors contributed equally to this work.

ceptors as the primary target, include signaling via mitogen-activated protein (MAP) kinases and phosphoinositol-3 kinase (PI3K). The insulin receptor (IR), the first step in these cascades, exists in two isoforms as a result of alternative mRNA splicing of the 11th exon of the insulin proreceptor transcript (Seino et al., 1989). The A type (IR-A), or Ex11⫺ (Ullrich et al., 1985), lacks whereas the B type (IR-B), or Ex11⫹ (Ebina et al., 1985), contains the respective sequence coding for 12 amino acids in the C terminus of the ␣ chain of the receptor. To date, no insulin-induced effect has been reported that discriminates signaling via A- and B-type receptors. In fact, the functional significance of these IR isoforms remains unclear. Recent studies have shown that the insulin-producing pancreatic ␤ cell is a target for insulin action, with insulin effects on transcription, translation, Ca2⫹ flux, and exocytosis (Leibiger et al., 1998a, 2000; Xu and Rothenberg, 1998; Xu et al., 1998; Aspinwall et al., 1999; Kulkarni et al., 1999a). In an animal model with a ␤ cell–specific knockout for IR, there is a decrease in glucose-stimulated insulin release and a decrease in the insulin content of the cell (Kulkarni et al., 1999a). In addition, disruption of insulin signaling in the ␤ cell at the level of insulin receptor substrate (IRS)-1 (Kulkarni et al., 1999b) or IRS-2 (Withers et al., 1998) leads to altered growth and function of the ␤ cell. Consequently, insulin resistance may not only affect the function of the “classical” insulin target tissues muscle, fat and liver, but also apply to the pancreatic ␤ cell and thereby affect ␤ cell function. In the present study, we show selective insulin signaling via the two isoforms of the insulin receptor (i.e., IR-A and IR-B) in the pancreatic ␤ cell. Insulin that is secreted by ␤ cells upon glucose stimulation up-regulates transcription of its own gene as well as that of the ␤ cell transcription unit of the glucokinase (␤GK) gene in an autocrine feedback loop. More interestingly, while the insulin gene is activated by insulin signaling via IR-A involving PI3K class Ia, p70 s6 kinase (p70s6k), and Ca2⫹/calmodulin dependent kinases, insulin-stimulated ␤GK transcription occurs via IR-B, PI3K class II–like activity, and protein kinase B (PKB/c-Akt). These results provide evidence that signaling via either IR-A or IR-B and the subsequent activation of different classes of PI3K and protein kinases (i.e., p70s6k and PKB) represent a mechanism for selective insulin action. We furthermore show a preferential activation of p70s6k and PKB as a result of insulin signaling via IR-A and IR-B, respectively, in insulin-producing and non-insulin-producing cells. Results and Discussion Glucose Activates Glucokinase Gene Transcription via Secreted Insulin Insulin, secreted upon glucose stimulation, is a key factor in the up-regulation of insulin gene transcription (Leibiger et al., 1998a). The promoters of both the insulin gene and the ␤GK gene contain many similar cis ele-

Molecular Cell 560

ments (Shelton et al., 1992; Leibiger et al., 1994a, 1994b; Watada et al., 1996). To test whether transcription of ␤GK is regulated by similar mechanisms as the insulin gene, we studied the role of glucose and insulin in regulation of ␤GK mRNA steady-state levels. Stimulation of cultured islets (Figure 1A) or insulin-producing HIT-T15 cells with 16.7 mM glucose led to an increase in ␤GK mRNA levels 60 min following start of stimulation. This is similar in time course to the effect of glucose to stimulate insulin mRNA levels (Leibiger et al., 1998a, 1998b). To define in more detail the dynamics of ␤GK mRNA, we analyzed the half-life time, stability, and transcriptional rate of the ␤GK mRNA pool. As shown in Figure 1B, the half-life time of ␤GK mRNA was ⵑ60 min and was not changed in the presence or absence of glucose. On the other hand, stimulation of HIT cells with 16.7 mM glucose led to an increase in ␤GK gene transcripts as early as 15 min and reached a maximum of transcriptional activity at 30 min in a nuclear run-off assay (Figure 1C). This effect of glucose on ␤GK transcription initiation was also observed in normal pancreatic islets (Figure 1D). To further corroborate these data, we established a reporter gene assay using the ␤GK promoter coupled to the green fluorescent protein (GFP) (pr␤GK.GFP). We used the rat ␤GK promoter fragment up to nucleotide ⫺278, since this has been shown to contain all cis elements responsible for both glucose-dependent and celltype-specific transcriptional control (Jetton et al., 1994, 1998). Stimulation with 16.7 mM glucose led to an increase in ␤GK promoter–driven GFP fluorescence in HIT cells, isolated primary pancreatic ␤ cells, and intact pancreatic islets (Figure 1E). As with the nuclear run-off assay, the dynamics of the activation of ␤GK promoter– driven GFP expression were similar, if not identical, to those of the glucose-stimulated insulin gene promoter (Leibiger et al., 1998a, 1998b). To determine whether glucose metabolism per se or secreted insulin is a requirement for the up-regulation of ␤GK transcription, we investigated the effect of insulin secretagogues on ␤GK mRNA steady-state levels and ␤GK promoter–driven GFP expression. Insulin secretagogues, like KCl or the sulfonylurea compound glibenclamide, stimulate insulin secretion by depolarizing the ␤ cell plasma membrane and provoking influx of extracellular Ca2⫹ through voltage-gated L-type Ca2⫹ channels (reviewed in Berggren and Larsson, 1994). As shown in Figure 2, stimulation with either 50 mM KCl or 1 ␮M glibenclamide for 5 min, at substimulatory glucose concentrations, led to an increase in ␤GK mRNA steadystate levels (Figure 2A) and to an elevation in ␤GK promoter–driven GFP expression (Figure 2C). Alternatively, preventing stimulus-induced insulin secretion by blocking L-type Ca2⫹ channels using nifedipine abolished upregulation of ␤GK mRNA levels (Figure 2B). We next studied the effect of exogenously administered insulin on ␤GK mRNA steady-state levels and ␤GK promoter–driven GFP expression at substimulatory glucose concentrations (Figures 2D and 2E). Addition of only 50 ␮U of insulin per ml to fully supplemented culture medium was sufficient to evoke ␤GK mRNA levels in pancreatic islets (Figure 2D). Interestingly, a more careful comparison of the necessary amounts of exogenous insulin to trigger promoter activities revealed that instead of 5–10 ␮U of insulin per ml, as is the case with the

