VIDEO: Kitchen Table Talk: Jesse Lai on Modulating TDP-43 condensates to treat ALS

Condensates.com welcomed Jesse Lai, Principal Scientist, Dewpoint Therapeutics to deliver our last Kitchen Table Talk of 2025.

Title: “Modulating TDP-43 condensates to treat ALS”

Abstract: Cytoplasmic aggregates of TAR DNA-binding protein 43 (TDP-43) occur in more than 97% of amyotrophic lateral sclerosis (ALS) cases. Aberrant cytoplasmic TDP-43 condensates represent an intermediate, reversible, pre-pathological state in neurons.​​ Over time, TDP-43 condensates can lose their fluid-like properties and convert into irreversible, toxic aggregates. Mislocalization of TDP-43 into cytoplasmic condensates results in toxic loss of TDP-43 splicing function in motor neurons. At Dewpoint Therapeutics, we have discovered brain-penetrant small molecules that potently resolve pathological TDP-43 condensation and rescue nuclear splicing function in ALS patient iPSC motor neurons. This series of molecules broadly reduces markers of neurodegeneration across three mechanistically distinct mouse models of TDP-43 proteinopathy. In this Kitchen Table Talk, we will describe our first-in-class condensate-modulating drugs (c-mods) that highlight the translational potential of condensate biology in ALS.  

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

TRANSCRIPT

Diana Mitrea: Hello, everybody, and welcome to Kitchen Table Talk number 41. Thank you very much for joining us today from all over the world. I’m Diana Mitrea, your host, joining you from Dewpoint’s Kitchen Table in Boston, Massachusetts. 

In the 2025 Kitchen Table Talk series, we focused on highlighting scientific advances in understanding the roles of condensates in physiology and condensate-centric mechanisms that drive disease and also how these insights inform therapeutic discovery. I’d like to thank today’s speaker, Dr. Jesse Lai, for kindly accepting the invitation to close this exciting series. 

But before we begin, we’d like to make a few housekeeping messages. For most of you, these are reminders. But as you know, this talk is a live event, and it will be recorded. Feel free to turn on your camera and follow the talk along with the speaker. However, please be respectful. Do not do any shenanigans on camera. 

We would prefer to keep the questions to the built-in Q&A breaks, but with that being said, if there’s any burning questions, Jesse has kindly agreed to answer them mid-talk. He will also stay a little bit longer, in case there is an extended discussion. Feel free to type in your questions in the chat at any time. During the Q&A, I will call your name, ask you to unmute your microphone, and at that point, you can ask your question directly. In the case that you are unable to speak, please let me know in the chat, and I will read the question for you. This event, as you know, will be posted on Condensates.com within a week or so, so you can refer to it at any time. And now, I would like to pass the microphone to Marty Fernandez, an Associate Principal Scientist here at Dewpoint, who will introduce Jesse. 

Marty Fernandez: Good morning, everyone. I’m delighted to introduce Dr. Jesse Lai. He earned his PhD in Hematology and Immunology from Queens University, where he investigated immune responses to the therapeutic biologic factor VIII in hemophilia A using humanized transgenic mouse models. He went on to a joint postdoc with Justin Achida at the University of Southern California Keck School of Medicine and Amgen. His research focused on utilizing human stem cells to model neurodegenerative diseases. He demonstrated that TDP-43 proteinopathy promotes cell death in a human organoid traumatic brain injury model and that KCNJ2 inhibition mitigates neurodegenerative processes in TBI in vitro and in vivo.  

Dr. Lai is a Principal Scientist at Dewpoint, where he has led the discovery biology efforts for the ALS program. He built the program and led the team modeling ALS using patient-derived iPSC neurons to drive discovery and development of small molecules that modulate TDP-43 proteinopathy and continue to lead the program through pharmacology studies. This work has resulted in DPTX3230, a first-in-class small molecule for the disease-modifying treatment of ALS. This program has been presented and has widely received praise at ALS conferences from advisors and external leaders in the field, including the Northeast ALS Consortium, and has been validated by the receipt of two grants from Target ALS. Today, he’ll outline this work on Dewpoint’s ALS program in his talk, Modulating TDP-43 Condensates to Treat ALS. Please join me in welcoming Dr. Jesse Lai. 

