The TaqI 'star' reaction: strand preferences reveal hydrogen-bond donor and acceptor sites in canonical sequence recognition.

TaqI endonuclease recognizes and cleaves its canonical sequence, TCGA, with complete fidelity under standard conditions. In the presence of some organic solvents, TaqI endonuclease introduced additional single-strand and double-strand cuts at sequences termed TaqI 'star' sites. Using 'middle-labeled' DNA, the relative rates of cleavage of each strand were simultaneously determined for several star sites. These star recognition sequences differed from the canonical sequence by a single base, and all potential star sites were either nicked or cleaved. Star sites within the middle labeled substrate represented ten of the twelve possible star sequences for each strand. For each group of identical star sites, one strand was consistently preferred for cleavage. Based on these preferences, a model for TaqI recognition of the TCGA sequence is proposed. According to this model, sequence discrimination is mediated by eight hydrogen bonds formed between TaqI and the cognate nucleotides within the major groove.