INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE CYTOKINE TUMOR NECROSIS factor- (TNF-) is an important mediator of communication between the central nervous system and the periphery. Elevated blood concentrations of TNF-, such as those that occur in inflammation and various cancers, are related to sickness behavior, and exogenous TNF- can induce sleep. Specifically, TNF- given intraperitoneally in rats just before the onset of the dark phase increases the time of non-rapid eye movement sleep in a dose-dependent manner (11). TNF- also attenuates fast waves and enhances slow waves (11). Vagotomy attenuates TNF--induced non-rapid eye movement sleep and abolishes -activity slowing, suggesting that vagal afferents mediate part of the action (12). However, persistence of the somnogenic effect of TNF- suggests that other mediators are also present. TNF- p55 receptor knockout mice fail to sleep more after the administration of TNF-, indicating that the p55 receptor is also essential for the somnogenic effects (6).

In normal mice, basal endogenous TNF- levels are below the limit of detection under physiological conditions (19, 21). However, in humans, TNF- concentrations show a biphasic elevation, peaking at 0730 and 1330 h, a pattern different from that of interleukin-2, interleukin-10, and granulocyte-macrophage colony-stimulating factor (23). The soluble p75 TNF- receptor also shows a circadian rhythm with a peak around 0750 h, preceding the peak of cortisol (13). By contrast, in cancer patients, the concentration of TNF- reaches its peak at midnight (3). The circadian rhythm pattern of TNF- concentrations is likely related to the circadian variations in the chemotherapeutic response to TNF-. The toxic effect of a lethal dose of TNF- and survival probability appear to follow circadian dynamics (10).

Circadian variation of the TNF- mRNA concentration in brain has been reported. There are regional differences, with the hypothalamus and hippocampus having TNF- expression higher in the light phase than in the dark phase (4). Similarly, the concentration of TNF- is highest in the hypothalamus, hippocampus, and cerebral cortex of the rat at the onset of light (7). This is consistent with the onset of sleep in rats in the light phase.

Thus both peripheral and central sources of TNF- could contribute to sleep. To discern the importance of each individual source, it is essential to evaluate the involvement of the blood-brain barrier (BBB) in the uptake of TNF-. Our laboratory has shown that TNF- crosses the BBB by a saturable transport system and that the p55 and p75 receptors are involved in this transport (9, 19). The BBB would provide the most direct and immediate means for a cytokine such as TNF- to affect cortical activity because of its large surface area of 100-150 cm²/g of brain, in contrast to the much smaller area of the circumventricular organs of ~0.02 cm²/g of brain (22). A circadian rhythm at the BBB is present for the nonapeptide  sleep-inducing peptide, which has its highest entry from blood to brain between 1200 and 1600 h (clock time, lights on 0600 h and lights off 1800 h) (2), and for another somnogenic cytokine, interleukin-1, which has its highest influx at 0800 h and lowest at 2400 h (1). Therefore, in this study, we assessed whether there is circadian variation in the permeability of the BBB to TNF-. We measured the uptake of radiolabeled TNF- by brain and spinal cord and compared it with that of a peripheral control (muscle).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Recombinant murine TNF- (R&D Systems, Minneapolis, MN) was radiolabeled with ¹²⁵I by the chloramine-T method, the reaction being stopped at 1 min, and the ¹²⁵I-labeled TNF- was purified on a column of Sephadex G-10. The specific activity of ¹²⁵I-TNF- was 80 Ci/g. Adult male CD1 mice (Charles River), weighing 19-21 g at arrival, were housed at the institutional animal care facility for 6 wk before study. The mice were kept under a 12:12-h light-dark cycle (lights on at 0600 h, lights off at 1800 h), with constant room temperature, water, and food. Each group of mice was transferred from the animal room to the procedure room immediately before their time of study and anesthetized intraperitoneally with 40% urethane.

