 So it's my great pleasure today to introduce Radhika Subramanian from Massachusetts General Hospital, Department of Molecular Biology, and Harvard Med School Department of Genetics, where she's an assistant professor. And so Radhika and I overlapped in the Gellis lab at Brandeis, where she was a graduate student, and I was a postdoc and we were just reminiscing. So I think Radhika was the last graduate student Jeff had work on kinesin, which many of you may know was Jeff's initial claim to fame were his studies on kinesin motility. And she was also at the time Brandeis, I think Radhika and I both have very fond memories of Brandeis. It was kind of, in many ways, a scientific kind of utopia because there were, I think, three labs in our own building, Kausal. For those of you who are familiar with it, it was Liz Headstrom's lab, the Gellis lab, and Melissa Moore's lab. And so it was a pretty close knit community. And, you know, it was full of quirks. And one quirk we were saying is I think Radhika was probably the final student in the Gellis lab to record her data on VHS tapes, which for the graduate students in the audience are these pieces of plastic about this big that contain a magnetic piece of film between them. And you insert them into something called a VCR player video cassette recorder, and you can record movies off them. And that was originally how a lot of microscopy data was recorded. And I remember you could overlay the computer monitor with transparency film, and then to plot the motion and trajectories on the monitors off of the VHS recordings. But the other thing I remember is that I think Radhika and I both enjoyed while we were in the Gellis lab, giving Jeff's longtime staff scientist Larry a hard time. But it was done, it was done with a lot of affection. So after Radhika graduated with her PhD from Brandeis, she went to Rockefeller to do a postdoc in Tarun Kapoor's lab, which was wildly successful. I just learned that her bay mate was Emily Foley, who some of you may now know from her job as a SRO at the NIH, particularly if you have an R35 grant. But during that time, Radhika kind of continued her scientific trajectory by not focusing just on the motor proteins, but also thinking about the tracks that those motor proteins work on. And she's expanded that into her own lab, where she's studying really interesting biology and really interesting biochemistry at the intersection of motor proteins and the networks these motor proteins construct and deconstruct and move on inside cells. And with that, I'll let Radhika take it away. And thank you for agreeing to give the seminar. Yeah, thank you, Aaron, for the invitation. And I sort of feel like one of the best sort of parts of like, you know, science life is visiting different places and giving seminars and in the process, like meeting friends from graduate school and do the postdoc time and sort of reminiscing about all the all the good stuff and all the weird stuff and everything else. And I'm really like sad not to be there in person today, but I'm happy to to share some of the new some of some new stories from the lab that we are currently working on. So, okay, so I told my, I'm sorry, I'm having some issues with my pointer. I need to figure out what's going on. Okay. So I remember like telling my PhD advisor, John Gallus at the end of my PhD that I, you know, I was so done with my materials, I never wanted to see them again. And clearly, you know, never say never because here I am still working on on my critubules, but and exploring different aspects of my critubule biology. So the inspiration and, you know, for for all the work that we do, when that gets me excited about comes from these beautiful pictures from our cell biology textbooks. And everything green on this slide is microtubules. And what I what the take home message from these images is that, you know, microtubules organized into very different architectures for different cellular functions. So for example, in neurons, microtubules have very distinct architecture in axons and dendrites. Similarly, ciliated cells have microtubules that are organized into axonies. And I'll tell you a little bit more about that in a bit. And then very close to my heart in dividing cells, microtubules organized into what I think is one of the most beautiful like cellular machinery, the mitotic spindle. All of these structures are built from a common building block, which is the which is a microtubule polymer, which is a polymer of alpha and beta tubulin that is organized into a hollow cylinder. So the question that we are interested in is how do these nanometer sized building blocks, which are like tubulin and associated proteins that organize microtubules give rise to structures that are much, much larger in size. Or in other words, how do you go from nanometer sized proteins to micron sized arrays? How do you build these structures? And how do you destroy these structures? And what are the mechanisms underlying these processes? So in my lab, we specifically focus on two biological problems. So the first one is the spindle and anaphase that plays a role in directing the cell division machinery. And the second one is it's sort of a new direction that I started when I started my lab, which is ciliated cells and particularly primary cilium in the context of hedgehog signaling. So today's talk will be in two parts. I will mostly talk about this new story from my lab. That I would love to get feedback on where we are trying to understand the role of the cytoskeleton in a developmental process hedgehog signaling. And part two is sort of a, it started as a really fun project that now I'm very like, you know, it was just something we just, we just wanted to try and retried it. And now we are very excited about what we can do with AFM. So I would like to share some of the new data from that study with you. So starting with the first part of the talk, I'm going to be telling you about cilium-dependent hedgehog signaling and a step in that pathway that we are trying to work out. But before I talk about hedgehog signaling and microtubules and all of that, I want to take a step back and just, you know, go back to the last step of any signal transduction pathway, which is gene activation. And when I think of transcription and when I think of transcription factors, and I must admit I don't think about these things often enough, I think of the nucleus. And because I was giving a talk and Erin is in the audience, I even have slicing in there. So this is, you know, the view of transcription factor in the nucleus. But really there is a lot of fine regulation of transcription factors in the cytoplasm and some of the best understood mechanisms are those that involve the membranes, such as Golgi and ER based mechanisms, where transcription factors are held and then activated and released to enter the nucleus at the right time in any given process. But in addition