PANCREATIC islet cells undergo a functional deterioration

Transcription

1 /99/$03.00/0 Vol. 140, No. 7 Endocrinology Printed in U.S.A. Copyright 1999 by The Endocrine Society Increase in PDX-1 Levels Suppresses Insulin Gene Expression in RIN Cells* RHONA SEIJFFERS, ORIT BEN-DAVID, YAEL COHEN, AVRAHAM KARASIK, MEIR BEREZIN, CHRISTOPHER B. NEWGARD, AND SARAH FERBER Endocrine Institute, Sheba Medical Center (R.S., O.B-D., Y.C., A.K., M.B., S.F.), Tel-Hashomer 52621; Tel-Aviv University Sackler School of Medicine (R.S., O.B.-D., A.K., M.B.), Tel-Aviv; and Bar-Ilan University (Y.C.), Ramat-Gan 52900, Israel; and the Departments of Biochemistry and Internal Medicine, University of Texas Southwestern Medical Center (C.B.N.), Dallas, Texas ABSTRACT RIN cells (RIN-38) exhibit a passage-dependent reduction in both basal and glucose-regulated insulin secretion, accompanied by decreased insulin content. In an attempt to explain the mechanism of the gradual decrease in insulin production in cultured cells, we analyzed the insulin promoter activity and the levels of an important trans-activator of the insulin gene, PDX-1, as a function of aging in culture. We demonstrate that the decrease in insulin content and secretion is reflected in decreased promoter activity and is associated with a decrease in E47 and BETA2 nuclear factors, but with a paradoxical 3-fold increase in PDX-1 protein levels. To dissect the effect of increased PDX-1 from the decrease in the additional transcription factors on insulin promoter activity, we overexpressed PDX-1 protein in low passage RIN-38 cells by recombinant adenovirus technology. PANCREATIC islet cells undergo a functional deterioration process that can be induced by aging or chronic exposure to nutrients and hormones (1 4). The rat insulinoma cell line RIN (RIN-38) serves as a cellular model for such a functional deterioration process (5). At low passage number, these cells are glucose responsive, but lose this ability progressively with time in culture along with a specific decrease in insulin, GLUT-2, and glucokinase (GK) messenger RNA (mrna) levels (5). An important issue is whether this functional deterioration process is reflected in and can be explained by alterations in -cell-specific transcription factors. The transcription factor PDX-1 has an important role in pancreatic islet differentiation (6 11). Cooperativity between PDX-1 and bhlh proteins in trans-activation of the insulin gene was demonstrated (12). The ubiquitously expressed E47 protein creates heterodimers with the -cell-specific HLH protein BETA2 (13). In terminally differentiated -cells, PDX-1 participates in mediating glucose stimulation of insulin promoter activity and trans-activation of other important -cell-specific genes, such as GK, GLUT-2, and IAPP (14 17). These findings suggest that alterations in PDX-1 levels could influence the expression of a host of islet cell genes that are important for normal function. Long term Received September 10, Address all correspondence and requests for reprints to: Sarah Ferber, Ph.D., Endocrine Institute, Sheba Medical Center, Tel-Hashomer, Israel. berezin@post.tau.ac.il. * This work was supported by a grant from the United States/Israel Binational Foundation (to S.F. and C.B.N.). PDX-1 overexpression did not reduce E47 and BETA2 levels, but was sufficient to suppress rat insulin promoter activity in a dose-dependent manner. The fact that PDX-1 levels participate in trans-activation of insulin promoter activity was demonstrated in HIT-T15 cells. Treating HIT-T15 cells with 1 2 multiplicity of infection of AdCMV- PDX-1 increased rat insulin promoter activity, whereas higher doses repressed insulin promoter activity in these cells as in RIN-38 cells. Our data demonstrate that PDX-1 regulates transcription of the insulin gene in a dose-dependent manner. Depending on its nuclear dosage and the levels of additional cooperating transcription factors, PDX-1 may act as an activator or a repressor of insulin gene expression, such that low as well as high doses may be deleterious to insulin production. (Endocrinology 140: , 1999) exposure to glucose reduces PDX-1 dosage or binding activity in pancreatic islets, cell lines, and islets of an in vivo model of noninsulin-dependent diabetes mellitus (NIDDM), the 90% pancreatectomized rat. These changes in PDX-1 activity have been correlated with reduced insulin and GLUT-2 gene expression (3, 18 25). In addition, dexamethasone induces a rapid decrease in PDX-1 expression in HIT-T15 cells, which correlates with a decrease in insulin mrna levels, whereas fatty acids induce a decrease in PDX-1 binding activity and a commensurate decrease in insulin, GLUT-2, GK, and somatostatin