VIDEO: Rui Tong Quek on Advances in Targeting Biomolecular Condensates for Small Molecule Drug Discovery

Condensates.com welcomed Dr. Rui Tong Quek, Scientist – Experimental Drug Development Centre, Agency for Science, Technology and Research, Singapore to deliver the first Kitchen Table Talk of 2025. During her talk, Dr. Quek discussed the potential of targeting condensates to develop novel antivirals and be about an expanded toolkit to accelerate c-mod identification for future crucial therapeutic targets.

Dr. Quek recently earned her PhD in Chemical Biology from Harvard University and is currently a scientist at the Experimental Drug Development Centre in Singapore where she works on projects related to RNA small molecule drug discovery. She earned her Bachelor of Science degree in Biochemistry, with honors from the Imperial College in London and is a two-time recipient of the A*Star National Science Scholarship along with multiple other accolades.

Click here to view the engaging talk and discussion below.

TRANSCRIPT

Diana Mitrea: Hello, everybody, and Happy New Year! Welcome to Kitchen Table Talk number 38. My name is Diana Mitrea. I’m Head of Scientific and Corporate Communications at Dewpoint Therapeutics and your host. I’m joining you from a packed Dewpoint Kitchen Table in Boston, Massachusetts. Thank you all for being here, and thank you to our guest speaker Rui Tong Quek, for kindly accepting our invitation to kick off a new year of exciting scientific progress. 

Before we begin, I have a few housekeeping messages. This talk is a live event, and it will be recorded. Feel free to turn on your camera and follow along with the speaker. However, be respectful. No on-camera shenanigans please. We would prefer to keep the questions to the end. With that being said, if there are any burning questions, Rui has kindly agreed to answer them mid-talk, and she’s also willing to stay a bit longer in case there is extended discussion. Feel free to type your questions in the chat as they come to you during the Q and A. I will call your name, and we will unmute your microphone, and at that point you can ask your question directly. If, for whatever reason you’re unable to speak, please let me know in the chat, and I will read the question for you. The event recording will be posted on Condensates.com within a week or so, so you can refer to it at any time. 

And now let’s move to the main event. It is my utmost pleasure to introduce Dr. Rui Tong Quek, a rising star in the field of applied condensate science as today’s invited speaker at the Kitchen Table Talk. Rui is a freshly minted PhD and is currently a scientist at the Experimental Drug Development Centre in Singapore. So, thank you so much for joining us at this ungodly hour for you. 

Rui works now on projects related to RNA small molecule drug discovery. Rui earned her Bachelor’s in Science in Biochemistry, with honors from the Imperial College in London and her PhD in Chemical Biology from Harvard University under the mentorship of Dr. Pamela Silver. She’s a two-time recipient of the A*Star National Science Scholarship along with multiple other accolades. I personally first became aware of Rui’s work when I was researching the literature for our Nature Reviews Drug Discovery perspective on opportunities to drug condensates, a preprint at the time. Now that paper is fully published, and the peer review was a collaboration between the Silver and Mitchison Labs at Harvard and Abbvie, and it was an elegant, fascinating study that leveraged phenotypic screening to identify novel antivirals. The study also produced one of the first pieces of evidence that supported the hypothesis that a condensate approach to drug discovery could be the key to developing broader acting therapies. I’m excited to hear about the evolution of the story, and, without further ado, join me in welcoming Dr. Quek. 

Rui, the floor is yours. 

Rui Tong Quek: All right. Thanks so much, Diana, and thanks everyone at Dewpoint for having me here. It’s a great honor and privilege to be able to share my work with all of you guys. And I’m really excited to talk to you today about sort of the broad topic of advances in targeting condensates for small molecule drug discovery. And before I begin, I just wanted to sort of frame the talk by saying that there’ll be two main parts to my story. So the first part of my story is sort of what Diana alluded to earlier, and the second part of my story is sort of a continuation from the work that we looked at in targeting viral condensates. And the second part will be a little bit more on technology development. So it’ll be a little bit more biological and then a little bit more on technology development in the field of drugging condensates. 

So just to recap some background, which I’m sure all of you guys are really familiar with. But I just wanted to highlight a few key background points that might be relevant to this talk. Condensates, as we all know, have tunable physical properties. So a given, for example, protein – condensing protein can exist in several different states in a cell. So it ranges from the soluble state, what we call, you know, diffuse, for example, diffuse in the cytoplasm to a condensed, liquid-like droplet state where these condensing proteins, along with other components of the cell, come together in a liquid-like droplet, and this can also mature or develop into more solid gel-like aggregates. So this is just illustrated in this cartoon here. Again, just for visual depiction in the cell. If you were to visualize this protein, for example, this protein here is tagged with GFP, you would be able to see these sort of distinct states by microscopy: so soluble state it’s diffuse, liquid-like state you might see spherical droplets, and in a gel or solid-like state these droplets often sort of form amorphous aggregates. 

