VIDEO: Abhinav Diwan on Condensates in Cardiac Myocytes – Effectors of Physiology and Therapeutic Targets

Condensates.com welcomed Abhinav Diwan, MD, FACC, FAHA, Chief of Cardiology, Saint Louis Veterans Affairs HealthCare System, Charlie W. Shaeffer, M.D. Professor of Medicine to deliver our September Kitchen Table Talk.

Title: “Condensates in Cardiac Myocytes – Effectors of physiology and therapeutic targets”.

Abstract: Condensates are increasingly recognized as critical membrane-less organelles that compartmentalize vital cellular functions. Cardiac myocytes are long-lived cells that contract throughout the human lifespan to generate cardiac output. The contractile unit, i.e. the sarcomere is a dynamic protein assembly, reminiscent of a membrane-less organelle that forms through phase separation, capable of self-repair and self-renewal without interrupting contractile function. The Diwan lab is interested in understanding the role of condensates in sarcomere homeostasis. Our work has focused on chaperones and co-chaperones in sarcomere protein quality control. We have discovered that CRYAB, a highly enriched heart shock protein in cardiac myocytes, can form condensates. Disease-causing mutations and stress-induced phosphorylation affects CRYAB condensate properties to induce a ‘condensatopathy’. In ongoing work, we are charactering the nature of these interactions to understand how chaperones promote sarcomere homeostasis.

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

TRANSCRIPT

Diana Mitrea: Hello, everybody, and welcome to Kitchen Table Talk number 40, today featuring Dr. Abhinav Diwan. I’m Diana Mitrea, your host, joining from Dewpoint’s Nucleolus conference room in Boston, Massachusetts. Pavithra Aravamudhan, Associate Principal Scientist is our co-host today. I know you’re all excited to learn about the roles of biomolecular condensates, and how they play into the proper functioning of the heart and how they contribute to disease. 

But before we get to the fun part of today’s talk, I have a few housekeeping announcements. So, as you know, this is a live seminar, and it’s being recorded. You’ll be able to see it after the talk, within a week or so, posted on Condensates.com. So, for that reason, please keep your microphone muted. You are welcome to keep your camera on, but no, on-camera shenanigans. Feel free to type your questions as they come to you in the chat, and also you can also raise your hand during the Q&A breaks. 

Abhinav kindly agreed to stay a few more minutes if there’s a heated discussion at the end, but in case you have any questions that come to you after the talk, always feel free to submit them through the Contact form on Condensates.com. Whenever you have questions, Pavi and I will prompt you to unmute your microphone and ask a question. With that, I will pass the baton to Pavi to introduce today’s speaker. 

Pavi Aravamudhan: Thanks, Diana. Good morning, everyone. I’m delighted to introduce Dr. Abhinav Diwan. Dr. Diwan earned his medical degree from the AIIMS Institute, which is renowned in India, and completed his internship, residency, and a fellowship in cardiovascular science at Bayer College of Medicine. Dr. Diwan currently serves as Chief of Cardiology at the St. Louis VA Medical Center and holds the Charlie Schaefer Professorship of Cardiology at Washington University School of Medicine. He’s a tenured Professor of Medicine with secondary appointments in cell biology and physiology, neurology, and obstetrics and gynecology.  

Work from Dr. Diwan’s lab has defined how the interplay between autophagy, lysosome, and mitophagy safeguard cardiac proteostasis, and findings from this research have opened translational avenues for heart failure and ischemic injury. Building on this foundation, his group now interrogates biomolecular condensates and phase separation as drivers and drug targets of cardiac remodeling. Recent highlights include work published earlier this year that Dr. Divan will talk about today in the Journal of Clinical Investigation, showing that phosphorylation of CRYAB induces a condensateopathy that exacerbates post-myocardial infarction remodeling, and this sharpens our lens on condensates as actionable biology. He has helped shape the sarcostat concept. This is how protein assemblies and condensates preserve myocyte function. 

Dr. Diwan’s work and service have been recognized widely, including highlights of election into American Society of Clinical Investigation, Association of University Cardiologists, and Alpha Omega Alpha. Today, he’ll speak on condensates in cardiomyocytes, outlining how modulating these dynamic assemblies can restore contractile integrity in heart disease. Please join me in welcoming Dr. Abhinav Diwan. 

Abhinav Diwan: Thank you, Pavi. That was a really kind introduction, and I hope everyone can hear me clearly. For the talk today, I’ll focus on how condensates are effectors of physiology and could be therapeutic targets in cardiac myocytes. And by way of introduction, briefly, so I’m a physician-scientist, and my interest has been in cell biology of age-related diseases, which includes cardiomyopathies, heart failure, Alzheimer’s disease, diabetes. All of these have a common, a shared set of risk factors. And it’s interesting, it’s the same human that develops heart disease, Alzheimer’s disease, diabetes, so we have artificially separated ourselves into specialists, but understanding the common cell biology can potentially prevent or treat a variety of these diseases at the same time. And that’s why we focused on lysosome as an organelle and acquired lysosome dysfunction, so our lab has primarily published in that area. 

