Paradoxical sleep mechanisms

Since there are excellent recent reviews concerned with paradoxical sleep (PS) or REM sleep mechanisms (1)(2) (3)(4), it is unecessary to give an overwiew of these reviews.
However the following topics need some discussion either because they have not yet received any unequivocal solution or because they were not examined in these reviews :

The review of transections, carbachol injections or lesions does not permit to answer unequivocally to this question :

1) It has been known for more than 30 years that a brain stem transection (A or B) in front of the pons does not suppress the periodical appearance of PS caudally to the transection. Moreover, there is no evidence of any signs of PS in front of the transection (5).

Transection at the junction of the spinal cord and medulla (E)(encephale isolé preparation) does not prevent all the signs of PS from occuring rostral to the cut, although atonia in muscles innervated by spinal motoneurons is disrupted by the lesion (6).

From the above it is safe to conclude :

That the rhombencephalon (caudal to the midbrain and rostral to the spinal cord) is sufficient for PS.

That there is no "humoral" link from the lower brain stem which would be able to activate the forebrain when PS occurs below a total prepontine transection

2) After a transection between the pons and medulla (D) there was no signs of PS caudally to the transection (no muscle atonia, no burst-pause discharge characteristic of PS in the unitary activity of the medial medullary neurons). However, according to Siegel et al. (7), there was a state "ressembling PS" rostral to the section characterized by a desynchronized EEG with PGO spikes occuring in irregular bursts. REM often accompagnied the PGO bursts but they were also dissociated from them. Curiously this state could last for hours. However, after similar transection, Webster, Friedman and Jones (8) found a total disruption and even a loss of PS in the brain rostral to the section.

3) Finally, after transection though the middle of the pons (section C) there was no evidence of any specific signs of PS on both sides of the transection even in chronically maintained animals (2). According to these data, it is evident that the pons might be sufficient to generate PS since there was no proof that the medulla alone contributes to PS signs (like atonia) and since "putative PS" could occur rostrally to a ponto medullary transection.

B) Carbachol microinjections in the pons

Carbachol microinjections in the medio dorsal pontine tegmentum (region of the locus coeruleus alpha and perilocus coeruleus alpha) induced consistently a high amont of PS with short latency (less than 5 min.)(9). However, after transection of the brain stem (C or D), the same microinjections were no longer able to induce PS even in chronically transected cat. In these preparations, Vanni-Mercier et al. (9) have also observed PGO activity during long period of desynchronized EEG. However, they believed that these periods represent waking since they could track the eye following reflex. Therefore, according to these authors, the pons is insufficient for PS generation and thus connection between the pons and the medulla are necessary for PS.

The cholinergic input to pontine "executive PS" structure are known to originate both from the dorsal pontine tegmentum and from the medulla. In the transected medio pontine or bulbo pontine cats the pontine cholinergic input is preserved while the bulbar one is suppressed. This could explain the disappearance of PS in transected cats. However, carbachol microinjection in the pons should have mimik the bulbar cholinergic input. This demonstrates that the cholinergic stimulation is insufficient to induce PS and suggest that "complex reciprocal" ponto-bulbar connections are indispensable for its induction (see figure 2).

C) Lesion studies

1) Coagulation :

The destruction of both cells bodies and pathways located in the dorso lateral portion of the pontine tegmentum (L 2,5 to L 4), in the nucleus reticularis pontis oralis (or in the locus coeruleus alpha and perilocus coeruleus alpha) may suppress PS for months (10). PS can also be suppressed by coagulations of the ventrolateral part of the nucleus reticularis pontis oralis and caudalis which leave intact the locus coeruleus area. On the contrary, coagulation of the medial part of these nuclei do not suppress PS. These results suggest that the connections between pontine and bulbar neurons which are essential for PS are located laterally (11)(12).

2) Neurotoxic lesion :

Extensive kaïnic acid lesion of the giganto cellular tegmental field neurons are without effect upon PS (11). However, kaïnic acid destruction of the entire locus coeruleus complex led to PS suppression. Immunohistochemical studies of the lesion demonstrated that in cats with the most extensive destruction of cholinergic neurons of the dorso lateral pontine tegmentum, PS was eliminated for 2 to 3 weeks.

This result did suggest that cholinergic neurons of the dorso lateral ponto mesencephalic tegmentum may be critically involved in the initiation and maintenance of PS (13).

2) Recently the medial medullary reticular formation was destroyed with the neurotoxin quisqualic acid injected into the medullary giganto-cellular and magno-cellular tegmental field in the cat (14). Following the cytotoxic lesion which affected about 60% of the cells bodies of these area, there was a significant but mild reduction of PS (64% of baseline) during the first week. The most ped effect of the lesion was a substantial increase in the amplitude of nuchal EMG during both SWS and PS. This was associated with movements of head, neck and/or limbs during PS.

