The role of the thalamus in sleep, pineal melatonin production, and circadian rhythm sleep disorders


Address reprint requests to James E. Jan, Diagnostic Neurophysiology, BC Children’s Hospital, 4500 Oak St., Vancouver, BC V6H 3N1, Canada.


Abstract:  The thalamus has a strong nonphotic influence on sleep, circadian rhythmicity, pineal melatonin production, and secretion. The opening of the sleep gate for nonrapid eye movement sleep is a thalamic function but it is assisted by melatonin which acts by promoting spindle formation. Thus, melatonin has a modulatory influence on sleep onset and maintenance. A remarkable similarity exists between spindle behavior, circadian rhythmicity, and pineal melatonin production throughout life. Together, the thalamic and chronobiological control of sleep leads to a new and improved understanding of the pathophysiology of circadian rhythm sleep disorders and also of the principles of sleep hygiene interventions.

Circadian and homeostatic mechanisms in sleep

Borbely [1] introduced theories of how circadian and homeostatic mechanisms influence sleep, which, subsequently, have been widely accepted. Sleep–wake cycles are controlled by circadian rhythm oscillators in the suprachiasmatic nuclei (SCN). In contrast, the homeostatic process regulates sleep need (propensity) which increases during the day and decreases during sleep depending on the amount of mainly nonrapid eye movement sleep (NREM) a person previously had [2]. The homeostatic mechanisms are able to overrule the circadian system because sleep deprivation leads to longer sleep. Adenosine is one of the main sleep-inducing factors in homeostasis [3], although there are also changes in neuronal activities associated with sleep propensity [4].

Over time, researchers have been increasingly able to explain various aspects of sleep, but the neurological basis of these highly complex mechanisms had remained unclear. Within this framework of circadian rhythms and homeostasis, the photic input into the SCN and the functions of these small structures responsible for circadian rhythms were strongly emphasized. However, the role of SCN, especially when it was viewed in isolation without their extensive neurological and neurochemical network, failed to explain the pathophysiology of circadian rhythm sleep disorders (CRSD).

In recent years, research into the anatomic and physiological aspects of circadian rhythmicity has blossomed. Major progress has been made in understanding molecular biology of the SCN [5] and at the same time, knowledge about the roles of the thalamus and other brain structures in sleep have expanded [6–11]. Sleep is a neurological function and the process of wakefulness and sleep involves the entire central nervous system. Similarly, modulators of melatonin secretion and the effects of this indoleamine are more complex than previously thought. Melatonin receptors, clock genes, sleep deprivation, the timing of circadian rhythms, a vast number of neurochemicals, hormones, the environment, and various neurological structures and other factors are acting as a symphony to influence melatonin physiology and the sleep–wake cycle [2, 5, 12–14].

The role of thalamus in sleep

The thalamus plays a critical role in processing, integrating, correlating, and relaying sensory and motor information [9]. Major auditory, visual, and somatosensory pathways, together with inputs from the limbic structures, the brainstem, and cerebellum have their final subcortical relays in this highly complex paired structure, which contains over 30 nuclei. The thalamic regions, particularly most of the ventral tier nuclei of the lateral area, have reciprocal connections with specific regions of the cerebral cortex [9]. The bidirectional thalamocortical (TC) network modulates the flow of sensory and motor information to and from the cerebral cortex. The TC projection is composed of excitatory relay and inhibitory neurons, neurons of the thalamic GABAergic inhibitory reticular (RE) and perigeniculate nuclei and local circuit interneurons. The TC relay neurons are located in thalamic nuclei and are associated with sensory, motor, and/or high order roles. The RE neurons are in a sheet-like nucleus that covers most of the rostral, lateral, and ventral parts of the thalamus. During early NREM sleep (its later and deeper stages are known as slow wave sleep), the RE neurons generate low- and high-frequency spindles, which represent synchronized oscillatory electrical activity between the RE and TC neurons. As the TC neurons undergo progressive hyperpolarization, the synaptic responsiveness is reduced and the bidirectional flow of information is interrupted.

The spindles play an active role in inducing, maintaining, and advancing NREM sleep towards deeper stages [6, 15–17] (Fig. 1). Absent, poorly formed, or a reduced number of spindles reflect inadequate protection of NREM sleep which in polysomnographic studies, usually is manifested as delayed sleep onset and frequent awakenings (impaired sleep maintenance). While the sleeping brain continues to be active and reactive, it loses its ability to be integrative. This feature is also important for the fading of consciousness in the early stages of sleep [18].

