 with starting it. And it appears that we are officially live. So hello everybody and welcome back to another session of our Sussex Vision Seminar Series, as always within the Worldwide Neuroinitiative. I'm Giorgio Caffegi, a master's graduate from Thomas Euler's lab and currently a PhD student with Tom Baden. And as your host for today, I would like to once again begin by thanking Tim Vogels and Panos Bozellos for putting forward this several expanding initiative towards a greener and much more accessible seminar work. Having said that, allow me of course to get back to the reason we all gathered here for today and introduce our guest from Northwestern University, Professor Tiffany Smith. Tiffany's scientific path reflects her growing interest in retinal ganglion cells that sense light directly. Following her undergrad degree in biology and psychology, Tiffany joined Paolo Kofugi at the University of Minnesota for her PhD on the diversity of form and function of these intrinsically photosensitive retinal ganglion cells, the IPRGCs as they are abbreviated. In 2010, she moved for some very successful postdoctoral years to Johns Hopkins University and the lab of Summer Hatter before joining Northwestern in 2014 and starting their own lab. She has remained there since and nowadays Tiffany holds the title of associate professor at the Department of Neurobiology and is the director of both the undergrad and the master's neuroscience programs. In her lab they are interested in how light regulates behavior and they tackle this question at different levels from how cells function to where they project in higher brain areas and to how or what actually they contribute to different visual behaviors. For such multi level questions they exploit the genetic accessibility of mouse models and employ different techniques from molecular to ify's and behavioral to elucidate the role of these IPRGCs in both image and non-image forming vision. It is such a pleasure to have Tiffany here with us today, taking us through some of the latest and I'm sure exciting findings in her talk entitled visual circuits for threat anticipation. So without any further ado from my side, please all welcome Professor Smith. Tiffany the stage is officially all yours. Thank you very much, George for the invitation to speak today it's great to be back let me just get my screen. Organized here. Okay. So, as George mentioned, our lab has really had a long standing interest in understanding how light affects, you know, behavior and we want to understand that question all the way from the retinas to the brain and kind of all of the circuits and cell types. So today I'm going to tell you about actually two ongoing projects in the lab both from Marco Saranda who's an incredibly talented postdoctoral fellow in the lab, going on the job market in the next year or two so watch out for him. And I'm excited to share this with you so. Light influences behavior through really two distinct systems that we have image forming vision which involves the detection and perception of color contrast form and motion. And you have non image forming vision which involves those functions that occur outside of our conscious perception. So really well known one non image forming function is circadian photo entrainment where organisms will align their daily activity rhythms to the environmental light cycle in order to regulate or in order to align an animal's activity and physiology of them to the environmental light dark cycle. Another really important non image forming function is the people are light reflects which is really key for regulating the amount of I that's entering our eyes and getting to our retinas across different luminances in the environment. And of course, in mammals all this information is getting to the visual system via the retina, which is this layered structure that sits at the back of the eye and consists of six different types of neurons. So the light comes into the eyes focused by the cornea and the lens onto the retina passes through all of these layers of neurons, falling on the outer segments of these rod and comb photo receptors, which will then convert that light information into an animal that's relayed back through this network of internal rounds to the retinal ganglion cells of the ganglion cell layer. And these ganglion cells project to 40 plus targets in the mouse brain which I'm showing you a few of here. And these different structures in the brain are then sort of differentially regulating these image versus done image forming behaviors. So this is part of the brain involved in conscious visual perception or image forming vision such as the dorsal lateral geniculate nucleus which is the visual thalamus, as well as the superior colliculus, and your brain regions devoted to non image forming functions. So the super cosmetic nucleus of the hypothalamus is one such region so this is the master circadian pacemaker, and this pacemaker is responsible for all of the other clocks in the body and there are many to this central pacemaker. And so light input from the retina to the SCN are what is then responsible for photo and training, your endogenous activity rhythms to that environmental light dark cycle. Another important non image forming region is the shell of the all very presexual nucleus which is important for driving pupil construction in response to increases in environmental illumination. So for a long time this was kind of this, the, you know, the way we thought things worked you have light detection and the rather comfort receptors relive that information by the ganglion cells to the brain and then you have these image and non image forming functions, driven by these various brain regions receiving ganglion cell input. It started to become clear back in the, you know, 80s and 90s that this really might not be the whole story that rods and cones may not be the only light sensitive cells in the retina. And these clues actually came from first patient data from clinician observations of blind patients who they noticed retained certain non image forming functions even though they were completely perceptually blind. So I'll show you one example of what that looks like here. This is work from Chuck sizler's group at Harvard in 1995. And here you have a normally cited subjects and their plasma plasma melatonin levels and melatonin is the so called sleep hormone. And so melatonin will be high during the night and low during the day in a normally cited subject and then if you give a person a brief light pulse in the middle of the night when their melatonin is high. You can see that their melatonin levels will sort of transiently and fairly rapidly decrease. This is one example of why it's really bad to give yourself a light pulse in the middle of the night as a side note because this is what happens to your melatonin hormone levels. You can see that when the lights do turn off though that level will rebound back up to sort of its normal level and then go back down when the day comes. That's normally cited subject but what they noticed was when they looked at these melatonin levels in blind patients who had degenerated rod and comb photoreceptors. So no known photoreceptors in the retina those blind patients retain the ability to have high melatonin during the night low melatonin during the day. And you'll notice that this blind individual also showed this transient decrease in melatonin levels to a light pulse in the middle of the night. So how do you have people lacking all known photoreceptors who are clearly still responding to environmental light cues, even though they're telling you they're completely perceptually blind, these people have no light perception. So it was these types of data that really include people in that there was likely a third class of photoreceptor. Others recapitulated these findings in mice that lacked all of the rod and comb photoreceptors. And so then the search was on for really the identity of this new class of photosensitive cell. So taking a cue from the fact that the