|Type||Kitchen Table Talk|
On March 24, rising star Shana Elbaum-Garfinkle, from the Graduate Center of the City University of New York, delivered a fantastic Kitchen Table Talk for Dewpoint and the Condensates.com community. Shana received her Ph.D. from Yale in molecular biophysics and biochemistry and then did a postdoc with Cliff Brangwynne at Princeton. She has since set up her own lab to study protein phase separation with the help of a prestigious career award from the National Institute of Neurological Disorders and Stroke. She uses a wide range of technologies including single molecule fluorescence, soft matter material science, and C. elegans genetics.
Shana’s career in condensates has already led to some amazing publications. Shana was the lead author on the highly cited 2015 PNAS paper which characterized the P granule protein LAF-1. She co-lead authored the very important 2015 Molecular Cell paper with Huaiying Zhang (see also Huaiying’s talk from last month) which demonstrated it is possible to fine-tune material properties of condensates using RNA. And she also collaborated on an important 2017 Nature Chemistry paper that further studied the role of IDRs in the function and behavior of LAF-1 and provides interesting ways to measure the mesh sizes of droplets.
Shana’s talk delves into her most recent work, published in Nature Communications, demonstrating how the different properties of arginine and lysine can drive tunable regulation of droplet formation and behavior. Watch the video below to join us in thinking about how arginine and lysine contribute to the miscibility of phases. Shana also raised the deep and profound question of how to pronounce “condensate.” At condensates.com, we don’t care how you pronounce it as long as you keep talking about condensates, but you can check out the results of our Twitter poll here…
Create an Account or Sign In to view the video.
Mark Murcko (00:00):
Hi, everybody. It’s great to see everybody here. It’s exciting today to have the chance to hear from Shani, who has of course now got her own lab at City University in New York, but she’s been doing amazing science for already so many years. She started out getting her PhD at Yale in biophysics and biochemistry, and then she did a postdoc with Cliff at Princeton, and then, of course, has set up her own lab and she’s, it’s of note that she’s won a very prestigious career award already from the National Institutes of Neurological Disorders and Stroke to help her get the lab up and running.
Mark Murcko (00:36):
She really is a rising star in this whole condensate field. Amazing accomplishments already. Publications on Tau, Alpha Synuclein on the use of microfluidic methods to study condensates. She was the lead author on a very highly cited PNAS paper in 2015, which looked at LAF-1 in P granules. She was the co-lead author on another 2015 paper in Molecular Cell and Cliff was on that and also Amy Gladfelter that demonstrated the nature of these critical poly-glutamine interactions with RNA, and also was important because it demonstrated that it was possible to fine-tune the material properties of condensates using RNA.
Mark Murcko (01:23):
The last Kitchen Table Talk speaker we had, Huaiying was also the other co-lead author on that paper, so obviously, we all here think that’s great paper. She’s also just been active in the field more generally. She’s a co-author on a very important Nature Chemistry paper from 2017 that studied the role of IDRs in the function and behavior of that same protein LAF-1. There’s a really interesting trick in that paper, the ability to measure the mesh size of condensates and really getting at the whole length scale question of diffusion, very important problem.
Mark Murcko (02:00):
And now at CUNY, her lab, she’s continuing of course to study phase separation, in particular, with an emphasis on neurodegenerative disease and the aggregation of proteins in that space. And it’s a wide variety of technologies that she uses, single molecule fluorescence, soft matter of physics, genetics in C elegans. Obviously, a strong emphasis on RNA granules, and she just continues to do great work and pump out really interesting results that all of us should really be paying very close attention to.
Mark Murcko (02:34):
So, it’s really an honor, Shani, to have you here today, and your talk is titled Emergent Material Properties of Biomolecular Condensates. So, thanks for doing this. It’s great to have you.
Shani Elbaum-Garfinkle (02:43):
Wow. Thank you so much, Mark. That’s really an incredible introduction. I don’t even know how I can move on from there, but thank you so much. It’s really an honor to be invited and included in this really, really great series, this Kitchen Table Talk series. And so today, I’m going to hop around a little bit, but mostly speak about a recent paper where we’ll use model condensates, but talk more broadly about our focus on emergent material properties and why we love them, why we care so much about them, and why we try desperately to measure them in many different ways…
Shani Elbaum-Garfinkle (03:21):
So, if there’s ever a talk that does not need an introduction to condensates, it would probably be this one hosted by condensates.com, and actually I think it was a talk or two ago in the series that I believe it was Mark, and I hope I’m not misquoting you who at some point jokingly said, “Well, all roads leads to condensates.” I think that’s probably true or at least seems true if you follow the literature.
Shani Elbaum-Garfinkle (03:48):
And so here this is sort of now this famous picture, this pictorial summary of which is already now probably rather outdated. But it was in this important review in 2017 that summarized in just short almost less than a decade the variety of different granules, bodies, transient clusters of molecules that then was coined to be termed biomolecular condensates. And so I believe this term condensates, right? So, inherent to this is the understanding that it’s really sort of an inclusive term, right? So, sort of borrowed from the field of soft-condensed matter physics.
Shani Elbaum-Garfinkle (04:35):
It’s inclusive in the sense that some of these granules, at least, initially were shown to have liquid properties such as the P granule and the seminal 2009 paper by Cliff and Tony Hyman. Cliff Brangwynne and Tony Hyman, and then, of course, there’s a variety of bodies now with a variety of different material properties. And so biomolecular compensates really is a nice inclusive term that takes into account this spectrum of properties.
Shani Elbaum-Garfinkle (05:00):
And so, again, as the name implies, condensates are inherently defined by these properties. These emergent properties and emergent behaviors, and so in the title, and I talk a lot about emergent behaviors, the word emergent of course denotes the fact that these features are novel. So, the features that we see in these condensates are novel, and they emerge on the macro scale. On the scale of this new bulk network of interactions that occurs, and so, of course, a classic example from again the 2009 Science paper would be the emergent property of wetting or flowing.
