Dorothy  Crowfoot Hodgkin (1983) - Insulin 1983

Mr. Chairman, ladies and gentlemen. This is a great pleasure again to be here in Lindau. I find it also a pleasure to be talking to this kind of audience about my present subject. Just about a year ago a book was published called the Discovery of Insulin by Michael Bliss, which described what really happened according to the laboratory notebooks of those who took part in the original discovery. Could I have the first slides, please. There was a fantastic two years of late 1921, 1922, during which it became possible to effect this extraordinary transformation. You hardly ever, I imagine, see a child and certainly not in any civilised country in this condition suffering from acute juvenile diabetes. A condition which was observed very anciently and described as one in which the flesh flows over the bones, as you can see here. And yet a short time afterwards through injections of insulin she was transformed back. The flesh grew again. Insulin is an extremely powerful growth promoting hormone. It came about through the operations of those two over there, two young men. One a young doctor of about thirty and Charlie Best who was half way through his medical course aged 22. Banting got an idea which was not correct about how it might be possible to isolate the diabetic hormone from the pancreas, which took him to Toronto, and he managed to persuade Macleod, the Professor there of physiology, to let him try it out. And Macleod gave him this very intelligent student to help with his operations and some very good advice about how he might extract the hormone. It’s clear that Banting and Best, though they were the real workers, wouldn't actually have got anywhere without Macleod’s guidance and assistance. And it’s only that reading their account, one feels they were so terribly ignorant, they hadn’t read the literature. They made all sorts of mistakes. They covered over their mistakes in order to make it clear that they really were getting somewhere. They were rescued essentially by Macleod and Colipp, who finally made the insulin pure enough to be used in the treatment of diabetes. And yet it’s clear that the myth that they were really responsible for the isolation of insulin is correct. It wouldn't have happened without their intervention. The account is one that anyone working on diabetes reads with great fascination and excitement. Now that everyone who took part in the original battle to succeed is finally dead. So we have this extraordinary transformation, which is still incompletely understood. How does insulin act on sugar metabolism to make possible this cure when the supply in the body’s pancreas is insufficient? I find myself thinking that JD Bernal and I were about the same sort of age, he a little older, both of us a little older, 34 and 24 respectively, when we got involved in taking the first protein x-ray photographs. On the next slide I show, just for history, beautiful crystals of pepsin, which came to us from Uppsala. And on the right hand side an x-ray photograph not of pepsin, the first ones which Bernal took in Cambridge, but of an insulin crystal taken very, very much later. I was given the 10mg of microcrystalline insulin by Robert Robinson who’s picture you saw yesterday. He had been given it by Boots in England as a result of advice that he gave them. And he didn’t know what to do with it, so he passed it onto me saying: So I did, I had to grow up the crystals and on the next slide - I also made a lot of mistakes. I let the crystals dry because they were quite biofringent to my eyes, when Bernal had shown it before it was better to have them wet. But I obtained the first x-ray pattern which was totally interpretable in terms of the size of the unit cell. And in this unit cell you could put, if you measure the density, a mass of about 36,000 protein units. I should have a photograph of Bernal but the one I used to show got perished in a large meeting in Canada a year or so ago. And so on the right, I put Bernal’s letter to me from Cambridge on my telling him the news that I had got the photographs of insulin and what the size of the unit cells of the rest of it was. And everybody thinks this is a very funny letter but it’s absolutely interpretable by any x-ray crystallographer. He has looked up the paper which D. A. Scott had written the year before in 1934 on how to crystallise insulin, which I had used, which showed that it was necessary to add zinc. And that Scott had also made a cadmium and a cobalt insulin. He had taken Scott’s figures and worked out what this meant, zinc per protein in the unit cell. And the implication is try to get photographs of cadmium insulin and see if you can’t see changes in the intensities which would guide the determination of the complete structure of insulin. Well, as I say, I was 24 at the time. I hadn’t worked out any crystal structure at all. The cadmium preparation turned out to be rather impure and I couldn't grow large crystals of it at all. All I could do was get a powder pattern which looked only too like the zinc crystal powder pattern. I had grave doubts as to whether the cadmium would be heavy enough to make the changes and I thought I should start on something simpler first before I tried to determine the crystal structure of insulin. At the same time that I got, that this was happening, when I published the mass of protein in the unit cell, 36,000, I got a letter from Freudenberg, which I value greatly, which said And of course the crystal has threefold symmetry, so that the implication there is that the molecular weight is probably 12,000. As everyone knows, insulin, the chemical structure was determined by Fred Sanger and his colleagues and it was found to be even smaller, only under 6,000. The structure two chains is written up here, disulphide bonds joining them and below a scheme for the synthesis of insulin. And this general scheme was followed in three different places and actual synthetic insulin made around about 1965 in Pittsburgh, in Aachen and in Shanghai and Peking a joint operation. And later it has been found that insulin, the two chains, varies according to the creature which produces the insulin, and