Optogenetics

00:00:11.29 Hello, my name is Karl Deisseroth. I'm in the Departments of
00:00:14.26 Bioengineering and Psychiatry at Stanford University.
00:00:18.13 I'll tell you about optogenetics today, what you can see
00:00:22.21 here is a painting, actually, done by one of my graduate
00:00:26.08 students, Emily, who you'll hear about later, representing
00:00:29.29 optogenetics. And then here on the right, you can see a
00:00:32.27 mouse that's actually in the process of undergoing optogenetic
00:00:36.22 intervention. Let's get into some detail on how exactly
00:00:40.12 this technology works and what it is.
00:00:42.19 Optogenetics is combining genetics and optics to achieve what
00:00:47.20 you might call a gain or loss of function of well-defined events.
00:00:51.24 But not of genes, instead of things like trains of action potentials
00:00:57.24 within specific cells of living tissues, even within freely moving
00:01:01.28 mammals. What you can see here are some of the organisms
00:01:07.00 which give rise to the single component microbial opsin genes
00:01:11.18 that we find to be useful in optogenetics. These are single genes
00:01:15.02 that both receive light and give rise to ion flow. All encompassed
00:01:21.03 within a single open reading frame. One thing we found
00:01:25.11 back in 2004 and 2005 was that we can take one of these
00:01:29.04 genes from this organism here, Chlamydomonas reinhardtii,
00:01:32.05 the gene was channelrhodopsin and we could put that into mammalian
00:01:36.20 neurons. And here are two separate neurons receiving the same
00:01:40.06 blue light pulse chain, and we can elicit action potentials
00:01:43.19 coming from these two neurons that are almost identical.
00:01:47.25 So this is what you might call a gain of function
00:01:50.05 of the neural code. Now, one can deliver loss of function
00:01:55.15 as well, in fact, one can deliver hyperpolarizing or inhibiting
00:01:59.28 interventions, for example, with this chloride pump.
00:02:02.17 The combination of these different kinds of capabilities give
00:02:07.14 rise to the following situations. Instead of a case with electrodes,
00:02:10.28 where you have high speed interventions but poor precision
00:02:16.02 in terms of the target, you can't control which of these circuit elements
00:02:20.14 is being controlled. With optogenetics, you can capitalize on
00:02:23.17 the specificity of introduction of gene into one element or
00:02:31.22 another and deliver precise excitation or inhibition.
00:02:35.03 Even bathing an entire tissue in light, one can turn cells on
00:02:39.14 or off, one can modulate biochemical events with substantial
00:02:42.14 temporal precision, and precision with regard to the kind of
00:02:46.29 intervention that's experienced by the cell.
00:02:49.00 Now, this encounters some technical challenges which took a few years
00:02:54.10 to solve. One issue is, how do you get light deep into mammalian
00:02:58.00 brain tissue? This is a scattering problem and an absorption problem.
00:03:03.11 For example, you can see in this lower curve here. There's a very steep
00:03:08.03 drop off of light powered density, or radiance, as a function of depth
00:03:12.10 in the tissue. It's a log-log scale and you can see you're down to
00:03:15.11 1% of your initial light powered density at only 1 millimeter depth
00:03:18.27 in the tissue. It's true, red shifted light will penetrate more deeply
00:03:22.27 than blue light. But still, the essence of the problem is the same,
00:03:25.23 as you can see there. The way we solve this is, for example,
00:03:30.20 with fiber optics. We can use fiber optics to access any area of the
00:03:34.04 brain that we want in freely moving mammals. And this
00:03:38.07 affords us substantial technical versatility. These tools originated
00:03:45.11 in 2007 and we published a recent review of the latest technology
00:03:49.24 in 2011. You can see the references there. Now there are other challenges,
00:03:54.15 as well. For example, how do you target with versatility and
00:03:57.25 generalizability with the tissue? Now there are just with the very fact
00:04:02.11 that you're introducing the opsin into the tissue, you gain some specificity
00:04:05.25 already. And that's illustrated in the top row here.
00:04:08.13 What you can see is if we introduce the opsin gene into, for example,
00:04:12.28 brain region B, we have specificity when we deliver light because
00:04:17.00 only the cell bodies that are present in that region are light sensitive.
00:04:22.01 The other cells that may be present elsewhere in the brain
00:04:24.21 that are sending in their axons, those axons will not be light
00:04:28.12 sensitive. And so you have a specificity that you would not have with
00:04:31.13 an electrode. An electrode introduced into that region will
00:04:33.29 drive those axons as well as the local cell bodies.
00:04:36.24 So there's already a kind of specificity, which is quite useful.
00:04:41.19 Now, one can take that to the next level. For example, as illustrated
00:04:45.24 in this middle row. One can use genetic targeting strategies,
00:04:49.11 such as promoters or recombinase driver lines and recombinase
00:04:52.22 viruses to make light sensitive only a subset of cells within
00:04:56.29 that tissue. And then what one can do is deliver light at only
00:05:01.03 the targeted cell population within that region that's being controlled.
00:05:04.10 Of course that requires some genetic knowledge, either of a promoter
00:05:08.20 or having a recombinase driver line. Another targeting strategy
00:05:13.23 which can work together with all of these, but requires no
00:05:17.00 genetic information at all, is what we call projection targeting.