functions that were retained in these blind individuals were non image forming functions. So what this person's group did are really took a really simple but really elegant approach I think to trying to uncover the identity of these new photoreceptors and what they did was retrogradedly labeled the ganglion cells in the retina that projected to this central circadian pacemaker. They're doing this in rats in this particular experiment. They put fluorescent beads in the SCN retro beads and this allows them then to identify the ganglion cells in the retina which is shown here and be that send input to this central pacemaker. What you can then do is target those SCN projecting ganglion cells for electrophysiology and see if those cells themselves might be light sensitive. So when David's group recorded from these SCN projecting ganglion cells. They looked at the light response. These are current clamp recordings and these steps up here are just when the lights turn on and then the lights turn back off. You can see that these SCN projecting ganglion cells if you look at the black traces first depolarize the response to the onset of that light stimulus. They maintain their depolarization throughout the light stimulus and actually long after the light stimulus turns off these ganglion cells continued to fire. And you know this in and of itself is not earth shattering a ganglion cells should respond to light if you give it a light pulse and record from it. But what made these SCN projecting ganglion cells incredibly unique and special was that the person lab then recorded from these same cells but in a cocktail of pharmacological blockers of all synaptic transmission in the retina. And that's what's shown in these top two traces in red. And you can see that even in the absence of any input from the rod and comb photoreceptors thought to be the only photoreceptors in the retina. These SCN projecting ganglion cells continue to respond to light and they do so with that same sustained depolarization that persists following stimulus offset. Indeed, even if they physically isolated the somata of these SCN projecting ganglion cells from the surrounding retinal tissue you can see that the soma itself of this particular cell continues to depolarize in response to light. So these ganglion cells were truly intrinsically photosensitive and we're given the very long name of intrinsically photosensitive retinal ganglion cells so we call them IPR GCs for short. And the protein that allows these IPR GCs to respond directly to light is a photopigment called melanopsin. So like rod and cone options melanopsin is a seven transmembrane G protein coupled receptor option called open for it is maximally sensitive to blue light. So at about 480 nanometers. And so IPR GCs are bonafide photoreceptors they express a photopigment and they respond directly to light. And one of the things that makes them really fun to study in the retina is that they're also bonafide ganglion cells in the sense that they are getting input relayed from these rod and comb photoreceptors that they are then integrating with their intrinsic melanopsin response which gives them some really interesting signaling properties that I'm not going to talk about today. And it's also important to note that these IPR GCs are actually incredibly well conserved we see them. Well, they others have seen them using melanopsin immuno labeling in human and on retinas. And there's good evidence to that melanopsin is affecting the behavior of humans as well as mice and other animals. So if we look at the projections of IPR GCs and compare them to conventional ganglion cells IPR GCs are shown in this light red and conventional ganglion cells and gray, you can see that IPR GCs project quite widely throughout the brain. This is one of the reasons that I love studying this system because IPR GCs project to so many different brain regions that I feel like by studying the IPR GC circuits we can actually learn a lot about many different visual circuits. And we know from a lot of work from Sam or Hattar's lab but also other labs that and some of my own work I guess that IPR GCs are required for multiple types of behavior. So we know they're absolutely required to relay any light information for non image forming functions so if you blight the IPR GCs for example, animals will lose the ability to photo and train. And they'll also lose the ability to constrict their pupils in response to light. They relay the rod, the cone and the melanopsin input to drive these non image forming functions. But we also know that IPR GCs are involved in pattern vision. You know, so they originally discovered as a sort of a circadian photoreceptor but as we've learned more about the system we now know that they're also involved in visual perception behaviors along with the conventional ganglion cells. So if you blight the IPR GCs or you knock out melanopsin you get deficits in an animal's ability to, for example, see contrast as well that they don't become completely blind like the conventional ganglion cells play a very important role in visual perception as well. They're really influencing a broad range of behaviors. And so before I transition into the actual data, I just want to leave you with this thought that, you know, we know about a few different behaviors that IPR GCs modulate but really their broad projections throughout the brain suggest that there are many levels of modulation of IPR GCs on different behaviors that we just have not uncovered yet. And so that's what I'm going to talk to you about first today. And so we got interested in the question of how do animals decide what types of behaviors they actually want to take when they're acting in sort of their naturalistic environment where they're not always safe. So one of the interactions that is most fascinating, I think in nature is interaction between predator and prey. So predators have evolved multiple mechanisms for better detecting and capturing prey. And likewise, prey animals have evolved a lot of mechanisms for better evading or escaping from predators. And of course, the visual system is often central to this ability of these different prey animals to avoid predators. But I think equally important to the ability to sort of avoid detection or escape from a predator once you've been spotted is when you're making the decision as an organism about how much risk you want to take when you're entering a given environment. So you can imagine the animals are constantly faced with the decision of do I enter this environment and if I do so, you know, I might get eaten by a predator. But I get to, you know, obtain food or mate, or do I avoid this situation because it's too dangerous that if that organism avoids that situation, it does so at the cost of getting food or of reproducing so they're constantly doing this calculation and of course in order to survive you have to be making the right types of calculation. So the system has to be really well calibrated and animals have to be able to predict whether a threat is actually going to appear. So Marcos got interested in this idea of how well an animal can actually predict the appearance of a threat. Thankfully in the visual system, we actually know of a stimulus for mice that is innately threatening meaning it doesn't require training for the animal to become afraid of it. This is called the looming stimulus. So again to mice this is innately threatening and it's actually quite simple to present to the mouse so you can put them in this arena. Once they're in the center of the arena you have a computer screen over the top, and you can play an expanding black disk over the mouse over multiple seconds one expansion per second. And mice will find this innately threatening and they will either freeze or they will escape and it's thought to be innately threatening for the animal because it makes the approach of a predator overhead which would of course take the form of this big shadow expanding in the mouse's visual field. So Marcos wondered if we could use the