 All right, everyone. Well, I'll keep today's introduction short, but I'm pleased once again to introduce Professor Mike Rosen on the second of the Steenbach Lectures today. And yesterday, we heard a really exciting story about the fundamentals of biological condensates and the types of changes to chemistry and reactivity that can occur inside these structures. And today, Professor Rosen is going to be telling us about some of the biological implications of these structures in the context of chromatin itself. So without further ado, I'll just pass it on. Thanks very much. It really has been just a spectacular couple of days for me here, just hearing so much wonderful, wonderful science from the faculty, frankly, especially from the students that I had lunch with today. Everybody had great stories, just been really, really stimulating for me, and they have some collaborations to go home with, so just great all around. And I'll reiterate what I said yesterday, what an honor it is for me to be giving the Steenbach Lectures, both because Steenbach is really one of the pioneers in nutrition science, which then led to biochemistry, which is my intellectual home, and also because of his strong commitment to the benefit of passing basic science onto the public good through commercialization, which again is something that I also feel quite strongly about. So thank you for the opportunity to be here. So the work I'm going to tell you about today is a very, very new area in my lab. We've always focused throughout my entire career on protein molecules. Nucleic acids kind of scared me. We've always focused on biology that occurs in the cytoplasm and really never ventured into the nucleus. But as I'll explain to you, our understanding of the kind of molecular species that undergo liquid-liquid phase separation recently has taken us into a new compartment, the nucleus, and a new kind of biomolecule chromatin. And so what I'm going to do today is just spend just a few minutes reiterating some of the points that I made at the very beginning yesterday and then tell you about our new work in this area of chromatin organization. So again, research in my lab has focused for about the last dozen years or so on these cellular compartments that we've taken to calling biomolecular condensates. Condensates, by definition, are compartments in cells that concentrate selected groups of molecules, both protein molecules, RNA molecules, probably small molecules. And the key thing is that they concentrate them in discrete both sides in the absence of a surrounding membrane. Condensates occur both on membranes. I said a lot yesterday about actin assembly being controlled by membrane-bound condensates. But then many, many of condensates are found either in the cytoplasm or in the nucleoplasm. Key feature of condensates is that the components are quite dynamic. They move around within the compartment. They exchange between the compartment and surroundings quite rapidly. Yet even in the face of that dynamics, condensates remain distinct from one another with distinct collections of components. Condensates are involved in diverse range of biological processes. I feel like every time we open up a journal lately, there's a new condensate that's involved in some new biological process. And similarly, there involves an increasing number of diseases spanning from neurodegeneration out to skin disorders and various cancers. I'm sure there will be more to come. And a key point from my lecture yesterday is that many condensates, but again, certainly not all, but many of them appear to form through this physical process of liquid-liquid phase separation driven by multivalent macromolecules. Now, over the last decade, there have been a very wide range of molecular types that have been shown to interact through weak multivalent interactions and consequently phase separate. Everything from multi-domain proteins that we've worked on a lot in my lab. Other people, Minji Zhang in Hong Kong has done a lot of work on the postsynaptic density. Its components are multivalent and phase separate. But a lot of work on intrinsically disordered regions of macromolecules that can also interact with each other through multiple adhesive elements connected by flexible tethers. RNA and DNA through repetitive base pairing interactions can also assemble and phase separate. And an example that I always like to give and becomes directly relevant today is all of the ideas that underlie this basically come from polymer chemistry. And a key point of polymer chemistry is the notion of multivalency driving assembly and phase separation is independent of the scale of the system you're talking about. And so a nice example of that is synaptic vesicles, where the vesicle is much, much, much larger in order of magnitude larger than any of these molecules. But nevertheless, if one generates vesicles coated with adhesive molecules, they will assemble sharply to form a phase separated liquid whose unit component is something even as large as a vesicle. And again, that will become relevant in just a moment in the context of chromatin. And importantly, again, this comes straight out of polymer chemistry. This notion of multivalency producing these structures leads very naturally to modes of regulation because polymer chemistry tells us that higher valence, the larger the number of adhesive elements in a given molecule, as well as higher affinity between the adhesive elements promote oligomerization and thus promote phase separation. And I spent a lot of time yesterday talking about a system here on the left that regulates actin assembly, where changes in phosphatiricine is effectively changes in valence and drives phase separation. And there are a lot of other examples in the literature as well, or disordered proteins being regulated this way. So with that as very, very brief prelude, this has taken us into a completely new area for my lab. And I'll introduce the problem, and then I'll tell you how we got into it through this lens. And the problem is the organization of chromatin in the eukaryotic nucleus. And the basic problem coming from the world of structural biology to me is just cool. And that is that every cell in the human body has about two meters of DNA in it. And that has to be compacted into the nucleus, which is on the order of 10 microns in diameter, which is a tremendous degree of compaction. And one thing I've done to try to appreciate this is to move it to kind of a human scale. And that is if you took all of the DNA in a human body, my body, and you lined it up end to end, how big would it be? And the answer is that would go from Earth to Pluto and back three