Regulatory, Integrative and Comparative Physiology

Physical activity alters the brain Hsp72 and IL-1β responses to peripheral E. coli challenge

M. Nickerson, G. F. Elphick, J. Campisi, B. N. Greenwood, M. Fleshner


Physically active rats have facilitated heat shock protein 72 (Hsp72) responses after stressor exposure in both brain and peripheral tissues compared with sedentary rats (11). This study verifies that physically active animals do not have elevated Hsp72 levels compared with sedentary animals in the hypothalamus, pituitary, or dorsal vagal complex. We then examined whether 1) physically active rats respond more efficiently than sedentary rats to a bacterial challenge; 2) peripheral immune challenge elicits brain induction of Hsp72; 3) this induction is facilitated by prior freewheel running; and 4) Hsp72 upregulation produced by peripheral immune challenge results in a commensurate decrease in the proinflammatory cytokine IL-1β. Adult male Fischer 344 rats were housed with either a mobile or locked running wheel. Six weeks later, rats were injected intraperitoneally with saline or Escherichia coli and killed 30 min, 2.5 h, 6 h, and 24 h later. Serum endotoxin and IL-1β, and peritoneal fluid endotoxin and E. coli colony-forming units (CFUs) were measured. Hsp72 and IL-1β were measured in hypothalamus, pituitary, and dorsal vagal complex. The results were that physically active rats had a faster reduction in endotoxin and E. coli CFUs and lower levels of circulating endotoxin and cytokines compared with sedentary rats. E. coli challenge elicited significantly greater time-dependent increases of both Hsp72 and IL-1β in hypothalamus, pituitary, and dorsal vagal complex of physically active animals but not sedentary animals. Contrary to our hypothesis, increases in Hsp72 were positively correlated with IL-1β. This study extends our findings that physical activity facilitates stress-induced Hsp72 to include immunological stressors such as bacterial challenge and suggests that brain Hsp72 and IL-1β responses to peripheral immune challenge may contribute to exercise-mediated resistance to long-term sickness.

  • exercise
  • immune
  • cytokine
  • pituitary
  • hypothalamus
  • nucleus of the solitary tract

regular, voluntary, physical activity, in the form of wheel running, protects organisms from many of the negative consequences of exposure to physical and psychological stressors such as exposure to footshock (27) and tailshock stress (37, 43, 73). Additionally, exercise protects immune cells from H2O2-induced apoptosis in vitro (5), prevents stress-induced immunosuppression (73) and enhances natural killer cell cytotoxicity in vivo (50). Finally, rats given access to a running wheel exhibit a faster and greater Hsp72 response to exhaustive exercise (11) and inescapable tailshock (11) in a number of central and peripheral tissues. This facilitated induction of Hsp72 that may contribute to the stress resistance exhibited by physically active rats (11, 58, 59).

Hsp72 is the stress-inducible member of a large family of heat shock proteins. It is a 72-kDa protein that is upregulated in response to a wide variety of stressors, and it facilitates cell survival. Specifically, Hsp72 prevents stress-induced cellular damage such as protein denaturation and aggregation (15, 58). Induction of Hsp72 creates a stress-tolerant environment, and organisms with greater Hsp72 induction are better able to cope with a variety of stressors. Among these are heat stress (2, 64), cold stress (21), light stress (8), exposure to toxins (33), oxidative stress (68), tail shock stress, and forced exercise (11, 41, 42, 66). Most of these data were derived from peripheral tissues. Given the ubiquity of Hsp72 across tissues and species, it is reasonable to suggest that Hsp72 may also play an important role in protecting brain tissue during stress.

The purpose of the following experiments was to determine whether an immunological stressor (i.e., intraperitoneal Escherichia coli) would elicit Hsp72 in major stress-reactive brain regions, whether Hsp72 induction in the brain would be facilitated by prior habitual wheel running, and whether an upregulation of Hsp72 would correlate with a more rapid cessation of E. coli-induced IL-1β. E. coli can be considered an “immunological stressor” because it elicits features of the stress response, including increases in plasma corticosterone (70), body weight loss (94), and the expression of proinflammatory cytokines (20). Freewheel running was used in this experiment rather than forced treadmill running because freewheel running can prevent stress-induced antibody suppression (71), learned helplessness behavior (43), alterations in brain monoamines (28), and suppressed NK cell function (27), and it is associated with a decrease in bacterial illness (29) and increased survival rate after bacterial infection (12). Forced treadmill training, on the other hand, elicits a series of physiological adaptations that define chronic stress. Among these are decreased thymus weight, adrenal hypertrophy, and decreased corticosterone binding globulin (72). Therefore, we used voluntary freewheel running in this study to determine the effects of physical activity on the brain’s Hsp72 response to immune challenge.

