Time-dependent mediators of HPA axis activation following live Escherichia coli

Z. R. Zimomra, V. M. Porterfield, R. M. Camp, J. D. Johnson

Abstract

The hypothalamus-pituitary-adrenal (HPA) axis is activated during an immune challenge to liberate energy and modulate immune responses via feedback and regulatory mechanisms. Inflammatory cytokines and prostaglandins are known contributors to HPA activation; however, most previous studies only looked at specific time points following LPS administration. Since whole bacteria have different immune stimulatory properties compared with LPS, the aim of the present studies was to determine whether different immune products contribute to HPA activation at different times following live Escherichia coli challenge. Sprague-Dawley rats were injected intraperitoneally with E. coli (2.5 × 107 CFU) and a time course of circulating corticosterone, ACTH, inflammatory cytokines, and PGE2 was developed. Plasma corticosterone peaked 0.5 h after E. coli and steadily returned to baseline by 4 h. Plasma PGE2 correlated with the early rise in plasma corticosterone, whereas inflammatory cytokines were not detected until 2 h. Pretreatment with indomethacin, a nonselective cyclooxygenase inhibitor, completely blocked the early rise in plasma corticosterone, but not at 2 h, whereas pretreatment with IL-6 antibodies had no effect on the early rise in corticosterone but attenuated corticosterone at 2 h. Interestingly, indomethacin pretreatment did not completely block the early rise in corticosterone following a higher concentration of E. coli (2.5 × 108 CFU). Further studies revealed that only animals receiving indomethacin prior to E. coli displayed elevated plasma and liver cytokines at early time points (0.5 and 1 h), suggesting prostaglandins suppress early inflammatory cytokine production. Overall, these data indicate prostaglandins largely mediate the early rise in plasma corticosterone, while inflammatory cytokines contribute to maintaining levels of corticosterone at later time points.

  • corticosterone
  • prostaglandin
  • IL-6
  • indomethacin

activation of the hypothalamus-pituitary-adrenal (HPA) axis is one of the critical brain-mediated sickness responses that enhance survival of an organism during an immune challenge. The resulting elevation in circulating glucocorticoids, mainly cortisol in humans and corticosterone in mice and rats, liberate energy necessary to mount a fever and have numerous immunomodulatory effects that reduce the risk of septic shock and increase the chances of survival during infection (52, 53). For example, glucocorticoids suppress proinflammatory cytokines (2, 17, 34) and prostaglandin production (27, 38), stimulate anti-inflammatory cytokine production (16, 20), upregulate Fc receptors and major histocompatibility complex class II molecules on phagocytes (24), increase cell adhesion molecules on endothelial cells (57), enhance acute phase changes in the liver (3), and mobilize immune cells to sites of infection (10, 40). However, it remains unclear how and when various immune factors contribute to the initiation and maintenance of HPA activation during an immune challenge.

Numerous studies have investigated the role inflammatory cytokines, TNF-α, IL-1, and IL-6 play in HPA activation (11, 55) since these are the first cytokines released following immune activation (typically within 60–90 min). Studies generally find that inflammatory cytokines are sufficient, but not necessary, to activate the HPA axis during immune challenge. Administration of any one of the inflammatory cytokines potently stimulates the HPA axis and results in elevation of circulating corticosterone (11, 55). However, blockade of any individual inflammatory cytokine via administration of a receptor antagonist (12, 13, 50), neutralizing antibodies (46, 60) or use of selective cytokine or receptor knockout animals (31, 37, 58) fails to block HPA activation following LPS challenge. IL-6 appears to play the largest role since administration of IL-6 neutralizing antibodies (60), and IL-6 deficient mice (58) show some attenuation of corticosterone levels following LPS administration; however, a robust HPA response is still observed in these animals. The overall conclusion has been that all the inflammatory cytokines contribute to HPA activation and blockade of any individual cytokine fails to prevent HPA stimulation due to duplicity in their actions.