fact that this is an innately threatening stimulus to see how well an animal can predict the future reappearance of a threat in an environment that it's previously experienced. This is the paradigm that Marcos designed to measure something we're calling visual threat anticipation or VEDA for short. So what Marcos does is he places the mouse in an arena for four minutes. The animal's never been in this arena before there's just a neutral contact for the animal. He'll then play after those four minutes have elapsed one set of a looming stimulus exposure over 10 seconds so 10 expanding disks of a single contrast over those 10 seconds. And then we just put the animal back in its own cage for two days. Two days later we come back and we just put the animal back in that arena and it gets no further visual stimulation. And we see if the animal is acting like it's experienced or that it's expecting the reappearance of this threatening stimulus that it experienced. So if it is anticipating the reappearance of that threat, the idea would be that that animal would avoid the threat zone where the threat had previously appeared may also reduce the amount of locomotion. So reduce the exploration of the arena. This is the anticipatory behavior that we're actually interested in measuring here during the test phase. So first we wanted to measure the looming responses of animals in this paradigm in our hands and we wanted to look at it to looming stimuli of contrast, you know, from very low to 100%. So this is the percentage of animals that respond to a looming stimulus in our exposure phase, this middle phase here. And you can see that as the contrast of a looming stimulus increase, the percentage of animals that are showing a detectable response to that stimulus in the form of freezing are increases. And then you can also just measure freezing time and you can see that as the contrast increases the amount of time animals spend freezing also increases for that 10 second period. So we can, we are able to generate a looming response. And it's important to note that this 55% contrast which is very low doesn't actually elicit a detectable looming response for the animals in our paradigm and this will become important in a minute here. This is sort of a sub threshold stimulus. Okay, so now what about the anticipatory response, visual threat anticipation, what does this look like in a control animal. So if you put the animal back in that context two days after its experience one exposure to a looming stimulus of a given contrast we chose this 55% contrast to start with and very surprisingly we found that even at this really low contrast so fairly low salience stimulus where we couldn't detect a looming response, we saw that animals would show anticipatory behavior. So on the left is an animal that received no looming stimulus is just a 0% controls you put it back in that arena two days later and it still will freely explore the entire arena. So the animal that was exposed to 55% contrast looming stimulus, you can see spends much less time in the threat zone, and a decreased distance travel shows showing really strong anticipatory behavior, even to this really low contrast stimulus. So the fact that this was happening at this stimulus here at 55% where we weren't even seeing a detectable looming response glued us into the idea that perhaps this was occurring through a set a circuit that was actually separate from the looming detection circuit itself. Because we know that IPRGCs project widely to some of these non-image forming brain regions, we know that they signal over incredibly long time scales. This made us think that maybe IPRGCs were actually a candidate for sort of the visual inputs to modulate this anticipatory behavior over these sort of day long days long time scales. So we wanted to know whether IPRGCs are involved in this visual threat anticipation behavior. So in fact, if IPRGCs are involved in visual threat anticipation, then we should see deficits in this behavior when we use mutants where IPRGCs are dysfunctional. So we started with a melanopsid knockout animal which removes that really long term light signal that you see in IPRGCs, their intrinsic light response. And we simply compared their behavior during the test phase of this paradigm. So first I'm just showing you the control experiment where we looked at just exploration during the test phase of these control and melanopsid knockout animals. So that's after two days after they've been exposed to, in this case, no looming stimulus. And both of them explore the arena at, you know, they spend the equal amount of time in the threat zone and equal amounts of total distance traveled. But what about after they've actually been exposed to this low contrast looming stimulus? Remarkably, when animals lack intrinsic light sensitivity in their IPRGCs, you essentially lose the, they essentially lose the ability to anticipate the reappearance of this threat. So while a control animal will now avoid the center after it's experienced a looming stimulus and show significantly less distance traveled, you see that melanopsid knockout animals show increased time spent in the threat zone and increased total distance traveled, suggesting they have deficits in this anticipatory behavior, and that they are in fact required for this visual threat anticipation. And so we wanted to make sure that this wasn't actually just a visual deficit and animals ability to detect the looming stimulus. IPRGCs are not thought to be involved in the looming detection circuit, but we wanted to make sure that they could see the stimuli okay. And so we also looked at an animal's ability to just respond to the looming stimulus during the exposure to that looming stimulus itself. And we found that especially at these lower contrasts, which is where we're operating here for the behaviors I'm going to show you, you can see that the percentage of melanopsid knockout animals that are responding to the looming stimulus is similar. And the freezing time as the contrast increases between melanopsid knockout and their control litter mates was the same up until you get to these higher contrasts. So down in these lower contrast regions, we don't think that the differences between the two genotypes is a function of a decreased visual ability of some kind. We think it's a deficit in the actual anticipation of the threat, especially again down here at 55%. So one other thing I just want to touch on that I'm going to leave for a minute and then I'll come back to you at the end of the discussion. So the biggest part of the work is that we saw some really interesting sex differences when we looked at this behavior. So initially we had been grouping all of our male and female animals together. And that was where we first saw the deficit that I showed you, but Marcos then took and split his data by male and female mice and remarkably when he did that he saw that it was actually just the male melanopsid knockout animals. That showed that increased time spent in the threat zone so that actually showed the deficit in anticipatory behavior. Whereas female melanopsid knockout mice had absolutely no difference compared to their control litter mates when we just put them by sex. And so we started off by pursuing again the circuit in the male animals and then I'll circle back to the females, but this was a really good lesson for us and starting to split our data by sex. I would actually urge other people to consider doing the same because we're finding the more we look the more we find in terms of sex differences and this was a pretty dramatic difference for us to see and I'm glad that he thought to split the data and look at it in this way. So now I know I pair just these are required for visual threat anticipation we have this new behavior and we wanted to figure out where in the brain are the I pair just he's actually sending this information. That's tuning this anticipatory behavior. We had no idea where to start because we didn't even know what this circuit was and so we wanted to