over the far side I have the insulin sequence grouped in kinds of creatures with the variations that occur in the aminoacids in the long chains. You can see that some regions, the beginning and end of the b-chain, are highly variable. It doesn’t almost seem to matter what is there. Whereas in other regions there’s been no change throughout time from the earliest, from the latest creature, I think man is at the top in these slides. The hagfish cyclostomes are down at the bottom. These invariant residues include the cystine residues, as you might expect, a lot of leucines, about six leucines and phenylalanine, and the aromatic residue is a highly conservative. When they change, they change usually between tyrosine and phenylalanine. Could I have the next slides, please. So we now set out to try to calculate the electron density map of insulin. And before that happened, I tried out the experiment on penicillin and I put the penicillin map up here for those who aren’t familiar with the situation because you can see what's involved. We calculate the electron density using the isomorphous replacement method suggested by Bernal, much more complicated in the protein case and first carried out by Max Perutz and John Kendrew. We calculated over the whole body of the unit cell and then draw it out in sheets and stacked the sheets together, which gives you this 3-dimensional fit. These at least are the kind of atoms you find in proteins, oxygen, carbon, nitrogen, sulphur or potassium iron. In fact the two zinc insulin structures we call it, was solved twice over, once by my group in Oxford and once in Peking and by Liang Dong-Cai and his group. It is being refined three times over, once by us in Oxford and we have stopped at a resolution of 1.5 Å, and once by Nori and Kiwako from Japan, who worked with us and who have taken it furthest to 1.1 Å. And I think Dong-Cai’s measurements end at about 1.2 Å and both of theirs are at low temperatures. This was taken two years ago of the meeting of the representatives of the three groups, comparingtheir maps, as you can see, in Peking, laying one on top of the other at a scale of 1cm = 1 Å. And you can see we were all very pleased with the observations we were making. Nori in the middle, Tom Blundell here, behind me. I have had very, very many other good collaborators of whom the most important historically are Guy Dodson, Eleanor Dodson and Vijayan from India. So I’ll now take you through the crystal structure of insulin very quickly to see the kind of molecule that finally turned up in these crystals. And so in the next slide I show two bits of the electron density map, the first map from China with the zinc in the middle and this is a little overlapping series of sections in which you can see the zinc and a group coming towards it unresolved. The resolution is 2.5 Å, and if you could read Chinese, you would see that that was histidine. And over here our latest map at 1.5 Å with all of the atoms now beginning to show as separate units and the histidine has broken up into the five peaks corresponding to its position. In both cases we are looking down the threefold axis and it’s just that one map is flipped over compared with the other. This is essentially this region here with the leucine residue at that position. On the next two slides the histidine map as seen in Nori and Kiwako’slatest map and now they have cut the electron density slab through the histidine ring and drawn the contours, I must say very close for beauty at about 0.1 of an electron per cubic Å. So you see here the separate atoms looking beautifully resolved attached to the zinc and the zinc also attached to the water and there are three round the threefold axis, making the zinc octahedral. And over here a different map, calculated with all the heavy atoms taken out of the contributions to the terms, so that you can see peaks representing the hydrogen atoms alone. And you can see this hydrogen is proteinated here, that is on the histidine attached to the zinc. Other little strays are still the necessary bits of experimental error that remain in the calculation. I must admit that not all of Nori and Kiwako’s map is as good as that. The atoms are much better well defined near the zinc positions in the crystal than they are at the edges of the molecule, where there is considerable movement certainly of the atoms within the crystal. And this smears and lowers the electron density in the crystal. Now, if from those maps you take all of the atomic positions in the protein and project them down the threefold axis, this is the picture that you get. An old picture, now the positions have been refined since, but it serves to show, the object in the unit cell is this object, which is the projection of the positions of six insulin molecules down the threefold axis. Zinc in the middle represents two zincs, one on top of the other and each connected with three histidines from the upper and lower molecules. You can see the molecules are in contact with one another along this line quite close, here’s an aromatic ring from one molecule. And there’s another one from the other molecule just about here, b-phenylanaline. They are touching across the gap there. And at this position you can see that there’s empty region water flows from one side of the molecule to the other. There are also sort of gullies across the top of the molecule into which water flows across into the region around the threefold axis. So that water is in fact continuous throughout the whole body of the crystal. The slide over here shows approximately the geometry of the hexamer, you can see there’s a roughly twofold axis along these regions. If you break the molecules apart around those axis, then you get a rather closely knit object here which corresponds to a dimer of insulin, and if you break these apart along this line, you get two separate molecules. And in solution ordinary - this in the presence of zinc the molecular weight of insulin was first measured as 12,000 via osmotic pressure measurements. Now, the next two slides show the dimers viewed perpendicular of the threefold axis, and here I have in fact put up three of them. And now this one in the middle is the one we see