VIDEO: Kitchen Table Talk: Josh Riback on Coordinating Bodies: How Disparate Leukemia Mutations Converge on a Shared Oncogenic Condensate

Condensates.com welcomed Josh Riback, Assistant Professor, Department of Molecular and Cellular Biology, Baylor College of Medicine to deliver our 43rd Kitchen Table Talk, on June 17, 2026.

Title: “Coordinating Bodies: How Disparate Leukemia Mutations Converge on a Shared Oncogenic Condensate”

Abstract: Acute myeloid leukemias driven by NPM1 mutations, nucleoporin fusions, and KMT2A rearrangements share common transcriptional programs and clinical targets, yet the mechanistic basis for this convergence has remained unclear. In this talk, Josh will present his lab’s discovery that mutant NPM1 (NPM1c) forms a previously unrecognized nuclear phase-separated condensate, which we term coordinating body, or C-body. Using quantitative live-cell microscopy, mutagenesis, and pharmacological perturbations across human cell lines, mouse models, and primary patient samples, we show that C-bodies selectively enrich a network of leukemogenic proteins, including XPO1, NUP98, KMT2A, and MENIN, at HOXA/MEIS1 chromatin. Critically, C-body formation is required to maintain leukemic gene expression, prevent differentiation, and support leukemic expansion in vivo; cytoplasmic NPM1c localization alone is not sufficient. Strikingly, nucleoporin and KMT2A oncofusion proteins found in other leukemia subtypes independently drive condensates that recruit the same interaction network and are biophysically indistinguishable from NPM1c-driven C-bodies. Together, these findings suggest that distinct leukemia mutations converge on a shared mesoscale structure that organizes oncogenic transcriptional programs and may represent a common therapeutic vulnerability.

Click here or on the video to view the engaging discussion and read the transcript below.

TRANSCRIPT

Diana Mitrea: Thank you, everyone, for being here today. This is Kitchen Table Talk 43. So, very excited about that. Welcome to everyone joining us from across the globe. My name is Diana Mitrea, your host, and I’m joining you from Dewpoint’s Nucleolus conference room here in sunny Boston. As a reminder for those of you who usually come to this talk and a notice for the ones who are first-timers, this talk is a live event, and it will be recorded. Feel free to turn on your cameras and follow along, but please no on-camera shenanigans. 

 We would prefer to keep the questions to the end of the talk, so feel free to type your question in the chat at any time. During the Q&A, I will call your name, ask you to unmute your microphone, and at that point, you can ask your question directly. If, for whatever reason, you cannot turn on your microphone, let me know and I’ll ask the question for you. The event will be posted on Condensates.com within a week. 

 So, without further ado, it is my great pleasure to introduce today’s speaker, Josh Riback. Josh earned his BA in Biophysics with honors from Johns Hopkins University, and his PhD in Biophysical Sciences from the University of Chicago, under the guidance of Alan Drummond and Tobin Sosnik, where he made his foray into the biomolecular condensates field with a cell paper identifying stress-triggered phase separation as an adaptive, evolutionarily tuned response in yeast. 

 He continued shaping our understanding of the molecular and biophysical mechanisms understanding underlying condensate regulation in live cells as a postdoctoral fellow with Cliff Brangwynne at Princeton University. And, actually, that is when I met Josh during his postdoc, when we collaborated on a paper describing how nucleolar composition adapts to changes in protein expression and the thermodynamics behind this adaptation. So, it is really fitting that we’re in the Nucleolus room today. Josh started his independent investigator career in 2021, when he joined Baylor College of Medicine as an assistant professor. Throughout his career, Josh’s impact has been recognized with an impressive number of awards and distinctions, including a Leukemia Research Foundation New Investigator Award, a Searle Scholar Award, and an APSUO Rosalind Franklin Young Investigator Award. 

 Today, Josh will talk about his exciting discovery and characterization of coordinating bodies, which are a new class of disease-driving condensates. But I won’t steal your thunder, so Josh, take it away. 

