Oliver Smithies:

[00:00:00] Here we are, August 12th, 1992, beginning and ending February of 1993, book psi.  Labeling of oligos, but the little airplane note at the beginning CFII renewed for two years.  That’s 35 flight-in flight instructor, flight instructor of an instrument renewed for two years on the basis of a graduation from refresher course, graduating from a refresher course. Trying to label the oligo, [00:01:00] beginning on the end of previous book.  First attempts at labeling.  Very borderline counts.  Why I have this labeling I have to find out, by going back, why the labeling was being done.

These experiments were really a little bit wild in the sense that I thought that if I had mismatched DNA, that that might encourage recombination, so the idea was therefore to make sequences that didn’t match completely, and to do that, it was helpful to have a [00:02:00] labeled sequence and use that in this type of experiment.

Trying to use the labeled DNA to look for mismatching.  Rough test of the mismatch idea on Tuesday, August 18th, page 7, saying that the HPRT oligo hybridizes well, and was well-labeled.  The [ollie?] oligo hybridizes correctly but [00:03:00] appears to very poorly labeled, et cetera, so having difficulties in making the idea work, but tested anyway and the result being that a bunch of ratios in the experiments, six experiment varying from 59 to 106, very doubtful if there were any differences.  So these are different oligo hybridizations. [00:04:00]

Wednesday, September 2nd, page 11 from Frank Dean, a bunch of nice things that he sent me from Sloan-Kettering.  Human single-stranded binding proteins, and an ADP-ribose polymerase, et cetera.  Frank Dean in Jerry Horowitz’s lab.

[00:05:00] Going on with the experiments, Wednesday, September 2nd, an experiment there that said extra reviews had different construct; however the conclusion, “Can’t detect a single-stranded binding by gel or by ES cell,” this was an attempt to see whether KH21, after exonuclease, with or without human, plus or minus human single-stranded binding protein, whether that made any difference, but the ratio of non-homologous to homologous didn’t vary much in these experiments, 35-49.  “Can’t detect human SSB by the gel, [00:06:00] or by ES cells.”

“PARP, poly-ATP ribose polymerase, PARP, binds to nicks, and try to nick the DNA and add PARP, and see what happens,” so this page 15, Thursday, September 13th, is nicking the DNA.

Experiment with PARP done and [00:07:00] as usual, “no significant effect.  Like human SSB, human PARP is negative.  Call Frank, he suggests” et cetera, was trying different things, but negative result.  Spent many hours of this type of work.

Exonuclease 1, rough tests on degrading to leave 5’ ends.  Exonuclease 1 again, [00:08:00] then acts specifically on single-stranded DNA, it should digest delta-hn exo extensively but not but not delta-hn, Hind.

Exonuclease 7 being thought of on page 21.  All of these different things.  Exo7, page 23, Friday, September 11th and electrophoresis has no effect.  And in the test of homologous versus non-homologous, again, nothing significant. [00:09:00]

Repeat of the 3’-5’ exonuclease 3 on d-delta-hn-Hind, Sunday September 13th, 24.  “The enzyme was too old,” was the conclusion.

Repeat of Exo3 experiment on September 14th, page 29.  “No effect.” [00:10:00] Conclusion: “With this more conservative enzyme action, Exo3, that is to say the 5’ overhang has no effect either on the ratio, or about two-fold on the absolute frequency.”  More tests of the same type, page after page.

ATP-dependent DNA, Monday the 21st of September. [00:11:00] Rather strange entry, on page 38, Monday September 21st, conclusion, “The virus is back,” something not very happy being thought of, but not so clear what it was.

I went to Ohio to do the annual on the Grob 109b motor glider, not without incident.  Don’t remember what the incident was, but there must have been some incident on that trip. [00:12:00] This is back in 1992, and so that’s 23 years ago, and I still have that same airplane, at least I have a share in it now.  Heat shocks, trying some ideas of heat shocks on the [00:13:00] ES cells, and looking at ES cells and photographs of the colonies.

So here was an experiment, heat shock post electroporation rough tests on page 49, Wednesday, September 30th.  One of the two non-heat shock ES cell plates was trypsinized and used it as a standard, and the other was from heat shocked cells.  The cells were zapped as usual, and incubated for three hours at 37 degrees, and then heat shock for 20 minutes at 45 degrees, and then looked to see if it made any difference. [00:14:00] Homologous recombination frequency was, if anything, poorer after heat shocking, changed from 128 non-homologous, rather non-homologous, homologous, and increase to 299 after the heat shock, so negative result, if anything, got worse.  And did a heat shock time series, Wednesday, October 7th.  At least planned. [00:15:00]