in two zinc insulin. And the noticeable thing about it is that the two molecules of insulin look very much the same but a little different from one another. We first noticed it at this point where one phenylalanine is across the position which should be a twofold axis between them. The two molecules are very close to one another, in fact they are hydrogen bonds between the long chains. And you can see the long b-chain here going into an alpha-helix and coming out here and the a-chain sitting on top of it. The other two are the dimer as found in the most ancient insulin so far studied, hagfish insulin, a cyclostome. And there you can see the two molecules are the same, related by crystal graphic twofold axis, both phenylalanine rings turned in this direction. And the essential pattern is the same as one of the two molecules here. We now call this molecule 1, we called it molecule 2 for accidental experimental reasons originally. But itseems to be the conserved molecule which stays the same in all the crystal structures we have observed. The third dimer is one that occurs if you have more zinc and sodium chloride in the crystallising solution. And then one molecule changes even more under the pressures of these conditions but the other one stays still the same as molecule 1 in this crystal. The slide over here shows the two separate molecules and you can see that they are very much the same but there are these little differences. The other point we always noticed the difference at is this one here where the lysine at the end of the b-chain goes straight up here and gets really lost about in the solvent on this particular molecule. The next two slides should show here just an idealised representation of the fold in two-zinc insulin, in all of these insulin molecules, with b-chain beginning straight, running into a long helix well defined, out again here. A-chain, two helixes, not so well defined, straight and second one disulphide links between the chains at the ends of the helixes down here and down here. And then this little loop in the a-chain with the other internal disulphide bond here. And over here one of the b-chains showing the general form of the b-chain as observed in this crystal with the lysine going up here and the b5 histidine sticking up on the end, at the beginning of the b-chain. On the next slide, what happens to the a-chain? Now, the a-chain in general form is represented on this side, one of the two molecules with the beginning of the helix straight, second helix, A19 bridging the gap between the top and the bottom halves. It rests on the folds of the b-chain. But over here I’ve got a different view of the a-chain looking down on the ring that contains the disulphide. And you can see that the projection of the beginning helix is quite different in the two examples. In fact they have very much the same arrangement of the side chains and everything, the difference corresponds to just turning about 30° at the beginning a-chain helix from one position to the other. And this produces a difference in the pattern of the hydrogen bonds across the 6/11 disulphide ring. You can see the different projections here and here. We can see that what produces the general fold in the insulin is the pattern of the residues along the chains some of which encourage the formation of helixes and others encourage the more extended form of chain. How they come together is illustrated by the next two slides, these show the centres of the molecule as seen in our latest maps with electron density contours representing the different groups that lie within a 6 Å sphere of the molecular centre. And the molecular centre there and there is empty but it’s packed around by all of the invariant leucines in the molecule. Invariant 6/11 disulphide bond, invariant A19 tyrosine. Invariant A2 isoleucine. But you can also see that the relative positions of these packing groups is different. This is caused by this difference in the a-chain confirmations, and the fact that this leucine is rather weakly defined, really suggests that it is being moved to a rather less favourable position than the original one. This is the molecule 1 confirmation. The next two slides show the packing in the centre of the molecule, in a different and modern view, using all modern conveniences of colour, graphics, you see here a view, a rather odd view looking down the b-chain helixes in the insulin molecule. And the arms of the helix coming out. The a-chain sitting a small compact object within the arms of the helix and all of these leucines, these non-polar residues packing in the centre between the two chains. So the chains, as you may imagine, are brought together by a tendency of like to aggregate to like, this non-polar region and then clipped into position by the disulphide formation of the disulphide bonds in the synthesis that was practiced by chemists. No doubt it’s a little bit different in the natural synthesis which passes through a single long chain precursor. And over here you get a packing view drawn which gives you the impression that this centre of the molecule is really a very tightly packed object in spite of its apparent emptiness over here. So when we look at the insulin molecule in isolation it becomes clear that one face is very flat. It’s studied again with essentially non-polar and aromatic residues. And these two faces of two molecules brought together will give us the dimer. So in the next two slides I show the groups that form the dimer. Aromatic residues bridging the gap between the two molecules and again these four aromatic groups and these two are all in contact with one another, criss-cross as if they were forming a little crystal. These other ones are on the edge of the molecule hanging out. And over here the very close packed views of the dimer presents to the outside world if you draw in slightly less than the van der Vaals radii with these aromatic residues hanging out on the outside here, two different views of the dimer. Having got the dimer, of course we must put it into the hexamer, and so in the next pair of slides I should show the hexamer. And here again you see centre zinc, the outline chain in red, the dimer forming residues are