Hello, my name is Angelika Amon, I’m a professor in the Biology Department at MIT, and I’m also an Investigator of the Howard Hughes Medical Institute. In the second part of my talk, I will tell you about our work in trying to understand what the consequences of aneuploidy are on organismal and cell physiology. In the first part of that talk, I defined the terms for you, I gave you a historical overview of the study of aneuploidy, and I ended with what the big questions are in the field, especially how aneuploidy relates to cancer. And so this part now, having set up this question how can it be that aneuploidy at the organismal level is highly detrimental, but at the cellular level it is also associated with a disease that’s characterized by unrestricted growth, cancer, how we can merge these two questions. I will start with the research question that we started about 10 years ago, asking, well, what are the consequences of aneuploidy on cell physiology. So this is basically the second part of this three-part series. How can we understand how aneuploidy impacts cellular fitness? And again, like the first part of my talk, I subdivided this research presentation into three parts. I will first explain to you what kind of model systems and cellular systems we’ve established and developed over the years to study the effect of aneuploidy on cell physiology. I will then tell you that basically aneuploidy causes two sets of broad phenotypes. One which I will call a gene-specific phenotype, where changing the dosage of one particular gene will cause a particular phenotype, and I will give you examples for this. And then the main part of my talk will deal with this more surprising discovery that we made in the lab, that sort of suggests that in addition to these gene-specific effects, there are general effects of aneuploidy, where changing the dosage of many genes simultaneously will actually impose a very significant burden on various cellular housekeeping functions, and therefore will lead to what we call an aneuploidy stress response. So let me start out by telling you about the model systems that we have developed over the years to study the effects of aneuploidy. So we have generally used two broad systems, one we define as the low-complexity aneuploidies, and what we mean by that is these are usually cells, yeast cells or mammalian cells, that carry one or two additional chromosomes. In our study of aneuploidy, we initially started with budding yeast, and we created yeast strains that carry one particular chromosome in two copies instead of one copy, and we created a series of 20 different strains that carry 20 different additional chromosomes. So this slide here shows you how we know that this particular yeast strain has an additional copy of a particular chromosome. So what you see here, each box presents a particular karyotype of a particular yeast strain. And what we’re showing here is we’re showing the DNA content of each of the chromosomes, and we’re ordering the chromosomes according to their numbers. The left arm of chromosome 1 is at the left end of the box, the right arm of the chromosome 16 (budding yeast has 16 chromosomes) is the right end of the box. And you see there’s the same amount of DNA for all of them, except for this particular stretch here shown in red. This, turns out, is present in two copies instead of one. And as you can see here, this corresponds to the entire chromosome 8. And so we now know from this DNA content analysis that this particular yeast strain has two copies of chromosome 8. And as you can see here, we generated many different strains with all sorts of different aneuploidies, and we began to study them. After we began to study the aneuploidies in budding yeast, and we actually identified a number of phenotypes that were shared among all these different budding yeast strains, we then were also curious to ask, these shared phenotypes that we see in yeast, do they exist also in other systems? And we were especially interested in whether they exist in mammalian cells. So we used some old genetic tricks to generate mouse embryos that instead of being euploid, they had one particular chromosome extra. So in the mouse, unlike in humans, all trisomies are lethal, but we can isolate embryos from pregnant mothers that have a particular trisomy. So initially we made four of them: trisomy 1, trisomy 13, trisomy 19, and trisomy 16. So we isolate these embryos, we then dissociate these embryos, make cell lines from them, we’re making mouse embryonic fibroblasts from them, and then we can study them. So these are the two systems that we use to study what we call “constitutional, low-grade aneuploidies.” We also have a way of generating more complex and heterogeneous aneuploidies. And here we take advantage of mutations that increase the frequency with which chromosomes are missegregated. For example, in budding yeast, we can use mutations that induce chromosome nondisjunction, or chromosome missegregation, both during mitosis and meiosis. We can also make triploid yeast strains by various