insulin gene, the addition of 20 ␮U per ml was required to gain an effect on ␤GK promoter activation (Figure 2E). Stimulation with 5 mU of insulin per ml of culture medium for 5 min led to an ␤GK promoter–driven increase in GFP fluorescence in isolated primary pancreatic ␤ cells (Figure 2C), HIT cells, and intact pancreatic islets (data not shown). Thus, our data support the view that the insulin gene and the ␤GK gene are both stimulated by insulin secreted in response to glucose. Interestingly, a higher concentration of insulin is needed to activate ␤GK transcription when compared with the insulin gene. Insulin-Stimulated Glucokinase Gene Transcription Utilizes Signal Transduction, which Is Different from that of the Insulin Gene Our studies on insulin-stimulated insulin gene transcription have shown the involvement of PI3K, p70s6k, and Ca2⫹/calmodulin-dependent kinase(s) in the signaling cascade (Leibiger et al., 1998a). Because previous data from others and our laboratory suggest that insulin- and ␤GK-promoters can bind the same transcription factors (Shelton et al., 1992; Leibiger et al., 1994a, 1994b; Watada et al., 1996) and both genes respond positively to many of the same stimuli (glucose, insulin, secretagogs) at the level of transcription, we questioned whether both genes might be regulated by the same signaling pathway. To test whether the same protein kinases that are involved in insulin-triggered insulin gene transcription contribute to insulin-triggered transcription of ␤GK, we studied the effect of pharmacological inhibitors on insulin-stimulated ␤GK promoter activity (Figure 3). We combined insulin stimulation (5 mU/ml for 5 min at substimulatory glucose concentrations) with the cotreatment of islet cells and HIT cells with inhibitors of protein kinase C (PKC; 150 nM bisindolylmalemide I [BIM]), PI3K (25 ␮M LY294002 [LY]), p70s6k (10 nM rapamycin [rap]), MAP kinases Erk1/2 (20 ␮M PD98059 [PD9]) and p38/ RK/SAPK2a ⫹ SAPK1/JNK (10 ␮M PD169316 [PD1]), IR tyrosine kinase (100 ␮M HNMPA-(AM)3 [HNMPA]), and Ca2⫹/calmodulin-dependent kinase II (CaMKII; 400 nM autocamtide-2 related inhibitory peptide [AC]). The efficiency of these inhibitors was verified by the respective protein kinase assays in cell lysates of inhibitor-treated and nontreated cells following glucose/insulin stimulation (data not shown). In agreement with the data on insulin-stimulated insulin gene transcription, insulinstimulated ␤GK transcription was not sensitive to inhibition of PKC or MAP kinases Erk1/2 and p38 but was sensitive to inhibition of IR tyrosine kinase by HNMPA(AM)3 (Figure 3A). However, to our surprise, insulin-stimulated ␤GK transcription was not inhibited by LY294002, rapamycin, or autocamtide-2 related inhibitory peptide, suggesting that signaling via PI3K/p70s6k and via CaMKII, respectively, is not involved (Figures 3A and 3B). To further confirm that insulin stimulates insulin gene transcription and ␤GK transcription using different signaling pathways, we established a technique that allowed monitoring of insulin and ␤GK promoter activities simultaneously in the same cell. In addition to pr␤GK.GFP, we generated an expression construct where the rat insulin I promoter (⫺410/⫹1 bp) controlled the expres-

Selective Signaling via A and B Insulin Receptors 561

Figure 1. Effect of Glucose on ␤GK mRNA Steady-State Levels, Transcription Initiation, and mRNA Stability (A) Elevation of ␤GK mRNA steady-state levels in isolated islets after stimulation with 16.7 mM glucose (15 min). (B) Dynamics of ␤GK mRNA stability in islet cells at 3 mM glucose (closed squares) and after stimulation with 16.7 mM glucose for 15 min (open squares). Actinomycin D (5 ␮g/ml) was present all the time under nonstimulatory conditions (closed squares), whereas in the case of stimulation (open squares) the inhibitor was added 45 min after start of stimulation. In (A) and (B), ␤GK mRNA values are presented as percentages of mRNA levels of the nonstimulated control at minute 0 (given as 100%). (C and D) Dynamics of ␤GK transcription initiation in response to glucose stimulation in HIT cells (C) and isolated islets (D). Transcription initiation was studied by nuclear run-off analysis. Elevation of RNA levels in stimulated cells is shown as the percentage of RNA levels of the nonstimulated control (given as 100%). In (A)–(D), all data are shown as mean values ⫾ S.E. (n ⫽ 3). (E) On-line monitoring of glucose-stimulated ␤GK promoter–driven GFP expression in transfected HIT-T15 cells, islet cells, and whole islets. Representative images of HIT cells (n ⫽ 40), islet cells (n ⫽ 40), and islets (n ⫽ 3) are shown 60 and 240 min after start of glucose stimulation. The pseudo-color images were created by converting the original “gray-scale” data using Isee software; the fluorescence increases from blue to red. Scale bars, 10 ␮m.

Molecular Cell 562

Figure 2. Effect of Secretagogues, VoltageDependent L-Type Ca2⫹ Channel Blockers, and Exogenous Insulin on Endogenous ␤GK mRNA Levels and ␤GK Promoter–Driven GFP Expression (A) Elevation of endogenous ␤GK mRNA levels in cultured pancreatic islets in response to stimulation for 5 min with either 50 mM KCl (KCl) or 1 ␮M glibenclamide (glib) at 3 mM glucose. (B) Elevation of endogenous ␤GK mRNA levels in islet cells in response to stimulation for 15 min with 16.7 mM glucose with or without 10 ␮M nifedipine (nif). In (A) and (B), data are shown as mean values ⫾ S.E. (n ⫽ 3), and amounts of ␤GK mRNA are presented as the percentage of mRNA levels of the nonstimulated control (given as 100%). (C) On-line monitoring of ␤GK promoter– driven GFP expression in islet cells. Islet cells were transfected with pr␤GK.GFP (␤GK) or with pRcCMV.GFP (CMV) as control and incubated with 16.7 mM glucose, 50 mM KCl, 1 ␮M glibenclamide (glib) or 5 mU/ml insulin (ins). Data are shown as mean values ⫾ S.E. (n ⫽ 8). (D and E) Effect of increasing concentrations of insulin added to the culture medium for 5 min, on (D) endogenous ␤GK mRNA levels in isolated pancreatic islets and on (E) ␤GK promoter–driven GFP expression and insulin promoter–driven DsRed expression in transfected HIT cells. In (D), amounts of ␤GK mRNA are presented as the percentage of mRNA levels of nonstimulated control (given as 100%), and data are shown as mean values ⫾ S.E. (n ⫽ 3). In (E), HIT cells were cotransfected with pr␤GK.GFP (open bars) and prIns1.DsRed (closed bars) and stimulated for 5 min with the indicated amounts of exogenous insulin. On-line monitoring data are presented as the ratio of fluorescence obtained at minutes 240 and 60 and represent mean values ⫾ S.E. (n ⫽ 7).

