Senior Research Associate, Dewpoint Therapeutics
|Type||Kitchen Table TalkFor Beginners|
It was fantastic to have avid condensate scientist from Dewpoint, Bede Portz, give a Kitchen Table Talk about the basics for the condensate community and the first class of Dewpoint interns on June 8. Bede received his PhD training at Penn State with David Gilmore where he studied RNA polymerase II and became obsessed with the biological and pathological implications surrounding condensates. His postdoc fellowship was with the BrightFocus Foundation at the University of Pennsylvania in James Shorter’s lab, where he worked on developing therapeutic RNAs to combat aberrant phase transitions and neurodegenerative diseases with noncoding RNA condensation. Now a Senior Scientist at Dewpoint, Bede demonstrates an incredible passion for science, in general, and specifically condensates, as you’ll see in his video below.
You can find links to all of Bede’s references further below; many of the best reviews are also listed in the ‘For Beginners‘ section of the site. If you’d like to have a conversation with Bede about condensates, feel free to email him at email@example.com
Create an Account or Sign In to view the video.
Alex Milot (00:01):
Hi, everyone. Thanks for coming. I have the pleasure of introducing my boss and critical member of oncology, Dr. Bede Portz. Bede received his PhD training in the Center for Eukaryotic Gene Regulation at Penn State with David Gilmore for a studied RNA polymerase II and became obsessed with the biological and pathological implications surrounding condensates. His postdoc fellowship was with the BrightFocus Foundation at the University of Pennsylvania in James Shorter’s lab, where he worked on developing therapeutic RNAs to combat aberrant phase transitions and neurodegenerative diseases with noncoding RNA condensation. He demonstrates an incredible passion for science in general, and specifically condensates and became a condensate expert to take this passion to new heights. So if you ever want to talk to someone about condensates, I recommend talking to Bede. He loves them as much as he loves his man bun, which we all absolutely adore. And with that, I’m going to let Bede give his talk.
Bede Portz (01:07):
Okay. Thanks, Alex. If you got something right, it is that I am enthusiastic about this topic, which is both condensates and Dewpoint, this place that we are together building. So for the aficionados, this might be an unseasoned dish, but the thrust of this talk is really going to be introductory. And in the audience here in Dewpoint are our first crop of interns. So welcome. And what I hope is that this talk gets you enthusiastic, but also lowers the activation energy to further explore the literature. So I’m going to talk about… I was tasked with two things. So talking about condensates and talking about Dewpoint, and each of these things transcends what could be covered in 45 minutes. The scope of this field has exploded in the last decade and indeed the scope of Dewpoint across multiple therapeutic areas and partnerships and disease programs also transcends what could really be covered in 45 minutes.
Bede Portz (02:13):
But I’m going to try to focus this on those concepts that most captivated my attention when I was new to this field. So on my slides to that end are some references that I have found particularly interesting. And this is an image from one of them and it depicts what is becoming an increasingly understood phenomenon in biology. And that is that cells are organized in via both membrane encompassed organelles, but also membraneless organelles, which have been come to be known as biomolecular condensates. So these things are widespread and this image depicts some of the more well-characterized condensates, and these are non-stoichiometric assemblies of proteins and nucleic acids…
Bede Portz (03:04):
So that is to say that in contrast to large complexes, like the ribosome that assemble through stereo-specific lock and key type of interactions with defined stoichiometries, condensates have a range of allowable stoichiometries of components. And they can come together in time and space to organize particular cellular functions or respond to particular stimuli. And so some of the examples that I’m going to touch upon in the talk today are the nucleolus–and so this is the site of ribosomal RNA, biosynthesis and processing and ribosome assembly–in the nucleus and stress granules in the cytoplasm.
Bede Portz (03:44):
And these are RNA and protein condensates that form dynamically in response to multiple different types of stresses. And so these things have been observed in cells for quite some time, and they form at least some of them in part through a process known as liquid-liquid phase separation, which can be described by a phase diagram. That’s what I’m depicting here. And the radical thing about this is that what phase separation does is integrates linear changes in the concentration of a biomolecule into switch-like changes in the behavior of the solution.
Bede Portz (04:27):
So on the phase diagram on the Y axis is the interaction strength between two molecules. And on the X axis, it’s the concentration of the biomolecule in question, in this case, an RNA-binding protein. And so as you move across the concentration axis, say by going from position one to position two, you might only change the concentration of the RNA-binding protein in question by a little bit, but you change the behavior of the solution by a lot. And that’s depicted by these images at the bottom. So at position one, you have a well-mixed homogeneous solution. And at position two, which again, might arise from a really subtle change in either the solution conditions or the concentration, you have this switch-like demixing resulting in a dense phase and a light phase.
Bede Portz (05:18):
And so what a phase transition is really is a density transition. But one of the things I want to highlight, and this is a pitfall that people encounter when they’re new to the field, is that the entire cell is dense. So what I’m referring to is a density transition with respect to the thing that you’re looking at. And so in a cell, in contrast to this cartoon, you have a density transition that serves to essentially sort biomolecules in space. You put things in a compartment that lacks a membrane with other things with which that molecule may collaborate to execute some function.
