Disrupted Sleep and Circadian Rhythms in Schizophrenia and Their Interaction With Dopamine Signaling

Recent studies provide evidence that sleep disruption may indeed play a causal role in the development of psychosis. It is clear that SCRD can precede the appearance of psychosis on the one hand, and exacerbate negative outcomes on the other. Fragmented sleep and blunted circadian rhythmicity of behavior have been observed in adolescents and young adults at risk for psychosis (Castro et al., 2015) and SCRD predicted the severity of psychosis symptoms and psychosocial impairment at 1 year follow-up in adolescents with clinically high risk for psychosis (Lunsford-Avery et al., 2017). With similar findings in bipolar disorder (Ng et al., 2015), now often considered to be on the same spectrum as schizophrenia, further long-term studies are required to assess the predictive power of SCRD in early diagnosis. Even in healthy subjects, psychotic episodes can be predicted by poor perceived sleep quality in combination with reduced sleep duration (Cosgrave et al., 2018b). Further, sleep deprivation, or restricted sleep to mimic insomnia in healthy subjects leads to symptoms associated with schizophrenia such as reduced prepulse inhibition, increased delusions and hallucinations, and psychotic episodes (Petrovsky et al., 2014; Reeve et al., 2018).

Disrupted sleep in schizophrenia is associated with poor quality of life, higher rates of relapse and suicide (Hofstetter et al., 2005; Pompili et al., 2009; Palmese et al., 2011), and the stabilization of sleep can have beneficial effects. A controlled trial administering cognitive behavioral therapy (CBT) to reduce insomnia in a cohort of over 3000 university students had the concomitant outcome of reducing the appearance of paranoid delusions and hallucinations in those subjects whose insomnia was reduced (Freeman et al., 2017). This approach has also been used in patients with schizophrenia where improvements in sleep are associated with reduced persecutory delusions and hallucinations (Myers et al., 2011; Waite et al., 2016). Pharmacotherapy for sleep stabilization includes traditional sedative-hypnotics such as benzodiazepines, and second generation sedating antipsychotics such as quetiapine, olanzapine and risperidone improve sleep quality, and pharmacotherapy for sleep also correlates with an improvement in negative symptoms (Kajimura et al., 1995; Yamashita et al., 2004; Kluge et al., 2014). Pharmacotherapy is administered with caution however, as many sedative-hypnotics impair slow-wave sleep and REM sleep patterns, sleep parameters which may already be impaired in schizophrenia.

Clinical studies examining melatonin use as an add-on therapy in schizophrenia suggest that it may also be effective at ameliorating sleep disruption in this disorder (reviewed in Bastos et al., 2019). Melatonin is a key hormonal output of the central clock, its rhythmic release acts to signal circadian time to the brain and peripheral tissues. Administered alongside antipsychotic medication, it results in improved sleep measures including sleep efficiency and duration (Shamir et al., 2000; Kumar et al., 2007). Furthermore, animal studies have demonstrated that it reduces schizophrenia-like behaviors in mice, suggesting that it may also be effective at treating symptoms such as cognitive impairment and social withdrawal (da Silva Araújo et al., 2017; Onaolapo et al., 2017). However, studies which have looked at the effect of melatonin on the psychotic symptoms in patients, although limited, show conflicting results. A reduction in symptom severity has been reported with add-on treatment of both melatonin (Modabbernia et al., 2014) and the melatonin receptor agonist, ramelteon (Mishra et al., 2020). In contrast, Romo-Nava et al. (2014) did not detect any improvement resulting from melatonin administration over placebo, although this study also included patients with bipolar disorder. Inconsistencies in melatonin efficacy may be due to deficiencies in its target regions or its receptors. Indeed, a polymorphism in the MT1 melatonin receptor gene is associated with schizophrenia (Park et al., 2011), indicating that the receptor expression may be altered in some patients.

Together the evidence from these studies suggests SCRD could indeed play a causal role in the development of schizophrenia and that the stabilization of SCRD has beneficial outcomes. As an extension, these findings suggest the SCRD could be a manifestation of a dysfunctional circadian system, which could underlie aspects of the pathophysiology of schizophrenia.

