The Islet β Cell-enriched RIPE3b1/Maf Transcription Factor Regulates pdx-1 Expression*

Pancreatic duodenal homeobox factor-1, PDX-1, is required for pancreas development, islet cell differentiation, and the maintenance of β cell function. Selective expression in the pancreas appears to be principally regulated by Area II, one of four conserved regulatory sequence domains found within the 5′-flanking region of thepdx-1 gene. Detailed mutagenesis studies have identified potential sites of interaction for both positive- and negative-acting factors within the conserved sequence blocks of Area II. The islet β cell-enriched RIPE3b1 transcription factor, the activator of insulin C1 element-driven expression, was shown here to also stimulate Area II by binding to sequence blocks 4 and 5 (termed B4/5). Accordingly, B4/5 DNA-binding protein's molecular mass (i.e. 46 kDa), binding specificity, and islet β cell-enriched distribution were identical to RIPE3b1. Area II-mediated activation was also unaffected upon replacing B4/5 with the insulin C1/RIPE3b1 binding site. In addition, the chromatin immunoprecipitation assay showed that the Area II region of the endogenous pdx-1 gene was precipitated by an antiserum that recognizes the large Maf protein that comprises the RIPE3b1 transcription factor. These results strongly suggest that RIPE3b1/Maf has an important role in generating and maintaining physiologically functional β cells.

Pancreatic duodenal homeobox factor-1, PDX-1, is required for pancreas development, islet cell differentiation, and the maintenance of ␤ cell function. Selective expression in the pancreas appears to be principally regulated by Area II, one of four conserved regulatory sequence domains found within the 5-flanking region of the pdx-1 gene. Detailed mutagenesis studies have identified potential sites of interaction for both positive-and negative-acting factors within the conserved sequence blocks of Area II. The islet ␤ cell-enriched RIPE3b1 transcription factor, the activator of insulin C1 elementdriven expression, was shown here to also stimulate Area II by binding to sequence blocks 4 and 5 (termed B4/5). Accordingly, B4/5 DNA-binding protein's molecular mass (i.e. 46 kDa), binding specificity, and islet ␤ cell-enriched distribution were identical to RIPE3b1. Area II-mediated activation was also unaffected upon replacing B4/5 with the insulin C1/RIPE3b1 binding site. In addition, the chromatin immunoprecipitation assay showed that the Area II region of the endogenous pdx-1 gene was precipitated by an antiserum that recognizes the large Maf protein that comprises the RIPE3b1 transcription factor. These results strongly suggest that RIPE3b1/Maf has an important role in generating and maintaining physiologically functional ␤ cells.
Targeting of the pancreatic duodenal homeobox factor-1 (pdx-1) 1 gene in mice has established that expression in a common progenitor cell population is essential for the development of both the endocrine and exocrine compartments of the pancreas. PDX-1 acts by stimulating proliferation, branching, and differentiation of the pancreatic epithelium (1-3). In contrast, all other characterized islet endocrine-(e.g. PAX6 (4,5), Ngn3 (6), BETA2 (7), and exocrine (PTF1-p48 (8,9)-enriched transcription factors act downstream of PDX-1 and are principally involved in islet or exocrine cell differentiation. Selective elimination of PDX-1 in mouse ␤ cells in vivo also results in a reduction in both insulin secretion and islet ␤ cell numbers (1). These animals become glucose intolerant and diabetic, largely because of their inability to synthesize appropriate amounts of PDX-1-regulated gene products that are involved in maintaining glucose homeostasis (1) (e.g. insulin (10,11), GLUT2 (12), and glucokinase (13)). Moreover, mutations in pdx-1 cause pancreatic agenesis (2, 3) and a form of maturity onset diabetes of the young in humans (14,15). These data have established an essential role for PDX-1 in islet ␤ cell development and function.
The recent success in reversing type 1 diabetes by islet transplantation has led to renewed optimism for this form of treatment (16). However, the availability of human islets is limited and will never be sufficient to treat all patients. Because islet-enriched transcription factors are essential for islet cell development, information valuable for generating transplantable cells will likely be gained by understanding how their expression is regulated. Therefore, efforts have recently focused on characterizing the transcriptional control regions in genes necessary for islet cell formation, including pdx-1 (17)(18)(19)(20)(21)(22)(23), BETA2 (24), pax6 (25,26), pax4 (27), and ngn3 (28). In specific regards to pdx-1, expression will likely be mediated by factors involved in both the differentiation and maintenance of functional ␤ cells.
