The dimeric transcription factor nuclear factor-κB (NF-κB) regulates several endogenous sleep-modulatory substances and thereby serves as a pivotal mediator of sleep-wake homeostasis. To further define the role of NF-κB in sleep regulation, we monitored sleep and temperature in mice that lack the p50 subunit of NF-κB [p50 knockout (KO) mice]. Compared with the control B6129PF2/J strain, p50 KO mice spend more time in slow-wave sleep (SWS) and rapid eye movement sleep (REMS) under normal conditions and show enhanced homeostatic recovery of sleep after sleep loss. p50 KO mice also show increased SWS and reduced REMS and temperature after the administration of lipopolysaccharide, yet they are behaviorally less responsive to challenge with influenza virus. These data support a role for NF-κB, and, in particular, for the p50 subunit, in the regulation of sleep in healthy mice and in mice experiencing immune challenge.
- nuclear factor-κB
- p50 knockout mice
the nuclear factor (NF)-κB/Rel family of proteins regulates transcription via binding to a common decameric sequence motif known as the κB site (5′-GGGACTTTCC-3′). Members of the NF-κB family (p50, p52, p65, c-Rel, and Rel B) share a conserved Rel homology domain and form homo- or heterodimers in various combinations in different cell types. In the basal state, NF-κB occurs as an inactive dimer (most commonly p50/p65, p50/p50, or p65/p65) that is located in the cytosol and is bound to inhibitory IκB proteins. Activation of NF-κB can be triggered by a variety of stimuli, including proinflammatory cytokines, oxidative stress, bacterial and viral products, ischemia, and ultraviolet radiation. Activation leads to rapid phosphorylation, ubiquitination, and subsequent degradation of IκB-α. This allows translocation of the dimer into the nucleus where it can initiate NF-κB-dependent gene transcription. NF-κB activation regulates a large and diverse array of genes that modulate varied physiological functions, including neuronal survival and immune function (1, 24).
In the central nervous system, NF-κB activation exhibits diurnal variation, with the greatest amount of activation present during the light phase (i.e., the rest phase for mice; see Ref. 6). Activation of NF-κB also increases during periods of sleep loss (6), and sleep deprivation increases p65 immunoreactivity in rat hypothalamus (5). In addition, administration of an NF-κB inhibitor peptide reduces both spontaneous and interleukin (IL)-1β-induced sleep in rats and rabbits (32). Furthermore, NF-κB modulates the expression of several endogenous sleep regulatory substances (SRS), including tumor necrosis factor (TNF)-α (4, 7, 29–31), IL-1β (4, 29–31), and nitric oxide synthase (NOS; see Refs. 62 and 63). Finally, many SRS in turn activate or inhibit NF-κB. For example, TNF-α activates NF-κB (6, 7) and promotes sleep (28, 29, 48), whereas glucocorticoids inhibit NF-κB activation (59, 60) and promote wakefulness (14). Thus NF-κB appears to participate in numerous biochemical interactions that regulate sleep and arousal.
Infectious and inflammatory challenges alter normal patterns of sleep in numerous species. For example, administration of the bacterial product lipopolysaccharide (LPS) elicits time-dependent changes in time spent in slow-wave sleep (SWS) and rapid eye movement sleep (REMS) and in electroencephalogram (EEG) delta power or delta wave amplitude (DWA) during SWS in rats, rabbits, mice, and humans (28, 38, 51, 57). These changes are likely to be mediated via LPS-induced production of proinflammatory cytokines such as TNF-α and IL-1β (16, 29, 30). LPS also activates NF-κB (46), which in turn promotes the production of TNF-α, IL-1β, and NOS (16, 62). Similarly, an active NF-κB signaling system is a prerequisite for influenza virus infection of human cell lines (39), and IκB kinase is a key factor in triggering production of inflammatory cytokines in influenza-infected airway epithelial cells (3). Because infectious challenges promote inflammatory reactions in part through activation of NF-κB, the transcription factor may also contribute to the alteration of sleep during infection.
