Controlling the Cell Cycle
Transcript of Part 3: Controlling the Cell Cycle: Anaphase Onset
00:00:03.25 Hi, I'm Dave Morgan from the University of California in San Francisco 00:00:07.07 and in this lecture I'm going to tell you a little bit about our work 00:00:09.26 on the regulatory system that drives the cell through the metaphase to anaphase transition. 00:00:16.12 The theme of this part of my lecture is going to be synchrony. 00:00:20.00 And by that I mean that in biology there are many occasions in which multiple entities 00:00:23.15 such as chromosomes or genes or cells have to be commanded or regulated 00:00:28.28 in such a way that they respond synchronously to a single input signal. 00:00:33.05 And so today I'm going to talk about how the chromosomes of the mitotic cell 00:00:36.24 are behaving synchronously and what the mechanisms underlying that synchrony are. 00:00:42.02 So we're going to focus in particular on the metaphase to anaphase transition. 00:00:45.23 This moment in the cell cycle when the cell goes from the metaphase state 00:00:49.07 where the sister chromatids are aligned on the mitotic spindle 00:00:51.24 and suddenly separates all of its sister chromatids more or less synchronously 00:00:55.21 and pulls them to opposite poles of the cell. 00:00:59.00 And the synchrony of anaphase is particularly well illustrated by this movie here 00:01:03.16 which shows a vertebrate cell in which the mitotic spindle is labeled in these reddish-orange colors. 00:01:08.27 The microtubules are reddish-orange. 00:01:10.10 And these green dots represent the kinetochores 00:01:12.29 where those microtubules are attached to the sister chromatids. 00:01:15.28 And the sister chromatids themselves are visible 00:01:17.22 as these dark shadows in the middle of the spindle. 00:01:20.06 And so when we start this movie what we'll see is this metaphase cell 00:01:23.06 progressing into anaphase through the 00:01:25.07 synchronous separation of the sister chromatids. 00:01:28.23 And so right about there the sister chromatids begin to separate 00:01:33.19 and are pulled to opposite poles of the cell. 00:01:35.09 And the key thing that I want you to focus on in this movie is how 00:01:38.01 synchronous sister chromatids separation is. 00:01:40.04 Once the sisters begin to separate 00:01:42.04 they all separate at more or less the same time. 00:01:48.15 So, what I'm going to tell you about today is a little bit about 00:01:50.26 the regulatory system that helps drive that synchronous sister chromatid separation. 00:01:55.11 And to do that I need to remind you about some of the basic features of cell cycle control. 00:02:00.13 As is well established, it is now known that the cell cycle control system involves 00:02:03.12 a series of Cdk-cyclin complexes that are activated at different cell cycle stages 00:02:08.01 and then that takes the cell up to metaphase, here 00:02:11.02 when the anaphase promoting complex or APC is responsible for targeting 00:02:14.24 specific proteins that bring about the metaphase to anaphase transition. 00:02:18.16 So the APC is the regulatory molecule that lies at the center of my talk today 00:02:23.13 and lies at the center of any discussion of the metaphase to anaphase transition. 00:02:27.15 Now, the APC drives anaphase through a fairly complex circuit 00:02:30.26 that's illustrated on this slide here. It has two major targets. 00:02:34.22 The first of which is a protein called securin 00:02:37.08 and securin is a tight binding inhibitor of a protease called separase. 00:02:43.09 So that when securin is destroyed in mitosis, separase is activated 00:02:47.17 allowing it to go to the sister chromatids and cleave a specific subunit of 00:02:51.00 a protein complex called cohesin that holds those sister chromatids together. 00:02:55.14 And so when separase is activated, that results in the loss of sister chromatid cohesion 00:02:59.10 which allows those sisters to be separated and then pulled apart by the mitotic spindle. 00:03:05.02 And so through this simple mechanism the APC directly initiates 00:03:08.11 the onset of sister chromatid separation and segregation. 00:03:12.15 Now, the other major target of the APC are the cyclins, 00:03:15.17 the regulatory subunits that activate the Cdks. 00:03:18.19 And so when cyclins are destroyed in mid-mitosis, that leads to the inactivation 00:03:22.17 of the associated Cdks which then allows dephosphorylation of Cdk substrates. 