More recently, focus has turned to the role prostaglandins play in activation of the HPA axis during an immune challenge. In contrast to the soluble protein cytokines, prostaglandins are lipid mediators that can be rapidly (within minutes) released following immune stimulation via the conversion of arachidonic acid to PGH2 by constitutively expressed cyclooxygenase-1 (COX-1) and then converted to PGE2 by PGE synthase; furthermore, later production of PGE2 can be synthesized by the inducible COX-2 (39). Administration of PGE2 is sufficient to stimulate the HPA axis (43, 61, 62), but in contrast to cytokines, a nonselective COX inhibitor completely blocks the early (0.5–1 h) corticosterone response after LPS challenge (14, 23). Similar to inflammatory cytokines, prostaglandins are released by innate immune cells in local tissue and have been shown to signal the brain via activation of peripheral nerves such as vagal afferents in the liver (36) or released directly by endothelial cells composing brain microvasculature (23, 49, 51). The rapid production of prostaglandins compared with soluble proteins following an immune challenge led to the hypothesis that PGE2 mediates the initial activation of the HPA axis, while inflammatory cytokines help maintain HPA activation at later time points.

We present a series of studies that examined the contribution of both lipid (i.e., prostaglandin) and protein (i.e., IL-6) immune products across time in mediating activation of the HPA axis following intraperitoneal injection of live Escherichia coli. Previous studies used purified LPS derived from the cell membrane of gram-negative bacteria to stimulate immune responses, but LPS is only one component in gram-negative bacteria cell membrane that selectively targets toll-like receptor-4, while whole bacteria possesses multiple pathogen-associated molecular patterns capable of activating several toll-like receptors and is more potent at activating other immune responses, such as complement proteins (5). Here, studies examined corticosterone levels at multiple time points (30, 60, 120, 240 min) following two concentrations of E. coli, and used indomethacin (a nonselective COX inhibitor) and IL-6 neutralizing antibodies to examine the contribution of prostaglandin and IL-6 in activation of the HPA.

MATERIALS AND METHODS

Animals

Adult male Sprague-Dawley rats (175–199 g; Harlan) were used for all studies in this investigation. Animals were single housed in Plexiglas cages with food and water available ad libidum and a 12;12-h light-dark cycle beginning at 07:00. Animals were given 1 wk to adjust to their new environment before any study was performed. Animals were handled once a day in the morning for 4 days before any study was performed to help minimize their stress response during the studies. All studies began in the morning (∼09:00–10:00 h) to minimize the effects of the natural rise of corticosterone in the early evening. Care and use of the animals was in accordance with protocols approved by the Kent State University Institutional Care and Use Committee. All studies were designed to minimize pain and the number of animals used.

Drug Concentration and Administration

Indomethacin (Sigma-Aldrich), a nonselective COX inhibitor, was injected at a concentration of 5 mg/kg in a volume of 100 μl of sterile 33% dimethyl sulfoxide. Rat IL-6 antibody (cat. no. AF506; R&D Systems, Minneapolis, MN) was injected at a concentration of 4.2 μg/kg in a volume of 250 μl of sterile saline. All drugs were injected intraperitoneally.

Bacterial Growth

E. coli (0111:B4; ATCC 15746; American Type Culture Collection) was rehydrated and grown overnight in 40 ml of brain-heart infusion broth (DIFCO Laboratories) at 37°C at 5% CO2. Bacterial cultures were then aliquoted into 1 ml brain-heart infusion broth with 10% glycerol and frozen at −20°C. All studies used bacteria from these stock cultures. One day prior to experimentation, stock cultures were thawed and cultured overnight in 40 ml of brain-heart infusion broth at 37°C and 5% CO2. Bacteria were quantified by extrapolating from previously determined growth curves. Cultures were centrifuged for 20 min at 3,000 rpm, and supernatants were discarded. Bacteria were resuspended in sterile 0.9% saline at a concentration of 1 × 108 CFU/ml for studies 1, 2, and 3, and at a concentration of 1 × 109 CFU/ml for studies 4 and 5. Rats were injected intraperitoneally with 250 μl of E. coli (2.5 × 107 CFU for studies 1, 2, and 3 and 2.5 × 108 CFU for studies 4 and 5).