Jesse Lai: Thanks, Marty, for the introduction. Welcome to this Kitchen Table Talk. I’ll be discussing how to modulate TDP-43 condensates to treat ALS today. Dewpoint is obviously very interested in condensation of proteins and RNA and nucleic acids into these phase-separated membraneless organelles. And you don’t really have to look very far within the cell to find where these condensates have a function. It can range anywhere from as fundamental to the cell as transcription all the way to how cells communicate to each other, so neurons at the presynaptic terminal. And what these condensates do is they organize the cell into these functional compartments, and they actually occur across all cell types and organisms. At Dewpoint, we have an end-to-end platform that really identifies different condensates from genetics and how they apply to disease, specifically human disease. And we have some sort of in-house mapping and painting mechanisms to really understand what the phenotyping function of these condensates are in aberrant or healthy condition. And then we have an in-house chemical library, coupled with high-content imaging and AI that really helps us discover ways to modulate these condensates, towards the benefit of health. And lastly, our Chemistry team iterates and optimizes these compounds, and we pass them into the best translational models available. And really what this has built is an in-house proprietary knowledge graph from which we can pull data from to uncover new insights into disease. 

So today is really all about neuro, and condensates are central to neurological and rare diseases, and likely this happens through a coalescence of genetic defects, aging, injury, environmental factors, in which these multi-protein and multi-nucleic acid structures become aberrant in neurodegeneration. And you can see on the right-hand side here all these different neurodegenerative diseases that all have a common condensate dysfunction. 

And today, we’ll focus on ALS. ALS is obviously a neurodegenerative disease that affects primarily motor neurons. It is inevitably fatal, sort of starting from progressive muscle weakness and then leading to death from respiratory failure. And these patients survive at an average of 2 to 5 years and the current treatments really have very little efficacy in extending patient lifespan. If you take a look at these images in the middle here, if you look at the ALS genetic spectrum, you can see that very little of ALS is actually genetically defined. About 10% is familial, 75% of that we actually know the genetics of, and then the vast majority is sporadic. 

But if you actually zoom out a little, you can take a look at the patients, post-mortem that have ALS, and 97% of them have TDP-43 pathology. So clearly, this protein plays a central node role in how disease pathogenesis occurs. And on the right-hand side here is just an image of post-mortem human motor neurons, where in brown, you can see sort of inclusions of TDP-43.  

So again, TDP-43 is now believed that this condensation property of this protein drives ALS. So on the left-hand side, starting from the top left, genetic and environmental factors lead to an aberrant sequestration of TDP-43 in these cytoplasmic condensates. As a result of this, TDP-43 is no longer in the nucleus, and you get loss of function, and probably best characterized is the misplicing of STMN2. 

And then furthermore, something happens to this cytoplasmic condensate that leads to sort of gain-of-function toxicity through the sort of irreversible transition to a protein aggregate. And ultimately, this all leads to motor neuron generation. On the far right-hand side here is sort of some of the more recently characterized aspects of TDP-43 loss of function, and these serve as potential biomarkers in the future. HGGFL2 is a cryptic peptide, which I’ll talk about in a subsequent slide, and then again, STMN expression seems to be very specific to the ALS case.  

Our overall hypothesis at Dewpoint for this program is that abnormal cytoplasmic TDP-43 pathology in the form of a cytoplasmic condensate. TDP-43 in green normally associates with nucleic acid in the nucleus. Under certain circumstances of genetics or stress this TDP-43 translocates into the cytoplasm as a condensate. And this is a reversible process. The condensates are very dynamic. Under certain circumstances, such as additional stress or genetics, these condensates can transition into TDP-43 aggregates, and this is irreversible and toxic. We’re focused on this dynamic phase where we can use condensate modulating drugs, or c-mods, to reverse this condensation, allowing TDP-43 to function in the nucleus. 

I’m going to start here with some in vitro biology. The approach we took here at Dewpoint was really to, number one, highly utilize patient-derived iPSC motor neurons, nd second, by using these iPSC motor neurons, we really reflect the genetic variability in this disease. We use a multi-stressor approach. Each individual one of these stressors may not represent ALS entirely, but a combination of them and the sort of intersection of them likely contribute to a better understanding of what might work across these diverse, genetic causes. So ultimately the molecules that I’ll describe today, or the series of molecules that I’ll describe today, have been optimized in ALS patient motor neurons from diverse genetic backgrounds under multiple stresses, with the objective of really finding that central intersection of disease. 

