Oliver Smithies:

[00:00:00] This is book omicron beginning April 28th, 1987 and going through February of the next year.  So here we are with the PCR machine and actually an image of it rather entertaining.  We still have that old board somewhere.  Hexapus mark one, MK1, with the three valves.  Altogether there are five valves on that that I can see.  One, two, three.  No, four valves and various timers and so on.  Anyway here it is with a 2-liter bath.  These are rather small water baths.  And the temperatures [00:01:00] of the water baths and inside the block, etc.  So the reaction temperature was aimed at 59 and the length of time that it took to get to that temperature is recorded.

So the water bath was pretty constant, all right, inside the block, starting at 32 degrees.  Reached the reaction temperature essentially in two minutes.  The melting temperature starting from the reaction temperature of 54 which was the end takes about two and a half to three minutes to get up to temperature.  So everything is rather very slow really.  Too long time.

[00:02:00] And so on the next page one of the baths is changed to a 7-liter bath and also improved design.  And one can see the improvements of the speed at which things are coming to equilibrium.  The tube takes about two and a half minutes to equilibrate.  The melt to anneal.  This was anneal to reaction.  Melt to anneal takes about one and a half minutes with a new block.  The block was changed.  The block that was made for me.

The block was modified by [Joe Svoboda?] who did all the work on making beautiful aluminum blocks [00:03:00] that matched the tubes very well and were a joy to work with.

So Monday, May 4th, page seven.  Retesting where the top of the base of the chamber had been modified to increase the area of contact to the water.  So improve the speed of reaction.  And the result on page nine showing that the tube changes are now about two minutes to get to something like the right temperature.  But not really.  Still not very good.

Still more cut down chamber tests on Monday, May 11th, page 11.  Acceptable rates.  [00:04:00] One reached the melting temperature approximately in 45 seconds.  Both the annealing and the cooling temperatures were fairly rapid.

Here was an attempt on page 13, Tuesday, May 12th that was rather entertaining.  To try to speed things up.  The temperature of the melting was a little difficult.  So to try to get over this problem I thought that I could use antifreeze, ethylene glycol, instead of water, and have a higher temperature than 100 degrees.  So I set the system up with some old antifreeze.  [00:05:00] Type used in car radiators and common in Wisconsin.  So this was set up with ethylene glycol.  So I had, I remember, gone to give a seminar somewhere I think in New York.  And I called back to the lab and found out.  Nobuyo answered.  Said that everybody was made sick with the ethylene glycol decomposition products, which were evaporating from this old antifreeze.  I said, “No, no, let it go on.  It’s only in your head.  The headache is only in your head.”  But it really was induced by the horrible material.

Nobuyo I think allowed the experiment to go to completion.  But with some somewhat [00:06:00] pointed comments on my return.  So here the apparatus was reset up on Monday, May 18th, page 15, with a bigger melt bath.  And distilled water.  And the results of the temperature changes and time scale are really quite good on the opposing page.  And there is a picture of the new machine where now there are two big water baths.  And only one small one.

Later even the small one got replaced by a big one.  But now we were talking about the modified cell.  So it was hexapus mark two.  With still a comment.  Larger melt bath.  But [00:07:00] use a bigger reaction bath.  And set the temperature higher.  Is the comment.

So various things that at least were planned are talked about on Tuesday, May 19th.  Modification that it would be a good idea to convert the control box to a digital form for each step.  Include a cycle counter, etc., etc.  To be able to do things more precisely and easier.  Plumbing should include a digital thermometer in the block, etc.

And find three uniform baths, not three different size baths.  And have a digital temperature, etc., etc.  And what one might try.  So anyway here is omicron 17 test [00:08:00] which is the reaction bath is 64.  The melt bath is 97.  The anneal bath is 36.  And we see a gel which I think was run by Hyung-Suk Kim on the left page showing what the products were.  What was obtained with 10 nanograms of plasmid in the presence of delta left primer, delta right primer, and a mixture of the two, hoping to find the specific band at 550 base pairs.  But not [00:09:00] seeing that in this initial test.

So trying different temperatures.  Tuesday, May 19th and Wednesday, May 20th, 37, 60 degrees, and 90.  Omicron 17.  And omicron 19.  Temperatures 45, 60, and 90.  So the anneal temperature was changed from 37 to 45.  The result being a very weak signal.  [00:10:00] Not much change from Tuesday.  Suspect some error.  Spurious (inaudible) not removed anyway.  So Thursday set up again with a larger reaction bath.  With many bands seen in all of the sample.