Experiments performed in transgenic animals have established that ␤ cell-selective expression of the pdx-1 gene is regulated by sequences 5Ј to the transcription start site (18,23). Control also appears to be largely mediated by those conserved between the vertebrate pdx-1 genes (19,20,22). Thus, ␤ cell-specific reporter gene expression was driven in transfection assays by areas of sequence identity shared between the chicken, mouse, and human genes (i.e. Area I, Ϫ2839/Ϫ2520 base pair (bp) (19), Area III, Ϫ1879/Ϫ1799 bp (19), Area IV, Ϫ6047/Ϫ6529 bp), 2 or only the mouse and human genes (Area II, Ϫ2141/Ϫ1961 bp (19,22)). In contrast, the 5Ј-non-conserved sequences were inactive (18,23). A pdx-1 gene fragment spanning Areas I and II also directed transgene expression to islet ␤ cells in vivo (termed PstBst, Ϫ2917/Ϫ1918 bp (18,23)), although only Area II, and not Areas I (22) or III (18), functioned independently in these in vivo assays. Collectively, these data strongly suggest that Area II represents the core of the mammalian pdx-1 transcription control region.
Mutational analysis of 17 conserved sequence blocks within Area II revealed sites for both positive-and negative-acting regulatory factors (22). Gel shift analysis performed on the activating B8 (Ϫ2068/Ϫ2060-bp) and B14 (Ϫ2006/Ϫ1996-bp) elements demonstrated specific binding to Pax6 and Foxa2 (formerly termed HNF3␤), respectively. Mutation of the Foxa2 binding site in Area II limited expression of the PstBst transgene to a subset of the islet ␤ cells in vivo (22). In addition, conditional deletion of Foxa2 specifically from ␤ cells decreased pdx-1 mRNA and protein expression in mice (29). The ability of Pax6 and Foxa2 to bind within Area II of the endogenous pdx-1 gene also strongly supports a direct role in mediating transcription in ␤ cells (22).
In the present study, we show that the B4/5 activator of Area II is RIPE3b1/Maf, a 46-kDa islet ␤ cell-enriched protein(s) essential for both cell type-specific (30,31) and glucose-inducible (32,33) transcription of the insulin gene. Moreover, an antiserum that recognizes the recently isolated large Maf protein(s) of the insulin C1/RIPE3b1 activator revealed binding to Area II of the endogenous pdx-1 gene in ␤ cells. We propose that RIPE3b1/Maf is required for transcription of genes critical to ␤ cell function.

MATERIALS AND METHODS
Transfection Constructs-The Area II and PstBst reporter constructs were made using human (Ϫ2141/Ϫ1890-bp) and mouse (Pst/Ϫ2917bp: Bst/Ϫ1890-bp) pdx-1 sequences (23), which were cloned directly upstream of the herpes simplex thymidine kinase (TK) promoter in a chloramphenicol acetyltransferase (CAT) expression vector, pTK(An) (34). The block transversion and insulin C1 (InsC1) substitution mutants in B4/5 were constructed in Area II:pTK and PstBst:pTK using the QuikChange mutagenesis kit (Stratagene). Each construct was determined to be correct by DNA sequencing.
Cell Transfections-Monolayer cultures of pancreatic islet ␤ (␤TC-3, HIT-T15, and Min6) and non-␤ (NIH3T3) cell lines were maintained as described previously (35). The LipofectAMINE reagent (Invitrogen) was used to introduce 1 g each of Area II:pTK or PstBst:pTK and 0.5 g of pRSVLUC. The activity from the Rous sarcoma virus enhancer-driven luciferase plasmid served as an internal transfection control for the pdx-1:pTK constructs. Luciferase (36) and CAT (37) enzymatic assays were performed 40 -48 h after transfection. Each experiment was carried out more than three times with at least two independently isolated DNA preparations.
