It was now clear that both types of nucleic acids were unbranched linear 3': 5'-linked polynucleotides, and that individual members differed from one another in molecular size and in the sequence of nucleotides present in them. Accordingly, methods both for synthesis and for sequence determination would in due course become important. We did indeed carry out some work on stepwise degradation of polynucleotides by chemical means, but it quickly became clear that such methods would be tedious in the extreme, and probably inaccurate. It seemed to me that little real progress was likely to be made until enzymes could be found which would chop up polynucleotide chains in a manner analogous to the specific enzymes which were known to attack polypeptides, and which were being used with such success in the case of insulin and other proteins by F. Sanger. Very few nucleotidases of such a nature were known at this time, and it seemed to me wisest to leave the further development of sequence determination to men like Fred Sanger, or to one of the numerous young men who were going out from my laboratory to build up new research groups in the nucleotide field in many countries. On the side of synthesis A. M. Michelson and I showed that the synthesis of oligonucleotides was quite feasible by synthesising dithymidine-3':5'-dinucleotide. I confess that I have never been much attracted to the kind of repetitive procedures involved in synthesising either polynucleotides or polypeptides, so I left further developments in synthesis to my 'offspring' including such men as A. M. Michelson, F. Cramer, H. G. Khorana and C. B. Reese. The spectactular results which later emerged culminating in Khorana's synthesis of a gene are well known. There still remain serious problems to be solved, however, even in the oligonucleotide field; especially is this true of the ribonucleotides, where the presence of cis-vicinal hydroxyl groups in the ribose residue presents a difficult problem for the synthetic chemist.
In the course of our studies on methods for the phosphorylation of nucleosides (i.e. nucleotide synthesis) we developed and used as the method of choice what has come to be known as the 'triester procedure', i.e. one in which one uses an active diester of, for example, phosphorochloridic acid as phosphorylating agent and subsequently removes selectively one or two ester groups from the resulting triester to give either a di-or a monoester of phosphoric acid. Although demonstrated by us to give good results in synthesising a dinucleotide, it appeared for some years to be superseded by other quicker but less selective routes based on direct diester formation. I find it sometimes a little difficult not to say 'I told you so!' when I see the triester procedure now being belatedly adopted for oligonucleotide syntheses as the method of choice where selectivity as well as yield is of importance.
The double helical conformation of DNA was advanced by Watson and Crick about two years after our clarification of the chemical structure which made it possible. I and my colleagues played no part in the development of the Watson-Crick model, largely because our interests at the time were essentially chemical and we really gave little thought to the physical conformation of the polynucleotide molecules in nature. A secondary reason may have been the almost total lack of contact between physics and chemistry in Cambridge - a lack of contact which is all too common in universities. It is true that before Watson and Crick were allowed to publish their paper Sir Lawrence Bragg, who was head of the Cavendish Laboratory, insisted that D. M. Brown and I should approve their model (which we did!). The reason for this insistence (which is mentioned in Watson's book The Double Helix but with no explanation) lay in the fact that only a year or two earlier Pauling had published the alpha-helical structure for a protein. Pauling sent copies of his manuscript both to Bragg and myself, and I well remember Bragg coming over to see me in the chemical laboratory (for the first time since my arrival in Cambridge) and asking me how Pauling could have chosen the alpha-helix from among three structures all equally possible on the basis of X-ray evidence, and all of which he (Bragg) had indicated in a paper with Perutz and Kendrew. He was quite shattered when I pointed out that any competent organic chemist, given the X-ray evidence, would unhesitatingly have chosen the alpha-helix. It was a direct consequence of this that he decided that no nucleic acid structure based on X-ray evidence would go out from his laboratory without it first being approved by me!
