Biochemistry. In the article “The TRAP220 component of a thyroid hormone receptor-associated protein (TRAP) coac- tivator complex interacts directly with nuclear receptors in a ligand-dependent fashion” by

We propose that an essential factor on the origin of genetic codes is a balanced accomplishment of robustness and changeability, two antithetical, but fundamental, properties for the survival and evolution of organisms. These measures are defined as the intrinsic properties of genetic codes. An evaluation of these properties explains the structural regularity of genetic codes, estimates the order of codon reassignment in deviant codes, and predicts the most probable deviant codes that exist. The enumeration of genetic codes that could have evolved from the standard genetic code under the selection pressure on robustness and changeability strongly limits the freedom of codon reassignments. The codon reassignments of all currently known deviant genetic codes belong to this predicted evolutionary path, and they generally give the highest improvements on robustness and changeability. We propose that requests for both robustness and changeability have a strong influence on the origin of the standard genetic code (SGC) (Table 1) and its evolution to deviant codes (Table 2). These are paradoxical requests, and whereas the robustness is related to the survivability of organisms, the changeability is related to their evolvability. The investigation indicates that this is a reasonable possibility. The robustness is defined by two properties: the m-robustness, which is the unalterability of phenotypes caused by a single base mutation of codons, where the phenotypes denote any of 20 amino acids and the stop codon; and the s-robustness, which is the robustness against nonsense mutations. The changeability is the alterability of phenotypes by a single base mutation of codons. These measures are intrinsic properties of genetic codes. The elucidation of an increasing number of deviant codes (Table 2), where some codons are reassigned to different phenotypes, suggests that SGC is their ancestor (1). No general theory, however, explains the structural regularity of SGC and why it has evolved to many deviant codes. Current hypotheses, such as distance minimization of the polarity of amino acids (2–4), coevolution of amino acids and the genetic code (6), and maximum resistance against single base mutations (7), explain only partially the structure of SGC, and fail to explain the origin of deviant codes, which have occurred independently a number of times at least in ciliates (8). On the other hand, the biased codon usage was proposed as a mechanism to originate the deviant codes (1). Under a strong GC (or AT) pressure, only the codon whose third base is GyC (or AyU) would be used to code phenotypes assigned with multiple codons. Unused codons were free to change without affecting the functionality of organisms, originating deviances in the code. This did not, however, explain why deviant codes had appeared. Robustness and Changeability of Genetic Codes The genetic code is a coding table between 64 codons and 21 phenotypes. Theoretically, 21 phenotypes are assignable to 64 codons to minimally reflect the mutations in a DNA sequence on amino acid sequences, to increase the robustness against the mutations. Genetic codes with high robustness imply a low probability of change in amino acid sequences, but for a fixed mutation rate, a high reflection of mutations is advantageous for exploring proteins with new functions and for following environmental variations. Because necessary changes are unpredictable, a high average changeability between all phenotype pairs is advantageous. Graph Visualization and DNA Mutation Model. The robustness and changeability of genetic codes are calculated based on their graph representation (Fig. 1). Some simplifications are made to specify the DNA mutation mechanism against which the genetic codes should be robust and changeable. First, DNA substitution models used in phylogenetic methods (9) are unused; for example, DNA substitution rates varying among lineages, because phylogenetic analysis treats the DNA sequences that are the result of a repetitive process of change in the DNA sequence and subsequent selection. We assume that the robustness and changeability of the genetic codes are related solely to the mutation of the DNA sequence because no environmental changes can be predicted. Therefore, mutations observed in pseudo genes are most appropriate. Second, nucleotide substitution is assumed to be the most influential mutation mechanism. Consequently, insertions and deletions, which are about 10 times less frequent than the nucleotide substitution (10), are ignored. Finally, unbiased codon usage is assumed because of the wide intraspecific variations in the codon usage among genetic systems using the same genetic code. For example, the GC content on the silent base varies between 2% and 59% among species using the deviant code MNe, and the variation increases with any increase in the size of the available DNA sequence data (11). We model the bias of the mutation rate between transition pairs (GC to AT and AT to GC), which is probably the primary cause of variations in the GC content in DNA (12). The existence of some unpredictable factors, such as tRNA abundance (13), is another reason to ignore the codon usage. Indeed, such a bias is easily modeled, as the graph structure is unmodified. Initially, we use an even mutation rate because the concept of the robustness and changeability of genetic codes becomes clearer. The transition-transversion bias and GC-AT bias affect neither the graph structure nor the concept. Then, biased mutation rates explain the detailed structures of the genetic codes, and reinforce our explanation on the possible origin of the genetic codes. m-Robustness. Let si be the set of codons in node i of a graph, and ni be the number of codons in si, denoted as the size of si. Then, the m-robustness ri of node i, denoted as individual m-robustness ri, is The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked ‘‘advertisement’’ in accordance with 18 U.S.C. §1734 solely to indicate this fact. © 1998 by The National Academy of Sciences 0027-8424y98y955088-6$2.00y0 PNAS is available online at http:yywww.pnas.org. This paper was submitted directly (Track II) to the Proceedings office. Abbreviation: SGC, standard genetic code. †To whom reprint requests should be addressed.