Figure 1.

 A normal stage 2 NREM sleep recording of a 13 yr old girl. The thin arrows point to spindles while the thick arrows indicate vertex waves in each hemisphere (left panel). A normal REM recording of a severely sleep deprived 8 yr old boy, just after falling asleep. The characteristic low voltage background is associated with marked eye movement artifacts (right panel).

The thalamic structures also play an essential role in generalized epilepsy. During absence of seizures, similarly to NREM sleep, the back and forth flow of information is arrested in the thalamus and, as a result, individuals temporarily become unaware of their environments [6, 7, 19]. This thalamic mechanism has been termed ‘spindle gating’ [7]. This research in epilepsy has been important because it has facilitated the development of concepts for NREM sleep gating mechanisms.

Intergeniculate leaflet and suprachiasmatic nuclei

The majority of research, conducted over the last 20–30 yr, focused on the effects of light exposure on the functions of the SCN located above the optic chiasm in the anterior hypothalamus. These small nuclei each contain about 10,000 highly specialized neurons. In contrast to the intensive investigation on the importance of photic input to the SCN, studies on nonphotic modulators of SCN activity were much less frequently studied until recently.

Extensive research based on neuronal tracers and electrophysiological techniques show that the thalamus has prominent nonphotic and even some photic inputs into the SCN through the polysynaptic pathway that includes the intergeniculate leaflet (IGL) [5, 20, 21]. The IGL has amazingly widespread bilateral and ipsilateral connections to the brain and the brainstem [5]. It contains excitatory and inhibitory modulators which originate from numerous areas of the neocortex, including the visual system. In addition to terminating on neurons in the SCN, the IGL also receives photic input from the retinohypothalamic pathways. These tracts originate from specialized ganglion cells of the outer retina and even from classical photoreceptors and appear to project widely to possibly all the retinorecepient regions of the visual system. Thus, the axons of the IGL neurons contribute to the photic regulation as well. Via these diverse connections, the thalamus seems to make major contributions to circadian sleep regulation and wakefulness.

The SCN contains specialized neurons which drive circadian rhythms in mammals, including humans [5, 22]. Some of the circadian oscillations are intrinsic to the SCN but they are entrained by daily environmental cues (Zeitgebers) provided by the retinohypothalamic and IGL projections to ensure that they function in phase with the day and night cycles. The retinohypothalamic pathways contain mainly an excitatory photic influence [23–25]. The afferent and efferent projections to and from the SCN are much more complex than previously thought. Inputs to the SCN are provided by neurons from the forebrain, midbrain, hypothalamic, and numerous other structures [5, 25]. The SCN has synaptic efferent connections to the hypothalamic, ventrolateral preoptic nuclei, and several other areas and also influence nearby tissues by diffusible substances [8].

Both the retinohypothalamic and IGL pathways terminate in the SCN. However, a significantly greater number of neurons respond to the IGL than to photic stimulation from the retinohypothalamic tract [25, 26]. The greater representation of IGL sensitive neurons underlines the importance of the nonphotic entrainment of circadian rhythms. The SCN is not homogeneous in structure or function. Portions of the neurons respond to photic stimuli, melatonin, or IGL inputs and only some cells exhibit intrinsic rhythmicity. However, the end result is such that the neurons are synchronized and oscillate with the mean period of the cell population as a whole [21, 22]. The generated oscillations are transferred to the pineal gland through a diverse multisynaptic pathway that eventually involves the peripheral sympathetic nervous system, which modulates melatonin production and secretion [27]. While the influence of photic and nonphotic regulators of the SCN on pineal melatonin secretion is well studied, melatonin is also secreted by other cells in a wide variety of tissues where paracrine and autocrine functions are present [13, 28].

NREM and REM sleep

Sleep patterns, like other neurological functions, mature with age. In premature infants, the electrographic features remain undifferentiated until 32 wk of gestational age. Thereafter, NREM and rapid eye movement (REM) sleep states can be distinguished by their behavioral correlates. In normal premature infants, recognizable patterns of NREM and REM are present by 34 wk. At 2 months of age, after full term birth, these two sleep states are well differentiated [29] (Fig. 1).

