Oliver Smithies:
[00:00:00] There is a little confusion in looking at these books, sometimes, because of my rather bad habit of sometimes starting some experiment from the back of the book towards the front at the same time as doing experiments that recorded from the front of the book to the back. So, records, physical one that we’ve talked about, actually has some experiments at the back which are related to insulin work. So, there is a note of December 1st. Now, we’re talking about 1953. December 1st, working with David Scott with DAS. Different solutions of insulin are being considered. And going backwards now, from December [00:01:00] 1st, we go backwards. And then December 1st, Tuesday, and going backwards again, Wednesday, December 2nd. And [zinc-free?] insulin again. Thursday, December 3rd. And that is the end of the backwards part of this book which then goes forwards in the book labeled “Insulin,” which starts overlapping. Wednesday, December 2nd is overlapping the other book which has Thursday, December 3rd in it. So, the end of the backwards part and the beginning of the insulin book are consecutive. But the physical book is ending [00:02:00] now, already, thinking about gel electrophoresis which is Sunday, January 24th. So, it’s not the same year. It’s now in 1954, beginning of the gel electrophoresis. So, we’re picking up book two, physical, is following from Tuesday, January 26th, the last entry in the physical book of electrophoresis. So, now, moving to book two of the physical 1954, it’s talking about different ways of making the gel more [attractable?]. In particular, trying to get rid of the fact that you looked at the gel [00:03:00] not in a horizontal plane, but in a vertical plane. You could see that the protein at the top of the gel, which was exposed to the air where it could evaporate and get more concentrated, trailed as the starch got more concentrated. So, in trying to get rid of that problem, Friday, February 5th is the first entry in 1954, book two. Which, again with the power from the DC supply of the university at 220 volt at 10% starch with a reasonable slot and trying to seal the top with polyethylene to avoid evaporation. [00:04:00] Sunday, February 7th, 8:15 p.m. Sunday evening, working at it. Monday, February 9th. Tuesday, new starch arrived. I don’t know where I was getting — new acid-treated starch arrived. These were starch according to [Lintnor?]. These were Lintnor starches partially hydrolyzed with HCl. And I was getting these samples from a company in Toronto. I don’t remember the name of the company. And they would have different lot numbers, apparently, which I later found were completely misleading. The lot numbers were changed in their bottles even though it was from the same [00:05:00] lot. Very confusing. But I wasn’t aware what was happening at that point. So, I have a comment that’s on Tuesday, February 9th. “The new acid-treated starch has arrived as a sample,” as a first sample. Later on, there was a lot of difficulty associated with different starches. A little comment here regarding the solution in which the protein sample is dissolved. In future, ensure that the insulin is in a medium of less conductivity than the starch because of noticing what happened when it was in a higher [00:06:00] conductivity. Now, I tried a vertical starch at one point. Then think about setting up, like this, vertically, to avoid the density effect because I’d already noticed that samples migrating in the slot came up to the starch gel and then made a concentrated solution which decanted to the bottom of the gel so that the entry was not uniform. And I was already thinking about a vertical electrophoresis, but I don’t know whether I did try it yet. But I drew a picture of [00:07:00] — think about setting up like this to avoid the density effect, and to enable sharper boundaries to be made, making a vertical electrophoresis. Saturday, February 13th. Then I did an experiment to help my friend, Gordon Dixon. He had been having problems with enzymes in cabbage that he was studying as a thesis problem. And so I have here a note that at 5:15, Saturday, February 13 — no. Monday, February 15th, set up Gordon Dixon’s enzyme. [00:08:00] Two percent in two times distilled water and diluted with [barbituary?] buffer, pH 8.25, with the starch containing undiluted buffer, a wide slit. The solution is colored. May be able to observe the movement by inspection. And the enzyme had been dialyzed. I must have been using copper electrodes because I have a note that — check at 7:45. The current is almost zero due to precipitation on the copper electrodes, parentheses must get platinum. And changed in the meantime to a nickel-plated steel and [00:09:00] readjusted the current. Tuesday, February 16th, off at 10:00 a.m. — that was Gordon’s — was set up on a different experiment. Tuesday, February 16th, off at 10:00 and no sign of protein. Set up again with a narrow slot, etc. And then off at 1:45. So, 10:45 a.m. to 1:45 p.m., a migration about seven millimeters. I could see two bands. One about a millimeter wide and the other about three millimeters