 Good morning. I hope everyone had a good weekend. Today we are going to talk about three related topics, synaptic inhibition. We've not discussed that at all, cable properties of neurons, and then putting it all together, electrical integration inhibition cable properties. We'll use as an example the cerebellum. Here are the readings, and you'll also see that many of the illustrations in today's talk come from Candela, and so you can peruse those chapters as well. We're going to talk about the pentameric, GABA, and glycine receptors. As we'll see, they look a lot like acetylcholine receptors at the nerve muscle synapse. As we've said over and over again, the nerve muscle synapse is a model for all sorts of aspects of synaptic transmission, both function and development. Indeed, as we'll see, the GABA and glycine receptors are not cation channels like acetylcholine receptors. They are permeable to anions, mostly chloride of course, which is the dominant anion, negatively charged anion in the body. They are in fact so close to the structure of acetylcholine receptors that you can change a cation channel into an anion channel with only one amino acid change per subunit. Remember, the story about the five Caltech professors with their knees in the middle of the channel, they all apparently rotate their knees or straighten them up simultaneously, and yet the anion going through the channel can only be a cation ion, or in the case of a GABA receptor, it can only be an anion. Changing one of those amino acids very near the knee does the trick, though the selectivity is not so good as what a real GABA or glycine receptor would do. So GABA, gamma amino butyric acid, is the major inhibitory transmitter in the brain. Glycine is the dominant inhibitory transmitter in the spinal cord and in the hindbrain. GABA, actually GABA A receptors are the ones that look like nicotinic acetylcholine receptors. There are also G protein coupled receptors that are activated by GABA, just as there are G protein coupled receptors that are activated by acetylcholine. The GABA A receptors are more variable than the glycine receptors in subunit composition and therefore in kinetic behavior. As far as we know there are no G protein coupled glycine receptors among vertebrates or among invertebrates. Now I had to talk by analogy with nicotinic receptors until about a year and a half ago, at which point the crystal structure of a GABA A receptor was published. Very exciting observation and you've seen enough pictures of nicotinic acetylcholine receptors to know, not surprisingly, that at this level of resolution, a GABA A receptor looks pretty much like an acetylcholine receptor. It's got an extracellular portion which consists mostly of beta strands where the ligand, where the transmitter binds. I'll show you some more views of that in a moment. It's got a transmembrane domain, mostly consisting of alpha helices. And then it's got an intracellular, so here we have out and in, it's got an intracellular part which is not really very well resolved in most of the x-ray crystallographic structures, just like unfortunately for nicotinic receptors. And in fact here it's been cut off to make a construct that will crystallize. Now if we, if instead of looking from the side of the membrane, we look from the top of the membrane, we see the usual pentameric structure. We see the transmembrane domain lining what appears to be an ion channel. And we see, well this part is not so important. So again it's a homomer, this particular one is a homomer of five GABA A subunits. Remember I told you there were lots of different kinds of GABA A subunits. This one is the GABA A beta 3 subunit. And so it is quite gratifying to see this sort of structure. Now you remember we also said when we talked about nerve muscle synapses that an excitatory a synapse pushes the membrane potential toward the reversal potential called eREV for the synaptic channels. And so we get a push more positive to the threshold for canion channels, more negative with a reversal potential of around zero. So we would expect and it is indeed true that at GABA and glycine receptors which are permeant to chloride, we push the membrane potential toward the Nernst potential for chloride, which is pretty near the resting potential in most cells around minus 70 millivolts. Remember we did say during the lecture and development that in some immature cells the reversal potential is a little more positive. We won't go into that now. So here we have a push toward the resting potential, perhaps a little more hyper polarized than the resting potential. And so this would prevent a cell from firing spikes and there's a great deal of analysis of the way excitatory and inhibitory influences some on a neuron. Any questions so far? Okay. I'll give you a little sampling of what you'll learn about for neuro pharmacology in January of 2017 when I teach by 155. But there is a very rich and useful pharmacology of GABA receptors. There are, first of all, the activators. GABA is itself an activator and agonist from the Greek word to act. It's the natural