now in blue here and all the yellow ones are the hexamer forming residues. The histidines coming into the zinc and the aromatic residues packing together on the outside, four of them here, other non-polar residues. And on the right hand side a very beautiful slide which I’ve only had in my possession about three days, which came just the morning I was leaving England. A present from Andrew Moorephew who made it, and this shows you the top of the hexamer looking very well packed but with three projecting disulphide bonds on the top. And there is a sort of a groove down these lines through which water flows. So we’ve reached the stage of the hexamer. Now, the next two slides illustrate the point about the hexamer and the way that it packs in the crystal. And this is the hexamer as first we, so to speak, saw it, at low resolution 6 Å. This is a very old slide I'm afraid, we are looking down the zinc here and you see three sort of bulges on the top which actually is the top of the a-chain and three dips, this groove I was talking about here and here. And when you think how to pack the hexamers into crystals then it’s simple. You just stack them one on top of the other and the bulge below, which occurs at this point fits into the groove above and everything looks fine. And it’s only when you get to high resolution you realise that there’s one trouble about this packing. This is represented, or the consequences of it are represented over here. This is a view of three molecules seen on its side and you see the b-8 histidine, b-5 histidine sticking up here and this actually occupies exactly the same space that the b-5 histidine from the one above would occupy if something didn’t happen in order to allow the rest of the molecule to fit nicely into this groove. What happens is the histidine moves aside, the top histidine up here is in a different relative position to this one here. There it is there, it packs quite nicely against it as histidines have been seen, as the rings have been seen to pack in crystals of histidine peptides. But in doing so it changes the hydrogen bonding pattern. The a-chain swivels around this bond, takes up a slightly different position in the crystal, everything settles down again but with the molecules different from one another, as we’ve seen all the way through. And over the other side, and whether it’s a consequence of the movement of the a-chain through the crystals or separate idea of its own, the lysine b-29 sticks up and makes contact with the b-30 carboxyl group of the molecule stacked on top of the crystal. So now we come to the last stage, we have these pillars of insulin molecules, we can pack them like pencils in the crystal and we only have the possibility of just moving around this pencil how we fit them together. And this is done in order to again bring like to like together between the molecules in the crystal. As I showed earlier, the outline aromatic groups make contact but also so does an enormous number of different ionic residues. In the next slides I show this ionic layer with the b-30 carboxyl group in contact with its own a-1 glycine LH3+ but also with this b-29 lysine from the molecule packed above it in the crystal. And at the same time the arginine from another hexamer over here managed to come into this ionic region. The slight complication of this slide is that the arginine is in two minds. Whether to make a contact over here with the a-21 carboxylic acid or over here with the a-17 carboxylic acid. And there’s also a certain amount of disorder, this other lysine residue. But, as you can see, an extraordinary number of different ionised groups come together within the layer between the hexamers in the crystal. Over here a rather pretty picture illustrating the water over the surface of the hexamer. It’s a slightly misleading drawing because the more diffuse an atom is, the larger it appears on this drawing, illustrating an uncertainty in its position. So the tightly bound protein atoms in red are well defined, the ones on the outside of the molecule are the loose chains on the surface. Now, the next two slides take me over to one of the crystals I mentioned earlier. And this is the four zinc insulin, which occurs just when 6% of sodium chloride is in the crystallising solution. It was grown by Schlichtkrull in Copenhagen, in order to have very regular crystals to administer to patients to produce a slowly absorbed insulin. And to our surprise, although the crystals look just the same and have very similar lattice constance, they turned out to be quite different inside. And the hexamer here is one in which there are four zincs per six insulin molecules. On the next slide I think you can see more clearly the difference, again one molecule stays steady, molecule 1, three histidines come into the zinc. The b-chain is extended, helical extended. But in the other one the b-chain helix is carried further, the b-5 histidine has been carried into quite a different place. It connects with the zinc and the b-10 histidine losing the zinc at the centre just turns around and makes contact with it. So that there are now four groups, two histidines and two water molecules round this zinc atom and four zinc atoms in the crystal to six insulin molecules. Now, some confirmation for our idea that it’s this trouble with the histidine packing that produces the change in position of the a-chain. You can see this is the position in two zinc insulin with the two b-5 histidines having different relative positions but being in fact in contact with one another. And over here there’s no b-5 histidine interfering with the position of the second one. So in the next two slides, when you look at the whole molecule up here are these two different positions of the beginning of the a-chain in two zinc insulin, but in the four zinc insulin this doesn’t have to happen. So the a-chain helix has maintained the same relative position in relation to the ring next door. The isoleucine has the same situation in space, instead of a quite different one as here. Other things have happened of course, owing to the b-chain movement which alters the positions down here. And over here is a rather complicated map taken from Cyrus Chothia’s