Development of Optogenetics

Top

00:00:06.29 Hello, my name is Karl Deisseroth. I'm a
00:00:09.20 professor of bioengineering and psychiatry
00:00:11.17 at Stanford and the Howard Hughes Medical Institute.
00:00:13.28 And today, I'd like to tell you a little bit about
00:00:16.27 optogenetics. Some of the humble origins of this
00:00:20.29 method in science. Some of its applications
00:00:23.23 and some of the interesting lessons it teaches
00:00:26.17 us about the process of biology.
00:00:28.26 Now, I'm also a psychiatrist and a lot of the
00:00:32.22 inspiration for this technology came from psychiatry.
00:00:35.22 And although this is an important field in medicine,
00:00:39.09 it's got a long way to go. In many ways, we haven't progressed
00:00:44.08 too far in our basic description of the core symptoms
00:00:47.21 and their underpinnings, their causal underpinnings
00:00:51.08 than we had hundreds of years ago. Here
00:00:53.27 is a depiction of some of the earliest medical model
00:00:57.19 type descriptions of depression and its symptom
00:01:00.23 that today we call anhedonia, the inability to enjoy
00:01:04.16 normally rewarding things. And John Haslam
00:01:09.13 in the early 1800s described in melancholy,
00:01:12.25 what we would call depression today,
00:01:14.21 they therefore neglect those objects and pursuits
00:01:17.17 which formerly proved sources of delight and instruction.
00:01:20.05 And this is about as precise as we can be today about
00:01:23.27 what anhedonia really is in psychiatry.
00:01:27.24 And what's a really amazing part of this story is
00:01:31.18 microbial organisms, single celled forms of life,
00:01:36.06 have given us tools that in the last few years
00:01:39.25 have allowed us to come to a deeper understanding of
00:01:42.18 this fundamental and important medical and scientific
00:01:47.02 question. And many others like it. Now this
00:01:51.23 also gives an opportunity to take a broader view
00:01:54.26 about what is optogenetics, where does it stand
00:01:57.20 in the broader context of interventions and experiments
00:02:02.07 that we could do in neuroscience. And in the brain, which
00:02:06.17 as we all know, is a very complicated system with many
00:02:12.19 interconnecting parts, we have historically had the ability
00:02:16.04 to use either magnetic or electrical interventions to
00:02:19.02 stimulate neurons. But one problem is that they cannot
00:02:23.24 discriminate individual kinds of neurons that might be
00:02:26.09 right next to each other. They all look the same to an
00:02:28.17 electrical or a magnetic intervention. What optogenetics
00:02:32.04 allows us to do is to put a particular sensitizers -- antennas
00:02:36.20 of a sort, for external sources of energy and information
00:02:40.19 into specific kinds of neurons. And then we can use light,
00:02:43.28 in this case, if we make light sensitive antennas or
00:02:48.28 transducers, and make neurons of a particular kind respond
00:02:52.07 to light, and we do that to allow different kinds of ions
00:02:55.16 to flow into cells and turn neurons on or off depending on the
00:02:59.15 experiment. That's the fundamental idea of optogenetics
00:03:03.04 and it's something people have wanted to do for a very long
00:03:06.01 time. The first person who seemed to have framed the
00:03:09.16 possible solution to this age old question was Francis
00:03:14.02 Crick in 1999. He described the essence of this problem
00:03:19.06 that we needed to control, turn on or off individual
00:03:21.27 kinds of cells. And he suggested the ideal signal
00:03:24.18 would be light. He didn't have an idea how to do it
00:03:27.24 but in his usual perceptive fashion, he had some pretty
00:03:31.04 good instincts, I would say. And what I'd like to do here is
00:03:34.14 to sort of pay homage to the many people who've
00:03:38.01 thought about this and done experiments, and contributed
00:03:40.28 both historically and up to the present in making all of this
00:03:44.27 ultimately work. Now there were early efforts dating back
00:03:51.12 to 2002, a pioneer in this field, Gero Miesenboeck
00:03:54.23 was able to take individual components from a metazoan
00:03:59.15 eye, the multicellular organisms we all know of course,
00:04:03.23 we have light sensitive cells in our retinas. And they achieved
00:04:07.13 light sensitivity with a cascade of proteins, and he was able to
00:04:10.20 get to the point where he could bring some of those
00:04:12.14 individual components of that cascade together and
00:04:16.10 confer light sensitivity on neurons. And the bright areas
00:04:19.05 of this plot are where the light was on. And there were other
00:04:23.23 groups who also took what you might call a multicomponent
00:04:27.11 approach to try to make neurons light sensitive.