This is really the only slide that I’ll show that is not going to be data from an IPSC motor neuron. This is a U2OS cell expressing TDP-43 GFP. What you’ll see here at the beginning, is that we’ll add a stress to these cells, and you’ll see the formation of these cytoplasmic puncta in basically white, forming. About midway through the video, we’ll add a compound, and you’ll see a rapid disollusion of these TDP-43 cytoplasmic condensates. Each frame of this video is about seven minutes, so this really attests to the dynamic nature of these TDP-43 condensates, and you can see within 14 minutes of time, these cytoplasmic inclusions will disappear. 

And if you take a look in motor neurons, there’s a selectivity factor that we’ve taken into account. In motor neurons, 99% of TDP-43 in the cytoplasm, as a condensate, co-localizes with G3BP1. So we’ve designed a sort of internal counterscreen within our pipeline to really find molecules that are selectively inhibiting cytoplasmic TDP-43 over stress granules labeled with G3BP1. So in the immunofluorescence image here, you can see, with stress,in white, in the merged image, you can see a co-localization of stress granules and TDP-43. However, with our lead molecule, DPTX3230, you can see that we’re able to dissociate or departition TDP-43 out of these cytoplasmic granules while leaving the stress granule intact. And we think this is really important, right? Because stress granules are the cell’s natural response to cope with stress. 

By really highlighting the selectivity of TDP-43 c-mods in this manner, we hope that there should be a better safety profile than just general widespread stress gradual inhibition. On the far right-hand side here is just a dose-response curve, showing the sort of separation between when we inhibit TDP-43 condensates and when we inhibit stress granules.  

How’s this working? If we start on the left side, lead series interacts with the complex of TDP-43 and UG-rich RNA. So just in a very simple TR-FRET assay, so this is a competition assay using donor and acceptor-labeled TDP-43 and UG-rich RNA. When TDP-43 and the RNA are together, there’s a fluorescence emitted. When you add compound, it dissociates this. So you can see here, on the bottom left, basically the percent inhibition of this complex, and we’ve tested this molecule across different species, and it basically correlates across. 

We can also take this series of molecules, or a prototype of this molecule, and label it with a diazirine and an alkyne that really enables us to UV cross-link the molecule with all of its interactors around it to see what it interacts with and also pull down the molecule subsequently with click chemistry. And what you can see here is that in this molecule, when we treat iPSC motor neurons with this molecule, we can pull down TDP-43, but not G3BP1. And that’s consistent with the data I showed in the previous slide, showing that we actually leave the stress granules intact. 

When TDP-43 is condensed, what happens? TDP-43 normally, again, functions in the nucleus, and it acts to splice out this little red bit in the pre-mRNA. So it acts to normally facilitate the removal of that in the final transcript, leading to the appropriate expression of protein. However, when TDP-43 is condensed or sequestered in the cytoplasm, as we see in these motor neurons, you have the inclusion of this little bit of genetic material. And what happens here is it can lead to different events. It can either lead to, if it’s in frame, it can lead to a protein with a novel segment or a novel peptide. It can lead to the premature degradation of that transcript, or it can lead to a truncated protein. Right now, in the field, all of these different aspects are being explored as potential biomarkers, to really track the progression of disease. 

So, at least in the stem cell field, one of the gold standards is to test it in multiple different lines and multiple different genetic backgrounds. On the left-hand side here, what we’ve done is we’ve taken a panel of 9 different donors for ALS. This ranges from familial to sporadic ALS. Each row, A through J, represents a different individual. 

And what we’ve done is we’ve looked at two different genes that are directly linked to TDP-43 function, so STMN2 and POLDIP3. What you’ll see here is that when you induce stress into these motor neurons, you get a deficit of splicing in STMN2 and POLDIP3, shown in white. And as you increase the amount of molecule DPTX3230, by dissolving these condensates, we actually are able to rescue the function of TDP-43, regardless of the genetic background of ALS. 