The groups were studied every 3 h (n = 5 mice/group, grouping = clock time). Mice were tested at clock times 0300, 0600, 0900, 1200, 1500, 1800, 2100, and 2400 h within a 24-h period by the same experimenter. Each mouse received 0.9 µCi of ¹²⁵I-TNF- injected into the left jugular vein in 100 µl of lactated Ringer solution containing 1% bovine serum albumin. At 10 min after this intravenous bolus injection of ¹²⁵I-TNF-, arterial blood was collected by transection of the right common carotid artery, and the mouse was decapitated immediately afterward. Blood was centrifuged to obtain serum. The whole brain, without pineal and pituitary glands, and spinal cord segments (cervical, thoracic, and lumbar regions) were collected. Part of the right gluteus major muscle was also collected as the peripheral control. The radioactivities of ¹²⁵I-TNF- in the weighed tissue and in 50 µl of serum were measured in a gamma counter. The uptake of ¹²⁵I-TNF- in blood and tissue was previously shown by HPLC at 10 min to mainly represent intact TNF- (9).

The ratios of tissue uptake of ¹²⁵I-TNF- at 10 min were calculated for the brain-to-serum ratio, the spinal cord-to-serum ratio, and the muscle-to-serum ratio and were expressed as microliters per gram [(cpm/g of tissue)/(cpm/µl of serum), where cpm is counts/min]. The circadian rhythm of tissue uptake was evaluated by cosinor methods. Because the circadian variation of the uptake of TNF- in a region did not appear to be sinusoidal, a multiple-component model, including harmonic terms, in addition to the 24-h fundamental component, was used. Several cosine curves in harmonic relation (24, 12, and/or 8 h) thus were fitted concomitantly. Parameter tests at a trial period of 24 h were also performed to compare TNF- uptake at different sites. The rhythms were characterized by the following parameters: 1) the midline estimating statistic of rhythm (MESOR), a rhythm-adjusted mean; 2) for each cosine component, the double amplitude, a measure of the extent of predictable change within a cycle; and 3) the acrophase, a measure of the timing of overall high values recurring in each cycle.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

There was no statistically significant circadian rhythm of ¹²⁵I-TNF- in serum [y = 56.2 + 5.0 cos(2t/24  3.51), percent rhythm (PR) = 12%, P = 0.088], where t is time. The PR is equivalent to R² and is the proportion of the variance (around the mean value), which is accounted for by the fit of the model. Similarly, there was no statistically significant circadian variation in the muscle uptake of ¹²⁵I-TNF- after intravenous delivery (Fig. 1), and there was no significant circadian rhythm of ¹²⁵I-TNF- entry into the brain (Fig. 2). In contrast, a statistically significant circadian rhythm was present for the uptake of TNF- into the spinal cord with a model that included cosine curves with periods of 24 and 12 h (Fig. 3).

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Lack of significant circadian variation of tumor necrosis factor- (TNF-) uptake by the muscle. Each point (mean ± SE) represents mice studied at that particular clock time (n = 5/group). The dashed curve was generated by the fitted model. PR, percent rhythm; t, time. Lights were on from 0600 to 1800 h (open bar). Hatched bar, lights off.

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Lack of significant circadian variation of TNF- uptake by the brain. The dashed curve was generated by the fitted model. Lights were on from 0600 to 1800 h (open bar). Hatched bar, lights off.

View larger version (12K): [in this window] [in a new window]   Fig. 3.   Significant circadian rhythm of TNF- uptake by the total spinal cord. The dashed curve was generated by the fitted model. Lights were on from 0600 to 1800 h (open bar). Hatched bar, lights off.

A circadian rhythm was also demonstrated for each spinal cord segment considered separately. Except for a statistically significant lower MESOR of the thoracic spinal cord (Table 1), overall the circadian amplitude and acrophase were very similar among the spinal regions. The peak ¹²⁵I-TNF- uptake occurred around 0456 h in the cervical spinal cord (Fig. 4), around 0324 h in the thoracic spinal cord (Fig. 5), and around 0228 h in the lumbar spinal cord (Fig. 6).

View this table: [in this window] [in a new window]   Table 1.   Circadian features of TNF- uptake by CNS: recurrent peak time

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Significant circadian rhythm of TNF- uptake by the cervical spinal cord. The dashed curve was generated by the fitted model. Lights were on from 0600 to 1800 h (open bar). Hatched bar, lights off. The peak 125I-labeled TNF- uptake occurred at 0456 h.

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Significant circadian rhythm of TNF- uptake by the thoracic spinal cord. The dashed curve was generated by the fitted model. Lights were on from 0600 to 1800 h (open bar). Hatched bar, lights off. The peak 125I-TNF- uptake occurred at 0324 h.

