Regulation of MafA Expression in Pancreatic β-Cells in db/db Mice With Diabetes

The −238 insulin firefly luciferase (−238 Insulin luc) expression plasmid contain rat insulin 2 gene sequences from −238 to +2 bp (32). The cytomegalovirus (CMV) enhancer-driven rat c-Jun, human JNK, p38α (CSBP2), and mouse MafA expression vectors were constructed by subcloning coding sequences into the pAdTrack-CMV or pcDNA3.1 expression vector. Myc-tagged MafA was produced by subcloning mouse MafA coding sequences into pcDNA6/myc-His (Invitrogen). −238 insulin LUC (0.25 μg) was transfected into cells with the control empty vector, pAdTrack, and/or the indicated expression vectors (1.0 μg in total) using LIPOFECTAmine (Invitrogen), together with the thymidine kinase promoter-driven renilla luciferase expression plasmid (20 ng phRL-TK; Promega Madison, WI) serving as an internal control.

Recombinant adenoviruses expressing c-Jun, JNK, p38, and MafA were prepared using the AdEasy system as described previously (33). An adenovirus expressing a small-interfering RNA against mouse c-Jun (Ad-si-cJun) was constructed using the following oligonucleotides as described previously (34): 5′-CGCGTAAGCTGATTACTGTCAATAAATTCAAGAGATTTATTGACAGTAATCAGCTTTTTTTTGGAAA-3′ and 5′-AGCTTTTCCAAAAAAAAGCTGATTACTGTCAATAAATCTCTTGAATTTATTGACAGTAATCAGCTTA-3′ (c-Jun sequences are underlined). These oligonucleotides were annealed and inserted into the MluI/HindIII sites of the pRNAT-H1.1/Adeno shuttle plasmid (GenScript, Piscataway, NJ). Adenovirus titer was roughly 10¹⁰ plaque-forming units (pfu)/ml after purification with the Adeno-X Virus Purification kit (Clontech), as estimated using the Adeno-X Titer kit (Clontech).

To isolate mouse islets, 1.5 mg/ml of collagenase plus protease inhibitor (Sigma-Aldrich) and hyaluronidase (Sigma-Aldrich) were injected into the pancreatic duct. The isolated pancreas was digested in a 37°C incubator for 20 min. After washing and precipitating with 0.25 mol/l sucrose, islets were hand picked.

Real-time PCR analysis was performed as described previously (34). Primer sets for mouse insulin 1 (numbering relative to ATG, forward −47 GACCAGCTATAATCAGAGACC, reverse +331 AGTTGCAGTAGTTCTCCAGCTG, 378 bp product), mouse insulin 2 (forward −57 AGCCCTAAGTGATCCGCTACAA, reverse +331 AGTTGCAGTAGTTCTCCAGCTG, 388 bp), mouse MafA (forward +757 TTCAGCAAGGAGGAGGTCAT, reverse +973 CCGCCAACTTCTCGTATTTC, 217 bp), mouse PDX-1 (forward +192 CATCTCCCCATACGAAGTGC, reverse +526 GGGGCCGGGAGATGTATTTG, 335 bp), mouse c-Jun (forward +736 TCCCCTATCGACATGGAGTC, reverse +881 TTTTGCGCTTTCAAGGTTTT, 146 bp), and mouse β-actin (forward +778 GCTCTTTTCCAGCCTTCCTT, reverse +945 CTTCTGCATCCTGTCAGCAA, 168 bp) were utilized to detect specific bands for each factor.

Total RNA or Poly(A)⁺mRNA was separated under denaturing conditions on 1% agarose gels. The transferred RNA was hybridized with a ³²P-labeled mouse MafA (314–490 bp; numbering relative to ATG), mouse insulin (35), mouse PDX-1 (622–852 bp), or mouse-β actin (53–549 bp) cDNA probe.

MIN6 nuclear extracts were prepared as described previously (34) at the indicated times or 60 h after the adenovirus infection. Islet nuclear extracts were prepared with a nuclear extraction kit (Active Motif, Carlsbad, CA). Western blot analysis was performed as described previously (34), with the following antibodies and dilutions: rabbit anti-MafA antibody (8), 1:1,000; rabbit anti–c-Jun antibody (no. 9165; Cell Signaling Technology, Danvers, MA), 1:200; rabbit anti–PDX-1 antibody (17), 1:2,000; and goat anti-actin antibody (no. sc-1615; Santa Cruz Biotechnology), 1:2,000.

