Cerebrum Article

The Miracle of Light

Karl Deisseroth, a winner of the 2021 Lasker Prize in Basic Medical Research for his pivotal work in optogenetics and author of the widely praised book, Projections: A Story of Human Emotions, explains this new instrument in a neuroscientist’s toolkit, its potential to help people, and what he’s learned as a practicing psychiatrist.

Published: April 15, 2022
Author: Carl Sherman
Photo of Karl Deisseroth, M.D., Ph.D., in his lab

Karl Deisseroth in his lab. Photograph by Steve Fisch

How is the brain put together? How does it work—what happens when we think, feel, act? What goes wrong in psychiatric and neurological illness?  A giant step toward answering these central questions of neurobiology emerged over the last 15 years with the development of optogenetics. This now-essential instrument in the neuroscientist’s toolkit represents a convergence of botany, microbiology, proteomics, genomics, and optics, and an accordingly wide array of contributions by an impressive cast of researchers.

A leader of the pack is unarguably Karl Deisseroth, M.D., Ph.D., professor of bioengineering and of psychiatry and behavioral sciences at Stanford University, one of three scientists awarded the 2021 Lasker Prize in Basic Medical Research for his pivotal work. Deisseroth, a practicing psychiatrist as well as a researcher, hit another milestone last year with the publication by Random House of Projections: A Story of Human Emotions, his widely-praised book that brings bench and bedside together with a detailed, empathic account of his work with patients, in the context of the brain dynamics underlying their difficulties—according to optogenetics.

CARL SHERMAN: Let’s talk about optogenetics. Can you give me a capsule version of just what it is?

KARL DEISSEROTH: Optogenetics is a way of using light to study how a biological system works. But it’s the opposite of how we normally use light, which is to collect information, to look at something and see what happens. Instead, optogenetics uses light to control what happens.

The way we do this is make some cells responsive to light, so we can turn them on or off with it. Then we can see what kinds of activity, what kinds of cells, and what connections between cells actually matter for the mysterious and wonderful things that the brain accomplishes; for sensation, cognition, action.

More specifically, how do you transform these cells?

We use genes from microbes, single-celled organisms, that for their own reasons have evolved to make light-activated ion channels, little molecules that sit in the cell membrane and move ions—charged particles—across it. They create a current in response to light.

In evolving these beautiful, light-sensitive proteins, these microorganisms have done an important job for us, because we can use genetic tricks with viruses to put their genes into neurons, and neurons use electricity to turn themselves on and off.  Then we can guide light into the brain, and only the cells that have this gene will be turned on or off by it, depending on which microbial gene we put in.

It can get very fancy. We can use holographic techniques to guide dozens or hundreds of spots of light to individual cells and control them.

I understand that discoveries by a number of scientists went into the development of this technique, starting with the identification of these light-sensitive proteins 150 years ago. What was your lab’s contribution? 

I’m a neuroscientist, but I’m also a biochemist by training. All through my career I’ve taken a molecular approach to systems-level questions. What my lab did was figure out how these molecules work, bridging all the way to their implementation in neuroscience. A lot of real basic science that enabled the applications came later.

First, we discovered the structures of channelrhodopsins, these light-activated proteins, which allowed us to see how they work, and exactly which atoms and which movements are important. That let us redesign them, to change the colors of light that they respond to, and the kind of ions that they let through. Along the way we discovered major new types of channelrhodopsins.

And we built the tools to make optogenetics work—the viral vectors to put these genes into neurons in the brains of mammals, and the fiberoptic methods to guide light and target cells and connections in different parts of the brain.

Just how do you target and activate specific neurons? You’re introducing the viral vector into an area of the brain, yet it only affects certain cells in that area. How do you make this happen?

The virus gets the gene into all the nearby cells, but the genes are not turned into functional protein (channelrhodopsins) in all of them. That step is achieved with bits of DNA called promoters and enhancers, which regulate gene function, and we design the virus so that the protein is only made by the specific, targeted type of cell.

We achieve additional specificity by guiding light spots to a subset of the cells in which the protein is made. And then finally, we developed projection targeting, where you deliver the opsin—the light-sensitive protein—to one part of the brain and deliver the light to another.

The reason this works is that the opsin protein fills the whole neuron; it gets trafficked out into the long axons that extend from one part of the brain to another. If you put your fiber optic in that other brain region, the only thing that’s sensitive to light will be the axons that started from where you injected your virus. So you get specificity by connection type.