Shani Elbaum-Garfinkle (05:45):
And so here these P granules that we see originally they sit and they wet the surface of the nucleus and upon shear stress, so this is in the adult C elegans gonad. We can see these flowing and dripping and wetting, and this was one of the first indications that these P granules since they have displayed this type of behavior can be considered some sort of condensed state or liquid-like state as it was called at the time.
Shani Elbaum-Garfinkle (06:13):
Of course, in 2011, with the nucleolus, Cliff and Tony went on to show that other properties such as relaxation to spherical shapes, formation of liquid bridges, coalescence, again, sort of these emergent behavior was seen in the nucleolus. Emergent properties also give rise to novel local environments, right? So, when you have phase separation that occurs, you now have a novel environment that is unique inside that condensate which can only occur if there’s a condensate, right. So, this is an emergent property.
Shani Elbaum-Garfinkle (06:49):
Finally, multi-phase coexistence first shown in the nucleolus in 2011, and now shown in several other systems, in vivo as well as in vitro for different proteins. This really also emerges from the unique properties such as viscosity and surface tension that give rise to this hierarchical organization of phases. So, how phases that have distinct properties then or self-organize amongst each other in a system.
Shani Elbaum-Garfinkle (07:22):
Okay. So, as I just mentioned, the term condensate or condensates, I guess depending, I don’t know if that’s a 50/50 split in terms of pronunciation. I actually flip flop between the two. I don’t know why. So, it’s intentionally inclusive I believe, and that denotes the fact that materials really exist like many things on a spectrum. And so on the one end of the spectrum, you approach something that’s sort of an ideal liquid or an ideal Newtonian fluid, which is defined always by this certain viscosity and a certain surface tension.
Shani Elbaum-Garfinkle (08:03):
The other side of the spectrum, you have sort of the idealized solid or Hookean solid. Something that’s defined by this elasticity, this constant elasticity, and then in the real world, you really have a spectrum of materials that fall somewhere in between where unique blends of viscous behavior and elastic behavior can exist and coexist and/or dominate at different length scales and under different conditions.
Shani Elbaum-Garfinkle (08:35):
So, in my group, we’re really obsessed with measuring and quantifying these properties. And we care about this for many reasons, of course all the reasons I just described, so material properties are inherent to the function and even just the definition of these things, these condensates. So, not only in the way they wet or coalesce or fuse, but also in the way the internal molecules are organized.
Shani Elbaum-Garfinkle (09:08):
Material properties are also inherent in the discussion of pathology. So, there’s increasing evidence that’s really sort of implying that changes in material properties–so, the exact definitions of those properties, how something can go from something liquid-like to something solid-like–this process is being implicated in a variety of diseases, including neurodegeneration which is one focus of the lab, which I’m actually not going to, unfortunately, really talk about today in this talk, but it’s something that we care a lot about.
Shani Elbaum-Garfinkle (09:43):
We also are really interested in understanding and quantifying properties from a design perspective. So, understanding the fundamental principles that govern how molecular level interactions translate to certain material properties is incredibly useful in thinking of leveraging that information in the design or engineering of novel materials that are bio-inspired. And this is just a classic example of the spider silk that has been just beautifully understood and engineered and even improved in bio-inspired models.
Shani Elbaum-Garfinkle (10:26):
Finally, perhaps most related to the mission of Dewpoint Therapeutics is we believe that quantifying emergent material properties is crucial for building therapeutic strategies to target condensates. And so I really consider this to be analogous to the way crystal structures have been invaluable in allowing for the development of targeted drug therapies to understanding of three-dimensional structure, the three-dimensional map of a protein allows for efficient targeting. So, too understanding the three-dimensional network configuration of the condensate, I believe is going to be crucial for targeting condensates for future therapeutics.
Shani Elbaum-Garfinkle (11:17):
And so for these reasons and because it’s really cool and beautiful, which is implied. We study and measure material properties of condensates. And so one big thing that we asked, so underlying all of these reasons as an approach or question we ask, which is: how do molecular interactions give rise to these network properties, properties that are novel that emerge from those interactions?
Shani Elbaum-Garfinkle (11:44):
This relationship is actually not very intuitive or easy to predict, and so we’re interested in really understanding that relationship. And so to do that, we do spend some time with reductionist approaches both on the protein level, so maybe a protein nucleic acid, and then even more reductionist where we just use polymers to really try to go to the very fundamentals and say, “What does one residue contribute to a condensate versus another?”
Shani Elbaum-Garfinkle (12:15):
And so one of the methods that we use a lot that we really like is passive microrheology. And so rheology, of course, is the study or science of material deformation and material response to force or stress, and in passive microrheology, it’s actually, it’s quite relatively simple. So, in passive microrheology, you don’t apply any external stress or external force, you just use leverage thermal fluctuations, and you look at particle diffusion within a material, and so this doesn’t really work for very, very, very solid materials and it works for things that are closer to the viscous spectrum.
Shani Elbaum-Garfinkle (12:58):
And so, here, these are just beads diffusing around in water, and so we understand the diffusion of beads and if we track the particles, and we plot the mean square displacement as a function of lag time, we expect this relationship. So, the MSD will scale with the lag time with this exponent alpha and alpha is equal to one if it’s a pure viscous fluid. So, if it’s a pure viscous fluid in two dimensions, the mean square displacement is just going to equal four Dt, the diffusion coefficient and. In this case, we can actually extract the viscosity directly using Stokes-Einstein.
Shani Elbaum-Garfinkle (13:34):
Now, it’s interesting to note. So, when you do have viscoelastic behavior or in the extreme you have elastic behavior, you start to lose that linear time dependence, and so in the extreme elastic case there’s no time dependence to the bead motion over time. And so using this method, we can extract information about the viscoelasticity and then we can explicitly define and quantify viscosities.
