The recognition sequences for individual transcription regulatory proteins vary among different promoter regions. The reasons for this diversity of recognition sequences are not well understood. The recognition sequences are frequently suboptimal binding sites for the regulatory proteins, which would appear to impede the already formidable task of transcription factors: to specify unique sites of transcription initiation in the genome. In some cases, replacement of nonconsensus recognition sequences by optimal binding sites eliminates their transcription regulatory functions (18). There are many possible explanations for this apparently imperfect nature of regulatory elements. One possibility is that they serve as binding sites for many different proteins and that the sequences of the elements serve to balance binding by alternative regulatory proteins. A second possibility is that each element contributes only a small fraction of the specificity required for promoter function and that the sequence variability prevents individual transcription factors from dominating the regulation of the promoter. A third possibility is that the sequence of the element influences the function of the protein(s) that binds to the element. Such functional effects might be mediated by changes in the conformation or the structural organization of nucleoprotein complexes (7,14,28,32,41,44,45).In eukaryotic organisms, transcription initiation is regulated by the concerted action of many transcription factors that form multiprotein complexes at promoter and enhancer regions. Assembly of these nucleoprotein complexes can be affected by the orientations of the individual regulatory elements (26,27,39,40). Some heterodimers that recognize palindromic DNA sequences also exhibit orientation-dependent transcriptional activities at regulatory elements that contain asymmetric base pairs either within the recognition site or in flanking sequences (6,10,35,36). This orientation dependence can be mediated by cooperative DNA binding with transcription factors that recognize adjacent binding sites within composite regulatory elements (10,35,36).Fos and Jun family bZIP proteins form homo-and heterodimers via a leucine zipper dimerization interface and bind to palindromic AP-1 recognition elements [TGA(C/G)TCA] via a basic DNA contact region. Fos and Jun make essentially identical contacts with the two half-sites in the X-ray crystal structure of the Fos-Jun-AP-1 complex (13). However, symmetry-related base substitutions in the two half-sites have distinct effects on Fos-Jun heterodimer binding (38). The differential effects of symmetry-related base substitutions indicate that they are recognized in the context of the asymmetric central base pair or flanking DNA sequences.The consensus AP-1 element is not perfectly symmetrical, since the central CG base pair results in two different overlapping half-sites (TGAC and TGAG). Fos-Jun heterodimers and Jun homodimers have similar, though not identical, binding preferences for variants of the AP-1 recogniti...