Scientist Stories: Aashish Manglik, Transmembrane Signaling Processes & Protein Engineering

All right, hello everyone and welcome to today's Cryo-EM Current Practices webinar. You're joining the 3 national centers established by the NIH Common Fund Transformative High Resolution Cryo-Electron Microscopy program. My name is Christina Zemanie. I'm a scientist and training liaison at the National Center for Cryo-EM Access and Training, NCCAT, and today I'm also your webinar moderator. I'm joined today by my colleague here at NCAT, Ed Eng, and also Theo Humphreys and Lauren Hilsbeck from the Pacific Northwest Cryo-EM Center, PNCC, and Michael Schmid from the Stanford SLAC Cryo-EM Center, S2C2. S2C2 has invited today's speaker, Dr. Ashish Manglik, and Mike will introduce him properly after my colleagues from each of the centers give a brief overview and update on the cryo-EM resources we offer at no cost to the research community. For those of you new to our webinar series, Cryo-EM Current Practices is an ongoing event that we host the last Thursday of every month at this same time. And we highlight particularly the methods that researchers are using to obtain and interpret the data they can collect at the national centers. So plug for next month's talk, James Chen and Brandon Malone from the Rockefeller University will be presenting presenting on "COVR: All the Bases: Structural Perspectives of SARS-CoV-2 RNA Synthesis." So you can save the date now for that talk. We are recording today's talk, and you'll find the recording along with those from past talks on the events page at cryoemcenters.org, along with the registration link for upcoming talks and other general information about the larger NIH Common Fund Cryo-EM program. A couple of final logistics. We'll be using the Q&A feature for questions. So you can ask your questions there and also upvote other questions that you see. If you have questions directed to logistics or access at the centers, our panelists will respond to those directly in that Q&A box, and we'll save questions for Ashish for the end of his talk. And with that, I will turn over to the representatives from each of the centers and Ed is up first. Take it away.

Hey, hello, and I'm here at New York City where NCAT is housed.

And we offer dedicated instrument access for high-end instrumentation on one of our four CRIS instruments, as well as a cross-training program.

Right now, we have an upcoming call for proposals July 1st.

But given the July 4th holiday, we'll accept some applications a little bit later.

Something to note, right now we are open for business and especially on-site access. So if you want to spend a couple days in New York for your CRIOS access or a few weeks for your embedded cross-training, we welcome you here to the Big Apple.

Hope to see you soon.

Thank you, Ed. Uh, Theo, you are up next.

Hello everyone, my name is Theo Humphreys and I work as a scientific point of contact for the Pacific Northwest Cryo-EM Center, or PNCC. At this time we offer one proposal type for both single particle analysis and tomography, and that grants you up to 480 hours of microscope time a year for up to 2 years. At the PNCC we have a monthly submission deadline for applications, and each approved proposal will be delegated a scientific point of contact such as myself to help answer scientific questions. The PNCC has 5 microscopes. We have an Arctico with a K3, a Krios with a Falcon 3 and a K3, 2 Krios with a Falcon 3 and BioQuantum K3, and 1 Krios with a Falcon 3 and BioContinuum K3. And for sample preparation, we have a VitraBot, like a GP2, and a VitraJet that's undergoing internal validation but will be offered to users once that's complete. While we're still under some On-site restrictions, we're offering one-on-one remote trainings and small remote workshops, and we hope to have visitors on-site soon again to offer workshops for sample preparation and microscope operation. We hope to work with you soon. Thanks. Thanks, Theo.

And Mike?

Hello, I'm Mike Schmid, and I'm one of the co-directors of the Stanford SLAC Cryo-EM Center. As for the other, as with the other two centers, we have access for high-end data collection and training. We have 3 Krios microscopes with K3 and Falcon 4 cameras. Some are equipped with energy filters. We have the normal selection of sample preparation equipment. And we are currently having some visitors on site for in-person training. We hope to have more people on site very soon for both in onsite training and to complement our remote remote access for data collection with, with on-site access for data collection. Please come and see us. The website is there. Thank you. Now, it gives me great pleasure to introduce our speaker for today, Ashish Manglik. Ashish received a specialist degree at Washington University in chemistry and biology, and while there received the Specter Award. He got his PhD and MD at Stanford University, right here, in 2014 and 2016. And he was working with, in the labs of Brian Kobilka and George Yorgos Kiniotis. He has received several awards for young and early investigator awards from Scientific American from the National Institutes of Health and from the International Narcotics Research Conference. His focus is on, as you'll see, cell signaling, especially the GPR system for cell signaling. There are thousands of GPCR receptors, most of which we don't know what they smell, and after they smell it, what they do. So he's very in, involved in figuring out this, these orphan receptors for which we do not know the, their function. His talk today will be Solving Molecular Puzzles in GPCR Signaling with Cryo-EM. Welcome, Ashish.

