 I hope the sun will be nice, okay, we are nice, nice. Okay, hello and welcome everyone. Thanks for meeting us today. Before we start, I would like to thank the organizers of the worldwide neuro initiative, and most specifically team boggles for providing the logistics that hello has to continue hosting neuroscience seminars despite the current event. And this also allows us to share the seminars or seminars with a larger audience engine. So we're very grateful with that. Thanks for meeting us today. Sorry for that. So I just wanted to say that those French times, give us the opportunity to explore new ways to communicate and exchange or work, and I really hope that we will consider keeping and upgrading this new format for academia in the future. Today is the first talk of the Sussex vision series where, as you have guests, we will focus on vision and retinal secrets. Obviously, if you have any question related to the talk, please use a YouTube chat, and we will answer them at the end of the To integrate this series. Let me introduce you to Dr. Jeffrey Diamond from the National Institute of Health in Bethesda, Maryland. He received his actual degree in science from Duke University in 1989, and his PhD from the University of California in San Francisco in 1994. There is to did excitatory synaptic transmission in the retina with David Copenhagen. During a postdoctoral fellowship with Craig jar at the volume Institute in Portland in Oregon, he investigated the effects of glutamate transporters on excitatory synaptic transmission in the campus. He joined the National Institute of neurological disorders and stroke as an investigator in 1999. He was then awarded by the presidential early career award in science and engineering in 2000. And it was finally promoted to senior investigator in 2007. So, with that in mind, I would like to thank Dr. Jeffrey Diamond for the great work that he has done on this series. And I would also like to thank Dr. Jeffrey Diamond for the great work that he has done on this series. So without further ado, please welcome Dr. Jeffrey Diamond. I would like to turn off this series, although I know in reality you just need a guinea pig to work out the kinks. But I also realized that that I had better turn off my YouTube thing so that I don't see any of the comments coming on during my talk. I hope, I hope everybody can hear me okay I hope everybody is doing okay in these extraordinary times and that a little science might be a good, a good respite from our sort of new schedules that we've all been forced into. I'm assuming that most of my audience is our vision neuroscientists with maybe a few general neuroscientists sprinkled in and the occasional person who stumbled across this in search of some kind of do it yourself video about how to unclog their, their toilet. So, you know please ask questions in the comments and we'll get to them at the end. So my lab is interested in how synapses and neurons and and and small circuits of neurons in code and process information that's required by by the surrounding network, and we focus particularly on on the retina to study these questions and the retina is a great system. To study these kinds of questions because we can stimulate it physiologically with light, and then a variety of different kinds of preparations, we can record from any of the neurons here in in the retinal circuit. And I am going to hang on. Hang on guys I'm getting a pointer here. I was getting a pointer. Anyway, so the retina circuit can be thought of in very simple terms as a series of parallel radial pathways where light is absorbed by the receptor cells the photoreceptors here. That light energy is transformed into a neural signal that's passed the bipolar cells and the bipolar cells then relay that signal through excitatory synapses down here in the inner retina to ganglion cells and ganglion cells send their, their axons through the optic nerve to the rest of the brain and the ideal signaling pathways are modulated by two laterally oriented inhibitory networks, one in the outer retina made up of horizontal cells and one in the inner retina comprising amicron cells. I should make a note just at the outset for people who aren't used to thinking about the retina that most of these cells within this network communicate with each other without the aid of action potentials most of them respond to light and signal these responses with small slow graded changes in membrane potential as opposed to action potentials ganglion cells of course fire action potentials to communicate through the optic nerve. Some amicron cells fire action potentials and very occasionally a bipolar cell will and some species but in general, this is an analog circuit. Now our interest in amicron cells is twofold. One, they're interesting from a visual processing standpoint because they confer a great deal of complexity to the visual signals communicated by the ganglion cells. As synaptic physiologists were interested in amicron cells because they don't operate like most neurons. Most neurons like the Prokinje cell rendered over here, receive their synaptic inputs in elaborate dendritic arbor shown up here. And this input is integrated in interesting ways down to the soma and eventually to the axon hillock where it's formed into a pattern of action potentials that's connect cut that's conducted along the axon to downstream targets. Amicron cells are different. They're so named amicron cells by kaha because they lack a long axonal fiber fiber so amicron is a mashup of the Greek for lacking along fiber. And they receive their synaptic inputs and make their synaptic outputs on the same dendritic processes. And we've been very interested in the lab how amicron cells do this how they transform their excitatory input that they get from bipolar cells into inhibitory output that they send to a variety of different targets and and how this transformation contributes to the visual processing that they've been tasked to do by the network. Now this is a rich question particularly in the mouse, where we now know, based on recent drop seek studies from Josh Sains's lab that there are about that there are probably give or take one or two 63 different kinds of amicron cells in the mouse retina and 45 of those are shown here from a connectomic study from Winfrey Dent's lab a few years ago and see that these are transverse sections of the retina with all of each amicron cell type that was observed in a in a small section of retina that was examined using serial block face electron microscopy. We all have a lot of time on our hands, but I'm not going to go through all 63 amicron cells, I'm going to focus on just one, the a two amicron cell. The a two is the most numerous amicron cell in the mouse retina makes up about 10% of the total population and the a two is a little guy it's, it's cell body is relatively small less than 10 microns in diameter so it's no more than about 30 or 40 microns from from head to toe. And you can see that it has two general sets of dendrites there are the lobular dendrites out here, and then there are the the arboreal dendrites