Oliver Smithies:[00:00:00] This is book little i. Begins July 3rd, 1984 and runs through February of 1985. It’s continuing the problem of trying to control the electroporation of cells with a high voltage pulse. So the first couple of pages are concerned with trying to find the way of estimating two things, the amount of killing, and what the effective voltage that should be used. So page one is talking about the eosin permeability that’s — and eosin is used as a measure of killing. If cells allow eosin in then they’re dead. [00:01:00] And so this is a time to determine the ratio between the voltage used for the zapping, the electroporation that’s to say, and killing. So 3,000 volts, 2,000 volts, and so on. That’s right. But point out rather interestingly that the 3,000-volt experiment for — the cell shattered. So rather dramatic effect. But the dead cells were quite variable. So 89% were dead at 2,000 volts. And at 3,000 volts it looked like there were more living cells than at 2,000. So there was a lot of variation [00:02:00] that was worrying me. And so I began to think about the time constant. Thursday, July 5th, page three. And looking at the voltage across a 1-ohm resistor discharging through the cells at 1,000 volts. And it was clear that there are different time constants in different parts of the system. Anyway I was unhappy with it and decided at this point that I’d better use a silicon-controlled rectifier system so that instead of having a spark at the time of making the contact and getting the variation [00:03:00] in effective pulse to do something where there were never a spark. Thursday, July 5th, page five. Just continuing these experiments.
And then as we come to page six and seven, it’s an interesting set of pages to me because it’s back to some electronic things which I’ve always enjoyed. And trying to understand how a silicon-controlled rectifier could be used to control high voltages. And so I went to the local Radio Shack store and found that they had a silicon-controlled rectifier called [00:04:00] 2N6173 and the specifications of that silicon-controlled rectifier were given. And it was particularly that it could be used with a voltage of about 400 volts. And was still usable as switching 400-volt. But since I was wanting to control at the order of 4,000 volts it would need 10 of these in series with some sort of circuit to make sure that they took equal parts of the load. And then there are attached to page six an enjoyable set of experiments that I did trying to understand it. And it starts with [00:05:00] stage one July 10th four o’clock. With a mock cell just using a 910-ohm resistor as a mock cell. Trying to understand how to control the discharge of a capacitor C1 through the 910-ohm resistor on closing voltage. And a little circuit diagram then said this fires the silicon-controlled rectifier as long as switch S1 is closed. And C1 could be anything from 0.02 to 10 microfarads. That was my stage one.
And then going up stage two July 10th at 5:00 p.m. [00:06:00] It was now two of these in series. And trying to find out what capacitors I needed to use between the two. And notes on this. And so on. It goes on and 6:00 p.m. on July 10th we get to stage five and stage six. Where I’ve got one, two, three, four, five of these silicon-controlled rectifiers in series. Beginning to learn how to get them to fire regularly. And stage seven and so on and so forth.
Using resistors to make sure that the voltage drop across each rectifier was uniform. With a comment that 1 megohm doesn’t work. [00:07:00] And also that 47,000 ohms works. But I remember ending up with 1 megohm to make sure that the voltage on each rectifier was — no one rectifier would have to take the load.
And comment likly tends to toggle if not allowed to charge fully, etc. And here we get to stage eight is now July 11th at 12:00 noon. Works reliably with one shot only and no back currents when the voltage is greater than 150, etc. And stage nine, July 11th at 1:30. And stage 10, July 10th at three o’clock. [00:08:00] And a final test stage July 11th at five o’clock. Where the circuit was — it managed up to 3,000 volts, it breaks down at about 3,500 volts, but is OK at 3,000 volts. About 15% higher than the rating for the silicon-controlled rectifier. And nothing gets clocked and it can be cycled in about 30 seconds. So that was the final circuit that I made and that I used. And then attached to the page seven is a circuit that was produced by our electronics shop. I asked them to make a circuit – it came out really fairly similar to my circuit. [00:09:00] And it was dated 7/10/84 and tested and worked fine. But I was happy to be able to use my machine for further work, which in fact was important for making things reproducible.
So the new zapper, Thursday, July 12th, page nine, 6 times 10 to the 7 cells available, etc. And zapping at 0 degrees centigrade. Not too happy with what was happening with the electrode. So I had the comment let’s redesign the cell next. And so retested however with the old type zapper cell and cleaned very carefully, etc., etc. [00:10:00] Note again on page 13, Friday, July 17th. So trying to make things work.
