Hi. My name is Jim Haber. I’m a professor of biology at Brandeis University, near Boston. I’m very interested in how cells repair their broken chromosomes. And I’m especially interested in how they use homologous recombination to preserve genome integrity. When we were born, we had 23 pairs of chromosomes from our mother and our father that had a particular chromosome arrangement, and that arrangement has stayed stable through all those rounds of mitoses in order to produce the trillions of cells that make up our body. The exception to this genome stability is what happens in tumor cells. And here, you can see at this microscopic level truncations, translocations, inversions… all sorts of chromosome rearrangements. And of course, many other alterations that you can’t see without going down to the level of DNA sequencing. Different tumors have different rearrangements, but all of them are somehow creating these kinds of chromosome alterations, many of which are not important in the development of the tumor. But sometimes these rearrangements turn out to drive the cancerous nature of these cells. One example is the so-called Philadelphia chromosome, which is found in chronic myelogenous leukemia, where two segments of genes are joined together. And what happens is essentially that a perfectly nice gene is turned on at the wrong time, and drives tumor growth, when otherwise this gene would be silenced. So, some of these translocations are actually responsible for some of the phenotypes that cancer cells have, but many of the other rearrangements you see are just the consequence of joining pieces of DNA together without any actual consequence for the life of the cell. The failure of these cells to maintain their stable genotype is because they have defects in the homologous recombination machinery that I’m gonna talk about. As I said, they have very efficient ways of joining segments of DNA together which have become broken, but they no longer have the ability to put these segments back together in an orderly fashion that preserves genome integrity. And that’s really what I’m gonna talk about for the rest of this talk. Okay. The source of the breaks that these chromosomes have comes from replication. It doesn’t come primarily from radiation or from other external agents. It’s the process of replication itself which is… which is incredibly accurate, but, nevertheless, every time this much DNA is replicated in the cells of your body there are breaks, and these breaks have to be repaired. So, an illustration of this… so, these are actually chicken cells, but if you deprive these cells of the key recombination protein Rad51, you see all of these chromatid breaks. And what these breaks represent is that… during the process of replication, either the Watson or the Crick strand didn’t get properly copied, and there’s an interruption. And that interruption is what is… what’s seen in these… in this image. And it’s the job of the Rad51 protein, a recombination protein, to patch up this break, which it’s going to do by copying the sequences from the intact template to patch up the break. And I’m gonna be talking in more general detail about how that process occurs. Okay. So, the source of most of the damage is the DNA replication machinery itself. This is just an image of the DNA replication fork. And really, what I’m telling you is that there are sources of damage — replication fork barriers and other instabilities — that cause the replication machinery to fail, maybe as many as a dozen times per normal cell cycle. Maybe more, as you’ll see. So, the simplest of these breaks that arise are just when one of the two strands is nicked. And as the replication fork comes through this sequence, it can’t copy across the nick, and this leads eventually to the formation of a double-strand break. That essentially means that one of the two sister chromatids has a break, and the other one is intact and could be used as a template to do its repair. But there are other sources of these breaks that I want to just mention. One is the consequence of UV exposure of DNA, which leads to the formation of cyclobutane dimers, here thymine dimers, where two adjacent thymine residues become covalently linked together. This leads to a really severe distortion of the double helix, and it prevents the normal DNA polymerase from going through this sequence. And that frequently, as you’ll see, leads to breaks. Another source of these breaks comes from what are called triplet repeat sequences, here CTGCTG repeated dozens of times. In the case of Huntington’s disease, sometimes hundreds of times. And these very simple repeats have the capability of forming quasistable secondary structures, which, again, block the formation of the replication fork and lead to breaks. And another example, which has become appreciated much more recently, are the formation of what are known as RNA:DNA hybrids, or R-loops, where transcripts of DNA, which ought to be liberated during the process of transcription so they can go off and be messenger RNA or other kinds