Controlling the Cell Cycle
Transcript of Part 2: Controlling the Cell Cycle: Cdk Substrates
00:00:02.03 So, hello, my name is Dave Morgan. I'm from the University of California in San Francisco. 00:00:06.04 And in this lecture I'm going to go over some of my own work on studies of 00:00:09.23 how the cyclin-dependent kinases drive the events of the eukaryotic cell division cycle. 00:00:15.29 Now it’s well established at this point that the major regulators 00:00:19.11 of the eukaryotic cell cycle are the cyclin dependent kinases or Cdks. 00:00:23.13 And the basic idea is that a series of Cdk-cyclin complexes are activated 00:00:28.11 in a specific sequence during the cell cycle 00:00:30.08 to trigger the events of the cell cycle in the appropriate order. 00:00:33.21 And so, for example, S-phase Cdk-cyclin complex is formed in late mitosis or in late G1 00:00:39.15 and are then activated in the beginning of S-phase to initiate DNA synthesis. 00:00:43.03 And then M-phase Cdk-cyclin complexes form at the end of G2 00:00:47.06 and are activated to initiate the events of mitosis and take the cell to metaphase. 00:00:52.17 So the big question that I want to address today is how is it that these 00:00:55.26 Cdks actually drive these cell cycle events? 00:00:59.05 Now obviously, Cdks are protein kinases, which means that the most likely mechanism 00:01:02.24 by which they promote cell cycle events is through the phosphorylation of other proteins 00:01:07.10 which then bring about those events. 00:01:09.15 And so over the past 10 or 12 years or so, we've dedicated quite a lot of effort 00:01:13.18 to identifying the substrates of the cyclin-dependent kinases 00:01:16.20 in the hope that that will lead us to an answer to this question of 00:01:19.17 how the Cdks actually initiate cell cycle events. 00:01:23.23 So in this lecture I'm going to tell you about two methods that we've used 00:01:26.23 to systematically and comprehensively identify Cdk substrates 00:01:30.09 and then in the second half of the lecture we'll go into some interesting ways 00:01:33.20 in which we use those lists of substrates to address some general questions of 00:01:37.19 Cdk function and phospho-regulation. 00:01:41.07 So the first method we use to identify Cdk substrates 00:01:43.22 began about 10 or 12 years ago in a collaboration with Kevan Shokat, a chemist here at UCSF. 00:01:51.00 Now, Kevan came up with an idea whereby it would be possible to label the specific targets of 00:01:57.04 a protein kinase in a crude cell mixture and this slide attempts to explain that basic method. 00:02:02.09 On the left...let's focus on the left first. 00:02:04.08 On the left is a wild type regular protein kinase like Cdk1 with its cyclin regulatory partner. 00:02:10.08 And typically when one wants to label the targets of a protein kinase, like Cdk1, 00:02:14.28 you simply provide that kinase with a version of ATP in which the gamma phosphate 00:02:20.00 is labeled with a radioactive tag. 00:02:22.28 And then when that protein kinase uses that ATP it will then transfer 00:02:27.24 that P-32 onto its substrates. 00:02:31.11 Now, unfortunately this cannot be used to identify unknown substrates 00:02:34.24 of Cdks because if you take a pure Cdk and some gamma labeled P-32-ATP 00:02:40.16 and put that into a crude cell lysate, you will get not only the labeling of the Cdk's targets 00:02:46.12 but the labeling of all the other protein kinase targets in that lysate because 00:02:49.15 that ATP can be used by any kinase. 00:02:52.22 And so Kevan Shokat's idea was to avoid this problem by using 00:02:56.15 so called analog-sensitive protein kinases. 00:02:58.29 And the strategy is based on the fact that protein kinases 00:03:02.07 tend to contain a large hydrophobic residue in the wall 00:03:05.19 of the adenine binding pocket of their active site. 00:03:08.21 And so the basic strategy is to mutate that large hydrophobic side chain there 00:03:12.28 to a glycine residue resulting in the formation of an extra pocket in 00:03:16.18 the side of that ATP binding site. And as a result 00:03:20.00 this mutant kinase can now use a bulky ATP analog in which extra moieties 00:03:24.25 have been added to the adenine base of the ATP. 00:03:27.28 And so for example, N6-benzyl-ATP can be used by the analog sensitive Cdk1 kinase 00:03:32.28 but cannot be used by a wild type kinase because that bulky ATP analog 00:03:37.04 can't fit into the wild type active site. 00:03:41.14 And so, of course, if you put a radio-label on the gamma phosphate 00:03:44.05 of this bulky ATP analog and then add this kinase to a crude cell lysate 00:03:48.19 what you hope to get is the specific labeling of just the direct targets of 00:03:52.23 that protein kinase and no other kinases in the cell lysate 00:03:56.00 because those other kinases can't use this bulky ATP analog. 