mrna levels in pancreatic islets (3, 4, 25a). On the other hand, Kajimoto et al. (26) documented that suppression of the transcription factor PDX-1 by administration of antisense oligonucleotides to MIN6 insulinoma cells caused no decrease in insulin mrna levels. Moreover, in islets from an alternate rodent model of NIDDM, the db/db mouse, loss of GLUT-2 expression was associated with a 3-fold increase in PDX-1 binding to this promoter (27). The seemingly controversial data regarding the effect of PDX-1 dosage and activity on islet cell function motivated us to analyze the role of PDX-1 in regulating insulin gene expression in the rat insulinoma cell line RIN-38. An important issue is whether this functional deterioration process is reflected in and can be explained by alterations in -cell-specific transcription factors. We demonstrate that the decrease in insulin production in RIN-38 cells that occurs with aging in culture is associated with a similar decrease in basal rat insulin promoter (RIP) activity, and a decline in the protein levels of the insulin gene trans-activators, E47 and BETA2 (12, 13), but is also associated with a paradoxical 3-fold increase 3311

2 3312 INCREASE IN PDX-1 LEVELS SUPPRESSES INSULIN GENE EXPRESSION Endo 1999 Vol 140 No 7 in PDX-1 levels. Systematic titration of PDX-1 levels in RIN-38 and HIT-T15 cells demonstrates that overexpression of PDX-1 alone, without reducing BETA2 and E47 levels, suppresses insulin promoter activity in a dose-dependent manner. Cell culture Materials and Methods RINr (RIN-38) cells were cultured in medium 199-Earle s salts (5.5 mm glucose) supplemented with 5% FCS, 100 mu penicillin, and 100 g streptomycin/ml (Life Technologies, Grand Island, NY). Cells were split once a week, using 0.05% trypsin-edta solution (Life Technologies) and kept under an atmosphere of 95% air and 5% CO 2 at 37 C. 293 and HIT-T15 cells were cultured in DMEM 25 and 5.5 mm glucose, respectively, supplemented with 10% FCS (Life Technologies) and antibiotics. Cell handling was as described for RIN-38 cells. Preparation of recombinant adenoviruses All recombinant adenoviruses were constructed according to the procedure reported by Becker et al. (28). The gene of interest was ligated into the paccmv.plpa plasmid followed by cotransfection with the adenovirus plasmid pjm17 and harvesting the recombinant virions as previously described (28). The recombinant adenovirus termed Ad- CMV-CAT was prepared by ligation of a BamHI/EcoRI restriction fragment containing the intact bacterial chloramphenicol acetyltransferase (CAT) gene distal to the cytomegalovirus (CMV) promoter in the paccmv.plpa plasmid. The AdRIP-CAT recombinant adenovirus was prepared by insertion of 410 nucleotides of the 5 -flanking region of the rat insulin-1 gene (supplied by Dr. Larry Moss) in place of the viral CMV promoter in the paccmv.plpa plasmid and ligation of the BamHI/EcoRI CAT gene insert distal to the insulin promoter fragment. The AdCMV-PDX-1 recombinant adenovirus contains a HindIII/BamHI fragment encompassing a complete coding region of the mouse homolog of PDX-1 [insulin promoter factor-1 (IPF-1)] (6) [PDX-1 complementary DNA (cdna), a gift from Dr. Christopher V. E. Wright]. AdRIP-PDX-1 recombinant adenovirus contains a HindIII/BamHI fragment encompassing a complete coding region of the mouse homolog of PDX-1 (IPF-1), which replaced the CAT fragment in pacrip.plpa plasmid used for the AdRIP-CAT recombinant adenovirus. Preparation of viral stocks 293 cells were cultured in 145/20 plates (Greiner Friekenhausen, Germany) and treated with recombinant adenoviruses at multiplicity of infection (moi) of 10 for 90 min in minimal volume (8 ml). Forty-eight hours later, cells and media were collected, cells were pelleted by centrifugation in 800 g, and the virus in the supernatant fraction was precipitated overnight in 20% polyethylene glycol (PEG 8000, Sigma Chemical Co., St. Louis, MO) and 2.5 m NaCl at 4 C. The virus-containing medium was centrifuged at 10,000 g, and the pellet was resuspended in physiological saline [137 mm NaCl, 5 mm KCl, 10 mm Tris-HCl (ph 7.4), and 1 mm MgCl 2 ]. Viral stocks were stored at concentrations of plaque-forming units/ml at 4 C. Determination of insulin promoter (RIP) activity Cells were plated in 12-well dishes at a density of 10 6 cells/well. Twenty-four to 48 h after plating, cells were incubated with 4 moi AdRIP-CAT or 1 moi AdCMV-CAT for 90 min. Thereafter, the virus was removed, cells were washed with PBS, and fresh medium was applied on the plates. Forty-eight hours after infection, cells were extracted in lysis buffer [luciferase assay system with reporter lysis buffer from Promega Corp. (Madison, WI) suitable for