Great, thank you, Michael, for the introduction, and, you know, for giving me the chance to come talk here, and also for the attendees that have come. I must fully admit the beginning of this talk that I personally am learning cryo-EM really through a remarkable collaboration with Yifan Cheng's group, as well as, you know, bolstered by both the S2C2, SLAC Stanford Cryo-EM Center, and the Pacific Northwest Cryo-EM Center, as well as the UCSF Cryo-EM Center. So, you know, my background really is in structural biology, trained as a crystallographer in Brian Kobilka's lab, and what I'll hopefully get across to you today is how cryo-EM has completely revolutionized our ability to peer into how GPCRs work. Now focus really on one system just to give you a sense for how, how quickly and how much more rapidly we're able to move compared to how, you know, where we were about 5 or 6 years ago. So again, the title of my talk is Solving Molecular Puzzles in GPCR Signaling, and really what I view, you know, modern structural biology is able to do, instead of being an end in itself where we get, you know, the first high-resolution view after someone's entire PhD or postdoctoral experience, the rate at which we which we can get structures of GPCRs and signaling proteins, is such that we can really start to fit in, you know, many puzzle pieces at once. And really, it's not just, you know, the structure itself that we get, but really gives us deep insight into the physiology or the biology of the system. This talk will probably focus a lot more on the biology than the cryo-EM, mostly because that's really the perspective that my lab comes at it from. Okay, so what are we interested in? My lab is broadly interested in G protein-coupled receptors., and this very huge family of proteins are incredibly diverse, and they sit on the cell surface and recognize a whole host of stimuli. So everything from hormones, proteins, ions, pH, light, or even force itself can activate one of these receptors. And the really neat thing about this broad family is that each one of these receptors, while they're slightly diverse, has a kind of common mechanism of activation where this receptor is going to change shape, and go on to engage a whole host of intracellular signaling proteins and signaling cascades like the cyclic AMP pathway, calcium pathways, or growth factor pathways like MAP kinase pathway. And perhaps what's even more remarkable is that, you know, mammals encode thousands of GPCRs, about half of them are just specialized for olfaction, as Michael alluded to, but the rest of these G-protein-coupled receptors control almost every aspect of physiology. So everything from neurotransmission metabolism, Feeding, uh, you know, you name it, um, is controlled by one of these receptors. And as a result, they've been perennially amazing drug targets for manipulating human physiology and restoring physiology in disease. So with this kind of importance in biology and in medicine, it's no wonder that there's a lot of effort to understand how these receptors function at the atomic level with the hopes that we can manipulate physiology more precisely. So what I'll focus on today is, you know, not our work on many of these other physiological systems, but really focused on one specific signaling system that's deeply important in development and that goes awry in certain cancers, and that's the hedgehog signaling pathway. Let's give you a little bit of background on hedgehog signaling. This is one of these core developmental pathways that was discovered almost now 40 years ago by pioneering studies by Christian Nusslein-Volhard and Eric Wachshaus. Where they were looking at, you know, mutations that would disrupt normal development of Drosophila embryos. And this is the COVID of Nature, you know, from their pioneering studies, which showed that they could discover mutations that would disrupt this very normal segmentation of a developing Drosophila larvae. Again, here's a more recent figure. So you can see, for example, there's these very clear repeating units that are really important for determining certain axes for how the Drosophila embryo develops. And the hedgehog mutant was named because it basically completely disrupted this, this very specific phenotype, leading to something that resembled this cute little furry animal, or, you know, in subsequent years, uh, you know, the video game, uh, which some of you may have played. Okay, so what does hedgehog signaling practically do? This is just in Drosophila. Hedgehog signaling is deeply important in guiding the body plans of essentially all animals. So for example, just, you know, just to give you one example, if you're thinking about how do the cells in the developing— I have to do this on camera somehow— you know, how do the cells of the developing limb bud figure out which side is going to be the pinky or the thumb? It's actually a hedgehog gradient of this hormone that determines how that's going to happen. In the same way, hedgehog gradients determine what's going to end up becoming your head or your feet, what's going to end up becoming your front and your back. All of this is determined by hedgehog gradients in combination with a number of other morphogen gradients. And in adult tissues, when hedgehog signaling goes awry, people develop perhaps the most common cancer in all of North America, or certainly in Europe and the United States, which is basal cell carcinoma, as well as a number of other cancers that are driven by aberrant signaling of this pathway. Now, how does this pathway work? And this is really where we operate, which is we want to understand at atomic level detail a number of the puzzles that have remained despite, you know, 4 decades or whatever of beautiful genetics and a beautiful biochemistry and cell biology to understand how this pathway works. So what we know at this point is that there's a number of components involved. We know that there's a protein called Patched that has very high homology to transporter-like molecules that putatively transport some sort of sterol, you know, and depletes it from the inner side of the cellular membrane., and the constitutive activity of Patched inhibits a GPCR-like molecule called Smoothened, and this is what keeps the pathway off. Now when the Hedgehog morphogen comes along and binds Patched, it's proposed that Hedgehog inhibits whatever Patched function is. This then releases the brakes on Smoothened, and now Smoothened turns on, and this activated Smoothened inhibits the degradation of this transcription factor Glee, allowing it to build up and then turn on a whole host of transcription cascades that are responsible for hedgehog signaling. So again, this is, you know, these parts are known, but really fundamentally what allows this communication to happen or how Patch works remains very mysterious. And part of this mystery lies in the fact that there's a very intricate cell biological dance for all of these components. So for example,— when the pathway is off, many of these components, for example, PACH and another G-protein-coupled receptor, GPR161, exist in this very unique cellular membrane called the primary cilium. And when the pathway turns on, PACH has been shown to be moved out of the primary cilium, smoothin moves into the primary cilium, this receptor GPR161 also leaves. So it's a very intricate dance that's happening, but at some level all of this is controlled by very specific molecular interactions that have to happen. So a big goal that we have is to really understand, you know, what are, you know, fundamental puzzles in hedgehog signaling. So for example, it's still not 100% clear how Patch at the basal level inhibits the function of Smoothened. Even the molecular function of Patch— while it's been proposed that it's a transporter, we don't have incontrovertible evidence that that's truly the case. We still to this day don't really deeply understand or have a full understanding of how Smoothened is activated and how it then trigger signaling cascades. And then finally, this orphan receptor GPR161, while it's known that it's a really important negative regulator of the pathway, we have very little understanding of, you know, what manipulates its function to serve as a regulator. So each one of these questions, you know, what we're hoping to do is to understand at a very deep mechanistic level And in many ways, while we could go solve structures of many of these components, and some of these components, you know, have already been structurally characterized, really our hope is not just to understand structures, but to build bridges between structure and function. And, you know, this is obviously a beautiful picture of the Golden Gate Bridge, and, you know, if you live in San Francisco like I do, you know, perhaps you'll recognize that the Marin Headlands are are even more beautiful, and if you'd like to get over, again, the analogy here is that perhaps the function side lives here, and, you know, the structure is just plain old San Francisco. Okay, so if we think about this now from the terms, from the perspective of structural biology, you know, what we're able to do pretty often in structural biology, I say cryo-EM, you know, while it's an incredibly enabling technique, is still limited by this, is that our view on structural biology is limited to low-energy states of signaling systems. You know, these are usually the stable states that are either crystallizable, or I would say, you know, they're imageable. They're the ones that average together and that we can see in large numbers when we do a cryo-EM experiment. But really where biology happens, where signaling biology gets exciting and interesting, is our states that are actually responsible for transducing the signal, or for states that are responsible for transducing unique signals. And these states on average are unstable and dynamic, and as a result they've resisted crystallography certainly, and even for cryo-EM, unless you're going through really extensive methods of classification, it's going to be pretty rare that you can see one of these states. So our approach to this has been a little bit unique in that we think that instead of trying to go, you know, collect enough images or collect enough crystal structures to go see one of these states eventually, perhaps we can find ways of actually stabilizing these states to actually image them or to crystallize them. And the idea here is one that, you know, we employed quite a bit when I was in Brian Kobilka's lab, which is that if you have a very dynamic protein, like for example a G protein-coupled receptor, perhaps you could identify an antibody fragment that recognizes that excited state of the protein, one of these active or intermediate states, and make it amenable to structural biology study. And this