Mice carrying a targeted deletion of the bHLH protein NeuroD1 fail to develop secretin-expressing enteroendocrine cells, indicating that this protein is essential for secretin gene expression. Paradoxically, in vitro studies suggest that NeuroD1 is a relatively weak activator of the secretin gene transcription by itself. Of the transcription factors that bind to the secretin enhancer, only NeuroD1 is relatively restricted to a limited number of cell types, as opposed to the ubiquitous expression of the others, implying a major role for NeuroD1 in regulating secretin gene expression. In the present work, we show that physical interactions between NeuroD1 and the ubiquitously expressed transcription factor Sp1 on the secretin gene enhancer significantly potentiate transactivation by this tissue-specific class B bHLH protein. We previously identified the protein Finb/RREB1 as a transcription-modifying protein that synergistically potentiates NeuroD1. Thus, the interaction of NeuroD1 with two widely expressed DNA binding proteins is one mechanism for potentiating the relatively weak transactivating function of NeuroD1.
Cell-type-specific gene expression depends on the presence of a combination of transcription factors, some of which are also highly restricted in expression, as well as the organization transcription factor binding sites controlling a gene. The proximal enhancer of the secretin gene is conserved in different mammalian species, maintaining a nearly identical arrangement of four cis
regulatory elements including the E-box and immediately adjacent Sp1 binding site. The close proximity of these two sites appears to be essential for the synergism between Sp1 and NeuroD1. Finb/RREB1, which binds to sequences upstream of the E-box, appears to be an unusual class of transcription-modifying protein. Although Finb/RREB1 is present in corepressor complexes with CtBP (31
), it serves to potentiate NeuroD1-dependent transcription despite the absence of an intrinsic activation domain or any direct activation of transcription by itself. In contrast to coactivators, Finb must bind to DNA to modify the activity of NeuroD1.
The interaction of NeuroD1 with Sp1 through regions close to the DNA binding domain of each protein may in part explain why synergistic transcriptional activation depends on the adjacent positions of the E-box and Sp1 binding sites. This interaction results in cooperative DNA binding and stabilized Sp1 DNA binding in the resultant ternary DNA-protein complex with NeuroD1 and E12.
The promoters of several muscle-specific genes including the regulatory region of human cardiac alpha-actin promoter and troponin I promoter have Sp1 binding sites in close proximity to E-boxes that bind to myogenic bHLH proteins (1
). The HLH domain of the myogenic bHLH protein myogenin, like NeuroD1, physically associates with the C-terminal zinc finger domain of Sp1. However, it is not known whether myogenin and Sp1 increase transcription of the cardiac actin promoter in an additive or synergistic manner. In this promoter, the E-box and Sp1 site are separated by an additional 10 nucleotides, or approximately one complete helical turn, compared to the secretin enhancer. Introduction of 10 additional nucleotides between the secretin gene E-box and the adjacent Sp1 site abrogated their synergistic effects on transcription.
Besides the secretin gene, relatively few target genes that are directly activated by NeuroD1 have been identified. These include the genes encoding the hormones insulin and POMC. Expression of both of these genes depends on interactions between bHLH proteins with lineage-restricted homeodomain transcription factors bound to nearby cis elements to potentiate the transactivating function of NeuroD1.
Lineage-specific transcription of the POMC gene in the pituitary gland is enhanced by the functional synergy between bHLH factors binding to an E-box and the pituitary gland-specific homeodomain protein Pitx-1 (25
). Pitx-1 indirectly modifies NeuroD1-dependent transcription by physically interacting with one of its ubiquitously expressed dimerization partners rather than by a direct association with NeuroD1. The bHLH domain of Pan1 (E47) serves as a protein-protein interaction domain with the homeodomain of Pitx-1. The NeuroD1 binding E-box in the POMC promoter is separated from the Pitx-1 binding site by 67 bp. This spacing may indicate that the interactions on the POMC promoter involve different surfaces of the bHLH proteins than in the case of the insulin and secretin promoter, where Pdx1 and Sp1 bind to sites immediately adjacent to the E-box.
Insulin-expressing pancreatic β cells arise from the primitive gut endoderm and are developmentally related to enteroendocrine cells, and yet insulin gene expression does not occur in the intestine. The expression of insulin and secretin genes, two NeuroD1-dependent genes, in distinct cell types depends in part on the expression of different sets of transcriptional activators in islets versus enteroendocrine cells and on the organization of transcription factor binding sites on each promoter. The major elements responsible for β-cell-specific transcription of the insulin gene are localized to a relatively small region of the promoter that contains an E-box and two TAAT-rich A elements that bind to the homeodomain protein PDX-1. In developing animals, PDX-1 is expressed throughout the proximal duodenum and the pancreas, but intestinal expression is largely absent in adult animals, whereas expression in islets continues.
The homeodomain of PDX-1 associates with several other proteins including E47, NeuroD1, and high-mobility-group protein Y1 to form a higher-order transcription-activating complex (24
). Unlike the specific interaction between NeuroD1 and Sp1 on the secretin gene, PDX-1 interacts with the ubiquitous dimerization partners of NeuroD1 as well as with NeuroD1. In the case of the insulin gene, synergistic transcriptional activation occurs with E47 as well as NeuroD1. In addition, high-mobility-group protein Y1 synergizes with the other members of the insulin gene-transactivating complex. The absence of PDX-1 in adult enteroendocrine cells may explain in part why insulin is not expressed in the intestine. Thus, NeuroD1-dependent expression of the insulin and secretin genes in pancreatic β cells and S-type enteroendocrine cells, respectively, depends on different sets of transcription factors that interact with NeuroD1 to potentiate its activity.
NeuroD1 plays an important role in regulating terminal differentiation of neurons and hormone-producing cell lineages. Relatively few target genes have been identified for this bHLH protein. Other DNA binding proteins potentiate transactivation of the genes encoding the hormones POMC, insulin, and secretin by NeuroD1. The enhancer of the secretin gene shares little with the organization of the insulin or POMC genes. Unlike the POMC and insulin genes, where NeuroD1 synergizes with homeodomain proteins specific for their tissue, NeuroD1 synergizes with the widely expressed proteins Sp1 and RREB1 to increase secretin gene transcription. The specific arrangement of factor binding sites in the secretin enhancer provides the critical context for the interaction with Sp1 and Finb that potentiates the activating function of NeuroD1.