 Josh Riback: Perfect. And thank you so much for that introduction, Diana, and thanks for the whole Dewpoint Kitchen Table Talk team, for this, unusually, but hopefully, I’m not going to jinx it, smooth set up and for the invite. I’m excited to tell you today about my lab’s new discovery of this new kind of condensate that results from multiple leukemia mutations, and also how some of the rules have impacted our understanding of condensates. And so, just to start, I thought I’d first do an overlay of the talk. I think there was also an option to do questions throughout, so I included it, but it could also be at the end, depending, and so for time, I’ll just jump into the introduction, where I’m going to start with my lab’s broad interests and then funnel into the key parts that led to the story that I’m going to tell you about today. 

 And so, my lab, like the room that Diana’s in, is really excited about the nucleolus. The nucleolus is the location of ribosome biogenesis, but in a cell, here’s how it actually looks. On the top left, you can see a picture of a nucleus, where the most abundant, nucleolar protein that you’ll all know way too much about by the end, is endogenously fluorescently tagged with mCherry. And then it’s also shown in fire mode, which really, for condensates, lets one really see the dynamic range of localization patterns. And then the scale bar is 2 microns. And so my lab’s really excited about how compartments like the nucleolus do their functions. And of course, the nucleolus, as I’m sure all who signed up for this are familiar with, isn’t the only condensate, and even within the nucleus, looking at chromatin, you can easily see how many other condensates are squeezing their way in between histones and pushing the chromatin away. Now, of course, not all condensates are not pushing directly on chromatin, and the one that’s going to be most relevant for the talk today is going to be these aberrant condensates, especially the ones that people have seen with nucleoporins, and that seem to recruit core regulators and drive cancer. 

 What’s really the excitement behind this and differs from sort of the nuclear body field is this notion that even though these systems are really complicated in the cell, that they can form like oil and water, but with much higher complexity. To summarize, my lab’s really excited about that concept, and how we really shape that into the rich biochemical functions that condensates have, both in normal physiology, but also in disease.  

 A lot of what we’ve learned has been from reconstituting simplified condensates in the test tube, where we’ve learned about things like weak interactions between intrinsically disordered regions and various substrate binding domain and then also the extension of multivalency and network interactions, like the PRM and the SH3 multivalent system. A critical part of this is the essence of phase separation, which is the idea that you can simplify your system in terms of a phase boundary or a phase diagram. At least for me and for my lab, a really critical part of this is this notion that unlike the membrane-bound organelles that have a strict boundary, condensates allow for steady-state partitioning, and that can be a good proxy for learning about the thermodynamics and what’s driving things to the condensate, or the condensate to form itself. 

 But while reconstituted systems have taught us a lot and have really been great for these basic principles, their simplicity really makes mapping them to complicated biochemical processes that happen within cells, really difficult. On the other side, for condensates and cells, their complexity that really holds that information also makes them difficult to perform measurements with the same quantitative rigor and depth of understanding. And really what drives my lab and my lab’s approach, is the hypothesis that there’s additional complexity that may arise in cells and under these more complex systems. For this crowd, I thought it’d be best to summarize this in two ways, and so I think that the emergent biophysics that could exist may yield additional principles of how cells can have new modes of biochemical control, but also, in the context of disease, may yield insights that sharpen how condensates are changed, or how aberrant condensates are formed, and that may provide key foundation for therapeutic insights. 

 And so how do we do this? My lab is really just focused on this concept of pushing the forefront in trying to understand condensate principles by directly measuring everything in cells. So, we have a very well-calibrated microscope. I actually just aligned it yesterday, after a year, because we take daily power meter measurements, so it’s sort of like our physical chemist’s calorimeter, and then we turn that into cellular measurements, and then interplay with quantitative microscopy, mostly inspired from physical chemistry concepts, and then through perturbations and mutations, or using therapeutic agents, such as drugs. We try to build models and really try to gain some mechanistic or therapeutic insights. The core elements of my lab are really this notion of building new tools and new models to understand condensate form and function of physiology and disease. 

 And so, today, I’m going to talk about, in particular, one new model that ended up being much bigger than we initially thought it would be, and that was spearheaded by a student, Gandhar, who is a recently graduated MD-PhD student. He recently started his residency at Harvard, and MGB, and he was a joint mentee between my lab and Peggy Goodell’s lab, and, additionally, key to this project, was Elmira, a postdoc in Peggy’s lab.  