Electrophoretic Mobility Shift Assays-Double-stranded Area II block 4 (B4, agcttTCTTTTTGCAAAGCACAGCAt), B5 (agcttAAAGCA-CAGCAAAAATATTAt), and B4/5 (agcttCTTTTTGCAAAGCACAGCA-AAAAt) sequences, in which the lowercase lettering corresponds to linker sequences, were excised from pBluescriptKS2ϩ and Klenowlabeled with [␣ 32 P]dATP. The InsC1 probe spans nucleotides Ϫ126 to Ϫ101 of the rat insulin II gene and was labeled as described (38). Nuclear extracts were prepared as described previously (39). Binding reactions (20 l total volume) were conducted with 5-10 g of extract protein and labeled probe (8 ϫ 10 4 cpm) in binding buffer containing 10 mM Tris-HCl, pH 7.4, 100 mM NaCl, 2 mM dithiothreitol, 1 mM EDTA, 10% glycerol, and 1 g of poly(dGdC) (final concentrations). The conditions for the competition analyses were the same, except that excess (see figures for amounts) of the specific competitor DNA was included in the mixture prior to the addition of probe. Anti-c-Maf antiserum (10 g, M-153, Santa Cruz Biotechnology) was added to binding reactions 10 min prior to addition of the probe for supershift analysis. This antiserum, referred to in the text as ␣c-Maf M-153, was made to an N-terminal region of c-Maf that is common to the other large Mafs and cross-reacts with each (i.e. MafA, NRL, and MafB). The samples were resolved on a 6% nondenaturing polyacrylamide gel (acrylamide:bisacrylamide ratio 29:1) and run in TGE buffer (50 mM Tris, 380 mM glycine, 2 mM EDTA, pH 8.5). The gel was dried and subjected to autoradiography.
SDS-PAGE Fractionation-␤TC-3 and Min6 nuclear extracts (30 g) were separated on a 10% SDS-polyacrylamide gel (SDS-PAGE) and then electro-transferred onto an Immobilon polyvinylidene difluoride (PVDF) membrane (Millipore). The extract lanes were cut horizontally FIG. 1. The conserved B4 and B5 sequence blocks regulate PstBst-mediated activation in ␤ cells. A, a schematic diagram illustrating the position of the Ϫ2560/Ϫ1880-bp PstBst region in the mouse pdx-1 gene. The location of the Area I and Area II control regions and the characterized conserved block mutants within Area II are also shown. The Foxa2 (19,23) and Pax6 (22) control elements were characterized previously. B, the normalized activity of the transfected mutant (MT) Area II:pTKCAT and PstBst:pTKCAT constructs is expressed as a percent activity of the wild type Area II and PstBst reporter Ϯ S.E. of the mean. PstBst:pTKCAT is ϳ2to 4-fold more active than Area II:pTKCAT in transfected ␤ cell lines. into 3-mm slices. The molecular mass range of each lane fraction was determined by comparison with colored Rainbow protein markers (Amersham Biosciences). The proteins from each fraction were eluted as previously described (38) and analyzed for B4/5 and InsC1 binding activity in electrophoretic mobility shift assays.
Anti-phosphotyrosine Immunoprecipitation-Immunoprecipitations using anti-Tyr(P) (4G10, Upstate Biotechnology, Lake Placid, NY) were performed as described previously (40). Briefly, SDS was added to a final concentration of 0.5% (w/v) to ␤TC-3 nuclear extract (100 g protein) in a buffer containing 10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 10% glycerol, 1 mM Na 3 VO 4 , and 2 mM dithiothreitol (final concentra-tions) and then heated to 65°C. After diluting the SDS to 0.05%, anti-Tyr(P) or control mouse IgG was added along with protein A-Sepharose beads. The washed beads were then resuspended in 1ϫ SDS-PAGE loading buffer and the immunoprecipitated proteins separated on a 10% SDS-polyacrylamide gel. After transfer to an Immobilon PVDF membrane, the 44 -47-kDa eluted proteins were assayed for B4/5 and InsC1 binding activity.
