In the 1920s Harvard University graduate student Clyde E. Keeler discovered two surprising facts about mice he had bred in his rented attic room. One, all the progeny were completely blind. Two, despite the animals’ blindness, their pupils still constricted in response to ambient light, albeit at a slower rate than did the pupils of sighted mice.

Many years later researchers extended Keeler’s observation, showing that mice genetically engineered to lack rods and cones (the light receptors involved in vision) nonetheless reacted to changes in light by adjusting their circadian clock—the internal timer that synchronizes hormone activity, body temperature and sleep. The animals performed the usual daytime activities when in daylight and nighttime activities when in the dark. They could do so even though their retinas lacked the photoreceptor cells that vertebrate eyes use to form images, although surgically removing their eyes abolished this ability. This phenomenon may be common to many mammals, including humans: recent experiments have shown that certain blind people can also adjust their circadian clocks and constrict their pupils in response to light.

One explanation for the apparent paradox is that the photoreceptors within the eye that are required for vision are not responsible for regulating the timing of daily activity; other receptors do that. But until quite recently, the notion that the eyes could possess photoreceptors other than rods and cones seemed absurd. The retina is one of the most thoroughly studied tissues in the body, and the only photoreceptors known to exist in the eyes of mammals were the familiar duo, rods and cones.

The evidence is now very convincing, however, that the eyes of mammals, including those of humans, do have specialized photoreceptors that are not involved in image formation. The light-detecting molecules in these cells are different from those in rods and cones, and the cells connect to different parts of the brain. Thus, just as our ears provide us with our sense of equilibrium as well as with our hearing, each of our eyes is also essentially two organs in one.

The discovery may lead to help for people who have trouble adjusting their biological clocks. Jet lag is the most obvious manifestation of circadian desynchrony—the loss of synchronization between the cycling of day and night and our internal clock. Working the night shift, a self-imposed form of the condition, is thought to raise one’s risk of cardiovascular disease, gastrointestinal distress, cancer and metabolic syndrome—a condition that can ultimately lead to type 2 diabetes and stroke. Some of history’s most infamous industrial accidents, such as the 1989 grounding of the oil tanker Exxon Valdez, the 1984 explosion at the Union Carbide plant (now owned by Dow Chemical Company) in Bhopal, India, and the 1979 near-nuclear meltdown at Three Mile Island, occurred during the night shift, when worker vigilance was compromised. Moreover, millions of people living at extreme northern or southern latitudes suffer from seasonal affective disorder, an often severe form of depression that also appears to be a response to lack of light during short winter days. Better understanding of how the third kind of photoreceptor controls circadian rhythms and emotions is already suggesting ways to minimize the negative effects of jet lag, night-shift work and long winter nights.

Light-Sensing but Overlooked
Biologists have long known of organisms that have light-detecting organs for purposes other than image formation. A change in illumination may signal to an animal that it has become exposed, which in turn indicates a vulnerability to predators or potential damage from ultraviolet radiation. Many animals have evolved adaptations, such as active camouflage or avoidance of light, to minimize the consequences of being exposed. Although these adaptations require some system of light detection, they do not require vision per se. For example, in 1911 Austrian zoologist and future Nobel Prize–winner Karl von Frisch recognized that blinded European minnows darkened when exposed to light. Damage to the base of the brain, on the other hand, abolished the response, leading von Frisch to propose the existence of nonvisual photoreceptors in the deep brain.

Many animal species possess such light-sensing cells. Sparrows, for example, can tune their circadian clocks even when deprived of their eyes, as shown in the early 1970s by Michael Menaker, then at the University of Texas at Austin. Follow-up experiments showed that the birds have light-sensing cells in their brain. It turns out that a surprising amount of light can penetrate the feathers, skin and skull of a bird to activate these cells.