It’s thanks to Rohit Pappu here at WashU and others that I got inspired to think about condensates, and it makes a lot of sense, as you will see during the talk. I do not have any particular disclosures. I put the cover from the journal on the slide here, we were very proud that we put condensates out there in a very common cardiac condition. So I’m going to talk a bit about that. And Moydul Islam, who’s a graduate student who led this particular work, he’s on this meeting as well, so Moydul, welcome.  

And, let me now go through the slides. I’m going to present the sarcostat as a conceptual framework. Then I’ll share with you a literature review on what is known about condensates in the cardiac field at this point. Then I’ll talk about our work on ischemic cardiomyopathy as a condensateopathy and tell you about some unpublished work on condensates in dilated cardiomyopathy. 

This is an electron micrograph of an adult mouse heart. An adult human cardiac myocyte looks very similar, so there are shared features between mammals. And what you see there most prominently are sarcomeres. So, as you will see here, these are the sarcomeres, and you’ll see these are Z-discs, M-lines and actin and myosin filaments there that help the muscle contract. I want you to appreciate the architecture, see how the mitochondria are tightly packed in between cardiac myocytes. And the reason was thought to be that they provide energy, they’re right next to the cardiac myocyte, but this theme will continue to play out. 

This is a very intricate and complex piece of machinery, so a lot of support systems need to be right there, next to this. It’s like if we had a critical industry here in St. Louis, and if something were to break down, we don’t want to be sending a message to Boston to fix it. We need everything available right here to be able to fix it, and that’s what a cardiac myocyte does beautifully, as you will see. 

We’ve focused on protein quality control, and lysosomes are what we focused on, but I put arrows there to show you a few lysosomes on this picture. And they are these dark osmophilic structures that you see. But I think you’ll appreciate that given the mass of proteins that’s out there what small fraction of the area is occupied by lysosomes. So perhaps there are other organelles that participate in protein quality control that we don’t appreciate. Some of our work has gone in that direction, and I’ll tell you about it. 

So, what is a sarcomere? I’ve thought of the sarcomere as a condensate. Is it a crystalline or paracrystalline structure, and perhaps this is how these proteins achieve the lowest free energy state, by coming together in this intricate molecular machinery? 

And shown here is a schematic from a review article giving you the scale, about a 14-centimeter heart size you can imagine in an adult human. A cardiac myocyte is 100 microns in length, so these cardiac myocytes are attached end-to-end in series. Multiple myotubes are attached in parallel in this muscle fiber, and these muscle fibers are wrapped inside the heart in various orientations. And they’re either circular or they’re wrapped in a spiral fashion. So that allows the ventricle to not just squeeze but also wring. The ventricle twists on itself, and that allows the hemodynamic force to efficiently push the blood out. 

So, as I explained to my patients, the ventricle, the heart, does two things. It fills, and it empties. And it turns out that both of these functions are very important, and abnormalities in either one of them can result in heart failure. So, for the purpose of this talk, we’re going to focus more on the squeezing function of the heart, which is contraction. But relaxation of the sarcomere is a very critical aspect as well. And here’s the sarcomere depicted. It’s a structure that’s quite invariant in length, between 1.8 to 2.2 microns, and all adult sarcomeres are roughly the same size. And as you see here, there are thin filaments and thick filaments, and Z-discs are a key site of action where proteins are anchored.  

And I’m going to bring in another schematic here to zoom in to the area of the Z-disc. And on the Z-disc, there is a wide bevy of proteins, some of whom are recognized to have intrinsically disordered regions, and I’ll show you a study for how Z-body formation was linked to condensate formation. 

But you see here how actin filaments are inserted into the Z-disc and actin constitutes five proteins – actin, tropomyosin, three proteins of the troponin complex. And the troponin complex responds to calcium, which allows for the inhibitory proteins to move out, and the actin to be available to bind to myosin. Myosin is the key protein in the thick filament. There’s also myosin binding protein C and titin, and we’ll get to titin in a minute. 

But this myosin is the motor, and myosin heads interact with actin, break down ATP, which allows for cross-bridge cycling, and this is what allows the sarcomere to contract. And that’s how the muscle achieves its function. But you need ATP to break these cross bridges, and that is why, in energetic insufficiency relaxation, which is breaking of cross bridges, is the first thing that is impaired, which is why diastolic function or relaxation happens to suffer early in any cardiac pathology, even before systolic dysfunction. 

Some of the proteins are listed here, and their roles and diseases are listed here. We’ll talk a bit about that as we move forward. On the next slide, we’ve put a movie that we generated to give you an idea of the large spaces in the cardiac myocytes. It’s a huge structure, and on the right here is the nucleus, and you see the myofibers, and you saw various organelles floating in there. 

