In 1937 the great neuroscientist Sir Charles Scott Sherrington of the University of Oxford laid out what would become a classic description of the brain at work. He imagined points of light signaling the activity of nerve cells and their connections. During deep sleep, he proposed, only a few remote parts of the brain would twinkle, giving the organ the appearance of a starry night sky. But at awakening, “it is as if the Milky Way entered upon some cosmic dance,” Sherrington reflected. “Swiftly the head-mass becomes an enchanted loom where millions of flashing shuttles weave a dissolving pattern, always a meaningful pattern though never an abiding one; a shifting harmony of subpatterns.”

Although Sherrington probably did not realize it at the time, his poetic metaphor contained an important scientific idea: that of the brain revealing its inner workings optically. Understanding how neurons work together to generate thoughts and behavior remains one of the most difficult open problems in all of biology, largely because scientists generally cannot see whole neural circuits in action. The standard approach of probing one or two neurons with electrodes reveals only tiny fragments of a much bigger puzzle, with too many pieces missing to guess the full picture. But if one could watch neurons communicate, one might be able to deduce how brain circuits are laid out and how they function. This alluring notion has inspired neuroscientists to attempt to realize Sherrington’s vision.

Their efforts have given rise to a nascent field called optogenetics, which combines genetic engineering with optics to study specific cell types. Already investigators have succeeded in visualizing the functions of various groups of neurons. Furthermore, the approach has enabled them to actually control the neurons remotely—simply by toggling a light switch. These achievements raise the prospect that optogenetics might one day lay open the brain’s circuitry to neuroscientists and perhaps even help physicians to treat certain medical disorders.

Enchanting the Loom
Attempts to turn Sherrington’s vision into reality began in earnest in the 1970s. Like digital computers, nervous systems run on electricity; neurons encode information in electrical signals, or action potentials. These impulses, which typically involve voltages less than a tenth of those of a single AA battery, induce a nerve cell to release neurotransmitter molecules that then activate or inhibit connected cells in a circuit. In an effort to make these electrical signals visible, Lawrence B. Cohen of Yale University tested a large number of fluorescent dyes for their ability to respond to voltage changes with changes in color or intensity. He found that some dyes indeed had voltage-sensitive optical properties. By staining neurons with these dyes, Cohen could observe their activity under a microscope.

Dyes can also reveal neural firing by reacting not to voltage changes but to the flow of specific charged atoms, or ions. When a neuron generates an action potential, membrane channels open and admit calcium ions into the cell. This calcium influx stimulates the release of neurotransmitters. In 1980 Roger Y. Tsien, now at the University of California, San Diego, began to synthesize dyes that could indicate shifts in calcium concentration by changing how brightly they fluoresced. These optical reporters have proved extraordinarily valuable, opening new windows on information processing in single neurons and small networks.

Synthetic dyes suffer from a serious drawback, however. Neural tissue is composed of many different cell types. Estimates suggest that the brain of a mouse, for example, houses many hundreds of types of neurons plus numerous kinds of support cells. Because interactions between specific types of neurons form the basis of neural information processing, someone who wants to understand how a particular circuit works must be able to identify and monitor the individual players and pinpoint when they turn on (fire an action potential) and off. But because synthetic dyes stain all cell types indiscriminately, it is generally impossible to trace the optical signals back to specific types of cells.

Genes and Photons
Optogenetics emerged from the realization that genetic manipulation might be the key to solving this problem of indiscriminate staining. An individual’s cells all contain the same genes, but what makes two cells different from each other is that different mixes of genes get turned on or off in them. Neurons that release the neurotransmitter dopamine when they fire, for instance, need the enzymatic machinery for making and packaging dopamine. The genes encoding the protein components of this machinery are thus switched on in dopamine-producing (dopaminergic) neurons but stay off in other, nondopaminergic neurons.

In theory, if a biological switch that turned a dopamine-making gene on was linked to a gene encoding a dye and if the switch-and-dye unit were engineered into the cells of an animal, the animal would make the dye only in dopaminergic cells. If researchers could peer into the brains of these creatures (as is indeed possible), they could see dopaminergic cells functioning in virtual isolation from other cell types. Furthermore, they could observe these cells in the intact, living brain. Synthetic dyes cannot perform this type of magic, because their production is not controlled by genetic switches that flip to on exclusively in certain kinds of cells. The trick works only when a dye is encoded by a gene—that is, when the dye is a protein.

