Systemic bacterial invasion induced by sleep deprivation

Carol A. Everson, Linda A. Toth


Profound sleep disruption in humans is generally believed to cause health impairments. Through comparative research, specific physical effects and underlying mechanisms altered by sleep deprivation are being elucidated. Studies of sleep-deprived animals previously have shown a progressive, chronic negative energy balance and gradual deterioration of health, which culminate in fatal bloodstream infection without an infectious focus. The present study investigated the conditions antecedent to advanced morbidity in sleep-deprived rats by determining the time course and distribution of live microorganisms in body tissues that are normally sterile. The tissues cultured for microbial growth included the blood, four major organs, six regional lymph nodes, the intestine, and the skin. The principal finding was early infection of the mesenteric lymph nodes by bacteria presumably translocated from the intestine and bacterial migration to and transient infection of extraintestinal sites. Presence of pathogenic microorganisms and their toxins in tissues constitutes a septic burden and chronic antigenic challenge for the host. Bacterial translocation and pathogenic sequelae provide mechanisms by which sleep deprivation appears to adversely affect health.

  • bacterial translocation
  • bacterial infections
  • immunity
  • immunocompetence
  • neuroimmunology

sleep is a biological requirement for survival, as has been well documented in laboratory animals (21, 35, 45). Although sleep deprivation is considered a risk factor for human disease, specific health impairments have not been definitively linked to sleep deprivation per se, and accordingly it has no descriptive clinical signs. Up to 64 h of sustained wakefulness in young, healthy individuals produces marked cognitive impairments (32) but only minor and nonspecific neurological and physiological signs (41). Prolongation of sleep deprivation to >8 days in four healthy young men produced several neurological and physiological signs, but these signs were not more revealing despite the fact that the men had reached a point “near the limit of their resistance” (36). Similarly, 9 days of voluntary sleep deprivation ended in collapse and hospitalization due to unspecified proximal causes (38). Sleep disruption can be profound in the general population. In the critically ill, for example, postsurgery slow-wave sleep is reduced to <1% of the nighttime recording for 5–8 days (44). The effects of severe sleep restriction in these patients have not been systematically studied despite the fact that deficiencies of basic biological requirements can have life-threatening implications. Most scientific reports of long-term sleep deprivation were conducted more than 30 years ago and employed physiological evaluations that were only rudimentary by the standards today, having predated major advances in fields such as neuroendocrinology and immunology. The lack of evidence for specific organic changes therefore does not indicate that sleep deprivation is harmless, but rather that its consequences are of a nature that have eluded discovery.

Recent findings in sleep-deprived animals indicate that host defense failure is central to the adverse effects of sleep deprivation (19). During 3 wk of sleep deprivation, rats develop a progressive negative energy balance, a progressive decline in circulating thyroid hormones due to altered central regulation, and sympathetic activation without overactivation of the hypothalamic-pituitary-adrenal axis (8,21, 23). These changes culminate in host defense failure, manifested by bloodstream infection at the onset of mild hypothermia, and a cachectic-like moribund state that comprises the last hours or days of survival (19). Marked tissue inflammatory reactions are absent. Other physical changes include early development of erythematous papules on the skin of the dorsal tail and plantar paws that develop into well-circumscribed necrotic plaques (26, 37). In most clinical conditions accompanied by cutaneous disease, dermatoses are considered outward manifestations of changes in the immune status of the host (48). Other signs of sleep deprivation include early and progressive increases in circulating neutrophil counts (26) and plasma alkaline phosphatase concentrations (25), the latter without histopathological evidence of liver damage (19, 28).

The proximal cause of death in sleep-deprived rats is bloodstream infection by opportunistic facultative anaerobes generally ascribed to gut origin, such as Pseudomonas aeruginosa, Klebsiella pneumoniae, Staphylococcus aureus, Streptococcus agalactiae, and Corynebacterium jejeikum (19). These organisms do not cause primary bacteremia and threaten life unless the host is immune compromised (46). Bloodstream infection by these microorganisms is highly lethal in humans, particularly when associated with hypothermia (50). In patients who develop septicemia and multiple organ failure but lack a primary focus of infection, the suspected causative events are translocation of microorganisms and endotoxin from the intestinal mucosa and their dissemination to extraintestinal sites (12, 27, 43, 49). The translocated bacteria and toxins are postulated to trigger a hypermetabolic and systemic inflammatory state resulting in sepsis without a septic focus (16).