down here. Okay, and you know the last 40 or 50 years of literature about a two amicron cells has shown a picture of these cells where they they get their excitatory input down on on these dendrites the arboreal dendrites and then they make their outputs up here on these lobular dendrites so there's there's some distance between the site of input and the site of output and in the first half of the talk. I'll talk about some experiments that we've done and published a year or two ago, examining how the a two transforms the input to this output. In the second half of the talk I want to talk about more recent work that we've been doing in the lab that hasn't been published yet, where we find that the a two also makes an output down here in these dendrites arising very close to where it receives its excitatory input for a much more direct and efficient transformation of input to output. Okay, so the a two plays a very important role in the mammalian retina circuitry. Particularly in the rod pathway so called that that that mediates night vision. Right when photons are very scarce they're absorbed primarily by rod photoreceptors which are much more sensitive than their cone photoreceptor counterparts. And rods make the majority the large majority of their synaptic connections on to rod bipolar cells there are about 14 different kinds of bipolar cells and the mouse retina. So only one of them is a rod bipolar cell but it makes up about 40% of the total population of bipolar cells because the mouse retina like our retina is is rod dominated. So that bipolar cells do not talk to ganglion cells directly instead, they go through the a to American cell. Okay, now the rod bipolar cell is called an on cell, because it is it is activated or depolarized in response to increments in the retina and it makes an excitatory synaptic connection on to the a to so the a to is also an on cell. Now the a to release relays this on signal to what we call the cone pathways, the on cone pathway it relays this signal through a sign conserving gap junction synapse here, the synaptic terminals of on cone bipolar cells, and it inverts the signal through an inhibitory glycinergic synapse under the synaptic terminals of off cone bipolar. Okay, and so in this way the rod pathway piggybacks onto the cone pathway to use common ganglion cells down here in the inner retina. And note here that that the inner plexiform layer the synaptic layer down here where the bipolar cells make all of their synapses can be functionally divided in into well many different layers but for the purposes of this talk to primary layers the lower layer here contains all of the synapses carrying on information and the upper layer contains the synapses carrying off information. Now, a to have also been shown to make inhibitory synaptic contacts up here in the off layer to off ganglion cells and we're interested in how the a to transmits its input from rod bipolar cells to its downstream targets, particularly because the visual synapses that we can record in a to suggest that they carry multiple kinds of visual information and you can see that here in this rather reduced experiment. That was performed by Nick Ash when he was a postdoc in the lab and Nick is recording from an a to amicron cell and an acutely prepared slice of rat retina. And everything was done under complete darkness using infrared illumination so the the retina has remained very sensitive. And what he's done here is he's he's voltage clamping an a to so that he can measure the, the excitatory synaptic input from the rod bipolar cells pre synaptic to it. And then he steps to a very dim background light here and you can see that this elicits a transient response that settles down to a more sustained response. And then when he goes to a brighter light level. He gets another transient response followed by a sustained response and it turns out that these transient and sustained responses encode different kinds of visual information. So the amplitude of the sustained response at any particular time, you can see, and this is a response of the amplitude of the steady state current as a function of of the visual, the, the, the luminance of the visual stimulus. And you can see that the response is the same, regardless of whether he arrived at that luminance from complete darkness as shown him in blue, or whether he stepped to that luminance from some other brighter background. Okay, and what this tells us is that the sustained component is reporting luminance instead of changes in luminance in other words it's it's blind to contrast, but it's a reliable indication of luminance. The transient component is very different. If you look at the amplitude of this transient that arises above the sustained component you can see that it's that it's greatly depressed when you step to higher light levels from some lighter background here. It's depressed in a very interesting way. If you look at the amplitude of these depressed events, you see that they encode quite nicely, the contrast between the two steps, and, and not just the difference in the luminance of the, of the two steps. This is called the Weber contrast. In other words, the difference in the two steps as a function of the preceding background, and this is in fact how psychophysically, we detect sensory input. The smallest detectable change and this is the first commandment of psychophysics, the Weber fechner relationship the smallest detectable change in the sensory stimulus is proportional to the background and I can show you that quickly here. Two spots on two different backgrounds, and the absolute difference in luminance this is obviously going to depend on on your monitor, the absolute difference in luminance is the same in in the two sides of the of the image here, but I think you would agree that you can see this spot on the left, easily than the spot on the right and that's because the Weber contrast or, or this relative the difference relative to the background is much greater than it is on the right and if we increase the luminance on the right so that the Weber contrast is the same. You can see that it's, it's similarly easy to see the two spots now. Um, we've been interested in how the a to transmits this information does it transmit luminance information sustained information, or the transient information to its downstream targets. And I want to focus on how in the first part of the talk how it transmits this information to the off cone bipolar cells and should say we're interested in this because the the connection between the a to and the off cone bipolar involves one of these nonlinear masses a glycinergic inhibitory synapse onto the off cone bipolar cell terminal. But the first question that we need to ask is which of the off cone bipolar cells is the a to talking to there are five options in the mouse retina five different off cone bipolar cell types and we don't we didn't really know, going into our study which of these types, the a to connected to all of them equally or did it did it favor one or the other. So to answer this question we collaborated with Kevin Brigham, who