Not many with delta left, 10 nanograms of different plasmids being tested.  It’s [00:11:00] not very well documented.  Page 21 still chasing this problem of the water bath.  So Friday, May 22nd an even larger reaction bath made so that the temperature increases faster and stays more level.  Because if you look back at the other ones the temperature would continue to rise after it had reached equilibrium because of some cooling that had occurred in the water bath that was being tapped.

Looking at what various odd bands were.  [00:12:00] For example 170-base-pair fragment.  Must require both the left and right since it doesn’t occur in either delta left or delta right alone.  Is this a loop out product was the question.  So the 170 band is something that one didn’t want.

Trying to understand the result.  So here on page 27 Tuesday, June 30th.  So that’s quite a delay from May 19th.  There’s a gap, the next entry being June 30th.  Meetings and vacation over.  And in that time several things had happened as we shall see.  But going back to work again with the [00:13:00] PCR machine.  Various temperatures were increased.  And it was much improved when an air leak was fixed.  And the temperature got up rapidly in the melt and rapidly down and everything stayed reasonably constant in the reaction phase.  So it was pretty good.

But checking with Kim on amplification some possible problems in the machine.  Not really likely to be the cause of the trouble.  On June 30th, [00:14:00] Tuesday, page 27.  So various reaction mixtures were set up using 2.5 nanograms of pSV2neo, 5.7-kilobase plasmid, as the test plasmid, which we were hoping to be able to amplify with the PCR machine.

Various comments on page 29.  Must increase these concentrations.  That’s of dNTPs.  As a result of some calculations suggested by Kim.  Or decrease the DNA substantially.  [00:15:00] Cetus uses 2 millimolar dNTPs with Klenow and I said let’s assume this is OK.  And working out how much primer one would need.

Need quite a lot of primer.  For example 5 micrograms or thereabouts of total primer would be needed.  So not realizing that you could run out of primer.  More temperature checks July 1st, Wednesday, page 31, acceptable result.  High concentration experiment on Thursday, July 2nd, page 33.  Results show no bands in any sample.

[00:16:00] Must be the new batch of enzyme.  More thoughts on the assay on Friday, July 3rd, page 35.  Thinking about the concentration of the dNTP with another image of hexapus mark two on the left with now two big water baths that are visible.  The third water bath I can’t see.

Here was found (inaudible) slow response still.  Concentration nucleotides being investigated [00:17:00] next page.  And now going to unlabeled DNA amplification tests Tuesday, July 7th, page 39 with Kim’s reaction buffer now.  And this experiment was done a couple of times with this.  And the result was satisfactory that after 20 cycles there was a visible band in one reaction and in the other reaction even heavier visible band in the correct place, 550 base pairs.  [00:18:00] So 10 cycles and 20 cycles both giving visible band.  Reaction setup.

Tests of other enzymes and concentration continuing on page 43.  Consider increasing the primer for the assay.  The conclusion that says though the gel is not very good, it’s clear that the experiments are working now.  More cycles and twice the primer on page 45.  Decreasing the temperature of the melt bath from 97.5 to [00:19:00] 95.5 didn’t cause any problem.  Response was faster now and the results are very easily seen.  Heavy bands.

Trying to work out what the expectation is.  The results are in 13 cycles the amplification is about 2,000-fold.  Theoretically it should be 8,000-fold.  But the [00:20:00] conditions are now acceptable.  Next try six, seven cycles.  Which give about 20 — oh.  Going from 13 cycles up to 20 cycles should give about a 20-fold increase in the amount of DNA.  It isn’t too far wrong.

Seven cycles should be 20x, 2 to the 7, 2, 4, 8, 16, 2, 4, 8, 16, 32, 64, seven cycles, yeah, it should be 128.  But [00:21:00] in fact it was about 20-fold but not too bad.  Continuing to try different conditions in the following page.  So page 51.  Beginning to reach the conclusion that it wasn’t necessary to use an anneal test at all.  So the temperature changes are shown [00:22:00] when there was only a melt and reaction temperature bath with an anneal temperature.  And concluding that it really wasn’t necessary to have an anneal temperature.

So this was tested on Sunday, June 12th.  Versus on Saturday the three-cycle result is shown.  And on Sunday the two-cycle result is shown.  And there is no difference between them.  So the conclusion is that anneal is not needed.