The crucial question now is whether failure of host defenses in sleep-deprived rats is only a proximal cause of death related to moribund processes or whether problems in host defenses develop early during protracted sleep deprivation and thus reveal underlying disease processes. The purpose of the present experiment therefore was to determine whether there exists abnormal control of endogenous bacteria in body tissues prior to advanced morbidity in sleep-deprived rats, and, if present, the time course of these abnormalities. Bacteria in body tissues of rats after different durations of sleep deprivation were quantified and identified. The tissues evaluated included the blood, four major organs, and six regional lymph nodes. Additionally, the numbers and subgroups of bacteria cultured from the ileum and the cecum indicated the degree of bacterial overgrowth, and bacteria cultured from tail skin were assessed for their contribution to microbial populations found in lymph nodes of the extremities. The principal finding was early infection of the mesenteric lymph nodes (MLNs) by both aerobic and facultative anaerobic intestinal bacteria. Migration of bacteria from the MLNs was indicated by transient, polymicrobial infections of major organs. We conclude that sleep deprivation induces a chronic infectious and antigenic state that precedes outward signs of poor health. These findings provide an explanation of how sleep deprivation may exacerbate disease or hinder recovery from illness.


Animals and surgical procedures. All procedures were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and with an animal study protocol approved by the University of Tennessee Animal Care and Use Committee. Male Sprague-Dawley rats obtained from Harlan (Indianapolis, IN) were housed until surgery in filter-topped colony cages in a barrier facility. At the time of surgery, these animals weighed 435 ± 32 (SD) g and were 22.5 ± 1.7 (SD) wk old. The rats were anesthetized with ketamine ⋅ HCl (100 mg/kg ip), xylazine ⋅ HCl (2.4 mg/kg im), and atropine sulfate (0.1 mg/kg im). Supplementary doses of ketamine ⋅ HCl (10 mg/kg ip) were provided as needed. A solution of 1–2% lidocaine ⋅ HCl (2.4 mg/kg) was administered subcutaneously in the abdominal incision site through which a low-frequency telemetric transmitter (Barrows; San Jose, CA) was implanted for continuous recording of body temperature as an aid in assessing morbidity. Electrodes to record the cortical electroencephalogram (EEG), hippocampal theta waves, and temporalis muscle electromyogram (EMG) were implanted to distinguish wakefulness and specific sleep stages. Electrode leads were soldered to a head plug assembly and insulated, and the head plug assembly was connected to a long recording cable.

Procedure for producing sleep-deprived and yoked animals. Sleep deprivation was produced by the method of Bergmann et al. (9) and is described and illustrated elsewhere in detail (9, 26). This method has been validated for the selectivity of sleep deprivation and minimal interference with normal waking behaviors (8, 9, 21, 26). In brief, the procedure was a yoked paradigm, wherein two animals were housed under identical experimental conditions. Detection of sleep onset in the rat to be sleep deprived resulted in 6 s of forced locomotion for both the sleep-deprived and yoked rats. The yoked rats could sleep whenever the sleep-deprived rat was engaged in behaviors other than trying to sleep, because there were no periods of forced locomotion. This was accomplished by housing the two rats on either side of a large, round platform (45 cm diam) that was divided into two sides by a fixed wall. Beneath each side of the platform and extending out to a large open-air Plexiglas enclosure was a pan of water, 2–3 cm deep. The recording cable connected to the head plug assembly of each rat was attached to a swivel commutator and counterbalanced boom assembly. This arrangement permitted each rat to move freely vertically and horizontally along the platform, as well as to step down into and roam the water to the perimeter of the cage. After initial exploration, the rats almost always avoided the water and stayed on the platform where they would eat, groom, and sleep. During the 7-day baseline period the platform was slowly rotated once per hour for 6 s (one-third revolution) to acquaint the paired rats with the platform movement that required them to walk along the platform before it rotated beneath the dividing cage wall and thereby avoid stepping down into the water. During the experimental period, each onset of sleep in the rat designated as sleep deprived caused the platform to automatically rotate for 6 s. Sleep onset was detected by a microprocessor programmed to recognize changes in the amplitudes of EEG, EEG theta, and EMG signals meeting criteria matched to both online polygraphic recordings and behavioral observations. The platform was stationary when the sleep-deprived rat was awake during bouts of feeding, grooming, and exploring, which allowed the yoked rat opportunities to sleep. This procedure results in a consistent amount of sleep loss for both subjects (19, 21, 24, 26). Sleep-deprived rats are awake 90–91% of total time, compared with 46–50% during the baseline period. Sleep obtained during the remaining 9–10% of the time is highly fragmented and is composed largely of transitional sleep from wakefulness, which is of little apparent biological value (39). Because of the paired conditions on the platform, yoked rats are partially sleep deprived. Yoked rats are awake 58–62% of time, and their sleep is highly fragmented. Yoked rats do not develop the severe morbidity of the sleep-deprived rats, indicating that the experimental conditions are tolerable with an adequate amount of sleep. The procedure is considered benign in comparison with other methods of sleep deprivation, because the locomotion that is required is slow and brief and the rats are relatively unconfined. The procedure does not produce signs of either behavioral distress or neuroendocrine and cerebral metabolic changes associated with behavioral stress (reviewed in Ref. 20), thereby providing a high degree of certainty that a relatively selective physiological deprivation is produced and that physiological sequelae can be attributed to the deprivation of sleep.