And on the right-hand side, I mentioned in the previous slide that little red chunk of nucleic acid that occurs in the final mRNA transcript. That is termed a cryptic exon. TDP-43 regulates hundreds and thousands of genes, and what we’ve done here is we’ve looked at 20 different cryptic exons, specifically looking at that nucleic acid section. What you can see is that with stress, again, motor neurons lose the ability to regulate that appropriate splicing, and as you increase the amount of molecule given, you can correct across all of these different cryptic exons. And we’ve collapsed it all into a single score here, shown on the right. It’s very consistent across these different targets. 

And lastly, in terms of TDP-43 function, obviously adding stress is an artificial environment. So we’ve cultured ALS lines, over time, and we’ve actually treated them in the absence of stress. And what you can see here is that D9817, which is a prototype, an earlier prototype of our lead molecule doesn’t have any effect on healthy cells, right, in terms of STMN2 splicing. But you can actually correct, 12 different sporadic and familial ALS lines with this molecule, which suggests that there was already an inherent loss of function in TDP-43. 

And to wrap up our phenotypic viewpoints in this program is to really look at neuroprotection. We care about neurodegeneration, so what we’ve done here is a five-day neurite retraction assay with a stress-induced model, showing that if you add compound on top of this stress, we get a really nice dose-dependent rescue of the neurite length over the course of these five days. And this is juxtaposed next to the standard of care, one of the two standard of cares. We actually have both, and they’re basically superimposed. However, we can greatly rescue the neurite length over the standard of care here. 

And on the right-hand side is just an image, depicting this rescue, so in gold are basically the cell bodies, and in purple are the neurite length. So I’m going to take a step back here and really talk about how we’re doing this, how this rescue is being achieved. There’s a couple options. TDP-43 function could be rescued by restoring the expression in the nucleus. It can be rescued by shuttling TDP-43 back into the nucleus, and it can be rescued by increasing the specific activity of TDP-43 itself, which is probably unlikely. 

What we did really quickly was taking a look at multiple different stressors and their effects on TDP-43 plus/minus compound. So right off the bat, you can see here that with compound, even though some of the stressors reduce the total amount of TDP-43, with the compound itself we don’t actually see any changes in the total TDP-43 levels. So these data suggest that the restoration of TDP-43 function is likely due to the relocalization of TDP-43 from the cytoplasm back into the nucleus. 

And we took this one step further. In this slide, if we start on the right, this is a graph curve showing POLDIP3, which is another gene that is directly related to TDP-43 function. And we’ve taken the most dynamic section here and looked at the nuclear cytoplasmic ratio of TDP-43. So, you can see that within this range, you have a very nice dose-dependent increase with compounds, showing that TDP-43 is more localized in the nucleus than in the cytoplasm. 

And then we took it one step further. If we block nuclear importers that are known to be associated with TDP-43 localization in the nucleus, that should negate the function, or the effect of our molecule. So we did exactly that. We took antisense oligonucleotides, and we knocked down some nuclear import importance, such as IPO7, IPO9, KPMB1. And what you see here is that with compound and stress alone in a control ASO, we get loss of splicing in this cryptic ELAVL3. So you get an increase in this cryptic exon. However, if you knock down some of the nuclear importers that are necessary to translocate TDP-43 back into the nucleus, you can see that you basically reverse the effect of the molecule, which really suggests that we’re allowing TDP-43 out of the condensate and back into the nucleus. 

I can take a pause here for any burning questions, before I switch to some pharmacology or in vivo pharmacology. 

Tapojyoti Das: I had a question related to the function of the compound DPTX3230. In the in vitro assay, you showed that it prevents TDP-43 binding to UG-rich mRNA. While the same binding, which is responsible for its function in splicing is enhanced, or in some way, it’s restoring the splicing activity. How is that feasible or possible? Do you have any guess? 

Jesse Lai: That’s a great question that we get pretty frequently. We’ve done a ton of work, actually. Given the mechanism of this compound, we’re obviously very concerned with it having on-target toxicity. My theory is that the stoichiometry of the distribution of RNA is actually helping us. There’s so much UG-rich RNA in the nucleus that we’re pretty much unable to compete with the molecule in the nucleus. So when TDP-43 is sequestered out in the cytoplasm in a condensate, there’s more space, there’s less RNA, the compound is actually able to dissociate that condensate, but once it gets into the nucleus, it just is out-competed with the rest of the RNA. That’s one theory and personally my theory. Out of all the studies we’ve done, I’ve never been able to show that the molecule alone is able to inhibit TDP-43 function in the nucleus in the absence of stress. 