00:05:21.00 And there, you introduce your opsin gene and one delivers light
00:05:25.23 not to the transduced region but to a downstream region, where
00:05:29.22 axons might be sent in. And this allows you to recruit the population
00:05:34.18 of cells that does indeed send axons to region A.
00:05:37.28 This is quite versatile, you only need to know the anatomy.
00:05:42.05 It's important to stress, as we have, that the entire cell may be
00:05:47.02 recruited. A cell that resides in B and sends its axon to A
00:05:50.12 will also send perhaps collaterals to other brain regions, as well.
00:05:55.02 And what you're doing is recruiting a cell to find by its projection.
00:05:58.25 One can deliver inhibition or excitation by this method, and it's
00:06:02.27 quite versatile. So this panoply of methods of targeting have turned out
00:06:09.08 to actually be practical in mammals. In 2007, we were able to
00:06:13.07 achieve gain of function of mammalian behavior. This was in
00:06:18.27 collaboration with our colleagues, Luis de Lecea and Antoine
00:06:22.14 Adamantidis, and Feng Zhang. Together, what we found was
00:06:25.14 we could target a population of cells deep in the lateral hypothalamus,
00:06:29.11 the hypocretin neurons. We used a virus with a specific promoter
00:06:33.27 that allowed us to target the brain cells that we thought were
00:06:37.19 modulating sleep-wake transitions. And we found that by playing
00:06:41.18 certain patterns of activity but not others, we could drive
00:06:47.05 this complex behavioral state transition. Now in this case, our targeting
00:06:51.09 strategy was, as shown in this virus, a 3 kilobase pre-pro-hypocretin
00:06:56.09 promoter fragment. One can be much more versatile, since the
00:07:00.29 number of promoters that are small enough, strong enough, and specific
00:07:04.02 enough to fit into a virus are quite small that are practical.
00:07:08.26 What we can do is capitalize on cre-driver lines. Here's the design
00:07:13.12 generated by Feng Zhang in the laboratory of a recombinase
00:07:17.01 dependent adeno-associated vector backbone, which is
00:07:21.03 double Floxed, inverted, and only expresses the opsin in the
00:07:25.01 presence of Cre recombinase. This strategy was shown to work
00:07:29.26 in behaving mammals in 2009 when he and his colleagues
00:07:34.10 targeted the dopamine neurons from the ventral tegmental area
00:07:37.12 in one paper, and targeted the parvalbumin neurons in the prefrontal
00:07:41.24 cortex in another paper. And this allows one to capitalize on the
00:07:46.00 large growing resource of Cre-driver lines. That was a gain of
00:07:51.17 function intervention. It took until 2010, until December 2010,
00:07:57.22 for a loss of function of mammalian behavior to be achieved.
00:08:01.15 And that's illustrated here, Ilana Witten in the laboratory was
00:08:05.01 able to achieve this. And what's illustrated here is delivery of
00:08:08.25 the halorhodopsin, a chloride pump that is an inhibitory, into
00:08:11.08 the nucleus accumbens. These cells here are the cholinergic cells,
00:08:14.25 the accumbens were the target in this case. And she used a Cre-driver
00:08:18.28 line that expresses Cre recombinase only in the ChAT cells
00:08:23.28 or choline-acetyltransferase expressing cells. And she used
00:08:28.07 a recombinase dependent virus and delivered this third generation
00:08:30.25 halorhodopsin, eNpHR3.0, into these cells to see if she could
00:08:36.03 inhibit them during free moving behavior. These cells had been
00:08:40.00 hypothesized to be linked to cocaine-dependent conditioning and
00:08:42.22 she was able to address the question, what if you inhibit these
00:08:45.07 cells in freely moving mammals, what happens to the behavior?
00:08:49.20 Now what is this 3.0? This is the third generation version of the
00:08:54.07 halorhodopsin, this is a trafficking enhanced version. These opsins
00:08:59.16 come from very diverse organisms that many of them don't have
00:09:03.24 endoplasmic reticulum or membrane trafficking like mammalian cells.
00:09:07.28 And what Viviana Gradinaru in the laboratory figured out was how
00:09:12.08 to add trafficking motifs to these opsins, to help them move
00:09:16.12 to the right place in mammalian cells. And this has helped
00:09:19.18 every opsin that we looked at, substantially. Both in terms of getting
00:09:24.07 into the membrane, getting down axons and dendrites, which of course
00:09:27.09 is very important for projection targeting. And for getting robust
00:09:31.03 behavioral effects. Now what happened in this case, well,
00:09:35.29 what we found was one could inhibit the cholinergic cells
00:09:40.23 in the nucleus accumbens. In a loss of function intervention it's useful
00:09:44.18 to come in, in many cases, with bilateral interventions into the
00:09:48.15 bilateral brain structures that are involved. To make sure you're
00:09:51.11 inhibiting both sides of the brain that might be involved in
00:09:54.23 and possibly giving rise to the behavior in vivo.
00:09:58.14 And what we found was, as illustrated here. Normally,
00:10:03.02 an animal will prefer to spend time in a chamber, as you can see
00:10:06.14 on the left here in the bottom row, where it previously received
00:10:10.07 cocaine. See the path tracing here, the animal spending more
00:10:13.05 time in the chamber where it had received cocaine.
00:10:15.11 That normal behavior in red on the bottom is contrast to the
00:10:20.01 experimental animal in blue on the top, where the cells were
00:10:23.14 inhibited. And what you can see is there's much less preference
00:10:27.08 for the chamber where cocaine was experienced.