So, just conceptually, if you imagine that Brownian motion was somehow going to allow this large structure to function, I think that’s asking for too much. And that is where we believe that there is a dark matter to this, which is made of condensates. And we don’t see it. Condensates organize the function within this large spatial cell type. And some of the evidence has started to come about of why that might happen.  

Then taking it to the next functional concept, which was briefly introduced. The sarcostat, we posited in a paper from Moydul as a review article back in 2020, where we conceptualized that all key pieces of this machinery, which is protein synthesis, repair and degradation have to be situated right next to the sarcomere, and we call that a sarcostat. And sarcostat is a concept. We believe that this entity exists right next to each sarcomere to allow for optimal functioning. And why is that important? Because it turns out sarcomeric proteins have short half-lives. 

The largest sarcomatic protein is cardiac titin, and titin is, like, 1.4 megadalton, I believe. It’s the largest protein known in humans. And the half-life of titin is estimated at 14 days. Nature wishes to have the highest quality of proteins at the cost of turning over even these huge proteins. So, we believe the sarcostat does that right next to the sarcomere. And condensates are a very logical concept, as if for membrane-less dynamic compartmentalization to allow the sarcostat to exist and function. 

With that, I’m going to segue into cardiomyopathy. Cardiomyopathy, by definition, is weakness of the cardiac muscle. And typically, that weakness is assessed as a contractile defect. But in the background, there is a relaxation defect, and if anything, the relaxation defect precedes the contractile defect. Listed here are various causes. So, cardiomyopathy is genetic, or inherited, and congenital, so that can be both in the genetic etiology or acquired. And in genetic cardiomyopathies, the typical ones, over time, there have been many iterations of this nomenclature, so to speak. The most recent one is this Padua classification that addressed it as dilated or hypokinetic, hypertrophic or restrictive, and scarring or arrhythmogenic. And these are based on the phenotypes of the heart. In dilated, the ventricle is dilated, the walls are typically thin. In hypertrophic cardiomyopathy, the heart size is not enlarged, but the walls are very thin and there is a contractile defect. In arrhythmogenic, there is scarring, there is replacement of the cardiac muscle with fibrofatty tissue. Then, also listed are a variety of other causes which are typically excluded from these definitions, but ischemic heart disease is a big one. Up to two-thirds of all cardiomyopathies have an ischemic component. So up to two-thirds of all causes of heart failure in the world have an ischemic component, so that is perhaps the most common etiology leading to cardiomyopathy.  

And I’m showing you this schematic from Ray Hirschberger from a review article. This is about 12 years old, so there may be a couple of new genes added to it, but I want you to appreciate that much progress has been made in linking genes to these different phenotypes. And as you can see, these genes span a wide variety of proteins, from the sarcomere to heat shock proteins. The big one here is BAG3, a co-chaperone protein, proteins in the nuclear lamina, enzymes, including metabolic enzymes, and ion channels. So, a wide variety of protein dysfunction in the cell can lead to a contractile defect in the cardiac myocyte. This is why cardiomyopathy and heart failure continues to be a challenging field. 

But what I also would like to emphasize is that all the drugs, the medicines that we have to treat heart failure right now do not treat the sarcomere. There are some drugs that have started treating the sarcomere in hypertrophic cardiomyopathy, for example. That is to change the myosin active state of the myosin head, which allows for reducing contractility in hypertrophic cardiomyopathy. Other drugs have come about as calcium sensitizers or, again, myosin head activators, but these are early in clinical development, and the studies that have been done are not overwhelmingly supportive of their use in terms of preventing bad outcomes like mortality. This is a huge area of unmet need, where we do not have drugs that directly target the sarcomere to treat cardiomyopathy and heart failure. So, some of our interest in condensate biology is clearly linked to finding such drugs. 

Let me briefly go through a review of the extant literature. The first paper I want to highlight is from Michael Gotthardt’s work, where they looked at titin’s life cycle. They very cleverly engineered mice that had titin tagged with a dsRed fluorescent protein at the end that pokes into the Z disc, and titin attached to GFP at the other end, where it attaches to the M line, which is in the middle of the sarcomere. And they generated these double heterozygous animals, and they looked at what, co-localized with these. 

Here’s the first image that shows titin on Z-disc in red, this is titin on the M line. And when they looked at ribosomes, the ribosomes correlated with titin on the Z-disc. And there’s subsequently multiple papers from Ben Prosser’s group and others that have shown that ribosomes and mRNAs are trafficked right next to the Z-disc. 

There is a component of the sarcostat that allows for local protein translation. How is it organized? Again, we speculate it is condensates, but data are lacking in that regard. Here’s now looking at the proteasome. When you look at the proteasome, again, it co-localizes with the titin at the Z-disc. 