So, as we know the interaction strength between all the molecules in this condensate is responsible for the state that the protein is in, whether it’s soluble, liquid, or a gel or solid-like aggregate, and anything that sort of perturbs interaction strength, such as post-translational modifications, pH cellular conditions, or mutations in these condensing proteins can affect its ability to form condensates, and also the state of the condensate. So I’ve brought up PTMs or post translational modifications because that’s going to be a big topic that I will talk about later on. 

So there, of course, are many roles for condensates in cells and in disease as well. This regulation of phase separation or condensation is involved in progression of various diseases, so one of the more obvious ones would be neurodegeneration and cancer, where persistent condensation or irreversible maturation of condensates can lead to disease states. So, for example, constitutive condensates in the nucleus that leads to aberrant transcription or formation of amyloids. So that’s sort of you know, disease state associated with condensation. 

Another important disease field that we look at is viral infections, and a lot of viruses actually produce proteins that are phase separating proteins, that are involved in various stages of a virus life cycle. So that’s where we come in. And we were like really interested in, especially the Coronavirus N protein, which is known to be a phase separating protein, and at the time it was the start of COVID. So we were all really interested in SARS-CoV-2, and studying this phenomenon of phase separation of N and trying to think in a bigger picture of drug discovery. So the SARS-CoV-2 N protein is thought to have multiple roles in viral infection. But specifically, in thinking about condensation of the N protein, there are sort of two main functions of N condensates. 

Firstly, it’s thought that these condensates in infected cells act as reaction crucibles for replication and transcription of the virus. So the N protein can actually interact with a lot of RNA. So be it viral RNA or host RNAs to form these phase separated condensates in infected cells and concentrate factors, both host as well as other viral factors. That’s important to form the viral replication transcription complexes for the virus to replicate and transcribe. So that is thought to happen in the liquid-like state where you know, the molecules within the condensates are mobile and dynamic, and reactions are able to happen. 

On the other hand, it’s also thought that the N protein can sort of mature into a more aggregated state. And this is important for organization of mature virus particles. So, for example, this happens when the N protein binds onto the viral RNA and then condenses into a mature virus particle. 

So, since condensation of the N protein is so important for viral replication, our hypothesis going in is that drugging and condensation sort of perturbing this sliding scale of biophysical properties of N may have a therapeutic outcome for viral infections. And so we thought that if we could try using small molecules to sort of arrest, phase separation that would have implications for therapeutics in terms of viral replication. Just as another sort of background aside, condensates, as we know, are challenging drug targets. And that’s because of what we call druggable binding pockets. A lot of FDA approved drugs are actually targeting druggable proteome, which are, for example, GPCRs, ion channels, kinases, nuclear receptors. These targets have distinct binding pockets that can basically engage a small molecule very specifically and very potently, and that can lead to very distinct therapeutic outcomes. 

On the other hand, condensates comprise multivalent interactions, and many of the components are intrinsically disordered. So they’re often really undruggable. And so it’s hard to sort of think about drug discovery for these types of targets. 

For druggable targets, it’s very common to do target-based screening where you have a certain protein target of interest, and you screen compound libraries for binding and for activity inhibition, however, for sort of a bigger phenomenon, like condensation, it’s often much better to go down the route of phenotypic screening where you’re looking for a cellular outcome instead of, you know, coming up with a specific target to begin with. 

 I’m going to talk about this phenotypic screening in the first part of my story. Here’s the overview of my talk. The first part of my story would be talking about our efforts in identifying small molecules that modulate SARS-CoV-2, and condensation and its effects on viral replication. So this is the biological story that kickstarted my interest in small molecule drug discovery for condensation. And so this part of the story is actually done in collaboration with Abbvie, together with our team at Harvard Medical School, and this was done with other members of the team as well with Pam Silver’s group as well as Tim Mitchison’s group. 

Okay, so phenotypic screening. We were interested in identifying compounds that promote N condensation or basically perturb the face separation of the N protein. So, this is again the graphic from earlier. And I’ve sort of listed on top the functions that the N protein is involved in environment. So transcription replication, when it’s more in the liquid-like state as well as packaging when it’s more in the solid state. And so one part of our screening assay was actually looking for what we call pro-condensation. Small molecules which are small molecules that promote the formation of condensates or promote the hardening of end condensates. To do this screening assay, we set up a screening assay pipeline that used the A549 cell line that’s constitutively expressing the SARS-CoV-2 N protein fused to GFP. This is not an infected cell line. It’s actually just a cell line that is constitutively expressing the N protein. Under regular conditions, just expressing the protein in this cell line here with DMSO, so that’s untreated, you can see that the N protein is sort of diffusely localized throughout the cytoplasm. 

We also have a positive control compound for our screen, MS023, that upon treatment of cells forms these puncta that can be visualized by fluorescence microscopy. This was just a positive control for the screening assay to look for pro-condensation small molecules. And so we, together with Abbvie, we actually screened first a pilot screen comprising 2,500 compounds from a chemical library at 10 micromolar, and so we were looking for no condensates to condensates. 