The assessment of mice that lack specific genes (i.e., knockout or KO mice) is an important strategy for determining the importance of genes and gene products in physiological processes such as sleep. Characterizing the sleep patterns of KO mice under normal conditions and in response to infectious and inflammatory stimuli could provide novel insights into the mechanisms by which NF-κB modulates sleep and arousal. Mice with genetic deletion of the p65 subunit of NF-κB die in utero (2), but p50 KO mice survive to adulthood (47). The normal diurnal variation of NF-κB activation, its increased activation after sleep deprivation, its role in regulating the production of several SRS, and its prominent effects during inflammation and infection led us to evaluate the sleep patterns of p50 KO mice. We assessed sleep under basal conditions, after 6 h of forced wakefulness, and after challenge with LPS or influenza virus. Furthermore, we hypothesized that a deficient NF-κB system would alter transcription of SRS like TNF-α.
Eight- to ten-week-old male B6;129P2-NfκB1tm1Bal (p50 KO), and genetically intact B6129PF2/J (F2) mice were purchased from the Jackson Laboratory (Bar Harbor, ME). The p50 KO mice were homozygous for deletion of a functional p50 subunit of NF-κB. Although the F2 mice provide only an approximate genetic match for the p50 KO strain, they are recommended for use as the control strain according to the Jackson Laboratory and were used on that basis. All mice were housed in temperature-controlled chambers (21 ± 1°C) with a 12:12-h light-dark cycle and free access to food and water. The Laboratory Animal Care and Use Committee of the Southern Illinois University School of Medicine approved all animal procedures used in this study.
Mice were surgically implanted with instrumentation to permit monitoring of the EEG, electromyogram (EMG), locomotor activity, and core body temperature. Mice were anesthetized with a mixture of ketamine (50 mg/kg sc) and xylazine (50 mg/kg sc) and were supplemented with additional anesthetic during surgery if needed. Standard aseptic techniques were used. In brief, four insulated stainless steel wires (Plastics One, Roanoke, VA) were visually positioned parallel to and under the skull in bilateral frontal (1 mm anterior to bregma and 2 mm to the left and right of midline) and parietotemporal (3–4 mm posterior to bregma and 2 mm to the left and right of midline) positions to serve as EEG electrodes. All electrodes were inserted in a pedestal that was secured to the skull with dental acrylic. One of the electrodes was made continuous with cable shielding and served as a ground; this electrode was not used for data acquisition. Two of the other three electrodes were referenced against each other in the combination that provided the best visual differentiation of three vigilance states (wakefulness, SWS, and REMS). EMG electrodes (Plastics One) were placed subcutaneously overlying nuchal muscles of the left and right sides of the body and were referenced against each other. During the same surgery, mice were also implanted with intra-abdominal transmitters (DSI, St. Paul, MN) to allow telemetric recording of locomotor activity and core body temperature.
After surgery, mice were housed in individual cages in a sound-attenuated temperature-controlled (21 ± 1°C) chamber under a 12:12-h light-dark cycle. Ibuprofen (0.2 mg/ml) was provided in the drinking water as an analgesic from 1 day before through 5 days after surgery. Mice were allowed a minimum of 14 days for recovery from surgery and habituation to the recording environment.
To permit collection of EEG and EMG data, mice were tethered to a six-channel electrical commutator with a lightweight flexible cable (Plastics One) and were acclimated to the tether for at least 3 days. Past work has shown that this acclimation period is adequate for obtaining stable recordings of sleep (52). Throughout all recording sessions, the mice could move freely in their cages and had continuous access to food and water. Monitoring of sleep, temperature, and locomotor activity in the absence of any experimental manipulation was initiated at light onset (lights on for 12 h beginning at 9:00 A.M.) and continued for 24 h.
To assess the homeostatic response to sleep loss, sleep was monitored for 24 h in undisturbed mice (n = 10/strain). Beginning immediately after light onset on the following day, mice were subjected to 6 h of enforced wakefulness. Wakefulness was maintained by gently stimulating the mice whenever they assumed a sleep-like posture or developed high DWA on the EEG. After the sleep deprivation period, recording continued for the remaining 18 h of day 2 without further disturbing the mice.