sion of the red fluorescent protein DsRed (Matz et al., 1999), prIns1.DsRed. Because of their different excitation and emission profiles, the signals generated by the two fluorescent proteins can be measured directly in the same cell. Following cotransfection of islet cells and HIT cells with pr␤GK.GFP and prIns1.DsRed, insulinstimulated insulin and ␤GK promoter activities were monitored as DsRed and GFP fluorescence, respectively. As shown in Figure 3C, stimulation with 5 mU/ ml insulin led to an elevation in both DsRed and GFP fluorescence, as expected. This increase in fluorescence was abolished when the cells were treated with HNMPA-(AM)3, thereby blocking the tyrosine kinase activity of IRs. By combining insulin stimulation with pharmacological inhibitors of either PI3K (LY294002), p70s6k (rapamycin), or CaMKII (autocamtide-2 related inhibitory peptide), we show that activation of the insulin promoter is abolished (no increase in DsRed fluorescence). On the other hand, no effect on insulin-stimulated ␤GK promoter activity (increase in GFP fluorescence) was observed in the same cell (Figure 3C). Thus, insulin activates the ␤GK promoter by em-

ploying a signaling pathway that is different from that utilized by the insulin promoter. Besides signaling via the MAP kinase and the PI3K/ mTOR/p70s6k pathways, insulin has been shown to exert its effect via the activation of PKB(c-Akt) (Coffer et al., 1998). To test whether stimulation with either glucose or insulin leads to the activation of PKB in pancreatic ␤ cells, we studied PKB activity following stimulation with either 16.7 mM glucose or 5 mU of insulin/ml. As shown in Figure 4, PKB activation was observed 5 min following stimulation with 16.7 mM glucose (Figure 4A) and 2 min following stimulation with 5 mU of insulin/ml, at substimulatory glucose concentrations (Figure 4B). Prevention of glucose-induced insulin secretion by treatment of insulin-producing cells with the L-type Ca2⫹ channel blocker nifedipine abolished glucose-induced activation of PKB, as did inhibition of insulin signaling by HNMPA(AM)3 (data not shown). These data suggest that PKB is activated in response to glucose-stimulated insulin secretion. Because of the lack of a selective pharmacological inhibitor of PKB, we tested its involvement in insulin-stimulated ␤GK gene transcription by transiently

Selective Signaling via A and B Insulin Receptors 563

Figure 3. Effect of Various Protein Kinase Inhibitors on Insulin-Stimulated Insulin and ␤GK Promoter Activity and Endogenous ␤GK mRNA Levels (A) On-line monitoring of insulin promoter–driven (closed bars) and ␤GK promoter–driven (open bars) GFP expression in transfected islet cells. Data are presented as the ratio of fluorescence obtained at minutes 240 and 60 and represent mean values ⫾ S.E. (n ⫽ 10). (B) Amounts of ␤GK mRNA are presented as the percentage of mRNA levels of nonstimulated control (given as 100%). Data are shown as mean values ⫾ S.E. (n ⫽ 3). (C) HIT cells were cotransfected with pr␤GK.GFP (open bars) and prIns1.DsRed (closed bars). On-line monitoring data are presented as the ratio of fluorescence obtained at minutes 240 and 60 and represent mean values ⫾ S.E. (n ⫽ 13).

overexpressing PKB␣/c-Akt1. Whereas overexpression of PKB␣ had no effect on insulin-stimulated insulin gene transcription, it led to a more pronounced effect on insulin-stimulated ␤GK promoter–driven GFP expression (Figure 4C). According to the current view, insulin-stimulated PKB activation involves the phosphorylation of PKB by the phosphoinositol-dependent kinase 1, PDK1 (Vanhaesebroeck and Alessi, 2000). Indeed, transient overexpression of PDK1 led to a pronounced stimulation of insulin-triggered ␤GK promoter activity, whereas overexpression of the antisense transcript of PDK1 abolished the stimulatory effect of insulin on insulin-triggered ␤GK promoter activity (Figure 4C). Noteworthily, transient overexpression of either PKB␣ or PDK1 did not lead per se to an increased basal insulin or ␤GK promoter activity (data not shown). Interestingly, the activation of PKB has so far been shown to be dependent on the activity of PI3K (Vanhaesebroeck and Alessi, 2000) and therefore to be sensitive to the independent pharmacological inhibitors wortmannin and LY294002. Whereas treatment of insulin-producing cells with 25 ␮M LY294002 clearly abolished insulinstimulated rat insulin I gene promoter activity, it did not block insulin-stimulated rat ␤GK promoter activity (Figure 3). When analyzing the effect of LY294002 on insulin-stimulated insulin and ␤GK promoter activity in a dose-dependent manner in cells cotransfected with pr␤GK.GFP and prIns1.DsRed, we observed that LY294002 inhibited the two promoters at different concentrations. Whereas 25 ␮M LY294002 blocked insulinstimulated insulin promoter activity, 100 ␮M LY294002 was needed to completely abolish insulin-stimulated ␤GK promoter activity (Figure 4D). The effect of wortmannin was similarly concentration dependent. Treatment of cells with 50 nM wortmannin was sufficient to inhibit insulin-stimulated insulin promoter activity,

whereas 150 nM wortmannin was necessary to block insulin-stimulated ␤GK promoter activity (Figure 4E). These data indicate that insulin-stimulated ␤GK gene transcription occurs by signaling via PDK1/PKB, whereas insulin-stimulated insulin gene transcription is mediated via PI3K/p70s6k and CaMKII. Insulin Signaling via IR-A Activates Insulin Gene Promoter Whereas Signaling via IR-B Activates ␤GK Promoter Previous data on insulin-stimulated insulin gene transcription (Leibiger et al., 1998a) favored signaling via IR but did not exclude the signaling via IGF-I receptors or possible hybrids of insulin- and IGF-I receptors. The loss of insulin effect when treating cells with HNMPA(AM)3, an inhibitor of the IR tyrosine kinase (Saperstein et al., 1989), supported the idea that signaling via IR is crucial. Consequently, we examined whether the expression of IR per se is an absolute requirement for insulin-stimulated insulin and ␤GK gene expression. Therefore, we analyzed insulin and ␤GK mRNA levels in response to glucose/insulin stimulation in isolated islets from ␤IRKO mice, a knockout model that lacks the expression of IR specifically in the pancreatic ␤ cell (Kulkarni et al., 1999a). Stimulation with either 16.7 mM glucose or 5 mU of insulin/ml led to an increase in both endogenous insulin and ␤GK mRNA levels in islets of wild-type mice, whereas no increase in insulin and ␤GK mRNA levels was observed in islets prepared from ␤IRKO mice (Figure 5A). These data suggest that the expression of the IR in pancreatic ␤ cells is an absolute requirement to gain the stimulatory effect by insulin on both insulin and ␤GK gene expression and that signaling via IGF-I receptors is unlikely to be involved. This is consistent with the finding that activation of IGF-I receptors by stimulation with 2.6 nM IGF-I did not activate