00:10:29.19 And this was a consistent finding against many animals.
00:10:32.13 This highlights a practical control aspect, as well, for optogenetics.
00:10:37.02 As illustrated here, in both cases, the animal is receiving
00:10:43.24 the light, has the hardware implanted and received the injection of
00:10:50.14 the same virus. So it's a good comparison. The only reason that this
00:10:54.14 is an experimental animal is it happens to be a Cre-recombinase
00:10:58.12 positive animal, so it's able to unlock the Cre-dependent virus
00:11:03.05 and express opsin. This animal is not. That illustrates the
00:11:06.20 kind of control experiment that's quite useful.
00:11:09.13 The other thing that's important is to build in many behavioral
00:11:11.27 controls. You can see the effect quite prominently here, but
00:11:16.21 control experiments show that there was no effect on
00:11:19.08 other forms of contextual memory formation. No effect on anxiety,
00:11:23.20 no effect on locomotion, and that the intervention itself was not
00:11:27.03 aversive, as shown here on the right. So this illustrates finally
00:11:31.20 loss of function, which complemented the earlier gain of function
00:11:35.17 for mammalian behavior. Now, we have of course, many more
00:11:41.10 things that we'd like to do. We'd like to do combinatorial control,
00:11:44.14 we'd like to control one population of cells alone or in combination
00:11:48.19 with another. And for this reason, we're able to bring to bear many different
00:11:53.28 kinds of opsins now. All at once. There are an incredible diversity of these
00:12:00.11 molecular machines out there in nature. They extend through all the major
00:12:06.05 kingdoms of life. Many of them are illustrated here, as you can see,
00:12:09.29 the channel rhodopsins down here in blue, some of the inhibitory
00:12:13.13 tools up there in yellow. All of these, it turns out, are substantially
00:12:16.20 helped by providing membrane trafficking enhancements.
00:12:20.01 Now, one thing one has to be careful of when thinking about
00:12:23.25 the effects of one's optogenetic interventions is that maybe having
00:12:29.25 within that targeted cell, as shown here, what might be happening
00:12:34.22 in the extracellular space and how that might influence
00:12:39.02 neighboring neurons that are not directly under optogenetic control.
00:12:43.14 And of course, one may have various kinds of light-driven transporter
00:12:50.10 pumps, but there are also many other kinds of transporters,
00:12:55.08 exchangers, that will be affected by the shifting concentrations of
00:12:59.19 ions, both on that cell and on other cells. Now this is no surprise
00:13:05.29 to physiologists. There are many such known processes that occur.
00:13:11.17 Changes in protons, changes in potassium ions, sodium ions,
00:13:15.05 chloride ions, calcium ions, can all have this sort of effect
00:13:18.00 whether in response to native activity or electrical stimulation
00:13:23.02 or optogenetic excitation or inhibition. But an important thing
00:13:27.02 to keep in mind. Now, one for example, they wished to be sure
00:13:33.29 that during one's exposure to light, that one is achieving the inhibition
00:13:37.23 that's needed. It's important to do recordings in various settings
00:13:42.24 whether intracellular or extracellular to assess not only the directionality
00:13:46.21 of the effect that's being exerted, but also it's stability and maintenance over
00:13:51.21 the course of experiment. And of course, what happens immediately after
00:13:57.05 the light goes off. This is important as we had pointed out back
00:14:01.07 in 2007, and continues to be important to think about.
00:14:04.12 Sometimes, as we've shown, you can see very brief changes
00:14:07.24 in excitability lasting for a few seconds after an optogenetic intervention
00:14:12.22 this kind of thing can be seen with native processes, as well.
00:14:17.10 It's important to simply design the experiments to be aware
00:14:20.17 of what happens both during and after the optogenetic intervention.
00:14:24.07 Now, this kind of opportunity that we have for optogenetics is
00:14:30.07 now expanding rapidly. We've now, together with our colleagues
00:14:36.14 at the University of Tokyo, Osamu Nureki and Hideaki Kato,
00:14:41.07 have been able to generate channelrhodopsin chimeras that allowed
00:14:46.27 us to see the crystal structure at high resolution for channelrhodopsin
00:14:51.04 which has created some interesting opportunities. Both for
00:14:54.29 improving the, what you might call the basic science of optogenetics
00:14:59.08 and improving its power, as well. And I'll explain how this happens.
00:15:04.13 This is the crystal structure of channelrhodopsin, it is a dimer.
00:15:07.21 The pore is formed not at the interface between the two monomers
00:15:13.16 but it's formed when the transmembrane helices 1 and 2,
00:15:16.14 which are tilted back further than they would normally be
00:15:21.06 in another seven transmembrane protein, form a pore and each
00:15:25.03 monomer has a pore. This, by the way, is all trans-retinal, it's
00:15:28.19 a small organic chromophore that is present in all vertebrate tissues
00:15:32.11 that we know of. One does not have to add this to get optogenetics
00:15:37.03 to work in vertebrates. However, in drosophila and C. elegans,
00:15:40.15 one typically adds all trans-retinal to the medium or to the food
00:15:45.28 of the organism. Now the pore here is presumably gated by
00:15:50.16 the conformational change of retinal that gets transferred to the rest
00:15:55.10 of the protein. But having the crystal structure gives us a number
00:15:58.14 of interesting opportunities. For example, we can see where we
