Beyond vision: effects of light on the circadian clock and mood-related behaviours

Introduction

Light perceived by the eyes passes through the lens and hits the retina at the back of the eye, leading to the formation of images and enabling the tracking of objects¹. Importantly, the eyes communicate information regarding light not only for vision but also to regulate various physiological and behavioural functions, many of which are independent of image formation, termed non-image-forming functions. These include the adjustment of the circadian clock to environmental light, suppression of sleep, and changes in alertness and mood. Hence, light is vital for both image formation and non-image-forming functions².

As the Earth rotates on its axis and around the Sun, light is not constantly present (except in polar regions during summer or winter seasons, respectively). However, due to the geophysical properties of the Earth, the change in lighting conditions is predictable. This predictability allowed organisms on Earth to develop a mechanism in the form of an internal clock with a period of about one day (circa diem) that can track and anticipate environmental changes i.e., day-night transitions but also seasonal changes in day length (photoperiod). This circadian clock is self-sustaining and continues to cycle even in the absence of light; however, it can adjust its phase according to environmental lighting³. Within an organism, the clock regulates biochemical and physiological processes, aligning metabolism and behaviour accordingly. Consequently, body function is synchronized with environmental demands⁴. This temporal alignment between light, the clock, and physiological processes is beneficial for an organism’s survival in a competitive environment^5,6.

One of the core symptoms in mood disorders, including seasonal affective disorder (SAD)⁷, major depressive disorder (MDD)^8,9,10, and bipolar disorder (BP)^11,12 is abnormal sleep/wake patterns, which serve as one of the diagnostic criteria for mood disorders¹¹. Sleep is a fundamental state of quiescence with reduced mental and physical activity, altered consciousness, and inhibited sensory activity. It is regulated on one hand by circadian clock mechanisms that determine sleep timing, and on the other hand by a homoeostatic component that determines sleep drive, which can override circadian timing¹³. Importantly, light can influence both circadian and homoeostatic components of sleep^14,15,16,17 and it suppresses secretion of melatonin, a molecule with a multitude of physiological actions¹⁸. However, it is unclear whether sleep disturbances are the cause or consequence of mood disorders.

In this review, we discuss studies in mice that describe the non-image-forming functions of light that affect the circadian clock and mood-related behaviours at the molecular level. The extension of these findings in mice to human disease relevance is also discussed.

How light reaches the brain

Image formation begins when photons of light hit photoreceptors (rods and cones) lining the retina at the back of the eye. Information regarding light is communicated via the optic chiasm to the lateral geniculate nucleus (LGN) and then the occipital lobe visual cortex (Fig. 1), where image processing takes place. Cones are primarily responsible for colour vision, functioning best in bright light (i.e., during the day), while rods are specialized for the detection of low light levels, enabling organisms to see during dawn and dusk¹⁹.

**Fig. 1: Retinal brain connections underlying the effects of light on visual functions.**

In recent years, non-image-forming pathways by which light influences the brain have been identified in the mouse. Intrinsically photosensitive retinal ganglion cells (ipRGCs) containing the photopigment melanopsin²⁰, send glutamatergic projections via the retino-hypothalamic tract (RHT; Fig. 1) towards the mammalian master clock, the suprachiasmatic nuclei (SCN)²¹. Here, the phase of the circadian clock is shifted by nocturnal light directly²² while processes such as hippocampal learning are located downstream and thus indirectly affected by light²³. Furthermore, various subtypes of ipRGCs project to the subparaventricular zone (SPZ) and intergeniculate leaflet (IGL), both of which are important for clock regulation, as mediators of SCN signals and as participants in broader networks of circadian regulation. Other targets include areas involved in sleep regulation such as the ventrolateral preoptic area (VLPO) and lateral hypothalamus (LH), or the medial amygdala and lateral habenula (LHb), which are implicated in the regulation of mood²⁴. Recent studies show that ipRGCs encoding light-intensity innervate the perihabenula (PHb). The PHb in turn projects to the LHb, affecting the ventral tegmental area (VTA), the nucleus accumbens (NAc) and the medial prefronatal cortex (mPFC), driving light-mediated alterations in mood (Fig. 1)^23,25,26,27. ipRGC-like properties which encode changes in ambient lighting levels (irradiance) have also been detected in human prefrontal regions revealing light-intensity-dependent responses, suggesting that the pathway by which light affects mood and cognition in mice may also exist in humans²⁸. Interestingly, a retino-raphe pathway, a direct connection between the retina and the dorsal raphe nucleus (DRN) has been described in the cat²⁹. ipRGCs as well as other types of RGCs project to the DRN³⁰ to regulate serotonergic and GABAergic activity³¹ ultimately affecting depression-like behaviour³². This pathway may also exist in humans as indirectly evidenced by variation of serotonin levels in blood samples collected in different seasons and lighting conditions³³. However, the molecular details of how light may affect serotonin and GABA levels in the DRN are unknown.