The same mice were used ∼1 wk later to assess the impact of LPS administration on sleep. Sleep and temperature were monitored in undisturbed F2 (n = 8) and p50 KO (n = 7) mice for 24 h, beginning immediately after light onset. The next day, mice were injected intraperitoneally with LPS (10 μg of Escherichia coli serotype O111:B4; Sigma, St. Louis, MO), and sleep recording continued for 24 h. Preliminary work in our laboratory has demonstrated that intraperitoneal injection with pyrogen-free saline (PFS) does not influence the normal sleep-wake patterns in F2 or p50 KO mice (Jhaveri KA, Ramkumar V, and Toth LA, unpublished data), consistent with our published findings in other strains (55, 56). Therefore, the sleep patterns that developed after LPS administration were compared with those obtained during the baseline period for each animal.
A separate set of mice were used to evaluate sleep during influenza infection. Basal patterns of sleep were recorded for 24 h without treatment. Immediately after light onset on the next day, mice were anesthetized lightly with methoxyflurane and inoculated intranasally with 25 μl of allantoic fluid that contained ∼1,000 hemagglutinating units of strain A/HKx31 influenza virus (H3N2; see Ref. 56). This dose approximates the Lethal dose (LD10) for 12-wk-old male C57BL/6J mice. At the end of the recording period (i.e., 72 h after inoculation), mice were killed by exsanguination under isoflurane anesthesia. The lungs were removed and frozen for subsequent culture to verify the presence of pulmonary infection. Previous work has shown that intranasal inoculation with uninfected allantoic fluid does not influence the normal sleep-wake patterns of C57BL/6J mice (52, 56). Therefore, the sleep patterns that developed after influenza inoculation were compared with those obtained during the baseline period for each animal.
Data acquisition and analysis.
EEG and EMG signals were processed through an eight-channel Grass polygraph. The amplified EEG signals were electronically bandpass filtered at delta (1–4 Hz) and theta (4–8 Hz) frequencies (Quality Software, Springfield, IL), and the filtered signals were digitally sampled (Quality Software) at a rate of 16 Hz. The sampled data were rectified and averaged in 10-s epochs. A computer-assisted scoring method employing custom software (Quality Software) was used to assign vigilance states to each 10-s epoch of the recording period. To accomplish this, EEG tracings for each animal were examined visually to determine a threshold DWA associated with SWS and a threshold theta-to-delta ratio associated with REMS. EMG tracings were examined to determine threshold amplitudes associated with movement. A computer algorithm used these thresholds to assign vigilance states to each 10-s epoch over the entire recording period. Three vigilance states were defined as follows: 1) SWS (DWA values above threshold in the absence of movement for any two consecutive 10-s intervals); 2) REM sleep (low DWA and high theta-to-delta amplitude in the absence of movement); and 3) wakefulness (all time intervals that did not conform to requirements for SWS or REMS). Two scorers visually reviewed the data to assign thresholds, and one scorer reviewed all computer-scored data to verify the accuracy of the computerized scorings.
The percentage of time spent in SWS and REMS was calculated for 12 2-h intervals during each 24-h recording period. Differences were analyzed with ANOVA to allow between- and within-group comparisons for effects of strain, phase (light vs. dark), conditions (baseline, postsleep deprivation, post-LPS injection, or postinfluenza infection), and interactions. The length and number of bouts of SWS and REMS and mean DWA during SWS (expressed as a percentage of the average value measured on the baseline day) were also calculated. All values are expressed as means ± SE.
The vendor provides calibration factors for each transmitter. The DSI software uses these factors to convert the frequency signals emitted by the abdominal transmitters into temperature values that are expressed in degrees Celcius. Temperature values were measured and stored every 10 min over the entire recording period. Locomotor activity is detected as movement of the animal (i.e., the abdominal transmitter) across the DSI receiver positioned under the cage. Activity counts, which reflect transmitter movement, were summed and stored across each 10-min interval for the entire recording period.
Serum and pulmonary measurements.
Blood was collected at light onset (baseline condition), immediately after 6 h of forced wakefulness, 4 h after injection of PFS or LPS, and 72 h after influenza infection. In all cases, blood was collected by intracardiac puncture from anesthetized mice. Mice were then killed by cervical dislocation without recovery from anesthesia.
In influenza studies, lungs and EDTA-anticoagulated blood were collected at the time of death (i.e., 72 h after infection). Lungs were frozen for subsequent measurement of viral titers and TNF-α. Whole blood was assessed for a complete cell count and differential analysis using a Hemavet hematology instrument (CDC Technologies, Oxford, CT). Plasma was analyzed for TNF-α (see below).