Genome-wide association studies have associated mutations in circadian clock genes with an increased risk of schizophrenia; single nucleotide polymorphisms (SNPs) in CLOCK, PER2, PER3, RORB, and TIMELESS have been associated with the disorder as assessed within relatively small groups of patients, numbering a few hundred (Takao et al., 2007; Mansour et al., 2009; Zhang et al., 2011). Whilst these associations with clock genes have not been replicated in larger studies and meta-analyses (Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2014; Pardiñas et al., 2018; Rees et al., 2019; Wang et al., 2019), these larger studies have implicated SNPs in genes with a role in dopaminergic/glutamatergic neurotransmission, mitochondrial function and immunity. All of these pathways also directly feed into the regulation of circadian rhythms or are regulated by the clock; highlighting multiple points of convergence with circadian rhythms in the pathophysiology of schizophrenia. For example, dopaminergic regulation of the clock is described in detail below and there is a growing body of literature on elevated neuroinflammation in the prefrontal cortex in schizophrenia. This has now been linked to increased nuclear factor-κB (NF-κB) transcriptional activation (Volk et al., 2019), and NF-κB function is regulated by CLOCK, with Clock deficient mice showing significantly reduced NF-κB activation in response to immunostimuli (Spengler et al., 2012).

Circadian disruption could result from aberrant synaptic networks and connectivity within the brain, which characterize both schizophrenia and bipolar disorder (Chai et al., 2011), or from a deficit at the level of clock gene expression. There is evidence for both. Blind-drunk (Bdr) is a mouse model of synaptosomal-associated protein (Snap)-25 exocytotic disruption that displays schizophrenic endophenotypes. Their circadian rhythms are fragmented and phase advanced, and there is some evidence that this arises from disruption of synaptic connectivity within the SCN, thus disrupting rhythmic outputs (Oliver et al., 2012). On the other hand, skin fibroblasts isolated from patients with chronic schizophrenia lose rhythms in the expression of the clock genes CRY1 and PER2 when compared with healthy controls (Johansson et al., 2016), and PER1/2/3 and NPAS2 mRNA levels were reported to be altered in white blood cells from schizophrenia patients (Sun et al., 2016). These studies show a deficit at the level of the molecular clock in schizophrenia that is cell autonomous, much like the clock itself. This has been replicated recently by a study analyzing gene expression in the human dorsolateral prefrontal cortex in schizophrenia and control subjects (Seney et al., 2019). By factoring time-of-death into transcriptome analysis, Seney et al. (2019) showed diurnal rhythms in gene expression of very different sets of genes in two groups, with circadian signaling showing strong rhythmic patterns in the control participants but not the schizophrenia patients. In contrast, mitochondrial signaling genes were rhythmic in the patients, but not the controls. The implications of these findings are as yet unclear, but point toward a mechanistic role for circadian rhythm, and consequently circadian disruption in schizophrenia.

Dopamine is one of the major regulators of sleep and wakefulness. Evidence from numerous studies spanning mammalian species and non-mammalian models, demonstrates the importance of DA in sleep control across phylogeny (Eban-Rothschild et al., 2018). The dopaminergic neurons in the ventral tegmental area (VTA) and dorsal raphe nucleus (DRN) are key components of the neuronal circuitry for the regulation of sleep and wake states in mammals, owing to their potent wake-promoting activity (Lu et al., 2006; Eban-Rothschild et al., 2016, 2018; Cho et al., 2017; Oishi et al., 2017; Yang et al., 2018). Selective optogenetic stimulation of VTA DA neurons in mice can induce behavioral arousal responses during anesthesia (Taylor et al., 2016), and induce wakefulness during NREM sleep, even following sleep deprivation when homeostatic sleep pressure would be high (Eban-Rothschild et al., 2016). Similarly, activation of DA neurons in the DRN induces wakefulness during NREM and REM sleep, while their chemogenetic inhibition reduces wakefulness and increases NREM sleep duration (Cho et al., 2017).