times. But yet it compacted into me and all of you. So it's just this incredible amount of compaction of this polymer chain of double-stranded DNA. But this must occur with a lot of restrictions. So it has to retain dynamics in order to permit regulation and function. Moreover, it has to occur non-isotropically. There have to be distinct functional domains that are on scales of, let's say, kilobases to megabases. And of course, this organization is known to control a huge range of nuclear processes, everything from transcription to replication to repair. It goes awry in a very large number of diseases, most notably many cancers. So it's a very important problem for basic biology and human disease. Now, if you look in canonical explanations of this, that appeared in textbooks, the organization of chromatin, the compaction, has been explained largely through a hierarchical assembly process. And the basic idea is you start with this long chain of double-stranded DNA. The first level of compaction is that the DNA gets wrapped twice around an octamer of histone proteins, forming nucleosome structures. And those nucleosomes have then been proposed to assemble into what's been called the 30 nanometer fiber. It's a two-start helix of self-satisfied nucleosomes. Those fibers then assemble into thicker fibers, which assemble into still thicker fibers, and eventually you get something like the mitotic chromosome, which is the most compact unit of chromatin. And there are some older data that support this, but as imaging has gotten better and better and better, and people have been able to look inside intact nuclei of eukaryotic cells, clearly one sees tons and tons of nucleosomes, but these higher order structures, if they're there, are not highly enriched. And so the question of how you go, how the cell goes from nucleosomes to mitotic chromosomes still is, I would argue, quite poorly understood. Now, there have been a variety of recent studies that have invoked liquid-liquid-phase separation as a mode of organizing chromatin. The earliest ones came from Gary Carpin and Gita Narlikar, where they showed that the critical heterochromatin protein, HP1-alpha, was able to undergo liquid-liquid-phase separation and recruit in DNA. And they proposed on that basis in a variety of other data that this phase separation of HP1-alpha is an important organizer of chromatin. There's been a lot of work done on various models of assembling transcriptional machineries. There are some who have argued for overt liquid-liquid-phase separation, particularly Phil Sharp and Rick Young. Robert Tegin and Xavier Darzak have a somewhat different model but that nevertheless invokes assembly of molecules at sites of active transcription. All of these models, though, essentially invoke proteins as the drivers. It's the assembly of the protein that creates the thing and then DNA kind of follows. There have also been a variety of computational models of chromatin organization that invoke differential interactions between different types of chromatin. They'll invoke U-chromatin interacting better with U-chromatin than between U and heterochromatin and heterochromatin interacting better with itself than with other types. And in that way, they've been able to generate simulations of overall organization of chromatin in the eukaryotic nucleus. And in some cases, they've been able to show that those match quite well actual microscopy data of different types of chromatin. An issue with these models is that they generally lack any kind of molecular mechanism. They say, heterochromatin likes heterochromatin, U-chromatin likes U-chromatin, but they don't involve any specific kinds of context that would produce those interactions. Now, we got interested in this area. I got interested in this area through this lens of multivalency-driven phase separation in the following way. The chromatin can be thought of as essentially a large, long multivalent polymer with the nucleosome as the basic unit. Now, the DNA wrapping around the histone core is obviously very positively charged. The histones all have these long tails that emerge from the faces of the nucleosome that are all quite basic. And so one can think of, perhaps, acidic basic interactions driving assembly of these systems. Moreover, of course, we know there are many different kinds of post-translational modification epigenetic marks that are put onto those histone tails, and many of the proteins that read out those marks have multiple reader domains. And so another way of multivalency-driven interactions that could produce assembly of chromatin. And so even though the scale is very different from modular domain proteins coming back to the theme I established a couple of slides ago, nucleosomes are gigantic compared to the protein domain. And moreover, even though the DNA linkers are much more rigid than protein linkers, and I'll come back to that in just a few minutes, even though there are those difference, fundamentally, the essential multivalency is the same. And so we thought long ago that perhaps this would be a system that could undergo face separation. And in fact, there's been a lot of work from a variety of labs that have shown that if one starts with different kinds of arrays of nucleosomes and adds, in this case, super physiologic magnesium, one can see precipitation of the chromatin. And there's been a little work, as far as we can tell, doing sort of light microscopy imaging of those precipitates and what one sees. For example, this paper from the Meshima Lab, sort of these aggregated structures, which might be consistent with some kind of certainly assembly process. And so with that as sort of framing of our thinking, a new postdoc in my lab, a terrific guy named Brian Gibson, who came to my lab from Lee Krause's lab and thus a background in genome organization. Brian was able to put together a dodechomeric or series of dodechomeric arrays of nucleosomes based on the positioning sequence, the Wittem 601 sequence and label H2B with a fluorophore. And sure enough, one takes solutions of those molecules and puts them in either physiologic monovalent salt or in this case, a little bit super physiologic divalent salt. One seems formation of these, I think quite beautiful spherical objects that have both histones in them, as well as we can use a stain for double-stranded DNA, also have DNA in them. And I should note that Gita Narlakar, UCSF independently made similar observations since our initial work here, a variety of groups have made also similar observations. Now the behavior of this system followed many features that we would expect for multivalency-driven phase separation. So