to these membrane, membrane scaffolds that regulate transcription factors, the other major scaffold in a cell is, of course, the cytoplasm, cytoskeleton, which form these long polymeric structures that can act as platforms for different cellular processes. There are a few examples now of how transcription factors are localized to cytoskeletal proteins. But the link between cytoskeleton and transcription factor regulation is very poorly understood at a mechanistic level. And so I got interested in this question in the context of the primary psyllium, because in the last two decades, it has become very clear that this tiny cellular organelle, so this organelle is less than a micron in diameter and just a few microns long, is turning out to be a major hub of cellular signaling. And of course, the backbone of this structure is made up of a microtubule array known as the axonine. And it turns out that this small microtubule based organelle is absolutely critical for hedgehog signaling invertebrates. So what is the hedgehog signaling pathway? I'm just going to give you like a very, very broad and brief introduction. So it's a major developmental pathway. It plays essential roles in tissue patterning during embryo development. And it also has mitogenic roles. This includes proliferation during adult tissue homeostasis. And as you can imagine, errors in the signaling pathway lead to multiple disorders that are developmental, but they're also associated with numerous cancers. So of course, I got interested in this pathway because of its dependency on the cellular and trying to understand what is the link between hedgehog signaling and the microtubule cytoskeleton per se. So I'm just going to tell you about one step of the pathway. This is the very last step of the pathway. So as I mentioned, the last step of any signal transduction is of course gene expression. And the major effector of the hedgehog pathway is this transcription factor called glee. It is glioma associated on the gene. And when the hedgehog pathway is off, glee is localized to the base of the cellium when it is turned off. When the pathway turns on, glee translocates to the very tip of the cellium. And there are processes here that we don't understand. It's a little bit of a black box, but glee is activated at the tip of the cellium and then it travels back out of the cellium and then it enters the nucleus where it transcribes gene. So there are these very distinct microtubule localizations of this transcription factor leading to the question of what is this role of the cytoskeleton in glee regulation? Is it a placeholder? If it is a placeholder, how is it forming the placeholder? And is there more to it than just being a spot where glee lands for other downstream functions? So we just wanted to begin to understand the links here. And the entry point for us was acillary kinesin KIP7. So the discovery of this kinesin in 2009 by three different labs actually was very exciting because finally there was a molecular link between the hedgehog effector proteins and the microtubule-based cellium. So at the molecular level, this seemed to be a good starting point to start thinking about the roles of the cytoskeleton in hedgehog signaling. So KIP7 is a conserved protein in the hedgehog pathway. And what we know about KIP7 is that, as you can see in this immunofluorescence image, when the pathway is activated by the small molecule called SAG in our tissue culture experiments, glee, the transcription factor localizes to the very tip of the cellium and KIP7 also localizes to the same place. In the absence and in vitro experiments from my lab that we published last year shows that KIP7 is a not-mortile kinesin. So when the kinesin was discovered as being part of this pathway, it was the first sort of model or the hypothesis was, oh, it must be a transport protein that carries the transcription factor to the tip of the cellium. And so we did some in vitro experiments and we showed that it is a non-mortile kinesin, but it can track the ends of a growing microtubule. So in the image below, I hope you can see the green comet decorating the end of a microtubule. And it turns out that, and we worked out the mechanism of this in this paper which shows that KIP7 has increased affinity for the GTP form of tubulin at microtubule ends. So what happens when KIP7 is missing from cells? So what we find is that when KIP7 is absent, glee no longer properly localizes to the very tip of the cilia, but instead is localized all along the axoneme in these puncta. So it leads to mislocalization of glee too, and that seems to affect both pathway activation and pathway repression in terms of transcription readouts. And finally, there were experiments published back in 2009 where multiple groups had also shown that perhaps glee and KIP7 directly bind each other based on pull-down experiment from tissue lipids. And we wanted to take a deeper look into this question of, is this kinesin really a transcription factor binding protein? And if so, how does it bind the transcription factor? Because there are not a lot of known examples of this cargo motor interaction. So this is the work of a postdoc in the lab, Farah Huck, and a postback student who has now moved on to graduate school, Christian, who together worked on this question of understanding how these two proteins bind each other. So both of these proteins are very large proteins. And so we had our first technical problem when we tried to express and purify them. So we decided to take a step back and just look at domain-domain interactions by doing co-principitations by over expression of these domains in XB293 cells. And we got our first surprise when we went on the glee side when we found that the zinc finger domain, now this is the DNA binding domain of the transcription factor is the one that is involved in binding the kinesin. On the kinesin end, it turns out that the first coil-coil domain, so this is the dimerization domain, the kinesins have a motor domain connected to a small neck linker. And these two motor domains get, two motor domains get dimerized by a small coil-coil domain. And this is followed by a large cargo binding domain. And it turns out that it's this very small coil-coil domain made up of about 40 amino acids that forms the glee binding site in this complex. So we wanted to see what kind of a complex they form and whether we could actually purify the complex. And so we conducted some binding assays where we found that the zinc finger of glee forms a very tight complex with KIP7 coil-coil domain with a KD of about 50 nanomolar. We are able to isolate these complexes and we can also determine the stoichiometry and what we learned was that one KIP7 coil-coil dimer binds to one molecule of the glee zinc finger. Now glee zinc finger is made of like five zinc finger repeats and I'll show you a structure of that. So basically the five zinc finger domains in this