MafA Stability in Pancreatic β Cells Is Regulated by Glucose and Is Dependent on Its Constitutive Phosphorylation at Multiple Sites by Glycogen Synthase Kinase 3

Abstract

Regulation of insulin gene expression by glucose in pancreatic β cells is largely dependent on a cis-regulatory element, termed RIPE3b/C1, in the insulin gene promoter. MafA, a member of the Maf family of basic leucine zipper (bZip) proteins, is a β-cell-specific transcriptional activator that binds to the C1 element. Based on increased C1-binding activity, MafA protein levels appear to be up-regulated in response to glucose, but the underlying molecular mechanism for this is not well understood. In this study, we show evidence supporting that the amino-terminal region of MafA is phosphorylated at multiple sites by glycogen synthase kinase 3 (GSK3) in β cells. Mutational analysis of MafA and pharmacological inhibition of GSK3 in MIN6 β cells strongly suggest that the rate of MafA protein degradation is regulated by glucose, that MafA is constitutively phosphorylated by GSK3, and that phosphorylation is a prerequisite for rapid degradation of MafA under low-glucose conditions. Our data suggest a new glucose-sensing signaling pathway in islet β cells that regulates insulin gene expression through the regulation of MafA protein stability.

Insulin, a polypeptide hormone, is a critical regulator of blood glucose levels and is produced exclusively by the β cells in the pancreatic islets of Langerhans. A rise in blood glucose concentration induces insulin secretion as well as transcription and translation. The molecular mechanism of β-cell-restricted and glucose-responsive expression of the insulin gene has been studied extensively, and we now know the identities of several critical cis-regulatory elements in the insulin gene promoter, as well as trans-acting factors, such as Pdx1 and Beta2 (for reviews, see references 39 and 53).

Pdx1 is a homeodomain transcription factor that binds AT-rich nucleotide sequence motifs (termed A1, A3, and GG2 elements) in the insulin promoter (21, 27, 34, 38). Beta2 is a basic helix-loop-helix factor that binds as a heterodimer with the ubiquitously expressed E47 protein to the E1 element (31). Gene disruption experiments with mice have revealed that both Pdx1 and Beta2 play critical roles in insulin gene expression as well as in islet development and function (1, 30). Furthermore, mutations in pdx1 and beta2 are found in patients with maturity onset diabetes of the young type 4 and type 6, respectively, and in some populations of patients with type 2 diabetes (12, 23, 26, 47).

Previous studies have shown that Pdx1 and Beta2 are also involved in glucose-regulated insulin promoter activation and that the DNA-binding activities of these factors to their respective cis elements is induced by glucose (8, 25, 27). It has also been demonstrated that a rise in glucose concentration induces nuclear translocation and increases the transactivation capacity of Pdx1 and Beta2 (6, 24, 36, 37, 40). However, the precise molecular mechanisms by which these proteins are activated are not fully understood.

Another cis-regulatory element, termed RIPE3b/C1, has been shown to be important not only for β-cell-restricted expression of the insulin gene but also for its up-regulation by glucose (42-44). Mutation of the C1 element abolished the glucose responsiveness of the insulin promoter in reporter assays. In addition, C1-binding activity in β-cell nuclear extracts, detected by gel mobility shift assay, increased in response to glucose. A β-cell-specific transcription factor, MafA, was subsequently identified as the C1-binding factor (15, 17, 28, 35).

MafA is a member of the Maf family of basic leucine zipper (bZip) transcription factors, which also includes MafB, c-Maf, and Nrl. MafA, together with Pdx1 and Beta2, synergistically activates the insulin promoter (2, 5, 55), and coexpression of these three factors in non-β cells induces endogenous insulin expression (16). Recently, it was also shown that ablation of mafA in mice leads to a deficiency in glucose-stimulated insulin secretion from β cells, progressive β-cell degeneration, and diabetes (54).

Despite this progress in characterizing MafA function, the mechanism of MafA regulation by glucose remains unknown. In this study, we examined the phosphorylation status of MafA in β cells and examined its relationship to the biological activity of MafA and to the regulation of MafA by glucose. We demonstrated that MafA protein stability is regulated by glucose. In β cells, MafA is constitutively phosphorylated at multiple sites by glycogen synthase kinase 3 (GSK3) and is rapidly degraded under low-glucose conditions but not under high-glucose conditions.

MATERIALS AND METHODS

Plasmids and reagents.

To construct the expression vector for hemagglutinin (HA)-tagged mouse MafA (m-MafA), a double-stranded oligonucleotide encoding the HA epitope was ligated to the open reading frame of the m-mafA cDNA (pGEM-T Easy/m-mafA) (17), and the resultant HA-m-mafA fragment was inserted into pHygEF2. All of the amino acid substitution mutants of MafA were constructed by site-directed overhang extension PCR mutagenesis (14) using pHygEF2/HA-m-mafA as the template. pHygEF2/m-mafAΔBsu36I was constructed by double digestion of pHygEF2/m-mafA (17) with ClaI (located in the multicloning site) and Bsu36I (located in the mafA open reading frame), followed by blunting and self-ligation.

To construct the IRES-EGFP cassette plasmid, a NotI-NcoI fragment from pUC/NotI/IRES (20), containing the internal ribosome binding site (IRES) of encephalomyocarditis virus, and an NcoI-NotI fragment from pEGFP-1 (Clontech), containing the entire open reading frame of enhanced green fluorescent protein (EGFP), were ligated and inserted into the NotI site of pBluescript II SK(−). The NotI fragment was then excised and inserted into the unique NotI site of pHygEF2 and pHygEF2/HA-m-mafA (the wild type [WT], the 4A mutant [S49A T53A T57A S61A], or the 5A mutant [S49A T53A T57A S61A S65A]) to generate pHygEF2/IRES-EGFP and pHygEF2/HA-m-mafA (WT, 4A, or 5A)-IRES-EGFP, respectively.

pHygEF2/HA-h-mafA, pHygEF2/FLAG-Pdx1, pHygEF2/FLAG-Beta2, h-ins-p-luc, and pEF-Rluc were described previously (17, 20). pG4x5/TATA/luc was a gift from Kazuhiko Igarashi (Tohoku University).

A DNA fragment containing the Gal4 DNA binding domain (DBD), Gal4-DBD, was amplified by PCR using the specific primers 5′-AGAGAATTCGTCATGAAGCTACTGTCTTCT-3′ and 5′-AGAGGTACCCAAGCTTCGATACAGTCAACT-3′ and pEF/Gal4-DBD (19) as the template. An amplified fragment was digested at the newly incorporated KpnI restriction site at its 3′ end, ligated to the SacII site of m-mafA by blunt-end ligation, digested with EcoRI and BssHII, and inserted into the corresponding sites of pHygEF2. The resultant plasmid, pHygEF2/Gal4-mafA, encoded Gal4-DBD fused with amino acid residues 4 to 148 of MafA, with an additional His Arg construct at the carboxyl terminus.