of RNA, remain stably base paired to the template from which they were derived. And that… these turn out also to be severe blocks for replication fork progression. The cell has ways of getting rid of these R-loops, either by unwinding them using an RNA:DNA helicase, or by degrading them using some nucleases. But quite frequently, these structures remain, and they are, again, sources of damage. So, all of these sources of damage are at least possibly going to lead to chromosome breakage. In humans, there have been dozens and dozens of fragile sites identified, where such breakage is likely to occur. One way of finding these fragile sites is by slowing down and disabling the normal DNA replication process, in this case by using a drug called aphidicolin. And so, if you treat the cells with a quite low dose of aphidicolin, so that replication is proceeding, but not as efficiently as normal, you see these fragile sites appear. Places where… again, you see sis… one of the two sister chromatids is broken because the replication fork has been unable to get through those sequences without some kind of consequence. One consequence of this replication fork stalling is a phenomenon called replication fork regression, and this will turn out to have some interesting consequences. So, here, the replication fork is moving from left to right. It encounters, in this case, a thymine dimer, and the replication fork can’t progress any further than that. But interestingly, this replication fork is capable of rearrangement so that the newly synthesized strand — this red strand, here, and this red strand, here — can unpair from their original template and can pair with each other to make this new structure. And this is a very odd structure because it’s got… it’s a 4-way junction. That’s not the normal thing you see in DNA. But what that leads to is the formation of these intermediate structures, which can be seen in the electron microscope, and which are often called chicken feet for obvious reasons. And the consequence of this is to move the replication fork backward from the place where the stalling occurred. One of the ways that… if this occurs, that means that maybe repair proteins can gain access to this site and can actually repair it. That they couldn’t do if all the replication machinery was jammed up against the site where stalling is taking place. So, this weird structure, this chicken foot structure, has been called a Holliday Junction after the scientist Robin Holliday who first postulated it in 1964. As the time, nobody had ever seen these structures, but Holliday imagined what they would look like and had very interesting predictions about how these Holliday junctions were going to be important in recombination. So, these structures are formed by base pairing, and they can be completely base paired. They don’t… the picture here has a little opening in the middle, but in fact every base pair in a Holliday Junction can be paired, as is illustrated in this more accurate picture. One of the features of Holliday junctions, which I just mentioned in a way, was the fact that they can migrate. And they can migrate because every time… you can form the same structure here as here. The only difference is whether these bases are paired or those bases are paired in this structure. And so, if you… it turns out that, energetically, it doesn’t require a lot of effort for this Holliday Junction to be able to migrate back and forth. Here’s another picture of that, illustrating the mobility of these Holliday junctions. So that they… when they form, they can branch migrate forward and back, and they can go away from that source of blockage and back towards that source as I mentioned. Okay, so here again is one way in which this replication fork… Holliday junction migration can be used. Again, there’s this chicken foot being formed. And in this case, the 5′ strand is longer than the 3′ strand. The 3′ strand can now copy what was the other newly replicated strand, as shown in the green image there. And then, if this branch migrates all the way back, it can bypass the thymine dimers by virtue of the fact that it has copied those sequences from a different template. Here, the DNA damage isn’t repaired, but the… but the replication fork can continue. The other feature of Holliday junctions which is of some interest is the fact that they can be resolved by endonucleases that will cut the Holliday junction apart. And it can cut the Holliday junction apart in two different ways. One way would leave the original strands in their parental configuration and be called a noncrossover. And the other way… which looks different, but remember that in three dimensions these two alternative structures are in fact very similar…but if it cleaves the other strands, then you end up with essentially a crossover, which is to say that part of one parental strand is linked to part of the other parental strand. And, of course, crossovers, which arise frequently in meiosis, turn out to come from this kind of resolution of intermediate structures. So, for