Immunohistochemistry-Pancreata from 6 -8-week-old mice were fixed 4 -5 h in 4% paraformaldehyde at 4°C, washed, dehydrated, embedded in paraffin, and then 5-m sections cut and mounted on glass slides. Double immunofluorescence was performed using guinea pig ␣-human insulin (Linco) and rabbit ␣-mouse c-Maf (␣c-Maf M-153) as primary antibodies at dilutions of 1:2000 and 1:100, respectively. Secondary antibodies were Cy3-or Cy5-labeled donkey anti-guinea pig and anti-rabbit IgG diluted to 1:500 (Jackson ImmunoResearch Laboratories). Fluorescent images were captured on a Zeiss LSM510 confocal microscope at an optical depth of 1 m, false colors were assigned, and the images merged in Photoshop 5 (Adobe). Immunoperoxidase staining was performed with Vectastain Elite kits (Vector Labs) and with 3,3Јdiaminobenzidine tetrahydrochloride substrate (Zymed Laboratories   FIG. 2. B4 and B5 comprise a single element that binds a ␤ cell-enriched factor. A, the B4, B5, and B4/5 probe sequences are shown. The conserved B4 and B5 block sequences are in bold. The mutant B4/5 competitor contains transversion mutations of the conserved bolded sequences. B, gel shift binding reactions were conducted with the B4, B5, or B4/5 probe using ␤TC-3 and Min6 nuclear extracts in the presence of a -fold molar excess of unlabeled wild type (WT) or mutant (MT) competitor to probe. The complexes labeled A and B are discussed under "Results." C, nuclear extracts from ␤ (␤TC-3, Ins-1, Min6, HIT-T15) and non-␤ (␣TC-6, RC2-E10, NCB20, MDCK, BHK, NIH-3T3, and H4IIE) cell lines as well as rat liver (rliver) were analyzed for B4/5 binding activity.
FIG. 3. The RIPE3b1 complex contains a protein(s) of ϳ46 kDa. Min6 nuclear extracts were electro-transferred onto an Immobilon PVDF membrane after SDS-PAGE. The proteins were eluted from membrane slices and assayed for B4/5 and InsC1 binding activity. Binding specificity was determined by competition with a 10-fold excess of unlabeled B4/5 or InsC1 (data not shown). Each fraction represents a different molecular mass range. The position of complex A detected in unfractionated Min6 nuclear extracts is indicated.

RESULTS
B4 and B5 Affect Area II Activity-Block mutations within conserved B2 (Ϫ2131/Ϫ2115 bp), B3 (Ϫ2110/Ϫ2102 bp), B4 (Ϫ2100/Ϫ2093 bp), and B5 (Ϫ2089/Ϫ2086 bp) reduce Area II: pTK activity in ␤ cell lines (Fig. 1B, HIT-T15, MIN6, and ␤TC-3) (22). To further examine the significance of these elements in Area II activation, each was mutated within the mouse pdx-1 'PstBst' region that spans Area I and Area II (Fig.  1A). In the context of the more active PstBst:pTK expression construct, the B4 and B5 mutants reduced activity to a greater extent than in Area II:pTK (Fig. 1B). Combining the B4 with B5 mutations in PstBst:pTK reduced activity further than either individual mutation (Fig. 1B). In contrast, the B2 and B3 mutants had less effect on PstBst:pTK activation (data not shown). The following experiments were designed to characterize the B4 and B5 activators in ␤ cells.
B4 and B5 Represent a Single cis-Element That Interacts with a ␤ Cell-enriched Protein(s)-To define the factors associated with B4-and B5-mediated regulation, gel shift experiments were performed with probes spanning B4, B5, and B4ϩB5 (B4/5), and ␤TC-3 or MIN6 cell nuclear extracts (Fig.  2). Identical results were obtained with MIN6 and ␤TC-3 cells, and they were used interchangeably in these analyses. Two common protein-DNA complexes were detected with the B4 and B4/5 probes (labeled as A and B in Fig. 2B), whereas no binding was found with B5 (data not shown). The binding affinity of B4 and B4/5 for these complexes was determined with the wild type and B4/5 double mutant site (B4/5 MT) competitors. As expected, both B4 and B4/5 reduced the levels of these complexes, although B4/5 was roughly 20-fold more effective (Fig. 2B). In contrast, B5 did not compete for binding (data not shown), whereas the B4/5 MT only competed away complex B, consistent with the conclusion that it is unrelated to activation (Fig. 2B). These results suggested that B4 and B5 define a single activator-binding site, which is regulated by the factor(s) found within the slower mobility complex A.
