MafA Is a Glucose-regulated and Pancreatic β-Cell-specific Transcriptional Activator for the Insulin Gene*

The insulin gene is specifically expressed in β-cells of the Langerhans islets of the pancreas, and its transcription is regulated by the circulating glucose level. Previous reports have shown that an unidentified β-cell-specific nuclear factor binds to a conserved cis-regulatory element called RIPE3b and is critical for its glucose-regulated expression. Based on the sequence similarity of the RIPE3b element and the consensus binding sequence of the Maf family of basic leucine zipper transcription factors, we here identified mammalian homologue of avian MafA/L-Maf, an eye-specific member of the Maf family, as the RIPE3b-binding transcriptional activator. Reverse transcription-PCR analysis showed that mafA mRNA is detected only in the eyes and in pancreatic β-cells and not in α-cells. MafA protein as well as its mRNA is up-regulated by glucose, consistent with the glucose-regulated binding of MafA to the RIPE3b element in β-cell nuclear extracts. In transient luciferase assays, we also showed that expression of MafA greatly enhanced insulin promoter activity and that a dominant-negative form of MafA inhibited it. Therefore, MafA is a β-cell-specific and glucose-regulated transcriptional activator for insulin gene expression and thus may be involved in the function and development of β-cells as well as in the pathogenesis of diabetes.

Insulin is the only polypeptide hormone that critically regulates blood glucose levels and is produced exclusively by ␤-cells of the islets of Langerhans in the pancreas. The molecular mechanism of the ␤-cell-restricted expression of insulin has been extensively studied (for reviews, see Refs. 1 and 2), which led to the identification of various cis-regulatory elements on its promoter region. The most important among them are the conserved E1, A3, and RIPE3b/C1 elements (3)(4)(5)(6). The isletrestricted transcription factors, Beta2/NeuroD and Pdx1/IPF1/ STF1/IDX1/GSF/IUF1, which bind to the E1 and A3 elements, have also been isolated (7)(8)(9)(10). Gene disruption experiments in mice have revealed that both Beta2 and Pdx1 play critical roles in insulin gene expression as well as in islet development and function (11)(12)(13). Furthermore, mutations in beta2 and pdx1 genes are found in some population of patients with type 2 diabetes (14 -16).
The third regulatory element, RIPE3b, has also been shown to play a critical role in ␤-cell-specific insulin gene transcription as well as in its glucose-regulated expression (4,17). Previous studies have identified a ␤-cell-restricted RIPE3b-binding factor, called the RIPE3b1 activator, that appears in response to glucose in pancreatic ␤-cell nuclear extracts (18). However, despite growing evidence of its importance in the regulation of insulin gene transcription, it remains to be cloned.
We and others have previously determined the consensus DNA-binding sequences of the Maf family of transcription factors as the 13-and 14-base pair palindromic sequences TGCT-GACTCAGCA and TGCTGACGTCAGCA, which we have called Maf recognition elements (MAREs) 1 (19,20). By searching the GenBank TM data base, we were previously able to make a list of putative Maf-regulated genes that have MARE-related sequences in their regulatory regions, including the insulin gene (19). Later, some of them, for example, erythroid lineagespecific genes (␣-and ␤-globin and porphobilinogen deaminase) and eye-specific genes (opsin and crystalline genes) were actually found to be downstream targets of Maf family members (see below).
Maf family proteins belong to the basic leucine zipper family of transcription factors, and the originally identified member, v-Maf, is an oncoprotein encoded by the avian transforming retrovirus AS42 (21). To date, several maf-related genes have been identified in various species including human, mouse, rat, chicken, quail, frog, and zebrafish. The Maf family members are divided into two groups depending on their molecular sizes (i.e. the large Maf and small Maf proteins).
In both mammalian and chicken genomes, three small maf genes, mafK, mafF, and mafG, have been identified (22,23). Their products lack a transactivator domain, but they activate transcription when they form heterodimers with members of CNC ("cap'n'collar"), another basic leucine zipper family (24). For example, by forming heterodimers with the erythroid-specific CNC member p45, they constitute the erythroid-specific transactivator NF-E2 and activate transcription of the ␣and ␤-globin and porphobilinogen deaminase genes (25,26).
