Circadian Photoreception in Humans and Mice

Introduction

Circadian rhythms are oscillations in the behavior and biochemical reactions of organisms, and occur with a periodicity of approximately twenty-four hours (cf Latin circa , or “about,” and dies, or “day”). A daily biological rhythm is widespread among organisms but not universal. Thus, humans, mice, and model organisms such as Drosophila and some cyanobacteria exhibit circadian behavior, whereas other model organisms, such as E. coli , S. cerevisiae, and S. pombe do not (1–3) . Three important properties of the circadian rhythm are: i) its innate nature, that is, the intrinsic rhythm that exists without any sensory input from the environment; ii) its capacity for temperature compensation, that is, the maintenance of the intrinsic period, phase, and amplitude of the rhythm despite external fluctuations in temperature, so long as these fluctuations do not interfere with physiological thermoregulation; and iii) its photoentrainment, that is, the synchronization of the phases of the rhythm with the external light-dark cycles of the solar day.

Circadian rhythms are thought to confer a selective advantage to organisms by enabling them to pursue levels of activity that are optimal for growth and development, and that minimize, by establishing favorable time/space niches, susceptibility to predation and competition (4) . This point is illustrated in the idealized representation of the circadian rhythms of man and mouse in Figure 1⇓. Humans are diurnal organisms (i.e., active during daytime), whereas mice are nocturnal (i.e., active during the nighttime); the two species can thus coexist within the same physical habitat while minimizing direct interaction (5) . Recently, important progress has been made in understanding the nature of the light receptor that entrains the circadian clock and ensures that humans remain diurnal and that mice remain nocturnal (6–8) . This review will focus on the circadian photoreceptors that regulate circadian rhythms in mammals.

View larger version:

• In this window
• In a new window

Figure 1.

Circadian rhythms in mouse and human. The figure shows daily oscillation of physiological and behavioral variables in response to an idealized cycle of twelve hours of dark followed by twelve hours of light (LD12:12). The black bar at the top shows the dark phase at 1800 to 0600. Both nocturnal and diurnal organisms secrete melatonin during the nighttime. Red and blue bars indicate locomotor activity in mouse and human, respectively. Both organisms are placed under constant darkness (DD; indicated by arrow) at end of day 4. Under constant darkness, the locomotor activity of mice indicates an intrinsic period, or free-running period, of 23.7 hours. In human subjects, the intrinsic period under DD conditions is 25.1 hours; a recent report suggests that in humans the periods is 24.3 hours. From reference (5) .

The Circadian System in Mammals

Traditionally, the circadian system has been conceptualized in terms of three components: an input component, a clockwork component, and an output component. Recent studies, however, have shown that there are considerable overlaps between the three components at both macroscopic and microscopic levels, so that the three components may be organized into multiple interconnecting pathways (9) . In mammals, the input component is mediated at the macroscopic level by the visual perception of light; the master circadian clock is located in the hypothalamus as a pair of neuron clusters known as the suprachiasmatic nuclei (SCN); and the output component is elaborated, particularly in the form of prokineticin (PK2) and perhaps also transforming growth factor-α (TGF-α ) (10, 11) , from the SCN, which may thereby engage, through neural connections, other regions of the brain. Biological functions influenced by the circadian clock in humans include sleep-wake cycles, neuroendocrine levels, mental alertness, physical strength, body temperature, blood pressure, and blood viscosity (12) . The phases of zeniths and nadirs of these functions can be affected by many factors, including feeding, emotional stress, and ambient temperature. Light, however, is the most dominant cue, or Zeitgeber , that synchronizes the circadian phases to the environment (cf German Zeit or “time,” and Geber , or “giver”) (Figure 2⇓).

View larger version:

• In this window
• In a new window

Figure 2:

Light reception by eye and signal transduction to SCN. Rods and cones, photoreceptor cells located in the inner retina, mediate the perception of light. There are other light-sensitive cells located in the inner nuclear layer (INL) and ganglion cell layer (GCL) that are involved in signal transduction to the SCN via the retinohypothalamic tract (RHT). The cryptochromes CRY1 and CRY2 are expressed in the INL and GCL, and melanopsin is expressed in a small fraction of ganglion cells. In this representation, we show only the axons that transmit light signals from inner retina and outer retina to the SCN. The ONL and INL are partially redundant for signaling to the SCN. Similarly, CRY1, CRY2, and melanopsin (and possibly other unknown photopigments expressed in ganglion cells) may be functionally redundant in the inner retina. The visual pathway, initiated from the outer retina, is omitted for simplicity.