had acquired a very nice data set of serial electron micrographs from the interplexa form of the mouse retina using serial block face electron microscopy. And you can see this is a movie stepping through these sections that are that are cut every 25 nanometers, and you can identify individual cells, and, and, and annotate them and get the skeletons of these cells, and identify their synaptic inputs and outputs. And, and Cole Graydon and Josh Singer, working together Josh's professor at the University of Maryland, who collaborated with us on this project found about a dozen a to amicron cells within this data set and and this is what the dendritic arbors of the a to look like looking above on top of the retina. These are convex halls showing the extent of the synapses made by these, these different cells, and this is a side view of their dendritic arbors. And within this data set, we can identify structurally both the synaptic inputs to the a to is the come primarily from the rod bipolar cells but also from cone by off cone bipolar cells, and also the synaptic outputs that they make. And when we focused on the synaptic outputs that these a to is made up here in the off layer of the interplexa form layer. We found two interesting things. Is that the a to strongly favor sending their outputs up here to bipolar cells 89% of their synaptic outputs are directed to off cone bipolar cells in the in the off layer only 11% are directed to ganglion cells. The second interesting thing is that a to's don't love all off cone bipolar cells the same. They strongly favor the type to cone bipolar cells sending about 60% of the output that they send to bipolar cells are sent to the type tos. Okay. So we were interested to understand how the a to's were signaling to type to cone bipolar cells and what kind of information they were transmitting, and this was aided very nicely by an existing mouse that expresses gfp under control of the snap to And in these mice. There are some ganglion cells that glow all of the horizontal cells glow, but in addition, the type to cone bipolar cells also glow, and even though there are there are multiple types florescing in the in the retina, it was easy to identify and to record from these type tos specifically. Okay, so this is nice because now we can record consistently from a single type of cone bipolar. And our strategy here was to record from the three principles in this part of the circuit the rod bipolar cell, the a to and the type to off cone bipolar to see how the signals are being transmitted through the through the network. Okay, so the first thing to do is to record from rod bipolar is an a to's and this is work that Cole did when he was a postdoc and This is similar to work that that we've done in our lab previously and others have done subsequently where you can record from synaptically coupled rod bipolar and an a to and gives a series of to polarizing steps to the rod bipolar and see the post snap to excitatory synaptic currents in in the a to and I'll remind you that these voltage steps don't look much like action potentials but and don't look perfectly like the light of both responses in the rod bipolar cell but rod bipolar cells do respond with slow graded changes to the firing potential within this range. So this is actually a fair approximation to the kinds of depolarizations that you would see in a rod bipolar cell in response to life. And when you make these steps when Cole made these steps, he saw a transient burst of release from the rod bipolar cell and and then an increase in sustained signal. And this increased with the size of the depolarization and in fact if you if you plotted the amplitude of the of the transient response here versus the membrane potential step in the rod bipolar cell, you see a relation here that looks very similar to the the voltage dependence of the pre synaptic calcium channels in the rod bipolar cell. Now, Cole could make an analogous recording from, hang on. Okay, from between a twos and off comb bipolar's and when he depolarizes the a to in a similar way, he sees quite analogous responses in the, the type two comb bipolar transient responses, followed by the sustained responses. Now the fun happens when he records from a rod bipolar and a type two off comb bipolar. And now the signal is being relayed from the rod bipolar to the off comb bipolar by the a to amicron cells in between, but we're not controlling those. And what you can see is that as before this transient component of the response is relayed quite faithfully through the a twos to the off comb bipolar. But if you look at the sustained component, you see a very different story. The sustained component, the background activity that we record in the off comb bipolar is is a good bit higher than in the other recordings. But what what we see is that the sustained response does not change at all with changes in the depolarization delivered to the rod bipolar cell. And this is despite the fact the sustained responses change reliably in the other two recording configurations. And so this actually this suggests under under these conditions that the a twos are transmitting only transient information and not the sustained information. And by analogy to light signals and and we can only make an analogy at this point. This would suggest that the a twos are transmitting only contrast information as opposed to sustained information. And we think the explanation for this is fairly straightforward and can be seen when you deliver a voltage steps to a rod bipolar and record the post synaptic potential, the change in membrane potential in the a to under current clamp. Okay, so this is not the the excitatory synaptic current. This is the post synaptic potentials. And you can see that this step this large step here elicits a sizable transient component, and then only a very small sustained component. In fact, you can see individual synaptic events here that may reflect the release of individual vesicles from the rod bipolar cell. Okay, but the average change in the membrane potential here during the sustained component really isn't that much different than the membrane potential here. And so we already know what the a to release function is because I just showed you those paired recordings between the a to and the off comb bipolar. So we know how much of this depolarization will cause a change in the a to release. And the expectation is that it shouldn't cause much of a change at all. So Paul tested this directly by taking this, this post synaptic response. Okay, putting it back into the computer and playing it into an a to as a voltage command. So he's voltage clamp the a to and he's, he's playing this, this wave form in as a voltage command and recording from a synaptically coupled off comb bipolar. Okay, and you can see, as you would have expected that only the transient component of the response is transmitted it's transmitted very faithfully actually on every trial you see a transient response but the average sustained response is not very impressive at all. And we think that's because the sustained component of the response