In contrast to the small Mafs, the large Maf proteins contain a transactivator domain in their amino terminus and activate transcription as homodimers (27,28). In chicken and quail, the c-maf, mafB, and mafA/L-maf genes have been identified as large maf members (28 -31). In mammals, c-maf, mafB, and nrl have been identified (32)(33)(34), but a homologue of mafA/L-maf has not yet been cloned. Nrl shows retina-specific expression and regulates opsin gene expression and retinal development (33,35,36); its mutation causes retinal degeneration in humans and mice (37,38). Chicken MafA/L-Maf, on the other hand, is exclusively expressed in lens and plays a critical role in the regulation of crystalline genes and lens development (29). As we have mentioned previously, insulin genes of various species, such as human, mouse, and rat, contain conserved MARE-related sequence in their promoters (19), and we have noticed that the MARE sequence overlaps with that of the RIPE3b element. In this report, we show that the RIPE3b1 factor contains a previously unidentified mammalian member of the large Maf family. Through GenBank TM data base searches and PCR amplification, we have isolated human and mouse homologues of the mafA gene and demonstrated that MafA is the RIPE3b1 factor.

EXPERIMENTAL PROCEDURES
Preparation of Nuclear Extracts-MIN6 cells (39) were a generous gift from Dr. Jun-ichi Miyazaki (Osaka University). ␤TC6 and ␣TC1 clone 9 (hereafter ␣TC1) cells were purchased from American Tissue Culture Collection. MIN6 and ␤TC6 cells were grown in Dulbecco's modified Eagle's medium supplemented with 15% fetal calf serum, and ␣TC1 cells were grown in F12K medium supplemented with 10% fetal calf serum. When controlling the glucose concentration, the medium was changed to Dulbecco's modified Eagle's medium containing various concentrations of glucose supplemented with 10% fetal calf serum. Nuclear extracts were prepared as described by Schreiber et al. (40).
Gel Mobility Shift Analysis-For the gel mobility shift analysis, nuclear extracts were incubated in DNA-binding buffer (final concentration 10 mM HEPES (pH 7.9), 100 mM NaCl, 2 mM EDTA, 8 mM dithiothreitol, 5% glycerol) with 1 g of poly(dI-dC)(dI-dC) on ice for 15 min. Then 0.5 ng of 32 P-labeled probe was added, and the extracts were further incubated at room temperature for 15 min. The reaction mixture was subjected to electrophoresis in a 0.5ϫ Tris borate-EDTA buffer plus 6% polyacrylamide gel at 4°C. The oligonucleotide probe containing the human RIPE3b element was made by annealing two 32-mer oligonucleotides, 5Ј-GATCCGGAAATTGCAGCCTCAGCCCCCAGCC-A-3Ј and 3Ј-GATCTGGCTGGGGGCTGAGGCTGCAATTTCCG-3Ј.
The core sequences of the mutated 32-mer RIPE3b-oligonucleotides used as competitors are shown in Fig. 2B. The complete sequences of MARE-related oligonucleotides were described in Ref. 19.
Anti-v-Maf serum was previously raised against a nearly full-length recombinant chicken v-Maf protein (41). Anti Cloning of Human and Mouse mafA-By searching the GenBank TM data base, we found a human genome contig (NT 023684) on chromosome 8q24 and several human expressed sequence tags that were most homologous to avian mafA. From the nucleotide sequence information, we designed primers (5Ј-agaacggTCCCGGGCGATGGCCGCGGAG-3Ј and 5Ј-agaacggcGTCCGGCGCCTACAGGAAGAAG-3Ј) and cloned the entire open reading frame of this gene by PCR. Human genomic DNA was used as a template because the chicken and quail mafA genes contain no introns. The same primers were used for cloning the mouse mafA homologue from mouse genomic DNA (Balb/c strain) by PCR. The PCR products were cloned into a pGEM-T-easy vector (Promega) to generate pGEM-T-easy/h-mafA and pGEM-T-easy/m-mafA.
Northern blot analysis was performed as described (21). The mafAspecific probe (pSae-hA) was constructed by deleting the BssHII-PmlI fragment from pGEM-T-easy/h-mafA to remove the GC-rich and CAC repeat region. To obtain the insulin probe, insulin cDNA was cloned from MIN6 total RNA by RT-PCR using the specific primers 5Ј-AGG-AATTCAACATGGCCCTGTTGGTGC-3Ј and 5Ј-CCTGGTGTTTT-ATCACAAGCTTCATAC-3Ј.