00:04:00.12 So this method...we developed this method in collaboration with Kevan Shokat 00:04:04.04 a number of years ago and applied it to the yeast Cdk1 kinase 00:04:08.28 and the results with that are shown in the next slide. 00:04:12.09 So this slide shows an autoradiograph of a protein gel 00:04:15.14 in which we've separated the reaction products 00:04:16.28 from three different reactions, two of which are control reactions 00:04:20.03 and the third of which is the experimental reaction. 00:04:23.09 In the first lane what you see is what happens when you add this radiolabeled N6-benzyl-ATP 00:04:28.16 this bulky ATP to a crude cell extract made from yeast. 00:04:33.05 And the result is that you get very little labeling of anything in that cell extract 00:04:36.14 because that ATP analog cannot be used by the protein kinases in that crude cell extract. 00:04:42.14 The next lane is another control in which we're mixing the purified protein kinase Cdk1-as1 00:04:48.23 together with a cyclin partner and then adding that to some N6-benzyl-ATP 00:04:54.01 that's radio-labeled on its gamma position and the result then is that you 00:04:57.10 see auto-phosphorylation of the cyclin subunit of the Cdk-cyclin complex. 00:05:01.21 And so that results in a background band in the experimental lane over here. 00:05:07.03 But the third lane is really the crucial lane in which all three components 00:05:09.22 have been added. And so we're adding a purified kinase...analog sensitive kinase 00:05:13.14 with the bulky ATP analog and the cell extract and the result 00:05:17.21 is that you see a whole raft of different proteins being radio-labeled 00:05:22.06 in the cell extract and those proteins are presumably the direct targets of 00:05:27.00 Cdk1-cyclin complexes in that lysate. 00:05:31.12 So we obtained this result a number of years ago and then 00:05:33.29 dedicated quite a lot of effort to identifying 00:05:36.03 these various radiolabeled bands in this cell lysate. 00:05:38.05 And to make very long story short, we ended up using proteomic libraries 00:05:42.24 to individually identify substrates and in the end 00:05:45.27 we came up with a list of about 181 proteins in cell extracts that 00:05:50.04 are rapidly modified by Cdk1-cyclin complexes. 00:05:53.27 And so this list of proteins provided us with our initial list of Cdk substrates. 00:05:59.23 These substrates are involved in a wide range of different cellular processes; 00:06:02.25 many of which are known to be connected to the cell cycle in some way 00:06:05.20 and are likely to represent important targets of Cdk1 throughout the cell cycle. 00:06:11.01 But for various reasons we decided that this list of substrates was 00:06:14.02 incomplete and also, because it was done in vitro we wanted to get 00:06:18.03 another approach that would allow us to identify, comprehensively, a larger number 00:06:23.09 of Cdk substrates that were modified in vivo by Cdk1. 00:06:26.23 And so the second method we've been using more recently has been to use 00:06:30.17 quantitative mass spectrometry approaches to identify all the phosphorylation 00:06:34.10 sites in the cell that are dependent on Cdk1. 00:06:37.18 In other words, phosphorylation sites whose levels decrease abruptly 00:06:40.18 when you inhibit the protein kinase activity of Cdk1. 00:06:45.12 And that method begins again with the analog sensitive Cdk1 mutant. 00:06:49.17 Now, another advantage of these analog sensitive mutants is that not only do they bind 00:06:53.24 bulky ATP analogs, but they also bind bulky inhibitors that can only fit into the active site 00:06:59.24 of the analog sensitive kinase but not the active site of a wild type kinase. 00:07:05.02 And so, for example, this inhibitor here 1-NM-PP1 binds with extremely high affinity to 00:07:10.05 analog sensitive Cdk1 but has essentially no affinity for the wild type kinase 00:07:14.06 or for any other kinase in the yeast cell. 00:07:16.23 And so we could use this analog sensitive Cdk1 to make a yeast strain in which 00:07:22.02 we can inhibit Cdk1 in vivo rapidly and specifically. 00:07:26.22 And so we did that a number of years ago. We created a yeast strain in which 00:07:29.12 the endogenous Cdk1 protein is replaced with the analog sensitive protein 00:07:33.23 and in that yeast strain it is now possible to almost completely and specifically inhibit 00:07:38.05 Cdk1 activity within minutes by the addition of 1-NM-PP1 to the culture medium. 