is not moving the a to very far along its release function. Okay, it's well it's released versus voltage function. Now an interesting twist in this story is evidence that was collected by Will Grimes and Greg Schwartz and Fred Rieke a few years ago, showing that the resting membrane potential of a to and reconciles varies with the visual conditions of the retina. If you if you change the background the ambient luminance bathing the retina, the resting potential of the a to varies and it can vary by five or six millivolts. Now five or six millivolts doesn't seem like a lot when you're talking about, you know, 100 millivolt action potentials but again remember in the retina in in this, this analog network five or six millivolts can be a significant change and we wondered whether these kind of changes could influence the kind of information being transmitted by the a tos to the to the type two cone bipolar. And so what Cole did first was he repeated this experiment, but he added a variable amount of bias voltage on this signal to change the resting membrane potential of the input artificially. And when he did this, he saw something very interesting now when the resting membrane potential was relatively hyper polarized, he saw very little sustained responses but as he depolarized that he started to see more sustained information. If he looked at those responses when the a to was more depolarized, you could see correlations between the signals in the pre synaptic a to and the signals in the post synaptic type two cone bipolar and these correlations were actually quite strong. They were close to minus one here and it's a negative correlation because these signals are outward and these signals are inward but actually it's a very strong correlation. And that's very interesting it indicates that the synapse can go from basically ignoring sustained information to being able to pass information about individuals synaptic events that are arriving the a to from the rod bipolar cells and that's a result of depolarizing the a to up into a different region of its release function here. Now another way to look at this is in the experiments where we're leaving the a tos to their own devices and recording from rod bipolar son off cone bipolar and and what Cole has done here is he's taken a response to a step in the rod bipolar this transient response that you see in the cone bipolar. And he's, he's, he's made it so that the average baseline is at exactly zero, so that when he integrates this this integrated baseline remains at zero, and you see a jump in the integral that is this transient component. But then following that you can see that the integral is flat and that that's because the average baseline result here is is no different than the preceding I'm sorry the sustained response here is no different than the preceding baseline. And as I, as I mentioned to you, this remains the same for a series of steps where we changed the pre snapped to grab bipolar cell membrane potential. But then we stole a trick that Josh Singer had had had shown by applying line operating which is a drug that blocks em currents the KCNQ channels in the a tos and to polarizes the a tos in Josh's hands by about six millivolts. And when we did this and to polarize the a to all of the a tos in the network by about six millivolts you can see now that the sustained component of the signal being transmitted from the rod bipolar. To the off cone bipolar encodes very nicely, the membrane potential of the rod bipolar so just with a few millivolts of depolarization we're able to completely transform the input output characteristics of these a to elephant cells. Okay, so I'm the experiments that I've shown you so far, examine how the a to might might transmit and we still have to do light experiments and showing what kind of light about signals. The off cone bipolar gets from the a to for a variety of reasons those are not completely trivial experiments, but these talk about a regime of night vision where we're able to, to make calculations of contrast that we have enough light in the background and in the foreground to make these comparisons. But I want to remind you that this circuit is is extremely sensitive and is able to respond to visual stimulus when there this photons are so scarce, the contrast really isn't a relevant parameter that that the network really is in a in a more linear regime since simply trying to get the photons that are falling on on the retina. In fact, will in a collaborative project with with Fred, Ricky has showed that that you know you can record responses in the a to, even to single photons being absorbed by one of the pre synaptic rod photoreceptors here. And so, you know, we've been interested in how the, the, the a to transmits this kind of very low light information I'm not going to be able to to go to the single photon level but let's just say down in the lower regions of night vision how these signals are transmitted downstream. And I should note an interesting feature of this is that you know so the a to is going to communicate to on the on pathway and also the off pathway in the common pathway in the ganglion cells and the off pathway is actually more sensitive than the on pathway you can see this in a recent paper from Petri Ali La Rila and Fred Ricky showing that off ganglion cells respond more sensitively at the at the lowest visual stimuli and then the linear fashion. Petri has gone on to show that the on pathway does mediate meaningful signals down near visual threshold, but we've been interested to understand how this circuit could accomplish this very high sensitivity and linear characteristics. Now the pathway that I've shown you so far involves, you know, the rod bipolar going through the a to and the a to going to the off cone bipolar and then, you know, the next step there of course would be going to the off ganglion cell. Now if we're going to critique the the the finer points of retinal circuitry, you know, the thing that this pathway has going for is that I've shown you that a to connect quite vigorously to off cone bipolar is particularly the type twos, but there seems to be a lot of problems with this pathway if you're if you're concerned about sensitivity. First of all, the inputs and the outputs of the a to are are are separated by a fair amount in space, even for a small cell like the a to aspen hard bite and his colleagues have shown recently that there is a significant amount of a signal attenuation from the inputs to the inputs. So that's going to hurt your sensitivity, as will all of these chemical nonlinear chemical synapses in this pathway. Now, this these nonlinearities might be good for limiting the signal or enhancing the signal to noise ratio of these signals but it would likely hurt the sensitivity. You know the outputs from these off cone bipolar cells are being received on the dendritic harbor of the ganglion cell. And so they're being integrated together with all kinds of other spontaneous and evoked inputs and it seems like a noisy situation that would limit the sensitivity of the response to a very to sparse visual