Twenty-five pairs of animals were studied, each consisting of one rat subjected to sleep deprivation and the other yoked to the experimental conditions. Each experiment on a pair of rats lasted 3–5 wk and included at least 7 days for recovery from surgery, 7 days of baseline conditions, and 5, 10, 15, or 20 days of sleep deprivation. Sham-comparison rats, which were matched to sleep-deprived and yoked rats by age and weight, and given the same surgery, recovery period, and baseline conditions, were also evaluated. The sham-comparison rats were housed in the experimental apparatus under baseline conditions that included an hourly period of 6 s of forced locomotion, but otherwise were permitted sleep ad libitum. Sham-comparison rats compose the time period of 0 days of sleep deprivation.

Food intake and body weight were recorded daily. The diet was balanced, purified, and augmented with protein (3.8 kcal/g metabolizable energy; Zeigler Brothers; Garners, PA), which previously has been shown to attenuate extreme hyperphagia and to reduce variability in survival time in sleep-deprived rats, compared with provision of a diet augmented with fat and calories (26). All rats were kept under conditions of constant light to diminish the amplitude of the circadian rhythm. Ambient temperature was maintained at 28°C by thermostatically controlled heat lamps positioned above the apparatus.

Procedure for analysis of live bacterial populations and species. On the designated experimental day, the rats were individually removed from the apparatus for necropsy and for collection and processing of tissues. Rats were anesthetized briefly with halothane and then maintained with ketamine ⋅ HCl (58 mg/kg ip), xylazine ⋅ HCl (5 mg/kg im), and atropine sulfate (0.14 mg/kg im). Supplementary doses of ketamine ⋅ HCl or xylazine ⋅ HCl were provided as needed to cause deep anesthesia and analgesia. The skin was shaved, disinfected with 70% ethanol, and lathered with antibacterial foam (Convatec; St. Louis, MO). The animal was secured to a sterile barrier field inside a sanitized biosafety cabinet. Cardiac punctures for exsanguination and collection of blood specimens were performed sterilely through the left ventricle after aseptic removal of skin, muscle, and diaphragm. One milliliter of whole blood was injected into each of two culture bottles containing 20 ml of brain heart infusion broth (Septi-Chek, Becton Dickinson; Cockeysville, MD) for aerobic and anaerobic culture and into a tube containing EDTA (Terumo Medical; Elkton, MD) for later plating on blood agar. The following tissues were removed from the exsanguinated animal by sterile technique, usually in this order: spleen; both kidneys; entire liver; MLNs; cervical lymph nodes (CLNs); axillary lymph nodes (ALNs); inguinal lymph nodes (ILNs); mediastinal lymph nodes; entire lungs (lobe dissection without piercing the trachea); popliteal lymph nodes (PLNs); ∼200 mg skin at dorsal base of the tail, which included dermatoses if present, and the terminal ileum (∼12 to 15 mm) and the cecum, including contents. Swab specimens (Culturette, Becton Dickinson) were taken from the peritoneum and the thoracic cavity in 21 of the 25 pairs, and from the throat in 10 pairs. Portions of each tissue were weighed and placed in a sterile container before homogenization for culture under aerobic or anaerobic conditions. Dissected tissues were transferred to a second tissue culture hood prior to dissection of the ileum and the cecum.

Tissues were homogenized in 10 parts of either brain heart infusion or thioglycollate broth medium for cultures of aerobic and facultative anaerobic microorganisms, respectively. Tenfold serial dilutions of homogenates were performed to 10 5 for most tissues, 10 7 for skin, and 10 8 or 10 9 for intestine. Aliquots of all samples were plated on blood agar (Edge Biologicals; Memphis, TN). Homogenates of intestine were also plated on eosin methylene blue (EMB) agar (Levine, Edge Biologicals) to differentiate classes of gram-negative bacteria. Homogenates of the lung and of the mediastinal lymph nodes were cultured only under aerobic conditions. Homogenization, dilution, plating, and incubation of tissues were performed in an anaerobic chamber for culture of facultative anaerobic microorganisms. Microbial growth is reported as the number of colony forming units (CFUs) in aerobic and anaerobic cultures on the second day of incubation. The lower limit of detection was 110 CFUs/g tissue. Culture plates for major organs and lymph nodes were incubated for up to 7 days in many cases to determine the presence of slow growing microorganisms, but none were found. Bacterial species were identified by conventional microbiological methods and grouped according toBergey's Manual of Determinative Bacteriology (31).

Data analysis. Unless otherwise noted, nine pairs of rats were evaluated at the 5-day time point, six at 10 days, and five each at 15 and 20 days. There were five sham-control comparisons. The number of tissues studied at each time point is unequal because some tissues were added to an expanded study design after initial subjects had been evaluated. Mediastinal lymph nodes, CLNs, ileum, and cecum from five pairs of rats, rather than nine, were analyzed on day 5. For ALNs, the last addition to the study design, there were five, one, four, and five pairs studied at the sequential time points, as well as two sham-control comparisons. The 10-day time point for ALNs was not used for group comparisons because data were available from only one sleep-deprived and yoked pair.