00:07:44.19 So we used that strain in 00:07:46.08 this quantitative mass spectrometry approach that I want to tell you about 00:07:49.16 which was done in a collaboration with Judith Villen and Steve Gygi of Harvard University. 00:07:55.11 And the basic approach that we used is illustrated in this slide and in the next two slides as well. 00:08:01.11 It begins, as I said, with the analog sensitive yeast strain cdk1-as cells 00:08:05.06 in which Cdk1 can be inhibited specifically with the 1NM-PP1 inhibitor. 00:08:10.21 And what we do is we grow two parallel cultures of this yeast strain. 00:08:14.14 One culture, the so called light culture, is grown in regular lysine and arginine 00:08:19.21 whereas the so called heavy culture is grown in a different form of lysine and arginine 00:08:23.11 in which carbon-13 and nitrogen-15 have replaced the usual carbon-12 and nitrogen-14. 00:08:29.21 And so, as a result, after growth in this medium for some time, all the proteins in these cells 00:08:34.03 have been labeled with slightly heavier than average 00:08:37.04 lysine and arginine residues which means that all 00:08:39.10 the peptides derived from this culture will have a slightly higher mass 00:08:43.01 in the eventual mass spectrometry analysis 00:08:45.21 and that will allow us to identify the peptides coming from these two lysates. 00:08:50.22 So we treat the heavy culture with the inhibitor 1-NM-PP1 for a brief period, 15 minutes. 00:08:56.03 And then we harvest these cells after the inhibitor treatment. 00:09:00.08 Harvest the cells, mix them together, lyse them, break them open, 00:09:04.03 and then treat all the resulting proteins in those cell lysates 00:09:08.18 with trypsin to break them all down into tryptic peptides. 00:09:12.06 And then Judit Villen in the Gygi lab has developed a wide range of 00:09:15.27 powerful methods for purifying the phospho-peptides out of that tryptic peptide mixture. 00:09:20.20 And then we then subject those phospho-peptides 00:09:22.25 to mass spectrometry as shown in the next slide. 00:09:26.23 There are two basic forms of mass spectrometry that are applied to these phospho-peptide mixtures. 00:09:31.07 The first, on top, is to use conventional tandem mass spectrometry methods 00:09:35.15 to actually fragment these peptides and use those fragments 00:09:38.07 to determine their sequence. 00:09:40.02 And so, by this approach we can determine the sequence of all the phospho-peptides 00:09:44.04 coming out of these yeast lysates and just as importantly, we can identify 00:09:47.10 the precise site of the phosphorylation on those peptides. 00:09:50.29 And so by doing this, Judit was able to produce a list 00:09:54.12 of about 10,000 phosphorylation sites on 2,000 different proteins in the yeast lysate. 00:10:00.25 And then, in addition to determining sequence, we also quantify all the peptides 00:10:06.12 and determine the relative amount of the so called light and heavy peptides. 00:10:10.19 What this means is that every peptide coming out of these phospho-peptide mixtures 00:10:15.15 comes in both a light form which originally came from the light medium culture 00:10:19.13 and a heavy form that originally came from the inhibitor-treated heavy culture. 00:10:23.17 And they can be distinguished based on this slight 00:10:25.08 mass difference of their lysines and arginines. 00:10:28.12 And what we're looking for, of course, are peptides that look like this: 00:10:30.29 where the heavy peptide is much less abundant than the light peptide. 00:10:34.16 And that means that that peptide's abundance was inhibited 00:10:37.23 or decreased as a result of Cdk1 inhibition and therefore 00:10:41.02 that phosphorylation site on that peptide represents a 00:10:43.23 Cdk1-dependent phosphorylation site in vivo. 00:10:47.23 And so by applying this approach to the many phosphorylation sites identified here 00:10:52.18 we came up with a list of about 547 phosphorylation sites on about 308 proteins 00:10:58.10 that were clearly Cdk1-dependent and represent likely candidates for Cdk1 targets in vivo. 00:11:04.13 This list of targets included many of the same proteins 00:11:07.02 we had identified in our previous screen in vitro 00:11:09.12 and so for those proteins at least we have very good evidence that these 00:11:11.29 proteins are kinase substrates both in vitro and in vivo. 00:11:17.11 Now the list of substrates includes a wide range of proteins involved in a wide range of processes. 