Shani Elbaum-Garfinkle (14:00):
And so just because, especially given the introduction and Huaiying’s talk yesterday, I thought I would just do two little throwbacks from the early application of this method in condensates. And so, as Mark mentioned, I initially worked on developing this method and applying it to condensates back in 2015 as a postdoc with Cliff Brangwynne, and in this case we were starting with just the P granule system, which at the time was just sort of liquid-like looking down into the embryo seeing it look like a liquid, and I don’t have time to talk about how fun 2015 or 2014-2015 was in terms of trying to build those first in vitro reconstituted systems, right? Which proteins are involved, which ones were actually formed? But long story short, we had identified the protein LAF-1, which phase separates in vitro.
Shani Elbaum-Garfinkle (14:50):
It has an RGG domain, an arginine/glycine rich domain, which was necessary and sufficient for phase separation. We were able to quantify the viscosity of those condensates very reliably, and importantly show that viscosity can be not only quantified, but then it can be tuned, and specifically it can be tuned by addition of RNA binding partners. In this case, just sort of short RNA polymers. In this case, make it more fluid and decrease the viscosity.
Shani Elbaum-Garfinkle (15:21):
Just after that in that very same year, I had a real pleasure to work with Huaiying and collaborate with Amy Gladfelter and working on a different system, which is the Whi3 protein that we just heard about last month I guess, and in this case, we show that not only can RNA tune the viscosity of a droplet, but actually the specificity of the RNA can specifically tune that viscosity leading to condensates with unique material properties, and that’s another really exciting area.
Shani Elbaum-Garfinkle (15:55):
And so LAF-1 as I said was driven by this arginine/glycine rich domain. Long story short, it is now known that arginine/glycine domains are really, really important. I couldn’t list all the references here nor all the proteins that have arginine/glycine rich domains. This is just a few in a nice review from the Forman-Kay Lab that talks about the arginine/glycine rich domain, but a variety of P granule proteins, nucleolar proteins, stress granule proteins have RGG domain or sometimes called an RG rich domain that is very important for phase separation and necessary for driving it.
Shani Elbaum-Garfinkle (16:34):
Of course, and now Kitchen Table Talk, I don’t know, a year ago or so from Rohit Pappu, if you don’t know about the sticker and spacer model, this is a very important theme in the area of understanding the molecular grammar, right? So, the grammar of understanding how sequences and how residues, and their identity, and their spacing are important for driving phase separation. And so arginine, of course, in this scenario is a sticker. So, it engages not only in electrostatic interactions, but can engage in cation-pi interactions and is showing up in a lot of phase separating systems.
Shani Elbaum-Garfinkle (17:18):
None the least of which are the ALS dipeptide, arginine rich repeats. So these associated with ALS C9orf dipeptide repeats that are arginine rich have been shown to alter both stress granule dynamics as well as nucleolar behavior and dynamics.
Shani Elbaum-Garfinkle (17:43):
So, arginines are important in driving phase separation. They’re also implicated in maybe perturbing phase separation. So, this is something that we became interested in and specifically because arginines and lysines seem to show very different behavior, and so this is one example, there are several more now that show. For proteins that do rely on their arginine rich domain to phase separate, arginine to lysine mutations are not conservative. So, this is an example from the Forman-Kay Lab on FMRP, which is a neuronal granule protein, and when the arginines are mutated to lysines you get differences in phase behavior.
Shani Elbaum-Garfinkle (18:29):
And recent work has also shown using FRAP and some coalescence that for dipeptide repeats as well as short synthetic sequences, swapping arginine to lysine leads to apparent changes in material properties or emergent properties. So, you get better recovery from FRAP, faster coalescence for lysine than arginine. And so this is really fascinating, and I think there is a lot of literature even outside of phase separation that points to the unique differences between arginine and lysine despite their conservation of charge, the distribution of charge in the guanidine group of the arginine does give it some sense of, some additional multivalency or additional versatility in interactions.
Shani Elbaum-Garfinkle (19:17):
And so we thought this was a great case to look at very specifically to understand and to quantify specifically the unique contributions of arginine and lysine, so we can understand, again, these principles from going from molecule to a material and how do these differences translate to these new emergent behaviors. And so in this case, we went really reductionist on this one. So, we really just took homopolymers of poly-lysine, and we took homopolymers of poly-arginine. Of course, glycines are very important, and we hope to build up to that understanding, but we wanted to look at this in just the purest way possible to compare these residues side by side.
Shani Elbaum-Garfinkle (19:58):
And so, in this case, we have homopolymers of each of increasing length, 10, 50, or 100, and an anionic binding partner of polyU RNA of either 10 or 50 length. And in some cases, UTP actually can also serve as an anionic partner and in some cases UDP. And so here’s just an example of polyK100 with UTP. It forms these beautiful droplets. And everything unless I state otherwise, at this point, since this is a complex coacervate system, we just make everything charge matched. So, if we alter anything everything is compared in a charge matched scenario.
Shani Elbaum-Garfinkle (20:44):
And so we apply our favorite technique. And so these are, I believe these are 500 nanometer beads. We actually use a few different sizes from 100 or 200 micron or so, and we can measure the viscoelasticity. And so here these are all of the beads that we track in this one very large droplet of poly-lysine. In this case, I think this is polyK100 and UTP. We see this slope of one, so we can extract the viscosity.
Shani Elbaum-Garfinkle (21:18):
Actually, we can also plot the distribution of displacements for all the average over all of the beads, over all of the in the system. And we see that these actually are Gaussian fits which suggests, which gives us another piece of information that this condensate is completely homogeneous. So, every bead is in a similar environment. This is consistent with idealized Brownian motion. And perhaps not surprisingly when we increased the length of the poly-lysine polymer, so this is in the scenario with UTP, but we’re just looking at poly-lysine 10, 50, and 100.
