Broken Chromosome Repair by Homologous Recombination
Transcript of Part 2: Molecular Mechanisms of Repairing a Broken Chromosome
00:00:07.23 Hi. 00:00:08.23 My name is Jim Haber. 00:00:09.23 I'm a professor of biology at Brandeis University, near Boston. 00:00:13.11 I'm talking today about how cells repair their broken chromosomes. 00:00:18.28 I previously presented a video about the overall mechanisms 00:00:23.07 by which this double-strand break repair takes place. 00:00:27.03 And what I want to do now is to talk about how we know many of those steps in molecular detail. 00:00:36.24 Just to remind you, I talked before about the fact that tumor cells are really distinct 00:00:43.08 from the rest of the cells of our body because they have failed to maintain the proper arrangement 00:00:48.24 of the 23 pairs of chromosomes that we have in our body. 00:00:53.02 And these tumor cells suffer hundreds of chromosome rearrangements. 00:00:59.08 And these rearrangements are done by joining, inappropriately, segments of DNA from 00:01:04.22 different chromosome breaks together to make these translocations. 00:01:09.00 And there's... as a consequence, what this means is that these tumor cells, 00:01:13.15 though they are efficient in repairing DNA in one way, they've lost the accurate 00:01:18.06 homologous recombination mechanisms by which double-strand breaks can be repaired. 00:01:22.20 And that's the topic that I'm pursuing. 00:01:25.16 And as I also mentioned previously, the source of most of the double-strand breaks that lead 00:01:32.19 to these chromosome rearrangements are not from exogenous sources such as X-rays. 00:01:39.06 They come spontaneously from the process of DNA replication itself. 00:01:44.02 And I illustrated that by the fact that if you deprive these cells of their key recombination 00:01:50.13 protein, called Rad51, within a single cell division what you see are all of these chromatid breaks. 00:01:57.14 And these chromatid breaks therefore represent cases where one sister chromatid is 00:02:02.19 completely properly replicated, but its sister has an interruption. 00:02:05.26 And it's the job of the Rad51 protein to patch that up. 00:02:10.06 And I'm going to talk, now, in more detail about one way in which this occurs. 00:02:16.25 And I'm gonna talk about it using a model organism, Saccharomyces cerevisiae, 00:02:22.17 where we have the highest level of resolution of understanding this repair process. 00:02:28.17 Okay. 00:02:29.17 I also mentioned that there are multiple different pathways by which double-strand breaks 00:02:35.07 can be repaired. 00:02:36.21 There is a so-called double Holliday junction mechanism, which can lead either to crossovers 00:02:42.02 or non-crossovers. 00:02:43.27 And there is a synthesis-dependent strand annealing mechanism, which only leads to non-crossovers. 00:02:50.23 I'm gonna be predominantly talking now about this mechanism, and the way we know 00:02:57.20 how each of these steps takes place. 00:03:00.01 Okay. 00:03:01.01 So, the model system that my lab has studied for quite some time is 00:03:07.11 this so-called mating-type switching system within Saccharomyces cerevisiae. 00:03:13.07 Saccharomyces... budding yeast has two mating types: one called a and one called alpha. 00:03:19.25 And they differ by a set of about 650 or 700 bases of unique information, 00:03:27.00 so that MATa has a set of sequences that MATalpha does not have, 00:03:33.01 and they specify whether the cell behaves sexually as an a or as an alpha. 00:03:39.08 But it turns out that yeast are homothallic. 00:03:44.11 This is a term that applies to quite a number of different organisms. 00:03:49.06 They are able to change from one mating type to the other. 00:03:52.19 And this process of mating-type switching turned out to be an opportunity for 00:03:57.24 learning about how double-strand break repair takes place in detail. 00:04:02.21 In the case of this mating-type switching, there is a site-specific endonuclease, 00:04:07.17 an enzyme encoded by a gene called a HO -- homothallism -- 00:04:12.20 that cuts the DNA just next to those red sequences. 00:04:17.06 And it leads to the replacement of the red sequences by a copy of blue sequences. 00:04:22.10 And obviously, for that to happen, there had to be a source for those additional DNA sequences. 00:04:28.24 And it turns out that there are two other copies of mating-type information. 00:04:33.09 Here, one called HMLalpha. 