Bede Portz (05:59):
So how does this demixing process occur? The Pappu lab has given us a really nice conceptualization known as stickers and spacers. So what I’m depicting here is a generic biomolecule. It could be a protein, it could be an RNA. And the sites on that molecule that are capable of interacting with other copies of itself are known as stickers, and the intervening sequences are known as spacers. And one of the concepts that you’ll encounter in the literature is that of multivalency, the idea that a lot of phase separating and condensing molecules are comprised of multiple stickers, and it is the summation of some amount of sticker-sticker interactions that leads to demixing or condensation.
Bede Portz (06:45):
And another theme that you’ll encounter in the literature is that of intrinsic disorder. So intrinsically disordered proteins are proteins that don’t adopt the stable three dimensional structure. And another point I want to emphasize is that intrinsic disorder is not a requirement for phase separation. So there’s a sometimes misleading link between the two, but intrinsically disordered proteins are definitely implicated in phase separation. In some cases, intrinsically disordered regions are required. In other cases, they’re modulators or tuners of condensate composition or behavior. And we’ll talk about that a little bit. But intrinsically disorder proteins are cool for multiple reasons and as an evolutionary solution to encoding multivalency, that they’re a pretty clever invention.
Bede Portz (07:36):
So if you have a protein that’s not under selective pressure to adopt the stable three-dimensional structure, evolution can iterate through the sequence space of that sequence and come up with solutions that can tune other functionalities to include condensations. So this is a theme that you’ll encounter quite a bit in the literature, intrinsically disorder proteins owing to their flexibility can facilitate phase transitions in multiple ways. So if you think about a molecule that is multivalent, that’s multiple stickers that can engage multiple other stickers. If that molecule is also flexible, it can interact with other stickers in different three dimensional orientations, which can facilitate condensation and a network of interactions.
Bede Portz (08:27):
So phase transitions are a concentration-dependent phenomenon. And so essentially what happens is at some threshold concentration known as the saturation concentration under a given set of solution conditions, sticker-sticker interactions become favored over that molecule’s interactions with the solvent that it’s in, the nucleoplasm the cytoplasm. And that leads to demixing. And so the higher the concentration in theory, the more likely these intermolecular interactions are to occur. And the more likely you are to get demixing. And again, I think the cool feature of phase transitions are that relatively subtle changes in the interaction strength or concentration of molecules can elicit dramatic changes in the way the solution behaves.
Bede Portz (09:16):
Once a condensate forms, these sticker-sticker interactions don’t necessarily stop. And so if you think about the probability of sticker-sticker interactions occurring in a dense phase, they’re more likely to occur than in the light phase where the concentration of those molecules is lower. And so as a function of time, more and more sticker-sticker interactions can occur. And thus you can get changes in the material state of a condensate.
Bede Portz (09:50):
This can have pathological implications in the form of protein aggregation, and also prion-like polymerization. So these are concentration-dependent phenomena as well. So you can conceive of situations whereby a liquid-like condensate might form, but as a function of time, more and more sticker-sticker interactions occur. And that leads to an aggregate that the cell has a difficult time resolving. Likewise, there are rare protein conformations that are capable of self-templating one another, such that when they interact with another copy of themselves, they induce that same conformation and so on and so forth. And you get these fibrillization type of scenarios, which can also be pathogenic. And this could conceivably be more likely to occur in an environment where there’s a lot more of those self-templating molecules around.
Bede Portz (10:45):
And so the cell has evolved means to regulate the material properties of condensates to include things like protein chaperones, RNA helicases. These are molecules that are machines. They do enzymology using ATP as fuel. ATP itself as a component of the cellular solvent can also impact the material state of condensates. And again, because this is a concentration-dependent phenomena, the rates of synthesis of decay of biomolecules can also impact the behavior of condensates.
Bede Portz (11:21):
Okay. So far these examples that I’ve depicted are that of a homotypic condensate. So that’s a single component demixing from the solvent. But in cells, of course, these condensates are going to be heterotypic. That is they’re going to contain multiple proteins. So in the context of the stress granules for example, that could be literally hundreds or thousands of different proteins and RNAs. And the grammar that encodes the composition and material state of condensates is an area of active research.
Bede Portz (12:00):
So if you allow for the notion of a heterotypic condensate, that invites all sorts of means of regulation. And so again, take our toy example of an RNA-binding protein that undergoes a phase transition, and now consider adding RNA into the mix. So on one extreme, if you have a ton of RNA around and a relatively small amount of that RNA-binding protein, RNA-binding protein binding to the RNA leads to this big molecule that may sterically occlude self-assembly. On the other extreme, you might have an RNA on which multiple self-assembly prone RNA-binding proteins dock, and that could nucleate a condensate. And these behaviors have indeed been identified for RNA-binding protein on RNA-binding protein systems, and described as opposite extrema of a process that really exists on a continuum.
