 Hello everyone, sorry for the delay. Welcome for the second episode of this third season of Vision in Success. This is an online web seminar series for vision and visual neuroscience topics. These talks, as you know, are part of the Worldwide Neuroinitiative, which as you should know by now, is a platform for neuroscientists in various fields to exchange on their most recent work. And we do so while we cannot still efficiently travel to conference or want to roll. My name is Maxime Zimmerman. I'm a former PhD in Tamba de Sade, and today I'm very glad to receive Steven Massé. Steve received his PhD in neuropharmacology from the School of Pharmacy at London University in 1990. He then moved to the US and completed postdoctoral work, first with Dinah Redburn, working on cholinergic macrosales in the retina at the University of Texas, and then with Robert Miller, studying glutamate receptors in the retina at Washington University in St. Louis. In 1995, Steve joined the faculty at the University of Texas, where he's now a professor and researcher director in the Ruse Department of Ophthalmology and Visual Science at McGovern Medical School. Steve's group studies road and con pathways in retinal secretaries, and the current focus right now being on photoreceptor coupling via gap junctions. And that topic we're going to assess today. So hello, Steve. Thanks again for being with us today. How are you doing? Okay, thanks, Max. Very nice introduction. It's great to be here. Thank you very much to the Tombard and Lab for putting these talks together. We've all really enjoyed them. The floor is yours. Whenever you're ready, you can start sharing your screen. Okay, I'll share my screen and we'll get going. We cannot lost your presentation here. Okay, I've lost it too. Sorry for the short delay. In case it's approved that we're doing those talk live, but I'm glad to see that there's a lot of people in the chat with us today. Steve, your mic is off. If you can just turn it on and I give you the floor. Now, you just remove your lizard. You had it. Mala Filler is giving you a wish. Okay, I think we're ready to go now. You got my screen and you can hear me, right? Yep, we can. Okay, good morning, everyone. Great pleasure to be here. This talk is going to be about work that I've done with Christoph Ribeliga here at the lab in Houston. This picture here shows the seven sisters. I grew up not far from the Brighton lab in Sessex and the South Downs are just marvelous for hiking. If you go visit Leon and Tom, you can walk up here and hike around. It's fantastic. I highly recommend it. I put this in here because Dr. Richard Ruiz was the founding chair of our department and he passed away last week. He was great for our research group. He supported the basic science research group in ophthalmology, studying retinal circuitry for more than the last 25 years, and we certainly appreciate his support. So in memory of Dr. Ruiz today. Okay, I wanted to put this up to acknowledge the people who've worked on this project. We've had a lot of help from people around the world. Specifically, I want to draw your attention to a couple of people. In Christoph's lab, Nanga made all the recordings and Jay raised the mouse lines. Joyce in my lab just all the immuno. Nori Ishibashi does the image analysis. These are my faculty collaborators. When it was time to do the serial block face, we had a lot of encouragement from David Burson. Christian Barron's shared data with us. Josh Singer, Catherine Morgan, and Sue H. Provider data sets for us. Their connection 36 flocks line was originally made by Friso Postma in David Paul's lab, and we got Rod and Cone lines from Jason Chen. So particularly, I wanted to mention Christian Barron's and Philip Barron's for sharing their data with us and helping us to get started with this project. The E2006 data set that I'll show you later is from Maurice Helmstadter and is hosted at Max Planck. And these are our funding sources. So today, we're going to talk about photoreceptor coupling. So if we're going to talk about gap junctions, we have to have a little introduction. So a gap junction is an electrical synapse. They're made of connections. Six connections make a connection or a hammy channel. And two hammy channels dock to make an intercellular pour. Now, the docking is required to make a gap junction. If there's no connections on one side or the other, then there'll be no gap junction. You must have connections on both sides. And then they frequently make a raise, something like this cartoon drawing. But these gap junctions can have a variety of shapes. The connections are 20 genes in the mouse retina, but primarily today, we're interested in connection 36. And that's because connection 36 is one of the relatively few neuronal connections. Actually, it's the dominant neuronal connection. And in retina, it's famous for its presence in A2 amicron cells, as well as photoreceptor coupling. Now, some of the functions of electrical synapses, they make coupled networks, and they make direct pathway connections, like the A2 to comb bipolar pathways in the inner retina. And they're responsible for signal averaging, noise reduction, synchrony, things like that. So this is our introduction about gap junctions. Today, let's think about a couple of retinal circuitry questions. We know that the cone connection is connection 36, but there's been some controversy about the identity of the rod connection. And then we'd like to know what's coupled out of the photoreceptors. This is going to be