00:11:23.20 I'm not expecting you to see or read any of the gene names on these lists here. 00:11:26.27 This slide is simply meant to illustrate that we have lists of proteins involved in 00:11:30.14 a wide range of interesting processes. Some of these processes are totally expected. 00:11:35.00 For example, DNA replication, spindle behavior, kinetochores and cytokinesis are all 00:11:40.08 processes in which we expect Cdks to be involved in regulating some aspect of those processes. 00:11:45.10 There's also a few surprises here as well. 00:11:47.27 Protein translation, chromatin structure, and nuclear transport and the secretory pathway 00:11:53.09 all have a number of Cdk substrates involved 00:11:56.08 in those processes and so one might imagine 00:11:58.15 that this will lead to some new understanding of how Cdks might control those processes 00:12:02.17 as well as the more conventional cell cycle regulated processes. 00:12:07.14 But for the rest of this lecture today, I'm not going to talk in detail about 00:12:10.21 any specific substrates or processes, but instead I'm 00:12:13.14 going to tell you how we used our lists of substrates 00:12:16.02 to address some interesting general questions on how cell cycle progression is 00:12:20.10 controlled by Cdks in general. 00:12:23.13 And so we're going to address two questions in the remainder of this lecture. 00:12:26.14 The first one of which is shown on this next slide. 00:12:30.15 And that question is this one: How do different cyclins trigger different cell cycle events? 00:12:35.10 So I told you at the beginning of the lecture that 00:12:37.02 S-phase cyclin Cdk complexes initiate S-phase and mitotic Cdk-cyclin complexes 00:12:42.00 initiate M-Phase and there's good evidence from yeast genetics and elsewhere 00:12:46.00 that S-phase Cdk-cyclin complexes have a better intrinsic ability to 00:12:49.21 initiate S-phase than a mitotic cyclin-Cdk complex. 00:12:54.01 So there's something different about cyclin-Cdk complexes that are activated at S-phase 00:12:58.28 that allows them to more effectively activate the onset of S-phase. 00:13:02.05 And so, what is that difference? 00:13:04.12 Well, one obvious possibility is that the cyclin that associates with the Cdk 00:13:07.21 helps determine the substrate specificity of that Cdk. 00:13:11.20 So in budding yeast, for example, where there's only a single Cdk 00:13:14.20 associating with all these different cyclins, 00:13:16.08 one can imagine that associating with an S-phase cyclin 00:13:19.20 at the beginning of S-phase might target that Cdk for specific substrates involved in S-phase. 00:13:26.01 And so we decided we could address this question on a more global level 00:13:29.15 by actually analyzing the relative phosphorylation rate of 00:13:32.09 a wide range of Cdk substrates using purified S-phase Cdks and M-phase Cdks. 00:13:38.15 And specifically we carried out these studies using the S-phase cyclin Clb5 from budding yeast 00:13:44.08 and the M-phase cyclin Clb2 from budding yeast. 00:13:47.15 And Mart Loog, a post-doc in the lab, basically purified these two kinases 00:13:51.14 and then tested their activity towards about 150 different Cdk substrates 00:13:55.17 to look for substrates that were highly specific for one or the other. 00:13:59.21 Some of his early results are shown in this next slide 00:14:02.07 which gives you an illustration of the sort of thing we found. 00:14:06.03 Here we're looking at autoradiographs of protein gels in which three different proteins 00:14:11.00 listed across the top--Mcm3, Orc2, and Orc6 00:14:14.00 have been treated with either the mitotic cyclin-Cdk complex on the left 00:14:18.14 or the S-phase complex on the right. 00:14:20.22 And you can see, quite clearly, that these three proteins are all phosphorylated 00:14:24.08 much more rapidly by the S-phase Cdk-cyclin complex Clb5. 00:14:29.21 So Mart went ahead and did this exact same reaction with about 150 proteins as I said 00:14:35.06 and the results from those experiments are shown on this slide. 00:14:37.24 So this slide summarizes everything that he found. 00:14:40.15 What we're looking at here is a plot of about 150 proteins 00:14:44.06 each one of which is represented by these little circles on this plot. 00:14:47.10 And these circles are plotted according to the rate of their phosphorylation 00:14:52.02 by Clb2 on that axis and Clb5 on this axis. 00:14:55.22 And so most of the proteins are falling along the diagonal of this plot 00:14:59.21 indicating that they are equally well phosphorylated by both kinases. 