 Do you want me to turn it on now or wait till I go? Well, it's because they turn it on now and make sure it's okay. Here we go. Life is good. Okay, I think we'll get started. It's my pleasure to introduce today's seminar speaker, Professor Brian Freeman from the University of Illinois, Urbana-Champaign. Brian and I met over Zoom, actually, when I gave a seminar at University of Illinois during the COVID pandemic. And while we work in different areas, I found them really fun to talk to you. Because he seems so curious about so many topics, which actually makes sense because chaperones are involved in virtually every aspect of cellular function. So you have to be curious about a lot of things, I think, to do good work in that area. Brian got his bachelor's in microbiology from University of Michigan Ann Arbor and his PhD in biochemistry and biophysics at Northwestern. So he is a real midwesterner. But after a successful postdoctoral stint with Keith Yamamoto at UCSF, he started his own independent career at University of Illinois in 2002, where he's now a professor in the Department of Cell and Developmental Biology. He also has affiliate appointments in both the cancer and genome biology institutes, attesting to his breadth of curiosity and his institutional commitment. And he has some very interesting ideas about promoting DEI initiatives at his institutions. And so, again, the breadth of thinking about different things and incorporating service into his academic life is impressive. He's the recipient of several honors during his career, most recently being elected a fellow of the American Association for the Advancement of Science, although the honors I was most envious of were the single-digit scores on multiple NIH grants, which I have never seen. So Brian's long, for myself, I mean, I've seen him for other people because I've served on study section. Brian's longstanding interests are in the conserved mechanisms by which chaperones guide the dynamic protein environment in our cells with an emphasis on central nuclear processes. And today, though, he's going to discuss recent progress on defining the cellular substrates of HSP90s. Awesome. Thank you so much, Catherine. It's been an absolute pleasure today. It was a wonderful meeting, Catherine, by Zoom and really getting to know her in person afterwards. And I've had just a terrific day talking with people. I think, if anything, the meetings were all too short because, as she said, I do like to chat and maybe talk a wee bit too much. That being said, thank you for the lovely introduction, Catherine. You're looking at, I will talk briefly about it today. And I put it up there because, honestly, I love this movie. It's something that keeps me thinking about how stuff works. And what you're looking at, I'm sure most of you realize this is budding yeast. This is mom down here. This is a daughter cell that forms. The nucleus upon budding is a closed nuclear environment. We have a cherry marking the nuclear envelope and the outer envelope of mom and daughter. And the green dot is actually one site of the DNA that's labeled with a GFP-LAC fusion. And what really fascinates me about it and one of the areas that my group studies, and I will talk a little bit about today, is watch when it gets over to the daughter cell and then suddenly it sort of zips all over the nucleus. And that's when the genome's being reorganized. And we know so little about how that process works. And my work on chaperones got us into that area. But I would have to admit, if someone says the word molecular chaperone to me, I don't think anything about the nucleus whatsoever. That's just not where my mind drifts. My mind really goes over. That got bright in a hurry. And the classic sort of chaperone biology, and that is really, I see Betty back there just did beautiful work along with Bill Welch showing that HSP-70 binds the nascent chains as they come out of the ribosome. And it's not that they guide folding, but they provide an environment, a friendly environment for nascent chains to fold so that those nascent chains as they grow won't aggregate with other nascent chains. And that's a beautiful function for certainly HSP-70. And of course, if a protein can't fold to it, and we'll talk more about this, the seeming native state, HSP-70 helps deliver that to degradation machinery such as the proteasome where Aaron Chekinover has spent a lot of time showing that that's another nice chaperone role. And of course, even if a protein does fold to a seeming native state, there's still chaperone usage. Proteins typically don't work as a single subunit. They work in protein complexes. Many of the chaperones contribute here, including GroYale or TREC homologs. And then there's the area that my group likes to spend a lot of time on. And we started with supporting metastable proteins where HSP-90 has a large role to help create an activatable state with sort of beginning hypothesis. 90 doesn't do this by itself, but it does it with a whole slew of co-chaperones. But here we already have four wildly different roles for chaperones in the cell. And that sort of starts to beg the question of like, all right, well how are chaperones doing folding, degradation, metastable proteins, complex assembly? And I think to get an idea of how is the chaperone functioning in the cell, I always go back to sort of an old school principle of in protein folding. As the nascent chain comes down its energy landscape, it often will encounter unproductive folding intermediates. And that is where the majority of us, including myself, think about chaperones. That chaperone binding or activity gets us over these energy barriers and leads us to my favorite misnomer, the native protein. And I say misnomer because proteins don't have a singular native state. They always have multiple native states. And those multiple native states are usually indicative of the work a protein does inside a cell. And so we think of there being energy barriers again between those different native states when they're trying to accomplish work. And you might be thinking like, what exactly are you talking about? And my group, when I started my own independent group, I started with what I thought was the easiest one to sort of get a handle on. And that is if we think about transcription factors or DNA binding proteins, we don't think of them as being non-native whether they're free or bound to DNA. These are both native proteins. And yet there's a clear energy barrier between the free and bound state. If you take a purified transcription factor, bind it to DNA, this complex has a half-life of typically on the order of two hours. There's a clear energy barrier, and I say that