|Home | About | Journals | Submit | Contact Us | Français|
NeuroD1/BETA2 is a key regulator of pancreatic islet morphogenesis and insulin hormone gene transcription in islet β cells. This factor also appears to be involved in neurogenic differentiation, because NeuroD1/BETA2 is able to induce premature differentiation of neuronal precursors and convert ectoderm into fully differentiated neurons upon ectopic expression in Xenopus embryos. We have identified amino acid sequences in mammalian and Xenopus NeuroD1/BETA2 that are necessary for insulin gene expression and ectopic neurogenesis. Our results indicate that evolutionarily conserved sequences spanning the basic helix-loop-helix (amino acids [aa] 100 to 155) and C-terminal (aa 156 to 355) regions are important for both of these processes. The transactivation domains (AD1, aa 189 to 299; AD2, aa 300 to 355) were within the carboxy-terminal region, as analyzed by using GAL4:NeuroD1/BETA2 chimeras. Selective activation of mammalian insulin gene enhancer-driven expression and ectopic neurogenesis in Xenopus embryos was regulated by two independent and separable domains of NeuroD1/BETA2, located between aa 156 to 251 and aa 252 to 355. GAL4:NeuroD1/BETA2 constructs spanning these sequences demonstrated that only aa 252 to 355 contained activation domain function, although both aa 156 to 251 and 300 to 355 were found to interact with the p300/CREB binding protein (CBP) coactivator. These results implicate p300/CBP in NeuroD1/BETA2 function and further suggest that comparable mechanisms are utilized to direct target gene transcription during differentiation and in adult islet β cells.
The mouse pancreas develops as an outpocketing from the embryonic gut. Cells lining this evagination differentiate and segregate into exocrine and endocrine tissues. Both of these pancreatic tissues arise from a common, but limited, set of multipotential endodermal precursors (2, 12, 29, 61). The hormones produced by the endocrine pancreas appear sequentially during development. Glucagon- and insulin-producing cells are first observed at 9.5 days postcoitum in mice, followed by somatostatin- and then pancreatic polypeptide-producing cells. The pancreas develops from precursor cells that coexpress these hormonal gene products, with expression both selectively increased and extinguished during islet maturation, resulting in the production of the single hormone, α (glucagon)-, β (insulin)-, δ (somatostatin)-, and PP (pancreatic polypeptide) cell types (20, 21, 42, 59). The endocrine pancreas progenitors also coexpress a subset of markers of neuroectodermal cell differentiation, such as tyrosine hydroxylase (59), an enzyme in the catecholamine biosynthesis pathway. This observation suggested that there may be shared lineage-specific regulators within pancreatic and neuronal cells. Recently, this relationship was assessed by analyzing the effect of elimination of various genes encoding transcription factors enriched within both cell types on pancreatic development. Such experiments have clearly demonstrated that inactivation of the genes encoding PAX-4 (57), PAX-6 (58), Isl-1 (1), or NeuroD1/BETA2 (36) profoundly influences islet development. Although these studies showed that these proteins play important regulatory roles during development, the molecular mechanisms involved are poorly understood.
NeuroD1/BETA2 is in the basic helix-loop-helix (bHLH) family of transcriptional activators (26, 37) and functions in a complex with the more generally distributed E2A (i.e. E47, E12, E2-5)-encoded proteins (3, 10, 18, 54) and HEB-encoded proteins (41). Dimerization between bHLH proteins depends on the HLH region, while protein-DNA interactions are mediated by the basic region. The presence of a tissue-enriched and a ubiquitously distributed bHLH factor in the activator complex is characteristic of other tissue-specific members of this family, the best characterized of which are involved in myogenic (17, 32) and neuronal (8, 23, 56) differentiation. These activators bind to the consensus sequence CANNTG (termed an E-box), with heterodimerization increasing DNA binding and activation capacity (25, 32, 37).
