Sleep-dependent changes in cerebral oxygen consumption in newborn lambs

Authors


Giovanna Zoccoli, Dipartimento di Fisiologia Umana e Generale, Università di Bologna, Piazza di Porta San Donato, 2, 40126 Bologna, Italy. Tel.: +39 051 2091726; fax: +39 051 251731; e-mail: giovanna.zoccoli@unibo.it

Summary

During rapid-eye-movement (REM) sleep in adult subjects, the cerebral metabolic rate of oxygen consumption (CMRO2) is as high as that during wakefulness. We investigated whether CMRO2 during active sleep is already at the waking level in newborn life, to support the role of active sleep as a state of endogenous brain activation during early postnatal development. Newborn lambs, 2–5 days old (n = 6), were instrumented with electrodes for sleep-state scoring, catheters for blood sample withdrawal and pressure monitoring, and a transit-time ultrasonic blood-flow probe around the superior sagittal sinus. At the age of 19 ± 3 days, blood samples were obtained simultaneously from the carotid artery and the superior sagittal sinus during uninterrupted epochs of wakefulness, quiet sleep, and active sleep. The arteriovenous difference in blood oxygen concentration was multiplied by cerebral blood flow to determine CMRO2. CMRO2 during active sleep (47 ± 5 μmol min−1) was similar to the value in wakefulness (44 ± 6 μmol min−1) and significantly higher than in quiet sleep (39 ± 5 μmol min−1, P < 0.05). These data show that active sleep provides newborn lambs with brain activity at a level similar to that in wakefulness in terms of cerebral oxygen metabolism. The high CMRO2 during active sleep supports its functional role during early postnatal development, when time spent in active sleep is at a lifetime maximum, albeit constituting a metabolic challenge for newborns, because of the impairment of systemic and cerebral vascular regulation in this sleep state.

Introduction

The ontogeny of mammalian sleep is a subject of debate, particularly concerning altricial species, like the rat, whose nervous system is very immature at birth (Frank and Heller, 2003). Conversely, in human subjects and in precocial mammals, like sheep, electroencephalographically determined states akin to adult rapid-eye-movement (REM) sleep and non-REM sleep appear in utero (Frank and Heller, 2003; Szeto and Hinman, 1985). In human subjects, the states of neonatal active and quiet sleep may be considered homologous to the adult states of REM and non-REM sleep, respectively, although significant differences occur, e.g. concerning thermoregulation (Bach et al., 2002).

In human newborns, both the total and fractional daily amounts of active sleep are at a lifetime maximum even when compared with those of adult REM sleep (Roffwarg et al., 1966). The developmental decrease in active sleep that occurs postnatally is mainly restricted to the daytime (Fagioli and Salzarulo, 1982), so that active sleep decreases in reciprocal proportion to the developmental increase in wakefulness (Peirano et al., 2003).

The stimulus-induced brain activation that occurs during wakefulness is critical for brain development (Jain et al., 2003; Tagawa et al., 2005 for recent references). Intriguingly, electrophysiological (Steriade, 1995) and metabolic (Madsen et al., 1991) evidence indicates that brain activity in REM sleep is at waking levels in adult subjects. If brain activity in active sleep were also at waking levels in newborns, the abundance of active sleep early in postnatal life would provide the newborn's brain with endogenous activation at a time when waking life is limited in duration and scope (Peirano et al., 2003; Roffwarg et al., 1966; Shaffery et al., 2002). In addition, the assumption of endogenous brain activation in neonatal active sleep is central to Crick and Mitchison's (1983) hypothesis on the developmental ‘reverse learning’ function of this sleep state. However, this assumption has never been directly tested by comparing brain metabolic activity during active sleep with that during wakefulness in newborn life. In the present study, we tested this assumption by comparing the cerebral metabolic rate of O2 consumption (CMRO2) in active sleep with that during wakefulness and quiet sleep in newborn lambs.

Methods

Six lambs (Merino/Border-Leicester cross) were surgically prepared for the study at the age of 2–5 days, once feeding independently (lamb milk replacer: Veanavite; Shepparton, Australia) and gaining weight normally. All procedures were performed in accordance with the guidelines of the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes established by the National Health and Medical Research Council of Australia and were approved by the Monash Medical Centre Committee on Ethics in Animal Experimentation.

