Figure 1. http://www.metmuseum.org/collection/the-collection-online/search/437125. This link will lead to one of the most popular paintings from the Morning on the Seine series.

In the Fall of 2000, neuroscientists at UTHealth were among the first to hear about an exciting new development in vision research. We had invited Dr. David Berson from Brown University to give a seminar about his research on the anatomy and physiology retinal ganglion cells, the neurons that convey information from the retina to the brain. Ganglion cell bodies are located along the inner margin of the retina, and their axons exit the eye as the optic nerve. Ganglion cell dendrites extend into the inner plexiform layer, a neuropil where they receive excitatory inputs from the photoreceptors via bipolar cells and inhibitory inputs from amacrine cells. There are about twenty different types of retinal ganglion cells, and they differ in their morphology, light responses and their targets in the brain. David’s seminar was very interesting to vision researchers here because we were doing very similar work, ourselves, but we were unprepared for the revelation at the end of David’s talk. He had found that one type of mouse retinal ganglion cell expressed a photopigment and generated responses to light without any synaptic input from other retinal neurons. The responses mediated by the photopigment were relatively insensitive and had slow kinetics, beginning after a long delay and out lasting the stimulus by minutes. The intrinsically-photosensitive ganglion cells projected to the hypothalamus, where they provided the visual input to synchronize the master circadian clock in the brain with the daily changes in illumination [1]. In 2002, David and his collaborators published the results in the prestigious journal Science, and that paper was later selected as one of the runners-up in their contest for “breakthrough of the year”.

In retrospect, we should not have been surprised. Dr. Ignacio Provencio, now at the University of Virginia, had isolated the photopigment called melanopsin from the light-sensitive melanophores in toad skin, and in 2000, he had localized it to a subset of retinal ganglion cells in rodent and primate retinas. Based on their large size and sparse distribution, he suggested that these ganglion cells projected to the hypothalamus. This provided an explanation for the ability of otherwise blind humans and other animals to entrain their circadian clocks to light even though all their rods and cones had degenerated [2]. Previously, visual scientists had argued that a few surviving conventional photoreceptors accounted for this effect, despite the evidence suggesting that it was mediated by a different photopigment than those found in the rods and cones. The intrinsically-photosensitive ganglion cells project to the pretectum, where they provide input to the neurons that control the diameter of the pupil, and also to other nuclei in the brainstem. Initially, most of the research on intrinsically-photosensitive ganglion cells dealt with their roles in visual functions that do not require formation of images, and this remains a very active area of investigation. But the focus of this essay is on two newer sets of results indicating that intrinsically-photosensitive ganglion cells project to targets within the retina and contribute to conscious perception of light and image-forming vision in humans and other mammals.

The first study to suggest that signals from intrinsically-photosensitive ganglion cells contribute to conscious visual perception was done by Dr. Dennis Dacey of the University of Washington and his collaborators. They showed that the intrinsically-photosensitive ganglion cells of monkeys also project to the dorsal lateral geniculate nucleus of the thalamus, a relay for signals to the primary visual cortex. These ganglion cells depolarized and fired action potentials in response to increments in dim stimuli, indicating that they received input from rods. At higher light intensities, the polarity of their responses depended on the wavelength of the stimulus. The ganglion cells were inhibited by short wavelength, blue, stimuli and excited by long and middle wavelength, yellow, stimuli, findings suggesting that they contributed to color perception. In collaboration with the Dacey Laboratory, we showed that intrinsically-photosensitive ganglion cells of macaque retina received inputs from bipolar cells and amacrine cells that would account for these responses. There was also a slow, sustained component of the responses to bright light mediated by melanopsin. In humans and other primates intrinsically-photosensitive ganglion cells comprise a small fraction of the total, but because their dendritic fields are very large, they cover the entire retina [3].