Four of 57 operated animals were excluded from study for the following reasons: unexplained poor health (n = 2), sudden death after receiving anesthetic drugs before necropsy, and detachment of a recording head plug assembly 3 days before the planned completion of the experiment. Exclusion of individual data points from analysis included occasional, documented instances in which the tissue may have been inadvertently contaminated during processing and instances when it was not possible to accurately count the number of CFUs because, for example, the bacteria Proteus had spread uniformly over the agar plate.

The numbers of CFUs per gram of tissue were transformed to log10 for analysis. A three-factor ANOVA was used for most comparisons of microbial growth in tissues of sleep-deprived and yoked rats, the factors being time, paired dependency, and treatment. Two-factor ANOVAs with factors of time and treatment were used in those instances in which there was a poor distribution of microbial counts across the pairs of rats (e.g., liver, see Fig. 3), and the factor of pairing appeared to bias the estimates of least-squares means. CFUs are expressed as the geometric means ± SD of log-transformed data. Food intake and body weight changes were averaged within successive 5-day blocks, first for individual subjects and then across subjects within a group. As the duration of the experimental period increased, fewer pairs contributed to the group means for food intake and body weight. Therefore, statistical differences were tested by a two-way repeated measures ANOVA for an unbalanced, split-plot design. A two-way ANOVA, unbalanced design, was used to determine group differences in the weight of organs, the number of bacterial species contained in each organ (excluding skin and intestine), and the number of infected tissues in each subject. Two-by-two contingency tables were constructed for individual tissues to compare the proportions containing live bacteria (i.e., incidence) in sleep-deprived vs. yoked groups by Fishers exact two-tailed tests. For all analyses, statistical significance is indicated by the conventional value of P < 0.05. Bonferroni corrections for multiple comparisons were not applied in this study, because in most tests P < 0.002 would be required for statistical significance at the 95% confidence level. This approach to statistical significance seemed intolerably conservative because microbial populations in extraintestinal body tissues are considered de facto as biologically significant. Nonsignificant statistical differences are designated as NS. Unless specifically noted, variance is given as the means ± SD. Specific notations for use of the standard error include the abbreviation SE.


Physiological variables. During the baseline period, food intake and body weight were stable in all rats whether designated to be sham controls or to serve as sleep-deprived or yoked rats. During the experimental period, food intake and body weight changed progressively in sleep-deprived rats. By 16–20 experimental days, food intake in sleep-deprived rats had increased to 204 ± 21% of basal amounts and body weight had decreased by 12 ± 5%, compared with an increase to 168 ± 30% and a decrease of 9 ± 3% in yoked rats (food intake, group difference P < 0.0001; body weight loss, group by time interaction P < 0.01). These findings are consistent with those of previous studies showing a marked negative energy balance during sleep deprivation (21, 26). Skin dermatoses appeared more severe for sleep-deprived rats than for yoked rats, and the appearance of the dermatoses was consistent with that described previously (37). Diarrhea was not observed in any subject. During necropsy, no abscesses or other infectious foci were found on internal tissues. Weights of organs that could be removed in their entirety (kidneys, spleen, liver, lung, ALNs, PLNs, and cecum) did not differ significantly between sleep-deprived and yoked rats except that sleep-deprived rats had heavier kidneys [sleep deprived 3.26 ± 0.07 (SE) g, yoked 3.03 ± 0.05 (SE) g, P < 0.0003] and heavier PLNs [sleep deprived 0.12 ± 0.01 (SE) g, yoked 0.10 ± 0.02 (SE) g; group by time interaction P < 0.04] than did yoked rats.

Microbial populations and species in body tissues. All blood specimens were sterile, and therefore microbial populations detected in body tissues were not likely associated with bloodstream infection that eventually develops (19). Paired analyses showed that sleep-deprived rats had significantly more infected tissues than did the yoked rats (aerobic cultures, P < 0.008; anaerobic cultures, P< 0.003). Microbial species identified in aerobic or anaerobic cultures were grouped in Table 1 according to Bergey's classifications (31).

View this table:
Table 1.

Species identified in sleep-deprived and comparison animals classified according to Bergey