Blood Collection

For studies 1 and 5, rats were euthanized by decapitation and truck blood was collected in 10 ml EDTA-coated vacutainer tubes (BD PharMingen) and stored on ice until the completion of the study. Blood was centrifuged for 10 min at 4,000 rpm at 4°C, and plasma was collected and stored at −80°C until the time of assay. For studies 2, 3, and 4 tail vein blood was collected at designated time points. Rats were gently wrapped in a towel and lightly secured with a Velcro strap prior to tail blood collection. A lateral tail vein was nicked at the posterior end of the tail using a scalpel blade, and the tail was stroked from anterior to posterior to facilitate the movement of blood. Approximately 200–300 μl of blood was collected in an Eppendorf snap-cap tube within ∼60–90 s of removing the animal from the cage. Blood was centrifuged for 10 min at 10,000 rpm, and plasma was collected and stored at −80°C until time of assay.

Tissue Collection

For studies 1 and 5, a small specimen of liver was quickly harvested from the rats and frozen in liquid nitrogen, and then stored at −80°C until time of use. Each liver specimen was sonicated in Iscove's Dulbecco Medium with 2% aprotinin for ∼10 s. Specimens were centrifuged at 10,000 rpm for 10 min at 4°C and the supernatant was collected for measurement of cytokines (IL-1β, IL-6, and TNF-α) by ELISA. Cytokine measurements were corrected for protein by Bradford assay.

Measurement of Peripheral Mediators

Cytokines were measured using available ELISAs for IL-1, IL-6, and TNF-α (R&D Systems). Plasma ACTH also was measured by ELISA (ALPCO). PGE2 and corticosterone were measured by enzyme immunoassay (Cayman Chemical and Assay Designs, respectively). All ELISAs and enzyme immunoassays were performed according to manufacturer's recommendations. PGE2 samples were diluted 1:20 in the manufacturer's enzyme immunoassay buffer, and corticosterone samples were diluted 1:50 in dH2O and heated for 1 h in 70°C water bath.

Experimental Design

Study 1.

Animals (n = 6–8/group) were injected intraperitoneally with either saline or E. coli (2.5 × 107 CFU) and euthanized by decapitation 0.5, 1, 2, 4, or 24 h later. One or two saline-control animals were euthanized at each time and presented as a 0-h time point. Trunk blood was collected for measurement of cytokines, ACTH, corticosterone, and PGE2. A small specimen of liver was quickly harvested, and cytokines was measured by ELISA.

Study 2.

A baseline sample of tail vein blood was collected prior to drug and E. coli administration. Animals (n = 6/group) were injected intraperitoneally with either vehicle or indomethacin (5 mg/kg) 30 min prior to intraperitoneal E. coli administration (2.5 × 107 CFU). Tail vein blood was then collected at 1, 2, and 4 h after E. coli challenge for measurement of corticosterone. PGE2 was also measured in blood samples to verify blockade of PGE2 formation by indomethacin.

Study 3.

A baseline sample of tail vein blood was collected prior to drug and E. coli administration. Animals (6–9/group) were injected intraperitoneally with either saline or IL-6 antibody (4.2 μg/kg) 30 min prior to intraperitoneal E. coli administration (2.5 × 107 CFU). Tail vein blood was collected at 1, 2, and 4 h after E. coli challenge for measurement of corticosterone.

Study 4.

A baseline sample of tail vein blood was collected prior to drug and E. coli administration. Animals (6–9/group) were injected with either DMSO or indomethacin 30 min prior to intraperitoneal E. coli challenge (2.5 × 108 CFU). Tail vein blood was collected at 1, 2, and 4 h after E. coli challenge for measurement of corticosterone.

Study 5.

Animals (6–9/group) were injected with either DMSO or indomethacin 30 min prior to intraperitoneal saline or E. coli (2.5 × 108 CFU) administration. Animals were killed by decapitation at 0.5 and 1 h, and trunk blood was collected for measurement of cytokines, corticosterone, and PGE2 by ELISA or enzyme immunoassay. A small specimen of liver was quickly harvested for measurement of tissue cytokines.