agonist of GABA receptors. And just like acetylcholine and nicotine acetylcholine receptors, GABA actually binds at the interface between two subunits. How that actually causes the gate to open, we're not sure yet, but it does so. In addition, there is a very interesting large class of molecules that binds at the interface between other subunits. So this particular GABA receptor is a hetero pentamer to alpha one subunits to beta two subunits and one gamma two subunits. Remember, we said that the GABA receptors have all sorts of subunits and have varied kinetics and varied localization. Well, Valium, also called diazepam, binds not at the interface where the transmitter binds, but at another interface. And it is, you may have learned, and by eight by nine, about allosteric modulators. You certainly will learn this in biochem 110. So Valium is an allosteric modulator diazepam, sorry, is an allosteric modulator of the GABA receptor. There are many allosteric modulators of GABA receptors. Some of the trademarks are Ambien and Lunista. As usual, we don't remember trademarks in this course because there are various types of GABA receptors. Some of the diazepam-like molecules decrease anxiety. Others are pretty good sleeping pills and others have more of an sedative effect. So there's a rich field of ligands of molecules that bind to GABA receptors at the allosteric site. And this, what an allosteric modulator does is not to open the channel by itself, but to help the natural ligand GABA open the channel. Also, phenobarbital binds to GABA receptors and activates them. It's not certain yet where phenobarbital binds. There are also blockers for GABA and glycine receptors. They can bind in one of two places. Again, you'll learn more about this in by 155. Some of them can bind at the agonist site, at the GABA site, and prevent it from binding. Those would be competitive antagonists. A good example is called bicuculine. And as at other ion channels, there's a class of blockers that binds like a cork in a drain or in a wine bottle. And among those blockers are picrotoxin, which seems to sit right inside the channel and prevent conduction. Those GABA A blockers are relatively toxic drugs. The good example, for instance, about picrotoxin is that it causes spasms and convulsions. So they're not very useful clinically, but they are useful for research, for proving what sort of receptor or channel we have, or for emphasizing excitation over inhibition. Any questions? And again, we discussed the controversies about how the receptor transduces binding into channel gating, whether it is those professors rotating their knees or straightening their legs. Not yet clear, both ideas are also in play for GABA A and glycine receptors. The main problem is that we don't have x-ray crystallographic structures for a given receptor in both the open and the closed state. But it's really remarkable that we do have these structures in the first place. So we have pretty much completed our survey of synaptic receptors. We've talked about the family that includes acetylcholine, serotonin, and GABA. There are other members of the family, which we mentioned, for instance, the day the Nobel Prize for Ivermectin was announced. We noted that there are invertebrate glucial channels that are allosterically activated by Ivermectin, and then there are some other invertebrate channels. We also talked about most glutamate receptor channels, which have a totally different topology, a totally different number of subunits, except for the invertebrate glucial channels. And then we mentioned that there is a third class of synaptic receptors, synaptic receptors for ATP, and we're not going to have a chance to discuss those in any kind of detail in this course. They have a different membrane topology per subunit again, and they also have only three subunits per receptor. Quite interesting molecules. Now we begin to put together our knowledge about individual molecules, and we start thinking about how a neuron works. And so here in a figure that's very much like the one in Candel, but rotated, we talk about parts of two generalized central nervous system neurons, the presynaptic neuron and the post-synaptic neuron. And all of the parts are here. We have the cell, the, first of all, the inputs, the dendrites, dendrites comes from the Greek word for tree, and the apical dendrites are the furthest away there at the apex of the cell. We also have the basal dendrites that are nearest to the cell body. We have the initial segment of the axon, which is where the action potential is mostly generated. That's called the axon hillock, because it's a little wider, a little hill. We have the cell body itself. That's the soma. We have excitatory terminals, either on the apical dendrites, not drawn here, or on the basal dendrites. We have inhibitory terminals. So the excitatory terminals would be secreting glutamate. And the inhibitory terminals would be secreting GABA, or in some cases glycine. And then we have the axon, including the myelin in many CNS neurons, and the space at roughly one millimeter spacing is