One of the hallmarks of NREM sleep is the appearance of sleep spindles immediately after entering early NREM sleep (stage 2) from drowsiness. This transition occurs within seconds, and represents the opening of the NREM sleep gate [7]. Spindles play a critical role in the induction and maintenance of sleep. The various modalities of stimuli involving different areas of the neocortex influence the location of spindles. This suggests that spindle activity not only prepares the entire brain for sleep but also prepares individual regions while taking into consideration the level of previous sensory inputs [30]. The spindles originate from the thalamus but their expression depends on the intactness of the cerebral hemispheres. This is a very important point because it explains why so many individuals with disturbed central nervous systems have persistent sleep difficulties. NREM sleep can also be assisted by sleep active neurons in the preoptic and basal forebrain structures; still this sleep state is a thalamic function [11].

The cholinergic onset and aminergic offset neurons and the connecting circuits within the pons and medulla influence REM sleep. The neural and neurotransmitter control of NREM and REM sleep are complex and multiple areas of the brain are involved. Although the two systems are separate, they are closely integrated [8, 31, 32]. Faulty transitions from NREM to REM may occur and give rise to night terrors, sleep walking, and other parasomnias [33, 34].

Similarities between pineal melatonin production and spindle formation

Spindle formation has a surprisingly close relationship with circadian rhythmicity and melatonin production [35, 36]. In mammals, including humans, the SCN is already sufficiently mature in early prenatal life to respond to maternal melatonin and light [37]. In spite of early maturity, circadian rhythmicity and melatonin secretion only develop at 2–3 months of age, which has been attributed to a delay in the maturation of the widespread SCN neural circuitry [38, 39]. It is important to note that circadian rhythmicity, pineal melatonin secretion, and the transition from rudimentary to well-developed spindle formation appear at the same time. Spindle evolution, like melatonin production, is also an index of neural development [29, 40]. A delayed appearance of spindles is associated with impaired brain maturation, delayed onset of circadian rhythms, and deferred melatonin secretion [41].

Normative data of spindle maturation show that, at the age of 5 yr a plateau is reached, which remains to around 16 yr of age [16]. Thereafter, with aging and in the elderly, spindle density, amplitude, duration and frequency are reduced. Again, here is a remarkable similarity to melatonin production and secretion, which parallel spindle activity throughout life. During the night, low frequency spindles peak close to the height of melatonin levels and/or to the nadir of body temperature [42]. Peak nocturnal blood melatonin levels reach their maximum at 4–5 yr of age. High melatonin levels remain constant until puberty, after which there is a steady decline, and as a result, many elderly individuals have measurable relative melatonin deficiency and associated sleep difficulties [43]. It has been proposed that the decline of melatonin secretion in the elderly is the cause of the spindle changes [44]. Indeed, exogenous melatonin in the elderly enhances spindle activity [45]. Pineal melatonin secretion and spindle formation are currently the most accurate circadian phase markers in humans and they are a reliable means of estimating the maturation of circadian mechanisms [15,16].

Spindles and melatonin in brain disorders

A large number of diffuse brain disorders are associated with impaired spindle activity [16, 46]. This is a very common finding when individuals with neurological disorders are monitored with electroencephalograms. These patients frequently have disturbances in their sleep initiation and maintenance as a result of impaired pineal melatonin secretion. Severe neurodevelopmental disorders with attenuated blood melatonin levels are the commonest cause of persistent sleep difficulties in children [46]. Again, a shared mechanism is suggested between spindle formation and pineal melatonin secretion. When a melatonin deficiency is present, administration of melatonin may correct sleep difficulties, except in profoundly brain damaged individuals whose SCN may also be functioning suboptimally [47]. In contrast to bilateral brain damage, destruction of only one hemisphere does not abolish spindle activity and melatonin secretion is generally normal.

A paramedian thalamic stroke presents with decreased spindles, sleep fragmentation, and a number of other neurological deficits [48]. Analysis of sleep architecture shows that only NREM sleep is disturbed. This acquired sleep disorder is most severe with bilateral rather than unilateral damage. Unilateral thalamic stroke does not reduce sleep spindles ipsilaterally, rather it produces a bilateral diminution of their number [49]. Another neurological disorder, fatal familial insomnia, is caused by a prion disease due to a mutation in a gene encoding the prion protein. In this condition, the most severe tissue changes involve the thalamus. Patients develop a marked diminution in their spindles, absence of NREM sleep, severe sleep fragmentation, interestingly loss of circadian rhythmicity, and eventual death [50, 51]. Portaluppi et al. [50] have followed two individuals with fatal familial insomnia with polysomnographs and carried out repeated blood measurements of melatonin. NREM sleep was never recorded, REM sleep was not abolished and, as the disease progressed, plasma melatonin levels waned.