Pulmonary virus titers were measured using a standard plaque assay. Confluent monolayers of Madin-Darby canine kidney fibroblasts in 12-well tissue culture plates were inoculated with 10-fold serial dilutions of lung homogenate. After 60-min incubation at 37°C, the inoculum was removed, the cells were overlaid with 0.9% agar/media suspension containing 1 μg/ml of trypsin, and plates were incubated at 37°C in 5% CO2 for 3 days. The number of plaques present in the monolayers was then counted. Data were expressed in terms of plaque-forming units (PFU) per lung.
Concentrations of TNF-α in sera and lung homogenate were measured using a Quantikine ELISA kit for mouse TNF-α (R&D Systems, Minneapolis, MN) according to the manufacturer’s recommendations.
Statistical analyses were performed using ANOVA and the Student’s t-test with the SPSS (Chicago, IL) statistical package. A three-factor ANOVA, with strain treated as a between-subjects variable and days and phase (nested in days) treated as within-subjects variables, was used to analyze the data for each dependent variable. If a significant interaction was obtained between treatment condition and phase differences, a stepdown analysis was used to isolate the treatment condition(s) that generated the interaction. Further analysis of simple effects (separately for light and dark phases) was conducted on the isolated treatment conditions to determine the manner in which they differed from the other treatment conditions. Bonferroni corrections were applied to all tests subsequent to the ANOVA. All values presented are means ± SE for indicated sample sizes. An α level of P ≤ 0.05 was considered to indicate a statistically significant effect. Error bars and asterisks shown in text and Figs. 1–4 represent SEs and significance levels, respectively.
Food and water consumption of p50 KO and F2 mice.
Body weights measured after acclimation to the recording conditions revealed that the KO mice weigh significantly less than do F2 mice of the same age (F2 mice, 28.0 ± 0.6 g; p50 KO mice, 24.5 ± 0.3 g; n = 16 for each strain; P < 0.05). Although the consumption of food (F2 mice, 5.1 ± 0.8 g/day; p50 KO mice, 5.4 ± 0.5 g/day) and water (F2 mice, 7.7 ± 0.9 ml/day; p50 KO mice, 7.5 ± 1.1 ml/day) for two consecutive days under undisturbed conditions was not significantly different between the two strains, the food intake per gram body weight is 21.8% higher in the p50 KO mice compared with F2 mice (P < 0.05).
Spontaneous patterns of sleep in p50 KO and F2 mice.
Both strains of mice showed clear diurnal variation in sleep, spending more time in both SWS and REMS during the light phase vs. the dark phase of the 24-h cycle (Fig. 1). Compared with F2 mice (n = 16), p50 KO mice (n = 14) spent significantly more time in SWS during the initial 6 h after light onset (Fig. 1; P < 0.01). The number of SWS bouts was significantly greater in p50 KO mice during this period, but the SWS bout length did not differ significantly between the two strains (data not shown). Compared with F2 mice, the KO mice spent significantly more time in REMS during both the light and dark phases (Fig. 1; P < 0.01). The greater amount of REMS in the p50 KO mice reflected a significantly greater number of REMS bouts during both light and dark phases (Table 1). The greater time in REMS and SWS together generated an additional 85 min (26% more) time spent asleep in p50 KO mice vs. F2 mice over a 24-h period. Relative to the 24-h average value for each strain, DWA during SWS was significantly higher in p50 KO mice vs. F2 mice during the initial 6 h after light onset (Fig. 1; P < 0.01). Locomotor activity and core temperatures also showed diurnal variation in both strains. At some time points, p50 KO mice were less active than F2 mice (P < 0.05) and had lower core temperatures (P < 0.01; Fig. 1).
Sleep patterns after forced wakefulness in F2 and p50 KO mice.