00:04:30.13 Now the only thing that held back these elegant methods
00:04:33.18 was in many ways their multicomponency which contributed
00:04:37.11 to difficulties in targeting, and they of course would be the first
00:04:41.27 state showed beautifully that some delay in the onset of
00:04:47.26 the responsivity, the action potentials that were fired
00:04:50.26 by the cells in response to the light. And what allowed
00:04:56.01 this field to take off was the microbial approach as I
00:04:59.24 alluded to earlier. Now these microbial organisms,
00:05:03.04 they're small, they don't have a lot of space to work with
00:05:06.04 and nor do they have to do as complex processing
00:05:08.16 as other organisms do. And so they had the leeway
00:05:13.00 evolutionarily, it seems, to make very efficient all in one
00:05:16.22 single component systems for both detecting light
00:05:20.18 and generating ion flow. And that allowed the generalizable
00:05:24.19 aspect of this solution to happen. And this is a field that
00:05:28.07 has its roots back, even as far back as the early 1970s.
00:05:32.21 Dieter Oesterhelt and Walter Stoeckenius in 1971
00:05:36.06 identified bacterial rhodopsin and over decades
00:05:41.15 they and a thriving community of researchers were able
00:05:45.24 to sort out how this amazing protein works, how it responds
00:05:49.18 to light, how its photocycle of the individual protein allows
00:05:54.02 ions to be transported across the membrane. And this
00:05:59.03 pump type class of opsins, bacteriorhodopsin being an example
00:06:02.29 is followed by the discovery of halorhodopsin,
00:06:05.05 where instead of protons being moved, chlorine ions
00:06:07.24 are moved instead. Peter Hegemann and his colleagues
00:06:10.17 beginning with some early predictive work in 1985
00:06:14.26 leading to a very important paper in 2002 showed
00:06:18.10 that there were also channel type microbial opsins that
00:06:20.28 opened actual pores in the membrane instead of pumping
00:06:23.29 one ion at a time. And then many groups, here
00:06:28.18 Feng Zhang in my laboratory was able to find a
00:06:31.08 red light activated channelrhodopsin in 2008.
00:06:34.07 And a great number of different groups, Feng Zhang,
00:06:37.22 Peter Hegemann, our colleague Ed Boyden, Georg Nagel,
00:06:40.17 Ernst Bamberg, John Spudich, and many others
00:06:42.23 have been identifying amazing, amazingly diverse
00:06:46.10 microbial opsins with different properties from across
00:06:50.26 the phylogenetic spectrum. Now, what is striking as
00:06:55.05 you'll notice from this, is how long we've known about
00:06:57.20 these and this happens to be my biochemistry textbook
00:07:01.19 The beautiful book written by Lubert Stryer in 1988,
00:07:04.23 this was the version that I used, and you can see this is
00:07:07.17 textbook type knowledge. Descriptions of the photocycle
00:07:11.25 of bacteriorhodopsin, for example. Part of the foundational
00:07:16.08 knowledge of biochemists and neurobiologists. Now there
00:07:20.07 were people who did try to put these microbial
00:07:23.27 opsins in different settings, in fact, successfully.
00:07:26.04 This is a very interesting paper from 1994, where
00:07:29.28 the bacteriorhodopsin was indeed transferred along with
00:07:34.20 light sensitive flow of ions into a eukaryotic organism.
00:07:40.14 But nobody had yet tried to actually create a tool for
00:07:47.24 neurobiology, to turn on or turn off individual kinds of
00:07:51.17 cells. Neurons are complicated, they're vulnerable,
00:07:55.12 they're sensitive, they're different but in principle,
00:07:58.01 it was an experiment that could be done. And there were
00:08:00.27 many people trying to do it. Now the path to this
00:08:06.03 in my own lab started very near to the time when I
00:08:11.03 had started setting up my independent laboratory.
00:08:15.10 And I had a very humble approach, where as you can see
00:08:21.02 I used very simple tallies to assess how the experiments
00:08:24.14 were working, and it's interesting to look back on the
00:08:28.06 simplicity of this early experiment. But what I was doing
00:08:30.17 here, I had a broader view in the lab to try different kinds
00:08:36.26 of proteins to see if we could turn neurons on or off
00:08:40.04 with different kinds of stimuli. I had acquired the channel
00:08:44.24 rhodopsin gene, I'd written to a group in Germany
00:08:46.25 that had worked on the initial clone from Georg Nagel
00:08:49.14 and he sent me the clone right away, which was very
00:08:51.27 generous of him. I was also looking at different kinds of
00:08:54.20 potassium channels, to see if we could turn on or off,
00:08:58.16 constitutively active or dominant negative potassium
00:09:00.10 channels. And I was putting all of these into separate groups of