00:04:35.08 And here, one called HMRa. 00:04:38.20 And these serve as a library, or reservoir, of sequences that can be used to replace 00:04:45.05 the red sequences by the blue sequences, 00:04:47.24 and it turns out, also, to replace the blue sequences back to the red sequence. 00:04:51.11 So, the organism can switch in both directions. 00:04:55.23 What makes these donor sequences so unusual is that they are 00:04:59.24 unexpressed copies of mating-type information, 00:05:03.20 and they are kept in a highly heterochromatic and silent state by 00:05:08.28 very highly positioned nucleus zones that go across this region, 00:05:12.23 which requires a special histone deacetylase called SIR2. 00:05:17.17 So, these sequences are not transcribed, but they can be used as donors in this repair event. 00:05:26.14 And all of the expression of the mating-type information comes from the MAT locus itself. 00:05:31.28 Okay. 00:05:33.18 So, Ira Herskowitz's lab made this all possible by making a galactose-inducible version of 00:05:43.14 the HO endonuclease. 00:05:44.20 So, this enzyme, now, is only turned on when we add a particular sugar to the medium. 00:05:50.15 And when you add the sugar to the medium, very rapidly this sequence is cleaved by the enzyme. 00:05:57.19 And you can see that by using a Southern blot, which is shown here, where the... 00:06:03.03 originally, the MATa locus is cut into a smaller fragment, which you can see here. 00:06:08.17 And then, after some time, what we discovered is we could follow the appearance of the product, 00:06:15.06 because the blue sequences have different restriction sites than the red sequences. 00:06:19.17 And so, on a Southern blot, we were able to sort of follow, in real-time, 00:06:24.08 how this switching process took place. 00:06:26.26 So, the key element of this is that the endonuclease is incredibly efficient. 00:06:31.20 Almost all the sites are cleaved in a very short period of time, 20 minutes. 00:06:35.21 And after that there's a rather lengthy period of time before we see the appearance of the product. 00:06:43.05 And that allowed us to identify lots of slow steps in this process, each one of which represents 00:06:49.12 a key intermediate step in the process of DNA repair. 00:06:54.04 Okay. 00:06:55.04 The first of these that we identified was the fact that the ends of the double-strand break 00:07:01.00 are chewed away by exonucleases. 00:07:04.16 And Charles White, who was a postdoc in my lab, showed that he could see 00:07:11.09 if he separated the DNA on a denaturing gel, so that the two strands -- Watson and Crick strands -- 00:07:15.22 are separate from each other, he could see the appearance of these strange, 00:07:20.13 what looked like partial DNA digests in the... in this Southern blot. 00:07:26.12 And they came because the enzyme, StyI, that he was using cannot cut single-stranded DNA. 00:07:33.07 So, it could cut the sequence at the beginning, but once the DNA was being resected 00:07:40.07 by these exonucleases, now this site became unavailable for cleavage. 00:07:44.14 And that means that we could see a DNA fragment that started at the next StyI site, 00:07:50.03 or even the next StyI site, and he could identify the fact that resection was taking place. 00:07:55.27 And this was the first real demonstration that the resection was happening in a polar fashion. 00:08:01.09 It turns out that, as far as we can tell, all resection goes in this 5' to 3' way. 00:08:08.14 It turns out there are at least three major DNA resection machines, if you will. 00:08:16.25 One of them involves a set of proteins called Mre11 and Rad50, and another protein 00:08:23.02 called Sae2 or CtIP. 00:08:25.26 It does very short resection, and it then allows other resection machineries to be 00:08:31.18 more efficiently loaded onto the broken DNA, one of which is an enzyme called Exo1, 00:08:37.06 which chews off individual nucleotides as it goes down the DNA. 00:08:41.18 And the second is a complicated quartet of proteins, which uses a helicase called Sgs1, 00:08:50.27 which is the homolog of the Bloom's helicase in people. 00:08:54.18 It's prying open the DNA, and then an enzyme called Dna2 is clipping off the DNA so there 00:09:00.20 is what is called a helicase and an endonuclease, which is cleaving off chunks of DNA at a time. 00:09:06.19 But by one of these means, then, we end up with single-stranded DNA that the Rad51 protein 00:09:12.01 is going to use to carry out the next steps in recombination. 00:09:16.04 Okay. 00:09:17.15 Here's just another illustration of this kind of resection. 