Therefore it was concluded that the neurons of the medial medullary reticular formation contribute to, but are not necessary for the generation of PS. In order to explain the discrepancy between the results of the bulbo-pontine transections which eliminate PS, and the results of medullary neurotoxic lesions which do not impair significantly PS, the following hypothesis have been proposed by Holmes and Jones (14) : Either the transection induced a retrograde alteration of pontine neurons responsible for PS or the neurotoxic lesion of the medulla was too restricted and did not affect medullary neurons responsible for PS (mostly in the lateral part of the medulla which cannot be destroyed by necessity for the survival of animals).

Another hypothesis, which would explain the lack of effect of microinjections of carbachol in the pontine tegmentum in pontobulbar transected cats, would be the following :

Cholinergic input to executive PS pontine cells would come mainly from the pons. Cholinoceptive "executive PS pontine cells" would have to stimulate non cholinergic bulbar neurons which in turn would activate other pontine non cholinoceptive executive cells. Thus, in normal animals, carbachol would activate both descending ponto bulbar neurons, then bulbo pontine non cholinergic neurons. In transected animals, carbachol would not mimik the unknown bulbo pontine neurotransmitter.
In summary, it is still impossible to answer positively to the question : is the pons both necessary and sufficient to generate PS ? But It is evident that we can answer negatively to the question : Are the cholinoceptive cells of the pons sufficient for PS ?

II - Is the rebound of paradoxical sleep dependent upon "stress" ?

The concept of paradoxical sleep rebound was initially proposed by DEMENT (15) to describe the increase of PS which follows its instrumental suppression. This concept has been extended to encompass any increase of PS above the baseline which follows alterations of the sleep waking cycle.
The main hypotheses which have been proposed for explaining PS rebound (PSR) can be summarized as follows :

• PSR is due to increase in PS factor(s) (presynaptic or hydraulic hypothesis)
• PS deprivation (PSD) induces alteration of receptor sensitivity which might explain PSR (postsynaptic hypothesis)
  - PSR is caused by various stressful situations (strain) which occur during PSD (Strain hypothesis).

The "hydraulic theory of the rebound" or the rebound considered as intrinsic to PS mechanism :

Instrumental privations of PS (either by the swimming-pool or flower pot technique, or by arousing animal or human subject at the onset of PS) are usually accompanied and followed by the following symptoms : On the one hand, an increased "need" or "pressure" for PS which are expressed by an increasing frequency in PS onsets (up to every min after a PSD lasting 24 h in the cat). On the other hand, an increase in PS above baseline level which occurs after the PSD. The duration of the rebound is usually proportional to the duration of PSD so that the deficit of PS (or the debt) is partially repaid (50 to 80%).

The discovery of the rebound of PS after PSD opened two new concepts :

1) It led to the belief that PS should be an important functional state since there existed a "need" for PS and since its suppression had to be "repaid" by a subsequent (compensatory or homeostasic ?) increase.

2) It led also to a first tentative of explanation of the mechanisms of the rebound which was rather similar to the hydraulic theory of instincts : During PSD there should exist the accumulation of some PS factor(s) (REM juice) (16). This accumulation would then be responsible for the PS "pressure" until the factor would be "used up" during the rebound. According to this theory the same factor(s) would be responsible both for PS and its rebound. This theory initiated the endless search for PS or sleep factor(s) in the brain or the CSF either after total sleep deprivation or PSD. Despite the discovery of many PS enhancing factors or peptides, it is evident that no specific factor, necessary and sufficient for PS has yet been isolated (17). There is indeed no experimental proofs that the putative suppression of any of these factors is followed by a total and long lasting suppression of PS.

Paradoxical sleep rebound and alterations of receptor sensitivity.

In the late 1970, PSD was suggested to improve depression in man. Since chronic antidepressant drugs could alter the sensitivity of monoaminoceptive receptors, the effects of PSD upon the sensitivity of these receptors were then studied. In some cases, the alteration of receptor sensitivity has been correlated with the alteration of behaviour or drug responsiveness which occur during PSD. Since PSR could sometimes be abolished by administration of receptor antagonists, it was also implicitely or explicitely admitted that PSR could result from the increase sensitivity of monoaminoceptive or cholinoceptive receptors (see Review in 18).

Paradoxical sleep rebound and "stress"

A - Paradoxical sleep rebound may not result obligatorily from the suppression of PS

The following experiments demonstrate that PSD is not a necessary prerequisite for the development of a subsequent PSR.