00:15:02.24 In other words, they're not cyclin specific targets. 00:15:05.09 However, we found a quite large number of proteins over here on the right 00:15:09.17 especially these red circles here that represent proteins that 00:15:12.25 are far more rapidly phosphorylated by Clb5-Cdk1 than they are by Clb2-Cdk1. 00:15:19.18 So these proteins, and note by the way that this is a log phase scale here 00:15:23.03 so some of these proteins are 10 or over 100 or even 1000 fold more rapidly phosphorylated by 00:15:28.02 Clb5-Cdk1 than by Clb2-Cdk1. So these clearly represent proteins that are highly Clb5 specific. 00:15:35.05 That the cyclin is somehow determining or increasing 00:15:37.29 the rate of phosphorylation of these proteins. 00:15:40.14 So what are these proteins? Well, we were satisfied to see that at least five of them 00:15:46.07 are proteins known to be involved in DNA replication, especially Sld2 here. 00:15:50.07 Sld2 is a protein whose phosphorylation is known 00:15:52.25 to be crucial for the initiation of DNA replication. 00:15:55.16 And so these proteins make perfect sense as Clb5 specific targets because those are 00:16:00.04 the proteins that we need to phosphorylate early in S-phase 00:16:02.22 to help drive progression through chromosome duplication. 00:16:07.27 So what this list of cyclin-specific substrates in hand, we next addressed 00:16:12.10 the mechanism underlying this cyclin specificity. 00:16:14.25 Why is it that Clb5-Cdk1 phosphorylates these proteins 00:16:18.06 so much more rapidly than Clb2-Cdk1? 00:16:21.13 Through kinetic studies we discovered that the reason for this higher rate of phosphorylation was 00:16:26.15 that these substrates have a much higher affinity for the Cdk-cyclin complex 00:16:30.05 when Clb5 is associated, suggesting that they might associate with that cyclin subunit. 00:16:36.16 In fact, there's previous suggestions of what 00:16:38.04 the mechanism for this association might be. 00:16:40.25 And those are based on the known crystal structures of Cdk-cyclin complexes from human cells. 00:16:46.06 So this shows the crystal structure of a Cdk-cyclin complex from humans 00:16:50.20 that illustrates very nicely the basic parts of the Cdk-cyclin complex 00:16:55.08 and where cyclin substrates typically associate with this complex. 00:16:59.01 Over on the left is the Cdk catalytic subunit and between these two lobes here 00:17:02.29 is an active site cleft in which you can see this ATP molecule binding right here. 00:17:08.19 Typically a protein substrate would bind along the surface of this protein kinase right here 00:17:12.29 in a way that the serine or threonine hydroxyl would be positioned in such a way 00:17:17.25 to allow the transfer of phosphate from that ATP onto the hydroxyl residue. 00:17:23.10 So, the primary site of substrate association with the Cdk-cyclin complex 00:17:27.00 is of course the active site, the place where that serine or threonine 00:17:30.21 associates with its sequence contacts to be phosphorylated. 00:17:35.16 However, this is probably not the only site of substrate association in a Cdk-cyclin complex. 00:17:39.28 There is considerable evidence from mammals and from yeast as well 00:17:43.13 that there is a docking site on this cyclin itself 00:17:46.20 that can also associate to some extent with parts of the substrate. 00:17:50.10 And this docking site is mostly composed of this large alpha-helix here 00:17:54.18 that contains a number of hydrophobic residues 00:17:57.06 that are together called the hydrophobic patch. 00:17:59.16 It is involved in associating with certain substrates 00:18:01.23 and enhancing activity towards those substrates. 00:18:07.02 So, we obviously hypothesized that perhaps this docking site on Clb5 00:18:10.16 exists on Clb5 and that this docking site is required 00:18:13.26 for the cyclin-specific phosphorylation that we saw in our experiments. 00:18:18.03 And so to test that the obvious approach was to mutagenize this docking site 00:18:21.09 through a number of single point mutations and then test whether that 00:18:25.08 has any impact on cyclin specificity and that is shown in this slide. 00:18:28.20 And the answer was a definite yes, that mutation of that docking site 00:18:32.27 completely abolishes the Clb5 specificity that we had seen. 00:18:35.23 So here again we're looking at autoradiographs of protein phosphorylation 00:18:39.27 by purified Clb2 on the left two lanes and Clb5 on the right. 00:18:43.20 And we're looking at the phosphorylation of 5 highly Clb5 specific proteins. 