At the molecular level, the clockwork of the SCN involves several proteins that participate in positive and negative transcriptional feedback loops. BMAL1 (for b rain and m uscle a ryl hydrocarbon receptor nuclear translocator (A RNT)-l ike protein 1 ) and CLOCK are transcription factors that contain two basic helix-loop-helix domains and bind E-box elements (CACGTG) in the Period and Cryptochrome clock genes and thereby effect a positive feedback loop of circadian rhythm regulation (13–15) . The mammalian PERIOD proteins (PER1 and PER2; collectively designated PER) and CRYPTOCHROME proteins (CRY1 and CRY2; collectively designated CRY) act as negative regulators of transcription driven by the BMAL1/CLOCK heterodimer (16–19) (Figure 3⇓). PER and CRY form heterodimers that interact with casein kinase Iε (CKIε ) and then translocate into nucleus where CRY acts as a negative regulator of BMAL1/CLOCK–driven transcription (20) . This basic molecular clock is present in all tissues, but the peripheral clocks are synchronized to the master clock of the SCN through humoral and neural signals.

View larger version:

• In this window
• In a new window

Figure 3:

Molecular model of mammalian circadian clock. The clock is made up of positive and negative transcriptional regulators. The CLOCK–BMAL1 heterodimer activates transcription of the Per and Cry1 genes. The PER proteins interact, at their C-terminal domains, with the CRY proteins; central sequences of the PER proteins interact CKIε , and the resulting ternary complexes translocate into the nucleus, where they negatively regulate the transcription of Per and Cry genes (indicated by “oscillating” transcript). The PER and CRY proteins also positively regulate transcription of Bmal1 (62). A recent study suggests that all clock genes form a “supercomplex,” or “timesome,” on the Per promoter, with the PER proteins acting as a scaffold for assembly (63) . Adapted from (3) .

Circadian Photoreception

In some vertebrates, such as birds and reptiles, photoreceptor molecules for resetting the circadian clock are located in multiple structures: the eyes, the pineal gland, and mid brain (i.e., deep-brain photoreceptors). In iguana, in addition to the eyes, a photosensory organ on the top of the head (i.e., the parietal eye) acts to synchronize the clock with the environmental light-dark cycle. In Drosophila and zebrafish, virtually every cell is photosensitive and capable of synchronizing the cell-autonomous circadian clock with external light-dark cycles. In humans and other mammals, such as mice, however, the eye is the only organ that functions as a light receptor for the circadian clock. Removal of the eyes, or severing the optic nerve, causes both visual and circadian blindness in the human and the mouse (12) . Similarly, mice genetically lacking optic nerves cannot entrain their biorhythms to light cycles (21) .

Although the mammalian eye plays a central role in both visual and circadian photoreception, subsequent processing of photic input is handled by distinct photoreceptive systems. Visual information is processed by the visual cortex (which may constitute as much as thirty percent of total cortex), whereas photic input into the circadian system is transmitted to the SCN in the hypothalamus. In addition, the photosensory pigments (i.e., rhodopsin and color opsins) located in the rods and cones of the outer retina are, remarkably, irrelevant to circadian photoentrainment (Figure 2⇑). Humans who lose all rods and cones at advanced stages of inherited retinal degeneration syndromes (e.g., retinitis pigmentosa) have no conscious sense of light; nevertheless, many such individuals respond normally to environmental light-dark cycles, as determined according to a variety of criteria, with perfect circadian photoreception and phototransduction (22, 23) . Similarly, three-month-old mice with the retinal degeneration (rd ) mutation, which inactivates the rod-specific cGMP phosphodiesterase (24) and thus blocks rod-mediated phototransduction, lose all rods (as well as virtually all cones due to secondary degeneration) and are visually blind according to both behavioral and by electroretinographic tests; however, these animals exhibit normal to moderately reduced circadian photoreception (25) . Histological analysis of the retina of these animals shows that the outer retina is completely replaced by scar tissue, with minimal cone cell bodies, whereas the inner retina, comprising amacrine cells, interneurons, and ganglion cells, is intact. Extensive studies of rd mice and genetically engineered rodless and coneless mice lend very strong evidence to the interpretation that pigment(s) located in the inner retina are the photoreceptors for circadian photoreception (26) . Thus, the outer retina is used for vision, and the inner retina is used for circadian photoreception.