Bede Portz (12:56):
So in this top panel, we’ll revisit this idea of concentration-dependent phase transitions. And here you just have a homotypic case where you’re titrating in increasing amounts of the RNA-binding protein. And at this so-called saturation concentration, you get the switch-like demixing and the solution now has two phases. In the bottom example, you’re not changing the concentration of the RNA-binding protein. You keep that constant and you’re instead titrating in the RNA. On one hand, you could get nucleation, a situation where you have multiple RNA-binding proteins docking onto a relatively small number of RNAs. And on the other extreme, you can dissolve that phase by synthesizing perhaps more RNA and creating a situation where you don’t have such close opposition of RNA-binding proteins. And so this behavior is known in literature as reentrant phase separation. And that’s also a concept you might encounter with pretty profound biological implications.
Jill Bouchard (13:53):
Hey Bede, we have a quick question about IDRs in the audience. Muthu, would you like to unmute yourself and ask?
Muthu Venkat (14:03):
Hi. Just a quick question regarding IDRs. Is it always that IDRs promote condensate formation? Or if not, is there a particular idea that promotes condensate formation and there are certain ideas that do not promote condensate formation?
Bede Portz (14:21):
Yeah, that’s a great question. So let me preface the answer by saying that there are a whole lot of things that I’m talking about, the nucleolus, stress granules, IDRs, these were active areas of research onto themselves well before phase transitions entered the biological lexicon. So we’re not replacing the other roles for these entities or IDRs. So now more specifically to your question about IDRs. IDRs are absolutely not required for phase separation. They absolutely do not equate to phase separation. I’m going to highlight briefly an example from Allen Drummond’s lab in an upcoming slide where the protein PAB1 is a stress sensor in cells, and it has an IDR, but it actually doesn’t need the IDR for its phase separation. So there’s an example of where the IDR is dispensable. There are plenty of other examples where it’s essential. And so really this is case by case, and there are many, many flavors of IDRs that drive different types of interactions. And this is also an area of active research. And if you want to email me, I could email you some papers that I think might be accessible entry points to this topic.
Muthu Venkat (15:36):
Fantastic. Thank you.
Bede Portz (15:37):
No problem. Okay. So there are some folks who are skeptical about the implications of condensates. And I want to take a moment to highlight some examples of diverse and adaptive functions for condensates in the cellular context. So briefly I alluded to the nucleolus. This is the factory for ribosome biosynthesis, and some really elegant work by the Kriwacki and Brangwynne labs, including by our own Diana Mitrea who’s here at Dewpoint revealed that the nucleolus is actually a nested layer, it’s essentially a multi-layered condensate. The center of which is the site of ribosomal RNA biosynthesis, and the other layers serve essentially to chaperone the processing and folding of ribosomal RNAs in a flux, a directional vector out of the nucleolus. So whether or not condensates are essential for this process is immaterial because ribosomes are pretty important and the nucleolus is where they’re getting made and the nucleolus is as bonafide to condensate as exists.
Bede Portz (16:55):
And then I touched upon this work from the Drummond lab, this stress sensing demixing process. So there’s this protein, poly(A)-binding protein. It binds RNAs, but it senses thermal stress. And when it does that, it releases RNAs and it demixes, and this changes the available pool of RNA and this process is adaptive. So if you mutate the poly(A)-binding protein in ways that inhibit its ability to respond to stress via condensing, the organism yeast is a less fit. So there’s an adaptive evolutionary role for condensates here.
Bede Portz (17:33):
And then a paper that I was fortunate to contribute to when I was at Penn State is it involves RNA polymerase II. So this is the machine that’s responsible for transcribing protein-coding genes. So it’s pretty important. And affixed to it is a long IDR known as the C-terminal domain or CTD. And if you express the CTD in cells fused to GFP, it forms condensates, they zip around the nucleus and they’re dynamic. If you make mutations to those GFP-CTD fusions that abrogate the ability of the CTD to form condensates, or allow it to form condensates with altered material properties, they’re less dynamic and they’re less mobile, and you move those mutations into the context of the RNA polymerase itself, the fly doesn’t develop, but you can make considerable mutations to this very conserved sequence. And as long as it can still phase separate comparable to the wild type version of the protein, not only does the polymerase function, but you can get a fly, which involves a highly complex developmental process that needs to be tightly regulated. And so here’s a situation where, again, this is likely adaptive and we posit that the function here is to localize the CTD to sites on the genome where transcription is occurring.
Bede Portz (18:58):
So disease is essentially biology gone awry. And so if you can consider adaptive evolutionary functions for condensates that begins to invite us to think about roles of condensates in disease. And I’m going to start with neurodegenerative diseases, which are essentially the index case for the intersection of the condensate and biomedical fields. So it’s long been known that many neurodegenerative diseases, including the devastating neurodegenerative disease, ALS are linked to protein aggregation. These are the pathological hallmarks of ALS. And so what I’m showing you here is a figure from a paper that shows protein aggregates in the cytosol of a neuron in a deceased patient. ALS has been historically a pretty intractable problem. 10% of ALS cases are so-called familial; there’s a genetic link. 90% of them are sporadic; there’s no known genetic link. So the underlying genetics makes this a difficult problem. And it’s fortunately a relatively rare disease.