do a screen to try to identify IPRGC targets that might be involved and which brain regions we should start to try to manipulate with different circuit tracing techniques. So, to do this, we started off really simple with a stefos based screen so we put our animals through our visual threat anticipation paradigm they got either no looming stimulus or that low contrast 55% stimulus. Put them back in their home cage for two days again and then put them back in the arena for four minutes, like they were like we would do their test phase, then we just put them in the dark for 90 minutes, using as the animals and immunolabeled their entire brain for and then we looked at the different IPRGC targets throughout the brain with the idea that if a brain region is important for tuning visual threat anticipation, then the differences in seafoss activation should really mirror the differences that we see in behavior, where the 0% control and the 55% exposed melanopsin null animal will show similar seafoss, and then this exposed control animal would be different from the other two. For us, only one brain region showed this expected pattern of differences in seafoss labeling, and that was the perihabenular region. So here you're looking at seafoss density from immuno labeling and you can see that the melanopsin knockout and the unexposed control have similar levels of seafoss, but this 55% control has significantly less than both. The perihabenular, we're not familiar, is a region that was actually identified by David Berson and Samaritar's labs. It gets strong IPRGC input and it kind of borders the lateral perihabenular, but is actually a distinct region from the lateral perihabenular itself. But now we have a place where we know IPRGCs are projecting that seems to be differentially activated during this anticipatory behavior across our different conditions. So we wondered whether these PHB perihabenular projecting IPRGCs are required for visual threat anticipation. If this is the case, then we should be able to rescue our melanopsin knockout deficits by activating the perihabenular projecting IPRGC specifically, right? So we can specifically activate just those IPRGCs that go to the PHB and that should rescue our melanopsin knockout deficit if these cells are actually required. So to do this, Marcos used a GQ excitatory dreads based approach where he injected bilaterally into the perihabenular accredependent GQ dread virus into these melanopsin Cree-Cree animals, which are melanopsin knockouts. So by doing this, he's expressing the excitatory dreads only in the IPRGCs that go to the perihabenular. And then activate the dread receptors on these cells using a ligand called CNO, which is specific for cells that are expressing the dread receptor because these are excitatory dreads, CNO will actually activate those cells, okay? So, excitingly, when Marcos specifically activated these PHB IPRGCs, animals went from having, you know, this more, these deficits in threat anticipation where they're kind of crisscrossing the center to having significantly less time spent in the threat zone. Okay, so this is the rescue of the melanopsin knockout deficit by activating just these PHB IPRGCs. And we did this by first injecting CNO both when the animals were experiencing the looming stimulus, as well as when the animal was just put back in that context two days later. So then we started to wonder, well, when are IPRGCs required during this really long paradigm? Are they required when the animals forming the association between the context and the stimulus? Are they required or also, or instead required, I should say, when the animal is put back in that context and needs to recall that that's where it experienced the threat? So, to test this, Marcos did a really clever experiment where he just excited the dread receptors using CNO administration at either the exposure or the test phase to look at whether a single CNO application or injection could actually rescue the behavior. That was the exposure phase when the animals actually experiencing this looming stimulus for the first time. And when he activated the GQ dreads during just this phase, he saw no difference between a vehicle injected littermate and a CNO injected littermate. However, when he injected CNO at only the test phase and vehicle during the exposure phase, that was when he was able to see a significant rescue of the anticipatory behavior where the animal would spend less time in the threat zone, just like a control animal would, as long as we were exciting their PHBI IPRGCs. This tells us that IPRGCs are required for tuning visual threat anticipation appropriately. And the time that they're required during this behavioral assay is actually when they're having to recall that this is a context where I previously experienced this threatening stimulus, which I think is a really exciting nuance of this paradigm. So as a second test of whether this circuit was actually necessary for proper anticipatory behavior in visual threat anticipation. Marcos also did an experiment where he silenced the neurons in the perihibonyl that get IPRGC input, which are primarily the GABAergic neurons of the perihibonyl. And so in this case, he took the GAD2-CRE, this is a GABA synthesis enzyme, where GABAergic neurons in the PHB will express CRE, and then took a CRE-dependent inhibitory dread this time and bilaterally injected it into the perihibonyl. So now instead of rescuing the behavior, he's going to silence the perihibonyl. If the perihibonyl is required for this behavior, then what we should see is increased time spent in the threat zone and increased distance traveled in animals where the PHB is silenced because they are no longer anticipating that threat. And that turns out to be exactly what he saw in the males. So you can see here, the vehicle injected animals versus animals that had the inhibitory dreads activated by CNO. And so when these PHB, when the PHB neurons are inhibited, you can see increased time spent in the threat zone, increased distance traveled. These animals have deficits in their anticipatory behavior. So now we know the PHB itself is also necessary, which it should be if the PHB, IPHB-GCCs are involved. And so, yeah, and this is just to remind me to remind you that this deficit is mimicking what we see in the melanopsin knockout animals. So silencing that region mimics what we see when we take away the melanopsin from the IPHB-GCCs. So now we know that IPHB-GCC input to the GABA-ergic neurons of the perihibinol are really required for proper anticipatory behavior of these male animals. And so then we wondered what neurotransmitter IPHB-GCCs might be releasing at the perihibinol to actually tune this behavior. In work I didn't introduce today, we have previously shown that IPHB-GCCs are not only glutamatergic, which has been well established by many labs. But a subset of IPHB-GCCs also releases GABA, and we know that they do this in STN and likely in the OPN to modulate photo entrainment and pupil constriction. But we haven't identified other behaviors that these cells are involved in as of yet. And a postdoc in the lab, Chen Liang, was able to use a mouse genetic reporter to show that these GABA-ergic IPHB-GCCs do send a projection to this perihibinol region. So we wanted to see whether they were releasing glutamate or GABA or maybe both to actually impact this visual threat anticipation behavior. Lucky for us, there are conditional knockouts that allow us to remove either glutamatergic signaling by knocking out V-glute 2 or by removing GABA synthesis by knocking out GAD2, specifically in the IPHB-GCCs. So we have conditional knockouts that are OPM4-cree V-glute 2-flocs-flocs or OPM4-cree GAD2-flocs-flocs, and these will remove the glutamatergic or the GABA-ergic signaling from IPHB-GCCs, respectively. So we simply started by just taking these animals and comparing their threat anticipation to that of their control littermates. And really exciting, and this is, again, in-mails. And really excitingly, we saw that, first of