because inside a live cell, that complex has a half-life of about a half a second. So there's clearly a difference between in vitro and in vivo. Workthrough in my group demonstrated that a small molecular chaperone called P23, it also happens to be a co-chaperone for 90, could accomplish the change in those states. And what you're looking at is DNA anisotropy experiment where transcription factor called MCM1 is bound to the DNA. If we challenge this fluorescent label complex with a non-fluorescent DNA oligo, this thing happens in the anisotropy value, and that's because it's a very happily stable complex, not dissociating at all over an hour period. And yet if we challenge this P23, we quickly drop down to the anisotropy value of the naked DNA. And that drops down sort of on the kinetic values of what we see in live cells. And it turns out that P23 itself regulates about 70% of all yeast transcription factors, which is a very sort of chaperone-esque role inside a cell. And I say that chaperones are generally doing very broad activities inside a cell. All right, but what about HSP90, which is the topic of today? So 90 was originally discovered because it forms stable complexes of metastable proteins, typically signaling proteins, be it kinases or aposteroid receptors. A lot of beautiful work from Dave Toffbill, Pratt, early work, led to a model where HSP90 will bind two clients in conjunction with recent work from Johannes Buchner's group, showing that 70 will deliver those clients along with various co-chaperones to 90. With ATP and additional co-chaperones, we will eventually carry out late-folding steps to lead to an activatable state of those clients. And then when we have client activation, say, hormone addition, it will destabilize this complex and it will return to its cycle. For 90 to do this, it doesn't work alone. I've shown a few co-chaperones. In budding yeast, there are over a dozen co-chaperones. In human cells, there's over 20. It's a very complex system. And what was more intriguing to me is that HSP90 itself is a highly abundant protein. It's about 2% of your protein mass, way outnumbering the number of metastable proteins in your cell. So what are we doing with all that extra HSP90? And we didn't really have a good grip for that. We knew that 90 had a large interactome. So there are 1,400 out of 6,000 open reading frames in yeast that are linked to HSP90. But that linkage wasn't what you would think for a chaperone. It wasn't a physical linkage. We didn't know that these were physically associating. Most of these interactions were based on just genetic interactions between the various open reading frames and 90 inhibition. And they came out of three primary papers all really focusing on the genetic or chemical interactions between 90 and various open reading frames in budding yeast. Now, the genetic interactions have been useful for identifying what does 90 physically interact with? So my group has certainly taken a large advantage of trying to figure out what does 90 do in the cell? What does all that 90 do? What are its clients? One of the things that we focus on were telomere biology, where telomere's linear ends of chromosomes and eukaryotes have to be capped to protect them from degradation. 90 will actually attack this cap, much like P23 attacking transcription factors and dissociate it. But 90 does additional things at the telomere. It actually helps deliver telomerase, the enzyme that will extend the end, and facilitates elongation of that DNA. So it's doing a fair number of things at that DNA. And a lot of it being transient interactions. 90 will work with other things. It will work with chromatory molars. But here, the chaperones are doing different things. Before I invented P23, it kicks things off DNA. Here P23 doesn't kick it off, but actually greases this machine to slide the nucleosome along. And then 90 will then attack risk, and this is from the risk chromatory molar, at the DNA binding subunit to kick it off the DNA. So we were able to use the genetic interactions to establish what does 90 physically interact with, but those two quick little examples was five to ten years of work. We have 1400 things to truck through. Did we really want to keep taking that approach? The answer was obviously no. That was just, it was non-tenable. How can we get at what 90 physically interacts with? Because if we looked of 90 interacting with the capping complex or the chromatory molaring complexes, we wouldn't find it as stably associated. When it does what it does, it does it in a very transient way. So we needed a way to stabilize what 90 physically interacts with. And I was once again inspired by Betty, Craig. So Betty had used a non-natural amino acid called BPA to look at HSP70's interaction with its partners in live cells. And I thought, well, that, I like that cross-linker. Typically I'm not a big fan of cross-linkers, especially with chaperones. I always worry about, like, once I break them open, I'll capture artificial interactions. But here, BPA, we're going to integrate it into the HSP90 structure. We're going to activate it with UV light, and that will create our cross-linker. And we only have the cross-linker when we are activating it with the UV light. And so we're controlling when and where we're going to have a cross-linker. And we took advantage of the crystal structures, our crystal atomic information from various groups, including Dave Agar and Lawrence Pearls, to pick where we're going to place the different moieties of BPA within the HSP90 structure. 90, this is a dimer of 90. Here's one protomer. Here's the other one. N-terminal domain, middle domain and C-domain. And we did try a lot of other spots, but they seemed to interfere with function of 90. These seemed benign, and they seemed like really great spots for us to utilize. And one of the things that drove where we picked, what locations we picked, was we knew they were close to sites of protein-protein interactions, be it co-shaperones or the really limited amount of client interactions of where 90 would bind to a client. The CDK4 kinase was one of the few that we had information on. So we thought we could then take advantage of this BPA screen, incorporate mass spec to answer the following four questions. We could identify, first off, what are all the proteins HSP90 physically interacts and build out a really good solid physical interactome? We could answer a really seemingly simple question that had not been answered. How does HSP90 interact with so many different clients? And that's why we actually were covering the three different surfaces. We didn't know where all, what HSP90 used. There was some good clues that the middle domain was