Growing evidence suggests that Hsp72 may indeed protect neural tissue from damaging effects of stress, and one of these ways may involve interactions with the proinflammatory cytokine IL-1β. For example, exposure to a variety of stressors, insults, or immunological challenges increases concentration of the proinflammatory cytokine IL-1β in the brain (18, 45), and elevated IL-1β in the brain has been reported to produce sickness behaviors (54), and chronic elevations of IL-1β can damage tissue (7, 88). IL-1β in the brain can also induce Hsp72 (22), and one role of Hsp72 induction may be to protect cells from cellular insult and to downregulate the IL-1β via intracellular signaling cascades.

For example, Hsp72 could protect neurons from excessive IL-1β by acting indirectly on the nuclear factor-κB (NF-κB) signal transduction cascade and, therefore, as an anti-inflammatory agent (74). NF-κB is a cytoplasmic molecule that is bound to the inhibitory-κB (I-κB) molecule until there is an inflammatory stimulus [i.e., peripheral bacterial challenge (9, 80)]. When activated, NF-κB translocates into the cell nucleus and initiates the critical transcription of proinflammatory cytokines (6). The proinflammatory cytokine IL-1β can be very potent when invoked centrally and may mediate sickness behavior (54), and the short-term induction of IL-1β is necessary for survival during illness (56, 90). In contrast, long-term exposure to IL-1β can cause tissue damage (7, 88) and prolonged sickness behavior (54). Therefore, Hsp72 may contribute to the protection of neurons by promoting I-κB binding to NF-κB in the cytoplasm, thereby inhibiting NF-κB signaling and decreasing the expression of proinflammatory cytokines in the brain (24, 47). Because IL-1β transcription is directed by NF-κB nuclear translocation (3, 13) and the NF-κB pathway is affected by Hsp72, there may be an interaction between Hsp72 and IL-1β in the brain, and such an interaction may be evident in response to a peripheral E. coli challenge.

Many brain regions and endocrine tissues have been shown to respond to peripheral stressors, including immune challenges (55). The current study examined the hypothalamus, pituitary, and dorsal vagal complex (DVC) for several reasons. First, the pituitary and hypothalamus mediate stress-induced endocrine responses, sickness-induced anorexia (81, 82), and febrile responses (62). Second, the DVC was examined because it contains many stress and immune responsive nuclei, including the nucleus of the solitary tract (NTS). The NTS is of interest because it receives afferent vagal information, responds to peripheral bacterial challenge (30), and modulates inflammatory activity in the brain (93).

Although no previous studies have tested the effect that bacterial challenge might have on Hsp72 in the brain, it is clear that the brain is affected by bacterial challenge (67, 83). For example, bacterial challenge has been reported to induce a c-Fos response in the brain (30, 92), and Hsp72 is detectible in similar brain regions in response to a variety of nonbacterial stressors (51, 8385). Thus, given the strong links between immune function and the central nervous system (38, 40), it is reasonable to suggest that bacterial challenge may also induce Hsp72 in the brain. Therefore, we hypothesize that physically active (PA) rats, despite similar baseline Hsp72 levels, will demonstrate a faster and greater brain Hsp72 response to E. coli challenge than sedentary (Sed) animals. In addition, we predict that PA animals will return to baseline levels of brain Hsp72 more quickly than Sed animals, signifying a more efficient response. We expect a greater or faster brain Hsp72 response to result in a decrease of brain IL-1β in response to bacterial challenge in PA vs. Sed animals. Finally, as a result of this more efficient central response, we expect that PA animals will exhibit a faster clearance of E. coli.