Molecular Cell 564

Figure 4. Role of PKB and PI3K in InsulinStimulated Insulin Gene and ␤GK Transcription (A and B) Dynamics of PKB activities following stimulation with 16.7 mM glucose (A) or 5 mU of insulin per ml in HIT cells (B). Activities of PKB are presented as the percentage of the activity of the nonstimulated control (given as 100%). Data are shown as mean values ⫾ S.E. (n ⫽ 3). (C) On-line monitoring of HIT cells cotransfected with pr␤GK.GFP (open bars), prIns1.DsRed (closed bars), and either kinase-inactive mutant of PKB␣, i.e., PKB␣⌬308/437 (mock), wild-type PKB␣ (PKB), wild-type PDK1 (PDK1), or an antisense construct of PDK1 (PDK1antisense). Data are presented as the ratio of fluorescence obtained at minutes 240 and 60 and represent mean values ⫾ S.E. (n ⫽ 8). (D and E) Effect of different concentrations of PI3K inhibitors LY294002 (D) and wortmannin (E) on ␤GK promoter–driven GFP expression (open bars) and insulin promoter–driven DsRed expression (closed bars) in cotransfected HIT cells. Data are presented as the ratio of fluorescence obtained at minutes 240 and 60 and represent mean values ⫾ S.E. (n ⫽ 7).

insulin promoter– or ␤GK promoter–driven reporter gene expression in insulin-producing cells (data not shown). Employing RT–PCR with subsequent DNA sequence analysis, we have previously observed that insulin-producing cells express both IR-A and IR-B (Leibiger et al., 1998a). Transient overexpression of IR-A led to a pronounced effect of insulin stimulation on insulin promoter activity, whereas overexpression of IR-B had no effect. To test a similar effect for insulin-stimulated ␤GK transcription, we cotransfected islet cells and HIT cells with pr␤GK.GFP and prIns1.DsRed in combination with either IR-A or IR-B. To our surprise, we found that overexpression of IR-B led to a pronounced activation of the ␤GK promoter while overexpression of IR-A had no effect (Figure 5B). Moreover, overexpression of an inactive IR-B mutant (IR-Bm), described as M1153I (Levy-Toledano et al., 1994), did not lead to a further effect on r␤GK promoter–driven GFP expression (Figure 6B). To test the involvement of IR-B in insulin-stimulated ␤GK promoter activation in more detail, we treated islet cells and HIT cells, prior to glucose/insulin stimulation, with an anti-IR-B antibody. This antibody selectively binds to the ␣ chain of IR-B and therefore selectively blocks insulin binding to IR-B and signaling via IR-B.

Employing coexpression of pr␤GK.GFP and prIns1.DsRed in the same cell, we observed that treatment with the B-type receptor-specific antibody (␣IR-B) abolished insulin-stimulated pr␤GK.GFP expression, whereas it did not affect insulin-stimulated prIns1.DsRed expression (Figure 5C). In addition, treatment of insulin-producing cells with ␣IR-B abolished elevation of ␤GK mRNA levels following stimulation with either 16.7 mM glucose or 5 mU of insulin/ml at substimulatory glucose concentrations (data not shown). As expected, treatment of transfected cells with an antibody that blocks insulin signaling via both receptor isoforms (␣IR-AB) suppressed insulin-stimulated activation of ␤GK- and insulin-promoters (Figure 5C). Accordingly, treatment of insulin-producing cells with this antibody also abolished insulin-stimulated elevation of ␤GK mRNA steady-state levels (data not shown). By contrast, treatment of transfected cells with an antibody that blocks signaling via IGF-I receptors (␣IGF-1R) did not affect insulin-stimulated activation of ␤GK promoter– and insulin promoter– driven reporter gene expression (Figure 5C). These data indicate that insulin stimulates the insulin gene promoter through IR-A, whereas it stimulates the ␤GK gene promoter via IR-B.

Selective Signaling via A and B Insulin Receptors 565

Figure 5. The Role of Insulin Receptors (A) Effect of stimulation with either 16.7 mM glucose (15 min) or 5 mU of insulin per ml (5 min) on endogenous ␤GK mRNA steady-state levels in isolated pancreatic islets obtained from ␤IRKO mice (␤IRKO) or control animals (wild type). The values of ␤GK mRNA are presented as percentages of mRNA levels of the nonstimulated islets (given as 100%). Data are shown as mean values ⫾ S.E. (n ⫽ 3). (B) On-line monitoring of ␤GK promoter–driven GFP expression (open bars) and insulin promoter–driven DsRed expression (closed bars) in cotransfected islet cells. Cells were cotransfected with pr␤GK.GFP and prIns1.DsRed and either expression constructs for wild-type isoforms of IR-A, IR-B, or the M1153I mutant of the respective receptor isoform, i.e., IR-Am and IR-Bm, respectively. Data are presented as the ratio of fluorescence obtained at minutes 240 and 60 and represent mean values ⫾ S.E. (n ⫽ 10). (C) Effect of antibodies that block signaling through IR-A and IR-B (␣IR-AB), through IR-B (␣IR-B), and through IGF-I receptors (␣IGF-1R) on insulin-stimulated ␤GK promoter–driven GFP expression (open bars) and insulin promoter–driven DsRed expression (closed bars) in cotransfected islet cells. Cells were incubated with a 0.67 ␮g/ml concentration of the respective antibodies 30 min prior to stimulation and throughout stimulation. Data are presented as the ratio of fluorescence obtained at minutes 240 and 60 and represent mean values ⫾ S.E. (n ⫽ 10).

To start to understand the molecular mechanisms that underlie the selectivity in insulin signaling via the two IR isoforms, we aimed to explain the different sensitivity for PI3K inhibitors we observed between insulin-stimulated insulin- and ␤GK-promoter activation (Figures 4D and 4E). One possible interpretation would be that the same PI3K is involved in the transcription of both genes but that a lower PI3K activity is sufficient to trigger the cascade that activates ␤GK gene transcription via PKB. If this is the case, inhibition of PI3K-mediated ␤GK transcription should require higher concentrations of wortmannin or LY294002 to be fully effectuated. This should also be reflected upon by an inverse relationship between required inhibitor concentration and sensitivity of the respective promoter activity to insulin. If the same PI3K is sufficient to lead to the activation of both insulinand ␤GK-promoters, then the promoter activity that is least sensitive to the inhibitors (i.e., ␤GK) would be expected to require less insulin to become stimulated. As demonstrated in Figure 2E, this is not the case. On the contrary, more insulin is needed to stimulate ␤GK promoter–driven GFP expression. Another interpretation is that the different sensitivity in vivo could be due to a different accessibility of the inhibitor for the same type of PI3K as a result of a different distribution/localization of the two IR isoforms, or it could be due to the involvement of different classes of PI3K, exhibiting a different sensitivity to wortmannin and LY294002 as described for PI3K classes I and III versus class II (reviewed in Fruman et al., 1998). To test whether IR-A and IR-B exhibit a distinct distribution in vivo, we tagged both