rod, rod, rod, cone, cone, or all of the above. Can we find any rod, rod coupling? It was really reported by Tsukamoto and Peter Sterling, but we'll look for that today. And what's the function of photoreceptor coupling? And this is what we're going to do today. We're going to have some confocal microscopy, some mouse genetics, some patch clamp, some EM, and some RNA seat. So that's what's coming up. So we can flip this around where we do this work because we're interested in retinal circuitry, but these circuits can also tell us quite a bit about gap junctions, and we're interested in that aspect, too. Gap junctions make these electrical synapses, but we all know that neurotransmission is really dominated by chemical synapses. And the electrical synapses tend to be overlooked, and that's partly because they're not very well understood yet. Notably, they're mostly absent in these large-scale datasets that people have produced recently. So these are the things that we need to know about gap junctions if we're going to understand them a little better. So connectivity, size, et cetera, et cetera. In particular, we'd like to know what they're going to do. We'd like to know what the open probability is. That's how many channels out of a gap junction plaque are actually open. And usually, we think of that as a very low number, sometimes less than 1%, and I'll come back to that later. If we know the open probability and the size of the gap junction, we can calculate the minimum and maximum conductance, and then we can get the modulation depth. So that gives us some idea about the plasticity provided by gap junctions. And if we know these things, then we can plug these numbers into neuronal models and try to make some sense of the circuits that use gap junctions. So these are the things that we're going to do today. We're going to find the location with confocal microscopy. The expression will be mass genetics. The connectivity will be serial block face. The conductance and the pharmacology from the paired recordings. And we'll be able to find the size from some i-resolution FIB data. Once we know all this, we'll be able to calculate some of the other properties that we'd like to know about. So this picture on the right actually shows connects in 36 and the mass retina. And you can see that the connects in 36 is restricted to the two plexiform layers. In the inner plexiform layer, there's a lot of big gap junctions. And those are predominantly A2 gap junctions, which are very well known. Out here in the outer retina, you'll notice there's a lot of very small little dots. These are tiny gap junctions. And that's what we're talking about today. Okay, this is a low-power EM picture, but it's just to show you what the OPL looks like. There really are two major structures. These big structures are cone pedicles. They're very recognizable. And above and between the cone pedicles are these smaller round rod spherules. And of course, the rod spherules vastly outnumber the cone pedicles. So this is in the OPL between the outer nuclear layer and the inner nuclear layer. And this was taken from E2006 from Helmstadter again. Okay, confocal microscopy. So if we stain the cones with an antibody against cone arrest and we can see the pedicles, and you'll see an exuberant spray of fine dandrites leave the cone. The cone pedicles. And the Konexin-36 is really co-localized with the telodandrea from the cone pedicles. Now, there's a little Konexin-36 staining underneath each cone as well. And that's not going to be part of our interest today. Those are actually on bipolar terminals. And we know this because if we look at the co-localization, you can see that all these, now shown in white, are co-localized. But the stuff that's underneath the cone terminal is in a separate location. And really, I won't be talking about that anymore today. What we're interested in is the cone telodandrea. Okay, and then when we turn on the rod spherules, which we can stain with the glutamate transporter, and we start to look at this, you'll see that all these co-localized Konexin-36 stats that we know are on the cone telodandrea, they're all reaching out to touch individual rod spherules. And so this is really our first idea that there's a lot of rod-cone coupling present in the outer plexiform layer. So these are individual rod spherules, and let's visit that in a little more detail. And then we get this kind of picture. Here's some rod spherules, and now they've been outlined with a dotted line. Here are the cone telodandrea decorated with Konexin-36, and they're at the base of each rod spherule. Now, a rod spherule has two big holes in it when it's stained this way, and we want it to be sure of what the structure was. So let's look at that in a little more detail. Here's a single rod spherule that we're focused on. Now, if we stain for M-glue R6, which is present on the tips of rod bipolar dandrites, we see that the bottom compartment has these M-glue R6 structures. So this is the postsynaptic compartment. When we look over here, we see that the top hole has a large mitochondrion in it, stained for Tom 20. Here's the ribbon between the two synaptic... Excuse me, between these two compartments. And here we have mitochondrion ribbon and M-glue R6, and we put all this together. So that's really what a rod spherule looks like in these preparations. If we take 18 different rod spherules, and we cut them out and stack them up and align them and then signal average across that group, this is what we find. It's a single rod spherule. All