 Now that I’ve told you about my biophysical interests, I’m going to briefly mention the background, for the focus of what’s going to become this new condensate, which is hematopoiesis and acute myeloid leukemia. Hematopoiesis is basically the generation of new blood from blood marrow. Blood stem cells undergo a standard process in the marrow into our blood, where the stem cells go from progenitor cells to precursor cells, and finally myeloid and lymphoid cells. Now, in the context of acute myeloid leukemia, that doesn’t quite work as well, and that’s because certain cells, due to the mutations, get stuck in the progenitor-like phase, and in the case of acute myeloid leukemia, they get stuck in this myeloid leukemic state. And so, AML is particularly a big clinical burden. It’s about 20,000 new cases per year in the U.S, and it doesn’t have a really great 5-year survival rate. 

 And so that makes it of interest, and additionally, from the basic cancer biology side, the particular class of AMLs, sometimes called high HoxA9 AMLs, have really focused on a few sets of leukemia mutations that are very different, but seem to act effectively similar in terms of the downstream cellular and therapeutic properties. And so, notably, it’s NPM1, so there’s nucleolar protein mutations, along with nucleoporin, right here, the R is rearranged, or nucleoporin oncofusions, and also KMT2A, where some of you may be familiar with it as MLL rearrangements or oncofusions. What’s really surprising about these is that they seem to activate a very common gene profile that sort of keeps them in this leukemic state. Additionally, from the clinical side, two sets of inhibitors have really shown some promise in all of these contexts. And so, there’s this broader idea that these few things may relate to each other. Initially, we had this idea that there’s some common mechanism of action. In case you forgot where you were – spoiler alert – condensates are going to be involved, and that’s where these C-bodies are going to come into play. And of course, I tried to warn you up front, you’re going to know too much about this NPM1 protein, and that’s where the story is going to start. 

 NPM1 has a basic structure, very common to condensate-forming proteins. It has an oligomerization domain that allows it to form a self-pentamer, an intrinsically disordered region, and an RNA binding domain. And good both for its ability to phase separate, but also in terms of imaging and microscopy, it’s one of the highest proteins in the majority of cells, so about top 10 in most cancer cells, and top 50 in most cells in general, or I’ll say most human cells. Additionally, in the context of adult AML, it’s not just some minor rare mutation, it’s one of the top, if not the top, most common driver mutation in adult AML, and it really results from a small frame shift that changes very little of the amino acid sequence. Initially, it was found so easily because NPM1 itself is used as a histological marker in cancer pathology, and so when stained, it actually stains the cytoplasm, and so it’s historically had this name, cytoplasmic NPM1. Later it was found out that that’s because this frameshift causes a de novo site to bind the major exportin XP01 that can export it into the cytoplasm. Later results really seemed to suggest that it somehow was very directly regulating these leukemic genes, and it really was a confusing result how the cytoplasmic protein could do that. Of late, studies were coming out with this notion that maybe it is in the nucleus. That’s the backdrop to the story, and really yields the question of what’s going on? What’s happening here? 

 And so, just to summarize the first part, the introduction, I briefly told you about my lab’s interests to find out how condensates work in more complicated systems, this idea that what’s going on with these distinct AML drivers that are so different but seem to act very similarly, and then some depth on NPM1, which is a very unclear case of how this is working.  

 Now we’ll get into the crux of it, which is the backdrop of this C-body, and so my lab is really focused on really trying to create new models and using quantitative microscopy. We rely heavily on patient-derived cell models and related, patient-derived systems to do our work. The first steps were to take some of these cell lines from patients who had NPM1 mutant AML and also are still dependent on NPM1c for growth and other features that we’ll discuss later on. 

 But here you can see it is an endogenously tagged live cell with NPM1 wild-type labeled, and you see what you expect, which is NPM1 is in this ugly smushed chocolate chip cookie region. And then it’s also found, albeit less strongly, in the nucleoplasm labeled with N. And so now, if we compare it in the same cell where NPM1c is also in endogenously edited with a fluorescent tag, we see what would be expected. It’s in the cytoplasm, and what we could expect, given it’s binding to XPO1, which is it seems to have some enrichment on the periphery, suggesting that it’s at the nuclear-poor, slightly more than the cytoplasm. But unexpectedly, we additionally find that it’s in these bright nuclear puncta, and really not showing any preferential enrichment in the nucleolus. And this isn’t just this one cell, but throughout the population. 