And at last, beginning of what I’m thinking is that, better work was carried out in my work with gene targeting, which was a new idea, first stated here on page 53, of this book, 1992, Sunday, October 11th.  AGT, angiotensinogen, and other genes idea.  It’s worth thinking about because it’s the motivation for much of what has happened since.  It was stimulated by the paper from Jeumaitre, et al, 1992, where they were looking at the molecular basis of human hypertension, [00:16:00] and the role of angiotensinogen, and they collected families in which there were at least two members who had at least two siblings, and then these were families in which at least one person was hypertensive, and then they asked whether there was any difference in the angiotensinogen genes in the individuals who were hypertensive, and they found a molecular variant, methionine/threonine difference, and were able to correlate the hypertension [00:17:00] with the presence of the threonine change.  But, they also noted a change in the amount of angiotensinogen, in these persons, so one could not tell whether the change in angiotensinogen level was caused by the genetic change, or by the blood pressure change, and so, I began to think about this, and this was the start of that type of thought.  So, it says it’s stimulated by their paper over the last few days, I’ve considered how to test the human methionine-235 threonine change, versus the concentration of angiotensinogen, versus [00:18:00] hypertension in mice.  Because I began to think that the MT chains maybe only a tag, and not the actual important change, because actually p, position 235 methionine is not present in the mouse; it’s a species-specific substitution.  And I felt that the basic hypothesis of Jeumaitre et al, however, is that the concentration of AGT is the critical variant, not the M-to-T change.  And this can be tested in vivo even without knowledge of most of the controlling elements [00:19:00] by changing the number of copies of the gene.  And this was a new idea to me, and I think, a new idea in general.  That if I used a suitable targeting vector, with an O-type recombination, I could get a chromosome in which the gene was now duplicated, and compare that with the normal unduplicated chromosome, so I could have a two-copy normal person, a three-copy with a 1+2, and a four-copy with 2+2.  And if I made a knockout, I could use the heterozygote as a one copy, so I would be able to test heterozygous knockout one [00:20:00] copy, the normal one, one animal, one copy-one copy, and one and a duplication two, two-two.  This was the beginning of my start of a long series of experiments in which I varied the concentration of a gene product, the amount of a gene product, rather than changing the sequence, or knocking out.

And as I say on page 55 there, the idea can be tested on any of the genes we are studying, and gives the opportunity also to investigate what the compensating genotypes might have, real tests of animal models, the genetics of hypertension.  So, [00:21:00] the beginning of the idea of altering the amount of a gene product, rather than just knocking it out, or altering its sequence.

Going back to the current experiments on Wednesday, October 21st, cold shock, trying to sense the heat shock source, that homologous recombination, before or shortly after the zapping clearly decreases the homologous recombination more than the non-homologous recombination, so try, what happens if I take them down to -5 degrees.  And that experiment was done, cold shock, with the usual result, nothing changed. [00:22:00]

So page 63, and 62 [00:23:00] shows heat shock cells before zapping, heat shocking the cells before zapping, 20 minutes at 45 degrees.  And the ratio before with no treatment was 42 non-homologous per homologous and the heat shock only increased that number. [00:24:00]

Thinking some more about the heat shock experiments, and coming to the conclusion that it was worth doing synchronization of the cells would be abandoned.

Thursday, November 19th, page 67, ACE-duplication being begun.  Kim is doing the AGG duplication, but ACE becomes another good candidate, because of the paper bag, Cambien et al., that had been published in Nature, in October 15th, that the deletion polymorphism [00:25:00] in the angiotensin converting enzyme is a potent risk factor for myocardial infarction; in other words, that polymorphism in the Ace gene changes the level of the amount of the protein, so more angiotensin converting enzyme individuals are at greater risk for myocardial infarction.  So, ACE was a good candidate, again, for what we now began to call a “gene titration,” effect of different doses of the gene.  So, on Thursday, November 19th, page 67, [00:26:00] I talk about that.

Working with old-type recombinant, and large deletion, but knowing from Vicky Valencius’s work, that big gaps can be repaired, but by her work, was quite important in being confident about using constructs with big gaps, and I’ll just look up that reference.

That we are, as I said, November 19th, [00:27:00] in 1992, and Vicky Valencius’s work was published in 1991, and she had shown, published a paper, “Double Strand Gap Repair in a Mammalian Gene Targeting Reaction,” a very pretty paper.  And she said in her abstract, “To better understand the mechanism of homologous recombination.  We’ve analyzed the recombination reaction that it sets a plasmid into a homologous chromosomal locus, using partially-deleted HPRT gene.”  Her experiment showed that targeting plasmids that carry a double-strand break, or a gap, produce [00:28:00] many more recombinants than plasmids that are uncut, and that the gaps could be 200, 600, or 2.5kb, and efficiently repaired during the integrative recombination reaction, so that this is why we began to feel quite happy about using plasmids that had big gaps in them to repair, or to duplicate genes that were too big to include in the construct.  Let me say that again, in a different way.  To get the gene duplication, one has to crossover within the gene with a plasmid carrying another copy [00:29:00] of the gene.  But if the gene is a big gene that exceeds the capacity of the targeting vector, but a double-stranded gap can be repaired; therefore one doesn’t need to make the targeting vector as large, and this is shown quite clearly in the diagram on page 67, Thursday, November 19th, how to get gene titration with the ACE gene, and as we’ll see, that produced some very interesting results.