U0126, MG132, and lactacystin were purchased from Calbiochem. Cycloheximide (CHX) (Sigma), lithium chloride (LiCl) (Nacalai tesque), and SB216763 (Tocris) were also obtained commercially.

Cells and transfection.

MIN6 cells were a generous gift from Jun-ichi Miyazaki (Osaka University) (29). βTC6 and NIT-1 cells were purchased from the American Tissue Culture Collection. The In1024 cell line was established from a BK virus-induced hamster insulinoma (48). MIN6, βTC6, In1024, NIH 3T3, BHK21, and CHO cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 15% (for MIN6 and βTC6 cells) or 10% (for the others) fetal calf serum. NIT-1 cells were grown in F12K medium supplemented with 10% fetal calf serum. The medium contained 25 mM glucose unless otherwise indicated. When glucose conditions were varied, the medium was changed to DMEM containing the indicated concentrations of glucose and supplemented with 10% fetal calf serum. Primary islets of Langerhans were isolated from 5-week-old Jcl:ICR mice (CREA Japan, Inc.) by a standard collagenase digestion method (9) and grown in RPMI containing 10% fetal calf serum.

For transfection, In1024, NIH 3T3, BHK21, or CHO cells grown in 24-well plates were transfected with a total of 0.8 μg of plasmid using 2 μl of Lipofectamine 2000 reagent (Invitrogen). For MIN6 cells, 3.2 μg of plasmid and 8 μl of Lipofectamine 2000 reagent were used.

Preparation of cell extracts and immunoblotting.

Nuclear extracts were prepared as described previously (41). Whole-cell extracts were prepared by direct addition of 4× sodium dodecyl sulfate (SDS) sample buffer (200 mM of Tris-HCl, pH 6.8, 8% SDS, 400 mM dithiothreitol, 0.2% bromophenol blue, 40% glycerol) to the cells or by cell lysis in radioimmunoprecipitation assay buffer (10 mM Tris-HCl, pH 7.5, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS, 150 mM NaCl) containing protease inhibitor cocktail (Nacalai tesque), followed by sonication for 20 s. For calf intestine alkaline phosphatase (CIAP) treatment, 20 U of CIAP (Takara) was added to aliquots of nuclear extracts (∼1 to 2 μg) in 25 mM Tris-HCl, pH 7.9, 100 mM NaCl, 5 mM MgCl₂, and 10 mM dithiothreitol and incubated at 30°C for 30 min. For metabolic labeling with ³²P, a MIN6-derived stable cell line expressing HA-MafA-IRES-EGFP (clone no. 5) was grown in 12-well plates in DMEM containing 0.3 or 20 mM glucose for 12 h. Cells were washed with phosphate-free DMEM and then labeled by adding 0.5 mCi of [³²P]orthophosphate (MP Biomedicals) for 6.5 h. For pulse and chase experiments, MIN6/HA-MafA-IRES-EGFP cells (clone no. 5) were grown in DMEM containing 0.3 or 20 mM glucose and then washed with methionine- and cysteine-free DMEM. Cells were labeled by adding 100 μCi of Redivue PRO-MIX (Amersham Biosciences) for 2 h and then incubated in DMEM containing 0.3 mM or 20 mM glucose.

For immunoprecipitation, anti-HA-agarose (Roche) was used. Immunoblotting was performed as described previously (17). To detect endogenous MafA protein, anti-c-Maf (M-153; Santa Cruz Biotechnology) was used because it cross-reacts with large Maf subfamily members (c-Maf, MafB, and MafA). It is referred to as anti-pan-Maf serum in this study. Anti-v-Maf (18) was used to detect MafAΔBsu36I. The other primary antisera used in this study were as follows: anti-HA (MBL), anti-phospho-p44/42 mitogen-activated protein (MAP) kinase (Erk1/2) (Thr202/Tyr204) (Cell Signaling), anti-GSK3 (0011-A; Santa Cruz Biotechnology), anti-phospho-GSK3β (Ser9) (Cell Signaling), anti-Akt (Cell Signaling), anti-phospho-Akt (Ser473) (Cell Signaling), anti-Gal4 (DBD) (Santa Cruz Biotechnology), anti-GFP (Clontech), and anti-TF-IID (TBP SI-1; Santa Cruz Biotechnology).

In vitro phosphorylation.

To express recombinant HA-tagged MafA (rec-HA-MafA) in Escherichia coli, a DNA fragment containing HA-m-mafA was excised from pHygEF2/HA-m-mafA by digestion with NcoI (located at the initiator methionine) and EcoRI (located in the vector sequence downstream of the open reading frame) and inserted into the NcoI and XhoI sites of pET-14b (Novagen). The resultant plasmid, pET14b/HA-m-mafA, encoded rec-HA-MafA that lacked the amino-terminal His tag but contained its own His cluster, thus enabling one-step purification using Ni-nitrilotriacetic acid agarose (QIAGEN) under urea-denatured conditions. Purified rec-HA-MafA was dialyzed against HGKEDN buffer (20 mM HEPES, pH 7.9, 10% glycerol, 100 mM KCl, 0.2 mM EDTA, 1 mM dithiothreitol, 0.1% NP-40).

For the in vitro kinase assay, 2 ng of rec-HA-MafA and 1 μg of MIN6 nuclear extract or recombinant GSK3β (Calbiochem) were combined in reaction buffer (final concentration of 5 mM HEPES, pH 7.9, 100 mM NaCl, 5 mM KCl, 0.8 mM dithiothreitol, 4.5% glycerol, 1 mM MgCl₂, 10 mM Na₃VO₄, and 2.5 mM ATP) and incubated at 30°C for 60 min.

Luciferase assay.

In1024 cells grown in 24-well plates were transfected with a total of 0.8 μg of plasmid DNA (0.1 μg luciferase plasmid, a total of 0.6 μg expression plasmids, and 0.1 μg pEF-Rluc) using 2 μl of Lipofectamine 2000 reagent (Invitrogen). Cells were harvested 24 h after transfection. Firefly and Renilla luciferase activities were measured using a dual luciferase assay system (Promega). Data represent the averages ± standard errors of three independent experiments.

RESULTS

MafA is phosphorylated by GSK3 in β cells.