The first demonstrations that genetically encoded dyes could report on neural activity came a decade ago, from teams led independently by Tsien, Ehud Y. Isacoff of the University of California, Berkeley, and me, with James E. Rothman, now at Yale University. In all cases, the gene for the dye was borrowed from a luminescent marine organism, typically a jellyfish that makes the so-called green fluorescent protein. We tweaked the gene so that its protein product could detect and reveal the changes in voltage or calcium that underlie signaling within a cell, as well as the release of neurotransmitters that enable signaling between cells.

Armed with these genetically encoded activity sensors, we and others bred animals in which the genes encoding the sensors would turn on only in precisely defined sets of neurons. Many favorite organisms of geneticists—including worms, zebra fish and mice—have now been analyzed in this way, but fruit flies have proved particularly willing to spill their secrets under the combined assault of optics and genetics. Their brains are compact and visible through a microscope, so entire circuits can be seen in a single field of view. Furthermore, flies are easily modified genetically, and a century of research has identified many of the genetic on-off switches necessary for targeting specific groups of neurons. Indeed, it was in flies that Minna Ng, Robert D. Roorda and I, all of us then at Memorial Sloan-Kettering Cancer Center in New York City, recorded the first images of information flow between defined sets of neurons in an intact brain. We have since discovered new circuit layouts and new operating principles. For example, last year we found neurons in the fly’s scent-processing circuitry that appear to inject “background noise” into the system. We speculate that the added buzz amplifies faint inputs, thus heightening the animal’s sensitivity to smells—all the better for finding food.

The sensors provided us with a powerful tool for observing communication among neurons. But back in the late 1990s we still had a problem. Most experiments probing the function of the nervous system are rather indirect. Investigators stimulate a response in the brain by exposing an animal to an image, a tone or a scent, and they try to work out the resulting signaling pathway by inserting electrodes at downstream sites and measuring the electrical signals picked up at these positions. Unfortunately, sensory inputs undergo extensive reformatting as they travel. Consequently, knowing exactly which signals underlie responses recorded at some distance from the eye, ear or nose becomes harder the farther one moves from these organs. And, of course, for the many circuits in the brain that are not devoted to sensory processing but rather to movement, thought or emotion, the approach fails outright: there is no direct way of activating these circuits with sensory stimuli.

From Observation to Control
An ability to stimulate specific groups of neurons directly, independent of external input to sensory organs, would alleviate this problem. We wondered, therefore, if we could develop a package of tools that would not only provide sensors to monitor the activity of nerve cells but would also make it possible to readily activate only selected neuron types.

My first postdoctoral fellow, Boris V. Zemelman, now at the Howard Hughes Medical Institute, and I took on this problem. We knew that if we managed to program a genetically encoded, light-controlled actuator, or trigger, into neurons, we could overcome several obstacles that had impeded electrode-based studies of neural circuits. Because only a limited number of electrodes can be implanted in a test subject simultaneously, researchers can listen to or excite only a small number of cells at any given time using this approach. In addition, electrodes are difficult to aim at specific cell types. And they must stay put, encumbering experiments in mobile animals.

If we could tap a genetic on-off switch to help us find all the relevant neurons (those producing dopamine, for instance) and if we could use light to control these cells in a hands-off manner, we would no longer have to know in advance where in the brain these neurons were located to study them. And it would not matter if their positions changed as an animal moved about. If stimulation of cells containing the actuators evoked a behavioral change, we would know that these cells were operating in the circuit regulating that behavior. At the same time, if we arranged for those same cells to carry a sensor gene, the active cells would light up, revealing their location in the nervous system. Presumably, by rerunning the experiment repeatedly on animals engineered to each have a different cell type containing an actuator, we would eventually be able to piece together the sequence of events leading from neural excitation to behavior and to identify all the players in the circuit. All we needed to do was discover a genetically encodable actuator that could transduce a light flash into an electrical impulse.

To find such an actuator, we reasoned that we should look in cells that normally generate electrical signals in response to light, such as the photoreceptors in our eyes. These cells contain light-absorbing antennae, termed rhodopsins, that when illuminated instruct ion channels in the cell membrane to open or close, thereby altering the flow of ions and producing electrical signals. We decided to transplant the genes encoding these rhodopsins (plus some auxiliary genes required for rhodopsin function) into neurons grown in a petri dish. In this simple setting we could then test whether shining light onto the dish would cause the neurons to fire. Our experiment worked—in early 2002, four years after the development of the first genetically encoded sensors able to report neural activity, the first genetically encoded actuators debuted.