Adult male Fischer rats (n = 36; 12 wk, 300–350 g) were individually housed in Plexiglas cages (60 cm × 30 cm × 24 cm) with food and water ad libitum. Colonies were maintained in a pathogen-free barrier facility with a 12:12-h light-dark cycle. After arrival, animals were housed for 1 wk before the onset of experimentation. A 2 (Sed, PA) × 4 (30 min, 2.5 h, 6 h, 24 h) experimental design was used. PA rats were housed in cages with running wheels (circumference 1.081 m) and had access to these wheels at all times. Sed animals were housed with locked running wheels. Running was monitored by computer using Vital View software (Mini Mitter, Bend, OR). Basal levels of Hsp72 and IL-1β were measured in nonbacterially challenged Sed controls. All animal procedures were approved by the University of Colorado Institutional Animal Care and Use Committee.

Baseline Hsp72 Levels

Although it has been previously reported that PA alone does not result in altered baseline Hsp72 levels (9, 11), we did a separate study to verify that PA does not alter brain Hsp72 in Fischer rats in our environment. In this study, PA rats (n = 8) were housed in cages with running wheels, as described above. Sed animals (n = 8) were housed with locked running wheels and kept in the same room. After 6 wk of running or exposure to a locked running wheel, animals were decapitated and brains were microdissected for hypothalamus, pituitary, and DVC. Tissue was collected throughout the day to account for circadian factors and to best match the time points of the E. coli experiment.

Bacterial Culture

E. coli (ATCC 15746) were purchased from American Type Culture Collection (Bethesda, MD). Vial contents were rehydrated and grown overnight to maximal densities in 30 ml of brain-heart infusion (BHI; Difco Laboratories, Detroit, MI) at 37°C, 5% CO2. Cultures were then aliquoted into 1.0 ml BHI supplemented with 10% glycerol and frozen at −70°C. These aliquots were used as stock cultures. One day before experimentation, stock cultures were thawed and cultured overnight (∼19 h) in 35 ml of BHI (37°C, 95% air+5% CO2). Previously determined growth curves were used to quantify cultured bacteria (10). Cultures were then centrifuged for 15 min at 3,000 rpm, the supernatants were discarded, and bacteria were resuspended in sterile PBS.

Bacterial Challenge and Time Course

Rats (n = 36) were given an intraperitoneal injection of 2.5 × 108 colony-forming units (CFUs) or an intraperitoneal injection of PBS vehicle (n = 6) and were returned to their home cages after injection. Only Sed rats were tested as controls because it has been previously reported and verified in the study above that basal brain Hsp72 is not elevated by wheel running alone (11, 91) and that IL-1β is not expressed in the brain due to wheel running alone (unpublished observations). To control for any changes in Hsp72 or IL-1β across time, nonbacterially challenged controls were killed at each time point (2 per time point). Body weights were monitored daily, and no significant difference was found in weight gain between groups before immune challenge (data not shown). Animals were killed 30 min, 2.5 h, 6 h, and 24 h after injection (4 PA, 4 Sed, and 1–2 vehicle controls at each time point).

Brain Microdissection

At the various time points after E. coli or saline injections, animals were decapitated, and brains were quickly removed. Rongeurs were used to crack the skull and to expose the brain. With the use of curved forceps, the optic nerve was severed, and the brain was eased out of the skull onto a chilled glass plate. From the base of the skull, pituitary was removed from membranous case with INOK-5 forceps (Fisher Scientific, Pittsburg, PA) and immediately flash-frozen. The remaining dissections occurred on a chilled glass plate.

With the use of a razor blade, the brain was partitioned just above the dorsal vagal complex. With the use of the “V” shape of the NTS area on the dorsal side of the brainstem, the DVC was trimmed out (excess tissues on the sides was removed and only the dorsal half of tissue retained). With the razor blade, the next partition of the brain was made; a slice just rostral of the optic nerve was made to remove the periform cortex from the central regions. The remaining tissue was then positioned so that the ventral side was facing the dissector, and remains of the optic nerve were removed. With the use of INOK-5 forceps, the hypothalamus was removed; the points were used like razors to slice and pinch 2–3 mm into the tissue below the mammary bodies, and attention was given to avoid thalamic tissue. Each dissected region was flash frozen in liquid nitrogen. Dissections were completed within 4 min. Trunk blood was collected in a 15-ml conical tube and stored on ice. Serum was collected and frozen for later measurement of serum cytokines.

