 This is Caltech Teach Week in Cassandra Horry, Dr. Cassandra Horry, who is head of the Caltech Center for Teaching, Learning, and Outreach, has invited some guests to the lecture, and I'm happy to have them here. Today we are going to discuss advanced electrophysiology. There are actually three topics for today. Two of them are brief. One is inward rectifiers, and remarkably they are mentioned inward rectifiers roughly ten times in candle, but not explained at all. Second is glia. Again, we have mentioned them in this course and we will go over just the types of glia. Most of the emphasis today will be on contemporary recording and stimulating techniques, and this summary or this list is not at all in candle. And as you will see, it is very much a Caltech story in the sense that Caltech is very good at the, excuse me, I have to figure out why this PowerPoint is doing what it's doing on its own. Transitions. There it is. Okay. Institutes have wonderful groups that develop tools, biological tools. I think Caltech is preeminent in this regard, and so I'll show you some recent papers from Caltech that have been done in some cases by people associated with this course. So first of all, inward rectifiers. In a sense, the mechanism of inward rectification in a channel is in some cases, in some cases, in some cases the mechanism of inward rectification is the only mechanism for getting of the channel. It's a it's a weird one. So let's go through it. It involves the concept that we've discussed before in this course of a plug in a drain or in a pathway. And in fact, that is how the channel is often controlled. It also involves two molecules present in the cytosol in the intracellular solution. Both of them are polyamines. That is, they have more than one, in fact, more than two amine groups. And they both are typically protonated at a physiological pH, spermine and spermidine. The total concentration in most cells is fairly high around a millimolar. And using the concepts that you learned in chem one and a typical forward binding rate of 10 to the eighth times a millimolar gives you an effective forward rate constant for binding of around 10 to the fifth per second. Or one over 10 to the fifth is 10 to the minus fifth seconds or 10 microseconds. And so on the usual timescale of electrophysiology, these channels get plugged up by intracellular polyamines too rapidly to reserve to observe. So the unblocked potassium channel passes current as usual, but when the cell tries to generate outward current through the potassium channel, the spermine or the spermidine gets plugged in the channel and current does not flow. And so now you see why we call this a rectifier because it's the kind of rectifier that you in fact learned about in physics practical tract. What good is this? Well, from lecture one, we learned about potassium channels. Potassium channels that are the three dimensional structure of potassium channels. This is great for the visitors. This is terrific. We learned that they have a A in which potassium ions travel and in fact the cork can get stuck in that potassium channel. You may remember the two trans membrane domains on the left and on the right that cork gets stuck in the channel and we know a lot about where that cork binds. So what happens then is that the reason that inward rectifier channels are so useful comes back to our equivalent circuit for a nerve cell. As you'll remember, we have a lot of individual sodium channels and their conductances and in parallel, that's the little gamma NA. We have a lot of individual potassium channels, their conductances and in parallel. So we have a macroscopic GNA and a macroscopic GK and if all goes well, then the current that flows out of a sodium channel flows back into a potassium channel. Now, that is just fine. It sets the membrane potential according to the equation that we have used in this course in which the membrane potential is a weighted sum of the conductances times the Nernst potentials. However, with an inward rectifier channel, what happens then is that most of the potassium channels get blocked by this cork in the drain. And as a result, the potassium conductance is much lower and therefore to depolarize the cell, for instance, to fire an action potential or to release transmitter, we need only a few sodium channels to counteract only a few potassium channels. So as a result, if we have the potassium channels open at rest to make a robust resting potential, but then we want to depolarize the cell to fire a spike and use as little energy as possible, but the inwardly rectifying potassium channels accomplish this. So as an example, cardiac tissue, as we've mentioned before in this course, is actually depolarized for 50% of one's life. And in fact, most potassium channels in the heart are inward rectifiers. That is, in this wave form of the heartbeat, the inward