Managing editor, Condensates.com
|Type||Kitchen Table Talk|
The Dewpoint scientists and Condensates.com community joined Keren Lasker from the Scripps Research Institute for a lively Kitchen Table Talk on August 18. She got her start with a PhD in computer science, where she spent time studying between Tel Aviv, UCSF, and Max Planck in Munich. She extended her expertise to in-cell experiments during her postdoc with Lucy Shapiro at Stanford University, where she helped lay the foundation of our understanding of membraneless organelles in bacteria.
The broad multi-disciplinary skillset Keren has picked up along the way is perfectly suited to studying condensates. Her new lab at Scripps uses high resolution imaging, computer modeling, and cellular biology to study the structure and dynamics of bacterial condensates, with an eye toward engineering condensates for eukaryotic cells. During her talk, she tells us about the factors that play a role in bacterial condensates and how they can be leveraged in mammalian cells.
Create an Account or Sign In to view the video.
Jill Bouchard (00:00:00):
I’m going to switch gears to introduce the woman of the hour, Keren Lasker, from Scripps Research Institute. Awesome, thanks, Keren. She received her PhD in Computer Science from a plethora of universities, really, Tel Aviv University, she worked through UCSF, and did some time at the Max Planck in Munich. And the highlight, she says, is that she got to solve the 26S proteasome structure. Wow, that was tough to say, a mouthful, I guess. Then she went on to Stanford University to work with Lucy Shapiro during her postdoc, where she really laid some foundations on showing the functional role of phase transitions. And most recently, Keren’s been working on designing condensates, as many of you have probably seen on her bioRxiv preprint that came out earlier this year.
Jill Bouchard (00:00:56):
So, Keren’s actually just getting her lab up and running this summer at Scripps, where she’ll be spearheading research on the structure and dynamics of bacterial condensates, with an eye toward engineering condensates for eukaryotic cells. And to do this, her lab uses really cool high resolution imaging techniques, complemented with powerful computer modeling. And so we’re all really excited to see what the future has in store for her at her lab. And so, thanks, Keren, for joining us today. We’re all super eager to hear about your work today on the bacterial and designer condensates. Please take it away.
Keren Lasker (00:01:29):
Thank you. So, hello everyone and thank you, Jill, and Dewpoint for this kind invitation. I am super excited for the opportunity to present my work as part of the Kitchen Table Talks. And today, I’m going to tell you about cellular organization in bacteria, and in particular, about one biomolecular condensate that resides at the pole of the bacterium, Caulobacter crescentus, and how it can be used for synthetic applications in mammalian cells…
Keren Lasker (00:02:00):
So, just to give you an orientation of what we’re going to be doing today, I’m going to introduce cellular organization in bacteria. Then, I’m going to introduce Caulobacter crescentus and the polar organizing protein PopZ. Then, we’re going to talk about PopZ condensation and its molecular grammar. We’re going to talk about the function of PopZ in Caulobacter, and finally about how we can use PopZ as a synthetic biology tool.
Keren Lasker (00:02:32):
So, cells are organized across many length scales, from the organization of atoms into proteins, proteins into molecular machines, and the organization of these machines within distinct subcellular membrane-bound organelles. We now know that in addition to these membrane-encased organelles, cells harbor a variety of compartments that lack a casing. These protein-and-RNA condensates dynamically assemble and disassemble, and they selectively permit entry of enzymes and substrates to carry out specific cellular functions.
Keren Lasker (00:03:09):
But what about bacteria cells? The size of a bacteria cell is a third of a mitochondria. Do they also use these intermediate organizational scale? So, until not long ago, it was actually thought that a bacteria cell is simply a bag of enzymes. But with advances in multiple imaging modalities and the ability to look at dynamics of individual molecules inside the cytosol, it became very clear that the cytosol of the bacteria cell is both highly crowded and highly organized. And I’m going to list three such examples of such structure, although many more have been discovered and characterized in detail.
Keren Lasker (00:03:57):
So, first, with the aid of nucleo-associated proteins, the bacterial chromosome hierarchically folds into compact DNA, while remaining accessible for protein complexes involved in replication, transcription, and DNA repair. Second, the tubulin-like protein, FtsZ, assembles at mid-cell to initiate cell division. Additional essential proteins are then recruited to the division site and activate cell envelope constriction. And finally, chemoreceptor arrays allow bacteria to assess their chemical environment and bias their movement towards a more favorable conditions.
Keren Lasker (00:04:46):
And really exciting for us is that we’re now starting to appreciate, really, the last few years that in addition to this highly ordered structure that have been studied for a long time, the bacteria cytosol actually includes several biomolecular condensate akin to their eukaryotic counter partners. These include, among others, RNA degradation bodies, RNA polymerase bodies, phosphatase granules, and carbon-fixing bodies.
Keren Lasker (00:05:17):
My work is focused on aspects of cellular organization in the bacterial model system, Caulobacter crescentus. Caulobacter crescentus is a freshwater bacteria, part of a highly abundant and diverse alpha-proteobacteria class. Caulobacter goes through asymmetric cell division. So, what do I mean by that? One life form of a Caulobacter cell is a multi-flagellated swarmer cell that cannot replicate its chromosome and is in a G1 phase. The swarmer cell then differentiates into a stalked cell by shedding its flagella, attaching to a surface, and synthesizing the stalk at the same pole where the flagella once resided.