00:03:28.09 That dephosphorylation of Cdk substrates 00:03:30.10 is essential for normal progression through 00:03:32.26 the stages of late mitosis. So for example, 00:03:36.10 normal anaphase spindle behavior and the disassembly of the spindle and cytokinesis 00:03:40.24 all require dephosphorylation of Cdk substrates. 00:03:45.21 So the question for today is: What is it about this regulatory system 00:03:48.13 that drives the synchronous separation of 00:03:51.02 multiple sister chromatids all at about the same time? 00:03:54.01 And to appreciate one of the answers to that question that we've discovered 00:03:57.16 there's one more component in this system that I need to mention, and that is 00:04:01.13 a protein phosphatase that contributes to the 00:04:03.12 dephosphorylation of Cdk substrates in late mitosis. 00:04:06.19 Now obviously the inactivation of a kinase alone 00:04:08.26 is not sufficient to cause the dephosphorylation of a protein. 00:04:12.10 You need a phosphatase to take that phosphate off. 00:04:15.19 And in budding yeast, the system we work in, the major phosphatase 00:04:18.23 that takes off Cdk substrate phosphates is a protein called Cdc14. 00:04:23.21 And Cdc14, this phosphatase, is activated in early anaphase 00:04:27.25 at about the same time as cyclins are destroyed 00:04:29.28 and by this mechanism, by activating the phosphatase 00:04:33.01 at the same time as you inactivate the kinase, 00:04:35.03 that leads to a highly concerted dephosphorylation of Cdk substrates. 00:04:40.17 Now what's interesting about Cdc14 is that itâ€™s activated in early anaphase 00:04:44.06 by a mechanism that depends on separase. 00:04:47.09 In other words, through a mechanism that's really not very well understood, 00:04:50.09 the activation of separase 00:04:51.29 leads not only to sister separation, but also to the activation of Cdc14. 00:04:57.15 And so this results in a really interesting regulatory circuit in which the APC 00:05:02.06 uses multiple pathways essentially to lead to the dephosphorylation of Cdk substrates 00:05:06.24 at the same time as itâ€™s causing this separation of the sister chromatids. 00:05:12.08 But the question for today is: What is it about this regulatory system that drives 00:05:15.20 synchronous separation of all the sister chromatids at the same time. 00:05:19.09 Now, an obvious possibility is that separase activation 00:05:21.26 has to ... assume certain characteristics 00:05:25.08 that allow separase to be activated in large amounts very suddenly 00:05:29.08 so that it can spread very quickly through the cell to all the sister chromatids 00:05:32.21 and basically cleave the cohesin more or less simultaneously on all those sister chromatid pairs. 00:05:39.12 But there's no real indication in this system 00:05:40.29 of any such sort of switch like separase activation. 00:05:44.25 In other words, one might expect that there might be positive feedback 00:05:47.11 in this system somewhere that could generate a highly abrupt 00:05:50.13 and complete activation of separase which could then 00:05:53.04 spread rapidly through the cell to generate 00:05:55.06 sister separation in a synchronous fashion. 00:05:58.27 And so I'm going to tell you today about some of our work that led to the discovery 00:06:02.14 of a potential positive feedback loop in this system that helps generate 00:06:05.14 rapid and steep separase activation in the yeast cell. 00:06:09.26 And these studies began innocently enough when we were looking at 00:06:13.24 the phosphorylation state of this protein here, securin. 00:06:16.22 Now, securin had been identified in our earlier studies as a potential Cdk substrate, 00:06:21.25 something that might actually be phosphorylated by the Cdks. 00:06:24.19 And itâ€™s well known to be a good Cdk substrate in vitro 00:06:28.15 and so we decided to analyze its phosphorylation 00:06:31.03 state in the cell using mass spectrometry. 00:06:34.21 And those studies which were carried out in collaboration 00:06:36.17 with Andrew Krutchinsky at UCSF 00:06:38.07 led to the identification of these four phosphorylation sites on securin 00:06:42.01 that are all found at Cdk consensus sites: SP or TP residues 00:06:46.10 that are known to be Cdk consensus sites. 00:06:50.09 Interestingly, three of these sites had been identified previously by others 00:06:53.23 as involved in various ways in the regulation of separase by securin. 