rectifiers keep the channel very nicely at the normal resting potential. And then when the heart is beating, we get this spike of depolarization, which goes overshoots, and then the plateau actually is a prolonged depolarization, roughly half a person's life is spent during these plateaus. But it actually requires very few open sodium channels because the potassium channels have nearly closed, and this saves energy for the sodium potassium ATPase. So, Hilla, who wrote a very nice book on ion channels and was here about a month ago when we had our symposium, says that an inward rectifier functions like a latch on a cabinet door. So, a latch on a door takes a great deal of effort to open, but once it's open, it swings freely. So, depolarizing a cell that has lots of inward rectifier potassium channels is analogous. It takes a lot of energy when the cell is at its resting potential, but very little after that. Now, in fact, some inward rectifier channels are additionally gated by other signals in the cell, and we will discuss these later, but they are also gated by the polyamines that I told you about. Okay, so that was topic number one for today, which is inward rectifier channels. Topic number two is the types of glial cells. There are three major types, and actually, if you want to talk about the myelinating glial cells, you could say there are two major types, because the types in the central nervous system and the peripheral nervous system have different names or rise from different precursors. But their functions are very similar. In both cases, their major function is to produce myelin. Here is the myelin, these sausages. And in between these sausages are the nodes, the nodes of Ranvier. And in each case, the myelin helps the nerve impulse to hop not along the unmyelinated axon, but from one node to the next, vastly increasing the conduction velocity of the impulse. So, Candel points out, and others pointed out, that there are a great many demyelinating diseases in the nervous system. Perhaps the most common place is multiple sclerosis. The oligodendrocytes become unhappy. They lift off from the axon. The conduction velocity becomes much slower. The axon has to generate, has to devote lots more energy simply to conducting the impulse. The second major function played by astrocytes is a support role. The typical glial, and that kind of glial cell is called an astrocyte. It's mostly in the central nervous system. Here is a capillary. Here is the astrocyte. Well, it looks starse shape. That's why we call it an astrocyte. And here is the typical cell body of a neuron from a figure in Candel. So we have end-feet of astrocytes on the neuron cell body and on the capillary. And a fair number of support functions are subserved by these end-feet. So if we, oh yeah, and the glia simply means glue. So that's why we call these glial cells. The name for oligodendrocyte simply means several branches. So it has several branches like a tree and an individual oligodendrocyte or an individual swan cell can then myelinate several axons. And then you will remember that the archetypal case of the swan cell, which is that a single swan cell can cover many nerve muscle synapses at the nerve muscle synapse. And swan cells are much of the reason. Swan cells and oligodendrocytes are much of the reason that we very rarely see a bare neuron. There's usually a surrounding glial cell, for instance, in this case. And in fact, there is very little extracellular space in the central nervous system. And so the typical diagram that we have showed you in this course, which is a neuron hanging out in lots of extracellular solution, is really not true. Most of the space in the nervous system is occupied by cells. Roughly 5% of this volume in the CNS is occupied by the processes of astrocytes. And as we'll see, they provide supporting pathways to maintain the extracellular space. In more detail, here's the capillary in the brain. Here are four astrocytes sitting around the capillary. And you may remember that previously in this course we have talked about the blood-brain barrier. In the periphery, even in the peripheral nervous system, the endothelial cells have spaces between them so that small molecules and drugs in the capillaries can reach the extracellular space. This is not so in the central nervous system. There are tight junctions between the endothelial cells. As a result, molecules cannot diffuse out unless they are membrane permeant. And drug companies have special ways to make drugs membrane permeant by adjusting their chemistry. And the body has special ways also. In addition, here are the end feet of the astrocytes. They tightly surround the capillary, but not so tightly as the endothelial cells do. So probably, although the astrocyte and feet provide a, to some extent, a diffusion barrier, drugs can diffuse around them. The main blood-brain