for example, if we look at the presence of phase separation, that's what the black dots are here or the absence, which are the unmarked intersections, one can see that over relatively small changes in salt, either monovalent or divalent, we go from no droplets to droplets or no structures to structures at this point. We can see the same thing as we add, increase the concentration of the nucleosomes themselves, relatively sharp transitions to forming these things. Moreover, we see a quite strong dependence on the valence of the array. So a tetrameric nucleosome array doesn't ever, in the conditions we've been able to find, form large macroscopic droplets. A hexanucleosome array though does, but you gotta go to a bit higher salt concentration and the dodecameric array, phase separates a whole lot better than those others. So again, some exactly what we would expect from multivalency-driven phase separation. Interesting thing that we were pleased about is that when we look at the concentration within the droplets here, A, it's the degree of concentration is enormous. It's about 10,000-fold in the droplets versus the surroundings. Just thought that was cool. It was the record we had seen in our lab. More substantively though, the concentration of nucleosomes within the droplets is at least within the range of nucleosome concentrations that have been measured in eukaryotic nuclei, which of course doesn't mean this is what's going on in nuclei, but at least we're not towards a magnitude off. So this kind of compaction or this kind of a process could at least contribute to what's going on in cells. These are very clearly liquid-like structures. If we photobleach the center of a large one, we can watch the recovery on, quantified on time scales here of about 10 minutes. That's slower than other phase-separated systems, which suggests that these are more viscous droplets than observed elsewhere, but clearly this is liquid. There's been a bit of controversy in the literature there with other groups, one other group in particular seeing different things. We're quite confident that if you handle the droplets correctly and you don't photocross-link them, that this is really quite general. We've been able to do this in a wide range of buffers and we basically always see this same kind of behavior. So we think this really is intrinsic to these systems. We can also assess the dynamic properties, the liquid-like properties here by making two solutions of droplets, one labeled red, one labeled green, and if we mix them together, you can see what happens. We can see very rapid fusion of the different colors. And this is about a 25-minute movie and what I hope you can see is that the fusion itself occurs very rapidly. As soon as droplets hit each other, they round up and become spherical. What's interesting though is that the pole of the incoming color remains for quite some time. You can see the green pole remains for quite a bit of time here. I think I may have lost my laser pointer. Okay, I will do my best just to point things out. So we can see this a little bit better if we look at snapshots from individual points in the movie. So if you look at this one over here, what you can see is at 30 seconds, the two droplets have encountered each other. By the time a minute's gone by, boom, they've rounded up. So there's very strong surface tension in these systems, but the polar cap of green exists all the way out to, let's say, 15, 20 minutes or so. So consistent with the idea that the viscosity within these structures is very high, but nevertheless the surface tension is very high that can overcome that and cause rapid rounding. So high surface tension, also high viscosity. We've looked a little bit at the chemical determinants of this phase separation process, and what we see is if we chop off the tails, the image on the left here with proteases, that essentially the nucleosomes we know remain intact in this experiment, I'm not showing it, they're intact, but we completely lose the ability for these things to assemble and phase separate. Moreover, there's what's been called the basic patch on the histone H4 tail that plays an important role in organizing chromatin in cells. If we mutate that patch on H4 and our octomers, again, we lose the capacity to phase separate. And that's consistent with previous data on the precipitation of chromatin arrays by magnesium. This is not an identical dependency, though, in those other experiments, because when people have looked at precipitation, there's also an acidic patch on the face of the nucleosome that plays an important role there, and at least in our hands, if we mutated here, the droplets can still form. We haven't quantified whether the phase separation boundary has shifted, so there may be quantitative difference, but not qualitative differences. So just to summarize this part of my talk, what I've shown you so far is that polynucleosomal rays will undergo liquid-liquid phase separation when moved into physiologic salt concentrations on the order of 100-millimolar monovalent salts and zero to a couple-millimolar of divalent salts, calcium or magnesium. The phase separation depends on the length of the nucleosome arrays, and the phase separation condenses the chromatin to concentrations within a physiologic range. Should be the droplets are dynamic, but nevertheless very highly viscous. And finally, the interactions of the histone tails are clearly important for this process. We don't know how at this point we're hoping. I'll show you a little bit of cryoEM data. We're hoping to begin to understand it through direct structural studies. Now, one of the questions that we asked as we were going through this is is there any place in cell biology where these kinds of behaviors of chromatin have been observed experimentally? And the answer is that we're aware of at least three situations where it does seem to occur. So one of them is during apoptosis because there's a large body of data showing that one of the early things that happens in apoptosis is DNA condenses in the nucleus. And that's shown in the images on the upper left there. In addition, during carotinocyte development or more generally during processes of enucleation. Thank you. During processes, whoops. I'll use it in a pointer. Let me see the little arrow there. In processes of enucleation, so where the nucleus is kicked out of the eukaryotic cell. It's one of the things that happens in the later stages of carotinocyte development. Also during red blood cell development, for example, that also occurs. And in all of those processes, what's been reported is that the DNA condenses in the nucleus. And here's an