Remote-Controlled Flies
More recently, investigators have enlisted other light-sensing proteins, such as melanopsin, which is found in specialized retinal cells that help to synchronize the circadian clock to the earth’s rotation, as actuators. And the combined efforts of Georg Nagel of the Max Planck Institute for Biophysics in Frankfurt, Karl Deisseroth of Stanford University and Stefan Herlitze of Case Western Reserve University have shown that another protein, called channelrhodopsin-2—which orients the swimming movements of algae—is up to the job. There are also a variety of genetically encoded actuators that can be controlled via light-sensitive chemicals synthesized by us and by Isacoff and his U.C. Berkeley colleagues Richard H. Kramer and Dirk Trauner.

The next step was to demonstrate that our actuator could work in a living animal, a challenge I posed to my first graduate student, Susana Q. Lima. To obtain this proof of principle, we focused on a particularly simple circuit in flies, one consisting of just a handful of cells. This circuit was known to control an unmistakable behavior: a dramatic escape reflex by which the insect rapidly extends its legs to achieve liftoff and, once airborne, spreads its wings and flies. The trigger initiating this action sequence is an electrical impulse emitted by two of the roughly 150,000 neurons in the fly’s brain. These so-called command neurons activate a subordinate circuit called a pattern generator that instructs the muscles moving the fly’s legs and wings.

We found a genetic switch that was always on in the two command neurons but no others—and another switch that was on in neurons of the pattern generator but not in the command neurons. Using these switches, we engineered flies in which either the command neurons or the pattern-generator neurons produced our light-driven actuator. To our delight, both kinds of flies took off at the flash of a laser beam, which was strong enough to penetrate the cuticle of the intact animals and reach the nervous system. This confirmed that both the command and pattern-generating cells participated in the escape reflex and proved that the actuators worked as intended. Because only the relevant neurons contained the genetically encoded actuator, they alone “knew” to respond to the optical stimulus—we did not have to aim the laser at specific target cells. It was as if we were broadcasting a radio message over a city of 150,000 homes, only a handful of which possessed the receiver required to decode the signal; the message remained inaudible to the rest.

One nagging quandary remained, however. The command neurons initiating the escape reflex are wired to inputs from the eyes. These inputs activate the escape circuit during a “lights-off” transition, as happens when a looming predator casts its shadow. (You know this from your fly-swatting attempts: whenever you move your hand into position, the animal annoyingly jumps up and flies away.) We worried that in our case, too, the escape reflex might be a visual reaction to the laser pulse, not the result of direct optical control of command or pattern-generating circuits.

To eliminate this concern, we performed a brutally simple experiment: we cut the heads off our flies. This left us with headless drones (which can survive for a day or two) that harbored the intact pattern-generating circuitry within their thoracic ganglia, which form the rough equivalent of a vertebrate’s spinal cord. Activating this circuit with light propelled the otherwise motionless bodies into the air. Although the drones’ flights often began with tumbling instability and ended in spectacular crashes or collisions, their very existence proved that the laser controlled the pattern-generating circuit itself—there was no other way these headless animals could detect and react to light. (The drones’ clumsy maneuvers also illustrated vividly that the Wright brothers’ great innovation was the invention of controlled powered flight, not simply powered flight.)

We also engineered flies with light switches attached only to neurons that make the neurotransmitter dopamine. When exposed to the laser’s flash, these flies suddenly became more active, walking all around their enclosures. Previous studies had indicated that dopamine helps animals predict reward and punishment. Our fly findings are consistent with this scenario: the animals not only became more active, they also explored their environment differently, as if reacting to an altered expectation of gain or loss.

An Unexpected Forerunner
Three days before the paper reporting these experiments was scheduled for publication in the journal Cell, I was flying to Los Angeles to deliver a lecture. A friend had given me Tom Wolfe’s recently published coming-of-age novel I Am Charlotte Simmons, thinking I would enjoy its depiction of neuroscientists, not to mention the material that had earned the book the Literary Review’s Bad Sex in Fiction Award. On the plane I came across a passage in which Charlotte attends a lecture on the work of one José Delgado, who also remotely controls animal behavior—not with light-driven, genetically encoded actuators but with radio signals transmitted to electrodes he has implanted in the brain. A Spaniard, Delgado risked his life to demonstrate the power of his approach by stopping an angry bull in midcharge. This, Wolfe’s fictional lecturer declares, is a turning point in neuroscience—a decisive defeat of dualism, the notion that the mind exists as an entity separate from the brain. If Delgado’s physical manipulations of the brain could change an animal’s mind, so the argument went, the two must be one and the same.