Shani Elbaum-Garfinkle (21:58):
We can increase the viscosity with length and this is not surprising, but we were just playing with our system. We can increase it both with poly-lysine length as well as increasing the anionic partner. And so we asked how this compares to poly-arginine, and so before measuring the properties we actually noted there are some important differences. So, there are differences in the phase propensity altogether. So, in the case of UTP where poly-lysine of each length formed very nice droplets.
Shani Elbaum-Garfinkle (22:31):
Actually, the poly-arginine 100 form aggregates and that was the only case where we saw that which was a difference, and interestingly, poly-lysine was not able to form condensates with UDP, but poly-arginine was. Nice liquid droplets with UDP. And so this already suggests that the increased strength, interaction strength and/or an increase of valency of the poly-arginine and the nucleotide interactions is giving rise to new behavior. So, sort of perhaps the polyR100 and UTP the interactions are so strong with the increased length, you actually get an aggregate and we don’t see droplets at all and UDP is not multivalent enough to form droplets with the lysine, but can form with the arginine. So, perhaps the arginines contribute that valency. We don’t have a model for this right now, but that’s what we think.
Shani Elbaum-Garfinkle (23:29):
Then, we can construct a phase diagram just to sort of test what we would suspect, which is that the arginine droplets here in purple are much more stable in response to salt, so they can persist with higher salt. So, the interactions are stronger and able to withstand higher salt than lysine.
Shani Elbaum-Garfinkle (23:49):
Okay. So, then we go to the rheology, which is the more exciting thing, because we want to quantify things. We want to have metrics that we can compare directly. And so when we do this, we initially, it looks like the beads are not moving at all. This movie, two movies are playing, but in the arginine case, we hardly see any bead motion. But you zoom in and you do see that they’re moving but just ever so slightly.
Shani Elbaum-Garfinkle (24:11):
And so what we do find, of course, we started this project expecting there to be a difference and hoping to quantify it, but what we were really surprised to see was how big that difference really, really was. And so the difference, this is all on a log scale. The difference that we see, so in the case of UTP between poly-lysine and poly-arginine is several hundred, a factor of several hundred. I think that anywhere between 100 and 300 higher viscosity for poly-arginine than poly-lysine in situations where everything is otherwise perfectly analogous And of course, this is a much bigger difference than just increasing the length of the polymers.
Shani Elbaum-Garfinkle (25:01):
So, something that was really surprising, and so the poly-lysines are maybe something on the order of a syrup, where it can flow, it’s not too bad. But the arginines are really more like a paste or something very, very thick. And so at this point, this was actually meant to just be originally like a control of some sort, and we were going to move on to other things and build on this, but we were so interested in this big spectrum of viscosities, we thought this could be interesting from a design perspective to be able to span such a large range. Could we mix or combine our lysine and arginine materials to form a more complete spectrum?
Shani Elbaum-Garfinkle (25:44):
So, could we sort of occupy and make, sort of tune a viscosity range between these extremes just using these few components? The answer is no. We couldn’t. Because whenever we did that, we either got exactly, no matter what combination we mixed them. We either got exactly the viscosity of the arginine itself as if the lysine wasn’t there or the data looked terrible and we got nothing.
Shani Elbaum-Garfinkle (26:14):
And so one thing I should add, one nice thing about microrheology is that since we really only care about the particles and the particles are fluorescent, in these assays, we actually don’t care, we don’t need the protein to be labeled for these assays or the droplet to be labeled, because we can see that with just DIC. And then we can see that the droplets are in there.
Shani Elbaum-Garfinkle (26:34):
Actually, that’s really, really nice because that’s a good way to test for label artifacts, so we can do it with label and without. And that’s just a plug. This is just a microrheology plug I’m making for some reason, but because of this, we thought, “Well, this is really strange. What’s going on?” And so we decided, we labeled our lysine and arginine. And I guess if you’ve already seen the social media plug for this event you know that this happens. So, this was really shocking, and this wasn’t at all expected going into this project.
Shani Elbaum-Garfinkle (27:05):
The arginine and lysine, they do not mix. So, the phases are, they’re so different and they’re different enough and they’re unique enough that they actually form coexisting liquid phases where here the arginine is labeled with 594. The poly-lysine is FITC-labeled. And the arginine is always nested within the lysine. And so initially, when we were trying to like dial in all these different viscosities, either we were just looking at the arginine phase, which was always in the middle and sometimes the lysine is just the shell or we were looking at beads stuck at all these interfaces, and so the data didn’t make any sense.
Shani Elbaum-Garfinkle (27:41):
And so this was beautiful and I had this printed out on my desk for a really long time and just stared at it. I continued to stare at it. And again, sort of very surprising, and this sort of goes back to my intro statement which is that these relationships are very hard to predict. So, it’s very hard to predict how you go from an amino acid, a particular amino acid to a viscosity. And it’s even more so not necessarily known, certainly, how and when you would get immiscible phases considering these are both like charges, and they’re both interacting with anions.
Shani Elbaum-Garfinkle (28:13):
All right. I’ve made my point. This is exciting. Okay. And so we know more information. So, since the arginine is nested within the lysine that already tells us something about the relative surface tensions. And so we wanted to test this by doing a droplet coalescence experiments. And so here the arginine is not labeled, but the lysine is. You can see that the arginine is fusing much less, much more slowly than the lysine.
Shani Elbaum-Garfinkle (28:43):
And so this is quantified here. So, poly-lysine fuses much more quickly than the poly-arginine. Actually, on the right here, the bottom right, we can actually compare the coalescence of the droplets when they’re just alone or when they’re in the mixed system, and that’s symbolized by the circles and the crosses and they actually align perfectly and so somewhat perfectly. And so we believe that the properties in these separate phases are consistent with the co-existing phases, and this also suggests that the surface tension of the poly-arginine is 10 times higher than the surface tension of the poly-lysine, and we would expect it to be in the, nested within.