Bede Portz (20:12):
Of the genetic risk factors, there’s this long tail of genetic risk spread across a number of genes that interestingly code for similar proteins. These are RNA-binding proteins with disordered domains, a specific type known as a prion-like domain. And interestingly, one commonality of these proteins in addition to their architecture is that they’re all enriched in stress granules. And so this invites us to consider not each of these proteins as individual drug targets in decreasing numbers of patients, which makes developing drugs increasingly challenging. But it instead invites us to look at the condensate itself as an integrating node of pathology. And so all of these proteins are involved in stress granules.
Bede Portz (21:12):
And one of them, FUS, was identified as being able to undergo a phase transition. So this is data from Avinash Patel who’s also now at Dewpoint. That work he did with one of our founders, Tony Hyman and collaborator Simon Alberti. And so what Avi did is purified FUS and shows that under certain conditions in a test tube, it can form condensates and those condensates mature as a function of time. And some of them end up turning into these gnarly looking aggregates, reminiscent of the protein inclusions found in the cytoplasm of neurons of deceased patients with ALS. That’s pretty interesting. Moreover, a single amino acid point mutation in the disordered region of FUS hastens this liquid to solid aberrant phase transition that gives rise to these gnarly looking aggregates. So this paper and a paper from Paul Taylor’s lab that came out at the same time, really opened up this idea of condensates as nodes of disease and aberrant phase transitions as potentially representing the etiology of various protein aggregation disorders.
Bede Portz (22:31):
That is highlighted on this slide. So each one of these proteins was from that list of ALS-associated, stress granule-associated proteins that I mentioned earlier, and each of them by multiple groups in some instances has been shown to undergo some sort of aberrant phase transition, and often ALS-associated single amino acid point mutations alter the properties or dynamics with which these aberrant phase transitions occur.
Bede Portz (23:04):
Neurodegenerative diseases and cancer are often considered as completely different diseases. Neurodegeneration is a disease where cells die when they shouldn’t and cancer is a disease where cells don’t die when they should. And this slide hints at a link between these two, otherwise disparate diseases. So FUS, TAF15 and EWSR1 are stress granule proteins that aggregate in certain patients with ALS. Interestingly, the disordered domains of each of these proteins is also involved in cancer. So there are instances where there are DNA breaks and the cell stitches the genome back together, and in so doing sometimes you get these chromosomal trans locations that link DNA binding domains from one protein to the disordered domain of other proteins to include FUS, TAF15 and EWSR1. And that creates this aberrant chimeric transcription factor that can drive an oncogenic gene regulatory program in cancers like sarcoma. That’s bad.
Bede Portz (24:17):
And also the architecture of RNA-binding proteins is shared with transcription factors and I’ll get there in a second. So as an aside, this is a picture from the version of the classic textbook Molecular Biology of the Cell by Bruce Alberts that I had when I was an undergrad. And this is how transcription activation was explained to me. You have activator proteins, these are transcription factors. They bind some distal regulatory element known as an enhancer, and then through some sort of magic, they fold over onto these general transcription factors. So activator proteins, transcription factors, these these are sequence-specific DNA binding proteins. They regulate relatively small subsets of genes. They do so in a tissue- or lineage-specific way in contrast to this general transcription machinery, which is more or less required for all protein-coding gene transcription.
Bede Portz (25:16):
And so how does this cell integrate these tissue- and lineage-specific, sequence-specific transcription factors with this general stuff that’s required everywhere? That’s a question. And this was sometimes explained via this looping, this long range three-dimensional genome interaction, which I want to stress is very important for gene regulation. But this was explained as magic when I was an undergrad. This loop somehow stabilizes this massive complex and hundreds of polypeptides must come together at precise time and space, especially during development and stress responses to get appropriate gene expression. And somehow this loop was the key to this as somehow organizing and stabilizing this, enriching for the right stuff at the right time. But biology doesn’t occur on the printed page. And a loop turns sideways is also a hoop and stuff can fall through it. So this never made a lick of sense to me.
Bede Portz (26:20):
And now, again, back to this link between neurodegeneration and cancer, RNA-binding proteins, transcription factors, they have this common modular architecture. There’s some sort of nucleic acid binding domain fused to some sort of intrinsically disordered domain. It’s long been known that intrinsically disordered domains in the context of transcription factors function as their activation domains. And they can be these aberrant activation domains in the context of fusion proteins, like I alluded to. So again, this hints at a link between the condensate behavior of these proteins and their regulatory function.
Bede Portz (26:54):
And this has been interrogated by many groups, including some people also that have ties to Dewpoint. And so first there was this proposal that super-enhancers, which are clusters of enhancers, which are, again, these distal regulatory elements to which transcription factors bind, are themselves condensates. So this was modeled. So super-enhancers, these coalescence of enhancers lead to switch-like amplification of gene expression in cells, much like condensates involve the switch-like, self-assembly and demixing of constituent proteins.
Bede Portz (27:33):
And so this work was followed up on experimentally by Anne Boija and Isaac Klein who are also now here at Dewpoint, back when they were in Rick Young’s lab at MIT, who’s one of our scientific founders. And I’m going to walk you through some of their findings, which I think are particularly interesting. So super-enhancers are characterized in part by an enrichment of one of these general factors known as mediator, specifically the subunit of mediator known as Med1.