F2 mice (n = 9) and p50 KO mice (n = 10) were subjected to forced wakefulness during the initial 6 h of the light phase. Compared with the baseline period, F2 mice showed greater DWA during SWS during the initial 2-h period in which sleep was permitted (P < 0.05), indicating a deeper plane of sleep. However, the time spent in SWS did not change (Fig. 2). In contrast, the p50 KO mice spent more time in SWS during the first 4 h after sleep deprivation (P < 0.01; Fig. 2). This increase was because of the greater number of SWS bouts (64.9 ± 2.7 vs. 74.8 ± 4.2; P < 0.05) and was accompanied by a significant increase in DWA during SWS (P < 0.01). Together, these changes indicate a deeper plane of relatively consolidated sleep in the p50 KO mice. Time spent in REMS did not change significantly after sleep deprivation of either p50 KO or F2 mice (Fig. 2).
LPS-induced changes in sleep and temperature in F2 and p50 KO mice.
Administration of 10 μg of LPS to F2 mice immediately after light onset significantly altered the time spent in both SWS and REMS (phase-by-day interaction, P < 0.01, n = 8; Fig. 3). F2 mice spent more time in SWS during the first 4 h after LPS injection (53.0 ± 5.3% on the baseline day vs. 72.0 ± 7.0% after LPS injection; P < 0.05). The increase in SWS was the result of a greater number of SWS bouts (Table 2), although the average length of SWS bouts was reduced, indicating fragmented sleep. REMS was reduced for 6 h after LPS administration (P < 0.01; Fig. 3). This reduction in REMS was because of a reduced number of REMS episodes; bout duration was unchanged. LPS administration reduced the EEG DWA during SWS in the initial 6 h after administration (P < 0.05). F2 mice also developed a transient hypothermia during the initial 2 h after LPS administration; this was followed by a modest and prolonged fever (P < 0.05; Fig. 3).
p50 KO mice injected with 10 μg of LPS developed a marked increase in SWS time that persisted for up to 18 h (P < 0.01, n = 7; Fig. 3). During the light phase, the number of SWS bouts increased, whereas the duration of SWS episodes was reduced, indicating fragmented sleep (Table 2). DWA during SWS was reduced during the initial 4 h after LPS administration (P < 0.05; Fig. 3), suggesting a relatively light plane of sleep. p50 KO mice also showed a large and prolonged reduction in time spent in REMS (P < 0.01; Fig. 3). This change was because of decreases in both the duration and number of REMS episodes after LPS administration (P < 0.05; Table 2). p50 KO mice also developed marked hypothermia that began within 2 h after LPS administration and persisted for the duration of the recording period (P < 0.05). Sleep and temperature responses of F2 and p50 KO mice to LPS administration differed significantly at most postinjection time points.
NF-κB is a prominent transcriptional regulator of TNF-α expression.
TNF-α production and release are elicited by LPSchallenge (16), and TNF-α promotes SWS (48). We therefore measured serum concentrations of TNF-α in both strains of mice at 4 h after saline or LPS administration. In both strains, serum concentrations of TNF-α were below limits of detection under basal conditions and after 6 h of sleep deprivation (data not shown). In contrast, values measured at 4 h after LPS administration were significantly higher in p50 KO vs. F2 mice (F2 mice, 336 ± 48 pg/ml; p50 KO mice, 1,567 ± 55 pg/ml; n = 6/strain; P < 0.01). Serum TNF-α concentration measured after saline administration was below the limits of detection in both strains.
Sleep and temperature responses of F2 and p50 KO mice after influenza infection.
F2 mice (n = 4) inoculated intranasally with influenza virus developed hypothermia (P < 0.05) and spent significantly more time in SWS (P < 0.05) and less time in REMS (P < 0.05) compared with the baseline period (Fig. 4). Infected p50 KO mice (n = 5) also showed hypothermia (P < 0.05) but did not develop significant alterations in sleep time (Fig. 4). However, after infection, KO mice did show shorter and more frequent bouts of SWS, indicating fragmented sleep (P < 0.05; Table 3). Pulmonary viral titers were higher in p50 KO mice (44 ± 8 PFU × 102/lung, n = 5) than in F2 mice (9 ± 3 PFU × 102/lung, n = 4; P < 0.001). Total white blood cell counts and numbers of neutrophils, lymphocytes, and monocytes in circulation did not differ between strains [cells/μl for p50 KO (n = 5) and F2 mice (n = 4), respectively, were as follows: total white blood cells, 5,690 ± 850 vs. 6,880 ± 1,660; neutrophils, 2,050 ± 260 vs. 3,230 ± 860; lymphocytes, 3,200 ± 710 vs. 3,080 ± 810; monocytes, 410 ± 60 vs. 510 ± 270]. Serum concentrations of TNF-α were below limits of detection at 72 h after influenza infection and were not different in lung homogenates of KO and F2 mice (KO mice, 205 ± 88 pg/ml, n = 5; F2 mice, 79 ± 12 pg/ml, n = 4; P = 0.21). Values varied considerably across individual mice of each strain (range of values: F2 mice, 56–107 pg/ml; KO mice, 60–633 pg/ml).