 Even more striking, these small puncta seem to actually enrich NPM1c better than NPM1 wild type is enriched in the nucleolus. That was really surprising, and before proceeding, we also wanted to make sure that this wasn’t just a fluke of the cell line. Cell lines are pretty far from actual patients, and so we tested other cell lines, patient-derived xenografts models, and even antibody, stained with cutting-edge anti-NPM1 antibodies in near-direct, patient cells. And throughout all this, we see the same, consistent features. And so, what’s going on? How does this spatial separation occur?  

 First, why am I really asking that? And so, some background. Of course, it’s cancer, and so a key part of that is you need to divide pretty frequently, and a key part of division is actually breaking down most of your compartments, including essentially all, condensates. For NPM1, and Diana did a lot of work on this, NPM1 actually becomes phosphorylated, and that results in it not being a pentamer, and it contributes to nucleolar disassembly, where it goes for monomers. And after cytokinesis, and then you have two daughter cells, then it gets dephosphorylated, reforms pentamers, and helps the nucleolus reform. 

 In the context of NPM1c and NPM1 mutant AML, this creates a little bit of a seeming problem based on our initial results, and that’s that NPM1c and NPM1 wild-type seem very separated into distinct pentamers, one seemingly, in these foci or condensates, or eventually we’ll call C-bodies, and then NPM1 in the nucleolus, and so when they both get phosphorylated, because again, NPM1c is 95% of the sequence, and so those sites have not changed. They’re on the whole other side of the protein, and then the question becomes – how do they get back together?  

 The reason why that seems a little non-intuitive is because you’d expect that they should mostly just randomly mix, because the oligomerization or pentamerization domains are exactly the same between the two proteins, and so you can imagine there’s some oligomer sorting that happens over time, or maybe they actually phase separate pre or during that pentamerization process. That’s really where we started to try and gain some insights into what’s going on and how this happens.  

 I have a cool video of this. I hope people can see it okay on their screens. Here you can see NPM1 and NPM1c go through mitosis, so here’s metaphase. Every second here is about an hour, and then you can see rapidly when they undergo cytokinesis, they separate and form condensates. As that was really fast, I’ll quickly walk it through. As expected, mitosis resets NPM1c and NPM1 wild-type and removes them from condensates, and then largely they’re diffuse throughout and during metaphase. However, during anaphase and telophase, they both enrich on what’s called the chromosomal sheath, and so they’re essentially just adjacent to chromatin. You can see that here, and NPM1c is also there, albeit a little bit less. And then rapidly they seem to go to very distinct places where NPM1 is starting to form what are called pre-nucleolar bodies, and maybe we’ll call it pre-C-bodies. But, very distinct locations, and then additionally what was also apparent is that in early G1, a lot of NPM1 wild types seemed to be still in the cytoplasm. 

 From doing this on a bunch of cells, and with fast imaging without toxicity, we could do some quantification, and you can see a summary in this plot that shows NPM1c enrichment with NPM1 wild-type, and shows that progressive enrichment together, and then rapid, within nearly 3-minutes, separation or demixing that would be consistent with it separating during the oligomerization process. Additionally, once the periphery is established, NPM1c and NPM1 wild-type really have different allowances, per se, to go into the nucleus, where NPM1c really has no change in its nucleoplasmic and cytoplasmic levels, whereas NPM1 wild-type slowly enters the nucleolus, potentially more consistent with a slow, second stage of self-sorting that occurs. And note, this is really different than what you see in other cell types. For example, here’s U2OS, where this normally happens on the order of 15 or 10 minutes, not 3 hours. 

 For those of you who could be skeptical, like one of our reviewers was when we overexpress NPM1c in U2OS, it actually does recapitulate this delay, shifting it, and note the difference in timescale. And so now that we’ve really established that these form, and they seem to really happen rapidly and consistent with the signs you would expect for a phase-separated condensate, we wanted to take a step back and ask how do we distinguish exactly what’s here to basically understand more about how it works, but also to know if it forms independent of the NPM1c. 