Shani Elbaum-Garfinkle (29:34):
Okay. So, everything I’ve just showed you now is in this charge-matched situation, which is sort of predicated that this part of the talk with. So, there’s enough anion to satisfy the amount of cation, even in the case where we add the two polymers. And so we asked, “Well, what if we had different amounts, insufficient or excess?” And so when we add excess, in this case, it’s UTP. So, when we add excess UTP, we continue to see the same thing, but interestingly, when we have insufficient UTP, so there’s not enough UTP to satisfy the coacervation of both the arginine, lysine polymers, we actually only see the poly-arginine droplets. I guess, so, yeah, which is consistent with the fact that the arginine nucleotide interactions are much stronger than poly-lysine, and so it’s really outcompeting the poly-lysine.
Shani Elbaum-Garfinkle (30:34):
Then, we thought, “Well, if it’s such, if they’re really dominating the nucleotide interactions so strongly, what happens when we look at the system out of equilibrium?” And so what we’re doing here is we’re making poly-lysine droplets with enough UTP to make nice beautiful poly-lysine droplets, and then, in this case, instead of mixing them all together initially and looking at them, we’re adding poly-arginine to the system and seeing what happens.
Shani Elbaum-Garfinkle (31:06):
And so this is essentially the limiting case once you add the arginine. So, there’s only enough UTP to satisfy half of the total polymer. And so even though these poly-lysine droplets are already formed, once you add the poly-arginine, you see them really just sort of going straight for the droplet, scavenging whatever UTP is there, and essentially releasing the lysine to the buffer and replacing the droplets. We call this droplet inversion. Very pretty.
Shani Elbaum-Garfinkle (31:41):
And so we can quantify this here and so in the top, we can see that the slide coverage changes. So, the lysine disappears and the arginine sort of takes over. And then in the bottom panels, if you just zoom in into one droplet, and this is apparent in the movie, we see that the arginine is actually nucleating within the lysine droplet. And actually on the right what we can do is we can quantify the poly-lysine, the FITC-poly-lysine signal inside and outside of the droplet. And it’s actually hard to see, because of course once the lysine goes to the buffer it’s much more dilute, but if we quantify it, we, of course, see that we’re getting this burst of poly-lysine out into the surrounding buffer as this process is happening and arginine is replacing it. So, poly-arginine is outcompeting the poly-lysine for the available UTP.
Shani Elbaum-Garfinkle (32:41):
Then, we were curious, in this case, so what about when the UTP is not limiting and even in excess, in this case? So what do the non-equilibrium dynamics look like? And so here all these movies are playing at the same rate. And here we’re repeating the experiment, and we have either just enough or excess UTP, and so we know we end up at equilibrium with the same two-phase state, but the difference is that here, the kinetics of this inversion have changed. And so it gets slower and slower for 3 millimolar and 4 millimolar UTP, and then at 15 millimolar there’s so much extra UTP, we can’t really capture it at this time scale.
Shani Elbaum-Garfinkle (33:25):
So, we can actually quantify this kinetics. And so what we’re essentially showing here is that increasing the stoichiometry of UTP is a way, in this system, to modulate the kinetics of this inversion, but also modulate the kinetics of the release of the poly-lysine into the buffer. So, we like to think about the release aspect of it, because we think from a design perspective if this were built into a more complex system this could be a way to tune the release kinetics of some say compound or other molecule that associates with one phase versus another. And that’s something that we’re working towards developing or thinking about more.
Shani Elbaum-Garfinkle (34:09):
And so we also asked if we can tune this kinetics in other ways, because UTP concentration is a little extreme. What if we just change the length of what we were looking at? In this case, we’re repeating the same experiment, except here the only difference is that instead of UTP being the anion, we’re just changing to pU10 versus pU50. And so here all the concentrations are exactly the same, but just increasing the length of the anion in the system drastically alters these kinetics as well. And that is another interesting sort of viable knob at which to tune a system like this, which we thought was really fascinating. That’s just quantified here is that we can play with these kinetics, and they can be increased or decreased.
Shani Elbaum-Garfinkle (35:07):
Okay. So, I think timing okay. So, I started by saying that these relationships are not easy to predict. And I think this story really highlights that, because we expected differences between arginine and lysine; we didn’t necessarily expect them to be so different, so different that they would be immiscible in most cases. And we found that we were able to actually control the dynamics of these coexisting multiple coexisting phases by playing with just either the length or the concentration of these very minimal components.
Shani Elbaum-Garfinkle (35:43):
And so this really lends insights into the unique roles of arginine versus lysine. And so let’s go back to the situation where arginine to lysine mutations alter phase separation. Well, you bet. I mean based on this you would really expect based on the differences in phase behavior that arginine and lysine would contribute to very different behavior. And we’re not even considering post-translational modifications yet, in this case, and that’s clearly another way to build a complexity in the system.
Shani Elbaum-Garfinkle (36:14):
Additionally, we’re thinking this can provide a framework for innovative use of arginine and lysines, and designing tunable droplet systems that could be orthogonal, and so we’re sort of interested in that as well from a design perspective. And so I’m just going to close with a summary again of sort of our, how we approach these problems in the lab. We measure these material properties and we try to understand these emergent behavior both to understand function and dysfunction. I didn’t… I’m glad. Okay. We didn’t really have time to talk about pathology here design, and then ultimately therapeutics down the line.
Shani Elbaum-Garfinkle (36:56):
And so this is the group and so I just want to thank. So the bulk of this work was done by Rachel Fisher, a very talented postdoc in the lab. And her work was just featured, she won the first place in the Biophysical Society Art Contest, and this was using the same system, so compete or coexist using the poly-lysine, poly-arginine system. And with that I will take questions.