Going back to some digestion experiments, theta 133 3, etc., etc. Delta beta 117 digestion of some of the plasmids which we were going to use. In fact delta beta 117 became our final useful product.
New cell for zapping designed on Monday, July 23rd, page 17, which in the end was the one that was used for all the experiments, which had a platinum face and a spacer [00:11:00] which controlled the amount of solution that was going to be zapped, 1 square centimeter cross section. And another electrode block on the other side. So two platinum faces with a spacer. But notice that tested with PBS and gassing or boiling of the cell occurred at any voltage greater than 700. But sparking occurred if there’s any air in the cell, etc. So that was the design that eventually was used. Testing at ice temperature, page 19. And calculating the relative — what was relative to a 1-centimeter-1-centimeter-1-centimeter [00:12:00] cubed, etc. Various calculations. And getting rid of droplets.
Testing pulse length. Looking at some issues and trying petrolatum but not liking it very much, page 21. Now much of the work for a little while is going to be on this problem. Zapping again Wednesday, August 1st, page 25, 5-millimeter cell unsealed, etc. And so on we go with Monday, August 6th, page 27, a 500-microliter cell. And the percentage dead [00:13:00] at different voltages. In this case quite low voltages, 1,000 volts, 750, 500, 250, and a very nice range of death from 5%, 10%, 69%, and 89% dead. So that there was good control over cell death. Aiming basically at getting something on the order of 50% survival being something like optimum by guess rather than by knowledge.
And now I’m back to Raju Kucherlapati’s material on Thursday, August 23rd. Unselected DNA. That’s DNA from cells that had [00:14:00] not been selected with any antibiotic. So these are bladder carcinoma cells transformed with DNA delta beta 117 cut with BstXI and then unselected in G418 with the assay and the phage assay again. Total number of phages was 6.3 times 10 to the 8 and the number that would survive on C1A were 2,520 of which 125 were positive for the beta and one was positive for the IVS2 probe. [00:15:00] Could be more on longer exposure. In other words one positive scored for the recombinant fragment. So that is unselected DNA. Also referring back to Mike Koralewski’s phage F24.
Candidate minis for my 129. Conclusion. It’s OK. But rerun, etc. So it looked indeed as if everything was OK because it was tested on C1A, C600, C600 suppressor F8 or DP50 supF. [00:16:00] Which is the standard that we later used, etc. And then a marker. And conclusion, it’s OK, rerun. Diluted a different degree. And rerun the Xba digest on page 33. Friday, August 24th. With no comment but the experiment shows the result.
So thinking about Tuesday, September 25th, page 35. Trying to rescue the old huge Cosos 17. [00:17:00] Which there was some DNA left from a previous experiment. And trying to regrow. And on the following page Tuesday, October 9th, page 37. Looking at minis growing up the Cosos again. Yields had been low previously and so on and so forth. Trying to rescue that plasmid. And so on page 39, on opposing page 38, the yield was about 138 micrograms. In cells made without chloroform. No. Without chloramphenicol rather. [00:18:00] And 200. But less pure when chloramphenicol was not used. And that was tried again with and without chloramphenicol on Friday, October 12th, page 41. Conclusion. It doesn’t do any harm. But the best was without chloramphenicol, 138 micrograms from 3 liters, plus as mentioned earlier, plus chloramphenicol 200 micrograms but not clean. And so redesigning Cosos 17. Still thinking about that big beast on the following page.[00:19:00] In order to make it comparable to delta beta 117 being cut at a single site within the region of homology. I needed to generate such a site. Not so easy. But trying to do. Put it in a SalI site in the right place was the idea. Thought I got 1 milligram but concentrations were wrong. Continuing this modification of Cosos 17 Friday, November 2nd, page 45. And on the next page as well. [00:20:00] So the conclusion on page 47 or 46 Saturday, November 3rd is that decided to proceed with putting in an Apa site. Continue. But proceed with Apa after 18 more hours, etc. And so on page 49 [00:21:00] Tuesday, November 6th this is actually rather looking to see whether I ApaI will digest as a single site. There must have been a single site within the region of — within the Cosos that I don’t anymore remember. Trying to get a single cut with ApaI. And together with the mini minutes digestion. So continuing with Apa on Tuesday, November 6th page 49. [00:22:00] And this goes on following page. Following several pages. Retest with Apa nathaphenol and restart of the whole thing on page 53. Conclusion that this Apa is much more potent but still has problems. Start again. Decision was correct. No. Thinking about purification of these large molecules I got interested in pulsed-field — pulsed electrophoresis. The idea being that you get electrophoresis with a pulse of voltage and then allow the DNA to relax again and then pulse again and so you can get large molecules that migrate through a gel. [00:23:00] Larger molecules than normal would be possible without waiting between pulses. It was an idea which was popular for a while. But at least in my hands never went very far. Although I tried it. And so we made a setup of this type on page 55, Friday, November 30th.