Keren Lasker (00:06:04):
At this point, the cell switches to its S phase and starts replicating its chromosome to grow and to synthesize a new flagella and to the cell pole opposite stalk. This process takes approximately 90 minutes in minimal media. Once DNA replication is completed, the cell enter the G2 phase and cytokinesis of the inner membrane using those FtsZ proteins that I’ve mentioned compartmentalize the replicated chromosome into nascent swarmer and stalked progeny, giving rise to two daughter cell, the swarmer cell and the stalked cell. The swarmer cell, again, will swim around, eventually differentiate, and the stalked cell will immediately go through another round of the cell cycle.
Keren Lasker (00:06:57):
So, how is this… And this process happen every cell cycle. It’s very predictable and beautiful to see. So, how is this process being regulated? Especially given that cytosolic protein diffuse, on average, on a rate of .5 micron squared per second, and the Caulobacter has no membrane-bound organelles, and that the cytosol of the cell is only 4 micron long, prior to diffusion? So, I mean, wouldn’t all possible signals just be scrambled in the cytosol of the cell? So, it turns out actually, that through many years of work of many groups, we now know that Caulobacter achieves this exquisite assymetry by sequestering antagonistic signaling proteins to the two poles of the cell.
Keren Lasker (00:07:48):
And what happens is that signaling proteins that promote these stalked cell fate that I mentioned are recruited to the old pole. And signaling proteins that promote the swarmer cell fate are recruited to the new pole, such that upon division, the nascent swarmer cell inherits the signaling pathway that will promote the swarmer cell fate, while the mother cell inherits the signaling pathway that promotes the stalked cell fate. Now, this sequestration of at least 13 different types of signaling proteins is mediated by direct and also transient binding through a disordered protein, PopZ, which is the star of the show today. And what’s happening is that PopZ self-assembles at the poles of the cell.
Keren Lasker (00:08:39):
So, what do we know about PopZ? PopZ was co-discovered in 2008 in the labs of Lucy Shapiro and in the lab of Christine Jacobs-Wagner. It is a cytosolic protein, 177 amino acids long. It consists of conserved helical N-terminal region, which was later shown to be used for client binding, and a helical C-terminal region, as well as a disordered region that is enriched with acidic amino acids and prolines. What else do we know about PopZ? Loss of PopZ results in multiple problems with the cell in morphogenesis, polarity, and cell cycle control. The cell will survive, but grow very slow, would lose all of these assymetry hallmarks, such as the stalk and flagella, and essentially, it’s just a mess.
Keren Lasker (00:09:38):
Also something that is very interesting for us now is the realization that PopZ is conserved within alpha-proteobacteria but is not found outside of alpha-proteobacteria. And again, alpha is a very, very diverse set of bacteria, living in all kind of conditions, and all of them have this PopZ protein encoded. And here, what I’m showing to you, is a snapshot of the alpha phylogeny. And so far, PopZ from four species across different order of the phylogeny have been studied. And in all cases, PopZ was shown to self-assemble, to form a microdomain, and to play a critical role in cytosol organization.
Keren Lasker (00:10:24):
So, we really believe that we’re hitting here on a very important function and a very important process in cellular organization of alpha. But we also know that in Caulobacter and in some other bacteria species, PopZ was shown to exclude ribosome and DNA from entering the microdomain. So, when I was in Lucy’s lab for my postdoc, we wanted to understand those driving forces for selective entry. So, now, studying PopZ in Caulobacter is very challenging. Why? The size of the PopZ microdomain is 250 nanometer, which is within the diffraction limit of optical microscope. So, can’t really do much, we think, but to overcome this challenge, we decided to perform single molecule tracking experiments of cytosolic proteins inside and outside the microdomain.
Keren Lasker (00:11:30):
And to perform these experiments, we collaborated with a super talented PhD student at the time, Lexy von Diezmann, and she was a PhD student in the lab of WE Moerner at Stanford. And what I’m showing you here are single PopZ molecules colored by their density, with red indicating low density and white indicating high density. And to start, we decided we’ll start with just a simple control, nothing really interesting, just to measure the baseline diffusion inside and outside the microdomain. And for this experiment, we localized a heterologous PIF protein fused to YFP, with the assumption that a protein like that that is not coming from Caulobacter would not bind any Caulobacter proteins, and because of that, it would freely explore the microdomain.
Keren Lasker (00:12:26):
But to our surprise, when we localized individual fPIF-eYFP proteins, they did not enter the microdomain. Now, that actually was pretty striking. For ribosomes not to enter the microdomain–sort of the mental model we always had in our head–is that it might be that PopZ creates some sort of a mesh and large obstacles like ribosomes cannot penetrate. But here, we just have a very small protein, and we’re still unable to enter this microdomain. But it’s not that all small proteins cannot enter. When we look at additional cytosolic protein, specifically, we look at Caulobacter-native cytosolic protein, ChpT, here in orange, CtrA, in green, both fused to eYFP. They did enter the microdomain as indicated by this red rectangle, while fPIF-eYFP and eYFP alone did not enter the microdomain.
Keren Lasker (00:13:31):
So, what is going on there? What is the mechanism behind this entry into the microdomain? Very suspicious. So, we thought, okay, maybe it’s a matter of size, as I mentioned, because of the ribosomes. But actually, we saw that size is not a determining factor, because larger proteins actually enter, again, indicated by those red rectangles, and small do not. Similarly, PopZ is very negatively charged with this acidic disordered region. So, we thought that maybe entry might be mediated by electrostatic interactions. However, charge does not seem to be a determining factor, as proteins that do enter are at two ends of the charge distribution.