Bede Portz (28:02):
So this leftmost panel, these are ChIP-Seq on browser tracks. So this is a whole bunch of cells ground up, and you’ve mapped the location of various proteins to specific sites of the genome. The rest of the images are individual cells in contrast to this ChIP-Seq data. So at sites of the genome where this general factor Med1 is enriched, a very specific factor, OCT4 is enriched. OCT4 is a transcription factor that’s involved in pluripotency, the maintenance of stem cells.
Bede Portz (28:34):
And if you look at individual cells, sites where OCT4 forms these clusters, you also find nascent transcripts, the RNAs themselves emerging from the underlying genes that are regulated by OCT4. And these genes are also associated with super-enhancers. So the idea here is that these clusters of enhancers recruit Med1, which is a very general factor that somehow integrates the signal of these very specific factors.
Bede Portz (29:05):
So mechanistically, how does that work? So again, condensates: concentration dependent phenomena. As you titrate in increasing amounts of GFP-labeled OCT4, the image gets brighter, more GFP, but you don’t see demixing. OCT4, at least under these conditions is not phase separating or condensing by itself. But Med1, the IDR of Med1, the intrinsically disordered region, by contrast it does.
Bede Portz (29:31):
And Med1, not only does it form condensate, it can co-condense with OCT4. So this hints at a way by which these general factors and specific factors might come together. So in this experiment, GFP-labeled OCT4 and mCherry labeled Med1 are mixed together, and you get these yellow drops, co-condensates. And if you make mutations in the acidic activation domain of OCT4, this is the IDR of OCT4, that abrogate its ability to co-condense with Med1. And you move those mutations into a luciferase reporter assay in cells, the mutations that impair the ability of OCT4 to co-condense with mediator impair the ability of OCT4 to activate transcription from this reporter assay. So this is but one link in this paper between condensates and co-condensates and gene activation. And I’m not doing this paper justice, there’s other examples.
Bede Portz (30:35):
So what I’ve described is this through line between the behavior of RNA-binding proteins and transcription factors between transcription factors and the activation of gene expression. And that then invites us to think about condensates as targets for cancer therapeutics, much like we might think of stress granules as integrating nodes of ALS pathology and thus targets for neurodegenerative disease drugs.
Bede Portz (31:01):
So another transcription factor is known as the estrogen receptor. This responds to the hormone, estrogen. Estrogen is a ligand for the estrogen receptor. It binds the estrogen receptor. It leads to estrogen responsive gene activation. Interestingly, if you make Med1 condensates and you add to it GFP-labeled estrogen receptor, there’s a modest recruitment of estrogen receptor into those Med1 condensates. But remarkably to me, if you add estrogen, it increases the co-condensation of the estrogen receptor with Med1 condensates in a way that hints at this estrogen responsive gene activation in cells. This result is one of a relatively small molecule, estrogen, binding a protein, regulating its condensate behavior. And that invites us to think again about small molecule drugs that might do the same.
Bede Portz (31:59):
And again, in work from Isaac and Anne when they were in Rick Young’s lab, starts to dig into this concept. So Tamoxifen is a breast cancer drug. It is somewhat a mimic of estrogen. And so far as it binds the estrogen receptor, but it does so in a way that prevents its activation. And here what they show is that the addition of Tamoxifen prevents this co-condensation of the estrogen receptor with Med1 condensates. So if you add Tamoxifen, fewer of these condensates are yellow because it’s not the co-localization of the estrogen receptor in Med1. Now, patients and their cancers evolve resistance to Tamoxifen treatment. And one of the ways in which that occurs is through mutations to the estrogen receptor that impair its ability to bind Tamoxifen. And if you do the same experiment with those mutant forms of the estrogen receptor, they’re less affected by the addition of Tamoxifen.
Bede Portz (33:04):
So there’s another route to Tamoxifen resistance. And that is the overexpression of Med1 in cells. And so Med1 is over expressed in Tamoxifen-resistant cells and the addition of extra Med1 to these condensate experiments in a test tube, counteracts the ability to prevent this ER-Med1 co-condensation. So condensates might explain the integration of sequence-specific and general factors. They might in part explain how genes are activated in response to particular ligands. And condensates might also explain, in this case, multiple routes to Tamoxifen resistance in the context of breast cancer.
Bede Portz (33:55):
So condensates sort biomolecules in time and space, they bring certain subset of the proteome and transcriptome together. The result is chemical environments that must be somewhat unique to the interior of individual condensates composed of different stuff. And that led this group to posit that drugs might differentially target different condensates by virtue of exploiting differences in the chemical environment. And I’m going to highlight two of the most striking examples.
Bede Portz (34:30):
I’ll actually talk about one of them, JQ1. So in this experiment, a bunch of different condensates are made for different proteins. The one to pay the most attention to is BRD4. The cancer drug candidate JQ1 binds to the folded domain of BRD4, a domain known as a bromodomain. In this experiment, a fluorescently labeled version of JQ1 gets enriched in BRD4 condensates. But the radical thing about this result is that in this experiment the folded bromodomain that is the target of JQ1 binding is absent. These condensates are comprised of just the intrinsically disordered region of BRD4. So what this result suggests is that the chemical environment in this condensate seems to preferentially enrich for the partitioning of this drug JQ1, and that has radical implications. Perhaps we have all along, been selecting for drugs with favorable condensate properties, unbeknownst to us. And I think that was a pretty profound insight.