signals from the a to. Now, a more recent pathway has has cut out one of the middlemen here, the the off cone bipolar by a to synapsing directly on to the off ganglion cells and this of course has the advantage of cutting out one of these synapses. So it still has some of the same problems as the previous pathway, in that the inputs are coming into the dendrites of the ganglion cell. And in addition, we found anatomically that a to is really don't connect very strongly to ganglion cells up here in the off layer. Okay. So more recently we've considered a third option in which the a to since I mean it's got all these processes down here anyway maybe it makes direct contacts on to the ganglion cells down here. Okay. Now this would be nice because again it would cut out the the off cone bipolar's, but even better is that if the a to is we're making outputs here they would be making their outputs very close to where they're receiving their inputs from the rod bipolar cells so there would be much more direct transfer of information. And of course, it would also be nice to make these inhibitory synaptic inputs near the soma of the ganglion cell where they would be much more powerful and exert likely exert almost like a veto action on all of the other inputs coming in from the dendrites. Now evidence is is an unfortunate detail that needs to be dealt with. But actually, if you look in the literature of the anatomy of a to and this is in mouse on the left and in rat on on the right. This is the connectomic study that I alluded to earlier from tanks group, and every once in a while you do see an a to process sneak down into the ganglion cell layer. And a hard bite and colleagues have shown in the rat when they filled fluorescently filled several dozen a to's they see occasionally that some of these a to's send processes down into the ganglion cell layer, and in serial EM reconstructions of mouse a to's Tsukamoto and only actually explicitly identified synapses made by the a to's down here. In all three cases, I think that the authors and recently so dismiss these as as perhaps anomalous and not indicate indicative of any kind of specific pathway. But we were interested to pursue this, and we are going to receive a strong glycinergic input from a to's they need receptors on their cell bodies or near them to receive this input. So what our biologist in the lab did some immunohistochemistry to test this and and looked at the co expression of an antibody for the alpha one subunit of the glycine receptor here with an antibody that labels ganglion cells and you can see that I mean all ganglion cells express some of the receptor on their on their cell body. But a subset actually seem to express them much more strongly and while co stained this with a marker for alpha ganglion cells which are the largest ganglion cells by cell body. It's going to be easy to pick out, and you can see that these alpha ganglion cells in particular seem to express a high concentration of glycine receptors and our collaborator my name who at the University of Wisconsin looked at this in a different way by looking at the glycine receptor on individual filled ganglion cells and saw that off alpha this is an off sustained alpha ganglion cell express a large number of glycine receptor clusters on the cell body. And even the proximal dendrites and the axon, as opposed to an on off directional selective ganglion cell which which has a much lower expression. We had our serial block face em data, and all of those reconstructed amicron cells. And so Morgan must grow who was a post back fellow in the lab and is now a graduate student with Josh singer, mostly as a result of this very nice look down here in in the, the arboreal dendrites of these a twos to see if she could find synapses with ganglion cell bodies and in fact she could. She was able to locate one off sustained alpha and one off transient alpha ganglion cells and the data set, and each of them received synaptic inputs directly on their somas and surrounding areas from a twos and you can see these are serial actually every other section in a series of sections showing a synaptic connection between an a two and green and an off sustained alpha cell body in red. And Morgan took the trouble to construct this cell and you can see that there's a large number of glycinergic synaptic inputs from a twos made on the soma the proximal dendrites and even, and even the axon, and Morgan traced the terminal dendrites of the pre synaptic a twos and saw that some a twos actually make a very powerful multi synapse connection with this ganglion cell, all of these green processes that I'm showing you here arise from a single way to the cell overall receives inputs from a number of a twos, but several of them make very powerful connections on to the cell. And there's a graph showing all of the a twos in this in this EM data set that she found that made synapses on to ganglion cells up in the off layer so the red symbols show the total number of synapses that each amicron cell makes in the off layer, and the x axis is the relative location of each one of these amicron cells of their cell of their, their primary axon or dendrite shaft coming into the inter plexiform layer in XY space relative to the location of the off alpha ganglion cell in the data set. Okay, and you can see that that most of these a twos make some synapses in the off layer to this off alpha ganglion cell you can see that the number of these synapses decreases somewhat as you go further away in terms of lateral distance. The black symbols show synapses made from the a twos and to the the soma or or the the proximal dendrites in the on region of the inter plexiform layer and these kind of synapses are only made by a twos that lie directly above the ganglion cell in the off alpha ganglion cell and this makes sense geometrically it's it's only right there that the off alpha ganglion cell has any processes in the on layer of the inter plexiform layer. So, so this would suggest that synaptic connections ought to be particularly strong, coming from the a twos that lie directly above the off ganglion cell, and will grinds and Milo said let's check to two. Will's the staff scientists and Milo says that a senior postdoc approach this problem using a mouse that was expressing channel adopts and in a to American cells and you can see they filled the an off alpha ganglion cell this is experiment that will did with the green die you can see the a twos in red. Actually, there are two kinds of of American cells stained here, but you can distinguish them from each other afterwards morphologically so that we can identify which ones are a twos and the idea here was to flash a spot of blue light on the cell bodies of these different cells and see if we got a synaptic response in the off alpha and of course blue light excites the retina if you shine blue light, even in an area where there's no American cell you get a light response in this cell, but you can block that pharmacologically by blocking the synaptic inputs to bipolar cells. And under these conditions now, if we go around. If will went