Whole-cell extracts were obtained from >100 islets from each group of mice or adenovirus-infected MIN6 cells by treating for 24 h at 4°C in acid-ethanol. The insulin content of the extract was determined with the Mouse Insulin ELISA kit (Morinaga Biochemicals, Yokohama, Japan). Insulin concentration was normalized with total cellular protein, as measured using the Bio-Rad protein assay kit (Bio Rad Laboratories, Richmond, CA).

Pancreata were dissected and fixed overnight in 4% paraformaldehyde. Fixed tissues were processed routinely for paraffin embedding and sectioned. After treatment with 1% blocking goat or donkey serum, sections were immunostained for MafA, c-Jun, and insulin with the following antibodies and dilutions: rabbit (8) and our newly produced goat antibodies against MafA, 1:1,000 and 1:200, respectively; rabbit anti–c-Jun (no. 9165; Cell Signaling Technology), 1:200; guinea pig anti-insulin (no. A0564; DAKO, Glostrup, Denmark), 1:2000; guinea pig anti-glucagon (Linco, St. Charles, MO), 1:500; sheep anti-somatostatin (American Research Products, Belmont, MA), 1:500; and guinea pig anti–pancreatic polypeptide antibodies (Linco), 1:500. For the single detection of MafA or c-Jun, a Vectastain Elite ABC kit (Vector Laboratories) was utilized. For the combined staining of c-Jun, MafA, and insulin, we used Alexa Fluor 488 donkey anti-rabbit IgG, Alexa Fluor 555 donkey anti-goat IgG (Molecular Probes, Eugene, OR), and donkey Cy5-conjugated anti–guinea pig IgG (Jackson ImmunoResearch Laboratories, West Grove, PA), respectively, at a dilution of 1:200. For the costaining of glucagon, somatostatin, and pancreatic polypeptide, we used donkey Cy5-conjugated anti–guinea pig IgG and donkey Cy5-conjugated anti-sheep IgG as secondary antibodies at a dilution of 1:200. Fluorescent images were captured using a BIO-RAD Radience 2100 confocal microscope.

Sixty hours after the infection of adenoviruses or the transfection of expression plasmids, MIN6 cells were fixed on Lab-Tek chamber slides, and mouse islets were fixed on a poly-d-lysine cell culture plate with 4% paraformaldehyde. After the treatment with 0.1% Triton X-100 for the induction of cell permeability, the slides were incubated with 1% blocking goat serum. To detect MafA, MafA-myc fusion protein, c-Jun, and insulin, we performed immunocytochemistry with rabbit anti-MafA at 1:1,000 dilution (8), mouse anti–c-myc at 1:200 dilution (no. sc-40; Santa Cruz Biotechnology), rabbit anti–c-Jun at 1:200 dilution (no. 9165; Cell Signaling Technology), and guinea pig anti-insulin antibodies at 1:2,000 dilution (no. A0564; DAKO). Cells were then exposed to the secondary antibodies (Alexa Fluor 596 goat anti-rabbit IgG for MafA and Alexa Fluor 596 goat anti–guinea pig IgG for insulin at 1:200 dilution), and images were captured using a Nikon microscope digital camera (DXM1200F). Other secondary antibodies were used for detection with the Bio-Rad Radience 2100 confocal microscope (Alexa Fluor 546 goat anti-mouse IgG for myc, Alexa Fluor 488 donkey anti-rabbit IgG (Molecular Probes) for c-Jun, and donkey Cy5-conjugated anti–guinea pig IgG (Jackson ImmunoResearch Laboratories) for insulin at 1:200 dilution.

Data are expressed as means ± SD. Statistical analysis was performed using one-way ANOVA followed by the Scheffe test. A value of P < 0.05 was considered to be statistically significant.