Blood was spun at 2,000 rpm in a refrigerated centrifuge for 15 min. After serum was pipetted off, it was aliquoted and stored at −20°C until time of assay. Blood was spun at 2000 rpm in a refrigerated centrifuge. After serum was pipetted off, it was aliquoted and stored at −20°C until time of assay. Undiluted serum samples were tested in IL-1β (R&D Systems, Minneapolis, MN) and IL-1ra (Biosource, Camarillo, CA) ELISA assays. Assays were performed following manufacturers’ instructions. Blood and peritoneal endotoxins were measured by quantitative chromagenic limulus amebocyte lysate (LAL, Cambex, Westborough, MA). In brief, 50 μl of sample were pipetted into a 96-well plate, and LAL was added. After 10 min, 100 μl of substrate solution are added, and the reaction is stopped after 16 min and read at 405–410 nm on a plate spectrophotometer. Serum samples were not collected in an endotoxin-free environment. Therefore, vehicle animals exhibit some endotoxin contamination above zero.

Live Bacterial Assessment

Live bacterial load in the peritoneal cavity and in the blood was determined using MacConkey II agar plates. These plates inhibit gram-positive bacterial growth and are selective for aerobic gram-negative bacilli, such as E. coli (Fisher Scientific). Peritoneal cavity samples were obtained by sterile lavage. Fluid was placed into a sterile snap cap and kept on ice until diluted from 1:10–1:109 in sterile PBS. One milliliter of each dilution 106-109 was placed under a sterile hood on an agar plate. Plates were incubated at 37°C, 5% CO2 for 22 h. CFUs were counted using an electric handheld colony counter (Fisher Scientific). Numbers from plates with less than 70 CFUs were used to determine total bacterial peritoneal cavity load. Dilutions factors 108 and 109 were used in statistical analysis.

Tissue Processing and Cytokine Characterization

Sonication buffer was added to each tissue sample in the following volumes: dorsal vagal complex, 500 μl; hypothalamus, 250 μl; and pituitary, 500 μl. The sonication buffer was composed of extraction buffer, 10× enzyme cocktail and phenyl methyl sulfonyl fluoride (PMSF). The extraction buffer was Iscove’s media [Gibco (Carlsbad, CA) 12440–020 lot #15K2952] with 5% fetal calf serum (Gibco 16000–044 lot #33N3051). The 10× enzyme cocktail was 13.1% 1 M amino-n-caproic acid [FW 131.2, Sigma (St. Louis, MO) A-7824 lot #75H0458], 3.722% 100 mM EDTA (FW 372.2, Mallinckrodt 4931 lot #4931 KMBR), 0.783% 50 mM benzamidine (FW 156.6, Sigma B-6506 lot #95H7705), and 82.485% ddH2O. A ration of 0.348% 20 mM PMSF [Böhringer Manheim (Mannheim, Germany) 837 091 lot #83654521] was dissolved in isopropanol.

Sonication was performed in the original storage tubes with a Fisher Sonic Dismembrator, model 100 (Fisher Scientific). All samples were kept on ice before and after sonication. Immediately after sonication, samples were spun at 4°C for 10 min at 10,000 rpm. Supernatant was carefully pipetted into new 1.5-ml snap caps and stored at −20°C until ELISA procedure. Undiluted homogenates were used in Hsp72 (Stressgen, Victoria, Canada) and IL-1β (R&D Systems, Minneapolis, MN) assays, which were performed per manufacturer’s instructions.

Total protein was measured to equate total tissue assayed using a Bradford assay. For the Bradford assay, tissue dilutions were made with ddH20 as follows: DVC, 1:20; hypothalamus, 1:30; and pituitary-1:20. 10 μl of each diluted sample was pipetted in duplicate into a 96-well plate. Bradford solution was added (250 μl/well) and read after 15 min at 595 nm. Protein values were calculated using linear regression from a standard curve of known protein concentrations.

Statistical Analyses

Baseline Hsp72 levels.

A Student’s t-test was used to determine whether the Sed and PA basal levels of Hsp72 were different. Sed animals displayed significantly higher levels of Hsp72 than PA animals in the pituitary (DF = 14, P = 0.0005). There was no significant difference in Hsp72 levels between Sed and PA animals in the hypothalamus or in the DVC. Levels of IL-1β were undetectable.