around and stimulated these individual cells, he could measure responses in the ganglion cell that resulted from, you know, direct synaptic input from those cells. What he saw was that he saw direct synaptic input from a twos only if they were located in retina space near to the cell body of of the ganglion so you can see that here that the response that he would record in the ganglion cell fell off with a length constant of about 40 microns. Okay, so these powerful snap the connections are only being made by a twos that are very close and XY space to to the ganglion so so this suggests that if if the synaptic input from the rod bipolar is able to depolarize these these synapses down there sufficiently that there should be a good transfer of of synaptic inhibition to to the off alpha ganglion cell. Okay, and so this is going to depend on the relative location in these terminal dendrites of the synaptic outputs here and the synaptic inputs to the a twos from the rod bipolar cells and so Morgan looked at these processes and identified where the synaptic inputs from rod bipolar cells arise and she's shown that here is these little red dots I hope you can see that. And you can see that in many cases, the rod bipolar cell inputs arrive very close on the on the process to the outputs to the a two. And what I did was I took these data from Morgan and and graft it in a way that allows us to look at this so these are each one of the terminal a to dendrites that make contact with the off alpha ganglion cells. What I've shown is I've aligned each one of these dendrites I stretched them out linearly and I've aligned them so that the the end of the dendrite is is here at at a distance of zero. Then I've plotted the relative position of each of the rod bipolar cell inputs in blue, and the outputs of the a to to the off ganglion cells in gold. And what you can see here is that the off the sorry the outputs are made largely within the last 10 microns of the, the process, and the inputs from the rod bipolar cells are received when about the last 20 microns. Okay, so this is actually an ideal situation for the transfer of signals and I can show you this with a simple neuron model. Okay, which this is extremely simple there's a cable here that I have a soma over here just for fun. And I'm recording the post synaptic depolarization the two different parts one is down here near the sealed end of the cable, and another one is up here near the soma. And I have a relatively small excitatory synaptic conductance that I can move up and down the dendrite here and record the post synaptic response. And you can see that when the synaptic input is very close to the end of this cable, the post synaptic depolarization is quite large. And that has to do with that's contributed to by the effects of this, this sealed end of this cable because when the charge comes in through the synaptic conductance, instead of having, you know, two directions to flow through the dendrite it only has one because because one end is sealed and this results in a larger depolarization here. Now, I mean this is just a model but it suggests if we map this depolarization on to our a to release function that we collected before with the paired recordings, it would suggest that even a relatively small excitatory input ought to be able to evoke release from the a to Now, there's a little bit of a fly in the ointment here with regard to the the a to is doing this in the literature and that is that previous examinations of a to and this is a very nice paper from the Rio Pratsis lab. This paper was published from almost 20 years ago, showing that a to showing very nicely that a to express one kind of voltage gated calcium channel CAV 1.3 so sustained L type calcium channel but when they they imaged calcium signals in an a to and this is using a CCD camera and imaging the the entire cell. The majority of the calcium signals if they depolarized the a to the large majority of the calcium influx through calcium channels occurred up here in the soma and in the the proximal lobular dendrites. And they saw a very little signal down here in the arboreal dendrites and our mechanism would would be greatly aided by the presence of calcium channels down in these arboreal dendrites and and so We wondered whether they might have been these authors might have been missing those signals because they have a relatively insensitive approach to examining the calcium. This was this was very nice for for the tech for what they had then but we re approached the question using laser scanning to photon microscopy. And filling an a to American cell with a calcium indicator die and you can see when when Miloche filled one of these cells the lobular dendrites and then the arboreal dendrites here and we could, we could image down here in these arboreal dendrites to polarize the a to and see if we saw calcium signals in these arboreal dendrites and in fact we did. In some of these arboreal dendrites we saw a very large calcium signals and others. They were smaller and we think that this indicates that there are calcium channels down on these arboreal dendrites, and that they're able to respond to depolarization by admitting calcium into the cell. We don't think this is just passive diffusion down from the top of the cell because these calcium signals are every bit as fast as the signals that we record in the upper reaches of the cell. And also if it were passive diffusion we would expect to see very similar signals in all of these processes and in fact sometimes we see large responses and other times, we see small responses. Now, it's a little hard to see here but if we look at the subset of these processes that are located very close to a large cell body this is very likely to be an alpha ganglion cell. You can see that the processes that are that are adjacent to the cell body exhibit very nice calcium signals. As a consequence of this strong response, this strong synaptic input to off ganglion cells, when will records light evoked responses in off alpha ganglion cells and here's a response from an off sustained ganglion cell and this is the inhibitory synaptic conductance, elicited by relatively a small spot, small spots of varying strength but but but but quite quite gentle visual stimuli he sees a much higher synaptic gain in the, the off inhibitory conductance in an off ganglion cell, then in the on excitation in on ganglion cell. He does this experiment a little bit differently by using slightly larger, brighter stimuli but using bars to go across the receptive field of the cell he can see that these bars can elicit very powerful inhibitory synaptic conductances in the off ganglion cell that are capable of completely shutting off the ongoing background dark spike rate in an off sustained ganglion cell. Now the other thing you might notice here is that the, the, the width in space over which one of these bars elicits and inhibitory response is quite narrow. It's, it's width here is only about 100 microns, half width this, this is rather strange when we think about classical receptive fields because usually we think about in