C57 BLKsJ db/db (db/db) mice have been established as a type 2 diabetic model with obesity. In db/db mice, hyperinsulinemia is observed to compensate for the increased insulin resistance. However, after ∼12 weeks of age, plasma insulin levels begin to decrease, as shown in Fig. 1,A. This alteration of insulin levels is similar to that observed in type 2 diabetic patients. To evaluate whether the key insulin transcription factor MafA is also affected under diabetic conditions in vivo, we performed immunohistochemistry with an anti-MafA antibody (Fig. 1 A). In db/db mice, MafA expression in pancreatic islet nuclei was markedly decreased after 12 weeks of age and was not clearly detected at 17 and 22 weeks of age, whereas control m/m mice retained MafA expression at least up to 22 weeks old. These findings were consistent with the profile of plasma insulin and blood glucose levels.

Since c-Jun activity and protein levels are increased in response to oxidative stress (23,24), which leads to suppression of insulin promoter activity in vitro, we evaluated in this study whether c-Jun expression is upregulated in the islets of diabetic mice that are exposed to oxidative stress. As shown in Fig. 1,B, c-Jun expression was clearly detected in duct cells and a small number of islet cells in control m/m mice at 7–22 weeks of age and diabetic db/db mice at 7 weeks of age. However, the number of c-Jun–positive cells gradually increased with age in the islets of db/db mice. At 12 weeks of age, several c-Jun–positive islet cells were detected, and at 17 and 22 weeks of age a larger number of c-Jun–positive cells were detected. Interestingly, this expression pattern of c-Jun paralleled the loss of MafA expression (Fig. 1,A). Quantitative real-time PCR analysis, using freshly isolated islets from m/m and db/db mice at 17 weeks of age, clearly demonstrated that both MafA and insulin mRNA levels were markedly decreased (Fig. 1 C). In contrast, c-Jun mRNA level in islets was significantly increased, which was consistent with the expression pattern of MafA and c-Jun observed by immunohistochemistry.

Our results implied that the increased level of c-Jun caused a decrease in MafA and insulin gene expression in older diabetic mice. To evaluate these association in further detail, we first performed immunohistochemistry with MafA, c-Jun, and insulin antibodies using 7- and 17-week-old diabetic mice (Fig. 2,A), as MafA expression levels were markedly decreased between 7 and 17 weeks of age (Fig. 1,A). Normal mice expressed a large amount of insulin and MafA without enhanced c-Jun expression as expected. In contrast, in db/db mice, insulin expression was decreased in some cells in which MafA was undetectable or weakly expressed compared with control m/m mice. Interestingly, MafA and insulin expression was suppressed in most c-Jun–positive cells, which became clearer in 17-week-old db/db mice. KK-Ay mice, in which the Ay mutation is introduced onto a KK strain, are also known as an obese diabetic model. Since KK-Ay mice get severe diabetes after ∼20 weeks, we used older mice than db/db mice. In the KK-Ay islets, the number of c-Jun–positive cells was increased with marked hyperglycemia (∼200–300 mg/dl fasting glucose) at 21 weeks of age, and both MafA and insulin protein levels were decreased in those cells, as observed in db/db mice (Fig. 2,B). To examine the distribution of c-Jun in islet cells, immunohistochemistry was performed with antibodies against c-Jun and non–β-cell islet hormones, including glucagon, somatostatin, and pancreatic polypeptide (PP). In the islets of 17-week-old db/db mice, many c-Jun–positive cells were observed without costaining of non–β-cell hormones (Fig. 2,C), indicating that c-Jun was expressed in islet β-cells. To quantify the protein levels of MafA, c-Jun, and insulin in the islets of normal and diabetic mice, Western blotting using islet nuclear extracts and enzyme-linked immunosorbent assay (ELISA) for islet insulin were performed (Fig. 2 D and E). c-Jun protein level was clearly increased in the islets of 17-week-old db/db mice and of 21-week-old KK-Ay mice. In contrast, MafA protein level was decreased in those islets. Consistent with the suppressed expression of MafA, insulin content in the islets of 17-week-old db/db mice and 21-week-old KK-Ay mice were significantly decreased compared with control mice. These findings suggest that c-Jun is involved in the suppression of MafA and insulin expression under diabetic conditions.

To evaluate the effects of c-Jun on insulin gene transcription, we examined the effects of c-Jun on insulin promoter activity in the MIN6 β-cell line by reporter gene analysis. As shown in Fig. 3 A, c-Jun overexpression markedly decreased insulin promoter activity to 30% of that observed in control, which was consistent with previous reports (29,30). In addition, although it is known that several kinases such as JNK and p38 are activated by ROS in islet β-cells in vivo (15,16), their overexpression did not show significant effects.