The data presented here demonstrate that: 1) under normal conditions, p50 KO mice spend more time in SWS and REMS than do F2 mice, 2) p50 KO mice show a more robust homeostatic response to sleep loss than do F2 mice, 3) administration of LPS causes mild fever, increased somnolence, and elevated serum TNF-α in F2 mice but marked hypothermia and significantly greater somnolence and TNF-α elevations in p50 KO mice, and 4) influenza-infected p50 KO mice do not show enhanced SWS, despite higher pulmonary viral titers and fragmented sleep. Taken together, these data indicate that the p50 subunit of NF-κB modulates sleep under physiological conditions and after immune challenge in mice.
Although p50 KO mice weigh less than the control mice, the two strains are not different in terms of food and water consumption. However, the food intake per gram body weight appears to be ∼22% higher in the p50 KO mice. In addition, at some time points, p50 KO mice sleep more and are less active than are control mice. These data suggest differences in basal metabolic rate in the p50 KO and F2 mice. Some reports (43, 65) suggest a role for NF-κB in the regulation of metabolic rate via its modulation of cytokine production. However, these studies do not address the metabolic impact of specific subunits of NF-κB either in adipocytes or after administration of peroxisome proliferator activator receptor-γ during sepsis. Nonetheless, these findings collectively suggest a role for NF-κB (and possibly for the p50 subunit) in the regulation of basic metabolic rate, but additional studies are needed to corroborate this possibility.
NF-κB modulates a number of proinflammatory genes that also modulate sleep. Furthermore, NF-κB is constitutively expressed in neurons and during sleep (6, 22). Given the ubiquitous presence of NF-κB in brain, together with the varied functional systems impacted by NF-κB, we had expected that the loss of p50 subunit would reduce time spent in sleep. However, we unexpectedly observed an apparent “gain of function” phenotype (i.e., increased sleep) in the p50 KO mice compared with genetically intact F2 mice. Current evidence suggests that p50 homodimers, which lack a transactivation domain, can inhibit expression of NF-κB-dependent genes (23). Thus the absence of p50 may facilitate gene transcription (25, 64) or promote formation of alternative dimers that modify normal patterns of transcription (17, 58). For example, the p52-p65 dimer forms in p50 KO mice, but this alternative dimer does not compensate functionally for the absence of p50 (12, 17). Furthermore, the transcription of specific genes is influenced not only by the subunit composition of the NF-κB family of dimers (17) but even by single nucleotide differences in the κB sites of the gene (34).
Such intricate regulation of the transcription of individual genes could influence the expression of SRS such as TNF-α, IL-1, and NOS in p50 KO mice, with subsequent interactions of these substances that indirectly alter sleep in a complex manner. Therefore, higher basal levels of specific SRS [e.g., NO via NOS (19) and TNF-α (64)] in the KO mice could represent a possible mechanism for their increased sleep. Supporting this possibility, our previous data show that p50 KO mice have higher basal levels of inducible NOS than do F2 mice (19). A recent study showed that expression of Rel/NF-κB-related factors in hamster suprachiasmatic nucleus is correlated to κB binding activity and implicates NF-κB activation in photic entrainment and the regulation of circadian rhythms (36). However, the diurnal rhythm of sleep is comparable in p50 KO and F2 mice under basal conditions.