receptor isoforms with GFP and DsRed at the C terminus of the ␤ subunit. Tagging both IR isoforms did not interfere with their physiological function (e.g., overexpression of the tagged IR isoform led to a pronounced insulin effect on the respective promoter activity to the same extent as the untagged IR) (data not shown). Whereas transient coexpression of the same, but differently tagged (IR-AⵑDsRed/IR-AⵑGFP and IR-BⵑGFP/ IR-BⵑDsRed), IR isoform led to a complete colocalization (data not shown), coexpression of the differently tagged IR-A and IR-B in either combination (IR-Aⵑ DsRed/IR-BⵑGFP and IR-AⵑGFP/IR-BⵑDsRed) clearly showed IR isoforms that are not colocalized. This pattern of distinct IR isoform distribution was observed in insulin-producing cells HIT (Figure 6A), INS1, and MIN6, as well as in non-insulin-producing cells HEK293 and COS7 (data not shown). To test whether the two IR isoforms do utilize different classes of PI3K, we overexpressed either IR-AⵑGFP or IR-BⵑGFP in HIT cells and studied the sensitivity of PI3K activity to wortmannin in vitro following immunoprecipitation with GFP antibodies. Whereas the PI3K activity in the IR-A immunoprecipitate was inhibited by wortmannin in the lower nanomolar range, as typical for PI3K class I and III, the PI3K activity in the IR-B immunoprecipitate was only inhibited at higher concentrations (Figure 6B), as described for PI3K class II (see Fruman et al., 1998). To test whether insulin-stimulated insulin gene transcription involves IR-A-mediated insulin signaling via PI3K class Ia, we combined insulin stimulation with the transient overexpression of the dominant-negative form of the PI3K

Molecular Cell 566

Figure 6. Molecular Mechanisms Involved in the Selective Insulin Signaling via IR-A and IR-B (A) Distribution of IR-AⵑGFP (green) and IR-BⵑDsRed (red) in HIT cells obtained by laser scanning confocal microscopy. Areas in yellow indicate colocalization of the two IR isoforms. This is a representative image out of a total of 25. (B) PI3K activity in GFP immunoprecipitates obtained from insulin-stimulated (150 nM insulin for 5 min) HIT cells overexpressing either IRAⵑGFP (closed bars) or IR-BⵑGFP (open bars). The amount of wortmannin included in the in vitro assay is indicated. Data are presented as mean values ⫾ S.E. (n ⫽ 3). (C) Effect of overexpression of dominant-negative p85 PI3K subunit, ⌬p85, on insulin-stimulated ␤GK promoter–driven GFP expression (open bars) and insulin promoter–driven DsRed expression (closed bars) in cotransfected HIT cells. Data are presented as the ratio of fluorescence obtained at minutes 240 and 60 and represent mean values ⫾ S.E. (n ⫽ 10). (D and E) Analysis of insulin-stimulated p70 s6 kinase and PKB activities in HIT (D) and HEK293 (E) cells following transfection with IR-A or IR-B. Cells were stimulated with insulin for 10 min and lysed after a further 10 min. Data are represented as percentages of the nonstimulated, mock-transfected control, set as 100%, and presented as mean values ⫾ S.E. (n ⫽ 3).

class Ia adaptor protein p85 (i.e., ⌬p85). Whereas transient overexpression of ⌬p85 totally abolished insulinstimulated insulin promoter activity, this approach had no effect on insulin-stimulated ␤GK promoter activation (Figure 6C). Taken together, these data suggest a selectivity in insulin signaling in insulin-producing cells via IR-A

through PI3K class Ia and p70s6k on the one hand, and via IR-B through a different PI3K activity, very similar to that of class II, and PKB on the other. When separately overexpressing IR isoforms in HIT cells, we observed a more pronounced activation of p70s6k in cells overexpressing IR-A in response to insulin stimulation, while cells overexpressing IR-B showed

Selective Signaling via A and B Insulin Receptors 567

Figure 7. Selective Activation of Insulin and Glucokinase Gene Transcription by Selective Insulin Signaling via A- and B-Type Insulin Receptors The scheme illustrates the coupling between insulin exocytosis and the activation of transcription of insulin and glucokinase genes.

a trend toward a higher PKB activity (Figure 6D). To test whether this effect is purely specific for the pancreatic ␤ cell, we analyzed both PKB and p70s6k activities in non-insulin-producing HEK293 cells that were transiently overexpressing either IR-A or IR-B. Most interestingly, insulin-stimulated p70s6k activity here was also more pronounced in cells overexpressing IR-A, whereas insulin-stimulated PKB activity seemed to be more tightly coupled with the overexpression of IR-B (Figure 6E). Conclusion Selectivity in insulin signaling is currently discussed as the result of the activation of specific signal transduction pathways. This selectivity may be gained by activating specific adaptor proteins (i.e., IRS and Shc proteins) that “channel” the insulin signal in a more defined way by specifically interacting with downstream located effector proteins (Myers and White, 1996; Virkama¨ki et al., 1999). Whereas the importance of IRS proteins in achieving insulin effects in different tissues is currently under extensive investigation, the possibility of selective insulin signaling via the two isoforms of the IR has been neglected. Studies on general and tissue-specific IR knockout models have demonstrated that a defect IRmediated insulin signaling leads to a type 2 diabetes–like phenotype (reviewed in Taylor, 1999). However, these knockouts do not discriminate between the two IR isoforms. This is of importance, since earlier studies clearly established differences in tissue-specific IR isoform expression as well as in their activation profile. Whereas IR-B, which shows a 2-fold lesser affinity for insulin in comparison to IR-A (Mosthaf et al., 1990; Yamaguchi et al., 1991; McClain, 1991), is predominantly expressed in liver and muscle, IR-A is mainly expressed in brain (Moller et al., 1989; Seino and Bell, 1989; Mosthaf et al., 1990). Although it is tempting to link the phenotypes of the liver- and brain-specific IR knockouts (Bru¨ning et

al., 2000; Michael et al., 2000) to the failure in IR-B or IR-A function, respectively, a direct proof of the involvement of IR isoforms remains to be shown in IR isoform– specific knockout models. Attempts to correlate tissuespecific expression of IR isoforms with diabetes mellitus has generated conflicting results (Mosthaf et al., 1991; Benecke et al., 1992; Norgren et al., 1993) that do not clarify the functional role of either isoform. Little is known about selective insulin signaling via Aand B-type IR. Besides the affinity for insulin, differences in their kinase activity (Kellerer et al., 1992) as well as internalization and recycling (Vogt et al., 1991; Yamaguchi et al., 1991) have been described. These data have implied differences in the function of either IR isoform, but no isoform-specific insulin-induced effect has been reported so far. In the present paper, we provide a “readout” system for discriminating selective signaling via the two IR isoforms. We have demonstrated that the molecular basis for this selectivity could be provided by the different localization of the two IR isoforms in the plasma membrane and their different sensitivity for insulin. Mechanistically, this enables preferential activation of IR-A/PI3K Ia/p70s6k in pancreatic ␤ cell glucose/ insulin–stimulated insulin gene transcription and IR-B/ PI3K class II–like/PKB in glucose/insulin–stimulated ␤GK transcription (Figure 7). That this specificity in signaling has a wider implication than to the pancreatic ␤ cell is demonstrated by the data obtained in non-insulinproducing HEK293 cells. Thus, our data clearly demonstrate that selectivity of insulin signaling can be gained by signaling through the two IR isoforms and reinforce the concept of the pancreatic ␤ cell as a target for positive insulin action. Experimental Procedures Materials Bisindolylmalemide I, PD98059, wortmannin, LY294002, rapamycin, HNMPA-(AM)3, autocamtide-2 related inhibitory peptide, and nifedi-