MLNs were sterile in sham-control rats. Sleep deprivation resulted in microbial invasion of these lymph nodes, and both incidence and numbers of CFUs were higher in the sleep-deprived groups than in the yoked groups (incidence, 18 of 25 sleep deprived, 6 of 23 yoked, P< 0.004; aerobic CFUs, P < 0.005, anaerobic CFUs, P< 0.02) (Fig. 1). By experimentalday 5, the numbers of live bacteria in the MLNs under both aerobic and anaerobic culture conditions were significantly increased in sleep-deprived rats but not in yoked rats (aerobic CFUs, P< 0.003; anaerobic CFUs, P < 0.001). After day 5, the numbers of CFUs in MLNs of yoked rats were few or absent, whereas the average numbers for sleep-deprived rats ranged from 460 to 1,630 CFUs/g under aerobic or anaerobic culture conditions at each time interval. MLNs of sleep-deprived rats also contained more species than did those of yoked rats (average species number in MLN 1.7 ± 0.3 SE in sleep-deprived and 0.5 ± 0.3 SE in yoked groups, P < 0.008). The predominant bacteria in MLNs of sleep-deprived rats wereEnterobacteriaceae, principally K. pneumoniaeand Enterobacter cloacae. Escherichia coli, considered the most common translocating bacterium, was found in MLNs only once in a sleep-deprived rat on day 20 and once in a yoked rat onday 5. After the Enterobacteriaceae, the next most prevalent bacteria in MLNs were Staphylococcus species, principally S. aureus and S. sciuri. Other bacteria identified in the MLNs of sleep-deprived rats included those belonging to Group 20 (Actinomyces, Corynebacterium, andEubacterium); Group 4A, (Acinetobacter spp., F. breve, P. aeruginosa, and X. maltophilia); and occasionally to Group 19 (Erysipelothrix andLactobacillus), Group 6, and Group 8. All bacteria identified in the MLNs of sleep-deprived rats can be found in mammalian intestine.

Fig. 1.

Number of colony forming units (CFUs) in mesenteric lymph nodes (top), ileum (middle), and cecum (bottom) of sham-control, yoked, and sleep-deprived rats, cultured under aerobic (left) and anaerobic (right) conditions. Geometric means ± SD.

Bacterial invasion of the MLNs was associated with bacterial overgrowth in the intestine (Fig. 1). Populations of live aerobic bacteria were significantly greater in the ileum and cecum of sleep-deprived rats than in those of yoked and sham-control rats (ileal CFUs, P < 0.005; cecal CFUs, P < 0.0001). Populations of facultative anaerobic bacteria also were significantly greater in ileal specimens from sleep-deprived rats than in those from yoked and sham-control rats (ileal CFUs, P < 0.05; cecal CFUs, NS). Intestinal homogenates cultured on EMB plates revealed large numbersEnterobacter (E. cloacae and E. agglomerans) and K. pneumoniae at each duration of sleep deprivation; on day 10, these levels were ninefold higher than those of sham-control rats (Fig. 2). Populations of E. coli increased dramatically in the ileum of 20-day sleep-deprived rats cultured under anaerobic conditions to levels 21-fold and 13-fold higher than in sham-control and yoked rats, respectively (Fig. 2). The number of CFUs of cultures from cecum showed a similar pattern. Under aerobic culture conditions, cecal gram-negative bacteria populations cultured on EMB plates from 20-day sleep-deprived rats were 37-fold higher than those of sham-control rats. Overgrowth of non-lactose-fermenting bacteria also was apparent, as can be seen by comparing the differences in the total number of CFUs on blood agar (Fig. 1) with those on EMB agar (Fig. 2).

Fig. 2.

Average number of CFUs of gram-negative bacteria cultured on eosin methylene blue (EMB) plates from homogenates of ileum of sham-control, yoked, and sleep-deprived rats across experimental period. Total number of CFUs represented by each bar is comprised of those bacteria under aerobic (A) and anaerobic (B) culture conditions that were identified as Klebsiella pneumoniae or Enterobacter cloacae (solid bars) and Escherichia coli (stippled bars).

Colonization of the kidney, spleen, and liver is shown in Fig.3. The number of CFUs in liver was highly variable. Cultures of liver in sleep-deprived rats showed a strong tendency to contain more species of bacteria than those of yoked rats (average species number in liver, 1.3 ± 0.4 SE in sleep-deprived and 0.2 ± 0.4 SE in yoked groups, P < 0.08). Most frequently detected were bacteria from Group 4 (F. breve and occasionallyAlcaligenes, Pseudomonas, Xanthomonas, andBacteroides). Next in prevalence were microbes from Group 19 (Erysipelothrix and Lactobacillus), Group 17 (Staphylococcus and Streptococcus), Group 5, and occasional bacteria from Group 20 and Group 6. The kidney and the spleen were infected infrequently in sleep-deprived rats and rarely in yoked rats; the differences between groups were not significant. Bacteria species cultured in kidneys from sleep-deprived rats wereKlebsiella, and those from spleen were nearly all from Group 17 (S. aureus, S. haemolyticus, S. sciuri, S. xylosus, Aerococcus,Enterococcus), except for one instance from Group 4B. All but 1 of 48 swab specimens of the peritoneal cavity were sterile, indicating colonization rather than contamination. The swab specimen containing bacteria was collected from the peritoneum of a yoked rat on day 5, and the bacteria were identified as S. capitis, an organism that was not detected in other tissues.

Fig. 3.

Average numbers of CFUs in liver, kidney, spleen, and lung under aerobic (left) and anaerobic (right) culture conditions in sham-control, yoked, and sleep-deprived rats across experimental period. Geometric means ± SD.