00:09:04.18 neurons in culture to see if I could get control over their
00:09:09.08 activity level. And I was using a very simple straightforward
00:09:13.04 read out of membrane depolarization, the phosphorylation
00:09:16.28 of a transcription factor called Creb, which I had done
00:09:19.14 a lot of work on previously. And the amazing result was that
00:09:24.13 in these neurons, you can see by the red staining of the nucleus
00:09:27.16 of the cell in a stimulated compared to an unstimulated
00:09:30.14 neuron. Both of these were expressing the channel
00:09:33.25 rhodopsin coupled to a yellow fluorescent protein,
00:09:36.19 so I could see a number of things here. I could see
00:09:38.14 that the gene that I'd put in was being expressed,
00:09:41.02 it was being transported to the membrane, that the
00:09:44.02 neuron was healthy, it hadn't exploded as a result of
00:09:46.12 putting this high levels of this membrane protein into
00:09:49.05 the cell. Which can happen to be sure. And moreover
00:09:54.06 that the light activated, the light treated neuron revealed
00:09:59.18 that it had been stimulated. That its membrane had been
00:10:02.29 depolarized and calcium flux had occurred. And this
00:10:06.28 biochemical readout was telling us that this was happening.
00:10:10.07 And in neurons that were expression a lot of this channel
00:10:13.05 rhodopsin, I could see a big difference in the number
00:10:16.17 of cells that were reporting on this depolarization
00:10:21.07 event, and overall, this was statistically significant.
00:10:24.10 And although this was a very simple, cheap, humble
00:10:28.24 experiment, this showed us that it was possible.
00:10:32.01 And this was in early July of 2004, but things went
00:10:35.29 very quickly after that. This is what our small
00:10:38.26 group looked like at the time, you can see the two grad
00:10:42.07 students, Feng Zhang and Ed Boyden, now at MIT.
00:10:45.14 Feng Zhang now also very well known for his work on
00:10:48.06 CRISPR-Cas9 systems. But we knew that we had to do
00:10:53.06 a lot more than just show some neurons getting stimulated,
00:10:56.22 we had to show that this was unique, that it was
00:10:58.25 a potentially generalizable or versatile, that it could
00:11:01.26 potentially be applied in broader systems. And there was
00:11:05.04 rightly so a lot of skepticism at the time. What the students
00:11:09.16 did was some really brilliant and creative work. Feng Zhang
00:11:13.09 designed the viral systems that would allow us to put
00:11:16.17 the channel rhodopsin in a very versatile way into different
00:11:19.22 brain regions in vivo, and the fact that it's just a single gene
00:11:23.12 allowed us to be confident that this would work well.
00:11:26.12 Ed did a number of beautiful recordings, showing how fast
00:11:29.26 the process could be, which was another important feature
00:11:32.04 of the system. Both its potential generalizability and its
00:11:35.20 speed. And there were many other groups working on this
00:11:39.29 at the time, Stefan Herlitze, Zhuo-Hua Pan, Yawo's group
00:11:45.04 in Japan. But we were able to put the different pieces
00:11:49.11 together in time to get the paper out, and it was published
00:11:52.16 together with two students, Ed and Feng, as well as Georg
00:11:56.02 Nagel and Ernst Bamberg. And this, though, was, I
00:12:01.23 wouldn't even say this was yet a fully optogenetics because
00:12:04.15 there was a great deal more work that needed to happen.
00:12:07.23 It took about five more years of very intense effort to
00:12:11.01 show that we had actually created something that was
00:12:14.05 generalizable, useful, and broadly applicable, which was
00:12:17.12 the key goal. So optogenetics didn't start overnight,
00:12:21.15 experiments were humble and simple, but it took a number
00:12:27.01 of years to actually say that we finally discovered how to
00:12:30.10 make this work. Part of the problem was how do you get
00:12:32.22 light into big mammalian brains. And this was an early
00:12:37.04 sketch of Feng Zhang's, where he designed the fiber optic
00:12:40.20 interface. And this was from 2006, but then this looks very
00:12:44.09 much like how we now do it today. So as you can see
00:12:47.28 he's a good artist and a great scientist. And this
00:12:51.13 fiber optic interface has since allowed us to do many
00:12:54.12 things, but I'll show you one other first experiment.
00:12:57.21 This was two years later in 2007, Feng was working
00:13:03.22 with some mice that we had made in collaboration with
00:13:06.13 Guoping Feng and George Augustine, and these
00:13:08.22 were mice designed to have channelrhodopsin
00:13:11.04 expressed in the brain and what Feng did was he implanted
00:13:15.03 into this mouse, a fiber optic on the right side of its