00:09:23.28 We realized, and other people realized, that you could see this resection much more dramatically 00:09:28.20 if you prevented the recombination from taking place. 00:09:31.25 And so either by deleting the HML and HMR sequences, so there are no donors, 00:09:37.23 or in this case by knocking out the Rad51 protein itself, so there is no recombination. 00:09:42.19 Now you could see these bands appear on the... on the gel because resection is just inexorably 00:09:49.03 going down the DNA, making one site after another not able to be cut by the... 00:09:55.17 by this restriction enzyme, and leading to these apparently large fragments of DNA. 00:10:01.02 So, that allowed us to sort of look at this process and to figure out that, in general, 00:10:06.06 the nuclease is moving about one nucleotide a second, four kilobases an hour. 00:10:11.23 So, it's a pretty slow enzyme compared to many biochemical reactions. 00:10:19.27 Okay. 00:10:22.12 Another thing that is shown here is that if you get rid of the two machines 00:10:26.26 -- the Sgs1-Dna2 machine and the Exo1 machine -- 00:10:30.11 now basically all that resection stops, except for a little bit, 00:10:33.13 which is being done by the... by the last enzyme in this set, the Mre11 complex. 00:10:44.10 In order for this to take place in chromatin... because, after all, we're not really interested 00:10:50.23 in what happens in naked DNA, but what's happening in chromatin in the nucleus... 00:10:57.04 it turns out that there have to be other components. 00:10:59.19 Because these nucleases have a very hard time getting past highly positioned nucleosomes. 00:11:06.06 And so along comes another enzyme, which is known as FUN30. 00:11:12.12 Its name came from its original identification as "Function Unknown Now", 00:11:18.21 and the name hasn't been changed. 00:11:21.09 But what FUN30 turns out to be is a chromatin remodeler, a member of the family of Swi2/Snf2 homologs, 00:11:29.15 which can open and push nucleosomes around so that it is possible for these nucleases 00:11:36.25 to get past these nucleo... these barriers. 00:11:42.04 And so, we've looked at that kind of a process in another way of measuring resection. 00:11:47.18 And that is to use small PCR reactions to say when... to say, what is the state of 00:11:57.09 the DNA at this location versus what's the state of DNA, for example, here? 00:12:02.18 And of course, what happens is that a PCR will copy any template, but in this region 00:12:08.17 here it's all... there are two copies of the template. 00:12:12.06 And in this region here, there's only one copy of the template. 00:12:15.11 So, the intensity of the PCR reaction is going to go down. 00:12:19.04 And you can see over time that the amount of resection is just getting broader and broader 00:12:24.14 as the nucleases keep chewing away the DNA at this rate of about 4 kilobases an hour. 00:12:32.13 If you do this same reaction in the fun30 enzyme deficient cells, which is shown here, 00:12:39.12 now you can see that the resection is highly confined, because it can't just go past 00:12:44.00 all of those other nucleosomes. 00:12:45.27 And there's about a four-fold reduction in the rate of resection. 00:12:50.04 So, what I've talked about so far is the fact that there's a break; the break is made 00:12:55.12 by a site-specific nuclease; the nuclease-induced break is then chewed away by a series of exonuclease 00:13:02.09 enzymes to make single-stranded DNA. 00:13:04.14 Now, I'm gonna talk about the next step in the process, which is how the Rad51 recombination protein 00:13:10.04 gets loaded onto the DNA, and then how it carries out this really critical part 00:13:15.16 of the repair process, which is this strand invasion step, where this single strand of DNA 00:13:20.06 finds a homologous template elsewhere in the genome, with which it can make 00:13:25.08 the base pairs as the initial step in the DNA repair process. 00:13:29.21 Okay. 00:13:30.21 So, I mentioned before that the Rad51 protein and the... and its E. coli RecA homolog 00:13:40.28 were understood by the really brilliant X-ray crystallography of Nikola Pavletich's lab, in order to be able 00:13:47.13 to see how the single-stranded DNA is bound inside this filament of Rad51 proteins. 00:13:55.01 And then, inside this filament of Rad51 proteins, it effects an exchange of base pairings 00:14:02.08 such that what was originally a single strand of DNA now is duplex, and it has displaced 00:14:07.20 from the original duplex DNA a single strand of DNA. 00:14:14.03 To be able to really see how Rad51 binds to broken DNA in vivo, we needed another technique. 