the use in CAT and protein assays] and analyzed for CAT activity and protein concentrations. CAT and luciferase activity assays CAT activity was measured with a Promega Corp. kit in the presence of butyryl coenzyme A (Sigma Chemical Co.) and [ 14 C]chloramphenicol (ICN, Irvine, CA), according to the instructions provided by the manufacturer. Luciferase activity was measured with a Promega Corp. kit. Western blot analysis of nuclear proteins Preparation of nuclear extracts from mammalian cells. Nuclear extracts were prepared according to the method of Leonard et al. (29). Approximately 10 7 cells were washed in ice-cold PBS and collected in 400 l ice-cold hypotonic buffer [10 mm HEPES (ph 7.9), 10 mm KCl, 1 mm EDTA, 1 mm EGTA, 1 mm dithiothreitol, 0.1 mm p-amino benzoic acid, 1 mm phenylmethylsulfonylfluoride, 10 g/ml leupeptin, 1 g/ml pepstatin, and 50 g/ml aprotinin] and left to swell on ice for 15 min. Twenty-five microliters of 10% Nonidet P-40 (Sigma Chemical Co.) were added to the suspension, the cells were vortexed vigorously for 10 sec and centrifuged at 12,000 g for 30 sec at 4 C, and the supernatant (cytoplasmic fraction) was separated from the pellet (nuclear fraction). Nuclear proteins were extracted in 200 l high salt buffer [20 mm HEPES (ph 7.9), 400 mm NaCl, 1mmEDTA, 1 mm EGTA, 1 mm dithiothreitol, 0.1 mm p-amino benzoic acid, 1 mm phenylmethylsulfonylfluoride, 10 g/ml leupeptin, 1 g/ml pepstatin, and 1 mg/ml aprotinin] for 15 min with gentle rotation. The nuclear extracts were then centrifuged for 5 min at 4 C, and the pellet was discarded. The extracts were kept in 50- l aliquots at 80 C. Protein concentration was determined using the Bradford method (30). Western blot analysis. Forty micrograms of nuclear extract proteins were resolved on 12% SDS-PAGE and electroblotted onto 0.2- m nitrocellulose membranes (PROTAN BA 83, Schleicher & Schuell, Inc., Dassel, Germany) (31). The membranes were blocked for 1hina10% solution of 1% fat liquid milk (Tnuva, Tel-Aviv, Israel), and 0.05% Tween-20 in PBS and incubated with a 1:10,000 dilution of a rabbit polyclonal antibody against the N-terminal region of mouse PDX-1 (11) (contributed by Christopher V. E. Wright) in blocking buffer for 1 h, followed by horseradish peroxidase-conjugated goat antirabbit secondary antibody (Amersham, Arlington Heights, IL). Membranes were stripped and rehybridized with anti-beta2 (1:500; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), E47 (1:500; Santa Cruz Biotechnology, Inc.) and anti- -thymopoietin (1:10,000; contributed by A. Simon and G. Goldstein) antibodies. Immunoreactive proteins were visualized using the enhanced chemiluminescence detection system (Amersham). Statistical analysis Data are presented as the mean the se. Differences among means were analyzed using one-way ANOVA, followed by Bonferroni s method for pairwise multiple comparison. P 0.05 was considered significant. Results Passage-dependent decrease in insulin promoter activity in RIN-38 cells Insulin content and secretion decrease by 50% in RIN-38 cells at passage 45 compared with the same cells at passage 20 together with a substantial decrease in insulin mrna levels (Refs. 5 and 32 and our unpublished results). This motivated us to analyze insulin promoter activity in these cells as a function of passage number. RIN-38 cells of low (passage 20), intermediate (passage 30), or high (passage 45) passage number were cultured in 5.5 mm glucose, treated with AdRIP-CAT or AdCMV-CAT, and harvested 48 h later. Cells of passages 30 and 45 demonstrated 39% and 72% decreases in basal RIP activity, respectively, compared with cells of passage 20 cultured under the same conditions (Fig. 1). Cells treated with the AdCMV-CAT control virus exhibited no significant reduction in CAT activity compared with cells of passage 20. Therefore, CAT activity under control of the RIP promoter was normalized to that under control of the CMV promoter for each set of conditions. As the profound decrease in promoter activity was specific to RIP, we pos-