calculations of how all the movements happen all the way through the molecule, which requires, because at one end there’s a histidine b-5 interfering and having to move. Right at the other end of the molecule a phenyl group has swung around into a different position. Now, very quickly I want to go over to one or two points about the biological activity of insulin. On the next slides this is still one more crystal, a cubic form of pig insulin, which was found first of all in 1926 by Abel when first he crystallised it, and has quite a different arrangement of the molecules in the insulin zinc-free pig insulin two identical molecules, another symmetrical dimer. That will come into the story a shade later. But on the next two slides we go back to the actual sequences of these insulins and the positions of the invariant residues within the molecule which suggested certain ideas about the biological activity to us. That this end of the chain didn’t matter. That the aromatic residues B-24 mattered. This was shown by the Chinese particularly taking off the whole of this chain enzymatically, and then building back down here, if you remove everything almost inactive. When you move back to B-24 activity returns. One of the last crystal structures I want to show a little off is a very interesting molecule in which the whole of this chain, des-pentapeptide, the first groups have been removed which is still biologically active. The other part that seems to be concerned in the activity is this front A-chain region here. The next two slides just illustrate this actual, there’s the A-chain face over here with the invariant residues in it, which stays the same even through these transformations. And over here the part of the chain that can be removed and played with and at the top here just the actual change that occurs in human insulin if you want to transform pig to human insulin, you only have to change the end residue. Now, I would like to show, just before I give the very last few slides, two or three interesting points. If I may, perhaps I should tell this story first, this is for Professor Gilbert,of course. It is possible to take the end of the b-chain off of pig insulin and replace it by threonine, alanine to threonine, and then you have human insulin. But you can also carry out the synthesis through bacterial engineering as was first done I think in Prof. Gilbert’s laboratory, but now can be done and has been done in various different places with various different strategies. This is the human Eli Lilly insulin made by making the two chains separately and then joining them together. So very beautiful crystals, not absolutely the same to our eye, which is an interesting crystallographic phenomenon. The actual habit over of the Eli Lilly insulin is not one I’ve seen naturally though it’s one that is recognisably that of a rhombohedralpolar crystal. And probably implies slightly different impurities in the two preparations which may account for some of the small troubles in comparing them. But essentially they are both human insulin and the last two slides just illustrate this crystal graphically. Two identical x-ray photographs and over there a different map over the two different ends of the b-chains, because of this complication I’ve been talking about, which show that when you leave out the end of the B-chain from the calculation in which you’ve inserted all the other atoms, you see evidence that there is three in the crystal and not pig insulin the coordinates of which were used to obtain the primary solution for the crystal structure. Now, the interesting thing that we have come across, of course, is the way the insulin molecule varies or doesn’t vary, as the case may be the parts that don’t vary, as you change its surroundings, and this matters in relation to the actual sequence of reactions of the activity of insulin. Because the insulin receptor which many groups are working on, the isolation, is now known to be a large molecule, something like the immunoglobulines. Of a molecular weight about 350,000, two different sized subunits, 90,000 and 130,000 and a little object of 6,000, you can see, must occupy a small region in the folds of this sort of molecule. So you might expect that different residues in different parts of the molecule may affect the activity. Now, we have here just got comparisons between how much insulin changes in different conditions. This is the two pig insulins seen one in two zinc insulin and the other in the cubic insulin. This you might see, most of the molecule stays sensibly the same and the only thing that markedly changes its position is the beginning of the b-chain which hangs out from the body of the molecule in the two-zinc insulin molecule. It’s clearly free to pack, it isn’t involved in the activity in any serious way. Now, this one is the hagfish insulin, a- and b-chains. And, as you can see, in spite of very considerable residue variation over a long evolutionary time, the insulin main confirmation is maintained over most of the b-chain and over the a-chain. The last one, I get all these beautiful things from York, where Guy and Eleanor Dodson are now working. This is a very interesting little, des-pentapeptide insulin. And two different crystal structures of this have been solved. One in China with the assistance of David Stewart from our country, one in York with the assistance of Bi from China, two different crystals and you get essentially the same answer again. Part of the insulin structure is wholly conserved although this doesn’t form zinc complexes or anything like that anymore and the whole end of the b-chain is taken off it. It’s still biologically active as a monomer which corresponds to the general view that it is the monomer that is active insulin. I should stop there and thank you very much, I'm sorry if I go on too long. The trouble is this story grows in the telling. I have to make one interesting point in relation to the next lecture, having heard a word. Insulin reacts first with the receptor, and then it seems that they both pass through into the cell and are concerned with phosphorylation and with the sequence of phosphorylation reactions. End of that story, thank you.