1) Instrumental (flower pot technique) or pharmacological (chlorimipramine-amphetamine) suppression of PS during 10 hr is not followed by a significant PS rebound in some genetic strain of mice which present normal amount of PS (BALB C and SEC) while it induces a large increase of PS in other strain (C57BL)(19).

2) If PS is suppressed in the cat by non aggressive (or stress provoking method), it is not followed by any PS rebound : Thus, if a cat is gently aroused at the very onset of PS and kept awake for 6 min (the average duration of PS), there is no evidence of higher frequency of PS onsets nor of PS rebound even if the PS suppression last for several hours (20).

3) When rats are given access to a running wheel, they can run up to 7,5 km per night ; this particularly high level of locomotion occupying a great part of the dark period is reported to be without significant influence upon sleep (21).

These experiments demonstrate that it is not PS suppression per se which is responsible for PSR. For this reason, it is likely that it is the qualitative aspect of PS deprivation (stress or strain) which could be responsible for PS rebound. This hypothesis is supported by the following experiments.

B - Paradoxical sleep rebound depends upon the stressful quality of the PSD

An immobilisation stress (IS) of 2 hours applied to rats at the beginning of the dark period (i.e. when the animals are more active) induces during the 10 consecutive hours a significant rebound of PS (+ 92 %) while slow wave sleep (SWS) is little affected. On the other hand, 2 hr of PSD realized by the platform technique or by maintaining the animals awake with a gentle handling do not significantly affect subsequent SWS or PS. It should be emphasized that both IS and PSD were associated with similar increase in corticosterone whereas only IS was followed by PSR. Finally when repetitive IS are inflicted to the animals (one IS of 2 h every 3 days) an attenuation of PS rebound is observed (22).

These data suggest that a qualitative aspect of the waking state as in intense stressful situation might be the source of hormonal or neural process inducing PS excess. This hypothesis is strengthened by the following experiments in which the neurohormonal link of "stress" have been excluded.

C - Paradoxical sleep rebound is eliminated by the suppression of the hormonal and the neuronal loop of stress

1) The destruction of both the arcuate nucleus and of the pituitary suppresses the PS rebound which follows an instrumental PSD in rats which have otherwise normal level of PS. However, the destruction of only the arcuate nucleus or hypophysectomy alone does not suppress the rebound of PS (23). For this reason, it is likely that either the pituitary-adrenal axis which is involved in the "classical stress mechanisms" or central neural system originating from the arcuate nucleus and innervating the brain stem are involved in the mechanisms of the "compensatory" rebound of PS.

2) In chronic pontile cats with isolated hypothalamo-pituitary island (in which both arcuate nucleus and pituitary are separated neuronally from the brain stem), the instrumental suppression of PS up to 10 h (obtained by delivering a foot shock at the onset of every PS) is never followed by any subsequent PS rebound (24).

Finally the problems of the mechanisms and the signification of stress related PSR should be briefly outlined.

1) PSR is apparently not related with the classical hypothalamo-pituitary-adrenal axis involved in stress since PSR still occurs in hypophysectomized rats (or cats) and since there is no correlation between the increase in corticosterone and PSR.

The following putative neuronal loop could be involved. Increase liberation of serotonin in the Arcuate Nucleus has been observed during stress with voltammetric technique. This could facilitate the release of some peptides located in the Arcuate nucleus and derivated from the proopiomelanocortine family. Since Des-acetyl Ó-MSH and CLIP have been shown to have strong hypnogenic activity they would close the loop in inhibiting the activity of the dorsal raphé (by increasing 5 HT liberation at the dendritic level : autoinhibition) (25).

2) It has been proposed that, during stress, glucocorticoids protect the organism by counteracting rather than enhancing stress activated defense reaction thereby preventing them from causing damage by overshooting (26). If PSR plays the same role for the brain that glucocorticoïds play for the organism , then PSR may be assimilated also to some protective or compensatory mechanism. Why a short (2 hr) "punitive" immobilisation stress would necessitate for the brain such a compensatory PSR, while a long-lasting (10 hr) "gentle PS" suppression would not, is still an unsolved problem.

III - Is paradoxical sleep energy gated ?

The ultradian rhythm of paradoxical sleep which represents the duration of the intervals between two succesive PS episodes is correlated both with brain size and general metabolic activity (27). Thus, in mice this interval is about 10min, whereas it is 24 min in the cat, 90 min in man and around 120 min in the elephant. These data suggest that in some direct or indirect way PS is related with metabolism or energetic mechanism.
The relationships between energetic mechanisms and PS could be summarized as follows (see references in Giuditta (28) and Siesjö (29)) :

The brain utilizes mostly glucose. Glucose is taken from the blood by special system which transport it in glia cells (astrocytes). Therein it is metabolized into pyruvate - the main fuel of neurons. In the neurons, the pyruvate may be processed along two different metabolic pathways according to the availability of oxygen (02).