Luciferase Assay-To construct the reporter plasmid h-ins-p-luc, the human insulin promoter region was amplified by PCR using primers (5Ј-caggtaCCCCGCCCTGCAGCCTCCAGCTC-3Ј and 5Ј-agaagcTTCT-GATGCAGCCTGTCCTGGA-3Ј) from human genomic DNA. The fragment was inserted into the KpnI-HindIII site of the pGL2-basic plasmid (Promega) after digesting with KpnI and HindIII. Mutations were introduced into the RIPE3b/MARE element by site-directed overhang extension PCR mutagenesis (45) using the primers 5Ј-TGCAGCCgac-taCCCCAGCCATCTGCC-3Ј and 5Ј-ATGGCTGGGGtagtcGGCTGCAA-TTTC-3Ј. pG4ϫ5/TATA/luc plasmid that contains five copies of the Gal4-binding sequence was a gift from Dr. Kazuhiko Igarashi (Hiroshima University).
NIH3T3 cells grown in a 35-mm dish were transfected with a total of 1.5 g of plasmids (0.4 g of luciferase plasmid, a total of 1.0 g of expression plasmids, and 0.1 g of pEF-Rluc (48)) using 10 l of Polyfect reagent (Qiagen). Cells were harvested 24 h after transfection. MIN6 cells grown in a 35-mm dish were transfected with a total of 1.0 g of plasmids (0.3 g of luciferase plasmid, 0.6 g of expression plasmid, and 0.1 g of pEF-Rluc) using 8 l of LipofectAMINE and 6 l of PLUS reagent (Invitrogen). Twenty-four hours after transfection, the medium was changed to Dulbecco's modified Eagle's medium containing 10% fetal calf serum and 0 or 20 mM of glucose, and the cells were incubated for another 24 h. The firefly and Renilla luciferase activities were measured using the Dual Luciferase Assay System (Promega).

Similarity of the DNA Binding Specificities of RIPE3b1 and
Maf-As we have previously described (19), the conserved RIPE3b element in the promoter region of the insulin genes of human, mouse, and rat has a nucleotide sequence that is very similar to the consensus MARE sequence (Fig. 1A). This finding prompted us to examine whether RIPE3b-binding complexes in pancreatic ␤-cells contain Maf family members.
To explore this possibility, we prepared nuclear extracts from insulin-producing mouse ␤-cell lines, MIN6 and ␤TC6, grown in the presence or absence of 20 mM glucose, and subjected them to gel mobility shift analysis using an oligonucleotide probe containing the human RIPE3b element. As shown in Fig. 1B (lanes 1-4), a DNA-protein complex that appeared in the presence of glucose was detected both in MIN6 and ␤TC6 nuclear extracts. In contrast, we did not detect this complex in nuclear extracts prepared from the non-␤-cell lines ␣TC1 (␣cell) (lane 5), NIH3T3 (embryonic fibroblast) (see Fig. 2C), C2C12 (myoblast), and RAW264 (monocyte) (data not shown). Therefore, this ␤-cell-specific complex seemed to be identical to the previously reported RIPE3b1 complex (49,50).
We next tested the DNA binding specificity of this complex. As shown in Fig. 1C, a 100-fold molar excess of unlabeled human and mouse/rat RIPE3b-containing oligonucleotides specifically competed for the binding of this factor to the probe (lanes 1-3). The DNA binding specificity of the complex, as determined by the addition of excess cold oligonucleotides bearing various nucleotide substitutions (mut-A to mut-D) (lanes 4 -7), was very similar to what has been previously reported (49,50). These results again confirm the identity of this complex as the previously reported RIPE3b1 complex. Furthermore, the complex formation was inhibited by the addition of oligonucleotides containing the consensus MARE sequence (Fig. 1C, #1, lane 8) or a related sequence that can bind to Maf family members (#11 and #7, lanes 9 and 10) but not of oligonucleotides that do not bind to Maf (#17 and #23, lanes 11 and  12). Therefore, the RIPE3b1 factor showed similar DNA binding specificity to the Maf family of transcription factors (19).
Identification of the RIPE3b1 Factor as MafA-We then tested whether the RIPE3b1 complex contains Maf family members using a gel mobility shift analysis with a series of anti-Maf antibodies. As shown in Fig. 2A, the RIPE3b1 complex was supershifted by the addition of antisera that broadly react with large Maf members (anti-v-Maf and anti-c-Maf (M-153), lanes 1-4), whereas anti-small Maf antisera (anti-MafK and anti-MafK/F/G) did not react with the RIPE3b1 complex ( lanes  8 and 9). The addition of antisera that broadly react with Fos and Jun family members, possible heterodimeric partners for Maf family proteins (19,20,22,28), had no effect (lanes 10 and 11). These results clearly indicate that RIPE3b1 does contain a large Maf. However, none of antisera that specifically react with c-Maf (anti-c-Maf N-15), MafB, or Nrl affected the RIPE3b1 complex (lanes 5-7). These results strongly suggest that the RIPE3b1 complex contains an unidentified member of the large Maf family that reacts with anti-v-Maf and anti-c-Maf (M-153) sera.