NeuroD1/BETA2 is expressed in pancreatic islet endocrine cells (34, 36), the intestine (34), the pituitary (43), and a subset of neurons in the central and peripheral nervous system (26). This factor was independently isolated and characterized by its ability to activate neurite formation upon ectopic expression in Xenopus embryos (termed NeuroD and referred to here as NeuroD1 ), and insulin reporter gene transcription in transfected β cells (termed BETA2 ). NeuroD1/BETA2 also stimulates secretin (34) and proopiomelanocortin (POMC ) transcription in the intestine and pituitary gland, respectively. Activation by NeuroD1/BETA2 is potentiated by the p300/CREB binding protein (CBP) coactivator (35, 45). Although the exact mechanism involved in p300/CBP-mediated transcription is unclear, it may result from bridging through direct interactions the activator to the basal transcriptional machinery and/or from promotion of a transcriptionally active state to targeted genes through its intrinsic histone acetyltransferase activity (15, 55). Since p300/CBP also modulates the activity of a number of key activators involved in regulating cellular proliferation and differentiation (15, 55), including the myogenic bHLH factors (16, 50, 64), its association with NeuroD1/BETA2 may be important for transcriptional signaling during development and in the adult.
Gene targeting experiments established an important role for NeuroD1/BETA2 in pancreatic development. NeuroD1/BETA2−/− mice have a marked reduction in insulin-producing β cells and fail to develop mature islets (25a, 36). In addition, secretin- and cholecystokinin-producing enteroendocrine cells were not present in homozygous mutant mice (36). These animals develop severe diabetes and die within a few days of birth. In contrast, homozygous E2A (5, 65, 66) or HEB (65) null mice do not display any change in endocrine pancreas morphology or function (22, 52). The nervous system also appears to be normal at the gross level in NeuroD1/BETA2−/− mice, presumably due to the presence of a compensatory factor.
Collectively, these results indicated that NeuroD1/BETA2 is required during the development of specialized pancreatic and enteroendocrine cell types arising from gut endoderm, as well as being involved in differentiated gene product expression (i.e., insulin, secretin, and POMC). Furthermore, because both NeuroD1/BETA2 (26) and NeuroD2 (31) can convert epidermal cells to neurons, NeuroD-like factors also appear to be important in the development of the nervous system. In this study, the amino acid sequences of NeuroD1/BETA2 that are necessary for insulin gene transcription and neurogenesis were identified. The stimulatory properties of the mammalian and Xenopus NeuroD1/BETA2 proteins were compared. These proteins share approximately 71, 96, and 85% identity in their N-terminal (amino acids [aa] 1 to 99), bHLH (aa 100 to 155), and C-terminal (aa 156 to 355) regions, respectively (26, 31, 37). The sequences spanning the C-terminal region of NeuroD1/BETA2 were required for insulin gene activation and neuronal differentiation. Stimulation appears to be mediated by the p300/CBP coactivator through contacts within aa 156 to 251 and aa 300 to 355. We propose that the interaction of p300/CBP with NeuroD1/BETA2 plays an important role in E-box activation by NeuroD1/BETA2 in adult islet cells and during differentiation.
The hamster BETA2 (37) and Xenopus NeuroD (26) protein coding sequences used in constructing the full-length and deletion mutant GAL4 chimeras were derived by PCR. Amplification was performed according to the method of Saiki et al. (49) with Thermus aquaticus DNA polymerase (Perkin-Elmer Cetus). The resulting BETA2 and Xenopus NeuroD sequences were ligated into the simian virus 40 (SV40) promoter-enhancer-driven GAL4 expression plasmid, pSG424 (28), to create in-frame GAL4-fusion proteins. Each construct is named according to the N- and C-terminal amino acid sequence of BETA2 and Xenopus NeuroD present in the construct. The BETA2 and Xenopus NeuroD sequences were subcloned into pCS2+NLSMT (60), which contains the simian cytomegalovirus (CMV) promoter and SV40 T-antigen nuclear localization signal fused to six copies of the myc epitope recognized by the 9e10 monoclonal antibody. The inserts were cloned in frame and downstream of the myc sequences. The structure of all plasmid constructs was confirmed by sequencing. The E47 activation domain (AD) mutants, AD1, AD2, and AD1/AD2, were described previously (39). The activation domain activity of E47 has been significantly compromised by a change in aa 19 (Leu to Arg) and aa 22 (Phe to Arg) in the AD1 mutant and aa 337 (Val to Glu) and aa 338 (Leu to Arg) in the AD2 mutant. The wild-type and mutant E47 cDNAs were expressed from the SV40 enhancer-promoter in pJ3Ω (30). The p300 dl10 mutant (deletion of aa 1680 to 1811) (27), p300:herpes virus acidic activation region (VP16) fusion (p300 Q:VP16) (aa 1945 to 2377) (45), insulin FF chloramphenicol acetyltransferase (CAT) (19), and (GAL4)5 E1bCAT reporter (28) constructs have been described previously.
