I have written to Corey that I thought it important to make an immediate and vigorous
attack on the insulin problem, even though it involves duplication of work being done
by Chibnall and his collaborators. Would you go to see him, and talk over the problem
of what might be done at once. Chibnall spoke here before the Alembic Society last
night. I judge that a good bit of what he said has been published already. The amino
acid analysis of insulin indicates very clearly that there is a molecule of sub-molecule
with molecular weight 12,000. The free amino groups have been determined by a method
(due, I believe, to Saenger) involving treatment at about pH 7 with dinitrofluorobenzene,
which couples dinitrobenzene groups onto the free amino groups, without changing the
molecule otherwise. Then the molecule is hydrolyzed and analyzed by paper chromatography,
the bright color of the dinitrobenzene group permitting easy identification of the
substituted amino acid. It is found that the two lysine residues present in the molecule
have their epsilon amino groups free, because these are coupled with the reagent.
Also there are two glycine molecules that turn up as coupled derivates, and also two
phenylalanine derivatives. It is accordingly concluded that there are four polypeptide
chains in the molecule of molecular weight 12,000 (which contains 106 amino residues),
two of these chains having glycine residues at their free amino ends, and the other
two having phenylalanine residues at the free amino end.
There are six cystine residues in the molecule, and the four polypeptide chains are
presumably held together by these sulfur-sulfur bonds of the cystine residues. These
can be reduced, but the more effective method of destroying the bonds
[page 2]
is oxidation with performic acid, also carried out by Saenger. The oxidation produces
sulfonic acid groups in place of the sulfide bonds, the residues then being cysteic
acid. An ultracentrifuge study has been made by Gutfreund and Ogsten here at Oxford,
which is reported to show that the boundary just barely moves, indicating molecular
weight slightly less than 5,000. The molecular weight of four equal residues from
the 12,000 molecule would be 3,000, and later evidence indicates that the two chains
containing glycine at the end have molecular weight 2500 and the other two have molecular
weight 3,500.
Saenger has separated the two kinds of chains. He has obtained a 30% yield - that
is, 30% of the original protein, which would be about 75% yield - of the glycine-end
polypeptide. He has carried out an amino-acid analysis, and has found that most of
the simple amino acids are in this molecule, whereas the complicated ones are in the
other polypeptide chains, with molecular weight 3,500. Each of the polypeptides seems
to contain about 26 residues, the difference in molecular weight resulting from the
difference in complexity. He has begun the analysis of the glycine-ended polypeptide
by coupling it with the dinitrofluorobenzene reagent and then hydrolyzing, and then
separating the various peptides by paper chromatography. He can identify the peptides
that come from the end of the group where the glycine residue was by their color.
He finds that the dipeptide is the substituted glycine attached to isoleucine. The
tripeptide is the substituted glycine attached to isoleucine and valine, and the tetrapeptide
is similar but with glutamic acid also. A pentapeptide has also been investigated.
Thus it is found that in this peptide containing 26 amino acid residues the first
four are, in this order, glycine, isoleucine, valine, glutamic acid, and the fifth
one is believed to be serine.
It is clear that there is already considerable progress made on the job of a complete
structure determination of insulin. However, there
[page 3]
is still a very great deal of work that remains to be done, and I do not think that
it is assured that the British school will finish the job. I believe that this is
the problem that we should begin to work on, with as much vigor as possible, under
our insulin project. I would accordingly suggest, and I ask you to talk the matter
over with Corey, that the thing to do is to get insulin, and to begin its degradation
by essentially the same methods as those used by Chibnall, with such variations as
seem reasonable to us. In particular, I think that the preparation of the 26-peptides
should be carried out in quantity, and that the two (presumably two) kinds of molecules
should be isolated in as pure form as possible. I suggest that the men working under
you on the insulin project do this, and that the material obtained be turned over
to Corey for crystallization. Chibnall has not succeeded in crystallizing either
of the 26-peptides. It might be a very big job to do this crystallization, and I
think that a specialist should have the job, namely Dr. Schroeder or some similar
man. The effort should be made to grow crystals of many different possible kinds,
in the hope that one derivative of the 26-peptide would crystallize, and in a suitable
form for x-ray examination. It might not be necessary to determine by chemical methods
the complete order of residues in the chain, because the x-ray structure determination
should provide this information.
I would think that you might want to try to split the 26-peptides into half-size molecules
by enzymatic methods. If these were made and crystallized their x-ray investigation
would be, of course, much easier than that of the 26-peptides.
With best regards, I am
Sincerely yours,
Linus Pauling:par
cc: R. B. Corey