Complex A Contains an Approximately 46-kDa Protein(s)-To estimate the size of the protein(s) in complex A, Min6 nuclear extracts were separated by SDS-PAGE and transferred to a PVDF membrane that was cut into slices to represent distinct molecular masses. The separated proteins were eluted from the membrane slices, renatured, and tested for binding to the B4/5 probe. The binding specificity of fraction 8 was identical to complex A found in unfractionated Min6 extracts ( Fig. 3; data no shown). The molecular mass range of the proteins in fraction 8 was 44 -47 kDa. These results indicate that complex A is composed of one or more proteins of ϳ46 kDa.
The 46-kDa Complex A Protein(s) Corresponds to the InsC1 Activator, RIPE3b1-Because the RIPE3b1 protein(s) that binds to and activates the InsC1 control element has the same cell-restricted distribution (38) and molecular size (see Fig. 3 and Ref. 38), we compared the binding properties of B4/5 to InsC1 (Fig. 4). Both InsC1 and B4/5 competed effectively for complex A binding when either B4/5 or InsC1 were used as probes (Fig. 4B). In addition, RIPE3b1/complex A activity was affected in the same manner by B4/5 or InsC1 mutations that either modestly (e.g. InsC1mt1, Ref. 38) or profoundly (e.g. InsC1mt3, Ref. 38) (Fig. 1B, B4/5MT) affected activity. The resulting competition patterns were consistent with each element binding the same factor(s) (Fig. 4B).
RIPE3b1 binding activity is inhibited by the actions of a tyrosine phosphatase (40). To test whether complex A formation on B4/5 is also regulated in this manner, Min6 nuclear extracts were incubated in the presence or absence of CIAP and a general (sodium pyrophosphate, NaPPi) or phosphotyrosinespecific (sodium orthovanadate, Na 3 VO 4 ) phosphatase inhibitor. B4/5 and InsC1 binding activities were monitored in the treated extracts. The binding characteristics of complex A were affected in exactly the same manner with both probes (Fig. 5A). Complex A mobility was shifted upon incubating the ␤ cell nuclear extract at 30°C with both the B4/5 and InsC1 probes, presumably because of the actions of an endogenous tyrosine phosphatase (40). CIAP treatment reduced binding to each probe, an effect blocked by addition of NaPPi or Na 3 VO 4 (Fig.  5A). In addition, B4/5 bound specifically to the 46-kDa fraction immunoprecipitated from ␤TC-3 nuclear extracts with the anti-phosphotyrosine immunospecific monoclonal antibody, 4G10 FIG. 5. Complex A binding to B4/5 is sensitive to tyrosine dephosphorylation. A, B4/5 and InsC1 binding reactions with Min6 nuclear extracts were incubated at either 4 or 30°C, either alone or in the presence of CIAP, CIAP ϩ 10 mM Na 3 VO 4 , or CIAP ϩ 10 mM NaPPi. B, ␤TC-3 nuclear extract was immunoprecipitated with either the anti-phosphotyrosine antibody 4G10 or normal mouse IgG. The immunoprecipitated proteins (labeled 4G10 and IgG) and whole nuclear extract were then fractionated by SDS-PAGE and transferred to PVDF membranes (Immobilon). Protein fractions 1 (53.7-62.7 kDa), 2 (41.7-53.6 kDa), and 3 (29.9 -41.6 kDa) were eluted and used in B4/5 and InsC1 gel shift assays along with unfractionated ␤TC-3 nuclear extract. (Fig. 5B, compare 4G10 immunoprecipitate binding to InsC1 (40) and B4/5). Collectively, these results strongly suggest that RIPE3b1 binds to both the pdx-1 B4/5 and InsC1 elements.