3 INCREASE IN PDX-1 LEVELS SUPPRESSES INSULIN GENE EXPRESSION 3313 FIG. 1. Passage-dependent rat insulin promoter-1 activity in RIN-38 cells. Cells at the indicated passages were plated in 12-well dishes (Nunc) at a density of 10 6 cells/well. After 24 h, cells were treated with AdRIP-CAT (4 moi) or AdCMV-CAT (1 moi). Forty-eight hours after adenovirus treatment, cells were extracted and analyzed for CAT activity and protein. Data represent the mean SE for eight independent samples for each treatment (n 8). The asterisk indicates that RIP activities in cells of passages 30 and 45 (p. 30 and p. 45) were significantly different from each other and lower than that in cells of passage 20 (p. 20), at a level of significance of P As CMV-CAT activity was unaffected by passage number, RIP activity was normalized to CMV activity in each passage. tulated that the decrease in RIP activity with time in culture was a function of alterations in the activity or the levels of transcription factors interacting with this promoter. Passage-dependent alterations in nuclear factors: PDX-1 levels are markedly elevated, whereas E47 and BETA2 levels decrease in high passage RIN-38 cells The alterations that occur in the expression levels of several genes in RIN-38 cells with time in culture could be due to alterations in -cell-specific transcription factors. An attractive candidate factor is the PDX-1 homeobox gene product, which is known to trans-activate the insulin, GLUT-2, and GK promoters (14, 16, 17). Therefore, we measured PDX-1 protein levels in nuclear extracts of RIN-38 cells by Western blot analysis as a function of passage number (Fig. 2A). Densitometric analysis of several experiments and exposures revealed that cells at intermediate and high passages (passages 30 and 40) contained % and % more PDX-1 protein, respectively, in their nuclear fraction than RIN-38 cells at passage 20 (Fig. 2B). In contrast, levels of immunodetectable E47 and BETA2, two additional transcription factors that have been implicated in control of insulin promoter activity, decreased in aging RIN-38 cells to 25 35% of their levels at passage 20 (Fig. 2, A and C). Overexpression of PDX-1 in RIN-38 and HIT-T15 cells suppresses RIP activity As a passage-dependent decrease in basal insulin promoter activity was associated with alterations in the levels of several transcription factors, we sought to analyze whether the increase in PDX-1 levels would be sufficient to affect RIP activity. We used recombinant adenovirus to overexpress PDX-1 in RIN-38 cells of low passage and in high passage HIT-T15 cells, both cultured in 5.5 mm glucose. As shown in Figs. 3 and 4, nuclear extracts from control RIN-38 (passage 20) and HIT-T15 cells contained immunodetectable PDX-1. Treatment of cells with increasing amounts of AdCMV-PDX-1 revealed that the transcription factor is correctly expressed in the nucleus in increasing quantities that correlate with the moi in both cell lines tested. Binding to the cognate locus on RIP was elevated in proportion to the extent of overexpression (data not presented). Moreover, AdCMV-PDX-1 treatment did not reduce BETA2 and E47 nuclear proteins levels in RIN-38 (Fig. 3) or HIT-T15 (Fig. 4) cells, which allows us to dissect the separate role of PDX-1 dosage on insulin promoter activity (the slight increase in BETA2 or E47 protein levels in RIN and HIT cells, respectively, was not consistent). To evaluate the specific effect of increased PDX-1 dosage on RIP activity, RIN-38 (Fig. 5A) and HIT-T15 (Fig. 5B) cells were treated with constant concentrations of AdRIP-CAT (4 moi), or as a control with AdCMV-CAT (1 moi), plus increasing amounts of AdCMV-PDX-1. CAT activity was determined 48 h after infection. PDX-1 expression dramatically inhibited RIP activity in low passage RIN-38 cells. One and 10 moi of Ad- CMV-PDX-1 inhibited RIP activity to 66 3% and 14 7% of its activity in untreated cells, respectively (Fig. 5A). The fact that PDX-1 participates in trans-activation of insulin promoter activity is demonstrated in high passage HIT-T15 cells cultured in 5.5 mm glucose (Fig. 5B). One and 2 moi of AdCMV-PDX-1 increased RIP activity to % and %, respectively, but higher doses suppressed RIP activity in these cells in a manner similar to that observed in RIN-38 cells. Treatment of HIT-T15 cells with 20 moi of AdCMV-PDX-1 inhibited insulin promoter activity to 20% of its maximal activity (at 1 moi treatment). CMV promoter activity was not significantly altered by PDX-1 overexpression (Fig. 5, A and B). Similar results were obtained when RIN-38 cells were treated with an adenovirus containing the cdna encoding the rat homolog of PDX-1 (somatostatin transcription factor-1; data not presented). To demonstrate that high titers of adenovirus per se were not responsible for altered RIP activity, we treated the same cells with increasing multiplicities of infection of AdCMV-Luc instead of AdCMV-PDX-1. As shown in Fig. 5C, this manipulation proportionally increased luciferase activity (Fig. 5C, lower panel), but did not affect RIP activity (Fig. 5C, upper panel). Use of AdRIP-PDX-1 adenovirus to moderately overexpress the PDX-1 protein, in levels that mimic the passagedependent alteration As even 1 moi of AdCMV-PDX-1 resulted in a substantial increase in PDX-1 nuclear protein levels compared with those in untreated cells due to the potent CMV promoter (Fig.