00:16:01.15 think the pore is. It's lined with charged polar residues. And this
00:16:06.28 has led us actually to build useful tools already. For example,
00:16:10.17 mutants that have mutations along the pore and have zero
00:16:14.27 photocurrents, now can be used as control tools. They can be expressed
00:16:18.14 and not conduct current in response to light, but occupy space
00:16:24.13 in membrane, perhaps change membrane capacitance. Perhaps
00:16:27.14 affect to the extent that this happens, the distribution of other ions
00:16:32.22 within a cell -- of other receptors in the cell. And this is an important
00:16:39.16 control that's actually better than simply expressing a fluorescent
00:16:44.02 protein, because you actually got a lot of protein that's being
00:16:48.10 loaded into the membrane but not generating photocurrents.
00:16:52.14 So the crystal structure has already been useful in that regard.
00:16:55.19 It also may allow us to generate different kinds of conductivity.
00:17:01.09 For example, it may be able to buy us the conductance of the
00:17:05.13 pore to potassium ion or protons. In fact, nature's already shown
00:17:09.18 us that it's possible for microbial opsins to generate a ion
00:17:13.28 selective channel. We found from Dunaliella salina, a channel rhodopsin
00:17:20.03 that seems to be selective for protons. And so this is illustrated
00:17:24.29 here, Feng Zhang identified this and we published this in collaboration
00:17:28.13 with Peter Hegemann's group. This shows that it is possible
00:17:33.01 to generate ion selectivity with a microbial opsin. The diversity
00:17:37.29 of microbial opsins out there also affords us the possibility
00:17:41.15 of combinatorial control. Controlling one cell population alone
00:17:45.18 or in combination with another. This organism, Volvox carteri,
00:17:50.09 is a multicellular green algae. It as Feng identified in 2008,
00:17:58.21 allows one to generate red shifted light activation. The action
00:18:06.06 spectrum is illustrated here, as you can see, there's a nice
00:18:09.07 band in the amber, where VChR1, Volvox channelrhodopsin
00:18:13.09 can be recruited, not touching a channelrhodopsin 2 expressing
00:18:16.11 cell at all, which is normally controlled by blue light.
00:18:19.18 Now it took a while to improve the photocurrent, over years we
00:18:24.22 carried out various kinds of molecular modifications that improved
00:18:29.03 the expression of VChR1 until it became a very potent
00:18:32.26 molecule that we call C1V1. In collaboration with Peter Hegemann,
00:18:37.15 we were also able to sharpen the absorption spectrum, as shown here,
00:18:42.04 and create a more purely red light activated tool. You can see
00:18:45.19 this now leads to the ability to drive action potentials with red
00:18:50.09 light. And as we've shown, this now allows combinatorial
00:18:55.02 excitation even in vivo, in freely moving mammals. Driving one
00:19:00.00 population alone or in combination with another. And the tools
00:19:03.19 and the methods are described in the reference. The large diversity
00:19:09.16 of these tools is now becoming difficult to organize and keep track of. We've
00:19:14.11 put forth a few such organizational tables and strategies. This is one of them.
00:19:21.19 What you can see on the y-axis is the visible light action spectrum,
00:19:25.10 and on the x-axis is a very useful kinetic parameter that we call the
00:19:29.13 Tau off, this is the time constant of deactivation of the microbial
00:19:34.05 opsin or the optogenetic tool after you turn off the light.
00:19:39.04 Just to orient you, here is channelrhodopsin 2, it's got about
00:19:44.15 a 10 millisecond time constant of deactivation. We've been able to
00:19:48.03 speed it up with the Cheetah mutants, which are about 4 milliseconds.
00:19:51.23 Here are the C1V1 tools that are red shifted and also a little
00:19:55.00 slower, as you can see on this log scale. And then all the way,
00:19:59.29 many orders of magnitudes slower, you can see some extremely
00:20:04.01 persistent tools that keep their currents going for very long after
00:20:10.00 the light is turned off. This may seem counterintuitive, but I'll explain
00:20:14.26 why these tools are useful. But all of the ones that are represented
00:20:17.24 here share these key properties. They're single component, which greatly facilitates targeting, speed, robustness of the
00:20:25.22 system, you don't need to add chemicals to make any of these
00:20:28.00 work, even in behaving mammals. They all come from biology,
00:20:31.27 which enhances their physiological tolerability. They're fast
00:20:36.07 and as you can see, there's substantial spectral diversity and tunability.
00:20:40.29 So what about these slow opsins? How are they useful?
00:20:45.11 Well, first of all, they're generated by mutating this cysteine residue
00:20:49.24 and the aspartic acid residue as shown here, and that takes this
00:20:53.01 very rapid deactivation after light off, 12 milliseconds, as you can see
00:20:56.26 here, to some very slow, very persistent patterns of activity
00:21:01.23 that are seen in these mutants here. We call these step function
00:21:05.20 opsins, they become more bistable-like, or step-like in their
00:21:08.21 properties. And this has a couple of interesting effects. First,
00:21:12.13 it makes the cells that are expressing these opsins more light
00:21:17.29 sensitive for long light pulses. This is because they effectively become
00:21:23.10 photo-integrators. The opsins don't turn off, so very low levels of
00:21:26.13 light can be delivered for a long period of time. Photons accumulated by