image for a paper from Elaine Fuchs's lab. She actually in this paper was focused on the phase separation of the molecule here, shown in green. But kind of a side note is you can see clearly that the chromatin has condensed here. The other place that one has observed has been observed for many, many years, decades now, is in the macronucleus of tetrahymena where the heterochromatin is organized into these roughly spherical structures that have been called chromatin bodies. And the surrounding lighter regions in this EM micrograph is the eukromatin in these organisms. Now, the question is, is there anything that unites these three biological processes? And one thing that jumped out to us is they all occur when DNA is fragmented, is in small pieces. So in fact, during apoptosis, that the event that is believed to cause this condensation is nucleus activity that chops the chromatin up into small pieces. And in fact, this same paper studying apoptosis showed if you isolate nuclei of cells and you treat them with a variety of nucleases, you can see the chromatin condensin. It doesn't have to be during apoptosis. DNA is fragmented during enucleation. And the reason that telomerase was discovered or was looked for in tetrahymena is because tetrahymena naturally stores its genome in relatively small pieces of DNA. So there's lots of n's, lots of telomeres. And so we think that what this is telling us is that in all of these cases, the long range constraints on the chromatin polymer, crosslinking with molecules like cohesins and condensins, other kinds of interactions that produce long range interactions, those get lost when the chromatin is chopped up. And so without those constraints, we think that this liquid like behavior really is intrinsic to the chromatin polymer. And so when you remove them experimentally, what you see is this inherent condensation. And a corollary of that is we think that the dynamics that we are observing in the droplets and the chemical interactions that we will ultimately discover that, figure out that cause these, reflect short range dynamics in chromatin because there's a variety of experimental data, everything from FRAP studies to single molecule imaging, single nucleosome imaging that show that DNA, that chromatin is quite dynamic on short length scales, less than let's say 100 nanometers. But then if you look on longer length scales is absolutely not dynamic. And so again, what we think is that and what we're seeing in the movement in our droplets likely reflects those short range interactions, those short range motions of bonafide chromatin because we have again relieved those long range constraints that produce the lack of motion on longer length scales. More than happy to talk about those ideas afterwards. So along these lines of where in biology might all of this matter, we've had a really wonderful collaboration with Daniel Gerlich at the INP in Vienna and in particular with his really terrific graduate student, Max Schneider. Because Daniel has been interested in, for a number of years, in an interesting question is how is it that chromosomes are pushed by microtubules away from the spindle poles during mitosis? So if you're like me and you don't know a heck of a lot about chromatin, one thinks of what happens during mitosis is the chromatin has to be pulled into the two daughter cells. Well that's what happens late in mitosis. Early in mitosis though, the chromatin's got to be aligned at the metaphase plate so it's got to be pushed from around the nucleus to line up in the center of the cell. And so that's the question that Daniel was trying to answer. And a specific way of phrasing this is and the pushing is known to occur by microtubules. So you go from this chromatin throughout the nucleus on the left in interface during mitosis to being pushed again into the metaphase plate, what happens by microtubules? So the question is how is it that chromosomes can be moved by the microtubules? Why don't the microtubules just penetrate throughout the chromatin matrix which would then render it unable to be moved around? And so to get at this, Max did a really interesting experiment. And I think the truth of the matter is Max did this experiment first and then tried to figure out what it all meant in terms of microtubules moving around. But Max took a mitotic cell and micro-injected into it a nuclease. And here's what he found. So you do the micro-injection and within minutes the chromatin basically rounds up and forms these spherical, oh shoot, I hate when it does that. Forms these beautiful spherical structures. You can watch them, I'll play it again, you can watch them move around. You can watch them clearly fuse with each other. You can watch them move out and spread out and separate as little liquid droplets. Really looks for all intents and purposes like formation of a chromatin liquid in the cells. And moreover, if you photobleach one of these droplets, you can see the movie here and the data below it, you bleach and the recovery occurs very fast and you can see quantification here on roughly 10 seconds basically you fill in the bleach spot. So I think it's very clear that once you chop up the chromatin, at least in mitotic cells, you create a liquid like structure. Now the density of the chromatin in intact, or the density of intact chromatin and the density of these droplets generated by nucleases turns out to be identical with an experimental error and this is quantification from a number of cells. And so what that says is that the physical properties, the interactions that contribute to this phase separation are very likely the same processes that contribute to compaction of a real bonafide native chromatin. I should note for those in the audience if you think about condensin depletion which does play important roles here has no effect on density. And so this doesn't seem to be something that has to do with condense and mediated long range crosslinks. Again, we think that this is largely due to these short range interactions intrinsic to the chromatin polymer. Now, not only does the fragmented chromatin have the same density as native chromatin, it can also be pushed around by microtubules in analogous ways. So again, here on the left is what native chromatin looks like as the cell goes into mitosis. It gets pushed around, pushed around and ends up aligned here on the metaphase plate. In the cell on the right, the microtubule network has been disassembled with nicotazole digested with nuclease and then the nicotazole was washed out. And what you're looking at is microtubules in green and you can see that as the microtubule networks reform, those, that liquid