4 3314 INCREASE IN PDX-1 LEVELS SUPPRESSES INSULIN GENE EXPRESSION Endo 1999 Vol 140 No 7 FIG. 3. Immunoblot analysis of PDX-1, E47, and BETA2 protein in RIN-38 cells overexpressing PDX-1. Forty-eight hours after Ad-CMV- PDX-1 treatment, 40 g nuclear extracts were resolved on SDS-PAGE and subjected to immunoblot analysis as indicated and as described in Fig. 2. Lanes 1 3, RIN-38 passage 20, untreated or treated with 1 and 10 moi of Ad-CMV-PDX-1, respectively. FIG. 4. Immunoblot analysis of PDX-1, E47, and BETA2 protein in HIT-T15 cells overexpressing PDX-1. Forty-eight hours after Ad- CMV-PDX-1 treatment, 40 g nuclear extracts were resolved on SDS- PAGE and subjected to immunoblot analysis as indicated and as described in Fig. 2. Lane 1, Untreated intermediate/high passage HIT-T15 cells (the endogenous PDX-1 is above the visualized band and is very weak, the visualized band is nonspecific, as it is present in HeLa cells as well). Lanes 2 and 3, Proteins in nuclear extracts from HIT cells treated with 1 and 10 moi of Ad-CMV-PDX-1, respectively. FIG. 2. PDX-1 protein levels in RIN-38 cells of increasing passages. A, Representative Western blot analysis. Forty micrograms of nuclear extract protein from cells at indicated passages were resolved on 12% SDS-PAGE and electroblotted onto 0.2- m nitrocellulose membranes (PROTAN BA 83, Schleicher & Schuell, Inc.). (a), Western blot analysis was performed using rabbit antimouse PDX-1 (43-kDa band) antibodies at 1:10,000 dilution as described in Materials and Methods. The same membranes were washed and rehybridized with (b) rabbit antihuman E47 ( 73-kDa band; (c), goat antimouse BETA2 (50- to 60-kDa band) (both b and c at dilution of 1:500, Santa Cruz Biotechnologies, Inc., Santa Cruz, CA), and (d) mouse antihuman -thymo- 6A), we decided to construct an additional adenovirus (Ad- RIP-PDX-1) that could direct a moderate increase in PDX-1 nuclear protein levels. Three moi of this adenovirus resulted in a 3-fold increase in the homeobox protein levels (Fig. 6A). This manipulation did not inhibit insulin promoter activity in RIN-38 cells (Fig. 6C). Ten moi of AdRIP-PDX-1 resulted in an additional 2.4-fold increase in these protein levels above those achieved with the 3 moi treatment. The 2.4-fold increase in PDX-1 protein levels between 3 10 moi of AdRIP-PDX-1 treatment decreased RIP-CAT activity by 44% (Fig. 6C). PDX-1 overexpression in this experiment, as in the above experiments, did not reduce the E47 and BETA2 poetin, nuclear factor (53 kda band) at dilution of 1:10,000. Lanes 1 3 RIN-38 cells of passage 20, 30, and 40, respectively. B, Densitometric analysis of eight separate PDX-1 immunoblot experiments was performed using the Bio-Rad Multi-Analyst/PC version 1.1. Endogenous PDX-1 protein levels were normalized to -thymopoetin and were all statistically different from each other. Data are mean SE, n 8, *P C, Densitometric analysis of four separate immunoblots for E47 (black bars) and Beta2 (open bars), normalized to -thymopoetin. The levels of both proteins at passage 30 and 40 are statistically different than that of passage 20. Data are mean SE, n 4, *P 0.05.

5 INCREASE IN PDX-1 LEVELS SUPPRESSES INSULIN GENE EXPRESSION 3315 protein levels (data not presented). Importantly, 10 moi of AdRIP- -gal did not affect RIP-CAT activity, meaning that the effect is specific to PDX-1 overexpression and not to an increase in the insulin-1 promoter copy number (data not presented). This process of passage-dependent decrease in RIP activation is clearly affected by both the increase in PDX-1 and the decrease in E47 and BETA2 protein levels (and possibly additional factors). However, using a molecular approach we demonstrate that the increase in PDX-1 dosage by itself is sufficient to confer a decrease in RIP activity. Discussion In the current study we have employed RIN-38 cells as an in vitro model of -cell functional deterioration. At low passage number, these cells exhibit a relatively differentiated phenotype and are glucose responsive. However, they lose this ability gradually with time in culture coincident with a specific loss of GLUT-2 and GK gene expression and a substantial decrease in insulin mrna levels (5, 32). In the present study we demonstrate that these cells also exhibit a gradual decrease in basal insulin promoter activity. Surprisingly, we found that PDX-1 protein levels substantially increase as insulin gene expression falls with increasing passage numbers, whereas the levels of BETA2 and E47, both insulin gene trans-activators, decrease. That increased PDX-1 expression is related to the fall in insulin expression is clearly established by our finding that overexpression of the transcription factor suppresses insulin promoter activity in a dose-dependent fashion in low passage RIN-38 cells without substantially altering the endogenous BETA2 and E47 nuclear proteins levels. Moreover, using AdRIP-PDX-1, we are able to closely mimic the dosage alterations in PDX-1 levels that occur in RIN-38 cells with an increase in passage and to demonstrate that these alterations may mediate a large proportion of the decrease in insulin gene transcription. In contrast, it was reported that the increase in passage of HIT-T15 cells is associated with reduced PDX-1 binding and decreased RIP activity (33). Thus, only in these cells, but not in RIN-38 cells, did a moderate overexpression of PDX-1 (1 and 2 moi of AdCMV-PDX-1) activate the insulin gene, whereas higher levels of overexpression resulted in a repression similar to that observed in RIN-38 cells. This indicates that enhancement of insulin promoter activity by PDX-1 is specific for cells low in this protein (HIT-T15 cells) and is not a generalized phenomenon for all -cells. Most studies correlate reduced insulin gene expression with a decrease in PDX-1 levels (3, 20, 23, 24, 34). When viewed in this context, our results with RIN-38 cells may seem paradoxical. Nevertheless, our findings are in accord FIG. 5. RIP and CMV promoter activities in RIN-38 and HIT-T15 cells overexpressing PDX-1. RIN-38 (A) and HIT-T15 (B) cells were plated in 12-well dishes at a density of 10 6 cells/well and treated with either 4 moi of Ad-RIP-CAT (black bars) or 1 moi of Ad-CMV-CAT (hatched bars). In addition, cells were treated with increasing amounts of Ad-CMV-PDX-1 as indicated. Forty-eight hours after viral infection (AdCMV-PDX-1), cells were harvested, and CAT activity was measured as described in Fig. 1. C, RIN-38 cells at passage 25 were plated and treated by Ad-RIP-CAT as described for A and B, with increasing amounts of Ad-CMV-Luc instead of AdCMV-PDX-1. CAT and Luciferase activities were measured in the same samples and normalized to protein as described in Materials and Methods. #,No significant difference among the mean values of the Ad-CMV-CAT infected by Ad-CMV-PDX-1 or between them and that of PDX-1 uninfected controls (hatched bars); *, mean values are statistically different from each other and from PDX-1 untreated cells (black bars in both A and B; P 0.05). Mean values of the 2 moi Ad-CMV-PDX-1 treatment in B are significantly different from those of the 10 and 20 moi treatments (P 0.05), but not from control values and that of the 1 moi of Ad-CMV-PDX-1 treatment (black bars). Data are the mean SE (n 8 for each condition).