The conclusion that p50 KO mice show a greater recuperative response to sleep loss than do F2 mice is tempered in that the sleep rebound we observed in F2 mice (Fig. 2) was somewhat less robust than reported previously by others in inbred C57BL/6 or 129 strains (9, 10, 18). At least three issues are relevant to interpretation of these results. First, the strain of mice studied can influence many features of both normal and induced changes in sleep (45, 49). In particular, the homeostatic response to sleep loss is under genetic control (11). The well-known hybrid vigor of F1 and F2 intercrosses could render them less susceptible to external physiological stressors such as enforced wakefulness. Supporting that possibility, two independent studies reported sleep rebounds in B6129P F1 or F2 mice that are similar in magnitude to those we report here (50, 61). Second, ambient temperature influences sleep differently in different strains of mice (45). In our study, mice were housed at an ambient temperature of 21°C, whereas others report ambient temperatures ranging between 22.5 and 25°C (9, 10, 18). Preliminary data collected in our laboratory using C57BL/6J mice (a background strain for both F2 and p50 KO mice) revealed similar recuperative responses to 6 h of enforced wakefulness at ambient temperatures of 22, 26, and 30°C (20). In that preliminary study, the recuperative sleep response of C57BL/6J mice was similar in magnitude and duration to that reported here for p50 KO mice. These observations suggest that F2 mice show diminished responses to sleep loss, whereas p50 KO mice show “normal” responses. Additional studies are necessary to distinguish between these possibilities.
A third consideration is that, during recovery from sleep loss, F2 mice showed a significant increase in DWA during SWS despite the absence of a significant increase in SWS time (Fig. 2). Furthermore, the changes in delta amplitude that we report here are consistent with the magnitude of changes in delta power reported by others (50, 61). Related to this, several studies report that, during the light phase (i.e., when spontaneous sleep is relatively high), rodents that are challenged with sleep-promoting stimuli may show modest increases in sleep time in conjunction with more prominent increases in DWA or delta power (33, 40, 41, 53). The converse (i.e., a robust increase in sleep time in conjunction with little or no increase in DWA during SWS) occurs when rodents are challenged with the same stimulus during the dark phase (i.e., when spontaneous sleep time is normally low; see Refs. 33, 40, 41, 53). Our finding of a significant increase in DWA without a concurrent increase in time spent asleep is consistent with this relationship.
Compared with F2 mice, p50 KO mice have altered immune responses to LPS challenge (47) and altered behavioral changes, as shown in our data. Changes in sleep in p50 KO mice could be related to changes in either the brain NF-κB system, with resultant alterations in the availability of SRS, or to changes in the peripheral host defense response generated by LPS administration. Mice that lack p50 display several deficiencies in immune responsiveness and an associated alteration in the production of immune modulators (e.g., cytokines, NOS; see Refs. 13, 25, and 64) that also regulate sleep (19). Our data show that a low dose of LPS has a greater pathophysiological impact on p50 KO vs. F2 mice, supporting a previous suggestion that activation of NF-κB p50/p65 heterodimers confers protection against the detrimental consequences of LPS exposure (13). Thus the marked behavioral and physiological responses of p50 KO mice to peripheral administration of LPS are likely to reflect their sensitivity to the toxic effects of LPS. A preliminary assessment of the responses of F2 and p50 KO mice to doses of 10, 20, and 80 μg of LPS revealed greater sensitivity to debilitation, shock, and death in p50 KO mice compared with the F2 mice, such that the qualitative and quantitative impact of 10 μg administered to p50 KO mice were not observed even at the 80-μg dose in F2 mice (data not shown). A striking feature of this sensitivity is the markedly divergent temperature responses of p50 KO and the genetically intact F2 mice (Fig. 3). F2 mice showed an initial hypothermia that lasted for 2–3 h, followed by fever. In contrast, p50 KO mice developed profound and prolonged hypothermia. Similarly, F2 mice showed increased SWS time 4–8 h after LPS administration, whereas the increases were greater in both magnitude and duration in p50 KO mice. Treatment of p50 KO mice with a higher (20 μg) dose of LPS profoundly suppresses sleep, DWA, and temperature (data not shown), as would also be expected to occur in F2 mice treated with yet higher doses (>80 μg) of LPS.