like chromatin gets pushed around out to the cell periphery in ways that seem analogous to the way, again, native chromatin gets pushed around when it's intact. And so our conclusion from this is that the material properties of chromatin seem in this particular context, in the pushing context, to really be governed by this liquid liquid phase separation process. So the question is sort of how? And on this we're a little bit less clear, but it seems to have to do with the fact that for chemical reasons, microtubules and tubulin are both excluded from both chromatin in cells and also from phase separated chromatin droplets. And so here are cryo electron tomograms, slices from a tomogram of a real cell. The chromatin is here outlined in darker colors and I hope people can see that there are a variety of microtubules here. And what one sees if one looks at a blown up region here is that the microtubules, most of the time here, extend to the surface of the chromatin but don't penetrate into it. And in this larger view that more closely matches the image here, again you can see when they've outlined a lot of microtubules, again they really don't penetrate very much at all into the mass of the chromatin. In vitro, one can do experiments where you can label tubulin with a color and what you see is that tubulin gets excluded from the droplets formed by nucleosome arrays and if one does this in a way that allows the microtubules to form, they basically form a matter that is excluded from the droplets. Now for technical reasons we weren't able to do an experiment where we could create an array of microtubules that would push on the droplets, we were hoping to be able to do that and just again for technical reasons it never worked out but at least looking at these non-dynamic systems, it seems pretty clear that both tubulin and microtubules don't get into our droplets paralleling what one sees in vivo in cells. And I should note that in both cases, both in vivo cells and in vitro, this exclusion appears to be based on the high negative charge on tubulin, which kind of makes intuitive sense, you've got very negatively charged DNA and negative charged molecules don't like to go into those droplets. And so essentially what we think is that the material properties of the phase separated chromatin enable it to be pushed by microtubules. So this has then led to a model where chromatin is condensed both by intrinsic liquid-liquid phase separation interactions between nucleosomes on short length scales and again on long length scales, condensin will hold longer loops together but that the resistance to microtubule forces depends mostly on again the phase separation properties, the short range interactions between nucleosomes within chromatin. So in my last couple of vignettes here, I wanna tell you a bit more about the work that we've done in vitro on these systems and in particular to look at how different regulatory factors that have been known to modulate chromatin in cells interact with our nucleosome arrays. And so there are two things I'll tell you about this afternoon. One of them is the spacing between nucleosomes and the other one is the actions of histone tail acetylation and histone tail readers. So there's an interesting observation that's been in the literature since the late 1970s. And that is if one looks on a genome-wide scale at the positions of nucleosomes and the length of the linkers between nucleosomes, one sees in essentially all eukaryotes that this has been looked at so far, everything from yeast out to mice and to humans as well. One sees this interesting oscillatory pattern if you look at the frequency of a given linker length. Both in yeast you see it, you also see it in mice. And if you look at the periodicity here, basically what you see is that for 10 N spacing where N is an integer. So 10 nucleotides or 10 base pairs, 20 base pairs, 30 base pairs, those are all depleted on a genome-wide scale. 10 N plus 5, so 15, 25, 35 base pairs, those are all the ones that are enriched on a genome-wide scale. And again, this is seen across the eukaryotic spectrum and the reasons for it really remain entirely unknown. Now, if you think about this, DNA has 10.2 base pairs per turn, which means that 10 N essentially align successive nucleosomes facing the same way, right? So 20 base pairs, 30 base pairs, nucleosomes at either end of this will aim the same way. 10 N plus 5 aims them exactly opposite. So 15, 25, 35 aims them in opposite directions. Now, the structural biology that's done, crystallography and cryoEM has been done on the 10 N spacing arrays. And what one sees there are these beautiful 30 nanometer fibers where every other nucleosome stacks on itself. So one, two, three, four, five. So even one stack on themselves, odd ones stack on themselves. So the 30 nanometer fiber is made with 10 N spacing, but that's the one that's disfavored in cells. So the question is, what is different about 10 N plus 5 spacing? And you might have guessed what the answer is. Turns out that 10 N plus 5 linkers produce beautiful phase-separated droplets. And we've looked at 15, 25, 35 actually out to 45, they all phase-separate. 10 N spacing, phase-separates much more poorly. So 20 base pairs will a little bit, 30 we don't see anything, 40 we don't see anything. This is with monovalent salts, if we add a little bit of divalent salts, one millimolar divalent salts, we can then begin to see phase separation of 30 and 40, but it's still much, much less than we see with the 10 N plus 5 series. And so what this suggests to us, this is the spacing that is enriched in cells. Perhaps again, this supports the idea that phase separation, interactions between chains of chromatin may be used to organize cells and in vitro with our short fragments, it manifests as phase separation. And of course this says that chromatin remodelers that push around nucleosomes and dictate their relative spacing may be something that can be used to control not just nucleosomes along one strand, but the association between different strands of chromatin, giving higher order organization. Now, one thing, again, I'm a structural biologist at heart, that's where my training is. That's one of the things that I love to do. And so we become very interested in understanding the internal organization of phase separated droplets at large. Chromatin droplets are particularly amenable to this because in cryo-electron tomograms, cryo-ET images, an individual nucleosome is big enough to be able to be seen. Whereas most of the other kind of droplets that I'm aware of, the individual particles are too small and so you just get kind of a dense gray image. Here