going to be used, but we needed to study it in further depth. And then also I will say I naively thought we'll be able to use this BPA as a tag to identify where 90 interacts with a client. Because BPA is a non-natural amino acid, I thought that would just fall right out of the mass spec data and you'll see later we have some success with it, but it certainly wasn't to the level I was hoping. And then finally, of course, we wanted to discover new pathways that were governed by HSP90 by knowing what it interacts with. We could then figure out what it regulates. And so we started all this with a proof of principle screen with Matthias Meyer at Heidelberg University to figure out exactly what were the samples that we wanted to test, how did we want to go about it. And then eventually we did a large scale screen with Alex Campos at Sanford Burnham Presbyterian. And what we ended up deciding to do is take a wild type HSP90. We do grow it in the presence of BPA, but this doesn't have any integration of BPA. And this would allow us to isolate those stable, the meta-stable proteins that stable interact with HSP90. And then once it contained BPA, we would capture both the stable interactions and the transient interactions. And then, of course, importantly, all these are histagged. We also use a non-histag to screen out what were sort of our background hits on our beads and were interactors. Through this approach, we were able to identify about 1,100 hits roughly half of those are dependent upon BPA. When I say dependent, we either have to have BPA there or we at least enrich over wild type levels. And if you read around the circle, you'll see HSP90 is really interacting. Anything that's going on inside a cell, we're hitting. And unfortunately with my level of curiosity, I like to float all over the cell and try to address where is 90 working within the cell. One of the things I was really happy to see is that it was about split half and half of clients in the cytoplasm and the nucleus. And I say that because it wasn't too long ago that chaperones weren't even believed to have a presence in the nucleus. All right. So first things first, how does 90 interact with 1,100 different proteins? Client interaction surfaces that were known are shown here when we started this work. Tau, glucocorticoid receptor ligand binding domain and CDK4 kinase. The contact residues with HSP90 on one of the protomers primarily in the middle domain, a little bit for Tau off in the end domain and a few client interactions down in the C domain. We thought, well, what we chose should hit most of those surfaces. But the prevalence really seemed to be on the middle domain. And when we look and ask where do we, what residues are capturing what clients, all the surfaces seem to be used. There is a small bias for the middle domains where they have about 350 hits, but the end domain is 225 and the C domains are around 200. And notice, you know, in this quasi-Venn diagram there's a lot of sort of co-usage of the different surfaces. And I think this is how 90 achieves such a broad binding capacity with so many different clients. It's basically a combinatorial use of binding surfaces. We have clients that are dependent only on the end domain, middle domain, or C domain. We have clients that will use both the end and middle, and C, or even the end and the C. We have some clients that use all three domains. And then, of course, this is all within a single protomer. We have ones that will cross the protomers. And that is how you have one protein that is able to reach out, or at least in part reach out to almost 20% of all the proteins inside a cell. All right. Now, let me move on to what I thought was going to be a simple straightforward question. We have this non-naturamino acid. It's a chemical cross-link to its client. This is going to get us, what is 90 recognizing on a client to make it a client? And you might think, oh, we already know what chaperones recognize. They recognize short hydrophobic peptides. Yes, for HSB 70. And that's the only chaperone we actually know the rules for. We had no idea how 90 was recognizing a client, and we thought we could get there with that. And I totally convinced my student that this was going to be easy and doable. These were our non-naturamino tags coming out of 90 where we put the different BPAs. And when she started looking, nothing. And we had to search and go through so many different computational approaches until she hit upon MIROX. And John B. Coley is a grad student doing this, and she just dug in until she found a program that could at least recognize the cross-link peptides. And that got us basically the needle in the haystack because mass spec data, there are just hundreds of thousands of peptides. And finding the ones that represented cross-link peptides was really difficult, but it didn't get us... We had a lot of false positives in that, a lot of sort of nonsense information out of that. She figured out we had to add two more criteria. And then we could finally, with confidence, identify BPA-containing cross-link peptides. STI-1 is a classic co-shapron for HSP-90, and by adding in these two, we could read across the sequence in STI-1 across to an X that's BPA, and then more sequence density from HSP-90 itself. STI was a really nice one to hit because we knew it's one of the best well-behaved co-shaprons in terms of forming stable interactions with 90. I mentioned that 90 kicks risk chromatary myelin complex off the DNA. It does so functionally through risk-3 and we capture risk-3. So we're gaining sort of confidence like, okay, these hits are actually real. We also hit B&I-1. B&I-1 regulates actin dynamics. I'll get to that story in a little bit. It's one of the clues that help us how chromosomes are moved in interface cells. To really sort of drive it home for us, were we hitting or were we getting valid hits? If we look at where we placed in the cross-linkers in STI-1, I mentioned as being a really good co-shapron, when we dock it on with the known sort of atomic structural information, wherever the yellow is where we get a cross-link in the structure of the co-shapron, and this is the red is where we have a BPA moiety. And so these two moieties under here cross-link to those two, there wasn't any structural information for STI at the tail, but we did capture the tail with a BPA there. And so all three hits made sense where the co-shaprons should interact. And we hit a lot of different co-shaprons and chaperones, really convincing us and hitting some chaperones we didn't really expect. We actually, CCT is the trick, Gro-Yell homolog and Budding East. We didn't expect to hit them all. 90 really does look like it's a central hub for chaperone interactions. And when we dock any of the ones where we had that information, we could always make good sense of that's why we captured them, where we captured them. And so our confidence that this looks like this is working and capturing physiologically relevant things was making us happy. Using Javi's sort of combination of Mirox and her additional criteria, we have 1100 hits, we didn't get all the BPA peptides, but we did get out to 180, well over 180. And that I felt was enough for us to start figuring out at least some of the rules for how HSP90 might interact with either co-shaperone, chaperones, or even clients. 