00:13:19.25 brain, which as you know would control movement
00:13:22.04 toward or attention to the left side of the world. And
00:13:25.14 when he turned on the light, as you can see here
00:13:28.14 the animal immediately starts moving, starts turning
00:13:33.07 to the left. And this was the moment I would say
00:13:36.12 we finally knew we had made something that was going to be
00:13:40.08 broadly applicable and generalizable. And that was followed
00:13:44.16 quickly by a collaborative set of experiments with
00:13:48.04 Luis de Lecea, Alex Aravanis, Antoine Adamantidis,
00:13:51.28 and others, where we applied this fiber optic interface
00:13:55.00 to target a deep structure in the brain, in lateral
00:13:57.26 hypothalamus. Feng designed some promoters
00:14:01.19 that would fit into a lentivirus and allow us to control
00:14:03.28 just one kind of cell deep in the hypothalamus.
00:14:07.16 And with this work, we were able to play in certain kinds
00:14:11.10 of activity and show some of the neural codes that were
00:14:14.06 causal in sleep-wake transitions. And then, even then though
00:14:19.20 this required some very specific fragment of a promoter,
00:14:23.15 the hypocretin promoter, which would fit into a virus.
00:14:25.27 And there was still greatly so some skepticism as to how
00:14:29.18 broadly generalizable this would be. But between 2009
00:14:34.02 and 2010, we finally were able to show that there were
00:14:38.26 a broad array of different tools you could use to
00:14:41.21 generalizably target different kinds of cells. And by this
00:14:45.08 point, I would say by 2010 or so, people were able
00:14:49.23 to see that this could be applied to virtually any system.
00:14:53.18 A good example of how that all worked was in our
00:14:57.16 studies in anxiety. Now anxiety is a normal part of
00:15:02.01 life, it's a healthy part of life, of course it can become
00:15:04.16 excessive and it can be pathological in some cases.
00:15:07.19 But what we were able to do is to use optogenetics
00:15:12.00 to play in patterns of activity or suppress patterns
00:15:14.10 of activity in particular cells or projections in and around
00:15:18.01 structures relating to the basolateral amygdala,
00:15:21.23 and the bed nucleus of the stria terminalis.
00:15:24.24 And the way we did some of these experiments is we
00:15:27.00 injected a virus, an adeno associated viral vector into
00:15:31.06 the structure of interest, the GNST, but then we were able
00:15:35.00 to position our fiber optics in other downstream structures.
00:15:38.14 By doing so, we were able to recruit cells to find by having
00:15:42.09 a particular projection pattern, which as you can see
00:15:45.04 is a generalizable strategy. You don't need a special
00:15:47.26 kind of animal or a special promoter, you just need to know
00:15:51.00 your anatomy. And this works because the opsins are
00:15:53.21 trafficked down the axons of the cells, particularly if we
00:15:56.23 help them out with little trafficking motifs that were
00:16:01.00 discovered in the lab by Viviana Gradinaru and her
00:16:04.01 colleagues. And I'll show you an example of how
00:16:07.23 this sort of thing works. This is an experiment,
00:16:10.02 I'll play this movie, Kay Tye in the lab's work which
00:16:13.27 was published in 2011. This is a what's called an elevated
00:16:18.15 plus maze, this so-called closed arm has these elevated
00:16:23.28 walls and you'll see that the animal prefers to spend its
00:16:28.16 time in the closed arm of the maze. Very rarely venturing
00:16:31.23 out into the open arms of the maze, which is like walking
00:16:34.23 the plank for the animal. Both in our studies of anxiety
00:16:38.11 related behaviors, what we found and what Kay was
00:16:41.07 able to show, and we followed this up with a great deal
00:16:44.11 of other work, was that there were anti-anxiety
00:16:47.24 pathways buried within the brain. When the blue letters
00:16:50.23 come on here, you'll see when she started driving one
00:16:53.16 of these, as we discovered, anti-anxiety pathways.
00:16:56.13 And you'll see for the first time, the mouse is willing
00:16:59.15 to go out into the open arms and explore, which it
00:17:03.01 never had done before. Nothing else changed
00:17:04.15 about the animal, the speed or anything else,
00:17:06.23 it simply was willing to accept this risk. And this was
00:17:10.00 instantaneously reversible and this of course was only
00:17:13.00 one feature of anxiety, apprehension in the absence
00:17:16.01 of immediate threat. But there are other features, too.
00:17:18.01 Respiratory rate changes, subjective feeling,
00:17:21.08 anxiety feels bad and resolution of anxiety feels good.
00:17:24.27 And in later work, we were able to show different
00:17:27.25 projections coming from the bed nucleus of the stria