E. coli challenge.

A 2 × 4 ANOVA [2 (Sed, PA) × 4 (30 min, 3 h, 6 h, 24 h)] was performed to determine group effects of time or of physical activity and interactions between time and physical activity status. In addition, a one way ANOVA was used to determine differences between groups. Alpha was set at 0.05. In areas that revealed a significant difference, Fisher’s protected least significant difference (PLSD) post hoc test was used. StatView was used for all analyses, and figures show group means with SE bars. In figures, * symbolizes a significant change from vehicle baseline, and # marks significant differences between Sed and PA animals at a given time point.


Baseline Hsp72 Levels

Baseline Hsp72 levels were not different between PA and Sed animals in the hypothalamus or in the DVC. There was a reduction of Hsp72 in the pituitary of PA animals compared with Sed animals (P = 0.0005). Results are not on the same scale as results from the E. coli challenge experiment and are, therefore, expressed on a separate graph (Fig. 1). This discrepancy may be due to interassay variability. We were unable to detect IL-1β or IL-1ra in these samples (data not shown).

Fig. 1.

Sedentary (Sed) animals are represented by the light-colored bars, and physically active (PA) animals are represented by the solid bars. Baseline heat shock protein 72 (Hsp72) levels were not altered in the hypothalamus by PA alone. *There was a slight reduction of Hsp72 levels in the pituitary (P = 0.0005) of PA animals compared with Sed animals. PA did not alter baseline Hsp72 levels in the dorsal vagal complex (DVC).

E. Coli Challenge

Activity and body weights.

On average, animals ran 13.6 ± 2.86 km/wk. Animals at all time points of death ran equal distances. Sedentary and PA animals had similar body weights after initial 6 wk of locked wheel or freewheel exposure. After E. coli challenge, Sed and PA animals lost weight to the same degree (data not shown).

Peritoneal Cavity


Figure 2A depicts the rate of endotoxin clearance in Sed and PA animals. There was a main effect of time on the clearance of E. coli from the peritoneal cavity [F(3,24) = 17.92, P < 0.0001] and a main effect of physical activity [F(1,24) = 29.81, P < 0.0001]. This resulted in a significantgroup by time interaction [F(3,24) = 3.136, P = 0.0441]. Levels of endotoxin in Sed animals were significantly greater than PA animals at 30 min (P = 0.0017), 2.5 h (P < 0.0001), 6 h (P = 0.0062) time points. Compared with vehicle controls, Sed animals had significant levels of endotoxin controls at 30 min (P < 0.0001), 2.5 h (P < 0.0001), and 6 h (P < 0.0001) time points, and PA rats had elevated peritoneal cavity endotoxin at 30 min (P = 0.005) and 6 h (P = 0.0389) after E. coli injection. Both groups had cleared the endotoxin from the peritoneal cavity by 24 h.

Fig. 2.

A: endotoxin levels from the peritoneal cavity. All animals had more endotoxin than vehicle controls (#). PA animals cleared endotoxin from the peritoneal cavity faster than Sed animals and had significantly less endotoxin at 30 min, 2.5 h, and 6 h (*). B: levels of live Escherichia coli in the peritoneal cavity after E. coli challenge. Vehicle and 30 min timepoints were unavailable. It is clear though that at 2.5 h, 6 h, and 24 h, PA animals had less live E. coli in the peritoneal cavity compared with Sed animals (#).


Figure 2B shows that PA animals had significantly fewer CFUs of living E. coli. There was a main effect of time [F(3,22) = 13.7, P < 0.0001] and a main effect of physical activity [F(1,23) = 8.4, P = 0.03], but there was not a significant interaction. PA animals exhibited significantly fewer CFUs at 2.5 h (P = 0.03), 6 h (P = 0.04), and 24 h (P = 0.007).



There was no main effect of time on blood endotoxin levels. There was, however, a main effect of PA [F(1,24) = 7.716, P = 0.01]. PA rats had lower endotoxin in the blood than Sed rats. Sed rats challenged with E. coli had elevated serum endotoxin compared with vehicle controls (P = 0.0029) and compared with PA animals at the 24-h time point (P = 0.007), as shown in Fig. 3A.

Fig. 3.

A: circulating levels of endotoxin were significantly greater in Sed animals after 24 h compared with vehicle (*) animals or PA animals (#).B: serum IL-1β was significantly elevated in Sed animals above vehicle (*) and PA (#) animals at 30 min, 2.5, and 6 h. PA animals exhibited significant elevations of IL-1β at 2.5, 6, and 24 h.