tandem repeat bind one coil-coil dimer. So how does this actually work? So we we were fortunate to have a structure of KIP7 bound to sorry glee bound to DNA from Nicola Palmlett to just work and essentially glee the glee zinc finger is now every every every helix in this in this pink structure that you see on the screen is part of a zinc finger and there are five of these and the zinc fingers clasp around the double stranded DNA sort of holding the DNA in this conformation that's seen in the structure. And then we look when we model the KIP7 coil-coil based on numerous coil-coil structures that are available in the structure database and and you know it didn't seem unreasonable that the size and the shape of a coil-coil of this lens is is quite similar to double stranded DNA. So then we we looked into this a little bit more carefully and we looked at the electrostatic of the between DNA and glee that formed the interaction and it turns out that that glee has a very positively charged C-shaped structure that then clasps a very negatively charged DNA molecule and so we looked at the electrostatic surface of the KIP7 coil-coil and it turns out that it is a highly negatively charged rod-like structure. So in principle it it essentially looks like a piece of DNA from a charge and size and shape perspective. So we then went ahead and modeled the complex and when we when we did that we our our best fit model was the one where essentially the zinc finger clasps around a double coil-coil the same way as it clasps around a double stranded DNA in in the structure of the transcription factor with DNA and in this slide I've shown a comparison of the two two two structures. Okay so so this is a structural modeling work so we wanted to then go ahead and test this model more rigorously and I'm just going to show you two two things that we did. So one of them was extensive site directed mutagenesis that validates the model but also helps us to to definitively place the zinc finger on the coil-coil. The coil-coil is highly negatively charged throughout but we find that it is a very specific region that the complex formation occurs and the other thing we learned from this was that the residues in the zinc finger that are involved in DNA binding are also required for kip-7 binding so if we mutate those specific residues then the interaction is lost. So next we wanted to directly test if there is a competition between DNA and the kip-7 coil-coil so or in other words do these two molecules bind the same site on the GLE2 zinc finger. So this is a competition I'd say black is the curve binding curve between kip-7 and GLE in the absence of DNA when we add in DNA the entire curve shifts to the right and we only get binding once there is an excessive GLE. So this suggests that there is mutually exclusive binding of the kip-7 coil-coil and DNA to the GLE2 zinc finger. So together this suggests that this mode of interaction between a kinesin and a transcription factor falls into the small category known as DNA structural mimicry which is a mechanism that has been recognized as a way to regulate DNA binding proteins that are about 30 or 40 known so far or reported so far but most of these are prevalent in bacteria and viruses and they as you can imagine they play very interesting roles in evading host cell defense and on the on the screen is like one of my like favorite structures which is a protein from mycobacterium that pretends to be a DNA and inhibits a host DNA gyrase. In eukaryotes there are very very few examples and they're typically nuclear localized and so we think this is an example of DNA mimicry uniquely like the unique aspect of it is that it is a eukaryotic DNA mimicry mechanism but that it happens in the cytoplasm and it is a way of tethering perhaps a transcription factor to the cytoskeleton in this case. So this is you know the first part of this work led us to this sort of insight into the mechanism by which these two proteins bind each other but there was a question of you know what what what do what do these proteins like can kip seven actually recruit glee to micro tubules so we wanted to test that and we used a turf microscopy based assay to do these experiments so we immobilized micro tubules on a glass surface it's fluorescently labeled and yes i eventually graduated doing fluorescence work in my postdoc we label the kip seven with gfp and we we have the zinc finger from the transcription factor labeled with an alexa dye so we can visualize all three components in the system and what we see here in this slide is you can see the micro tubule channel and i'm showing you the glee channel in the absence and presence of kip seven and as expected you can see that kip seven is is is is recruited to the micro tubules by sorry so the transcription factor is recruited to the micro tubules by the kinesin so this is not unexpected but it was nice to like see that this this work that expected but the surprise came when we actually looked at the kinesin and here now it's the same experiment but now we are looking at the kinesin instead of the transcription factor and as you can see even in the absence of glee there is kip seven on the micro tubules but when we add glee when we add the transcription factor there is now an increase in the amount of kip seven on micro tubules so what this says is that glee is not a passive cargo it's not just that the kinesin sits on the micro tubule and recruits glee the glee actually positively regulates the kinesin micro tubule interaction and you can imagine that this will set up a positive feedback loop to concentrate both the transcription factor and the kinesin on micro tubules as as would be beneficial for dynamically relocalizing these proteins to a particular region of of a cytoskeletal array so we then looked at how this recruit how this binding of kip seven changes with the concentration of glee so now this is an experiment where we hold the concentration of kip seven constant and we look at how much kip seven is on micro tubules with increasing amounts of glee and you can see there's a dose dependent response that eventually saturates so what this suggests is that the amount of kip seven on micro tubules depends on the amount of glee and so this sets up a loop where there is now basically kip seven is sensitive to the concentration of glee in solution and and and the the readout is that both kip seven and glee two concentrations can increase on micro tubules in a dose dependent manner sensitive to glee concentrations so so how did this happen how is it that the information from this coil coil domain that is far away from the motor domain transmitted to the motor domain itself so how did the binding to the coil coil change the motor micro tubule interaction so this was a puzzle that took a really long time to solve and the experiments here are like little complicated so i'm just going to give you the answer here so the first