The possibility that at least some mammals might also have light receptors not involved in vision first drew the attention of biologists when Keeler reported on his home-bred mice in the 1920s. Because the anatomy of the mammalian retina was so well understood, the assumption was that the missing light-sensing organ must be located somewhere other than in the eyes. But by the early 1980s studies on eyeless rodents by Randy J. Nelson and Irving Zucker, both then at the University of California, Berkeley, seemed to call this hypothesis into question. Those animals were unable to adjust their circadian rhythms to the cycle of night and day, suggesting that the light-sensing receptors had to reside within the eye.

Menaker, who meanwhile had moved to the University of Oregon, set out to investigate whether mouse eyes had a role in light-sensitive responses that do not require the formation of images. He and two of his graduate students, Joseph Takahashi and David Hudson, looked at mutant mice that lacked functional rods and cones, except perhaps for a few minimally active cones. To the researchers’ surprise, these blind mice could restrict their activity to the nighttime and remain relatively inactive during the daytime, just as fully sighted mice do.

One possible explanation for this behavior was that the few sickly surviving cones were somehow able to maintain nonvisual responses to light. But in 1999 a team led by Russell Foster, then at Imperial College London, used mutant mice completely lacking rods and cones to show that these cells were not necessary for nonvisual responses to light. This finding left only one ex­planation: the eye must contain a yet to be discovered type of photoreceptor.

This was a heretical proposition. The cells in the retina involved in the formation of images had been known since the mid-1800s. The notion that another light-sensitive cell in the retina had been overlooked for almost 150 years seemed absurd.

Promoting Heresy
And yet research Mark D. Rollag and I began in the mid-1990s at the Uniformed Services University of the Health Sciences eventually helped to prove Foster right. Rollag was interested in a different form of nonvisual photoreception: amphibian camouflage. Pigmented cells in the tails of tadpoles darken under light, an adaptive response that helps to conceal the animal when it is exposed. These cells, called dermal melanophores, maintain their response even when removed from the animal and cultured in a dish. Rollag and I identified a novel protein in the cultured cells that is strikingly similar in composition to the class of protein pigments called opsins, which enable rods and cones to detect light. We named the new protein melanopsin.

The similarity with the known opsins strongly suggested that melanopsin was the molecule that triggered the darkening response. Wondering if melanopsin also played a role in other light-detecting cells, we searched for it in other frog tissues known to be directly light-sensitive—such as particular areas of the brain and the iris and retina of the eye. As it turned out, neither rods nor cones contained this new light-sensitive protein. But, to our surprise, it did turn up in retinal neurons called retinal ganglion cells that were not previously believed to be light-sensitive.

The vertebrate retina is an elegant three-layered structure. The deepest layer contains the rods and cones, so light must travel through the other layers before it is detected for vision. Information from the rods and cones is then transferred to the middle layer, where it is processed by several different classes of cells. Finally, these cells communicate the processed signal to the surface layer, which is primarily composed of ganglion cells. Long, signal-conveying axons extend from these ganglion cells to carry information through the optic nerve and to the brain.

In 2000 my colleagues and I found the first hints that a very small fraction of these ganglion cells were directly sensitive to light. We then discovered that 2 percent of mouse retinal ganglion cells contain melanopsin and that a small percentage of these cells in humans also contain it. In 2002 experiments by David M. Berson and his colleagues at Brown University confirmed our view. They incapacitated the rods and cones and filled the melanopsin-containing ganglion cells with a dye. Next, they removed the retinas from the eyes of the mice and showed that the stained nerve cells fired when exposed to light. Given that the rods and cones were disabled, the response meant that, beyond relaying signals from rods and cones, these particular ganglion cells were able to detect light on their own.

The hypothesis garnered support from evidence found in 2002 by other teams. Samer Hattar of Johns Hopkins University and his co-workers showed that some axons from the mouse retina connect to the suprachiasmatic nucleus—the area of the brain that regulates the body’s internal clock—whereas others connect to the area of the brain that controls the constriction of the pupils. And the ganglion cells connected to those areas are precisely the ones that contain melanopsin. All these discoveries pointed to the same solution to our riddle: photosensitive ganglion cells would enable mice with no functioning rods and cones to constrict their pupils and to keep their bodies in tune with the light/dark cycle. But eyeless mice, which lacked retinas altogether, would lose those abilities.