of the connects in 36 is found at the base by the opening to the postsynaptic compartment. The cone telodendria are gathered at the same place. While if we put a spline curve around there and plot it in a linear fashion, then we get this data. You'll see that the synaptic opening is shown by a drop in the glutamate signal. So we know that the opening to the synaptic invagination is right here. And you'll see that this is where all the connects in 36 is in two peaks because we crossed over the synaptic opening. And what I really want you to notice is that really all of the connects in 36 is within a micron of the opening to the postsynaptic compartment of a rod spherule. When we count all these numbers up, we get the distribution of gap junctions at each rod spherule. And the numbers run between one and six, but the mean is about 2.4. So two to three gap junctions per rod spherule all within a micron of the synaptic opening. And I want you to remember that within a micron. We'll come back to it. Okay, if we think about the cones, here are some stained cones. They're all stained with cone arrestin, but the cone in the middle is also labeled with GFP. And if we put this into motion and just concentrate on the GFP label cone, now we're going to turn on all the gap junctions. But really we only want the ones that are co-localized with this cone pedicle. And it looks like that. And then once we've done this, we can have the software tell us how many connects in 36 clusters are on each cone. And the answer is about 50. So approximately 50 connects in 36 clusters per cone. So now we know about the rod end and we know about the cone end. Okay, let's switch gears now to some mouse genetics and worry about connects and expression. Here are the, here are the mouse lines. Here's the rod line crossed with a reporter and here's the cone line. So when we look at the rod line, the reporter is shown in green and basically all of the rods are labeled. There's a bunch of holes up here, high in the outer nuclear layer and they're all filled with cone arrestin labeled cones. When we look at the cone line, the cones are all labeled for the Cree reporter and they're all double labeled for cone arrestin. So basically these two lines will let us label all the rods are all the cones. So we can use these lines and fax sort the cells and then use the cells for RNA seek. When we do that and look for the connections out of the whole family of connections, the only one we find is connection 36. So what this tells us is that both rods and cones express connection 36 and no other connections. Okay, we can use these mouse lines in the rest of our work, specifically if we cross them with the flux connection 36 line, we can make rod or cone specific connection 36 knockouts. And we need to think about this a little bit to think about what the consequences are. So if we knock out connection 36 in cones selectively, then it's rather obvious that cone-cone coupling is going to be deleted. But so is rod-cone coupling because you need a connection on both sides. The rod-rod, if there is any rod-rod coupling, should remain. Conversely, when we knock out connection 36 and the rods selectively, if there's any cone-cone coupling, it should remain. But the rod-cone will be deleted because you need connections on both sides. And obviously the rod-rod will be deleted. So we can use these mouse lines as tools to try to figure this out. Okay, this is going to show connection 36 immunolabeling in the retina. And here's the wild type. And lots of labeling in the inner retina. It's very bright and faintly labeling in the OPL. Here's the pan connection 36 knockout. And there's no connection 36 labeling in the OPL or the IPL. These structures are blood vessels. Now let's go to the cone-specific knockout. And first of all, you can see that the IPL is relatively unchanged. That tells us it's really a cone-specific mutant. But in the OPL, there's almost nothing. It looks like the pan knockout. And in the rod-specific knockout, again, the IPL is unchanged. But the OPL, there's almost nothing. So our interpretation of that is that because connections are required on both sides of a rod-cone-gap junction, if you knock out either side, you remove nearly everything in the OPL. Our interpretation is that means that nearly all of those gap junctions are rod-cone-gap junctions. If we look at this in whole mount, this is a whole mount preparation showing a field of cone pedicles. And really, they have exuberant cone telodendria that blanket the OPL. If we turn on connection 36, you can see really there's a lot of it and it's co-localized with the telodendria, as we would expect. Okay, now let's look at the cone and rod-specific knockouts. And in the cone-specific knockout, there's really nothing. In the rod-specific knockout, which should reveal any cone-cone connects in 36, there's a few little dots left. And that might be a little bit of cone-to-cone coupling. When we quantify this, here's the wild type and we calculated this as the number of gap junctions per cone. There's around 45 or 50 connects in 36 plaques per cone pedicle in the wild type. In the pan knockout, it's effectively zero. And in the cone-specific knockout, it's effectively zero. In the rod-specific knockout, there's just a little bit, and this is actually significantly different than the cone knockout. So we think this means the vast majority of the coupling is rod-cone gap junctions. There might be a little bit of cone-to-cone coupling. So 50 plaques per cone pedicle, maybe two