Figure 3B shows that there was a significant effect of PA [F(1, 24) = 8.08, P = 0.001] and a significant effect of time [F(1,18) = 18.04, P < 0.01] on circulating levels of IL-1β. Post hoc analyses revealed that E. coli challenge reliably increased IL-1β above vehicle in Sed (P = 0.0344) but not PA after 30 min. This induction remained elevated at 2.5 h (P = 0.009) and at 6 h (P < 0.0001) in Sed animals and returned to baseline levels by 24 h. PA animals had high levels of IL-1β after 30 (P = 0.035) min, and these levels remained significantly elevated after 2.5 h (P = 0.046) and after 6 h (P = 0.05). IL-1β levels returned to baseline in PA animals by 24 h. IL-1β levels were significantly greater in Sed animals compared with PA animals 30 min (P = 0.033), 2.5 h (P = <0.001), and 6 h (P < 0.0001) after E. coli challenge.



There was a significant effect of Hsp72 induction in the hypothalamus across time [F(3,22) = 6.891, P = 0.0019] and a significant effect of exercise status [F(1,22) = 22.979, P < 0.0001]. A group-by-time interaction was also revealed [F(3,22) = 5.611, P = 0.0052]. Sed animals, compared with vehicle controls, revealed a significant induction of Hsp72 at the 30-min time point (P = 0.04). PA animals exhibited significantly elevated levels of Hsp72 in the hypothalamus after 30 min (P = 0.0027) and 6 h (P < 0.0001). As depicted in Figure 4A, Hsp72 induction was greater in the PA than Sed animals only at the 6-h time point (P < 0.0001).

Fig. 4.

A: hypothalamic Hsp72 was elevated above vehicle (*) in Sed animals 30 min after E. coli challenge and in PA animals 30 min and 6 h after E. coli challenge. At the 6-h time point, PA animals had significantly greater Hsp72 than Sed animals at the same timepoint (#). B: IL-1β was induced above vehicle (*) in Sed animals 30 min after E. coli challenge and in PA animals 30 min and 6 h after E. coli challenge. PA animals had significantly greater IL-1β at 6 h than Sed counterparts (#).


IL-1β levels in the hypothalamus were higher in PA animals challenged with E. coli compared with Sed counterparts. There was a significant group effect of physical activity [F(1,26) = 8.438, P = 0.0074], but not a group-by-time interaction (Fig. 3B). Post hoc analyses revealed that E. coli challenge reliably increased IL-1β above vehicle in both Sed (P = 0.0344) and PA (P = 0.036) groups after 30 min. This induction returned to baseline after 6 h in Sed animals. On the other hand, PA animals had high levels of IL-1β after 30 min, and these levels remained significantly elevated after 6 h (P = 0.022). IL-1β levels returned to baseline by 24 h. Six hours after E. coli challenge, IL-1β levels were significantly greater in PA than Sed rats (P = 0.004).



As shown in Fig. 5A, PA animals displayed a significantly higher level of Hsp72 induction in the pituitary after intraperitoneal E. coli challenge at all time points [F(1,24) = 27.009, P < 0.0001], but there was not a main effect of time or a group-by-time interaction. Post hoc tests revealed a significant induction of pituitary Hsp72 at all time points among PA animals compared with control (30 min, P = 0.0023, 2.5 h, 6 h, and 24 h < 0.0001). Hsp72 was elevated in the pituitary of Sed animals after 30 min (P = 0.0390) and 2.5 h (P = 0.05) compared with vehicle. Reflecting a persistence of elevated Hsp72 in PA animals, protein values were significantly higher than Sed counterparts 6 h (P < 0.0001) and 24 h (P = 0.0050) after E. coli challenge.

Fig. 5.

A: Hsp72 in the pituitary was significantly induced (*) in Sed animals above vehicle controls 30 min and 2.5 h after E. coli challenge. PA animals exhibited a significant Hsp72 induction at all time points examined. Compared with Sed animals, PA animals had significantly more Hsp72 at 6 h and 24 h after E. coli challenge (#). B: IL-1β was not induced in Sed animals at any time point examined and significantly induced (*) in PA animals 2.5 h and 6 h after E. coli challenge compared with vehicle controls. PA animals had significantly more pituitary IL-1β than Sed animals 2.5 h after challenge.