the rest of the talk, I’m gonna talk a little bit more in detail about different mechanisms of homologous recombination that can be used to patch up the double-strand break. All of these mechanisms have one common principle, which is that the broken ends of the DNA are going to be able to be repaired by base pairing with a template sequence, to recognize a sequence that is identical or nearly identical, with which it can then effect repair. The mechanism I’m gonna start talking about is break-induced replication, which is, in a sense, the simplest of these repair mechanisms to look at. So, I mentioned that there were these chicken feet intermediates and that a chicken foot intermediate, in addition to being branch migrated, could also be cut. It’s a Holliday junction after all. It can be cut by Holliday junction cleaving enzymes. And what that does is to leave one of the of the chromatids intact and the other one is essentially broken at the site where the replication was blocked. So, that broken replication fork can then be used to restart DNA replication by using a process of homologous recombination. And the steps in this that I wanna go through are illustrated here. So, first we have a broken replication fork, as illustrated on the top left. The next thing that happens to this end is that enzymes — exonucleases — chew away one of the two strands of the DNA, leaving long 3′-ended single strands of DNA. And those 3′-ended strands of single-stranded DNA are then bound by a recombination protein called Rad51 in eukaryotes and RecA in bacteria. And this recombination protein forms a filament on the single-stranded DNA, and then does this incredible step of locating, elsewhere in the genome, homologous sequences with which it can make those alternative Watson-Crick base pairs to recognize the template, as illustrated on the lower left. And this base paired intermediate, then, serves as a place where replication can be restarted. Okay. So, here’s another illustration of this break-induced replication. The break is made. The end is resected. The Rad51 protein helps to invade into the donor template, making Watson-Crick base pairs. And then there’s new DNA synthesis to restart the replication fork. And in this case, what’s illustrated is that the replication fork isn’t quite normal. Normally, we would assume that a replication fork would have leading and lagging strand synthesis happening at the same time. But it turns out in break-induced replication — at least in the one case that we can study in great detail, which is in budding yeast — the two strands of DNA synthesis are actually discoordinated. And this accounts for the fact that there is much more error associated with this replication process than you would see in normal DNA replication. In fact, in a third video that that will be part of this series, I’ll talk a lot about the errors that are produced by this kind of repair mechanism. Okay. So, Rad51 or RecA binds as a filament onto the DNA, and then effects this search for homology. If we look at these proteins in more detail, we discover that RecA and Rad51… by each subunit of these molecules, binds three bases of the single-stranded DNA, but makes a long and continuous filament which are illustrated here. One of the consequences of that is to stretch the DNA by almost 50%, so that the DNA is much more extended than it would be under normal B-DNA form. And this stretching open of the DNA I think is very important in the way in which this search for homology takes place. Okay. So, if we want to just define what’s going on inside the filament, if you imagine just cutting through the middle of the… of the filament, here’s a single strand of DNA which is being bound by the RecA or Rad51 protein. And here it binds to double-stranded DNA. And if it binds in the right way, then essentially all that’s happening during the strand exchange process is to exchange one base pair. And that’s happening at every step along the DNA. But we go from having a single strand of DNA and a double strand template to having a strand exchange intermediate and a displaced single strand. And you can see that, also, in this biochemical example that is shown here. So, here, in this experiment, RecA has covered a single-strand DNA template, which is homologous to a double-stranded linear DNA, which is shown here. And then Rad… RecA drives the strand exchange process, forcing the pairing of one of these two strands with the single strand, and displacing the other one. So, at the end of this process, there’s been an exchange of one strand. The complementary strand binds to the single strand of DNA, and the other single strand is now displaced and is liberated as the opposite product. A great deal of insight as to exactly how this is happening came from the brilliant crystallographic work of Nikola Pavletich’s lab, who figured out a way to crystallize and analyze this RecA protein bound of DNA by hooking up a whole bunch of RecAs together so that it made a uniform object for crystallographic study. And when they did