Oct 22d 1953
Dear Pauling,
I am playing with complex organic molecules (what I never did before!), am getting
some amusing results, and would like to hear your opinion about it.
Ever since I read the article of Watson and Crick last June, I was trying to figure
out how a long number written in a fourdigital system (i.e. nucleic acid molecule) can determine (uniquely) a correspondingly long word based on 20-letter-alphabet (i.e. an enzyme molecule). Along the lines of key & lock ideas, one could
think about different amino acids as fitting into quadrangular loops formed by four
bases in the DNA chain. But there are only five different types of such loops ie such as [diagram of types], etc. *
For a while I have thought that things can be helped by permitting 1-4 (excited) and
3-2 (excited) bonds since, indeed, this would lead to twenty different types of loops. But it would lead to a complete break up of the original
W&C model of replication process and is apparently inacceptable.
1-Adenine, 2-Thymine
3-Guanine, 4-Cytosine.
It just [occurred] to me today that another way of get[t]ing 20 different loops would be the use of triangular combinations with three arbitrary bases at vertices: [drawing of triangular combinations]
etc. 4x5x6/1x2x3 = 20.
And, indeed, the W&C model permits such shapes because of helical nature of DNA.
On the surface of a cylinder the bases will form a system of rombs with three independent and one dependent and one dependent vertices. A, B, and C
can be any of the four bases, while D is determined completely by B (D=2 if B=1, ect)
[drawing of cylinder]
Thus, there are 20 different types of such rombs, and I wonder [whether] the 20 amino
acids "vital for life" are just those which would fit into these 20 different "locks".
*) What do you think about it? It would be wonderful if it could be true!
Yours
Geo Gamow
*) The shape of these loops should also explain the fact that only L-amino acids could
be used for building of proteins by DNA.