Melatonin, a modulatory hormone produced by the pineal gland is inhibited by light^34,35, enabling the encoding and communication of day length (photoperiod) across seasons and may play minor contrasting roles in sleep vs activity in diurnal³⁶ and nocturnal mammals³⁷, respectively. For an in-depth review for melatonin in circadian rhythms and sleep see ref. ¹⁸. The importance of ipRGCs in non-image forming processes is evident in visually blind humans, who lack rods and cones but not ipRGCs, and can still detect light and inhibit the secretion of melatonin³⁴. Furthermore, mice lacking rods and cones can still align to the day-night cycle (termed photoentrainment)^15,38,39,40. Interestingly, however, mice which lack intrinsically photosensitive ipRGCs due to the loss of the photopigment melanopsin show slightly attenuated photoentrainment (but also in non-image forming responses i.e., pupil constriction) in response to monochromatic (one colour) light, however, as clock phase-shifting responses are not completely abolished this indicates that other photoreceptors (i.e., rods and cones) play a role in photoentrainment. While changes in light levels during the day are prominent, another change that occurs is the spectral (colour) composition of light, with the ratio of blue-to-yellow colour shifting as the sun sets⁴¹. Notably, the absence of this change in blue-to-yellow ratio shifted rodent body temperature and SCN electrical activity to earlier time points⁴², indicating a role for cones in photoentrainment. Importantly, mice lacking all three photoreceptors cannot align their activity to the day-night cycle⁴³. Surprisingly, the complete loss of ipRGCs produces this same phenotype⁴⁴, indicating a central role of ipRGCs in communicating non-image-forming light responses by relaying extrinsic information about light from rods and cones but also intrinsic light information coming from the pigment melanopsin⁴⁵.

Light and the circadian clock

Environmental lighting aligns the internal circadian clock and the sleep-wake cycle to the solar day. This defines the organism’s activity to the correct temporal niche, either during the day (diurnal) or during the night (nocturnal). Hence, light acts as a time-giver (Zeitgeber) to the SCN (Fig. 1)⁴⁶, aligning the circadian clock, whose period length differs slightly from 24 h, to the day-night cycle. Artificially induced light changes, as experienced during shift-work or as the results of jet-lag, can directly affect the SCN molecular clock. Temporal changes in light levels induce adaptive behaviours, either advancing or delaying an organism’s activity phase relative to when the light is presented, as illustrated in the phase-response curve to light (Fig. 2)^47,48. Hence, light late in the dark period (i.e., night time) advances clock phase, whereas light in the early part of the dark period delays it.