So this is the degradation aspect. We know lysosomes also sit right next to the Z-disc. Again the speculative concept of the sarcostat is being borne out by the experimental evidence. Then, in this schematic that I cut and pasted here, they looked at how the two ends of the titin are synthesized and how are they put in place, and they found out that it happens in a stochastic manner. 

Earlier, it was thought that all synthesis started at the Z-disc and then it extended to the M line. That is partially true. It turns out there is a pool of soluble or free-floating titin there that can be integrated into the place where it needs to be. Conceptually, if you looked at a mature sarcomere, you can imagine this huge protein that is running from the Z-disc to the M line, so a 1-micron long protein. And you can imagine it being synthesized in pieces, and those pieces are there insoluble, so when a certain portion of the titin is damaged, it can be replaced. How is that orchestrated? I would posit that condensates have something to do with it. 

But again, this is early work that starts to get at that concept. The other paper, again, from Michael Gotthardt’s group, was another amazing discovery, where they tagged a biotin ligase at the Z-disc end of titin, and they looked at what proteins would come out in an unbiased proteomics approach. You all can look at the paper in detail on the publication, but I want to cut to the chase and get to the main idea from the paper. What they found was that proteins that do proteostasis, ubiquitin proteosome system, lysosomes signaling proteins, metabolic enzymes, they’re all located right next to the Z-disc. So again, another pointer to how the sarcostat might be organized, and I would posit that condensates have a role in doing that, but all data need to be generated to support that. 

So with that, we’ll segue to this really nice paper from Ben Sabari’s group, looking at how condensates might drive cardiovascular cell specification. And they looked at myocardin, which is a co-activator that does not have transcriptional activation activity itself, but it binds with SRF. And it brings RNA polymerase II and P300 to start transcription of sarcomatic genes and allows for differentiation of the cells into a myogenic cell line. It can go either cardiac myocyte or skeletal muscle. And what they found was that during early differentiation of these cells towards muscle cells, there is an increase in concentration of myocardin. And myocardin turns out to have an intrinsically disordered region in the C terminus. That is in the transcriptional activation domain, where it binds to SRF as well as RNA polymerase II and P300. And what they showed was that myocardin can form condensates. And if you remove this domain that allows for it to form condensates, the muscle cell specification does not occur. The differentiation into muscle cells does not occur. They could then replace this by the IDRs of other completely unrelated proteins. And that was able to restore the condensate formation, as well as muscle cell differentiation. Again, pointing to how structure begets function. 

But there is a need for myocardin to bring together this transcriptional complex. An interesting aspect of this paper was they found that when they put an IDR that is not very rich in aromatic amino acids, so here’s a display of the amino acid composition of these various IDRs, and as you look at this orange color, which is aromatic, if you look at CDT1 IDR, it’s really not that rich in aromatic amino acids. And this did not confer condensate formation, or transcriptional activation. So that was a key finding, that you need automatic amino acids. Getting to what specific molecular grammar might be needed in this instance. 

Then I’ll skip quickly to another paper, and this was an interesting observation. This was early in 2021, where they looked at this protein FATZ, which is there in Z bodies and Z-discs, and it has a long name based on its different interactors. But they found that FATZ protein has IDRs, and it can form condensates, and FATZ is responsible for bringing together various elements of the Z-disc. So that’s shown in the schematic. There is alpha-actinin, it binds to actin, telothonin, CAP-C protein that caps the actin, and titin. And FATZ protein has to bring it together. And they found that the IDR region of FATZ was essential for this to happen. And they showed that the interaction between FATZ and actin was in these droplets, the condensate, and by changing the arginine concentration, they could modulate this interaction. 

This was early work hinting at how molecular grammar might be critical in formation and resolution of these condensates. And that’s a very interesting point, because what they showed in this paper is that the condensate, or the ability for FATZ to form condensates is important to bring alpha-actinin to the right location. But then, with increasing concentrations of actinin, it was important for the condensate to dissolve as well. This showed that how a dynamic compartmentalization was critical for bringing actinin to start forming the Z-bodies. This is where the Z-discs are synthesized. So again, a very elegantly done paper. More work along these lines needs to be done to understand how sarcomere biogenesis might happen.  

Now shifting to another publication. This one, I’m sure folks at Dewpoint are very familiar with. This was looking at RBM20 and RNA-binding protein mutations in which are implicated in human cardiomyopathy, and this was very nice work in a gene-edited pig model that showed that when these RBM20 mutations result in mislocalization of RBM20 to the cytosol, where being an RNA-binding protein, it traps transcripts in these condensates, and that creates a sarcomere deficiency situation, because the proteins cannot be translated. So that’s one mechanism. So this was an example of an abnormal condensate, a condensateopathy-driving disease. 