Surgical procedures

The lambs were anesthetized (halothane 1–2%, N2O 60%, balance O2) and instrumented using sterile surgical techniques under artificial ventilation. Pairs of Teflon-coated stainless-steel wires were implanted on the parietal cortex 1 cm anterior and 1 cm lateral to lambda (electrocorticogram), at the inner and outer canthus of the left eye (electrooculogram), and in the dorsal musculature of the neck (nuchal electromyogram). Non-occlusive saline-filled catheters were inserted into the carotid artery for arterial blood sampling and the femoral artery for arterial blood pressure recordings. A catheter was inserted under the dura mater for intracranial pressure recordings. After the removal of a 2-cm × 2-cm section of the skull at the intersection of the lambdoid and sagittal sutures, a transit-time ultrasonic flow probe (2-mm diameter; Transonic Systems, Ithaca, NY, USA) was positioned around the superior sagittal sinus (Grant et al., 1995). To sample cerebral venous blood, a non-occlusive catheter was inserted into the superior sagittal sinus via a guide needle passing through the caudal border of the cranial window, and the catheter tip was arrested caudal to the flow probe. A rigid cap of dental acrylic was formed over the flow probe and sinus catheter to stabilize them and to replace the section of the skull that had been removed.

Following surgery, a 1-mL intramuscular dose of procaine penicillin (250 mg mL−1) was administered daily for 5 days then every third day. Catheters were flushed with heparinized saline (1000 units mL−1) at least every other day to maintain patency.

Measurements

The lambs were studied for 1–3 days at the age of 19 ± 3 days (mean ± SEM). Food was available ad libitum and room temperature was maintained between 22 and 25 °C. During the study, the lambs’ cages were partitioned to prevent the lambs from turning around while allowing freedom to move forward and backward and to stand up and lie down. The lambs did not show any sign of discomfort caused by the cage divider during the recordings. The femoral and subdural catheters were connected to calibrated strain-gauge manometers (Cobe CDX III, Cobe Laboratories; Lakewood, CO, USA) and pressures were referenced to the midthoracic level when the lambs were lying down. The flow probe was connected to a flowmeter (model T101, Transonic Systems). All signals were digitized and stored on a personal computer (CVSOFT Data Acquisition and Analysis Software, Odessa Computer Systems, Calgary, Canada).

The scoring of the wake–sleep states was performed according to behavioral and polygraphic criteria as previously described (Silvani et al., 2005). Quiet wakefulness was identified when the lambs were lying down and when low-voltage high-frequency electrocorticographic activity, eye movements and electromyographic tone were present. Active sleep was identified when low-voltage high-frequency electrocorticographic activity and REMs were present and electromyographic tone was absent. Quiet sleep was identified when high-voltage low-frequency electrocorticographic activity was present, eye movements were absent, and electromyographic tone was lower than during wakefulness.

During stable wake–sleep states, 1.5 mL of blood was withdrawn simultaneously in heparinized syringes from the carotid artery and the superior sagittal sinus catheters to remove saline from the catheters’ dead spaces. Blood samples (0.5 mL) were then simultaneously drawn from each catheter, collected in heparinized syringes and analyzed to measure the partial pressures of O2 and CO2 (ABL500 Radiometer Pacific, Copenhagen, Denmark) and the concentration and O2 saturation of hemoglobin in the arterial and cerebral venous blood (OSM2 Hemoximeter, Radiometer Pacific, Copenhagen, Denmark). Finally, the blood volume withdrawn before the sampling was re-infused to minimize the animals’ blood loss.

Data analysis and statistics

The recordings were reviewed offline to exclude data from subsequent analysis in case arousals, state transitions or major movements occurred during blood sampling. With the exception of one lamb during wakefulness, in which only one blood sampling fulfilled these criteria, the analysis was based on 2–6 samples (average 3) for each lamb and wake–sleep state.

The cerebral arteriovenous difference in O2 concentration was computed from the concentration and saturation of hemoglobin in the arterial and cerebral venous blood. The cerebral O2 extraction fraction (EO2) was computed as the ratio of the cerebral arteriovenous difference in O2 concentration and the arterial O2 concentration. CMRO2 was computed as the product of the cerebral arteriovenous difference in O2 concentration and the mean value of cerebral blood flow over the 1-min interval that preceded the blood sampling. Cerebral perfusion pressure was computed as the difference between arterial and intracranial pressure over the same time interval. Because of technical problems, we could not obtain cerebral perfusion pressure data in active sleep in one lamb.

Mean values were calculated for each animal and wake–sleep state, and data are reported as the mean over all lambs±SEM, with n indicating the number of animals. The significance of the preplanned comparisons between active sleep and wakefulness and between active sleep and quiet sleep was assessed by Wilcoxon matched-pairs signed-ranks tests (SPSS, http://www.SPSs.com) with P < 0.05 considered to be statistically significant.

