How a protein from green algae lets scientists control chosen brain cells with light

In Short. This year's Nobel prize in medicine rewards algal proteins that let researchers turn chosen nerve cells on with light. Most of the work has been in animals, and human use is limited to small early trials to restore sight.

A SINGLE-CELLED green alga called Chlamydomonas reinhardtii swims toward light. Peter Hegemann spent years trying to identify the molecule it uses to sense the light. The answer, which he and Georg Nagel published in 2002 and 2003, was a protein in the alga's outer surface that lets charged particles through it when light hits it. On October 5th the Nobel Assembly at the Karolinska Institutet gave the 2026 prize in physiology or medicine to the two of them and to Karl Deisseroth of Stanford University and the Howard Hughes Medical Institute, who showed that the same protein could make nerve cells in a dish, and then in living animals, fire when a researcher shone a light on them. Hegemann is now at Humboldt University of Berlin and Nagel at the University of Würzburg.

An algal protein won a medicine prize because of a problem that held back brain research for most of the 20th century. Scientists could record which parts of the brain were active while an animal moved or fell asleep, or destroy a region, give a drug or pass a current through an electrode and watch what changed. In its scientific background to the award, the Nobel committee writes that these methods were good at linking brain activity to behavior but "lacked the precision to manipulate specific cell types within intact circuits on the millisecond timescale relevant to neural computation."

The difference matters because the brain packs many types of cell close together. An electrode excites whatever lies near its tip, including fibers that are merely passing through on their way somewhere else. Finding that a group of cells is busy while a mouse freezes in fear shows only that the activity and the fear happen at the same time. To show that those cells cause the fear, a researcher needs to activate that one type of cell, or silence it, at a moment of their choosing, and see whether the behavior follows. Francis Crick, one of the discoverers of the structure of DNA, had suggested that the best way to do this would be with light, an idea he called "rather far-fetched", according to the committee.

A gate that opens in light

Nerve cells, or neurons, carry messages as a brief electrical pulse that travels along the cell. This pulse is called an action potential, and the rest of this article calls it simply a nerve signal. A nerve signal starts when tiny gates in the cell's surface open and let electrically charged atoms, called ions, flow into or out of the cell. Each such gate is a protein known as an ion channel. In a neuron, most of these channels open in response to a change in voltage or the arrival of a chemical message from a neighboring cell.

Light is detected by another family of proteins. The rod cells in the human eye hold a protein that changes shape when it absorbs light, with the help of a small molecule called retinal. That protein is called rhodopsin. In the eye, rhodopsin does not let ions through itself. It passes its signal on to other molecules inside the cell.

Hegemann's group found a gene in Chlamydomonas for a protein related to rhodopsin. To test what it did, Nagel put the gene into the eggs of a frog, Xenopus laevis, which are large enough to record from, and supplied retinal. In the paper they published in Science in 2002, green light at about 500 nanometers drove a current into the eggs and red light did not. The current was "carried by protons", which are hydrogen ions, and when the light went off it died away over tens of milliseconds, with time constants of 35 and 150 milliseconds at 19°C in one egg. Most of the measurements were typical of five eggs. The authors concluded that the protein was "a combined photoreceptor and ion channel", and they named it channelrhodopsin-1. A channelrhodopsin, then, is a light-sensing protein that is itself the gate.

The second gene, published in PNAS in 2003, mattered more. Channelrhodopsin-2, as it was called, responded most strongly to blue light at about 460 nanometers, measured in three eggs. It let through many kinds of positive ion, including sodium, potassium and calcium, but no negative ions. Using laser flashes lasting 10 billionths of a second, the team found that the current rose within about 200 microseconds, or a fifth of a millisecond. Under steady light the current fell to a lower but stable level, which the authors described as typical of 24 eggs. The team also put the gene into human kidney cells (more than ten recorded) and hamster kidney cells (more than six). Those cells became sensitive to light, and blue light changed the voltage across their surface by "tens of mV" in five recordings. The authors wrote that the protein could be used to change the voltage of "small or large cells, simply by illumination."

Neurons that fire when lit

A change in voltage is what starts a nerve signal, so the obvious test was to put channelrhodopsin-2 into neurons. In 2005 Edward Boyden, Feng Zhang, Ernst Bamberg, Nagel and Deisseroth, working in Deisseroth's lab at Stanford, reported in Nature Neuroscience that they had done so. To get the gene into the cells they used a lentivirus that had been altered to carry a chosen gene into cells. A virus modified in this way to deliver genes is called a viral vector. Because the gene can be paired with a stretch of DNA that is active only in one kind of cell, a viral vector is what lets researchers put the protein into one type of neuron and leave its neighbors untouched.

The neurons came from the hippocampus of newborn rats and were grown in dishes. The team lit them with blue light between 450 and 490 nanometers through a microscope, in pulses of 5, 10 or 15 milliseconds. Of 18 neurons recorded, 13 fired reliable nerve signals in response to the pulses, and the other 5 changed their voltage but stayed below the level needed to fire. In 10 neurons given steady light, the first nerve signal peaked about 8 milliseconds after the light came on, and that delay varied by only about half a millisecond from one trial to the next.