general idea, you know, found a lot of traction for for the, uh, for crystallographers, and we'd argue that it actually again enables us to build bridges between the structures that we can now solve much more rapidly by cryo-EM, but really what their functional relevance is. Now, just to give you a little primer on antibodies, um, this is a structure of a conventional antibody. It's a pretty complicated molecule. It's composed of a heavy and a light chain, a heterodimeric protein, uh, and it turns out that camelids like alpaca or llama or camels themselves, make these really cool miniature antibodies that are just comprised of the single immunoglobulin domain of the heavy chain. But what's really cool about these miniature antibodies is that, similar to conventional antibodies, they can recognize an almost infinite range of antigens. But on top of that, they're single little proteins that are ultra-stable, and they're very similar to human antibodies. And, you know, for therapeutic purposes, they're They're very easy to make, and they're relatively cheap to make. Now, similar to conventional antibodies, which have about 6 very highly diverse loops that allow them to lock onto a bunch of different antigens, nanobodies have 3 loops. But what's really unique about nanobody loops is that they can inject themselves deep into the crevices of proteins that are either the active sites or the functional hearts of proteins. And as a result, they've served as really great ways pushing and pulling on signaling systems. Now in the past, most people have discovered antibodies by actually immunizing these large alpaca or llama, and in the past couple of years working with Andy Kooz's lab at Harvard, we've developed a number of completely in vitro methods to discover antibodies. So the idea here is that we start with very large repertoires of camelid antibodies, just to understand how to How does nature build antibodies? And what we've done over time is build very complex libraries, about a billion variants, billion unique antibodies, that we display on the surface of a yeast cell. So here, you know, I'm showing 10 or whatever antibodies. The unique thing is that in each one of these yeast is a plasmid that encodes for the antibody that's on the surface. Again, this library is about 10 to the 9th, or a billion unique antibodies. What we can do is go into this library, and for any given protein that we're interested in, we can do this kind of process. We can basically identify antibodies that are specific to our protein by basically pulling on those yeast that have a nanobody that binds to our protein, discarding all the rest, amplifying those yeast, and then repeating the process. And for us, this takes about on the order of about 2 days to run the cycle once. We run the cycle about 4 or 5 times, and thus we're able to find antibodies that are specific to any protein that we desire. And importantly, we've made this technology available for free to basically every nonprofit group. And I think last time I checked, we've distributed this, our libraries, to about 350 groups across the world. So with that, our hope is to again understand the structural biology of the hedgehog signaling pathway and, you know, in many cases, many other GPCR signaling pathways. But really our lens into this and what I'll describe today is that we use these single domain antibodies or nanobodies to basically push and pull on these various systems to understand how they respond, and as a result, what we hope to do in the atomic level is understand how— is understand the structural biology of one of these proteins, but then with functional studies, build bridges between a specific conformational state that we see and a specific function elicited by that conformation. Okay. So, let me dive into the first example of our work, and that's really to understand the function of this protein called Patched. Now, Patched had been proposed almost 20 years ago to catalytically influence the function of smoothen, and pretty early on there was a proposal by Phil Beachy's group at Stanford that Patched perhaps sequestered a lipid modulator of SMOOthen. And then the general model that's been proposed in the field is that Patch perhaps moves a sterol like cholesterol away from SMOOthen, and that's what keeps SMOOthen inactive. Now, over the past couple of years, the labs of Yanming, Shukun Li, and Vladimir Korkov, along with Phil Beachy and Yifan Cheng, have determined a number of structures of Patch. And some of these structures are Patch alone, some of them are with this morphogen hedgehog bound, And what these structures have generally revealed, of course, is the general topology of Patched, which is, you know, which is again very consistent with the transporter-like molecule. And they've also revealed how Hedgehog binds to Patched, and the model now is that Hedgehog, you know, binds this extracellular region of Patched, but then there's this N-terminal bit of Hedgehog that goes— dives deep into Patched, and the proposal is then that that hedge— that this little long bit of hedgehog basically occludes the pathway whereby some sort of molecule that would be transiting Patch needs to move. But that doesn't fundamentally answer the question of whether Patch is a transporter. At the most basic level, you know, some of the really pioneering work by Oleg Jardetzky and Peter Mitchell almost now, you know, 60 or so years ago, suggests that for a transporter to be a transporter, fundamentally it has to complete a conformational cycle. So it has to, you know, in a simple mode, you know, basically have to go from an outward open to an inward open state, or basically go through a complete conformational cycle in order to move some sort of substrate. So we had a simple idea on how to basically approach this, approach this question of whether PASH is a transporter. We wondered, could we develop a molecule that would trap Patch in a single conformation, and there are two possible outcomes. One possible outcome is kind of like a G-protein-coupled receptor. We trap it in one, in one conformation, and perhaps you trap it in an active or an inactive conformation that would lead to either pathway activation or inactivation. The other model is really consistent with the transporter, which is if we stabilize any conformation along a conformational cycle, that would inhibit the function of this protein, it's a transporter, and thereby turn the pathway on again because of this double, uh, this double negative control pathway. Okay, so this is a project led by Yunxiao Zhang and Fulvici's lab, and what Yunxiao ended up doing is using our nanobody library to find nanobodies that would bind to Patch in only a specific conformation. So what he did is he started again with our nanobody library, and he identified a mutant version of Patch that was locked to be in a completely inactive state, so that was incapable of doing its— whatever its function— putative function is. And what he would do over multiple rounds of selection is identify antibodies that were selective to this mutant form of Patch over the wild-type form of Patch. Ended up with about 15 clones, but I'll tell you about one of them today. Now, really critical— one thing I forgot to mention is that really critical to this process is, uh, an experiment that goes like this, which is, um, what Yousha and what we've done subsequently a lot in the lab is, uh, purify, for example, two different versions of the same protein. So in this case, uh, well, what Yousha did is, uh, he could purify a version of Patch that, um, in this case was inactive, you know, that's in the green color, purify another version of Patch that's in this case the wild-type protein in red, and then in order to find these very rare nanobodies that specifically bind to one conformation of PASH, what he does is mixes these two proteins that are each labeled with a different fluorophore with a pool of yeast, and then you can do a simple flow cytometry experiment where on the y-axis would be all the yeast that are binding to conformation 2, on the x-axis are all the yeast that are binding to conformation 1, and then by simply drawing this red little gate and selecting out those clones,— you could identify antibodies that would be selective again to one conformation versus the other. So, we did this process for Patched again, and he identified an antibody that binds to Patched, binds only again to this mutant form of Patched, and over multiple rounds of, you know, further improving its affinity, was able to get a molecule that not only binds to Patched, but turns the pathway on. So, this is data from some signaling data. This on the y-axis is just whether the pathway turns on, and the x-axis is concentration, and the best clone here is in blue here, which basically in a dose-dependent manner turns the pathway on to about 80% the activity of Hedgehog itself. We could also show kind of in an acute experiment that, you know, you could administer this kind of molecule in an animal and get pathway— to get the pathway to turn on. So really bona fide, it's a bona fide hatched patched inactivator and pathway agonist. Okay, so, again, our hope is to build bridges between function and structure, and working with Yifan Chen's group, again, one of our very— one of my more early forays into cryo-EM, we were able to determine a cryo-EM structure of patched bound to this nanobody, which we call nanobody Ti23. And what you can hopefully see here is that this nanobody binds an extracellular domain on PACHD, and here again is the kind of ribbon diagram of it. Again, in these structures, similar to many other structures of PACHD that have previously been solved, we see a number of sterile-like densities, again suggesting, very strongly suggesting, that PACHD is a sterile transporter. But here's where it gets really cool. When we compared where this nanobody binds, to the binding site of Hedgehog, we find that their interfaces are completely overlapping. That's what's being shown on the left here. So, in blue again is Hedgehog, and yellow is this nanobody, and you can see that they both basically landed on the same interface. I'll remind you that these selections that we did didn't— we didn't use Hedgehog as part of those selections. So, really what happened there is that this conformational selection that we do with our yeast display allowed the yeast, or the yeast library, to identify an allosteric hotspot on Patch, Okay, so what has this taught us? Now, if we compare the structures of Patched on its own that have been solved previously, in this case by the Beachy lab, or Patched bound to this nanobody, what we saw is that this critical region, which we've termed the switch helix, has two different conformations, you know, pose 1, and for Patched alone, and pose 2 when the nanobody binds. And