To further examine the impact of c-Jun on MafA in islet β-cells, we generated an adenovirus expressing both c-Jun and green fluorescent protein (GFP) (Ad-c-Jun). We first performed Western blotting using nuclear extract of MIN6 and islet cells and found that c-Jun protein expression was almost undetectable in uninfected cells (data not shown). As shown in Fig. 3,B, adenoviral c-Jun overexpression markedly suppressed MafA protein level and its DNA-binding activity 60 h after the infection in MIN6 cells. The protein level and DNA-binding activity of PDX-1 were not apparently altered under these conditions. Northern blotting and real-time PCR analysis also showed that c-Jun overexpression significantly suppressed MafA mRNA level (supplemental Fig. 1, available in the online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db08-0693/DC1/) (Fig. 3,D). Adenoviral overexpression of c-Jun in isolated mouse islets also markedly suppressed MafA mRNA and protein levels (Fig. 3 C and E). Consistent with these results, the mRNA levels of insulin 1 and 2 were suppressed by c-Jun overexpression in both MIN6 cells and islets. A much smaller defect on PDX-1 mRNA was also observed under these circumstances.

To directly observe and visualize the effect of c-Jun, we performed immunocytochemistry using MIN6 and islet cells. Adding only a low level of Ad-c-Jun (1.0 μl of 1 × 10⁹ pfu/ml to 1 ml of culture medium) enabled us to compare MafA and insulin expression in c-Jun–expressing and nonexpressing cells. In c-Jun–positive MIN6 cells, both MafA and insulin expression were reduced (Fig. 4,A), while the negative control Ad-GFP had no effect (supplemental Fig. 2). We also evaluated whether MafA and insulin were affected by c-Jun overexpression in islets isolated from the m/m mouse. As observed in MIN6 cells, protein levels of MafA were markedly reduced in most Ad-c-Jun–infected islet cells, while control Ad-GFP showed no effects on MafA (Fig. 4,B). To quantify the suppressive effects of c-Jun on insulin content in isolated islets, a large amount of Ad-c-Jun (1.0 μl of 1 × 10¹⁰ pfu/ml to 1 ml of culture medium) was used to overexpress c-Jun in entire islet cells. Consistent with the results of immunocytochemistry, insulin content measured by ELISA was significantly decreased in c-Jun–overexpressed islets (Fig. 4 C). These findings directly demonstrate that c-Jun suppresses the expression of both MafA and insulin.

It was previously reported that MafA protein level was suppressed at the posttranslational level under chronic high-glucose conditions in vitro (19). To examine how c-Jun suppresses expression of MafA, protein and mRNA levels of MafA were compared at different times after exposure to Ad-c-Jun in MIN6 cells (Fig. 5,A and B). Quantification of mRNA and protein levels showed that MafA protein levels are markedly decreased with relatively small suppression of mRNA levels within the first 48 h after exposure (Fig. 5,C and D). To further evaluate the possibility of posttranslational suppression of MafA, 24 h after Ad-c-Jun or Ad-GFP infection, MIN6 cells were incubated with 10 μg/ml of cycloheximide (CHX) for another 24 h. As Fig. 5 E shows, protein level of MafA was markedly decreased by the 24-h exposure to CHX with overexpression of control GFP (lane 5). In the same condition, the loss of MafA was more intensified with overexpression of c-Jun (lane 6), although c-Jun had no effect on protein level of MafA within first 24 h (lane 2). These results suggest that c-Jun possibly reduces MafA levels via a posttranslational mechanism, although it was also observed that c-Jun significantly suppressed MafA at the mRNA level.

Since c-Jun decreased the expression of both MafA and insulin, we speculated that c-Jun decreases insulin gene transcription at least partially by mediating the suppression of MafA expression in addition to E1 and CRE binding transcription factors (29,30). To investigate this speculation, reporter gene analysis was performed using MIN6 cells. Insulin promoter activity was reduced by c-Jun, which was significantly restored by MafA overexpression (Fig. 6). Importantly, MafA single overexpression had no additional effect on insulin promoter activity in MIN6 cells. These findings suggest that the downregulation of MafA is the main cause of the suppression of insulin gene transcription by c-Jun.