**Fig. 2: The Phase Response Curve (PRC).**

How exactly circadian behaviour is linked to molecular events has been an intensely researched topic. The first convincing evidence demonstrating the governance of the circadian clock by genetic mechanisms was carried out in chemically mutated fruit flies, which lacked circadian activity behaviour in constant darkness⁴⁹. The cloning of specific genes in the fly and the demonstration of an autoregulatory feedback loop as the basis of a circa 24-hour oscillation in the behavioural activity of the fly^50,51,52 led to the elucidation of the underlying clock mechanism in mammals⁵³. Like in the fly, the mammalian circadian clock mechanism consists of transcriptional and translational processes that generate a cell-autonomous autoregulatory feedback-loop with a period of ~24 h. The core clock genes are Bmal1 and Clock (and its homologue Npas2), which bind to specific promoter elements called E-boxes, activating transcription of Per and Cry genes. PER and CRY proteins, in turn, repress their own transcription by inhibiting BMAL1 and CLOCK complexes (Fig. 3). This, combined with the activity of the nuclear receptors REV-ERBα and RORα, regulates the expression of clock components Bmal1, Clock and Npas2, repressing or activating their transcription, respectively, ultimately generating the clock mechanism with a periodicity of ~24 h⁵⁴ (Fig. 3). The period of the clock is fine-tuned by many additional levels of regulation, including the modification of DNA (epigenetic mechanisms, EPM) through methylation, acetylation, and non-coding RNA intervention, or the modifications of clock proteins themselves (post-translation modification; PTM) e.g. with phosphorylation and sumoylation^55,56. Importantly, phosphokinases, which drive the proteasomal degradation of PER proteins through phosphorylation⁵⁷, may provide a mechanism to couple external stimuli (e.g., light) with the clock. Indeed, mice lacking protein kinase-C have defects in light-induced resetting of the clock⁵⁸.

**Fig. 3: Cellular and molecular mechanism of the circadian clock.**

The timing information generated by this autoregulatory feedback-loop can align important molecular processes within the organism. Indeed, many genes encoding rate-limiting enzymes or those involved in the regulation of key physiological processes contain E-box promoter elements; hence, they are activated in a clock- dependent fashion and can be referred to as clock-controlled genes (Ccgs; Fig. 3). Thus, important biological processes are temporally controlled, including metabolism and neurotransmitter expression. Consequently, the peak activity of these processes is aligned appropriately to the time of day when their function is most crucial⁵⁹. Hence, the output of the circadian clock is the regulation of physiology and behaviour.

As described above, circadian activity behaviour is modulated by light. Since behaviour is regulated by the clock, behaviour must be altered by light. IpRGCs activated by light promote the release of several neurotransmitters involved in photic stimulation from the RHT at target brain regions, including glutamate, pituitary adenylate cyclase-activating polypeptide (PACAP) and substance P. Glutamate from the RHT generates an influx of Ca²⁺ into SCN cells by activating N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors⁶⁰, ultimately activating several intracellular signalling pathways. This involves changes in activity of various kinases, including cyclin-dependent kinase 5 (CDK5), dopamine and cAMP-regulated phosphoprotein of 32 kD (DARPP-32), protein kinase A (PKA), and Ca²⁺/calmodulin-dependent kinases (CaMK), culminating in the phosphorylation of cyclic AMP response element binding protein (CREB) (Fig. 4)^{61,62,63,64,65,66}.

**Fig. 4: Molecular model of light-induced signalling pathways for clock phase shifting.**

This promotes phosphorylation⁶⁷ and acetylation of chromatin by recruiting cAMP-regulated transcriptional co-activator 1 (CRTC1) and the histone acetyltransferase CREB-binding protein (CBP), which involves the clock protein PER2⁶⁸. Consequently, immediate-early genes such as cFos and clock genes such as Per1 are induced (Fig. 4)^69,70,71, causing a rapid shift in the phase of the clock autoregulatory feedback loop in oscillating SCN cells⁷². This manifests at the behavioural level in delaying activity onset (Fig. 2, between CT12-18). Furthermore, CRTC1 is expressed rhythmically in the SCN, peaking during the middle of the day and light during the early or late night induces strong nuclear accumulation of CRTC1, leading to the induction of salt-inducible kinase 1 (SIK1) in addition to Per1 (as mentioned above^73,74). Negative regulation of CREB occurs via SIK1 through the phosphorylation and inactivation of CRTC1, forming a negative autoregulation feedback loop. Importantly, SCN knockdown of SIK1 enhances phase-shifting responses to a phase-delaying light pulse and enables rapid re-entrainment to a shift in the light cycle i.e., as experienced in jet lag⁷⁴. The advancement of activity onset (phase advance as seen in Fig. 2, between CT18-24), however, involves another signalling pathway that is not yet fully understood, except for the observation that L-type voltage gated calcium channels (L-type VGCCs) are important^75,76. Additionally, cholecystokinin-expressing neurons may play a role as their specific activation induced phase advances⁷⁷. Taken together, it appears that light signals perceived in the early portion of the dark phase elicit rapid phase delays via CREB signalling, PER proteins, and T-type VGCCs^66,75, whereas light perceived in the late part of the dark phase elicits rapid phase advances via L-type VGCCs and an unknown signalling pathway. This quick adaptation of the clock to a light signal is referred to as clock resetting, as visualized in the PRC of mice (Fig. 2). However, of note is that the human PRC does not have a dead zone and hence is only similar but not identical to the mouse PRC⁷⁸.