InsC1 Can Substitute for B4/5 to Drive Area II Activation in ␤ Cells-Considering the interchangeability of B4/5 and InsC1 in gel shift assays, it was surprising to find only modest sequence identity between human (h) and mouse (m) B4/5 and mouse InsC1 (Fig. 6A). However, methylation interference assays over B4/5 suggested some similarity in contact nucleotides for RIPE3b1 with InsC1 (Ref. 30 and Fig. 6A; data not shown). Because of sequence dissimilarity between B4/5 and InsC1, we tested whether InsC1 could substitute for B4/5 in the context of the PstBst:pTK reporter. Replacement of B4/5 with InsC1 maintained the same high level of activation found for wild type PstBst in Min6 ␤ cells (Fig. 6B). Furthermore, mutants in B4/5 (B4mt, B5mt, B4/5mt) and InsC1 (mut3) that compromised complex A/RIPE3b1 binding also reduced PstBst activity only in Min6 cells. These data strongly suggest that the ␤ cell-enriched RIPE3b1 transcription factor activates the B4/5 control element in Area II.
A Large Maf Transcription Factor within ␤ Cell Nuclei Binds to Area II-The RIPE3b1 transcription factor was recently isolated and shown to be a member of the large Maf transcription factor family, most likely MafA (33,46). 3 To determine whether the B4/5 binding complex A contained a large Maf protein, Min6 nuclear extract was preincubated with a poly-clonal antiserum raised to N-terminal sequences of c-Maf shared with other members of the large Maf family. This c-Maf antiserum, termed ␣c-Maf M-153, cross-reacts with MafA, MafB, and NRL (46). 4 Complex A was completely supershifted by ␣c-Maf M-153, whereas IgG had no effect (Fig. 4C). These results strongly suggested that complex A contains the RIPE3b1/Maf protein. Immunohistochemical analysis performed with ␣c-Maf M-153 on adult mouse pancreas also showed that the large Maf protein(s) of the RIPE3b1/Complex A activator was nuclear and expressed almost exclusively in insulin-producing ␤ cells (Fig. 7).
To directly determine whether RIPE3b1/Maf binds within Area II of the endogenous pdx-1 gene, a chromatin immunoprecipitation assay was performed using formaldehyde crosslinked chromatin from ␤TC-3 cells. The cross-linked DNA was precipitated with the Maf antiserum and PCR-amplified with Area II and PEPCK promoter-specific primers. The Maf antibody was capable of immunoprecipitating Area II sequences, whereas the control IgG could not (Fig. 8). However, the Maf antiserum did not immunoprecipitate transcription control sequences from the PEPCK gene, which is not transcribed in ␤ cells. These results demonstrate that RIPE3b1/Maf occupies the Area II region of the pdx-1 gene in ␤ cells.

DISCUSSION
Area II, when compared with Area I and III, was the only pdx-1 control region capable of independently directing pancreatic ␤ cell-selective transgene expression (18,22). This property and the uniqueness of this domain within the human and mouse genes imply that Area II represents the core of the mammalian pdx-1 transcription unit. Mutagenesis of conserved sequence blocks within Area II revealed sites for both 3  positive-and negative-acting regulators, including the B8 and B14 control elements that bind the Pax6 (22) and Foxa2 (20,23) activators, respectively. However, the rather general distribution of these two developmental regulators suggests that a more ␤ cell-restricted factor(s) likely contributes to the expression pattern observed for Area II-driven reporters. The objective of this study was to determine whether the B2, B3, B4, and/or B5 activators had such properties. Our analysis revealed that the B4 and B5 mutants reduced expression driven by the mouse PstBst fragment spanning Areas I and II in all ␤ cell lines tested. The B2 and B3 mutants had a lesser affect and were not analyzed further. B4 and B5 were found to comprise a single regulatory element activated by RIPE3b1/Maf, a ␤ cell-enriched nuclear factor known for its important role in insulin gene transcription.
Mutation of either B4 or B5 reduced activation of the PstBstdriven fragment, and all of our subsequent analysis focused on characterizing their activator(s). Using B4, B5, or B4/5 in gel shift assays revealed that a specific, ␤ cell-enriched complex bound to these sequences (Fig. 2). The results of the transfection studies performed with B4, B5, and B4/5 mutants in PstBst were also consistent with this conclusion (Fig. 1). Because the size and cellular distribution of the specific B4/5 binding complex was similar to the InsC1/RIPE3b1 complex, the binding and functional properties of InsC1 were compared with B4/5. InsC1 not only specifically and effectively competed with B4/5 in gel shift assays but also was functionally indistinguishable in transfections performed with InsC1 substitution mutants in PstBst (Fig. 6). In addition, the apparent dependence upon tyrosine phosphorylation for RIPE3b1 binding to InsC1 (40) was also found with B4/5 (Fig. 5).