all, knockout of this particular glutamate transporter V-glute 2 for IPHB-GCCs caused a decreased anticipatory behavior in the form of increased time spent in the threat zone. So this mimics what we saw in the melanopsin null animals, right? The animals are spending more time in the center during that test phase two days later. Whereas these GAD2 conditional knockout animals actually showed enhanced anticipatory behavior. So they spent less time in the threat zone than did their control littermates, suggesting that GABA signaling from IPHB-GCCs is involved in tuning this behavior in kind of the opposite direction to glutamate release. Now we know that IPHB-GCC glutamate and GABA are differentially tuning VEDA behavior. We are, at this moment, completely assuming, I should say, that both of these signals are going through the perihabinula. But that's kind of our current working model that we would like to test more specifically. Okay, so now I want to circle back to the sex difference. I remember I told you that this sort of increased time spent in the threat zone, this deficit in anticipatory behavior, was really only in the males when we split by sex. And the female melanopsin knockouts didn't show any difference compared to their control littermates. But Marcos started to wonder if maybe this is something that females could be there, but maybe it's just something we're not seeing because we're not looking across the estrus cycle. He did what was, I think, kind of a heroic experiment where for every single female he tested, he went through and retested a bunch of females, or tested a bunch of new females, I should say. He took vaginal swabs and looked and identified the stage of estrus of that female across every single day of the paradigm. So across all four days for every single female. And he classified each female as either sexually receptive or not receptive on the actual day of the test itself. If he did that, that's where we actually saw differences between the knockouts and the controls. So sexually receptive females, we saw no significant difference in terms of the time they spent in the threat zone, which is already really low. However, for these non sexually receptive females, Marcos actually saw a significant decrease in the amount of time the animal spent in the threat zone. And if you're feeling a little confused, it's because for the males, the phenotype was the exact opposite. For the males, not often knockout animals spend more time in the threat zone, or non sexually receptive females, they actually spend significantly less. Not only is there a sex difference, but the effect is the complete opposite between these two different groups. That's dependent and estrus dependent, just to make it like super complicated, but also very interesting. Okay, so is this effect in the females happening to the same circuitry as the males? Or is it something kind of completely different? So we started with some of the parallel experiments to the ones I showed you in males. And the first was to try to rescue the behavioral deficits we see in females. Oh, that's the opposite graph. I should ignore that graph. That's like the male graph, not the female graph. If this is important for females, then activating these PHB IPRGC's with GQ dreads could rescue the anticipatory behavior by actually increasing it. And so that turns out to be exactly what Marcos saw. So if you activate these PHB IPRGC's in females, they go from avoiding the center of the arena during the test phase to now kind of crisscrossing it a little bit more than they were. Right. So you have this significant increase like we saw for the female or to rescue the female deficits and threat anticipation. But this means that the same IPRGC PHB circuit is functioning in males and females to differentially drive this anticipatory behavior. So this was just a reminder of what we saw in males is the actual correct graphs for males here. So you have the rescue for males resulted in sort of a decrease in time spent in the threat zone back down to control levels. Rescuing this melanopsin knockout deficit that we saw in the males. And I should also mention that we see the same effect when we silenced the perihabinula where the females in the males go the opposite direction. So what we've been able to identify is a new circuit where IPRGC's are involved in anticipatory behavior where they're anticipating the reappearance of a threatening stimulus where they're in an environment that they previously experienced that threat before. And we're seeing that in males and females the IPRGC seemed to play different roles in tuning that threat anticipation and in the males we have glutamate and GABA release presumably in the perihabinula region to differentially tune this behavior. So we're really excited to figure out what's downstream of the perihabinula. Marcos is doing a lot of work to figure that out to specifically look at releasing the perihabinula and even record in people activity to see what's happening when the animals behaving at these different phases. So now I just want to switch gears and tell you briefly about a second project actually also from Marcos where we're looking at IPRGC diversity. So I've had a like since the beginning I feel like I've been interested in IPRGC diversity and sort of what are the properties of these cells, what are their roles and behavior. I've always been interested in what actually drives that diversity developmentally what makes these cells diverse and then just generally for neurons you know what are the things that drive subtype the development of subtype defining features for different neurons. So we got interested in this question for IPRGC right so we know that they express melanopsin. But we also know that there are, you know, at least six subtypes of IPRGC that have been identified. These cells differ across their physiological features in almost every measurement you can make there'll be a difference across the subtypes. They differ in their melanopsin expression level so M1 cells have the highest melanopsin expression levels, then M3 then M2 and then M4, 5 and 6 have the lowest levels of melanopsin expression. They differ in their morphologies so M1 cells are the stimplest and have some of the smallest dendritic arbors, whereas M4 cells have the largest and most complex dendritic arbors. They differ in their stratification and one cells are off stratifying and twos and fours and fives are on and threes and sixes are by stratified here. So lots of different cells with lots of different properties and what I like about using IPRGC as a system for understanding cell diversity is you have represented here. A lot of the different features that you'd find in a given ganglion cell. You know, and then you have the fact that IPRGC is project so widely throughout the brain, driving a diverse array of behaviors that we can measure really nicely so we are really poised to figure out a lot of things about IPRGC development if we can start to get it. What the underlying genetic programs might be. We also know that IPRGCs are transcriptionally diverse so work from Nick Tran and Josh Sainz's lab has shown that you know multiple different IPRGC clusters fall out when you do 10x single cell RNA sequencing experiments. Our collaborator UA Yang has performed a dimensionality reduction and then just is shown we're showing here just the IPRGC clusters kind of remapped and I've moved the M5 from in their paper where they put it so we think it's down here based on some of Greg Schwartz's data on the Pixar cells to the M2, the M1, M2 and M4 clusters here and then kind of provisionally labeling these clusters but we don't have any verification of that yet. So they're also not only morphologically and physiologically distinct, distinct in their projections but also in their genetic programs and the genes that they're expressing. So how do these diverse IPRGC properties actually develop. One thing that interested me right from the start was the fact that brain through be which is a transcription