If you talk to folks in the ALS field, we will all admit that all of the models are objectively awful. But you still have to do them. And each model tells you something different about how your molecule is working. I’ve outlined three models here that we’ve conducted at Dewpoint, and I’m going to start with the first one here. 

TBI is a quick model, it induces endogenous TDP-43 proteinopathy. And what we’ve seen in this model is that there is TDP-43 loss of function, and then there’s disease biomarkers that are relevant to ALS, such as neurofilament light, GFAP, myelin-associated glycoprotein. And what we like about this model is that it’s really fast, and it’s really just directly looking at the pathology of TDP-43. 

On the left-hand side, TBI is a known risk factor for ALS. Association studies have shown that individuals that have had a TBI have a four to six-fold odds ratio increase in developing ALS, or some sort of other neurodegenerative disease. And TBI provokes mislocalization of the endogenous mouse TDP-43 into the cytoplasm, and it’s phosphorylated, and there’s condensates detected. In the bottom immunofluorescence image here, you can see that three days in a cortical impact, so this is basically a piston that is brought onto the mouse cortex. You can see in these neurons really dramatic TDP-43 leakage into the cytoplasm. 

What we did here is a study design. We took a seven-day study where we administered the compounds subcutaneously every other day. And this compound is blood-brain barrier penetrant, and the PK actually facilitates — we could do one weekly dosing if necessary. But basically, on the right-hand side here, it really shows the progression of this model. So immediately after injury, between one to two days, you get an influx of neurofilament light, neurofilament heavy, GFAP, all these inflammatory cytokines and chemokines into the CSF and the plasma. 

As the pathology progresses, you start getting cytoplasmic and phosphorylated TDP-43 inclusions, you get astroglia and microglia activation, and over time you get neurodegeneration. We administered the TBI and the molecule around the same time, so the molecule was given about 2-4 hours before the injury itself. You can see immediately, seven days after injury, on the left-hand side, we can reduce the amount of cytoplasmic TDP-43 in these neurons. In the immunohistochemistry image here, we zoom in on the penumbra of the injury site. And what you can see is that you have this sort of perinuclear halo of TDP-43 in brown, that is stained that sort of pushes up against the nucleus. And if you add compound at 3 mgs per kg and 1 mg per kg every other day from the onset of injury, you get an almost complete reversal of this TDP-43 pathology. 

Moreover, if you look at the only approved biomarker for ALS, or a filament light chain, in the CSF, you again see a dose-dependent decrease in the amount of neurofilament in these animals. So you get a really large increase, suggesting that there’s neurodegeneration in general, and that with compound, we can reduce that neurodegeneration. Similarly, in the plasma, we have two other analogous markers of nonspecific neurodegeneration through NFH, as well as GFAP, which is a sort of astrocyte marker, which suggests a reduction in inflammation. 

If we take this one step further and conduct a full dose response in these animals, you can see a really clear, dose-dependent reduction in neurofilament over time, and you can see at 1 mg per kg here, we get an 80% reduction in the amount of neurofilament in the CSF of these animals following traumatic brain injury. And again, we see a reduction in neurodegeneration biomarkers in the plasma as well. So collectively, we have biomarkers in the CSF, biomarkers in the brain, and biomarkers in the plasma that all point to that this DPTX3230 molecule is able to rescue TDP-43 condensates and reduce the amount of neurodegeneration in these animals. And perhaps, most important in terms of biology here is really looking at the function of the molecule. One thing that I didn’t mention is that the targets of TDP-43 between mouse and human are not conserved. The binding motif is conserved, but the location of these binding motifs is not conserved between different genes. So the genes that I mentioned earlier, STMN2, POLDIP3, ELAVL3, are not genes that can be used necessarily in mouse. One of the conserved genes is Sort1. So, if you took a look at this inset here in orange, you can see that if you knock down TDP-43, in mouse primary cortical neurons, you get an increase in the expression of this exon 17b of Sort1. 

And then we’ve done qPCR, so this is RNA-seq, obviously, and then we’ve done qPCR to confirm these results that if you knock down TDP-43, you get an increase in expression of this exon 17b. And what we see here is we did in situ hybridization of this specific exon, and we see, again, or an increase, first of all, with TBI. So all of these little pink dots represent a cell or a transcript that is caused by TDP-43 loss of function. And you can see here that with compound, even at 1 mg per kg, we get an almost complete rescue in the hippocampus of these animals. 