thing we checked was perhaps the glee binds the motor domain itself and the answer was no in solution there was no interaction between the motor domain and and and glee the coil coils bound glee when we have now a a dimeric construct containing both the coil coil and the motor domains what we found was that there was not one but two molecules of glee so this was very confusing so one site was where the coil coil is and then there is a second site and we didn't know where it was going to be and it turns out that it is close to the motor site and it's it's formed by both the motor domains so what we found was that if we now take motors so to test this what we did was we took the motor domains and we concentrated them on micro tubules and now the same motor domain that was unable to bind glee in solution can recruit the transcription factor so i'm happy to to talk more about this later but for now i just want to give you what what we think happens which is that in fact kip seven has two sites for glee binding one is at the coil coil and and one is at the form by the motor domains of kip seven so in order to be certain about this and and and really try and make sure that glee does touch the motor domain of kip seven we collaborated with Ron Milligan's lab at Scripps and we solved the cryo en structure of kip seven bound to micro tubules in the presence of glee and what we see is indeed there is now we can clearly see glee bound to a motor we cannot resolve the second interaction site on the second motor but we can we can we find that there are freezing fingers that can be fit into this density so uh so we we had now you know so far what we had learned was that there were two sites on kip seven for glee binding so there is the coil coil site and then there's the motor domain site and we wanted to look at what the charge charge surface of these sites looked like and as you can see here these are both highly negatively charged surface and the kip seven is involved in hedgehog signaling and glee binding and we wanted to ask you know is this you know where we see this with every kinesin or you know is this somewhat specialized and to to just begin to get at that question we we now looked at a kinesin that is a ciliary kinesin a closed homolog of kip seven which is in fact thought to arise from a gene duplication event and and there are two kinesins one is kip seven the other is kip 27 kip 27 does not bind glee and it is not involved in hedgehog signaling and when we look at the the structural models for kip 27 coil coil and motor domain we find that the charge surface electrostatic surface charge is is completely different from kip seven even though they're both ciliary kinesins and very closely related so we think that the glee binding sites are specialized on kip seven but this also you know so you know it was it was interesting to find the two sites but then there was the question of there is one site that's the coil coil and the other site which is at the motor domain which of these sites plays a role in this graded response of kip seven binding in the presence of glee so how so which of these binding interactions is important to change the kip seven microtubule binding so now that we had these two proteins we could we could we could play some chimera games and we could do some domain swapping and that's what we did so in this experiment we took the coil coil from kip 27 and we replaced it in the kip seven protein so that we now have a motor that's kip seven and a coil coil that's kip 27 and this is the the data wild type protein data so again as you see as they increase the glee concentration the amount of kip seven on microtubule changes so plotted on the y-axis is the kip seven gfp intensity on microtubules and with the chimera we saw something that initially sort of surprised us and then we worked out what's going on but essentially what we see is that if the kip seven coil coil domain is missing then the protein is constitutively strongly bound to microtubules and it can no longer be regulated by glee so it becomes glee independent it can still recruit glee but the amount of binding is independent of the transcription factor so just to summarize this part of the talk we think there is a there are these these two binding sites on the kinesin together set up a system that allows for synergistic accumulation of both kip seven and glee in response to to glee level so it's sort of you know and can think of it as some kind of a rheostat like system where the amount of glee the kinesin is sensitive to the amount of glee and responds to glee concentrations which in turn then changes how much transcription factor is on the microtubules so this was all in vitro and we we we wanted to ask the question is the synergistic accumulation of glee to glee and kip seven on microtubules relevant in cells and also in the context full length proteins because all the vitro work was done with you know smaller constructs that were more amenable to the biochemistry so this is now a very simple heterologous overexpression experiment in in non-celliated cell so the top panel here is a cell in which we've transfected glee alone with a neon green tag and you can see when you overexpress just the transcription factor it is mostly localized to the nuclear when you overexpress the kinesin it is it localizes in this perinuclear fashion in the in the cytoplasm but when we co-express both the proteins we see a very dramatic change and both of these proteins now localize strongly to microtubules and glee is out of the nucleus so this says that the synergistic accumulation of both proteins on the on the cytoskeleton is is is something that can that is like seen with full length protein in the cellular context but that was a you know heterologous overexpression system and and what the real question was does this matter for the ciliary localization so just to like take a step back what I had shown you in the introduction was was this data where I had shown you that in wild type cells in the presence of the hedgehog activator both kip7 and glee are at the tip of the cilia so that's the that that's color coded in red here and green is the protein signal in the absence of kip7 glee is no longer at the tip of the cilia but localized all along the actually so what we wanted to ask was but what our data told us was that this doesn't go one way this goes the other way too and it told us that if you if you don't have glee then our data predicts that kip7 would also not be at the tips of the cilia so in order to to test this we were very fortunate because Rob Lipinski's group had made these cell lines that have the different isoparms of glee knocked out and what I'm showing you here are immunofluorescence images where we did the immunofluorescence with directly labeled kip7 antibody to to try to do this as quantitatively as possible and in pink is kip7 and in green is the axonene and those are the two