now if we go back and look at all the structures of PAFs that have been previously solved, you know, either bound to Hedgehog or not bound to Hedgehog, what we see is that this switch helix is in two different conformations, and in fact this actually doesn't correlate with Hedgehog binding at all. So, some of these structures in conformation A are bound to Hedgehog, others are not, and same for conformation B, some are Hedgehog bound or not. But, um, so what we think is happening is that this nanobody stabilizes the switch helix in this specific conformation, And that this is associated with a constriction of this conduit in Patched. That's right here. So again, in green here is a structure of Patched, a conduit in Patched without any nanobody, and in blue here is what happens when the nanobody binds. Again, this just reflects that switch helix moving, which basically then again creates a new cavity and a constriction. So again, even in the absence of that long little lipidic, you know, tail of Hedgehog, this nanobody, by stabilizing a conformation of Patch in this region, is able to inactivate Patch and turn the pathway on, which again really suggests even more strongly now that Patch functions primarily as a transporter, and simply by changing the conformational ensemble of Patch, one's able to influence its activity and thereby influence pathway activity. Okay, so hopefully that gives you a little bit of vignette of, you know, how we're using single domain antibodies or nanobodies to probe transporter function. What I'll tell you about now is our efforts to understand how the pathway is turned on and understand this G-protein-coupled receptor called SMOOthened. So there's still fundamental questions, or some fundamental questions that remain in the field, one of which is how does PASH actually regulate the activity of SMOOthened, both at a distance, or, you know, what does PASH actually remove or sequester from SMOOthened to keep it off? And then fundamentally, once SMOOthened turns on, how does that activated stimulus then get transduced inside the cell. So, we had a goal to try to understand, you know, early on what is, you know, how does SMOOthen actually turn on, what is the active conformation of SMOOthen. And while there's now a number of structures of G protein-coupled receptors in the on and the off states, a couple of years ago we had a very poor perspective on this, this SMOOthen family of proteins, or what we call family F GPCRs, and what their activation mechanism is. Now, one of the things that we've learned from a whole host of structural biology and biophysics, including a lot of spectroscopy work, is that G-protein-coupled receptors are incredibly dynamic proteins, both in the off-state, but also in the on-state while they're bound to agonists. What we know now is that they're really quite conformationally complex, and if you think about this from, from kind of an energetics perspective, it's really the the presence of a transducer molecule like a G protein or beta-arrestin that causes these receptors to become in one conformation. So, in the absence of one of these transducers, these receptors explore many different conformations, and that's really what's made it very challenging to assess conformations of G protein-coupled receptors, and I'd argue many other transmembrane signaling proteins, in the, you know, by simply doing structural biology. I think really what you need to do to understand these excited states is to stabilize them in some way. Now, in order to really understand smooth end function, a couple of years ago, Ishan Deshpande in my lab did a screen, a very similar screen to what I described previously for Patch, but here the idea was to again start with one of our nanobody libraries, and then progressively select for nanobodies that would only bind to activated smooth ends. So, here Again, it's a very similar scheme, but what Yishan did is looked for nanobodies that would bound, that would selectively recognize agonist-occupied smoothen and would not bind inactive or antagonist-occupied smoothen. So, he identified a number of clones which we've subsequently characterized. The first one, and this was, you know, when we were far more adept at crystallography than cryo-EM, we were able to get a structure with x-ray crystallography, and this structure actually revealed quite a bit about how this pathway, how this receptor turns on. And it's published work, so I won't spend a ton of time on it. But, you know, here what I'm showing is a structure of Smoothened, again, in an active conformation, bound to one of these nanobodies, in this case nanobody Smoothened8, which binds to the bottom of this receptor and again props it open and keeps it on. What we learned from the structure, and that's, you know, one of the more remarkable findings from this study, is that while activated smooth end again binds to this SAG, this synthetic agonist that we actually put in the mixture, it actually also comes down with a cholesterol molecule that's deep inside the heart of smooth end. So, the field previous to this had primarily believed that cholesterol binding to this extracellular site was important for function, and again, solving the structure of the activated conformation of the excited state in this case revealed a cholesterol bound deep in the heart of SMUDEN that's actually, we think, probably more important for how the pathway works, and is really the cholesterol that Hatched probably sequesters away from SMUDEN. Now, at the same time, you know, while we were focused on trying to get a crystal structure of an activated SMUDEN with one of these nanobodies, the labs of Cheng Jiang and Shaoqun Li have determined a structure of— cryo-EM structure of SMUDEN in this case bound to a heterotrimeric G protein of the Gi class. And for those of you who have been following the, the GPCR cryo-EM field, you'll recognize that there's been an absolute revolution in the structural biology of G protein-coupled receptors by cryo-EM over the past, you know, 4 to 5 years. And that's really driven in part by the remarkable ability to very rapidly generate complexes of GPCRs with heterotrimeric G protein. So, in some of our own work led by Antoine Cole, who was a graduate student at the time, and as well as when I was in the lab of Brian Kobilka and working with Georgios Keniotis, you know, what you can hopefully see, you know, from this, from even from this early work, is that, you know, a 2D class average of a GPCR bound to a G protein is pretty darn easy to analyze, and you can see the receptor clearly there, different components of the G protein, there, as well as an antibody fragment in this case that we discovered for this complex, that's found a lot of use subsequently, that then, you know, eventually gave rise to a pretty nice, pretty reasonable 3D reconstruction to enable structure determination. And really, this remarkable revolution in structural biology for GPCRs by cryo-EM is driven by the fact that most GPCRs, by their fundamental names, G-protein-coupled receptors, couple the G proteins, and if you can make this complex between a receptor and a G protein, this G protein essentially serves as a fiducial to drive alignment of the GPCR. So, um, so what we've seen very dramatically is a proliferation of structures of GPCRs bound to their G proteins, and this is, this is really exciting because it enables us to see, uh, receptors, you know, bound to agonists or bound to their endogenous hormones. But one of the big challenges is, um, that still remains is, what if you want to determine structures of G-protein-coupled receptors in their off states? And really, if you want to understand how proteins function, you know, you really need to understand multiple conformations in order to understand really what drives changes in conformation. So, this has still been perplexing, and one part of the reason why this has been challenging is because, like, like many other very small membrane proteins, G-protein-coupled receptors have almost no extracellular or intracellular real estate that sticks out of the micelle or out of a, you know, a nanodisc. And as a result, it's very hard to drive particle alignment, both from a size perspective, but also from the fact that these particles are essentially spherical. So, we've been also pondering and working on using our single-domain antibodies to essentially serve as fiducials for inactive state structures of G-protein-coupled receptors. And this is not a new idea, you know, people have proposed this quite some time ago, and, you know, Yifan Cheng, our main collaborator in all this work, has proposed that, you know, using antibody fragments or FAB fragments for small membrane proteins may function as far as, you know, a decade or so ago. But we wondered whether we could use again these, these nanobodies that were coming out of our screens and determine structures of of G-protein-coupled receptors, with the hopes that perhaps again we could use structures of these antibodies bound to their targets to build bridges between a conformation that we could observe and a function in a cellular context. So again, going back to this original screen that, that Ishan in my lab did, a different antibody that he found from this screen we've worked up now, and this is a pretty curious antibody. We call it antibody SMUDN12., and what's really unique about this antibody is that it's actually a partial agonist of the pathway. This is pretty rare. It turns out that finding antibody modulators or antibody signaling modulators of G-protein-coupled receptors remains a pretty challenging feat, and what Yifan, what Yishan has found is that this antibody can on its own very weakly turn the pathway on. So what you're seeing again on the y-axis is just whether the pathway is on or not, and on the x-axis is a dose response of this antibody. We know that it's a partial agonist because if we fully turn the pathway on with Hedgehog and then dose in the nanobody, we see a decrease in signaling back down to this, you know, 25 or 20% level. So, that's really very convincingly suggests that this nanobody, by doing whatever it does, clamps signaling at a very specific level in the pathway. So, we wondered how does this nanobody actually do that. And, and I'll be honest, you know, when Ishan in my lab, working with Brian Faust, another graduate student, wanted to approach this by cryo-EM, you know, I, I really didn't think this was going to work. And I'll also say that, you know, we'd spent quite a bit of time getting the structure of nanobody 12 bound to, to smoothened by X-ray crystallography with a maximal resolution that we were able to obtain of about 2.8 angstroms. And, you know, well, that was probably good enough, we probably didn't need a better structure than that. And what's really remarkable is