Statistical Analysis

Data from study 1, which examined the effects of E. coli on various mediators of HPA axis activation, were analyzed using a 1 × 6 ANOVA across time (0, 0.5, 1, 2, 4, and 24 h). In studies 2, 3, and 4, tail blood was taken from each animal to examine the effects of drug administration on corticosterone release across time. A repeated-measures ANOVA was used to examine the effects of each treatment across time (0, 1, 2, and 4 h), and significant interactions were further analyzed using one-way ANOVAs at each time point. Necessary post hoc analyses were done using Fisher's least significant difference test. P < 0.05 was used in all cases as a measure of significance to reject the null hypothesis.

RESULTS

Study 1: Time Course of Mediators Involved in HPA Activation

To establish a time course of HPA activation following E. coli administration and to determine the factors that might be involved in HPA stimulation at each time point, animals were injected intraperitoneally with 2.5 × 107 CFU of E. coli or saline and killed 0.5, 1, 2, 4, and 24 h later. After E. coli challenge, plasma corticosterone levels peaked at 0.5 h and decreased in a time-dependent manner, thereafter returning to baseline levels by 4 h [F(5,44) = 14.720, P < 0.001]. Post hoc analysis revealed that corticosterone was significantly elevated at 0.5 h (P < 0.001), 1 h (P < 0.001), and 2 h (P = 0.020) following E. coli challenge compared with saline-injected controls (0 h) (Fig. 1A). Plasma ACTH also increased across time [F(5,42) = 7.655, P < 0.001] following E. coli challenge. ACTH began to increase at 0.5 h (P = 0.072) although levels did not reach significance until 1 h (P < 0.001) and returned to baseline levels by 2 h (Fig. 1B).

Fig. 1.

Sprague-Dawley rats were injected intraperitoneally (IP) with 2.5 × 107 CFU Escherichia coli and euthanized 0.5, 1, 2, 4, or 24 h later for measurement of plasma corticosterone (A), plasma ACTH (B), plasma PGE2 (C), plasma IL-6 (D), plasma TNF-α (E), and liver IL-1β (F). Baseline (BL) levels were determined from animals injected IP with sterile saline. Data points represent means ± SE (n = 6–8/group). *Significant differences P < 0.05 compared with baseline values.

Plasma PGE2 showed a similar pattern following E. coli challenge as plasma corticosterone with peak levels at 0.5 h and levels slowly decreasing across time [F(5,42) = 17.318, P < 0.001]. Post hoc analysis revealed that plasma PGE2 is rapidly elevated by 0.5 h (P < 0.001), continues to be significantly elevated at 1 h (P = 0.004), and returns to baseline levels by 2 h. Interestingly, PGE2 significantly decreased at 4 h (P = 0.015) compared with baseline levels before returning to basal levels by 24 h (Fig. 1C). Not surprisingly, the inflammatory cytokines took longer to elevate in plasma following immune challenge. Plasma IL-6 levels increased across time [F(5,45) = 53.588, P < 0.001] with levels significantly elevated at 2 h (P < 0.001) (Fig. 1D), but no longer significantly elevated at 4 h (P = 0.072). Plasma TNF-α also increased across time [F(5,43) = 11.723, P < 0.001] with TNF-α levels elevated at 1 h (P = 0.193) and significantly elevated by 2 h (P < 0.001), and returned to baseline levels by 4 h (Fig. 1E). Plasma IL-1β was measured, but there were no significant interactions across time [F(5,42) = 1.768, P = 0.141] with only low levels of IL-1β being detected (figure not shown).

Liver IL-1β showed a significant interaction across time [F(5,45) = 8.840, P < 0.001] following E. coli challenge. Post hoc analysis revealed that liver IL-1β levels did not rise until 2 h after E. coli challenge (P < 0.001). Liver IL-1β was still elevated at 4 and 24 h, but the values were not significant (P = 0.075 and P = 0.357, respectively) (Fig. 1F). Liver IL-6 and TNF-α levels were not elevated at any time point [F(5,47) = 1.189, P = 0.331] and [F(5,47) = 0.318, P = 0.899], respectively (figures not shown).