where there's nothing. So it's called a node. And the question is, who is the scientist who discovered the node of Ronvie? But I'm learning my code. Okay, so then the axon bifurcates and synapses onto a postsynaptic cell, where that's now presynaptic terminals, and we have the whole process that goes on again. Any questions? So as usual, we like to take a nucleus or a region of the brain and emphasize that region of the brain. So we've discussed the midbrain in the context of nicotinic receptors, and we've discussed the hippocampus in the context of glutamate receptors. Today we're going to talk about the cerebellum in the context of GABA-ergic transmission. There certainly is GABA-ergic transmission in the midbrain, the dopamine-ergic neurons, and there certainly is GABA-ergic transmission in the hippocampus. But in a sense, the most regular nucleus in the brain, the most regular structure in the brain, is the cerebellum. Here it is in most rodents. It's in back of the cortex. Here in the human brain, it's actually below the cortex, and here's the spinal cord. It has highly regular structures, highly regular layers, and a stereotyped set of projections from one cell to another all over the cerebellar cortex. In fact, one of the cell types in the cerebellum, the so-called granule cell, contains a plurality. It represents a plurality of the cells in the human central nervous system or in any central nervous system. Not a majority, but it is a more common cell type than any other cell types. In fact, probably 10% of the neurons in the central nervous system are cerebellar granule cells. Well, if you have that many, they're all individually very small, which is why they're called granules. Those granule cells have dendrites locally, and the granule cells are located fittingly enough in the granule cell layer. Then there is another hero of the story, very few of these, but very large cells, the Purkinje cells, and you know who discovered the Purkinje cells. And then the outer layer of the cerebral cortex is called the molecular layer, not because it keeps molecular biologists busy or that because that's where molecules live or that's because molecular biologists live. But in the old days, with just light micrographs, there were all of these little dots. People said, oh, this is the smallest component, so we'll call it molecular too bad, but that's what it's called. And then down below, projecting to other regions of the central nervous system is white matter that is myelinated axons. Typically, white matter in the brain, not the one that you saw that Ralph passed around that was pretty well fixed, and you couldn't tell. But in most cases for unstained brains, myelin is much lighter in color than the cell bodies. So there's this wonderful circuit that has really played a large role in understanding neuroscience. The Purkinje cells, those giant cells, are quite interesting because you can think of them as highly elaborate hands. If you look at them face on, the dendrites ramify over hundreds of microns, all within a given plane. Those dendrites all intersect the axons of the granule cells up here in the molecular layer of the cerebellum. But if you look at them sideways, all of those hands line up this way and they look very thin. So here we are sideways, here we are face on. The cerebellar folia do twist and turn, so what is face on from one view may be sideways from another view, but this structure on a local scale exists. Here are in an electron micrograph is the molecular layer. Each of these slightly sub micron diameter structures, roughly half a micron, are in fact the axons of individual granule cells. These are not synaptic vesicles, they're axons, unmyelinated obviously, packed as closely together as they can be. Any questions about the structure of the cerebellum? Obviously highly interesting and entertaining structure. But as Ralph Adolf said, as interesting as the cerebellum is and is regularly organized and as well studied, a somewhat embarrassing point is that we're not actually sure what it does. So people have lived without the cerebellum. And animals live without the cerebellum. Obviously it's there to fine tune one or more functions of the nervous system, probably sensory, but we're not sure exactly what. Now, candle shows in one of the early illustrations in chapter 10. He shows types of synapses. There's type one on a spine. There's type one on a shats dendritic shaft. And then there's a type two on the cell body. So type one is. Axo dendritic. Sorry, type or Axo spine. The type two is typically exosomatic. Now, if you look very carefully in the electron micrograph, you can see differences between the type one and type two. The synaptic vesicles in the type two are oval and in the type one they're round. That's pretty much an outmoded terminology. So we're not going to mind the type one and the type two stuff. But we should know that there are synapses on dendrites called Axo dendritic. And synapses on the cell body on the soma called exosomatic. Okay, so we're not going to emphasize the type one. All right, so we have a general