The PRC and the mechanism described above by which the clock responds to light are important for the short-term effects of light, mainly during the dark phase under a constant photoperiod (in the laboratory, 12 h of light and 12 h of darkness are commonly used). However, slow seasonal changes occur in environmental lighting throughout the year. As a result, the length of the light and dark (photoperiod) changes during the 24-h day, influencing the molecular clock. The clock is entrained to this change in photoperiod, inducing sustained plasticity in the SCN⁷⁹. A recent study investigated the transcriptomic plasticity of the circadian clock in response to photoperiod⁸⁰. Interestingly, no significant changes in the expression levels of core clock genes were observed. However, prominent differences across photoperiods were found in genes that regulate SCN neuropeptide signalling and light responses. Within this study, molecular elements shown previously to be involved in photoperiodic plasticity in the SCN, such as GABA-A receptor signalling and chloride co-transporters NKCC1 and KCC2^81,82,83, were observed to be important. This appears to be accompanied by light-dependent reprogramming of DNA methylation⁸⁴, producing persistent alterations in the SCN and altering its acute response to light pulses^85,86,87. Ultimately, this indicates that entrainment to the photoperiod may overlap mechanistically with the acute response of the clock to light pulses, as described previously. This is supported by the identification of acute light-responsive genes involved in photoperiod entrainment, including dual-specificity phosphatase 4 (Dusp4), ras related dexamethasone induced 1 (Rasd1), and GTP binding protein overexpressed in skeletal muscle (Gem)^80,88, of which Gem gene induction is regulated by CDK5 in response to acute light⁶⁶. These genes regulate or are regulated by the CREB-signalling pathway in the SCN and are probably important for the daily alignment of the clock to seasonal changes in photoperiod. An alternate light-inducible signalling pathway involving ERK signalling, culminates in FOS/JUN binding to AP-1 sites on target genes (Fig. 4)⁸⁰. Dexras1 was shown to potentially act as a link between NMDA receptors and ERK activation as the light-induced activation of ERK signalling was attenuated in mice lacking Dexras1. Furthermore, NMDA produces phase shifts in SCN neuronal activity which can be blocked through the pharmacological blockage of G_i/o, however, this effect is absent in Dexras1 knockouts, demonstrating a link between NMDA receptors, G_i/o and ERK⁸⁹. While only a fraction of the light-regulated transcriptome can be linked to photoentrainment functionally, the role of a large amount of this transcriptome still remains unclear.

Mood-related behaviours and light

Mood is an affective state influenced by many factors such as motivation, alertness, attention, and anxiety. The various components comprising mood can be regulated by neurotransmitters like dopamine, serotonin, and noradrenaline⁹⁰, illustrating the complexity of mood states.