Always interested in electrophoresis. Not very pretty results on Saturday, December 1st. Looks rather poor. And Sunday, December 3rd doesn’t look much better either. And Monday, December 4th a little bit better. Things look as if they’re migrating now. [00:24:00] At least it isn’t all distorted. And having various thoughts on how long to apply voltage in each direction.
And returning to the old conditions for a check on page 63. Gel full slowed it off put back. Don’t know the extent of the damage. But amazing separation for six hours at this voltage. This electrical arrangement is good. Repeat overnight. So arrangement of the electrodes was good. And separating many polymers. Not perfectly. But obviously separating. [00:25:00] But didn’t repeat very well because next pages look pretty poor with a comment most odd. It looks as if separation are all set up at the entry region, etc., etc. Problems. Continuing this type of experiment over the next few pages. Next few pages, here I am, still looking at it, Wednesday, December 5th, page 71. Very local gel load is OK. Etc., etc. So trying to improve things. Longer gel on the following page. Load and dilution and so forth. Until that’s Friday, December 7th, page 75.
Looking back at some religation making a lambda phage ligation, [00:26:00] lambda cb2 rather to I think this is basically to get a substrate for testing the material rather new subject. To get a material to test the pulsed-field electrophoresis again. So this was taking the bacteriophage lambda cb2 DNA and ligating it to get polymers. And the conclusion is the dilutions are weird with this ligated stuff. But the separations are terrific. So I was very happy with the separations that we were getting. [00:27:00] And a longer run on Sunday, December 9th. The polymers are there but the separation is poorer. New solutions tested on December 11th, page 81. Testing with and without pulses. And the conclusion being that all is working well. I have an arrow pointing to a rather reasonable gel versus the one below where there were no pulses and no separation.
Even better on page 83. With the conclusion that the separation ranges are very easily controlled by the pulse time. So I was feeling happy about the — [00:28:00] so I’m testing Dixie Mager’s samples on page 85, Thursday, December 30th and with the comment that her DNA is partly degraded but signs of cutting with SfiI. So using. But this is an ordinary electrophoresis. Not pulsed-field. Just an ordinary 0.8% half TBE buffer. With a comment that the separation is very much voltage gradient-dependent. Not unexpected.[00:29:00] So plans being pointed out on Friday, December 14th. So I was worried on page 87 whether the material that I was using for these experiments was good or not. And thinking of having a better primary molecular weight standard. Test material for Nobuyo or Nobuyo’s material but with not much indication of what the samples are. Regular marker mix. CGS and MKS and CGN, MKN, etc., etc. [00:30:00] Looking to see whether the Klenow was working. Again trying to get material to get standards.
Better markers then still being sought Friday, December 14th, page 91. And then being tested, the first mapping experiment, on Saturday, December 15th, page 93. The conclusion that this is a separation of various materials that were available. And remarkably good for a first try. Looks as if cooling is an improvement. [00:31:00] This was using the pulsed-field system. Some samples being tested. Markers 1.6% agarose. And it’s a straightforward buffer. Just [00:32:00] with 6-second pulse in the two directions. Electrodes 4, 5, 1, 9, 2, as they were numbered in this system. And they were quite straight-looking gels now. And looking again with markers page 97.
Progress is poor. I was worried about whether the polymers were real or being manufactured in the machine. And came to the conclusion that the Klenow worked really because the [00:33:00] no dynamic polymers. And that the ligation is working slowly. New try at a marker Wednesday, December 19th. Conclusion: discard. Still more marker attempts. Page 103 Friday, Saturday, December 22nd. Retry with ligation first after heating. Conclusion. Excellent ligation. But then some note that it’s not really so.