tricks, sporulate them, and the progeny of these triplet meioses are highly aneuploid yeast strains. And as I said in both the mouse and humans, we just use mutation. Down below here you see an example of what we, for example, do in mammalian cells. So this is a mammalian cell that’s about to undergo chromosome segregation, and we actually treat it with a drug that inhibits a protein that’s critically important for faithful chromosome segregation. So we give the cells the drug, and as a result, this protein is inactive. And now, the cells will missegregate its chromosomes and will result in aneuploid cells. So, the advantage of these sort of random, high-complexity aneuploidies that we induce with various mutations or chemicals, is that these chromosomal aneuploidies are very similar to the kinds of aneuploidies that we see in cancer. And from that perspective, they’re very interesting. However, these types of aneuploidies are very unstable, they’re a moving target, so they’re very difficult to study. So what we usually do is we usually conduct our initial studies and the discovery process in the low-complexity aneuploidies that I told you about before, these disomic yeast strains and these trisomic mouse embryonic fibroblasts. And when we discover a phenotype that we’re particularly interested in, where we want to know whether there are general phenotypes, we then ask, do these phenotypes that we see in these low-grade aneuploidies also occur in these various aneuploidies when we induce massive chromosome missegregation? And to our great delight and satisfaction, so far this has always been the case. If we discover a general phenotype among the various low-grade, constitutional aneuploids (the disomic yeast strains, the trisomic MEFs), we also usually see them in these high-grade, random aneuploidies. So, having introduced to you the model system that we use to study the effects of aneuploidy on cells, let me now move on and sort of very briefly highlight for you the sort of gene-specific effects. The fact that aneuploidy duplication of particular genes can lead to a particular phenotype has been known for a very long time. And I’m showing you here one example from the work of Judith Berman’s lab, where she was actually studying a pathogenic fungus called Candida albicans. Candida albicans is a fungus that can become pathogenic to humans, especially immunocompromised humans. And so if you are immunocompromised and you become afflicted with this yeast infection, what doctors usually do is they start giving you an antifungal drug called fluconazole. Unfortunately, when you are treated for long periods of time with that drug, the fungus eventually develops resistance, so it will no longer be responsive to the drug. And Judith Berman’s lab showed here is that when you duplicate a part of chromosome 5, that can lead to fluconazole resistance. And she actually then figured out which gene in that particular region was responsible, so a specific gene on this part of chromosome 5 can provide a new trait, it can provide resistance to a particular drug. Down here you see an example from our lab, where we asked, being disomic for a particular chromosome, can that confer a new trait? What we’re looking at here is we’re looking at benomyl resistance. Benomyl is actually a microtubule depolymerizing drug, it sort of depolymerizes microtubules, it kills all cells basically. And what Eduardo Torres, a postdoctoral fellow in the lab, observed: that this particular strain here, you can see this is a yeast strain that has an extra copy of chromosome 16, it is actually able to grow much better in the presence of the drug, than the wild-type euploid cells. So it acquired the ability to grow better in the presence of high concentrations of benomyl. So it is quite clear that gene-specific effects exist, that duplicating particular genes on a particular chromosome can confer either a drug resistance or some other new trait. So, there’s many more examples for this, I actually mentioned one in the first part of my talk: APP duplication in early onset Alzheimer’s and Down syndrome; and this list goes on and on and on. What to us was much more interesting is that our basic studies of these aneuploid yeast strains revealed that, in addition to these gene-specific effects, there were very general effects. And those are sort of listed here. What we find is that aneuploidy causes proteotoxic stress. Now I’m going to tell you in a minute what this is all about, and what proteotoxic stress means. We found that aneuploidy causes a transcriptional response. And then aneuploidy also causes a delay in G1. And so we basically, since these are all signs of a stress response — you will see very quickly where these stresses are coming from, I’m going to explain this to you — we sort of generally refer to this as the “aneuploidy stress response.” Okay? So what I’m going go do next is I’m actually going to show you the data that led us to the conclusion that aneuploidy causes proteotoxic stress, and I’m going to show you the data that led us to the