To investigate the role of posttranslational modification in the regulation of MafA function, we examined whether MafA was phosphorylated in β cells. We prepared nuclear extracts from two mouse insulinoma cell lines, MIN6 and βTC6, treated them with CIAP, and subjected them to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis using an anti-pan-Maf antibody. Multiple bands of approximately 48 kDa were detected in both of the insulinoma cell lines (Fig. 1A). Treatment with CIAP caused a shift in mobility to a more rapidly migrating form of 40 kDa, which indicates that MafA is present in multiple phosphorylation states in β cells and that these different states can be resolved by determining mobility during SDS-PAGE.

FIG. 1.

Phosphorylation of MafA by GSK3 in β cells. (A) Nuclear extracts prepared from MIN6 and βTC6 cells were treated with (+) or without (−) CIAP. Endogenous MafA was detected by Western blot analysis using anti (α)-pan-Maf antiserum, which reacts with the large Maf family members MafA, MafB, and c-Maf. The asterisk indicates weakly expressed c-Maf. (B) MIN6 cells and a stable transformant that expresses HA-tagged MafA (see Fig. 6E) were labeled with [³²P]orthophosphate, and cell extracts were subjected to immunoprecipitation (I.P.) with anti-HA agarose. The precipitates were analyzed by SDS-PAGE and autoradiography (right). In a parallel experiment, unlabeled total cell extracts were prepared and analyzed by Western blot (W.B.) analysis using anti-HA and anti-TATA-binding protein (TBP) (as a loading control) antibodies (left). (C) MIN6 cells were treated with 25 μM of the MEK inhibitor U0126 for 6 h, and whole-cell extracts were analyzed by Western blotting using pan-Maf antisera or anti-phospho-Erk (Pi-Erk1/2). (D) GSK3 inhibition in MIN6 cells affects the mobility of MafA. MIN6 cells were treated with the indicated concentrations of SB216763 or LiCl for 20 h. Whole-cell extracts were prepared and subjected to Western blot analysis using anti-pan-Maf serum. (E) Mouse (βTC6 and NIT-1) and hamster (In1024) insulinoma-derived cell lines were treated with 20 μM of the GSK3 inhibitor SB216763, and then the mobility of endogenous MafA was analyzed by SDS-PAGE as described for panel D.

To confirm that MafA was phosphorylated in cells, we metabolically labeled MIN6 cells or a stable clone of MIN6 cells that expresses HA-MafA (see Fig. 2A) with [³²P]orthophosphate. Cell extracts were subjected to immunoprecipitation using an anti-HA antibody and subsequently analyzed by autoradiography. As shown in Fig. 1B, ³²P was incorporated into HA-MafA. This result was similar to previous results demonstrating the incorporation of ³²P into HA-tagged quail MafA in HeLa cells (3).

FIG. 2.

Identification of phosphorylated residues in MafA in β cells. (A) HA-MafA and MafAΔBsu36I were expressed in In1024 cells, and the cells were treated with (+) or without (−) 20 μM SB216763. Cell extracts were then analyzed by Western blotting using anti (α)-HA or anti-v-Maf, which cross-reacts with MafA. (B) Identification of phosphoacceptor residues of MafA. Expression plasmids for HA-MafA-WT or the indicated Ala substitution mutants were transfected into In1024 cells, and whole-cell extracts were analyzed by Western blotting using anti-HA antibody. +S72/S342A at bottom indicates that the mutants also contained Ala substitutions at both Ser72 and Ser342. (C) Summary of phosphorylated residues of MafA and comparison with the other Maf family members. hu, human; mu, murine; ck, chicken; zf, zebrafish. (D) Comparison of the mobilities of single and multiple substitution mutants of MafA. Mutants of HA-MafA with single (S49A, T53A, T57A, S61A, and S65A) or multiple (2A, 3A, 4A, and 5A mutants) Ala substitutions at the indicated residues were expressed in In1024 cells and analyzed by anti-HA Western blotting. (E) Effect of GSK3 inhibition on mobility of MafA mutants. In1024 cells transfected with expression vectors for HA-MafA or the indicated mutants were treated with SB216763 (20 μM) for 4 h. Cell extracts were analyzed by Western blotting using anti-HA antibody. Mutants containing additional Ala substitutions at Ser72 and Ser342 are indicated by +S72/342A (bottom).

We next attempted to identify the kinase(s) responsible for phosphorylation of MafA. An avian homolog of MafA was shown to be phosphorylated by Erk in vitro and in vivo in primary lens cells and HeLa cells (3, 32, 45). We treated MIN6 cells with the MEK1/2 inhibitor U0126, which also inhibits subsequent phosphorylation and activation of Erk. Treatment with U0126 had no effect on the mobility of MafA (Fig. 1C). Treatment of the cells with other MAP kinase inhibitors (SB203580 and SP600125, inhibitors of p38 and Jun N-terminal protein kinase, respectively) alone or in combination also had no effect (data not shown). These results indicate that phosphorylation of MafA, as determined by mobility during SDS-PAGE, is not due to the activity of members of the MAP kinase family in β-cell lines.

We then screened several other types of kinase inhibitors in MIN6 cells for their abilities to alter the mobility of MafA during SDS-PAGE and found that SB216763, a specific inhibitor of GSK3, caused the rapidly migrating form of MafA to accumulate in a dose-dependent manner (Fig. 1D). Another GSK3 inhibitor, LiCl, had a similar effect (Fig. 1D). Treatment of other mouse (βTC6 and NIT-1) and hamster (In1024) β-cell lines with SB216763 also resulted in similar changes in MafA mobility (Fig. 1E). These results suggest that MafA is phosphorylated by GSK3 in β cells.

Identification of conserved Ser and Thr residues at the amino terminus of MafA.

To determine the residues of MafA that were phosphorylated in β cells, expression vectors encoding HA fused to full-length MafA (HA-MafA) or an amino-terminal truncation of MafA (MafAΔBsu36I) were transiently transfected into In1024 cells, and MafA mobility in these cells in the presence of SB216763 was examined. The change in mobility of HA-MafA following SB216763 treatment was similar to that of endogenous MafA, whereas the mobility of MafAΔBsu36I was unaffected (Fig. 2A). These results indicate that the GSK3 inhibitor-sensitive site(s) of phosphorylation of MafA is located in the amino-terminal region of MafA.