Monolayer cultures of BETATC3, baby hamster kidney (BHK), HeLa, and HIT T-15 2.2.2 cells were maintained as described previously (39). The transfections were performed by using either the calcium-phosphate coprecipitation procedure (HIT T-15, BHK, and HeLa) (62) or the electroporation procedure (BETATC3) (47). A luciferase (LUC) reporter gene recovery marker, pSV2 LUC (13), was cotransfected with the CAT reporter plasmid. Four hours after the addition of the calcium-phosphate DNA precipitate, BHK and HIT T-15 cells were treated with 20% glycerol for 2 min. The transfections in βTC3, BHK, HeLa, and HIT T-15 cells with the GAL4:ND plasmids (or the GAL4 DNA binding vector alone [termed GAL4 1–147]) (8 μg) also included (GAL4)5 E1bCAT (2 μg), and a luciferase (LUC) reporter gene recovery marker, pSV2 LUC (1 μg); the transfection analysis (total DNA concentration 10 μg) in HeLa cells of the CMV NeuroD1 constructs was conducted with (or without) the E47 expression plasmid (2 μg), FF CAT (1 μg), and pSV2 LUC (1 μg). Cells were harvested 40 to 48 h after transfection. The CAT activity from the reporter plasmid was normalized to the LUC activity of the cotransfected internal control plasmid. LUC and CAT enzymatic assays were performed as described by de Wet et al. (13) and Nordeen et al. (38), respectively. Each experiment was repeated several times with at least two different plasmid preparations.
HeLa nuclear extracts were prepared from transfected cells as described previously (51). Fifty micrograms of protein extract was resolved on a sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel (12.5% polyacrylamide) and electrotransferred to Immobilon polyvinylidene difluoride membrane (Millipore, Bedford, Mass.). The membrane was probed with the monoclonal anti-myc tag 9e10 antibody (ATCC CRL1729). The positions of bound antibodies were detected by autoluminography with the ECL (enhanced chemiluminescence) detection kit (Amersham).
Albino embryos were obtained by in vitro fertilization, dejellied with 2% cysteine 1 h later, and injected at the two-cell stage with in vitro-synthesized, capped, and polyadenylated RNAs. All RNAs were synthesized from CS-2 vectors (60) by using SP6 polymerase. Microinjection of RNA was performed by injection of approximately 4 to 5 nl of RNA (100 pg/nl) into one cell of two-cell stage Xenopus embryos at two positions in the animal hemisphere. The injected embryos were allowed to develop in 0.1× modified Barth’s saline until they reached between stages 20 and 24. Embryos were fixed in Dent’s fixative (20% dimethylsulfoxide, 80% methanol) and stained with rabbit anti-N-CAM antibody (1:500 dilution) (6) as described previously (26). Primary antibody was detected with alkaline phosphatase-conjugated goat anti-rabbit secondary antibody (diluted 1:2,000; Boehringer Mannheim).
To begin to map the domain or domains of NeuroD1/BETA2 that mediate activation, we employed a GAL4 DNA-binding domain (DBD)-dependent reporter system, in which regions of hamster NeuroD1/BETA2 were fused to the DBD of yeast GAL4. The GAL4 protein fusion plasmids, together with a CAT reporter plasmid containing five GAL4 DNA binding sites upstream of the E1b TATA, were cotransfected into cell lines that express both insulin and NeuroD1/BETA2 (HIT T-15 and βTC3) and those that do not (BHK and HeLa). CAT enzyme activity from the GAL4 chimeras was normalized to the activity of the cotransfected internal control plasmid, pSV2 LUC.
In general, the expression patterns of the GAL4:NeuroD1/BETA2 (GAL4:ND) chimeras were similar between different cell lines (Fig. (Fig.1).1). We observed that while constructs containing only the N-terminal region or both the N-terminal and bHLH regions were inactive [see GAL4:ND(1–99) and GAL4:ND(1–155)], those that contained the C-terminal region were effective activators [see GAL4:ND(156–355), GAL4:ND(189–355), and GAL4:ND(239–355)]. Gel shift analysis also revealed that similar relative amounts of each GAL4 fusion protein were synthesized between cell lines (data not shown).