00:17:31.13 terminalis recruit different features of anxiety using
00:17:34.29 this projection targeting approach, where we position
00:17:37.29 the fiber optic in a different location than the injection
00:17:41.14 of the virus. And this sort of thing allowed the
00:17:45.02 generalizability of optogenetics to become clear.
00:17:48.00 And so, in this sort of work, if you think back to
00:17:51.12 those very early humble experiments in culture,
00:17:54.04 you can see it took a number of years to finally put
00:17:56.14 all the pieces together. I'm often asked what's the
00:18:00.11 potential clinical impact of optogenetics, and although
00:18:03.28 I'm a psychiatrist, I run a purely basic science laboratory.
00:18:06.11 I'm not working on clinical translation in my lab.
00:18:10.08 But it's a natural question to ask, and other people
00:18:12.24 are working on it. The group of Antonello Bonci
00:18:16.03 found in 2013 that they could turn down cocaine seeking
00:18:19.29 behavior in rats with optogenetic stimulation of the
00:18:23.10 medial prefrontal cortex. A very striking result that
00:18:26.17 they published in 2013. And that was then followed up
00:18:30.16 with Antonello Bonci and some clinical colleagues with
00:18:35.09 a optogenetically guided clinical intervention.
00:18:38.12 They used transcranial magnetic stimulation, which can
00:18:41.03 be non-invasively delivered to awake human beings.
00:18:44.17 Of course, human brains are a lot bigger but they have
00:18:47.02 a lot of the same basic reward circuitry and what they found
00:18:50.05 was guided by optogenetics, they were able to
00:18:53.18 have a group of cocaine addicted human beings
00:18:56.02 no longer seek cocaine. They had an experimental group and
00:18:59.13 a control group, and they had a very powerful effect.
00:19:02.24 This sort of thing shows how far we've come,
00:19:06.14 that we can directly affect with a principle-guided
00:19:11.12 approach, these very powerful and important fundamental
00:19:15.25 behaviors in human beings. And this makes, of course
00:19:19.16 coming back to the initial inspiration that we have
00:19:23.13 starting to go down this path, now we do understand
00:19:27.06 in a causal and precise fashion many of the cells
00:19:30.08 and connections across the brain that do affect
00:19:34.08 these very important and both basic and disease
00:19:39.13 relevant behaviors. Now this, of course, had very humble
00:19:44.28 beginnings, as you recall, those initial tallies in the lab
00:19:48.05 notebook. And we've come a long way since then.
00:19:50.03 But in a way, it's even more striking when you think
00:19:53.02 how these very simple organisms themselves
00:19:56.19 have created these tools that let us make this
00:20:00.02 progress. There's a whole other world that has
00:20:03.06 happened as well, in terms of discovering how these
00:20:06.17 basic microbial opsins work. We've been able
00:20:09.10 to adjust the speed of the opsins. Their sensitivity to light.
00:20:12.26 We've been able to engineer them to conduct
00:20:15.02 different kinds of ions to turn cells on or off.
00:20:17.20 Guided by getting to the crystal structure of
00:20:20.26 one of these channel rhodopsins. And we can change
00:20:23.28 the color that they respond to, as well.
00:20:26.11 This has allowed us to do something very important
00:20:29.11 which is to tune the amount and the timing of the activity
00:20:34.16 that we play into precisely what's naturally occurring
00:20:38.14 in the organism, even that same individual during a very
00:20:42.16 well-defined behavior. And so this elegance and precision
00:20:47.20 of the method is all the more striking when you think back
00:20:50.18 to how crude some of the early experiments were.
00:20:52.29 And how long it took to get the whole system working.
00:20:57.03 But in the end, that's what makes biology fun
00:20:59.29 and beautiful. And I hope to have conveyed some of that
00:21:03.09 to you today.

This Talk

Speaker: Karl Deisseroth

Audience:

Educators of H. School / Intro Undergrad
Student
Educators of Adv. Undergrad / Grad
Researcher
Educators

Recorded: September 2016

Talk Overview

Dr. Karl Deisseroth gives a retrospective on the development of optogenetics, a technology that has revolutionized neuroscience. Optogenetics uses pulses of light to alter an animal’s behavior by taking advantage of light-sensitive ion conductance regulators called microbial opsins, which convert a photon of light into an electrical signal. By using genetics to express opsins in particular neurons, scientists can control the activation or suppression of neurons with pulses of light and measure the resulting behavioral changes in an animal.

Speaker Bio

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