6 3316 INCREASE IN PDX-1 LEVELS SUPPRESSES INSULIN GENE EXPRESSION Endo 1999 Vol 140 No 7 FIG. 6. Moderate overexpression of PDX-1 using AdRIP-PDX-1 confers a decrease in RIP activity. A, Immunoblot and densitometric analysis of PDX-1 protein. RIN-38 cells of passage 25 were untreated (lane 1) or treated by 3 and 10 moi of AdRIP-PDX-1 (lanes 2 and 3, respectively) or 1 moi of AdCMV-PDX-1 (lane 4), as described in Figs B, RIP and CMV promoter activities upon AdRIP-PDX-1 treatment. RIN-38 cells of passage 25 were untreated or treated by 3 and 10 moi of AdRIP-PDX-1 or 1 moi of Ad-CMV-PDX-1 in addition to AdRIP -CAT (black bars) and AdCMV-CAT (hatched bars), as described in Fig. 5. Data are the mean SE (n 8 for each condition). *, P with other observations that relate PDX-1 dosage to the level of insulin expression. 1) PDX-1 levels are generally higher in transformed -cell lines than in pancreatic islets, yet all such lines contain substantially less insulin than found in primary cells (2, 14). 2) Using transient transfection methods, it was demonstrated that overexpression of PDX-1 in pancreatic islets affects insulin promoter activity with a bell-shaped curve, such that low levels activated and high levels suppressed RIP activity (14). 3) In the rodent model of NIDDM, the db/db mouse, loss of GLUT-2 expression was associated with a 3-fold increase in PDX-1 binding to the GLUT-2 promoter (27). 4) In addition, in MIN6 cells, decreasing PDX-1 dosage had no effect on insulin or GK mrna levels (26). The correlation between decreased RIP activation and low E47 protein levels visualized in high passage RIN-38 cells is in agreement with previous data (35). Moreover, E47 and BETA2 overexpression in HIT-T15, TC-6, and RIN 5AH cells resulted in trans-activation of the insulin promoter (12, 13, 33, 35, 36). The value of the current study is that it presents a cellular model in which a decrease in insulin gene expression is associated with a 3-fold increase in PDX-1 dosage, whereas E47 and BETA2 levels decrease. Moreover, it clearly demonstrates that increased PDX-1 levels repress insulin gene expression and provides a systematic analysis of the relationship between PDX-1, E47, and BETA2 expression levels and insulin promoter activity in two -cell lines. Our experimental system is able to dissect between the effect of increased PDX-1 dosage from that of decreased BETA2 and E47 levels on insulin gene transcription. PDX-1 overexpression did not alter the levels of these two insulin gene trans-activators. In addition, it should be taken into consideration that the decrease in BETA2 and E47 levels (and possibly additional transcription factors) may further potentiate the decrease in RIP activation demonstrated in high passage RIN-38 cells. Our data suggest that PDX-1 may play a dual role in insulin gene expression, and that low as well as high doses may be deleterious. The idea that homeobox genes have multiple roles depending on their relative dosage is not unique to PDX-1. The paired box Pax proteins are crucial regulators of organogenesis in thymus, kidney, thyroid, pancreas, and eye. Thus, overexpression of these genes in tissues in which they are normally expressed may lead to tumorigenesis and abnormal development, suggesting that doses of Pax proteins are critical for their normal function (37), and that squelching of transcription factors can naturally occur when inappropriately high levels of such factors are reached. The dosage of the homeobox genes and the cellular context of the available nuclear factors are two of the most important parameters responsible for specificity of HOX action (38). The increased PDX-1 levels could directly induce nonproductive cooperation between insulin gene trans-activators, which may compete for binding to relevant regulatory loci on the insulin promoter. In addition, its increased levels could induce an increase in levels of transcription factors, such as C/EBP or Ids, which may indirectly mediate dysregulation of RIP activation (39 41). C/EBP was identified as a repressor of insulin gene transcription in conditions of supraphysiological glucose levels in HIT-T15 and INS-1 cells. Inhibition of RIP activation occurred by direct protein-protein interaction with the basic helix-loop-helix transcription factor E47. This interaction was suggested to lead to the inhibition of E47 binding to the E elements of the insulin promoter, thereby reducing the trans-activation potential of E47 on insulin gene transcription (39). We conclude that PDX-1 is sufficient to affect the insulin gene rate of transcription in a dose-dependent manner. However, our results also suggest that the relative quantities of cooperating insulin gene trans-activators may play an important role in regulating the expression of this gene. Moreover, the perturbation of the balance between these interacting factors could contribute to -cell dysfunction in islet cell lines or in animal models of diabetes. The exact mechanism by which PDX-1 converts from a trans-activator of the