wide, with [00:10:00] clear demonstration of two components. However, the staining is extremely weak with amido black 10B, even with 20% trichloroacetic acid added. But Gordon Dixon took some for activity tests. So, he’s now going to see if he can get activity in the gel after the electrophoresis. And I was going to try bromophenol blue and make a comment that result approximately equivalent to amido black. And Gordon Dixon tested, even though there was very little migration there — the three millimeters and seven millimeters — he tested the two fractions. And all of [00:11:00] hydrolytic activity which is what he was interested, was in the fast zone and not in the slow. So, already, he began to realize that it was a useful method for him. And so I diverged from insulin and helped him for a while. Measuring the pH. Thursday, February 18th on an insulin experiment again. Little bit of a component behind the insulin. [00:12:00] Back to Gordon again. Friday, February 19th, 2:40 p.m. Set up with barbiturate buffer, which people, at that time, were using for filter paper electrophoresis. And put on about one milligram of Gordon Dixon’s enzyme in a half-strength barbiturate buffer by filter paper technique. And ran in a little buffer as well. And the pH was 8.78. And after ten minutes, a very narrow band was visible. [00:13:00] Two millimeters. And this band widened rapidly, so it may pay to reduce the currents. So, now I’m seeing after one hour that the protein has migrated a distance of 7.5 millimeters and were two visible bands. Set up for a much longer run. Now can see at least three proteins, possibly four. One, two, three, and four are extracted with buffer centrifuge and tested for activity. There were two enzymatic activities that Gordon was interested in. One was hydrolytic and the other was a transfer activity which at this point I would have to go back to the literature to find out what those were. But [00:14:00] after one hour experiment, you could see four visible bands. And say the next run is to be more concentrated for three hours, and see what we can get. Saturday, again, 1:00, Gordon Dixon enzyme set up, etc. etc. Migration now about three centimeters. Here is February 23rd. The comment of setting up his enzymes again. And Gordon Dixon has results. [00:15:00] Now, [an x amount?] of insulin, again, tucked in the — I mean, begin to start thinking about cutting samples into multiple slices. We’re using razor blades assembled in a rack so that each blade was about two or three millimeters apart. And many of these blades set up in a slicing machine. I still have those racks of razor blades [00:16:00] which were used to cut these gels. For example, one could use them — cut those slices and then use half of one slice for protein and half for activity. Talking about the same slice. And Thursday the 25th, the results on February 27th, sectioned and stained one piece. The general staining. Now I’ve got the band at one millimeter, nine millimeters, and 2.1 centimeters. Visual staining of the slices quite satisfactory. [00:17:00] And, at last, a plot of the results showing the optimal density at 536 millimicrons, diluted with 10 milliliters of ninhydrin. The results of an insulin protein experiment, Saturday the 27th. And one page would be something with the insulin. The next page is results of Gordon Dixon experiment at 6.45 pH. I got three bands. There is no fourth visible. And none going backwards. Anyway, slice for analysis. [00:18:00] A little note on the — I’m evidently concerned about weighing these samples. Typical Smithies comment. “Think about a glass torsion balance for weighing the dried slices the correct protein estimation.” Checked in the idea with a rough one. The torsion balance. [00:19:00] Noticing a change of pH along the gel, that the pH at the negative electrode was 7.84 and the pH at the positive electrode was 6.5. This shift of pH seems to occur with insulin, but not with Gordon Dixon’s enzyme query due to HCl still present in the insulin so that the chloride ion displaces the barbiturate. More likely, I’d say, from the comment on the 3rd, due to electrode products. [00:20:00] Weighing these units of the gel corrected for not having uniform slices, so there’s a pH Gordon — on Monday, March 1st, Gordon Dixon pH 6.5 units of gel and then weighing each slice. So, they varied in weight from about 69 milligrams up to as high as 102. [00:21:00] Tuesday, March the 2nd, something to do with insulin measurements in different slices with a comment that “too many accidents. Abandoned the experiment.” Decided to make a large insulin gel run overnight and get out five approximately equivalent sections for analysis trial. One for ninhydrin, one for Follin, one for Follin (inaudible), and some spare. [00:22:00] And insulin doesn’t migrate very well, evidently, at pH 7.5 because on Wednesday, March the 3rd, I have a sketch of things with a comment that the usual sharp trailing edge with insulin at this pH. So, it was not very nice at that pH. [00:23:00] Still trying to get a quantitative estimation of protein into slices of the gel. Thinking about alkaline hydrolysis for some real reason which is now a little obscure. [00:24:00] Beginning to use a borate to buffer March 10th. And later on, that became the most often used buffer. 