factor important for ganglion cell specification actually distinguishes different subsets of IPRGCs, and it does so in the adult. So we know that brain through be positive IPRGCs include all of the non M1 IPRGCs which are thought to be involved in pattern vision, they all express brain through be as do a subset of the M1 IPRGCs that are absolutely required for pupil construction. And then you have a small subset of about 200 M1 cells that are brain to be negative that we know are sufficient to drive circadian photo entrainment. Give us really early express transcription factor that's for some reason being maintained into adulthood in a large proportion of IPRGCs. And so we wondered whether brain through be expression levels might correlate with some of these features that really define the different IPRGC sub types. And so this work was started by Stokely and then has been taken over since by Marcos. And she started looking at whether these IPRGC properties actually correlate with brain to be expression levels. So here's what she did first was just label for the brain to be protein using an antibody stain. And she found the brain to be fluorescence intensity is really not a binary thing so we have to present it when we talk about the mouse genetic models as like they have brain to be or they don't. And I guess that's true, but the ones that have it have a pretty broad range of amounts of brain to be protein. You can see generally the M1 cells have less than the non M1 cells which are just plotted here in pink. But really interestingly, when, oh, and this is just like a great to be negative cell, a weekly expressing cell and a highly positive M1 IPRGC. But Stokely actually filled each of these cells and looked at their morphological properties as well. And she found that among the M1 IPRGC that she filled there was actually a correlation between the amount of brain to be and the size and the complexity of the dendritic arbor just within this one sub type. And we can quantify that here so she's looking just at the correlation for the M1 cells we just plotted the non M1 for kind of reference, you can see dendritic length and dendritic field both significantly correlate with the amount of fluorescence intensity within the M1 cells themselves. So this was a clue for us that bring through be really might be driving some of at least these morphological features for sure within the M1 cells and then maybe across some of these other IPRGC sub types as well. So we have brain to be expression having this nice positive correlation with dendritic arbor size and complexity. And we also, she also observed I'm not showing you the data here that the brain to be negative M1 had the strongest levels of melanopsin expression and actually the highly brain to be positive M1 cells had lower melanopsin expression. So she's observing this interesting inverse correlation between melanopsin and brain to be as well. So now we know this is happening, you know, in a correlational manner with multiple features of the M1 IPRGC is. But what about the other IPRGC sub types. Okay, so we know that these are all brain to be positive we're just like vague about how much brain to be right. And so do we see the same relationship. So first, what you a did was just look at the levels of melanopsin expression using the same data set versus brain to be expression and you can actually see this really nice inverse correlation between the different clusters of melanopsin and brain to be that you know, they're nicely inversely correlated. So if you look at just the properties that we already know of about M3 and two and M4 IPRGCs, we actually know that the dendritic harbor complexity increases the degree of synaptic input these cells increases and the melanopsin levels decrease across these different sub types and we know that the different levels are increasing from M1 and three and two and M4. So some really interesting relationships started popping out across a lot of different features that got us excited about the idea that bring through be might be a really important regulator of the development of IPRGC sub type features. So this is when Marcos comes in and what we really wanted to do was test this in a causal way instead of just to find you know identifying these relationships is brain to be actually causally affecting these different properties. So we worked to in collaboration with a neighboring lab, Dr. Yu Yang and then his student Omar and a staff scientist in the lab to Moco, and actually another student just joined on Shin Chiro who I don't have pictured here. And so we wanted to know whether altering brain to be expression in the IPRGCs will influence the development of their properties and we kind of started really broad with the idea that like, it could be physiological morphological melanopsin genetic programs. So all these things we know differ. So we want to test them all and what we did what we decided to do to test them was to either remove brain to be completely from all the IPRGCs using a conditional knockout approach with a mouse line from Tutor Bedia's lab. Or we generated a new mouse line that would over express brain to be in all IPRGCs by expressing brain to be in a pre dependent manner at the Rosaloca so now we can dial brain to be all the way up or all the way down in IPRGCs and we can see if we see a change in the features. And the really simplistic idea we had from this was that if we take away brain to be the features and the genetic programs and the morphology of IPRGCs would shift toward the M1 cells. And if we dialed up brain to be the features the genetic programs and the physiology would all shift toward, you know, more toward the M4 cell population. And so, like I said, this is really ongoing work and I'll show you some of the data that we have so far. So, first, we wanted to look at melanopsin expression in these brain to be conditional knockout and conditional over expressing lines. And so we used RNA scope fluorescent into hybridization to look at M RNA levels for melanopsin and brain to be in individual IPRGCs. First here on the left is brain to be and you can see in the conditional knockout we have almost no brain to be per IPRGC which is good that means our knockouts working, and you can see a significant increase in brain to be in our over expressing line. And then you can see that consistent with our hypothesis. Melanopsin goes in the opposite direction of brain to be right so if we knock out brain to be we see a significant increase in melanopsin mRNA in individual IPRGCs. And if we over express brain to be that suppresses the melanopsin expression levels and it's significantly less than it is in control. Now like melanopsin goes all the way to the ceiling or all the way to the floor, but it shows us that brain to be is tuning the level of melanopsin, which is a really important subtype defining feature of the various IPRGC subtype. So we've also looked at protein expression to see if this is mirrored and antibody standing of individual IPRGC is again and you can see that when brain to be is removed we have an increase in melanopsin which is shown here in the middle panel. And when brain to be over expressed we have a suppression of the levels of melanopsin, an individual IPRGC is which is easy to appreciate from this image here on the right. This is the brain to be an important modulator of melanopsin expression levels in IPRGCs. What's actually really cool, but I have time to put it in here is that even if we remove the brain to be or over express it in adults, so we can do this at P 60. We actually see similar changes in the melanopsin expression levels so it's not just a developmental thing but it's something that's being sort of actively maintained by brain through be in some way in the individual IPRGCs. This is a pretty targeted approach looking at the melanopsin expression levels just because we know it's a gene that whose expression differs across the subtypes. So let's take a more unbiased approach and just look at the genetic programs of IPRGCs more