Try now a different reaction temperature from 65, try 70 degrees.  So [00:23:00] Sunday, July 12th showed us that you really didn’t need to do an anneal.  Subsequently we stopped using it completely.  And there’s a test now at 70 degrees reaction temperature on page 53, Monday, July 13th.

The result 70 degrees is OK for now.  Could be a tad low.  But specificity is good, we can accept it.  And so it’s transferred in a Southern blot for Kim to hybridization.  Hybridization very very strong positive on the band which we can see visually in the ethidium bromide gel [00:24:00] anyway.  So it’s rapidly becoming unnecessary to do the hybridization.  But testing at 75 degrees.  Rough test without annealing.  But reaction temperature 75 shows nothing really visible on the ethidium bromide.  And even the hybridization extremely low amount so that there’s a discrimination in the temperature.

And repeating then on Thursday.  That was with annealing at 75 degrees.  And returning on Thursday, July 16th to annealing again.  No annealing.  But 70-degree reaction.  For a better [00:25:00] Southern.  An enormously positive Southern and a very good band on the ethidium bromide.

The system is working now.  So controls on the following day delta left and delta right primers.  But getting worried because there’s something there that shouldn’t be.  Contamination is a probable cause because delta left delta right is approximately equivalent to delta left or delta right.  So that there must be some contamination.  And we began to realize how critical it is to be sure that the products of the reaction are never anywhere near the region [00:26:00] where the setup of an amplification is being carried out because one or two molecules is all it takes to get a false positive.

So where do we go to on this?  Page 60 is here now.  Page 61.  Friday, July 24th.  New solutions and looking to amplify 1-kilobase- and 2-kilobase-long fragments.  So now have some primers that should be able to do 1 kilobase and 2 kilobases.  Primers called [R1000?] and [R2000?].  With no carrier.  And the result occurs on the next page, [00:27:00] July 24th, Friday, page 63.  And the results show that the 2-kilobase fragment did amplify.

So now we’re on book pi, March 17th, 1988.


Hold on just a minute, Oliver.  [00:27:23]


[00:00:00] OK.


So I’m going to return to page 61 and pick it up again.  Try that a bit better.  So page 61, Friday, July 24th.  Here I’m trying to think about ways of trying to get bigger amplification.  Bigger pieces.  Bigger recombinant fragment pieces, 1-kilobase- and 2-kilobase-long.  So the mixture is the mixture called omicron 61 reaction mix without primers.  [00:01:00] And that is the composition of the mixture with no carrier DNA.  And then the primers.  What primers should be used.  And then on the left-hand page delta left primer and delta right primer.  And two new primers called [R1000?] and [R2000?] the right-hand primers, attempting to get that to work.

And the following page.  Page 62.  With a comment that the autoradiograph results after transfer show that the 2-kilobase fragment did amplify.  But even the buffer control had a problem.  [00:02:00] With these primers.

And so that those are the various solutions being used, discussed on Friday, July 24th, page 63.  It’s an attempt to get a different product.  So here we are on page 65, Saturday, July 21st, comparing the 550-kilobase normal previously used recombinant fragment with a 1-kilobase, 2-kilobase.  And the control delta left and delta right.

And it says the results are OK, 1-kilobase is OK.  And nothing was obtained with the 2-kilobase one.  [00:03:00] But the standard reaction mix was better.

So continuing these tests on Sunday, July 26th, page 67, 60-degree reaction temperature now with 550 [00:04:00] base pairs versus 1-kilobase and 2-kilobase and delta left and delta right.  With a result.  [00:04:11]


[00:00:00] Then you have to go and have your lunch.  So where was I, Jenny?  Page.


I’m not sure.


So picking up anyway on page 67.  Sunday, July 26th.  Reaction temperature was 60 degrees.  And looking to get 550-base-pair fragment versus 1-kilobase and 2-kilobase and delta left delta right.  For neo I think.  Ran 20 cycles at 60 degrees.

[00:01:00] With various mixtures.  Results still show nothing on [R20?].  So remake the mix and try again.  So again 60 degrees Sunday, July 26th.  Thinking there must have been some contamination of things.  Or that in some way I’d been stupid or something.  Page 68.  Or been stupid underlined twice.