 Initially, we had a hunch that it would relate to those chromatin-regulating proteins that are targeted. In particular, we focused on XPO1,NUP98, and KMT2A, and MENIN, because they’re the primary targets, and so here you can see that when we fix these cells and immunostain for those markers, in each case, those proteins are also in bright, are also contained within, or enriching within those condensates, and here you can see a radial enrichment from the center of the NPM1c signal, outwards for the other proteins. Then we thought, if this is a chromatin-regulated component, maybe it’s also involved in activating the chromatin. And consistently, we see polymerase II signal there, active marks, and, just as a negative control for what the split would look like, we’ve also ruled out a bunch of your standard, other nuclear bodies or nuclear condensates. 

 Finally, we’ve additionally confirmed that not only are those proteins there, but it’s also found on the HOX genes, itself. And so, it seems to be that the C-body is directly activating, or at least that’s our model, and this is consistent, or we were able to show that these four proteins are also co-enriching with your more standard chromatin localization sequencing assays, like cut and run. Now that we know some things that are there and have gained some insight, we are really able to ask, is NPM1c actually necessary for any of this that’s going on? Peggy’s lab, prior to this, had published an inducible degradation model of NPM1c, where when you add this molecule called dTAG-13, it induces acute degradation on the order of 3 to 6 hours, you get about 80% degradation, as shown here. And then now we can ask, what happens to those key leukemic regulator proteins? And so, here’s one of them, the one that interacts with NPM1C, XPO1. You can see, whereas in the control, there are these bright, co-localized foci, those actually go away. And this is not just true for this one cell and for XPO1, but it’s really true for every cell and for these four regulators, and because of that, we started calling these coordinating bodies with the notion that it’s coordinating the co-focal enrichment of these proteins. And so from now on, because it’s been hard, I’m just going to call them C-bodies. 

 This shows necessary, but is it sufficient? The notion is that right phase separation is this thermodynamic minima, where you minimize your system and separate into two phases, a dilute and a dense, and so the question is can we now add back and then degrade and see if there’s a rough concentration boundary? And in cells, especially the Degron models, where not every cell is degrading at the same level, it’s a little tricky. But in general, we did see signs where, at the lowest cells, we did not see condensates. Here’s a fraction of cells where one plane is showing an identifiable condensate above a threshold, and that seems to get bigger, as you overexpress. Additionally, because it’s difficult, we also went to other genetic systems that don’t have NPM1c to ask the reverse. Can we induce NPM1c formation? And we found it was quite independent of genetic context, and so here’s a common suspension line, HL-60, human CD34 blood cells with NPM1c overexpressed. Additionally, in all these cases, including in U2OS, where I put here, you don’t have the four  standard leukemic markers showing strong focal enrichment, but when you add an NPM1c now, even in these unrelated to NPM1 mutant contexts, you now get, them being enriched together and, consistent with a phase boundary, and so from here, I’d like to close, this section. 

 Diana Mitrea: I actually have a couple of questions. Anyone else in the room? 

 Josh Riback: Perfect. I think we have time for a minute or two of questions.  

 Diana Mitrea: It was really cool that the pentamers sort themselves out. I was wondering if you went deeper into the mechanism and whether binding to RNA has something to do with it given that the NPM1c has an RNA binding domain that’s defective of binding nucleic acids. 

 Josh Riback: That’s a great question. So we have done a little bit of follow-up on it, actually both anecdotally in my lab, but also things I’ve seen before my lab, and so even if you’re a standard protein, let’s say just delta RBD, you actually still will go to the chromosomal sheath and actually behave, just a little bit weaker, like NPM1 wild type. And so, it doesn’t seem, at least, that it’s rRNA-dependent, if that’s what you were meaning by RNA. In the context of the C-body, we don’t yet know if RNA is critical there or not for its formation. 

 Diana Mitrea: Got it. Thank you. 