barrier then is the endothelial cells. So astrocyte membranes have lots of transport properties. They have transporters for glutamate and GABA and for several other neurotransmitters, as we've seen previously. And so this provides part of the delta function that I've told you about during so-called fast synaptic transmission. They also have transporters for other nutrients, sugars such as glucose and lactate. And this brings, obviously, nutrients from the restricted extracellular space. They also have permanently open potassium channels, although in some cases those permanently open potassium channels are also inward rectifiers. And so when a potassium ion appears outside a cell because nerve conduction has taken place, that potassium ion can either be pumped back in or can diffuse through the capillaries, sorry, through the astrocytes to the capillaries. Since the flow through the capillaries is continuous and has a low potassium concentration, there is, in fact, a diffusion gradient for those potassium ions to diffuse out of the brain into the capillaries. So these two processes, uptake and pumping of potassium, but also in cases of maintained activity, diffusion out into the capillaries are quite important. And in fact, we will discuss fMRI, functional magnetic resonance imaging, later in this course. But the major fMRI signal comes from more rapid blood flow. And part of the selective advantage of this more rapid blood flow is that it clears potassium ions out rapidly. Okay, so now we're going to talk about advanced physiology, mostly advanced electrophysiology. We'll talk about extracellular recording with pipette electrodes, with so-called tetrodes. We'll talk about wireless recording, micro devices, direct imaging. And we won't talk much about single unit recording in humans, but actually Ralph Adolf does this in his lab. And so he'll tell you about it a bit. And again, this is a very exciting topic for Caltech undergrads and grad students, since it's so well represented here in so many labs at Caltech. The general idea of pushing forward how one records from and stimulates neurons. So the simplest kind of recording in the central nervous system is of course extracellular single unit recordings. Extracellular recordings typified by this mouse, which has a single pipette electrode pushed into its brain, are the way that people recorded for decades in the central nervous system. Here is an example, we've blown up this mouse with a pipette in its brain, and here we're in the midbrain. The midbrain is a place where we have discussed examples of nicotine and its action on acetylcholine receptors. Here in the midbrain, we've diagrammed the tip of the pipette and two neurons. One of them is dopamine ergic, that's what the DA stands for. One of them is GABAergic, so it's inhibitory. We do know from various other experiments that the GABAergic cells primarily inhibit the dopamine ergic cells. And so even single unit recordings sometimes can be used to distinguish neuronal types in vivo. Here's an example of a single unit recording with one electrode in which one can excise the extracellular action potentials due to neurons firing. Because they are extracellular, they are very much smaller than the 100 millivolt or so action potentials that one sees intracellularly. And so they are typically about a tenth of a millivolt, and they are due to the currents flowing across the extracellular resistance. I times R gives you a little bit of IR. So interestingly enough, the currents when the point of the electrode, the currents produced by a dopamine neuron action, so these are extracellular action potential, actually have a different waveform from the ones produced by a GABAergic neuron. And so it becomes in some cases relatively straightforward to say, ah, that's a spike from a dopamine ergic neuron, or ah, that's a spike from a GABAergic neuron. And so here, for instance, is an experiment where the mouse received an injection of nicotine at this time point. One recording separated into GABAergic and dopamine ergic. And you can see some features of the firing frequency. So this is the number of spikes which we classify as dopamine ergic, or the number of spikes which we classify as GABAergic. Not necessarily, there certainly was only one dopamine neuron here, there were probably several GABAergic neurons. You can see that the nicotine, not surprisingly based on what we've said, causes the dopamine ergic neuron to begin firing, but in a leisurely sort of way. And when we looked at the GABAergic neuron, we realized that the GABAergic neuron fires more rapidly. Look at the different frequency scales, the GABAergic neurons fire at up to 25 hertz. The dopamine ergic neurons rarely fire at frequencies less than 10 hertz, more than 10 hertz, sorry, and most of the time they start off firing at one or two hertz. But the nicotine produces