the textbooks. The center surround receptive field has an excitatory center and an inhibitory surround. And when you combine these you get this sombrero function here. That is very useful in sensory systems and other places for, for further distinguishing the center from the surround. What you hear is, is sort of an inversion of that where the inhibitory center is actually narrower than the excitatory surround. This is a little odd, and Miloche wanted to take a closer look at this so he recorded the excitatory and inhibitory receptive fields in, in off sustained ganglion cells and off transient ganglion cells this is an off sustained. What he did was he recorded the, the excitatory receptive field using play presenting dark bars on a gray background dark bars because this is an off cell and we want the excitatory input. So we put them in different, different places a lot across the receptive field and at different angles. And this allowed using a procedure that was developed by Lee and lignado, and sort of stealing from CAT scan math rate on transforms. This allowed you to recreate the two dimensional receptive fields of these cells and so this is the excitatory receptive field of this cell and then you can get the inhibitory receptive field by using light bars and you can see here that the inhibitory receptive field is in fact smaller than the excitatory receptive field. Okay, and this is shown for a group of cells down here. And the inhibitory receptive field is in general smaller than the excitatory receptive field. And in addition, that the inhibit it's a little bit hard to make a strong statistical argument here because the collection of dendritic size was was relatively tight, but the inhibitory receptive field didn't really vary with the size of the dendritic harbor and usually, usually it does and it like and as the is the case for the excitatory input, larger dendritic arbors tend to have a larger receptive field, but that's not the case for the inhibitory input. In fact, the size of the inhibitory receptive field is very similar to the receptive field of a twos. It's very narrow. And you can see the average receptive field here of a to American cells the excitatory receptive field is similar to that of the inhibitory receptive fields in these cells suggesting that the inhibitory receptive field of these cells is dominated by strong inhibition from the a twos located immediately above it. Okay, so I have shown you that American cells play favorites in the outer retina they connect primarily to type two combi polar cells, and depending on the state of the network they can transmit either transient or transient and transient information. And a twos also make powerful glycinergic synaptic connections on to the soma's of off ganglion cells and this enables a very, a very powerful light evoked inhibition of spiking and off sustained ganglion cells. So we have more to do on this but it suggests, you know, a very direct, a much more direct route for signals under very low light conditions, and I have talked longer than I wanted to. So I'll just acknowledge the members in the lab for particularly Hua and Morgan and Miloche and will who did the experiments that I talked about today in collaboration with with Josh singer Rinaldi Hoon and Kevin Brugman. And so I'll stop there and. Yeah, sorry so that I guess the first, the first YouTube talk went a little longer than expected. Yeah, thanks a lot. I'm the only one clapping but. And now I will. Now I will go and. Oh good you guys. So thanks for everything. I'm looking forward to your upcoming publication. So I'm looking forward to see that we have a large audience we have a global audience I saw people from Sydney from Jerusalem from the US all across Arab is very nice. I guess they're all clapping now and we will move to questions. I have a couple here. If you have more question, please use the YouTube chat. I have one question from an avlast states. You were saying that it to American cells are I pass filtering the RBC input, at least when they are more hyper polarized. So these things like it can be accomplished via passing filtering. What do you think would be the mechanism for specific transmission or more transient signals, just that they are much larger. Yeah, and I that's what I tried. That's what we we tried to address. Thank you Anna for the question. I hope you're doing well and doing that. That's what we tried to address with the, the paired recordings between the a to and and the off cone bipolar. So, we don't think it's just passive passive filtering, you know that that would that would if it were just passive filtering that would favor a low pass situation we think that the high pass nature of the transmission has to do with the fact that the transient component is able to to move the membrane potential along. You know to change the activation of the calcium channels and change the amount of release, whereas the sustained component is not unless you're able to move the resting membrane potential of the a to more depolarized into a more linear of a more linear range. Now see the nice thing about this is that I'm unable to see the the the confused look on my face. I can give you a confused look if you want. Oh, thank you. Yeah. I have a really good question from Anna, and she also asking, what affects a to depolarization is this going to be affected by amount of adjunction coupling, or are the off layer dendrites quite isolated from that. Yeah. So we wouldn't expect that the dendrites would be particularly isolated because all of this is happening very close to where the gap junctions are being made. We don't know yet really what what is causing these I would say we don't, we don't have a perfect idea about what's causing these light dependent changes in the membrane potential okay you could certainly imagine that that the depolarization that you see as as what comes up is due to increasing input from from rod bipolar cells that's that seems easy enough to believe but I don't know what the subsequent hyper polarization at higher light levels. It's been suggested that that could be due to inactivation of release from the rod bipolar cells inactivation of the pre synaptic calcium channels. In our hands we see very very little inactivation of release under over physiological membrane potential ranges. So, I don't really know. What is causing that entire trajectory that that they've reported yet. Thanks for that. Can I ask a question. Sorry, this is Helen from Singapore. Okay, I don't know what's coming from but let's go for it. Hi. Yeah, so, um, maybe I didn't really follow that well. There's a slide that you show the receptive or receptive few for the a two neurons. That while you mentioned a two neurons has a relative small receptive few similar to the RGC. So, is that the part that at the center shows the receptive few where you have the one color. So basically that that's that's a color code of the size of respond that's a very processed piece of data where we're recording the size of the responses that we get to bars at different at different lateral locations across the