view of the synapses in the nervous system, but they look like where they are. Now we're going to talk about integration of signals by synapses. Remember the typical CNS neuron has about a thousand synapses. In most cases, the output of a synap of a neuron is a firing pattern. And so the neuron really integrates all of those inputs and either does fire a spike or a train of spikes or does not. And to some extent, this depends on whether those inputs exceed threshold or don't exceed threshold for firing a spike. Everyone clear on this? Okay, so there are several types of synaptic integration. There is temporal integration and temporal integration between synaptic inputs itself can have two causes. It can be caused by molecular lifetimes and it can be caused by the fact that the membrane is a capacitor and filters signals spreads the mountain time. We'll give examples of that. Then we can have spatial integration. Think of the Purkinje cell with all of those dendrites. So we would have integration between an input on one of the dendrites and an input on another dendrite and the cable properties of those two dendrites would play a role in how they sum. And finally, we can have what we have learned today. We can have integration of excitatory and inhibitory inputs so that in the absence of inhibitory inputs, many excitations would get to be super threshold. But in fact, because of inhibitory inputs, many fewer get to be super threshold. And there are regions of the brain where a majority of the synapses are in fact inhibitory. Because there are so many excitatory granule cells in the cerebellum that's not true in the cerebellum, it is true in regions of the midbrain. Now, so let's go over the method, the kinds of integration. In a previous lecture, we talked about synaptic integration due to molecular lifetimes. That is, the concentration of acetylcholine at the nerve muscle synapse is a delta function in time. It's less brief than a millisecond. And yet the current in the synaptic cleft or the current due to the receptors is longer than a millisecond because those channels all get opened nearly synchronously, but they close stochastically with a time constant of about a millisecond. We talked about the transition from an open state to a closed state, and we said that that transition is rather leisurely, at least on a time scale of milliseconds. In some cases, channels stay open with transmitter bound for a millisecond, others for 10 milliseconds, and MDA receptors even longer for tens of milliseconds. That all has to do with the transitions between open and closed states and the eventual dissociation of the transmitter. All of this works like a single exponential because of acetylcholine esterase, the enzyme which hydrolyzes acetylcholine into acetate plus choline and is very, very efficient. So as molecules of acetylcholine leave the receptors, acetylcholine esterase can take care of them and hydrolyze them. They don't act again as long as acetylcholine esterase is present. If acetylcholine esterase is not present or it has been inactivated as in some insecticides or nerve gases, then the consequences are very severe. Okay, so we've talked about integration due to molecular lifetimes. We've talked about the units. Now, that was the story for acetylcholine esterase, which is present at high enough densities so that it can hydrolyze the acetylcholine. There is also usually a delta function of glutamate and GABA at CNS synapses. Now, there is no enzyme like acetylcholine esterase at CNS synapses. Nothing hydrolyzes GABA and glutamate. Instead, the GABA and glutamate signals are terminated just about as rapidly as acetylcholine by transporters. In fact, by sodium coupled transporters, which harness the difference between extracellular and intracellular sodium to drive GABA or glutamate. And those two are present at densities of more than a thousand per square micron. I'll show you how we know that later in the talk today. Here in this schematized view of the CNS synapse, we have the glutamate transporters. Actually, not surprisingly, there are several genes for glutamate transporters. Some of them are actually on presynaptic cells. Some of them are on glia. The same is true for GABA transporters. There are several families of GABA transporters, some are on glia, some are on presynaptic cells. Those transporters that are on presynaptic cells obviously help to replenish the transmitter, which then gets pumped back up into vesicles. But there's also this very efficient elimination of transmitters that occurs on glia. It can't be reused, but it does sharpen up signal transmission in the central nervous system. So we have these delta functions of transmitter, which typically produce inwards synaptic currents. And then we have the capacitive filtering that simply is due to the fact that once you put charge on a membrane, because the currents have flowed and charged up the membrane, it takes a while for that charge to leak back off the membrane, and therefore the voltage for the depolarization to decay back to zero. So the simple fact that membranes are