00:06:57.15 But this one site over here had not been studied 00:06:59.27 in any detail and we found this one to be interesting 00:07:02.18 because itâ€™s located right next to a sequence 00:07:04.12 called the D-box or destruction box of securin. 00:07:07.22 And that destruction box is known to be required 00:07:09.26 for the recognition of securin by the APC. 00:07:13.08 Likewise, there's another putative Cdk site 00:07:16.11 that we didn't identify down here next to another sequence 00:07:19.04 called the KEN box and that KEN box, 00:07:21.14 likewise, like the D-box is known to be involved in recognition 00:07:24.27 of securin by the APC. And so an obvious hypothesis was perhaps that phosphorylation 00:07:30.12 of this serine right next to the D-box 00:07:31.29 could interfere with recognition of this protein by the APC. 00:07:35.19 In other words, Cdk-dependent phosphorylation 00:07:37.20 of securin inhibits its destruction via the APC. 00:07:42.10 So we tested that hypothesis by the most direct means possible 00:07:46.13 which is simply to test the ability of phosphorylation to inhibit the ubiquitination of securin 00:07:51.08 in an in vitro assay using purified components. 00:07:54.15 And so this gel, this is a polyacrylamide gel autoradiograph showing 00:07:58.28 the ubiquitination of a radio-labeled securin protein here by purified APC in vitro. 00:08:04.13 And so what we do for these reactions is produce securin 00:08:06.22 by in vitro translation in a radiolabeled form 00:08:10.04 and then incubate it with purified APC 00:08:12.16 and all the other components you need to get ubiquitination of a protein. 00:08:15.22 And that results in this ladder of bands that forms above the securin 00:08:19.17 which represents the addition of multiple copies of this ubiquitin protein. 00:08:24.17 And so in the absence of kinase, clearly securin is a good APC target as we expect. 00:08:29.15 And the question then becomes: How good a substrate is it 00:08:32.06 for the APC after itâ€™s been phosphorylated? 00:08:34.17 And the answer to that is shown here. Once securin is phosphorylated, 00:08:37.24 its mobility on the gel shifts but it is no longer a good substrate for the APC. 00:08:43.01 Phosphorylation of securin has therefore greatly inhibited its ubiquitination by the APC, 00:08:48.02 just as we predicted from the position of the phosphorylation sites. 00:08:53.18 Then of course, we were also able to restore ubiquitination of the protein 00:08:57.04 by treating that phosphorylated protein with a phosphatase, Cdc14 00:09:01.22 and so by adding Cdc14 to the reaction we can now restore ubiquitination of securin 00:09:06.15 indicating that this phospho-regulation is reversible by the appropriate phosphatase. 00:09:13.08 So what does all this mean for the regulation of anaphase? 00:09:17.03 Well, it has a number of interesting implications. The first of which is 00:09:20.05 that we can now add another regulatory arrow here suggesting that cyclin-Cdk activity 00:09:25.20 is capable of inhibiting APC-dependent securin destruction. 00:09:29.23 And this might have implications for the order of events in anaphase. 00:09:32.19 It might suggest that some level of cyclin destruction 00:09:35.07 is required before securin destruction can occur. 00:09:39.04 But, more importantly, we can also add another arrow here 00:09:43.02 suggesting that Cdc14 can feedback and stimulate securin destruction. 00:09:47.27 And this is interesting because securin destruction 00:09:49.20 leads indirectly to Cdc14 inactivation. 00:09:52.27 In other words, what we have here is the potential for a positive feedback loop. 00:09:56.22 And the idea would be, therefore, at low levels of securin destruction that might 00:10:00.21 lead to the low levels of Cdc14 activity which could feedback 00:10:04.12 and promote additional securin destruction 00:10:06.19 and one might imagine that under certain conditions that would lead to the switch-like, 00:10:10.09 abrupt activation of separase in the cell. In other words, the sort of behavior 00:10:14.16 we were looking for to explain the synchronous behavior of anaphase chromosomes. 00:10:20.04 This next slide illustrates that same point in a more graphic fashion. 00:10:24.01 What we're seeing here is that if there is no feedback in this system 00:10:26.22 a linear increase in APC activity would be expected 00:10:29.16 to lead to a linear increase in separase activity. 