When I saw the Watson-Crick model that day in their laboratory, I at once recognised that, by a brilliant imaginative jump, they had not only solved the basic problem of a self-replicating molecule, but had thereby opened the way to a new world in genetics. Maybe it is a pity that the physicists and the chemists were not closer at that time, but, even if they had been, we might at best have enabled the physicists to make the imaginative jump a year or so earlier, but probably not much more. Arising from our synthetic studies on simple nucleotides, I had long since learned that the nucleosides were effectively flat, and their stereochemistry indicated that, linked together by phosphate residues, they must form some kind of helical structure. We also knew, from X-ray studies of some of our materials by W. Cochrane and his group, that nucleosides and, indeed, their parent pyrimidines and purines were strongly hydrogen bonded. I recall telling Astbury of these views as early as 1947. I knew, of course, that the DNA molecule must contain some kind of code if it were to transmit hereditary characteristics, but, save in a very desultory way, I never considered the matter very seriously. Thus, although well aware of Chargaff's analytical findings in 1950 and 1951, I never gave any serious thought to their possible significance as part of a physical arrangement of DNA which could provide the basis of the genetic code. All of which is just an illustration of the way in which scientists are very often blind to matters which happen to lie outside their own specific field of interest. A further striking example of this last point is to be found in our work on organic phosphates, a facet of our nucleotide coenzyme studies. One of the problems we had to face quite early in our work aimed at coenzymes, most of which were unsymmetrical pyro- or triphosphates, was that the initial phosphorylation of a nucleoside by our normal method (using dibenzyl phosphorochloridate) gave rise to a triester, from which one esterifying group had to be selectively removed under very mild conditions which would not damage other parts of the molecule. One of our most successful devices for this purpose was to make use of (a) the electrophilic character of the CH2-grouping in the benzyl residue and (b) the fact that diesters of phosphoric acid are strong acids, i.e. they have very stable anions. Thus, when a triester of phosphoric acid containing a benzyl (or, for that matter, an allyl) group is treated with a nucleophile such as a tertiary base, or an anion such as chloride or iodide, the nucleophile attaches itself to the benzyl or allyl group liberating the anion of a diester of phosphoric acid. The more powerful the nucleophile and the stronger the diester of phosphoric acid produced in the reaction the better it goes. Now, of course, ethylenic compounds are nucleophilic, but rather weakly so, and accordingly were not of any value to us in the nucleotide work. But, around 1952, my colleague, F. R. Atherton (now with Roche Products Ltd and one of the world's experts on organic phosphates) decided to explore their use in this reaction. In the course of his work he found that geranyl diethyl phosphate was quite stable, but geranyl diphenyl phosphate was unstable and underwent cyclisation to give a mixture of cyclic terpenes, diphenyl phosphoric acid (a very strong acid) being expelled. What had happened was, of course, that the isolated double bond in the geranyl residue was sufficiently nucleophilic to attack the allylic carbon intramolecularly. This was quite interesting, and was in accord with expectations. What we failed to realise was that he had, in fact, discovered the way in which nature makes carbon-carbon bonds! Our interest lay in the behaviour of the phosphate - not the geranyl residue. During the 1950s there was an increasing interest in the mechanism by which nature synthesises compounds like terpenoids and steroids containing recurring 'isoprene units' in their carbon skeleton, starting from acetate. The discovery of mevalonic acid as an intermediate by Karl Folkers and his group in 1956 gave a big fillip to research in this area, and several of my friends -J. W. Cornforth and G. Popjak in England, Konrad Bloch in the United States and Feodor Lynen in Germany - were deeply involved. I paid no more than passing attention to this field, being absorbed in nucleotide coenzyme studies as well as work on vitamin B12 and aphid colouring matters. In August 1958, however, I went on holiday to Lugano with my family and one sunny morning, having just swum in the lake, I was sitting on the hotel terrace with my wife having a Campari when a somewhat decrepit car drew up close by and much to our astonishment the Bloch family emerged from it. Greetings having been exchanged, they joined us on the terrace for a gossip. While thus engaged Konrad said he thought it might interest me to know that there appeared to be a phosphate group in the precursor of the terpenes, which was produced in nature from mevalonic acid. I said to Konrad 'I'm not surprised, but I would bet that the intermediate will be a pyrophosphate' and left the matter at that. It was only later that I remembered Atherton's work, and realised that it really held the key to the problem. Subsequently, of course, the intermediate was identified as iso-pentenyl pyrophosphate and this led to the beautiful work on terpenoid and steroid biosynthesis carried out by Bloch, Cornforth and Popjak and by Lynen.