Among the large maf family members identified in vertebrates to date, c-maf, mafB, and nrl genes have been isolated in mammals, but a homologue of avian mafA has not yet been identified. By searching the GenBank TM data base, we found a human genome contig on chromosome 8q24 and several human expressed sequence tags that were most homologous to avian mafA. From this nucleotide sequence information, we cloned the entire open reading frame of this gene and its mouse homologue (human and mouse mafA genes) by PCR. From the deduced amino acid sequence of mouse MafA, we raised a MafA-specific antiserum by immunizing a rabbit with a MafAderived peptide and added it into the gel mobility shift analysis. As shown in Fig. 2B, the RIPE3b1 complex was supershifted by the addition of the anti-MafA serum. We also confirmed that recombinant human and mouse MafA reacted with anti-v-Maf and anti-c-Maf (M-153) sera but not with antic-Maf (N-15), MafB, or Nrl sera (data not shown). Furthermore, transfection of an MafA expression vector into NIH3T3 cells resulted in the appearance of a DNA-protein complex of very similar mobility to the RIPE3b1 complex (Fig. 2C). These results clearly indicate that MafA is a component of the RIPE3b1 factor.

Structure of Human and Mouse
MafA-Nucleotide sequence analyses of human and mouse mafA (GenBank TM accession numbers AB086960 and AB086961) have revealed that they encode proteins of 351 (human) and 359 (mouse) amino acids that exhibit the highest homology to avian MafA (Fig. 3, A and  B) (29,30). As shown in Fig. 3B, the carboxyl-terminal basic domain and leucine zipper of human and mouse MafA that should serve as the DNA-binding and the dimerization domains, respectively, are completely conserved and are highly similar to other Maf family members. They also contain an amino-terminal acidic amino acid/serine/threonine/proline-rich domain, which may serve as a transcriptional activation domain (27). The middle part of MafA is relatively divergent but contains clusters of glycine and histidine as do all of the other large Maf members except Nrl.
We determined the chromosomal location of the human mafA gene by fluorescent in situ hybridization. Specific hybridization signals were observed on chromosome 8q24.3, and no other hybridization sites were detected (Fig. 4). This result is consistent with the map position (8q24) of the mafA-containing human genome contig (NT 023684).
Restricted Expression of mafA in Eye and Pancreatic ␤-Cells-We next prepared total RNA from MIN6, ␤TC6, and ␣TC1 cells and examined the expression of mafA mRNA by RT-PCR analysis. As is clearly shown in Fig. 5A, mafA mRNA expression was detected in two insulin-producing ␤-cell lines, MIN6 and ␤TC6, but not in the glucagon-producing ␣TC1 cell line. Moreover, we could not detect mRNAs for retina-specific nrl and for c-maf and mafB in MIN6 and ␤TC6 cells (data not shown), although c-maf and mafB are known to be expressed in a wide variety of tissues (28). Therefore, among the large Maf family members, insulin-producing ␤-cells selectively express mafA.
We also examined tissue distribution of mafA mRNA expression in adult mouse by RT-PCR. As shown in Fig. 5B, we could not detect mafA mRNA in any tissues we have examined except the eye. Expression of mafA in the eye makes sense in that avian mafA/L-maf is specifically expressed in the lens (29,30).
Although mafA is expressed in ␤-cell lines as shown above, we could not detect mafA mRNA in the pancreas, probably because of the very low abundance of ␤-cells in whole pancreas or, alternatively, because of degradation of the RNA source.