So set for 30 cycles.  Faint band at 2 kilobases being detected.  [00:02:00] So various tests with retests of various earlier mixes on Monday, July 27th.  Continuing to have problem.  More careful check of the buffers.  Results.  Oh my God.  Monday, July 27th.  [B3?] and [N7?] are both negative.  [B1?] and [N5?] are both positive.  Etc.  So evidently quite confused with these results.  So next page.  Usual business when one is confused.  Tuesday, July 28th, page 77.  Review of the result.  And the overall conclusion.  At 70 degrees [00:03:00] with the three available enzymes expect 30 cycles to be poorer in false positives and only a little better in true positives.  At 60 degrees expect 30 cycles to be very poor on false positives.  So still trying to refine the conditions.  New set of trials 70 degrees which Aida was going to do on Tuesday, July 28th.  Repeats of her test Wednesday, July 29th without any carrier, 70 degrees.

With the results on Thursday, July 30th, page 85 and 84.  [00:04:00] With the conclusion no amplification in any of the buffer sample.  The problem must have been contamination.  So nothing is positive in any of these tests.  Except for the 30-cycle results.  Working out the difficulties.  [00:05:00] Test of new polymerase.  Or recheck of a new Taq polymerase on page 87.  Then now looking on page 89, looking at 65 degrees.

With a conclusion.  Etc., etc.  That says just repeat the whole set at 70 degrees.  And on Friday, July 31st, anneal at 70, reaction temperature 71-degree bath, and melt at 94.  So it’s basically a two-step reaction.  And the results are [00:06:00] very satisfactory.  The comment that the super result, 70-degree is better than 65-degree, 0.5-kilobase, 1-kilobase, and 2-kilobase are close to perfect.  No visible buffer signal.  Etc., etc.

So very happy result.  The gel is very pretty.  Homing in on the result.  With a comment in pencil.  Gloves.  No templates until the [B?] series is closed.  Etc.  So realizing that contamination really can be a problem with Taq polymerase.

One has to be very careful not to let any product get [00:07:00] near the reaction mixture.  Otherwise you get false positives very easily.  And 20 cycles may be enough.  Page 95.

Testing different concentration of dNTP page 99.  And so on and so forth.  Now beginning to want to test with very small amounts of DNA Tuesday.  Or Thursday rather.  October.  No.  Tuesday, August 25th.  [00:08:00] Here.  Let’s just check this.  Looks like Tuesday, August 25th.  Aim to take 10, 100, 1,000, 10,000 cell.  Make DNA in micro volume for amplification and see if one can make the experiment work with very small numbers of cells.  I know that in the end we were able to do it with a single cell so that this must be working up towards that.

Some fairly big gap between the August 25th and the next entry is Friday, October 30th.  Must have been teaching.  [00:09:00] Some ideas with Tom Doetschman and Nobuyo Maeda on various inserts.  Thinking about maybe testing metallothionein insert.  In a lab submeeting on page 107, Monday, November 2nd.  This is planning the [dimian?].  It turned out to be very useful.  Tom would take — Tom Doetschman that is.  Would take HPRT- male mice which have come from Martin Hooper and will prepare fibroblast cell lines, etc.  And can be used to compare this [00:10:00] and the other (inaudible) HPRT- going to HPRT+ so this really is the beginning of what eventually became our first successful experiments because we got these cells, HPRT- mice actually from Martin Hooper.  This is a result of Nobuyo and me going to a meeting in which Martin Evans talked about HPRT- mice obtained by insertion of retrovirus and Martin Hooper talked about HPRT- mice obtained by selection for a deletion.  So one was [00:11:00] inactivation by a retrovirus, Martin Evans.  And the other was a deletion.  And Nobuyo realized that the various tests we’d been doing with HPRT construct meant that we could hope to correct either HPRT-.  Either the one of Martin Evans or the one of Martin Hooper.  Going from – to +.  That turned out in the end to be very wise because you can go from HPRT+ to HPRT- in numerous ways.  But going from HPRT+ to – can be many ways.  Going from HPRT- to HPRT+ is much more restrictive.  Especially when the – is of a [00:12:00] deletion, because it means there won’t be any false positives.  And when Nobuyo and I approached both Martin Evans and Martin Hooper and suggested we try an experiment jointly Martin Evans said, “OK.  Give me the construct and I will try it with my ES cells,” back in his lab.  And Martin Hooper said, “Oh, well, I’ll give you the ES cells of the mice and you can do the experiment in your lab.”  And so we in a sense divided the two approaches.  It turned out that Martin Evans going back from – to + had a much higher background because of the retrovirus elimination whereas – to + in Martin Hooper’s [00:13:00] was zero background.  So anyway here a lab submeeting Monday, November 2nd talking about this.  There’s a big gap between the two because we’d been to Europe and been to these meetings.

