Genetic Code: What Nature Missed

The genetic code encodes 20 amino acids in almost all living cells; the only known exceptions are a few organisms that use selenocysteine [1] or pirrolysine [2]. In order to acquire chemical properties not represented in this set, proteins must be modified post-translationally, be made non-ribosomally, or use co-factors. In a series of papers, Peter Schultz’s group has described how they carried out a directed expansion of the genetic code to include unnatural amino acids in living prokaryotic [3,4] and eukaryotic [5,6] cells. The most recent of these papers [6] reports the incorporation of five artificial amino acids into the yeast genetic code (Figure 1). In this study, Chin et al. [6] selected mutant tRNA–tRNA synthetase pairs, based on a suppressor tyrosyl-tRNACUA synthetase from Escherichia coli, that were evolved to incorporate unnatural amino acids in response to amber stop codons in living yeast cells. Their laboratory developed a similar approach in bacterial cells only two years ago [3,4], and it is exciting that they have already extended this methodology to eukaryotes. Chin et al. [6] began by choosing a suppressor tRNA and tRNA synthetase pair ‘orthogonal’ to yeast — defined as one in which the components do not crossreact with any part of the yeast translation machinery. They then generated a library of 3.2 million tRNA synthetase mutants with possible altered amino-acid specificity. Mutant cells that recognized a desired unnatural amino acid in response to amber stop codons were selected in a three-step process based on the GAL4 gene and the HIS3, URA3 and lacZ reporter genes. This procedure was repeated for five unnatural amino acids, each with distinct chemical properties and potential experimental applications. The incorporation of amino acids with desirable properties in defined positions of proteins has several applications for determining protein function and structure. Special amino acids can be inserted to improve site-directed mutagenesis studies, to create biophysical probes, or to introduce photoreactive side chains to study reaction mechanisms (for review see [7]). The incorporation of new amino acids can also provide a way to enhance the properties of proteins: by allowing new chemical characteristics not available from the canonical set of amino acids, it is possible to design proteins with new functionality. Among other applications, cells with these new proteins might prove useful in bioremediation; cells with artificially evolved proteins using only the twenty natural amino acids are already being developed for this purpose [8,9]. Several groups have developed methods to incorporate artificial amino acids into proteins in vitro and in vivo (see [10] for a review). In most of these approaches, however, the tRNAs must be chemically charged with the desired amino acid before use. This is particularly troublesome for in vivo studies, where the charged tRNAs must be microinjected into individual cells. A few studies have addressed these problems recently, and advances in the field are accumulating [10]. The approaches described by the Schultz lab [3–6] and other groups [11–13] avoid these problems by incorporating into the cell all the components necessary to exploit the unnatural amino acid, namely a new aminoacyl-tRNA synthetase–tRNA pair. Impressively, Mehl et al. [4] also recently developed a strain of E. coli that is capable of both coding for and autonomously synthesizing an unnatural amino acid. Strains of yeast with this capacity cannot be far behind. The development of these strains will also allow investigators to study the natural mechanisms of genetic code evolution. A variety of organisms have naturally reassigned some codons during evolution [14], and many theories have been proposed to explain how and why these reassignments occurred [14,15]. Experiments such as those performed by the Schultz lab provide a new way to artificially reassign a codon and to observe its evolution and fitness consequences for the organism. Once the researcher-induced selection on the tRNA–tRNA synthetase pair is relaxed, will cells quickly lose the ability to utilize the unnatural amino acid? Or will it be incorporated into other proteins? Will the cells begin to convert amber stop codons to ochre or opal to prevent readthrough? How many generations will it take for these types of global changes to be observed? And what costs are involved? Answers to these questions may be found using experiments similar to those performed by Richard Lenski’s lab, in which E. coli cells are grown under controlled conditions for thousands of generations, and then their fitness can be compared with those of the original cells [16,17]. Experiments of this kind using E. coli or yeast with an expanded code could determine if an extra amino acid confers an advantage to the cell, or if the current machinery required is too cumbersome to maintain. The use of different growth media and conditions could determine under which conditions the modified cells show more plasticity than wild-type cells. One extension to the methodology, as reported by Anderson and Schultz [18], is the development of other orthogonal tRNA–tRNA synthetase pairs in E. coli, using the opal stop codon and the four-base codon AGGA. Improvements in these techniques will soon permit the simultaneous incorporation of more than one unnatural amino acid into the code, allowing even further embellishment of the diversity of chemical Current Biology, Vol. 13, R884–R885, November 11, 2003, ©2003 Elsevier Science Ltd. All rights reserved. DOI 10.1016/j.cub.2003.10.052