00:21:30.23 the cell in turning on opsins, and you can elevate the photocurrents
00:21:35.15 to the maximal levels that have been experienced during the brief higher
00:21:39.23 intensity pulse of light. Now that is a big effect, many of orders of
00:21:45.09 magnitude increase in light sensitivity as shown. But you may say,
00:21:49.27 well, you can't turn these off, so how useful are they?
00:21:53.02 In fact, you can turn them off. This is a very simplified
00:21:58.01 version of the photocycle of these opsins, starting from a dark
00:22:01.05 state, D470, that's closed, and absorbing the blue photon,
00:22:05.09 and goes to an open state where the step function opsins are
00:22:08.26 stuck, this is where they stay and stay open. But you can actually
00:22:13.20 shortcut back to the closed state with a green photon. So you can
00:22:16.20 actually turn them on with a blue photon and off with a green
00:22:20.08 photon. And this behavior is illustrated here, this allows us
00:22:24.11 to in a minimally invasive fashion, we can even take away the
00:22:29.24 fiber, the light delivery source, after we deliver a pulse of excitation.
00:22:33.18 You can deliver a stable increase in excitability of a targeted
00:22:38.08 cell population that can be persistent and that can be terminated when you
00:22:43.09 want with a pulse of different color of light. This is illustrated
00:22:46.24 here, it's got a patch clamp neuron expressing one of the opsins
00:22:49.16 receiving five identical stimuli played in through the patch
00:22:55.00 pipet. But you can see the cells are only responding with
00:22:57.21 action potentials in the first, third, and fifth epoch. And
00:23:01.02 the reason is that we've flipped on and off, on and off,
00:23:06.27 on and off, the step function opsin with all three different pulses
00:23:09.25 of blue and green light. Flipping a cell into and out of an
00:23:13.26 excitable state. And this has a number of advantages. First,
00:23:18.27 it allows the physiology and behavior to happen in an epoch
00:23:24.19 where there is not light being continuously delivered.
00:23:27.05 This can be useful for a number of reasons, including not having
00:23:30.16 optical hardware, not continuously depositing energy into the
00:23:33.29 tissue, and that's turned out to be useful for social behavior
00:23:38.05 studies as we've published, and other studies as well.
00:23:40.10 Another advantage, of course, is that these tools are very
00:23:43.22 -- these tools make the cells very light sensitive, and this
00:23:47.21 allows you to recruit larger tissue volumes for complex
00:23:51.08 behaviors in mammals. And then finally, what you can see here is
00:23:55.23 that you're not prescribing a defined action potential pattern,
00:23:58.20 the spiking pattern that the cell is eliciting is in some way capturing
00:24:04.02 the native synchrony or asynchrony of the activity coming in.
00:24:08.05 And this basically means that you are sensitizing a cell population
00:24:13.22 to its native pattern of activity without prescribing a defined
00:24:20.13 spike pattern. And so all of these properties, we think, make these
00:24:25.06 tools particularly useful. We've shown they work in rodents,
00:24:28.24 also Krishna Shenoy's lab showed that they work quite
00:24:32.23 well in awake behaving Rhesus macaques. You can deliver
00:24:37.16 blue light pulse to motor cortex and get elevated spiking
00:24:40.11 that can be terminated with a pulse of yellow light.
00:24:43.04 Now, this is a summary of some of the optogenetic tools. Now
00:24:49.04 is a good time to talk a little bit about applications, particularly those
00:24:54.17 that have used projection targeting, One illustration of that is
00:25:00.26 illustrated here. This is a rat, where a virus carrying channelrhodopsin 2
00:25:08.20 was introduced into motor cortex, but light was delivered not to motor cortex
00:25:14.02 but to thalamus, a downstream projection target. The other
00:25:18.15 interesting thing about this experiment, conducted by Jin Lee and
00:25:21.09 Remy Durand in the laboratory, is that the readout is fMRI.
00:25:28.09 And so this is an integration of optogenetics and fMRI,
00:25:31.29 this is called ofMRI. And this allows one to, with all the disadvantages
00:25:38.24 and caveats of fMRI, including as we all know, somewhat poor
00:25:45.00 spatial and temporal resolution, at the same time it's unparalleled
00:25:48.12 for its non invasiveness and its globality and simultaneity
00:25:51.27 of the readout one can obtain. And so one can drive a fine
00:25:56.18 population of cells and one can look throughout the entire
00:26:00.02 brain at the recruitment of different brain regions. And what we
00:26:04.02 could see, as we reported here, is not only are we recruiting a cell
00:26:10.14 that's defined by having its cell body and motor cortex, its projection
00:26:15.10 to thalamus, and by the way, there's a promoter fragment used
00:26:21.08 in the virus here that allowed us to pick out a subpopulation
00:26:24.11 of those cells within the motor cortex. So very precisely defined
00:26:28.06 population of cells. But we can look at the entire brain and
00:26:31.09 see the pattern of activity that's recruited across the entire brain
00:26:35.09 by this population. We can see local activity in thalamus, bold
00:26:39.25 signals. But also back at the motor cortex site, we can see
00:26:45.10 recruitment ipsilaterally, as well. And we can map out
00:26:48.21 the hemodynamic response functions that show very similar
00:26:53.07 patterns, even corresponding to those that are seen in
00:26:58.08 behaving human beings. And of course, of sensory or behavioral
00:27:01.18 stimulation, including the onset/offset kinetics undershoot recovery kinetics.