Mark Murcko (37:28):
Great. Thank you, Shani. Wonderful lecture. So, I think Jill wanted to solve this question once and for all. Is it CONdensate or conDENsate or condenSATE? So, everyone has to vote. We’ll all vote while the questions are coming in. There’s already a few.
Shani Elbaum-Garfinkle (37:51):
Does it have to be constant? I mean-
Mark Murcko (37:53):
Shani Elbaum-Garfinkle (37:54):
Mark Murcko (37:55):
It could be that if we’re talking about stress granules, it’s CONdensate, and if we’re talking about P bodies, it’s conDENsate or something. We can make it up exactly however we want to. So there’s a couple of questions already that have come in. Maybe, Jill, if you could unmute Carsten. Carsten has a really interesting question about the ability to reverse the behaviors you’re seeing.
Carsten Donau (38:22):
Yes. Thank you. So, really nice and clear talk. Thank you for that really. So, I would have a question. So, you said the droplet inversion happens because the arginines basically have a higher interaction strength, and they compete basically for the poly-anion. I was wondering could you reverse the process by adding excess of poly-lysine?
Shani Elbaum-Garfinkle (38:43):
Carsten Donau (38:45):
Because typically coacervates are formed because salt or counter ions are expelled and you can reverse this process by adding excess of salt, so I was wondering can you do the same with poly-lysine?
Shani Elbaum-Garfinkle (38:58):
Yeah. So, are you asking if we did the reverse? So, if we add excess poly-lysine to poly-arginine droplets-
Carsten Donau (39:03):
Shani Elbaum-Garfinkle (39:04):
Right. So, yeah. We did that. I think it’s in the supplementary. Nothing happens. So it doesn’t work when you do it the other way. You eventually will get a two-phase system, but you don’t get the inversion. So, if you had pre-formed poly-arginine droplets, and then you add in poly-lysine, you don’t see this.
Carsten Donau (39:33):
Interesting. Yeah. Thanks.
Shani Elbaum-Garfinkle (39:34):
Yeah. Good question though. Yeah.
Mark Murcko (39:36):
That is interesting and that leads to a really good follow-up that Charlotte had and I had a similar question about spacers. So, Charlotte, why don’t you ask yours?
Charlotte Fare (39:43):
Yeah. Thanks for a really great talk. I’ve enjoyed seeing all these images on Twitter, and I was wondering if you had tried anything where you leave polyK as is, and then introduce spacers into polyR so that you sort of dilute out the effect of the arginine and if that could affect the extent of immiscibility?
Shani Elbaum-Garfinkle (40:05):
Yeah. So, no. That’s a great question. So, we’ve got a bunch of COVID stalled projects doing exactly that. So, introducing spacers, but also oscillating between lysine and arginine and a single peptide to try to understand this and build the framework a little bit deeper into complexity, but we… Yeah. That’s TBD. But, yeah. It’s a great question, and this is sort of an extreme case where we have just the homopolymer and certainly having the spacers in there. I mean I guess I would speculate that having more spacers would also maybe allow for the two to be more miscible, right? Maybe the immiscibility wouldn’t be as extreme, right? And you’d be able to have more flexibility and maybe determining where that threshold is and, yeah. It’s something that we’re interested in looking at.
Mark Murcko (41:04):
That’s great. That leads to I think another question that Anthony has about what you might expect to see in more complex situations. So, Anthony, why don’t you ask your question?
Anthony Vega (41:17):
Right. So, thank you for the great talk. I’m just curious. This was a really interesting study looking at well-defined condensates, but I was just curious how… I guess, the diffusion that you measure with your microrheology measurements might change as you look at more complex condensates. Do you still get a single viscosity or do you see heterogeneity within a condensate?
Shani Elbaum-Garfinkle (41:44):
Yeah. That’s a good question. So, in our in vitro systems we haven’t gone more complex than a few components, and so far in terms of looking at the homogeneity versus heterogeneity across the droplet is I think what you’re asking. We don’t really see heterogeneity unless there’s maybe an aging scenario and we see a progression towards a different state. And in some cases there, we do see some heterogeneity and that’s work that hopefully will be out soon-ish. So, yeah. I mean I think you definitely will, should, would get, you get complexity as you increase components and as you increase complexity of conditions.
Anthony Vega (42:34):
Mark Murcko (42:35):
So, Diana, you had a couple of interesting questions too.
Diana Mitrea (42:45):
Yeah. So, I was wondering that… Very beautiful talk, Shani. That was great. Thanks. So, I was wondering about the implications in signaling of your findings. So, for instance, the nucleolus, it’s highly enriched in arginine-containing proteins, and then the nucleoplasm is highly enriched or more enriched in lysine containing proteins, and those are involved in transcription, so I’m just wondering what [inaudible 00:43:16] the nucleolus’s role as a signaling hub? When under stress, it gets dissolved and a whole bunch of those arginine containing ones are released, and what effect might have on dissolving the more labile transcriptional condensates or other-
Shani Elbaum-Garfinkle (43:38):
Yeah. That’s really interesting. So, I actually, I don’t even think I appreciated the fact that there are arginine-rich and lysine-rich things that are either separated in some situations or are forced to possibly co-mingle in other situations. That is… Yeah. I mean I think this would have implications for that. It’s hard to know exactly how without more details about the situation. But, yeah. I mean I think certainly, I mean there’s so many cases.
Shani Elbaum-Garfinkle (44:12):
I mean even in the case of histone tails which are rich in arginines and lysines, even within [inaudible 00:44:17] and so you can expect… Yeah. I mean I believe this would have implications for many of those systems. Yeah.
