Logo IMG


Ode to the Code

Brian Hayes

Code On, Codon

Solomon W. Golomb of the University of Southern California, who was a central figure in the first round of speculations about the genetic code, has summed up the spirit of that era: The approach taken in those days was to ask, "How would Nature have done it, if she were as clever as I?" Now that we know how nature has done it, you might think that the period of freewheeling conjecture would be over, but I am pleased to report that there is no lack of adventurous ideas about patterns and structures in the genetic code. Here are just a few of the ideas in circulation.

One of the themes of the earlier period was the need to find some compelling relation between the numbers 64 and 20. And this quest had spectacular successes: In at least two schemes, the 64 codons could specify exactly 20 amino acids, neither more nor less. The mathematics was so beautiful, it was hard to believe nature would pass up an opportunity to make use of it. Pierre Béland and T. F. H. Allen of the St. Lawrence National Institute of Ecotoxicology in Montreal argue that nature did not miss the opportunity. They propose a primordial genetic code in which information was read from both strands of the DNA at once, and all messages were palindromic, so that they could be read in either direction. Under these conditions, meaning can be assigned to only 20 of the 64 triplets.

A double-stranded translation system may sound outlandish, and yet there are hints that the "antisense" strand of DNA may be more than just a placeholder. Jaromir Konecny, Michael Schöniger and G. Ludwig Hofacker of the Technical University of Munich point out that a rough symmetry of the genetic code creates a kind of antigene opposite every normal gene. Wherever the sense strand calls for a hydrophilic amino acid, the antisense strand (read in the opposite direction) is likely to code for a hydrophobic one. It's even possible that some of these antisense pseudogenes are transcribed in vivo. William F. Pendergraft III and six colleagues at the University of North Carolina at Chapel Hill have recently detected immunological reactions to one such antisense protein.

More generally, there is growing recognition that the genetic code may encompass more information than just the simple mapping from codons to amino acids. Synonymous codons may not always be completely equivalent. It's certainly true that codon frequencies are not random or uniform. Among the several codons that specify a given amino acid, some may be common and some rare, and these usage biases can vary both within and between genomes. The biases probably help to regulate the rate of protein synthesis: If the transfer RNA that matches a codon is rare, then transcription of genes including that codon will be slowed. For some proteins there is evidence that such pace-setting codons help ensure correct folding of the amino acid chain.

Another fertile area is the search for symmetries and patterns in the genetic code. The standard table of codon assignments derives from the obvious representation of the triplet code as a 4×4×4 cube. Several authors, observing that 64 is equal not only to 43 but also to 26, suggest organizing the codon table as a six-dimensional (2x2x2x2x2x2) hypercube. A mutation is a movement from one vertex to an adjacent vertex in this structure. The geometry is intriguing, and there are interesting connections with Gray codes and even with the I Ching, but I'm not so sure that biologists will find the concept useful.

Click to Enlarge Image

Not every interesting idea takes the form of a paper in the Journal of Theoretical Biology. Another quite different geometrical interpretation of the genetic code has been presented to the world in the form of a design for a toy. Mark White, a physician and inventor in Bloomington, Indiana, discovered that the genetic code can be represented succinctly on a dodecahedron (a solid whose surface consists of 12 pentagons) or its dual the icosahedron (made up of 20 triangles). Each face of the dodecahedron is labeled with one of the four nucleotides, each of which appears three times. Any grouping of three adjacent faces, read in the right order, generates the appropriate amino acid. White has made prototypes of toys that incorporate this design. He observes that the icosahedral model is closely related to the very first proposal for a triplet genetic code, the "diamond code" devised in 1955 by George Gamow. This neatly closes the circle and takes us back to the beginning of the story.

© Brian Hayes

comments powered by Disqus


Subscribe to American Scientist