To identify the phosphorylated amino acid residue(s) in the amino-terminal region of MafA, we generated a set of substitution mutants of MafA, in which each Ser or Thr residue in this region was replaced with Ala, and examined the mobilities of the point mutants by SDS-PAGE. Mutation of Ser49, Thr53, Thr57, Ser61, or Ser65 (S49A, T53A, T57A, S61A, or S65A, respectively) increased MafA mobility (Fig. 2B, top). We also found that Ser72 and Ser342 of MafA were phosphorylated but that GSK3 was not involved in the phosphorylation of these residues (data not shown). In the background of a Ser72/Ser342 double substitution mutant, the mobility changes of S49A, T53A, T57A, S61A, and S65A mutants were more evident (Fig. 2B, bottom). In this set of experiments, we used the hamster insulinoma-derived cell line In1024 because of its high transfection efficiency compared to those of the other mouse β-cell lines (MIN6, βTC6, and NIT-1). We observed essentially the same mobility patterns for the Ala substitution mutants in MIN6 and NIT-1 cells (data not shown).

Attempts to identify the phosphorylation sites of MafA by mass spectrometry analysis were unsuccessful, most likely due to technical difficulties associated with detecting peptides containing multiple phosphoserine and phosphothreonine residues. However, the putative phosphorylation sites that we identified in the amino terminus of MafA (Ser49, Thr53, Thr57, Ser61, and Ser65) are well conserved throughout evolution among Maf family members and are spaced every 4 amino acids (Fig. 2C). It has been shown previously that phosphorylation by GSK3 requires prior phosphorylation of a Ser or Thr residue 4 residues carboxy terminal to the target site by another kinase (priming kinase) and that GSK3 sequentially phosphorylates every fourth Ser or Thr residue amino terminal to the priming site (7). Therefore, it seemed plausible that Ser65 of MafA functions as a priming site and is phosphorylated by an as-yet-unidentified priming kinase and that GSK3 sequentially phosphorylates Ser61, Thr57, Thr53, and Ser49. Consistent with this hypothesis, we observed that there were differences in the mobilities of the mutants (Fig. 2B). For example, the mobility of S65A was greater than that of S61A, most likely because Ser49, Thr53, Thr57, and Ser61, in addition to Ser65, were not phosphorylated in the S65A mutant, whereas at least at Ser65 was phosphorylated in the S61A mutant.

To test this hypothesis, we constructed MafA mutants with multiple mutations at each of the putative GSK3 phosphorylation sites and compared their mobilities. The mobility of a single substitution mutant was similar to that of a mutant with multiple substitutions at more-amino-terminal residues (Fig. 2D, top). For example, the mobility of the S61A mutant was similar to that of the 4A mutant (S49A T53A T57A S61A). This behavior of the mutants was more clearly evident when Ser72 and Ser342 were also mutated (Fig. 2D, bottom).

We also examined the effect of a GSK3 inhibitor on the mobility of the MafA phosphorylation site mutants. Treatment of cells with a GSK3 inhibitor altered the mobilities of WT MafA and the S49A, T53A, and T57A mutants to a form that appeared similar to that of the S61A mutant, whereas the mobilities of the S61A and S65A mutants were not affected (Fig. 2E). These results provide further evidence that Ser65 is the priming site and that Ser61, The57, Thr53, and Ser49 are then subsequently phosphorylated by GSK3.

Phosphorylation of MafA by GSK3 requires priming phosphorylation.

Obtaining direct evidence of MafA phosphorylation by GSK3, by methods such as an in vitro kinase assay using recombinant MafA and GSK3, is not possible, since the priming kinase that phosphorylates Ser65 is unknown. To overcome this limitation, we performed an in vitro kinase assay using nuclear extracts prepared from MIN6 cells.

Full-length rec-HA-MafA was purified from bacteria and subjected to an in vitro kinase reaction using MIN6 nuclear extract. The mobility of purified rec-HA-MafA was similar to that of CIAP-treated HA-MafA expressed in MIN6 cells (Fig. 3A, lanes 2 and 3, form “U”). After the in vitro kinase reaction, a slower-migrating form of MafA that had a mobility similar to that of the hyperphosphorylated forms of HA-MafA in MIN6 cells (Fig. 3A, lane 1, form “F”) appeared (Fig. 3A, lane 4). Thus, the results of the in vitro kinase reaction mimicked conditions observed in vivo. The addition of recombinant GSK3β instead of MIN6 nuclear extract into the reaction mixture had no effect on MafA mobility (Fig. 3A, lane 6).

FIG. 3.

Phosphorylation of MafA in vitro. (A) Nuclear extracts of MIN6 cells stably expressing HA-MafA were treated with (+) or without (−) CIAP (lanes 1 and 2). In lanes 3 to 8, rec-HA-MafA was subjected to in vitro phosphorylation using nuclear extracts (N.E.) prepared from untransfected MIN6 cells (lanes 3, 4, 5, 7, and 8) and/or recombinant GSK3β (lanes 5 and 6) for the indicated time period. GSK3 inhibitor SB216763 (20 μM) or vehicle (dimethyl sulfoxide [DMSO], 5%) was added to the samples in lanes 7 and 8, as shown. Both rec-HA-MafA and transfected HA-MafA proteins were visualized by Western blot analysis using an anti-HA antibody. Unphosphorylated and fully phosphorylated forms of MafA are shown, and the arrowhead indicates the partially phosphorylated “primed” form of MafA. Asterisks indicate nonspecific (n.s.) cross-reacting material in the MIN6 cell extracts. (B) Phosphorylation of “primed” recombinant MafA by recombinant GSK3β in vitro. rec-HA-MafA was subjected to in vitro phosphorylation using MIN6 nuclear extracts in the presence or absence of SB216763 as indicated and was purified from the kinase reaction mixture by anti (α)-HA immunoprecipitation (I.P.). It was then subjected to a second kinase assay using recombinant GSK3β protein. (C) MIN6 cells stably expressing HA-MafA were treated with or without LiCl (20 mM) as indicated and then subjected to immunoprecipitation using an anti-HA antibody. Immunopurified HA-MafA was used as the substrate for an in vitro kinase assay in the presence (lane 3) or absence (lanes 1 and 2) of recombinant GSK3β and then visualized by anti-HA Western blotting (top). In a parallel kinase reaction, [γ-³²P]ATP was included instead of cold ATP, and the reaction mixtures were analyzed by SDS-PAGE and autoradiography (bottom).

The fully phosphorylated form of MafA was not detected when SB216763 was included in the in vitro kinase reaction (Fig. 3A, lane 8), which indicated that the kinase in MIN6 nuclear extracts that is responsible for phosphorylating recombinant MafA is GSK3. Instead, a faster-migrating form of MafA accumulated in the presence of the GSK3 inhibitor (Fig. 3A, lane 8). We purified rec-HA-MafA from the reaction mixture containing SB216763 by immunoprecipitation and subjected it to a second round of in vitro phosphorylation using recombinant GSK3β. When the products were analyzed by SDS-PAGE, we detected the fully phosphorylated form of MafA, which indicated that GSK3β directly phosphorylates MafA (Fig. 3B, lanes 4 and 5). When untreated MafA was subjected to an in vitro phosphorylation reaction using GSK3β, there was no change in mobility (Fig. 3B, lanes 1 and 5), demonstrating that a priming phosphorylation event is required for MafA phosphorylation by GSK3.