Glucose-regulated Expression of MafA Protein and mafA mRNA-In order to further confirm that MafA is the RIPE3b1 factor, we examined the expression of MafA protein in ␤and ␣-cells. Nuclear extracts prepared from MIN6, ␤TC6, and ␣TC1 cells grown in the presence or absence of glucose were analyzed by Western blotting using anti-MafA antiserum. As shown in Fig. 6A, a protein of 48 kDa was specifically detected in MIN6 and ␤TC6 nuclear extracts (lanes 2 and 4). A protein of this molecular weight was also detected using anti-v-Maf and antic-Maf (M-153) antisera (see Fig. 6B and data not shown), indicating that the 48-kDa protein is MafA. MafA protein was barely detectable when MIN6 and ␤TC6 cells were grown in the absence of glucose (lanes 1 and 3), and it was not present in an ␣TC1 nuclear extract even when the cells were grown in the presence of glucose (lane 5). These expression profiles correlated quite well with the appearance of the RIPE3b1 DNAbinding complex (see Fig. 1B), again supporting the idea that MafA is the RIPE3b1 factor. Furthermore, we found that the expression level of MafA protein (Fig. 6B, top panel) and mafA mRNA (second panel) in MIN6 cells was dependent on glucose concentration, which correlated well with the insulin mRNA level (third panel). Induction of MafA protein expression by glucose was also evident with immunofluorescent staining of MafA in MIN6 cells (Fig. 7) and in ␤TC6 cells (data not shown). MafA protein was FIG. 4. Chromosome mapping of human mafA gene. The biotinylated DNA probe for the human mafA gene was hybridized to chromosomes with replication bands prepared from normal donors. After hybridization and washing, hybridization signals were amplified using rabbit anti-biotin (Enzo) and fluorescein-labeled goat anti-rabbit IgG (Enzo). The chromosomes were counterstained with propidium iodide.
FIG . 5. mRNA expression of mafA. A, RT-PCR analyses. Total RNA (1 g, bottom panel) prepared from MIN6, ␤TC6, and ␣TC1 cells was subjected to RT-PCR analysis using specific primers for mafA, insulin, and glucagon. The aliquots of the amplified fragments were separated by agarose gel electrophoresis and were visualized by ethidium bromide staining. B, tissue distribution of mafA mRNA in mice. One microgram of total RNA (lower panel) isolated from various tissues of 7-week-old male mice was analyzed by RT-PCR using mafAspecific primers (upper panel). detected in the nucleus when these cells were grown in the presence of glucose.
Activation of Insulin Promoter by MafA-We next investigated the effect of MafA on insulin gene promoter activity. A luciferase reporter plasmid containing the human insulin promoter was transfected into NIH3T3 cells together with increasing amounts of the MafA expression plasmid. The insulin promoter activity was very low in NIH3T3 cells as compared with the activity of pG4ϫ5/TATA/luc that contains five copies of the Gal4-binding sequence and TATA-box (Fig. 8A). Co-expression of Pdx1 or Beta2, which are expressed in ␤-cells and have been shown to bind and activate the insulin promoter, resulted in dose-dependent activation. When MafA was expressed, the insulin promoter activity was greatly enhanced in a dose-dependent manner, indicating that MafA is an efficient transcriptional activator of the insulin promoter.
When the insulin promoter reporter plasmid was transfected into MIN6 cells, its promoter activity was significantly high relative to pG4ϫ5/TATA/luc, and was regulated by glucose (Fig. 8B), reflecting endogenous insulin gene transcription. Mutation of the RIPE3b/MARE element resulted in a decrease of both the basal and glucose-induced activities of the promoter, which has been reported previously (49,50), and confirmed the importance of the RIPE3b/MARE in basal and glucose-regulated insulin gene expression. In order to evaluate the possible role of endogenous MafA protein on insulin promoter activity, we constructed a dominant-negative mutant of MafA (HA-SID-ND-MafA), whose amino-terminal putative activation domain was replaced with the heterologous transcriptional suppressor domain of Mxi1 (SID). SID is known to actively suppress transcription by interacting with Sin3 protein and by recruiting the histone deacetylase complex (46 -48). As shown in Fig. 8B, expression of this mutant MafA protein in MIN6 cells resulted in a significant decrease in insulin promoter activity both in the presence and absence of glucose in the culture medium, suggesting an essential role of MafA in the insulin gene transcription. DISCUSSION In this report, we first demonstrated that the RIPE3b1 activator is related to a large Maf family member in its DNA binding specificity and its reactivity to a series of anti-Maf antisera. We then molecularly cloned human and mouse homologues of the mafA gene and showed that MafA is the RIPE3b1 activator. Anti-MafA-specific antiserum reacted with the RIPE3b1 activator in gel mobility shift analyses, and mafA mRNA was detected only in insulin-producing ␤-cell lines and the eye, but not in other cell lines and tissues as far as we have examined. Consistent with the previously suggested role of RIPE3b1 factor in glucose-regulated insulin gene expression, glucose induced the expression of mafA mRNA and protein in ␤-cell lines.