a robust response in the GABAergic neuron. And this is consistent with the leisurely firing in the dopamine neuron is consistent with this inhibitory effect of GABAergic on dopamine ergic neurons. Eventually the GABAergic neuron stops firing rapidly, probably because of the phenomenon that I told you about called desensitization. The receptors in the dopamine ergic neurons do not appear to desensitize so rapidly to nicotine as the ones in the GABAergic neurons. So eventually the dopamine ergic neuron begins to fire at spikes and the decay of the dopamine neurons firing actually probably has to do more with clearance of nicotine by the liver than by any kinetics of the nicotine receptors. And so you can see that simple extracellular recordings give you or at least confirm a great deal of information about characteristics of individual neurons. But it's possible to do much better than that. And so here from Thanos Theopis' lab, here at Caltech, is a tetrod. And so this tetrod is an array. They are so large that they go best on rat brains. And the point of this, the goal is certainly to get recordings from as many neurons as possible. Let's expand one of these arrays. And what you see now is that each of these inputs here is a small wire going to amplifiers. But if we look more carefully, we realize that each of these small wires is actually twisted from four different wires. So there are several dozen tetrods in this, in the previous picture, all going to slightly different regions of the rat's brain. And since there are four tungsten wires, you get four signals at once. And this is rather important because if we have a neuron over here next to the blue electrode, then the spikes from that neuron will be stronger at the blue electrode than they will at the black electrode, which is on the other side of the tetrod. Likewise for near the green and near the red. And so based on simultaneously recording, that's why we call this a tetrod. It has nothing to do with the old vacuum tube. Based on simultaneous recording from four wires at once, we can actually distinguish a number of neurons in the preparation. In this one, for instance, they have distinguished four neurons, neuron number one, two, three and four, simply by the relative amplitudes. And then if one records for hours, you get a pattern of number one versus of signals on the blue channel versus signals on the red channel or signals on the green versus the blue, etc. And these all cluster in various places. Not only are these tetrods useful, but they are very stable when designed correctly. So here, for instance, is a rat that's had tetrods in its brain for a couple of months. And over a period of six days, the waveforms, the relative waveforms are rather stable. And so using analyses like this, you can actually begin to cluster individual neurons so that you know which one is here. So here is a cluster of firing that we think come from one neuron, and it's distinguished in most of the graphs, but not in this one. But since we know when all the red ones occurred, we can put them, we can make them red in this pair as well. Likewise, the dark blue ones can be easily clustered, and those clusters are stable. So this obviously takes a lot of computing power. So I want to tell you about Dan Cagle, who graduated in 1986. Dan was one of the pioneers of electrophysiological instrumentation from his position in Blacker House. You can tell that hairstyles were quite a bit different than they are today, and here he is at a Blacker House party. Now, if you go to Dan's webpage, Cagle's webpage, it's very impressive in terms of his CV and what he has done. The story I'm going to tell you has, however, been sanitized from Dan Cagle's webpage, and you won't find it. In late 1983, he showed up in the lab and said, can I borrow your notebook computer over vacation, please? I'm installing a radio link, so Jerry Pine and I said, oh, sure, why not? Well, here is what he used the notebook computer for. He sat up on the hill overlooking, well, the first thing he did was to scale the wall to the Rose Bowl and put a receiver on the open collector bus leading to the scoreboard in the Rose Bowl. And then he sat up on the hills overlooking the scoreboard, and he practiced downloading bitmap graphics to the scoreboard for several nights. And then during the Rose Bowl, he actually, he and his friends actually did accomplish quite a nice stunt, but why read it from me? You can read it in the New York Times. Here we are, 100 glorious years of the Rose Bowl. This was 2013, and here we are, Caltech, zoom in. In 1938, MIT 9, their game plan explains how a university appeared in the Rose Bowl without ever actually losing the Rose Bowl. And they explained the electronics, which was actually a project and a class at Caltech and set up the portable computer. And then they got more and more ambitious. Hi, mom, and