To confirm the results of the in vitro kinase reaction, we treated MIN6 cells that stably expressed HA-MafA with the GSK3 inhibitor LiCl, and the “primed” form of MafA was purified by immunoprecipitation with an anti-HA antibody. The mobility of MafA was faster when the cells were treated with LiCl (Fig. 3C, top, lanes 1 and 2), similar to previous results (Fig. 1D). Following in vitro phosphorylation with recombinant GSK3β, the slower-migrating forms of MafA were apparent (Fig. 3C, top, lane 3). When the kinase reaction was performed in the presence of [γ-³²P]ATP, ³²P was incorporated into MafA (Fig. 3C, bottom). These results indicate that GSK3-mediated phosphorylation of MafA is dependent on a prior priming phosphorylation event.

Phosphorylated forms of MafA are unstable.

Previously, it was shown that Erk phosphorylates serine and threonine residues of avian MafA which correspond to Ser14, Thr53, and Ser65 of mammalian MafA and that phosphorylation by Erk enhanced proteasome-dependent degradation of MafA in cultured lens cells (32). To examine the relationship between MafA protein stability and phosphorylation status in β cells, we treated βTC6 cells with the proteasome inhibitor MG132. In the presence of MG132, there was an accumulation of endogenous MafA protein (Fig. 4A). Similar results were observed for MIN6 cells that expressed exogenous MafA following treatment with MG132 or another proteasome inhibitor, lactacystin (Fig. 4B). Accumulation of exogenously expressed HA-MafA upon MG132 treatment was also observed in the β-cell lines βTC6 and NIT-1 but not in NIH 3T3 fibroblasts (Fig. 4C). These results together suggest that MafA is degraded via the proteasome pathway in a cell-type-specific manner.

FIG. 4.

Characteristics of phosphorylated MafA in vivo. (A) Accumulation of endogenous MafA by MG132 treatment in β cells. βTC6 cells were treated with (+) or without (−) 5 μM MG132 for 8 h, and whole-cell extracts were analyzed by Western blotting using anti (α)-pan-Maf serum. Anti-TATA-binding protein (TBP) was analyzed as a loading control. (B) MIN6 cells stably expressing HA-tagged MafA were treated with MG132 (5 μM) or lactacystin (lact.) (10 μM) for 9 h. (C) Cell-specific degradation of MafA. βTC6, NIT-1, and NIH 3T3 cells were transiently transfected with an expression vector encoding HA-MafA and treated with MG132 as described for panel A. (D) Sensitivity of MafA mutants to MG132 treatment. MIN6 cells transfected with expression vectors encoding the indicated MafA mutants were treated with MG132 (5 μM for 8 h), and whole-cell extracts were analyzed by Western blotting using anti-HA antibody. (E) Degradation rates of MafA and MafA mutants. HA-MafA-WT, HA-MafA-4A, or HA-MafA-5A (Fig. 2D) was expressed in In1024 cells. Twenty-four hours after transfection, nascent protein synthesis was blocked by treatment with 10 μg/ml of CHX for the indicated time period, and cell extracts were analyzed by Western blotting. (F) Transactivation properties of WT MafA or the 4A or 5A mutant. A luciferase reporter plasmid driven by the human insulin promoter (h-ins-p-luc) and expression plasmids for MafA (WT) or its derivatives (4A or 5A mutant) were cotransfected into In1024 cells with or without expression vectors for FLAG-Pdx1 and FLAG-Beta2. Luciferase activity was assayed 24 h after transfection and is expressed as severalfold activation over luciferase activity in cells that received empty expression plasmid. Variability in Renilla luciferase activities, which represents transfection efficiency, ranged from 10 to 23% in each set of experiments. The inset indicates expression levels of transfected MafA and its derivatives.

To test whether phosphorylation by GSK3 affected the stability of MafA, several of the Ala substitution mutants of MafA were expressed in MIN6 cells. The Ala substitution mutants were less sensitive than WT MafA to MG132 treatment (Fig. 4D), which indicated that they were more resistant to proteasome-mediated degradation. We also compared the degradation rates of WT MafA and the 4A and 5A mutants, which contained Ala substitutions of four and five phosphorylation sites, respectively (Fig. 2D). In1024 cells were transfected with WT or mutant MafA and then treated with the protein synthesis inhibitor CHX. As shown in Fig. 4E, both the 4A and 5A mutants disappeared more slowly than WT MafA in the presence of CHX. These results suggest that phosphorylation of MafA by GSK3 confers susceptibility to proteasomal degradation.

We next examined the transactivation activities of the MafA phosphorylation site mutants. A luciferase reporter gene driven by the human insulin promoter was cotransfected into In1024 cells with an expression plasmid encoding WT MafA or the 4A or 5A mutant of MafA. MafA alone was a weak activator of the insulin promoter, probably due to the high basal activity of the insulin promoter in the insulinoma-derived cell line, and the 4A and 5A mutants had comparable transactivation activities (Fig. 4F). Coexpression of Pdx1 and Beta2 had a synergistic effect on MafA activity (Fig. 4F), which was in good agreement with previous observations (2, 5, 16, 55). Both the 4A and 5A mutants also functioned synergistically with Pdx1 and Beta2 and exhibited higher luciferase activities than WT MafA. However, the expression levels of the 4A and 5A mutants were higher than that of WT MafA (Fig. 4F, inset); thus, additional experiments are needed to quantitate the relative intrinsic transactivation potentials of MafA and the mutants. At least, these results indicate that amino-terminal phosphorylation is not essential for transactivation by MafA.

To determine whether the amino-terminal region of MafA was sufficient to trigger both phosphorylation and degradation, this region was fused with the DBD of Gal4 (Gal4-MafA-WT) (Fig. 5A). This system also enabled us to examine the transactivation activity of the amino terminus of MafA, as this region has been shown to contain a transactivator domain (15, 35). The corresponding regions of the 4A and 5A mutants were also fused to Gal4-DBD to create Gal4-MafA-4A and Gal4-MafA-5A, respectively. The fusion proteins were expressed in In1024 cells, and cells were treated with SB216763 or MG132. Gal4-MafA-WT was phosphorylated by GSK3 and sensitive to MG132, whereas Gal4-MafA-4A and -5A were not phosphorylated by GSK3 and were less sensitive to MG132 (Fig. 5B). Analysis of CHX treatment showed that both Gal4-MafA-4A and -5A also degraded more slowly than Gal4-MafA-WT (Fig. 5C). All three proteins exhibited significant transcriptional activities, which correlated with their expression levels (Fig. 5D). These results suggest that the amino-terminal region of MafA contains the domains that are sufficient for transactivation as well as phosphorylation-dependent degradation.