Keren Lasker (00:14:21):
But what we did find consistent is that binding pathway to PopZ, meaning that if the protein directly binds to PopZ, which is ChpT, we’ve shown through SPR directly bind to PopZ, or has a binding pathway to PopZ, in this case CtrA binds to ChpT but then binds to PopZ, it will enter the microdomain. So, we were thinking, can we actually force a protein into the microdomain by changing its binding affinity to PopZ as a way to test this hypothesis. So, to answer this question, what we done, is that we used optical dimer iLID. And in this system, the iLID LOV domain has a low binding affinity to SspB. Those showed here as the bait and the prey.
Keren Lasker (00:15:20):
And upon blue light activation, the iLID changes its conformation, exposing a peptide that bind the SspB bait. So, what we’ve done is that we’ve fused fluorescent protein to this iLID and we fused SspB to PopZ, and we asked whether we can force fluorescent protein into PopZ by simply by shining blue light. And indeed, we observed exclusion of the fluorescent protein from the microdomain in the dark and rapid recruitment into the microdomain upon blue light activation of the binding interaction. And we’ve also observed that the fluorescent protein left the pole at the time proportional to the half-life of this iLID-SspB binding.
Keren Lasker (00:16:08):
So, that leads us to a working model in which entry into the microdomain is mediated by either direct or indirect binding to PopZ. And we are now starting to work to understand the molecular features that drive this process. We also think that there might be competition involved between the different clients. There might be limited places to bind on the PopZ matrix. And we’re happy to get feedback and ideas later in the discussion session.
Keren Lasker (00:16:44):
Okay, so, going back to our scale diagram that I’ve shown you before, so far, I’ve told you about PopZ localization in the cell and its ability to selectively recruit proteins. Now, we’re going to think about PopZ as a biomolecular condensate and about the molecular features that are driving these properties. So, we’re heading now to the third section of the talk. And I’m going to show you that PopZ forms droplets in vitro, in Caulobacter, and in human cells. So, we’ll start with the in vitro data. So, in collaboration with Ashok Deniz and his postdoc Daniel Scholl, we recently show that PopZ creates droplets in vitro, as can be seen by this nice DIC picture.
Keren Lasker (00:17:47):
What we’ve also seen is that when we’re building a phase diagram of the PopZ condensate in sodium phosphate buffer, we build a phase diagram as a function of PopZ concentration and magnesium concentration. We find that PopZ condensation in vitro requires the presence of either magnesium or calcium. And building on a result that I’m going to show you later in the talk, we suspect that magnesium or calcium bind charged residues of the disordered region and modulate these PopZ-PopZ interactions. And just for reference, PopZ concentration in the cell is around 4 micromolar, and a recent paper measured concentration of ions in gram-negative bacteria where Caulobacter is one of those. And they’re in the range of 130-270 millimolar. So, those two range combined within the regime that we see nice droplets in vitro.
Keren Lasker (00:18:48):
Okay, but that was in vitro. What’s happening actually in Caulobacter? So, we would’ve loved to investigate those material properties of PopZ in Caulobacter cell. However, again, PopZ binds membrane proteins, accumulates at the poles of the cell, and creates these 250 nanometer microdomain. So, it’s very hard to do any kind of in vivo assay, really. But luckily for us, really just by accident, we found that in cells with deformed shape, we were able to observe PopZ leaving the pole as one droplet as the shape of the cell deforms, and immediately adopting a spherical shape that continues to bounce around in the body of the deformed cell. And here I’m showing you just one movie of this event, and this movie is in loop and taken in one frame per minute. And what you can appreciate is that this polar PopZ, as the cell deforms, is just leaving the pole into this extended body of the cell, immediately adopting the spherical shape, and it’s going to continue to bounce around.
Keren Lasker (00:20:02):
This suggests that PopZ is a liquid-like microdomain, adopting a spherical shape to minimize surface tension. Further, as Caulobacter is quite small, we decided to use mammalian cells as a test tube to study robustly PopZ material properties in vivo. For that, we started to collaborate with Steven Boeynaems, who is a postdoc in the lab of Aaron Gitler at Stanford. And Steven really is the most creative and talented scientist and a great collaborator, and I hope you’ll invite him to give a talk here sometime. So, when Steven was expressing PopZ in U2OS cells, we saw that it forms droplets outside of the nucleus–so many droplets but all of them outside of the nucleus–and further that these droplets fuse and merge, suggesting that, indeed, PopZ phase separate into these liquid-like droplets.
Keren Lasker (00:21:05):
And another property that was really cool is that those selective features that I’ve described to you before in Caulobacter were actually retained in the human cells. So, first of all, PopZ phase separates separately from stress granules, but if we were to express the cytosolic ChpT protein from Caulobacter in U2OS cells, they are immediately recruited into the PopZ droplets. And another good thing is that… I’m going to show you very soon a series of mutations, and all of the material properties that we measured for those series of mutation in Caulobacter actually matched what we’ve seen in the U2OS cells, meaning that those really are intrinsic properties to the protein and can really be used as an engineering tool.
Keren Lasker (00:22:05):
Okay, so I’m hoping that by now we are convinced that PopZ creates liquid-like condensates. And we next want to ask what is the molecular grammar of PopZ? Okay. So, it was shown in the past that the helical C-terminal domain of PopZ, which is shown here as three helices, but now I’ve done AlphaFold on it, it may be two helices, but we don’t know yet, is the one that drives PopZ oligomerization. And indeed, if you are to delete the C-terminal region of PopZ, either in Caulobacter or in mammalian cells, you lose PopZ condensation, so great. Further, deletion of another conserved… the conserved N-terminal domain does not affect PopZ condensation both in Caulobacter and in U2OS cells. In Caulobacter, it would affect the ability of clients to bind. This is not something we’re going to cover today.