or three, oh, this is a typo, should be two or three per cone pedicle. And we can't detect rod-rod coupling. So no rod-to-rod coupling, mostly rod-to-cone coupling, and a very small amount of cone-to-cone coupling. Okay, let's look at paired recordings with these knockouts. So this is a schematic to show what we're going to do. Here are two big cone pedicles and the overlying rod spherules. Well, we're starting to get the idea there's a lot of cone-to-rod coupling that's, you know, connects in 36 on both sides. The question is, really, is there any rod-to-rod coupling? Okay, so this is all from Kristoff's lab recorded in Spy Nanga. And so here we've got rod-rod paired recordings. And we find a small amount of rod coupling, which is eliminated in the Konexon 36 knockout. So it does tell us there's a little bit of rod coupling, and it is Konexon 36 dependent. The rod-cone conductor, this is substantially larger, over 300 pika signals in this experiment, but eliminated in the Konexon 36 knockout. And then the cone-cone conductance, these recordings are made from the pedicles. There's a small amount of cone-to-cone coupling, and again, it's absent in the knockout. So in the wild type, we have about 300 pika signals for rod-to-cone coupling, and a little bit less for cone-cone and rod-rod. In the pan knockout, everything is eliminated. So this is all connects in 36 coupling. If we look at the rod knockout, the rod-cone coupling is eliminated. And the rod-rod coupling is eliminated. There's a little bit of cone-cone coupling, and that must be direct. Cone-to-cone, that is. If we look in the cone connects in 36 knockout, rod-to-cone coupling is eliminated. Cone-to-cone coupling is eliminated. Ah, but so is rod-to-rod coupling. So rod-to-rod coupling requires connects in 36 in cones also. So the conclusion of this is that rod-to-cone coupling is direct, and a small amount of cone-to-cone coupling is direct, but rod-to-rod coupling is indirect. That is to say, it's rod-to-cone to rod. In other words, this is indirect coupling via the network. So, and this is another way of telling us really that the vast majority of the connects in 36 in the OPL is for rod-to-cone connects in 36 gap junctions. That's a summary of what I've just told you that there's no rod-to-rod coupling, or we can't detect any, and really what we're looking at is rod-to-cone coupling. So this is a summary of this first part of the talk. The connects in 36 is on the cone tealodendria. Rods and cones both have connects in 36. A rod-cone conductance is around 300 picosemons, and rod-cone-rod coupling is indirect via the network. And other people have reported that rod-to-cone coupling is dependent on connects in 36, particularly Astridi and Kangeana and Ingram, Sampath and Fein. And again, this just shows a home-out view of connects in 36 gap junctions. Okay, now we're going to move on to serial blockface EM, and I wanted to explain what this is. So I have to turn off my cursor. Okay, here's a block of retina embedded in E-PON, and we take a picture and then slice that off. And now you can go back and take another picture and you just keep on going maybe for several months as long as you need to and acquire a massive 3D data step. And this methodology was developed by Maurice Helmstadter, Kevin Brigham and Winfred Dank. Okay, we have a lot to thank, a lot to thank Christian Barron's, Philip Barron's, Thomas Euler, because when we got started with this, I read this paper and they'd analyzed the cones and bipolar cells and foolishly I thought, well, you know, we can do that too. And with a little help from these guys, they were very, very happy to share their data with us and we were able to get started. And this is all reconstructions that were taken from this E2006 data set from this paper in nature. We used E2006 because it's one of the few data sets that has the entire retina, including the OPL. And when we look at it, it's huge, 8K by 7K by 3K, you know, high resolution. That's two times 10 to the 11th voxels, 200 billion voxels. That's a big number. Usually you only hear numbers like that in the federal budget. There are 164 cone pedicles, six of them are blue cones that were identified by the Barron's group. But unfortunately, because this data set was fixed for tracing neurons, there are no visible gap junctions. But I'm going to tell you how we get around that problem. This is what the data set looks like. Here's the outer nuclear layer. Here's the inner nuclear layer. And here's the OPL. And it really is a 3D data block. And we can pass backwards and forwards through this data block. Here's the OPL. And now we're going to build it up from the bottom. And you'll start to see these large structures like this. That's a cone pedicle. And there is a whole mosaic of them. If we go a little bit further, the cone pedicles start to fade away. And if we go a little bit further still, you come to a depth where essentially all of these structures are rods ferules. And then if we keep focusing deeper and deeper, now we come to the outer nuclear layer. And we're through. We'll go back. And here's the OPL again. So this is what we're going to use to analyze this data. In the confocal microscope, a single rods ferule might look like this. And in this case, it has telodendria from two separate cones. This one is labeled for GFP. And this one is labeled for cone arrestin. Let me say that again. They're both labeled for cone arrestin. But in addition, this one is labeled for GFP. So two separate cone pedicles, both touching this rods