The results suggested that C-terminal region sequences from 189 to 355 spanned the transactivation domains of NeuroD1/BETA2. Further analysis indicated that two distinct activation domains were located within this region, between aa 189 and 299 (termed activation domain 1 [AD1]) and aa 300 and 355 (termed AD2). Because GAL4:ND(189–355) and GAL4:ND(239–355) had comparable activities (Fig. (Fig.1B),1B), we considered that the amino-terminal boundary of AD1 might be closer to aa 239. However, GAL4:ND(239–299) was found to be much less active than GAL4:ND(189–299) (Fig. (Fig.1B),1B), indicating that aa 189 to 238 also contribute to AD1 activation. The GAL4 constructs spanning the homologous region of Xenopus NeuroD1 also showed a similar selective expression pattern (Fig. (Fig.1C1C and Table Table1).1). These results suggest that the mechanisms important in transactivation domain function are conserved between the Xenopus and hamster NeuroD1/BETA2 proteins.
We next sought to identify the sequences in NeuroD1/BETA2 that were necessary for insulin E-box element-mediated transcription in HeLa cells. NeuroD1/BETA2 specifically stimulates transcription by binding to the E-box element at bp −239 to −228 within the insulin enhancer-driven FF CAT construct, which contains from bp −247 to bp −197 of the rat insulin I gene (19, 39). Because HeLa cells lack NeuroD1/BETA2, as well as other β-cell-enriched transcription factors important for insulin gene expression (10, 11, 19, 24, 40, 54, 62), this system provides the opportunity to evaluate the transactivation potential of mutant proteins in isolation. The bHLH region was retained in each of the NeuroD1/BETA2 mutant constructs, since it has been clearly demonstrated to be essenconstructs, since it has been clearly demonstrated to be essential for dimerization and E-box DNA binding within this family of proteins (9, 25, 33). A myc epitope tag and nuclear localization signal were inserted at the N terminus of each NeuroD1/BETA2 protein to facilitate monitoring of protein expression levels and direct appropriate localization. These modifications do not affect the function of the protein (26, 31).
HeLa cells were transfected with CMV-driven expression vectors encoding wild-type NeuroD1/BETA2 and C-terminally truncated constructs that are missing sequences from both AD1 and AD2 [ND(1–155) and ND(1–251), respectively] or only AD2 [ND(1–299)]. Insulin gene activation from FF CAT was analyzed for each NeuroD1/BETA2 construct in the presence or absence of the E2A-encoded protein, E47. Activation was not observed in cells transfected only with the NeuroD1/BETA2 construct spanning the N-terminal region, ND(1–155), whereas full-length NeuroD1/BETA2 effectively stimulated FF CAT expression (Fig. (Fig.2).2). In the absence of AD2, ND(1–299) retained 37% of NeuroD1/BETA2 activity, while the AD1/AD2 mutant, ND(1–251), retained 10%. In addition, removal of the region of aa 1 to 99 from ND(1–251) did not significantly affect activity [see ND(100–251) in Fig. Fig.2].2]. Western blotting also revealed that similar amounts of each NeuroD1/BETA2 protein were synthesized in transfected cells (Fig. (Fig.3).3). The same activation pattern was also obtained with comparable regions of Xenopus NeuroD1/BETA2 (data not shown). These results are in general agreement with the mapping of the transcriptional activation domains to aa 189 to 299 and aa 300 to 355. Although we did not expect to find any stimulatory activity from ND(1–251), since a GAL4 fusion construct spanning the C-terminal activation domain region sequences of this NeuroD1/BETA2 construct was inactive in HIT T-15 and HeLa cells [see GAL4:ND(156–251) in Fig. Fig.4].4]. Together, these observations imply that the activation domain functions of NeuroD1/BETA2 are important, but not essential, for insulin E-box-stimulated transcription.