00:14:22.13 This is too small to see microscopically. 00:14:25.21 Even with fluorescent proteins, you can't see enough detail. 00:14:30.16 And so we turned to a process called chromatin immunoprecipitation, which I'll briefly describe here. 00:14:38.23 It turns out that if you treat cells with formaldehyde, which is a cross-linking reagent. 00:14:44.12 It will join proteins to proteins. 00:14:47.19 It will also join proteins to DNA. 00:14:50.17 And they are held in this covalent state. 00:14:52.24 If the DNA protein complexes that have been cross-linked in this way are fragmented, 00:14:59.05 either by sonication or some other means, now what you end up with are relatively small pieces of DNA 00:15:05.19 to which proteins are cross-linked. 00:15:08.07 And if you have an antibody against a particular protein -- in this case, Rad51 -- 00:15:14.21 you can immunoprecipitate only those segments of DNA to which Rad51 is cross-linked. 00:15:20.22 And so you pull down and purify a small subset of all the DNA to which Rad51 is cross-linked. 00:15:29.00 Then magically, by just raising the temperature to 60 degrees and holding the cells... 00:15:34.02 the solution for a while, these cross-links reverse. 00:15:37.06 And as a consequence, you can then get rid of the protein and purify the DNA 00:15:41.26 in such a way that you can ask... either by PCR or by sequencing, you can ask what pieces of DNA 00:15:48.04 have been preferentially recovered by chromatin immunoprecipitation. 00:15:54.05 Okay. 00:15:54.19 So, we did that to look at, when does the Rad51 protein become bound to the single strand of DNA 00:16:01.20 which is generated by resection, as illustrated here? 00:16:06.06 And so, here's a Southern blot. 00:16:08.20 It's spread out at the beginning to show very early times, and to show you how efficiently 00:16:13.05 the HO endonuclease is cleaving the DNA into this smaller fragment. 00:16:18.09 And then on the top is PCR directed against the sequences adjacent to the mating-type locus 00:16:24.13 to ask, when did we... when... at what point could we immunoprecipitate Rad51-bound DNA 00:16:31.23 and then analyze it? 00:16:35.00 And you can see here that we could get a robust signal about a half an hour after we started 00:16:41.00 the process. 00:16:43.18 So, that's interesting, because we can see the cut at twenty minutes, and we don't see 00:16:51.26 the Rad51 binding for another ten minutes. 00:16:55.10 There's a... there's another series of slow steps that happens before the Rad51 is 00:17:00.10 firmly bound to the single-stranded DNA. 00:17:04.25 What that other step is it turns out -- and I’m not going to show you the data for this -- 00:17:08.13 is that before Rad51 binds another protein, a single-strand DNA-binding protein 00:17:14.28 called RPA first binds to the single-stranded DNA, and then it's replaced by Rad51 in a process 00:17:21.09 that requires some additional proteins, one of which is called Rad52, and in humans is called BRCA2. 00:17:28.19 And I'll show you a little bit more about that in a minute. 00:17:33.07 Okay. 00:17:34.07 So, here is the mammalian version of this kind of experiment, but now without 00:17:41.07 being able to see a specific DNA sequence. 00:17:44.19 This is just irradiating cells, and so every cell has chromosome breaks after X-ray irradiation 00:17:50.24 or... but the breaks are random with respect to any given DNA sequence. 00:17:56.09 And the advantage of the yeast system was we were looking at a specific site, 00:18:00.12 and could really define it in some detail. 00:18:02.16 But you can see here that if you... you can see the appearance of fluorescent foci, 00:18:09.10 which is by done by looking at an antibody against the Rad51 protein. 00:18:14.04 And so when you treat the cells by irradiation, you can see the formation of these foci. 00:18:19.24 Those foci don't happen very efficiently when you prevent the resection of the DNA ends. 00:18:26.17 They require single-stranded DNA for Rad51 to bind. 00:18:30.03 So, if you block the expression of a protein called CtIP, which is one of those enzymes 00:18:39.05 that is necessary for resection to take place, now you don't see these foci. 00:18:45.22 And as I said, Rad51 requires help in order to load itself onto the DNA. 00:18:51.22 In mammals, the key protein that is necessary for this is the BRCA2 protein. 00:18:59.02 This is the protein that's often mutated in women who have a predisposition to breast cancers, 00:19:05.21 a familial predisposition to breast cancers. 