channels to really focus on and you can see that when when there is when in the glee two null lines there is less kip7 and in the glee to glee three null lines there is basically no kip7 at the cilia tip and this is also something that we we can show by quantification that this is indeed the case so what this tells us is that this idea that the microtubule cytoskeleton is a platform and the kinesin on the microtubule is a site form sort of a placeholder for recruiting the transcription factor is not entirely true the transcription factor in turn actually regulates the cytoskeleton so there is a there is a there's a feedback between the transcription machinery and the cytoskeletal localization of the kinesin and the feedback loop is perhaps it makes intuitive sense to me in the context of a signal transduction pathway where you want to very finely regulate the amount of proteins in a particular site at a particular time so we so this was just like something we we were we were very surprised by our heterologous expression experiment and and how how well it worked and we said okay so that was with the full length protein but can this small coil coil domain of kip7 as I said it's about like 45 amino acids or so can it be used as a tool to regulate the nuclear localization of glee and why might one want to do that one it might be a we we are thinking of using this as a cell biological tool to move glee around and one can imagine like doing different types of experiments to tether glee to to outside of the nucleus wherever one wants to and the other context and isn't isn't the cancer context where what is seen is that in a lot of different cancer types there is over expression of glee and some of these happen through hedgehog pathway mediated ways and some of them are hedgehog independent so a direct way to individually would be advantageous so we just we were just curious if this like small peptide would would even work so this is just a proof of concept experiment and this was essentially the experiment we over expressed glee in the with the with the fluorescent tag and then we took this small coil coil and we put an er tag on it and we asked can this small construct sequester glee in the cytoplasm or can it like bring it basically prevented from entering the nucleus and to our surprise the answer maybe it's not a surprise and retrospect but we were very excited when this actually worked so what I'm showing you here are images on the top panel shows glee over expression alone and as you can see it's in the nucleus but then when we add this small coil coil domain tether to the er glee is now in the cytoplasm and this can be quantized as well so to summarize this talk we we started this project by asking you know can we learn something about how a kinesis interacts with the transcription factor and the mechanism by which the the phytoskeleton acts as a as a tether in the cytoplasm for recruitment of transcription factor and and regulation and we stumbled upon some unexpected findings that we had not anticipated the first one was the mode of interaction it was it was interesting to see that glee uses the same mode of mechanism that it uses to bind DNA to to bind the bind the kinesin and in some ways the kinesin just masquerades as a piece of DNA in the cytoplasm to hold it in the cytoplasm unless until it's properly activated but it's not as simple as just forming the tether what we what we see here is that the transcription factor actually is not just a passive cargo but it regulates the amount of kinesin on the microtubules so there's a feedback loop between the transcription factor concentration and and and its localization to microtubules by changing the k7 microtubule interaction and finally this was sort of like a fun experiment that we did but we are kind of excited about it now where we want to see if we can build a tool to to to to move and and sequester the the transcription factor away from the nucleus whenever we want to in order to answer questions pertaining to the cell biology of hedgehog signaling and and perhaps this this will also be useful as a as a way to control glee levels in cancer cells so that's the first story that I wanted to talk to you about and in the last 10 minutes or so I will switch gears and I will tell you about a direction that I never like anticipated I would like go in when I started my lab but it's uh it's been a very fun uh fun process and a very fun project so I would love to share it with you so you know in in my lab so one of the questions that I'm really interested in is how do you build these complex microtubule arrays and how are these complex arrays remodeled um and you know when we think of dynamics of multi microtubule arrays they they fall into sort of two broad categories so the first is sort of very large scale remodeling and disassembly where so for example every cell division the spindle has to be disassembled and rebuild rebuilds and similarly the axoneme is disassembled and rebuilt every cell cycle so these are very dramatic large scale events and at the same time there are dynamics on much finer scale to control the precise size and geometry of these structures and you know one reaction that plays a role in both of these types of remodeling events is microtubule depolymerization so this is a reaction where tubulin is lost from the ends of microtubules and here you see a classic like textbook image of peeling microtubules from the ends of a ends of a polymerized microtubule that leads to microtubule shortening and so one of the questions in the field is basically how is it that the same reaction which is the loss of tubulin gives rise to these two very different outcomes and it's been anything to study with an array of microtubules has just been very complicated because of a technical challenge that I'll illustrate here so on this slide on the left is a field of single microtubules we've added a depolymerase and when we record this depolymerization in real time you can very nicely see every single microtubule undergoing depolymerization in on the field now on the right is the same depolymerase but we've added a cross linker so now we have a cross-linked microtubule bundle and I hope you can see that the intensity across the bundle uniformly drops but there is no way to know where the ends of each microtubule is and how microtubules are bundled relative to each other so when it comes to a microtubule array we don't have a good way of looking at remodeling or destabilization in real time at the single microtubule resolution now similarly on another scale the same problem exists so now if we zoom into a single microtubule there are all these protofilaments in the microtubule so when the microtubule is depolymerizing what is happening to every protofilament in real time is something that is not possible to to image because of the resolution restrictions of uh life microscopy so a postdoc in my lab Sathara