The team also played trains of pulses at irregular intervals to seven neurons. In two such series, 76% and 85% of the pulses produced a nerve signal, and when a series was repeated, more than 95% of the pulses gave the same result on both trials. In 13 neurons, regular trains of 20 pulses drove firing at rates from 5 to 30 per second, though more pulses failed at the faster rates. The researchers added no retinal to the dishes, so the neurons supplied their own. They also looked for harm. In 18 neurons with the protein and 18 without, the electrical resistance of the cell surface and the resting voltage were about the same, and dye that enters only dying cells was taken up by 1 of 56 neurons with the protein and 1 of 49 without.

According to the committee, the method was given the name optogenetics in 2006: a light-sensitive protein is put into a chosen type of neuron by a viral vector, and light is then delivered to those cells to activate them. In 2007 Deisseroth's group put channelrhodopsin-2 into one type of neuron in the region of the mouse brain that controls movement, and lit it through a thin optical fiber passed through a small hole in the skull. The light made the mice's whiskers move. The same year, two groups, one in Deisseroth's lab and one led by Xue Han and Boyden, reported the reverse tool, a pump called halorhodopsin, taken from a microbe, that moves chloride ions into a neuron under yellow light and stops it firing.

From mice to monkeys

A monkey brain is much larger than a mouse brain, and a monkey's immune system might attack a foreign algal protein. In 2009 Han, by then working with Boyden at the Massachusetts Institute of Technology, reported in Neuron on tests in two rhesus macaques. The team injected a lentivirus carrying channelrhodopsin-2 into the frontal cortex, near the front of the brain, and paired the gene with a stretch of DNA that is active in excitatory neurons, the type that tells its neighbors to fire. Each injection labeled a roughly spherical patch of cortex about 1.4 millimeters across. In tissue from one of the two monkeys, all 127 of the labeled cells that were checked carried a marker of excitatory neurons, none of 78 checked carried a marker of inhibitory neurons and none of 84 carried a marker of support cells called astrocytes.

The light came from a blue laser at 473 nanometers through an optical fiber 200 micrometers wide, placed beside a recording electrode. During 200-millisecond pulses of light, 50 recorded units fired faster and 20 fired more slowly. (A unit is a single neuron or a small cluster of them picked up by one electrode, and the paper does not say how many units were recorded in total.) In 15 well-isolated excited neurons, the average firing rate rose from 6.5 signals per second at baseline to 37 in the first 20 milliseconds of light. At 32 sites in cortex that had not received the virus, light never changed the activity. Over many months of repeated injections and lighting sessions, the team saw no damage to neurons or their support cells and "no immune reaction at the cellular or antibody level."

The same paper shows the main physical limits of the method. One injection labeled about a cubic millimeter of brain, which the authors noted is "comparable to the volume illuminated by single optical fibers." The light from the tip of the fiber changed the activity of neurons a little over 1.2 millimeters away. To reach deeper or larger areas, light has to be delivered through fibers implanted in the brain, and the gene has to be delivered first by injecting a virus. The authors wrote that "future viral, promoter, injection, and illumination innovations will need to be developed to match the manipulations possible in mice."

Cause, then sight

In mice, the method has been used to test which cells cause a behavior. The committee's background cites a 2007 study by Antoine Adamantidis, Luis de Lecea, Deisseroth and colleagues, who put channelrhodopsin-2 into neurons in the hypothalamus that make a chemical messenger called orexin. Lighting those cells through a fiber made it more likely that sleeping mice would wake up. It also cites a 2012 study by Xu Liu, Susumu Tonegawa and colleagues, who put the protein into neurons in the hippocampus that had been active while mice learned to fear a particular cage. When the researchers lit those cells later, in a different cage, the mice showed a fear response.

The first use in people has been in the eye. Retinitis pigmentosa is an inherited disease that destroys the eye's rods and cones, the cells that detect light, while some other cells in the retina survive. In 2021 José-Alain Sahel, Botond Roska and colleagues reported in Nature Medicine on a single blind patient with the disease. A viral vector carrying a related light-sensitive channel called ChrimsonR was injected into one eye to make retinal ganglion cells, which normally pass signals from the retina toward the brain, respond to light. The patient wore goggles with a camera that detects changes in light and projects matching light pulses onto the retina. Wearing the goggles, the patient "perceived, located, counted and touched different objects" using the treated eye alone, and recordings from the scalp showed activity over the brain's visual area linked to the objects. Before the injection, or afterwards without the goggles, the patient could not see any of the objects.

This is one person, in a trial whose protocol planned to treat 12 to 18. The paper's supplementary material reports no adverse effects in the eye or elsewhere in the body across 15 visits over 84 weeks. The committee notes that several trials of the approach in retinitis pigmentosa are under way. Two decades after the frog-egg recordings, optogenetics lets researchers switch a chosen type of nerve cell on or off with light to test what those cells cause, and in people its use is still limited to small early trials that aim to restore some sight.