Study 2: Effects of Prostaglandins on HPA Axis Activation

Based on study 1, the time course of plasma PGE2 best matched that of plasma corticosterone as both were shown to rapidly elevate following E. coli challenge, with peak values reached within 30 min. To examine the role PGE2 may play on HPA activation and ultimately corticosterone secretion, animals were injected with either DMSO or indomethacin prior to E. coli administration, and plasma corticosterone was measured across time. A repeated-measures ANOVA revealed a significant effect of treatment across time [F(2,17) = 16.354, P < 0.001]. Administration of E. coli resulted in a rapid increase in plasma corticosterone with the greatest increase at 1 h and then levels steadily decreasing across time. Indomethacin alone had no significant effects on plasma corticosterone at any time point, but completely blocked the rise in corticosterone 1 h after E. coli challenge, but not 2 h following E. coli challenge (Fig. 2A). To verify that the effects of indomethacin did not wear off by 2 h, plasma PGE2 levels were measured at 1 and 2 h after E. coli challenge. Indomethacin completely blocked the rise in plasma PGE2 at both 1 h [F(2,17) = 11.305, P = 0.001] and 2 h [F = (2,14) = 15.525, P = 0.001] following E. coli challenge (Fig. 2B).

Fig. 2.

Sprague-Dawley rats were injected IP with either DMSO or indomethacin 30 min prior to 2.5 × 107 CFU E. coli and blood samples collected from a tail vein 1, 2, 4, and 24 h later for measurement of circulating plasma corticosterone (A). Circulating plasma PGE2 1 h and 2 h after E. coli administration is also presented (B). Baseline blood samples were collected immediately before saline or indomethacin administration. Data points represent means ± SE (n = 6/group). *Significant differences P < 0.05 compared with baseline values (or indomethacin controls in Fig. 2B). #Significant differences P < 0.05 compared with E. coli-injected animals.

Study 3: Effects of IL-6 on HPA Axis Activation

As demonstrated in study 2, plasma corticosterone was elevated at 2 h following E. coli challenge, despite continual blockade of plasma PGE2. Since plasma IL-6 was significantly elevated at 2 h and IL-6 has previously been shown to contribute to LPS-induced HPA activation, we examined the contribution of IL-6 on E. coli-induced plasma corticosterone via immunoneutralization techniques. Administration of rat IL-6 antibody alone had no significant effects on plasma corticosterone compared with baseline levels across time; however, administration of IL-6 antibody significantly attenuated plasma corticosterone levels at 2 h following peripheral E. coli. A repeated-measures ANOVA revealed a significant effect of drug treatment across time [F(2,17) = 30.023, P < 0.001] with post hoc analyses revealing that administration of IL-6 antibody only attenuated plasma corticosterone levels 2 h after E. coli challenge (P = 0.004) (Fig. 3).

Fig. 3.

Sprague-Dawley rats were injected IP with goat anti-rat polyclonal IL-6 antibodies 30 min prior to 2.5 × 107 CFU E. coli and blood samples collected from a tail vein 1, 2, and 4 h later for measurement of circulating plasma corticosterone. Baseline blood samples were collected immediately before saline or antibody (Ab) administration. Data points represent means ± SE (n = 6–9/group). *Significant differences P < 0.05 compared with baseline. #Significant differences P < 0.05 compared with E. coli-injected animals.

Study 4: Effects of Prostaglandins on HPA Activation After Higher Concentration of E. coli.

To further investigate the role of PGE2 in mediating HPA responses to higher concentrations of bacteria, we injected animals with either DMSO or indomethacin prior to a 10-fold higher concentration of E. coli (2.5 × 108 CFU) and measured plasma corticosterone across time. Administration of E. coli resulted in high levels of plasma corticosterone at 1 and 2 h with levels slightly lower at 4 h. Unlike with the lower concentration of E. coli, indomethacin only attenuated plasma corticosterone at 1 h (P = 0.044) compared with animals injected with E. coli alone as corticosterone levels were significantly elevated compared with baseline levels (Fig. 4). There were no other significant differences between the two treatments at any other time point.

Fig. 4.

Sprague-Dawley rats were injected IP with either DMSO or indomethacin 30 min prior to 2.5 × 108 CFU E. coli and blood samples collected from a tail vein 1, 2, and 4 h later for measurement of circulating plasma corticosterone (A). Baseline blood samples were collected immediately before saline or indomethacin (Indo) administration. Data points represent means ± SE (n = 6/group). *Significant differences P < 0.05 compared with baseline values. #Significant differences P < 0.05 compared with E. coli-injected animals.