24 of those hits are coming from co-shaperones and chaperones. When we ran for sequence conservations or any commonality, there wasn't any. So it wasn't simple like HSP70, it was recognizing a short hydrophobic peptide. But along the way we realized like, oh, intrinsically disordered regions, if you're a chaperone or co-shaperone, you're not so likely to interact with 90 through your intrinsically disordered region. But if you're a client, you sure are. So there's a preponderance of client interactions at intrinsically disordered regions. And some of that is shown here because I have to say the field in intrinsically disordered regions can be a little rough. It's still a developing field. If you run, there are many different software prediction programs for where IDRs are found. We were fairly dissatisfied by using one. So we typically run about eight. And this is where we got a mass spec hit for not nine. And all of them are predicting it intrinsically disordered region. The nice part about not nine is we actually have crystal structural information. And sure enough, it's a disordered region. And usually what we see is 90 is right next to not just an intrinsically disordered region, but it's usually right next to a structural motif like an alpha helix or a beta sheet. And we see that for wherever we had structural information, that is usually what we see except for PRI1 where that is truly just within an IDR itself. All right. So that was actually, to us, fairly huge because it's telling us how 90 is now recognizing this client. And for us, it was making a nice sort of contribution to the chaperone biology. 90 is typically thought of as a protein that interacts with seemingly native proteins. And the question always was, well, how's that happening? If it's recognizing IDRs, that's a really nice substrate for chaperones to bind to. And also may explain why so many proteins contain intrinsically disordered regions. Prediction of software usually runs about 30% of all proteins have sizeable IDRs. So are these sort of co-evolving so 90 can regulate those native proteins and sort of feedback on each other? Moving past that, we wanted to really drive home our confidence and our hits meaningful. And right away, I was pretty pleased that about two years ago, we published a paper using DNA seek to show that DNA footprints when you inhibit 90 are radically modified. But we didn't have a physical interaction with, no one had shown physical interactions with all these transcription factors. Our network of transcription factors is rather large. Almost all of the hits in the transcription area are simply DNA binding proteins. The work we had done before is within a DNA hypersensitivity site. If we look at the footprints in wild type cells, we see lots and lots of footprints. Some of them have consensus motifs. If we inactivate 90 for just 15 minutes, it's a radical change in those footprints. We use that information to look across a genome at all footprints. And at the 15 minute time point, we see typically a dramatic loss in the number of footprints. And that changes when we get out to six hours. It's a totally different sort of event inside the cell. And what we see is later on, we're basically just losing those transcription factors. And some of that came out in the biochemistry. If we look at steady state levels of transcription factors, late, there's a loss of falling inactivation of HSP90. In this case, we're using a temperature-sensitive allele of HSP90 called G170D, isolated by Sue Lindquist lab. Or if we use a small molecule inhibitor called radicicol at the six-hour time point, there's a loss of any transcription factors. But at 15 minutes, there isn't any loss. And yet we lost DNA footprints. And so it was a bit of a quagmire, like, well, they're not being lost, yet they're not binding DNA. And it turns out, 90s require for a late folding step for many transcription factors. Within the transcription field, people often just go with, well, I can't express my transcription factor in E. coli because the recombinant proteins are notoriously inactive. We went ahead and purified a few, RAP1, CBF, ABF, and MET31 in E. coli. And when it's just a transcription factor, there isn't any DNA binding activity. Because that transcription factor has never seen a eukaryotic version of 90. If we simply expose it to yeast HSP90, we gain DNA binding activity. And so we think that's the issue. 90 was required to physically fold these proteins. But if we look and ask, can we see 90 associated with any of these, we never could. And the BPA screen was really nice for that because it captured all these transient interactions and told us that 90 is actually folding all these proteins. Not all transcription factors require 90. RAP1 was a really nice negative control. All right. A couple of years ago, we published a really nice paper, I think, anyways, on chromosome motion. And we didn't mention in that work what really facilitated us to delineate the chromosome motion pathway was this BPA interactome. That was classically a really difficult and challenging area. People were trying to figure out how do chromosomes move within interface cells. We had a fairly large contingent of players thought to be involved in chromosome motion. And chromosomes, of course, I'm sure most of you already know, have a really high level organization inside of cells. They are organized not only by chromosome territories where each chromosome is placed in the same spot after every cell division. And they're also organized, in this case this is nomenclature related to high C experiments of compartment A and B. You can think of compartment A and B as heterochromatin and uchromatin. Always well positioned after every cell division going back to the same location. And the question became how does that happen? How is such unique precision achieved? And one of the things that's, you know, to bear in mind are chromosomes are huge and heavy polymers within the cell. So it's not a little task to position these long polymers. For us to address does 90 have a