Bede Portz (35:47):
And it’s been followed up on. So more recently, there’s this amazing preprint from Broder Schmidt who describes the mechanism of action of a particular cancer drug known as oxaliplatin. So this is a so-called platinating agent, and the mechanism of these drugs is thought to be through just indiscriminate DNA damage. But yet some of these platinating agents work better for some types of cancers than others, despite the fact that all cells have DNA. And so what Broder showed is that in contrast to this, the indiscriminate damaging of DNA, what oxaliplatin actually does is messes with this layered architecture of the nucleolus, which again is that condensate responsible for making ribosomes. So here again is a drug that’s been used around the world since the ’90s that might have a condensate related or even specific MOA in contrast to how people thought it may have been working.
Bede Portz (36:53):
So a summary of the condensates 101 part of the talk is that condensates may act like integrating nodes of disease. This is attractive to Dewpoint because instead of looking at relatively small subsets of patient populations, we can perhaps look at aggregating patients and reaching more unmet need through this notion of perturbing a condensate. I’ll talk about ways in which we’re thinking about doing that. I also mentioned that small molecules can modulate condensate properties and this has now been shown by multiple groups, and that small molecules may concentrate in condensates. And we might be able to leverage this by designing drugs that preferentially partition into condensates in which their targets reside. Someone has their hand up. This is not a bad time for questions. So go ahead.
Anoop Arunagiri (37:51):
Hi, nice talk so far. This is Anoop from Michigan. So in the last two slides, you were showing that small molecules can affect the condensate properties. So how specific are they? Because you said in the beginning that it’s not necessary that the protein need to have an intrinsically disordered sequence or something. So wouldn’t the molecules affect other condensates within the cells, or why do they specifically act upon certain condensates and not others?
Bede Portz (38:32):
No, it’s a great question. So a couple of things. First, again, we’re not going to fall into the trap of IDRs equal condensates or don’t or whatever. And it’s good that you brought that up again. Secondly, the Klein paper that showed JQ1, for example, partitioning favorably into BRD4 condensates, that paper looked at a relatively narrow range of cancer drugs, and a relatively narrow range of condensates relative to the vast space of cancer drugs and as yet fully unexplored space of condensates. So the fact that there are any drugs that do that, given that we didn’t design them to do that is actually I think pretty amazing.
Bede Portz (39:14):
I think the chances of that happening, naively prior to reading the paper would be small. So there’s likely some specificity there. Broder’s paper shows it in cells in the context of an endogenous heterotypic condensate, the nucleolus. So I think this is a real phenomenon. Now, you asked about the specificity of this phenomenon. Broder hints at this in his paper, which I encourage you to read. I don’t want to spoil it, but what gets me stoked is that if we did this by accident, think of how much more specifically we could do it if we did it on purpose. You know what I’m saying?
Anoop Arunagiri (39:58):
Yeah. I’ll go through the paper. Thanks.
Bede Portz (40:02):
So I guess this is a nice segue to one or some of the things we’re trying to do here at Dewpoint, which is really translate this emerging field of condensate biology into unmet medical need. And we’re doing that with a pretty amazing collection of people. So I talked about work that are pretty major contributions to the field of condensate biology done by people who now work at Dewpoint, but those people alone aren’t enough. And so what Dewpoint has done is synthesized experts in this emerging field of condensate biology, with experts of equivalent renown in drug discovery, business development, and so on. And it’s this synergy that is going to translate this really emerging field of biology into drugs.
Bede Portz (41:05):
So we think about condensates as integrating nodes of disease. And what that in turn does is opens up the druggable proteome. It expands the druggable proteome potentially by an order of magnitude. So a really small subset of proteins are thought to be druggable by the classic definition, they have some sort of active site or allosteric site to which a small molecule can bind and modulate their behavior. But if you think about a condensate as a target, and you think about the growing number of proteins and nucleic acids that are involved in condensates, that invites us to radically expand the definition of a druggable proteome. And we’re thinking about doing this using diverse therapeutic modalities that could attack condensates in various ways.
Bede Portz (41:53):
So we could conceive of molecules that modulate the scaffolds of condensates–these are the things that are thought to bring the condensate together, modulate their compositions, say by departitioning particular components. For example, let’s extract the estrogen receptor from a Med1 condensate. We could modulate the conformation or interaction landscape of condensate proteins. Because condensates are ultimately a concentration dependent phenomena, degraders of particular condensate components could radically alter their properties. There’s likely to be myriad layers of regulation that converge on condensates, classically druggable enzymatic-type proteins that we could conceive of drugging. Based on Isaac and later Broder’s work, we could think about optimizing the partitioning of drugs into condensates in which their targets reside. And then I highlighted these examples where there’s, in some instances, a liquid-to-solid or a liquid-to-aggregate transition, and we could conceive of drugs that alter the material properties of condensates in desirable ways to elicit particular outcomes.