00:10:32.02 But if the feedback, if positive feedback exists in the system, 00:10:35.04 then one might imagine that there would be an initial delay 00:10:38.14 in securin destruction until some threshold level of destruction is achieved 00:10:42.18 where Cdc14 positive feedback kicks into gear and essentially flips the switch 00:10:46.29 from low separase activity to high separase activity. 00:10:50.05 And this steep separase activity might be expected 00:10:52.25 to contribute to the synchrony of anaphase chromosome behavior. 00:10:57.19 So, how do we test that possibility? 00:10:59.15 Well, what we needed to do was disrupt this positive feedback loop specifically 00:11:04.25 in an intact cell and then look to see how anaphase occurs in that cell. 00:11:08.21 And to do that we developed the Securin-2A mutant 00:11:10.26 in which we've mutated the 2 Cdk-phosphorylation sites, 00:11:14.22 the one that we actually identified plus we also mutated 00:11:16.23 the other site that is a Cdk consensus site 00:11:19.22 in case that is also phosphorylated in vivo. 00:11:22.09 And so we found that, sure enough, when you mutate these two phosphorylation sites 00:11:26.08 that abolishes the ability of Cdk1 to inhibit ubiquitination of this protein. 00:11:31.18 So we then expressed this protein in a yeast cell. 00:11:34.00 We, in fact, we replaced the endogenous securin gene 00:11:36.11 with the securin-2A mutant gene, thereby resulting in a yeast cell that 00:11:41.03 should not contain this positive feedback loop 00:11:43.01 and then we analyzed anaphase in those cells. 00:11:46.09 The first result is that those cells appear quite normal at first glance. 00:11:49.26 They grow normally on yeast plates and so on. 00:11:52.03 But as you'll see, upon detailed analysis of their mitotic events, 00:11:56.01 we found that they do indeed have some serious defects as I'll describe in the next few slides. 00:12:02.13 Of course, the obvious hypothesis is that expression of the securin-2A mutant might 00:12:06.17 somehow reduce the synchrony of anaphase in these cells. 00:12:09.19 So then the question becomes: How do we test the synchrony of anaphase? 00:12:13.21 And to explain what we did this slide illustrates the underlying thought process 00:12:18.23 that's going into the experiments I'm going to show you. 00:12:21.03 Now one might imagine that when separase activity rises in early anaphase 00:12:25.14 that will result in slightly asynchronous separation 00:12:28.21 of different sister chromatid pairs because 00:12:30.07 one might imagine that different sister chromatid pairs would have different sensitivities 00:12:34.19 to separase. One chromosome might separate at a lower level of separase than another. 00:12:40.10 And so as a result of this one can imagine that there should be some small amount of time 00:12:44.07 between the separation of two different sister chromatid pairs 00:12:47.01 which reflects the synchrony of anaphase. 00:12:51.05 However, our prediction is that in the securin-2A mutant the activation of separase 00:12:56.15 is going to be less steep than it is in the wild-type cell. 00:12:59.27 And as a result there should now be more time between 00:13:02.16 the separation of two different chromosomes. 00:13:04.15 In other words, a loss of synchrony will be reflected in the fact that there is more time 00:13:09.01 between the separation of two different chromosomes. 00:13:11.29 And so what we needed to do of course was measure the relative timing 00:13:15.06 of separation of two different chromosomes in a yeast cell. 00:13:18.18 And this is not a simple task for the very simple reason that yeast cells are extremely small 00:13:24.03 compared to the giant animal cells whose mitoses I've been showing you in my previous slides. 00:13:29.12 So itâ€™s not possible to do this sort of high resolution microscopy on a yeast cell 00:13:33.02 for the simple reason that itâ€™s so tiny. 00:13:36.10 So what we needed to do was specifically label certain sister chromatid pairs 00:13:40.11 in the yeast cell using fluorescent tags and observe the timing of their separation that way. 00:13:45.14 And to do that we developed this yeast strain that began with 00:13:49.03 the construction of a red fluorescence tag on the spindle poles. 00:13:52.15 So using red fluorescent protein attached to an important spindle pole protein, Spc42 00:14:00.08 that resulted in the appearance of a red fluorescent tag on the spindle poles. 