I almost fell out of my seat. Was Delgado a fictional character, or was he real? Immediately after landing in L.A., I did a Web search and was directed to a photograph of the matador with the remote and his bull. Delgado, I learned, had been a professor at my very own institution, Yale, and had written a book entitled Physical Control of the Mind: Toward a Psychocivilized Society, which appeared in 1969. In it, he summarized his efforts to control movements, evoke memories and illusions, and elicit pleasure or pain [see “The Forgotten Era of Brain Chips,” by John Horgan; Scientific American, October 2005]. The book concludes with a discussion of what the ability to control brain function might imply for medicine, ethics, society and even warfare. Against this background, I should probably not have been surprised when the phone rang the day our paper was published and a U.S.-based journalist asked, “So, when are we going to invade another country with an army of remote-controlled flies?”

The media attention did not stop there. The next day the headline of the Drudge Report screamed, “Scientists Create Remote-Controlled Flies,” topping news of Michael Jackson’s latest court appearance. I assume it was this source that inspired a sketch on the Tonight Show a week or so later, in which host Jay Leno piloted a remote-controlled fly into President George W. Bush’s mouth—the first practical application of our new technology.

Since then, researchers have used the light-switch approach to control other behaviors. Last October, Deisseroth and his Stanford colleague Luis de Lecea announced the results of a mouse study in which they used an optical fiber to deliver light directly to neurons that produce hypocretin—a neurotransmitter in the form of a small protein, or peptide—to see whether these neurons regulate sleep. Researchers had suspected that hypocretin plays this role because certain breeds of dogs lacking hypocretin receptors suffer sudden bouts of sleepiness. The new work revealed that stimulating hypocretin neurons during sleep tended to awaken the mice, bolstering that hypothesis.

And in my lab at Yale, postdoctoral fellow J. Dylan Clyne used genetically encoded actuators to gain insights into behavioral differences between the sexes. The males of many animal species go to considerable lengths in wooing the opposite sex. In the case of fruit flies, males vibrate one wing to produce a “song” that females find quite irresistible. To probe the neural underpinnings of this strictly male behavior, ­Clyne used light to activate the pattern generator responsible for the song. He found that females, too, possess the song-making circuitry. But under normal circumstances they lack the neural signals required for turning it on. This discovery suggests that male and female brains are wired largely the same way and that differences in sexual behaviors arise from the action of strategically placed master switches that set circuits to either male or female mode.

Light Therapy
Thus far investigators have typically engineered animals to carry either a sensor or an actuator in neurons of interest. But it is possible to outfit them with both. And down the road, the hope is that we will be able to breed subjects that have multiple sensors or actuators, which would allow us to study assorted populations of neurons simultaneously in the same individual.

Our newfound authority over neural circuits is creating enormous opportunities for basic research. But are there practical benefits? Perhaps, although I feel they are sometimes overhyped. Delgado himself identified several areas in which direct control of neural function could lead to clinical benefits: sensory prosthetics, therapy for movement disorders (as has now become reality with deep-brain stimulation for Parkinson’s disease), and regulation of mood and behavior. He saw these potential uses as a direct and rational extension of existing medical practice, not as an alarming foray into the ethical quicksands of “mind control.” Indeed, it would seem arbitrary and hypocritical to draw a sharp boundary between physical means for influencing brain function and chemical manipulations, be they psychoactive pharmaceuticals or the cocktail that helps you unwind after a hard day. In fact, physical interventions can arguably be targeted and dosed more precisely than drugs, thus reducing side effects.

Some studies have already begun to probe the applicability of optogenetics to medical problems. In 2006 researchers used light-activated ion channels to restore photosensitivity to surviving retinal neurons in mice with photoreceptor degeneration. They used a virus to deliver the gene encoding channelrhodopsin-2 to the cells, injecting it directly into the animals’ eyes. The patched-up retinas sent light-evoked signals to the brain, but whether the procedure actually brought back vision remains unknown.

Despite their theoretical appeal, optogenetic therapies face an important practical obstacle in humans: they require the introduction of a foreign gene—the one encoding the light-controlled actuator—into the brain. So far gene therapy technology is not up to the challenge, and the Food and Drug Administration is sufficiently concerned about the associated risks that it has banned such interventions for the time being, except for tightly restricted experimental purposes.

The immediate opportunity afforded by our control over brain circuits—or even other electrically excitable cells, such as those that produce hormones and those that make up muscle—lies in revealing new targets for drugs: if experimental manipulations of cell groups X, Y and Z cause an animal to eat, sleep or throw caution to the wind, then X, Y and Z are potential targets for medicines against obesity, insomnia and anxiety, respectively. Finding compounds that regulate neurons X, Y and Z may well lead to new or better treatments for disorders that have no therapies at the moment or to new uses for existing drugs. Much remains to be discovered, but the future of optogenetics shines brightly.

For charts and graphs related to this story , click the images below to enlarge


Note: This article was originally printed with the title, "Lighting Up the Brain".