FIG. 5.

Analysis of MafA phosphorylation carried out using the Gal4 fusion system. (A) Schematic representation of the Gal4-MafA fusion constructs. (B) Expression of Gal4-MafA fusion proteins and the effects of SB216763 or MG132. In1024 cells were transfected with expression plasmids for the indicated Gal4 fusion constructs and then treated with SB216763 (+S) or MG132 (+M). Whole-cell extracts were prepared and subjected to Western blot analysis using anti (α)-Gal4 serum. (C) Degradation of Gal4-MafA and its derivatives. In1024 cells expressing the indicated Gal4-MafA fusion constructs were treated with CHX for the indicated time periods, and the amounts of Gal4 fusion protein were determined by anti-Gal4 Western blot analysis. (D) Luciferase assay. Reporter plasmids containing multiple Gal4-binding sites (pG4x5/TATA/luc) were transfected into In1024 cells together with an expression plasmid for Gal4-DBD or the indicated Gal4-MafA fusion proteins. Transfected cells were grown for 24 h and assayed for luciferase activity. Variability of Renilla luciferase activity within each set of experiment ranged from 5 to 20%. The right panel indicates the expression levels of transfected Gal4 fusions. TBP, TATA-binding protein.

GSK3 phosphorylation is a prerequisite for MafA degradation under low-glucose conditions.

MafA has been shown to regulate insulin gene expression in response to glucose (17, 42-44). As we reported previously, growth of the glucose-sensitive β-cell line MIN6 under high-glucose conditions resulted in an accumulation of MafA protein (Fig. 6A), which correlated with DNA-binding and transcriptional activities (17). To examine the relationship between glucose-induced increases in MafA protein levels and the degradation of MafA, we treated MIN6 cells with MG132 under low- or high-glucose conditions. Accumulation of endogenous MafA protein was augmented by MG132 treatment under low-glucose conditions to a level similar to that seen under high-glucose conditions (Fig. 6A). In contrast, MG132 had little effect on MafA levels in cells grown under high-glucose conditions. These results suggest that glucose-mediated regulation of MafA protein levels occurs mainly at the posttranslational level and that under low-glucose conditions MafA is rapidly degraded by the proteasome, whereas under high-glucose conditions degradation is suppressed.

FIG. 6.

Phosphorylation of MafA by GSK3 is a prerequisite for glucose-dependent regulation of protein stability in β cells. (A) Loss of the glucose responsiveness of MafA by inhibition of the proteasome. MIN6 cells were pretreated with (+) or without (−) 5 μM of MG132 for 2 h and then grown in the presence (20 mM) or absence (0.3 mM) of glucose for 16 h. Endogenous MafA was detected by Western blot analysis using anti (α)-pan-Maf serum. (B) Effect of GSK3 inhibition on the regulation of MafA stability by glucose. MIN6 cells grown in the presence (20 mM) or absence (0.3 mM) of glucose with or without 20 μM of SB216763 for 14 h were analyzed for MafA expression levels. (C) Expression plasmids for WT MafA or the 5A mutant of MafA were transfected into MIN6 cells. Twenty-four hours after transfection, cells were switched to high- or low-glucose conditions for an additional 12 h. Exogenously expressed MafA was purified by anti-HA immunoprecipitation and analyzed by Western blotting using anti-HA antibody. (D) Purified mouse primary islets were cultured in low-glucose (2 mM) or high-glucose (15 mM) medium with or without 5 μM of MG132 or 20 μM of SB216763 for 18 h and then analyzed by Western blotting using anti-pan-Maf antibody. The arrowhead indicates nonspecific (n.s.) cross-reacting material and provides a loading control. (E) Stable clones of MIN6 cells expressing HA-tagged MafA (WT or 4A mutant) and IRES-EGFP (shown schematically at top) were established. Two independent clones were grown in the presence or absence of glucose, and the expression levels of the transgenes were examined by Western blotting using anti-HA or anti-GFP antiserum. (F) Expression levels of endogenous (endo) and exogenous (exo) Maf proteins in MIN6 clones. Whole-cell extracts prepared from MIN6 cells stably expressing HA-MafA-WT (clone no. 5) or the 4A mutant (clone no. 2) (Fig. 6E) were analyzed by Western blotting using anti-HA, anti-pan-Maf, anti-EGFP, and anti-TATA-binding protein (TBP) antisera. (G) NIH 3T3, BHK21, and CHO cells were transfected with expression vectors for WT MafA, the 4A mutant, or the 5A mutant linked to IRES-EGFP (Fig. 6E) and then treated as described for panel C. Exogenous MafA and EGFP expression levels were analyzed by anti-HA and anti-EGFP immunoblotting, respectively.

To examine the relationship between glucose-induced changes in endogenous MafA levels and phosphorylation by GSK3, MIN6 cells cultured under low- or high-glucose conditions were treated with SB216763 to inhibit GSK3. Similar amounts of hypophosphorylated MafA accumulated under both conditions (Fig. 6B). These results suggest that degradation of MafA under low-glucose conditions requires phosphorylation by GSK3 and that hypophosphorylated forms of MafA escape glucose regulation. In agreement with this, the levels of the 5A mutant of MafA, lacking all five of the phosphorylation sites in the amino-terminal region of the protein, were similar under both low- and high-glucose conditions when transiently expressed in MIN6 cells (Fig. 6C). In mouse primary islets, glucose also regulated the abundance of endogenous MafA protein, and this ability to regulate MafA was blunted by MG132 or SB216763 treatment (Fig. 6D).

To confirm these results, we established MIN6 clonal cell lines that stably expressed wild-type MafA or the 4A mutant fused to IRES-EGFP under the control of the constitutively active EF-1α promoter (Fig. 6E, top). The expression of exogenous wild-type MafA was comparable to that of endogenous MafA (Fig. 6F), and exogenously expressed MafA protein accumulated in response to glucose (Fig. 6E, bottom). In contrast, the 4A mutant was more abundant than endogenous MafA (Fig. 6F) and was insensitive to glucose (Fig. 6E, bottom).