Continuing in the same vein on Sunday the 23rd with a rather unusual comment. Conclusion damn it, still not ligated. [00:34:00] Conclusion again, no change, scrap. Hopeful. On Monday, December 24th. After Christmas Eve celebrations. But it wasn’t any good. Abandon written across the bottom with try a new phage prep. No good, exclamation mark. Usual business over and over again.
Next page. Looks awful. Clean out the gel and repeat. So here we go. Now page 111 still the same sort of thing. Page 113. And then going to think about fractionating the DNA now. [00:35:00] In different ways using DEAE paper. And what is available from the zapping now. So we’re going back to size fractionation of material that had been prepared by zapping. So this is necessary to put this a little bit more into context because we’re getting to the important part of the experiments. This one is trying to take material that Ron Gregg had made by zapping with different preparations of material. Either delta beta 117 or Cosos 17. And selecting for neo-resistant cells. [00:36:00] He had in this — had prepared two different preparations, one of which was with delta beta 117, and the other with Cosos 17. The one preparation which was called 1C through — 1C dash 4C was equivalent to 740 colonies. And 1B through 4B was approximately equivalent to 1,131 colonies. And the amount of DNA available was recorded, 1B, 2B, 3B, 4B. And the aim was to get 40 micrograms of size-fractionated product [00:37:00] in a range 5 through 10 kb to include the 7.6-kilobase target. Etc.
Now this is the motivation of the type of work. Was described in the final paper that we published about targeting. We were concerned about the possibility that the recombinational events that we were observing occurred during the assay rather than in the mammalian cell. So the thought was if we — and we knew what sizes of recombinant fragment. What size the recombinant fragment would be after Xba digestion. So the fragment would be about 8 kilobases if it was unmodified. [00:38:00] Sorry. About 8 kilobases if the DNA had already been targeted. And would be about 11 kilobases if the DNA had not been targeted.
So by prefractionating the DNA into the different sizes one could determine whether the targeted size of about 8 kilobases was already in the DNA before the phage assay. So here is again to say again the motivation was to take the material obtained from the mammalian cell and fractionate it into two sizes, about up to 8 kilobases, and 10 kilobases or thereabouts upwards. [00:39:00] So that if the targeting had already occurred then we should get colonies in our phage assay from the 8-kilobase fraction. And we should not get colonies from the larger fraction.
If the targeting occurred — if the recombinant fragment was obtained in the bacteriophage bacteria assay then it would be — we would get positive colonies in the larger fraction. So it was an important – so this was from Ron. Was DNA from about — equivalent of about 4,500 neomycin-resistant colonies. And it was the idea being to separate it into two fractions after digestion with XbaI.[00:40:00] And so page 117. No. Pardon as it were. Was checking 117 DNA. Cosos DNA, etc. To see if the digestion had worked all right. And DEAE membrane purification being attempted on this material beginning on Thursday, January 24th, page 119. But not too happy with the result. Second try on Thursday, January 24th. Worrying about the marker mix. This marker mix is wrong. [00:41:00] But not happy with the product. So continuing on page 123, Thursday, January 24th. To get these materials of different molecular sizes and checking the fragments of size fractionation on page 123, Thursday, January 24th. Three preparations, 123 fast, 123 middle, and 123 slow. Fairly clear separation of the three fragments. Three results. But so with a note that the Xba digest [00:42:00] looks as if at least one of the DNAs could be worth more processing.
Third prep gel following page, Friday, January 25th. Comments on use of this DNA. Attempting some more tests of elution material. Different size fractions obtained. Gels not looking too good. Saturday, January 26th. [00:43:00] Day Nobuyo came back. Trying to obtain the separated material. But the gel shows that things are getting better.
So that was with the comment at the bottom of page 131. Pool number 2 and number 3 and proceed with two i121. Number 2 is best. Virtually no 10.5-kilobase. And number 3 is also good. [00:44:00] And S, i125S, is good for the 10.5 fragment. So getting happier with having these two types of DNA.
Back to digesting some bacteriophage DNA as recipients for the purified DNA with what was called delta delta Xba. Two preparations. One 146 micrograms per ml and the other 306 micrograms per ml. But the gel was overloaded [00:45:00] and had to be looked at again. Continuing the DNA fractions. And the bulk phage digest on Monday, January 28th.
So in the end the pool of material called delta delta X was obtained. Total of 330 micrograms as material in which to clone the purified mammalian cell DNA. Looking at the digest on page 137.