conclusion that there’s a stress response. I’m not going to show you the data for that, I’m just going to mention to you that in addition to all these other phenotypes, cell proliferation is impaired in these aneuploid cells, it causes a G1 delay. So, let me tell you how we initially developed the hypothesis that budding yeast strains are under proteotoxic stress. So what is proteotoxic stress? Proteotoxic stress is a condition where, when the proteins in the cell are misfolded or they’re not working right, that’s obviously bad news for the cells. And the cells try to fix it. And usually what it does is it sort of activates chaperones and ubiquitin-dependent protein degradation to sort of eliminate these unfolded and malfunctioning proteins, and sort of making new, better ones, okay? And so we initially realized that there perhaps was proteotoxic stress in these aneuploid cells because compromising proteasome function, either by chemical or genetic means, was more detrimental in the aneuploid cells than in the euploid cells, so lowering the proteasome function really affected these aneuploid cells in a substantial manner. And the other thing that we realized early on was that the aneuploid yeast strains, all of them, were sensitive to a protein synthesis inhibitor called cycloheximide, and they were also sensitive to temperature, high temperature. High temperature is known to unfold proteins, causes a lot of stress for the cells, and so raising the temperature in these particular disomic strains had actually quite detrimental effects, okay? And so these types of results led us to hypothesize that the protein quality control systems of the cells are either not functioning right, or they are overburdened, okay, and therefore can’t catch up with proteotoxic stress. So to address this question in more detail, we actually wanted to look at markers for proteotoxicity, and I’m showing you one marker here. We’re looking here at a chaperone, a disaggregase known as Hsp104, and it’s fused to GFP, and normally, this protein is sort of diffusely present throughout the yeast cells. But when aggregates are forming in these yeast cells, the disaggregase, the chaperone, will go to these places and you will start seeing these foci accumulating in yeast strains. And here’s an extreme case where the cell is actually heat- shocked at 37°, which causes a lot the proteins to unfold and aggregate, and see, now these cells have a large number of these Hsp104 foci. And what we found that was very exciting is that every single disomic yeast strain (so here’s a wild type, this is the percentage of normal wild-type cells that harbor Hsp104 aggregates… and here are all the disomic yeast strains), and every single one of them has a higher percentage of cells that harbor these Hsp104 aggregates, indicating that endogenous proteins in these disomic yeast strains tend to aggregate. We also looked at another marker for proteotoxic stress, this is a reference protein called VHL. VHL is actually a human protein that, when you express it in yeast, it doesn’t fold right, and it’s actually very quickly degraded by the ubiquitin-proteasome system. But if the ubiquitin- proteasome system and some other chaperones are either not working right or overburdened, what will happen is that VHL will start accumulating in the cells, and again it will start forming aggregates. And so again you can see, whereas wild type only a small percentage of cells has VHL aggregates, you can see that all the disomes have much higher levels of aggregates, indicating that indeed the protein quality control systems in the cells are compromised. This is not just a unique feature of these constitutional, low-grade aneuploidies, these disomic yeast strains that we constructed. We also see protein aggregates in these random, high-grade aneuploids that we create when we make progeny of a triploid strain, that will create highly aneuploid progeny. So we’re looking at Hsp104 here. This is sort of the distribution that you get in euploid cells. Upon sporulation, this is the distribution that you get in these highly aneuploid cells, and the same is true for VHL. And then finally we’re very interested in addressing the question of how quickly do these protein aggregates start accumulating upon chromosome missegregation, upon generating aneuploidy. And so to address this question, we took advantage of two mutants that missegregate chromosomes at a high frequency. For the purpose of my talk, it doesn’t really matter what ndc10 is and what ipl1 is. The only thing you need to know is that these are temperature-sensitive mutants, and when these strains are grown at the intermediate temperature, they will missegregate chromosomes at a very high frequency. So we take these cells, we arrest them in G1, and then we’ll send them through the cell cycle, and then ask two hours later, after they’ve missegregated their chromosomes, have they started to accumulate these aggregates? And the answer here is very clearly yes, it’s elevated in ndc10 mutants, it’s elevated in ipl1 mutants, compared to the