Exogenously expressed MafA and the 4A and 5A mutants accumulated to similar levels under both low- and high-glucose conditions in NIH 3T3 (mouse embryonic fibroblast), BHK21 (Syrian hamster kidney), and CHO (Chinese hamster ovary) cells, (Fig. 6G), which suggests that phosphorylation-dependent regulation of MafA stability in response to glucose is specific to β cells.

MafA degradation rate, but not phosphorylation, is regulated by glucose.

Previous studies have shown that GSK3 activity is negatively regulated by Akt-mediated phosphorylation (4) and that Akt is activated by glucose in β cells (46). We hypothesized that glucose inhibited GSK3 activity through Akt, leading to MafA hypophosphorylation and accumulation. The levels of phosphorylated Akt (active form) and phosphorylated GSK3β (inactive form) both increased under high- glucose conditions in MIN6 cells (Fig. 7A). However, the phosphorylation status of MafA appeared to be unchanged by glucose (Fig. 7A; see also Fig. 6A and B).

FIG. 7.

Overall phosphorylation of MafA is unchanged by glucose. (A) Effect of glucose on GSK3 (Pi-GSK3β) and Akt (Pi-Akt) phosphorylation. Whole-cell extracts from MIN6 cells grown in high (20 mM) or low (0.3 mM) glucose for 24 h were analyzed by Western blotting using the indicated antibodies. (B) Effect of glucose on the overall phosphorylation status of MafA. MIN6 cells stably expressing HA-MafA were grown in high- or low-glucose medium, and various amounts of whole-cell extract were subjected to immunoprecipitation (I.P.) and Western blot analysis using anti (α)-HA antibody. (C) Effect of glucose concentration on ³²P incorporation into MafA in vivo. MIN6 cells stably expressing IRES-EGFP (−) or HA-MafA-IRES-EGFP (+) were grown for 12 h in high- or low-glucose medium and then labeled with [³²P]orthophosphate for 6.5 h. Cell extracts were analyzed by immunoprecipitation with anti-HA antibody, followed by SDS-PAGE and autoradiography (top). In a parallel experiment, unlabeled cell extracts were prepared and the expression levels of HA-MafA, EGFP, and TATA-binding protein (TBP) were analyzed by Western blotting. (D) Nuclear extracts (NE) from MIN6 cells grown in high- or low-glucose medium were assayed for kinase activity by using rec-HA-MafA protein as a substrate. The arrow indicates the phosphorylated form of MafA.

To confirm this, a stable MIN6 line that expressed HA-MafA was grown in high glucose, and cell extracts were subjected to immunoprecipitation using an anti-HA antibody. Extracts from increasing numbers of cells grown under low-glucose conditions were also subjected to immunoprecipitation, and immune complexes were analyzed by Western blotting. There were no gross changes in phosphorylation status of MafA in response to glucose, based on mobility during SDS-PAGE (Fig. 7B). We also labeled HA-MafA-expressing MIN6 cells grown in the presence or absence of glucose with [³²P]orthophosphate and found that ³²P incorporation into HA-MafA correlated with protein levels (Fig. 7C). Although this type of analysis does not distinguish changes in phosphorylation at individual phosphorylation sites, these results suggest that the overall phosphorylation status of MafA is unchanged by glucose.

We also examined MafA kinase activity in MIN6 nuclear extracts in vitro. Nuclear extracts were prepared from MIN6 cells grown in low- or high-glucose medium, and an in vitro phosphorylation reaction was carried out using recombinant MafA protein as a substrate. The efficiencies of phosphorylation of MafA were similar for nuclear extracts from cells grown under both conditions (Fig. 7D). These results suggest that while glucose stimulates the Akt-GSK3 pathway in MIN6 cells, MafA phosphorylation activity is unchanged by glucose.

We next examined the degradation rates of MafA under low- and high-glucose conditions. MIN6 cells stably expressing HA-MafA-WT were labeled with [³⁵S]methionine and [³⁵S]cysteine for 2 h in low- or high-glucose medium, and then the medium was changed to low- or high-glucose medium. Cell extracts were subjected to immunoprecipitation, followed by SDS-PAGE and autoradiography. As seen in Fig. 8A, ³⁵S-labeled HA-MafA disappeared more rapidly under low-glucose conditions than under high-glucose conditions.

FIG. 8.

Control of MafA degradation rate by glucose. (A) Pulse-chase experiment. MIN6 cells stably expressing HA-MafA-WT were grown in DMEM containing 0.3 or 20 mM glucose for 12 h and labeled with l-[³⁵S]methionine and l-[³⁵S]cysteine for an additional 2 h. After extensive washing, the medium was changed to DMEM containing 0.3 or 20 mM glucose for the indicated time period. Whole-cell extracts were prepared, subjected to immunoprecipitation (I.P.) using anti (α)-HA antibody, and analyzed by SDS-PAGE. Autoradiograms were quantified with a BAS2500 image analyzer (Fujifilm). Data represent the results of two independent experiments (bottom). (B) Half-life of MafA under high- and low-glucose conditions. MIN6 cells stably expressing HA-MafA-WT or MafA-4A were grown in high- or low-glucose medium for 12 h. Cells were then treated with CHX for the indicated time period. MafA was purified from whole-cell extracts by anti-HA immunoprecipitation and then analyzed by Western blotting using anti-HA antibody. Data were quantified and represent the results of two independent experiments (bottom). (C) MIN6 cells were treated with SB216763 (20 μM) for 8 h in high- or low-glucose medium and treated with CHX, and endogenous MafA protein levels were analyzed by Western blotting using anti-pan-Maf antibody. TBP, TATA-binding protein; inh., inhibitor.

To determine the effect of phosphorylation on the degradation rate of MafA, nascent protein synthesis was blocked in stable MIN6 cell lines expressing HA-MafA-WT or -4A by CHX treatment for various lengths of time, and the amounts of MafA protein were measured by Western blotting. For Western blot analysis, extracts were subjected to immunoprecipitation with an anti-HA antibody, because the amount of wild-type MafA under low-glucose conditions was too small to quantify directly. Wild-type MafA disappeared more rapidly under low-glucose conditions than under high-glucose conditions (Fig. 8B), which indicates that the rate of MafA protein degradation is regulated by glucose. In contrast, the 4A mutant was very stable under both low- and high-glucose conditions. These results suggest that phosphorylation of MafA is a prerequisite for glucose-regulated degradation.

When we examined the effect of a GSK3 inhibitor on the half-life of endogenous MafA, we found that the rate of MafA degradation was decreased both in the presence and in the absence of glucose (Fig. 8C). These results provide further evidence that phosphorylation by GSK3 is required for the rapid degradation of MafA and that the degradation rate, not the phosphorylation status, of MafA protein is regulated by glucose.