Ligations now started. [00:46:00] Tuesday, January 29th, page 139. Ligating half of one and a quarter of the other. About 20 micrograms of each. Into the phage. So these are the purified smaller and larger prefractionated cellular DNA into the test bacteriophage DNA. Being careful not to get anything mixed. [00:47:00] Ligation test looked at on page 141. Conclusion. The gel was too red but proceed anyway. Ligation is OK. So here now is one of the very key pages in working towards the successful result. It’s page 143 and opposite 142. Wednesday, January 30th. The large-scale packaging and plating of the DNA of the two types. The smaller pool and the larger pool prefractionated. And [00:48:00] all of the details are there of the packaging with the sonic extract and the freeze-thaw lysate and how to do it. And then at the top of the page on page 143 it says two positive on the large pool. Number 4 and number 7A. And that two has obviously overwritten some smaller number. I remember very well that I found one doubly hybridizing phage that was grown on the right bacteria and was positive for the [00:49:00] beta probe. And then Nobuyo saw another almost off the film. So the number one was changed to two. But the comment there is two positive, number 4 in the large pool. Eventually. See i161, 62. Meaning that we later showed that they were correct. But this is illustrative of what was needed to get the system working. And I often show this in talking to students at the current time. And the main message from this is not that this was obtaining the positives — although that is the real message. In the bottom right-hand corner of this [00:50:00] page is a little comment that in growing the bacteria, the bacteria which was C1A grown 28th through 29th, that’s Monday and Tuesday. I thought that maybe they had been made too long ahead of time. Stationary culture.
So I diluted them to 1/4 and grew them up again, regrew them for a little while, so they were in the exponential phase. Now the significance of that comment is not immediately obvious. But it’s a very important note because it turned out that for quite some time afterwards things didn’t work, as we shall see, as I turn to the next page, [00:51:00] 145 and so on. There was continuing this type of experiment with in this case getting 1,420 C1A plaques from the pool ligation and from the S ligation 1,756. But it’s very difficult to get anything to work well.
I remember very distinctly what was happening. [00:52:00] So here is Thursday, January 31st, page 147. One possible plaque. But it didn’t light up with the second probe. Trying again Thursday, January 31st, page 149. Packaging and plating of i141F ligation, the fast fragment. [00:53:00] But very difficult to see the plaques. I remember the problem.
And that experiment 0.51 times 103 C1A plaques. Not a very high number. So looking back wondering what was it that had made the experiment work. On page 139, 139 ligation. [00:54:00] The results were on page 143. What had made the experiment work on page 143 whereas I was having trouble?
And this sort of comment here that the i139 pool had one very clear IVS2-positive single isolated plaque, etc. And two very doubtful spots. But one of which turned out to be right, etc. So this was a review of what was happening. [00:55:00] Still having trouble. So Monday, February 4th making new many translated probes. New try pool ligation on page 155, Monday, February 4th ligating, etc. [00:56:00] And the conclusion that the ligation is approximately equivalent to 139. Continue. But on the top is a dead loss. See the following page.
So on the following page is no phages at all on C1A on either plate. So less than 10 to the -6 titration. Showed packaging of control was a little low. And etc. Even lower. But no plaques on C1A. Still haven’t discovered what the problem is.[00:57:00] Dead loss page 156. Trying further to work out what was happening on Wednesday, February 6th, page 159. The old method and new method, etc.
But nonetheless some candidates being looked at, i149F candidate continued from i159. Clearly two positives on film briefly. [00:58:00] Phage minis of these two candidates.
And I don’t know where this will return. But I still remember very well what the solution to getting the experiment to work again was. To use freshly grown bacteria. Exponentially grown bacteria. Rather than steady-state cultures, saturated cultures.
The exponentially growing bacteria gave much bigger plaques. By finding that out by looking to see how the experiment had worked on 139. But here we are on Wednesday, February 6th and Friday, February 8th just checking these candidates. [00:59:00] Redigested on Sunday, February 10th, page 163. With the conclusion that 4 and 7 are OK, 5 and 6 are within the range, etc.
So the critical comment is there in rather small letters that 4 and 7 are OK at 7.6. Meaning the fragment is 7.6 kilobases long. The recombinant fragment. And 5 or 6 are within the P range. So the correct fraction was plated. [01:00:00] I made it for this stage. And I’m happy that the phages — that the recombinant event had occurred in the mammalian cells. And not during the assay. So beginning to really believe things. [01:00:30]