Vijaratne thought well maybe we can use AFM to bridge this resolution gap between light and electron microscopy and perhaps try to to to image bundles of microtubules so very briefly AFM is it's a very simple technique in concept it's simply a cantilever with a very sharp tip that is tapped along a surface and any obstruction in the form of a sample leads to a deflection of the tip and and the deflection can be read out giving a very fine detail of the surface topography so the the z axis resolution in this method is actually much better than the x and y so in this way also it complements some of the other methods we use in the lab so this is just a single microtubule biotomic force microscopy you can see actually like see all the protofilaments here as there are different types from the mica surface on which the sample was here it was a very exciting day for us because it took a while to get here but this is not what we really wanted to do what we really wanted to do was look at how bundles are organized and can we really look at every microtubule in a bundle by this method so we used a microtubule crosslinking protein called prc1 it's an anti-parallel microtubule crosslinking protein that has roles in spindle organization and here is an AFM image of a prc1 crosslinked array and you can probably see you know so every rod here is a single microtubule so it's it was it would be we could very nicely image microtubule arrays by AFM but this is a static image and that is not what we wanted to look at and so as our model system we went to we decided to start the deep polymerase and we asked can we look at the dynamics of individual microtubules in this array in real time when we add deep polymerases and the deep polymerase we mostly chose to work with is this kinesin 13 called mkac and the only thing I would like to point out about this is this is a fast deep polymerase it can deep polymerize microtubules from both ends and the and to just to have a comparison we we contrasted its activity with a different kinesin which is a plus and directed motor that walks to the ends of microtubules and depolymerizes microtubules from the ends one end of the microtubules so so these are the two kinesins we deep polymerizing kinesins we work with but I will mostly focus on the kinesin 13 in the interest of time so this is the first image that we acquired of a prc1 crosslink bundle after adding mkac and so what you can see here every striped here every large bun every every rod here is a microtubule and this is color coded by height and if you look at the regions that are highlighted in the white rectangles you can see that you can see some stripedness and these stripy features actually come from exposed protofillaments of partially deep polymerized microtubules so what we and because there are different heights from the surface and now we can we we were excited because we felt okay we could probably like see individual protofillaments within a microtubule within a larger array this is now a 3d rendition of the same data and as I play the movie I hope you can see it would be easier if you focus on the the pink arrows on the slide where you can see deep polymerization from the ends of microtubules and you can see individual protofillaments being lost and what we learned from this experiment was actually that the loss of protofillaments from a filament end is asynchronous and the difference protofillaments are actually lost at different rates with this deep polymerase and this was interesting because nearly 20 years ago using artificial substrates people suggested that that is that m cat might be able to use and recognize individual protofillaments as defective substrates but there was no way to actually test that in the context of a microtubule but this technique now gives us a a tool to do that we can also see propagation of lattice defect these are now defected in the middle of the microtubule lattice and we find that this enzyme can actually propagate these defects and you can again see the striping and protofillament based level deep polymerization of the defects so now I just want to contrast this with this other kinesin kinesin 8 and I just will quickly summarize the the main findings which is that we do not see the stripy deep polymerization we do not see any defect propagation and we and we also see different differences in how it responds to prc1 crosslink but I am not going to talk about that but mostly I would like to to emphasize the difference in the overall difference in deep polymerization from the ends of microtubule and the middle of microtubule suggesting that single protofillaments are probably not effective substrates for kip3p and kip3p prefers to deep polymerize from the ends it does not propagate defects so as you can imagine proteins that are like m-casks are probably better suited to be large-scale remodellers whereas those like kip3 which act at the end and do not cause destruction from the middle of the lattice are better suited to act as fine tuners of microtubule length so we also wanted to ask what happens in the context of a different microtubule array so how these these experiments pointed to properties of the two kinesins being different at the level of the protofillament structural dynamics so how does that translate to array remodelling of a completely different array and here we decided to look at axonemes the axonemes are made of doublet microtubules and that is shown in the bottom right hand side of the slide where there's a complete tubule this is a side view where you see a complete tubule and which is the atubule and a half tubule which is called the b-tubule and the and the junction between the a and b tubule is made up of a different protein it's a non-tubule and junction on the on the top so these doublet microtubules organize into this larger cylindrical array of nine microtubules nine doublet microtubules in the at the circumference so we wanted to ask you know can we image these by AFM and axonemes look absolutely boring by AFM they are featureless because they're quite tall for AFM so that wasn't particularly interesting but doublet microtubules have been imaged by AFM before and we were able to find ways in which we could clearly distinguish the a and the b tubule when when they were on a micro surface so again we just like added the deep polymerases and we asked what happens so in order to do this experiment we found that the best way to do it was to partially dissociate an axonene so that we formed an axonemal array this axonemal array is composed of doublet microtubules that are linked so if you'd on the right hand side the green and the pink in the zoomed in view highlight the doublets within each doublet and then they're forming a larger array and when we have when we add mkac i hope you can see that mkac deep polymerizes doublets and one tubule goes away faster than the other and we think it's the b tubule that goes away