then they put two Caltech beavers. You can see the beavers actually downstairs in the Rothschiller at the Athenaeum. They appear on that sign right in the Rothschiller more than they appear on the PowerPoint that I showed you. Then the only ones not amused with the police, their lawyer whose fees were paid by a Caltech alumnus. Notice that the name of the Caltech alumnus is not in the New York Times. It is not so carefully hidden. It is not me. However, you have seen his picture in this course, and I won't say any more than that. Okay, so moving on now to more technical accomplishments at Caltech. The next possibility is not to have a single contact on each of those wires, but to make those contacts into devices, each of which is a silicon device and has a number of contact on it. And so each of these silicon devices then can have a couple of dozen contacts. Or in later years, it is now being possible to push up to 256 contacts per device, sometimes even a thousand contacts per device, a couple of tens of microns in diameter. And so one pushes them into interesting regions of the brain, and obviously the multiplexing and the electronics are fairly complicated. And because you're pushing them into the brain at random, rather than adjusting each one individually to get a good cell, the yield of cells is rather low. But here are the traces from one of these devices, and you can see these little blips here are action potentials. And as expected, neighboring pads give similar waveforms, allowing one to say yes, this is one neuron or this is another. And so this is in the hippocampus. Here is a bit of the electronics associated with one of these nanofabricated multiplexed electrode arrays. Let's see. Du was a Caltech postdoc. Reed Harrison was a Caltech graduate student in CNS. So Tires Mazmanidis, who led this study, was a Broad fellow and a postdoc at Caltech and is now a professor at UCLA. And Kim Spunk Scott analyzed the data when she was a Caltech undergraduate and is now a graduate student at MIT. So this is definitely the waveform of the future with regard to trying to understand many neurons at once. And you can put these, this particular array actually has two prongs, like a fork. You can get them with up to four prongs and put them in separate brain regions to find out who's talking to whom. And let's go one step further and make this all wireless so that it's not connected to a cable, but has a battery on the back of the, in this case, the rat. And has a tetrod array. Now we're back to tetrodes on the head of the rat. And so these experiments were done again by Thanos Siappus and by Marcus Meister, who is now also in the Caltech faculty. And back in 20, in the late, in the first decade of this century, this device weighed 40 grams using 20 or five technology. And so it was more suitable for a rat. But now you could probably do it with two grams and a much smaller battery and various clever, either low energy Bluetooth or other technologies. And so again, you can do the typical tetrod recordings and these are stable, even as the rat is outside in an open area broadcasting wirelessly. These are spikes recorded either when the apparatus is connected up with a wire or when it's wireless. They look precisely the same. You can do the kinds of clustering of individual neurons that I showed you for the wired tetrod. Now you can do it wirelessly. And the signals are good out to 60 meters from the receiving device. So clearly, we'll be able to understand what an animal is doing in its brain while it is doing natural operations foraging, having social interactions, etc. Now, what I have told you mostly has been electrical recording, but it's also possible to do a combination of electrical and optical recording. And you may remember that we've seen fluorescent calcium indicators in this course. And you'll also, of course, remember that although a cell tries to keep its interest rate or calcium low, when it does go high, that's usually a signal of activity. And that can be sensed by calcium indicators either injected into the cell or genetically encoded. And so here is a device, a chamber, which sits now on a mouse's head. And it is possible to put one electrode in to stimulate, but then one can record optically from many neurons at a time, typically with GFP. And so this is a paper by David Tank, and before he pioneered this technology, he also invented fMRI. So here, now in order to get good images from the cortex of a mouse, one needs to use two-photon microscopy. And the mouse, actually in this case, is not roaming around. He has his head fixed, and mice do not mind this at all. So his head is fixed, but he is watching a performance on a video screen. And so he has been trained to move around. And so he's watching the video screen. You can see he is grooming himself. He's connected now simply to a star. He is