On the left is the scatterplot for the results from our initial screening assay. On the Y-axis, we’re plotting percentage activity versus the positive control. At 100%, that means that these compounds are equally active as MSO23 in inducing puncta, and at 20%, for example, it means that the compound induces 20% as many condensates as MSO23. 

From our screening assay we found that the bulk of the library was inactive at 0%, but there were also several hits that were active at inducing N condensates in this cell line, and so some of the hits are highlighted. But for this talk I would like to focus on CP21R7 and autophnib, which are two sorts of hit compounds that we got really interested in. So these are the chemical structures of the two compounds and below it is the SARS-CoV-2, N expressing cell line upon treatment with these compounds, where you can see formation of the N factor very robustly. 

What we found was that these compounds were actually GSK3 inhibitors, and there is prior knowledge in the literature that GSK3, which is a kinase, it’s involved in phosphorylating N protein. So in this schematic below the green squiggly line, that’s the N protein. It’s known that GSK3 actually phosphorylates N protein at several parts of the protein, but mainly in the Serine, Arginine rich middle portion of the N protein. It adds it phosphorylates the N protein there. And so what happens is when we add GSK3 inhibitors to the cells that prevents GSK3 from phosphorylating the N protein, and in the unphosphorylated state, the N protein is much more prone to phase separation, or condensation. We can see that by the puncta that’s present in the cells upon treatment of GSK3 inhibitors. So to keep it simple, when GSK3 inhibitors are added, we see puncta. Otherwise we don’t see any puncta. 

Okay, so we were really interested in how broadly active these GSK3 inhibitors were at inducing N condensation in other Coronavirus N proteins. So on top of SARS-CoV-2 there are also several other coronaviruses that are involved in also just the regular cold that is circulating around. And these viruses also have structurally similar N proteins. 

So here are cell lines expressing the seven proteins from seven different human coronaviruses. And again, these are not infected cells. They’re just cell lines that are constitutively expressing the N protein. And so this is the untreated cell line. When we add CP21R7 to all these cell lines, we see a robust formation of N puncta across all seven N proteins. 

The same was seen with the other GSK3 inhibitor. We were again interested in confirming this phenotype. So we actually tested a whole bunch of other GSK3 inhibitors, of which some were ATP competitive GSK3 inhibitors and some allosteric GSK3 inhibitors. 

So just zooming in and focusing on one cell line, SARS-CoV-2, and applying all these GSK3 inhibitors to the same cell line. Here’s a dose response curve showing the formation of number of puncta per cell upon addition of different concentrations of these GSK3 inhibitors, so we can see that for ATP competitive GSK3 inhibitors at some point puncta always forms in the cell. However, for the allosteric inhibitors where the inhibitors don’t bind in the ATP binding pocket, these compounds are not active at inducing puncta in this cell line and this was also repeated across all seven cell lines. So the activity is very similar. 

Now we wanted to know if these GSK3 inhibitors sort of change the biophysical properties of condensates. To do this, we sort of approached it by looking at what we think of as two different forms of N condensates. So one way you can form the N condensate is again by treating cells with the GSK3 inhibitor. So we start off with diffuse N. When we put in GSK3 inhibitors, it actually pushes N towards condensation. So that’s from middle to left. 

Another way of inducing N condensation is by adding polyIC.  PolyIC is a double stranded RNA mimic that interacts with the N protein through RNA binding domains and also induces robustly the formation of N condensate. So two different ways we can achieve N condensation. The green way or the teal way on the left is with GSK3 inhibitors, and then the pink way on the right is with polyIC. 

With these two ways of inducing condensation, we perform FRAP, which is fluorescence recovery after photobleaching. We wanted to look at whether the dynamics of the N condensate is any different between these two types of condensates. So what we do is we bleach individual condensates in the cell and monitor the recovery of GFP into the bleached condensate over time. Faster rate of recovery of the bleached spot indicates that the molecules within the condensates are mobile, so that unbleached molecules can easily return to the condensate for fluorescence recovery. With polyIC condensates, we see that there is some level of recovery of fluorescence over time after bleaching condensates in the cell. 

However, we found that with one of these, one of the GSK3 inhibitors from our library, that the recovery was extremely slow, and there was a very low recovery, suggesting that the condensates formed by GSK 3 inhibition are really solid, and they’re almost irreversible. In this assay format it becomes more of a gel, gel-like, solid state as opposed to a dynamic liquid-like state. 

And so the final question that we had was, okay, this is all good. How can we link this to antiviral activity? We have so many GSK3 inhibitors. We also know that there was a wealth of antiviral data that’s available in the literature. So this is again the dose response curves from earlier. Just looking at SARS-CoV-2 But we had dose response for all the different viruses, and we found that there were several GSK 3 inhibitors in the set that we tested, where there was also antiviral activity reported in the literature against different coronaviruses, so these three here I’ve highlighted. We have found that these inhibitors are also active at achieving antiviral efficacy for different coronaviruses, and importantly, LY – so this compound here, the one that is the most potent, meaning that it requires the least amount of compound to the lowest concentration to induce puncta in cells also achieves antiviral efficacy at a very low concentration. So the antiviral IC50s actually track very nicely with what we observed as the N condensation EC50s. 