View larger version (13K): [in this window] [in a new window]   Fig. 6.   Significant circadian rhythm of TNF- uptake by the lumbar spinal cord. The dashed curve was generated by the fitted model. Lights were on from 0600 to 1800 h (open bar). Hatched bar, lights off. The peak 125I-TNF- uptake occurred at 0228 h.

There was no statistically significant circadian rhythm in the uptake of TNF- by the brain or muscle. Parameter tests comparing the brain and spinal cord showed that the MESOR of the brain was larger than that for thoracic, lumbar, or total spinal cord (P < 0.001) but did not differ from that of the cervical spinal cord.

When the data were expressed as a percentage of each series mean value to counteract the large difference in MESOR among muscle, serum, and the central nervous system regions (i.e., MESOR = 100%), the relative amplitudes and acrophases could be compared directly across all regions in the test of equality of amplitudes and of the amplitude-acrophase pairs. Analyses showed that the brain had a circadian amplitude significantly smaller than that of any of the spinal cord regions, whereas no intraregional difference was found in the spinal cord. The difference in the amplitude-acrophase pair with respect to brain is a further validation of the absence of a circadian rhythm in the brain and the presence of a circadian rhythm in the spinal cord. Only a numerical estimate of amplitude for brain could be referred to, as the rhythm could not be detected. The control tissue, muscle, had no statistically significant circadian rhythm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Accumulating evidence suggests that circulating TNF- participates in sleep regulation. The variations of plasma TNF- concentrations in certain human subjects seem to assume a similar pattern with the amplitude of electroencephalographic -wave frequency (5). Reciprocal interactions of TNF- and melatonin have also been observed, suggesting that TNF- stimulates melatonin secretion, whereas melatonin in turn inhibits TNF- release into blood (14). One would expect a circadian variation in the availability of TNF- to the brain by penetration of the BBB. On the contrary, we found no statistically significant circadian rhythm of TNF- uptake by the brain. However, a circadian rhythm of TNF- uptake by the spinal cord was detected.

The circadian variations in the uptake of ¹²⁵I-TNF- into the spinal cord could not be described by a simple sinusoidal relationship. Several cosine curves in harmonic relation were fitted concomitantly, and the zero-amplitude assumption was tested for each of the components included in the model. Basically, the presence of harmonic terms in the model expresses the departure of the rhythmic waveform from a pure sine curve. In addition, a P value is provided for the model as a whole, which is an F test comparing the variability accounted for by the composite model with the residual variation. A significant rhythm was present in the total spinal cord as well as its individual segments. The reasons for the seemingly multiphasic responses, however, are unclear. Perhaps the secondary peaks seem more apparent because of the short duration of the main peak around 0200-0300 h. Nevertheless, despite the statistical differences in acrophase and recurrent time among cervical, thoracic, and lumbar segments, parameter tests at a trial period of 24 h (two-way ANOVA) did not show a statistically significant difference among the spinal cord regions.

The presence of a circadian rhythm for the BBB permeability of TNF- in the spinal cord but not in the brain also occurs for leptin, an ingestive polypeptide similar in size to TNF- that is produced in the periphery but exerts its potent satiety signal in the brain. In the spinal cord of the mouse, the influx rate of leptin peaks around 2400 h and reaches its nadir around 0800 h (17). Yet the blood concentration of leptin is high enough under normal physiological conditions to account for partial saturation of the saturable transport system in the brain. In contrast, TNF- is usually not detectable in the blood of normal mice and thus should not have saturated the transport system for TNF- at the BBB. The lack of circadian rhythm of TNF- uptake in the brain indicates that alternative pathways of access (e.g., diffusion to the hypothalamus by circumventricular organs and vagal nerve afferents) may explain the peripheral effects of TNF- on the electroencephalogram.

A regional difference in the BBB permeation of TNF- is present in mice, with the spinal cord having a higher influx than the brain (15). This differential permeability also is apparent in p55 or p75 receptor knockout mice, which have a significantly decreased uptake in the spinal cord but not in the brain (19). The blood-spinal cord barrier also is more susceptible to regulatory processes such as spinal cord injury (18, 20) and experimental autoimmune encephalomyelitis (16), in which the transport system for TNF- is upregulated. Thus the presence of a circadian rhythm for TNF- adds to the unique features of the endothelial blood-spinal cord barrier.