Previous reports have demonstrated that the RIPE3b1 activator is a protein of 37-49 kDa (49,50), which is consistent with our estimation of the molecular size of the MafA protein (48 kDa) by Western blotting analysis. During preparation of this manuscript, Olbrot et al. (51) reported that they purified a 47-kDa RIPE3b1-binding protein from MIN6 cells and identified it as MafA, which is again consistent with our observations. Olbrot et al. (51) reconstituted the RIPE3b1 factor from the purified 47-kDa protein alone in gel mobility shift assay. Therefore, despite the fact that the MafA protein possibly forms heterodimers with other basic leucine zipper proteins, including the other large Maf proteins, we believe that the RIPE3b1 factor is a homodimer of MafA. This idea is supported by our observations that the DNA binding specificity of the RIPE3b1 factor is very similar to that of the Maf homodimer (Fig. 1C) and that recombinant MafA protein had very similar mobility to the RIPE3b1 complex on gel shift analysis (Fig. 2C). Moreover, none of mRNAs for c-maf, mafB, and nrl were de- tected in MIN6 and ␤TC6 cells by RT-PCR analysis (data not shown).
Olbrot et al. (51) detected mafA mRNA in mouse islet and insulinoma cell lines but not in glucagon-producing ␣TC1 cells, which is consistent with our results (Fig. 5). Recently, Planque et al. (58) have shown that mafA mRNA is expressed in pancreas and lens of quail embryo, which suggests the conserved expression profiles and roles of MafA in birds and mammals. Planque et al. (58) have reported that c-maf mRNA is detected in ␣TC cells and that Pax6 and the large Maf family members synergistically activate the glucagon gene transcription. Therefore, interestingly, distinct members of the large Maf family (MafA and c-Maf) are expressed in distinct types of pancreatic islet cells (␤-and ␣-cells) and regulate insulin and glucagon gene transcription, respectively. It is worth examining whether the large Maf proteins are also expressed in islet ␦and ␥-cells and whether they are involved in transcription of somatostatin and pancreatic polypeptide genes.
Previous studies have identified a set of transcription factors that specifically bind to the cis-regulatory elements on the insulin promoter, including Beta2 and Pdx1, as well as Isl1 and Lmx1 (7-10, 54, 55). It has also been shown that these transcription factors synergistically activate the insulin promoter (54). Furthermore, some of them have been shown to physically interact (56,57). In contrast, MafA alone strongly activated the insulin gene promoter in NIH3T3 cells (Fig. 8A), and we did not see additional enhancement of MafA transcriptional activity with co-expression of Beta2 or Pdx1 as far as we have examined. 2 Together with our (Fig. 8B) and previous (3)(4)(5)(6) findings that mutation of the RIPE3b element greatly reduces the basal and glucose-induced insulin promoter activity in ␤-cells, MafA seems to be one of the most important transcriptional activators for efficient insulin gene expression.
Most of the previously identified transcription factors required for insulin gene expression have also been shown to be involved in development of ␤-cells as well as the other islet cells. For example, targeted deletion of beta2 in mice resulted in a decreased number of ␤-cells and the failure of mature islets to develop (11), and pdx1 gene disruption caused ablation of the pancreas (12,13). MafA may also play important roles in ␤-cell development and maintenance of ␤-cell function as well as in expression of insulin. Generation of mafA knockout mice should elucidate the role of mafA in the development of ␤-cells and its functional relationship with the other islet-specific transcription factors. Furthermore, taking into account that mutations in beta2 and pdx1 genes are associated with type 2 diabetes mellitus in humans (14 -16), a defect in MafA function may also cause diabetes in humans as well as in model animals. However, as far as we know, the chromosome location of the human mafA gene (8q24.3) is not assigned as a susceptibility locus of both type 1 and type 2 diabetes.
It has been suggested that tyrosine phosphorylation of the RIPE3b1 activator stimulates its DNA binding activity (52). Although avian MafA has been shown to be phosphorylated on serine residues at the amino-terminal transactivator domain (53), there is no evidence of tyrosine phosphorylation of MafA to date. It is worth examining whether the MafA protein is tyrosine-phosphorylated and what effect that would have on its DNA binding activity, and such post-translational regulation of MafA remains to be examined. As we have shown here, expression of MafA is regulated by glucose at least at the level of transcription (Fig. 6), which might in turn be responsible for glucose-regulated insulin gene expression. However, the glucose concentration may also regulate MafA activity by other mechanisms, such as post-transcriptional and/or post-translational modifications, which, if any, may help to elucidate the molecular mechanism of insulin transcription activation by glucose.