Dopamine neurons in both these regions, the VTA and DRN, exhibit differential firing patterns between sleep/wake states, although there are differences in the activity patterns between the two regions. In the DRN, DA populations are primarily wake-active, and interestingly the net increase in their activity at wake onset positively correlates with the duration of the subsequent wake episode (Cho et al., 2017), as it does in the striatum (Dong et al., 2019). Therefore DA levels at wake onset may determine the length of the following wake period, and subsequently influence sleep/wake cycles. While in the VTA, there is also higher DA activity during wake, with an increase in population burst firing compared to NREM sleep, however the activity of these neurons is further increased during REM sleep compared to wake (Dahan et al., 2007; Eban-Rothschild et al., 2016), suggesting distinct roles of these two DA regions in arousal. The increase in firing of VTA neurons is accompanied by an increase in extracellular DA levels in their projection target regions, the nucleus accumbens and prefrontal cortex (Léna et al., 2005), leading to sleep/wake-state dependent changes in DA levels.

However, not all DA populations are arousal-inducing, there is also evidence for a sleep-promoting role. Selective lesions of DA neurons in the substantia nigra, which project to the dorsal striatum, leads to increased wakefulness, reduced NREM and REM sleep, and fragmentation of sleep/wake states in rats, while optogenetic stimulation of these neuron terminals leads to increased NREM sleep (Qiu et al., 2016). Furthermore, low doses of D₂ receptor (D₂R) agonists have been reported to induce sedative effects and increase sleep in humans and rodents (Monti et al., 1989; Hamidovic et al., 2008; Laloux et al., 2008). Although this may be due to activation of pre-synaptic autoreceptors and the subsequent inhibition of DA neurons in arousal-promoting regions (Monti and Monti, 2007).

Further evidence for the role of DA in arousal is provided by the effects of stimulants (Boutrel and Koob, 2004). Genetic and pharmacological studies have demonstrated that the primary mechanism underlying the wake-inducing action of the stimulant modafinil is an increase in DA (Wisor et al., 2001; Qu et al., 2008). This is the preferred treatment for hyper-somnias such as narcolepsy, and is thought to act through inhibition of DA reuptake. Similarly, selective DA reuptake inhibitors have also been shown to promote wakefulness in rodents and dogs (Nishino et al., 1998; Luca et al., 2018).

Collectively these studies have demonstrated that DA signaling has a key role in the regulation of sleep/wake states. Generally it serves a wake-promoting role, however its effects are bidirectional and can promote both wakefulness or sleep depending on the brain region and receptor subtypes involved. Its role in sleep/wake control is therefore complex, and the precise molecular and neuroanatomical mechanisms through which it acts are still unclear, yet it is evident that normal DA signaling is important for stable sleep/wake cycles.

The sleep/wake cycle is also under circadian control to ensure that its timing is aligned with the external light/dark cycle, and the two systems are closely linked. There is growing evidence that DA is also an important modulator of the circadian system.

Dopamine signaling is involved in the functioning of two key components of the circadian system: the retina and the SCN. In the retina, it is necessary for functions including light adaptation and visual acuity (Jackson et al., 2012). Importantly, it is also one of the main modulators of retinal circadian activity (Green and Besharse, 2004) and can influence the core circadian clock, controlling the amplitude of rhythmic clock gene expression (Yujnovsky et al., 2006). There is also evidence that DA via D₂R drives rhythmic expression of the photopigment melanopsin in intrinsically photosensitive retinal ganglion cells (ipRGCs; Sakamoto et al., 2005), which are the primary photoreceptors responsible for light input to the SCN (Peirson and Foster, 2006). Therefore retinal DA is a key modulator of non-visual light detection, and may be important for photic entrainment of the master clock. Indeed, Doi et al. (2006) demonstrated that genetic deletion of D₂R in mice leads to a deficient masking response to light, whereby light suppresses locomotor activity during the dark period when the mice are usually active. However, the D₂R null mice did exhibit normal entrainment to a light/dark cycle, and photic responses in the SCN and pineal gland were intact. These data suggest that D₂R is required for behavioral responses to light, but not for other non-visual photic responses such as photic entrainment of the SCN. Although in this study D₂R was deleted throughout the brain, so while retinal DA signaling may be involved, the involvement of DA signaling in other regions cannot be ruled out.