RC filters and have time constants spreads out the synaptic potentials and allows them to sum more distantly in time than one might think naively. So here we have temporal summation in a cell. We have two synaptic currents, each of which are relatively brief about a millisecond with the decay that I told you about. But the synaptic potentials can have time constants in some cells as long as 100 milliseconds. So even though the synaptic current channels remain open for about a millisecond, the voltage in the membrane stays depolarized for tens of milliseconds. So if we have two synaptic currents, they can add partially so that the second one can reach threshold, even though the first one, by adding to what remains from the first one, even though they are so separated in time. If the membrane time constant is briefer only 20 milliseconds, for instance, then these two synaptic potentials would not sum with each other. So the concept of spatial summation in a cell for good or for bad depends strongly on the passive characteristics of the membrane. Any questions about this? And then we can have spatial summation. So here we have a cell body and it's usually easier to place an electrode in a cell body and to record from that cell body than it is in the dendrites, although later on in the talk I'll show you that you can do that. And also important from the encoding properties of a neuron. So if we have a synaptic current at point A and a synaptic current at point B, there is a possibility that they can sum right here in the soma if the length constant of the cable is long enough. Typically in a neuron, the length constant is on the order of a millimeter, sometimes half or a third of a millimeter, usually not very much larger than a millimeter. And so that will determine whether or not those two inputs on the dendrite can sum to reach threshold. So a short length constant does not allow them to sum a long length constant does allow them to sum. We're doing okay so far. Now actually there are additional delightful complications. If dendrites were completely passive, they would act like leaky cables. So if for instance, but as you'll see they're not, but let's assume for this slide that they are, if we put an electrode in the soma and have an excitatory postsynaptic potential right here near the soma, we get an EPSP, we can measure it, so that measured in the soma, the EPSP decays and is much smaller out here. And so we have a length constant of something like a millimeter here. On the other hand, if we have an input here at number six and measure the EPSP, then locally the EPSP is much larger, but we won't be able to measure it very well in the soma. So we get this peak right here and then it decays pretty rapidly. Any questions? Okay, so with these concepts you would expect dendrites to passively integrate inputs without any fancy sodium channels. So if we have two simultaneous co-localized excitatory postsynaptic potentials, then we expect them both to look like this in individual trials. One is a little larger than the other. On the other hand, if they sum, if they occur during the same trial, they can sum pretty well and they might get over-threshold when recorded in the cell body. If they occur simultaneously but at different places on the axon, they can also sum in the cell body, although the rising phase is a little more broad, and they could, so these are spatially distinct. And so it's possible just using the fundamental equations of electricity to simulate stuff like this. And there are lots of programs out there in the literature that do that. So here we have a cell body with three dendrites. This is a simulation program called Neuron, but there are others. And what we do is to have these three simulated inputs, they are occurring at different times, and one of them is able to produce an impulse. Let's run it again. Check on your browser, click on your browser's refresh button. Let me decrease the, let's see if clicking on refresh works. Yes, it doesn't. We have to go back, and this is the current, the voltage in time, and this is the voltage in space at various regions along the dendrite and parametrically in time. So let's run it again. Big depolarization was an action potential. So the more we know about the characteristics of a neuron and the more we know about the ion channels in that neuron, the more we can simulate about the encoding properties of that neuron. And there was just an article by a group in Europe called the Blue Brain Project, which is simulating an entire column in sensory cortex, all of the neurons in that column. But the fundamental units of that simulation are very much like the ones that we have talked about today. There is a complication and this complication is that the dendrites are not passive structures, simply with resistors and capacitors and with neurotransmitter gated channels. Actually, if you look at them more carefully, and you can do that now because you can fill a cell with a fluorescent dye by patching onto it with a, onto the soma with a patch pipette that contains a fluorescent dye. So here's the