sitting on a styrofoam ball, which is blown up against him. And so you can monitor the mouse's movements, as well as his brain activity, while he is glued to this, which is a microscope objective. So this is done in a couple of labs at Caltech now as well. And of course, I do chair the Institute Animal Care and Use Committee. And all experiments that take place on mice and rats and other animals at Caltech are carefully monitored with regard to scientific justification and with regard to minimizing harm to animals. So what happens then when we look at a mouse like this? I think we may have some. This is a frame. That's a slow motion. This is actually a frame from one of these movies. Each of these dots is a cell expressing the indicator. And they do blink during a typical movement. I'm sorry that I can't show you that. And here then are the calcium transients as a function of time from individual neurons. They are significant in the sense that they really do correlate with electrical activity. When a mouse is running, the neuron that's monitored has a certain pattern of activity. When it's grooming, it has another pattern of activity. So we can make correlations between behavior and the activity of single neurons while an animal is doing important behaviors. So in addition then to being able to record, it's also possible to be able to stimulate selectively these days. I'll give you a couple of examples, one of which is pacemakers for a heart that needs help beating. I don't think we need to emphasize that in this course. Another example that you've seen other people use presumably is transcutaneous stimulation for back pain. A person wears a belt with a little stimulator which tingles when it starts. And so you are directly stimulating neurons there. I'm going to tell you about deep brain stimulation for Parkinson's disease. A bit today and a bit later on in this course, there are also examples of cochlear implants and of retinal prostheses. And Richard Anderson's lab here at Caltech has developed some electrode arrays that actually go on the brain of patients. And allow them to decode their movements using electrical recordings and then to stimulate regions of motor cortex to get crude motions. And so I will also tell you about transcranial magnetic stimulation, pharmacological silencing or activation of neurons and optogenetics. So as usual, we speak about a particular brain region. I want to concentrate here on Parkinson's disease which takes place in the midbrain in the dopamine neurons that I've told you about. And in these dopamine neurons, we have a degeneration in Parkinson's disease. And this degeneration prevents dopamine from being released and prevents proper movements. And the tremor, again we're going to discuss Parkinson's in more detail, but the tremor in Parkinson's probably arises from a malfunctioning feedback loop involving the substantia nigra. That's the handlebars of the handlebar mustache that I told you about. The striatum and other structures in the brain. So there's a pretty tightly controlled loop that goes awry. And in deep brain stimulation, and this is a now surgery that's performed thousands of times around the world every year these days, when puts an electrode wand into the brain and keeps it there, typically into a region called the subthalamic nucleus, which is here, the STN, or into the globus pallidus, the internal region, and one stimulates that region. And so here is an MRI showing actually, that's an x-ray actually, showing the four bands of electrodes in that apparatus, as well as the wires. The wires lead to connectors, which in turn lead to a pulse generator that the patient typically wears on his or her shoulder. And the implanted stimulating electrodes obviously cannot bring the dopamine neurons back to life, but they apparently retune this loop. And so I think probably later on in this course, let's see if I can find the video today, kids, your dog or your software, we will send it to the VLC player, which always works. Even that did not work. No matter how many times you rehearse these, they sometimes just don't work. So what one sees here in a clinic in Germany, and I will get at working in future sessions, is a patient with a tremor either moving his hands uncontrollably, or moving his arms uncontrollably, unable to walk. Now this patient has implanted a deep brain stimulating apparatus, but it's been turned off and the patient is quite debilitated with Parkinson's disease. When the physician turns on the stimulator, and this is done with a magnetic control placed near the shoulder, the patient's tremors stop, the patient can put his arms back and forth, the patient can follow commands, and then at the end of the video, the patient gets up and leaves the room. So it is really quite a dramatic change due to the deep brain stimulation and you can see many examples of deep