A new method is reported for the synthesis of oligodeoxyribonucleotides containing 2-aminopurine residues at selected sites. This method involves protection of the 2-aminopurine ribonucleoside, reduction to the deoxyribonucleoside and standard preparation of the 5'-0- (4,4'-dimethoxytrityl)-3'-O-(2-cyanoethyl)-N,N- diisopropylphosphoramidite. The 2-aminopurine phosphoramidite prepared by this method couples with high efficiency and is stable under standard automated synthesis conditions. The presence and location of the 2-aminopurine residue is easily verified by treatment of the oligodeoxyribonucleotide with hot piperidine. The mechanism for selective hydrolysis of the 2-aminopurine residue in alkaline solution is predominantly direct cleave of the glycosidic bond.

Thymine residues in the DNA of eucaryotes may be replaced occasionally by uracil (U) or 5-(hydroxymethyl)uracil (H) as consequences of dUMP misincorporation or thymine oxidation, respectively. In this study, we constructed a series of 44-base oligonucleotides containing site-specific U or H residues and 5'-fluorescein labels in order to probe the influence of such modifications on sequence-specific DNA-protein interactions using several type II restriction endonucleases. We find that substitution within the recognition sites of several restriction endonucleases increases initial cleavage velocity by up to an order of magnitude. These results contrast dramatically with several previous studies which demonstrated that U substitution in short oligonucleotides inhibits or prevents nuclease cleavage. We propose that this apparent paradox results because the rate-limiting step in the cleavage of longer oligonucleotides is product release whereas for shorter oligonucleotides substrate binding is most probably rate-limiting. For longer oligonucleotides and DNA, more rapid release of the cleaved, substituted oligonucleotides results in more rapid turnover and a faster apparent cleavage rate. The sequence length at which the transition in rate-limiting step occurs likely corresponds to the size of the enzyme footprint on its DNA recognition site. We conclude that both U and H do perturb sequence-specific DNA-protein interactions, and the magnitude of this effect is site-dependent.