4:45 p.m. on Wednesday, March 10th, 220 volts for Gordon Dixon’s enzyme using a 0.02 molar total borate buffer, pH 8.85. And making sure that I didn’t have electro products by having two chambers on each end of the electrophoresis equipment. [00:25:00] Stained (inaudible) bands. Getting a better cutting apparatus (inaudible) with the new supported cutter. [00:26:00] Thursday, March 18th, worrying about putting insulin into the gel with or without filter paper, or with different filter papers. Thursday, March 18th, I tried a slit. Just letting a little bit of insulin go into a slit in the gel without any support or number 54 paper, the thin paper, on number three, a thick paper. [00:27:00] And got staining, as you might expect, just heavy on three, medium on 54, and almost invisible on the slit. So, I’m beginning to understand the importance of how you insert the sample into the gel. And I’m also beginning to understand that a buffer solution, in a filter paper wick used at the end of the gel, that it caused distortion of the gel [00:28:00] so that you need to be careful about how the filter paper wick going to the electrode vessels, it contacts the gel. All of these different little problems. Just try to get rid of some of this electrode difficulty. I even have one test where there are three different buffer chambers on each end of the gel with borate in the buffer, 0.02 molar borate. [00:29:00] And then 0.1 molar borate in the next chamber and 0.2 molar sodium chloride in the last chamber, to try to get rid of problems at the end of the gel. And then rather — this is Monday, March 22nd, doing the multiple gel. And then on Tuesday, March 23rd, there is a rather important note here that serum on — with same solutions, just for a rough test at 10:00 p.m. on Tuesday. Oh, no, I’m sorry. At — [00:30:00] yes. I think it is at 10:00 p.m. on Tuesday. Just for a rough test. And took it off after an hour and a half, and I could see four bands from the serum. One that I was pretty sure was albumin. And then I thought the next one was alpha one or alpha two globulin. I couldn’t tell which. And beta globulin. There is multiple overwriting, so I’m not sure what the bands are, but there is a [pink?] band at two millimeters. And the albumin has migrated six millimeters. And the gamma globulin has gone backwards, which I evidently thought that this was promising because I set up, at 12:00 midnight, again. And [00:31:00] ran overnight from 12:00 midnight. That’s an important page at the beginning of my — beginning to think that it would be useful to look at serum protein. And so this is Tuesday, midnight. And then on Wednesday, March 2nd, I looked at the result. And the result was many bands. I have a sketch here, and I could see one, two, three, four, five, six, seven, eight, nine — something like ten or eleven bands. In fact, I have that comment on the opposite page. “Total of eleven components.” Very [00:32:00] promising. So, that was the beginning of my change from studying insulin to studying serum proteins. I went to see David Scott and told him about this and said, “Scotty, I’d like to stop working on insulin and work on serum proteins because I can see so many bands. And people normally talk about five proteins in serum and I can already see eleven bands. So, it’s something very interesting.” And he agreed, and I very quickly stopped working on insulin and began to work on serum proteins. Not completely stopping the insulin. There was another experiment on March 25th with [00:33:00] insulin. March 26th, Friday, looking at these bands. And the albumin has migrated about 12 centimeters. And in front of albumin are two components that I can see ahead of albumin with a note that, are they new or are they hemolytic due to hemolysis. Albumin and onset (inaudible) alpha one, alpha two, alpha one something. New bands. I didn’t know what they were. Confused about labeling because I’ve got too many bands to label. And it all looks better, however, but not so sharp as before. Perhaps 0.03 buffer would pay. [00:34:00] It may be an anomaly at 0.02. But I got some new serum from MLC. That’s [George Connell’s?] sister, obtained from the hospitals where she worked. And ran at pH 7.9 to check for hemolysis. This sample was free from visible hemolysis. However, I make the note that it was — unfortunately, it froze in the cold. Must have been pretty cold weather in Toronto in March 26th if the sample froze. Here I am setting up the next day, Gordon Dixon’s enzyme with a complicated setup. [00:35:00] And I have a comment on Sunday, March 28th. “Better use something instead of filter paper because the gamma globulin is not in the gel at all.” Moving backwards. And I think it’s due to electrophoresis and then I cross it out. Electroosmosis must (inaudible) in the higher salt gels, as could have been predicted that there’s a big X across that statement on Sunday, March 28th. So, I’m still puzzling