In the SCN itself, there is a high number of dopaminergic neurons, with Drd1-expressing neurons comprising approximately 60% of cells in rodents (Smyllie et al., 2016). Early studies demonstrated that DA is required for entrainment of the developing SCN, acting to synchronize fetal-maternal circadian rhythms (Mendoza and Challet, 2014). More recently, studies have suggested that DA modulation of the central clock persists into adulthood (Jones et al., 2015; Landgraf et al., 2016; Smyllie et al., 2016; Grippo et al., 2017). Studies report that Drd1-expressing SCN neurons are able to entrain and set the period of circadian rhythms in mice (Jones et al., 2015; Smyllie et al., 2016). However, this does not provide direct evidence for a role of DA itself, as the Drd1-positive cells overlap with the majority of arginine vasopressin (AVP) and vasoactive intestinal peptide (VIP) neurons, other key SCN subpopulations involved in entrainment.

However, Grippo et al. (2017) did demonstrate DA modulation of the adult SCN clock via D₁R signaling. They showed that DA signaling in the mouse SCN is required for normal re-entrainment following a jet lag-like phase shift in the light/dark cycle (Grippo et al., 2017). Retrograde tracing suggested that this was mediated by direct DA innervation from the VTA, with enhancement of this DA input leading to accelerated re-entrainment. Interestingly, this facilitation of resynchronization by DA required light input, in line with previous studies that showed that administration of D₁R agonist alone during constant conditions does not elicit any phase shifts in hamster (Duffield et al., 1998; Grosse and Davis, 1999). This suggests that direct midbrain DA innervation of the SCN can modulate the light-responsiveness of the master clock, enabling quicker re-entrainment. Importantly, this study also demonstrates that the level of DA tone can have a direct impact on the central clock, and therefore provides a mechanism whereby aberrant DA signaling, such as an inappropriately timed elevation in DA tone, could influence the master clock. However, this study did not detect changes in several circadian parameters following modulation of SCN DA signaling, such as the period length during constant conditions and the phase angle of entrainment. Therefore, it appears that while DA via D₁R has a modulatory role of the light-responsiveness of the clock, it is not necessary for its normal functioning. Studies looking at additional measures of the circadian clock other than locomotor activity are required to elucidate the involvement of DA in the clock, given the essential role of DA in motor function (Ryczko and Dubuc, 2017).

There is also evidence for a role of DA in clock regulation outside of the SCN. As previously discussed here, DA is a major modulator of the retinal clock, in addition to this it also acts on the striatal clock. The dorsal striatum is heavily innervated by dopaminergic projections from the substantia nigra. Evidence from in vivo pharmacological and lesioning studies suggest that DA signaling via D₂R is required for an intact rhythm in Per2 expression in the dorsal striatum (Hood et al., 2010). Extracellular DA levels were found to oscillate here over 24 h with a peak during night, preceding the peak in Per2 expression. However, this was measured in rats maintained under a light/dark cycle, so it is unclear whether this is an endogenous circadian rhythm that persists under constant darkness or whether it requires light input.

There is also evidence that DA can regulate circadian rhythms in hormone production and behavior. It is thought to be necessary for the rhythmic release of prolactin in the arcuate nucleus of the hypothalamus (Bertram et al., 2010), and exerts an inhibitory modulation of melatonin synthesis in the pineal gland (González et al., 2012). Melatonin is an important output of the SCN for the regulation of circadian and seasonal rhythms throughout the body, and also feeds back to the SCN itself to modulate the central clock. DA is also thought to control rhythms in activity; lesioning of dopaminergic regions leads to a lengthening in the period of wheel-running and drinking activity during constant conditions (Isobe and Nishino, 2001; Tanaka et al., 2012). However, as DA itself is under circadian control (Chung et al., 2014) as discussed below, it may be that in these cases DA is an intermediate signal between the circadian clock and clock-controlled processes, rather driving the circadian rhythms itself. Nevertheless, it appears to be an important component in the regulation of circadian-controlled processes.