Dorothy Crowfoot Hodgkin (1983)

Insulin 1983

Dorothy Crowfoot Hodgkin (1983)

Insulin 1983

Comment

According to the Statutes of the Nobel Foundation, each Nobel Laureate has to deliver a lecture describing the work that has been rewarded by the Nobel Prize. The published texts of these Nobel Lectures constitute a valuable document for historians of science, culture and politics, since it often happens that this is the only time the Laureates officially tell the stories behind the rewarded discoveries, inventions, etc. But since the start of the Lindau Meetings in 1951, many Nobel Laureates have adapted their Nobel Lectures to a somewhat younger audience and told their stories over and over again. This is so for Dorothy Crowfoot Hodgkin, who first came to Lindau in 1970 to give a talk on “Structure of Insulin” and now in 1983 for the third time gave very much the same talk under the heading “Insulin 1983”. Since the major part of the audience in 1983 was not present in 1970 (probably being still in school), this repetition of the story of the structure of insulin mainly reached new ears and minds. But through the discovery of these historic Lindau tape recordings, we can now hear the story again and we can follow the changes that have been made, the stress put on different parts of it and also the change in the speaker, who in 1983 became 73 years of age. My own impression is that the speaker has realized the composition of the audience better than in 1970 and now gives a talk which is easier to follow and therefore more enjoyable. We know that Dorothy Crowfoot Hodgkin was still active in trying to understand not only the structure of insulin, but also the way it functions in the living organism, and some of the new realizations are also pointed out in the lecture. Finally, I was interested to hear her strong support of the much disputed 1923 Nobel Prize in Physiology or Medicine to John Mcleod, who shared it with Frederick Banting for the discovery of insulin, while Banting’s younger coworker Charles Best was left out. As Dorothy Crowfoot Hodgkin explains, without Mcleod guidance, the other two would never have discovered insulin at all!

Anders Bárány

Cite


Specify width: px

Share

COPYRIGHT

Cite


Specify width: px

Share

COPYRIGHT