38 THE BIOSYNTHESIS OF PROTEINS 



for proline in the polypeptide also suggests that the coding sequences on 

 RNA for leucine and proline are closely related. 



Data which may be of crucial significance for the coding problem have 

 been obtained from comparative studies on DNA. Two DNA species of 

 slightly different densities can be separated by sedimentation equilibrium 

 in a gradient of caesium chloride (Meselson et al., 1957). Sueoka et al. 

 (1959) found it possible to separate by this method DNA preparations 

 difi^ering in their A-T/G-C ratio. It was observed that DNAs from 

 various bacteria differ very much from one another in their base composi- 

 tion but that within one species the characteristic composition is main- 

 tained even within rather small fragments of DNA (Rolfe and Meselson, 

 1959). If, as it would seem, the gross composition of the proteins is not 

 very different between the various species, the very marked differences in 

 DNA composition would indicate that the code is not the same for all the 

 bacterial species. Another possibility is that the code does not use 4 letters 

 but only 2. It is conceivable for instance that as far as coding is concerned 

 the important matter is the presence of either NH2 or OH in position 6 of 

 the purines or the corresponding position of the pyrimidines; adenine and 

 cytosine would then have the same meaning in terms of amino acid 

 (Sinsheimer, 1959). Vielmetter and Schuster (1960) were able to deaminate 

 preferentially the individual amino bases of the DNA of T2 phage. They 

 came to the conclusion that deamination of adenine and hydroxymethyl- 

 cytosine can cause mutations, whereas deamination of guanine can be 

 lethal but does not cause mutations. It is striking that the NH2 of guanine 

 is in position 2 whereas that of adenine and of cytosine is in position 6. This 

 indicates that substitution of a keto group for an animo group in position 6 

 changes the information, and since deamination of adenine gives hypoxan- 

 thine which is not found in DNA, this strongly suggests indeed that the 

 information might be written in a two-digit system : 6-keto or 6-amino. In 

 such a case, each coding unit should contain 5 nucleotides for coding one 

 amino acid. 



Another question which can be raised in connexion with coding, is 

 whether there are enough nucleotides in a genetic locus to code for all the 

 individual amino acids of a protein molecule. Benzer tried to translate the 

 distances between mutation sites, as derived from genetic recombination 

 experiments, into number of nucleotides in DNA. He found that the ele- 

 ments separable by recombination in bacteriophage T4 might be as small 

 as two nucleotide pairs, and that a 'cistron' might contain of the order of 

 400 nucleotide pairs. Pontecorvo and Roper (1956) made similar estima- 

 tions for several loci of Aspergillus and of Drosophila, and they found, in 

 each case, values ranging from 1000 to 8000 nucleotide pairs. The cal- 

 culation involves assumptions which cannot be completely checked at 

 present, and the results must of course be considered a rough estimate. 