In contrast to their heightened sensitivity to LPS administration compared with F2 mice, p50 KO mice inoculated with influenza virus did not develop enhanced SWS or reduced REMS, although their SWS became fragmented. Although p50 KO mice had significantly higher pulmonary virus titers than did F2 mice, the biological significance of this difference is probably minor, since titers in both strains were relatively low. Consistent with this, the hypothermia, peripheral blood changes, and lung TNF-α concentrations measured after influenza infection were comparable in both strains. Compared with LPS, the role of NF-κB has been less investigated with respect to influenza infections. Nonetheless, an active NF-κB signaling system is a prerequisite for influenza virus infection of human cell lines (39), and IκB kinase is a key factor in triggering production of inflammatory cytokines in influenza-infected airway epithelial cells (3). The pulmonary infection, typical hypothermia, and behavioral suppression that characterized p50 KO mice indicate that they are susceptible to influenza infection and display many of the associated pathophysiological consequences. The data further imply that p65/p65 homodimers are sufficient to permit viral penetration of cells. The failure of p50 KO mice to develop enhanced SWS suggests a crucial role for the NF-κB family of proteins in this aspect of the response.
The known sleep-modulating properties of NF-κB and the ability of LPS to activate NF-κB p50/p65 heterodimers in brain in vivo (15) led to our hypothesis that LPS could influence sleep via NF-κB-dependent mechanisms. LPS-induced activation of NF-κB in turn induces a variety of SRS, including TNF-α (16) and NO (62), but the pattern of gene transcription may diverge from normal in a system that lacks the p50 subunit. Thus, in the absence of p50, production of some NF-κB-regulated SRS could be excessive, whereas induction of others could be reduced or absent (13, 17, 25, 47, 64). Analogously, lack of the p50 subunit could contribute to the elevated TNF-α concentrations we measured in serum after LPS administration, as reported previously by others (13, 25, 64). Under some conditions, TNF acts as an endogenous cryogen (26, 35). Thus the fivefold increase in circulating TNF-α concentration that develops in p50 KO mice after LPS administration could contribute to the prolonged hypothermia observed in those mice in our current study. Although thermoregulation and sleep are tightly coupled (44), temperature regulation can be dissociated from sleep under certain situations (27). The higher circulating levels of TNF-α could also contribute to suppression of REMS (8), which occurred after LPS administration in both strains but was particularly robust in the p50 KO mice. TNF-α that is relevant to regulation of sleep may be increased in various brain regions (hypothalamus, brain stem) responsible for sleep regulation after immune challenge. However, the stimulus for elevated central TNF-α originates in the periphery after LPS administration and influenza infection. We and others have speculated that the peripheral signal eventually impacts the brain and triggers local induction of TNF-α that then modulates sleep (37, 42). Supporting this idea, a technique similar to ours was used to show increased serum TNF-α levels in mice during the first hour after LPS administration (13). Similarly, elevated serum TNF-α is induced within the initial 1–3 h after influenza inoculation, but levels then decrease rapidly (21). Variable TNF-α levels in lung after influenza infection likely reflect variation in the severity of inflammation across mice. In our study, TNF-α concentrations were below the limits of detection under basal conditions and after sleep loss. Others are similarly unable to detect TNF-α in serum from healthy mice (13). However, failure to detect measurable serum TNF-α under baseline conditions and after sleep deprivation do not imply that TNF-α is not important in regulating sleep under these conditions, since central rather than peripheral TNF-α may be the more critical variable (8, 48).
In summary, our data show that, compared with F2 mice, p50 KO mice demonstrate more SWS and REMS under normal conditions and a relatively heightened homeostatic recovery of sleep after sleep loss. They are also more sensitive to challenge with LPS, which elicits increased SWS and marked suppression of REMS and temperature, yet do not display increased SWS during infection with influenza virus. These data reinforce a pivotal role for NF-κB in the physiological and pathophysiological regulation of sleep, perhaps via its modulation of various endogenous SRS, and highlight the diversity of influences that NF-κB can exert under physiological and pathological conditions.
This work was supported by National Institutes of Health Grants RR-017543 and HL-070522 and by the Excellence in Academic Medicine program of the Southern Illinois University School of Medicine.
We thank Lisa Cox, Sharon Lyons, and Joel Reichensperger for excellent technical assistance, the Southern Illinois University School of Medicine Division of Laboratory Animal Medicine for valuable help with animal care and maintenance, Tom Gardiner for the important contribution in developing the sleep scoring software, and Dr. Larry Hughes for guidance with statistics.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society