So there’s some key takeaways from this first part of our talk. You know small molecules can alter N condensation states, and likely with antiviral efficacy and importantly, also, pan-human coronavirus and condensation. Modulatory activity can be achieved with small molecules like GSK3 inhibitors suggesting that among families of viruses there may be conserved mechanisms for regulation of condensation, and that can be a very good antiviral therapeutic strategy. 

And another point that we actually took away from this screening assay that is potentially very useful in thinking about future pandemic viruses is that screening can actually be performed on isolated viral pathways instead of on whole virus replication. So all of our screening here was done in cell lines that express the N protein and not in infected cells. So this allows for us to rapidly screen against emerging pandemic viruses, and also it makes screening accessible without the need for a whole virus replication, especially for viruses that are BSL3, BSL4. 

Okay, so that was the first part of my talk. And I will now move on to the second part of my talk, which is developing alternative condensate screening essays. And this was sort of born out of, you know things that we learned from the first part of the first project that we worked on. And importantly, this is also done through participation in Eikon Therapeutics technology access program. So we collaborated very closely with Eikon Therapeutics and used their microscopy techniques for this part of my of the project. 

Okay, so just sort of recapping what I talked about earlier with GSK3, which is a kinase, that phosphorylates N protein.  If we inhibit GSK3 with GSK3 inhibitors, we end up forming N condensate. So that’s sort of from earlier. Again, this is the dose response curves. And I want to draw your attention to just three of the compounds from earlier that we’re going to use as two compounds for developing new methods for screening, for condensate modulators. So the first compound in dark teal LY, that’s the most potent compound that I alluded to earlier, where we need only really little or really low concentrations to induce N condensation. 

CP was the first compound that was pulled out in our screen. That’s also an active compound at inducing N condensation, but at slightly higher concentrations and tideglusib. That’s the allosteric inhibitor that’s completely inactive. So three different types of compounds, two compounds that I will be talking about now. 

Previously, when we did screening with the cell lines that I talked about it was all high content screening. So high content screening is basically fluorescence microscopy based where we treat cells in 384 well plates with compounds, and then we fix cells. And then we have to image with fluorescence microscopy, and do image analysis to count the number of puncta per cell. And that is what gives the final readout of puncta per cell. And so we determine if a compound is an active hit or not based on this quantification. 

Image analysis is also a really convoluted process. So you know, you have your raw image. You have to segment cells, ID cells, and then you have to, using certain pipelines, do a spot counting. So that’s generally how high content screening was performed and is typically performed, especially for phenotypic screening. For example, when we are trying to, in general, look at drugging condensates, however, with high content screening, there’s like there’s several limitations associated with it. So firstly, there’s a lack of information on dynamics. So oftentimes high content screening is done on fixed cells. So it’s a defined endpoint where you fix the cells, image the cells. You can only see puncta or no puncta at an arbitrary time point post treatment. So there’s no information on what’s en route or what’s going to happen or what has already happened. 

There’s also an issue with throughput, because imaging plates takes a lot of time, especially in comparison to, for example, a plate reader assay. There’s also optical resolution limitation. So you can only really see puncta that’s above the diffraction limit. And there’s always errors in image analysis where you may pick up puncta. That’s background, or you may miss out puncta. That is real. So the image analysis, pipeline, is not always accurate. 

So to sort of address these limitations that we faced in our initial screen, we developed two alternative screening technologies to identify condensate modulators, and these are not necessarily better technologies. They are all complementary technologies. But I like to introduce two different ways to go about looking for condensate modulators. So the first assay screening assay technology that I will talk about is high throughput, single molecule, tracking or SMT. And this was done in collaboration with Eikon Therapeutics. The second technology that I’ll talk about is proximity based condensate biosensors. So these biosensors are plate reader compatible assays. So that will address the throughput limitation. 

I’ll talk about SMT first. The concept of SMT is basically instead of, you know, looking at a sort of about condensate readout, we want to track the diffusion and trajectory of single N protein molecules. Whether or not they’re in condensates or not, we want to look at single protein level. So what our hypothesis is that you know, when you are sort of en route to condensation, or you’re in a condensed state, the diffusion is going to be different, and that’s what we hope to read out from this assay. To do this, we generated cell lines that are expressing the N protein. But this time, instead of being fused to GFP, it’s fused to a Halo tag. And what we can do with the Halo tag is, we can sparsely label the cells or sparsely label the proteins with it. We add a very limiting concentration of a fluorescent dye that binds to the Halo tag, so not every protein, not every N protein in the cell is labeled so we can track single molecules. 

And so what happens is the molecules, the N proteins moving around in the cells, and we’re imaging every so often. But what we can really see are just single molecules instead of the bulk phenotype. And so we image at every 10 ms, for example. At every point there will be an image taken. So this is how the cell line generally looks in the SMT channel. If you zoom in further and further, you can start tracking single molecules. Here’s a movie that I’m going to play. I’m not sure how well it will look, but just shout out, if anything doesn’t look right. But we can start looking for jumps. Or we basically start tracking single molecules over time to generate trajectories and determine the diffusion rate. 