The RIPE3b1 activator was recently independently isolated from ␤TC-3 (data not shown) and HIT T-15 cells (46) using a biochemically based InsC1 affinity matrix chromatography strategy. MafA, a member of the large Maf family of basic leucine zipper proteins, was identified by mass spectrophotometric analysis of the purified fractions. The presence of Maf in the RIPE3b1 complex was further demonstrated using an antiserum that cross-reacts with N-terminal epitopes common to all the proteins in the large Maf family (Fig. 4C). These results formed the primary basis for concluding that MafA is the RIPE3b1 activator (46). This is likely to be correct for ␤ cell lines; however, we have used specific molecular and immunological reagents to show that other members of the large Maf family are expressed in the adult islet ␤ cells, specifically c-Maf and MafB (data not shown). Because MafA, c-Maf, and MafB are equally capable of binding and activating B4/5-like sequences in vitro (Ref. 48 and data not shown), these results suggest that several members of the large Maf family may control RIPE3b1-like-mediated transcription in the ␤ cell. This broader composition of the RIPE3b1/Maf activator is also consistent with the importance of the large Maf proteins in myeloid (49,50) and lens (51-53) cell differentiation.
Importantly, the large Maf-recognizing antiserum demonstrated that a protein(s) in this family binds to Area II control region sequences in intact ␤ cells (Fig. 8). In addition, this antiserum also supershifted the ␤ cell-enriched A/RIPE3b1 complex formed with the B4/5 probe (Fig. 4C) and immunohistochemically localized Maf protein in the nuclei of mouse islet ␤ cells (Fig. 7). We and others have also shown that the large Maf proteins can activate islet target gene expression in transfection assays (33,46). 5 As a consequence, we conclude that RIPE3b1/Maf binds to B4/5 and activates Area II-mediated transcription in ␤ cells. Furthermore, because only MafA was identified during the biochemical isolation and characterization of the RIPE3b1/InsC1 binding complex, it is most likely the principal large Maf activator in ␤ cells.
Collectively, the data presented here and elsewhere demonstrate that insulin and pdx-1 are bona fide transcriptional targets for RIPE3b1/Maf control. Inspection of the consensus large Maf binding motif (TGC (N) 6 -7 GCA, Ref. 48) also revealed a greater similarity of B4/5 to InsC1 than was initially apparent (Fig. 6). Interestingly, the presence of this consensus binding site in the transcription control region of other selectively expressed genes suggests a general, but significant, role in controlling ␤ cell-specific expression (e.g. islet-specific glucose-6-phosphatase catalytic subunit related protein (54), Ϫ183/Ϫ165; islet amyloid polypeptide (55)(56)(57), Ϫ540/Ϫ528 bp). It is likely that the ␤-cell-enriched RIPE3b1/Maf activator acts in concert with more widely distributed factors (e.g. Pax6 and Foxa2) to mediate selective expression.
The functional cooperativity observed between the large Mafs and Pax6 in activating lens gene expression may also be of relevance to both insulin and pdx-1 in the ␤ cell. Thus, Pax6 directly regulates c-Maf gene expression during lens development (58,59), and together they function cooperatively to stimulate crystalline gene expression in differentiating lens fibers (58 -61). It is also intriguing to consider that the loss in ␤ cell function upon eliminating insulin receptor (62) or insulin receptor substrate-2 (63, 64) signaling may be mediated, at least in part, by directly effecting MafA activation. Thus, because signaling by all of these effectors in ␤ cells is imparted by tyrosine phosphorylation, large Maf activation would be directly influenced in the absence of insulin receptor and/or insulin receptor substrate-2 function, resulting in reduced insulin and pdx-1 transcription as seen in islets of Irs2 Ϫ/Ϫ mice (47). In support of this theory, ␤ cell mass and function are restored upon transgenic expression of PDX-1 in Irs2 Ϫ/Ϫ mice (47). Our studies are currently focused on determining the nature of the interactions between RIPE3b1/Maf and other factors that are required for assembly of the pdx-1 and insulin transcription complexes and, more broadly, the significance of RIPE3b1/Maf in pancreatic development.