role in chromosome reorganization? We took advantage of a system developed by Jason Brinkner at Northwestern University. He started this when he was a postdoc with Peter Walter. He's using the eno-1 gene locus that he can activate just simply starving cells for anocetal. When you starve for anocetal, you go from the center where you're inactive to the nuclear periphery really associated with the nuclear pore where you are then active. And he's incorporated a lac-i response element array where lac-i fusigee-f-p will bind. This is where that movie was I came up with in the beginning. And we can starve and move the lac-i over to the nuclear periphery. And it always moves to the same nuclear pore, which is really fascinating. It's a really precise motion mechanism. The downside of this system is, you know, yeast is a relatively small nucleus. This is a rather large spot. So the relative enrichment never looks great, but it's a very consistent two-fold enrichment. It's enough to follow. And one of the nice parts of this system is if we impair different chaperones. Here we're impairing P23. It's that chaperone that kicks transcription factors off the DNA. We don't get any movement. We saw the same thing with HSP90, except now we're inhibiting with a small molecule inhibitor. We don't get any movement when the inhibitor is there. But if we take those live cells and simply wash it out, it moves. So 90 is actively contributing to the motion of that DNA locus. I'm not going to go too deep into this. This was a rather complicated story. And I'm going to just tell a couple of sort of points highlighting 90 and highlighting what we use the sort of interactome to dissect out how this pathway works. But the E1 locus has different transcription factors, you know, 2 and 4, that will recruit in chromatary molars and will recruit in RNA polymerase. And then putt3 is a transcription factor. Jason had shown binds a very unique DNA element called the DNA zip code. And that is the element that tells this locus to go from the middle of nucleus to that specific nuclear pore. We know, you know, 2-4 binds to these and putt3 binds to, well, you know, 2-4 recruits in chromatary molars. This was information that was known before. You know, 80 is going to have a really important role in motion. And then putt3, we found recruits in a myosin motor. And that myosin motor is one of the substrates for HSP90, and so is you know, 80. And both of those fell out of our interactome. HSP90 is going to work with both P23 and another Coshack around for she4 to recruit in our nucleate actin at this site. And the myosin's then going to join the actin. It's going to be held there in part by the chromatary molar. And what 90 does to facilitate this pathway, if we do what's called an actin spin-down assay, where we're just adding this eno80 interact with actin, and it does, it will spin down with actin into the pellet, because eno80 contains actin-related protein subunits that interact, that are known to interact with actin, although the complex itself wasn't known before this. 90 really directs it towards whether it's interacting or not interacting with actin. And it changes its behavior quite a bit, along with P23 will contribute to get a very different behavior pattern for eno80 interacting with actin. And so that was one of the bits of this story and how we were able to show it. What I found really intriguing is it's not just a stable interaction. And we found that out by going to a single molecule approach where we were trying to recapitulate the chromosome motion with purified components. And so what you're looking at here, we did single molecule work with Paul Selbin's group. Actin is in purple and in green are nucleosomes. In this movie, it's just the mixture of those two. And not surprisingly, you don't see nucleosomes interacting with actin. We didn't really expect to, it was really more of a negative control. If we add eno80 with labeled histone, there's a little bit more interaction, but not much. If we then add in 90 to that same one, now you see a lot of interaction of that labeled nucleosome bound to the remodeler, but it's really dynamic. It's constantly on and off, but it's always returning to the actin. This we think will become a very important characteristic to the chromosome motion. And I'll talk about that in just a sec. So that's where 90's role with eno80 is coming into play to favor interactions with actin. The other one is myO3. We would have never really hit upon myO3 if we hadn't already isolated it in our BPA interactome. MyO3 is a myosin motor. It's to move cargo along actin. In this assay, this is called an actin gliding assay. If we've got our glass surface, we pin the myosin onto the glass surface and then we put down fluorescently labeled actin. If the motor's moving, the actin will slide across that glass surface, which is why it's called an actin gliding assay. And we isolated myosin from yeast. We put in the actin gliding assay and we were severely disappointed. Nothing happened. So it seemed like we just destroyed our myosin motor potentially, but if we put in 90 and she4, the motor's perfectly fine. It's just always actively dependent upon 90 being there. And we think that is why we hit myosin. And the way this system works is we have short actin polymers, and I'm not showing that data to you today, but it's a highly dynamic, polymerizing, depolymerizing actin network in the nucleus. It's short actin polymers. That transient interaction of 80 with the actin is important to move along that actin polymer as the myosin power stroke goes, pulling it over to that nuclear pore. What we're yet to identify is how that locus knows which pore to go to and that stuff that we're working on now. So those are things that either were published before we did the network, things that facilitated us publishing work, but that's already out. I wanted to tell you guys something new in terms of what 90 is doing. And the one I'm going to focus on today is in translation. 