This was another paper that I put in looking at a protein called HIP-55. This turns out to be a negative modulator of beta-adrenergic signaling. So, persistent beta-adrenergic signaling is deleterious in heart failure. Acutely, beta-adrenergic signaling is critical for the fight-or-flight response. But in heart failure and cardiomyopathy and heart failure, beta adrenergic signaling is persistently turned on, and that causes cardiomyopathy. That’s being directly implicated, co HIP-55 is a protein that down-regulates, and they found that AKT-mediated phosphorylation of HIP-55 was essential in resolution of these condensates, which prevented persistent beta-adrenergic signaling. This was, again, an interesting concept of how post-translational modification might affect condensates affecting cardiac function. And some of our work has also shown that.  

I’ll keep moving, and folks who are interested can look at these papers. This was another very recent observation looking at how mineralocorticoid receptor condensate formation might be triggering contractile dysfunction in donor hearts. As we all know, donor heart supply is one key limiting factor in cardiac transplants based on some policy changes on cadaveric donor hearts being available. The total number of transplants that had peaked at a ceiling of 2,500 in the last decade are now up to 4,000-something, because there is somewhat more availability of donor hearts. But still, there are many, many more patients with advanced heart failure who need transplants where donor hearts are not available. And there are many instances where, because of the short time window, the donor heart that is procured is no longer ideal for transplantation and ends up going wasted. 

And a key component of that is the cryopreservation injury. In this elegantly done paper, they described how the mineralocorticoid receptor, which has intrinsically disordered regions in the N-terminal domain, how it might be forming condensates in the hyperosmotic environment, which happens with donor heart cryopreservation together with hypoxia. So these are experiments that suggested that there are two pathologic drivers of this mineralocorticoid receptor condensation. It’s hypoxia and it’s hypertonicity, and both of these cause the mineralocorticoid receptor to form condensates. And they showed how this affects transcription of metabolic enzymes, and how that negatively affects the donor heart function. And in an elegant experiment, they showed a mineralocorticoid receptor antagonist, canrenone, was sufficient to inhibit condensate formation in this setting, and treatment with canrenone improved donor heart function, and that could be a clinically applicable way to salvage more donor hearts for transplantation. So again, an elegantly done study implicating condensates in myocyte biology. 

With that, I’m going to segue over to our work, and a lot of it is published, so I’ll walk through the key findings. We got interested in the role of small heatshock proteins and co-chaperones, including BAG3. Why invoke condensate biology? As it turns out, there are about 180 dedicated chaperone proteins in humans, and as we know, 20,000 or more proteins that they need to work with. There has to be some degeneracy that allows one chaperone to work with multiple client proteins and multivalent interactions through IDRs offers that conceptual framework for how a chaperone protein might be able to work with so many different client proteins. 

But working with them and folding them also needs compartmentalization. Again, invoking Brownian motion and random occurrences for protein folding to happen, I think, is expecting too much. You would speculate that membraneless compartmentalization is the ideal setting, where the chaperone can bring its client, fold it, and then perhaps release it. So with that notion in mind, we looked at crystalline, and shown here in the schematic is, again, a view into a cardiac myocyte, but I want to highlight a key protein here, Desmin. Desmin is this protein in yellow; it’s an intermediate filament protein that scaffolds everything in place. If you go back and remember an earlier slide showing the electron micrograph, you see how everything was so beautifully arranged. The myofibers with the mitochondria packed in between, it turns out Desmin provides that scaffold. And crystallin, or alpha-B crystallin, which is the most abundantly expressed protein in the heart. It’s been estimated that 3-5% of the weight of the heart is crystallin. So it’s that abundant a protein. It’s also abundant in the eye, in the skeletal muscle. Mutations in humans lead to cataracts, autosomal dominant cardiomyopathy, skeletal myopathy.  

Fortunately, these are rare causes of mutations, but as I’ll show you, we’ve exploited these mutations to understand the effects on condensate biology of crystallin. These are the domains of crystallin, so there is the alpha crystallin domain, which I’ll show you on the next slide, how it forms an ordered structure. But then the N-terminal and C-terminal domains do not form an ordered structure. When we looked at the bioinformatics approaches for intrinsically disordered regions, there are such regions present in all three domains. I’ve highlighted two key residues here, aspartic acid at position 109 and arginine at 120. These form a salt bridge. In humans, autosomal dominant mutations in either one of these by changing the charge causes cardiomyopathy. 

We also focused on this one serine. There are two other serines here, and I’ll walk you through that. This is a stress-induced phosphorylation event at this serine, and we found how this affects the condensate properties of crystallin. Here’s the amino acid composition. Again, there is that molecular grammar present there that most of you are familiar with that might drive condensate formation. So, to put things in perspective, crystallin dimerizes, and this is the alpha crystallin domain, and you see here the aspartate and the arginine interact through salt bridges, and it forms the groove in between, and serine 59 is within that groove. Later in the talk, it will become clear how that groove is important. But the N-terminal and C-terminal portions of crystallin are not shown here, that fly into this groove, so to speak, and they modulate the multimer composition. So, this can exist in a 32-mer or 64-mer, and this is how this is thought to chaperone other proteins. 