broadly and see how those might change when we knock out brain through be. So, these are just the re plotted UMFs of the subtypes and you can see if we just plot melanopsin expression again this is from the same data set which has just been like the most useful thing for like there are research lately and it's such a gift to the genetic in general, but you can see the melanopsin levels and the brain to be levels the inverse correlation is really evident when you look across the IPRGC clusters. And then what's plotted down here in the cyan are sort of the top genes expressed in the low brain to be IPRGCs so m one and m two clusters you can see are more enriched for these genes. And then here you guys just plotted the top brain to be high defining genes here in this pink color just value important later. And then we decided to compare the genes that were being actively translated in IPRGCs in control animals and animals were brain to be had been removed so we use a trap seek approach which I will not go into the details of but if you're interested I can talk about after the fact to just look at bulk IPRGC gene expression in our conditional knockout and our control animals. And that is there's that and then what we see is that where did my graph go thank goodness. What we see is that the, when we take away brain through be the gene expression programs that are enriched in these brain to be high cells decreases in just the whole IPRGC population. So the cells overall are becoming less and for like, when we take away brain to be. In the meantime, IPRGCs in these brain to be conditional knockouts are also turning up genes that define those brain through below IPRGCs. So in other words they're not only becoming less and for like, but they're also becoming more and one like the gene expression programs are indeed shifting more towards that of the m one the m two cells rather than the m four cells in the absence of brain to be. So what I was telling us is that brain to be is not just affecting Melanopsin expression but it's actually potentially a sort of a master regulator of the molecular identity the genetic identity of these different IPRGC sub types. So we're going to do follow up experiments in the future with NIH is blessing hopefully someday here to like, look at how this is changing in a really nuanced way and try to understand what the actual genetic programs are that are underlying this gift. So that's sort of the genetic expression in these cells but does removal of brain to be actually shift some of the morphological and physiological properties of IPRGCs. So we've started with morphology for now and we've done this for brain to be conditional knockout animals. Again, the idea is you take away brain to be and everything's going to shift from the sort of large complex dendritic arbors to the simpler, smaller dendritic arbors to become more and one like. So, Marcos has been able to do this using sparse viral labeling of m four and m two IPRGCs and we can of course distinguish these two based on whether they express this marker SMI 32. So m fours are SMI 32 positive and m twos are SMI 32 negative. Both are on stratifying. And so we compared m four and m two cells. We kept the region of the retina constant in control and brain to be conditional knockout animals. And really excitingly the m four cells in the brain to be conditional knockout and the m two cells you can see both showed smaller and simpler dendritic arbors compared to their control counterparts. So here's a more expanded analysis for the m four cells you can see the show analysis showing decreased dendritic arbore complexity, significantly smaller dendritic field diameter. We have trends, you know, this is still a data set we're collecting towards smaller total dendritic length and even smaller soma diameter when we take away brain to be and m two cells likewise show a significant decrease in dendritic field diameter. So, this also seems to be important for the morphological developments of these cells. What I've shown you so far is that brain to be is really shaping, you know, some of these subtype defining features of IPRGC is we have brain to be levels varying and a predictable gradient with IPRGC morphological and functional properties. They're inversely correlating with Melanopsin expression levels. And then when we manipulate brain to be expression which I should mention is happening post mitotically we've confirmed that use an experiment that enough time to show you this alters the Melanopsin expression IPRGCs. It's important for the molecular identity in shaping the molecular identity of IPRGC subtypes, and it's influencing dendritic development and even soma size which I didn't have time to include today. And so, lastly, in the last three slides, I just want to talk about the projection patterns and behavior as well, because we have really differential innovation of various brain regions based on whether IPRGC is a brain to be positive or brain to be negative and of these different subtypes. So what Marcos did was, use a pre dependent virus to label IPRGCs so that we can trace their axons to the brain. And then he looked at the contralateral and ipsilateral opn and also dlgn and then looked at pupil constriction, as well as contrast sensitivity. This is the control animal you can see the contralateral you have the core is really nicely filled here of the opn and then ipsilateral you can see the shell. And when brain to be is removed. Here you can see that you have a kind of a disorganization of the contralateral side but I think what's really interesting is this ipsilateral projection where you can see, you've essentially lost the fell of the opn projection and now it's almost like you have another projection to the core on the ipsilateral side. And so Marcos has done some people constriction experiments as well and so if you look with increasing light intensity at the pupil area, which of course gets smaller as the people will constrict you can see at bright light intensities. So the animals show a significant deficit in pupil constriction where they don't constrict their pupils as much as litter mate controls. It's not a massive deficit but there's definitely still a projection to the opn, but there is a significant decrease in pupil constriction. So what about an image forming brain region so Marcos has also mapped the projections to the visual thalamus the dorsolateral geniculate nucleus you can also see here IGL and VLGN. The ipsilateral projection and then of course or the sorry the contralateral projection and then you of course have this ipsilateral projection here as well. And so, interestingly, in brain to be conditional knockouts, and we know that these brain to be positive cells go to the DLGN, you can see a marked reduction in the projections to the dorsolateral geniculate nucleus on the contralateral side here. So you have decreases in projections to this image forming brain region. We haven't been able to do a test of conscious visual perception yet but we have done a test of reflexive vision which, in our experience and other models near they tend to mirror each other in their deficits. And we've looked at contrast sensitivity using this optokinetic tracking paradigm from Glenn prosky's lab the optomotory system. And in that assay we see a significant decrease in the contrast sensitivity of brain to be conditional knockout animals where I peer to see is lack brain to be compared to their control litter mate. So we're going to follow up on this and do some tests of conscious visual perception as well but based on this I would expect to see deficits especially based on the anatomy that we're seeing with these animals. So it seems that I peer to see or that the brain to be in the I peer to see is also really important for you know their projections and their behaviors as well based on these initial experiments anyway. And so with that I want to stop and take any questions and first want to just thank people that actually did the work so here's Marcos super talented to super interesting