Cultures from lung were positive for bacterial growth in 23 of 25 sleep-deprived animals, compared with 16 of 23 yoked animals and 3 of 5 sham-control rats (incidence, sleep-deprived vs. yoked, P < 0.07). Exceptionally high bacteria counts in lungs of three animals raised the group means shown in Fig. 3 for the 5-day yoked group and the 20-day yoked and sleep-deprived groups. Otherwise, the number of CFUs in positive cultures ranged from 100 to 11,000 CFUs/g in sleep-deprived and yoked animals and from 220 to 3,600 CFUs/g in sham-control animals. Lack of correspondence between species found in the lung and those found on swabs of the throat and trachea suggested that aspiration during exsanguination procedures was not the means of introduction of microorganisms in the lung. In 75% of all cases, one or more species identified in lung was also identified in liver or a lymph node. Enterobacteriaceae were identified in 13 of 23 positive cultures from lung in the sleep-deprived group, in 7 of 16 positive cultures in the yoked group, and in 1 of 3 positive cultures in the sham-control group. Enterobacteriaceae cultured were predominantly E. cloacae and less frequently Klebsiellaspp. Staphylococcus was identified in 37% of positive cultures of lung among groups. Thirty-six percent of positive cultures contained bacteria from Group 4 (principally F. breve), which are considered to be of low virulence. Swab specimens of the thoracic cavity were positive in 6 of 49 specimens (3 from sleep-deprived rats and 3 from yoked rats). Each of the six corresponding cultures from lung contained species not detected on the swab (e.g., K. pneumoniae), further supporting colonization rather than contamination.

Microbial populations in ALNs showed a late trend toward an increase in bacteria on aerobic cultures from sleep-deprived rats at 20 days (Fig.4). ALNs of sleep-deprived rats contained a larger number of aerobic species than did those of yoked rats (average aerobic species number in ALN, 1.3 ± 0.3 SE in sleep-deprived and 0.5 ± 0.3 SE in yoked groups, P < 0.03). These included only a few species from Groups 4, 5, or 17 in the sleep-deprived rats at 20 days. Significantly greater microbial populations and more species were found in the PLNs of sleep-deprived rats than in those of yoked and sham-control rats (aerobic CFUs, P < 0.04; anaerobic CFUs, NS; average species number in PLNs, 2.7 ± 0.3 SE in sleep-deprived and 1.3 ± 0.4 SE in yoked group, P < 0.008; Fig. 4). Species identified in the PLNs of sleep-deprived rats were predominantly from Group 17 but often included members of Group 5 (principally E. cloacae), Group 20 (principallyCorynebacterium), and Group 4 (principally F. breve). ILNs of sleep-deprived rats showed higher incidence, significantly greater numbers, and more species of bacteria than did those of yoked or sham-control rats (incidence, 80% of sleep-deprived and 43% of yoked rats, P < 0.016; aerobic CFUs, P < 0.0005; anaerobic CFUs, P < 0.05; average species number in ILNs, 2.2 ± 0.3 SE in sleep-deprived and 0.7 ± 0.3 SE in yoked groups, P< 0.002; Fig. 4). By day 20, ILNs of all sleep-deprived rats were infected and contained 330–40,700 CFUs/g tissue; only one yoked rat had infected ILNs (660 CFUs/g tissue). Species were predominantly Group 17, followed in order of frequency by species classified under Groups 4, 5, 19, and 20. The highest incidence of an individual species in the ILNs of sleep-deprived rats was that ofCorynebacterium, followed by F. breve and S. aureus. Skin dermatoses on the dorsal tail and plantar paws are likely portals of entry contributing to microbial loads in the lymph nodes of the extremities. Each skin sample contained from 3 to 10 species of bacteria that did not differ between sleep-deprived and yoked groups, except for Actinomyces and Lactobacillus. The latter were identified in the skin of about 25% of the sleep-deprived rats, in only 0–9% of the yoked rats, and in 0% of sham-control rats. The most common species were gram-positive cocci (Group 17), found in all specimens from both sleep-deprived and yoked rats. Twelve species of Staphylococcus were identified, principally S. aureus and S. sciuri, followed by S. simulans and S. haemolyticus. Gram-negative aerobic rods and cocci were frequently detected (principally F. breve), as were irregular, nonsporing gram-positive rods. Less frequent were facultative anaerobic gram-negative rods classified under 11 genera, followed by the regular, nonsporing gram-positive rods,Erysipelothrix and Lactobacillus.

Fig. 4.

Average numbers of CFUs in axillary, popliteal, inguinal, cervical, and mediastinal lymph nodes (LN) in aerobic (left) and anaerobic (right) cultures in sham-control, yoked, and sleep-deprived rats across experimental period, expressed as either geometric means ± SD or as log10 ± SD.