00:18:47.20 And you can see that the wild type Clb2, the wt here, phosphorylates these proteins 00:18:52.13 rather poorly whereas wild type Clb5 phosphorylates them extremely well. 00:18:57.18 Once again, indicating how specific these proteins are for Clb5. 00:19:01.03 However, if you mutate the hydrophobic patch or the docking site 00:19:04.15 on Clb5 you find that that specific phosphorylation is almost completely lost 00:19:09.12 indicating that that site is really required for the increased affinity 00:19:12.25 that Clb5-Cdk1 has for these substrates. 00:19:18.20 So we conclude that an interaction, probably simultaneous between 00:19:22.01 this docking site and the active site, allows specific Clb5 substrates to interact 00:19:26.19 with the Clb5-Cdk complex in a high affinity fashion 00:19:30.10 that allows more rapid phosphorylation of those proteins. 00:19:34.20 And so that leads us to at least a partial answer for the question that I first posed 00:19:39.04 which is: How do different cyclins drive different cell cycle events? 00:19:43.07 Well, part of the answer appears to be that the associated cyclin that associates 00:19:47.01 with the Cdk helps target that Cdk to specific substrates. 00:19:51.01 And so S-phase Cdk-cyclin complexes when they're activated 00:19:54.14 at the end of G1 tend to phosphorylate more rapidly the proteins 00:19:57.25 that are most important for initiating S-phase. 00:20:02.06 OK, now I want to turn to an entirely different sort of general question 00:20:05.05 that we also used our substrate lists to address. 00:20:09.13 And in particular, we used our recent mass spectrometry analysis 00:20:12.11 and our 547 Cdk1-dependent phosphorylation sites to address this question. 00:20:17.20 And this is a much more general question that just...that goes beyond 00:20:21.28 issues of simple cell cycle control but reaches into areas involved in 00:20:25.26 the general issues of phospho-regulation. And the question is this one: 00:20:30.10 How is it that phosphorylation changes the function of a protein? 00:20:33.10 How is it that the addition of a phosphate group to a protein changes that protein's function 00:20:37.20 in a way that allows it to initiate cell cycle events or do other things? 00:20:41.27 And there are typically a couple of different approaches 00:20:44.09 or different mechanisms that are thought to be involved in changing protein function. 00:20:48.03 And the first and possibly most commonly imagined mechanism is this one here 00:20:51.25 the so-called allosteric switch. And the idea with this mechanism is that the placement 00:20:56.13 of a phosphate on a protein in a very specific location 00:20:59.06 causes a precise conformational change in that protein that then 00:21:02.24 initiates some change in its function, its enzymatic activity or its association with something. 00:21:08.21 Now, this mechanism, of course, requires that the position of that phosphorylation site 00:21:11.29 is extremely precise and conserved. In other words, 00:21:15.10 you can't put a phosphate just anywhere on a protein and 00:21:17.23 cause this very precise conformational change. 00:21:20.08 It has to be extremely well positioned 00:21:21.21 and because of that it’s very difficult to evolve that sort of phospho-regulation. 00:21:26.23 That kind of phosphorylation cannot appear randomly very easily 00:21:31.22 and achieve the kind of regulation that is required. 00:21:35.08 So the alternative mechanism is this one--which I call bulk electrostatics. 00:21:40.08 And this mechanism suggests that the position of the phosphorylation 00:21:43.14 does not require such precise position of the phosphorylation. 00:21:47.27 The basic idea here is that the placement of clusters of phosphorylation sites 00:21:51.22 on the surface of the protein, typically on a loop or a disordered region on the surface 00:21:56.01 can result in interesting regulation such as interference with association with another protein 00:22:01.16 or for that matter, promotion of association with phosphate binding proteins. 00:22:06.04 And so this very simple mechanism of phospho-regulation 00:22:08.18 can occur by the placement of clusters of phosphates in a general region 00:22:13.19 of a protein but the exact position of each of those phosphates is not absolutely important, 00:22:18.06 not critical and therefore the position of those phosphates can shift during evolution 00:22:22.05 in different proteins. And so for that reason this mechanism is much more easily evolved. 