00:19:10.17 And you can see again that in the... in the irradiated cells you can see all these Rad51 foci, 00:19:16.11 but in the absence of the fully functional BRCA2 protein, now these foci don't appear. 00:19:24.08 And the same turns out to be true for another protein, which is a so-called paralog of Rad51 00:19:31.03 called XRCC2. 00:19:32.03 So, you need these additional proteins in order to get Rad51 to load, just as you do 00:19:37.25 in yeast. 00:19:38.25 So, in yeast, we did this by chromatin immunoprecipitation, as I showed you before. 00:19:44.14 And it turns out that in the absence of the Rad52 protein, you don't get this loading, 00:19:50.15 because Rad52 is a necessary part of the loading process. 00:19:54.12 On the other hand, another radiation-sensitive mutant in yeast, called Rad54, 00:20:00.16 turns out not to be required for this. 00:20:02.28 In the absence of Rad54, we have Rad51 loading. 00:20:10.28 This cross-linking technique, this chromatin immunoprecipitation technique, turned out 00:20:15.02 to be also... very... incredibly valuable for looking at the next step of this process. 00:20:21.08 So, after Rad51 has bound to single-stranded DNA, the next step in the process is to 00:20:28.24 search the genome to find a partner sequence with which it can make these base pairs. 00:20:34.13 And of course, in the mating-switching system, that would mean that the broken DNA at MAT 00:20:38.21 is going to find either HML or HMR, those two donor sequences, 00:20:42.18 in order to do this base pair exchange and to set up the rest of the repair, to switch one... 00:20:49.14 from one mating type to the other. 00:20:51.02 And it turns out that since you can cross-link protein to DNA, not only could we see 00:20:58.17 the cross-linking of the Rad51 protein to the original break, but once the strand invasion 00:21:04.10 had taken place, now we could see that the cross-linked proteins would not only immunoprecipitate MAT, 00:21:11.03 but they would immunoprecipitate the HML locus, the donor locus as well. 00:21:16.24 And that's shown here. 00:21:19.03 So, MAT... 00:21:22.10 PCR analysis showed that the Rad51 protein was easily visible by a half an hour 00:21:28.10 after we initiated the double-strand break. 00:21:30.20 But you don't see Rad51 bound to the HML locus until considerably later. 00:21:35.20 So, the search to find the HML locus and to be able to facilitate this kind of 00:21:42.28 repair event is a time-dependent process. 00:21:46.14 And we could see that by using chromatin immunoprecipitation. 00:21:52.02 So again, just to review what we've said so far, there's a break. 00:21:56.10 The break gets resected. 00:21:58.14 The Rad51 protein, with the aid of Rad52 and other helper proteins, loads onto the single-stranded DNA, 00:22:06.16 initiates a search across the genome in order to be able to find its partner, 00:22:12.05 makes those new Watson-Crick base pairs, and now the next step in the process would be to 00:22:17.26 initiate new DNA synthesis to be able to copy the new sequences that are going to be used to 00:22:23.11 replace the sequences originally at the MAT locus. 00:22:27.22 And we worked out a way to see the new DNA synthesis. 00:22:33.27 So, here, what we realized is that if we had a pair of PCR primers, one of which is specific 00:22:40.15 for the donor and one of which is specific for the MAT locus, 00:22:44.08 they would originally be 200 kilobases apart. 00:22:47.22 And there would be no possibility of getting a PCR product. 00:22:51.04 But once the strand invasion has taken place, and just 50 base pairs of new DNA synthesis 00:22:56.22 has happened, now there's a covalent piece of DNA which will link these two primers together. 00:23:03.04 And therefore you could see the appearance of this PCR product as an indication of... 00:23:08.05 when had new DNA synthesis occurred? 00:23:10.24 As you can see by just looking at this... at this gel. 00:23:14.17 So, we could figure out when exactly this primer extension 00:23:20.24 -- the recruitment and use of DNA polymerase -- had taken place. 00:23:27.21 Here, we got another surprise. 00:23:29.28 And that is that the Rad54 protein, which is another chromatin remodeling protein, 00:23:35.21 turns out not to be able to do this step. 00:23:39.06 And so, even though Rad54 was not needed to do this initial strand invasion step, 00:23:44.11 as I showed you before, it turns out to be critical for doing this next step, 00:23:48.17 which is the primer extension. 