Intraperitoneal E. coli challenge induced a time-dependent increase in IL-1β in the pituitary [F(3,24)=1.584, P = 0.039, Fig. 5B]. Post hoc analyses revealed that PA animals had a significant increase in IL-1β compared with vehicles 2.5 h (P = 0.0002) and 6 h (P = 0.0436) after E. coli challenge. The induction of pituitary IL-1β was significantly greater in PA animals than Sed animals 2.5 h (P = 0.015) after E. coli challenge.



In PA animals, E. coli challenge resulted in a significant Hsp72 induction in the DVC of PA animals [F(1,24) = 38.456, P < .0001]. As shown in Fig. 6A, post hoc analyses revealed that Hsp72 levels in PA animals were elevated compared with vehicle controls after intraperitoneal E. coli injection: 30 min (P = 0.0001), 2.5 h (P < 0.0001), 6 h (P = 0.0003), and 24 h (P < 0.0001). At no point was Hsp72 significantly induced among Sed animals. After E. coli challenge, levels of Hsp72 in the DVC of PA animals were significantly elevated above Sed animals at all time points (30 min, P = 0.0034; 2.5 h, P = 0.0008; 6 h, P = 0.0208; 24 h, P = 0.0011).

Fig. 6.

A: Hsp72 in the DVC was not induced above vehicle at any time point in Sed animals. However, Hsp72 was significantly induced in PA animals after E. coli challenge at all time points examined compared with both vehicle controls (*) and Sed counterparts (#). B: IL-1β levels illustrated were not induced at any time point after E. coli challenge in Sed animals compared with vehicle controls. PA animals displayed a significant induction of IL-1β 30 min and 24 h after E. coli challenge compared with control animals (*) but never had significantly greater IL-1β compared with Sed animals at the same time points.


The effect of E. coli challenge on DVC IL-1β can be seen in Fig. 6B. There were no significant main effects of group, time, or group-by-time interaction.


Animals that were physically active had less endotoxin in the circulation and faster clearance of the E. coli than sedentary rats after intraperitoneal E. coli challenge. This suggests that PA rats more efficiently responded to bacterial challenge. Both live CFUs and dead (endotoxin) measures of E. coli were more rapidly reduced in PA animals compared with sedentary animals. The more rapid clearance of E. coli from the peritoneal cavity of PA animals was marked by lower circulating proinflammatory cytokines over time. Circulating IL-1β peaked in Sed animals 6 h after challenge and was not elevated in PA animals. Because both groups of animals were injected with the same number of E. coli, the primary activator of the immune system in this experiment, it is probable that central mechanisms controlling sickness responses contribute to the accelerated clearance exhibited by PA animals.

The induction of central IL-1β is essential in the initial phase of an immune challenge. Because PA animals had a greater and more rapid upregulation of central IL-1β, they may have had a more robust early fever that aided in the rapid clearance of E. coli from the peritoneal cavity. Further, PA animals have increased levels of circulating natural immunoglobulin (31), and this may be attributed to centrally mediated alterations of peripheral IL-1β and IL-6 interactions (48, 63).

Interestingly, IL-1β was significantly elevated in the brains of PA animals 30 min after E. coli challenge, but PA rats had no significant elevation of circulating IL-1β. These data suggest that it is unlikely that circulating IL-1β is the primary source of central IL-1β (11). Although acute upregulation of brain IL-1β may be beneficial to host defense, persistent central IL-1β may compromise long-term neural health. Therefore, the anti-inflammatory role of Hsp72 is highlighted in this experiment, in which a rapid induction of brain IL-1β in PA animals elicited a commensurate induction of Hsp72 and a faster recovery from sickness. PA animals, which displayed higher levels of IL-1β and Hsp72, had faster endotoxin clearance and expressed lower levels of circulating cytokines.

This is the first study to show that an intraperitoneal E. coli challenge elicits a brain Hsp72 response and that voluntary wheel running facilitates the brain’s Hsp72 response to immune challenge, despite similar baseline levels of Hsp72. These results are especially intriguing because they highlight the wide variety of stressors capable of inducing Hsp72 (10, 33, 35, 36) and demonstrate the generalizability of the potentiating effect of physical activity on stress-induced Hsp72. We hypothesized that an increase in Hsp72 would result in a decrease of IL-1β. Our results, however, support a parallel increase of Hsp72 and IL-1β in most brain regions examined. In fact, brain IL-1β responds almost as robustly as Hsp72 during the initial 6-h time period in the hypothalamus and pituitary of PA animals but not DVC.