Keren Lasker (00:23:14):
However, we were very puzzled with what happened when we were deleting the disordered region. So, although we didn’t see an effect on polar localization in Caulobacter, in U2OS cells, PopZ looked more like beads on a string versus these large condensates we see in wildtype PopZ. And that’s led us to think about and to try to characterize this disordered region more carefully. So, we started to look at the amino acid composition of the PopZ IDR, and these very high percentage of prolines, around 25% of the linker is prolines and 25% of the linker is negatively charge residues. This looked, to us, quite unique. And indeed, when we projected this sequence composition of PopZ homologs, shown here in red, and the sequence composition of disordered region from the entire human proteome, shown here in gray, during a PC analysis, what we found is that PopZ creates a distinct cluster, indicating that this type of amino acid composition of the PopZ IDR is rarely visited by human IDRs.
Keren Lasker (00:24:38):
So, you might say, “Oh, well. This is just because it’s a bacterial IDR. So, maybe they all look like that.” But now in blue, we are projecting all the disordered regions from all Caulobacter proteins, and we see that this nice separation that we had before is lost and that PopZ clusters away also from most Caulobacter IDRs, with a few notable exceptions, including, for example, the disordered region of DnaA, which activates initiation in DNA replication in most bacteria. So, there might be a really interesting thing here as well to look at DnaA sometime in the future.
Keren Lasker (00:25:23):
Okay. So, our observation is that something is very unique with PopZ IDR and we decided to investigate the sequence features of this IDR further. And as we all know, it’s very hard or it’s almost impossible to do sequence alignment of disordered region, because they don’t align, because they are disordered. But what’s very cool is that there are collective properties that are actually conserved. So, what I’m going to show you now are four such collective properties. So, the first one is the length of this disordered region. So, in all or many of the PopZ homologs in alpha-proteobacteria, the length of the linker is conserved to be around 80 amino acids. And also, all of those homologs have around 25% of prolines in their sequence, again, not alignable, but just collectively they have 25%. Further, they have 25% of negatively charged residues. And finally, and really nicely, there is a nice mixing between those prolines and negatively charged residues along the sequence. And this mixing is conserved.
Keren Lasker (00:26:48):
So, we’re starting to think that maybe these four properties are important for something, and when we started this project, we really didn’t know what it’s important for. So, to understand why evolution maintained those four properties, we turned to do all-atom simulation of the linker region in collaboration with Alex Holehouse. And what Alex’s simulation showed us is that those properties lead to a very high scaling exponent of 0.7, which means there is a very extended linker and it’s self-repulsive. So, that led us to a model in which there is balancing of… I’m going to show you that in a few slides. But there is a balancing act between attraction forced implemented by this C-terminal oligomerization domain, shown here by those hands, and repulsion forces implemented by the linker.
Keren Lasker (00:27:53):
So, to test this type of hypothesis, what we’ve done is that we’ve created mutants where we’ve simply halved or doubled the length of the linker, and we found that that dramatically affect the microdomain viscosity ass measured by fluorescent recovery after photbleaching. And these experiments now are done in mammalian cells because it’s much easier to get robust FRAP data. And what shown here is the fraction of recovery of the bleached spot after 60 seconds. In gray is the wildtype PopZ that shows a recovery of about 70%, but when we doubled the linker, the recovery is complete, suggesting a very liquid material. And when we halved the linker, the recovery reduced.
Keren Lasker (00:28:46):
So, what was very nice is that we actually replaced the wildtype PopZ with either of those variants, and we wanted to see how is that affecting the viability of the cell using a spotting assay. And what we see is that in this double linker mutant, it’s extremely toxic and the cell is showing little to no growth, while halving the linker shows slightly slower growth than the wildtype. So, what is going on here? So, let’s look at the localization of these linkers in Caulobacter. So, here, I’m showing you the two different mutants fused to mCherry. And what you can notice, that in the wildtype and in the short linker, PopZ is still residing at the poles, but in this super liquid double-linker mutant, PopZ is not maintained at the pole anymore, which is really the first time we see it.
Keren Lasker (00:29:47):
And I’m going to show you one movie that just shows you the extreme that this thing can go to, and in this movie, this is an over-expression. But this liquid mutant is not just liquid within itself, but it’s also just swimming around the cell, filling the entire cytosol of the cell. Now, you remember that I told you that ribosomes do not enter the microdomain. So, what is happening in this liquid mutant? So, to answer this question, we collaborated with Elizabeth Villa and her student, Vinson Lam. And what they’ve done was cryo-electron tomography of a Caulobacter cell. So, they’re freezing Caulobacter cell, then they slice it with a focused ion beam, and image a thin slice by cryo-electron tomography, allowing us to look at the finer details inside the cell in a very high resolution.
Keren Lasker (00:30:44):
So, as a baseline, here is a picture of the wildtype PopZ with a polar PopZ that is residing at the pole and excluding ribosomes that I’ve told you before. Those ribosomes are segmented in gold, and also here you see a segmentation of the inner and outer membrane. But now, at the picture to the right, we’re actually able to capture this liquid PopZ as it’s just cruising across the cell, and we confirmed it with a correlative fluorescent imaging. And what is super cool is that now we have this liquid droplet that, again, is not maintained at the pole anymore. And it’s just sliding around the cell excluding ribosome and DNA. So, what we believe that one of the reason that this double linker is so toxic is that this complete loss is… that change in the material properties of PopZ leads to complete loss of cellular organization and constant reshuffling of the ribosomes and nucleoid.
