Oliver Smithies:
If a modern person were to decide to understand the availability of information regarding a scientific topic, or even an ordinary topic, one would go online and dial up Google and do a Google search with a given topic. In that way, one could look at a vast amount of information and publications if one went in the right form related to the subject. In 1953, when I went to the University of Toronto to start a new job, my immediate boss was Dr. David A. Scott. Dr. Scott was one of the early scientists in Toronto working at Connaught Medical Research Laboratories to start production of insulin as a commercial product. Insulin was discovered in Toronto by Banting and Bess in the early 1920s. And Dr. Scott crystallized the insulin in the form of a zinc salt, which turned out to be very insoluble. The zinc salt crystals therefore could be used as a slow release form of insulin to help treat individuals lacking insulin, type 1 diabetics. When I applied for a job in Canada, he was a person who was willing to have me join him. He worked by himself then, in a small lab, and did experiments primarily in the morning, and then in the afternoon went playing golf with his friend, whose name perhaps will come back to me later. But so when I asked him what I might work on, he said “Well, you can work on anything you like, Dr. Smithies, as long as it has something to do with insulin,” which was his primary love. So I did the equivalent of a Google search, as possible in 1953, and the result is contained in a stenographer’s notebook that I still have, which is labeled O.S. Planning Notes, 1953. And I challenge a person to recover their Google search sixty years later, as I can produce these notes from 1953.
The search possibilities in those days, primarily in science, were to go to chemical abstracts and quarterly cumulative medical index. These were some large books which were published yearly and kept in the library that would enable one to search for abstracts of articles under a given heading. And so I decided to see what was known about insulin, since I was going to work on something related to insulin, and I went back to 1945 first. I went to the chemical abstracts book and opened it at the place where insulin was the search word, and then it would produce, as we see on the first page of the planning notes, a series of numbers: 2144, P1023, 1734, etc. etc. In 1945. So these were abstracts of articles which contained the word insulin somewhere in the article, or perhaps in the abstract, I’m not sure whether it was just in the article or just in the abstract, probably in the abstract. So then I would look at the first one, which was 2144 with a superscript 4, which I don’t know what it means anymore, and look and see if the title of the abstract and the information available looked promising. And if it wasn’t I crossed it out. 2144 was not interesting, P1023 was not interesting, and evidently 1731 was not even correct. 3877, etc.
But I see when I got to 1430 a little checkmark alongside it and the note that says “the statistics of an assay,” saying “extended crossover design and its use in insulin assay, by K.W. Smith, published in 1944.” So I thought, well, at least I looked at that and thought there was a possible interest for me in that journal and that article. And then going down about 5 or 6 more, one comes to 1148, a Physiological Review article, 1944, by R.E. Haist. “Factors affecting the insulin content of the pancreas.” And so on it goes. If we look at the next page, one of them was interesting enough to me to write out more details, saying “glycogen and adipose tissue by Wertheimer, in the 1945 article Insulin induces glycogen synthesis in adipose tissues,” and so on it goes. And then going to 1946: not very much success there because nearly everything is crossed out, except one comment here on a review on newer methods for estimating amino acids insulin in detail and so on. Next page, esterification of proteins, etc. So these were the things that were looked at by me in trying to decide what to do next. I’ll see if there’s anything particularly noteworthy as I go through this. So I’m turning over the pages, “Degradation of insulin with pepsin and trypsin,” notes on that, and so on and so forth. A British patent here on one page, December 27, 1944, “Removal of fat-like fractions containing insulin enzyme degrading stimulants,” etc. I have no idea what it’s about. Another patent, USSR, and so on. “The action of insulin” in Physiological Review, 1947, a review on theory, and so on, page after page of this type of search. 1948: another maybe 20 or 30 articles, most of which are crossed out, but one or two are selected to be written about in more detail. One on contra insulin. These people obtained impure in lipid soluble fraction of beef pancreas and tested it in dog, rabbit, rat, mouse, and frog. Active orally or by injection, etc. So different people trying different things at a time when methods were not so sophisticated but people could still do careful experiments. Butler in 1948, for instance, finds that crystalline trypsin has little effect on insulin. Chymotrypsin acts rapidly, releasing about 15% of non-protein nitrogen, and then slowly up to 100%. I go on page after page of this material. One here, a pure insulin and solubility test. That must have rung a bell because of my work on solubility of lactoglobulin during my postdoctoral period. One by Gutfreund, and I knew Gutfreund because he used osmotic pressures and molecular weight determinations, Biochemical Journal 1948. I don’t even write out a detail because I already knew the article from my thesis days. Or I knew Gutfreund from my thesis days. Paging and paging through here to see whether there is anything particularly noteworthy. I have one comment here, underlined, as I evidently thought that this was important. It was a review from the Brazilian biological journal. They had got a very sensitive test for insulin and so on. I see nothing particularly notable. And then I do have again recollections because here is an article by Gutfreund and Sandy Ogston, back in Biochemical Journal, 1949. Finding fraction A of performic acid oxidation of — I can’t really understand it but it was used in diffusion studies. Gutfreund and Ogston.
