Our results suggest that the presence of aminoglycoside-binding sites on RNA molecules may not be a useful trait for determining evolutionary relatedness. Instead, the fact that RNA molecules can bind these 'low molecular-weight effectors' may indicate that natural products such as aminoglycosides have evolved to exploit sequence- and structure-specific recognition of nucleic acids, in much the same way that lexitropsins have been designed by chemists to recognise specific nucleic acid sequences.

Several different computational problems have been solved using DNA as a medium. However, the DNA computations that have so far been carried out have examined a relatively small number of possible sequence solutions in order to find correct sequence solutions. We have encoded a search algorithm in DNA that required the evaluation of >16 000 000 possible sequence solutions in order to find a single, correct sequence solution. Experimental evaluation of the search algorithm revealed bounds for the accuracies of answers to other large, computationally complex problems and suggested methods for the optimization of DNA computations in general. Short oligonucleotide substrates performed substantially better than longer substrates. Large, computationally complex problems whose evaluation requires hybridization and ligation can likely best be encoded and evaluated using short oligonucleotides at mesophilic temperatures.

The RNA species SHR1 reacts with biocytin (epsilon-biotinoyl-L-lysine) in the presence of Ni(2+) or Pt(2+) to produce a metal-bridged complex that migrates more slowly than unreacted RNA in the presence of streptavidin (StrAv) on denaturing polyacrylamide gels. Mapping of reverse transcription pause sites identified G79 as a reactive nucleotide. G79 is near the 3' end of a 37 nucleotide core motif that is nearly as reactive as SHR1. SHR1 reacts with biocytin in the presence of Pt(2+) to yield a product that comigrates with the Ni(2+) product but that is much more stable, suggesting that the metal ion used in the reaction is present in the product, possibly linking the RNA to the amino acid. In support of this model, SHR1 shows a strong affinity for Ni(2+) in immobilized metal ion chromatography.