We conducted our pharmacology study, TBI, where the molecule was administered around the same time as the injury. But obviously, in the clinic, you’re probably treating patients after some of the symptoms have already started. So, the obvious next step was to really see if we could reverse any neurodegeneration. 

Really what we’ve talked about is TBI as a whole. TBI is obviously one of the leading causes of death and disability in the U.S. There’s no disease-modifying treatments, really, and the existing therapies only really benefit within the first three hours of injury. And these are just some stats to show how prevalent it is. What we’ve done here is we actually injured the animals on day zero, and we took a look at day one. So at day 1 in the CSF, you can see that NFL has dramatically increased with traumatic brain injury, which suggests that there’s neurodegeneration. What we did then is we actually started dosing at day 1. We gave a single dose on day 1, after this injury has already occurred, and we took a look at what happens at day 7. And what you can see here is that neurofilament light chain is still up with TBI at day 7, but with compound, we’re actually able to either reduce the amount of progressive neurodegeneration, or revert some of that neurodegeneration. 

The next model that I’m going to talk about is the dNLS TDP-43 Dox-inducible model. So really what this is, is it is a Dox-suppressible, so it’s a humanized TDP-43, with a mutated nuclear localization signal. When dox is given to the mice, these animals do not express this TDP-43. When dox is removed, these animals start to overexpress TDP-43. This is almost purely a gain-of-function tox model. And as far as we know, there’s no TDP-43 loss-of-function in this model. However, we can use disease biomarkers such as CSF and CSF-NfL, and it’s very aggressive. One thing that I will note is that given that our molecule is not changing autophagic flux, we’re only targeting the condensate phase, or the dispersed phase, of TDP-43. We didn’t expect to be able to overcome overexpression.  

This was a two-month study. Basically, we dosed DPTX3230 every other day for two months, and it was very well tolerated. You don’t see any change in body weight in these animals. Again, this is subcutaneous, twice a week. And what you can see in the middle here is that we can actually reduce CSF-NfL. And what we see is that actually the molecule seems to prevent the aggregation of the dNLS TDP-43 and disperses it more throughout the cytoplasm. And by doing so, we believe that this is reducing the neurofilament shown here. And while not significant, we have two doses of DPTX3230 that actually seem to mitigate the decline in grip strength. This is obviously a very preliminary study, sponsored by Target ALS actually, that we were really excited about. 

The last model that I’ll talk about is this adeno-associated virus injection of G4C2, so 149 repeats of G4C2, which is the hexanucleotide expansion shown in C9 or 72 ALS. And this is the most prevalent genetic form of ALS. This model has endogenous TDP-43 aggregation in the cytoplasm, and it’s been shown that this TDP-43 co-localizes with condensate markers such as G3BP1. There is the potential for TDP-43 loss of function, but it hasn’t been shown yet. With lower number of repeats, I think it was about 60 repeats of G4C2, it’s been shown that there is no TDP-43 loss of function. But again, there are disease biomarkers such as neurofilament, light chain, and this is a very long model. So we took this model out until 9 months, where we started dosing at six months of age. And again, this was sponsored by Target ALS. 

And what you can see is over three months of dosing, sub-Q, twice a week, again, it’s well-tolerated, there’s no change in body weight over this time. What really astounded us is the degree to which DPTX3230 was able to reduce the amount of neurofilament in the CSF at the end of this study. It’s greater than 50% reduction here at two doses. And similarly, if you take a look at just the sheer number of neurons in the cortex, the images on the right-hand side here are just a snapshot of a segment of the cortex, and you can see here is that with the AAV treatment, with C9, you get a reduction in the number of neurons. 