To evaluate whether MafA overexpression prevented insulin protein loss by c-Jun, we transfected MafA and/or c-Jun expression plasmids into MIN6 cells. Sixty hours after the transfection, immunocytochemistry was performed (Fig. 7,A). As observed with the adenoviral c-Jun overexpression in Fig. 4,A, c-Jun overexpression by a plasmid also reduced insulin protein levels (Fig. 7,A, top panels) but was restored by MafA overexpression even in c-Jun–positive cells (Fig. 7,A, bottom panels). To quantify the effects of MafA on restoration of insulin protein, c-Jun and MafA were simultaneously overexpressed with the adenoviral method. As indicated in Fig. 7,B, c-Jun–induced suppression of insulin content was significantly prevented by MafA overexpression. Next, MafA was overexpressed with Ad-MafA in isolated islets from 17-week-old db/db mice, since both MafA and insulin expression levels were suppressed in these islets. Compared with control diabetic islets infected with Ad-GFP, both insulin 1 and 2 mRNA levels were significantly increased by adenoviral MafA overexpression in isolated diabetic islets (Fig. 7,C). To clarify that the increased expression of endogenous c-Jun in diabetic islets is the cause of suppression of insulin and MafA, c-Jun expression was suppressed by infection of adenovirus-expressing siRNA of c-Jun (Ad-si-cJun) using islets isolated from 14-week-old db/db mice in which c-Jun expression was upregulated. As shown in Fig. 7 D, c-Jun mRNA level was effectively knocked down, and mRNA levels of insulin 1 and 2 and MafA were significantly increased in diabetic islets. These results clearly indicate that expression level of MafA and insulin are suppressed by physiologically increased c-Jun in the islets under diabetic conditions. Taken together, it is likely that overexpression of MafA or siRNA of c-Jun restores the c-Jun–induced suppression of insulin gene transcription and its protein level.

The functional domains of c-Jun are well characterized, and previous reports (30,36) have demonstrated effective mutants that eliminate the activity of each domain. We generated expression plasmids and adenoviruses of various c-Jun mutants, namely c-Jun transactivation domain (TAD), c-Jun basic region (BS), and c-Jun leucine zipper domain (LZ), which contain a deletion or mutation of in the transactivation, basic DNA binding, and leucine zipper dimerization domain, respectively (Fig. 8,A). We first confirmed that each plasmid and adenovirus expressed proteins of the mutant c-Jun expected size and at adequate amounts by Western blot analysis (data not shown). As shown by Western blotting in Fig. 8,B, adenoviral expression of c-Jun TAD and c-Jun BS demonstrated that these mutants did not possess the activity of c-Jun against the MafA protein, while c-Jun LZ retained this activity. Furthermore, as shown in Fig. 8 C, insulin transcription activity was also affected by these mutants in the same way as their alteration of MafA protein levels. In addition, to confirm that the leucine zipper domain is not required for this activity, we generated another mutant c-Jun that has two mutations in amino acids 290 and 297 of the leucine zipper domain. However, this mutant c-Jun also decreased insulin promoter activity to the same degree as wild-type c-Jun (data not shown). These findings indicate that the transactivation and basic region domains are critical for c-Jun to suppress MafA and insulin expression.

It is well known that insulin gene transcription is downregulated under diabetic conditions, although its precise mechanism is still unclear. One possibility is that ROS induced by hyperglycemia and/or hyperlipidemia suppress insulin gene transcription and biosynthesis, as we and other groups previously suggested. In this study, we showed that protein levels of c-Jun, whose amount and activity are upregulated by ROS, are increased in diabetic islets in vivo, and that, in contrast, MafA protein levels are markedly decreased (Fig. 1). In addition, immunostaining of MafA and c-Jun in diabetic islets showed that both MafA and insulin expression are markedly suppressed in most c-Jun–positive cells (Fig. 2). These findings imply that c-Jun is involved in the suppression of MafA expression. We further directly demonstrated that c-Jun overexpression with an expression plasmid or adenovirus markedly decreases MafA and insulin expression in isolated islets as well as MIN6 cells (Figs. 3 and 4) and that overexpression of MafA markedly, but not completely, recovered insulin promoter activity suppressed by c-Jun (Fig. 6). These results indicate that increased expression of c-Jun suppresses insulin gene expression via the decrease of MafA expression, besides its direct suppressive effect on insulin gene promoter activity through the cyclic AMP response element as reported previously. Since MafA appears to not only regulate insulin expression but also to be involved in insulin secretion (37,38), it is likely that the suppression of MafA protein levels by c-Jun leads to insulin secretory defects that are often observed under diabetic conditions.