that, then they could trace the contour of the single-strand DNA inside this RecA filament. And what they saw was really quite remarkable. What they saw was that the single-strand DNA was stretched. We already knew that from electron microscopy. But what they saw was that the stretching was not uniform, that rather than all the bases just being pulled apart by one and a half times each three bases that were bound by one subunit of this recombination protein are still in roughly a B-form of DNA, and then all the stretch happens in between those three bases and the next three bases. And this led to the understanding that the searching for homology and the mechanism by which the strands are actually being exchanged is actually done in… somehow, in groups of three inside each one of these subunits of the… of the recombination protein filament. Okay, so just to summarize what this means… it’s that you start with a single strand of DNA and a double-stranded template. And when they have lined up properly, one of those strands — the complementary strand — now can start to form Watson-Crick base pairs with the original single strand, and there’s the displacement of the other strand in this process. And of course, what that means in real terms is that the… there’s a formation of what we will call a displacement or D-loop. Here’s the Watson-Crick base pairing that the Rad51 filament has made. And this is the displaced strand, which is part of this duplex template. And that provides the initiation for new DNA synthesis and for this repair process to take place. So, here’s the recruitment of the DNA polymerase, and then the initiation of this process. One of the things that we learned in studying this in budding yeast, which is the place… the only organism where these kinds of detailed molecular studies can be done so far, is that this process requires DNA replication components that are not essential for normal DNA replication. So, one of these opponents is called Pol32. It’s a non-essential subunit of the DNA polymerase complex, not needed for normal DNA replication but essential for this replication restart mechanism, break-induced replication. And we think that this Pol32 protein is allowing DNA polymerase delta to work as a more processive enzyme than it would under normal circumstances. If you know the current views about DNA replication, the leading strand of normal DNA replication is done by DNA polymerase epsilon, and Pol-delta is doing the Okazaki fragments, which are very short. So here, Pol-delta has to work in a much more extended way, and requires this Pol32 protein. The second thing I can tell you about this Pol32 protein and this mechanism is that it isn’t just a yeast-specific mechanism. It also happens in humans. And this mechanism is called alternative lengthening of telomeres. Many cancer cells become immortal by reactivating an enzyme called telomerase, which adds TTAGGG, over and over, to the ends of chromosomes, their telomeres. But some tumors don’t reactivate telomerase. And in order to keep their telomeres at a necessary length, they use recombination mechanisms of the sort that I’m showing here. They recombine from one telomere to another to make alternative lengthening of telomeres. In yeast, we showed that this process required the Pol32 protein. Much more recently, it’s been shown that this break-induced telomere synthesis also requires the homolog of Pol32, called POLD3, and is simply an illustration of the fact that these mechanisms have been conserved all the way from Saccharomyces to humans, you know, an enormous evolutionary distance. Okay. So, I… I… that’s what I wanted to say about break-induced replication. Now, I’ll say something about another process, which is called gene conversion. And the difference here is that both ends of the double-strand break can participate in the repair event. And the result of this is that instead of needing to synthesize a huge long distance, as happens in break-induced replication, just a little patch of new DNA synthesis is required to patch up the broken chromosome. And a mechanism by which this happens is illustrated here. And it involves the formation of an intermediate we haven’t seen so far, which is not one Holliday junction but two. And so, you form… by first the break, then the resection of the broken ends, then the loading of rad51, and strand invasion… all those steps are the same. But now, after a little bit of new DNA synthesis, which is illustrated in light blue, you end up with a structure which has two Holliday junctions. And this double Holliday junction can again be acted upon by resolving enzymes, by nucleases, to end up as a non-crossover or as a crossover, and can carry out this repair process, and just uses a little bit of new DNA synthesis, the parts illustrated in light blue. So, these double Holliday junctions can be, again, dealt with by nucleases that can cleave these structures. And depending on the orientation of how these structures are cleaved, you can end up with either crossovers or non-crossovers. I want to