Two circadian photopigments in the inner retina have been proposed. One is melanopsin, which is presumed, like all other opsins, to employ retinal as a cofactor (27) . The other is CRY, a flavin- and folate-containing pigment (28) . We have proposed that, in addition to the duality of vision and circadian photoreception that exists at the anatomical and histological levels, there is a duality at the molecular level in that vitamin A–based opsins are mainly used for vision, whereas vitamin B2–based cryptochromes are exclusively used for circadian photoreception (29) . The evidence for this model is discussed in the following sections.

Candidate Circadian Photoreceptors

Melanopsin and cryptochromes are the only known photopigments in the inner retina of mice and humans, and are hence the only candidates for the circadian photoreceptors, although the formal possibility of other photoreceptors in the inner retina has not been eliminated. Below, we briefly discuss the properties of melanopsin and cryptochrome.

Melanopsin

Melanopsin, first discovered in the dermal melanophores of Xenopus, is known to be directly photosensitive (30) . In addition to melanophores, melanopsin was also found in the Xenopus retina, pineal gland, and hypothalamus and in the retina of mice, humans, and other vertebrate species. Curiously, melanopsin has higher sequence homology to nonvertebrate than to vertebrate opsins (27, 30) . Several features of melanopsin make it a strong candidate for being a circadian photoreceptor. First, it is expressed only in a subset of cells in the ganglion cell layer (GCL) (27, 30) . Second, these cells, as revealed by patch-clamp analysis, are intrinsically sensitive to light (31) . Third, the action spectrum (i.e., action potential responsiveness as a function of wavelength) of these ganglion cells resembles that of opsin (with a maximum at about 500 nm), and is typical of the circadian clock of the mouse (31) . Finally, these ganglion cells are unique in containing the neuropeptide PACAP (for p iturity a denylate c yclase a ctivating p eptide), and they possess processes that project into the SCN. Their projections also extend into brain regions involved in nonvisual photoresponses (e.g., pupillary constriction and suppression of locomotor behavior) that are maintained in rodless and coneless mice (32–34) . These findings lead to the conclusion that melanopsin is a circadian photoreceptor in mammals.

Cryptochrome

CRY is evolutionarily related to the light-dependent DNA repair enzyme photolyase (5, 35–37) . Photolyases are enzymes of 55–65 kDa and contain two noncovalent cofactors, flavin adenine dinucleotide (FAD) and (usually) methenyltetrahydrofolate (MTHF)[or in some rare cases, 8-hydroxy-5-deazariboflavin; (38) ]. Photolyase repairs UV-induced cyclobutane pyrimidine dimers in the following manner (38) : The enzyme binds to the pyrimidine dimer in a light-independent manner; catalysis is initiated when the folate (or deazariboflavin) cofactor absorbs a blue-light (375–450nm) photon and transfers the excitation energy to the catalytic FAD cofactor; the excited, reduced FAD molecule donates an electron to the thymine-dimer, thereby splitting it into two thymines; concomitant electron transfer back to the flavin restores catalytic functionality; and then enzyme and repaired DNA dissociate. In plants, a protein with about thirty percent sequence identity to the E. coli photolyase was found to be a blue-light photoreceptor that regulates growth and development; that this protein photoreceptor had been long sought after by plant biologists is reflected in its name, “cryptochrome” (5, 39, 40) .

Photolyase is found in many organisms but is lacking in many plant, animal, and microbial species (38) . Significantly, photolyase is absent in humans, mice, and other placental mammals (41) . Nevertheless, the human genome project uncovered a putative protein with sequence silmilarity to photolyase (42, 43), and we subsequently discovered a second such human sequence (28) . We found that proteins encoded by the two sequences contain, as does photolyase, FAD and folate cofactors, but are devoid of photolyase activity (28) ; therefore, we concluded that these proteins might function as circadian photoreceptors in a manner analogous to the cryptochromes of Arabidopsis (28, 37, 44) . We have designated the two mammalian proteins CRYPTOCHROME 1 and 2 (CRY1 and CRY2).