soma. Or you can light up all of the branches of the cell by having just that one cell or just a few cells express GFP, and so everything in the cell will light up. And then, of course, this is done deep inside a several hundred micron diameter slice of brain. These days it can also be done in a living animal, but the success rate is very, very low. But if it's done correctly using two photon microscopes, you can image the dendrites of a neuron. Not yet it's axon. The axon is really too thin to be visualized correctly. But you can visualize the primary dendrites of a neuron and therefore come in with a patch pipette and come in and patch on the primary dendrites of a neuron a few microns in diameter and therefore record voltage from those primary dendrites. And lo and behold, there are action potentials in those dendrites and in some cases they overshoot from over zero. And so it's pretty sure that there are actually electrically gated sodium channels in those dendrites. And to be absolutely sure, here's an experiment in which a person patched onto a dendrite, did not break into the dendrite, but kept the patch pipette extracellularly and recorded single channels. And so here now you're only depolarizing a local region of the dendrite, not the entire dendrite. And it looks for all of the world as though when you depolarize in the, and I showed you simulations of this sort of experiment a couple of weeks ago, and when the experiment is done with depolarizations, you do see individual single channels. These are blocked for instance in the presence of tetrodotoxin. And when the experiment is done repeatedly over a time course of several minutes, the waveform looks just like the waveforms of voltage clamped sodium channels. It depends on the depolarizing voltage. Here is a small depolarizing voltage. Here's a large depolarizing voltage. Here's an intermediate one. So clearly there are sodium channels in dendrites. They are certainly at the axon hillock. In fact, the region of the axon hillock, which is going to give rise to the axon, is the densest sodium channel population. But out here on the dendrites there are also sodium channels. And so that means that voltage gated sodium and calcium channels in dendrites allow the action potential to back propagate into the dendrites and produce much larger signals than they might otherwise. So here is an overshooting action potential at greater detail in a dendrite. You can't patch individual spines, but you can patch the dendritic shaft. And so this implies that parts of cells can actually process signals semi-independently. So not only does the neuron itself make a decision about how to fire a spike and go to the next cell, but the neuron can also have individual isolated parts that process signals. So this is very exciting, and we should stay tuned to see how this works out. Now, let's talk about the veto principle of inhibitory transmission. Because of the way currents flow, inhibitory synapses work best when they're near-electrically the excitatory event that they're going to inhibit. Near means less than one cable length. So this means that on dendrites inhibitory synapses do an excellent job of inhibiting EPSPs, and excitatory post-synaptic potentials on nearby spines, but typically not on far away spines, perhaps on different dendrites. And it also means that inhibitory synapses on cell bodies and initial segments do a very good job of inhibiting spikes wherever they might arise in a cell. So in some cases, there's a selective advantage of local excitation and inhibition in a part of a neuron. In other cases, there's a selective advantage of vetoing everything in a neuron by an inhibitory synapse. The best example of a veto, well, one example of a veto, we're going to move away from the cerebellum for a couple of minutes, but in the cerebral cortex, there is the so-called chandelier cell. The chandelier cell is an inhibitory neuron. It makes GABA. You can see why it's called a chandelier cell. It seems to have these little candles. So this is one cell, one chandelier cell. It's axon branches. The axon is this little thin thing here. And it branches many times, produces these little candles at many different pyramidal excitatory neurons. So here, this is the schematic of this. You can clearly see the dendrites of the chandelier cell. There are apical dendrites and basal dendrites. But the axon is this very thin branched device here. And the axon makes these cartridges on the axon hillock, lots of pyramidal cells. The axon hillock is where spikes arrive. So this is a great example of the veto principle in which an inhibitory neuron can prevent an excitatory neuron from firing spikes. And in fact, this inhibitory chandelier cell can prevent many excitatory neurons. Simultaneously from firing spikes. Any questions about this figure? So there are veto synapses in the cerebellum, back to the cerebellum. The most obvious veto synapse would be right here on the cell body of a Purkinje cell. And that is where we're going to look. In fact, the inhibitory veto GABA-ergic synapses are so obvious around a