So you can use optogenetics to discover neural connections that hadn’t been mapped before and delineate them with more precision?

Exactly. You could demonstrate functional connectivity very efficiently, and on different scales. On a local circuit scale, you can zip your light around to all the cells individually or in combination, while listening to or recording activity. You can see which of the cells you stimulate, which types of cells, which locations, which combinations give rise to the activity of interest.

On the other hand, if you test all the cells and combinations of cells with the broad, high-content recording technologies we’ve developed, you can be quite global and unbiased about the kind of information you collect.

Also, there have long been ways to slowly ramp up or ramp down the activity of individual neurons—chemical methods, for example. But with those methods you don’t have naturalistic timing information, which turns out to be critical. With optogenetics you can be very, very rapid and swift.

Why is timing so important?

The brain is designed to work on millisecond time scales. The speed, the brevity of action potential transmission, is important for survival. We see this all over neuroscience. If you give a stimulus the frequency of 10 Hertz, you’ll get a totally different result than if you stimulate at 80 or 100 Hertz.

Could you give me examples how optogenetics has been applied in neuroscience? Some discoveries it’s facilitated, processes it’s clarified, etc.

One thing is anxiety. We all know anxiety; we know it’s got these different parts to it. We know it feels bad. We know it affects our body: It makes us breathe more quickly and our heart beat faster. And we know it affects our choices: When we’re anxious, we don’t do the risky things that might otherwise happen. All these features, that are so different, come on together and go away together. So somehow our brain has built a way to create this state, to make it coherent and unitary, to have anxiety happen and then go away.

We used optogenetics in my lab to sort out how this happens, how the different features are obtained and wrapped up together. This was done using projection targeting. You can show that there’s a part of the extended amygdala called the bed nucleus of the stria terminalis (BNST) and that different projections coming out from the BNST, going to different parts of the brain, access these different features. There’s a projection that goes to the parabrachial nucleus, part of the brainstem that controls respiratory rate, the breathing changes associated with anxiety. But it doesn’t affect the behavioral decisions, or the positive or negative feeling-state, at all.

There’s another projection that goes from the BNST to the lateral hypothalamus, and that controls the behavioral decisions but doesn’t affect the breathing rate at all, or the subjective valence, the feeling-state.

And there’s a third that goes to the ventral tegmental area. This is the part of the brain where dopamine neurons live, that seems to set the negativity of anxiety and the positivity that comes from the release of anxiety. But it doesn’t affect respiratory rate or behavioral decisions. So, these different features can be cleanly broken apart, and they’re assembled by these projections.

Just how did you parse the anxiety response—how did you show projections from the BNST to different brain regions to be responsible for distinct facets of anxiety?

We did this in mice, which have many of the same structures that we have and exhibit robust anxiety responses, using the projection targeting method that I described. We delivered excitatory or inhibitory channelrhodopsins to the BNST, and then we delivered fiberoptics to different downstream regions that we knew were connected to it but didn’t know how. And we explored what the causal impact was on which parts, which behaviors were elicited. We observed this clean deconstruction of the different features of anxiety, as a result of controlling these different projections from the anxiety control region of the BNST.

To take another example, Catherine Dulac at Harvard did the same kind of thing five years later for parenting. This is quintessential: There’s not much that’s more important to a mammal than parenting, right? As it turns out, as complex a state as it is, aspects of parenting can be broken down into component parts that map onto physical structures in the brain. What Catherine Dulac showed, using optogenetic projection targeting, was that there’s a connection that starts from one part of the brain and goes to another, that mediates the grooming of the young, the physical care that all mammals do. But it doesn’t affect the motivation to go out and retrieve the young that had strayed from the nest. That was controlled by a different connection, a different projection, which didn’t affect grooming at all. And there are many such examples.

You’re also a practicing psychiatrist. What kind of work do you do?

Well, I really do three kinds of work. There are the extremely acute cases that come to the emergency room. They’ve been brought in by police, or by family, or they bring themselves in because they know they’re in danger. I’m the attending physician, and I work with the emergency room doctors and others, to sort out the right care.

I also run an outpatient clinic.  I see those patients once a month, or some, if they’re quite ill, more often, even every week. I focus on very severe cases, hard-to-treat illnesses, particularly two things: treatment-resistant depression and autism.

And finally—and this is more recent—we’re doing clinical trials, trying to understand the internal brain processes, the dynamics that underlie complex altered brain states like dissociation, which is a very important and mysterious brain state that shows up all across psychiatry.