DISCUSSION

In this study, we show that the rate of degradation of the transcription factor MafA in β cells is regulated by glucose and leads to an accumulation of MafA protein under high-glucose conditions. We also present multiple lines of evidence that multiple Ser and Thr residues in the amino-terminal region of MafA are constitutively phosphorylated by GSK3 and that loss of phosphorylation, either by inhibition of GSK3 or by mutation of the phosphorylation sites of MafA, results in escape from degradation under low-glucose conditions.

Our data suggest that MafA is phosphorylated in β-cell lines at seven Ser and Thr residues (Ser49, Thr53, Thr57, Ser61, Ser65, Ser72, and Ser342) (summarized in Fig. 2C). Previous studies have shown that chicken and quail homologues of MafA are phosphorylated by Erk at Ser14, Thr57, and Ser65 in vitro and in vivo (in primary lens, primary neural retina, and HeLa cells) (3, 32, 33). It was also shown previously that Thr57, as well as Thr113 and Ser272 (equivalent to Thr131 and Ser342 of m-MafA, respectively), is phosphorylated by p38 MAP kinase in vitro and in 293T cells (45). While these phosphoacceptor sites are conserved in several species (Fig. 2C), treatment of β cells with pharmacological inhibitors of the MAP kinases Erk, p38, and Jun N-terminal protein kinase, either alone or in combination, had no effect on the mobility of endogenous MafA during SDS-PAGE (Fig. 1C and data not shown), indicating that the contribution of MAP kinases to MafA phosphorylation in β cells is insignificant. Our data suggest that in β cells, GSK3 phosphorylates Ser49, Thr53, Thr57, and Ser61 of MafA (Fig. 1 and 2). Ser65 appeared to be phosphorylated by an unknown kinase and to serve as a priming site for subsequent phosphorylation at additional sites by GSK3. It should be noted that GSK3 phosphorylates MafA without a priming phosphorylation event with low efficiency in β cells. For example, a small fraction of the S65A mutant appeared to be fully phosphorylated and was sensitive to the GSK3 inhibitor (Fig. 2D and E). The candidate phosphorylation sites that were identified in the current study, Thr53, Thr57, and Ser65, as well as the other phosphorylated residues of MafA in β cells (Ser72 and Ser342) or in other systems (Ser14 and Thr131), are followed by a Pro residue. Thus, they can potentially serve as phosphoacceptor sites for proline- directed kinases, including MAP kinases, and may be phosphorylated by different kinases depending on cell type. The kinase(s) responsible for phosphorylation of Ser65, Ser72, and Ser342 in β cells has yet to be identified.

We also present evidence that GSK3-phosphorylated MafA undergoes rapid degradation in β cells. This is consistent with previous findings demonstrating that mutation of chicken L-Maf at sites of Erk phosphorylation (Thr57 and Ser65) resulted in significant accumulation of L-Maf in primary lens cells (32). Many proteins, including transcription factors, are reported substrates of GSK3, and one of the functional consequences of GSK3 phosphorylation is degradation of the target protein. For example, phosphorylation of β-catenin, c-Myc, and c-Jun targets these proteins for polyubiquitination by specific E3 ligases and degradation by the proteasome (10, 22, 50, 51). We show here that phosphorylated forms of MafA in β cells accumulate in response to treatment with the proteasome inhibitors MG132 and lactacystin, which suggests that phosphorylated MafA is also degraded by the proteasomal pathway.

MafA has been identified as a trans-acting factor that binds to the glucose-responsive cis-element RIPE3b/C1, and glucose-dependent accumulation of MafA in β cells has been demonstrated previously (15, 17, 55). We and others showed previously that both mafA mRNA and MafA protein levels increase in response to glucose (11, 17, 49), which indicates that glucose regulates MafA at least at both the transcriptional and posttranscriptional levels. Based on our current observation that glucose-dependent regulation of MafA was almost lost when degradation was inhibited by MG132, we propose that glucose regulates MafA mainly by controlling protein stability. Additional studies are needed to determine the relative contributions of transcriptional and posttranslational regulation. Harmon et al. (13) showed that MafA protein loss by chronic exposure to oxidative stress is mediated at the level of protein stability. In the current study, we show that pharmacological inhibition of GSK3 in MIN6 cells results in an almost complete loss of MafA regulation by glucose and that phosphorylation-deficient mutants of MafA are very stable, even under low-glucose conditions. These results suggest that glucose-mediated regulation of MafA requires phosphorylation of MafA by GSK3 and that GSK3 and an as-yet-unidentified priming kinase are the major kinases in β cells that affect MafA stability. It is of note that the GSK3 inhibitor affected MafA abundance even in the presence of glucose (Fig. 6B). The reason for this is not clear, but it is possible that GSK3 inhibition may stimulate translation by affecting inhibitory phosphorylation of the translation initiation factor eIF2B (52).

We demonstrated that the overall phosphorylation status of MafA and the MafA-directed kinase activity in MIN6 cells are unchanged by glucose. Therefore, we propose that MafA is constitutively phosphorylated in β cells regardless of glucose concentration. We found that the half-life of MafA was prolonged by glucose in β cells. Exogenously expressed MafA protein was phosphorylated similarly in other types of cells, such as NIH 3T3, BHK21, and CHO cells, but MafA protein abundance was not regulated by glucose. This result is contradictory to a report by Kajihara et al. in which MafA abundance was regulated by glucose concentration even in NIH 3T3 cells (15). The reason for this discrepancy is not known. From our observations, we speculate that glucose-dependent degradation of MafA is cell type specific. The precise molecular mechanism(s) of regulation of the degradation of MafA in response to glucose is unknown. An as-yet-unidentified phospho-MafA-specific E3 ubiquitin ligase might be expressed specifically in β cells and may be regulated by glucose concentration. Recently, Vanderford et al. reported that glucose induces MafA protein expression via the hexoseamine biosynthetic pathway and O-linked glycosylation (49). The putative phospho-MafA-specific E3 ubiquitin ligase might be a target of this pathway. Our analysis of changes in the mobility of MafA during SDS-PAGE does not exclude the possibility that undetectable levels of phosphorylation or other modifications of MafA account for its selective degradation under low-glucose conditions in β cells. We have examined the O-glycosylation status of MafA by using an anti-O-GlcNAc antibody and have found no evidence for O glycosylation of MafA (unpublished observation).

MafA plays an important role in insulin gene transcription and β-cell function and is a promising therapeutic target for β-cell regeneration and differentiation. Therefore, identification of the molecular components of MafA degradation and their function is very important for clarifying the link between MafA and glucose sensing in β cells and for understanding β-cell biology and the pathogenesis of diabetes.