Keren Lasker (00:31:57):
Okay, so remember I told you that we have this model of attraction forces and repulsive forces, and what we’ve actually done by making the linker longer is essentially changing the stoichiometry between those attraction and repulsion forces. So, what we wondered is, can we rescue this phenotype by correcting the stoichiometry? And to do that, what we’ve done is that we’ve created… The wildtype oligomerization domain is this trivalent, three helices. We extended it by repeating helix three and four. And what was very nice to see is that if we now fuse this double linker but use this pentavalent oligomerization domain, we’re actually able to go back to the material properties as we have in the wildtype. And as a control, if we just use the pentavalent with the wildtype linker, we’re actually getting a pretty solid PopZ microdomain.
Keren Lasker (00:33:18):
And also very nice is that, if you’re going back to a viability assay, we can see that restoring the stoichiometry actually restores the viability of the cell. So, we actually continued to doing these type of experiments, because so far, I just talked to you about the length of the linker. But we actually also looked at the percentage of charge residues and prolines and their mixing between them and positioning and so on and so forth. This is in the bioRxiv paper, you can see. But this slide really summarizes a lot of different mutants. And I think that this is really an important point.
Keren Lasker (00:34:05):
So, on the x-axis is a calculated mobile fraction. So, the more right you go, you’re more liquid, and the left you go, you’re more solid. And the y-axis tells you the fitness, and the higher the number, you’re more fit. And what we see is that there is this type of Goldilocks effect where you have to be… If you’re too liquid, the cell is not going to survive, and if you’re too solid, also the cells are not going to survive, probably from a different reason. So, we believe that the reason evolution maintained these properties of the disordered region is really to tune the material properties to be just at the right spot to create those microdomains, which I’m going to tell you the additional function in a second, to create those microdomains that are required for cellular organization, but also keep them intact at the poles of the cell, at the right size, and the material properties.
Keren Lasker (00:35:13):
So, to summarize this section, I hope that you now agree with us that PopZ employs this modular design, and we are breaking it into three modules. There is the driver domain, which is this end to it, and C-terminal domain, which is the oligomerization domain. There is the disordered region, which tunes the material properties, and the N-terminal domain, which I haven’t talked about a lot here, but there is really nice work from Grant Bowman’s lab showing that the N-terminal domain is required for client binding. So, we see it as a domain that can give you function, and later, if we’re thinking about designer condensate, of course, you can change this functional domain to be whatever you want, but you’re still going to have those two modules of the driver for oligomerization and the tuner.
Keren Lasker (00:36:09):
Okay, so now I’m going to give you a little bit more information about what is PopZ actually doing in Caulobacter. And for that, we actually have to go back to this assymetry story. And if you remember, I told you that the whole beautiful reason that PopZ maintains this assymetry is that PopZ allows for sequestration of antagonistic signaling proteins to the two poles of the cell. But now, let’s go a little bit deeper in mechanism. So, it turns out that what largely determines a swarmer cell fate versus a stalked cell fate is the availability of a massive transcription factor, CtrA. And active CtrA, which makes phosphorylated CtrA, would promote the swarmer cell fate and inhibit the stalked cell fate.
Keren Lasker (00:37:09):
And what is happening is that those signaling proteins that I told you, those that are residing at the new pole, would actually lead to activation of CtrA, and those that are residing at the old pole will actually lead to the deactivation of CtrA, such that once a cell divides, you have a swarmer cell that is full of active CtrA, which would promote expression of genes that are needed for swarmer cell fate, and the stalked cell would be cleared of CtrA, allowing initiation of DNA replication.
Keren Lasker (00:37:43):
So, the sole source of phosphate for CtrA is a membrane-bound autokinase called CckA. And CckA is one of those PopZ clients, and it binds to PopZ, and we’ve shown it, and it also acts as a kinase specifically at the new pole of the PopZ microdomain. Critically, CckA can also function as a phosphatase by distinct biochemical mechanism. And CckA acts as a phosphatase everywhere else aside from the new pole. And the last player is ChpT. So, for CckA to pass its phosphate to CtrA, it goes through this intermediate cytosolic protein, ChpT. So, when CckA acts as a kinase, its autophosphorylating, passes its phosphate to ChpT and then to CtrA, and the back transfer of phosphate is from CtrA to ChpT to CckA, permitting a dephosphorylation of the pathway. And then, when we actually imaged those three proteins with respect to the PopZ microdomain, we saw that they are all enriched within the microdomain.
Keren Lasker (00:39:02):
So, I’m going to give you two conclusions from our previous Nature Microbiology paper about the functional implication of sequestering this whole pathway inside a PopZ microdomain. And the first one is that PopZ defines a microdomain for high kinase activity of CckA, reinforcing the assymetry, and that localization of CckA pathway to the microdomain is required for generating a robust gradient of CtrA activity, which is needed for maintaining assymetry. So, I’m going to show you first why concentrated CckA inside PopZ is important for its autokinase activity.
Keren Lasker (00:39:50):
So, Tom Mann, which was at the time a PhD student in Lucy Shapiro’s lab, created a synthetic liposome in vitro to test the effect of surface density of CckA membrane-bound autokinase activity. Now, it’s important to note that CckA is a tetramer in vitro. And what he did, he loaded purified CckA onto the surface of these liposomes via his His-tag-NTA attachment. And he kept the total amount of CckA fixed but spread them onto increasing amounts of liposomes and measured the autophosphorylation of CckA through a radioactive ATP. And really, his beautiful result is that the autokinase activity of CckA, but actually very importantly it’s not the phosphatase activity, is density dependent, meaning that for CckA to achieve high kinase activity, CckA molecules need to be clustered together, and PopZ is the one that’s going to help them do that.
Keren Lasker (00:40:54):
So, when we went back to our single-molecule experiments with Lexy, we constituted single CckA molecule localization for many cell. We observed that PopZ defines clustered zone of CckA, meaning that CckA cluster is defined by where PopZ is, it’s not going to extend, and that the surface density of CckA is highest at the new pole microdomain. And when we are looking at… And what’s really cool is that the concentration of CckA at the PopZ new pole microdomain is matching what we see in vitro as leading for… It’s concentrated enough for autokinase activity, but everywhere else, the concentration is not sufficient for autokinase activity. So, by having PopZ allowing for CckA molecules at the new pole and maintaining PopZ at a specific size, we can reinforce this autokinase activity at one position, patterning the autokinase activity of CckA.
Keren Lasker (00:42:15):
We also saw that now, when we move to do more detailed imaging for the dynamics of those three proteins with respect to PopZ, we saw that all those CckA molecules that are inside the PopZ microdomain are mobile. They do have a reduced diffusion inside the microdomain but they still move. And even more strikingly, the cytosolic protein ChpT and CtrA, upon entering the microdomain, they slow down by at least tenfold in their diffusion, leading to increased dwell time within the microdomain. They’re still moving, they can still find their partners for those phospho-transfer, but the movement is slowed down, and we have this increased dwell time. So, why is it important now to have those three proteins inside the microdomain with this increased dwell time?
Keren Lasker (00:43:16):
So, to understand that, we’ve done reaction-diffusion modeling, and we started by… We wanted to understand the benefits of sequestering this density-dependent kinase, CckA, and those substrates inside the microdomain. So, based on published literature, we know that CckA kinase is mostly at the new pole and ChpT phosphorylation is also mostly at the new pole and CtrA creates a gradient of phosphorylation. But we’ve shown is that this polar sequestration… So, now we’ve tried modeling to see what would happen if we actually removed this polar sequestration of ChpT and CtrA to the microdomain. And what we see through our modeling is that if those secondary transfer from… If ChpT and CtrA are not going to be in the microdomain, we’re actually going to completely lose the patterning of the CtrA gradient, which is, again, required for assymetry, and less profound, but still we’re going to see this effect, if just one of the other is going to be inside the microdomain.
Keren Lasker (00:44:36):
So, to sort of test… To do some perturbation of this localization in cells, we over-expressed PopZ, and what it did, it led to enrichment of those three proteins at the old pole now versus the new pole. And what was interesting is that we then wanted to see if that would actually affect the ability to activate CtrA. So, we’ve done a qPCR of CtrA-regulated genes, and we’ve seen that if we’re actually changing now the relative sizes and the position of the PopZ microdomain, and as an effect the localization of this pathway to those two microdomain, we’re actually reducing the expression of CtrA-regulated genes, suggesting that, indeed, the composition of the PopZ microdomain affects the patterning availability and activity of CtrA.
Keren Lasker (00:45:39):
And circling back to the previous part of the talk, when I was talking about material properties, we’re now starting to connect material properties to this pathway. So, what I’m showing you here is actually those five mutants that I told you about that have different material properties of PopZ, and I’m showing you to the right now, the expression of one of those genes is a proxy for CtrA activity. And I think it’s very compelling to see that, indeed, we again see this relationship between this Goldilocks kind of effect where the material properties would affect the amount of activation of this master transcription factor. So, we think that… I mean, it’s just the beginning. Of course, PopZ has many functions and talks to so many proteins in the cell that I showed you at least two ways in which the material properties affect cellular organization, both in the distribution of ribosomes and also in availability of active master transcription factor.
Keren Lasker (00:46:54):
So, to summarize this section, what I’m hoping you’re now convinced is that coordinated signaling within this PopZ condensate facilitate assymetry patterning of this master transcription factor, CtrA, and specifically that entry into the PopZ microdomain is selected and requires the binding pathway, and that selective sequestration into the PopZ microdomain reinforces spatial regulation of assymetry. And finally, for the last part of the talk, I want to give you sort of a primer of how we’re thinking about PopZ as a synthetic biology tool
Keren Lasker (00:47:33):
So, I told you that we have this modular view of the PopZ protein, with this driver, tuner, and actor. And for synthetic biology applications, we think that we can actually just now start playing with the actor to do different things. So, this driver, which is essentially the C-terminal domain, we call it PopTag, the tuner is this disordered region that we now have a lot of information of how to connect sequence properties to the resulting material properties, and in the N-terminal domain, we can fuse novel functionalities. And here are a few more experiments that Steven did. So, first, I’m just going to show you that we can direct PopTag to different cellular addresses in these U2OS cells.
Keren Lasker (00:48:33):
So, first, here I’m just showing you that the PopTag itself condenses in U2OS cells. You don’t need the entire protein. I mean, you don’t need the modulator and you don’t need the actor; you just need the driver. But then you can start condensing different things. So, here, for example, the PopTag is recruited to actin filaments and we can start to see a condensation of those actin filaments. Similarly, here it’s recruited to microtubules, and we can start to see condensation of microtubules. And this is my personal favorite, is that if we tag it to lipid droplets, we can actually create PopZ shells around lipid droplets. And of course, there are many other cellular addresses that we can use. This is just to give you a highlight or a taste of what we can do.
Keren Lasker (00:49:35):
Okay, so, so far, I’ve shown you that the connected PopTag would condense in U2OS cells and that we can address directly to different cellular addresses. But what about the composition? I mean, can we actually change composition? And this is where the actor, the N-terminal domain, is coming. So, what we’ve done here is that we essentially fused a nanobody to the N-terminal region of the PopTag. And by doing that, we can now control droplet composition. So, in the first experiment that Steven did, he show that we can actually use a GFP nanobody and recruit GFP proteins into PopZ condensates. And just to show you that this is not just GFP, we can actually now recruit stress granule proteins–which before I showed you the phase separate separately from PopZ–we can actually recruit stress granule proteins inside PopZ.
Keren Lasker (00:50:52):
And also very cool is that we can deplete the nucleus from proteins by asking a protein that’s supposed to be in the nucleus to bind to PopZ. PopZ would be outside of the nucleus, but then we can actually push the protein away from the nucleus keeping it away from its spot. So, this is sort of the setup. We’re not working on various applications to use this system for engineering. And we’re always happy to talk about possible ways to use it, and we’re now also working on seeing how changing the material properties, given everything we know about the PopZ linker, can affect function in U2OS.
Keren Lasker (00:51:38):
So, to summarize this talk, what I’ve hope I’ve showed you is that PopZ creates liquid-like droplets in Caulobacter cells with selective entry properties, that PopZ architecture is conserved across Alphaproteobacteria, that PopZ can be engineered to accommodate a range of behaviors, PopZ creates a zone of selective activation and assymetry determining kinase, essentially reinforcing assymetry in Caulobacter, and this PopZ can be engineered to drive phase separation in human cells. And acknowledgement, there are so many people I should say thank you to, first of all and foremost, to my postdoc advisor, Lucy Shapiro, who was very supportive throughout the entire postdoc process, taking a person that did a PhD in computer science and allowing her to start pipetting in the lab, and various people from my lab, especially Tom and Xiaofeng and Daniel, to my awesome collaborators, Lexy, Steven, Elizabeth, Alex, and most recently, Ashok and Daniel, and for all the funding agencies along the way.
Keren Lasker (00:52:55):
And I also want to give a very, very special and warm thank you to the wonderful community of condensates and especially to my close friends at IDPSIG and IDPSeminars who really were instrumental to my career. And I feel that being a part of this community was just really a game-changer for everything that I was thinking of doing. And we do have an early career virtual poster session coming up, so please check that out. And lastly… Oh, we don’t have the slide. Oh, no. Sorry. And lastly, I just started of my lab at Scripps. If you are interested in this type of work that we are doing, please consider joining us. We already have two members in the lab, Xia and Asma, which are a research associate and a postdoc. And yeah, that’s it. I’m happy for the discussion. Thank you.
Jill Bouchard (00:54:04):
Awesome. Thank you so much Keren. That was a pretty awesome talk. I must say that my favorite was the video of the really liquid linker condensates. That’s like magical-looking. Our first question, Charlotte, are you still around? Do you want to ask your questions about the IDR and…
Charlotte Fare (00:54:25):
Jill Bouchard (00:54:26):
Charlotte Fare (00:54:27):
Can you hear me okay?
Jill Bouchard (00:54:27):
Yup, you sound great.
Charlotte Fare (00:54:29):
Keren Lasker (00:54:29):
Charlotte Fare (00:54:29):
Hi. So, my first question, you showed… I was also struck by the video of the extended IDR, and I was wondering if under non-over-expression conditions if it still is properly localized?
Keren Lasker (00:54:44):
No. So, this is the picture that I’ve shown before. It still creates those droplets outside of the pole.
Charlotte Fare (00:54:54):
So, do you think if you were to… Like analogous to how you doubled the amount of helices on the C-terminus, if you were to double the N-terminal portion, do you think you would recover localization?
Keren Lasker (00:55:07):
Yeah, we’ve shown that. We can recover.
Charlotte Fare (00:55:12):
Okay, and then my second question is sort of also relevant to the last portion of your talk with the more synthetic condensates. And I was wondering if the kinase activity of the phosphorylation cascade that you were showing, is that affected by the localization and properties of the PopZ condensate and then sort of in a synthetic context, do you think that you can then tune the length of the IDR to sort of modulate enzyme activity?
Keren Lasker (00:55:45):
Yeah, so, you’re hitting exactly of where we’re going. And this is the reason I wanted to show this, because I think we now have a system which is not too complicated but complicated enough to really start teasing out this relationship between phospho-transfer and material properties. I do think that there is something there, because in mild mutation, where we just half the linker, to be half the size, and it’s not really dramatic, we can still see effect on [inaudible] activation of this master transcription factor, CtrA. So, I think that there is something there, but we’re really now, I think, can start and figure these things out in more detail.
Charlotte Fare (00:56:28):
Keren Lasker (00:56:31):
Jill Bouchard (00:56:32):
Very fair. I see we’re at the top of the hour, so if anybody has to drop off, you’re free to go. But everybody else, stay along for the discussion. We have a question from Guanhua He. I’m sure I’m slaughtering your name, so I’m sorry. But if you’d like-
Guanhua He (00:56:48):
Jill Bouchard (00:56:48):
Guanhua He (00:56:50):
Yeah, you’re pronouncing correctly.
Jill Bouchard (00:56:52):
Guanhua He (00:56:52):
So, yeah, I saw that you changed the lengths of the IDR to make it half or double, and obviously, the double like messed up with the material property and also the growth, but I wonder how adding two helix repeats, like three and four, could rescue this phenotype in terms of the growth. I just don’t get the logic here.
Keren Lasker (00:57:11):
So, what happens is that… This was Charlotte’s question as well. I probably should have shown that in the slide, but once we are adding those two helices, what actually happens is that the localization is going back to be polar. So, first of all, we’re actually losing this liquidity. So, there has to be some valency relationship here. And what’s happening with… Okay, so what we think that is happening is that, potentially, extending the linker actually inhibit the ability of PopZ to bind other PopZ molecules. And by extending now this oligomerization domain, we’re adjusting back the valency to allow for interaction between PopZ to PopZ molecule.
Keren Lasker (00:58:10):
So, pieces of data that we now have, our first that if we just by FRAP, we are recovering the material properties close to wildtype just by extending that. Second is that in Caulobacter, the localization of these pentameric double extended linker is back to like wildtype. So, between those, what we’re actually now trying to do more detailed experiments to really understand what is the relationship between the IDR and the oligomerization domain. Is it that the IDR is masking out potential binding interfaces on the oligomerization domain or is it something else that is happening between IDR to IDR? So, we have a series of experiments that we’re going to try to do in the future. But just by studying material properties, we can see that we are back to the viscosity of the wildtype.
Guanhua He (00:59:16):
I see. Yeah, this is also, I think, puzzling because I thought the IDR is not essential for oligomerization, not essential for helix interaction.
Keren Lasker (00:59:25):
Guanhua He (00:59:25):
But somehow it messed up, is the…
Keren Lasker (00:59:27):
Yes. So, okay, so what we’re thinking that is… We have other section of our work that I didn’t present here today, but we’ve done a couple of simulations, and we can show that there is a competition between PopZ to PopZ interaction, inter- versus intra- IDR oligomerization domain. And you can essentially have the disordered region bind to the oligomerization domain, and at this point, you’re going to lose valency to additional PopZ molecules. And that would essentially break your condensate.
Keren Lasker (01:00:19):
So, we were able to show that when we were actually changing the distribution of the charge residues to be close to the oligomerization domain or further from the oligomerization domain. I didn’t cover these parts today, but those are sort of the clues that we have now, and we’re now going to do some single-molecule FRET and other experiments to really tease out this relationship between intra- versus inter- PopZ interactions.
Guanhua He (01:00:54):
I see. Thank you. Looking forward for the updates. Thank you.
Keren Lasker (01:00:57):
Jill Bouchard (01:00:59):
Looks like Yifan has a related question. I’m not sure if you answered it, about the… Yifan, are you still around?
Yifan Dai (01:01:05):
Yeah, yeah. Thanks for the talk. So, I’m wondering like the PopZ-PopZ interaction, if it only depends on the helix interaction and this is going to be a fixed stoichemical ratio for such interaction. So, how can you differentiate colocalization with phase separation?
Keren Lasker (01:01:27):
So, first of all, regarding your comment on fixed… So, what I didn’t mention to you at the talk today, is that at the end of this C-terminal domain, there’s actually a patch of positively-charged residues that are highly conserved and are important for creating additional valency. So, we’re not sure, again, how many ways there is for inter- PopZ-PopZ interaction. And second, regarding the localization versus phase separation, I mean, yeah, that is a very important question. What actually makes PopZ localized to the pole of the cell?
Keren Lasker (01:02:23):
We do have some additional unpublished data that shows that a lot of it has to do with other forces that are happening in the cell, such as the DNA and the material properties of PopZ and how those, sort of, opposing forces work together to localize PopZ. So, this is why we’re using this mammalian cell system, to show you that PopZ does phase separate. In Caulobacter, it also localizes to the pole for various reasons that have to do with cellular organization. But when we’re changing the material properties to, for example, be to liquid, those consideration are just off, out the window, and PopZ is just losing its localization abilities, even though it can still bind to all those membrane polar proteins.
Yifan Dai (01:03:30):
Jill Bouchard (01:03:32):
So many variables, it sounds like, to me. We have a really cool question from Shruti. Shruti, are you around? About the flagella and stalk.
Shruti Jha (01:03:45):
Jill Bouchard (01:03:47):
We can hear you. Go ahead.
Shruti Jha (01:03:50):
Yeah, the reason I was curious, because you know, these two cell types change wherein the flagella becomes the stalk, right? Like, the flagella finds like a place to attach and then it becomes a stalk, and all of a sudden, it’s a different cell altogether. So, I was just thinking where does this change arise from? Could it be like once the flagella is not really operating, because how flagella operates it’s the whole new story of its own, like it has its own ways of operating. Once that stop operating or it becomes a stalk, could it be that something happens to the PopZ condensate that causes to promote different kind of protein content in that condensate? I was just thinking in that direction, could that change?
Keren Lasker (01:04:41):
Yeah, so, in this differentiation from swarmer to stalked, actually a lot of things are happening, and various clients are being destroyed and new clients are being synthesized and enter. So, what’s happening during this differentiation is actually also changing of the client composition within the PopZ microdomain.
Shruti Jha (01:05:09):
Mm-hmm (affirmative), okay.
Jill Bouchard (01:05:09):
Cool. Awesome. So, Talita, it looks like we have the last question. Do you want to unmute and ask?
Talita Duarte Pagani (01:05:17):
Yes, I was wondering if you know if the IDR portion of PopZ expressed alone allows the bacteria to survive or have you tried to express this IDR alone to see what happens in the bacteria?
Keren Lasker (01:05:37):
We haven’t expressed just the IDR by itself. It’s also, it’s just going to be diffused based on first one that we’ve done. But we haven’t checked for what it does to viability. But it’s not going to create any polar condensate.
Jill Bouchard (01:06:04):
Very good. Well, that was very enlightening, and I hope it was very inspiring for other people and their next experiments, because there are so many details that you covered, and we just really appreciate you sharing them all with us. So, thank you so much, Keren. It’s been really lovely to hear your talk. And I guess you’re seeing rounds of applause from the globe. So, we’ll close it down, and thanks again for coming.
Keren Lasker (01:06:33):
Jill Bouchard (01:06:35):
See you all next time.