The first thing we did was to, I had by this time introduced two dimensional; NMR into the lab, I'd recruited, against the opposition of some of my cryptographic colleagues, NMR structure determination, got a man called David Neuhaus to come here. In the lab... I wanted to do this for many years while Sydney [Brenner] was director but he couldn't bring himself to do it, he didn't really... himself feel the need for it. Also, the crystallographers object to that, they used to say... well, can you solve a structure more than 20,000 daltons? And of course, you couldn't at that time. They said, 'Well, that's not of interest to us.' But once I was director I found some allies, Tom Creighton for example wanted to know the structure, different folding forms, intermediate forms of BPTI used in the folding of a protein, so I had a bit of support in the lab I couldn't one sidedly just fix the... the lab is... well I was director but acted more like a chairman. So I did recruit structural studies so we had David Neuhaus in the laboratory and so we decided to solve a structure of a zinc finger.

Now, at that point I made an unwise decision, I said, 'Look, we're not going to need to know the structure of a zinc finger, we need to know how the zinc fingers are joined together by this linker.' It seems to be that it would be highly flexible and we didn't know that, so I suggested he solve the structure of a two-zinc finger construct which we made, two-zinc fingers is Y5. Unknown to me at the Scripps Institute Peter Wright had... was working on the structure of the single-zinc finger, so they published first and the NMR structure of a zinc finger showed very clearly what we had proposed, that the zinc was holding the two parts of the protein together and the two cysteines came off a beta sheet of two turns, two beta strands and the two histidines came off a helix, three amino acids apart, we could predict that, I could see that something like that was about to happen. And David Neuhaus republished later showed that the two-zinc fingers were totally independent solution, you could solve them as two separate fingers and you could work out their different orientations from the NMR and that covered the whole range, so what he proved was the two fingers had identical structures, to within less than a fraction of an angstrom, though the sequences were somewhat different. Of course, they have seven conserved residues and so it was I had basically predicted, zinc bound to the two cysteines, the two histidines and the three hydrophobic acids formed the kind of core, structural core, so... and Neuhaus showed that the linker was flexible so therefore it could end position itself on any sequence of DNA.

And then we also showed by... we'd also... before that I've got to say we'd already showed by EXAFS that the environment of the zinc was two cysteines and two histidines, we knew that from EXAFS, that happened in Manchester. We did this in Daresbury, by EXAFS, but then we still didn't have a crystal structure of the zinc finger with DNA and by this time a lot of people had joined in this field. Jeremy Burke had proposed a structure based upon the idea of three conserved hydrophobic groups and cysteines and histidines and based upon the structure, pherodoxin, Jeremy was at Johns Hopkins, proposed a structure which turned out to be basically correct in outline and the NMR structure showed that he was correct, so it's not often that you get a theoretical direction but he used the three conserved hydrophobics and showed that they packed... that preceded... the NMR work. I must say, by this time, a lot of people had joined in and Carl Pabo who was then the MIT, and there was one that Chris Lock was working on, transcription factors together with a young man called Nikolai Pavletich, a student who has then gone on to great heights, solved the structure of the... complex of the three-zinc finger peptide from a gene called ZIF268 which was a DNA binding domain. The DNA binding domain was ZIF268 which was an early response gene and showed very clearly that the three fingers bound in the major group of the DNA, they wound and wound the major group and that each finger bound to three bases and... three successive bases so that the three fingers bound to three base pairs. And so it's one finger recognises and binds to three base pairs. These were all on one strand of the DNA and that the fold, of course, the fold was a tremendous mark that the fold was NMR work, that was correct. Now, the three sites on the helix is the recognition helix, and positions minus one, position three and position six on the helix, are where the amino acids are, so you then can of course generate diversity by putting different amino acids onto the three positions. But the structures showed that it couldn't be alphabetically, because there isn't enough, and so we... so for example, the structure showed that although one... if we had a guanine it could be recognised, a guanine in the DNA sequence it could be recognised by either arginine or by a lysine. That was if you had them in positions one or three of the triplet, but if you had position two in the triplet then you had to use a histidine but there's not enough room for a long residue to get in.