Although not detected with current study methodology, it remains feasible that a more rapid reaction of brain Hsp72 to peripheral challenge sets the stage for constrained transcription of brain proinflammatory cytokines later in the course of an infection or in response to subsequent challenge (47, 87). Hsp72, as a preemptive anti-inflammatory mechanism, may be more accurate and less costly to the organism than a solely postscriptive anti-inflammatory mechanism. PA animals have a more robust central Hsp72 response to intraperitoneal E. coli than Sed animals, and it is initiated as quickly as 30 min and persists up to 6 h in all brain areas examined. Hsp72 remains elevated up to 24 h in the pituitary and DVC of PA animals.

The mechanisms through which physical activity status influences the brain Hsp72 stress response remain unknown. One possibility is that neurotransmitters altered by freewheel running, for example, NE (25), serotonin (43), dopamine (39), and glutamate (32), are responsible for a facilitated response of Hsp72 in the brain. Work done on brain noradrenergic responses reveals that PA animals display increased concentrations of brain NE in some regions (26, 28). NE is a viable candidate as an induction and/or releasing signal for Hsp72 because it is sensitive to physical activity status (25) and alters an array of intracellular processes when bound (17, 34, 53) and has been reported to mediate Hsp72 induction in peripheral tissues (69, 79).

One further possibility is that physical activity alters Hsp72 induction through pretranscriptional mechanisms. Specifically, the transcriptional factor heat shock factor 1 (HSF-1) mediates Hsp70 transcription. The expression of HSF-1 is upregulated by increases in PGs (49), estrogen receptor binding (77), and protein kinases (76). There is evidence for both noradrenergic (1, 61) and exercise (4, 14, 19, 52) interactions with PGs, estrogen receptors, and protein kinases. It is unknown, however, whether E. coli challenge alters these relationships.

Future studies will determine which specific brain cell types upregulate Hsp72 after E. coli. This is important because expression of Hsp72 is cellular and stressor specific (60). It is clear that Hsp72 can be found in neurons after a variety of stressors, including administration of kainic acid (46, 89), glutamate toxicity (85), cerebral ischemia (16, 57), and hyperthermia (65). Glial cells are also capable of expressing Hsp72 (23, 65, 75, 78). Interestingly, there is some evidence to suggest that mature neurons are incapable of making Hsp72 and that glial cells may make Hsp72 and then transfer it to neurons (44).

In conclusion, the results of the current study support the hypothesis that PA animals generate a faster and greater Hsp72 response to an immunological stressor (intraperitoneal E. coli) than Sed animals. The results extend our previous work using tailshock stress and exhaustive exercise stress to include bacterial challenge and suggest that the potentiated Hsp72 response found in PA animals is generalizable to a wide range of stressors. The altered central IL-1β and Hsp72 response may contribute to the faster bacterial clearance and reduced sickness exhibited by PA animals (10, 31). Hsp72 induction can downregulate IL-1β production in the same cell (86). The data reported here, however, do not agree with this past cellular work in that both Hsp72 and IL-1β changed in a nearly parallel fashion. This discrepancy in our findings could be due to our procedures (i.e., microdissecting and sonicating whole brain regions). Perhaps the isolation of individual cells and examination of coexpression of these proteins per cell across time may be needed to detect the predicted reciprocal relationship. Future studies will aim to isolate the cellular signals responsible for brain Hsp72 induction and to better understand the relationship between brain Hsp72 and IL-1β mRNA.

It is of interest to consider the implication of these findings from a broader perspective. It is unclear which of the tested environments (Sed. vs. PA) represents a normal physiological state. Perhaps, the sedentary condition is the abnormal physiological state and that the physically active condition is the “normal” physiological state. Given this perspective, our results would be equally demonstrative of the stress susceptibility of being sedentary. It is feasible to hypothesize that the lack of physical activity has hindered the expression of adaptive cellular mechanisms such as Hsp72 induction that are in place to successfully cope with stress. Regardless of one’s interpretation, the results of the current study clearly suggest important differences in stress physiology between physically active and sedentary organisms at the cellular level.


  • 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.


View Abstract