E47 stimulated the activity from each of the NeuroD1/BETA2 expression plasmids, yet did not effect FF CAT expression alone (Fig. (Fig.2)2) (39). Since dimerization between NeuroD1/BETA2 and the E2A (3, 10, 18, 54)- or HEB (41)-encoded protein is required for E-box binding (37), these results indicated that the levels of the generally distributed partner were limiting in cells transfected with NeuroD1/BETA2 alone. To determine if the activation domains of the E2A and HEB proteins were also important in FF CAT stimulation, mutants within these conserved regions of the E47 protein were cotransfected with ND(1–155) and ND(1–251). The two distinct activation domains of E47 are located between aa 1 to 83 (termed AD1) (4, 30) and aa 345 to 411 (AD2) (4, 46). FF CAT activity was reduced by approximately 50% in the AD1 or AD2 mutants of E47, and the AD1/AD2 double mutant activity was reduced to near the level of the NeuroD1/BETA2 construct transfected alone (Fig. (Fig.5).5). These results indicate that E-box stimulation is also mediated by the activation domains in the E2A- and HEB-encoded proteins.
Recent studies have demonstrated that p300/CBP can bind to and potentiate the activity of NeuroD1/BETA2 (35, 45) and the HEB- and E2A-encoded proteins (16, 45). p300/CBP can physically and functionally interact with the bHLH (35) and C-terminal region (aa 156 to 355) (45) of NeuroD1/BETA2. Since the results described above indicated that aa 156 to 355 were more important in selective activation, we sought to define more precisely the sequences within this region that were required for binding to p300/CBP. In this analysis, the abilities of GAL4:ND fusion proteins spanning aa 156 to 251, 189 to 299, and 300 to 355 to functionally interact with p300/CBP or the p300 mutant (p300 dl10) were compared to that of GAL4:ND(156–355) in HIT T-15 cells. Although p300 dl10 lacks the histidine- and cysteine-rich region (termed C/H3) between aa 1680 and 1811 which is important in adenovirus E1A binding (Fig. (Fig.6A),6A), it can functionally substitute for p300 in assays with NeuroD1/BETA2 (45). p300 dl10 (Fig. (Fig.6B)6B) and p300 (data not shown) stimulated the activity of GAL4:ND(156–251) and GAL4:ND(300–355) to the same level as the GAL4:ND(156–355) positive control. However, the activity of GAL4:ND(189–299) was not affected to the same extent.
We have recently shown that NeuroD1/BETA2 binds to the glutamine-rich region of p300/CBP located between aa 1945 and 2377 (termed the Q domain) (45). To determine if this region of p300/CBP also binds to the 156 to 251 and 300 to 355 (AD2) domains of NeuroD1/BETA2, we analyzed whether a p300 Q:VP16 activation domain fusion protein could interact with and stimulate GAL4:ND(156–251) and GAL4:ND(300–355) activity. The level of p300 Q:VP16 stimulation was comparable to that observed with GAL4:ND(156–355) (Fig. (Fig.6C).6C). In contrast, p300 Q:VP16 had only a minimal effect on GAL4:ND(189–299) activity. Together, these results strongly suggest that interactions between the Q region of p300/CBP and aa 156 to 251 and 300 to 355 (AD2) of NeuroD1/BETA2 are important for activation.
Ectopic expression of NeuroD1/BETA2 (26) or NeuroD2 (31) in developing Xenopus embryos results in cell fate conversion of ectodermal cells to neurons in the epidermis. To determine if the neurogenic activity of NeuroD1/BETA2 involved the same regions of the protein found to be important in stimulating insulin E-box transcription, we injected Xenopus embryos at the two-cell stage with Xenopus NeuroD1/BETA2 RNA encoding the wild-type protein and mutants either capable or incapable of inducing insulin gene expression. Each construct contained the NeuroD1/BETA2 bHLH region and a nuclear localization signal. Injections were conducted on one side of the embryo, with the other side serving as a control. The neurogenic conversion activity of the NeuroD1/BETA2 constructs was evaluated by their ability to induce expression of a neural cell-specific marker in Xenopus, neural cell adhesion molecule (N-CAM) (26), by whole-mount immunohistochemistry. An anti-myc tag antibody, 9e10, was used to confirm that most of the ectodermal cells on the injected side of the embryo expressed the myc-tagged BETA2 fusion protein (data not shown).