you can see, and this is one slice of a tomogram that we've taken of a droplet that's sort of irregular in shape because in part it's stuck to something on the matrix, and in part because we use standard blotting techniques which distort these. But nevertheless, you can see that within this structure, there are a whole lot of individual nucleosomes. So what I'm gonna do is run through individual stacks of this in a movie. And you'll see, again, we can see lots and lots of different nucleosomes. You can see that many of them, if you try to look carefully, they're in kind of random orientations, but there's also some regular structures within them. I'll highlight some of those at the end. We get these very interesting filament-like structures that are about 10 nanomies in diameter, which is just about the diameter of an individual nucleosome, and they look kind of like stacks of coins. And the truth is we're not sure what to make of these. We're still doing a lot of work to analyze these images and figure out what those structures are. What's interesting is that Clota O'Shea, who's a wonderful chemical biologist at the Salk Institute, has developed ways, she calls it chromium T, of looking at nucleosomes in intact nuclei. And while she doesn't see a lot of 30 nanometer fibers, she sees these weird things that look like stacks of coins that are about 10 nanometers thick. And so we think that perhaps this kind of structure does in fact represent something that's going on in eukaryotic cells. So stay tuned. We've got a lot of work going on on all the different kinds of droplets we're making, and hopefully we'll understand them over the next few years with cryo-ET. So the last thing I wanna tell you about this afternoon is our work to understand post-translational modifications and particularly acetylation of chromatin and its role in this whole process. So it's widely known that acetylation of histone tails opens chromatin in vivo. It decreases the density of chromatin in cells and also enables processes like transcription and DNA repair. And so my post-doc, Brian, rigged up, I would say, largely artificial system where we can recruit the catalytic domain of the P300 histone acetyl transverse into our chromatin droplets through binding of a fusion protein to the TETR transcription factor to a TETO sequence in the DNA. And so we can, if we label this histone acetyl transverse with GFP, we can see that initially it gets into the droplets quite well. So here is an image of droplets looking at the histone. Here is those same droplets now looking at GFP. And you can see we get very high concentrations of this enzyme into our droplets. And if we add acetyl CoA, which is a cofactor for acetylation, you can see as the acetylation proceeds, the droplets become a little dimmer, dimmer, dimmer. And eventually they crumple up and go away. And so what this suggests to us is that the chromatin could be controlled, phase-separated chromatin could be controlled through acetylation of the histone tails, which evidently weaken nucleosome-nucleosome interactions. We don't know how yet. Then obviously they quench the charge. And that ultimately causes lower density and ultimately leads to complete dissolution of the system here. So the last thing we wanted to look at was what about these histone tail readers? So there is a widely used domain called a bromodomain that binds selectively to acetylated histone tails on chromatin. And we created an entirely engineered protein, a pentobromo protein. We took a single bromodomain and strung it together in a ray. We also looked at a natural chromatin regulatory protein called BRD4, which contains two bromodomains as well as an IDR region, and it's determinants that can self-assemble. And in both cases we did the following experiment. We started off with chromatin droplets. We acetylated them to a high degree with the P300 acetyltransferase, and again the droplets went away. And in both cases then, if we added the bromodomain protein to it, we reintroduced a new phase that contains both the bromoprotein as well as chromatin. We can see that for the pentobromo as well as for BRD4. Now in both cases, we know that the interaction or formation of this new phase requires multivalency. A single bromodomain protein absolutely does not at any concentration cause reformation of the separated phase. And if we block binding to the acetylated lysine tail, either through a mutation or through a small molecule called JQ1, again in both cases we block formation of this acetyl bromomains. Now the last piece of data I wanna show you is what we call our chromatin caterpillar. We did an experiment that I really honestly did not think was going to yield something that's interesting, but we asked what if we start with both acetylated chromatin and unacetylated chromatin? So this is a single tube that has both in it. The unacetylated chromatin assembles into droplets. The acetylated chromatin does not, it's uniformly distributed. And if we add the pentobromodomain protein into it, we can see that again, we reintroduced that acetylated bromophase, but what's cool is it does not coalesce with the unacetylated chromatin. So if we show both colors here together, we can see that we get these two distinct phases. They adhere to each other, so there's enough attraction to cause adhesion, but not so much adhesion that the molecules coalesce and form a single dense phase. And so what that tells us, at least in physical terms, is that phase separation can produce distinct chromatin domains, but obviously have distinct physical properties and by implication could have distinct functions. So again, I wanna be very clear what we know and what we don't know. What we know is that in vitro, there are distinct physical properties of these kinds of chromatin. We don't know of course that this process contributes in a meaningful way to chromatin inside cells, but it certainly says that the physical properties alone are sufficient to give distinct domains. And so it gives us things that we are currently pursuing. My postdoc, Brian, is currently pursuing in my lab and also he's moving to St. Jude to start his own lab this summer to try to understand whether this physical incompatibility between these different kind of phases really could contribute to the distinctions between say U-chromatin and heterochromatin in cells. So I'll just finish up by summarizing what I've told you is that liquid-liquid phase separation is an intrinsic property of chromatin fragments of polynucleosome arrays. And this can give mechanisms of very, very strong compaction but allow the chromatin to remain dynamic in that state. There are a