Purkinje cell. Here, you can imagine that there's a Purkinje cell surrounded by these terminals. Actually, in the light microscopes, they look like paintbrushes surrounding the cell. And so the French neuroanatomists who discovered this called the pesso or paintbrushes. So we are going to study, we're going to show you some diagrams now. of GABA-ergic paintbrushes or pesso. To do this, we're also going to address the question, the conclusion that I gave you about 20 minutes ago, which is that there are large numbers of GABA transporters, large enough, locally dense enough to remove all of the GABA when it has, after it has acted on a GABA receptor. So here is a fusion protein between a GABA transporter. At the time these, actually it's still true that nobody knows the three dimension, the X-ray structure of the GABA transporter, but we know it's topology well enough to know that there are 12 transmembrane domains, which way they go, which ones are important. This particular GABA transporter has been fused to a molecule of GFP, even though it's blue here, take my word for it, it's a molecule of GFP. And this allows one to visualize the GABA transporter wherever it occurs. And this is a knock-in mouse, a transgenic mouse, whose GABA transporters have all been replaced by GABA transporters fused to GFP in the head. So here is a sagittal section, remember the sagittal plane is this way, it's of the head of the mouse, the central nervous system, the brain. Here we have the olfactory lobe. Here we have the cortex, which in a rodent is not too impressive. Here is the hippocampus looking like the curled seahorse. Here's the midbrain. Here's the cerebellum, the hero of today's talk. Here's the hindbrain leading to the spinal cord. So you've heard about all of these structures. They're all fluorescent green in this knock-in mouse containing fluorescently labeled GABA transporters because there are GABA urgic synapses and GABA transporters many different places in the brain. Any questions about what this slide means? So what we're going to do is to zoom in on the cerebellum. That's where we want to be counting the density of GABA transporters. So here is a individual folium of the cerebellum. And you can see, if you concentrate on this cerebellum, that there are layers of bright green. Those are really the cellular layers and layers of background. And that's the white matter. That's where the axons are, not the GABA transporters. If we zoom in a little more, here we have the... This is an antibody to the GABA transporter. Not very quantitative, but nonetheless it gives us an image of where the GABA transporters are. Here in the granule cell layer, below the purkinje cell layer, there definitely are GABA transporters. You can see some reticulate gray here, not so many. Here is a purkinje cell. There are five of them in this picture. Here are the pesto, the paintbrushes, which are inhibitory terminals containing GABA transporters to take that GABA back up. It stains very heavily. Here's the molecular layer, the so-called basket cells, which are also our GABA retic. And so what we've done here is simply to take a negative photoshop the image, so that the parts that stained most brightly are now lightest. And here is a light micrograph of one of the knock-in mice with fluorescent GABA transporters. The great advantage of having the fluorescent GABA transporters is that they can be counted and quantified much more precisely than immunocytic chemistry. But you can see that the general localization seems to agree well between the antibody staining and the GABA transporter. There is a little bit of reticulate staining in the molecular layer, lots in the pesto, and a fair amount in the molecular... sorry, a little bit in the granule cell layer, lots in the pesto, and a fair amount also in the molecular layer, which is because there are lots of inhibitory terminals on the dendrites of the Purkinje cell. And in fact, if we zoom in and look at the dendrites of a Purkinje cell, we're zooming in again, what we see is the presynaptic terminals of GABA-ergic neurons. Now you can see why synaptic terminals are often called butons, because they're like little buttons. So this now is 50 microns across, and the butons are on the order of a micron or two across. These light areas that look a little bit like the Milky Way are actually the radial glia cells, which Ralph told you about, that serve both as functions of glia and also developing neurons migrate along them. You can see that the butons are connected in many places by strings of green, so actually each presynaptic basket cell has an axon, and on that axon are strings of butons. The axon itself also has GABA transporters in it. And so it is now possible to use quantitative confocal microscopy, calibrating this with individual fluorescent GFP molecules to show that the transporter density is around 1,000 per square micron. And if we now zoom in on an individual buton, well, that's interesting too, but not in the context of today's talk. We'll stop here. So my office hour is as usual. The red door. See you on Wednesday.