receptive field and a different orientations. So we're showing bars at five different orientations and a different lateral lateral positions, and then we dump it all into the computer and it spits out this two dimensional receptive field. And so you, it gives the images are a little odd they look kind of like a star and and and the the radiating lines are simply artifacts from from the technique. So it's, it's the, it's the color in the middle that that can be used to to quantify the size of the receptive field and we fit, we fit that to a two dimensional Gaussian. Oh, so does it means that all because it's a Gaussian so actually the bars outside is not that relevant things. The bars that's just that's, that's, oh, I'm sorry, I cut you off. Go ahead. No, it's because the color for the a two neurons, it has a higher background compared to the RGC neurons. So I wonder whether it is a kind of feature that can explain something, or as I said, just a tail of the Gaussian curve so it doesn't really matter. Yeah, I wouldn't read too much into into that, the far periphery of, of, of those, of those data. We are looking at the nature of the surround of the a two receptive field. That's, that's an interesting story in and of itself that we're working with Josh together on. No, the, you know, you might have different background levels and in that analysis, but you know, that's just that that could be due to just differences in the sort of the steady state activity in the cell we're looking at synaptic, you know, post synaptic currents and so you could see, you could see different levels and different cells. And that if you're interested in that. In that procedure, Lee and like Nato's group has published a very nice paper a few years ago in the Journal of Physiology. We get it. Thank you so much. It's a nice talk. Thank you. Thanks. Thanks. So we'll move to a couple more questions if you're happy with that Jeff. Sure. Here's a quick one from Tom badden. He has a question. He loves the upside down Mexican hat. How general do you think it might be there are just the types of the species. You know, I, I think that. For sure we need to see whether whether you know this is it is it play in other species I would say though that the that the marked similarities in the rod pathway in different species different mammalian species suggests that that I would expect this kind of thing to be present in other in other species that had that had these cells in them. It's interesting to think about whether this kind of thing could arise in other parallel circuits I mean there are, there are quite a few narrow field glycinergic amicron cells that tile the retina, you know, very densely that that could contribute to this kind of receptive field structure and other cell types. You know, we just, we just haven't. It would be interesting to look into, you know, it looks like the, you know, from our very small sample size of this small, this small data set it looks like the particular arrangement with a twos is confined to a limited subset of ganglion cells. Thanks for that. Thank you indeed. I have a question from Chase Elmer. So, Dr diamond you showed a powerful effect from a single road bipolar cell to a single a to a macron cell, because the nature of the dual patch exclude the effects of conventions for many RBCs. Do you think that this would be close and underestimate of a to output. Yeah, no, so that's, that's, that's a good that taps into a lot of interesting features of this of this system. And it is that there's a very large variation in the strength of connections between rod bipolar cells and a tos. A tos receive input from. I know I forget what it is maybe, maybe 2020 rod bipolar cells and but not, not all equally. Okay, some, you know, just one or two by polar cells make much stronger connections than others and this has been shown we've seen this scenario and data. So, Sukamoto has, has, has seen this, this appears to be a feature of the circuit and it's interesting, because it makes for a way that that very weak signals perhaps even signals elicited by responses and individual rods can be communicated with some fidelity downstream. And that is that you have a very strong synaptic connection from some rod bipolar's to a tos. And, you know, so you might lose a lot of those very sparse signals, but the ones that get through will get through with the with the higher fidelity. So, I think that we are, you're right, we're underestimating the total input to the a to, but we're being a bit more realistic with regard to the signal being transferred by a single rod bipolar cell, which is relevant under under very low vision conditions when photons are extremely sparse. So there's many questions pulling up so I'm just going to pick up a couple and if if you agree to look at them later and so on the chart directly on by mail. But I think we have to close soon. So I'm going to take a question from Petri. So Petri like your talk and he's asking if you would like to comment or speculate on the potential functional role of this a to type of CBC pathway in relation to better all. Yeah, so so this this particular pathway seems seems well suited to to transmitting very sparse signals whereas the off come bipolar seems much more useful for integrating. signals when they become more plunder fall in the higher in the higher ranges of scotopic vision. So I think that there's going to be a, you know, there, there ought to be a transition in, you know, the relative weighting of this of this central input to to the to the sustained input and I would, I would, you know, I would note that that the receptive input shows the relative location and space of the inputs from the center and the surround but when you have a distributed surround input you are activating quite a quite a number of the synapses from from the off come bipolar so so that will that will play a significant role in the higher ranges of scotopic vision. I'm waving my arms a lot. One last from. I'm sorry there's so many. I'm going to pick one from Thomas solar. You see the inverted heart in both sustained and transient of alpha cells question mark. We've done off transients a lot, a lot less often, but I believe we see very analogous analogous results maybe, maybe will can can can put something in the chats but I believe that that it's quite. It's quite analogous in the off transients I don't think we have any reason to believe that there are major differences between the two. Well, we're going to stop here so if I didn't pick everyone's question but there's a lot planning up I guess we can continue this discussion on the chat. Thank you J3 for for the stock. Thanks everybody to join this online seminars. Is it possible, just as a, as a, is it possible for people to come on afterwards to zoom. Yeah, we can share, we can share link and invite them to the zoom, so we can continue on offline. I guess we have to close now. Okay, thanks everyone. And I just want to thank so much as an it is that helped me setting up this channel. So, see very soon, we, I'm just going to share this zoom room and we can continue on. Okay. Bye bye.