number of different regulatory factors that will modulate this process. I told you a little bit about our work to understand different rinker lengths and the sedlation of bromine domains. We've also looked at histone H1, again a very important chromatin regulator. I'd be happy to answer questions about that. And so the chromatin regulators can act on this liquid phase and one of the things they can do is produce distinct domains. And so what we think is that this intrinsic property of chromatin could then sort of a basis set of interactions that biology will then act on through all of these different regulatory factors to give the various chromatin structures and chromatin functions in cells. So I'll thank the people involved in this work. All the previous work in the lab again was done by four graduate students who've all left the lab at this point. The story today on chromatin was really driven by Brian Gibson. It's just absolutely fabulous, insightful postdoc in my lab. Brian is here in the back right of this image. Also working with our lab manager, Linda Doolittle, who really was essential. There are so many reagents to put together for all of this and Linda was just spectacular in providing all of the resources needed to do these experiments. And again, I mentioned our ongoing collaboration with Daniel Gerlich and Max Schneider. Also, Cy Redding was involved in some of the early stages of this whole story. And so I think a lot of the biological implications that we have at this point really owe significantly to Daniel and Max. It's been great to work with them. So again, I'll just conclude by reiterating it's been a great couple of days for me here. I've really enjoyed meeting so many people. Interactions have been wonderful. Dinner last night was great. So thank you for the opportunity. And again, thanks for listening to me for another hour. That was a wonderful talk. The experiments where you injected the nuclease were really cool. Yeah. And it made me wonder, first, do you know how many sites for ALU-1 there are in the genome and how many pieces you're cutting it into? Because I'm just wondering if you can inject different nuclases and say, oh, you get this behavior if there are 10,000 sites, but not 1,000 sites or something. So interesting, I'm pretty sure ALU-1 is a forebase cutter. Oh, so it's a lot. So it's a lot. The problem is that you're doing this individual cells. So it's not only you can run a gel and see what the distribution of fragments is. We had talked about, but I honestly don't remember the reasoning, but Max didn't go look at six nucleotide cutters or rare cutters to see if there were differences. I think the argument was just that it's so hard to know what you've got anyway. It might be hard to interpret that. What happens to the cells after this happens? You showed them they begin to push the DNA to the edges, but then what next? Do they just undergo apoptosis? Or do they stick around for a while? I don't know the mechanism by which they croak, but yeah, they basically die. See, they don't live a whole lot longer after you've done that to them. Yeah. I should say one interesting point along the lines of what you're asking about, the size dependent. One of the things that Brian has done recently that Kenny's taking away into his own lab is he's been able to isolate nuclei from, we buy gazillion gallons of helos cells, isolate nuclei from them, chop them up with nucleases, and then isolate. Forgot a beautiful way of getting most of the chromatin out of the cells. We can isolate. He can isolate in quite pure form in a biochemical prep. And what we've seen that, A, you can add salt to that, and it phase separates. But that fragments on order of, I think it's the 4,000 to 5,000 cut, phase separate at lower salt than if you take things that are in the 2,000, 3,000 cut. So even with native chromatin, the notion of longer arrays be getting phase separation is better. It's not quite the experiment you would like to see, but it's along the lines of size dependence, even with natural chromatin. Perhaps you mentioned it, but if you put into an interphase cell, do you also get, I mean obviously there's difference in just condensed states of chromosomes, but what happens for an interphase cell? Do you get the blobs? Good question. So you get small little bits, small condensed droplets. And we don't know what those are, but not nearly to the same extent. And we think probably it's because one of the things that happens as cells go into mitosis, you know better than I, is they become deacetylated to a very large degree, and there's a lot more acetylation going on in an interface. They didn't do an experiment where they did that injection also with an acetyltransferase inhibitor. So kind of assume that the reason we don't see the same kind of massive condensation is because there's a lot more acetylation. Sure. And you also have a hyperphosphorylation on S10. I was wondering if Brian or someone has made the recombinant proteins and done that in vitro. Does S10 promote condensation? We haven't tried that yet. And then the HAD experiment. So one of the preferred substrates for P300 is K16 on H4. So you could just be like, well, the reason why the HAD is acting like the basic patch is because it's hitting the same. But are there other license? There's so many license that P300 hits. Do you know which ones are actually important for this? Any students who want to come to postdoc and do that experiment, I'll make it a little bit further than that. So the answer is no, we don't know which sites get hit. We don't know if it's really the collection of sites or if it is K16 that's critical. Moreover, in the reforming the structures, we don't know whether there are certain sites that the bromodomain binds to that are better to reform. It doesn't have to be that the sites that are used to blow up the chromatin are the same sites that are used to reform it. Those could be orthogonal. So again, there's like a PhD thesis or a postdoc to do with that. And we just haven't done it yet. The pandemic has slowed down. Very interesting results. So what does addition of H1 do? Well, I'll just tell you. I had a couple of slides that I took out for time. So H1 does two things. First of all, it compacts the droplets. So the fluorescence intensity goes up by about 25%, consistent with what H1 does in cells. It also basically solidifies the droplets. So if we add H1 to existing droplets or we form droplets in the presence of H1, we see same kind of round structures form, but they're absolutely solid. So they won't fuse, in other words? We'll see them contact each other and they stick to each other, but they