euploid control. The gray bars here are an important control. What we did here is we actually prevented these cells from undergoing mitosis, so we prevented these cells from missegregating their chromosomes by basically depolymerizing their microtubule cytoskeleton. So even though they want to segregate, they can’t because they don’t have any microtubules. And you can see under those conditions, we do not see an increase in aggregate formation. So from these experiments we concluded that aneuploidy causes proteotoxic stress, it causes protein aggregation, and that occurs very quickly, very soon after chromosome missegregation. We then also asked: Hsp104, this disaggregase that’s very important for cells, does the amount of disaggregase that you make, is that actually correlated with the degree of aneuploidy that occurs in the cells? And so what I’m showing you here is I’m showing you the amount of expression of Hsp104 on the y-axis, with respect to what the percentage of aneuploidy in a particular strain. And as you can see, this actually tracks very nicely, again indicating that there is a very general response to the aneuploid state that’s proportional to the amount of disaggregating protein. So in yeast cells this suggests very clearly that there’s proteotoxic stress in these strains. Do we have evidence that this proteotoxic stress also exists in mammalian cells? The answer is yes here. I’m not going to show you the data, but instead I’m just going to summarize the data for you. Here, I’m showing you that aneuploid mouse cells are sensitive to drugs that cause proteotoxic stress, they’re sensitive to this drug called autophagy inhibitor chloroquine. We actually find that the basal levels of autophagy (that’s a protein quality control pathway that clears protein aggregates from cells) are increased in trisomic mouse embryonic fibroblasts. When we inhibit chaperones by drugs, in this case 17-AAG inhibits the chaperone Hsp90, again then trisomic mouse embryonic fibroblasts are more sensitive to this than euploid cells. And then very interestingly, the downregulation of a subset of protein-folding factors is actually delayed upon transient heat shock, indicating that the protein quality control pathways in these cells are either not working right or they’re again overloaded. There’s one piece of data that I want to show you that, like in budding yeast, this proteotoxic stress is seen very quickly upon the creation of aneuploid cells. This is the system that I showed you before, where we take a dividing mammalian cell, we treat it with a drug that induces chromosome missegregation, that leads to the formation of aneuploid cells. And here on the bottom, we’re looking at a marker for autophagy. I told you autophagy is this protein elimination system that gets rid of aggregated proteins and misfolded proteins. And as you can see, in the drug-treated cells shown in red, these autophagy markers are dramatically increased, compared to in green, the control cells. So a key question that arises from these findings is, where is this proteotoxic stress coming from? Obviously, a very simple hypothesis is there are additional chromosomes in these cells, these chromosomes make RNAs and proteins, and perhaps these proteins then change the proteome in the cell, and therefore lead to changes in protein homeostasis and increased aggregation. That’s clearly the simplest hypothesis, and the hypothesis that we favored. However, before we test this hypothesis, we actually needed to ask a much more basic question, which is: Are these additional chromosomes actually active? Do they actually make RNAs and proteins? And this experiment here shows you that this is indeed the case. I’m showing you here a strain again, I’m showing you the DNA content, the RNA content, and the protein content, of a yeast strain that carries an extra copy of chromosome 5. Remember, this is sort of the stretch of the DNA that’s present in two copies. And you can see very clearly, there’s a transcriptional response. You see there’s much more biological noise here that we actually understand where this is coming from, but what I want you to draw from this picture here is that, generally speaking, the amount of RNA is made according to gene copy number. And about 80% of the proteins are also produced according to gene copy number. So we now understand that these additional chromosomes are indeed active. So now can we acquire evidence to indicate that it is actually the fact that these additional chromosomes are active that caused the problems? And I’m not showing you the data here, I’m just summarizing the pieces of data here that are actually both the most telling (we’ve done many more experiments, but these I think are the most critical data). First of all, in yeast, we have the ability of introducing human DNA or mouse DNA into these cells, and because the splicing machinery is so fundamentally different between yeast and mouse (and humans), these pieces of DNA, they’re called “yeast artificial chromosomes,” will make few if any gene products. And so one can now