that, you know, I think on their first try, Yishan and Brian working with, you know, the S2C2 Center, were able to get a structure of SMUDEN bound to just this, this nanobody. So, again, in green here is the 7-transmembrane domain of SMUDEN, which a little bit of the extracellular region, and in orange here is this nanobody. Just to give you some kind of statistics, you know, this is a pretty small normal protein complex of only about 75 kilodaltons. I'd wager that the structured part is probably even smaller, probably approaching about 50 or 60 kilodaltons, and the overall, you know, kind of nominal resolution of this structure is really quite good, about 2.8 angstrom resolution. So, just a couple of other highlights— if we can get to the next slide. This structure was solved with a pretty high affinity antagonist called SNAPT-1, discovered again in Phil Beachy's lab about 15 or 20 years ago. And what's remarkable to me is that, you know, this structure with just this nanobody has a pretty good density for this molecule, so that's sufficient that it's completely unambiguous to resolve it. And for us, in other parts of our work in the lab, we're very interested in using, you know, structures to discover new small molecule modulators, and this is a very exciting space for us, because in the past it's been very challenging to crystallize GPCRs with low affinity molecules, and we're hopeful that the ability to get a cryo-EM structure of SMUDN with this nanobody, you know, relatively easily may enable us to now actually go discover new molecules and, and see how they bind, especially low affinity molecules that are pretty, you know, early hits in a campaign. Now the other exciting thing about this specific nanobody is that I told you this nanobody is a partial agonist, and in many ways it actually behaves as an inhibitor in a more endogenous system, in more endogenous signaling systems. And when we map the interface of this nanobody on the smooth end, which is again shown in these orange, in this orange cartoon here, against the surface of smooth end, which is colored by conservation, What we see is that parts of this nanobody engage pretty highly conserved regions of SMUDN, and the reason why we think this is great and pretty exciting is that there's a lot of interest in developing SMUDN inhibitors for a variety of cancers, but a subset of those patients that take SMUDN inhibitors develop resistance to the inhibitors that have thus far been developed. So, the exciting thing is that this nanobody targets a completely new allosteric site that hasn't been described to date, And we think that perhaps targeting this new site may be a great way of overcoming the resistance that's been observed in the clinic thus far. Okay, so I told you now about, you know, a couple of small mysteries that we've solved. You know, we've worked a little bit on how Patch actually functions. A lot of work that we've dedicated to understanding how SMUDN is activated, both at the level of, you know, what is the conform— what are the conformational changes in SMUDN, that occur upon pathway activation, and how can new, you know, poorly described allosteric sites impinge on pathway activity, with some exciting results perhaps for using, you know, very small membrane protein fiducials to drive particle alignment. The last thing I wanted to highlight, that's again all unpublished work, is our work to try to understand an important negative regulator of the pathway. This is a molecule called GPR161. And about 8 or so years ago, the lab of Peter Jackson at Stanford and really Saiket Mukhopadhyay discovered that this orphan receptor GPR161 is a really important negative regulator of the pathway. And the way that this works is that when the pathway is off, GPR161 constitutively turns on the GS signaling pathway to keep the pathway off, and then when hedgehog signaling turns on, when the hedgehog morphogen's present, GPR161 has to leave the primary cilium in order for the pathway to turn on. So while it's clear that GPR161 is a very important part of the pathway, it's been very unclear what actually controls GPR161 activity, and it's been proposed that, you know, GPR161 is a constitutively active GPCR, but more broadly, You know, one of the things that's really unique about the GPCR, about all GPCRs, is that there's about 100 or so GPCRs for which we have absolutely no idea what their endogenous ligands are. So this is a nice— on the left here is a nice dendrogram from the GPCRDB that highlights, you know, major classes, all major classes of G protein-coupled receptors. And on the bottom left, this kind of big group here are all of the Class A orphan receptors. Now, out of all of these orphan receptors, there's basically only one that's been structurally characterized today. And that's one of the big challenges is, you know, if you don't have a stimulus, if you don't know what the stimulus for a receptor is, it's very hard to go characterize it, both at the cell biological level or the physiological levels, but also at the structural biology level. It's very hard to understand, you know, how a receptor functions. So, we've started to develop new biochemical approaches to purify orphan receptors and to start to work with them, with the hopes that we can use biochemistry and perhaps even structural biology as a way of, as, you know, a kind of a completely orthogonal way of getting into how orphan receptors function. And for GPR161, we've had a pretty exciting early success. So, you know, Ishan Deshpande again in my lab working with with a really talented rotation student, Simone Harrison, another graduate student in my lab, Nick Koppi, we're able to, you know, purify GPR161 again bound to, in this case, the GS— this is a type of GS heterotrimeric G protein, and one of these nanobodies that the Kobilka lab had developed many years ago to kind of lock everything together. And we're able to, again, with working with the Stanford S2C2 Center, determine a pretty reasonably good cryo-EM structure of GPR161 again bound to its G protein. Now, the structures revealed a few very interesting highlights. The first is if we look at the extracellular side of the receptor where a stimulus normally would be, what we note is that— let me get this thing to play well. Okay, there we go. So, what we know is that what I've highlighted in red here is that the extracellular loop 2 of this receptor folds down on top of itself, and what we think is happening here is that this receptor has basically become self-activating, and there's a little bit of precedence for this kind of concept from another orphan receptor, in that in order for this receptor to function, what it basically does is is in the absence of any kind of exogenous stimulus, this loop folds down on itself and basically keeps this receptor constitutively active. But on top of that, another really exciting thing that the structure has revealed is that we observe a mystery density that's consistent with a sterol kind of appended to the side of this receptor. And if we look at all of the amino acids that surround this sterol, they're highly conserved all the way down to echinoderms. Which again suggests that we've potentially uncovered another sterol-regulated receptor in this pathway. And again, this opens a new frame to try to understand how sterols regulate hedgehog signaling biology. So again, this project is at a very early stage, but we're hopeful that, you know, with having identified this potentially new modulator, we might be able to really understand how GPR161 controls hedgehog signaling. Okay, so with that, I'll, I'll kind of start to summarize. We think we're at a very unique time in structural biology, certainly for the GPCR field, in which our ability to gain structures, you know, far outpaces our ability to understand what they actually mean. And what this has led to is that structural biology for us has largely become not an end in itself, but really a means to see something for the first time and ask, you know, what is a relevant function of it. And certainly for orphan receptors, we're finding this more commonly that, you know, we make observations for new structures and often actually reveals new biology. A big push for us remains to try to build bridges between structural biology and function, both in the context of cellular biology, but also physiology. And for us, these antibody fragments are really a great way of doing that for probing from structure at the atomic level, but also interrogating what they do functionally to understand what, what their role, what the role of that conformation is in physiology or cell biology. So, with that, I'll acknowledge again some of the key people that have enabled all this work. Almost all of our cryo-EM work is done in deep collaboration with Yifan Cheng's group, who has been deeply generous in almost every aspect, both training members of my group, you know, jointly advising a student, in this case Brian Faust, who's again led a lot of the effort here. Almost all of the hedgehog biology work in my lab in some ways led by a really talented postdoctoral fellow, Ishan Deshpande, and Nick Hoppe has been a really remarkable graduate student working at the interface of trying to understand orphan receptors and also using cryo-EM. We're a really remarkable lab of trainees that are, you know, willing to work on pretty challenging projects, And, you know, all of our hedgehog work is done in collaboration with Phil Beachy's lab at Stanford and Ben Myers, who's now at the University of Utah. Yunxiao Zheng's work I already talked about, and one person I didn't get a chance to talk about is Kaihua Zhang, who's a postdoc fellow in Yifan Cheng's group and has done really beautiful work on hedgehog signaling as well. All of our nanobody work is done in collaboration with Andy Kruse's lab and his postdoctoral fellow Connor McMahon. And, you know, we're obviously very grateful for a variety of funding sources, but for this audience, what I'll give, you know, a couple of key shoutouts to, you know, the uses of cryo-EM facilities has been incredibly helpful, David Bulkeley in particular, but really access to a number of cryo-EM centers, both the Stanford SLAC S2C2 Center, which is responsible for a lot of this work, as well as the PNCC Center, for which we've been able to get amazing data, but is work on a completely different project, as well as the NCI Cardiovascular Institute. All of these national facilities have really enabled us to get data at a rate that really I couldn't have imagined a couple of years ago. So really want to thank all these groups for making these resources so available and for the training and just the expediency of being able to get collected at these sites. So with that, I'll stop and take questions if there are any, and thank you for your time.