Bede Portz (43:10):
And we’re doing this through this collection of condensate expertise, drug discovery expertise, with a fully integrated pipeline that spans from genetics, where we identify condensates as novel integrating nodes of disease, through medicinal chemistry, where we have a team of people that can modify chemical scaffolds and drive them to the clinic. And so with that, I’ll take any questions.
Jill Bouchard (43:37):
Awesome, Thank you so much, Bede! Cool. It looks like we already have one question in the chat from Ashish. If you would like to unmute yourself and ask. Anybody at the table, you can start prepping your questions too.
Ashish Bihani (44:00):
Yeah. So in these cancer cells where this treatment is done, oxaliplatin and all these other drugs, do the control groups show aggregates?
Bede Portz (44:15):
Do the cell show aggregates? Wow, that’s a good question. So a pitfall born of this early profound result in the context of multiple neurodegenerative disease proteins that undergo this liquid-to-aggregate transition is this idea that liquids are good and aggregates are bad, but that is not a paradigm that we should embrace. And so there are probably routes to aberrant condensates that don’t necessarily result in aggregation. Likewise, there might be context where aggregation is an adaptive process the cell undergoes to prevent a worse thing from happening. And so are there aggregates in cancer cells that are treated by some of these drugs? Not necessarily, but there are multiple routes to, let’s say, pathological condensates. And so revisiting the Tamoxifen resistance example, one route to Tamoxifen resistance is the overexpression of Med1 that tunes the stoichiometry of a condensate, but it doesn’t necessarily lead it to some aggregated endpoint. So I hope that answers your question.
Ashish Bihani (45:36):
Jill Bouchard (45:39):
And it looks like we’ve got Muthu again, if you’d like to ask.
Muthu Venkat (45:42):
Yeah. Excellent talk, really enjoyed it. I have so many questions, but I will restrict myself with two questions here. One very naive question. Is there a way we could say that in disease cases there are more condensate formation or less condensate formation? Can we say something like that? Or is it too stupid to call it off like that? I’m thinking in cancers, there is more condensate formation. So if we reduce the number of condensate formation, then we make sense. Can we say that? And the second question with respect to your degraders in your two last slides. Yeah. Thank you.
Bede Portz (46:29):
So let me address this first question of, are there cases where there’s more condensates or less condensates? So there are examples. So as you move across the phase boundary in the concentration axis, you get this initial switch-like demixing, and as you add protein to the system, what happens? You get more condensates, larger condensates. So you could envision a situation where in disease, there’s this increased predilection for condensate formation that gets you, say, more condensates. And there are examples in the context of point mutations in the prion-like domain of neurodegenerative disease, proteins, where that’s probably the case, but there are actually other examples where those same proteins point mutations lead to impaired ability to form condensates. And so that would suggest that there’s probably cases where fewer condensates is bad. And I’m answering this question based on literature precedence.
Muthu Venkat (47:40):
Is there an example you were speculating what are the different ways you can control the condensates in your 3 last slides? And you mentioned degraders. So is there an example for that or is it just an idea?
Bede Portz (48:00):
So think about this simple case of an RNA-binding protein with this aberrant forward drive to form a condensate due to some point mutation. So how do you solve that problem? You could fix the point mutation. That’s hard. Or you could potentially design a drug that pushes back on that forward drive. That’s probably easier. And so this phase diagram is instructive in two ways. One is this concentration-dependent phenomenon. The other is the interaction strength bit. So if you increase the interaction strength between two molecules, say mutant version of FUS, compared to the wild type, you need less of it to get the same condensate endpoint.
Bede Portz (48:49):
So this is divorced from the material state, just to your point about more or less. And so how do you fix that problem? Again, you can’t necessarily fix the point mutation, but if you degraded the protein a little bit and use the x-axis, the concentration axis to compensate for the gain in interaction strength imparted by the mutation, that seems like a fix. And so that’s where the degrader-like concept comes in. That if you have this very switch-like behavior based on subtle changes in concentration or subtle changes in interaction strength, then it follows that subtle changes in concentration could elicit these switch-like behaviors in the opposite direction. Hence, degraders. That’s the rationale for that particular modality or concept.
Muthu Venkat (49:38):
Yeah. So I was just wondering if you want to recruit a proteasome or a degradation machinery into your condensate, but you are suggesting we could probably degrade the mutant protein so that it doesn’t get recruited to the condensate. Am I right?
Bede Portz (49:55):
I am an employee of a private company who can only answer questions to a certain extent. Suffice to say, what I’ve attempted to do today is ground diverse therapeutic modalities in this biophysical truth as defined by a phase diagram. And I want to highlight for people who might be aficionados, who are watching, that this phase diagram is an entry point to understanding condensates, but it breaks down as a function of heterotypic condensates and additional complexity. And I don’t want to mislead anyone into thinking that it’s quite this simple and actually apply that preface to much of what I said.
Muthu Venkat (50:41):
Thank you. Thank you very much.
Jill Bouchard (50:44):
And we have another question from Sangram if you’d like to ask and unmute.
Jill Bouchard (50:56):
We’ve got a mic. We’ll get you a mic.