And as you increase the dose of DPTX3230, you’re actually able to restore or prevent the loss of neurodegeneration, simply just looking at the number of neurons, so that kind of coincides with the neurofilament readout here. Overall, what we’ve done in this program is taken in vitro condensate biology and we’ve translated it in different models in vivo. If we start on the left-hand side, if we just review this real quick, cytoplasmic condensates of TDP-43 are dissolved with c-mod. DPTX320 allows it to be dissolved and returned to the nucleus, with a selectivity in the bottom left here for TDP-43 and not stress granules itself. By doing so, we allow TDP-43 to translocate back to the nucleus, and rescue cryptic splicing across diverse patient backgrounds as well. And then we’re able to reduce neurodegeneration in vitro. And then if we take it into the TBI model of endogenous TDP-43 pathology, we’re able to rescue biomarkers of neurodegeneration, as well as the pathology of cytoplasmic TDP-43 itself.  

Lastly, what I won’t get into today is this is well-tolerated across multiple species. I’ve mentioned earlier, across three different mouse models, this molecule has been very well tolerated. We have good tolerance in rat and mini-pig as well. At this point, I’d like to thank the multitude of folks that have touched this program over the years. I obviously did not do this all myself. And, you know, everyone that has been involved in this knows who they are. And it’s been phenomenal. I’ll stop here for questions, and thank you for tuning in. 

Diana Mitrea: Thank you so much, Jesse, for this really fantastic talk. If anyone in the audience on Zoom has questions, feel free to either raise your hand or type your question in the chat. 

Laura Behrendt: Great talk. Can you tell us something about how TDP-43 mislocalizes in response to stress, are there any PTMs involved? Or how does it end up in the cytoplasm? And also, do you observe mislocalization in the ALS iPSCs without stress, or do you also need to hit them to see that? 

Jesse Lai: I don’t know if anyone really knows how different stressors pull TDP-43 out exactly, but depending on the stress, you can actually see different PTMs form, but I think that forms after it gets pulled out of the nucleus. For example, proteasome inhibition will cause phosphorylation of TDP-43, but I don’t exactly know if that happens before it translocates or after, likely after. The second question was, do we see TDP-43 mislocalization without stress. That is something that is not consistent in the field. Some folks see mislocalization. And also, it depends on how you make the motor neurons, apparently. In some cases in the field, people see a difference in nuclear cytoplasmic distribution of TDP-43, and in some other cases, they don’t see anything. 

In our hands, it seems to depend on how long you culture the neurons for, and it also depends on which genetic background you use. If you have enough different lines and enough diverse subtypes of ALS, you will see some lines just by default have more cytoplasmic TDP-43 than others, so really what you need is probably a large cohort of controls and ALS patients to see what subsets show that. 

Andrew Mouland: Are you making any effort to increase potency by modifying compound backbone, etc.? 

Jesse Lai: The data that we’ve presented here is the culmination of probably about 1,500 molecules that we’ve tested in motor neurons over the years. We’ve explored different scaffolds, we’ve explored different substituents, and what I presented here today is already an optimized and potent molecule from those learnings. 

Diana Mitrea: You mentioned that the three animal models all come with their different challenges. By knowing how the different models are different, and what they lack in terms of representing the various mechanisms that TDP-43 contributes to ALS, and knowing the mechanism of the compound, can you infer the weight of the loss of function versus the gain of toxic function from those results? 

Jesse Lai: This is my opinion. In terms of gain and loss of function, the field in general is more interested in loss of function, especially given all the data that’s been shown post-mortem. That’s not to say that gain of function isn’t contributing to neurodegeneration. I’m a little biased, because I like the TBI model. It’s fast, and it lets you cycle through molecules a lot faster and test more things. It has gain-of-function aspects, it has loss-of-function aspects, but it also has a lot of other stuff that’s associated with TBI that likely has not much to do with ALS. In my opinion, I think the loss of function aspect is probably more relevant to the MOA of this molecule. And I think that’s a consistent line from our in vitro work to in vivo work that shows rescue of loss of function. And in terms of our MOA, I would place more emphasis on that model compared to the other two models of overexpression and an AAV overexpression that don’t actually show any loss function at all. 

Shruti Arya: Do you see a difference in how the compound responds under different stress conditions that you have looked at, or does the compound perform similarly to different stressors? 

Jesse Lai: What we’ve seen is actually the neurons respond differently to different stressors in terms of what genes are changed. So, TDP-43 obviously regulates hundreds, thousands of genes. And while some of them are changed in one stress, another subset are changed in another stress. But there is a core group of genes that are changed across all of the different stressors. And actually, our molecule seems to rescue all of these different stressors, regardless, so it rescues the splicing of genes across all these different stressor conditions that we’ve tested. 