Study 5: Effects of Indomethacin on Cytokines Production Following a Higher Concentration of E. coli.

Because indomethacin only attenuated the early HPA response after a high concentration of E. coli, we further investigated possible mediators that could be responsible for early activation of the HPA axis following a high concentration of E. coli. Animals were either injected with DMSO or indomethacin 30 min prior to intraperitoneal saline or E. coli administration, and animals were killed 0.5 h or 1 h later. Again, E. coli administration significantly elevated plasma corticosterone compared with DMSO-injected controls at both the 0.5 h (P = 0.001) and 1 h (P < 0.001) time points. There was a significant interaction between the treatments at both time points [F(2,15) = 8.351 P = 0.005 and F(2,14) = 22.812, P < 0.001, respectively] with post hoc analysis revealing animals injected with indomethacin prior to E. coli challenge had significantly attenuated plasma corticosterone levels at both 0.5 h (P = 0.024) and 1 h (P = 0.001) compared with E. coli-injected animals (Fig. 5A). E. coli administration also significantly elevated plasma ACTH compared with DMSO-injected controls at both the 0.5 h (P < 0.001) and 1 h (P < 0.001) time points. There was a significant interaction between the treatments at both time points [F(2,16) = 15.799, P < 0.001 and F(2,19) = 14.721, P < 0.001, respectively] with post hoc analysis revealing that indomethacin administration prior to E. coli challenge significantly attenuated plasma ACTH levels at both 0.5 h (P < 0.001) and 1 h (P < 0.001) compared with E. coli-injected animals (Fig. 5B).

Fig. 5.

Sprague-Dawley rats were injected IP with either DMSO or indomethacin 30 min prior to saline or 2.5 × 108 CFU E. coli and animals euthanized 0.5 h or 1 h later for measurement of plasma corticosterone (A), plasma ACTH (B), plasma PGE2 (C), plasma IL-1β (D), plasma IL-6 (E), plasma TNF-α (F). Data points represent means ± SE (n = 6–9/group). *Significant differences P < 0.05 compared with saline-injected animals. #Significant differences P < 0.05 compared with E. coli-injected animals.

Plasma PGE2 also showed a significant interaction between treatments after E. coli challenge at 0.5 h [F(2,17) = 7.926 P = 0.004] and 1 h [F(2,17) = 6.142, P = 0.010]. Post hoc analysis revealed animals injected with E. coli had significant increases in plasma PGE2 compared with control animals at both 0.5 h (P = 0.008) and 1 h (P = 0.011), and indomethacin completely blocked the elevations in PGE2 at both time points (Fig. 5C). Similar to that observed following the low concentration of E. coli presented above in Fig. 1, plasma TNF-α, IL-1β, and IL-6 were not significantly elevated at 0.5 h or 1 h following the high concentration of E. coli. Interestingly, a significant effect of treatment was observed for plasma IL-1β at 1 h [F(2,17) = 4.066, P = 0.035], plasma IL-6 at both 0.5 h [F(2,17) = 4.057, P = 0.039] and 1 h [F(2,17) = 10.007, P = 0.001], and plasma TNF-α at both 0.5 h [F(2,17) = 6.816, P = 0.008] and 1 h [F(2,17) = 8.537, P = 0.004]. In all cases, animals injected with indomethacin prior to E. coli challenge had significantly elevated plasma cytokines compared with control animals and animals injected with E. coli in the absence of indomethacin (Fig. 5, D–F).

Liver IL-β showed a similar pattern with a significant effect of treatment at 0.5 h [F(2,14) = 6.080, P = 0.015] and 1 h [F(2,14) = 14.814, P < 0.001]. Indomethacin treatment prior to E. coli significantly increased liver IL-1β at 0.5 and 1 h compared with controls (P = 0.005 at 0.5 h and P < 0.001 at 1 h) and animals receiving E. coli in the absence of indomethacin (P = 0.047 at 0.5 h and P = 0.011 at 1 h) (Fig. 6A). Liver IL-6 displayed no significant interactions between treatments at either time point (Fig. 6B). No elevations in liver TNF-α were reported at the 0.5-h time point, but there was a significant difference between treatments at 1 h [F(2,17) = 4.743, P = 0.022]. Post hoc analysis again revealed that only animals injected with indomethacin prior to E. coli challenge had elevations in liver TNF-α at 1 h, and these elevations were significant compared with both control animals (P = 0.015) and animals injected with E. coli alone (P = 0.017) (Fig. 6C).