 Josh Riback: Perfect. If there’s nothing else, then I’ll just continue for now. Now that we’ve really established that NPM1c is necessary and sufficient, really the question was how do these C-bodies work? What do they depend on? And how is this seven amino acid or 5% rough difference between the two proteins causing such a striking, difference in protein phase separation? And so, notably, for context, sorry, I skipped over that. But anyway, this is the third section, while condensates were initially discussed in analogy to oil and water, proteins and nucleic acids are much more complicated than oil, and so, therefore, they’re never exactly perfectly fitting into that. The simplest case is that there’s many components in the cell, and so then, so you’re going to depend on the ratio of those two components, and they’re favorable to interact together, which is usually called in the field heterotypic interactions, or with each other, homotypic interactions. And so previously in Cliff’s lab, we developed this assay where it allowed you in cells to overexpress one protein, and then ask how the partitioning or related, free energy of transfer changed as you overexpress that protein. Despite the nucleolus getting much, much bigger, NPM1in this case, doesn’t actually like the nucleolus as much, despite that size difference. And so that roughly is able to quantify the heterotypic interactions.  

 Additionally, I’ll briefly mention there’s even more complexities emerging in the literature with this notion that networks interactions and inhomogeneity and some other exciting things from multiple groups, and it creates behavior at small scales that maybe we’ll briefly touch upon, but largely, I’m going to stick with introducing this idea of heterotypic, destabilization.  

 First, we want to know what is driving these C-bodies? Is it just predominantly NPM1c interactions with itself, or is there really a role for heterotypic interactions? And so, a way we thought to do this and also for those of you who are more maybe in the clinical direction of relevance to agents that are used to some extent, depending on which one. In the clinic, we focused on those. And so those are, as I mentioned previously, XPO1 inhibitors, here shown in blue. This is Eltanexor, which is one of the ones available, and VTP, which is a MENIN inhibitor, which binds the interface between KMT2A and its partner MENIN, but called MENIN inhibitors. So, if one adds Eltanexor it’s well known that then – sigh of relief – it isn’t just messed up in some unknown way, but it actually will go into the nucleolus, which again, you can see by, maybe not as bright, but still smushed chocolate chip cookie. 

 And so, of course, under those conditions, when that XPO1, NPM1c interaction is broken, now we also additionally, again, get that the core chromatin regulators we looked at don’t go too bright. Puncta are seemingly that interaction is essential for C-body formation. Additionally, we applied VTP or MENIN inhibitors, and in that case, we did not see noticeable loss in condensates. However, when we looked at the individual proteins and their recruitment, all were still recruited to C-bodies, except for MENIN. So it seemed like the MENIN inhibitor was able to push MENIN off the C-body, or that MENIN was only in the condensate, presumably due to that interaction, and thus compositionally changing it. And so, this really shows at least one heterotypic interaction between XPO1 and NPM1c. That’s important. Additionally, I’ll just summarize that we do not just analyze n equals 1 pretty cell, but do this on dozens of cells to really show that it happens at a population scale, but for time, I’m just going to gloss over that and refer you to the paper, and hopefully you’ll trust me. If not, you’re free to read up more there. 

 Now that we’ve established it’s heterotypic, we thought, maybe we can go back to that idea of heterotypic destabilization and try to really tease apart, or begin to open up, how important C-bodies are for things like cell growth and things like that. First, we overexpressed NPM1c without degrading the endogenous, similar to what I showed before with NPM1 wild-type, in different contexts, and you can see the condensates get really big, but also does the localization in the cytoplasm. If we quantify that with overexpression level, what we see is that the C-body becoming destabilized doesn’t actually strongly increase nucleoplasmic levels, it allows for better export of the NPM1c protein, a sort of indicator of a weaker C-body formation. And not only that, but heterotypic destabilization also results in selection, and so the cells actually grow slower that are expressing high levels of NPM1c, shown here. They get selected out of the population, which makes them a little bit more difficult to study, and anecdotally, we’ve been told that others have had difficulties studying NPM1c for that reason in the endogenous context. 