In addition to an approximately 500-residue photolyase-like domain, CRY1 and CRY2 contain a C-terminal extension of 100–200 amino acids that are thought to be essential to effector function (45) . In contrast to the opsins, CRY1 and CRY2 are not restricted to photosensitive organs or neural tissues, but rather are expressed in many mouse and human tissues (29) ; nevertheless, the following findings are taken as evidence that the two proteins are circadian photoreceptors in mammals: First, CRY1 and CRY2 can be assigned, on the basis of sequence similarity, to a family of blue-light photoreceptors (e.g., photolyase and plant cryptochromes) with photoactivated functions. Second, CRY1 and CRY2 are expressed at high levels in tissues that are known to contain the circadian photoreceptor (i.e., the ganglion cells and inner nuclear layer of the inner retina) (29) . Third, nutritional studies in Neurospora and Drosophila indicate that a vitamin A–independent pigment might be the circadian photoreceptor in those organisms (5) , and recent studies show that the major circadian photoreceptor in Drosophila is a cryptochrome (46) . Finally, the mammalian gene that encodes CRY1 (Cry1) is highly expressed in the mouse SCN (29) , and significantly, its expression exhibits circadian oscillation, indicating that is might also have a light-independent function as well.

Genetic Analysis of Circadian Photoreception

To analyze the roles of cryptochromes, classical opsins, and melanopsin, mice lacking one or more of these pigments have been generated by knockout technology and tested for circadian photoresponses. The results of these studies indicate that circadian photoreception depends primarily on cryptochromes, with some dependence on opsin-based pigments as well.

Cryptochrome Knockout Mice

In the absence of light, animals should manifest SCN that lack photic gene induction; a light pulse given during the dark phase, moreover, should fail to reset the animal’s circadian clock. Indeed, knockout studies have indicated that mice lacking either the Cry1 or Cry2 gene are significantly impaired with regard to both the induction of clock genes and circadian rhythmicity. In mice lacking both Cry1 and Cry2 , the induction of both the c-fos gene and the Per1 gene was drastically reduced (but not abolished); Per2 induction was essentially normal (18, 19, 47) . With saturating light pulses lasting for six hours in the dark phase, phase shifts in the single mutants were drastically enhanced, indicating abnormal photosensitivity of the clock. Hence, these results are consistent with a model in which cryptochromes act as circadian photoreceptors; however, these studies also reveal that cryptochromes cannot be the sole photoreceptors for the circadian clock.

Behavioral analysis of mutant mice reveals that the CRY proteins play a variety of roles in the circadian clock, both as photoreceptors and as components of the transcriptional system that engenders the molecular clock. The Cry1^-/- and Cry2^-/- null mutants have, respectively, a shortened and lengthened free-running period relative to wild-type mice and thus confirm that the CRY proteins participate in the biological clock (18, 19, 47) . Most importantly, the Cry1^-/-Cry2^-/- double mutant is arrhythmic under constant conditions, thereby underscoring a central role for the CRY proteins in the clock (18, 19) . Under light-dark (LD) cycles, Cry1^-/-Cry2^-/- animals do exhibit rhythmicity in that they are active only in the dark phase, most likely reflecting the so-called “masking” effect of light, whereby activity is suppressed by light in a clock-independent manner, even in SCN-lesioned animals. In accordance with their arrhythmic behavior, Cry1^-/-Cry2^-/- mice express the clock genes Per1 and Per2 constitutively at high levels (19) indicating that CRYs negatively regulate Per gene expression in all tissues independent of their photoreceptor function in the retina. Thus, photic gene induction in the SCN is consistent with cryptochromes being the major circadian photoreceptors and show that cryptochromes are core components of the molecular clock. In addition, there are non-cryptochrome photoreceptors transducing photic signals to the SCN. To explain the latter finding, the following possibilities were considered: 1) a non-classical opsin such as melanopsin expressed in the inner retina is a circadian photoreceptor. 2) A non-opsin and non-cryptochrome pigment in the inner retina is a circadian photoreceptor. 3) Both cryptochromes and classical opsins function as circadian photoreceptors with cryptochromes playing the predominant role; in the absence of cryptochromes, RHT phototransduction is drastically reduced, whereas loss of classical opsins affects it only marginally or not at all. 4) Finally, it is possible that CRY and classical opsins in the outer retina, and novel opsins in the inner retina, all contribute to circadian photoreception to varying degrees. The analysis of other photoreceptor mutants clarifies the issue. (See below.)