Diana Mitrea (44:29):
Mark Murcko (44:31):
Great. So, let’s go to Jerelle, because I think it’s still interesting, all of us I think are still trying to wrap our heads around the idea that there’s this immiscible, this property. I mean, yes, we all understand that arginine is aromatic and lysine is not. We know they’re different, but still the fact that they’re immiscible, it’s just kind of weird. And so Jerelle has a really good follow-up question about that.
Jerelle Joseph (44:55):
Yeah. Thank you very much, Mark, and thank you, Shani, for a brilliant talk. I was wondering if you have a system that is instead of made up of a homo polyR or polyK polymers, if you have a case where you have one polymer that is made up of 50/50 arginine, lysine or like the first half is arginine and the second half is lysine, would the system be still immiscible, say for example for micelle-like architectures? Because I know that… Well, not I know, but I’m thinking that with this kind of scenarios there would be a length dependence that may be factored into.
Shani Elbaum-Garfinkle (45:33):
I mean that’s literally what Rachel has been looking at now is looking at sort of mixing these as maybe, in sort of a blocky, sort of cold-block way or interspersing the arginines and lysines, and so we don’t really know yet. But we are seeing interesting things and, yes. Since they are so different, well, we do suspect that there would be some really cool things that you can do depending on the sort of separation or the clustering of the residues along a sequence.
Jerelle Joseph (46:11):
Shani Elbaum-Garfinkle (46:12):
Yeah. Hopefully, we’ll see some-
Mark Murcko (46:15):
Yeah. There’s all these wonderful material science kinds of questions that you can ask which may or may not have anything to do ultimately with CONdensates, or conDENsates, but still, nonetheless, it’s just really interesting basic science. And so, actually, Patrick has along the same lines, Patrick has a question about what happens if you start to introduce anions into the system. So, Patrick, why don’t you ask yours?
Patrick McCall (46:39):
Shani, lovely talk. Thanks so much.
Shani Elbaum-Garfinkle (46:41):
Patrick McCall (46:42):
Yeah. My question is about giving, adding additional polyanion species. So, you said that if you have a single species of polyanion that poly-arginine can basically out-compete the poly-lysine for that. My question is if you have two different species, so let’s say polyU10 and polyU50, and then also the poly-lysine and the poly-arginine. So, four different anions, but everything charge-matched. Do you expect that the polyanions would partition differently in the two phases? I’m assuming that they’ll still be de-mixing, but you’ll have three phase coexistence effectively. But my question is can the arginine basically dictate which anion it gets?
Shani Elbaum-Garfinkle (47:22):
Super interesting. So, no, we haven’t tried that specifically, but one thing related to that that we’re interested in pursuing and hoping might be true is that, so if this…. We hope that certain molecules will preferentially partition in one phase versus the other, because then that would really open up really cool possibilities in terms of controlling what is being stored, what is being released?
Shani Elbaum-Garfinkle (47:54):
And so if everything and assuming that there are differences between partitioning them, then we could leverage that. But we haven’t tried that specific experiment, so, right. If there is enough for both, but maybe the interactions with the longer polyU are strongest, and so those end up partitioning say, right, it would be a possibility into the arginine phase, and then the shorter one would end up in the other phase. That’s a really cool thing to try actually.
Patrick McCall (48:23):
Yeah. I think also somewhat related like still thinking about RNA, but just if you have the power to specify which polyanion you want, then going back towards a more biological system where you have structured RNAs, for instance, more like the Whi3 and mRNA systems, then you could maybe start to get a sense of why you might partition one versus the other depending on the charge content of the protein.
Shani Elbaum-Garfinkle (48:46):
Yeah, and we follow Amy Gladfelter’s work really closely in thinking about how RNA structure contributes to condensate properties, and we’re really interested in that as well. The other thing I’ll just add, I mean this is sort of another ongoing project. So, on the one hand, we can think of using the polymers to create different phases. We’re also trying to see if we can use lysine, arginine like as tags. So, if we fuse them to other proteins, and then maybe that’s where we get the specificity. So, maybe we don’t get the specificity from the polymer itself, but if the arginine and lysine can act as tags that create unique environments, but they’re fused to something else that imparts specificity, then that can also be very powerful and maybe more robust, and so that’s something that we’re looking at as well.
Patrick McCall (49:31):
Sounds exciting. Thanks so much, Shani.
Mark Murcko (49:34):
That’s great. So, Mike, you had an interesting question about the role of sterics.
Mike Fenn (49:38):
Yeah, and I think I may have just found the answer to my question reading your paper really quickly, Shani, but, no. Yeah. I was just asking in regards to the steric interactions of arginine–a bit more bulky than lysine. So, it seemed that lysine would maybe pack a little bit more efficiently, and therefore be more viscous. Now, then maybe there’s some role like Laplace effects and Ostwald ripening, and things moving around to make larger droplets with lysine versus arginine, but I think the answer to my question originally on the sterics, the kind of bulkier group of arginine and tell me if I’m wrong or right, is that arginine has this kind of more delocalized structure and is able to undergo more pi-pi interactions, and therefore creates a more viscous droplet. Is that correct?
Shani Elbaum-Garfinkle (50:29):
Yeah. I mean that’s what we think. So, we don’t have any modeling to really back this up, and if anyone wants to look me up after this and talk about that, we can. I think that’d be really interesting to do. But we definitely think that’s what’s happening. And especially the UTP and UDP example I think is also really telling is that like UDP, we see this like all or nothing effect where the poly-lysine just cannot, under the conditions we looked at, form anything with UDP, but the poly-arginine can.
Shani Elbaum-Garfinkle (51:02):
And so I mean that would actually be really interesting to model more explicitly to try to understand that.
Mike Fenn (51:11):
Cool. Great talk.
Shani Elbaum-Garfinkle (51:15):
Mark Murcko (51:16):
Great. So, let’s go to Pradeep.