Results

The mean blood concentration of hemoglobin was 6.8 ± 0.2 g dL−1; the other results of the arterial and venous blood analysis are summarized in Table 1.

Table 1.   Results of the arterial and cerebral venous blood analysis
 WakefulnessQuiet sleepActive sleep
  1. Pa and Pv, partial gas pressures in the arterial and cerebral venous blood respectively; [O2]a and [O2]v, oxygen concentrations in the arterial and cerebral venous blood respectively. *P < 0.05 vs. wakefulness; P < 0.05 vs. quiet sleep. Data are mean ± SEM, n = 6.

PaO2 (mmHg)105.2 ± 5.5106.5 ± 5.6100.7 ± 6.5
[O2]a (mmol L−1)3.9 ± 0.14.0 ± 0.24.0 ± 0.2
PvO2 (mmHg)38.4 ± 0.537.5 ± 1.241.2 ± 0.9*
[O2]v (mmol L−1)1.9 ± 0.11.8 ± 0.22.1 ± 0.1*
PaCO2 (mmHg)42.3 ± 1.641.7 ± 1.844.2 ± 1.7
PvCO2 (mmHg)53.6 ± 1.552.2 ± 1.854.4 ± 1.5

No statistically significant difference occurred between active sleep and either quiet sleep or wakefulness in the arterial O2 partial pressure or concentration, whereas the respective values in the cerebral venous blood were significantly higher in active sleep than either in quiet sleep or in wakefulness.

The arterial and cerebral venous partial pressure of CO2 showed a slight increase in active sleep, which did not achieve statistical significance either with respect to quiet sleep (P > 0.07) or to wakefulness (P > 0.12).

Differences in CMRO2, EO2 and cerebral blood flow during active sleep with respect to the values in wakefulness and quiet sleep are summarized in Fig. 1.

Figure 1.

 Percentage differences in cerebral blood flow (CBF), cerebral O2 extraction fraction (EO2), and cerebral metabolic rate of O2 consumption (CMRO2) during active sleep with respect to values in quiet sleep (white bars) and wakefulness (black bars). Data are mean ± SEM, n = 6; *P < 0.05.

During active sleep, CMRO2 was 47 ± 5 μmol min−1 and increased significantly with respect to quiet sleep (39 ± 5 μmol min−1). Although CMRO2 was higher in active sleep than in wakefulness (44 ± 6 μmol min−1), the difference between these states was not statistically significant.

Changes in CMRO2 between wake–sleep states were paralleled by changes in cerebral blood flow, which was significantly higher in active sleep (25 ± 2.8 mL min−1) than either in quiet sleep (18.6 ± 2.3 mL min−1) or in wakefulness (21.3 ± 2.1 mL min−1). Such a rise in cerebral blood flow during active sleep occurred in excess of O2 metabolism, as indicated by the significant reduction in EO2 in this state (0.47 ± 0.01) with respect to either values in quiet sleep (0.54 ± 0.02) or in wakefulness (0.52 ± 0.02), and was not because of increases in cerebral perfusion pressure, which was remarkably constant between states (55.4 ± 3.7, 54.4 ± 1.2, and 54.6 ± 0.8 mmHg in active sleep, quiet sleep, and wakefulness respectively).

Discussion

We compared CMRO2 during active sleep with that during wakefulness and quiet sleep in newborn lambs, to test the hypothesis that active sleep is a state of endogenous brain activation during early postnatal life. Our results support this hypothesis, because CMRO2 during active sleep was similar to that in wakefulness and significantly higher than that during quiet sleep.

Brain activity requires metabolic energy, in particular for action potentials and postsynaptic potentials (Attwell and Laughlin, 2001). CMRO2 thus provides a suitable index of brain activity, especially because oxidative catabolism is the most efficient route for energy production, with a yield of 15 times more ATP per mole glucose than anaerobic catabolism. In this respect, the reported similarity in CMRO2 during REM sleep and wakefulness in adult human subjects (Madsen et al., 1991) indicates that brain activity during REM sleep is approximately at waking values in adulthood. No published data are available on changes in brain metabolism during active sleep with respect to wakefulness in the course of development. A study in newborn lambs (1 day old) was not successful in obtaining CMRO2 measurements during active sleep because of the instability of this sleep state early after birth (Richardson et al., 1989). Moreover, although CMRO2 has been measured in fetal sheep during the low-voltage electrocortical state (Czikk et al., 2001; Richardson et al., 1989), which is considered the counterpart of postnatal active sleep in precocial mammals (Czikk et al., 2001; Szeto and Hinman, 1985), no metabolic data are available during the rare episodes of fetal arousal (Szeto and Hinman, 1985). Our study is novel in that we could measure CMRO2 in active sleep during early postnatal development and compare it with both the values in wakefulness and in quiet sleep.