00:23:50.06 And we think that there's some kind of rearrangement of the chromatin in the donor locus that requires 00:23:55.28 the Rad54 protein in order for the recruitment of new DNA polymerase to come, and to be able 00:24:01.28 to initiate this repair process. 00:24:04.28 So, if we do this in... sort of... with more time points than I've shown you in this illustration, 00:24:12.13 it takes about 10 minutes from the time we can see the double-strand break until 00:24:16.08 there's the recruitment of Rad51. 00:24:18.16 It takes at least 20 minutes after at that point... when the Rad51 filament can find 00:24:24.15 its partner, which is located 200 kilobases away on the same chromosome. 00:24:29.19 It takes a lot longer to find its partner when the partner is in another, 00:24:34.28 different chromosome. 00:24:36.08 And then after that, there's still another 20 minutes after this strand invasion 00:24:40.17 has taken place before the Rad51 protein can be gotten out of the way and the new DNA polymerase 00:24:47.08 has been recruited in order to initiate new DNA synthesis. 00:24:50.13 So, there's a whole series of relatively slow steps that we could identify as important 00:24:57.06 in this repair of a broken chromosome. 00:25:01.02 Okay. 00:25:03.07 I mentioned that repairing the break when the donor and the recipient are on the same chromosome 00:25:10.09 is faster than when the donor and the recipient are on different chromosomes. 00:25:15.16 And we can see that by using that PCR reaction... as to how much slower 00:25:21.06 the interchromosomal recombination event is relative to the intrachromosomal event. 00:25:27.18 But that's just one place. 00:25:29.02 I mean, we put a donor somewhere, and we asked, how did the repair take place? 00:25:34.16 But you'd like to know more globally, what does the position of the chromosome relative 00:25:39.18 to the donor have as a way of influencing this process? 00:25:45.06 And so here we took advantage of a work from Noble's lab, in which they used what is called 00:25:56.09 chromosome conformation capture, by using, again, formaldehyde cross-linking of DNA to DNA, 00:26:02.12 to ask, which pieces of DNA are more likely to be close together, 00:26:08.03 and which pieces of DNA are likely to be more far apart within the yeast nucleus? 00:26:11.22 And so, this is a cartoon of what... where all the chromosomes are inside the yeast nucleus. 00:26:17.24 We already knew from lots of cytology that all the centromeric regions are 00:26:24.07 clustered together at something called the spindle pole body, the centriole. 00:26:27.25 And so all those sequences are clustered together. 00:26:30.22 But then the question is, does it really matter if we make a break at this red sequence, 00:26:36.03 for example... does it matter whether the donor is here or here or here? 00:26:40.17 In other words, does this cartoon really predict how efficient recombination will be? 00:26:47.08 And so, to do this, we changed the system. 00:26:51.09 We put the HO cleavage site inside just another piece of DNA, not the original mating-type sequence, 00:26:57.04 which has now been deleted, but a gene called LEU2. 00:27:01.12 And then we placed a donor sequence that shares about two kilobases of homology 00:27:07.13 with the broken segment elsewhere in the genome. 00:27:10.00 And we just measured by viability how efficient was this repair event. 00:27:14.11 If it's very efficient, then the cells plated on galactose, where the HO endonuclease is induced, 00:27:20.03 should be very similar to what happens on glucose where all the cells can grow. 00:27:25.12 But if there... if repair is very difficult, then this number on galactose plates is going 00:27:31.00 to be a low number. 00:27:32.23 And so, to do this experiment, we put a double-strand break in one location, and then we made 00:27:38.08 20 strains where the donor was in many different locations. 00:27:41.09 And they were chosen in part because some of them were estimated to be close together, 00:27:46.25 and some of them were estimated to be far apart, relative to the site which is being broken. 00:27:53.06 And so, this is the result we got. 00:27:55.05 It turns out that the likelihood that two sequences are close together before 00:28:01.04 we induce double-strand break has a very strong effect on how efficient the repair event is. 00:28:06.28 Some of these events are 50 perfect efficient, and some of these events are a few percent efficient. 00:28:12.13 And that's all dependent on location. 