00:27:09.22 This is one class of intervention, another that uses projection targeting.
00:27:14.24 Another experiment used behavior as a readout with projection targeting.
00:27:20.06 This was an experiment carried out by Viviana Gradinaru, Murtaza Mogri,
00:27:24.29 and Jaimie Henderson in the laboratory. Here, we asked
00:27:31.29 how deep brain stimulation might be involved in, and the mechanism
00:27:37.23 by which deep brain stimulation might be involved in restoring
00:27:41.05 normal function after generation of Parkinsonism in animals.
00:27:45.16 Now, deep brain stimulation is commonly delivered in people to
00:27:50.17 a deep brain structure called the subthalamic nucleus, but it
00:27:54.08 had been mysterious as to what its initial direct target stimulation
00:27:58.05 is. An electrode, of course, will not pick apart one cell type
00:28:02.07 or another, one circuit element or another. But you can imagine
00:28:06.19 now the kind of experiments one could do optogenetically, and
00:28:09.04 one could deliver an inhibitory, like this halorhodopsin, the eNpHR,
00:28:12.19 or halorhodopsin. And one can use a promoter fragment like
00:28:16.10 this CaM kinase 2 alpha promoter fragment to target excitatory
00:28:19.08 cells, for example. But you can imagine all the different combinations
00:28:22.28 one could do, turning different circuit elements on or off,
00:28:26.01 and seeing in these animals, which we made hemi-Parkinsonian,
00:28:29.29 which delivered a therapeutic effect. Now this was an interesting
00:28:36.09 process, what many of the different experiments that we did had no effect.
00:28:39.15 These are hemi-Parkinsonian animals, so we have a rotational
00:28:43.01 readout. And for example, inhibiting local cell bodies, as you can
00:28:48.03 see here, had very little effect on the rotational pathological
00:28:52.08 behavior. But when we did a projection targeting experiment,
00:28:56.13 as illustrated here, where the only light sensitive element
00:28:59.26 within the subthalamic nucleus was incoming light sensitive
00:29:03.21 axons coming from elsewhere, including from motor cortex,
00:29:06.26 there we saw a very robust and reversible therapeutic
00:29:11.06 effects. Reduction in rotational behavior, also the animals
00:29:14.29 which were initially bradykinetic and were barely moving at all,
00:29:17.16 with high frequency stimulation, could move quite freely
00:29:20.12 in reversible fashion. This suggested that the initial direct
00:29:24.12 target of deep brain stimulation may be the incoming axons
00:29:28.12 that are coming into the subthalamic nucleus from elsewhere.
00:29:31.16 And supporting this interpretation, we found that there was a great
00:29:35.02 deal of specificity with regard to the precise temporal pattern
00:29:39.26 of that activity, that mapped onto the clinical experience quite well.
00:29:44.03 High frequency stimulation was efficacious, low frequency
00:29:47.14 stimulation, 20 Hz or beta-band stimulation, was not only not
00:29:51.10 efficacious, but pathologically harmful. And this is what is seen clinically
00:29:56.27 as well, in fact, the crossover between therapeutic and pathological
00:30:01.09 occurred at about 70Hz, which is just what's seen clinically.
00:30:05.26 And this highlights the importance of cell type specificity.
00:30:11.27 But also of the temporal precision of the optogenetics, you can
00:30:14.24 see, if we didn't have temporal precision, you would be
00:30:18.16 quite confused as to what's going on here. Fundamentally
00:30:22.10 different effects are seen at a higher or lower frequency.
00:30:25.13 This was another example of projection targeting, this of course,
00:30:30.09 is only the tip of the iceberg. And this particular behavioral
00:30:34.19 model is only the tip of the iceberg. Now you can carry out not only
00:30:39.06 a host of basic science studies, but also look at a broad
00:30:42.27 range of disease related phenotypes. This is an animal
00:30:47.13 that is hemi-Parkinsonian, but one can do, for example,
00:30:53.25 psychiatric disease related assays. One can put the animals
00:30:56.21 in a forced swim test. Melissa Warden in the laboratory
00:30:59.16 has generated this integrated optogenetics forced swim test
00:31:04.15 apparatus. Animals that are in this apparatus will swim freely,
00:31:09.19 when they are tired of swimming or they don't want to swim
00:31:12.21 anymore, they'll float in the water. And one can assess the transition
00:31:17.28 between the swimming and the floating behavior, which is linked
00:31:21.13 to states of behavioral motivation and effort on the one hand,
00:31:26.09 and behavioral despair and mobility on the other hand.
00:31:31.04 And that in turn is affected by stress and antidepressant treatment.
00:31:35.01 This interesting behavioral test can now be richly interrogated
00:31:39.13 with a complex suite of inputs and readouts. For example, a magnet
00:31:47.00 can be attached to the hind paw of the animal, and induction coil
00:31:50.05 placed around the cylinder of water, and each kick of
00:31:53.12 the animal can be picked up with millisecond precision
00:31:56.12 while rich recordings are being generated in the brain of the
00:32:00.20 animal, multiunit recordings with tetrodes. And while millisecond
00:32:06.17 scale cell type specific optogenetic interventions are being carried out
00:32:09.26 through the fiber optic. All this happening at once in the same
00:32:13.05 freely moving animal. One can do projection targeting in rats
00:32:16.26 quite well. Of course rats don't have the same rich diversity
00:32:24.14 of genetic targeting tools that mice have. But we started to