Pradeep Natarajan (51:20):
Oh. Thanks for the great talk, Shani. I think you mentioned at some point that you’re trying to tune the properties of, to the viscosity of the poly-cation and RNA system, and you sort of said you played around with different combinations of arginine and lysine, and you found that the properties then very much, did I get that right?
Shani Elbaum-Garfinkle (51:43):
So, do you mean, so when I said that we try to see if we mix them could we get intermediate viscosities? Is that-
Pradeep Natarajan (51:49):
Shani Elbaum-Garfinkle (51:50):
Pradeep Natarajan (51:51):
So, did you play around with the composition of a single polymer where you vary the lysine and arginine ratio or did you just like mix polyR and polyK?
Shani Elbaum-Garfinkle (51:57):
Right. That’s what we’re doing now. So, yeah. So, it was probably naive to think that we can just mix these and get out this intermediate spectrum of viscosities, but perhaps by altering them just right within a single polymer that we can achieve it that way, and so, right. I think… Is that what you’re getting at?
Pradeep Natarajan (52:20):
Yeah. I guess, and like maybe a follow-up to that idea maybe potentially is that, so, most of your results show that arginine seems to be interacting a bit more strongly with the RNA for whatever reason compared to lysine, and so if you have like-
Shani Elbaum-Garfinkle (52:35):
I mean it’s known in other like unrelated to phase separation or anything, it’s known that arginine interacts more strongly with nucleic acids than lysine and there’s a host of other areas including, even just like anti-microbial peptides or things where like arginine can do things that lysine just can’t, and for reasons that aren’t well understood other than that we know that there are essential differences, but in terms of, there are a lot of functional differences that emerge from that even outside of phase separation, although if all roads lead to condensates maybe all those processes also involve phase separation, but that’s to be determined.
Shani Elbaum-Garfinkle (53:09):
All right. Back to your question.
Pradeep Natarajan (53:11):
Yeah. Sorry. Yeah. I guess I was just thinking out loud about like if you have a copolymer of arginine and lysines then you could sort of potentially play around with the relative amounts and like the patterning and such and provided that if arginine interactions are much, much stronger then maybe you have to have a much larger lysine to arginine ratio in your polymer before you start seeing if the fact of the lysine RNA interactions itself is compared to the arginine RNA interactions and-
Shani Elbaum-Garfinkle (53:39):
Yeah. I think that’s a-
Pradeep Natarajan (53:41):
… say you have some sort of like a steep decrease in the viscosity as you sort of like vary the arginine, lysine ratio in your polymer, but, yeah.
Shani Elbaum-Garfinkle (53:48):
Yeah. I think that’s a sound hypothesis. I would agree with that, that you… Right. It wouldn’t sort of be linear in terms of ratio, and you would probably need more lysines to sort of break the arginine, arginine interaction effects, but, yeah. We’ll see what happens.
Pradeep Natarajan (54:05):
Mark Murcko (54:08):
Great. So, let’s go back to Charlotte who’s continuing to think through very systematically this whole question of what we’re learning about this, the balance here between what lysines do and what arginines do. Interesting question about aging. So, Charlotte.
Charlotte Fare (54:23):
Yes. I was sort of thinking the question that brought up the nucleolus. I was wondering if you looked at aging specifically if the components of the heteropolymers age differently than they would on their own like if the lysine component of the lysine-arginine condensates sort of forms more of a gel faster than it would otherwise, and thinking sort of in a biological context if this might be a way to explore how different enzymes might modulate interfacial regions to maintain each one’s individual material components.
Shani Elbaum-Garfinkle (55:09):
Yeah. That’s such a great question. So, one thing I’ll say is that the poly-lysine droplets are just magical. So, they just do not age and they will just bounce off the surface, and then you can just resuspend them, and they’ll just form again. It’s unlike anything I’ve ever seen with the protein systems, and so, yes. The poly-lysine are a beautiful example of like non-aging. I don’t know. I think, I don’t know if we’ve seen them age at all without possible evaporation or other issues happening in the sample, whereas the arginine definitely does something.
Shani Elbaum-Garfinkle (55:51):
We haven’t characterized really the aging of the arginine specifically, but you certainly can’t do those same things with the arginine droplets. And so I think, yeah. I think you’re definitely on to something, and that’s a really interesting thing to consider is not just the difference in material properties, but, yeah, affecting the kinetics of potential aging. Yeah.
Shani Elbaum-Garfinkle (56:19):
But it’s interesting, I mean I don’t know, and I’m just asking this question to you, to the audience. I mean is it true… That’s something to test. So, is it true that viscosity scales with kinetics of aging? I don’t know if that’s true, if it’s that simple, things that are just more fluid inherently age more slowly the things that are viscous, and I think that I don’t know if that relationship is known to that extent, but it’s interesting to think about.
Mark Murcko (56:51):
Yeah. I bet it’s a lot more subtle than that.
Shani Elbaum-Garfinkle (56:58):
Mark Murcko (57:01):
I bet it’ll be so case specific. Yeah. Well, I think we’ve covered all the questions I see in the chat, and we’re just about at the top of the hour anyway. So, we’ll have to continue some other time the debate over CONdensate versus conDENsate although it appears that there is a bit of a preference for the first syllable being the stressed syllable, but a longer-term debate I think.
Mark Murcko (57:27):
So, I would just like to, again, thank you, Shani, wonderful talk. Great questions. Great participation from everybody. So, thank you all very much. Wonderful to have your lecture.
Shani Elbaum-Garfinkle (57:41):
Great. Thank you.
Mark Murcko (57:43):
All right. Thank you, Shani, and thanks everyone.
Jill Bouchard (57:45):
Mark Murcko (57:46):
Jill Bouchard (57:46):
Mark Murcko (57:46):
Shani Elbaum-Garfinkle (57:49):