ferule with connects and 36 plaques at the point of contact. So that's what it looks like in the confocal. We're quite familiar with it. This is the serial blockface data. Here's a single rods ferule. You can see the synaptic opening is right here. And you can see these postsynaptic conclusions. And there are contacts from cone telodendria. Now, these don't look like gap junctions because they're just not visible in this material. But this location within a micron very close to the postsynaptic opening is exactly where the connects and 36 is located. So we mentioned that all of these contacts, these cone contacts with a rods ferule, they all contain a gap junction. Sometimes you'll see a rods ferule on the top of a cone pedicle. And again, there's a contact. And we think these contain gap junctions because we can find connects and 36 in this location too. So this is the kind of data that we can develop from E2006. And right, so here's a single green, a single green cone pedicle with all its telodendria followed. And here it is in sections. So now the axon rises through the urban nuclear layer. And the black and gray dots show all the rods ferules that this cone contacts with. And the number is about 40. Here's a blue cone. You'll remember the blue cones we could identify in E2006 because of the work from Christian Barons. And the blue cones reach out and touch all the rods ferules within that dendritic field, just like the green cones do. When we plot all these dendritic fields, we get this massive overlap of cone telodendria. The overlap is about 1.5 or so. Here are the outlines of 13 cones that we analyzed in greater detail out of a group of 29. We analyzed the center so we wouldn't get any edge effects. So there are 13 cones and 361 rods. I don't think there were any rods in this area that did not receive contact, usually from multiple cones. This shows the divergence from one, from a cone to one, or two, or three rods ferules. Most rods ferules receive input from more than one cone, with the mean being about 1.9. In other words, two cones. This is very similar to a number reported from Peter Sterling's lab a long time ago. When we look at the convergence, that is, the number of rods connected to each cone, you see the number as the mean is 43 for a group of 29 cones, the same for the group of 13 we analyzed in detail. And the five blue cones that we could identify in E2006 have very similar numbers. So we don't think there's any color coding here. We think that green cones are connected to rods by gap junctions, and so are blue cones in exactly the same way. This summarizes our data a little more. Rods per cone is about 44, 43. Cones per rod is just under 2, 1.9. The mean area of a pedicle is 102 square microns, and most cones receive input from several rods. Okay, let's move on now and talk about segmentation and 3D reconstruction. Here's a single cone pedicle. I used to think these were just little triangles, but really there's this exuberant spray of fine telodendria that reach up to talk to the overlying rod spherules, and it looks like this. Now the red things are contacts, and as we rotate this, now we'll put all the rod spherules in place that talk to this individual cone, and the number is about 40. That seems like a lot. Now as we turn the rod spherules off and rotate the cone pedicle, and you can see looking down from the top all the contact pads, now we put the rod spherules back and rotate it the other way. Now you can't see anything because there's a cone pedicle in the way, but if we remove that, now you can see the contact pads on the rod spherules, and here's a single rod spherule that we've zoomed in on, and what I want you to see is that here in the middle is the entrance, here in the middle is the entrance to the postsynaptic compartment, and the contact is right around the postsynaptic compartment, exactly where the Conexon 36 is located. So we think that these contacts probably contain gap junctions. Here's a summary of this data, I just showed you this picture. There are approximately 50 contacts per cone, and it's very similar to the number of gap junctions, and it suggests that most of the contacts probably contain a gap junction. We were also very taken with the curved appearance of these contacts because it reminded me of this old data from Raviola and Galula, this freeze fracture EM that shows that this row of particles here, which are thought to be gap junctions, they make these concentric curves around the invagination and the postsynaptic inclusions that are right here. So very similar to the freeze fracture data, which was from 50 years ago, amazing. Okay now we're going to switch gear again to the FIB SCM. FIB is focused iron beam, and it's really another way to produce serial block phase data, but really at high resolution, especially in the z dimension, these are four nanometer voxels in the isometric, which is very useful for analysis. Before we go into that, here's a data set from Josh Singer, and this is really nicely fixed. Here's a single cone, and here's a rod spherule on the roof, as we say, and look at that, that's a gap junction. Over here at higher magnification, and you can see that the two membranes are merged, and there's some high density chromophilic material, that's a gap junction, and of course it's coincident with the location of Conexon 36. Okay let's switch over to the FIB, and now here's a single rod spherule. We know we're very close to the entrance to the invagination, because we can see these post-synaptic conclusions, and