So Yen Choo started using... perio-phage display to try to build up... a repertoire of sequences of zinc fingers which would bind given sequences of DNA and the way we did that was to fix... to take the Z1F268 which we based it on and change the middle DNA triplet into the desired binding site. And so, we had two fingers, fingers one and three, binding to the DNA sites one and three and the DNA target, that was put into the DNA sequencing... the DNA binding site, subsite that should be, into subsite and then we varied the DNA sequence just because we didn't know how representative the crystal structure was, in all nine positions of the recognition helix. And we found miraculously that all the selected binding sites all basically we got consensus of binding in the positions minus one, three and six which was rather very gratifying. But you see you have to know a structure. Now these libraries weren't so big as we were varying... we were varying nine positions and our libraries they only had, at that time, 107 or 108 members. But later on, we built bigger libraries and varied a smaller number of sites, only four sites, and not just positions minus one, three and six, but minus one, two and six. Minus one, two, three and six and I'll come to say how site two was involved.
But in the meantime, in the laboratory, Louise Ferrel who was the assistant of Daniela Rhodes and they started working on zinc fingers and Louise tried to crystallise a zinc finger peptide. There were only two-zinc fingers, it's from Tram Track which is a gene in drosophila which was being worked on by Andrew Travis in one of the other divisions. And Louise managed to crystallise this, binding to DNA and the structure was solved and published in Nature in 1993. Which was about 18 months after the pair position paper. Now what this showed were two things. The one was it broadened the amino acids which would bind cells: for example, it showed that you could... that if you had an adenine which wasn't represented in the Z1F268 structure, then you had to have an asparagine to bind to it and so it extended the code. But the other thing it showed very clearly, something we had suspected from our phage display, but we found with our phage display experiments, there were certain positions which we couldn't determine because it was clear they were being coupled to something else. And so, the structure not only gave us some new elements in the stereochemical code but solved the mystery. What the structure showed very clearly was that, from position two in the recognition helix there was an interaction with a site on the second strand of the DNA which was part of the... sorry base of the second strand of the DNA which was part of the subsite of the preceding finger. So in other words there was a... it was an interesting business because that site, that DNA position had to... was part of a double helical and so its partner had to be recognised by the preceding finger. So what this meant, a very important discovery, it meant that although the fingers were structurally independent they were synergistically involved and that made a big difference. But there was still nevertheless... it was still one amino acid per one... one amino acid per one DNA... sorry to one base. Now I didn't say this before... but in the many transcription factors being worked which all... where there's homodimers or variant heterodimers it isn't like that. The interface between the DNA recognition helix and the DNA is more complex. In most cases there's one amino acid interact with more than one base and conversely one base can interact with more than... more than one amino acid, so we haven't got this simple stereochemical code, we've got zinc fingers, even with this extra interaction to the second strand it's still one amino acid per... per base and vice versa and one base per amino acid. Now this meant, and it was at this point, that I decided that really this is a marvellous system. We should be able to develop it for intervening in gene regulation. Particularly as once we'd worked out the stereochemical code, so it was a very important discovery, the second crystal structure.