 With this in hand and that heterotypic destabilization seems to be a feature, we thought, maybe we can push this and sort of ask which interactions and how do they work together, which sequence features of NPM1c actually drive C-body formation? And so to do that, inspired by studies, in and domain assignments, mostly from Richard Kriwacki’s lab that Diana was a part of. We did a full truncation screen of all the regions in the IDR in the C-terminal domain that are classically annotated. The idea is that when expressed at low levels, or extrapolated to low levels, we could look at the enrichment or the amount in the C-body versus in the nucleoplasm as a measure of the stability and the role of that domain that’s being deleted in the interaction. And the notion, broadly, in the field, is this idea of weak interactions spread along the sequence of a protein, what I’m going to call non-cooperative interactions. And when we look at the data, we were really shocked to see something really not consistent with that. And in fact, most, if not all but two of those truncations result in a more than tenfold decrease in their ability to enrich in C-bodies. And that’s more consistent with a cooperative network of interactions, not weak spread across, but really all working together. 

 From here, we can also do the other assays that I showed on the overexpressing the full length, and so again, only those that still partition well are the ones that really hurt viability and result in the strongest response of the endogenous to be pushed into the cytoplasm –  strong signs of heterotypic destabilization. Of note, you can see that if you focus on the three, that really were those outliers from the tenfold impact and partitioning, and the reason why I’m highlighting them is because now we then asked, what happens when we actually degrade the endogenous NPM1c?  Which ones can rescue, so are able to form condensates, and so here’s the percentage of cells that I showed previously versus overexpression level. And again, only three, including the sort of wild-type full-length, are able to form condensates. Additionally, here I’m showing differentiation marker after a week in culture, showing that only those that are able to form condensates are able to  maintain the leukemic state, so they continue to grow, and you can see that here, where I’ve highlighted, that only those that are able to form condensates, and where those few that were partitioning highly, are still now the only ones that are able to grow. 

 From here, Peggy’s lab took one mutant in particular that seemed to be able to still be in the nucleus, and a little bit in the nucleolus, and additionally was still in the cytoplasm, but  was not able to localize the C-bodies to use as a strong tool to ask about the role of C-bodies in maintaining various leukemic features, and so we compared, overexpressing those, just in OCI-AML3 after degradation, and looked transcript-wide and saw that only the only, the construct that forms C-bodies was able to maintain the leukemic transformation. Additionally, we looked at the ability to stall differentiation when switched to an appropriate media that is supposed to induce the differentiation. Only those with C-bodies were able to delay that substantially. Peggy’s lab actually engraphed and showed the ability for in vivo expansion of these constructs in mouse. And again, only C-bodies seemed to be necessary broadly in these assays for maintaining the leukemic state.  

 Here you can see the fundamentals of how the C-bodies worked and also that allowed us to look at their role in maintaining the leukemic state in these assays. Finally, this idea of disparate mutations converging on C-bodies begins with this notion that Rick Young and Richard Kriwacki’s labs have been, along with others, have been pushing, which is this idea of fusion oncoproteins and also other pathogenic variants being prone to condensate formation. And really, there’s a wide depth there, and I think there’s a lot of interest in that space, and want to phrase it to preface maybe a little bit bigger, where we’re going to go at these in the narrow AML field, of this broader idea of how do we think about that. It almost seems like a daunting idea that there could be so many different ways, or so many different types of oncogenic condensates on one hand, or maybe there’s some way to converge them or bridge them together. And so, again, could that be the C-body for our nucleoporin oncofusions and these KMT2A or MLL rearrangements? First, we asked, in our hands, did they in U2OS when overexpressed independently? So no NPM1c here, and they’re each expressed alone, do they form, bright puncta? And they did, and I’ll show it at the end that they seem to be consistent with having a Csat.  

 Additionally, we were able to see that the same core proteins were being recruited: XPO1, MENIN, KMT2A, NUP98, to these de novo condensates. And here, we were initially like maybe that’s good enough, but of course, they could be acting either as more like clients or more like scaffolds. It doesn’t necessarily tell us if they’re actually forming via the same mechanism. And so, we thought about this a little bit in a different way. The way we were thinking about this is if the proteins are – imagine NUP98 is just FG repeats, and NPM1c is more unfolded, RBD interactions, and they actually don’t really need to interact that well, but can actually drive phase separation independently or distinguishably, then you could, with high overexpression in the same cell of NPM1c or of an oncofusion form distinct condensates that would both recruit the downstream leukemic proteins, but actually form condensates in a biophysically distinguishable way, or on the other hand, they could be indistinguishable, so they could actually be sort of enhancing similar interaction networks, and, therefore, one would never get different condensates within the same cell, and only have compositionally dependent condensates, depending on the ratio between the two drivers. 