CLNs contained bacteria in all subjects (group differences NS; Fig. 4) and yet those of sleep-deprived rats contained significantly more species of aerobic bacteria than did those of yoked rats (average species number in CLNs, 2.7 ± 0.2 SE in sleep-deprived and 1.9 ± 0.3 SE in yoked groups, P < 0.039). Staphylococcuswas the principal genus, identified in 40 of 46 specimens, and S. aureus was by far the principal species. S. agalactiae andE. cloacae were identified frequently, but there was a very low rate of coincidental detection in the lung. Mediastinal lymph nodes contained live bacteria in 20–30% of subjects within each group (Fig. 4; NS differences). No species was predominant, and most were gram-negative aerobes (Group 4) and gram-positive cocci (Group 17).


The present results indicate that sleep deprivation causes invasion of normally sterile body tissues by indigenous pathogenic bacteria within the first several days of sleep deprivation. Throughout the course of sleep deprivation, abnormal control of bacteria was indicated by bacterial overgrowth in the intestine, polymicrobial infection of the MLNs, and increased invasion of extraintestinal body tissues. This chronic internal septic state was associated with progressive development of a deep negative energy balance and is known to culminate in lethal bacterial invasion of the bloodstream (19). The findings indicate that processes underlying bacterial disease begin early after the onset of prolonged sleep deprivation and that infection of normally sterile tissues precedes overt signs of morbidity.

MLN cultures of sham-control comparison rats were sterile but contained bacteria after 5 days of sleep deprivation (Fig. 1), which was the shortest duration studied and less than one-fourth of the expected survival time. Detection of live enteric, oxygen-tolerant bacteria in MLNs is considered key evidence of bacterial translocation from the intestine (6, 17). The presence of translocated bacteria during discrete trials, such as in this study, indicates continuous migration of bacteria from the intestine, rather than a locally established infection (7). The main species identified in the MLNs (e.g., E. cloacae, K. pneumoniae, S. aureus) are considered among the chief translocating bacteria for their facility in escaping the intestine, for virulence, and for pathogenic consequences once invasion of extraintestinal sites has occurred (52). Infections of MLNs in sleep-deprived rats were polymicrobial, indicating translocation of a variety of bacteria that would include commensals, symbionts, and additional pathogens. Combinations of organisms can produce more virulent infections (4). Detection of enteric bacterial invasion of the liver, lung, kidney, and spleen suggests migration of bacteria to these sites where the bacteria are subsequently killed or pass on to other sites. In yoked rats, the presence of substantial numbers of bacteria in the MLNs of three of seven animals during the first 5-day period, but not during subsequent intervals, reflects the fact that this interval represents their period of most severe sleep deprivation (with total sleep time decreased below basal amounts by an average of 42% for any given 24-h period; Ref.18). Increased infection of lymph nodes of the extremities of sleep-deprived rats contributed to the septic burden and likely resulted from bacterial invasion through skin dermatoses, which are common to infectious disease states (48).

Bacterial translocation in sleep-deprived rats coincided with overgrowth of oxygen-tolerant species in the ileum and the cecum. Although bacterial translocation can occur without enteric overgrowth (15, 51), the numbers of bacteria of a particular strain that translocate to MLNs has been correlated with populations in the intestine (30, 47). The total bacterial populations in the cecum and ileum are strongly and positively correlated with the numbers that adhere to the ileal and cecal mucosa (34), which is the first step in the translocation process. In sleep-deprived rats, K. pneumoniae and E. cloacae composed a large proportion of the oxygen-tolerant bacteria in the ileum and the cecum and were the predominant species that were detected in the MLNs.

A certain level of antimicrobial control by the sleep-deprived animal is indicated by the observations that bacterial populations in extraintestinal sites did not progressively increase or appear established and that microbial invasion of the bloodstream had not yet occurred. As the septic insult continues, the sequence of resistance events, many of which have interdependent interactions and overlapping functions, would be expected to expand (10), directly or indirectly augmenting the survival potential of the host in the face of underlying deficits. For instance, the number of CFUs detected in the mesenteric lymph node tissue of rats deprived of sleep for 20 days was smaller than at preceding times, yet sustained bacterial translocation was likely given that cecal populations of bacteria cultured on EMB plates under aerobic conditions were sevenfold greater, whereas those cultured under anaerobic conditions were 37-fold greater than the respective populations in sham controls. If indeed there was increased local control of bacteria in MLNs in 20-day sleep-deprived rats, such control likely was short lived and achieved at great metabolic expense to the host. At this stage in the deprivation, the animal outwardly appears in poor health (21), hypercatabolism is maximal (19), and final failure of host defenses resulting in lethal bacteremia is imminent (19). Failure of host defenses in the sleep-deprived animal likely occurs sequentially as the residual capacity of defense system becomes exhausted. The sites through which bacteria eventually infect the bloodstream may vary among individuals and could include, for instance, the gastrointestinal tract and the lung.