70 has a big role with ribosomes, nascent chains. It was a known nascent chain binding protein, and the data's not out there because it was all negative data, but people looked hard. Does 90 have a role with nascent chains? It doesn't. But it does have seemingly important role with the translating ribosome. And that's where I'd like to go. There was a few publications that said 90 had a role and with the elongating ribosome, I never like to chase what somebody else has done, although I know there's a lot of important things in there for validation. I wanted to chase something we didn't know anything about. And a lot of our hits were for initiation factors. We didn't know anything about 90's role with initiating ribosome, so I was excited. The student got excited and we asked, okay, what goes on when a ribosome initiates? Why might we get hits there? A lot of things go on. There's a lot of sort of reasons for protein. Protein protein dynamics, protein RNA dynamics, but we have the 40S that's going to form primarily with what's called the multi-factor complex. And this is going to then interact with the RNA capping proteins to form an initial complex. And then this structure is going to move along the RNA until we get to an AUG. The multi-factor complex is particularly important for finding that AUG. Once we get to the AUG, we'll have some more protein remodeling events. We'll have the 60S will join, we'll have the 80S, and off we go for elongation. All right, so there's a lot going on. What got us excited is that our BPA Interactum had a lot of hits within these complexes, and I thought, okay, cool. This looks good. Let's move forward sort of in a cautious way and ask, well, is 90 associated with the 40S or the monosomes? And so we just did your sort of standard sucrose gradient as Betty would have predicted. 70 is found both with all the 4060, the monosome, and the polysome. HSP90 is primarily, though, with the 40S, a little bit with 60 monosome, and then it teeters off pretty quickly. But it's there. And that gave us some, you know, okay, let's keep going. Let's see, is there a role and an impact of 90 on finding the AUG? We took sort of a classic approach where we put in plazins that are going to express runeloluciferase with a canonical AUG or non-canonical start codon with Fireflyuciferase. So this will get barely used in wild-type cells. But when we inactivate 90, we're going to use a G170T temperature-sensitive allele and shift to 37 degrees. And inactivate 90 and just ask, how does the usage of that non-canonical go up or down? It goes up. So we start that ribosome cares less and less about starting at a proper AUG. Great. All right, let's move forward into, you know, a more high-throughput ribosome profiling approach. And because there's a lot of other non-canonical start sites that could be used. So we joined forces with Natalia as a postdoc with Judith Freeman group. And we did the ribosome profiling. They did the bioinformatics. The first time in my career, usually we do the bioinformatics. Ribosome profiling was a whole new nut for us. That was not so easy to crack, especially at the first question we wanted to ask. What is the ribosome profiling over the AUG? So if we do a meta-analysis just in wild-type cells, here's the 5' UTR. All the peaks are where the ribosome is pausing. And of course you have the highest occupancy over the AUG. And very nicely, when we do G170D, after 15 minutes we lose that peak. And so when you inactivate 90, the ribosome really doesn't care so much anymore about finding an AUG. It's there on the RNA. It will lock on to the RNA and it will start making peptides. And we thought, well, who are the targets that are sort of contributing to this loss of finding the AUG? The multifactor complex is one that I had mentioned. This is associated with the 4DS and it's important for finding the AUG. Jami's going to use a sucrose cushion to look at monosomes and the 4D and 6DS. What components are there? And in G170D, both without temperature rising or with the temperature rising, peer T1, which is EIF3B, homolog in budding yeast, is always clipped. So it's not even dependent. And I have to say G170D is not the happiest mutant. That non-happy mutant already has peer T1 clipped. And if we inhibit it in a different way with a small molecule inhibitor, we actually just see degradation of peer T1. And so we think that's one of the contributors to why it's no longer caring so much about finding a start site. But of course, there begs the question of, well, is translation even happening when you inhibit 90? So we label nascent chains. And much to our surprise, 15 minutes at 37 degrees where G170D gets inactivated, yeah, we're making nascent chains. If anything, we're making more nascent chains. We looked in our ribosome profiling data and I'll take you a bit through this. This is the proper frame where you see in a heat map highest level of occupancy and wild type. These are read lengths. The ribosome does have different size footprints along that RNA. They can tell you a little bit about what's going on. But for in the coding region, in general, wild type is in frame as expected. And in G170D, after just 15 minutes, it's no longer in frame. That's across the entire coding region. It doesn't matter what footprint it is. And so that told us, oh, this is a significant problem that may actually correlate to some old clinical data. And it's not even terribly old clinical data. About five years ago, five to 10 years ago, there were 60 clinical trials targeting HSP90. Those clinical trials all failed with the same sort of phenotype. All the patients exhibited a heat shock response. You can't inhibit a heat shock protein if you're making more heat shock protein, which explains why it failed. But it wasn't quite clear why it failed. At the time of that, there was one paper that suggested HSP90 kept heat shock factor 1. It's the main transcription factor for delivering a heat shock response to cells in a quiescent state. And that heat stress would cause release and HSF would go to the nucleus and up-activate genes. And so by inhibiting 90, they thought, well, maybe that's why we get a heat shock response. But the funny part is, no one really believed that paper. And I thought, well, is our data telling us why we would get a heat shock response? If we're making nonsense peptides, that essentially would mock a heat shock response. If that's true, then we should be able to sort of tweeze that out. And we took a shot at tweezing that out with the following experiment. Where we used cyclohexamide in budding yeast as a way to inhibit translation. And we're doing heat shocks after 15, 30, or 60 minutes. These are our controls. Heat shock, we're producing a lot of the yeast HSB70, allele, SSA4. If we treat with cyclohexamide, the only thing we do is exacerbate that heat shock response. Not surprising. You want to produce HSB70 to turn off the heat shock response. We're just keeping it on because we can't translate anything. The stress response can be brought on by more than just heat stress. You can add a proline analog, azide. And azide will get a heat shock response, but since it's a proline analog, when you add cyclohexamide, you don't get a heat shock response because you never incorporate that proline analog, so there's no sort of mal-folded proteins to get you a heat shock response. If we do that same thing with