first now this observation was exciting for a couple of different reasons to us the first was that it was known that you know b tubules are less stable to things like high salts or you know some very chemical and mechanical perturbations but enzymatic depolymerization of doublets has never been looked at and it was not even clear if these enzymes would these deep polymerize because they are some of the most stable microtubule structures that are seen in the cell so we were we were excited to find that in fact it destabilizes microtubules and there's a difference in the rate at which b tubules are depolymerized with respect to a tubules which again points to interesting ways in which the size of different tubules in an axonene could be differentially regulated by enzymes so but what happens in the context of an axonene when there are there are linked doublets so shown on the right it's a schematic so you know in an axonene there are linked doublets and we asked okay if b tubules go away first then how is the axonemal stability impacted and so on the left is a movie that i'm going to play and as i said axonemes are not pretty by afm but what we see is sort of this very dramatic unfurling of the entire structure and very rapid sort of it just falls apart and we think it's because once the b tubules start to depolymerize there is a loss in the linkages that hold the structure together it's sort of like cutting the string that holds the bouquet together and the whole thing falls apart and then the context of thinking about you know how a structure like the axonene composed of lots of proteins that are like very very stabilized doublet microtubules how you know in terms of destabilizing such a structure if we think of depolymerization from the end slowly looting like different microtubules it seems ineffective so i think mechanisms like this might play a role in destabilizing the structure enough to then accelerate the way in which the structure falls apart so in conclusion we we we wanted to find a way to look at microtubule destabilization of arrays and we essentially wanted a way to look at individual microtubules within arrays and we we asked the question uh you know how do how how are these arrays remodeled in different ways uh can we visualize them and can we learn something about the mechanism underlying these processes and we find that depolymerizes that are as such these quite similar they're both like kinesin depolymerizes they undergo two different they can lead to very different outcomes and we think that this depends on their activity at the protofilament level so an enzyme that can effectively recognize and depolymerize individual protofilaments has a greater ability to disrupt the structure in a in in a more drastic way whereas proteins that don't have that activity but have restricted activity to the ends are better suited to be length regulators and we can get insights into how how such sort of dichotomy is is built into array remodeling by by by imaging them so i think i summarized all of this so i will just stop here so i would like to thank our collaborators funding sources but most importantly i would like to thank my lab for uh and all the people who were brave enough to join a new lab now a few years ago and these are like the first sort of stories from the lab so i'm very excited to share them with you so thank you and i'll take questions great thank you radhika that was an amazing talk so um i'm clapping so people can can post questions in the in the chat and um if you'd like to i can call on you to um to speak as you go a question um i guess i can start off it's like oh i guess jill has a question i see her hand so jill do you want to go sure that that was a fantastic talk i loved it so i had a the glee story is really interesting and the positions where glee binds are also really striking as well and given that these are not really places where cargoes typically bind to kinesin motors i was wondering how glee binding with the implications and might be for glee binding to kinesin when it might be kif seven we might be attached to cargoes does that make sense yeah so so uh i i think sorry sorry i never mind these are not how about if i don't say cargoes but when they might be um bound to other proteins yeah so um so you're talking about when the kinesin is bound to other proteins and what are the implications so i i think i think there is more going on with kif seven at the sea terminus domain so there's actually a paper that suggests that there is another cargo i i don't know whether to call it even a cargo but um it's a there's another protein interaction site at the sea terminus and that is thought to be important for for kinesin entry into the cilia so so there is there is there's more than one binding partner and it remains to be seen how these different binding interactions all result in activation of the kinesin entry and glee binding and all of that we don't know yet so i had uh questions so i so i notice you had a kif seven double knockout cell line and so i was wondering what the the phenotype was then for signaling um so clearly like the cells are viable but is there what's the effect on the hedgehog signaling pathway so the kif seven knockout has hedgehog signaling defects so in terms of the knockout mice uh they it's not embryonic lethal so the but the the i guess the pups are sick and they die in a few weeks so they have developmental defects that are all connected to hedgehog signaling defects and then the like the flip side the question is um do all cells that have a glee seven have like an axon or a um like cilia like some sort of comparable cellular structure yeah so that so so whether there are more functions of kif seven that are not hedgehog or cilia related it's sort of an open question so there is there is a paper there's exactly one paper uh that talks about it and in in in some kind of a non-ciliated lung cell they found that there are proliferation defects when kip seven is absent but that's the extent to which we understand anything about that so we don't know the kip seven uh it's also a cilia genesis factor so if you don't have kip seven the the cilia are found to be hyper elongated and that is consistent with some of our in vitro work suggesting that in addition to binding glee and all of that it also regulates microtubule dynamics and it slows down microtubule growth so it has multiple functions during cilia genesis and um but the predominant phenotype in vivo in in the mouse model were these hedgehog related datasets great i think scott has a question hello i was i really enjoyed it i just had a question maybe more about the the glee biology or the glee activation mechanism so it has to get transported from the base to the tip of the cilium in response to signal and then something happens and it goes on to activate transcription what exactly happens and um or if it's known it might not be known but