Bacterial translocation from the gastrointestinal tract to the MLNs, spleen, liver, and kidney increases if the immune system is suppressed (6, 52). Sleep-deprived rats do not have anatomic defects of the intestine that might contribute to bacterial translocation before the onset of advanced morbidity (19, 28, 37). Structural abnormalities are not a necessary condition, and bacterial translocation can occur by direct intracellular penetration of enterocytes (1). Others have shown that a combination of oral antibiotic-induced bacterial overgrowth in the intestine and immunosuppression promote bacterial translocation. This combination resulted in lethal sepsis in mice within 14 days and was associated with only a few hundred bacteria in each mesenteric lymph node (7). Deficiencies of mucus and antibacterial secretory IgA are related to increased bacterial adherence to enterocytes but by themselves do not permit bacterial translocation, as has been determined through studies of food deprivation (11) and protein malnutrition (34). Gut barrier permeability is increased by nonlethal doses of endotoxin, which promote bacterial translocation in a dose-dependent manner (13) without apparent changes in intestinal morphology (14, 29), resulting in lethal systemic infection (13). Such a vicious cycle of bacterial translocation and endotoxemia is the proposed pathogenesis in experimental models of burn wound and systemic gut-origin sepsis, and endotoxemia occurs in human burn patients with sterile burn wounds and no infectious foci (3). Bacterial translocation in sleep-deprived rats implies the presence of increased endotoxin exposure and deficient local immunity but also systemic immunosuppression. Corroborating signs of an immunosuppressed state include the absence of abscesses on internal tissues despite colonization by Staphylococcus, an absence of fever despite an elevated hypothalamic thermoregulatory set point (42), and eventual primary bacteremia induced by endogenous opportunistic microorganisms, similar to the pathogenesis of sepsis in immune-compromised patients.

The clinical implications of bacterial invasion of body tissues are much broader than the mere presence or absence of bacteria suggests because of generated endotoxin, exotoxins, and cell wall fragments (51). Measurement of live bacteria generally is assumed to greatly underestimate both the number of bacteria that actually translocate and the biological impact on host defenses (2). For example, in a study of critically ill patients, 64% had microbial DNA in their blood but only 14% had positive blood cultures (33). The pathogenic microorganisms detected in the tissues of sleep-deprived rats are known to elicit a variety of defensive reactions to exert control over microbial presence, proliferation, and migration and to promote clearance of toxins (5). This internal battle between host and invading microorganisms generates cellular products that trigger cytokine and complement cascades and activate other cells downstream (10). The immune response can be enhanced or suppressed, depending on which cytokines are produced. For example, gram-positive bacteria such asStaphylococcus and Streptococcus act as superantigens and stimulate the production of the cytokines, tumor necrosis factor, and interleukin-1 (40), whereas endotoxin can stimulate a suppressive circuit mediated by PGE2 that downregulates the activity of macrophages and other effector cells throughout the body (13). Certain cardiovascular, metabolic, endocrine, and nutritional consequences of infectious processes further are expected (5). For these reasons, the presence of live bacteria in various tissues of sleep-deprived rats is a definitive sign of malfunction of host defenses, with dynamic and extensive physiological involvement.


The presence of live bacteria in lymph nodes and other tissues indicates abnormal control of indigenous bacteria and production of a chronic antigenic challenge to the sleep-deprived host. Immune suppression is necessary for this type of bacterial disease to develop, although the specific mechanisms responsible for immune suppression are yet to be determined. Sleep quickly reverses the abnormalities in neuroendocrine parameters and energy expenditure of sleep-deprived rats and health is regained (22), implying that the restorative nature of sleep may include reversal of immune suppression and bacterial disease processes before the final events associated with septicemia occur. Because humans and animals exhibit similar consequences of prolonged sleep deprivation, where such comparisons can be made (20), it may be considered that abnormalities of host defenses including immune suppression and chronic antigenic challenge would render ordinary sleep-deprived individuals susceptible to disease and exacerbate existing disease and complicate recovery in sleep-deprived patients.


We thank Mike Straign for identifying microbial species in culture, Elizabeth Tolley for consultation on statistics, Kenneth A. Kudsk and Thaddeus S. Nowak, Jr. for manuscript critique, and, Lawrence Clay and Jason M. Scott for technical assistance.


  • Portions of this work were presented at the 11th Annual Meeting of the Association of Professional Sleep Societies and the 14th Congress of the European Sleep Research Society.

  • The principal research support was provided by the National Heart, Lung, and Blood Institute Grant HL-59271. Additional support was provided by National Institutes of Health Grants NS-26429, NS-25378, and CA-21765, and the American Lebanese Syrian Associated Charities.

  • Present address for L. A. Toth: Dept. of Pharmacology, Southern Illinois Univ. School of Medicine, Springfield, IL 62794.

  • Present address and address for reprint requests and other correspondence: C. A. Everson, Neurology Research 151, VA Medical Center, Milwaukee, WI 53295 (E-mail: ceverson{at}

  • 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. §1734 solely to indicate this fact.


View Abstract