ask… we can stick a chromosome’s worth of human or mouse DNA into these yeast cells, and ask, well, what’s the phenotype of that? And the fact is that all these phenotypes that we see in the aneuploid strains that have extra yeast DNA are not seen in the strains that have extra mouse and extra human DNA. The perhaps most telling experiment that we did that it’s actually the fact that these additional chromosomes are active is what’s critical here, is the observation that ploidy “buffers.” What do I mean by that? What we routinely observe is that a haploid strain that carries an extra copy of a chromosome has much more severe phenotypes than a diploid strain that carries an extra copy of a particular chromosome. What that result actually tells you is that what the cells care about here is relative ratios. That is, if you double, in the case of a haploid, the gene products by adding an extra copy, that’s significantly more worse than if you just add an extra third, in the case of a diploid, where you add an extra copy. Okay? So from these, and many more, experiments that I don’t have time to tell you about is, this is our current working hypothesis. We believe that aneuploidy leads to excess protein production. And what we propose is that this causes, of course among other deleterious outcomes, proteotoxic stress, because overproduction of certain proteins saturates the protein quality control pathways of the cell. And so the question is, how can you saturate the protein quality control pathway of the cells? There’s several possibilities. Of course, there’s some proteins that are now made in excess in these aneuploid cells because you doubled their gene copy number, that are obligatory chaperone clients, or that obligatorily require some protein quality control for their function: protein kinases, WD40-repeat proteins, tubulin, actin, and so forth, many proteins. So that could cause an additional burden on the protein quality control pathways of the cell. And secondly, what we really believe is at the heart of the aneuploidy problem is protein stoichiometry imbalances. And I want to illustrate this here with this particular slide. I want you to visualize a protein complex that’s made out of protein A and made out of protein B, and they need to function together. So generally speaking, it’s very difficult for a cell make exactly the same amount of protein A and protein B. And so the way the cell solves this problem is, they make approximately the same amount of each, and then they neutralize excess subunits, for example “A” here, by using protein quality control pathways and sometimes feedback mechanisms. And there’s wonderful examples for this in the literature: α- and β-tubulin, ribosomal subunits, histones, and so forth. Now imagine that you’re doubling the gene dosage of hundreds if not thousands of proteins at the same time, by introducing an extra copy of a chromosome. All of a sudden, you do not have 5% excess of protein A, you have a 110% excess of protein A. And what we propose is that that causes an increased burden on the protein quality control pathways of the cells, and it also leads to other phenotypes that I don’t have time to tell you about. It is there’s clearly energy stress, and there’s also cell proliferation defects. And we speculate that these phenotypes are also caused by protein stoichiometry imbalances. Where we’re going with this project right now, clearly we are very much interested in understanding how the protein quality control systems are impacted, so we’re now probing the activities of individual protein quality control systems in these various aneuploid strains. So I’m showing you one example here. We’re looking at one chaperone, it’s called Hsp90, and what I’m showing you here’s a biological assay for Hsp90 activity. Doesn’t really matter how the assay works, but if there’s a lot of bands here on this gel, that means that the chaperone is working well, okay? And I hope you can appreciate that there’s many disomic yeast strains in which the Hsp90 chaperone is actually not working very well, so indicating that indeed in some of these aneuploid cells, there’s a profound impact on at least one chaperone system. We’re now going about trying to understand what the other ones are. So this is what I wanted to tell you, protein quality control. The second general phenotype that I wanted to briefly discuss is the realization that there is a transcriptional response to the aneuploid state. And we initially discovered this by simply querying the RNA gene expression profile of all the disomic yeast strains that we initially created. So what I’m showing you here is I’m showing you here all the disomic yeast strains that are sort of listed here. Each column represents a yeast strain. Each row represents a gene. And you can see here red- and green-colored images here. Red means the gene is downregulated, green means it’s upregulated. And you can see here very clearly all these disomic yeast strains, there’s a cluster of genes that’s downregulated, these are genes associated with growth (ribosome