Making a probe, 5’ ACE probe to get, isolate the gene. [00:30:00] 3’ candidate phage for the 5’ ACE gene, on page 75, Tuesday, December 8th.  Continuing testing. (laughter) On Tuesday, December 8th, tests of the product, page 77, on the opposite page, the conclusion, “Obvious, no recovery of the DNA.  Scrap that experiment.  Redigest and try again.” On page 81, and continuing on the following [00:31:00] page with looking at the various minis on page 85, Tuesday, December 15th, 3’ ACE phage candidates, and conclusion, “It doesn’t look good.”  I decide to go back and rescreen the pics that yielded the 5’ and 3’ ACE candidates.  So, looking at the candidates of the 5’ part of the gene, and 3’ part of the gene. [00:32:00]

So, Thursday, December 17th, page 91, I’m looking at the candidates for the 5’ ACE gene, the ACE gene being basically a two-headed gene, because a 5’ ACE and a 3’ ACE, each of which was functional, one test and one somatic, partly, or more, and so on.  So, looking at on page 91, at the 5’ ACE candidates which were blue, on page 93, the 3’ ACE gene candidates [00:33:00] which are red and trying to prove which were correct.

In the meantime, beginning tests of the synchronization of cells, with the Monday, December 21st, page 99, looking at tests of electroporation during the S phase.  Conclusion, “As usual, no obvious change in the ratio after [FED?] calling, but about a  10-fold or more decrease in the ability [00:34:00] to survive.”

Looking at the phage minis on page 103.  And the red ones, again, for the 3’ gene, page 105, December 24th, Christmas Eve, and Southern blots, and looking back at the results on Sunday, December 27th, and so forth, continuing to find the ACE gene. [00:35:00] 5’ ACE pics revisited, code blue, 5’ Tuesday, December 29th, page 113.  Three candidates, et cetera, but not very strong.

Making new primers and probes for the ACE gene, tests on the probes, page 121, Saturday, January 9th, continuing the work to try to isolate the genes.  Nothing particularly striking, just slogging away.

Thursday, January 14th, page 129, repeat of the 3’ with genomic DNA.  Looks OK. [00:36:00]

Bulk PCRs being used to make some probes on page 131, Friday, January 15th.  “Good stock now available.”

Page 133, Monday, January 18th, we’ve had good expected 5.2kb, BamHI, [00:37:00] with the long 5’ ACE probe.  Need to get a reproducible Southern prior to the phage, et cetera.

Paper suggesting a way of improving the efficiency of positive-negative selection in homologous recombination, published in Nucleic Acid Research, 1992, Bernet-Grandaud, et cetera, et cetera, from Godet’s lab, using an idea to increase the frequency. [00:38:00] It’s an idea that hairpens might protect the (inaudible).  So I can think about that too.

And tests with – so, the idea, this so-called “Splinker [00:39:00] test,” that I had, page 143, Sunday, January 23rd, the final test selected which planned to be, and there’s a diagram.  The TK gene alongside the usual small test vector, delta-hn.  The idea will be to copy Bam, ligating a phosphorylated Bam splinker, exonuclease, and reprecipitate to get a fully covalent structure.  So the idea is to close the ends of the DNA by ligating them into a circular end, so, as it were, make one continuous DNA [00:40:00] with no ends.  The diagram is quite clear, at the bottom of page 143, what the idea is.  The Bam Splinker, Bam-S-linker, “Splinker.”  And, the various experiments to try to get there. [00:41:00]

So we get to a test, check on BHN-TK, the TK gene, alongside the usual small delta-hn sequence, and a couple of candidates, and our TK in the same orientation.

And, the book is ending with, on page [00:42:00] 161 with some new ACE probes that are going to be used to capture the ACE gene.  And so, ending the book, on page 162, Wednesday February 3rd, with a nested amplification of the current 429 product, various ideas on trying to improve PCR by nested amplification.  The result being, conclusion, “Excellent PCR.  The time increase made a difference.  Run a preparative gel, so there is amplification of [00:43:00] the sequences needed to make a probe.”  The end of book psi.