Melanopsin Knockout Mice

Because of its expression in nearly all ganglion cells that are photosensitive and that have connections to the SCN and other non-visual photoresponse centers, melanopsin was seriously considered as a possible circadian photorecepor. Melanopsin knockout mice have recently been generated to analyze this hypothesis (48, 49) . The animals entrain normally to LD cycles, show phase shifting in response to short light pulses, and manifest normal photic induction of clock genes in the SCN. It appears, however, that under dim light, the magnitude of phase shifts is moderately reduced relative to wild-type animals. Assuming that this effect is not due to differences between mouse strain backgrounds, it appears that melanopsin either directly or indirectly plays a minor role in circadian photoreception.

3-Blind Mice: rd/rd Cry1^-/-Cry2^-/-

Because mice devoid of CRY1 and CRY2 retain significant retinohypothalamic tract (RHT) phototransduction, and given that melanopsin makes only a minor contribution to this pathway, the possibility must be considered that rods and cones might be the source of the residual photoreception in Cry1^-/-Cry2^-/- mice. When rd/rd Cry1^-/-Cry2^-/- (i.e., “triply blind”) mice were generated (50) , the resulting animals were about 100-fold less sensitive, as measured by photic gene induction in the SCN, to circadian entrainment (Figure 4⇓). Behaviorally, most of these animals become arrhythmic under cycles of alternating twelve-hour–light and –dark periods (LD12:12). The simplest explanation of these data is that CRY1 and CRY2 are the major circadian photoreceptors and that classical opsins play a back-up role. The residual photic sensitivity in rd/rd Cry1^-/-Cry2^-/- mice could arise from cones that survive the rd mutation or to melanopsin, and possibly to unknown opsins or even a non-opsin non-cryptochrome pigment located in the inner retina. In any case, the major finding that emerges from the triple mutant mice is that CRY and classical opsins play redundant roles in circadian photoreception. In these mice, the lack of CRY affects another non-visual photoresponse: Light-induced pupillary constriction, which is presumed to be in part mediated by the same photopigment that mediates circadian photoreception, is decreased over tenfold relative to rd mice (51) ; this observation provies further evidence for the photoreceptive function of the CRY proteins.

View larger version:

• In this window
• In a new window

Figure 4:

The induction of c-fos in the SCN as a function of genotype and irradiance. Mice were exposed to light pulses of indicated intensity during the night phase, and the level of c-fos induction was determined by in situ hybridization. Each light intensity induced c-fos in both wild-type and rd mutant mice, whereas Cry1^-/-Cry2^-/- significantly reduced c-fos induction under both dim and bright light. Significantly, c-fos induction was severely reduced in rd/rdCry1^-/-Cry2^-/- mice under all light conditions. From reference (50) .

Mice Devoid of the Retinol Binding Protein

Because there are residual circadian photoresponses in rd/rd Cry1^-/-Cry2^-/- mice, and because melanopsin does not seem to be a major contributor to circadian photoreception, other opsin-based pigments in the inner retina (as mentioned above) might be responsible for some aspect of photoreception. To address this possibility, mice that lack the retinol binding protein (RBP) were generated (52) and placed on a vitamin A–free diet (53) . The RBP transports retinol from the liver to the retina and other peripheral tissues. Mice with knockout of the RBP gene cannot mobilize retinol to the eye and consequently become blind within four months. When placed on a vitamin A–free diet after ten months, the majority of the animals lack any detectable retinal in the eye (<0.2% of wild-type level). Nevertheless, these animals exhibit normal (or marginally diminished) light-induced expression of the clock genes Per1 and Per2 in the SCN (53) . To investigate the molecular basis for these findings, RBP^-/-Cry1^-/-Cry2^-/- mutant mice were generated; these animals lack all molecular and behavioral indicators of circadian photoreception (Thompson, Selby, and Sancar 2002; unpublished observations), consistent with the following model: CRY1 and CRY2 are the main circadian photoreceptors in mammals; rods and cones provide a redundant function to circadian rhythmicity; and melanopsin performs a lesser role, possibly along with other minor opsins of the inner retina. It must be acknowledged, however, that despite these strong genetic and phylogenetic data implicating CRY1 and CRY2 as the major circadian photoreceptors, it is formally possible that the CRYs function primarily in signal transduction, downstream of a non-opsin photopigment. The biochemical demonstration of a cryptochrome-signaling photocycle is the last frontier in the circadian photoreceptor hypothesis in mammals.

