Definition of the sequence requirements for binding of the EBNA-1 protein to its palindromic target sites in Epstein-Barr virus DNA.

Interaction between the trans-acting DNA-binding protein EBNA-1 and cis-acting sequences in the ori-P region of Epstein-Barr virus DNA is required for maintenance of the viral plasmid state in latently infected B cells and is involved in the regulation of transcription during latency. In the Epstein-Barr virus genome, a total of 26 EBNA-1-binding sites occur within three clustered loci referred to as the family of repeats and dyad symmetry locus of ori-P and the separate BamHI-Q locus. Incubation of a bacterially expressed carboxy-terminal domain of EBNA-1 (28,000-molecular-weight EBNA-1 [28K EBNA-1]) with synthetic monomer and dimer consensus binding sites gave characteristic DNA-protein complexes in a mobility retardation assay. A similar approach with the naturally occurring Q locus confirmed that it contains two distinct but low-affinity binding sites. We then examined the precise sequence requirements for EBNA-1 binding, using a set of 30-base-pair oligonucleotides designed to contain symmetric point mutations within both halves of the palindromic target site. Analysis of all possible single substitutions between positions 1 and 10 in the consensus half-palindrome sequence revealed that positions 9 and 10 did not contribute to EBNA-1 binding and that considerable flexibility could be tolerated at positions 1 and 2. Positions 3 through 8 of the recognition site had the most stringent requirements, with transversions at these positions either reducing or eliminating binding. The relative spacing of the halves of the palindrome was also critical, since the addition or removal of 2 base pairs at the center of the sequence abolished binding. Similar results were obtained when a partially purified preparation of intact Raji EBNA-1 was substituted for the 28K EBNA-1, and the results were further supported by methylation interference studies which indicated contact points between EBNA-1 and the guanine residues at positions -8, -7, and +3 of the binding site. The three naturally occurring EBNA-1-binding loci have previously been shown to differ in their relative affinities for EBNA-1. The present study indicates that the sequence variations occurring within the family of repeats would not affect binding affinity, whereas certain base substitutions within the Q and dyad symmetry sites would be predicted to contribute to the observed lower affinities of these sites. An apparent Kd of 1.5 x 10(-11) M for binding of 28K EBNA-1 to a consensus recognition site was calculated from Scatchard analysis.