biogenesis and RNA metabolism). And that there are some genes that are upregulated, and they all fall into the category of stress. This signature has been observed before by David Botstein and colleagues, and they call it the “environmental stress response” in yeast, the ESR. And so, a graduate student in the lab, Jason Sheltzer, was very much interested in asking whether this environmental stress response in also seen not just in these disomic yeast strains that we created, but also seen in all sorts of other aneuploid organisms. And the first thing he did is he asked… There’s a collection of yeast strains called the Yeast Knockout Collection, and for reasons that are not important here, there’s among these actually a large number of strains that are aneuploid, naturally occurring aneuploidies. And he asked, is this ESR also seen in this collection of aneuploid yeast strains made while the Yeast Knockout Collection was made? And more importantly, does the strength of this transcriptional response correlate with the amount or the degree of aneuploidy? And that’s shown here. We’re showing here the percent of aneuploid, meaning the percent of excess gene products present in the cells, we can calculate this in yeast. And here we’re at the intensity of the stress response, and you can very clearly see that there’s a striking correlation. What was very exciting to us, he not only saw this in budding yeast, he saw also say that in fission yeast, another related yeast. And he also saw that in plants, indicating that at least among fungi and plants, there’s a general transcriptional response. He was also curious to see whether the transcriptional signature for aneuploidy was also observed in mammalian cells. And what he did here is he used data collected by others that were basically gene expression arrays of various trisomies in both mouse and humans, and the question that he asked here is, if we look at the percent of genes that are upregulated for example here in trisomy 21, what’s the likelihood that these same genes are upregulated in other trisomies, too, in this case trisomy 13, 16, and 19, for example? And you can see very clearly that again there’s a correlation. If something is upregulated in one trisomy, the likelihood of this gene being upregulated in other trisomies is very high. And these are all the various data here for trisomies in the mouse, and down below here for trisomies in humans. And the question is what kind of genes are upregulated and downregulated in these mammalian cells? And to our great satisfaction, we saw that the same types of categories of genes are upregulated and downregulated as in fungi and in plants. So for example, you see that upregulated are genes that are involved in stress response and inflammatory response. And genes that are downregulated are genes involved in cell proliferation and cell growth, DNA replication, and cell division. So these results to us suggested that aneuploidy is also detrimental at the cellular level. I gave you examples for gene-specific effects that undoubtedly existed, and then I summarized for you the evidence that led us to conclude that, in addition to these gene-specific effects, aneuploidy causes a set of phenotypes that’s independent of the identity of the aneuploidy, and is indicative of cellular stress. And we generically call this of phenotypes that are shared among many different aneuploid cells, the “aneuploidy stress response.” And with that, I would like to end this part of my presentation, giving you some insights into the kind of questions that we’re addressing, to get at how does aneuploidy affect cells. And now in the third part of my talk, I will address the questions of how aneuploidy impacts human diseases. And I will also address this, to us, very exciting possibility that aneuploidy could actually be developed as a new therapeutic. So let me end by acknowledging the people who’ve done all this work. On the top row are the yeast researchers in the lab, Eduardo Torres, a postdoctoral fellow in the lab, now has his independent faculty position at UMass Worcester. He actually developed the aneuploidy system in yeast in the lab many years ago. The work on the transcriptional response that I showed you was work by a graduate student in the lab, Jason Sheltzer. The work on proteotoxic stress in aneuploids was work done by another graduate student in the lab, Ana Oromendia. I want to especially acknowledge Bret Williams. He was a postdoctoral fellow in the lab who actually was brave enough to start studying mouse cells in an entirely yeast lab, and so he really set up the mouse aneuploidy system in the lab. Stefano Santaguida, another postdoc in the lab, characterized the effects of aneuploidy on autophagy. And Yun-Chi Tang did some of the work of proteotoxicity in mammalian cells. And of course, I would like to acknowledge the funding sources. Without, none of this would have been possible. We’ve been generously supported over many years by NIGMS and also by the Howard Hughes Medical Institute. And I thank you very much for listening.