And I’ll be showing you a bit about how this might affect the condensate biology. To see if crystallin can form condensates, we adapted the optodroplet system. And here’s the CRY2 protein only. I think most of you are aware, CRY2 being an Arabidopsis protein, which is light-sensitive protein, it homodimerizes, so it increases the chances for proteins to come together in condensates if they have a propensity to do so. But CRYT2 by itself does not form condensates. 

Here’s the positive control, which is the fuse protein. As we know, mutations in fused are implicated in neurodegenerative conditions. Here’s CRYAB. And again, this was work that Moydul did in the optodroplet construct, and you see CRYAB can be induced to form condensates. Shown here are all three regions of crystallin in this light-induced optodroplet system, and you can see all of them can form condensates in the right conditions. We did purified protein studies to validate all of this. 

But then we were interested in the human mutation, and the role of how phosphorylation might affect it. So here’s the human mutant, first of all. So that movie is playing. Crystallin has come together in these large aggregates, and you do not observe any light-induced dynamicity. When you look at serine 59A, a phospho-deficient construct, it, again, affects condensates formation, but you still see molecular movement in this video when you change it to a phosphomimetic S59D. This movie is playing, but there really isn’t much motion, and there really isn’t much new condensate formation with light. 

But when we hypothesized that hyperphosphorylation at serine 59 was driving the pathology in the R120G mutant, we generated a mutant where this serine could no longer be phosphorylated, and as you can see, it restores some dynamicity to that R120G mutant crystallin molecule. So, we subsequently evaluated the molecular motion using FRAP, and again, this work is published. 

But I want to draw your attention to this wild type, where you see recovery of the fluorescence within the condensate, and here’s the R120G. So wild-type here is shown in blue, the recovery of fluorescence, the R120G is in red. When we make that serine non-phosphorylatable, we restore dynamicity. When we change that serine to a phosphomimetic serine, it’s no longer dynamic. So this led us to conclude that phosphorylation at this site changes the behavior from a liquid-like to a gel-like biophysical state within the condensate. And we focused our attention on myocardial infarction. So, you recall I mentioned that up to two-thirds of all heart failure is driven by ischemic etiology, typically in a human. This is due to atherosclerosis, which is, you know, deposition of atherosclerotic plaques within coronary arteries. And typically, a plaque rupture causes a clot to form, and the area of the muscle that’s supplied by that blood vessel dies. The best therapy we have at this point, if someone’s having myocardial infarction, is to rush them to the cardiac cath lab, put in a cath through a catheter, put in a wire and a balloon to blow open this narrowed area, and then put in a stent to keep it open. So that’s the best therapy we’ve got. 

But that introduces reperfusion injury on top of ischemia. So despite the state-of-the-art development, which has been around for decades now, we are unable to salvage myocardium to the degree that we would like to. And this myocardial infarct, if it is beyond a certain size, then the myocardial infarct triggers what’s called adverse ventricular remodeling. And this is depicted on this slide, which is driven by multiple factors. 

But it leads to cavity dilatation of the ventricle, wall thinning, and systolic dysfunction, which is depicted here as a decline in ejection fraction. So this is typically how a human post-MI might progress from an asymptomatic state to symptomatic heart failure. We can model that in mice. First, we looked at myocardium from human patients with ischemic cardiomyopathy, where the ventricles are dilated and dysfunctional. And we focused on sarcomeric or sarcomere-associated proteins. I think it’s very evident that the sarcomere structure is disrupted. These proteins are now in aggregates, and they’re no longer in the striations. 

We focused on crystallin because crystallin is the chaperone for these proteins, and we saw a hyperphosphorylation of crystalline at serine 59. And again, all of this work is published, but we then generated mice where we created a knock-in of a phosphorylation-deficient mutant and a phosphomimetic mutant. Turns out, these mice had normal cardiac structure function by echocardiography as a young adult. So we did myocardial infarction modeling and showed that if the serine could no longer be phosphorylated, the post-MI ventricular dilatation and dysfunction was attenuated. So, suggesting that phosphorylation of crystallin at this site was creating abnormal condensates and was responsible for the post-MI cardiomyopathy. But as it turns out, when you look at their myocardium, the phosphomimetic mice bearing homozygous alleles for the phosphomimetic clustering have increased polyubiquitinated proteins and p62 in aggregates, and you can see these aggregates on here. 