A recent study of human color perception by Dr. Brian Wandell and his colleagues at Stanford University provided direct evidence for a contribution by melanopsin in the peripheral retina, but not in the center of the fovea, where visual acuity is the greatest. In other words, in addition to the red, green and blue photopigments of the cones, a fourth, turquoise pigment expressed by the intrinsically-photosensitive pigment is required to account for the sensitivity of human vision to bright lights in the periphery. A photopigment with these characteristics also mediates some of the effects of light adaptation on the human cone electroretinogram, a field potential recorded from the eye. This result indicates that some of the effects of intrinsically-photosensitive ganglion cells are mediated within the retina [4].

Previously, mammalian retinal ganglion cells were thought to make synapses only in the brain, but it now appears that intrinsically-photosensitive ganglion cells are presynaptic within the retina. There is now considerable, but indirect, evidence that their targets include dopaminergic amacrine cells. The retina is able to work over an enormous range of ambient light intensities, and to accomplish this, the neural circuits have to change in subtle ways as the illumination changes. Dopamine plays an important role in this process, changing the strength of chemical and electrical synapses on virtually every type of retinal neuron. Until recently, it was unclear how the dopaminergic amacrine cells could detect the absolute intensity of the ambient light because the rods and cones adapt, themselves.

We now know that melanopsin enables the intrinsically-photosensitive ganglion cells to transmit information about the absolute intensity of light stimuli. The underlying physiology has been studied most extensively in mouse retina, where stimulation of intrinsically-photosensitive ganglion cells has the same effects. Dr. Douglas McMahon and his colleagues at Vanderbilt University have shown that intrinsically-photosensitive ganglion cells release the excitatory neurotransmitter glutamate and generate slow, sustained responses to light in a subset of dopaminergic neurons [5]. Dr. Dacey and his colleagues have shown that, in both rodents and primates, there are branches from the axons of the intrinsically-photosensitive ganglion cells that run in the same layers of the retina as the dendrites of the dopaminergic cells3. In primate retinas, contacts between intrinsically-photosensitive ganglion cells and dendrites of dopaminergic cells have been observed. The synapses made by the axons of intrinsically-photosensitive ganglion cells have never been observed in the electron microscope, however, and we are beginning a series of experiments to accomplish that.

Last January, during a break from my reading and experiments, my family and I went to the Museum of Fine Arts, Houston to see their exhibition “Monet and the Seine: Impressions of a River.” After seeing a series of Monet’s “Mornings on the Seine” paintings, I realized that they might provide some primary data about melanopsin’s role in visual perception. Monet had a houseboat-studio anchored on a tributary of the Seine near his home in Giverny. He woke up at 3:30 AM, well before sunrise, and rowed there to paint exactly the same scene every morning. He worked on several canvases at once, each representing what he saw at different times of the day, pre-dawn, sunrise, early morning and late morning, and under a variety of weather conditions. The canvases were numbered, and he had an assistant put them up on the easel, in sequence, as he went along. Monet was very meticulous, taking the years 1896 and 1897 to complete a series of fifteen magnificent paintings, among others with different subjects. There are many different interpretations of what Monet was trying to accomplish with the “Mornings on the Seine” series, and they are not mutually exclusive. For example, he may simply have been trying to make a living. Many more French people were decorating their homes with oil paintings at that time, and landscapes were particularly popular. The exhibition including these paintings was very well-received by critics and financially successful [6].

Even though the paintings are somewhat abstract, I believe that Monet was systematically exploring the effects of light on the appearance of natural scenes and conveying what he really saw at different times of the day. In the paintings done earliest in the morning, Monet’s rods and cones would have both been active. This would account for the predominance of blues and purples in those paintings and for the absence of fine details. After sunrise, the rods would saturate, and the cones would predominate, and this would explain why more reds and oranges appear later in the series. But the cones would quickly adapt to the ambient light intensity, and Monet would have barely noticed that the absolute intensity of the light had changed. If rods and cones were the only photoreceptors, there should have been no further changes in the appearance of the scene as the day progressed, except for the direction of the light and the appearance of the shadows. However, there is a dramatic change in the paintings from later in the morning. Bright greens tend to predominate, and many more details are visible, particularly small movements such as ripples in the water. The color of sunlight changes as the angle between the sun and the horizon increases, and the wind tends to come up later in the day, and these would all be visible using rods and cones. But I believe that some of the other changes Monet saw were attributable to an increase in the melanopsin-mediated component in the signals of his intrinsically-photosensitive ganglion cells.