right here at the cone telodendral contract is a gap junction, and here another telodendral contact, and another gap junction. Here's another example. Now this is isometric data, so we can rotate this in 3D and get a flat amount view of the gap junction or the density that represents the gap junction, and then from this data we can get size measurements. Sometimes we find several gap junctions at the base of a rod spherule, and here's an example that there are two small gap junctions. You can see the high density staining shown at these arrows, but when we trace this through a series of sections, we find that these two gap junctions are actually connected. This is a viewpoint from underneath, and now we see that they're connected, and it looks like this. There's a massive gap junction right around the base of this rod spherule, and it looks like this if we make a little movie. Here's a single rod spherule with the mitochondrion, here's the ribbon and the horizontal cell processes at the bottom, the bipolar cell dendrites, and right here around the base in red is a large gap junction. We'll just rotate that. These are the post-synaptic processes exiting the base of the rod spherule, and right around the base of the rod spherule is a very large gap junction. This is scale bar is 0.3 microns, so we think this is about a 1.5 micron gap junction. That's well above the resolution limit for the confocal microscope, and in fact in the confocal pictures of Konexon 36, you can find these occasional large horseshoe-shaped gap junctions, and that's really what they are. If we quantify all this data, this is what we get. We get a number of gap junctions from the FIB from a sample of 42 rod spherules. The mean was 3.2 gap junctions per rod, with a range of one to six. This is the distribution, this shows the distance from the opening to the post-synaptic compartment, and you'll see that about 90% of that is within a micron, which is coincident with the location of Konexon 36. There are a few outliers, and we did trace all these back, especially these top half a dozen, and they are outliers, but we were able to trace all of these rod spherule contacts back to a KonePeticle. So these very few numbers here, even they actually are rod to Kone gap junctions. If we look at the dimensions of the gap junctions that we measured with FIB, we find that they're very variable in length from about 100 nanometers out to these very large single gap junctions over a micron, but they're all at about the same width, very narrow, and we think that this is consistent with a string or ribbon-like structure, as shown by raviola and galula in their freeze fracture work. And there's some very nice work from John Rash and Jim Nakey's lab, also using freeze fracture that shows that gap junctions can come in a variety of shapes and sizes, so we take this as evidence that the OPL rod Kone gap junctions are strings, and this shows the distribution of lengths. When we look at the variation by gap junction number per rod, if there's only a single gap junction shown here, we get these very large single gap junctions. These are outliers, really. And then two, three, four, five, then the numbers add up, but they're all about the same. And the mean size, actually that shows the median, the mean size of a gap junction, a rod Kone gap junction, is 480 nanometers, and this shows the total gap junction size. It tends to rise as there are more and more gap junctions, but it kind of bends over. That's because the more gap junctions there are, they tend to be a little bit smaller. So the total length per rod is about 1.5 microns, but the mean gap junction size is just under 500 nanometers, and they're all very close to the entrance to the post-synaptic compartment. We couldn't find any rod rod or Kone Kone gap junctions, and I'll come back to that. Okay, we did find some Kone Kone contacts. Here's a reconstruction of six from the FIB data, and the entire dataset didn't contain a whole Kone particle, so these are partial structures. But we did find some contacts. When we find contacts, and there's one shown here, it doesn't look like a gap junction. In contrast, there's a nearby rod Kone gap junction, and that is what we'd expect to see. So we don't believe these occasional contacts are really gap junctions, but there might be a few from the evidence I've shown before. When we look at rod rod contacts, and we can find a few of those, well, there are a couple of things to say about that. First of all, the rod contacts, they don't look like gap junctions. Here's a rod Kone contact, and I think you can appreciate this looks like a gap junction, and this doesn't. Furthermore, the position of the rod rod contacts tends to be on the midline of the rod spherule, and that's not where the Kone X in 36 is. So we don't think that we can find any rod to rod gap junctions. We find a few rod rod contacts, but we don't think they're gap junctions. Okay, now we're going to do some calculations with all this data, and be patient with me as we go through this. So we've got 40 odd rods per Kone, and 50 odd gap junctions per Kone, approximately three gap junctions per rod spherule. And so we can calculate the divergence, and it's just a little bit less than two. So 3.2 gap junctions for every rod, but they're coupled to more than one Kone. So when we calculate the number of gap junctions per average rod Kone pair, it's 1.7. Now the number of channels is 480 nanometers and 10 nanometers per Konexon channel. That gives us 48 channels, and 48 times the