Molecular Cell 568

pine were purchased from Calbiochem. Actinomycin D was from Sigma. Microcystin-LR and PD169316 were purchased from Alexis Biochemicals. Rabbit anti-Insulin Receptor ␣ and Rabbit anti-Insulin Receptor B antibodies were from Biodesign. Mouse monoclonal IGF-IR␣ was from Pharmingen. Oligonucleotides were synthesized at Genset (France). Expression Constructs The construction of prIns1.GFP has been described earlier (Leibiger et al., 1998a). prIns1.DsRed was generated by exchanging the GFP expression cassette versus the DsRed-expression cassette from pDsRed1–1 (Clontech). pr␤GK.GFP was generated by exchanging the CAT expression cassette of pr␤GK-278.CAT (Leibiger et al., 1994a) versus the GFP-expression cassette obtained from pRcCMVi.GFP (Moede et al., 1999). Constructs for expression of the human IR-A and -B types, pRcCMVi.hIR(A) and pRcCMVi.hIR(B), were generated by subcloning the respective IR-A and IR-B sequences from pCMVHIR(A) and pCMVHIR(B) (kindly provided by A. Ullrich, MPI for Biochemistry, Martinsried, Germany) into pRcCMVi. GFP- and DsRed-tagged variants were generated by introducing a ClaI-site into the IR␤ subunit coding sequence 69 nucleotides in front of the stop codon and subcloning of the GFP or DsRed cDNA, respectively, in-frame. Site-directed mutagenesis to introduce the M1153I mutation into the IR cDNA was performed by employing the QuikChange Mutagenesis Kit (Stratagene). Adenovirus-based vector Ad.r␤GK.GFP was constructed by subcloning the r␤GK.GFP cassette into pAC.CMV.pLpA and performing homologous recombination as described in Moitoso de Vargas et al. (1997). All vector constructions were verified by DNA sequence analysis. Plasmids pCMV5.PKB␣, pCMV5.PKB␣⌬308/437, pCMV5.PDK1, and pCMV5.PDK1 antisense were kindly provided by D.R. Alessi (MRC Phosphorylation Unit, University of Dundee, UK), and pcDNA3Zeo.⌬p85 was a gift from C.P. Downes (Department of Biochemistry, University of Dundee, UK). Cell Culture and Transfection Isolation of pancreatic islets, the culture of islets, islet cells, and HIT-T15 cells, and their transfection have been described in Leibiger et al. (1998a; 1998b). Following transfection, HIT cells were cultured for 12–18 hr in RPMI 1640 medium containing 0.1 mM glucose, 10% fetal calf serum and supplemented with 100 U/ml penicillin, 100 ␮g/ml streptomycin, 2 mM glutamine at 5% CO2 and 37⬚C. Before stimulation, islets and islet cells were incubated for 2 hr in RPMI 1640 medium supplemented as above but containing 3 mM glucose. HEK293 cells were grown in DMEM containing 5.5 mM glucose, 10% fetal calf serum, 100 U/ml penicillin, 100 ␮g/ml streptomycin, 2 mM glutamine at 5% CO2 and 37⬚C, and were transfected by the calcium phosphate/coprecipitation technique as described earlier (Leibiger et al., 1994a). Islets were transduced with Ad.r␤GK.GFP as described in Moitoso de Vargas et al. (1997) and analyzed by laser scanning confocal microscopy. Cultured islets, islet cells, and HIT cells were stimulated with either 16.7 mM glucose for 15 min or 50 mM KCl, 1 ␮M glibenclamide or with various concentrations of insulin for 5 min at substimulatory glucose concentrations (3 mM for islets/islet cells and 0.1 mM for HIT cells). All experiments were performed in RPMI 1640 medium, supplemented as above. Protein kinase inhibitors, antibodies, or nifedipine were added to the culture medium 30 min prior to stimulation and were kept in the medium throughout stimulation. After stimulation, islets, islet cells, and HIT cells were cultured further at substimulatory glucose concentrations. For RNA analysis, islets and cells were harvested 60 min after start of stimulation, if not indicated otherwise. For analysis of protein kinase activities or tyrosine phosphorylation, cells were harvested immediately following stimulation. Nuclear Run-Off Analysis Nuclear run-off analysis was performed as described previously (Leibiger et al., 1998b), except that labeled RNA was hybridized to 2.5 ␮g of cDNA of glucokinase, insulin, ␤ actin, and control pBluescript DNA, which were immobilized on nitrocellulose filters. For nuclear run-off analysis on islets, nuclei from 2000 islets per experiment were used. Data were analyzed by phosphorimaging. Values obtained for ␤GK mRNA were normalized by ␤ actin mRNA values.