00:32:27.20 make headway on that front, as well. Ilana Witten in the laboratory
00:32:31.09 has generated Cre-driver rat lines that allow genetic targeting
00:32:38.14 of circuit elements defined by Cre-recombinase expression in
00:32:42.11 rats. And she and Liz Steinberg, and Patricia Janak, have
00:32:46.28 together worked to generate a suite of rat-specific equipment
00:32:54.00 and hardware. And also to come up with some very interesting
00:32:56.25 integrated transgenic rat and projection targeting experiments
00:33:01.25 that all work together at the same time. I'll give you a flavor
00:33:04.27 of how multiplexing these different targeting strategies has really
00:33:07.27 helped us advance, in this case, the science of reward or behavioral
00:33:13.00 conditioning. In this case, what we used was a tyrosine hydroxylase
00:33:17.29 Cre-driver rat, and this allowed us to photosensitize only the dopamine
00:33:23.17 neurons within the ventral tegmental area. And delivering light
00:33:27.12 by fiber optic to these rats, only when the rats generated a
00:33:32.02 nose poke, allowed us to see if this was -- appeared to be
00:33:34.27 rewarding for the animals. If they would work hard for light
00:33:38.25 pulses, bursts of light pulses, delivered to the dopamine
00:33:42.06 neurons. And as you can see, this was very motivating, the animals
00:33:47.28 would execute thousands of nose pokes per day. Only those animals
00:33:51.25 that were actually receiving the light that was efficacious, in other
00:33:56.15 words, animals were Cre-recombinase positive, only those animals
00:34:00.16 showed this behavioral response. And it was not a nonspecific
00:34:03.26 behavioral activation, the animals did not nose poke at an inactive
00:34:07.13 nose poke hole. And this showed all the typical properties
00:34:11.00 of acquisition, extinction, reacquisition, extinction by contingency,
00:34:18.14 degradation, reacquisition. Now this was with transgenic rats
00:34:24.23 and so it was giving rise to cell type specific targeting and
00:34:30.21 control within the rat. What one could also combine this with
00:34:35.21 projection targeting. So one could illuminate not in the ventral
00:34:38.16 tegmental area, but in one candidate downstream region, in the
00:34:41.22 case, the nucleus accumbens. And here these intersectional strategies
00:34:45.19 allowed one to see that there was motivated behavior, in this case,
00:34:50.16 as well, the animals will work, although not as well. The multiplier
00:34:54.04 of a hundred, instead of a thousand, but the animals will work
00:34:56.13 quite hard for light that is recruiting the dopamine cells
00:35:01.19 with their cell bodies in the ventral tegmental area that have
00:35:04.23 projections to the nucleus accumbens. So this kind of
00:35:08.25 integrated specificity gives one substantial experimental
00:35:14.02 power. Now, this is an ongoing process as more Cre and other
00:35:23.23 recombinase driver animals are generated, this will increase our
00:35:26.02 understanding and our capabilities in optogenetic experiments.
00:35:31.07 But already, a number of interesting experiments relevant
00:35:35.16 to both basic science and disease have been achieved.
00:35:38.12 Kay Tye, Lief Fenno, and Rohit Prakash, and others in the laboratory
00:35:42.11 have been able to generate projection specific studies of anxiety
00:35:47.02 related behavior. I told you about the cocaine conditioning experiments.
00:35:51.03 Ofer Yizhar and Lief Fenno have been able to carry out social
00:35:54.26 behavior work that are relevant to autism and schizophrenia.
00:35:58.12 We've talked about the deep brain stimulation work, together
00:36:02.04 with Susumu Tonegawa and others, we've been able to study
00:36:04.28 memory processes. There is actually a number of both disease
00:36:13.01 related and basic science questions that have already been addressed
00:36:15.12 but that's still in the early days. Much more to be studied.
00:36:19.13 In fact, optogenetics is not limited to the neuronal system.
00:36:24.24 A number of groups are now working on controlling glial,
00:36:31.00 controlling cardiac function, applications to stem cells
00:36:33.27 are also growing and expanding rapidly. And the versatility
00:36:38.21 of the technique we're just now starting to fully appreciate.
00:36:41.15 So this is a brief summary of what we have been able to do. Our
00:36:46.00 collaborators are shown here, our many talented students and
00:36:50.01 postdocs are illustrated here, who contributed to the work that I
00:36:54.18 have been talking about. In terms of resources, we have a number
00:36:59.00 of places that you can go to access protocols, primers, and other
00:37:05.06 tools. For protocol-type applications, we've published in Nature
00:37:11.20 Protocols a number of strategies, including step-by-step procedures
00:37:18.15 for giving rise to optogenetic interventions. Primers published recently
00:37:23.11 in journals that allowed us to go into detail on the caveats,
00:37:29.14 technical challenges, conceptual strategies for carrying out
00:37:33.25 optogenetic experiments. In this Nature Methods paper, there's this
00:37:38.03 detailed comparison of the optogenetic tools and quantitative
00:37:41.17 parameters of kinetics, light sensitivity, temporal stationary and others.
00:37:46.23 And then finally, our website, optogenetics.org, gives you access
00:37:53.01 to all the tools which we'll send freely to you. Just shoot us
00:37:56.23 an email, and we'd be happy to help.
00:37:59.03 So that's optogenetics. I hope you found this useful.
00:38:01.24 And I look forward to hearing from you if we can help you in any way.
00:38:06.17 Thank you.

