510 A. TSUGITA AND H. FRAENKEL-CONRAT 



the ainino acid ."sequence of its jirotoin coat and its symptoms in the 

 host. This was best illustrated by the mixed reconstitution experiments 

 described previously. This concept was further supi)orted by the study 

 of chemical mutants, since one or very few changes in the nucleotide 

 chain were able to effect changes in terms of biological symptoms and 

 protein structure. Conclusive proof for the direct messenger activity of 

 viral RNA has come from the recent experiments on protein synthesis 

 in a cell-free E. coli system. 



With a small virus like TMV, one should in principle be able to 

 study both the structure of the RNA and that of the corresponding 

 protein. If we assume that the linear nucleotide sequence in RNA is 

 related to the amino acid sequence of the protein coat, and that the 

 nature of the correspondence can be generalized, then TMV appears 

 ideal for the study of the "coding problem" of protein synthesis, since 

 we already possess experimental methods for the in vitro production of 

 mutants and have knowledge of the complete amino acid sequence of 

 this protein. Unfortunately, the chemical study of the RNA is only in 

 its infancy, and the localization of mutagenic chemical events on the 

 polynucleotide chain is not yet possible. Chemical studies of the protein 

 of strains, on the other hand, has yielded definitive data, as reported 

 in the preceding section. We shall now attempt to discuss and interpret 

 these experimental results in the light of current theories and experi- 

 mental facts pei'taining to the problem of coding and information 

 transfer. 



In 1954 Gamow publislicd the first theoretical approach to the 

 problem how nucleotide sequences might code for amino acid sequences. 

 The main assumptions of Gamow's scheme were the following. First, 

 that a trinucleotide in DNA determined one amino acid. Second, that 

 the trinucleotide fragment of DNA determining an amino acid and its 

 neighbor "overlapped" by two-thirds of their length. Third, that the 

 code was "degenerate," which means that more than one triplet may 

 code for the same amino acid. 



A triplet of nucleotides was assumed to code for one amino acid 

 because pairs of the 4 nucleotides gave only 16 possibilities, while 

 triplets gave 64 possibilities. The first number is not enough to code for 

 20 naturally occurring amino acids, while 64 possible combinations are 

 more than necessary and permit degeneracy. 



With the increase in our knowledge of amino acid sequences in many 

 kinds of proteins, Gamow's overlajiping scheme has been experimentally 

 ruled out. Brenner (1957) has given a proof of the theoretical impossi- 

 bility of any univci'sal ()\-crlapping triplet code. 



In 1957 Crick rf al.. introduced a non-overlapping triplet code. The 