Photoperiod not only induces neuroplasticity in the circadian system as described above⁹¹, but it also produces neurotransmitter switching in rats exposed to either short- or long-day photoperiods⁹². Under a short photoperiod, hypothalamic GABAergic interneurons switch to dopamine (DA) neurons, and the ablation of these neurons increases anxiety and depression-like behaviour⁹³. In contrast, long photoperiods have the opposite effect. This suggests that monoaminergic systems in the brain are altered by changes in photoperiod, resulting in changes in mood-related behaviours. The length of photoperiod is modulated by DA signalling in the NAc, affecting female and male mice differently⁹⁴. This is of interest for specific forms of depression in humans, e.g. SAD, which has a higher prevalence in females compared to males. As mice and rats are nocturnal, while humans are diurnal, the responses to light in nocturnal versus diurnal species may be different. Indeed, while the molecular mechanisms governing pacemaker activity in the SCN is conserved in nocturnal and diurnal mammals⁹⁵, downstream responses may differ. In support, the photoperiod affected the number of DA neurons in the grass rat, a diurnal rodent, in an opposite manner compared to mice⁹⁶. Interestingly, a short photoperiod evoked depressive-like behaviour in these animals reminiscent of SAD in humans. Therefore, Costello et al.⁹⁷, applied a light protocol similar to that used in bright light therapy (BLT) for humans to grass rats. The resulting higher wakefulness, increased nighttime sleep quality, and improved rhythm entrainment were comparable to those observed in humans⁹⁷. Interestingly, a similar BLT treatment had positive effects in nocturnal mice subjected to the forced swim test, a proxy for motivation-related behaviour. Mice subjected to a 30-min light pulse, two hours before daily light onset (ZT22), showed increased motivation⁹⁸, resembling the positive effects of BLT in humans. In conclusion, it appears that the initial light signalling pathway which evokes the positive effects of BLT is likely conserved between nocturnal and diurnal species. However, the interpretation of this signal may affect downstream pathways in an opposing manner (e.g. neurotransmitter regulation) to achieve identical behaviour in nocturnal and diurnal species. This has its prerequisite in the light-dependent delay or advance of clock phase, which is conserved in nocturnal and diurnal species⁹⁹. Importantly, SCN clock gene expression has the same phase in diurnal and nocturnal animals¹⁰⁰; however, outside the SCN¹⁰¹ and in peripheral tissues¹⁰², it is opposite. Therefore, it is likely that the differences between diurnal and nocturnal species are at the level of cellular and neuronal networks rather than at the level of clock genes. This view could be tested by studying a recently described mechanism in mice for adaptation to photoperiod¹⁰³ using a diurnal animal such as the grass rat.

Molecular mechanisms linking light, circadian rhythms and mood

As previously discussed, we see that light can affect circadian rhythms and mood. However, questions emerge as to whether light can affect both systems independently or whether they intersect at the molecular level with light regulating both pathways similarly. The third possibility would be that light affects the clock, which then regulates mood, resulting in mood being indirectly regulated by light via the clock.

The first evidence that clock components may be involved in mood regulation was obtained in the fruit fly when circadian clock genes were observed to be required for cocaine sensitization (per, clk, cyc, dbt, but not tim)¹⁰⁴. Similar observations highlighted the involvement of period genes (Per1 and Per2) in cocaine sensitization in mice¹⁰⁵. The self-administration of psychoactive substances such as alcohol¹⁰⁶ and fentanyl¹⁰⁷ may influence and/or be influenced by the clock particularly via Period genes and ultimately influence mood-related systems. Indeed, Per2 mutant mice show increased extracellular glutamate and increased alcohol intake, both of which are normalized following the administration of acamprosate, a drug used to prevent alcohol relapse¹⁰⁶. Furthermore, fentanyl was shown to suppress light-mediated phase shifts and was accompanied by the absence of Per1 gene induction in the hamster SCN¹⁰⁷. Interestingly, fentanyl was shown to stimulate Cdk5/p35 activity and inhibit the PKA signalling pathway through the activation of opioid receptors¹⁰⁸, consistent with the inhibitory effects of fentanyl on light-mediated Per1 induction¹⁰⁷ (Fig. 4). At the same time, the activated opioid receptors modulate MEK/ERK signalling through Cdk5/p35/25 activity¹⁰⁸. This pathway has been shown to regulate light-mediated gene expression¹⁰⁹ and clock resetting¹¹⁰. These observations suggest that psychoactive substances and light may modulate mood-related behaviours and clock resetting via similar signalling pathways involving clock components such as Per1. Recently, the beneficial effects of light on mood involving the Per1 gene were described⁹⁸. Specifically, light-mediated induction of Per1 in the LHb/PHb was necessary, as specific deletion of Per1 in the LHb/PHb abolished the beneficial effects of light. However, it is not clear if this effect is related to normal clock function or whether it is a Per1-specific function independent of the clock.

Associations between clock genes and mood-related disorders have been described in several human studies. Single nucleotide polymorphisms (SNPs) in clock genes are associated with affective disorders like SAD, including CLOCK¹¹¹, PER2, NPAS2, BMAL1¹¹², PER3¹¹³, and CRY1 and CRY2¹¹⁴. However, such small-scale SNP studies in humans have recently been discredited as false positives, because genome wide association studies (GWAS) did not confirm any single clock gene variant to be genome-wide significant (reviewed in ref. ¹¹⁵). However, assembling large cohorts as used in GWAS, there is often loss of phenotypic detail. Therefore, molecular links relating the circadian clock to depressive behaviour in humans are scarce. To close this gap, mouse models with genetic disturbances in clock genes are used to investigate their involvement in mood disorders.

