Abstract
Epstein-Barr virus (EBV) encodes a protein, ZEBRA, which enables the virus to switch from a latent to a lytic life cycle. The basic domain of ZEBRA is homologous to the Fos/Jun oncogene family, and both proteins bind the canonical AP-1 site (TGAGTCA). However, ZEBRA does not contain a leucine zipper dimerization domain which has been shown to be necessary for DNA binding of Fos/Jun proteins. Additionally, ZEBRA binds to sites which deviate from the AP-1 consensus sequence. Thus, it was of interest to define the domain of the ZEBRA protein required for DNA binding. We have determined by mutagenesis that ZEBRA residues 172 to 227, representing the basic domain and a putative dimerization domain, are required for specific binding to AP-1 and divergent sites. Mutagenesis of the basic amino acids 178 to 180 or 187 to 189 abrogates ZEBRA binding to all DNA target sequences. These residues are conserved in Fos and are also necessary for Fos DNA-binding activity. We have found that a Fos-GCN4 chimera and ZEBRA have different cognate binding specificities. The autoregulated BZLF1 promoter contains three divergent AP-1 sequences, ZIIIA (TGAGCCA), ZIIIB (TTAGCAA), and Z-AP-1-octamer (TGACATCA). ZEBRA binds with high specificity to ZIIIA and ZIIIB but weakly to the Z-AP-1 octamer. Conversely, the Fos-GCN4 chimera recognizes only the Z-AP-1 octamer. ZEBRA binds the ZIIIA and ZIIIB sites together in a noncooperative fashion, while Fos-GCN4 binds these sites as a higher-order complex. Additionally, we have found that flanking sequences influence binding of Fos-GCN4 to a degenerate AP-1 site (TGAGCAA). The characteristic binding specificities of ZEBRA and cellular AP-1 proteins suggest that they differentially affect viral and cellular transcription.