Either it is metabolized into lactate without 02 by the enzyme lactate deshydrogenase or it is metabolized by the oxydative pathway (the KREBS cycle) into CO2 and H20. This oxydative metabolism necessitates another enzyme, the pyruvate deshydrogenase complex which transforms pyruvate into acetylcoenzyme A.

Therefore, some system of the brain can work either without oxygen (the lactate pathway) or need oxygen (the KREBS cycle). Are these pathways selectively used either during waking and sleep ?

It is evident that the waking brain has to perform in situations of emergency during which there might be a decrease of 02 availability. It is well known also that arousal increases in such conditions. In fact most of the "waking systems" like the monoaminergic systems contain lactate deshydrogenase so that waking might operate in anaerobic conditions (i.e. like the muscle during effort). This anaerobic metabolism does not only occur in emergency situation but also at the cortical level during attention. This has been recently demonstrated by Fox and his coworkers with positron emission tomography (30). In volontary human subjects during visual attention, the occipItal. cortex utilizes more glucose (+ 50 %) but almost no more oxygen (+ 5 %) than during relaxed waking : thus, the pyruvate has to be metabolized into the lactate pathway.

During slow wave sleep, there is both a decreased utilization of glucose and oxygen. This is due to the decrease of the metabolism of the brain and of the body. The temperature of the brain usually decreases because of the decrease of general metabolism and also because of the inhibition of the sympathetic system which leads to vasodilatation (increase of heat loss).

Then, at the energetic level, slow wave sleep performs two main functions : it decreases brain metabolism (and brain temperature). At the same time, it was shown by Karnovsky et al. (31) that glucose which is not utilized is stored as glycogen most probably into glial cells (glycogenogenesis)

Thus, after a sufficient duration of slow wave sleep (which represents the ultradian rhythm of PS) there appears to be a positive energetic balance between the increase of energetic fuel (glycogen) and the decrease of energy utilization. This positive balance would be "weighed" by some still unknown mechanism and would permit PS to occur and utilize aerobically the energy which has been stored in the glia cells.

There are indeed many indirect evidence which suggest that PS A) utilizes at least as much energy as during waking and B) that such energy has to be utilized according to the oxydative pathway :

A) As shown by C14 deoxyglucose technique, glucose utilization is increased during PS at a level much higher than during slow wave sleep and even higher than during waking in some part of the brain (32)(33).

B) Although there has not yet been any simultaneous measurement of O2 utilization during PS (in relation with glucose or pyruvate utilization), the following indirect evidences suggest the importance of oxydative metabolism : PS is indeed decreased or suppressed selectively during hypoxic hypoxia (34)(35), whereas slow wave sleep or waking may be enhanced. On the contrary, increase of O2 availability in chronic pontine cats, kept at a fixed temperature (temperature clamp) induces a most significant decrease of the ultradian rhythm of PS (and an increase of PS up to 100%). In such a case, PSS ultradian rhythm appears to be gated by energetic mechanisms (36) (Figure 3).
Finally, why does PS necessitate the aerobic metabolism of pyruvate ? A possible answer is provided by some of its mechanisms. There are converging indirect evidences that PS depends upon cholinergic mechanisms (see above). Now, acetylcholine is one of the central neurotransmitter which is the most strongly dependent upon oxydative metabolism. Indeed, acetylcholine synthesis is effectuated both from choline and mostly acetyl Co A (the product of pyruvate through the pyruvate deshydrogenase complex during the aerobic pathways). For this reason it is well known that the decrease of 02 availability alters or suppresses cholinergic metabolism (37).
The importance of pyruvate deshydrogenase may also explain why most of the neurons which are involved in the so-called cholinergic "executive mechanisms" of PS are particularly rich in this oxydative enzyme. And it is not unlikely that some alterations of this enzyme might explain the suppression of PS in some human degenerating disease or in any condition leading to decrease availability of O2 in the brain stem.

Summary

1. A review of the effects of brainstem transections, carbachol microinjections and lesions does not permit to conclude that the Pons is necessary and sufficient for Paradoxical Sleep (PS) generation.
2. The different theories which explain the mechanisms of PS rebound are summarized. In some circumstances, PS rebound depends upon the stressful quality of PS deprivation.
3. PS appears to necessitate oxydative metabolism of glucose, and in special occasion PS may be increased by hyperoxia.

Keywords

Paradoxical Sleep - Pontine Cholinergic Mechanisms - Carbachol - Paradoxical Sleep Rebound - Hypoxia - Hyperoxia.