How does your clinical work relate to your research?  In your book, you suggest a kind of synergy between them.

It’s so important to be able to go to my trainees in the lab, my students and postdoctoral fellows, and to tell them, “Look, this is what anxiety is really like. This is what depression is really like. This is what matters to someone with autism. It isn’t just one symptom from a list; it really disrupts their life, and this is how they feel about it.” That is so helpful to guide the research in the lab.

It works the other way, too. In the lab, we found this very interesting altered brain state that was associated with dissociation, and we were able to go and find that same exact state happening in a human being. That’s now guiding our ongoing clinical work, including the clinical trial that I mentioned. So it really works both ways. The science enhances the patient work. The patient work enhances the science. I wouldn’t do it any other way now.

Speaking of the connection between clinical work and science, have there been any clinically relevant applications of optogenetics?

Optogenetics, first and foremost is a discovery tool. It lets us understand what actually matters in the brain—which cells, which activity, and which connections are important for sensation, cognition, and action, whether healthy and adaptive, or maladaptive. This is so powerful because once you understand at that level, then any kind of treatment becomes much more grounded. It could be a pill. It could be a brain stimulation treatment: Now you can target specific? components.

That would be the broadest and biggest impact, you could call it indirect application, guiding treatment. But my colleague and friend, Botond Roska in Switzerland, just published the first direct application of optogenetics in the human central nervous system this past year, in Nature Medicine. It was used to treat someone who was blind, and he was able to confer new light sensitivity onto the retina of this person with optogenetics. It’s pretty amazing. You can see videos of this on the journal site: A person sitting at a table who had been unable to see objects was now able to reach for them in a directed way.

Are there any clinical trials under way? 

There are a bunch involving the retina. There’s some work percolating, exploring pain treatment. And then there are a lot of optogenetics-guided clinical trials—where people have discovered something using optogenetics and are using that to design brain-stimulation therapies, medication therapies, and so on.

What are your thoughts about the immediate or intermediate-term future of optogenetics?  

Thousands of discoveries have been made with optogenetics, all over the world. It’s been applied in essentially every experimental animal system, to many different kinds of cells and questions. And just in the past couple of years it’s made this leap from the level of cell types to the level of single cells. That’s very exciting—mammalian behavior controllable at the single-cell level. And so the future will be understanding how many cells are needed for a sensation, or a cognition, or an action. Which cells, and which regions, work together? What’s essential, what’s not at the single cell level?

The other development is all the way at the other end of the scale—at the whole, intact-brain level. Getting to all the cells. We can now do this in the zebra fish, which is a small, transparent vertebrate. It’s small enough and transparent enough that we can see all the way through it, see all the cells during behavior, and during optogenetic control.

Of course, there’s a huge amount of data. Basically, the people in the lab have to become near-professional data scientists. But it’s got the appeal of not missing anything. We can take global, unbiased perspectives, which we definitely need in neuroscience because there are so many unknowns. In many cases, we can’t even frame hypotheses well. We want to take the big picture. We want to play the role of an astronomer centuries ago who built a new telescope and was pointing it at part of the sky to see what could be seen.

And this is already bearing fruit, this approach. The dissociation question I mentioned earlier: nobody really knew—we certainly didn’t—how dissociation might be implemented in the brain. What it actually is, physically, materially. And by taking that very broad, unbiased approach, we were able to see a signal that nobody anticipated. It popped out at us and turned out to be really important, both in mouse and in human for dissociation.

How did you do this? 

We used what we call wide-field imaging, where you can see almost across the entire brain. We gave a range of different strong, psychoactive drugs, including dissociative agents. We saw an amazing pattern that we hadn’t predicted: a rhythm in one part of the brain that all the dissociative drugs elicited, but none of the non-dissociative drugs. And then optogenetics came in, because then we could say “does it matter?” We provided that rhythm to those cells, and we saw if it caused the dissociative behavior. And it did. Then we were able to go and look for the same rhythm in human beings who were dissociating, and we found it.

Final thoughts about optogenetics?

It highlights how important basic, curiosity-driven research is. The deepest roots of optogenetics are in botany, 150 years ago. If people weren’t poking around with weird microorganisms a century or more ago, we wouldn’t be where we are today.

magazine cover, with image of two children facing each other across the line of a brain's corpus callosum

This article first appeared in the Spring 2022 issue of our Cerebrum magazine. Click the cover for the full e-magazine.