Sangram Parelkar (51:00):
Thank you. So I really loved your presentation, but there’s one thing I’m really confused about, and that’s aggresomes and condensates. And so would you suppose condensates are a completely passive process whereby diffusion, they come together and aggresome formation is more like an active process where you might harness the microtubule network?
Bede Portz (51:23):
So I think we’re anthropomorphizing biophysics a bit here with active or passive or whatever. There are instances where a liquid-like condensate precedes the formation of an aggregate. So why might that be? Take it to its logical extreme, an individual protein cannot aggregate. It can misfold, but it can’t aggregate. That’s a cooperative process that requires a partner. So if you have demixing and you create a dense phase in which there’s a ton of that aggregation prone-protein, and aggregation is a bit of a chain reaction, that seems like a route to aggregation. But there is work suggesting that in some instances, aggregates might occur in route to a condensate, where the biomolecule has gone down the wrong path. So there’s actually a great paper from Steven Boeynaems and Alex Holehouse in PNAS that amidst the various neat findings in that paper is this idea that condensates might lead to aggregation, but aggregation could also occur in route to the formation of a condensate.
Sangram Parelkar (52:45):
I see. I was more inclined… Not inclined, but rather thinking of aggresomes, for example, like in Lewy body diseases, like Parkinson’s disease, where it’s been shown that they harnessed the microtubule network, they form the centrosome. And so that’s what I was just trying to understand, when we are looking at, say, cells, how would you really distinguish between condensates and aggresomes?
Bede Portz (53:10):
Ah, okay. That is a great question. So naively, I think of aggregates as very not dynamic, so to speak. That’s more or less an endpoint. That’s not absolutely true, but whereas I think of a condensate as more dynamic. Now, that being said, there’s probably ways to form a condensate that’s also not dynamic. So these are generalities. But aggregates, clearly something has happened there. The cell failed to compensate for that network of interactions in some way. And so what I’ve talked about is passive to use your words in so far as I’ve described one or few molecules coming together, but layer on that all the things that regulate, for example, aggregation, (chaperones, helicases, so on and so forth). So the cell is trying to maintain some balance. It’s our job to understand what the balance is.
Sangram Parelkar (54:26):
Sounds good. Thank you.
Jill Bouchard (54:30):
Awesome. A question from the crowd, the Zoom crowd, Sinem Saka. As long as you’re not in the room, you can unmute yourself.
Sinem Saka (54:42):
Sorry. I might have some background, but I was wondering, are there also examples of drugs that become ineffective when they’re applied to cells because they enrich in condensates, right?
Bede Portz (54:53):
What was that? Drugs that enrich in condensates?
Sinem Saka (54:58):
That become ineffective because they enrich in condensates. Almost the opposite of what you [inaudible 00:55:06].
Bede Portz (55:06):
Right, right, right. I wrote about this in a review paper, speculatively. So our job is to understand what the condensate is doing. So if the condensate is the site of some activity we want to impede, then yeah, you want the drug to go in there because that’s where the action’s happening. But if the condensate is instead some buffer, some concentration buffer or storehouse or inactive entity and the action’s happening outside the condensate, then yeah, you don’t want to drive a drug into the condensate. This is why I’ve tried it at times in the talk to caution us against paradigms, embracing paradigms based on n number of instances. We can’t replace my obsession with this concept with thinking, for example.
Jill Bouchard (56:11):
Yeah. All right. Any other kitchen table questions? Nobody? Bede, I think I speak for everyone when I say we loved your talk and that was fantastic. And thank you for doing that for everyone.
Bede Portz (56:26):
Thank you, folks, for your patience and attendance.
Jill Bouchard (56:37):
It sounds like at least one of the questions might want to follow up with you by email. So are you okay if we put your email up with…
Bede Portz (56:42):
Jill Bouchard (56:43):
Okay. So we’re going to put the email for Bede on the post with his video when it’s up on Condensates.com. And for anybody who wants to look at any of the sources, I think we’re also going to put some stuff up on there for that. And there’s also a new ‘For Beginners‘ section in the resources section…section within a section on Condensates.com. So look for that if you want to find some of the important beginning work and some good reviews and some other things. So thanks again, Bede. And welcome again, interns; We’re happy to have you all here and we’ll see you guys again for another Kitchen Table Talk probably in July sometime. Thanks all for coming.
4) Choi et al. Annual Rev. Biophys. (2020)
5) Li et al. Nature (2012)
6) Portz and Shorter. JMB (2021)
7) Riback et al. Nature (2020)
8) Mitrea et al. Nature Communications (2018)
9) Riback, Katanski et al. Cell. (2017)
10) Lu, Portz and Gilmour. Mol. Cell. (2019)
11) Isaac and Churchman. Mol. Cell. (2019)
12) Soto et al. Nature Neuroscience (2003)
13) Patel et al. Cell (2015)
14) Mackenzie et al. Neuron (2017)
15) Mann et al. Neuron (2019)
16) Jawerth et al. Science (2020)
17) Kim et al. Nature (2013)
18) Hnisz et al. Cell (2017)
19) Boija et al. Cell (2018)
20) Klein et al. Science (2020)
21) Shmidt et al. Biorxiv (2021)