brain stimulation in Parkinson's disease on YouTube as well, before and after. So this is pretty much what I've just told you. Now one issue is how deep brain stimulation works. We'll get to that later on. Another way to stimulate the brain directly is using transcranial magnetic stimulation, and that's used here in the shimojo lab at Caltech. And so with transcranial magnetic stimulation, we're basically using electromagnetic induction, and so this is a magnetic coil, but it's a changing magnetic field, it grows and shrinks, and therefore it produces an electric field, as you learned in Phys 1, and as a result, electric currents in the brain, and if it's done with carefully controlled geometry, you can actually stimulate neurons selectively in a region of the brain by passing currents locally. And this can either stimulate or silence spiking in neurons. Resolution is around five millimeters. The safe frequency is around a hertz because when one does this at 10 hertz, it could provoke seizures in people who are susceptible. So mostly it is a very convenient technique for activating or inactivating a region of the brain, the human brain, in a research setting to ask questions about what region of the brain is taking part. So we'll skip this, and we'll talk about a topic that I also mentioned, and now the question becomes, how do you use pharmacology to activate and silence neurons? And you may remember that about two weeks ago, I was rather excited by the fact that a Nobel Prize had been awarded for the discovery of the drug ivermectin. And here we are using ivermectin as a tool. Just a couple of years ago, the receptor for ivermectin was determined. It looks a whole lot like the nicotinic receptor and like the GABA receptor. It is in fact a chloride channel, and it comes from an invertebrate. And you remember that ivermectin is the world's largest selling antiparasitic drug. It binds right here at the interface between two subunits and allosterically opens the channel. So we know a great deal about the mechanism of action of ivermectin. And so now, because the particular channel occurs only among invertebrates, we can turn it into a research tool for vertebrates. So in experiments like this one performed by a former head by 150 TA, Shauna Frazier, as well as by the person who originally made up the problem set about the squid on Mars, Eric Slimko. Actually, Eric Slimko is now back at the Jet Propulsion Laboratory, JPL, and he's running one of the Mars lander's programs, and also by David Anderson's lab. So this molecule, ivermectin, can be used to activate chloride channels, as though it were a GABA or glycine receptor, and therefore it can inhibit firing when we give ivermectin to a genetically defined population of neurons in an animal. And so these rather low concentrations of ivermectin, spelled IVM here, can actually silence neurons on command. This silencing lasts a couple of days. And for some purposes, that's wonderful. For other purposes, you would like to silence these neurons. So here we have the injected current. Here we have the mean firing frequency. Ivermectin does a very good job of silencing. For other reasons, other experiments, you'd like to silence neurons for only a couple of minutes, and now there are techniques available for doing that as well. So made by bacteria, antiparasitic, allosterically activates the glucial channels. There are other ways of engineering ion channels, and they involve light. One can make photoactivatable caged neurotransmitters. Here's a paper by Mikhail Shapiro, who is a professor in CCE here at Caltech, and by Shona Frazier, the former by 150 head TA. And so one can make neurotransmitters that are photocaged, either photoisomerizable or otherwise inactive until light strikes them. One can turn on neurons that way. One can also use these small molecule photo switches and incorporate them into ion channels, where they can activate or inactivate the ion channel. But the most triumphant use of optical activation was discovered by Shona Frazier. Sorry, by Viviana Gradi-Naru, who is now a Caltech assistant professor, when she was a graduate student at Stanford. This is so-called optogenetics in which one takes a molecule that evolved as a cross between rhodopsin and an ion channel, typically in single-celled animals, and one now engineers it to work very well in a neuron. So in some of the early experiments, here is so-called channel rhodopsin. This is channel rhodopsin number two expressed in cortical neurons. Here is a step of light. Here is a voltage clamp. We see inward current that has a blip at the beginning, but otherwise lasts as long as the light. One can vary the amount of current produced by varying the intensity of the light. And one can vary the duration of the current produced by varying the duration of the light in a nice so-called dose-response fashion. Here is the