I got to 1950 here and another 20 or so articles. So this was quite hard work to go through all these considerably slower than would be possible these days, but of course in proportion fewer things to look at. So I’ve often thought that it’s about equally difficult. Now one can go through a lot quickly, sitting at a computer and pressing a button, and there are many to do. But back then you had fewer to look at, but it took more time. So the equation comes out about equal. Here’s one by Fred Sanger on species differences in insulin, and ox, pig, and sheep show differences in the proportions of the amino acids. This is before it was possible to determine amino acid sequences. More by Fred Sanger, alpha amino acid groups free at the chain ends enables peptides to be identified. In the beginning of learning of how to sequence proteins, Fred Sanger worked out of a method which was very important in its day of labelling the end terminal amino acid of a peptide. And we used that later, or my colleagues and I, Gordon Dixon and George Connell, used that in our work on haptoglobin. So many pages here.
Back to 1951 now. Many more articles. Was it not with Gordon? It was. People trying oral insulin therapy, review, of course, still would be nice to be able to use it orally. But proteins at this point still can’t be given orally. Fred Sanger again. Many notes. Ammonium chloride. Let me see what that was. There was one article here which I evidently thought was very important by Malandra and Cavallero, purifying insulin. And I have some notes, evidently I’m thinking about the possibility of modifying insulin if necessary by tacking on to amino groups of lysine and histidine oil or longer alcohol or an ester to alter the extraction. They had electrophoresis, convection, etc., resins. Fairly long notes on this one. Particularly it was talking about the conditions of the pancreas which Banting had observed. He was very concerned in trying to avoid the proteolytic digestion of insulin by the enzymes of the pancreas. So they tried to find conditions which would prevent trypsinogen and chymotrypsinogen, which are the proenzymes made in the pancreas, from being activated. So I have a note about looking up conditions which cause trypsinogen and chymotrypsinogen to go to the enzymatic void and avoid, underlined, 0 degrees and pH 8 is ok. So I was evidently thinking also about what to do. Test the solubility of dried insulin in several dry organic liquids, and the test tube with polyethylene tops, thinking back to my postdoctoral days of using polyethylene seals of the test tubes using crystallization. And I’m evidently interested in resins because I’m saying “write the details of available resins for protein purification, may be used on electrophoresis convected material. Try extracting normal pancreas with aqueous sodium chloride buffer solution.” So the notes changing a little bit to thoughts about what to do. A little diagram here of spraying insulin into a chamber, and allow an animal to inhale them. I was thinking of the possibility that it might be possible to administer insulin by the lungs. Are the lung capillaries permeable? Anyway the lymphatics are present, a little diagram of how this might be done. And with a comment that evidently somebody must have tried something similar because there is a reference. This would do for any gut wrecked substance that can go through the lung capillary. A note here: “definitely implied carryout of paper electrophoresis on post pituitary before electrophoresis convection.” A query: “find a method to extract proteolytic enzymes first.” Various thoughts, so the planning notes have gotten away from abstracts now to specific ideas. Although it goes back again to articles that I’m looking at.