General deletions and mutations of clock genes in the entire organism

Mice overexpressing a truncated form of the CLOCK protein caused by a Clock gene mutation (Clock Δ19)¹¹⁶, displays decreased sleep and anxiety, and increased reward-related behaviour¹¹⁷ (Table 1). Since overexpression of the mutated protein may affect many cellular processes in a non-specific manner, it is unclear whether these observations are caused directly by the circadian clock or are of an indirect nature. Interestingly, deleting the Clock homologue Npas2¹¹⁸ reduced anxiety similarly to the Clock Δ19 mutation¹¹⁹. Furthermore, mice that express an unstable non-nuclear protein due to a mutation in the clock component Per2 (Per2^Brdm1)^120,121, show decreased despair¹²², similar to Clock Δ19 mutated mice (Table 1). In both mutations (Clock Δ19 and Per2^Brdm1), the mesolimbic dopaminergic system is affected, potentially due to changes in dopamine-synthesizing¹²³ or degrading enzymes¹²². In contrast to the decreased despair phenotype of Per2^Brdm1 mice, Per1 knock-out mice show increased despair-like behaviour, while animals with deletions in both Per1 and Per2 genes show increased anxiety¹²⁴, comparable to the Per1 knock-out animals⁹⁸ (Table 1). Similarly, mice with deletions in Cry1, Cry2 or both, displayed changes in anxiety; however, despair-related behaviour was unaffected^{125,126,127,128} (Table 1). Interestingly, restoring circadian rhythmicity in Cry1/Cry2 double knock-out (KO) mice by rescuing Cry1 expression in the SCN improved anxiety-like behaviour as well as metabolic deficits¹²⁹. Discrepancies between studies suggest that the genetic differences between mouse strains used, breeding conditions, and/or variations in animal facilities may influence behaviour. Particularly, this may be the case for behaviours sensitive to environmental or housing conditions, including tests used to measure depressive- or anxiety-like behaviour in mice. Indeed, significant differences in exploratory behaviour (and thus anxiety-like behaviour) is observed between males and female mice in the open field, with group-housed females spending more time in the open area than their individually-housed counterparts. This observation is inverted in male mice, with individually-housed males spending significantly more time in the open section than their group-housed counter parts¹³⁰. Thus, the sex used and the way animals are housed can greatly influence the observations between studies.

Table 1 Mood related phenotypes of mice lacking clock genes in all cells or specific cell types and/or brain regions (adapted from Imamura and Takumi¹⁵²)

Full size table

The nuclear receptor REV-ERBα, a negative regulator of Bmal1, Clock and Npas2 (Fig. 3), is implicated in the regulation of mood-related behaviours in mice. Animals with a deletion in the Rev-erbα gene show a similar phenotype to Per2^Brdm1 animals, with a notable elevation in striatum dopamine levels caused by increased transcriptional levels of the tyrosine hydroxylase gene¹³¹ (Table 1), which encodes the rate-limiting enzyme in dopamine synthesis. As Per2 regulates the rate-limiting enzyme of dopamine degradation, momoamine-oxidase A¹²², it appears that both clock components REV-ERBα and PER2 balance the synthesis and degradation of dopamine, thereby influencing mood (Fig. 5). Because PER2 protein was shown to interact with nuclear receptors, including REV-ERBα and NURR1¹³², it can be hypothesized that the level of PER2 protein may modulate temporal synchronization between the clock and the regulation of ccgs as shown in Fig. 5. However, REV-ERBα may not only influence the dopaminergic system but also the serotonergic system, as has been proposed recently^133,134.