Fig. 6.

Sprague-Dawley rats were injected IP with either DMSO or indomethacin 30 min prior to saline or 2.5 × 108 CFU E. coli and animals euthanized 0.5 h or 1 h later for measurement of liver IL-1β (A), liver IL-6 (B), and liver TNF-α (C). Data points represent means ± SE s (n = 6–9/group). *Significant differences P < 0.05 compared with saline-injected animals. #Significant differences P < 0.05 compared with E. coli-injected animals.

DISCUSSION

The present data demonstrate a shift across time in the factors that mediate HPA activation following bacterial challenge. HPA responses following peripheral E. coli challenge are extremely rapid, within 30 min, and occur prior to detection of circulating or liver inflammatory cytokines. Data presented here demonstrate the initial activation of the HPA axis is largely mediated by COX-induced prostanoid synthesis. Not only did the early rise in circulating corticosterone correlate with the rise in circulating PGE2, but the administration of indomethacin, a nonselective COX inhibitor, completely blocked the early rise in plasma corticosterone following a low concentration of peripheral E. coli and attenuated the early rise in plasma corticosterone following a high concentration of E. coli. Interestingly, corticosterone levels peaked long before peak circulating ACTH, suggesting prostaglandins may stimulate corticosterone release through an ACTH-independent mechanism. This would support previous work demonstrating PGE2 directly mediate corticosterone release from cultured rat adrenals (41) and cortisol release from human adrenal H295R cells (56).

Even while prostaglandin production was completely inhibited, corticosterone levels increased 2 h following E. coli challenge, indicating other factors stimulate HPA responses at this time. The later rise in corticosterone correlated with the rise of inflammatory cytokines in plasma (IL-6 and TNF-α) and in liver (IL-1β). Administration of rat IL-6 antibodies attenuated the later rise in corticosterone 2 h following E. coli challenge. It should be noted that the magnitude of the corticosterone response in the original time course study was less compared with subsequent studies for unknown reasons. One possibility is that some bacteria were lost during the decanting process for the original study, which would have reduced the number of actual E. coli injected. However, this does not change the interpretation of the individual studies, that following bacterial challenge prostaglandins largely initiate HPA responses, while inflammatory cytokines, at least IL-6, are likely mediators that maintain HPA activation at later time points.

Most studies in the literature have utilized purified LPS, a component in the cell membrane of gram-negative bacteria, to stimulate immune responses and examine HPA activation. Here, we chose to use live E. coli to more closely mimic a true bacterial challenge as whole bacteria have multiple antigenic epitopes that can stimulate a wider range of immune responses (5). In fact, the time course of HPA activation presented here following E. coli is different than that commonly reported following LPS challenge. Administration of E. coli resulted in a very rapid rise in circulating corticosterone that peaked ∼30 min after E. coli challenge and decrease thereafter, while studies that administer LPS report a slower rise in circulating corticosterone that peaks ∼2 h after challenge (regardless of dose and route of administration) (13, 23, 26, 32, 35, 60). While there appears to be a rather dramatic shift in the kinetics of HPA activation between E. coli and LPS, the mechanisms mediating HPA responses appear to be very similar. By use of genetic and pharmacological techniques, recent studies have demonstrated inhibition of COX-1 results in a significant reduction in corticosterone levels in the first hour after LPS challenge but has little effect on corticosterone thereafter (14, 15, 23). Our data demonstrate COX-mediated prostanoid synthesis is also critical for the initial induction of corticosterone following E. coli challenge, particularly following lower concentrations of bacteria. However, because of the kinetic differences between E. coli and LPS, this study reveals prostaglandins may play a larger role in HPA activation during a bacterial infection than suggested by LPS studies. In studies using LPS, COX inhibitors only block a very small fraction of the overall corticosterone response because very little corticosterone is released in the first hour following LPS challenge (14, 23), while following E. coli the bulk of the corticosterone response is observed within the first hour, especially following lower concentrations of bacteria.

