The insulin gene is expressed almost exclusively in pancreatic beta-cells. Previous work in our laboratory has shown that pancreatic beta-cell-specific expression of the rat insulin II gene is controlled by a number of positive and negative cis-acting DNA elements within the enhancer. We have shown that one element within the enhancer, located between nucleotides -100 and -91 (GCCATCTGCT; referred to as the insulin control element [ICE]) relative to the transcription start site, is controlled by both positive- and negative-acting cellular transcription factors. The positive-acting factor appears to be uniquely active in beta-cells. To identify the nucleotides within the ICE that mediate positive cell-type-specific regulation, point mutations within this element were generated and assayed for their effects on expression. Base pairs -97, -94, -93, and -92 were found to be crucial for the activator function of this region, while mutations at base pairs -100, -96, and -91 had little or no effect on activity. The gel mobility shift assay was used to determine whether specific cellular factors associated directly with the ICE. Several specific protein-DNA complexes were detected in extracts prepared from insulin-producing and non-insulin-producing cells, including a complex unique to beta-cell extracts. The ability of unlabeled wild-type and point mutant versions of the ICE to compete for binding to these cellular factors demonstrated that the beta-cell-specific complex appears to contain the insulin gene activator protein(s). Interestingly, the adenovirus type 2 major late promoter upstream element (USE; GCCACGTGAC) also competed in the gel mobility shift assay for binding of cellular proteins to the ICE. These results suggested that the cellular factor that binds to the USE (i.e., USF) also interacts with the ICE. This was directly demonstrated by showing that ICE and USE sequences completed for the USF required for adenovirus type 2 major late promoter transcription in vitro and by showing that reticulocyte lysate-translated human USF products bound to the ICE. However, the USE sequences were unable to stimulate beta-cell-type-specific activity in vivo. We discuss the possible relationship of these observations to positive and negative control mediated by the ICE.