One additional test was left to seal the case. I and others thought that if we bred mice that were normal except for lacking the gene for melanopsin, the mice, being unable to produce the pigment, would have no nonvisual responses to light. What happened next confirmed a favorite mantra in our laboratory: “Science is a cruel mistress.” Just when we thought we were about to nail down the answer to our mystery, we were absolutely dumbfounded to find that the melanopsin-free mice had little trouble adjusting their circadian clocks.

One Last Hurdle
To explain this setback, we considered the possibility that perhaps yet another nonvisual photoreceptor could be lurking in the retina. But this possibility seemed unlikely for a variety of reasons. Most significant, the complete mouse genome, which was sequenced around the time we completed the studies on our knockout mice, contained no other obvious photopigment genes.

The second hypothesis was that perhaps rods, cones and photosensitive ganglion cells acted together to control nonvisual responses to light. This last possibility was put to the test when we engineered mice that completely lacked rods, cones and melan­opsin. These “frankenmice” failed to show any visual or non­visual responses to light and behaved as though their eyes had been surgically removed. Finally, we were able to conclude that rods, cones and the melanopsin-containing ganglion cells all work together to bring nonvisual light information to the brain.

In fact, evidence is emerging that photosensitive ganglion cells also function as a conduit for transmitting nonvisual light information from the rods and cones to the brain, just like the other retinal ganglion cells transmit visual information to the visual areas of the brain. In 2008 three different groups, including ours, each devised a method to kill photosensitive ganglion cells in mice without affecting the rest of the organism. Although the mice retained their vision, they tended to get their days and nights mixed up and also had trouble constricting their pupils. In other words, the specialized ganglion cells are necessary to engender nonvisual responses to light, but the system has some built-in redundancy: these cells can either detect light autonomously or relay information from the rods and cones, or both.

So the puzzle was finally solved—at least as far as mice were concerned. But evidence has emerged that the same physiological mechanism may exist in humans, too. Foster and his collaborators published a study in 2007 of two blind patients who lacked functional rods and cones—the human equivalent of Keeler’s mice—but who could still adjust their circadian rhythms when periodically exposed to blue light. The wavelengths of blue light where their response was optimal were precisely in the same range that melanopsin can detect—as measured in studies by my group in collaboration with Berson’s in which we forced normally nonphotoreceptive cell lines to produce melanopsin. Those cells responded to light by firing in response to blue light.

Perhaps more interesting, we found that when struck by light, melanopsin initiates a chemical signaling cascade inside these cells that more closely resembles what happens in fly and squid photoreceptors than in mammals’ rods and cones. Again, this was not completely unexpected, because we had recognized years earlier that the gene sequence for melanopsin more closely resembled the gene sequences of photopigments in invertebrates than those in vertebrates. Thus, in mammals, melanopsin appears to be the photopigment of a previously unknown and primitive nonvisual photoreceptive system, one housed within the retina alongside its more “advanced” cousin, the visual system.

Aside from pure scientific interest, the discovery of this new, hidden “organ” may have clinical implications as well, because it points to a previously unappreciated link between eye health and mental health. Studies suggest that exposure to blue light may increase awareness, counteracting jet lag or sleep deprivation, and alleviate seasonal affective disorder—a common problem at high latitudes that can cause debilitating depression and may induce suicide. It seems natural to assume that light therapy is effective because it targets photosensitive ganglion cells. Other studies have shown that blind children suffering from diseases that affect retinal ganglion cells, such as glaucoma, seem to be at higher risk of suffering from sleep disorders than children who are blind for other reasons. Targeting the health of photosensitive ganglion cells could thus lead to a new class of treatments for a wide variety of conditions.