To determine CMRO2, we multiplied the difference in O2 concentration between the arterial and sagittal sinus blood by the sagittal sinus blood flow (Czikk et al., 2001). In sheep, the superior sagittal sinus is the site of choice to sample cerebral venous blood in long-term experiments (Czikk et al., 2001; Richardson et al., 1989; Rosenberg et al., 1982). This technique allows repeated measurements to be performed without perturbing the animals’ behavioral state. However, this technique does not allow EO2 measurements to be simultaneously obtained in different brain regions. Such brain regional data would be of great value for completing the description of sleep-dependent changes in CMRO2 and would help to clarify developmental differences between newborn and adult sleep. Our CMRO2 measurements only refer to brain regions drained by the superior sagittal sinus, which are mainly constituted by the frontal and anterior parietal lobes and represent approximately 35% of total brain mass (Grant et al., 1995). Based on the results of a validation study (Grant et al., 1995), our results in newborn lambs at 19 ± 3 days of age yield an estimated mean waking CMRO2 per unit mass of brain tissue of 308 μmol (100 g min)−1 or 6.9 mL (100 g min)−1, which is substantially higher than the value reported in 1-day-old lambs (170 μmol (100 g min)−1, Richardson et al., 1989), yet in good agreement with published value in 3- to 10-day-old lambs (6.00 mL (100 g min)−1, Rosenberg et al., 1982).

In our study, CMRO2 in the brain regions drained by the superior sagittal sinus significantly increased >20% during active sleep with respect to the value in quiet sleep (Fig. 1). It must be noted that the extent of sleep-dependent changes in CMRO2 is likely to vary on a regional basis in the brain. Similarly, the changes in blood flow during non-REM sleep and REM sleep differ quantitatively among brain regions in adult human subjects, being more substantial and robust across studies in thalamic and brainstem regions (Franzini, 2005). In this respect, the increase in CMRO2, we found in newborn lambs from quiet to active sleep is striking, as it was measured in a predominantly cortical portion of the cerebrum, although its extent may not be generalized to the whole brain. Finally, the extent of sleep-dependent changes in CMRO2 in a given brain region varies with the sleep episodes considered. The duration of the sleep episodes, the relative electrocorticographic spectral power in the delta and sigma frequency bands during quiet sleep, and the prevalence of tonic and phasic phenomena during active sleep are among the possible modulating factors. Nonetheless, our finding of a higher mean CMRO2 during active sleep than during quiet sleep confirms that the high-amplitude, widely synchronized mode of operation of thalamocortical networks in quiet sleep entails a low metabolic cost during early postnatal development as well as in adult life, notwithstanding marked developmental changes in the electroencephalographic power spectra (Jenni et al., 2004). Accordingly, CMRO2 is lower in the high-voltage than in the low-voltage electrocortical state in fetal sheep (Czikk et al., 2001; Richardson et al., 1989), lower in quiet sleep than in wakefulness in newborn lambs, 1 day old (Richardson et al., 1989), and lower in non-REM sleep than in wakefulness in adult human subjects (Madsen et al., 1991). On the other hand, we found that CMRO2 during active sleep was higher than during wakefulness, although the difference between the two states was not statistically significant (Fig. 1). These data broadly agree with the findings of Madsen et al. (1991) in adult human subjects, and indicate that endogenous brain activation during active sleep is quantitatively comparable with the exogenous brain activation of wakefulness during early postnatal development.

The assumption that neonatal active sleep is a state of substantial brain activation is the basis of most hypotheses concerning a developmental role of this sleep state (Crick and Mitchison, 1983; Mirmiran, 1995; Roffwarg et al., 1966). The abundance of active sleep early in life and its subsequent decline (Jenni et al., 2004; Peirano et al., 2003; Roffwarg et al., 1966) suggested that active sleep is involved in shaping and sustaining brain development by providing endogenous brain activation at a time when waking life is limited in scope and duration (Roffwarg et al., 1966). Endogenous brain activation during active sleep may thus play a role akin to that of stimulus-induced brain activation during wakefulness, which is essential for the development of cortical sensory areas (Jain et al., 2003; Tagawa et al., 2005). In agreement with a relevant developmental role of neonatal active sleep, its experimental deprivation exerts distinct short-term (i.e. increased synaptic plasticity Shaffery et al., 2002), and long-term (i.e. behavioral and REM-sleep abnormalities Mirmiran, 1995; Peirano et al., 2003) effects. These data cannot be taken to imply that the developmental roles of neonatal active sleep are identical to those of wakefulness (cf. Mirmiran, 1995), but rather suggest that endogenous activity generated during active sleep promotes brain development as well as exogenous sensory input during wakefulness does (Shaffery et al., 2002). Our study in newborn lambs is the first to provide direct support of the assumption that neonatal active sleep is a state of substantial brain activation, which essentially had rested upon electrophysiological (Steriade, 1995) and metabolic (Madsen et al., 1991) data obtained in adult animals or human subjects. Complete support for a developmental function of active sleep for the brain must await neurophysiological data that systematic propagation of impulses is occurring in maturing brain pathways during active sleep.