**Fig. 5: Model for balancing dopamine levels by clock components.**

The development of mood disorders is likely associated with changes in glucocorticoid levels influenced by stress. Glucocorticoids act by binding to glucocorticoid receptors (GR), and the subsequent translocation of GRs to the nucleus activates the transcription of target genes. Importantly, this process is influenced by REV-ERBα, which competes with GRs for the binding of heat-shock protein (HSP) 90 in the cytoplasm. Thus, these two nuclear receptors influence each other’s stability and nuclear localization in an opposing cyclic manner¹³⁵ (Fig. 5). This is paralleled by the opposing rhythmicity of REV-ERBα and GR target gene expression^136,137. However, these observations were made in liver. It remains to be seen whether the same mechanism applies to brain tissue. Furthermore, Chrono, a less well-known negative regulator of the circadian clock, appears to be involved in glucocorticoid signalling, as the lack of this gene resulted in higher corticosterone levels following restraint stress when compared to control animals also subjected to this stressor¹³⁸ (Table 1). Hence, clock components such as REV-ERBα and Chrono may affect mood via noradrenergic signalling. Although it remains to be tested whether Chrono KO mice display changes in mood-related behaviours.

As shown in Fig. 3, clock components are regulated by post-translational modifications (PTMs). Therefore, it is not surprising that post-translational modifiers such as FBXL3 or CDK5, which phosphorylate clock components, can modulate mood-related behaviours^{139,140,141,142,143} (Table 1). Taken together, there is good evidence that clock components directly or indirectly modulate pathways involved in mood regulation. Taking caution to conclude that the circadian clock regulates mood is warranted. The evidence described above is based on clock gene deletions that manifest during embryogenesis, and these deletions are in the entire organism. Hence, in mice lacking specific clock genes in the entire organism, a mixture of developmental and systemic effects, including sleep may impact on mood-related behaviours.

Localized deletion or suppression of clock genes in the brain

Since light information is transmitted to specific brain regions to affect certain clock genes and mood, it is of great interest to understand how clock genes and lack of them in specific brain areas would affect mood. Several laboratories have started to investigate this question by using either genetically modified mice that allow the deletion of a specific gene in specific cell types or by using viral mediated transfection systems that allow the deletion of genes in specific brain areas of the adult animal.

Bmal1 is considered the most important clock gene, because its absence leads to arrhythmic activity behaviour, indicating a non-functional clock. Therefore, knock-down (KD) of the Bmal1 gene through RNA interference in the SCN was studied. While SCN-Bmal1-KD animals showed circadian activity rhythms with a longer period compared to controls, these mutants displayed several characteristics of depression-related behaviours¹⁴⁴ (Table 1). This suggests that the clock component Bmal1 affects mood-related behaviour including despair. However, this is not necessarily related to the circadian clock, because these mice were not arrhythmic. Interestingly, lack of Bmal1 in glial cells did not affect despair^145,146, indicating that the despair-related phenotype observed by Landgraf et al.¹⁴⁴ is relevant to SCN neuronal populations. The deletion of Bmal1 in CaMK2a glutamatergic neurons of the mouse mPFC (Fig. 1) was associated with changes in behaviour as assessed in a chronic despair model¹⁴⁷, illustrating how the deletion of a clock component in a specific brain area can have specific effects.

Knock-down of the Clock gene in the VTA resulted in hyperactivity, less anxiety and increased depression-like behaviour¹⁴⁸ (Table 1). While hyperactivity and decreased anxiety have been observed in the Clock Δ19 mutant as well, depression-like behaviour was decreased in these animals¹¹⁷, which is just the opposite of the observation in the VTA specific KD of Clock. The reason for this is manyfold. First, Clock Δ19 mutant mice overexpress a mutant form of Clock (it is not a deletion of Clock), and therefore may affect cellular processes in an unspecific manner. Second, the mutation is present in all cells and not only in the VTA, and hence presence of mutant Clock in other tissues may contribute to the phenotype observed in Clock Δ19 mutant animals. This underscores the importance of studying gene deletions in a brain region specific manner.

Mice with a KD of Per1 and Per2 genes in the NAc displayed increased anxiety, as observed in the whole-body Per1/2 KO animals¹²⁴ (Table 1). This implicates NAc Per gene expression in the regulation of anxiety stress responses. Interestingly, however, deletion of Per2 in glial cells of the NAc did not affect anxiety, but significantly impacted despair¹⁴⁵, suggesting that the lack of glial Per2 in the NAc does not contribute to anxiety. This illustrates the potential of deleting clock genes in specific cell types and brain areas to understand the relationships between clock gene expression and mood regulation.