The Circadian Rhythm in Human Health

Because the circadian clock affects all aspects of human physiology, clock disorders may influence virtually all human diseases. However, several conditions are especially intimately related with clock disorders and these are briefly summarized below (12) .

Jet Lag

Global travel can place the individual into environments out of synchrony with his or her circadian rhythm. Jet lag manifests itself as tiredness, irritability, lack of concentration, gastrointestinal symptoms, and insomnia. The condition may be alleviated by exposure to strong light pulses to reset the circadian phase to the light-dark cycle of the new locale. Some individuals may benefit from melatonin at bed time, as melatonin levels, which are tightly regulated by the biological clock, are highest during the night phase.

Seasonal Affective Disorder

Seasonal Affective Disorder (SAD) is a relatively common disease that occurs in winter, especially in latitudes where the days become particularly short. The patients exhibit the classic symptoms of depression: lack of energy, morbid state of mind, oversleep, and sense of helplessness. Apparently, the insufficient exposure to sunlight causes clock entrainment problems that result in depression; SAD is more severe than the more common “winter blues” that affect almost everyone to some degree. SAD patients benefit from phototherapy with white light or blue light that, when administered properly, can synchronize the patient’s circadian clock to apparent LD 12:12 phases. Many patients also benefit from moving to latitudes where sunlight is stronger and less variable with the season.

Familial Advanced Sleep Phase Syndrome

The intrinsic period (free-running period) of the human circadian clock is twenty-five hours, or perhaps 24.2–24.4 hours according to a recent report (54) . Individuals with Familial Advanced Sleep Phase (FASP) syndrome have a shorter free-running period, and even with daily exposure to sunlight that would otherwise re-synchronize the clocks of normal individuals, the intrinsic clock defect persists. As a consequence, affected individuals go to bed typically in the late afternoon and wake up accordingly in the early hours of the day. In at least one type of this syndrome, a mutation in the Per2 site that is phosphorylated by CKIε has been found (55) . Significantly, a spontaneously occurring mutation in the gene that encodes CKIε in some inbred hamsters shortens the free-running period of these animals (56) . Whether FASP syndrome invariably involves a mutation in Per2 or other clock genes remains to be determined.

Circadian Disruption and Breast Cancer

The SCN (i.e., the circadian master clock) compose a neuroendocrine organ that regulates various endocrine systems by humoral and neural pathways. There are a number of indications that the “light noise,” or light at night (LAN), characteristic of modern living, may act as an endocrine disruptor through unnatural phototransduction to the SCN. A number of studies in mice exposed to constant light have established a correlation with higher incidences of mammary tumors (57) . In humans, several epidemiological studies have also found a correlation between LAN and increased risk of breast cancer (58–60) . For example, several studies from Scandinavian countries have reported an approximate twofold higher risk of breast cancer in sighted women, relative to blind women, and a more dramatically heightened incidence among sighted women who work at night, also compared to blind women. The exact mechanism of the light-dependent risk of breast cancer is not known, although melatonin is known to suppress estradiol production and may have a direct oncostatic effect on estrogen receptor–dependent breast cancers. Mice bearing a mutation in the Per2 gene (and hence severely compromised in circadian clock function) are consequently prone to cancer, and in particular exhibit increased susceptibility to tumor development from ionizing radiation, underscoring the role of a normal functioning circadian clock in cancer prevention(61) . Clearly, the effect of circadian rhythms on cancer development is an important aspect to the treatment of disease.