The continuous, rapid firing of the intrinsically-photosensitve ganglion cells would generate a sustained release of dopamine, and this would account for the changes in the color palette, long after the rods had saturated. It might also explain why fine details and subtle movements in the periphery became more apparent as the morning progressed. Dr. Robert Lucas from Manchester University and his colleagues have shown that, in mice, the visual system becomes more sensitive to movements and fine details when melanopsin is activated, and the same may be true in humans [7]. It should take an hour or two after sunrise for the effects to fully develop, and they might depend on the weather. The melanopsin pigment is relatively insensitive to light, and bright, mid-morning light on a clear day would be required to activate it maximally. The effects of the intrinsically-photosensitive ganglion cells on the dopamine cells might also take many minutes to develop if they are mediated by intracellular signaling pathways, and this is known to be true for the changes in retinal synapses mediated by dopamine.

There have been many reviews that emphasize the contributions of melanopsin to visual functions that do not require image formation, such as resetting the brain’s endogenous circadian clock and controlling the diameter of the pupil. Studies of the role of melanopsin in human visual perception are just beginning, however, and a lot more work will need to be done to understand how much it affects our vision. Does a fourth visual pigment in the peripheral retina influence our perception of color? Do the intrinsically-photosensitive ganglion cells contribute to the perception of form and movement? To answer these questions we will need to design modern versions of Monet’s experiments in 1897.

References:

1. Phototransduction by retinal ganglion cells that set the circadian clock. Berson DM, Dunn FA, Takao M. Science. 2002 Feb 8;295(5557):1070-3.
2. A novel human opsin in the inner retina. Provencio I, Rodriguez IR, Jiang G, Hayes WP, Moreira EF, Rollag MD. J Neurosci. 2000 Jan 15;20(2):600-5.
3. Melanopsin-expressing ganglion cells in primate retina signal colour and irradiance and project to the LGN. Dacey DM, Liao HW, Peterson BB, Robinson FR, Smith VC, Pokorny J, Yau KW, Gamlin PD. Nature. 2005 Feb 17;433(7027):749-54.
4. Human trichromacy revisited. Horiguchi H, Winawer J, Dougherty RF, Wandell BA. Proc Natl Acad Sci U S A. 2013 Jan 15;110(3):E260-9.
5. Melanopsin mediates retrograde visual signaling in the retina. Zhang DQ, Belenky MA, Sollars PJ, Pickard GE, McMahon DG. PLoS One. 2012;7(8):e42647.
6. Monet and the Seine: Impressions of a River. Aurisch HK, Paul T. Yale University press, New Haven.
7. Melanopsin-driven light adaptation in mouse vision. Allen AE, Storchi R, Martial FP, Petersen RS, Montemurro MA, Brown TM, Lucas RJ. Curr Biol. 2014 Nov 3;24(21):2481-90.

Download a copy of Monet and Melanopsin.

Bio: Dr. Marshak first became interested in color vision as an undergraduate majoring in anthropology at Cornell University, and also studied the visual system at the University of California Los Angeles, where he received a Ph.D. in anatomy.  He continued his training at Harvard University and then joined the faculty of the University of Texas Health Science Center at Houston. He has been there for over thirty years, teaching anatomy to the first-year medical students and various graduate courses while doing research on neural circuits in the primate retina.  He and his wife, Patricia Ann Ward, have two sons in their twenties. He is an avid reader, but this is his first attempt to publish anything other than a scientific paper.