Thank you, Arshif. We have a question from Nam Nguyen. Can you please comment on why is it possible to determine the structure of Smoothened in complex with the NbSmo12 nanobody to such high resolution, but not a GPCR with an ICL3 fusion?

Yeah, great question. So for those of you who are perhaps not familiar with what an ICL3 fusion is, one of the strategies that had worked incredibly well for the crystallography era of determining GPCR structures is that instead of trying to develop antibody fragments to enable crystallography, what many groups did is, and this is really good pioneering work by the Kobilka lab, is to basically create a fusion protein into the intracellular loop of a GPCR, and really the primary thing that that fusion led to, it was the ability to form crystal lattice contacts that then enabled us to get crystal structures. And the challenge with ICL3 fusions is that, you know, there's basically two points of attachment, so getting that entire thing to be rigid where the mass of the fusion protein is driving alignment of the GPCR is really quite challenging, and in my lab, you know, we've tried this quite a bit. It's worked to some extent, you know, we've heard rumors where this has worked a little bit better for others, so it's, so it's kind of, you know, it's a very bespoke thing for each specific project. I think the reason why this nanobody approach works really well is because almost all of the mass is very rigidly coupled to a 3-dimensional epitope on a protein, and that's why you know, we think that it basically drives particle alignment so well, and, you know, there's great examples of this, not just for GPCRs now, but for a number of other transporters, and certainly, you know, great examples of using full FAB fragments as well to drive particle alignment. So we think it's really a complex three-dimensional epitope that enables this.