divergence, excuse me, 48 times the number per rod Kone pair gives us about 80 gap junctions on average per rod Kone pair. Now we can multiply that by the single channel conductance for Konexon 36, and the total comes out to be around 1,200 picosemons. Now the important thing about this number is that we calculated it from our morphological data. So that's the maximum number for the size of gap junctions that we've measured. How does that compare to our physiology? Well, at rest, as we've seen before, the conductance for a rod Kone pair on average is about 300 picosemons. That comes out to be 25 percent of the maximum. In other words, the open probability is 25 percent. But we know that gap junctions can be modulated by dopamine, and if we close them with a D2 agonist, it goes almost to zero. The noise level is about 50 picosemons. That's about 4 percent open probability. But with a dopamine antagonist, we can increase the conductance to over 1,000 picosemons, and that's 90 percent, close to 100 percent, open probability. I've never seen numbers like this, because normally we think the open probability is very low. But this suggests that for these rod Kone string-like gap junctions, it suggests that really all of the connexons can participate. Very briefly, some of the functions of coupling are to let the rod signals into the cones. When we look at the connexon 36 knockout, you'll see the real cone responses are very sharp, whereas they're very slow. That's the rod component, and the sensitivity changes by about a log unit. And the single photon responses shown here at night, they actually get smaller. That's because the gap junctions are open at night. And so the coupling actually makes the single photon responses a little smaller. But that may be less important than reducing the noise in the photoreceptor network. Okay, so I want to summarize what I've just told you. 40 rod cone gap junctions per cone. Most rods are coupled to more than one cone and two or three per rod. The gap junction size is the strings of about 50 connexons. And when we do our calculations, we think it's about a 1200 peak assignment maximum conductance. And with dopamine modulation, we can vary that from close to zero almost to 100%. So we think all the channels may be switchable. The modulation depth, which is a measure of plasticity, is approximately 20. The function, they're open at night, which reduces the single photon responses. So we think that noise reduction is probably more important. It injects the rod signals then to cone pathways. And that's the basis for the secondary rod pathway. When we use these mass lines and record from ganglion cells, we can show this an intermediate input carried by the secondary rod pathway. This reference is some of the nice work that came before. And we're almost done. All this work is published in two papers. One is in science advances. And the other one was just submitted as a preprint to the bio archive. I found a tweet from Julie Haas about this. And I thought, wow, this is almost the perfect summary of what I've just talked about. So if you can't remember what I said for the last hour or so, then just remember an exuberant spray of fine cone telodendria and read on for the dopamine can drive gap junctions to 100% open probability. And thank you very much for your attention. Thank you, Steve. That was a very enlightening talk. Thanks for that. But before we start with with questions, I just want to pass on to your comments. We had from a audience today. First, I had Tiffany Smith somewhere. Yeah, Tiffany Smith, she really enjoyed your confocal image. Apparently she's your favorite confocal images. And we also had a comment from Malafiller that is really mind blown by your movies. So I guess I can speak for everybody here. That was very, very good pictures. We really enjoyed your display here. Thanks very much. We're glad you like them. They're very neat. I will then move to the question part. Before we start, I'd like to remain audience that if they want to join us right now, if they want to ask their questions themselves, they can join us in this room. After the question, we're going to move to a more informal discussion. So if you want to join us, do that during that question time. And so that we're going to just close the live stream and stay between us. So let's start. I have a question from... I'm going to say it's wrong. Jochen and Yu Wang. So regarding the FIB movie. So let's say cool movie. We're wondering which signups is established first, the chemical or the electrical? Oh, hi, Yu Chen. Well, I don't know the answer to that. We haven't looked at any developmental sequence, but it certainly is an interesting question. We have spent some time thinking about why the gap junctions are gathered at that spot. And I don't really have a good answer for that either. But it might be that there are some addition proteins that determine the specificity of connections. And they're responsible, as you've shown in some of your work, responsible for synapse formation. And maybe some of those scaffolding proteins are required also for gap junctions. I don't really know. It's speculation. But those are some of my thoughts on the matter. I have a question from your first set of confocal images you showed. Once again, from Jochen and Yu Wang. Why is a conductance for road cone coupling are different if they are using the same connexin-36? Why is the rod cone conductance different from, say, the cone cone conductance? Well, the conductance is determined by the type of connexin