00:22:28.10 It’s very easy to imagine that random mutations 00:22:30.09 could result in the appearance of phosphorylation sites 00:22:32.19 on the surface of certain proteins where they interact with other proteins 00:22:35.17 and that can result in regulatory possibilities that could be selected for. 00:22:41.08 And so, both of these mechanisms are known to be important in different cases. 00:22:46.12 There are examples of proteins that are regulated in both of these ways. 00:22:49.17 But we thought that perhaps our giant list of Cdk substrates would allow us to 00:22:53.11 address the relative importance of these two mechanisms more generally. 00:22:58.06 And so what we did is we took those 547 Cdk1-dependent phosphorylation sites 00:23:03.06 and aligned them with homologous sequences from other species 00:23:07.17 to see how well these phosphorylation sites are actually conserved. 00:23:10.27 And the basic idea here was that if we found that 00:23:13.11 sites are generally, extremely well preserved that might argue for this sort of mechanism, 00:23:17.24 but sites that drift during evolution might argue for this sort of mechanism. 00:23:23.23 So the next slide gives you an illustration of the sort of thing that we found. 00:23:27.14 Now, here we are aligning a bunch of different protein sequences 00:23:29.29 and I don't expect you to actually read these sequences. 00:23:32.17 The important thing is that there are these little yellow boxes that represent 00:23:35.15 a SP or TP di-peptide motifs that are the consensus sequences for Cdk phosphorylation. 00:23:42.09 And along the very top here is the sequence of a part of a protein 00:23:45.27 called Shp1 that we identified two phosphorylation sites in our mass spectrometry experiments. 00:23:51.24 Those sites are site A and site B. 00:23:53.26 One site is over here in a region of the protein that is known 00:23:57.05 to form into a globular domain and anther site is here in a region 00:24:00.21 that's predicted to form a disordered domain. 00:24:03.16 And these other sequences that lie below this top sequence from budding yeast 00:24:06.26 are the sequences of orthologous proteins from various yeast species 00:24:11.25 whose genomes have been sequenced 00:24:13.20 starting with the most closely related yeast here and 00:24:15.25 moving all the way down to the most distantly related yeast at the bottom. 00:24:20.03 And so these protein alignments tell us some very simple things. 00:24:23.19 First of all, site A here is very well conserved in evolution 00:24:27.10 and you can see that almost all of the orthologs of this protein in all these other yeast species 00:24:31.14 contain a likely Cdk consensus site at the exact same position in this highly conserved region. 00:24:38.25 So site A appears to represent an example of the kind of site I mentioned 00:24:42.09 in the left side of the previous slide, a site that is highly conserved in evolution. 00:24:46.27 But site B is not. Site B is very poorly conserved and disappears essentially 00:24:51.15 after a few species and is no longer found at that position in other orthologs. 00:24:57.24 However, if you look in this region of these other species' proteins, you find that 00:25:01.21 SP and TP di-peptide motifs appear scattered throughout this region 00:25:05.08 in a larger number of these yeast homologous proteins, 00:25:08.20 suggesting that even though the initial position here in budding yeast has not been preserved, 00:25:14.09 the Cdk phosphorylation of this region has been conserved in evolution 00:25:18.21 but the exact position of the phosphorylation sites 00:25:20.23 has been shifting dramatically over evolution. 00:25:23.27 So this is obviously consistent with the second idea, 00:25:26.13 that precise phosphorylation site positioning is not 00:25:29.16 required here because these sites might be involved in some more simple, general 00:25:34.03 regulatory mechanism involving association with phosphate binding proteins 00:25:38.03 or interference with protein binding. 00:25:41.11 So we did this exact same alignment for all 547 of our phosphorylation sites 00:25:46.09 and then in the next slide what I'm going to show you 00:25:47.28 is a somewhat complex graphic that illustrates the results that we found from that. 00:25:53.14 And this was done in collaboration with Brian Tuch, a graduate student 00:25:56.22 working in the laboratory of Sandy Johnson at UCSF. 00:26:00.09 And this top plot here represents a hierarchically clustered 00:26:02.27 clustergram as we call it that illustrates the conservation of the precise position of 00:26:08.26 phosphorylation sites in orthologs of the proteins we identified. 