And here is something that actually became quite important very shortly after I began the work with the insulin. Here is a reference to Kunkel and Slater in the Proceedings of Experimental Biology and Medicine. They use potato starch for electrophoresis. No trailing as with paper. Blocks for estimation. And we shall see later in my notebook of the first experiments that that was in my memory, that reference. And many pages of – I had more detailed information from the abstracts. Here’s a note that I know that I later on tried it, because here’s an abstract from a paper by Macheboeuf from Chem. Weekblad in 1953. Utilizes controlled water evaporation and electro-osmosis in filter paper electrophoresis to balance electrophoresis, thus protein comes to rest at a characteristic point. Good possibilities. And a diagram on the next page of how the method worked. I did in fact try it experimentally, as we’ll probably see when I go through the books on the experiments, it was difficult and it didn’t really work, but this is the paper that led me to think about it. Macheboeuf in 1953. Notes about what turned out to be a difficulty because evaporation must be low enough to prevent the buffer salts from crystallizing. I was somewhat fascinated by counter-current purification methods such as that talked about in an article 1952 by Lyman and Craig in the Journal of the American Chemical Society. Several counter-current tests on insulin show that the major component and varying amounts of a minor component in different preparations, and some of the minor components, I think its saying, have activity. Because they – as we see later, I decided later to look for a precursor of insulin and these things were part of what was leading me to get to that point. Again, Kunkel using paper and starch electrophoresis to study, in this case, oxytocins. Beginning to think about the real experiments and I have one page here that is talking about the usual volumes and units per milliliter etc., different stages of preparing insulin and how much activity per milliliter and saying that usually one unit per milliliter is used for a test. I presume that that’s in an animal, though it doesn’t say so. But diluted about 1:20 and ½ ml given per mouse. The assay in those days was quite a crude but effective assay, that mice were allowed to run, as it were, or maintain their positions in a rotating cylinder with a rather rough coat (?), so the mice could prevent themselves from falling out by as it were running up the hill in this rotating cylinder. But if they were given insulin then they would no longer be able to run properly and would fall out of the rotating cylinder, then a technician would immediately take the mouse and give it a big shot of glucose so that it would recover. This was the assay when the mouse came unable to run properly but was rescued by being given the insulin so it was not killed. Usually 1 unit per ml was used in the test, but diluted 1:20 and ½ ml given per mouse. So the range was 0.25:0.01. I’m not quite sure what the decimals are here, per mouse. “The cylinder test eliminates personnel trouble.” I don’t know quite what that means. I think it’s probably referring to the fact that the mice really didn’t have a hard time, so nobody was worried about hurting the mice seriously. And the extraction was with the acid alcohol, centrifuged, neutralized, filtered, concentrated at 50 degrees, and that passed on to purification. The activity drops from 2000 units per kilogram tissue to about 800 at the end. Ammonium hydroxide causes the loss. And one of the persons in the lab who I got to know quite well, Romans, introduced a new method of salt precipitation instead of ammonium hydroxide, and keeping the pH down around 5.5. So these were the methods that were being used for purification of insulin that I had to know something about. And so again the notes are changing into plans rather than searching the literature. And here’s a diagram of the electrophoresis on paper by evaporation technique. The Macheboeuf technique uses an apparatus like the following, and I drew a diagram of the apparatus. “Electrophoretic convection” I called it, I wouldn’t use that word anymore, I would say electrophoretic migration, but convection was used as a means of saying movement. Still thinking about salt precipitation and resin columns. Better avoid electrophoresis convection until the number of components is reduced by salt fractionation. Straight forward paper electrophoresis for crystalline insulin and then equilibrium electrophoresis. As we shall see, by no means is electrophoresis on filter paper simple.
And evidently I’ve been talking to Charlie Haynes about these things because I have a little note here “using washed paper – CSH”, that’s Charles, Professor Haynes, who was a friend and who had been the person who had actually gotten me in touch with Dr. Scott when I first was looking for a job in Canada. I thought of trying some of these ideas with glucose on paper and then staining it with silver nitrate in order to see what was happening with this evaporation technique. When working, practice getting insulin off the paper into a test animal. Notes on what Romans and Scott were doing. Romans, Scott, and Fisher were three of the staff at Connaught Medical Research Laboratory who were in the production side. Of course they became my friends. Scott was no longer in production but Romans was and Fisher was. A comment there: in other words, dissolve and dilute ascetic acid and add zinc and adjust the pH to allow crystallization. That’s obviously related to David Scott’s method of preparing crystalline zinc insulin and some precipitation curves. And I came across a reference which was related to my thesis work — I’m sorry, to my postdoctoral work — by Northrup, Kunitz, and Herriott, Crystalline Enzymes, 2nd edition, 1948. Page 293: “On the other hand, a single protein which is present in two distinct solid forms, for example amorphous and crystalline, may also give the solubility curve of a mixture,” which is what I found in my postdoctoral work. And the planning notes end. The last page is notes on the method of use, 10 grams of dry pancreas suspended in 500 milliliters of distilled water, to which sodium carbonate and hydrochloric acid are added to pH 7.4, etc., stand 12 hours, centrifuge, etc. etc. This is somebody else’s methods with the final comment: “they did not try a calcium phosphate, but Burke (or Bruke), 1861, says it will remove pepsin.” That’s the last note in this planning book.