And as it turns out, these are images from sham and ischemia reperfusion-treated hearts. And again, you see normal Desmin architecture in the sham for wild-type, and the phospho-deficient, but it is abnormal in the phosphomimetic, with accumulation of polyubiquitinated proteins. And we demonstrated that phosphorylation at this serine changed how tightly crystallin interacted with Desmin. A phosphomimetic version of crystallin increased the binding strength, versus a phosphorylation deficient reduced it. 

I’ll quickly also mention that we used 25-hydroxycholesterol, which was previously found in a screen, to dissolve cataracts with this R120G mutant protein, and it binds to this groove in the crystalline molecule. This is what I showed you before. It normally interacts through soft bridges with that arginine that was mutated, but it can also interact with other arginines. 

And this was shown through a molecular dynamics simulation and other biophysical techniques, that 25-hydroxycholesterol stabilizes the driver. We found that it reduces aggregate formation with this aggregate-prone version of crystallin. It also restores the condensate size and number and also increases the dynamicity of these condensates. So, in a way, we found 25-hydroxycholesterol attenuated the condensatopathy that we observed. So this was sort of the schematic that we proposed where phosphorylation of crystallin would be a tool to allow for increased affinity to bind to its client proteins and physiology. 

Akin to the stress granule hypothesis, that under stress, the phosphorylation of crystallin allows it to bind more tightly to its client proteins, but when the stress abates, the phosphorylation should go away. The crystallin should now get dephosphorylated, so it restores the original binding affinity, so it allows for modulation of crystallin’s binding affinity to the clients. But under stress, this phosphorylation is persistent. And whereby these clients are now sequestered in these condensates, or aggregates. And that is why this results in being a condensate topic. 

So, I think in the interest of time, I’ll come to this one slide of one published work where we have focused on BAG3, and mutations in BAG3 are implicated in causing cardiomyopathy. And BAG3, it turns out, has domains through bioinformatic approaches, which are IDRs. So we’ve modeled in optodroplet construct. And as you can see, BAG3 readily forms condensates using this optodroplet construct. And here, we only use the N-terminal domain, but we’ve made similar observations with full length BAG3. But here’s a mutation in humans, proline to lysine at position 209, and this causes protein aggregate cardiomyopathy, which is restrictive in phenotype. 

And you see how this mutant behaves. So it is already forming these large aggregates, and the dynamicity is reduced. Then we did FRAP analysis on this. This is how wild-type BAG3 condensate recovers, and you’ll notice that the mutant BAG3 recovery is slower. And we’ve quantitated that, so in the interest of time, I’m just showing you the key pieces of data. We also looked at a different mutant, which is E455K, so this is not implicated with forming protein aggregates, but it also leads to cardiomyopathy, because it is in the BAG region, which is where BAG3 interacts with HSP70, and the other co-chaperone complex. 

And this still allows for condensate formation, but in early experiments, we’ve noticed maybe the number of condensates are reduced. And then we performed this other experiment to look at how BAG3 might be interacting with heat shock proteins. In this experiment, the optodroplet is generated with the N-terminal of BAG3, but GFP-tagged crystalline is expressed together with it. And when we induce optodroplet formation with BAG3, crystallin comes along in many of those condensates. So this suggests that this is how this protein quality control machinery might be put together within condensates. 

I’m going to summarize and be happy to take questions. I think what we have learned is that by examining the mechanisms by which these proteins become aggregate-prone, that is perhaps a better approach to target disease pathology than focusing on aggregates. And condensates could be therapeutically targeted for that purpose. 

So that is my lab. You know, again, Moydul is pictured here, and Moydul is now doing his postdoctoral training at Boston, and he led this work along with multiple other members in the lab. I want to thank our collaborators, in particular Rohit, for guiding us through this work, and our funding agencies, and happy to take questions. 

Q&A 

Isaac Klein: Abhinav, thanks for a great talk. Question about Desmin. Have you examined what happens to crystallin condensate formation upon mutation of Desmin, which occurs in some percent of genetically defined DCMs? 

Abhinav Diwan: That’s a great question. We haven’t experimentally approached that. I would postulate, because many of these Desmin mutations lead to aggregate formation as well. So, I would postulate that Desmin might be altering the condensate behavior of crystalline and the other components. I’m sure these are multi-component condensates and, relevant questions are about, what is the chicken and egg, what brings it together to begin with. But I suspect once all these pieces are put together, mutations in the individual components can change either the entire condensate or zones within the condensate that might then encourage homotypic interactions and allow for protein aggregation rather than the dynamic function. 

Federica Accornero: Thank you so much, a beautiful seminar. I was wondering if you know or can speculate if RNA is necessary for crystallin condensation? 