projects and it's been just so much fun. And so exciting working with him on both of these. This is Lucy Lou one of the undergrads who helped with the project, Chen who showed the projections to the perihebene as well as the rest of the lab and our collaborators here. And of course I want to thank our funding as well. I'll take questions. Thank you. Thank you very much Tiffany for this amazing work, both in quality and in quality and you know working with non model space myself I cannot help but be very jealous of the resolution you have when it comes to, you know, manipulating things and checking what is the effect of this and that. So, let me go ahead and post the zoom room link for the audience in case they would like to start joining us already. And as a reminder after a first round of moderation of online moderation of the questions that appear, we will be continuing an offline fashion in the traditional get together informal zoom gathering that we have. So, waiting for questions to appear the first one that I have with respect to the second part of your talk because you said that it's not a developmental stage specific effect this dynamic balance between the two. And it also happens like postnatally, would you expect, like what would you expect the role of such active dynamic balance between the two would be in terms of melanopsin and beer and the factor the other. What's the role of that will be. Yeah, like to actively like the ongoing. Yeah, I, that's a good question. I don't know if it's like if it's helpful to the cell to have that dynamic balance or if it just like is and maybe that's what kind of gives it sort of the fine tuning. I would love to know if how important is the level of melanopsin expression for say like an M4 cell like if we over express melanopsin in that cell. I wouldn't do it using the brain through B model but if you just like specifically over expressed melanopsin like what would does is it worse like the contrast sensitivity of the cell or the animal actually affected. So if it's predictive it needs to be this well regulated then it probably matters a lot for the function but we haven't tested that yet. And, but it's a really interesting question. Yeah, because like I think that like what we know from classic tech development is that you know the parts kind of get into one trajectory and that's it. They cannot actively maintain a balance so yeah it's very interesting. So it's a cool example of like I think recycling right like, because you have this pathway that was important to turn things into a ganglion cell right and that was kind of what bring to be was thought to do, and then it's like all right now I'm going to use it for all these other things and just keep it turned on into, you know, adulthood, and I just love the repurposing. Yeah, going back to the first part of your talk about the threat anticipation. I was wondering and maybe like my naive understanding of how seafoes protocols work will be will be making my question very stupid but if I understood correctly, you manage to identify the perihabenular region based on the seafoes activity right, but if Gabba was the primary driver of such behaviors. The fact that seafoes only captures the activation, would it make like, would you miss some territory some brain territory some brain region. If Gabba was behind it. Yeah, I think we're very lucky that that worked. Actually, I couldn't believe that it did. I mean it was handy that we saw the patterns right but then it was like well unless we see a functional confirmation of this like I wouldn't just trust the seafoes and everything with the genetic manipulations mimics sort of the vglute to knock out and the melanopsin knock out and so that's the thing we don't quite know yet is whether the Gabba signaling is happening at the PHP and we have to do some fancy tricks to try to just knock it out in those PHP IPR to see if we want to get at that question. So we could be missing something, the things we did after that make me comfortable that it's like a major region but yeah for the Gabba release I agree and it would be hard to see the seafoes if it's inhibiting but then theoretically the other like areas would have had the opposite effect so we might have still been able to pick it out just based on the comparisons with the different genotypes. Yeah I see what you mean. Thank you. So the first question that appeared in the chat is from David Berson who uses a very nice interplexiform name for his YouTube account so he says fabulous talk if I remember there should be many other types of BRN3B expressing RGCs which are non IPR disease. Do you think the gene plays the same role there? I would love to know the answer to that question and I think if we, I think one thing I would love to do in the future, I just want to make sure we do it in like intelligent way in terms of which cell type we would pick and we would need a driver to do it would be to overexpress or knock out BRN3B in some other types of ganglion cells and just see if we can drive expansion of the arbor complexity by overexpressing BRN3B or something like that. It may or may not work. It's hard to, without genetic drivers it's harder to do depending on which cell type it is but yeah it could be right. It could be if it's kept on I would imagine it would do some of the similar things. Ponging back to the first part of your talk. So my question is like, because like this looming stimulus has a certain speed right? Would you expect a tuned curve of speed preference for these cells that drive this behavior? So like if you drive it at different speeds? I don't know if Marcos tried different speeds. He tried, this took us about a year to convince ourselves this was like a real thing and make sure that the paradigm was like repeatable and clean. So we haven't done like a curve of that like we have for the contrast because every animal we can only do one set of contrast and so we've generated kind of a full curve at this point for males and females at both stages of estrus actually. And so like, that's been huge, a huge data set to collect, but I think there's probably a lot we could tweak to try to figure out like, is there any difference in the preference of the strength of the inhibition or of the anticipation. Right. Thank you very much for addressing this. I have one last question and if there are no more questions appearing in the chat, given that people are already waiting in the zoom room, we could terminate the live broadcast and continue offline. So if no more questions appear after my last question, I will be doing that. So my last question is like because you quantify like in terms of behavior, how many times, or not sure, not how many times, how much time the animal would spend in the zone. Have you tried to normalize it by how many times it enters because it could be entering working around and spending some time or staying even stationary inside, but this could be happening once, while other animals might not like might do it 10 times, you know, compared to like control versus that. So Marcos has actually just popped into the zoom room so we can talk to him more about this offline. Like he's definitely done a much more comprehensive analysis and we're going to go back and like do some of this more nuanced analysis on the behavior just to try to see like what other parameters might be important. I know he did a principal component analysis to try to pick out sort of the top behaviors and those two were two of the main ones that we that we actually mentioned that popped out as principal components and those are the ones I presented here. But that would be one that I could imagine maybe being different as well. Right. Okay, thank you very much for addressing this. Thank you very much for this fantastic talk again and I'm so happy that Marcos is here so we can continue offline. And again, I would like to remind to our audience that if they want to be part of this offline discussion, just to make sure they click on the link that is available. Thank you very much.