However, DA is in fact able to act independently of the master clock to regulate rhythms in behavior through DA-driven pacemakers which are entrainable by two non-photic stimuli: food and methamphetamine. The DA-enhancing psychostimulant, methamphetamine, can induce circadian rhythmicity in locomotor activity in the absence of an SCN, via the methamphetamine-sensitive circadian oscillator (MASCO; Tataroglu et al., 2006). This finding led to the identification of a ‘dopamine ultradian oscillator’ (DUO), whereby ultradian oscillations in DA drive an ultradian rhythm in locomotor activity, which persists in Bmal1 null mice and is therefore independent of the circadian clock (Blum et al., 2014). Interestingly, this study used Slc6a3-/- mice to model a hyper-dopaminergic phenotype, due to their deletion of DA transporter (DAT) for DA re-uptake, a potential model for schizophrenia. Elevated DA in these mice led to aberrant ultradian activity rhythms with fragmented daily activity, which mirrors the disruptions to the rest/activity cycle observed in patients with schizophrenia (Wulff et al., 2012), suggesting a causal role for DA in these symptoms. This highlights the central role that DA has in the regulation of activity rhythms, and importantly, demonstrates a potential mechanism through which aberrant DA can lead to abnormalities in rest/activity rhythms seen in schizophrenia.

Dopamine is also necessary for the activity of the food entrainable oscillator (FEO). In the absence of an SCN this drives food anticipatory activity (FAA), whereby an increase in locomotor activity precedes mealtimes by 1 to 3 h (Liu et al., 2012; Gallardo et al., 2014). DA signaling via D₁R in the dorsal striatum mediates this rhythmic activity, suggesting that DA is required for daily rhythms in feeding activity. While the underlying mechanisms are still being elucidated, these SCN-independent oscillators likely feed into the SCN to regulate rhythmic behavior in concert with the central circadian clock.

There is evidence that DA interacts directly with other neurotransmitter systems that are involved in the regulation of sleep and the circadian clock, namely adenosine and glutamate via NMDA receptors. There is an antagonistic interaction between DA and the G-protein-coupled adenosine receptors via both D₁R and D₂R. The D₁R heteromerize with the A₁R, resulting in an inhibitory action of adenosine to dampen DA receptor response (Ciruela et al., 2011). Whereas D₂R form a complex with the A_2A subtype of adenosine receptors, resulting in a reciprocal antagonistic interaction whereby D₂R activation can also dampen the A_2AR response (Figure 2A). Activation of D₂R is thought to inhibit A_2AR activation of the cAMP-PKA pathway, therefore leading to inhibition of adenosine-induced protein phosphorylation and gene expression, such as of immediate-early genes (IEGs; Ferré et al., 1997; Svenningsson et al., 2000). Adenosine is involved in sleep/wake control; the accumulation of extracellular adenosine during wake and its dissipation in sleep is thought to underlie homeostatic sleep drive, or “sleep pressure” (Landolt, 2008). Furthermore, the stimulant caffeine acts primarily through antagonism of A_2AR (Huang et al., 2005). Therefore, DA-mediated modulation of adenosine signaling through receptor heteromerization is a potential mechanism through which DA can modulate sleep and wakefulness, although the role of this interaction in sleep control is yet to be elucidated.

It is well-established that DA modulates NMDA receptor signaling; the two receptors can form functional heteromers and their downstream signaling pathways also interact at multiple levels (Perreault et al., 2014; Klein et al., 2019). Their interaction is complex; DA can have both a potentiating and inhibitory effect on the NMDA receptor response. Activation of D₁R can enhance NMDA-mediated responses (Cepeda et al., 1993, 1998; Levine et al., 1996). Evidence suggests this is through D₁R-dependent phosphorylation of the NR2B subunit of the NMDA receptor, which is thought to enhance NMDA-dependent calcium influx and therefore downstream responses such as ERK signaling (Figure 2B; Pascoli et al., 2011; Murphy et al., 2014). On the other hand, heteromerization of both the D₁R and D₂R with NMDA receptors leads to inhibition of NMDA currents (Figure 2C; Perreault et al., 2014). NMDA receptor activation is one of the primary mediators of light input into the SCN and therefore a key mechanism for photic entrainment of clock gene expression in the master clock (Golombek and Rosenstein, 2010). Together, these represent intriguing pathways for future investigation through which aberrant DA signaling may impact neurotransmitter systems that are central to the regulation of the sleep and circadian system.