Complete destruction of both SCN without widespread damage to other brain structures is rare but, when it occurs, spindle formation continues to be present (J.E. Jan, personal observation). Severely damaged SCN lead to loss of circadian rhythmicity, marked sleep fragmentation, and melatonin deficiency, but the ability to produce NREM or REM sleep states is not abolished. Melatonin administration or hypnotics cannot restore the circadian rhythmicity.

DLMO and NREM sleep gate

Dim light melatonin onset (DLMO) is generally observed 2–3 hr prior to habitual sleep. DLMO is an approximate onset of melatonin synthesis and secretion [52]. Both DLMO and sleep initiation (opening of the sleep gate) are persistently delayed in many children with attention deficit hyperactivity disorder [53, 54]. This is an important observation because, without considering the modulatory influence of the thalamus on the SCN, these delays cannot be explained. Research has shown that the RE neurons can only become hyperpolarized and produce spindles when the various excitatory inputs to the thalamus diminish to a certain level [40]. The promotion of sleep hygiene in children with attention deficit hyperactivity disorder is based on the intent to reduce excitatory stimuli prior to habitual bedtime, as it is well known that it may shorten sleep onset. This strongly suggests that DLMO is influenced by thalamic inputs.

Functional magnetic resonance imaging studies show that endogenous or exogenous melatonin plays a major role in preparing the brain to fall asleep at the habitual bed time, or during the day [55, 56]. Other researchers have drawn similar conclusions based on electroencephalographic studies [36]. Daytime naps, with or without exogenous melatonin or following sleep deprivation, are also associated with elevated melatonin levels and the appearance of spindles [35, 45, 57, 58]. Hypnotics and anaesthetics induce sleep by the same process, but anaesthetics more rapidly lead to loss of consciousness [10]. Anaesthetics and large doses of hypnotics can cause deactivation of the neocortex and death. However, even huge doses of melatonin do not cause deactivation of the cortex as melatonin is not a hypnotic drug.

Sleep deprivation affects melatonin production and the circadian and homeostatic systems [57]. It may lead to earlier DLMO and earlier opening of the sleep gate. In turn, melatonin and the circadian and homeostatic systems influence spindle formation [58]. Sleep deprivation, which initially deprives individuals of NREM sleep, enhances the potency of anaesthetics, again suggesting the involvement of thalamic structures [10].

Melatonin influences REM sleep as well. In individuals with disturbed and reduced REM sleep, it increases the percentage of REM and improves daytime functioning [59]. In REM sleep behavioral disorders, melatonin treatment partially restores atonia as measured by polysomnography [60]. Infants and excessively sleep deprived individuals may enter REM sleep directly, without first entering NREM. Thus, mechanisms in the brainstem regions, which control REM sleep are responsible for the opening of another type of sleep gate. It is unclear why markedly sleep deprived individuals occasionally enter REM first rather than NREM sleep. In NREM sleep, the bidirectional flow of information is arrested within the thalamus, whereas most areas in the brain including the thalamus remain active in REM sleep. In this sleep state, short-term memory is transferred from the hippocampus and incorporated into memory banks via various pathways which may or may not pass through the thalamus. During this process, dreaming occurs [61, 62] which may be more vivid during melatonin treatment. Functional neuroimaging confirms that during NREM sleep the thalamus is deactivated whereas in REM the thalamic nuclei are active [63].

Altogether, the evidence is very strong that circadian rhythmicity, pineal melatonin production, spindle activity, and the opening of the NREM sleep gate are closely linked. The SCN and the thalamus via IGL influence melatonin secretion, which feeds back to these structures. These associations are not surprising because both the spindle activity and IGL originate from the thalamus [5, 25]. In spite of feedback mechanisms, long-term melatonin therapy, even at high doses, does not suppress endogenous melatonin secretion [64], although it may alter the timing of the rhythm [65].

Thalamic influences on circadian rhythm sleep disorders

Whenever a child has CRSD or other sleep difficulties, initial attempts are usually made to improve sleep habits. Sleep hygiene (promotion of sleep health) is defined as optimal environment, scheduling, sleep practice and physiological sleep-promoting factors. CRSD, which constitute a large group of sleep disorders, is presently defined as dissociation between the sleep–wake behavior and the environment. An important characteristic of these disorders is that they are associated with inappropriately timed or deficient pineal melatonin production and secretion [46]. The most common types of CRSD are jet lag and shift work, delayed sleep phase syndrome and day/night reversals, impaired sleep maintenance, advanced sleep phase onset and persistent early morning awakenings. Surprisingly, in the literature there is an absence of a clear physiological explanation why sleep hygiene therapy may be beneficial within a neurological framework.