The newborn lamb, which is a model widely studied to understand sleep-related cardiovascular changes during early postnatal development, also represents a suitable model to study whether the abundance of neonatal active sleep is associated with an endogenous brain activation during this state. Sheep are precocial animals, their brain development mainly occurring in utero (cf. Frank and Heller, 2003). Whereas the human species is usually considered altricial, the designation is essentially assigned on a motoric basis, because the development of the human brain at birth is comparable with that of a precocial animal (Clancy et al., 2001). Moreover, sheep undergo a developmental reduction in active sleep time, from 10% of recording time in newborn life to 2.5% of recording time after weaning (Jouvet and Valatx, 1962). Similarly, the fraction of total sleep time spent in active sleep reaches 50% in human newborns and declines to approximately 30% at 9 months of age (Jenni et al., 2004).

In newborn lambs, we found that cerebral blood flow was significantly higher in active sleep than either in quiet sleep or in wakefulness (Fig. 1), in the absence of significant changes in cerebral perfusion pressure between states. Moreover, EO2 modestly but significantly decreased in active sleep with respect to the other states (Fig. 1). Thus, in active sleep, cerebral blood flow rose in excess of cerebral oxygen metabolism because of a reduction in cerebrovascular resistance, which had been shown to be dependent on nitric-oxide in previous work (Zoccoli et al., 2001). A similar excess increase in cerebral blood flow with respect to CMRO2 was reported by Madsen et al. (1991) in adult human subjects during REM sleep with respect to wakefulness, and attributed to a slight hypercapnia during the former state. Although we observed a trend toward higher values of the arterial partial pressure of CO2 during active sleep, differences between wake–sleep states were modest and not statistically significant (Table 1). Moreover, in newborn lambs, cerebral blood flow per unit O2 consumption responds poorly to changes in the arterial partial pressure of CO2 compared with the response in adult animals (Rosenberg et al., 1982). At least with respect to quiet sleep, the increase in cerebral blood flow we observed during active sleep thus appears linked to the increase in CMRO2, although in excess of it. Interestingly, a fall in EO2 accompanies experimentally induced neural activation and is the basis of functional magnetic resonance imaging studies with blood oxygenation level-dependent contrast (Ito et al., 2005). Both in sleep and in wakefulness, excess increases in cerebral blood flow with respect to CMRO2 may be explained by the limitation of O2 diffusion from the blood to brain mitochondria (Lenzi et al., 2000). Hypoxic brain tissue microregions would then develop far from capillaries and enhance vasodilatory stimuli during brain activation. Nonetheless, the EO2 decrease we observed during active sleep with respect to wakefulness demands further investigation in light of the non-significant increase in CMRO2 in the former state. In this regard, the changes that occur during REM sleep in the activity of brainstem nuclei may exert neurogenic vascular effects and contribute to the cerebral vasodilatation in this state (Maquet, 2000).

In conclusion, our data show that neonatal active sleep is a state of endogenous brain activity comparable with that during wakefulness in terms of cerebral O2 metabolism. Previous work has shown that the neural events characterizing active sleep also impair systemic and cerebral vascular regulation in this state: central autonomic commands cause paroxysmal events of hypertension and tachycardia (Silvani et al., 2005), the enhanced blood pressure variability being transmitted to cerebral blood flow (Silvani et al., 2004), while on the other hand, cerebral autoregulation is less effective against blood pressure reductions (Grant et al., 2005). Thus, while the substantial endogenous brain activity during neonatal active sleep may be beneficial for brain development (Crick and Mitchison, 1983; Roffwarg et al., 1966), it constitutes a metabolic challenge for the newborn.

Acknowledgements

The study was supported by the National Health and Medical Research Council of Australia (to A. M. Walker and D. A. Grant); the Monash University Research Fund (to D. A. Grant).

Ancillary