don't do that. It's consistent with what's been observed for H1 in cells controlling the dynamics. So this may be more relevant to yesterday's talk, but what is it that limits the size of the droplets? Because you show they fuse, but do they dissociate as well? Yeah, good question. So in vitro, really nothing limits the size of the droplets. They basically, as long as eventually, after a day or two, for reasons we don't understand, most droplet systems stop fusing. But in general, they will get bigger and bigger and bigger because the thermodynamic state is to minimize overall surface tension. And that would be one enormous droplet. In cells, the sizes do seem to be limited. And there's some really wonderful, actually, somebody apparently you are all trying to recruit here to Wisconsin, a woman in Geraldine Saidou's lab showed that there's a molecule that interacts with the surface of one kind of droplet here that basically forms what's called a pickering emulsion, which sort of stabilizes a particular size. And so my guess is that there are a lot of different condensates in cells that have that analogous kind of emulsifier. But in vitro, if you just have the few purified proteins you were talking about yesterday, they just get bigger and bigger. They just get bigger and bigger and bigger. Oh, yeah. A new way of giving seminars. DeBion asks, thanks for the great talk. Is charge the only criterion for the LLPS? Is LLPS only restricted charged biological polymers such as DNA, RNA? So absolutely not. Molecules can come together through many, many different types of interactions. Obviously, some of it's charge-based. Some of it's hydrophobicity-based. There are specific interactions that tend to be higher affinity, long-lived. There are interactions with lower specificity that tend to have lower affinities and be more short-lived. All of those kinds of interactions can come together. I think that the key feature is simply multivalency. There have to be multiple adhesive elements to allow formation of larger structures, which ultimately drives the insolubility. All kinds of different interactions play in here. There's an experiment that people do in the literature that I really hate, which is you add 1-6 hexane dial to droplets and they go away. All you're doing is disrupting hydrophobic interactions across the whole cell. So it has nothing to do with phase separation per se. It relates to this many different ways to put these together. So relative to that, what is the temperature sensitivity of phase separation? So most, but not all, biological systems show what's called an upper critical temperature. And that is that as you increase the temperature, the concentration of the bulk phase goes up and the concentration of the droplets comes down. And then eventually you go past a critical point, a critical temperature. And above that, it becomes a single phase again. There are some that go in the opposite direction and it relates to the role of water people think, such that as you go up in temperature, you favor phase separation. For most biopolymers, as you go up, the droplets tend to go away. And so we've seen that a number of our systems here. So have you tried to ask if different transcription factors can interface with these droplets in different ways, for example, pioneer transcription factors? Yeah. We did one, again, this is one of the things that got clocked by the pandemic. We've looked at one transcription factor. One of the things we're interested in is if we add magnesium, as I said, we can coax both the 30 base pair and the 35 base pair to make droplets. They're easier for 35. We think that those droplets have different internal structures. The 30 base pair is probably making 30 nanometer fibers that assemble laterally. The 35 is probably making our 10 nanometer things. And we asked whether if we rig in a transcription factor binding site into those two, does it care? And with an N of 1, no it didn't, which was kind of disappointing. I was really hoping. One that a new postdoc in the lab would have worked on as a graduate student. ETV4? Yeah, no, sorry. I honestly, it was one of the EPS family transcription factors that has an animal or binding affinity for a specific site. And we could put that specific site into either, again, the 35 or the 30, and its partition coefficient was basically the same in two cases. But my gut feeling though is that there's going to be differences. Because these likely will have different internal structures and biology cares about structure. And so we should see functional differences there. Again, we'll get to that. So I was just thinking about the fact that enhancers and promoters can both have acetyl marks and that they often sort of come together even if not stably. So and to me with your data at the end where you showed that the unacetylated sort of forms a phase separate from the acetylated, whether you guys had thought or whether you had experiments planned that could help to test whether those sort of interactions were helping with that sort of transient enhancer-promoter interactions. Yeah, I mean, we've thought about that, but we don't have anything planned. Cliff Brangwyn, though, another person who works in this field, did a really wonderful work. Cliff has shown that he has a way of using light to stimulate these opto droplets. And he's shown that if the chromatin is relatively soft that he can make droplets that will bring together, when the droplets use, they will bring together distant regions of DNA. So the implication being that perhaps that's involved in sort of long-range enhancer-promoter interactions. I think we might do that later. I see. Yeah, and it sort of like attracts like, yep. Could be. Could be. Could be. I'll piggyback off that and ask a transcription question as well. So you have to move the histones off of the DNA in order to move a polymerase through it, right? But it seems like you would need to maintain some level of histones there to keep it in these droplets and have that multivalency. Is that contribute to the on-off rates of these histones? Does the higher order assembly change the ability of a remodeler to move histones? Or an interesting question. We don't know. We don't know. We haven't looked at it. If I had to guess, I would guess yes. Because with a remodeler just on a single piece of chromatin is only fighting against that nucleosome. Whereas if the nucleosomes you can see in our droplets are making many other contacts, it would have to fight against those as well. That's an interesting question. Well, if there's no more questions, I think we'll let's thank Dr. Rosen for a fantastic series of seminars. Thank you.