c-Jun is well known to induce apoptosis in a variety of cells. On the other hand, disruption of the c-jun gene in mice showed that c-Jun is required for cell proliferation and differentiation (39,40). These opposite effects of c-Jun on both the promotion of and protection from apoptosis might depend on the type, severity, and length of stimuli, as well as the effects of additional factors and/or cell types. First, to examine whether the suppressive effect of c-Jun on MafA and insulin expression is simply due to the induction of apoptosis, we isolated DNA from MIN6 cells infected with Ad-c-Jun for 1, 2, or 3 days and analyzed DNA fragmentation by electrophoresis. However, DNA ladders were not detected in these experiments (supplemental Fig. 3). Furthermore, since overexpression of MafA prevented the suppressive effects of c-Jun on insulin promoter activity (Fig. 6) and protein levels (Fig. 7), it is unlikely that c-Jun induced apoptosis, at least in our model. Interestingly, some of the increased Ki67-positive cells also expressed c-Jun in the islets of db/db mice (supplemental Fig. 4), which indicates that c-Jun–positive cells are able to proliferate. These results suggest that c-Jun–mediated β-cell dysfunction is not simply due to the induction of apoptosis, although the physiological reason as to why c-Jun is induced in pancreatic islets under diabetic conditions remains unknown.

As reported previously, MafA is a phosphorylated protein, and phosphorylation is critical for its activity (41,–43). In this study, p38 and JNK, both of which are activated by ROS, were evaluated by these overexpression experiments; however, we observed no effects on insulin and MafA expression. This is consistent with the previous findings that MafA is not phosphorylated by p38 in vitro (43). Concerning JNK, which phosphorylates and activates c-Jun, we predicted that the JNK pathway might also be involved in the alteration of MafA activity. However, the effects of c-Jun on insulin promoter activity and MafA protein levels were not inhibited by a dominant negative form of JNK (data not shown). Therefore, we assume that the induction of c-Jun expression, per se, is more important for the suppression of MafA expression. However, as shown in Fig. 8, the TAD of c-Jun, which includes its phosphorylation site by JNK, is apparently important for its suppression of MafA. These findings suggest that a relatively small amount of JNK is enough to activate c-Jun and suppress MafA.

In this study, we showed that the protein levels of MafA and insulin are simultaneously reduced in the islets of diabetic mice. These results are consistent with the previous findings that RIPE3b1-binding activity is reduced in diabetic islets (18) and that chronic high glucose decreases insulin promoter activity through the reduction of RIPE3b1 activity in HIT-T15 cells (10,19). However, this is the first report to show the increase of c-Jun expression and the decrease of both MafA and insulin expression under diabetic conditions in vivo and to directly demonstrate that the increased c-Jun expression suppresses both MafA and insulin expression.

In conclusion, the findings reported in this study suggest that the augmented expression of c-Jun in diabetic islets decreases MafA activity followed by reduced insulin biosynthesis and thereby explains, at least in part, the molecular mechanism for β-cell glucose toxicity that is often observed in type 2 diabetic patients.

This work was supported by a grant from the Juvenile Diabetes Research Foundation (Career Development Award 2-2005-946 to T.-A.M.), KAKENHI (no. 10379258 to T.-A.M.), the Japan Diabetes Foundation (to T.-A.M.), the Takeda Science Foundation (to T.-A.M.), the Suzuken Memorial Foundation (to T.-A.M.), and the National Institutes of Health (R01 50203 to R.S.).

No potential conflicts of interest relevant to this article were reported.

We thank Yuko Sasaki, Satomi Takebe, and Dr. Yoshihiro Tochino for excellent technical assistance and Chikayo Yokogawa for secretarial assistance.