Suppression (KD) of Rev-erbα in the NAc increased sociability and reduced anxiety-like behaviour but had no effect on depressive-like behaviour in female mice. In contrast, male mice did not show alterations in any of the behaviours tested¹⁴⁹, comparable to observations made in Per2 KO animals, whereby only female mice displayed strong phenotypes⁹⁸ (Table 1). These observations suggest that Rev-erbα in the NAc alone is not sufficient for the regulation of depressive-like behaviour. However, this needs to be experimentally tested by thoroughly deleting Rev-erbα in the NAc, because KD only reduces but does not completely eliminate expression of Rev-erbα mRNA.

The specific deletion of cryptochrome genes Cry1 and Cry2 in dopamine 1 receptor-expressing medium spiny neurons of the NAc⁹², resulted in mice displaying reduced susceptibility to stress-induced helplessness and increased neuronal activation in the NAc at night. This suggests a role for CRY in the regulation of the dopamine reward system.

As highlighted earlier, the LHb is an important brain region involved in the regulation of depression-related behaviours in mice²⁵. This brain region appears to be important for the beneficial effects of light on mood, because the lack of Per1 in the LHb abolishes the beneficial effects of light⁹⁸ (Table 1). However, mice lacking Per1 in the LHb do not display a depression-like phenotype compared to control animals⁹⁸, suggesting that Per1 in the LHb influences only the beneficial effects of light but is not involved in regulating mood in the absence of light used in bright light therapy. Per1 expression in other brain regions such as the NAc or VTA are potential candidates whereby mood can be affected directly. However, this requires clarification in future experiments using mice with specific deletions in these specific brain regions.

Overall, it appears that clock genes in specific brain regions and cell types can affect mood-related behaviour. While we know that light can affect mood by stimulating specific brain regions, including the induction of Per1 gene expression in the LHb, further studies are required to fully dissect the influence of light in brain regions implicated in mood and how light impacts clock genes in these brain areas.

Conclusions and outlook

Light has a profound influence on the circadian clock and mood, not only in rodents but in humans as well¹⁵⁰. Depending on the time when light hits the retina, different behavioural outcomes can be observed. This is visualized by the phase response curve (PRC) where light perceived early in the prospective dark phase elicits phase delays but when perceived in the late part of the dark phase it elicits the opposite response, namely phase advances (Fig. 2). Interestingly, a similar relationship between the timing of light and its effects on mood-related behaviours have been observed. Light in the early part of the dark phase appears to be pro-depressive²⁶, whereas later in the dark phase it seems to be anti-depressive^7,98. Therefore, it is intriguing to speculate that there may be a PRC for the relationship between light and mood-related behaviours. This potential mood PRC could possibly be used to develop treatments for neurobiological disorders such as depression.

The main gap in knowledge lies in understanding the molecular pathways triggered by light and how exactly these connect to the cells regulating neurochemical pathways. Light detection and signalling is a very dynamic process involving immediate effects (Fig. 4, delay or advance of clock phase) that later turn into a medium-term change in cellular signalling and metabolism (photoperiod adaptation, change of neurotransmitter signalling). How exactly light influences the balance (synthesis vs degradation) of dopamine metabolism (Fig. 5) is not understood. The core clock genes affected by light are the Per genes, particularly Per1. Therefore, an immediate change in their expression levels induced by light may have a downstream effect on Ccgs, which regulate key biochemical pathways. One scenario could be that protein levels of PER1 and PER2 and their respective ratio would determine which Ccgs are induced or repressed and hence which biochemical pathways are activated or repressed, thereby regulating physiological/behavioural outcomes. Furthermore, mutations in clock genes impact on nonvisual photoreception in the mouse¹⁵¹ and may by this avenue affect various behaviours modulated by light.

Our overall knowledge of specific aspects of light on behaviour is relatively well investigated. However, the molecular underpinnings are not understood. We have fragmentary knowledge of specific molecular/cellular processes of light on signalling pathways and the circadian clock but do not understand how these pathways are molecularly linked and what the chain of events is to generate the observed behavioural consequences initiated by light treatment.