Okay, so here are two treatment conditions: one is untreated and one is with LY, which is the GSK3 inhibitor that does induce N condensation. So again, the hypothesis going in is that you know, with condensation, your molecules start moving slowly. So I’ll talk a little bit more about that later. 

Mapping the trajectories, as I’ve shown earlier onto this image and then pulling out individual trajectories, representative individual trajectories from both conditions, we can see that with the LY treatment condition the molecules actually move, much more the jump. The jump lengths are much shorter, which means they’re not moving as fast as in the DMSO, the untreated conditions. So the N molecules are slowing down on you know here what we call slower tracks. 

And just to sort of give an idea of the scale, this can be done in a high throughput format in 384 well plates with, you know, multiple fields of view per well, and up to 100,000 tracks per field of view. There’s a lot of statistical significance in this data set. Okay? So this again, on the left is the raw data. What we see in the images, the representative trajectories of the single molecules jumping. We wanted to quantify this further, so we developed several SMT metrics to assess dynamics of the N protein. 

So again, firstly, this is just a plot of the raw jump lengths between the two treatment conditions. So with the GSK3 inhibitor, you can see that there is a huge shift towards the left in jump length, which means that the molecules are actually jumping a much shorter distance, which means they’re not diffusing as quickly.  And from that we can actually infer a diffusion coefficient state array. So on the X-axis, we have the diffusion coefficient, and then on the Y-axis is sort of like the proportion of molecules in that diffusion coefficient state. So again, we see a flattening and a leftward shift of the diffusion coefficient that are the median diffusion coefficient with treatment. 

Finally, we also devise sort of a subsequent metric which is slow or fast proportion. So with a boundary defined, what proportion of the molecules are actually what we call slow moving molecules, and what proportion are fast moving molecules, and from there we can see that there is a much higher proportion of slow-moving molecules in the treated condition. 

So sort of explaining this, or like coming back to the big picture, why does this even happen? We think that when the N protein is in the soluble state, when it’s phosphorylated, for example by GSK3, it is moving about in the cell freely. So it is diffusing very quickly and there are sort of two reasons why we see a slowdown in diffusion upon treatment of GSK3 inhibitors. One is that en route to condensation the molecules are oligomerized, and there’s just a bigger hydrodynamic radius so they start slowing down. Eventually they do form condensates, and, within condensates, the N protein is moving much more slowly than in the diffuse state. 

So there are several metrics that I talked about in this slide, jump length, sort of the diffusion coefficient as well as the slow proportion. So for the rest of the subsequent few slides, I’ll just be focusing on slow proportion as our SMT metric. Okay, so that was just done with one GSK3 inhibitor. So now I’m going to talk about the other two GSK3 inhibitors that I sort of talked about in the start of this section. Here is the data we got from high content screening. That’s from the very early part of the talk. This is from imaging from counting puncta. So we do see that LY is much more potent at inducing N puncta, CP is OK and tideglusib is inactive. With high throughput SMT we see sort of the same trends where the slow proportion increases at much lower concentration of LY compared to CP, and tideglusib is entirely inactive, so we see the same trends. And if we were to pull out the EC50s of this, they are also fairly comparable, so they’re in the same order of magnitude. 

Another interesting thing we found through doing kinetic assays is that high throughput SMT readout is actually really robust at very early time points. So now on the X-axis instead of concentration, it’s time, so this is all done at a fixed concentration of compound. We can see that actually, really early on after compound incubation, so just 5 hours after incubation, we can already see a huge difference in the slow proportion in the treated cells versus tideglusib, for example, which is an inactive compound. And also at 24 hours for this particular cell line, the ones expressing N fused to Halo, if we were to look at these cells by fluorescence microscopy through high content screening, we cannot robustly identify a lot of puncta in the cells just by looking at it. However, even without the visual confirmation of condensation in this cell line, we can see a huge difference in the slow proportion, which means that this assay is actually really sensitive to many steps prior to the actual phenotype of condensation appearing. 

So that’s for the high throughput SMT. So now I’ll wrap up this section by talking a little bit about the proximity-based condensate biosensors, which is compatible with a plate readout. So here the idea again, we’re using the same compounds, the same 3 tool compounds for this assay, but the idea here is that we have our condensing protein of interest, so N protein again, and we fuse it to a proximity activated biosensor domain that’s in blue here. In the diffuse state, these biosensors are actually far apart from each other, so they don’t produce a high signal, and the idea is that, upon condensation, these biosensor domains come together in close proximity in a condensate, for example, and they produce a high signal. And so that’s sort of represented by the fluorescence images that I’m showing here as well. So the concept is that we don’t have to look at the fluorescence image, we can just read the signal readout with a plate reader assay. 