involved, which is the same for both of them. But it's also determined by the size of each individual gap junction. So we think that the rod cone gap junctions are larger than the cone cone gap junctions. We have not been able to find cone cone gap junctions. But there's a little evidence that they're there. And if they're small, we think. All this work is from the mouse retina. There's good evidence for cone cone coupling and a couple of other species, specifically in primate retina and in ground squirrel retina. So it might be a general feature of the mammalian retina. But we haven't been able to find it in the mouse retina. I hope that answers your question. I guess I'll also answer George's question. I have one from Jeff Dimon. Hello, Jeff. Hey, Jeff. Yeah, please join us, Jeff. He's referring to your FIB data set. So based on the narrow width and raviola replica images, are these gap junctions, he calls that ribbons, one connexin-wide. He's referring to the lengths of that. So really, Jeff has put his finger on a major issue here. And I think the answer is yes, that these are strings. And I think the evidence for it comes from a couple of things. First of all, from Ilyo raviola from 50 years ago, showed that the strings of single connexons. I find that very convincing. But in addition, when we look at the fluorescent intensity, the gap junctions in the OPL, not only are they very small, but they're very dim in terms of fluorescent intensity. I mean, by a factor of 10 or 20, dimmer than, say, the A2 gap junctions in the IPL. So we think that that means that they are much smaller and it's consistent with a string-like morphology, as opposed to a plaque, which would have depth, which would have length and width. So yes, the short answer is that we think they're strings. We'd like to confirm this with some additional methodology. And I've sort of been thinking about that. But I think to go back and do freeze fracture EM, that's a very specialized technique. Whether we can approach this with super resolution microscopy or not remains to be seen. All right. Thank you for that. I think I covered all the questions I could find. Let me just check. Yeah, that's what I could find for the moment. Hello, Rob. Rob Smith is with us. If you want to join us, please do that now. We're going to close the live stream very soon. So if you want to join us for follow-up questions or discussion, please do that now. Hello, Rob. Hi, I've got a question. Hey, Rob, how are you? I'm well, thank you. Working from home a lot here in Philadelphia. People are starting to go back into the laboratories. But since I do a lot of computer work, I can do it from home a lot. So my question was near the end of your talk. You're talking about how the coupling at night reduces the single photon response amplitude. That's right. And it would seem to me that it would make more sense to have that coupling at mesopic levels, high scotopic or something, but in the dark of night to have the coupling be reduced so that you would maximize the single photon responses that the rod butt polars would see. Do you have any evidence about that at all? Well, I showed just a little data at the end, but it's really from Kristoff's data, the 2015 J. Physiol paper. And it's very clear that when the gap junctions are open, as you would predict, the single photon response is smaller. And so, you know, we spent some time thinking about that. And yeah, I understand what you're saying that, gee, wouldn't you think it would be better if the single photon responses were bigger? Well, that doesn't seem to be the way that it works. And so then it makes us think that maybe the most important thing is not really the size of the single photon response. And it may be that what's really more important is to reduce the noise. You can compensate for a smaller single photon response by turning up the gain later, but only if you get rid of the noise. And this all fits with, you know, what we know about dopamine being, you know, light dependent and circadian and dopamine levels, you know, high in the daytime and reduced at night. So I think it fits together. And yes, I do understand I've certainly spent a long time pouring over the Smith, Freed and Peter Stirling paper from 1986. So you guys were pioneers here. But we think that the gap junctions are open at night. So just a quick follow up. Do you think that there's any way that dopamine or some other modulator could could work differently in the darkest night versus high scotopic so that maybe something else is modulating something at the gap junctions besides dopamine at that point? Well, so my short answer to that is it doesn't seem so. But it's also true that we would do well to think about other mechanisms to modulate gap junctions. We think that these electrical synapses have a very rich repertoire in the same way that chemical synapses do. And there's probably a host of proteins at these electrical synapses that we don't really understand yet. So I don't want to close the door as it were. But we think it's, you know, dopamine is high in the daytime and low at night. So that's the way it works. All right. Well, thanks for your answer. That's been very clear. Appreciate it. All right. It's good to see you, Rob. Sorry. I'm just going to tell our audience that I'm going to turn off the live stream. So see you next week for our next talk. Thanks again, Steve, for being with us today. And we will now follow up with our more informal discussion. That was a great pleasure. Thank you. Okay. And we are off.