radicicol, we see the same thing. It's very translation-dependent. If we inhibit translation, we don't get a heat shock response. And we now think we're in a good position to sort of go back and ask, oh, in those clinical trials that failed, is there a way we can increase the fidelity of the ribosome for its AUG, and can we return to those clinical trials? All right. And I'd like to thank John B. Coley, the grad student who did the majority of the work. She's carried out all the BPA. She's carried out the translation work. I didn't have a chance to talk about Audrey's work. She looks at the role of the nucleoskeleton and genome organization. Anna Mankiewicz is another student who looks at the role of 90 with tail anchor proteins, where we think 90 actually contributes to insertion of tail anchor proteins in the membrane. Nietzsche does aging studies. Ancien Zlada did the former work on chromosomal reorganization or the transcription network. Former students that contributed to various areas of all of his work. Alex has been just a great collaborator in the mass spec work. Bill Breyer was our collaborator with Acton. Judith is with the ribosome profile and Matias helped us set up the right conditions for carrying out the BPA screen. Craig is a long-term collaborator on chromatory modeling, and the Brinkner's helped us immensely with one relocalization. And I need to of course thank you guys for your patience during my talk. Thank you kindly. Thank you. Can you hear me? Yeah. So I'll open it up for questions. Does anyone have any questions? Okay. So you implied that the effect on translation was due to direct interactions between HSP90 and the initiation factors rather than like an indirect effect with like the other 1200 proteins about that also interact with HSP90 and I was wondering if you've is there a way to disentangle that issue or even have you tried looking at initiation from an iris to see if there are changes for CAP independent translation? So we're now moving towards a more sort of we're certainly moving towards an in vitro system to get at that because it's always a concern for any of the chaperone work that we do we typically try to take it all the way to a purified system I know I won't be able to do that here but at least take it to at least like a retic system and ask where we're not going to have so many sort of competing influences. One of the things that sort of brings a little bit of doubt at least in my mind that it's not some indirect effect is we're inactivating after just it's a 15 minute heat treatment takes 5 minutes and activates a 90 so whatever is happening is going to happen within 10 minutes that certainly won't rule out kinases it's still one of the possibilities but that's it's the best I think that we have right now and I will feel better about it and hopefully you will too once we move to an in vitro system and ask okay can we see the same things? Does the oxygen system work with the HSP90 or is it too concentrated to decorate it? I think the oxygen system probably could work but the small molecule inhibitors are good specific inhibitors and when we can use different 90 inhibitors the one that we are using does have at least two other weak targets but we can use C-terminal inhibitors and also sort of tweeze it out that way the temperature is sensitive to 90 and the small molecule inhibitors do attack it in two different ways and say are we still seeing that same impact? Thanks can I ask you to elaborate a little bit more on HSP90 interacting with IDRs IDRs are all the rage now forming these liquid phase separation things in vitro or any in vivo system that you will examine whether HSP90 modulates the type of liquid-liquid phase separation that you see? I don't know of anyone who's looked at 90 per se because there wasn't any real reason to look before but probably half a dozen of other chaperones have been looked at and most of them do modulate that phase separation do they promote more or do they promote more LLPSs usually it's less so that's I would say the general take on it okay can I follow up on the IDR question so some IDRs are enriched in certain types chemical flavors of residues and I'm wondering if when you do your analysis do you see any no selectivity of sequence that was the thing we focused on the most in the beginning nothing came out of it even if we sort of looked at it in different ways of just categorizing and in classes of amino acid and there's no sequence specificity to it the only thing that we found was the presence of the IDR for the clients okay question about the I-080 and the interaction of the nucleus cells with the actin that was it molecule stuff so the I-080 I think it has some actin related protein subunits in it and I'm just wondering like you know which subunits of I-080 complex are responsible for that all the ARPs contribute sort of mildly to the overall affinity that was a bit of a disappointing answer you can't you can't get rid of all the ARPs in the I-080 but you can get one at a time and the overall affinity went down just a little bit depending on which ARP we got rid of but it's not particularly satisfying but I suppose biologically it might make sense especially given the way it interacts with actin in such a dynamic fashion having multiple sites that all contribute would make that sort of multi-valent behavior or readily occur Thanks Peter I think there was one question before you Dave Hi, great talk my question was also about IDRs so you said you have looked at the amino acid composition and I am really curious whether there is any hydrophobicity versus hydrophilicity pattern that you could identify especially because you are saying that many of the clients are binding DNA and there may be burial of non-polar surface from the IDRs perhaps so I thought that there could be a way perhaps for HSD the evidence of the whole interactome are not a large chunk of that interactome so the transcription pie of the pie chart that was almost all transcription factors that bind DNA but that is maybe I don't know it's not even 10% of the entire interactome and so there but there those IDRs aren't even going to be near the DVD either a few of them are but again we were left with no sequence sort of common signatures in terms of amino acid sequences where we found 90 binding not even a broad character like hydrophobic versus hydrophilic yeah that's what I was I guess poorly conveying we broke it down to that not even specific amino acids but let's just go with negative you know nope we got nowhere that was very disappointing in the beginning I don't remember I'm trying to remember how we even tripped upon to look at IDRs actually I believe John V was looking at okay let