RNA Analysis Levels of ␤GK mRNA were analyzed by comparative RT–PCR as described in Leibiger et al. (1998b) using primers 5⬘-GTTCCTACTG GAGTATGACC-3⬘ and 5⬘-CCTCCTCTGATTCGATGAAG-3⬘ for characterizing ␤GK mRNA in HIT cells, and primers 5⬘-TGGATGACA GAGCCAGGATGG-3⬘ and 5⬘-ACTTCTGAGCCTTCTGGGGTG-3⬘ for ␤GK mRNA in rat and mouse pancreatic islets and islet cells. Levels of ␤ actin mRNA were analyzed by RT–PCR using primers 5⬘-AACTGG AACGGTGAAGGCGA-3⬘ and 5⬘-AACGGTCTCACGTCAGTGTA-3⬘. PCR conditions were chosen that guaranteed the amplification of ␤GK and actin fragments within the linear range, as verified by testing various numbers of amplification cycles (10–35). PCR products were separated on a 6% polyacrylamide sequencing gel and analyzed by phosphorimaging. Quantification was performed with TINA software 2.07d (Raytest). Values of ␤GK mRNA were normalized by ␤ actin values. Analysis of PKB Activity Analysis of PKB activity was performed employing the Akt1/PKB␣ Immunoprecipitation Kinase Assay Kit (Upstate Biotech.) according to the manufacturer’s instructions. Analysis of p70s6k Activity p70 s6 kinase from cell lysates was immunoprecipitated using a p70s6k antibody (Upstate Biotech.). Analysis of p70s6k activity was performed employing the S6 Kinase Assay Kit (Upstate Biotech.) according to the manufacturer’s instructions. Analysis of PI3K Activity HIT cells were transfected with either pRcCMVi.hIR(A)ⵑGFP or pRcCMVi.hIR(B)ⵑGFP. Cell lysates containing 2 mg of protein were subjected to immunoprecipitation using anti-GFP antibody A-11122 (Molecular Probes). PI3K activity was analyzed in the GFP immunoprecipitates as described in Krook et al. (1997) using L-␣-phosphatidylinositol from Avanti Polar-Lipids, Inc. On-Line Monitoring of GFP and DsRed Expression and Detection of Fluorescence Detection of fluorescence by digital imaging fluorescence microscopy was performed as described previously (Leibiger et al., 1998a; 1998b). The following filter settings were used: for GFPS65T: excitation at 485 nm, a 505 nm dicroic mirror, and an emission band-pass filter of 500–530 nm; for DsRed: excitation at 558 nm, a 565 nm dicroic mirror, and a 580 nm long-pass filter for emission. On-line monitoring was initiated 60 min following the start of stimulation and the 60 min value of GFP or DsRed fluorescence was set to 1.0. Cells to be monitored were chosen randomly at minute 60 from six fields of vision, and fluorescence was monitored up to minute 240. By overlaying the fluorescence and phase-contrast images, the positions of cells on the coverslip could be checked. Fluorescence intensity was calculated by using the Isee software for UNIX (Inovision Corporation). Laser scanning confocal microscopy was performed using a LEICA TCS SP2 (Leica Lasertechnik GmbH) with the following settings: Leica HCX PL APO 63⫻/1.20/0.17 UV objective lens, excitation wavelength 488 nm (Ar Laser) and 543 nm (HeNe laser), a 488/543 double dichroic mirror, and detection of GFP at 505–525 nm and DsRed at 605–670 nm. Laser scanning confocal microscopy of transduced islets was performed as described in Leibiger et al. (1998b). Acknowledgments The authors wish to thank Drs. A.M. Bertorello, K. Michelsen, T. Schwarz-Romond, X. Wang, and N. Welsh for sharing material and unpublished data. This work was supported by funds from Karolinska Institutet; by grants from Juvenile Diabetes Foundation International (JDFI), the Novo Nordisk Foundation, the Nordic Insulin Foundation Committee, EC Biotechnology Project (BIO4-CT98-0286), the Swedish Diabetes Association, and the Swedish Medical Research Council (72X-12594, 03X-13394, 72X-00034, 72XS-12708, 72X09890); and in part by NIH GRASP Center grant DK34928 and NIH

Selective Signaling via A and B Insulin Receptors 569

grants DK31036 and DK09825-02. L. M.V. is funded by a Career Development Award from JDFI.

insulin gene transcription by glucose. Proc. Natl. Acad. Sci. USA 95, 9307–9312.

Received August 21, 2000; revised January 8, 2001.

Leibiger, B., Wa˚hlander, K., Berggren, P.O., and Leibiger, I.B. (2000). Glucose-stimulated insulin biosynthesis depends on insulin-stimulated insulin gene transcription. J. Biol. Chem. 275, 30153–30156.

References Aspinwall, C.A., Lakey, J.R.T., and Kennedy, R.T. (1999). Insulinstimulated insulin secretion in single pancreatic beta cells. J. Biol. Chem. 274, 6360–6365. Benecke, H., Flier, J.S., and Moller, D.E. (1992). Alternatively spliced variants of the insulin receptor protein. Expression in normal and diabetic human tissues. J. Clin. Invest. 89, 2066–2070. Berggren, P.O., and Larsson, O. (1994). Ca2⫹ and pancreatic B-cell function. Biochem. Soc. Trans. 22, 12–18. Bru¨ning, J.C., Gautam, D., Burks, D.J., Gillette, J., Schubert, M., Orban, P.C., Klein, R., Krone, W., Mu¨ller-Wieland, D., and Kahn, C.R. (2000). Role of brain insulin receptor in control of body weight and reproduction. Science 289, 2122–2125. Coffer, P.J., Jin, J., and Woodgett, J.R. (1998). Protein kinase B (c-Akt): a multifunctional mediator of phosphoinositol 3-kinase activation. Biochem. J. 335, 1–13. Ebina, Y., Ellis, L., Jarnagin, K., Edery, M., Graf, L., Clauser, E., Ou, J.H., Masiarz, F., Kan, Y.W., Goldfine, I.D., et al. (1985). The human insulin receptor cDNA: the structural basis for hormone-activated transmembrane signalling. Cell 40, 747–758. Fruman, D.A., Myers, R.A., and Cantley, L.C. (1998). Phosphoinositide kinases. Annu. Rev. Biochem. 67, 481–507. Jetton, T.L., Liang, Y., Pettepher, C.C., Zimmerman, E.C., Cox, F.G., Horvath, K., Matschinsky, F.M., and Magnuson, M.A. (1994). Analysis of upstream glucokinase promoter activity in transgenic mice and identification of glucokinase in rare neuroendocrine cells in the brain and gut. J. Biol. Chem. 269, 3641–3654. Jetton, T.L., Moates, J.M., Lindner, J., Wright, C.V.E., and Magnuson, M.A. (1998). Targeted oncogenesis of hormone-negative pancreatic islet progenitor cells. Proc. Natl. Acad. Sci. USA 95, 8654–8659. Kellerer, M., Lammers, R., Ermel, B., Tippmer, S., Vogt, B., Obermaier-Kusser, B., Ullrich, A., and Haring, H.U. (1992). Distinct ␣-subunit structures of human insulin receptor A and B variants determine differences in tyrosine kinase activities. Biochemistry 31, 4588–4596. Krook, A., Whitehead, J.P., Dobson, S.P., Griffiths, M.R., Ouwens, M., Baker, C., Hayward, A.C., Sen, S.K., Maassen, J.A., Siddle, K., et al. (1997). Two naturally occuring insulin receptor tyrosine kinase domain mutants provide evidence that phosphoinositide 3-kinase activation alone is not sufficient for the mediation of insulin’s metabolic and mitogenic effects. J. Biol. Chem. 272, 30208–30214. Kulkarni, R.N., Bru¨ning, J.C., Winnay, J.N., Postic, C., Magnuson, M.A., and Kahn, C.R. (1999a). Tissue-specific knockout of the insulin receptor in pancreatic ␤ cells creates an insulin secretory defect similar to that of type 2 diabetes. Cell 96, 329–339. Kulkarni, R.N., Winnay, J.N., Daniels, M., Bru¨ning, J.C., Flier, S.N., Hanahan, D., and Kahn, C.R. (1999b). Altered function of insulin receptor substrate-1-deficient mouse islets and cultured ␤-cell lines. J. Clin. Invest. 104, R69–R75.