[Speaker] Thank you. I have a question about the comparison of the hedgehog binding to the— to the nanobody binding. And you say that the constriction is driven by the movement of a helix in Patched itself rather than any obstruction that's introduced by Hedgehog itself, like that long tail you originally pointed to as something that was kind of a candidate for that. Can you then look at what parts of Hedgehog are disposable, that you don't have to have that long tail as long as you have something that moves that helix to that position of that in there? Does it give you more insight into what parts of Hedgehog really were important?

Yeah, so that's, that's a great question. I think I probably misspoke or spoke too fast when I was talking about this. So what the field has done, Michael, is a version of that experiment. They basically chopped up hedgehog and tried to figure out, you know, what is the business end of hedgehog. And what they've actually found is that this, that this peptide that digs deep into hedgehog is actually the most important part for, for inhibiting the function of Patch. And in fact, this, this region that binds up here is important for affinity, but it's actually dispensable for for the activity of Hedgehog. So, you know, that's why I think our study is— kind of flips that logic on its head. It's not just that that lipid part, that blocking the conduit is important. You can actually block the function of Patch, not just plugging the conduit, but by sterically preventing its conformational transitions. So, you know, Hedgehog itself actually probably doesn't necessarily limit this this switch helix movement in the same way that the nanobody does, because, you know, what we see is that the nanobody really stabilizes one pose of the switch helix, whereas in the prior structures of Patch that have been solved, you know, some of these red conformations or blue conformations, or both of the, you know, when Hedgehog binds, the switch helix has been observed in both states. So, clearly Hedgehog probably doesn't influence the conformation of the helix, but our nanobody, just by impinging on this helix, is able to recapitulate what Hedgehog does in a very different way, we think.