me take a look at crystal structures what do I see there and that's when she was finding like oh wait it's where 90 cracks there isn't actually structural information to go with and then we went back and asked what about IDR sort of structural prediction software I see and so within that context can you say whether the function of HSP90 among all of those 1100 substrates is mostly to help final stages of folding or maybe hiding binding surfaces or something else oh I think that's a mix of lots of different things just in the limited time that you know we've spent asking functionally what does 90 contribute to a client we see all sorts of different things you know for today I just happened to focus on DNA binding events and they're depending on the client 90 might kick it off the DNA or 90 might put it on the DNA for most of transcription factors it's putting it on the DNA and that's because it's partner p23 is kicking it off the DNA but then for chromatory modeling 90 off the DNA so it we don't have a rule for what will 90 do to a DNA binding protein let alone you know in a lot of ways what 90 does at Eno80 in actin is very comparable to what it does with some transcription factors in DNA actin is just a polymer and it's helping to modulate whether it is or isn't interacting with actin it helps more to facilitate or disrupt interactions rather than help folding would that be accurate? I think there are clients where it is certainly helping we found that with the transcription factors that's a very late folding step require for DNA binding I didn't show you today but by limited proteolysis we see conformational changes in those factors so that we believe is a folding one but for a lot of things it's helping to promote protein protein dynamics and protein nucleic acid dynamics so is HSP90 ATP dependent or not? HSP90 has an ATPase it's a very sad ATPase it hydrolyzes ATP about once every 15 minutes and most of that ATP in the hydrolysis of it is really to control the conformational sort of what Conformation 90 is so I showed you it formed the dimer comes together and I have an open surface this is nucleotide free this is ATP this is ADP it doesn't look like the chemical energy of the hydrolysis goes to the client but rather to the conformational change within 90 and how about P23? P23 has no ATPase activity so when it what it does to all of its clients it's just through binding energy and so how do you see that being dynamic I guess how does it change states and do you see any cold sensitive mutants in these proteins that get stuck? so P23 the deletion of P23 is non-essential in budding yeast is essential in mouse but the yeast are cold sensitive P23 does not directly when it interacts when it kicks something off the DNA it doesn't interact with the DNA binding domain it always interacts with an adjacent domain that is often known to allosterically alter the DNA binding domain a good example that was with steroid hormone receptors the ligand binding domain and steroid receptors is known to be able to shut down the DNA binding domain P23 requires the ligand binding domain to kick it off the DNA in an odd sense it does that not just with DNA binding proteins but GCN5 which is a sonocidal transferase has an odd two interaction domain next to the hat domain that is used to control hat activity that's where P23 regulates GCN5 through that domain it's always some sort of allosteric interaction never with the activity that we're directly following it's always using a sort of regulatory domain that's used elsewhere in the biology of that client but then to recycle it it's just presumably using thermal energy when it recycles once P23 kicks it off we can see P23 kicking off a transcription factor off the DNA we can't see it bound to the transcription factor once it's free in the DNA we have one nice example MCM1 and I showed the data for P23 kicking MCM1 off the DNA if we remove P23 MCM1 will not rebind to the DNA but if we give it 90 90 will convert that conformation into a DNA binding form and now it can happily bind to the DNA but 90 won't join it to the DNA bound complex it's just changing its conformation and allowing it to now bind to the DNA so that's the only one where we have this sort of a stable state of it's free but not DNA binding component 90 gets it to the competent state P23 will kick it back off the DNA and leave it in a non-DNA binding it's a competent way does HSP90's ATPase change when it's associated with clients its rate change it can go up or down as frustrating as that is so some clients will get the ATPase the usual ATPase will go up with a client but there are certain clients that will make the ATPase go down Presumably the ATP binding and ATPase activity of HSP90 are essential despite it being a function given it there was an initial publication that said ATPase was essential and Johannes Buchner's group a couple of years ago published another a follow-up paper that said the ATPase is not essential ATP binding is essential but not its ability to hydrolyze it those yeast are very unhappy and slow growers but life still goes on even if it can't hydrolyze ATPase so something as fundamental as whether or not this thing hydrolyzes ATPase is still somewhat controversial as to what it does I think the people that published that it was required agreed with Buchner and said yes its ATP binding is more important than hydrolysis can live if they can't hydrolyze ATP but it is an extreme slow grower so life isn't happy without that did you have another question I'm just wondering instead of thinking about HSP90 may have evolved to interact with a certain set of proteins maybe it evolved to just not interact with another set of proteins and so like if you look at the things that don't cross-link does that give you any insight we haven't done that to any extensive degree it's like the negative of the picture I think that's a great angle to sort of go at it and I think we might we sort of are playing with that idea with telanchor proteins which I didn't talk about today because we think for telanchor biology get345 is this complex that delivers telanchor proteins to the translocase get12 90 delivers has a bias set that it delivers but if you get rid of get3 it'll deliver those if you get rid of 90 get3 will deliver night so they can cross over but they certainly have preferences so I guess in some ways we're pursuing that but I don't have for you yet because that's relatively we're like halfway through that project I guess okay well let's thank Brian for a stimulating talk lots of questions so that was great yay hmm that's good so I sent you that oh thank you I have my name with the check okay I got bond or so bond well thank you