 Long story short, we have yet to see a case, where they were able to form distinguishable condensates, and here’s an example of what that looks like, where there’s just complete overlap. And when one is a little bit dimmer due to being slightly out of Z, so is the other. And so, it really seemed to suggest a shared scaffolding network, and maybe a potential, possible targetable entity.  

 And so that ends my final part showing that these other oncofusions can form puncta that recruit C-body-associated leukemic regulators, that they are biophysically indistinguishable from NPM1c-driven C-bodies and that together, these results really suggest that C-bodies may be a unifying feature in therapeutic vulnerability across multiple leukemia subtypes. And from that we’re done. I’ll just end with a little, quick summary. Hopefully I’ve convinced you that C-bodies are a little bit, unexpected, especially in the ability for NPM1c to form them separate from NPM1wild-type, and sort of the unexpected cooperativity, amongst its sequence for doing that.  

 Additionally, just to summarize maybe a little bit more of the big picture cancer / therapeutic elements, we discovered this C-body demonstrated that it seems to be essential for cell growth differentiated expansion. And then have proposed it as an emergent phenomenon that may drive these distinct leukemia subtypes. And from there, I want to conclude and, thank everyone involved and everyone in my lab, especially Gandhar, who, if you’re in the Boston area, is going to soon be looking for postdocs and, and, potentially industry during his residency, and so, I highly recommend him, and, definitely reach out, and, I hope you all enjoyed, and I’m really excited to answer any questions. 

 Diana Mitrea: Thank you, Josh. Are there any questions online? Anyone in the room? 

 John Manteiga: Hi Josh, thanks for the great talk. I was wondering if there are any known therapeutics out there that are known to target one subtype of AML that you could test to see if they kind of unexpectedly also target these other subtypes, given the multiple factors coming together in the same condensate to kind of support the therapeutic angle? 

 Josh Riback: We focused on MENIN inhibitors so strongly, because this was an emerging avenue, that was unclear at first if the MENIN inhibitors, which were more developed for the MLL or KMT2A rearrangement subtypes would work. But beyond that, we’ve done a little bit, but not as extensively. That’s a great question. 

 Diana Mitrea: Following up on John’s question. You showed that beautiful data with the XPO inhibitor that dissolved the NPM1c C-bodies. Do you see the same thing with other fusions? Do XPO inhibitors dissolve all the C-bodies, regardless of what fusion they originate from? 

 Josh Riback: We have some new results on this. We went into it a little bit in the paper, a little bit more in the context of the MENIN inhibitors, which did act this way and it does seem like XPO1 inhibitors in some of them will result in dissolution. 

 Diana Mitrea: So do you think that XPO1 is one of the scaffolding proteins, then? 

 Josh Riback: Definitely. I think it really highlights that NPM1c seems to really be hijacking it from that network of interactions. It really gives this picture of this shared underlying network that something like oligomerization and causing phase separation may be hyperactivating when cells are trying to actually turn it off and differentiate is the idea. 

 Diana Mitrea: I have one more. Have you looked at a broad bioinformatics or data science approach to figure out which fusions might contribute to C-bodies? And do all the proteins that you’ve seen so far that are driving C-bodies, do they have strong XPO binding sites? 

 Josh Riback: We’ve done this a little bit. These NES predictors have mixed strengths and weaknesses. It was largely not very clear. Notably, even in the context of Nups, even NUP98, right, it’s not clear that that would even come up as a strong signal in that assay or bioinformatic approach, I guess I should say. 

 Diana Mitrea: Oh, that’s fascinating. Join me in giving Josh one more round of applause. Thank you to everyone who joined us today online and in person. Thank you to the team that worked behind the scenes to make the KTT happen. Thank you to Jennifer Tallman for editing and marketing and Bobby Ingram for IT support. As I said at the beginning, the recording will be posted on Condensates.com and YouTube within a week. Take care, have a wonderful summer, and see you at the next one. Thanks so much, Josh!