The regions of NeuroD1/BETA2 that are important for regulating insulin E-box transcription were also found to be necessary for ectopic neuronal development. Thus, neurogenic activity was dependent upon the C-terminal region, because constructs encoding the sequences from 1 to 251 and 1 to 299 induced N-CAM staining, while the 1 to 155 construct did not (Fig. (Fig.7).7). The N-CAM staining pattern for XND 1–155 was indistinguishable from that in wild-type uninjected embryos (data not shown). As in the case of XND (Fig. (Fig.7)7) (26), the ectopic neurons induced by the XND mutants were confined to a subpopulation of ectodermal cells, as shown by the spotty N-CAM-positive staining pattern.
Because XND 100–251 induced neurogenesis (Fig. (Fig.7)7) and insulin gene transcription (Fig. (Fig.2),2), we concluded that the N-terminal region was not important in these responses, nor was the bHLH alone sufficient. These results also indicated that the sequences between 156 and 251 of NeuroD1/BETA2 alone could mediate these activities. Toward this end, we analyzed whether removal of these sequences influenced either insulin E-box-driven transcription or neurogenic activation. XNDΔ156–251 was capable of stimulating both insulin gene transcription (Fig. (Fig.2)2) and N-CAM staining (Fig. (Fig.7).7). This NeuroD1/BETA2 mutant expression construct possesses activation domain function, as concluded from an analysis of GAL4:ND(252–355) activity in HIT T-15 and HeLa cells (Fig. (Fig.4).4). In contrast, XND 100–251 and XND 1–251 do not appear to have a functional activation domain [see GAL4:ND(156–251) in Fig. Fig.44].
In contrast to the insulin E-box-driven assay, we were unable to quantitate differences in neurogenic potential between the NeuroD1/BETA2 mutants upon ectopic expression in Xenopus embryos (Fig. (Fig.8).8). This may be reflective of the significance of the C-terminal regulatory regions in activating E-box-mediated transcription of an insulin versus neuronal target gene, or may be simply a difference in the sensitivity of the assays. Importantly, these assays indicate that activation of both the neurogenic and the islet transcription programs is mediated by the same two distinct regions of NeuroD1/BETA2. Furthermore, the interaction of the p300/CBP coactivator with these regulatory regions, which are located between aa 156 to 251 and 252 to 355, could play an essential role in NeuroD1/BETA2 transcriptional activity.
In this study, we examined the basis for transcriptional activation by NeuroD1/BETA2 in islet β cells and the developing nervous system. Our objective was to identify and functionally characterize the domain or domains of NeuroD1/BETA2 that were important in these processes. Although we found that the bHLH region alone was unable to stimulate insulin E-box or neurogenic activation, C-terminal region sequences from 156 to 355 of the Xenopus or mammalian protein, when linked to the bHLH region, were active. Mutational analysis demonstrated that selective activation by the C-terminal region was mediated by two functionally independent domains, which spanned sequences 156 to 251 and 252 to 355. These results suggest that similar mechanisms are utilized by NeuroD1/BETA2 to direct E-box-mediated transcription of important regulatory genes involved in neuronal differentiation and adult islet β-cell function. Since p300/CBP, an essential coactivator of NeuroD1/BETA2 (35, 45), interacted with each domain (i.e., aa 156 to 251 and aa 300 to 355) associated with selective activation, its recruitment appears to be required for activity.
Our initial observation was that the transactivation domain function of mammalian and Xenopus NeuroD1/BETA2 was located within aa 189 and 355, as analyzed with GAL4:ND chimeras. More detailed analysis demonstrated that this region contained two activation components, with AD1 (aa 189 to 299) more active than AD2 (aa 300 to 355) in islet BETA cells. AD1 function was severely compromised in GAL4:ND(156–251) and GAL4:ND(239–299), suggesting that NeuroD1/BETA2 sequences within each mutant played an important role in defining AD1 activation. To examine the role of the AD1 and AD2 regions in transcriptional stimulation of a physiologically relevant target gene, activation domain mutants within the mammalian and Xenopus NeuroD1/BETA2 protein were analyzed for their ability to activate insulin E-box-driven enhancer expression. The bHLH region was present in each mutant construct, while N-terminal sequences preceding the bHLH domain as well as selected portions of the C-terminal region were removed. The data indicate that C-terminal region sequences of NeuroD1/BETA2, which lack a functional activation domain, mediate E-box activation. Thus, while NeuroD1/BETA2 constructs missing AD2 [ND(1–299)] and AD1 [ND(1–251) or ND(100–251)] function were less effective than the wild type, they were significantly more active than the N-terminal region construct [ND(1–155)] (Fig. (Fig.2).2). Stimulation from each mutant was potentiated upon cotransfection with E47, a heterodimeric partner in the insulin E-box activator complex (3, 10, 18, 54), implying that mutagenesis affected only the activation properties of NeuroD1/BETA2 and not its dimerization and DNA binding properties.