And so Tom will take HPRT male mice got from Martin Hooper’s female and prepare fibro cell lines, etc.  And Ron Gregg will test the HPRT- mutants for fibroblasts.  So we’re going to test these ideas on fibroblasts first, not on ES cell.

[00:14:00] So looking on Tuesday, November 3rd.  Various tests that might be made.  And realization that one had to be very careful when using feeder layers that the feeder layers can kill an experiment because you get positives in large capitals at the bottom of page 109.  Feeder cells kill this experiment.  Except with primer four or in the knockout direction.

So we had to do it in the right direction.  Otherwise feeder cell contamination would be a problem.  Thursday, November 12th is a little insert.  My having directed a week lab in Biocore course in Wisconsin.  It was very successful.  Where one [00:15:00] had the senior scientist participate in undergraduate teaching.

And they had been used to let students see the change in hemoglobins that occurs between embryonic day 11 through day 17 in mice.  And look at the different hemoglobin.  And also correlate that with the type of cells.  And so there’s a summary of those results on page 110 and 111.  The last time I was going to teach that course.

So a new work we thought of December 19th.  Some ideas with ES cell genomic DNA.  [00:16:00] Genomic knockout GENKO.  Generic knockout was the idea.  Try this more specifically.  Knockout using a generic knockout GENKO (inaudible) having something that was a generic way of knocking out a gene.

Still thinking about ES cell and growing ES cell cultures on the following pages.  And with images on experiments that were begun on [00:17:00] page 115 Monday, February 1st.  ES cell culture going to be prepared in such and such a way.

And doing this experiment beginning Tuesday, February 2nd.  Plating trypsinized ES cells from [Phil Howell?], etc.  And on February 3rd, Wednesday, page 121 one has the beginning images after one day in culture of these different balls of cells.  [00:18:00] Thursday in two days.  Clear colonies in both series.  But perhaps aggregates.  Not quite understanding yet what’s going on.  Three days.  More images of cultures.  Four days.  And five days.  Six days on page 131.  Seven days and so on and so forth.  Images of these colonies.  And the cells growing out from these colonies.

So it’s worth just looking at Tuesday, February 9th, page 133.  Now the [00:19:00] comments on each one.  For example o115.  That’s omicron page 115 dot point six.  Still got some free colonies, etc.  And all the different ones.  For example omicron 115.1.  Many poor discolored colonies that are partly embryoid.  And see over page for conclusion.

Nothing very striking.  Starting again on page 137, Tuesday, February 9th, set up [00:20:00] six bacterial p60 dishes with 1.5 ml of — and it gives the composition of what was plated.

Looking at them at day zero, February 9th.  And continuing the next few days.  So that up to six days page 147 Monday, February 15th.  The majority of the colonies in agar are still looking good.  Although some of the larger ones are getting dark centers.

One is looking even better, etc., etc., [00:21:00] looking at these colonies.  Seven days.  Page 149.  Eight days.  Page 151.  Wednesday, February 17th, omicron 137.  A.3.  Top quality.  Some darker but most are good.

And now the larger colonies are visible to the naked eye.  Feed once again on Thursday and see what happens.  Eight days.  This pair has now deteriorated and junked it.  Images of the colony.

[00:22:00] These colonies were growing so well that the pH would change.  So page 155 Wednesday, February 24th.  Vigorous metabolism which made the cultures yellow despite Ed Shesley feeding them on Sunday.  This is Wednesday.  But now know that we have an L-form or Mycoplasma infection.  It also prevents differentiation and [Taq?] selection.  So beginning to know about the difficulties of Mycoplasma infection.  [00:23:00] So here is an example on Friday, February 26th, 16 days.  Page 157.  Some comment.  Abandoned on 19 days.  The colonies were still alive.  And in a different pen.  Subsequently learned that these cells probably had Mycoplasma.  Hence the failure to differentiate, etc.

So learning the difficulties.  So later on when we were doing these experiments more routinely we always ran Mycoplasma tests occasionally to make sure that we didn’t have this hidden problem.

Much more difficult to recognize than a simple bacterial contamination.  [00:24:00] So selective stop of function tests being thought of on Monday, February 29th.  And the last two pages are Thursday, March 17th.  A simplified version of the controller to be built for publication purposes.  And so there’s a circuit diagram of the simplified version of the PCR machine.  Which did in fact get published.  And that’s the end of this book.  I’ll take a break (inaudible) Jenny.  [00:24:45]