And so there are sort of two different approaches that we tested out with this format. One is the NanoBIT assay, and one is the NanoBRET assay. The NanoBIT assay is basically a split luciferase, so we fuse one half of the split luciferase to the N protein and the other half of the split luciferase to the N protein as well, and we express both types of N protein in the same cell. 

For NanoBRET, it’s a similar concept where one on one half we have the luciferase fused to end, and on the other half we have a Halo tag as well as a ligand that is, an acceptor of the light that’s produced by the luciferase. So we’re actually measuring the fluorescence of the ligand as a readout of proximity. So both are proximity-based or proximity-activated biosensors and in theory should produce a high signal only when they’re in close proximity. 

Okay, so just really, quickly, the same dose response curves for both the NanoBIT and the NanobBRET assays we can see that LY, which is the most potent compound, is again active at producing a plate reader-based readout at much lower concentrations than CP. And tideglusib is also inactive in both assays. And we can see that with the EC50s again, they’re pretty much comparable to what we obtain with high content screening. In fact, with the plate reader assays it might be a little bit more sensitive at picking up, for example, dimerization or oligomerization events compared to high content screening. 

To sum up this section of my talk, there are several key takeaways. So with high content screening, which is a traditional way phenotypic screening is done, especially in the field of drugging condensates, there, of course, there are benefits and challenges to all the different techniques, but for high content screening, it’s adaptable for probing native condensate biology. So if you have, for example, an antibody that you can use to visualize condensates in the native cell, that would be useful in this case, because you don’t have to exogenously tag any protein. So that’s sort of one benefit of high content screening. You can also score for other known interacting partners in condensates by staining for high throughput SMT. You can actually report on interactions with known or unknown partners as long as they sort of change the hydrodynamic radius or the condensation state of the condensing protein. And you can also score for changes in condensation that’s not observable by traditional fluorescence microscopy. 

And lastly, for proximity-based condensate biosensors, it’s a really fast plate reader readout. It’s independent of imaging so it doesn’t come with all the limitations of image analysis. And you know, it’s basically very adaptable to studying many different proteins, condensing proteins as well. So importantly, the main differences between these techniques is for high content screening. It scores the emergent phenotype of condensation for high throughput SMT and proximity-based biosensors. We’re actually scoring molecular scale interactions, so we’re looking at sort of two different levels of condensation and depending on the application of target or of the biology one or another, technology might be more applicable for screening. 

So that’s the end of my talk. And I’d just like to sum up with several concluding words, in the first part of my talk we looked at GSK3 inhibitors as N pro-condensing compounds, and also found that pan-human Coronavirus, and modulatory activity can be achieved with antiviral efficacy. In the second part of my talk, I talked about alternative assays to high content screening with complementary advantages for identifying condensate modulators.  

With that I’d like to acknowledge the people who have contributed to this work. For the first part of my talk with Abbvie, we had a ton of help from Steve, Sujatha and Steve, another Steve at Abbvie, with screening and with a lot of the interpretation of data. From my lab Kierra, and then helped a ton with all the work and all the image analysis and just discussions. We have also Adrian and Hasiani, who helped out with a lot of the GSK3 assays, For the alternative condensate screening essays, Kierra and Sam from my lab, were very helpful with all assay designs, performing assays, data analysis. From Eikon, David, Cyna, Christina, and Philip were invaluable to helping me learn their technology and applying it to condensate biology, which I think is a really exciting application with, you know, a ton of promise for the future. 

So with that I will be happy to take any questions. 

Diana Mitrea: Thank you so much, Rui. This is a fantastic talk, and I don’t know if you heard there was clapping here at the Kitchen Table. So we have some wonderful questions in the chat. 

Harihar, you’re first. I’m going to unmute, so please read it or state your question. 

Harihar Milaganur Mohan: Yeah. And thank you. This was a wonderful talk. So I was just wondering how the LY drug affects other condensates in the cell. So like nuclear condensates, or someone else asked about stress granules. So have you looked at that? And then I was also wondering if cells treated with the compound are more stressed or like, they activate ISR pathways and also in terms of cell health what the consequences are. 

Rui Tong Quek: Yeah. So to answer your first question, we actually didn’t look at other condensates in the cell. But that’s actually a good point to look at. So one thing we know is that GSK3 is actually involved in a ton of cellular processes, and I think this also links to your second question. So, in fact, even other condensates like TDP-43 are also somewhat modulated by GSK3. So you’re right in that we should also, in terms of the biology, look at other condensates. But we personally haven’t done that. So that’s a great follow up. And it’s true that cells are a little bit less healthy upon treatment with these compounds. Just because GSK3, you know, again, it’s involved in like a lot of cellular processes. And that’s also partly why we didn’t pursue GSK3 therapeutic further, just because it’s also oncogenic. So there’s a lot of side effects. But you know we were really hopeful that this is a demonstration of sort of this method of screening for looking for future potential condensate modulators through this type of screening assay. 

Harihar Milaganur Mohan: Great. Thank you. 

Diana Mitrea: Okay. We have a question here in Boston. 