about what’s going on. [00:36:00] With a comment the next day, “She’ll have to use a hand stethoscope to check for hemolysis as it looks as if the new bands are much less and they narrow with the increased concentration gel,” etc. Trying different methods of inserting the sample. [00:37:00] Here’s a test of number three filter paper, cellulose. I don’t know what the cellulose is, exactly. And starch. It’s making a slot and then filling it with starch powder. And at seven minutes, the starch shows the sharpest forward edge. And the cellulose is next, and number three is the last. But the thickness affects (inaudible). I’m struggling with trying to find a way to put the sample into the gel. [00:38:00] I have one example of the setting up that is a rough test setup at 11:20 with dry starch powder in an approximately one millimeter slot. [00:39:00] And then run in 66% serum. Just letting it run into the slot, holding the slot open with starch powder. I’m recording here. I set up an experiment with four nucleotides in a new [trough?] and [00:40:00] measuring them with cytosine, adenosine, guanine, and thymidine. And just set those up and measure the [optimal?] density in different slices. See what was happening. Not particularly useful. Still trying to get the bands to be uniform. I have a note on Wednesday, March 31st that the results show [00:41:00] no band. But new components. I keep getting new components that I don’t understand. So there is a comment there that the results show no band, presumably salt, etc., but there are new components unlikely to be due to hemolysis. A nucleotide experiment done again, but with the comment, “Thrown out later.” [00:42:00] Trying whether sodium chloride versus borate would make any difference. So, borate takes a moment and the borate — sodium chloride. And I can see more components in the borate per sodium chloride. One, two, three, four, five, six, seven, eight, nine, ten. (inaudible) new. As I look at it, there isn’t very much difference. A total of ten plus [query?] one more in beta with the sodium chloride. Gordon Dixon again. Gordon Dixon’s enzymes again. A new cutter [00:43:00], made 31 slices, 2.98 millimeters per slice. All slices were made. Eighty slices including the filter paper, and into the deep freeze with the rest of the gel. And Thursday, April 8th, “serum received yesterday,” presumably from George’s sister, “containing no detectable hemoglobin.” And also red cells. [00:44:00] So, trying borate per sodium chloride sometime. Should try it sometime as it seems to improve. If I can’t get good results in borate, might try (inaudible) buffer, but that’s about the ninhydrin. So, I took the serum free from hemolysis and added different amounts of the hemoglobin [00:45:00] across the gel. So, none four milligrams, twenty milligrams, a hundred milligrams from [last?] red cell. And ran a gel with the different concentrations. And I could see, at two hours, visibly, a single band that got darker as you added more hemoglobin. And I have a sketch of the results. But with the comment that hemoglobin complexes with alpha two and most of alpha one and prevents resolution. And some are not due to [lytic?] products. And [00:46:00] beginning to think that it might be useful to split the gel longitudinally for staining, to take the top off. That was always a problem, which I later realized was due to the fact that when the gel was being poured, the top would tend to be dry and more concentrated. But here, I’m beginning to think about splitting the gel longitudinally to get better results. And another gel with hemoglobin in it. Three samples of hemoglobin added to serum. [00:47:00] Nothing added to serum and hemoglobin added to serum to look and see what’s happening. And not quite understanding the result. With a comment on the following page that “note the decrease in alpha one as hemoglobin increases.” I haven’t yet realized that there is a hemoglobin binding protein which is what I’m thinking of as alpha one. And that it binds to the hemoglobin and gives a band of what is — later, I’m going to understand — is haptoglobin combined with hemoglobin. Not yet on board with that. [00:48:00] Having some difficulty understanding the migration of albumin which is always migrating as a rather wide band, a very wide band, in the borate buffer. And sometimes the back of the band was sharp and sometimes the front. So, I have a rather helpful experiment which was carried out on or about Friday, April 9th, in which the buffer concentration was varied. So, if the buffer was 0.02 molar, then the back of the albumin was sharp. And at 0.03 molar, front and the back were about [00:49:00] equal. And at 0.04 molar, the front of the gel was equal. And later on, we used to calibrate starch to tell people what concentration of borate buffer was best for a particular lot of starch. So, this was the beginning of finding out that the sharpness depended on the concentration. So, it’s saying the usual albumin affects, with unexpected pre-albumin effect, the usual beta effect, etc. etc. The gamma is the same in all. But the staining is quite clearly — the sharpness is quite clearly borate concentration dependent. And on the next page, even more [00:50:00] dilute solutions to look at what’s happening. 