Abhinav Diwan: That’s a great question. We haven’t detected RRMs within crystallin, but I wouldn’t be surprised if it binds RNA, and it may not even directly bind RNA. There may be another component in that condensate. So, I think we’re inspired by the LiTEC innovation where now we are going after probing the components of these condensates. But we want to do it in a cardiac myocyte setting, because doing it in a non-myocyte cell type, I don’t think is that informative. So, we have generated IPS-based cardiac myocytes. But as you may know, the IPS cardiac myocytes is a very immature system. So, ongoing work is trying to make these into cardiac muscle. So, together with our collaborators, we can fashion cardiac microtissues. But again, cardiac microtissues are also an immature form of muscle. And the key there is to try and get all cell types and try to mimic the native muscle architecture. So far, we can put fibroblasts and immune cells in, but developing blood vessels and then putting flow in there, I think, is a whole order of innovation above. But, so at this point, we’re going to target microtissues with the LiTEC system. So it’s still relatively immature, but hopefully we’ll be able to probe crystallin condensates using affinity tagging approaches, such as the APEX system, which would then allow us to pull out proteins and nucleic acids. But I’m as curious, as you mentioned, about the role of nucleic acids in organizing these condensates. 

Diana Mitrea: That was a beautiful talk. I was wondering, so at the beginning, you, stated the hypothesis of there being a sarcostat, and then, you showed that really impressive map of mutations that are associated with the cardiomyopathy. How many of those genes are likely to be colocalized in the sarcostat or at least regulated by similar signaling pathways? 

Abhinav Diwan: Yeah, that’s a great question. There’s someone in the lab undertaking that effort. The short answer is no, we haven’t done it, but we want to do it. We want to put that together using bioinformatics approaches, to predict, you know, what may form condensates. But as we all know, bioinformatics approaches have limitations, and you need to experimentally test those, but we do want to generate a framework, at least to then start building experiments around that. It’s going to be very complicated, as you can see. There are so many proteins within the Z-disc itself. And then if you start layering in components of synthesis machinery, degradation machinery, repair machinery, I mean, all of a sudden, now you’re dealing with a large number of proteins. I’d be curious to get others’ views on unbiased approaches as well. My thoughts were to take a more focused approach. But I’d love to hear about unbiased approaches that people might suggest of how to go after the sarcostat condenses. 

Diana Mitrea: I think just starting by looking at the protein-protein interaction maps and the signaling maps would be a good starting point. 

Isaac Klein: Are there protective mutations in crystallin that could be hypothesized to alter its propensity to condense?  

Abhinav Diwan: Great question. So, it’s not been described as a human modifier mutation, for example, that was beneficially altering the course of disease with another different mutation, but that’s hypothetically possible. But to your point, we’re working with Rohit to be able to design the grammar crystallin and then understand how that might modulate the aggregate prone. So, take the R120G mutant and introduce, you know, designer changes in the grammar, that will need to make sure they don’t affect the chaperone function. But at the same time, if they can restore dynamicity to that condensate, that could be a therapeutic approach. 

Isaac Klein: The reason I ask is I believe there are protective mutations in BAG3 that have been well described. That might be an interesting tool to study its role in affecting BAG3 condensation as well.  

Abhinav Diwan: That’s a great suggestion. So, we’re also, you know, doing the LiTEC approach to probe native BAG3 condensates, and I’m aware that there is a lot of interest in this area, so I’m sure data will come about from multiple sources to understand this better. 

Laura Behrendt: So, which kinases phosphorylate crystallin? 

Abhinav Diwan: There’s a whole bunch. So, the ones I showed you are from the MAPK family, then there is protein kinase N. There’s a whole bunch of serine threonine kinases that can phosphorylate. People have probed the functional effects of this phosphorylation. The S59 locus has been the most hotly studied, but that does not exclude the other loci being important as well. But I think the key message that we learned is that these phosphorylation events need to be dynamic to maintain physiology. Once you make these events sort of less dynamic, or you have, you know, stress-induced activation of a kinase persistently. So, for example, the p38 kinase that phosphorylates serine 59. In heart failure, it is known that p38 is persistently activated. And using mouse modeling studies, it’s very clear that if you remove p38 in an acute stress setting, you make things worse, because you need that stress kinase to function through its phosphorylation activities to allow for an adaptive response to the stress. But the response becomes maladaptive when it’s persistently turned on, and in that setting, removing or inhibiting p38 is beneficial. So it’s like most other things in biology, there is a duality to understanding the roles of these proteins and kinases. 

Laura Behrendt: And any idea on the phosphatases that might dephosphorylate crystallin again? 

Abhinav Diwan: Great question. We haven’t gone down that route, but that’s a very interesting question. Because there are so many kinases, so many phosphatases, how are they brought together in that functional setting, right? Because you would presume, just like in the stress granule, right, that phosphorylation of crystallin increased the affinity to bind to the clients, but then a phosphatase must kick in when the stress goes away. So, we’re hoping that by the LiTEC approach, probing the components of native content sets might shed light on which ones, but we don’t know that yet. 

Diana Mitrea: Thank you very much. Join me in thanking our speaker, Dr. Diwan. Thank you for a very fascinating talk.