This Talk

Speaker: Karl Deisseroth

Audience:

Researcher

Recorded: August 2012

Talk Overview

Optogenetics is a combination of genetics and optics to achieve a gain or loss of function of biochemical events such as action potentials in a particular neuron or tissue. Opsin genes encode proteins that receive light and give rise to ion flow. This talk gives an introduction to optogenetics followed by examples of how optogenetics is being used to study the brain.

Questions

Which of the following is/are true about opsins (select all that apply)?
1. Activated by light
2. Give rise to ion flow
3. Activated by ions
4. Can inhibit cells
Briefly explain the benefits of using optogenetics over electrode stimulation.
What is τ off?
1. Describes the wavelength of light needed to activate the opsin
2. Describes the time constant of deactivation of the opsin after the light is turned off
3. Describes the time constant of activation of the opsin after light is applied
4. Describes the number of opsins needed to be expressed on a cell to achieve activation
Explain why an opsin with a long τ off can be useful.

Answers

Speaker Bio

Karl Deisseroth

Karl Deisseroth

Karl Deisseroth is Professor of Bioengineering and of Psychiatry and Behavioral Sciences at Stanford University, and an HHMI Investigator. He earned his MD/PhD in neuroscience from Stanford University. His lifelong interest in psychiatry led him to build a world-class laboratory in neuroscience where he developed optogenetics. His lab studies the control and mapping of neural… Continue Reading