In order to fall asleep, the ascending neuronal pathways in the reticular formation, which contribute to cortical arousal, must be inhibited [10]. The cerebral cortex also sends excitatory projections to the thalamus [66] and they must be suppressed as well. Excitatory inputs prevent the hyperpolarization of the RE neurons and, as a result, they cannot enter into a synchronized oscillatory state and induce spindles, which together with melatonin facilitate the opening of the sleep gate. Like the SCN, the thalamus integrates and synchronizes various sensorimotor inputs. Spindle oscillation can only occur when the summation of all integrated stimuli is reduced to allow hyperpolarization. In this process, even individual excitatory inputs can delay the DLMO and sleep onset. Emotional upheaval via the limbic connections to the thalamus, sitting in front of a computer or TV prior to habitual bedtime are associated with increased excitation leading to depolarization of TC and RE neurons [67]. Trying to sleep with an erect posture or vigorous nocturnal exercise, also produce depolarization of the thalamic nuclei [68–70]. Therefore, sleep hygiene practices, with or without melatonin therapy, should aim to eliminate all activities which stimulate the brain.

The chronobiotic effects of light and melatonin have been used to treat sleep disruption due to jet leg and shift work [71]. The timing of light exposure and its intensity, the dose of melatonin and time of administration, the degree of sleep deprivation, the age and neurological status of individuals all influence the chronobiotic effect on the SCN. However, the thalamic input in circadian entrainment is rarely mentioned even though it is substantial.

In individuals with attention deficit hyperactivity disorder who have a high arousal level, the DLMO, the preparation of the brain for sleep and the opening of the sleep gate by spindle formation are frequently delayed [53, 54]. At times, the onset of DLMO is appropriate but the sleep onset is delayed because of exposure to stimulatory activities later at night. Melatonin supplementation at bedtime, which advances sleep onset, likely promotes earlier sleep onset through both the SCN and thalamus.

As discussed earlier, persons with diffuse brain damage and impaired spindle formation often exhibit a relative melatonin deficiency due to impaired thalamic input to the SCN. As a result, these individuals tend to have difficulties with sleep onset, sleep maintenance and persistent early morning awakenings [46].

Free-running sleep–wake rhythms associated with total blindness appear to be more common in adults than in children, as total loss of sight is less common in younger individuals [72]. Melatonin administration at the desired habitual bedtime can anchor the shifting sleep pattern [73]. In the absence of photic input to the SCN, the relative strength of IGL becomes stronger. As melatonin modulates only the optic nerve input to the SCN and not IGL [25] and in view of the influence of melatonin on the thalamus, the anchoring could be achieved via the thalamus. Not all totally blind children exhibit this sleep disorder perhaps, because the more numerous nonphotic neurons may compensate for the lack of photic input. Also, it is well known that in congenital blindness the neurons, which would serve visual function do not atrophy, but change their ‘alliance’ to other sensory stimuli [74]. Therefore, the SCN neurons that respond to photic stimuli might change their ‘alliance’ to IGL input. Research still needs to show if these basic neurological principles and assumptions apply to SCN physiology.

In addition to the chronobiological principles, the thalamic influences on sleep are helpful in clarifying the pathophysiology of CRSD. Researchers must be careful in attributing the causes of CRSD to the SCN alone because in the majority of instances these small, but complex neurological structures, are simply adjusting to changes to photic and nonphotic inputs.

During the last 30 yr, phenomenal progress has been made in sleep research. We have come a long way from Borbely’s initial circadian and homeostatic concepts [1], which still hold true. Since the mechanisms involved in sleep are extremely complex, it is not surprising that the various scientific disciplines view sleep in different ways. Sleep in humans and animals is studied by molecular biologists, chronobiologists, electroencephalographers, anatomists, psychologists, sociologists, economists, educational specialists, medical and allied health practitioners, geneticists, and other professionals. The challenge is to combine the acquired knowledge into a balanced and meaningful viewpoint rather than overemphasizing one aspect of sleep research over the other. It must also be remembered that the understanding of the neurological basis of sleep is rapidly evolving and what appears to be clear today may not be accepted in the future.