Translating DNA into Synthetic Molecules

At some time almost 4 billion years ago, nature likely was faced with a chemical dilemma. Nucleic acids had emerged as replicable information carriers and primitive catalysts (Joyce 2002), yet their functional potential was constrained by their structural homogeneity and lack of reactive groups. These properties rendered nucleic acids well suited for storing information, but fl awed for mediating the diverse chemistries required to sustain and improve increasingly complex biological systems. It is tempting to speculate that translation emerged as the solution to this dilemma. Translation, defi ned here as the conversion of an informationcarrying molecule into a corresponding encoded structure, enabled the expanded functional potential of proteins to be explored using powerful evolutionary methods that depend on the unique ability of nucleic acids to replicate. A small but growing number of researchers have begun to tackle a modern version of this dilemma. While proteins and nucleic acids can be manipulated using powerful molecular biology techniques that enable their directed evolution, the size, fragility, and relatively limited functional group diversity of biological macromolecules make them poorly suited for solving many problems in the chemical sciences. Ideally, researchers would like to apply evolution-based approaches to the discovery of functional synthetic, rather than biological, molecules. A solution analogous to nature’s translation of mRNA into protein could, in principle, address this contemporary problem (Orgel 1995; Gartner and Liu 2001). If a laboratory system were developed that could translate amplifi able information carriers such as DNA into arbitrary synthetic molecules, the evolution of synthetic molecules using iterated cycles of translation, selection, amplifi cation, and diversifi cation would be possible. The translation of DNA into synthetic molecules is conceptually distinct from the use of DNA simply as a tag during the solid-phase synthesis of a molecule that is part of a combinatorial library (Brenner and Lerner 1992). The latter process uses DNA to record the history of a series of chemical reactions by cosynthesizing a portion of a DNA oligonucleotide during each step of a molecule’s solidphase synthesis. As a result, the identity of compounds that pass screening can be inferred by PCR amplifi cation and sequencing of the DNA associated with a given bead (Needels et al. 1993). The resulting DNA, however, cannot redirect the synthesis of active compounds. In contrast, the translation of DNA into synthetic molecules uses the sequence of nucleotides in a strand of DNA to direct the synthesis of a nascent molecule. As a result, a complete cycle of translation, selection, and amplifi cation can be applied to the discovery of synthetic molecules in a manner that is analogous to the processes that take place during biological evolution. DNA-templated organic synthesis (DTS) has emerged as one way to translate DNA sequences into a variety of complex synthetic small molecules (Gartner and Liu 2001; Gartner et al. 2002; Li and Liu 2004). In this approach, starting materials covalently linked to DNA templates approximately 20–50 nucleotides in length are combined in very dilute solutions with reagents that are covalently linked to complementary DNA oligonucleotides. Upon WatsonCrick base pairing, the proximity of the synthetic reactive groups elevates their effective molarity by several orders of magnitude, inducing a chemical reaction. Because reactions do not take place between reactants linked to mismatched (noncomplementary) DNA, DTS generates synthetic products in a manner that is programmed by the sequence of bases in the template strand. In a series of three papers in this issue of PLoS Biology, Harbury and co-workers describe an elegant new approach to translating DNA into synthetic peptides called “DNA display.” Their approach uses DNA hybridization to separate mixtures of DNA sequences into spatially distinct locations. The fi rst paper (Halpin and Harbury 2004a) reports the development of resin-linked oligonucleotides that effi ciently and sequence-specifi cally capture DNA containing complementary subsequences. This immobilization process is effi cient enough to be iterated, so that DNA sequences specifying multiple amino acids can be routed to the appropriate miniature resin-fi lled columns during each step. In the second paper (Halpin and Harbury 2004b), Harbury and coworkers detail solid-phase peptide synthesis performed on unprotected DNA 340mers bound to DEAE Sepharose. Optimization of amino acid side-chain-protecting groups and peptide coupling conditions enabled a variety of amino acids to undergo effi cient peptide coupling to bound oligonucleotides containing an amine group. The third paper (Halpin et al. 2004) integrates the routing and peptide synthesis described above into the translation of a library of 106 DNA 340mers into a corresponding library of up to 106 synthetic pentapeptides. To achieve chemical translation, the DNA library was subjected to iterated cycles of routing and solid-phase peptide synthesis. After each routing step, the appropriate amino acid was coupled to each DNA-linked subpopulation. DNA routing was therefore used to achieve the splitting step of “split-and-pool” combinatorial peptide synthesis. The completed library of peptide–DNA conjugates was then subjected to in