just take a moment to say something about the consequences of crossovers in mitotic cells. If these are homologous chromosomes which are undergoing repair, one of them being used as a template to repair the other, you can end up with crossing over between these two homologous chromosomes. If crossover occurs between two sister chromatids, there’s no genetic consequence, because they are in fact identical pieces of DNA or… it’s the identical sequence that is just being exchanged. But if there are crossovers between homologous chromosomes, there can be very severe consequences, namely something called loss of heterozygosity, which is illustrated here. So, here I’m illustrating what happens in the… when one of these two chromosomes carries a recessive mutation called rb. Cells that are heterozygous for this mutation don’t have a phenotype. But in cells where there has been a crossover between the two homologous chromosomes, this makes… this causes the possibility, after chromosome segregation, of ending up with a chromosome which is homozygous for this rb mutation. And this loss of heterozygosity is associated with the progression of this particular disease, which is called retinoblastoma. But this same principle applies to a large number of other human diseases. It turns out that a number of diseases — retinoblastoma; the deficiencies in breast cancer… BRCA1 or BRCA2, the two familially inherited breast cancer mutations; something called Lynch syndrome — are all intrinsically recessive mutations. They don’t… they don’t have a real phenotype when there’s a wild type copy around. But the fact is that these kinds of recombination events occur frequently enough that some cells in a tissue can become homozygous, have a loss of heterozygosity, so that you end up… the people who carry these mutations, even though they’re just heterozygous to start with, end up with tissue which becomes homozygous, has a loss of heterozygosity, and that is… it is in those tissues that these diseases become manifest. We can also actually look at recombination visibly, by using studies of sister chromatid exchange. Here, some fraction of the thymidines of the DNA are replaced by an analog called bromodeoxyuridine. And so, if you start with cells that have grown in the presence of bromodeoxyuridine for a long time, and then you take away the bromodeoxyuridine, after one round of DNA replication this is just the picture that Meselson and Stahl showed for normal replication in E. coli, namely that one of the strands is old and has bromodeoxyuridine, and the other strand is new and doesn’t have any bromodeoxyuridine. And then, if these cells go through yet another round of replication, only one of the four strands, and therefore only one of the two sister chromatids, has any bromodeoxyuridine label. But if there’s been an exchange event, a crossover, during the process of this… of this second round, now there will be some bromodeoxyuridine on one chromatid, but at the other end they’ll be bromodeoxyuridine on the other chromatid. And you can actually see this by staining these chromosomes for the presence of bromodeoxyuridine. If there’s no sister chromatid exchange, then you see a single continuous line of labeling on one of the two sister chromatids. But if there’s been a sister chromatid exchange, now some of that label is exchanged to the other sister, as you can see here. And this turns out to be a very potent way of understanding how often these events happen in cells. And it turns out they’re frequent, surprisingly frequent. You can see in virtually every replication cycle in human cells that there are a few of these sister chromatid exchange events. If you treat these cells with a DNA-damaging agent, so that they suffer lots of chromosome breaks that require recombination, you actually can produce these astonishing pictures of what are called harlequin chromosomes. And they’re called harlequin chromosomes because a figure from the Renaissance, a Commedia dell’Arte figure known as Harlequin, wore a costume of these, as shown here, that resembles this picture. So, this tells you that there can be many, many sister chromatid exchange events. Normally, cells can handle this, and therefore you only see a few of these exchange events under normal circumstances. Okay. And one of the reasons you only see a few of these events is that these intermediates have an alternative way of being resolved that I haven’t talked about until now. And that is that some of these double Holliday junctions can be dissolved rather than resolved. That is to say, they are not being cut by nucleases in the crossover and non-crossover outcomes. They’re actually being unwound and taken apart in such a way that they result in no crossing over. And it turns out that a key element in that unwinding process is a helicase, a DNA unwinding protein, called the Bloom helicase, which was identified by the fact that individuals lacking this helicase are cancer-prone and have many other problems, the so-called Bloom syndrome… which unwinds the structure so that