0.02 molar, 0.01 molar, and 0.005 molar, looking at what happens to the pre-albumin. Tried making gels with different concentrations and measuring the pH of the gel. 0.04 molar borate, it was 8.78. 0.01 molar, it was 8.05. And lower, the pH was all over the place. “Practically unbuffered by the borate,” is the comment, Saturday, April 10th. And then little note: “Don’t forget plus or minus sodium chloride.” [00:51:00] Began to think about getting material out of the gel. pH of frozen gel [expressate?] is 8.51. It contains very little starches judged by the (inaudible) test. And it looks like a way of getting protein out of the gel by freezing it. [00:52:00] I’m beginning to be interested in whether a plasma versus serum makes a difference. So, Wednesday, April 14th, I made some samples with or without oxalate as an anti-coagulant or an anti-clotting agent. [00:53:00] Or with calcium deliberately. Serum with oxalate, plasma with oxalate, or plasma with calcium. There’s a note and it looks like fibrinogen. I don’t know whether it was or not, but I felt it might be in that experiment, Wednesday, April 14th. I have a comment here that “since best results to date are with 0.015 molar, 9.19 washed starch, set up 150 grams plus one liter of 0.015 molar, 9.19 [00:54:00] borate to wash,” and so on and so forth. Washing the starch powder, or the starch grain. Borate gel plus or minus oxalate, plasma versus serum. [00:55:00] A comment here on the present slicing method, horizontal technique. I remember what that was. I made a little piece of [Perspex?] with two strips of Perspex making, in a sense, a trough, with the thickness of the Perspex being about half the thickness of the gel so that by putting the gel between these two strips, I could then take a razor blade running along the rails, and it would split the gel horizontally down the middle. And I could stain the surfaces, which were then free from any curvature effect. It was a very simple method of horizontal slicing that was used a great deal later on. [00:56:00] pH 8.6 experiment which is labeled as good with plasma as the sample, with a buffer containing 0.02 molar borate and 0.005 molar potassium oxalate. With some comments on query fibrinogen, trying to find fibrinogen. Now, worrying again about hemolysis, starting to do what I did later in a different way is, since fibrinogen [00:57:00] is showing so poorly, prepare the 50 milliliter syringe coated with a silicate, as it’s called — silane, I think, was the name for it — and withdrew 30 to 40 milliliters of blood through a wide needle. Transferred without bubbles to a tube, and centrifuge at 2,000 in a (inaudible) centrifuge to try to get plasma free from hemolysis. I later became — once I knew about haptoglobin, I became a real fanatic for getting samples free from hemolysis. And I would take a tube of, for example, a 50 milliliter centrifuge tube, and coat it on the inside with petroleum jelly [00:58:00] by heating the whole thing up and putting in hot petroleum jelly and then inverting it and letting it drain. So, leaving a very thin — or, relatively thin — film of petroleum jelly on the inside of the tube. And then withdrawing — having blood drawn with a wide needle. And then gently expelled, not through the needle, into the tube, and so it would not clot easily because it was in such a good environment as far as starting to clot. So, you would sometimes have to add a little bit of filter paper to encourage the clotting. The clot would then retract, including the red cells, and pull away from the petroleum jelly without any tearing of the red cells, so I could get very low hemoglobin content in these samples. I remember [00:59:00] being so fanatic about it that I drove from the hospital for sick children, where they withdrew my blood, and going back to Connaught in the car and not stopping the car in order to not shake things. But the intersection — but I got a ticket from the police with it. And I went to court on this and explained what I’d done and why I’d done it, why I hadn’t stopped because I didn’t want to shake the sample. But the judge didn’t let me off. I still had to pay the fine. But, anyway. (inaudible) I have a note saying, “The results show the fibrinogen OK.” That was when serum was compared with plasma. Wednesday, April 25th: “Fresh plasma,” [01:00:00] with a little comment, “OS,” meaning my plasma, “was prepared by a rapid blood withdrawal into siliconized syringe. Transferred, etc. etc. etc. Quite fanatical about it. But with filter paper insertion, it was no good. The fibrinogen stuck to the filter paper. [01:01:00] Saturday, Sunday. Doesn’t make any difference. Working whatever time of the day, whatever time of the week, whatever day of the week. So the last experiment in this book was on Monday, April 26th. And following an experiment on Sunday. So, we’ll be moving to a physical book, 1954, number three. (inaudible) [01:02:00] find the next book. It’s making sense, anyway.