00:14:04.19 Then, more importantly, we labeled a couple of chromosomes with green fluorescent dots. 00:14:09.04 So first we started with a green fluorescent dot on chromosome IV here 00:14:12.02 and to do that what we do is we insert an array of Lac operator repeats 00:14:16.29 on the arm of chromosome IV 00:14:18.15 and then express in these cells a Lac repressor 00:14:21.23 that's been tagged with green fluorescent protein. 00:14:23.19 And so that Lac repressor, when it binds to these repeats, these Lac operator repeats 00:14:28.19 results in nice green fluorescent dot on the arm of chromosome IV. 00:14:33.29 That dot will, during metaphase, will be a single dot but when they separate in anaphase 00:14:38.06 that dot will become two dots and so we can determine the timing 00:14:41.12 of separation of chromosome IV that way. 00:14:44.23 Likewise, we labeled another chromosome, chromosome V as well 00:14:48.00 and in this case we used Tet operator repeats and expressed in the cell 00:14:51.18 a GFP-tagged version of the Tet repressor and that results in a green fluorescent dot 00:14:55.26 on chromosome V. And as I've indicated here basically fortuitously 00:15:01.04 it turned out that this chromosome V green fluorescent dot 00:15:03.26 was much brighter than the dot on chromosome IV and that actually allowed us to distinguish 00:15:08.00 between the two chromosomes in our experiments. 00:15:11.06 And so what we've created is a yeast strain in which we have two different 00:15:13.22 chromosomes labeled with green fluorescent dots 00:15:15.26 and now what we do is use microscopy to measure the relative timing 00:15:19.22 of the separation of those two chromosomes as an indication of anaphase synchrony. 00:15:26.22 And so this next slide shows what happens in a wild-type cell. 00:15:30.28 And what we did for these experiments is we subjected these cells to 00:15:33.25 spinning disc confocal microscopy, in which we're basically taking images every 10 seconds 00:15:40.01 over a period of several minutes in an asynchronous population 00:15:42.27 and then we go back to the videos and look for mitotic cells 00:15:45.23 as they progress through the metaphase to anaphase transition. 00:15:48.23 And one such cell is shown here in this video montage. 00:15:52.15 And so we're starting out in the upper left with a metaphase cell in which there are 00:15:56.12 two green dots visible in the cell and then at this time point right here 00:16:00.16 at the 20 second mark that top green dot separates into two green dots 00:16:04.07 indicating that one chromosome has now separated and segregated. 00:16:08.22 And then we follow this time course along every 10 seconds 00:16:11.09 until at this time point right about here 00:16:13.17 the other green dot separates into two distinct green dots 00:16:17.00 and that of course represents the second sister chromatid separation event. 00:16:20.15 And then from that point onward, those separated sisters then migrate to opposite poles 00:16:24.20 of the anaphase spindle until by the end of anaphase here the 00:16:27.23 two green dots of each sister are now fully separated to opposite ends of the cell. 00:16:33.25 So there are a few surprises in these results. 00:16:36.22 First of all, the big surprise was the amount of time 00:16:38.28 between the separation of these two events. 00:16:40.23 Clearly, 90 seconds difference in the separation of these two sister chromatid pairs 00:16:44.29 represents an awful lot of time given that mitosis in these cells is only a few minutes long. 00:16:49.21 And so we were a little surprised to discover how asynchronous 00:16:53.02 anaphase actually is in these cells 00:16:55.03 not nearly as synchronous as we thought it would be. 00:16:57.24 That was the first surprise. The second was that using the relative intensity of the green dots 00:17:02.13 as I said, we could distinguish between chromosome IV and chromosome V 00:17:05.16 and that led to the surprising realization that in all these cells 00:17:09.02 chromosome IV is always separating before chromosome V. 00:17:12.23 In other words, chromosomes don't separate in a random order, but appear to be ordered 00:17:16.23 such that IV always goes before V. 00:17:19.23 So the big question, of course, is: What does this look like in the mutant securin-2A cells? 00:17:25.15 And in the next couple of slides I'll show you those results. 00:17:28.09 So first this slide gives a summary of all the results 00:17:30.19 with a number of different wild type cells that 00:17:32.07 were analyzed, simply looking at the time between separation in a large number of cells. 00:17:36.11 So this is a histogram showing the distribution of results in wild-type cells 00:17:40.04 and as I mentioned, the median time of separation is about 90 seconds in these wild type cells. 00:17:45.16 Then in the securin-2A mutant, sure enough, we saw a 00:17:49.26 significant delay in the timing of separation 00:17:52.09 of the two chromosomes such that the median time of separation was more like 00:17:55.08 170 seconds rather than 90. And so we're seeing a roughly 00:18:00.06 two-fold increase in the amount of time 00:18:01.28 between the separation of these two chromosomes suggesting that, sure enough, 00:18:05.10 this positive feedback loop is important for the synchrony of anaphase in these cells. 00:18:11.09 This also has an impact on the accuracy of chromosome segregation. 00:18:15.13 By looking at the green dots in these cells and counting the number 00:18:18.07 of dots in the population, we can actually get pretty good idea of the frequency with which 00:18:22.20 chromosomes are mis-segregated during mitosis and by that sort of approach 00:18:26.28 we estimate that 2.5% of the cells in the securin-2A population 00:18:31.12 have experienced a serious segregation error indicating that this positive feedback loop 00:18:36.26 is really important for the normal fidelity of anaphase as well. 00:18:44.05 And so in summary, the idea is that the positive feedback loop that we've identified 00:18:48.19 here, this dephosphorylation of securin is really crucial 00:18:51.26 for the synchrony of anaphase and probably also for the fidelity of anaphase. 00:18:56.11 And that these positive feedback loops like this one 00:19:00.04 and possibly others that we haven't yet identified 00:19:02.06 are important for the normal, high fidelity, high synchrony anaphase 00:19:06.00 that is observed in most cells. 00:19:08.09 Itâ€™s not clear at this point, whether these sorts of regulatory mechanisms 00:19:12.00 also exist in higher eukaryotes, although I think itâ€™s very likely that they do 00:19:16.06 and will be discovered in the coming years. 00:19:19.20 So why is all this important? Well, the fact is that cancer, 00:19:23.22 a very common disease of humans of course, 00:19:25.17 is in large part a disease that's brought about by genomic instability. 00:19:30.17 And errors in chromosome separation are likely to be very important 00:19:33.28 in providing genomic instability that leads to an enhanced rate of tumorigenesis. 00:19:39.20 And so understanding these detailed mechanisms of chromosome segregation is 00:19:43.03 really crucial for understanding the genesis of cancer. 00:19:47.23 And with that, I'd like to thank the people who did all the work 00:19:50.29 that I've told you about in this lecture and in my second lecture 00:19:53.17 because none of the things I showed you were actually done by me personally 00:19:57.10 they're all done by a variety of students and wonderful collaborators with whom 00:20:00.05 I've worked over the past few years. 00:20:02.27 Two of the people that did much of the early work with the Cdk substrates in my second lecture 00:20:07.03 Jeff Ubersax and Mart Loog were previous lab members who established a lot of that 00:20:10.28 important early work and then more recent studies of Cdk substrates were carried out 00:20:15.10 by Liam Holt a student who collaborated on the mass spectrometry project 00:20:19.09 for the identification of the recent Cdk substrates. 00:20:24.16 Liam was also responsible for the anaphase switch studies that I just told you about in this lecture. 00:20:31.13 Throughout these studies our collaborator Kevan Shokat at UCSF has been 00:20:35.08 instrumental in the analysis of Cdk substrates. 00:20:37.15 In my second lecture I told you about how the evolution of Cdk substrates might give some 00:20:43.27 interesting insights into mechanisms of phospho-regulation and that was all done 00:20:47.15 by a bioinformatics student Brian Tuch in Sandy Johnson's lab at UCSF. 00:20:52.16 Kurt Thorn helped us with microscopy in this anaphase project that I just told you about, 00:20:57.07 and Andrew Krutchinsky is a mass spectrometrist who helped us 00:20:59.19 with the phosphorylation sites on securin. 00:21:02.26 And finally, Judit Villen, a post-doctoral fellow in the lab of Steve Gygi 00:21:06.06 at Harvard performed this remarkable, large phospho-proteomic analysis 00:21:10.22 of Cdk substrates that I told you about in the second lecture. 00:21:14.05 And thank you for your attention.