00:26:13.29 And so what we're looking at here is a graphic in which there are 547 columns 00:26:18.04 in this graphic, each one of which represents a single Cdk1 dependent phosphorylation site 00:26:23.14 that we identified by mass spectrometry. 00:26:26.17 And then each row in this graphic represents 00:26:29.22 how that phosphorylation site aligns with its orthologs in other species, 00:26:34.06 the same yeast species that I showed in the previous slide 00:26:36.23 starting with the most closely related yeast species 00:26:39.03 and then working to the most distantly related ones. 00:26:41.28 And in each column a yellow box indicates that that phosphorylation site 00:26:47.06 is precisely conserved in its position in that orthologous sequence. 00:26:51.03 In other words, over here on the left this yellow box at the top moves down 00:26:55.16 for a few species and then disappears indicating that this site is only precise... 00:26:59.24 these columns here, these phosphorylation sites 00:27:02.12 are conserved only in the closely related 00:27:04.21 yeasts species but then are lost in all distantly related species. 00:27:10.03 And so by looking at this graph you can see that there's only 00:27:12.10 a small group of phosphorylation sites, these ones in here especially 00:27:16.06 and note particularly these ones here that are conserved 00:27:19.04 throughout all the yeast homologs that we identified. 00:27:21.23 And so this small number of phosphorylation sites, perhaps 30 or 40 of them are 00:27:26.12 at most, are preserved in large numbers of yeast species, indicating that 00:27:31.29 the precise position of phosphorylation 00:27:33.17 has been conserved in a relatively small number of cases. 00:27:37.16 OK, so how do we then test the possibility that instead of precise positioning 00:27:44.01 during evolution that we're looking at drifting phosphorylation site positioning? 00:27:48.28 And that required the development of another graphic which is shown below here 00:27:52.17 in which I'll take you through slowly because it’s a little bit complicated. 00:27:56.16 So in this case, once again, it’s another hierarchically clustered graphic in which 00:28:00.22 there are 547 columns, each representing a different phosphorylation site 00:28:05.15 identified in our analyses. But in this case... 00:28:08.17 and once again, the rows represent alignments with orthologs, orthologous proteins 00:28:13.00 from other yeast species. But in this case the yellow box 00:28:16.02 doesn't indicate precise positioning of phosphorylation site, but instead 00:28:19.18 indicates that the ortholog of that particular protein in these other yeast species 00:28:24.03 has a statistically enriched frequency of Cdk consensus sites, SP and TP motifs. 00:28:30.28 In other words, these yellow boxes represent proteins 00:28:33.21 in which the frequency of SP and TP motifs is far greater than that expected by chance. 00:28:39.00 And so these large numbers of proteins here represent proteins in which 00:28:43.08 even though the precise site of phosphorylation is not conserved 00:28:46.19 as shown up here, these proteins do contain a high frequency of 00:28:50.12 Cdk consensus sites whose positions are clearly drifting during evolution. 00:28:56.04 And so these large number of proteins over here on the right side of this clustergram 00:28:59.24 may represent proteins in which the precise position of phosphorylation 00:29:03.21 does not matter but the regulation of those proteins by phosphorylation 00:29:07.17 is conserved despite that. And so clearly we'd like to think that that evidence 00:29:14.00 tends to suggest that this mechanism on the right here 00:29:16.03 these easily evolved bulk electrostatic mechanism 00:29:19.05 is a major mechanism by which phospho-regulation can easily be evolved 00:29:23.10 and that drifting phosphorylation sites especially clusters of phosphorylation sites 00:29:26.19 on disordered regions is really crucial for 00:29:30.10 the regulation of many different Cdk substrates. 00:29:34.28 So with that, I want to leave us with the question that we started out this whole lecture with 00:29:39.20 and that is: How do Cdk's drive cell cycle events? 00:29:42.04 Well, clearly our list of Cdk substrates, the ones I'm showing here and 00:29:45.21 the many that aren't shown here, probably contain the answer to this question. 00:29:50.06 Clearly through the detailed analysis of large numbers of these substrates 00:29:53.03 and that will lead us to a much better understanding of how Cdks 00:29:57.17 drive the events of cell cycle and how they alter all these different processes 00:30:02.04 in the cell to initiate cell cycle events.