7 INCREASE IN PDX-1 LEVELS SUPPRESSES INSULIN GENE EXPRESSION 3317 insulin promoter to its repressor in a dose-dependent fashion requires further analysis. Acknowledgments We are indebted to C. V. E. Wright, Vanderbilt University, for generously providing antimouse and frog PDX-1 antibodies and IPF-1 cdna; to Gideon Goldstein and Amos Simon, Sheba Medical Center, for anti- -thymopoietin antibodies; to L. G. Moss, Tufts University, for the RIP DNA plasmid; and to H. Constandy, BetaGene, Inc. (Dallas, TX), for preparing the AdRIP-CAT recombinant adenovirus. HIT-T15 cells were generously provided by M. D. Walker, Wiezmann Institute. AdCMV- Luc was provided by Robert Gerard (UTSW Medical Center, Dallas, TX). References 1. Kahn CR 1996 New concepts in the pathogenesis of diabetes mellitus. Adv Intern Med 41: Ling Z, Pipeleers G 1996 Prolonged exposure of human cells to elevated glucose levels results in sustained cellular activation leading to a loss of glucose regulation. J Clin Invest 98: Gremlich S, Bonny C, Weaber G, Thorens B 1997 Fatty acid decrease IDX-1 expression in rat pancreatic islets and reduce GLUT-2, GK, insulin and somatostatin levels. J Biol Chem 272: Ogawa A, Johnson J, Ohneda M, McAllister T, Inman N, Alam T, Unger RH 1992 Roles of insulin resistance and -cell dysfunction in dexamethasone induced diabetes. J Clin Invest 90: Ferber S, BeltrandelRio H, Johnson JH, Noel RJ, Cassidy LE, Clark S, Becker TC, Hughes SD, Newgard CB 1994 GLUT-2 gene transfer into insulinoma cells confers both low and high affinity glucose-stimulated insulin release. J Biol Chem 269: Ohlsson H, Karlsson K, Edlund T 1993 IPF1, a homeodomain-containing transactivator of the insulin gene. EMBO J 12: Melloul D, Ben Neriah Y, Creasi E 1993 Purification of the -cell glucosesensitive factor that transactivates the insulin gene differentially in normal and transformed islet cells. Proc Natl Acad Sci USA 90: Offield MF, Jetton TL, Labosky PA, Ray M, Stein RW, Magnuson MA, Hogan BLM, Wright CVE 1996 PDX-1 is required for pancreatic outgrowth and differentiation of the rostal duodenum. Development 122: Jonsson J, Carlsson H, EdlundT, Edlund H 1994 Insulin-promoter-factor-1 is required for pancreas development in mice. Nature 371: Fernandes A, King LC, Guz R, Stein R, Wright CVE, Teitelman G 1997 Differentiation of new insulin-producing cells is induced by injury in adult pancreatic islets. Endocrinology 138: Guz Y, Montminy MR, Stein R, Leonard J, Gamer LW, Wright CVE, Teitelman G 1995 Expression of murine STF-1, a putative insulin gene transcription factor, in cells of pancreas, duodenal epithelium and pancreatic exocrine and endocrine progenitors during ontogeny. Development 121: Peers B, Leonard J, Sharma S, Teitelman G, Montminy MR 1994 Insulin expression in pancreatic islet cells relies on cooperative interaction between the helix loop helix factor E47 and the homeobox factor STF-1. Mol Endocrinol 8: Naya FJ, Stellrecht CMM, Tsai MJ 1995 Tissue specific regulation of the insulin gene by a novel bhlh transcription factor. Genes Dev 9: Marshak S, Totary H, Cerasi E, Melloul D 1996 Purification of the -cell glucose sensitive factor that transactivates the insulin gene differentially in normal and transformed islet cells. Proc Natl Acad Sci USA 93: Serup P, Jensen J, Andersen FG, Jorgensen MC, Blume N, Holst JJ, Madsen OD 1996 Induction of insulin and islet amyloid polypeptide production in pancreatic islet glucagonoma cells by insulin promoter factor 1. Proc Natl Acad Sci USA 93: Waeber G, Thompson N, Nicod P, Bonny C 1996 Transcriptional activation of the GLUT2 gene by the IPF-1/STF-1/IDX-1 homeobox factor. Mol Endocrinol 10: Watada H, Kajimoto Y, Umayahara Y, Matsuoka T, Kaneto H, Fujitani Y, Kamada T, Kawamori R, 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: Robertson RP, 1989 Type II diabetes, glucose non-sense, and islets desensitization. Diabetes 38: Poitout V, Olson LK, Robertson RP 1996 Chronic exposure of TC-6 cells to supraphysiologic concentrations of glucose decreases binding of the RIPE3b1 insulin gene transcription activator. J Clin Invest 97: Olson LK, Sharma A, Peshavaria M, Wright CVE, Towle HC, Robertson RP, Stein R 1995 Reduction of insulin gene transcription in HIT-T15 cells chronically exposed to a supraphysiologic glucose concentration is associated with loss of STF 1 transcription factor expression. Proc Natl Acad Sci USA 92: Robertson RP, Olson LK, Zhang HJ 1994 Differentiating glucose toxicity from glucose desensitization: a new message from the insulin gene. Diabetes 43: Zhang HJ, Petersen B, Robertson RP 1994 Variable regulation by insulin of insulin gene expression in HIT-T15 cells. Diabetologia 37: Olson LK, Redmon JB, Towle HC, Robertson RP 1993 Chronic exposure of HIT cells to glucose concentrations paradoxically decrease insulin gene transcription and alters binding of insulin gene regulatory protein. J Clin Invest 92: Zangen DH, Bonner-Weir S, Lee CH, Latimer JB, Miller CP, Habener JF, Weir GC 1997 Reduced insulin, GLUT-2 and IDX-1 in -cells after partial pancreatectomy. Diabetes 46: Olson KL, Qain J, Poitout V 1998 Glucose rapidly and reversibly decreases INS-1 cells insulin gene transcription via decrements in STF-1 and C1 activator transcription factor activity. Mol Endocrinol 12: a.Sharma S, Jhala US, Johnson T, Ferreri K, Leonard J, Montminy M 1997 Hormonal regulation of an islet specific enhancer in the pancreatic homeobox gene STF-1. Mol Cell Biol 17: KajimotoY, Watada H, Matsuoka T, Kaneto H, Fujitani Y, Miyazaki J, Yamasaki Y 1997 Suppression of transcription factor PDX-1 causes no decrease in insulin mrna in MIN6 cells. J Clin Invest 100: Bonny C, Roduit R, Gremlich S, Nicod P, Thorens B, Weaber G 1997 The loss of GLUT-2 expression in the pancreatic -cells of diabetic db/db mice is associated with an impaired binding activity of islet-specific transacting factors. Mol Cell Endocrinol 135: Becker TC, Noel RJ, Coats WS, Gomez-Foix AM, Alam T, Gerard RD, Newgard CB 1994 Use of recombinant adenovirus for metabolic engineering of mammalian cells. In: Roth M (ed) Methods in Cell Biology. Academic Press, New York, vol 43: Leonard J, Peers B, Johnson T, Ferreri K, Lee S, Montminy MR 1993 Characterization of somatostatin transactivating factor-1, a novel homeobox factor that stimulates somatostatin expression in pancreatic islet cells. Mol Endocrinol 7: Bradford MM 1976 A rapid and sensitive method of quantitation of microgram quantities of protein using the principle of protein-dye binding. Anal Biochem 72: Sambrook JE, Fritsch EF Maniatis 1989 Molecular Cloning: A Laboratory Manual, ed 2, vol 1 3. Cold Spring Harbor Laboratory, Cold Spring Harbor 32. Clark SA, Burnham BL, Chick WL 1990 Modulation of glucose-induced insulin secretion from a rat clonal -cell line. Endocrinology 127: Harmon JS, Tanaka Y, Olson KL, Robertson PR 1998 Reconstitution of glucotoxic HIT-T15 cells with somatostatin transcription factor-1 partially restores insulin promoter activity. Diabetes 47: Ahlgren U, Jonsson J, Jonsson L, Simu K, Edlund H Cell-specific inactivation of the mouse IPF1/PDX1 gene results in loss of -cell phenotype and maturity onset diabetes. Genes Dev 12: Vierra CA, Nelson C 1995 The Pan basic helix loop helix proteins are required for insulin gene expression. Mol Endocrinol 9: Sharma A, Henderson E, Gamer L, Zhuang Y, Stein R 1997 Analysis of the role of E2A-encoded proteins in insulin gene transcription. Mol Endocrinol 11: Mansouri A, Hallonet M, Gruss P 1996 Pax genes and their roles in cell differentiation and development. Curr Opin Cell Biol 8: Hayashi S, Scott MP 1990 What determine the specificity of action of Drosophila homeodomain proteins. Cell 63: Lu M,Seufert J Habener JF 1997 Pancreatic -cell-specific repression of insulin gene transcription by CCAAT/enhancer-binding protein. J Biol Chem 272: Prabhu S, Ignatova A, Park ST, Sun XH 1997 Regulation of the expression of cyclin dependent kinase inhibitor p21 by E2A and Id proteins. Mol Cell Biol 17: Cordle SR, Henderson E, Masuoka H, Weil PA, Stein R 1991 Pancreatic -cell-type-specific transcription of the insulin gene is mediated by basic helix loop helix DNA binding proteins. Mol Cell Biol 11: