Scientist Stories: Maddy Parsons, Imaging Cytoskeletal Dynamics

I know, sorry, 1 o'clock.

I won't say anything controversial now.

Could reach the millions. Are you welcoming, Alex?

Indeed.

Thousands?

Well, I'm hoping.

Also for the people who kind of can't get enough of imaging, they could join still the Royal Society symposium afterwards and tomorrow. I put actually the thing in the chat. So half, one my ears in the symposium and one is in Imaging One World.

Multitasking.

Okay, well, shall I start and then while people are joining? So, um, so welcome everyone to Imaging One World. So So you're aware the RMS has actually just updated its website, or is in the process of updating its website, so you can now get calendar links from the website if you want to join any of the talks, which is quite an advancement. So that's worth looking out for. I think it'll start appearing in emails as well. And if anyone has any ideas of future speakers, let us know. I think we've actually got a packed plan until May now, I believe, but we're always looking for new ideas. And today we're fortunate enough to have Maddie Parsons who is a professor of cell biology at King's and director of the Nikon Imaging Centre and is known for magnificent microscopy movies from— so, and I think I probably don't need to introduce her any further than that. So, oh, we know why we normally don't let Alex introduce people.

Thanks, Alex, for the lovely introduction. It was all lies, obviously, but thank you very much for having me. It's, it's an enormous privilege actually to be part of this because I recognize you've had some absolutely stellar names come through this platform in the last few months and I'm not one of them, so obviously I'm going to disappoint you today. But I can at least tell you about endpoints of Microscape use.

We gave them presents initially.

Where's my present, Zefi?

We talk about it later.

Okay, so let me just bring this up. So hopefully you can see that. No one's saying no, so okay, perfect. So thanks again for the invitation, it's an honor to be here today. So yes, I'm going to be talking to you about a couple of projects involving lots of microscopy to study cytoskeletal dynamics in invading cancer cells, and by way of a little upfront disclaimer, As I said, I've been in awe of all the speakers that have come through this program over the last few months, and the majority of them are sort of hardcore biophysicists and for the large part instrument development people or software development people. I'm a cell biologist, in fact I'm a biochemist by training, but I promise I'm— I think I'm only going to show you one western blot in the next half an hour or so. But I love microscopes and we use them a lot, so what I'm hoping to do is show you how we've applied lots of different types of microscopy approaches to study a particular protein that we've been interested in over the last few years, and that will give you a flavour of sort of what we do in the lab and why we're interested in using microscopy more generally. So we are really interested, or a lot of people in the lab are really interested in understanding this protein called actin. So many of you might know that actin is a critical scaffold protein, it's expressed in all cells in the body, and for our purposes and our interests we're really keen to understand how actin and the organisation of actin regulates the process of cell adhesion and cell migration. And that's really typified here in this movie of a fibroblast. So this is a single fibroblast just placed on a coverslip that's expressing GFP actin and a red RFP-tagged focal adhesion protein. And you can see the focal adhesions sort of form the feet of the cells, allow the cells to stick down to the underlying extracellular matrix. And the actin is very actively, dynamically turning over at the periphery. And you can also see the formation of lots of different types of structures that F-actin forms during this process of polarization. And we know quite a lot about these different types of actin structures that can form in cells. They've been very well characterized in lots of different cell types. So just to highlight a few of the things that you can see in this movie over here that lots of people have been interested in over time. At the start, we're interested, and lots of other groups have been interested in characterizing the lamellipodium, which is this sort of branched F-actin network that forms at the edge of a leading or a spreading cell. And this has been fairly well characterized in terms of the proteins that can control the branching of actin that really enables the cell to form these large sheets at the leading edge and allow it to sort of sense the extracellular environment. And then you can see these more bundled F-actin structures within the lamella, and these are essentially linking through to the focal adhesions that I talked about that allow the cell to link to the extracellular matrix and form this interference, sort of crosstalk between the extracellular matrix and the inside and the organization of the actin itself. And so these stress fibers can tether to proteins that are in the focal adhesions to allow the cells to couple the sort of traction against the extracellular matrix and allow the assembly of these focal adhesions in the right place at the right time. And you'll see that the focal adhesions are also dynamic, so they turn over over time as the cells are moving around during the course of migration. And one of the sort of more interesting and dynamic structures that we've been really interested in studying over the last few years are these things called filopodia. And these, you can see, are very, very transient linear parallel effacing bundles that form generally at the leading edge of cells, but all around, and they're thought to be sensing structures. So essentially the cells are throwing these filopodia out, and you can see them also in this movie at the leading edge of this cell here, and they're throwing them out in order to sense the environment and then bringing them back in again. So they're very, very rapidly adapting structures that allow the cells essentially to control the sensing machinery and determine where to lay down new F-actin over time to undergo the migration process. So the motivation in the lab really for understanding migration, or at least one of them, is in the context of tumour cell metastasis. And many of you will know that this is really important in many of the stages of metastasis. So there's the initial stage where the, these tumor cells have to leave the primary tumor and they essentially have to burrow through the basement membrane that surrounds the primary tumor. They enter the stromal environment that surrounds the tumor and this can be a very complicated environment, lots of different types of extracellular matrix proteins, lots of different types of mechanical interplay and different types of cells that the tumor cell has to navigate through. And upon that sort of process, once they've completed that invasion, they'll hit the vasculature and have to enter travel through the vasculature and exit and then migrate again to form this distinct and distant metastatic loci. So all of these processes require the cells to adhere and to migrate and to turn over these actin structures that I've explained, and all of those processes are critical to allowing the cells to go through this so-called metastatic cascade. So within that context, we've been really interested in this protein called fascin for a number of years now, and fascin was really the first identified protein that can cross-link parallel F-actin bundles. So it's a highly conserved protein, and we know through many, many studies from our group and from many others that it's really important in controlling the formation of these filopodia structures that I explained to you are really critical for the cells to sense their microenvironment. And we've shown that it also binds microtubules in addition to actin, and we think that those two things can happen in synergy to allow the cells to coordinate these different types of cytoskeletal structures as the cells are adhering and migrating. And what's also important to notice is that FASCN expression is highly restricted in terms of the tissues and indeed the cell types that it's expressed in. The knockout mice are just about viable, but they have lots of neuronal defects during development. So FASCN is very highly expressed in the developing neuronal system, and we've shown with collaborators in the past that that's important for the neurons to migrate during migration. What's really important for us and in the context of cancer is that it's not expressed, or it's very, very, very lowly expressed in normal epithelial cells, but it's very, very highly upregulated in pretty much all cancers that have been studied to date. So because most cancers are derived from the epithelium, this says to us that upregulation of FASN is really important for cancer cells for lots of different reasons to allow the tumorigenic process to occur. And because this process also correlates in patients, this high level of FASN correlates with high levels of metastasis, we think that it's important in controlling not only the tumor growth and the sort of initial tumorigenesis, but also the ability of cells to undergo the metastatic cascade. And indeed, we and lots of other people have shown that if you knock out FASCN, you reduce filopodia formation, and you also reduce the ability of cells to invade in various different assays, both in vitro and in vivo. So we think it's a really interesting protein to study in this context and understand further. So there, as I said, been a lot of work done by our lab and by a number of others in terms of trying to understand how FASCN is regulated. And this diagram just shows you a very simplified format to demonstrate that it's a relatively straightforward looking molecule. It's made up of these 4 so-called beta-trefoil repeats. And within that, they're separated by these very short linkers. And within these repeats, there have been a couple of critical phosphorylation sites that have been identified that control this ability of FASCN to bind to F-actin and to bundle it. So the first one that was identified was at serine 39 in the N-terminus, and what we and lots of other labs have shown is that if this site is phosphorylated by protein kinase C, it actually inhibits the ability of fascin to bind to F-actin on this side and also to bundle it. So fascin has to adopt this sort of compact structure in order to have two adjacent actin binding sites, and that's what allows Fascin to bundle F-actin within filopodia. And if you disrupt by triggering phosphorylation on this site, it disrupts binding on this surface. And we also discovered in collaboration with Brian Strohm's lab a few years ago that serine 274 on the other side in the B3 domain forms part of the other F-actin binding site. So if you, again, if you phosphorylate that site, it disrupts binding. So these are the two critical interfaces that allow Fascin to bind and bundle actin. However, despite all of this knowledge and really the recognition of the importance of Fascin in controlling cancer cell invasion and metastasis, there was still very, very little known about how Fascin was controlled in terms of its spatial and temporal and functional roles. And this was really something that we wanted to drill down into to understand what proteins might be regulating Fascin localization and ultimately Fascin function in different contexts. So the first project I'm going to tell you about was one that we really tackled sort of early on in terms of trying to understand how fascin function was localized, and really how this sort of precise control of fascin into the filopodia was regulated by potential upstream proteins and signaling cascades. And this was a project that Karen, who was a really exceptional postdoc in the lab, at the time, who's now gone back to Vienna to set up her own lab, took on a few years ago. And she had a huge amount of help from James Levitt, who was a postdoc in Simon Amirbek's team at the time, and he's now actually the manager of the Nikon Imaging Centre at King's. And Richard Marsh, who is a postdoc in Susan Cox's lab. And these guys are so critical in helping us in our collaboration here, because they are essentially— they were either building microscopes or they were developing new tools tools to allow us to address these very complicated questions that Karen was posing. So the first thing Karen did to try and understand how quickly FASCN was moving around in cells, and when it did move, where it was going, was to develop a HeLa knockout cell line. So HeLa express a lot of FASCN naturally, so she knocked it out, the endogenous gene out, and she replaced it with a form of FASCN tagged to this photconvertible protein called MEOS3.2. And the reason we wanted to use this protein is because it's a really nice tool to be able to track different populations of the same protein over time. So the way this works, if you haven't come across it, is to be able to photoconvert a subpopulation of this protein within a specific region of the cell, and you photoconvert it, in this case, from green to red. So in the non-converted state, it's green, it looks like GFP, and when you photoconvert it, it switches into the red fluorescent channel.. And that allows us to, as you can see in this example here, photoconvert an entire population of FASCN that's in these phyllopodia that are on this side of the HeLa cell, and then we can actually watch where the converted population of protein pops up across the cell over time. And that will tell us something about the dynamics and the localization of FASCN at the sort of whole cell level. And when she did this, you can see hopefully from this surface plot over here, that when she photoconverts the fascin on one side, it very, very quickly pops up in other filopodia structures on the other side of the cell. So indicating to us that it's traveling around the cell very, very quickly. And indeed what this assay allowed Karen to do was to start to track the kinetics and rates of movement of new GFP protein into converted filopodia, where we'd switched everything to red, and how quickly the converted red protein came out of filopodia and popped up somewhere else in the cell. And from this she could actually pull out t halves of movement of fascin into phyllopodia, and she demonstrated that the kinetics were very fast. So she was seeing movement of fascin on the order of about 2 seconds into phyllopodia upon conversion, and she was seeing a slightly slower movement of about a t half of about 6 seconds out of phyllopodia. So suggesting perhaps that there is a sort of more active transport mechanism that's recruiting fascin into phyllopodia from the bulk cell body, and there's a slightly slower and maybe less active or passive diffusion of Fassn back out of filopodia. But overall, we were very surprised by this because we were expecting it to be much slower and much more stably bound in the filopodia when it got there. So to drill down into this in a little bit more detail, what Karen did was to perform fluorescence recovery after photobleaching experiments. So this is where we're just using the GFP Fassn expressed in the knockout cells again. And what Karen did was to look at a single filopodia and bleach a region out using a confocal and then track the GFP fascin coming back in. And the goal of this experiment was to understand again, whether there was this difference in active transport from the cell body into the filopodia versus this sort of more passive retrograde flow of fascin coming back from the tip of the filopodia. So she did these on single filopodia and bleached a little region out here, you can see, and tracked the movement of the GFP fascin back from both directions. And this is a chymograph shown in black and white here, and also in pseudo-color, which is slightly easier to see, where you can see the pre-bleach up here, that's very obviously still GFP positive, followed by the bleach in this little area, and then we're tracking the movement of the fascin back in from the cell body or from the tip coming back. And you can see hopefully from this, there's a much faster recruitment of GFP fascin coming into that bleached region within the filopodia compared to the speed of fascin coming back from the tip. So you have a much shallower curve coming in here, again indicating it's being actively transported in from the cell body. So to get at this in much more detail, we collaborated with Rich Marsh and Susan Cox's team, who had built a two-channel single-molecule imaging system, and this was really allowing us to use this Eos phasin to photoconvert a very, very small subpopulation of the phasin within single filopodia, and then track the converted red population of single molecules within filopodia to understand the actual speed of single molecule movement of Fassim within individual filopodia structures. And because we still had the green there, we had a reference, so we know exactly where the filopodia is to start with. So when we did this, we found some very, very interesting and very surprising results. So the first is that there were 3 populations of Fassim that could be identified, broadly speaking, in the filopodia. And these are example single molecule imaging tracks that you can see in the movies here. So we had a very fast population, an intermediate population,, and a very slow retrogradely moving population of Fassin within these individual filopodia. And what was incredibly surprising to us was that the Fass population, of which represented about 20% of the total molecules that Richard analysed, were moving on the order of about 30 microns a second. And we were pretty gobsmacked by this. We haven't seen anything move this fast before or since. And if any single molecule trackers are out there and want to tell us that they found molecules that can move this quickly, I would love to hear from you. We we still don't really know or understand what it's doing when it's moving that quickly. About 70% of the molecules were moving with about an 8 micron per second speed, which is still extremely quick given that we think that this molecule, well, historically we thought this molecule essentially entered the filopodia and really held onto the actin for the length of the filopodia's lifetime. But based on these movements, that's absolutely not the case. So we think fascin is coming in and moving around extremely quickly in the filopodia, and he's still able to stabilize the parallel F-actin in those bundles. And what was really interesting is when we expressed a mutant of FASCN where we have mutated this critical serine 39 residue to an alanine, which means it can't be phosphorylated by its normal kinase, and that should render this, this FASCN molecule much more stably able to bind to F-actin. Actually, what we saw was that there was an increase in the diffusion constant of this this, so suggesting that the ability of fascin to bind to actin is somehow promoting this turnover and this sort of dynamic movement of fascin within the filopodia, which is pretty much the opposite of what we expected. So I would say we still don't understand quite what this means and why this is being controlled so rapidly. It could be down to the spatial confinement of the molecule within the filopodia, which are very small, they're less than a micron wide, or it could be lots of other factors that are controlling it. But this is something we're actively pursuing and trying to understand going forward. So the next question was really, how is the interaction between fascinin and actin being controlled? Because we think now that, based on that data that I've just shown you, that fascinin in filopodia is moving around very rapidly. So what does this mean to its ability to bind to actin directly? And historically, we've worked a lot with Simon Amirbek's team to use fluorescence resonance energy transfer, or FRET, to study protein-protein interactions using time-correlated single photon counting, or TCSPC. And we do this using a fluorescence lifetime imaging microscope that allows us to essentially monitor the lifetime of the donor molecule. And by doing that, understand if we see a reduction in the lifetime of the donor molecule, it tells us that there is a direct interaction between the donor molecule and the acceptor molecule. Historically, the detectors were extremely slow, and they essentially took 3 to 5 minutes to take a single image. And this meant that we couldn't do any dynamic analysis. But over the last few years, there have been quite a lot of developments in terms of the detector systems that are available for these systems. And Simon and James Levitt built this fantastic live cell imaging system that allowed us suddenly to be able to image on the order of taking a single image every second that will give us sufficient photons to do proper analysis to retrieve this sort of interaction data. So what this allowed us to do for the first time was to study the interaction of fascin and actin that we co-expressed in our cells on a 1-second time resolution. And this is the edge of the cell you can see here with individual filopodia, and where you can see these hot spots or sort of the redder pixels popping up, that's where we're seeing a direct interaction between fascin and actin, and that's based on the lifetime of the GFP fascin molecule.. And what this allowed Karen to do was to start to ask questions about where and when you were seeing interactions, specifically in high-level interactions between fascin and actin, within these different places and parts of the phyllopodia. And what's worthwhile noting, and what you can already see from this movie, is that this was a very transient interaction. So over time, we were seeing a huge amount of fluctuation around the lifetime as we were imaging these phyllopodia, and that again indicates that this is not fascin coming in and binding to actin and not letting go. This is a very dynamic turnover effect, which we would have predicted we should have seen because of the data that I showed you previously, where we think fascin is very, very dynamic within these filopodia, and it turns over much, much more quickly compared to the filopodia itself. So what Karen was able to show by analysing lots of filopodia was that we see a very interesting distribution of interaction between fascin and actin across the different types of filopodia. So filopodia can grow, and then of course they can also shrink, but they can also have this period of time where they're relatively stable and they don't really change in their length over the imaging period. And what Karen did was to isolate individual filopodia and categorize them into these different— three different classifications, and she then analyzed the base versus the tip of the filopodia and asked how much interaction there was between fascin and actin in these different areas. And what she demonstrated was there was a very high, much higher level of interaction between fascin and actin at the base of growing filopodia compared to at the tip. And what this suggested to us was that this might be the key site that fascin is essentially tethering actin in a much more stable way and at a much higher level in order for the cells potentially to grow this filopodia, what would be a relatively unstable structure to start off with requires that sort of tethering between fascin and actin actin to stabilise the structure, and then allow it to grow from the plasma membrane. And we didn't see, although we saw obviously interaction in shrinking and stable filopodia, we didn't see any change between the tip and the base, suggesting that that sort of interaction is not spatially regulated, unless the filopodia is actually growing. So the next question is whether fascin is being phosphorylated dynamically within these structures, and You know, we as biochemists like to understand phosphorylation in context, but that's a very difficult thing to do in a live cell because phosphorylation is something that you can't really readily detect by just expressing a GFP-tagged protein. So what Karen did was to design a biosensor to start to get at this question whether or not this critical serine 39 residue is really important in controlling not only fascin dynamics, but also controlling fascin conformation. So she did some modelling of the known structures of fascin, and what she determined from the modelling was that when it's in this very compact state, which as I showed you right at the start is the state that we believe is not phosphorylated and is the form that is essentially able to bind and bundle F-actin, that actually if you position two fluorophores on either side of the fascin molecule here, and here to act as a donor and acceptor, that if the Fassin was in a dephosphorylated state bound to actin, it should be within fretting distance between these two molecules, so around 5 nanometers. And if serine 39 was phosphorylated, we predicted that Fassin would open up its conformation, and that this would essentially mean that these two fluorophores were more than 10 nanometers apart, which is above the distance you can measure fret. So that would suggest that we may be able to design a biosensor that report on the phosphorylation and the conformational change of FASc within a live cell. So after about 2 years of designing many, many different forms of this biosensor, Karen actually managed to make one which worked, and we were able to rescue FASc knockout cells with this, demonstrating that it was fully operational and functional. And you can see here that this is a FRET ratio metric movie that Karen made of the cells expressing the biosensor showing that you've got a beautiful localization to filopodia of this, this conformational phosphorylation sensor, but also hopefully you can see by the changing, the rapidly changing pixel colours, that we also see extremely dynamic changes in the phosphorylation and the conformational state of this molecule in individual filopodia. And indeed she can quantify this, so she demonstrated that there was there was a degree of fret within any cell, so this was clearly reporting on the ability of the fascin to undergo dynamic changes in phosphorylation. If we mutate the serine 39 within this conformation sensor, she was able to show that you saw an increase in the fret, which is what we would expect if you were seeing this more compact structure that we were predicting in the absence of phosphorylation. And in fact, if you do the reverse and generate this into a phospho mimic, you see the opposite, so the molecule is stretched, indicating you see lower FRET. And if we treat cells with an inhibitor of PKC, which is the natural kinase for this, you can see we also see an increase in the FRET, indicating that this, this sort of assumption is right. So the dephosphorylated version is more compact. So this was a really nice tool and allowed us to demonstrate, I think for the first time really, that fascin was dynamically phosphorylated within filopodia, and that in fact it wasn't just getting in there and becoming dephosphorylated and gripping the actin in a very stable and non-dynamic way. So what we also wanted to do was to understand the upstream regulators of fascin. And so really, we wanted to understand if the proteins that are already known to control the assembly of F-actin structures that build the filopodia might also be important in regulating the movement of fascin into those structures. And that might explain in part some of the data that we show where we think we have active transport and active movement of fascin to the base of filopodia to allow it to do its job. And there's been a huge, huge number of studies over the years demonstrating lots of proteins are involved in regulating this branched F-actin structure that I explained to you right at the start, that is really essential in forming the foundation that then allows the cells to consolidate the F-actin and nucleate new F-actin to grow this filopodia as a parallel bundle from the plasma membrane. So the critical proteins in controlling this are the Arp2III complex and the formins, and there are about 15 mammalian formins that have been shown, a few of them have been shown to be really important in controlling assembly and stabilisation of these phyllopanial structures. And ARPC3 is really important in controlling this initial branched actin network from which the phyllopania are thought to emerge. And fortunately, there are a couple of very, very good small molecule inhibitors to these different classes that we were able to use in our various assays to start to ask the question as to whether or not these proteins were really important in recruiting fascin and controlling fascin dynamics. So what Karen did was to repeat her MEOS2 experiments, where she was looking at mobility and movement of fascin in the presence of these two drugs. So SMIFH2, that blocks the formin family, and CK66, that blocks the ARK2/3 complex. And what she was able to demonstrate was that blocking the formin family members with SMIFH2 significantly slowed the movement of fascin into filopodia. So suggesting that, in essence, and we were also seeing this coupled with a reduction in the mobile fraction from our FRAP analysis, suggesting that formins are really important in controlling this sort of initial active recruitment that we're seeing of fascin into the filopodia. Interestingly, we didn't see any effect with CK666, so that's not suggesting that the ARK2III complex isn't really important in controlling the sort of architecture, but it does suggest it's not directly involved in recruiting fascin into those structures. So what she next wanted to do was to study the individual formins that might be important in controlling the fascin-actin interaction. And to do this, she initially tested the SMIFH2 compound to understand if that was not only affecting the ability of fascin to arrive in the filopodia, but if it was also disrupting the ability of fascin to bind to actin. So she did did her live assay again, looking at FRET between fascin and actin, and what she was able to demonstrate, and hopefully you can see from the movies, because these are slightly cooler colours, was a significant reduction in the amount of FRET in the filopodia between fascin and actin when we blocked formins. And what she was also able to demonstrate through screening essentially all of the different formins that we know are expressed in HeLa cells, that was our model system here, to understand which specific formins might be controlling this movement of fascin. And she demonstrated that knock-out of FNL2, one of the very specific formins that is already known to be involved in the assembly of filopodia, specifically reduced the mobile fraction and the T half of fascin, whereas we didn't see that when we knocked out a couple of the other key formins that we know are highly homologous but also have been shown to control filopodia formation. Demonstrating that this FMNL2 is critical in controlling fascin movement and mobility up to the filopodia. And what she also managed to show was that this was through a direct interaction. So we did a lot of biochemistry to map the interactions between FMNL2 and fascin. And she was also able to demonstrate this through FRET, again, by expressing FMNL2 and fascin in cells, and showing that there was an interaction at the periphery here shown by by these warmer colours, and we didn't see that interaction with FMNL3 in Fascin, which would agree with this idea that if we knock out FMNL3, we don't see any impact on the ability of Fascin to move into filopodia. So what we also wanted to do is to understand if this really holds true in a 3D environment, and the cartoon that I showed you of metastasis at the start, you will have noticed, is a very 3-dimensional structure, and cells have to undertake lots of different morphology changes in order to make it through this very stromal environment. And we're trying to model this more and more in the lab to get away from this sort of 2D classical sort of surface analysis that people use. Of course, that's a very straightforward model to use, and it's very nice for imaging, but it doesn't really recapitulate what you expect to see in vivo. And that really sort of holds out here in this movie of a human breast cancer cell that's expressing GFP-LIFAC that marks the F-actin cytoskeleton. And this is exactly the same cell line that's plated on on 2D glass substrates that we all know and love and we like to work with versus in a 3D collagen gel. And hopefully it's blindingly obvious to you that these cells adopt very, very different types of F-actin structures in these two environments. And that's something that we are really trying to understand and to try and really focus much more on the 3D environment, which as I said, we think is much more physiologically relevant. But the question is really always whether what we're measuring in 2D is important for a 3D environment that we might expect is more important for our biology. Biological setting. So to start to get at this question, Karen actually, and we went over a couple of times to the Janelia Research Campus to use their Advanced Imaging Center there because they had this lattice light sheet system, which we thought was going to be particularly useful for us to study fascia and filopodia in 3D. And the reason being that the filopodia are very, very small and they're very hard to resolve using a confocal microscope. And so we were able to actually use this high-resolution light sheet system to start to study the filopodia in a three-dimensional context over time. And one of the first things Karen wanted to do was to understand if actually the formation of FAS-independent filopodia was important to tumor cells in terms of their manipulation and sensing of the microenvironment. And the way she did this was to embed beads into these three-dimensional collagen gels that we use as a model for our sort of 3D collagen stroma that might be surrounding a tumor microenvironment. And these little beads allowed us to track these beads individually over time as we made movies of cells to understand how much the cells were displacing these beads, and as a function of that, understand if they're able to manipulate their microenvironment. And we used PIV, or particle imaging velocimetry, in order to study the movement and the dynamics of these beads over time. And you can see that just based on these little tracks here surrounding the cells. And what she did was just to plate either— put or embed FasC knockouts cells or cells re-expressing GFP-FasCin into these 3D gels and then track the beads. And what she was able to demonstrate over time was that the movement of these beads was much higher when we had FasCin in the cells compared to when we didn't. So indicating, as we might expect, but as we haven't shown before, that filopodia are really important for tumor cells to manipulate their microenvironment in a three-dimensional context. And what she was also able to show was that if she blocked forming in this model, just in the GFP-FASCN-expressing cells, that also significantly reduced the ability of cells to move the microenvironment around. So again, suggesting formins were important in controlling filopodial function and FASCN function ultimately. And we didn't see any difference when we treated with the ARK2/3 inhibitor. And finally, what she also wanted to do was to understand whether this sensing in the 3-dimensional environment was affected by the stiffening stiffness of the collagen gels that we were placing these cells in. And that was really important because, particularly in breast cancer, there's been a lot of studies now showing that the tumor microenvironment stiffens during the course of tumorigenesis, and that that really impacts on the, the way that cells can invade. And arguably, the stiffer the microenvironment, the more invasive the cells become. And the question is really about how that translates into a change at the molecular level and how the cells are sensing that environment. So what she did was to change the stiffness of the matrix without changing the concentration by adding in ribose, which is a non-enzymatic crosslinker that allows us basically to stiffen to a known kilopascal range so that we can understand what that does in terms of the molecular dynamics of fascin within these 3D matrices. And we can label the collagen using a Cy3-Diamonds model so that we can actually see how the cells are interacting with the collagen and moving it around. And hopefully what you can see, and this is again movies taken on the last light sheet microscope, is that the cells assemble a huge number of filopodia, and they seem to particularly assemble more when we increase the stiffness using the ribose that you can see on the right-hand side here. But importantly, what we were able to do in this model was to actually use our MEOS 3.2 Fassion to photoconvert populations and do exactly what we've done in 2D in the 3D model and to test whether or not fascin was turning over and whether that was different in the different stiffness environments. And what she was able to show using her MEOS 3.2 photoconversion was that actually if you stiffen the matrix, fascin seems to move much more quickly. So it's highly dynamic in the soft matrices, but it's much faster in terms of its recovery kinetics when you stiffen the microenvironment. So it suggests fascin is somehow able to respond to this change in the microenvironment mechanics, and in doing that, it seems to make it much more responsive and much faster recruited into filopodia. And we assume that that's to allow cells to respond and adapt more quickly to that microenvironment. And what she was also able to show was that this fast recovery in the stiff matrices was dependent on FML2, as we've shown in the 2D environment. So again, demonstrating that FML2 and the forming machinery is really required to recruit FasCin into these structures and allowing it essentially to do its job in filopodia. Phyllopodia. So just to summarise this first part, hopefully I've convinced you that fascin is a super speedy molecule. It moves around extremely quickly, and we were very surprised by the speed that it moves at. It's continuously exchanging between the cytosol and phyllopodia, and we think this depends on formins. So formins are, and in particular FNL2, is recruiting fascin to, we think, the base of the phyllopodia and allowing that and enabling that interaction with F-actin to stabilise the growth of new phyllopodia. And we also think that fascin, and indeed, it's is binding to FNL2 is responsive to a change in the 3D microenvironment and the mechanical properties of that. And that's something we're actively looking at now. So I just want to also highlight that FASCN has other binding partners and other roles in the cell. And this was a study that Azia did in the lab a few years ago that we've published now a few years ago, but I think it's a really important sort of point to take home that there are lots of other functions for this critically important protein in the context of other compartments within the cell. And Azir really sort of went at this question trying to understand whether or not FasC could localize to other compartments, or if really its entire job in terms of controlling invasion came down to its role in filopodia. And he had, and we had a lot of help from Greg at Columbia and Guillaume, who many of you know is at UCL, in terms of trying to dissect out about these different roles of FASCN in different subcellular compartments. So cutting a very long story short, the lead into this was that we did a mass spec screen to look for novel binding partners for FASCN, and we pulled out this protein called Nespren2. And Nespren2 is a gigantic protein that essentially binds the nuclear envelope and allows the nuclear envelope to tether then through this very large sort of scaffolding protein, Nespren2, to the actin cytoskeleton, and also to many, many other proteins that bind within the numerous spectrum repeats that are found within the DESPRIN2 molecule. So we were a bit surprised by this because we'd never really looked at FASCN, or considered it as a nuclear envelope protein before, but indeed when Azir was validating the interaction between FASCN and DESPRIN2, he started to look more closely at FASCN localization, and found it at the nuclear envelope. And we used initially structural illumination microscopy, and you can see here the FASCN, this is endogenous at the nuclear envelope, and the Nespriin2 that B2B labels the nuclear envelope, and that obviously it's a nuclear envelope protein, so we knew it should be there, but we were sort of quite surprised to find that FASCN was there. We also found it in tumor models, so these are human tumor cells that we injected into mice and grew these tumors over time and then excised them, and we stained up for FASCN and other nuclear envelope proteins and Nespriin2, and we demonstrated that they were co-located as well in this in vivo model. And we wanted to understand if Nesp2 was required to recruit FASCN, or if actually it was the other way around, and FASCN might be important in recruiting Nesp2. So we did a very straightforward experiment knocking out either Nesp2 or FASCN and looking at the other protein. And what we demonstrated was if we knocked out Nesp2, we completely lost localization of FASCN to the nuclear envelope, whereas if we knocked out FASCN, Nesp2 Nesprin-2 was still completely fine. So that suggests Nesprin-2 is important in bringing FASCN to the nuclear envelope and enabling it to form this very sort of nice interaction there. And again, cutting a very long story short, we were able to map the binding site between FASCN and Nesprin-2. So this beta-3 domain within FASCN that I mentioned to you earlier has this second actin binding site, and it turned out to be really critical in controlling this interaction. And we were able, again using our favorite technique FRET, to demonstrate that these proteins interact directly at the nuclear envelope. So if we expressed Despin2-GFP and Fascin-RFP, we were able to demonstrate a direct interaction at the nuclear envelope. And if we then express each of the little domains of Fascin as a, a FLAG-tag protein, we can compete in the case of where we express express the beta-3 domain of FASCN, we can actually compete out the interaction between Nesp2 and FASCN just by expressing this little bit of FASCN protein, whereas we don't have any impact on the binding if we express the other domains. So this really demonstrates that this little beta-3 domain, this sort of quarter of the FASCN protein, is really critical in controlling the interaction between Nesp2 and FASCN at the envelope. So the question is really, why is this important and what's it, what's it doing? And there was some seminal work published by Greg Gunderson's lab a few years ago demonstrating that the movement of the nucleus itself is really important in controlling the direction of travel of cells. And what he demonstrated using a really nice model system was that if you wound cells and you essentially have a monolayer of cells, you generate a wound that allows them to migrate directionally into into it, what he saw was an immediate and rapid backward movement of the nucleus that was sufficient and enabling for the cell to subsequently polarize and move forward. And this was really critical on and dependent on Nespin2 and the interaction between Nespin2 and the actin cytoskeleton. And so we wanted to understand if this interaction between FASCN and Nespin2 was also controlled controlling the Nesp1II-dependent nuclear movement. And to do this, Azir went to Greg's lab to learn how to do this assay, and what he essentially did was to tag the nucleus of cells with an HDV-GFP, so he could watch the nucleus over time in this assay and see if it moved rearwards or not. And also he was able to look at FASCN itself, or to tag these different domains of FASCN that we think should either interrupt or not affect the binding between FASCN and Nesp1II. So this is the sort of time, a chromograph he gets from just looking at the nuclei. And you can see here, hopefully, this is the wounded site here. As we wound the cells, the nuclei moves backwards towards the cell rear over time. And when we express the beta-1 domain of fasciculin that we would predict doesn't affect fasciculin vesperin binding, we see this same rearward movement, and that's quantified over here. And if we express this beta-3 domain, we don't see any movement at all, backwards. So essentially these cells get stuck, the nucleus is no longer able to connect to the cytoskeleton and to move the nucleus rearward, and that's really required for the cells to migrate. And that's quantified here, you can see a significant reduction in the rearward movement of the nucleus when we specifically disrupt the fascin-desperin binding. And what he really also wanted to understand was functionally whether this was important, and if so why. And to do this, we collaborated with Guillaume Chéraou, where he built us these little microfluidic chambers that allow us to understand whether or not the cells are able to enter a specific channel of a given width. So this sort of cartoon shows you how that works. You plate the cells, you then put a chemoattractant at the top here, and you ask the cells to migrate into these ever-decreasing width channels over time. And you can then quantify the number of cells that are able to take it into the channels. And this is a little movie that shows you what these look like in reality. And so you can see in the largest width channels at 20 microns here, the cells do a pretty good job of finding their way in and migrating all the way to the chemoattractant end, and there's no real limitation. Whereas when you start to get down to these much smaller width channels of sort of below 5 microns, the cells really struggle to get into those channels and migrate. And the limiting step here is actually the nuclear membrane. Nucleus, so the cell is not able to move its nucleus and compress it to get it into that gap. That essentially stops it from moving in, and it basically just leaves its backside at the top and is never able to make it down towards the chemotractor. And we can quantify this over time, and you can see here, this is the movies of the labelled nuclei and also the FasCin GFP, and what you can see here is the nuclei either makes it into the channels over time or it doesn't, and we can quantify that based on the number of translocated nuclei that make it into the different channels, depending on the channel width. And what Azia was able to show was that very nicely in the control cells, or in the cells where we've re-expressed FASCN in the knockout background, you hit this sort of magic 7-micron window where these cells that contain fascin are able to move their nuclei into these narrow channels, whereas the fascin knockout cells can't. So essentially, at some point, the cells lose— if they don't have fascin— lose their ability to be able to move their nucleus or to manipulate the nucleus in order to enter into these narrow channels. And indeed, what he was able to show is, again, if we block this specific interaction between fascin and Spn2 by expressing adding this little bit of fascin itself to interfere with binding, we see exactly the same effect. So in essence, you're blocking the ability of fascin-dependent nuclear plasticity at these sort of narrow channel widths if you interrupt the binding between fascin and Nespin2. So again, suggesting that this interaction is important for that plasticity and the overall movement of cells. And finally, we were able to demonstrate this is important in the context of invasion. This is a 3D invasion assay that we and lots of other people use, where you plate cells at the bottom of a chamber and then allow them to invade into a 3D collagen gel over time. And then we take serial Z-stacks through the chamber in order to quantify the number of cells that are invading. And what Azir was able to show is that if we again interfere with this specifically fasciculin-SPR interaction by expressing beta-3, we were able to significantly reduce the ability of cells to invade invade into collagen. So suggesting that all of these nuclear movement defects we were seeing really played out in terms of a functional and important role in controlling invasion. So just to summarise that part, hopefully I convinced you that FASCN can bind to Nespin2 at the nuclear envelope via this specific beta-3 domain. And we think that this represents a novel role for FASCN in terms of regulating the interaction between Nespin2 and the actin cytoskeleton at the nuclear envelope. And this allows allows the cells to adapt and to change the nuclear shape over time to allow the cells to invade. And ultimately, that essentially means that they can move their nuclei as they're invading through very dense and complicated networks of ECM proteins. So in the final couple of slides, I just want to say where we're taking this next, and the next sort of murky world for us is the nucleus. Collaborated with Tina Tuttle a few years ago, who is a fly biologist who works at the University of Iowa, and we serendipitously discovered that both in the fly and in human cells that FASCN could be localized into the nucleus. And so we had this really nice collaboration with Tina figuring out how that works in the context of the fly, and she works particularly in specific cell types trying to understand how prostaglandins control FASCN function. But we're able to demonstrate both in human cells and in the fly that FASCN was definitely in the nucleus. And so we wanted then to understand what this meant and why FASCN was in the nucleus and what it did. So this is a project Campbell's doing right now, and we have a really nice collaboration with AstraZeneca and Sam Peel in particular, who's driving this for us at the AstraZeneca side, and he's been absolutely critical in delivering that side of the project. This is still very much ongoing, but I just wanted to tell you a little bit about the motivation for this and where we are now. So we know already that actin exists in G and F forms in the nucleus, so it can form filaments. There's more known about the G form than the F form. So G actin, we know, is transported into the nucleus, it's very dynamic, it's regulated by importin-9 and exportin-6, and in the nucleus it can control a number of different processes, but more classically it's been shown to assist in RNA Pol II dependent transcription. So essentially G-actin association with RNA Pol II can control that interaction with the DNA and control transcriptional regulation. There've been a number of studies over the last few years showing that nuclear F-actin is also important. It's again, not that clear quite what those structures look like compared to the cytoplasmic forms, but some have reported these short polymers, and that they can form essentially post-mitotically, and they might assist also with repairing DNA damage response after breaks. So we've been doing quite a lot of work in the lab looking at what fascin looks like in the nucleus and how it compares to the nuclear actin assembly and the F-actin assembly in particular. And you can see here endogenous fascin beautifully co-localizes with a probe that can detect the, or can generate nuclear actin filaments. So this is actually an actin probe tethered to a nuclear localization sequence. You can see here these really beautiful actin bundles in the nucleus. And we can also see these in live cells. So this is a nanobody that detects nuclear F-actin only. Post-mitotic cells are known to assemble F-actin, and we can also see that using this probe. So this movie will circulate again, but you'll see this cell divides, and immediately after division, you'll see the formation of these F-actin filaments in the nucleus that then disperse. And it's thought that that process is really important in controlling the compaction and the organization of chromatin during or leading into S phase. And we've demonstrated already that this process is required, and the actin bundling in the nucleus, it requires fascin. So if we knock it out, that doesn't happen, or it's much less efficient. And indeed, this post-mitotic assembly also requires fascin. And we've identified that fascin itself has a nuclear localization signal and a nuclear export signal, suggesting it is actively controlled in terms of the recruitment into and out of the nucleus. And what we've also shown using some nice nanobody tools that Jan Gessmann developed for us is that we can control the endogenous localization of fascin by expressing these nanobody tools that have got nuclear localization NLS sequence or nuclear export sequence on them. So the NLS, so essentially these are antibodies, but they're nanobodies, so they're expressible, they're much smaller, that we can express in cells that specifically bind to FASCN, and the NLS version will drag all of the FASCN into the nucleus, which you can see here, and on this western blot here, and the nuclear export sequence takes all the FASCN out of the nucleus and puts it in the cytoplasm, and you can see that also here. And these tools have been brilliant for us to start to ask questions about what happens if we force FASCN into the nucleus, which we predict would force a lot of effector assembly in the nucleus. And what we've seen is that this significantly reduces cell viability. So we can target and kill cancer cells if we promote the specific recruitment of FASCN and essentially maintain FASCN in the nucleus over time. And that's not the case if we exclude it from the nucleus using the NES, or if we just express the nanobody itself, or if we can target Cheri to the nucleus. So this is a very specific effect of targeting MORFASYN to the nucleus to promote this stable F-actin bundling events that we saw in the previous slide, and we think this is specifically able to promote and retain these F-actin bundles that allow cells— that don't allow cells to undergo normal division, and essentially they end up undergoing apoptosis. So the screens that we've been doing with AZ have allowed us to use small molecule libraries using quite an exciting high-content imaging platform that they have there to ask questions about nuclear fascin and nuclear actin and to target specific, identify specific targets that are essentially able to promote nuclear actin and nuclear fascin. And they will give us insight into which pathways are controlling the assembly these structures, and they might also give us some insight into whether these are drugs that we might be able to use in combination with other drugs to force nucleofascin specifically in tumour cells, in order to target these cells and to enable them to undergo apoptosis. Okay, so with that, I'll say a huge thank you to everybody. I think I've mentioned everybody as I've been going along, but huge thanks again to Karen Azir and Campbell, who've really been driving and have driven all of the fascinating work forward, all our brilliant collaborators, and all of our funding bodies who pay our large laser bills. And thank you very much to everybody and to you for your attention.

Thank you very much indeed. What we'll do, I think we'll just quickly get into the quiz, so I'll start sharing if I can get Here we go. Can people see that? We've got 5 people on already. See if we can get a few more. There's only 3 answers for each of these questions, so it's going to be quick on the draw for today if you want to win it. You've really got to be fast. Let's see if we can get a couple more people joining. Do you want to do a question while we're waiting, Alex?

You caught me on the hop there. I think the— wake up! Oh, my chat window just disappeared. So, well, do you want to— no, just do the quiz.

I've got the— and then we can— we've got 18 here. Let's just go for it. Okay. Right, okay, and let's see what we've got. So, quick on the draw to get the Foldscope: MEOS is useful for tracking pools of dynamic proteins, measuring pH changes, or ablating protein function. I think this one should be— I'll be disappointed if everyone doesn't get this one right. Oh man, so 9 correct winners, correct answers, excellent. Right, next question. Oh no, sorry, here's our leaderboard. Who's the winner? Bill, John, and Helena. Okay, well, it's all pretty neck and neck at the moment. Next question. You've got to be quick. FRAP enables— what does it enable? Measurement of direct protein-protein interactions, quantification of protein kinetics, or measurement of molecular conformation? Okay. Yep, 8 answers there. Right, right, let's see what happens. Let's see who's in the lead now. Bill is still there, followed closely by John and Frosty. All right, question coming up. FRET is a useful technique that relies on molecular spatial proximity of less than 10 nanometers, or photoconversion of acceptor molecules, or identical wavelengths of donor and acceptor molecules. 3, 2, 1, let's see who's answered this. And the correct answer is molecular spatial proximity. Here comes the leaderboard. See if Bill is still up there.

Oh, bang!

Maybe we have a new winner. John and Frosty. I think I know who John is. All right, next question. PIV can be used for measurement of direct protein-protein interactions, measurement of molecular conformation, or measurement of molecular flow. Your votes in now, and measurement of molecular flow is the answer there. Let's see where the leaderboard takes us now. John and Frosty are still neck and neck, but John, you still have a very slight lead, see if you can hold on, you better be quick. Question 5, last one. TC-SPC cannot be used for, so what can't it be used for? Analysing lifetimes of individual fluorophores, for non-invasive detection of functional phenotypes or identifying molecular interactions? What can't it be used for? Non-invasive detection of functional phenotypes. Okay, so for that, let's see where we end up at. Oh, Frosty and Homer are now fastest on that one, so Frosty, you are the winner. If you could let yourself identify yourself to the organisers, so just email the RMS or stick your name in the chat, you'll get a Foldscope through the post. All right, and with that, there you go, there's just a note in the chat there. Alex, I'm going to hand back to you for a moment.

So Gamal, I don't know whether still there, asked, is there a correlation between cilia function and actin organization?

That's a great question. We don't know because tumor cells don't really have cilia, but we are working with a collaborating group who've actually actually managed to demonstrate that FASID might be involved in that process. So I would say watch this space, but yeah, it's a very interesting question and I think it's got legs for sure.

Okay, actually Gamal asked another question, so I don't know whether you want us to ask it or fly just for speed. Do it? No?

Okay.

Philippodia is a selenium— are the BBS genes involved in forming the tip of the filopodium?

Yeah, so I think other people— we haven't shown that, but other people have shown if you— I think it's either BBS6 or BBS8 knockout cells have reduced filopodia and reduced migration, I think. But whether or not that's a direct relevance with the FASc interaction remains to be seen, but I think there is clearly a relationship between this, the formation of primary cilia and And then Sargon's asked 3 questions, so I don't know whether Sargon's there and just wants to ask them, or whether he's disappeared.

Still in the audience, that's for sure. Hopefully he's working hard. All right, in that case, have you looked at Mycenaean and the presence, absence of—

is that FASYN? Yeah, yeah, we haven't. So we did wonder if myosin might be one of the proteins that was shuttling FASYN around, but the way that we've done that is to use some myosin inhibitors or knock them out, and it doesn't seem to make any difference. So we don't think myosins are regulating it, but we haven't actually looked at myosin itself when we do the reverse, so we don't know.

If fasting controls myosin function. I'm going to jump to Jenny's question. Jenny, are you there? But it's quite— it's more of an essay than a question, so, you know, it might, uh, might be good if you ask it.

Hey Jenny, how are you? Hi, hi, hi! Lovely to hear that in a talk. Thank you. Because of all COVID things, I hope we'd have met by now, but it will happen. Yeah, I'm sorry, I had loads of of thoughts. So SMITH2, do you think it's the effect on the— any effect on the actin that comes with that, or do you think it's forming fascin directly?

It was kind of a first question. Yeah, it's a great question, and obviously they're quite hard to unpick. We were using SMITH2 at quite early time points. If you use it, we find if you use it for much longer, then you get, you know, the filopodia are massively affected. So we're only really looking quite quite early after SMFH2. We don't see at that time point a measurable effect on the filopodia, because we think it probably takes time to kick in. So I suspect this— that what we were measuring is much more to do with the fascin-formin interaction than it is to do with the actin itself. But yeah, it's very hard to unpick that without— what we didn't manage to do was to map the exact interactions between and make a mutant, for example, that would only, you know, potentially interfere with fascin binding and not interfere with formin's ability to build filopodia, so we weren't able to do that. So I don't know for sure that it's not just down to affecting actin.

And you answered the myosin, that was the other question I had. And then at this rate of movement, that— yes, so much faster than diffusion, is it?

I—

yes, it is. Yeah.

Yes.

And I mean, do you think it's to do with like tracks and, you know, the fact it kind of can jump, or, you know, that was what was occurring to me, this come kind of— I think of it as a glue, but I was sort of as a stable glue, but obviously it's not a stable glue, it's some kind of like—

Yeah, exactly. Yeah, I know we had all the same questions that we just thought, I don't know, I don't know if it is something specific to the fact that Phyllophagia is so confined and there's something weird with the flow in there, but if we just look at GFP, it doesn't look like that. So it's definitely not just some property of all molecules in Phyllophagia are doing this weird movement. So I don't know if it's something to do with the way that they, you know, if actin is somehow slightly torsionally controlled or something in Phyllophoria, and you might get some sort of effects on the binding partners or, yeah, I don't know, but it's clearly, it's definitely not static, it's jumping. We were trying to get enough resolution to understand if that jumping was actually periodic, so you were getting like precise, but it wasn't quite good enough for us to say with any certainty that it was anything to do the what would be expected to be the, the kind of folds of the F-actin bundles. But yeah, if we could do much higher resolution, I think we might get to that, but we don't know.

I mean, for comparisons, I wonder about DNA, like, okay, there's proteins that skate around DNA, like what—

yeah, yeah, yeah, no, you're absolutely right.

I think it is something very similar to that. Yeah, and your nanobodies would look fabulous. So So is that camelids, or what? I missed my—

I think I missed it. Yeah, so, so Jan Gessemann, who's in Belgium, he generates these as expressible fusion proteins. They are essentially camelids, but obviously he clones them then into an expressible lentiviral vector, and you can stick NLS and NES tags on them, or whatever else you want, and tag them with GFP, and they're absolutely brilliant. So that's an inducible system we've got there. So you express them under an inducible promoter, and that, you know, allows you move it around under doxycycline, for example.

Thank you. Thanks, Jenny. Okay, thank you. Has anyone else got any questions? If so, pop your camera on and speak up.

Oh, can I ask a question, please? Go ahead. So you were using GFP-RFP FRET pairs. Did you find— I mean, clearly they're working. Do you find they're a better choice than than the traditional CFP/YFP that, you know, was used a lot in the past?

Yeah, so that's a very good question. So the reason we didn't, when we started these, the problem was that, so CFP is not a good donor for lifetime analysis because it's multi-exponential decay, so it's a complete car crash when you come to actually analysing the data. So we've always avoided CFP, and it's just a bit rubbish anyway, but in the meantime, obviously, things like Teal and Turquoise have come out, and they are mono- exponential. And so they've become actually quite good. I think it's a better pair to have TIL and/or Venus or something like that than it is GFP-RFP. But because, you know, we've got all these constructs cloned into the GFP-RFP pair, that was our favorite one. GFP is still, I think, also the gold standard for doing time-correlated stuff.

So it's, yeah, it's easier to work with. Certainly you can really see the attraction of sticking with GFP. If you've already got a whole library of clones worked out.

Exactly, yeah. Great, great. Well, thanks for that. If there's no other questions, I thank you very much, Maddy. That was a brilliant talk. Yeah, that was fantastic. Yeah, no, fantastic.

And I think