Thank you. We have another question. Is it possible to form trimeric complex with B2-ArgS without ligand in the purified system?

If so, the GPR161 being self-activating might not be unique. That's the question. Yeah, that's a good question. So, it, you know, the biochemistry of whether you can form a trimeric complex and whether it's stable enough is, you know, remains to be done. So, I think I actually have done that specific experiment that Sungsoo Yoo is asking for at some point in graduate school, and I think that the answer is yes, you actually can form a complex even in the absence of an agonist. It's not a very stable complex, but it can be formed, and probably if one really desired to image it, probably could be imaged. We think that the GPR161 self-activating, you know, feature is unique because we don't see this— if I could go back to that slide easily. I'll have to play the darn movie. Hopefully it sticks for a second. For most GPCRs, we don't see this kind of, you know, this extracellular loop actually diving deep and occupying the same region that an agonist would occupy. So, we think that, you know, we think it's self-activating, not because we can form this complex, but because of the fact that this loop actually takes up the same space where normally an agonist would sit.

Thank you. So, we have a question from Paula Flicker.

You are using nanodiscs for the EM in, in these cases, correct? Paul, that's a great question, and I'm slightly embarrassed to admit that most of this work is actually with detergent. So, we're in the GPCR field, unfortunately, has been a little bit slow to adopt nanodiscs, and that's partly because, you know, in the past, you know, when I was in Brian Kobilka's lab, this is again work really led by Antoine Cole, working with Georgios Steniotis, we tried for a very long time finding an ideal detergent, an ideal setup to do cryo-EM for GPCRs. And we landed on was a very specific mixture of two detergents that's really, you know, and it's been incredibly enabling for the entire field to get structures of GPCR/G protein complexes. So mostly for reasons of expediency, we haven't gone back and done really careful work with with nanodiscs. My lab is slowly moving, you know, to using nanodiscs. There's a couple of great examples out of Gerhard Wagner's lab, as well as, you know, working with Alan Brown, as well as Dan Rosenbaum's lab of structures of GPCRs now in nanodiscs. But by and large, that, you know, still hasn't caught on in the field as kind of the de facto mode of getting these structures.

But great question, and something that we're striving towards. If I can ask a quick question. So you maybe touched on this, Going back to your nanobody-12 smoothened structure, you said the connection was not very flexible. So I was wondering if you could speak to sort of local resolution, given you're using that nanobody as a fiducial. Does it look good in that particular structure? And then given you've now looked at a couple of things, are there any trends in terms of local resolution when you're using these nanobodies as fiducials?

Yeah, so we— I will say that we— this is the only example that we've gotten. This has emboldened us to do more antibody-driven fiducial, you know. I really— and this work is very hot off the press. I think we only got the structure about 3 or 4 weeks, or maybe 6 weeks ago. We are now emboldened to try other antibodies that we've discovered over many years, and to see if they're sufficient to drive particle alignment of GPCR. GPCRs. My gut feeling is that because, um, you know, the center of mass of this nanobody is not that far away from kind of the center of mass of the GPCR, that we won't get these, like, very profound local resolution problems. For the G protein complexes, this has actually been a problem for the field, where a lot of the center of mass of the G protein, as you know, drives particle alignment for the G protein, and you get these beautiful high-resolution views of the G protein, which are basically the same for every GPCR complex. And then, you know, the part that you care a lot about, which is the drug binding pocket, has even just abysmal resolution, and you just can't even see what we care about the most. So we think that, you know, this antibody approach certainly, you know, will hopefully have less of that, but then again, we don't know how successful this will be over many different receptors and many different antibodies.

This might just be an edge case, or it could be, you know, more common. Awesome. Well, we're at time actually, so if there are any burning questions, people can type them in, but most people probably have somewhere to run off to. So thank you very much, Ashish. That was really gorgeous maps that we saw there. And we hope everyone joins us again at another one of these seminars. And, you know, from the Cryo-EM Centers, please let us know how we can help you, and hopefully we will see you again.

Thank you, Ashish. Thanks a lot, everyone.