Strikingly, neurogenic activation was mediated by the same regions of NeuroD1/BETA2 required in insulin E-box activation (Fig. (Fig.8).8). These results suggest the direct involvement of 156-to-355 region sequences in stimulating transcription from genes important in neurogenic differentiation and adult islet gene expression. The NeuroD1/BETA2 bHLH domain is unable to mediate selective activation in the absence of C-terminal region sequences, yet is still required for up-regulation of transcription of target genes by its ability to direct both heterodimerization with the E47, E12, or HEB protein and E-box-specific binding.
The sequences from 156 to 251 and 252 to 355 of NeuroD1/BETA2 defined separable and independently acting stimulatory regions, with only aa 252 to 355 containing activation domain properties. Interestingly, p300/CBP interacted within both the 156 to 251 and 252 to 355 regions, indicating that its recruitment was necessary for activation. Mutoh et al. (35) have recently shown that p300/CBP can also interact with the bHLH region of NeuroD1/BETA2. While our results demonstrate that the association of p300/CBP with the bHLH region is not sufficient for insulin gene or neurogenic activation, it is possible that this interaction is important in another context. Notably, the association of p300/CBP with the myogenic bHLH activator, MyoD, appears to be required for E-box activation and differentiation (44), and this key regulatory factor also has binding sites for p300/CBP within both its activation domain (50, 64) and bHLH (16) regions.
The high degree of sequence identity among NeuroD1/BETA2-related proteins (i.e., Nex1 , NeuroD2 , and NeuroM ) within their bHLH and C-terminal region sequences suggests that p300/CBP is also likely to be essential for their activation as well. Neurogenic activation by NeuroD1/BETA2 (26) or NeuroD2 (31) converts only a subpopulation of ectodermal cells to neurons (represented by the spotty N-CAM expression pattern in Fig. Fig.7),7), suggesting that other cell-restricted factors act in conjunction with this factor to mediate neurogenesis. The ubiquitously distributed p300/CBP protein may facilitate regulated E-box transcription in ectodermal cells, as well as islet endocrine β cells, by promoting interactions between NeuroD1/BETA2 and other tissue-enriched regulators. Recent studies have clearly demonstrated that p300/CBP utilizes such a mechanism to modulate the activity of a number of other key activators, including those involved in regulating cellular proliferation and differentiation (15, 55).
Our results suggest that the mechanisms important in NeuroD1/BETA2 activation of E-box-driven genes involved in neuronal development are closely related to those required for insulin gene expression in islet β cells and are likely related to islet endodermal cell differentiation. Furthermore, that stimulation is linked to the transcriptional signaling properties of p300/CBP. Results consistent with this hypothesis were recently obtained in both Caenorhabditis elegans (53) and mice (63), when it was shown that p300/CBP was essential for cellular differentiation. In mice lacking NeuroD1/BETA2, there is a severe effect on islet endodermal cell development (25a, 36), with specific losses in islet β-, α-, and δ-cell numbers of nearly 75, 40, and 20%, respectively. Experiments are in progress to test if NeuroD1/BETA2-mediated differentiation of islet cells involves the recruitment of p300/CBP.
We thank Mina Peshavaria, Gladys Teitelman, and Kevin Gerrish for constructive criticism of the manuscript, Lauren Snider and Kristin Swihart for technical assistance; and Ming-Jer Tsai for generously providing the hamster BETA2 cDNA.
This work was supported by grants from the National Institutes of Health (RO1 DK49852 to R.S., NIH RO1 N535118 to J.E.L., and DK7061 training grant to S.S.), American Diabetes Association (to A.S.), and Juvenile Diabetes Foundation (398212 to S.S.). Partial support was also derived from the Vanderbilt University Diabetes Research and Training Center Molecular Biology Core Laboratory (Public Health Service grant P60 DK20593 from the National Institutes of Health).