Sayantanee Niyogi: Hi, this is Sayantanee. Great talk. Question on the second part of the talk where you talk about the technical other ways of looking at condensate screening. So in the data where you showed in the kinetic assay that the incubation time with the compound that could affect how sensitive the assay was. Did you by any chance follow it up with trying to see if compound inhibition is also, or you know, the concentration of the compound that can also show sensitivity to the assay? 

Rui Tong Quek: So we didn’t look too much into that. But we definitely thought about it. So we thought about it in terms of like, could we actually screen at a much lower concentration and also see equally robust readout with a good like Z prime. So that’s a totally valid point. We didn’t look at it for the kinetic assay, but I would imagine that that might be another benefit, too. 

Sayantanee Niyogi: Okay, thank you so much. And just one last question is, when you were comparing the GSK3 beta inhibitors, I think, then, is one LY and the other CP. Did you do a side-by-side study to see all the complementary methods where the screening was, and then follow it up with the NanoBIT assay and then just, to see if the highest potent compounds showed the same effects in three complementary assays? 

Rui Tong Quek: Yeah. So for the side by side comparison, we only did it for the three compounds that I showed plus another one, so we didn’t do it for all. But we did it for four, and we could see that it tracks for the NanoBIT and NanoBRET we did it for sure, so not all of it. But that would be something a great complementary experiment to do as well.  

Diana Mitrea: Great. Thank you. Now we’re going to move back to Zoom, and we’ll ask Melissa to ask her question. 

Melissa Keener: Hello! Thank you for the great talk. My question is sort of related to the one that Harihar asked. You mentioned that the cells are unhappy with treatment with the compounds. Are the cells able to clear these solid-like condensates that are formed are these condensates themselves toxic to the cells? 

Rui Tong Quek: Yeah, so they are. The condensates are somewhat reversible. We’ve seen that by imaging we’ve also seen that with the high throughput SMT assay. So you can see sort of a reversal once you wash out the drug, so I’m not sure if down the line, like even after wash out, if the cells are severely affected, but it seems like it’s somewhat reversible at the very least. So I have a partial answer to the question. We didn’t actually dig really deeply down the biology post treatment. That’s a good thing to consider. If we were like, you know, taking this further for development for a drug.  

Diana Mitrea: Okay, thanks. So we have another question here in Boston. 

Gwyneth Welch: Hi, great talk. I’m curious about the relationship between the phenotypic assay and the high throughput SMT assay. So you said that at conditions where you see changes in kinetics with the SMT, you actually can’t detect condensate formation with a phenotypic assay. I’m curious about your thoughts about the relationship between those two assays. Would you in general expect compounds that have a positive phenotype in the IF assay to also show changes in the SMT and vice versa? 

Rui Tong Quek: Yeah. So I think condensation is also really dependent on the environment of the cell, and also the fusion protein itself. So with the IF assay it was in A549 cells and it’s tagged to GFP, which is somewhat it’s somewhat self-dimerizes. So there’s a certain amount that is, of condensation that’s actually favored because of the tag. And also, you know, cell line differences. So with the high throughput SMT assay we were doing it in U2OS cells, and it’s fused to the Halo tag. So there’s slight differences in the condensing protein itself. And we do know that GFP, you know, promotes the formation of condensates. So, it wasn’t exactly too surprising that they were different, but we were still convinced, because even in the high throughput SMT assay we could see puncta with treatment of the drug. It just wasn’t as dramatic as in the cell line where N was fused to GFP. But I think that’s like a thing for the field in general is also to just consider, like, you know, exogenous tagging of the protein is always going to alter condensation and propensity to form condensates. So that’s a really great point. 

Diana Mitrea: So then we’ll go to Bede. Go ahead and ask your question. 

Bede Portz: So that was a marvelous talk. I really appreciate it. For the first part of the talk did you stain for GSK3 and is it in the condensate? And then, sort of relatedly, do you think it is the phosphorylation of N that is responsible for modulating its condensation? Or do you think it is a protein-protein interaction, you know, could be GSK3, or it could be something else that’s recognizing the phosphate that is, that’s regulating its partitioning? 

Rui Tong Quek: Yeah. So we tried staining for GSK3, but I think there were technical issues with it. So not a very good answer to your question. But we weren’t very sure. To answer the second part of the question, I’m not entirely sure, based on the data that we’ve collected, that it’s whether it’s phosphorylation itself, or interaction with another protein, and I think, the biology of it is also really interesting, considering that this phenomenon is consistent across all 7 Ns. So I think it’s definitely something to look at. But as far as we know, the phosphorylation is responsible for changing the phenotype of condensation, but I’m not sure whether a secondary protein is involved. 

Diana Mitrea: So we’ll move to another question here in the Kitchen Table. 