intensity of the light versus the peak depolarization. And here is the duration of the light versus the amount of charge you can eventually put through it. And so this is a very well-controlled procedure. Since then, Gradi-Naru has actually developed ways to use other molecules as inhibitory molecules as well. And so a key experiment that you would very much like to do is to entrain, is to make a neuron encode frequencies of action potentials. Because as we know, the output of a neuron is primarily the frequency of action potentials. And so here, the number of action potentials versus the intensity of light. Here's an example in which low intensity light elicits few and higher intensity light elicits more action potentials. And the question is, if you give pulses of light, how rapidly can cells fire spikes in response to individual pulses? The answer is up to about 30 hertz. If you give pulse of light every 30 milliseconds or so, the neuron will still be able to fire that before inactivating and having refractory periods of the sort that we discussed earlier in the course. Now we return to Parkinson's disease. The problem with deep brain stimulation is actually, although it works very well, nobody knows how it works. In fact, people are not even sure whether it works, does retune the stimulating loop. This is a picture of the circuitry in the midbrain. It's fairly complex. Here are the dopaminergic neurons. They're in blue. Here are the glutamatergic neurons. They are in green. Here are the inhibitory neurons. And this is then one of the regions of the brain where inhibitory neurons are more plentiful than excitatory neurons. Here is actually a cholinergic nucleus from outside the brain, the midbrain. So the question is, how do all these come together to form the oscillations and how does one detune this feedback loop using deep brain stimulation? So here is the one possibility then is that we are exciting one or more of these neurons. Another possibility is that deep brain stimulation, which occurs at a frequency of about 160 Hz, is so strong that it is actually inhibiting some neurons. So here we have our axons passing through. And the question is, which of these occurrences is occurring? So when Viviana Gradinaro was a postdoc, she published a paper entitled, Optical Deconstruction of Parkinsonian Neural Circuitry. And I won't read you the abstract, but she used her invention, channel rodopsin, driven by a promoter, which was known to express mostly in excitatory neurons. And rather than some of her experiments were done in the midbrain, but the most interesting experiment was done by putting the channel rodopsin in the cerebral cortex, stimulating the channel rodopsin. And here in the conclusion is that what she was doing is to stimulate fibers of passage through the subthalamic nucleus rather than directly cells in the midbrain itself. And this will be very handy for developing better stimulation procedures for deep brain stimulation using light rather than electrodes. So summarizing what we've said today is that for modern neuroscience, there are a large number of techniques and these techniques run across rather a large range of time scales, 10 orders of magnitude in time scales, a large range in distance scales, 7 orders of magnitude in distance scales, and they range in invasiveness from very, very invasive to you can just see it using an external machine. So we've talked about intracellular electrodes, patch clamp. We've talked about microscopy, which allows one to visualize action potential. We've talked today about extracellular single unit or tetroderecordings. We discussed silicon arrays today, optical dives. Very handy, not discussed at all today, but is simply to destroy a region of the brain, microlesions. Also, we did not discuss today two deoxyglucose, which is no longer in fashion but is a way of telling whether a neuron has been active because it takes up more glucose. Ralph Adolphs is an expert on lesions to understand what is happening in the brain, obviously quite invasive. Typically, one looks at this in the context of human patients, but very informative. And we'll talk about PET scanning, SPECT, fMRI, and magnetoencephalography later on in this course. So we have lots of techniques available. What do we need now? For further progress, we need help with biochemistry to make better optogenetics, better biosensors. We need help with chemistry. As I showed you, some of this work goes on in CCE here at Caltech. A lot of developing new biochemical probes involves industrial-scale drug screening, so-called chemical neurobiology. Alice Su is involved in some of this now. We need help with mice for getting better techniques for more efficient genome engineering to put probes in mice, such as the CRISPR's technology. And most important, we need talented, exciting young people like you. See you Friday.