Noah Nathan Cohen: Do you see an opportunity to also target TDP-43 in the nucleus? 

Jesse Lai: That’s a good question. I think there’s an opportunity if you can increase the actual specific activity of TDP-43 in the nucleus. If you believe that it’s purely loss of function, then it doesn’t really matter how much is in the cytoplasm if you can restore the splicing in the nucleus in general. So that’s one option. I’m happy to brainstorm more, but in general, that would be my approach to that. In the nucleus, the condensation of TDP-43 is so important in the nucleus for function, and I think there’s papers that have described the importance of condensation function in the nucleus, so we probably don’t want to mess with that too much if at all. 

Mark Chen: Great talk. I’m actually not even an ALS researcher, I was just curious, is there, a fundamental biological reason why no good models exist? And in that context, what is a good in vivo model besides the TBI, if there is one? 

Jesse Lai: I think people think that the ALS models are bad. A lot of them are overexpression, and they haven’t really translated well into the clinic either. So for that reason, I think the field is interested in better models. Additionally, none of the models really reflect sporadic ALS, because, it’s hard to make a model if you don’t know what causes the disease in the sporadic cases. That’s part of the reason why I like the TBI model, because it doesn’t matter, it’s a wild-type animal. And I don’t think it’s due to the lack of conserved targets. I mean, it might be. There are plenty of companies now that are targeting STMN2, KCNQ2, ONC 13A, and these are all downstream targets of TDP-43. And in this case, you can’t really use the mouse models to study those, per se. They’re making new models with humanized STMN2, for example, but for a lot of those genes, there is no model for it. I think I can make a list of just as many pros as cons for TBI. It really depends on what you’re looking at. You’re not going to get every single readout, you’re not going to get every single biomarker to work in every single model, but I think it’s the collective learnings of different angles at which you approach TDP-43 proteinopathy that can give you the confidence to move this molecule forward. 

Christine Vande Velde: Great talk. I was just curious, then, if TBI is causing TDP to translocate to the cytoplasm, does it apply as a function there? And I’ll just also mention that with exotomy, either sciatic exotomy or facial nerve exotomy, TDP-43 also translocates to the cytoplasm, and it’s reversible as the injury resolves. If it’s part of the injury response, what do you think the impact of the c-mod might be on that? 

Jesse Lai: That’s a great question, similar to what I was alluding to earlier with the stress granule response being a natural response to the cell. I don’t have an answer to your question. It’s very possible that TDP-43 translocation following TBI into the cytoplasm is an injury response, and it’s necessary. However, in our scenario, obviously, the biomarkers suggest otherwise. What we haven’t done is taken the animals out beyond seven days. So seven days is still very acute, considering the severity of the injury. We don’t know, I guess is the answer, anything longer than seven days. But what I can say is, at the end of our study, short, seven days, it seems like the compound by disrupting the TDP-43 in the cytoplasm does not seem to be changing, or it seems to be rescuing neurodegeneration, rather than interfering with a natural response to injury. 

Khadijeh Alnajjar: Is the molecule exclusive to multimeric/condensate form of TDP-43? Is there still SG formation (G3BP1 puncta) after stress?  

Jesse Lai: In our in vitro system, in the TR-FRET assay, the molecule can dissociate TDP-43 from UG-rich RNA, just as a monomer. In vitro, it seems to be exclusive to the condensed form of TDP-43. And I say this because, again, I have tried to add the molecule to a system to show that we get on-target tox, meaning that it inhibits TDP-43 in the nucleus, and we don’t see that. So we have not been able to show that the molecule inhibits TDP-43 in the nucleus. In that regard, I believe it is affecting the condensed form in the cytoplasm.  

The second question is, is there still stress granule formation after stress? In my slide I showed that the compound is able to dissociate or pull out TDP-43 from the stress granule, while the G3BP1 stress granule puncta are still there. If your question is more so, can additional stress granules still form after that, I don’t know the answer to that. But what I do know is that when condensates are there, when TDP-43 and G3BP1 are coalesced together, the G3BP1 puncta remain after compound treatment. I hope that answers your question. Thanks, everyone. 

Diana Mitrea: Thank you so much, Jesse. And this is our last KTT for 2025, so wishing everyone a joyful holiday season, and we’ll see you in 2026!