Levy-Toledano, R., Caro, L.H.P., Accili, D., and Taylor, S.I. (1994). Investigation of the mechanism of the dominant negative effect of mutations in the tyrosine kinase domain of the insulin receptor. EMBO J. 13, 835–842. Matz, M.V., Fradkov, A.F., Labas, Y.A., Savitsky, A.P., Zaraisky, A.G., Markelov, M.L., and Lukyanov, S.A. (1999). Fluorescent proteins from nonbioluminescent Anthozoa species. Nat. Biotech. 17, 969–973. McClain, D.A. (1991). Different ligand affinities of the two insulin receptor splice variants are reflected in parallel changes in sensitivity for insulin action. Mol. Endocrinol. 5, 734–739. Michael, M.D., Kulkarni, R.N., Postic, C., Previs, S.F., Shulman, G.I., Magnuson, M.A., and Kahn, C.R. (2000). Loss of insulin signaling in hepatocytes leads to severe insulin resistance and progressive hepatic dysfunction. Mol. Cell 6, 87–97. Moede, T., Leibiger, B., Pour, H.G., Berggren, P.O., and Leibiger, I.B. (1999). Identification of a nuclear localization signal, RRMKWKK, in the homeodomain transcription factor PDX-1. FEBS Lett. 461, 229–234. Moitoso de Vargas, L., Sobolewski, J., Siegel, R., and Moss, L.G. (1997). Individual ␤ cells within the intact islet differently respond to glucose. J. Biol. Chem. 272, 26573–26577. Moller, D.E., Yokota, A., Caro, J.F., and Flier, J.S. (1989). Tissuespecific expression of two alternatively spliced insulin receptor mRNA in man. Mol. Endocrinol. 3, 1263–1269. Mosthaf, L., Grako, K., Dull, T.J., Coussens, L., Ullrich, A., and McClain, D.A. (1990). Functionally distinct insulin receptors generated by tissue-specific alternative splicing. EMBO J. 9, 2409–2413. Mosthaf, L., Vogt, B., Ha¨ring, H., and Ullrich, A. (1991). Altered expression of insulin receptor types A and B in the skeletal muscle of non-insulin-dependent diabetes mellitus patients. Proc. Natl. Acad. Sci. USA 88, 4728–4730. Myers, M.G., Jr., and White, M.F. (1996). Insulin signal transduction and the IRS proteins. Annu. Rev. Pharmacol. Toxicol. 36, 615–658. Norgren, S., Zierath, J., Galuska, D., Wallberg-Henriksson, H., and Luthman, H. (1993). Differences in the ratio of mRNA encoding two isoformes of the insulin receptor between control and NIDDM patients: the RNA variant with exon 11 predominates in both groups. Diabetes 42, 675–681. Saperstein, R., Vicario, P.P., Strout, H.V., Brady, E., Slater, E.E., Greenlee, W.J., Ondeyka, D.L., Patchett, A.A., and Hangauer, D.G. (1989). Design of a selective insulin receptor tyrosine kinase inhibitor and its effect on glucose uptake and metabolism in intact cells. Biochemistry 28, 5694–5701. Seino, S., and Bell, G.I. (1989). Alternative splicing of human insulin receptor messenger RNA. Biochem. Biophys. Res. Comm. 159, 312–316. Seino, S., Seino, M., Nishi, S., and Bell, G.I. (1989). Structure of the human insulin receptor gene and characterization of its promoter. Proc. Natl. Acad. Sci. USA 86, 114–118.

Leibiger, I.B., Walther, R., Pett, U., and Leibiger, B. (1994a). Positive and negative regulatory elements are involved in transcriptional control of the rat glucokinase gene in the insulin producing cell line HIT M2.2.2. FEBS Lett. 337, 161–166.

Shelton, K.D., Franklin, A.J., Khoor, A., Beechem, J., and Magnuson, M.A. (1992). Multiple elements in the upstream glucokinase promoter contribute to transcription in insulinoma cells. Mol. Cell. Biol. 12, 4578–4589.

Leibiger, B., Walther, R., and Leibiger, I.B. (1994b). The role of the proximal CTAAT-box of the rat glucokinase upstream promoter in transcriptional control in insulin-producing cells. Biol. Chem. Hoppe-Seyler 375, 93–98.

Taylor, S.I. (1999). Deconstructing type 2 diabetes. Cell 97, 9–12.

Leibiger, I.B., Leibiger, B., Moede, T., and Berggren, P.O. (1998a). Exocytosis of insulin promotes insulin gene transcription via the insulin receptor/PI-3 kinase/p70 s6 kinase and CaM kinase pathways. Mol. Cell 1, 933–938. Leibiger, B., Moede, T., Schwarz, T., Brown, G.R., Ko¨hler, M., Leibiger, I.B., and Berggren, P.O. (1998b). Short-term regulation of

Ullrich, A., Bell, J.R., Chen, E.Y., Herrera, R., Petruzzelli, L.M., Dull, T.J., Gray, A., Coussens, L., Liao, Y.C., Tsubokawa, M., et al. (1985). Human insulin receptor and its relationship to the tyrosine kinase family of oncogenes. Nature 313, 756–761. Vanhaesebroeck, B., and Alessi, D.R. (2000). The PI3K–PDK1 connection: more than just a road to PKB. Biochem. J. 346, 561–576. Virkama¨ki, A., Ueki, K., and Kahn, C.R. (1999). Protein–protein interaction in insulin signaling and the molecular mechanism of insulin resistance. J. Clin. Invest. 103, 931–943.

Molecular Cell 570

Vogt, B., Carrascosa, J.M., Ermel, B., Ullrich, A., and Ha¨ring, H.U. (1991). The two isotypes of the human insulin receptor (HIR-A and HIR-B) follow different internalization kinetics. Biochem. Biophys. Res. Commun. 177, 1013–1018. Watada, H., Kajimoto, Y., Umayahara, Y., Matsuoka, T., Kaneto, H., Fujitani, Y., Kamada, T., Kawamori, R., and Yamasaki, Y. (1996). The human glucokinase gene ␤-cell-type promoter: an essential role of insulin promoter factor 1/PDX-1 in its activation in HIT-T15 cells. Diabetes 45, 1478–1488. White, M.F., and Kahn, C.R. (1994). The insulin signaling system. J. Biol. Chem. 269, 1–4. Withers, D.J., Sanchez-Gutierrez, J.C., Towery, H., Ren, J.M., Burks, D.J., Previs, S., Zhang, Y., Bernal, D., Shulman, G.I., Bonner-Weir, S., and White, M.F. (1998). Disruption of IRS-2 causes type 2 diabetes in mice. Nature 391, 900–903. Xu, G.G., and Rothenberg, P.L. (1998). Insulin receptor signaling in the ␤-cell influences insulin gene expression and insulin content. Evidence for autocrine ␤-cell regulation. Diabetes 47, 1243–1252. Xu, G., Kwon, G., Marshall, C.A., Lin, T.A., Lawrence, J.C., Jr., and McDaniel, M.L. (1998). Branched-chain amino acids are essential in the regulation of PHAS-I and p70 s6 kinase by pancreatic ␤-cells. A possible role in protein translation and mitogen signaling. J. Biol. Chem. 273, 28178–28184. Yamaguchi, Y., Flier, J.S., Yokota, A., Benecke, H., Backer, J.M., and Moller, D.E. (1991). Functional properties of two naturally occurring isoforms of the human insulin receptor in Chinese hamster ovary cells. Endocrinology 129, 2058–2066.