what i wondered was like it seems like one of those initially kind of like weird activation mechanisms but i assume it takes some amount of time when a hedgehog signal is applied to get it to the tip so is it like the length of the cilium ends up kind of setting like a threshold for how much sustained signal is needed to get it to the tip and get it activated or is there anything like that so that's a really good question and there isn't a very simple answer but i'll do my best so the you know so what sets the time so it is found that it takes about three to four hours for glee to be maximally at the tip of the cilia and so so what happens at the end and this is but you know there's an increase and then it maximizes that about that time and it correlates with the amount of time it takes to fully defuzz for a late glee which is the multiple things happen glee activation is not one step there is defuzz for relation of glee glee is at the tip of the cilia it has a repressor called suppressor a few that needs to fall off keep the transcription factor and then i completely ignored the membrane but there are proteins bound on the membrane like there is receptor based activation mechanisms that that that have to do with ensuring so the full length unproteolized defuzz for related form of glee is thought to be the fully activated form and when the protein is either proteolytically degraded or it can be degraded to a repressor form that forms the repressor at the base of the cilia so there is a lot of things that need to happen like the ubiquitination system has to be suppressed to prevent degradation and the you know the activation by loss of sufu needs to occur so so but i i don't think the length of the cilia sets the sets the threshold because it's not even clear if it's actively you know transported or if it's more like a diffusion and capture because kine the kip seven doesn't walk uh on microtubules and the cilia is quite short so i i find the you know motility model a little bit difficult to understand or uh so so i but but what is thought what is speculated is that the volume of the cilia is what does the trick so you were taking very low copy there's so low copy number that it's kind of a pain to do any like cellular imaging taking these very low copy number proteins and like putting it and concentrating it in this very small volume and this concentration of all the factors at the same time in this concentrated volume is what results in um completely activation but it yeah it's there there's a lot of question marks there i think uh judith had a question yeah um yeah so i also really loved your talk it was really gorgeous and i had tons of questions but um so am i right in thinking that kip seven is not in the cilium until signaling did you say that and then it so i i didn't go over that in detail but the answer is that kip seven is at low levels and its level increases when the pathway turns on so it so it must be somewhere else that's the deduction i i drew in my mind and does that mean is it possible it's acting as a dna mimic for other transcription factors yeah so we are we are very curious about you know because if it's binding a zinc finger domain then it might be able to bind other tandem zinc finger repeats and i don't have an answer to your question so in a kip seven proteomic study there were a couple of other zinc finger proteins that were reported we don't we'll have to verify that biochemically and we're also thinking that it might be fun to just see if we can do some kind of across linking pull down to see you know what if you just have the small domain and over express it what are the different proteins that it can pull along with it um so yeah and also the question and reverse which is that you know are there other chimesans or coil coil proteins out there that do do this and is this uh because the cytoskeleton is full of coil coil domains so that's sort of an intriguing idea and i i don't i don't know the answer to that yet i'll let someone else ask a question i have i have tons of questions but i gather we're having cocktails so i can ask yeah so does anyone else have a question else well i'll i'll ask one more than if someone thinks of one then then they can jump in so so a zinc finger domains can also bind RNA and your transcription is great at all but RNA binding and transport could be could be where it's at um so so you know as a lot of RNA binding proteins contain tandem nucleic acid binding motifs and so i could easily envision some of these interacting with a kinesin and the others interacting with their cognate RNA and so i i just wanted to throw that idea out there that yes yes aren't we haven't ignored our day so i i've been thinking about it but i i don't i don't have you know any any data or anything but you know i was thinking about it like particularly in the context of things like neuronal transport whether it's like RNA is a cargo and you know we always think of cargo and cargo adapters and such but but perhaps that's not needed maybe you know the kinesins the kinesins themselves act as a scaffold for um so we we find some coils so we've actually now looked at a lot of different coil coils and then there are coil coils that have a very positively charged surface so i also feel like you know um so so you know one idea is that the RNA binding proteins could bind these protein domains but it could also be i i wonder if there could be direct like nucleic acid binding to the protein like i had no idea but can i ask one more question nobody else's is coming so i was really curious about the activation mechanism of glee and you say it has to go up to the tip and be concentrated and presumably it's you know for the act whatever this activation mechanism it then presumably binds there and is concentrated there so how is it released yeah that is that is a really great question for which i have no great answer but there has to be a way in which this this interaction has to be broken and so we are now doing some experiments where we are building this reconstitution so one idea is that there is a there's a suppressor that binds glee in general so that's suppressor of fuse and there is the three protein complex and it's possible that the that there is some multi-protein interaction that changes the kd of the kinesins interaction with the transcription factor the other thing that happens is the dephosphorylation and a lot of the phosphorylation sites are flanking the kinesin binding sites on glee so i think the sort of the two next steps is to to add the other major suppressor of the pathway and then the third and to add the kinase and look at how those change the interaction affinities to get it some idea of release mechanism it's very cool thank you thanks for a great seminar thank you so so why don't we end there so so rodic has a few minutes to relax before the meeting with the students so let's all thank rodic for the great talk