there are no crossovers. And so in the absence of the Bloom helicase, there’s a huge increase in sister chromatid exchange. Because nothing is being unwound, everything is being driven through the crossover pathway. And so if you do the same bromodeoxyuridine label that I showed before, now, when you look at the Bloom chromosomes, they’re harlequin chromosomes. They have dozens and dozens of these crossovers through their genome. And this tells you, actually, that there are lots of breaks in DNA during normal DNA replication, but almost all of those breaks are handled in a way that has no genetic consequence whatsoever. Okay. I’ll just add here that one of the things we don’t understand — one of many things we don’t understand in detail — is why these double Holliday junctions are mostly always resolved as crossovers rather than non-crossovers. This is something that people really are working hard to really understand. Okay. And then, just to complicate your life, this is not the only double-strand break repair mechanism where both ends can participate in the repair event. There’s yet another process called synthesis-dependent strand annealing. Again, the two ends are attacked by nucleases, become single-stranded DNA. Rad51 protein gets involved and drives the formation of these displacement loops that… by Watson-Crick base pairing. There’s the initiation of new DNA synthesis, as illustrated in the light blue. But here the new DNA synthesis that’s being generated is different from what happens in the other mechanism, because it’s being unwound from the template in the same way that RNA would normally be unwound from the… from its DNA template. And this unwound strand of DNA, the newly synthesized DNA, eventually is copied far enough so that can the anneal with its partner. And then it all gets patched up. The result of this is there never was a stable Holliday junction intermediate, and all of these events are resolved as non-crossovers. This mechanism turns out to be very important in mitotic cells. This mechanism, the second mechanism, turns out to be much more important in meiotic cells. In meiosis, crossovers are of course desirable to generate genetic diversity. But it turns out also that crossovers are necessary to hold these pairs of homologous chromosomes together for proper chromosome segregation. If there’s no crossing over, there is what is known as first division nondisjunction. If you look at Down syndrome individuals, who have an extra copy of chromosome 21, they arise on chromosomes that have not had proper levels of chromosome exchange. And so crossing over not only fulfills a diversification role, but it also turns out to be critical in terms of proper chromosome segregation. So, in meiosis, this unwinding pathway that I’ve talked about is disabled. There are mitotic-specific proteins that basically prevent the unwinding process from happening. That drives them into crossovers. And so, almost all the crossovers are generated by this double Holliday junction mechanism. And the non-crossovers turn out to be generated, for the most part, by a synthesis-dependent strand annealing mechanism. Okay. Just to finish up, I’ll say that there’s one more interesting homologous recombination process, and that’s called single-strand annealing. It’s really the simplest of all of these events, because it just involves the break and the resection of the break by nucleases until flanking homologous sequences are exposed, Watson on one strand, Crick on the other. And these can anneal to form a structure that is then trimmed. And the result of this is a deletion between two flanking repeated sequences. Sometimes these sequences can be dozens of kilobases or more apart, so you can make quite large deletions between these kind of flanking repeated sequences. These events don’t need Rad51, but they do need an annealing protein called Rad52. And the reason that these are important in people is that our genomes are littered with repeated sequences. There are 500,000 copies of a sequence called a Alu, 300 base pair chunks of DNA littered around the chromosome. And if you make a break in between them, they make deletions by single-strand annealing. And it turns out that many human diseases have the pattern of being… of these recurrent deletions, which are occurring between these flanking repeated Alu sequences, and turn out to be clinically very important. Okay. So, I’ve told you about several mechanisms of homologous recombination, which all play a role in maintaining the stability of the genome. When these are disabled, you end up with rearrangements which are driven by non-homologous recombination, which I haven’t talked in detail about at all. But these homologous recombination mechanisms are really the gatekeepers to the maintenance of genome stability. In the next video, what I will talk about is looking at these processes in more detail, at the molecular level. How do we know, in molecular detail, what I just told you in general terms? And I urge you to tune in and see what I have to say. Thanks a lot.