00:28:15.00 It turns out, in mammals, the probability that those rearrangements 00:28:19.19 -- those translocations that I showed you at the very beginning, that you see in cancer cells -- 00:28:25.11 is also very much indic... predicted by the proximity of the segments of DNA 00:28:30.25 that are undergoing rearrangements. 00:28:33.11 So, this seems to be a general feature of DNA repair, that the really slow step 00:28:40.27 in finding sequences to do the repair is searching for them over huge distances. 00:28:46.04 And the closer they are to begin with, the more efficient they are in doing this repair. 00:28:51.18 So, I mentioned previously that the Bloom's helicase plays an important role in the resolution 00:28:58.23 of some recombination structures, by being able to unwind double Holliday junctions 00:29:04.21 so that they don't end up being crossovers, and lead instead to non-crossover outcomes. 00:29:11.02 And the Bloom helicase carries this out in a... in a way that is very obvious from 00:29:18.28 this cytology, here. 00:29:20.15 Here, we see sister chromatids undergoing frequent exchange events, crossover events, 00:29:28.28 compared to the wild type. 00:29:31.08 Because this process of dissolving intermediates is incredibly efficient in mitotic cells. 00:29:40.26 And we showed in Saccharomyces that this same Bloom's helicase -- its homolog, called Sgs1, 00:29:47.12 has a similar role. 00:29:49.06 So, again, using the mating-type switching system and the inducible HO endonuclease... 00:29:57.25 except here we placed a donor on a different chromosome. 00:30:01.08 This donor has a mutation so it can't be cleaved. 00:30:04.27 And the restriction sites are arranged so that if you see a crossover, you can see that 00:30:10.04 that product is different from the non-crossover. 00:30:12.21 So, in wild type, there are very, very few crossover events 00:30:16.22 associated with this double-strand break repair event. 00:30:19.22 But when we get rid of the Bloom's helicase, Sgs1, or the topoisomerase, you can see 00:30:25.10 a very large increase, almost a four-fold increase in the level of crossing over, 00:30:30.23 which is associated with their absence. 00:30:33.08 And so this dissolving pathway is preserved all the way from yeast to humans. 00:30:41.04 So, I've gone through a number of steps that we are able to visualize in real time 00:30:48.28 by using the yeast mating-type switching system and related systems, with an inducible site-specific nuclease, 00:30:55.03 to be able to monitor in real time these kinds of events. 00:30:59.20 We sort of grouped them all into something we called in vivo biochemistry. 00:31:04.06 And we could describe each of these events in some detail. 00:31:09.14 In the next video, I'm gonna talk about another consequence of doing this kind of work, 00:31:15.21 which is to reveal how mutagenic this supposedly accurate repair event actually is. 00:31:22.18 It turns out that, compared to normal DNA replication, repair by these kinds of mechanisms 00:31:28.13 turns out to be almost 1000 times more mutagenic than a normal replicative event. 00:31:35.07 So, even though repairing in this way prevents the kinds of crazy translocations that I showed you 00:31:40.07 at the very beginning, it is not without risk. 00:31:43.27 And it turns out to be a reflection of the fact that this repair mechanism is 00:31:48.11 not using the same machinery, in detail, as the normal replication process. 00:31:53.08 So, all of this work that I've talked about was of course not just done... or, in fact, 00:31:59.03 very little of it done by me. 00:32:01.08 It was done by a huge number of people in my lab. 00:32:03.26 This is not a picture of my current lab, but rather a reunion of a large number of people 00:32:09.17 who have worked in my lab. 00:32:11.24 This slide just lists all the undergraduates, graduate students, and postdocs who have 00:32:18.04 worked in my lab over a very long career in order to learn a lot of the things that I've told you. 00:32:24.23 But in addition to these many people with... who have been in my lab, we also have collaborated 00:32:30.22 with an astonishing number of other labs in order to accomplish this work. 00:32:35.11 And these are simply listing all of the principal investigators at different labs. 00:32:39.24 Of course, they too had students and postdocs who were part of this process. 00:32:44.26 So, learning all of this material took quite a lot of effort on the part of many labs 00:32:51.00 to really define what we know about double-strand break repair. 00:32:54.14 Thanks very much.