Sidharth Sirdeshmukh: So yeah, great talk, it’s always really cool to see people doing research with a combination of high content imaging techniques. I enjoyed listening to your presentation. So I have a question about the single molecule tracking assay in particular. How do you know you’re measuring the same protein over time? Obviously, you know, two of the same proteins can get quite close to each other, and you may end up kind of switching over to a different track. And so I think that’s probably a big challenge for that technique. But another question is, how do you know you’re not picking up background? I know that you mentioned one of the limitations with the traditional high content phenotypic screening is that you could be picking up background. But it seems to be also a challenge for single molecule tracking. So that’s the first question on the technique. But the second question is about, related to Gwyn’s question like, you’re seeing this concordance between the high content, phenotypic assay and single molecule tracking with the curves. If you just look at the curves on the slide you present, they look exactly the same, and I just wonder, is the decrease in speed of the proteins as well as the puncta deformation potentially driven by an effect on cell viability? I think someone asked a similar question, but I think that dissecting the kinetics of those two effects would probably be really valuable to know what causes what. 

Rui Tong Quek: So to answer your first question, I’ll be honest and say that I’m not really the expert in the image analysis part of the high throughput SMT. So in terms of making sure that you’re matching the right molecule to the right molecule in each frame. So there’s a certain sort of boundary, and distance that is included in the pipeline to account for the matching of the molecules correctly. I think there is actually a publication that that sort of details this analysis pipeline in much more detail than I can explain. So I think if you might want to refer to that, and that might be that might help answer more of the questions that you have about the analysis itself. I will admit that I’m not the expert in the field for this, for this part of the analysis.  

And for your second question about cell health. So it is true that some of these compounds are somewhat toxic to cells at high concentrations. So for us broadly, we were also testing a lot of intermediate concentrations. So where the cell health is not impacted or not visibly impacted by various metrics, and we still see the trends are similar, or, you know, in concordance. So at high concentration, that’s where you see the maximal effect, but that’s also when you see some deterioration of cell health. So I think again to the question that someone posted earlier performing a lot of these assays at a lower concentration like 0.5 micromolar might actually be also very significant and important to the story. 

Diana Mitrea: Thank you. So we’ll take a question from Julian on Zoom, and then immediately after one, for from Hicham, which are kind of related. 

Julian von Hofe: Hi, yes. So I actually also had a question relating to the single molecule tracking pipeline, specifically when I was looking at the figure that you showed of the distribution of diffusion coefficients. I was just kind of curious how you were extracting these diffusion coefficient values. Are those being assigned to each individual track? Is there any like ensemble averaging that’s being performed? So I was just wondering if you could talk more about that pipeline. 

Rui Tong Quek: So I believe that the jump length is the raw data. So every count is an individual jump length of a single molecule. The diffusion coefficient state is an inferred distribution. So I believe that it’s not one molecule is one diffusion coefficient state. But again, I would also encourage you to refer to this publication, that we have a preprint on this on this paper that links to the original description of the pipeline, so that might help to answer your question much better than I can. 

Diana Mitrea: (reading Hicham’s question) For the SMT part, curious to know which of the different parameters measured would be robust enough to derive SAR for compound optimization? 

Rui Tong Quek: Any of the parameters would be fine to do. They all have pretty much almost the same Z prime, if you were to calculate it so I think any of these metrics would be good for doing SAR. More specifically for you know, condensate biology, we sort of derived the slow proportion metric here, as like a more direct readout of the biology that we want to look at. So what is slow, what is fast? So I think you know, any one of them would work, but just looking at slow proportion, that might be a simple and more direct metric. But any of them, would work, because they all have the same Z prime sort of dynamic range. 

Diana Mitrea: Great. Thank you. So we’ll take one last question from the Kitchen Table. 

Jesse Lai: Thanks for your talk. I’m really curious about what happens to the viral condensate. Do they have adjuvant properties? Are they more accessible by the immune system? Or are you just sort of delaying you know, when that condensate dissolves? Do you get the infection again or are you allowing the time for the immune system to catch up, so that once it dissolves, if it dissolves then you can sort of tackle that? 

Rui Tong Quek: So I think the viral condensates themselves are means for the virus to escape the host immune system. So like sort of protecting the double stranded RNA that would typically trigger an immune response in cells. So it’s involved in all the cellular processes of the viral infection. So I don’t specifically know if the condensates themselves, how they contribute to the progression of the virus or the immune response. But I think from the virus side of things it helps the virus a little bit. But it’s true also that the N protein, the N condensate itself has well, maybe not the N condensate, but the N protein itself has certain effects on stress granules. So I think someone in the chat asked about stress granules because it’s known that the N protein binds to G3BP1, which is a core stress granule protein. So we have seen with N condensates there is a suppression of the formation of stress granules, be it through, you know, N protein sequestration of G3BP1 or other methods, or through RNA. So there, I think there is some interplay with the immune system that the virus is sort of evolving. But yeah, that’s you know, as far as I can conjure. 

Diana Mitrea: Great thank you so much, Rui, for a wonderful discussion and a wonderful presentation. So please join me in thanking her, and if there are any additional questions that come up after the talk, feel free to use the contact button on Condensates.com, and then I will relay the questions to Rui. 

Rui Tong Quek: Thanks so much for having me. 

Diana Mitrea: Thank you. Bye everybody, and see you at the next KTT.