At later time points, inflammatory cytokines likely help maintain HPA responses. Many studies have examined the role of inflammatory cytokines and found that genetic or pharmacological blockade of individual inflammatory cytokines have little effect on attenuating HPA responses to LPS (12, 13, 31, 37, 46, 50, 58), likely due to duplicity in inflammatory cytokine signaling. One exception has been that the pretreatment with IL-6 antibodies often attenuates ACTH (46) and corticosterone (60) responses 2–4 h following LPS but not at early time points (18, 60). Here, we replicate this finding following E. coli challenge by demonstrating administration of goat anti-rat polyclonal IL-6 antibodies attenuate corticosterone levels 2 h following bacterial challenge. Together, these findings suggest prostaglandins mediate early HPA activation and IL-6 (along with potentially other inflammatory cytokines) help maintain HPA activation as needed.

Indomethacin failed to completely block the early rise in corticosterone following higher concentrations of E. coli, suggesting additional mediators may be involved in HPA activation as bacterial challenge increases. We examined whether a higher concentration of E. coli would stimulate a more rapid increase in inflammatory cytokines; but this was not the case as there was no significant increase in any circulating or liver inflammatory cytokines at either 30 or 60 min following the higher dose of E. coli. However, indomethacin administration resulted in exaggerated inflammatory cytokine responses to E. coli challenge as demonstrated by significant increases in liver IL-1β at 0.5 and 1 h, plasma IL-6 at 0.5 and 1 h, and increases in plasma TNF-α at 0.5 and 1 h and liver TNF-α at 1 h. This suggests inflammatory cytokines may elevate in an active compensatory fashion to stimulate corticosterone production, thereby, explaining why indomethacin only attenuates corticosterone responses at 0.5 and 1 h following a high concentration of E. coli. However, the cytokine levels at 0.5 h are very low, and it is unclear whether such values would be sufficient to activate the HPA axis. If these low cytokine levels are sufficient to stimulate corticosterone production, then one would have expected a further increase in corticosterone production as cytokine levels increased from 0.5 to 1 h, but this was not observed. One would also have expected a similar enhancement of inflammatory cytokines and attenuation of corticosterone responses in indomethacin-treated animals challenged with the lower concentration of E. coli, yet indomethacin completely blocked the elevation in corticosterone when administered following the lower concentration of E. coli. Our interpretation of these data is that additional factors contribute to HPA activation at higher bacterial concentrations. One possibility is activation of complement. Complement proteins are constitutively produced and continuously circulate in the blood, and following contact with the surface of bacteria complement proteins can be rapidly activated, thus available to rapidly activate HPA responses. Complement products C3a and C5a have been shown to elevate ACTH (21, 22) and activate paraventricular hypothalamic neurons (8) as well as act directly on the adrenal gland (21). Complement products can also trigger prostaglandin synthesis and release by Kupffer cells (45, 47) adding to HPA activation.

The fact that indomethacin pretreatment resulted in exaggerate inflammatory cytokine responses following E. coli challenge indicates that prostaglandin signaling inhibits inflammatory cytokine production. Prostaglandins can directly inhibit inflammatory cytokines, particularly TNF-α, as demonstrated in cultured Kupffer cells (19, 36, 54), peritoneal (1, 63), alveolar (48), and bone-derived macrophage (30, 59). Additionally, the dampened HPA response in indomethacin-treated animals likely reduced glucocorticoid inhibition of inflammatory cytokine production resulting in widespread increases in inflammatory cytokine production. These data are of particular interest because nonsteroidal anti-inflammatory drugs are commonly used for the relief of pain and inflammation, yet chronic use is associated with an increased risk of atherosclerosis and thrombosis (6, 25, 28). It is possible that COX inhibitors create a prothrombotic state by elevating proinflammatory cytokines responses, which upregulate adhesion molecules on endothelial cells promoting leukocyte adhesion and aggregation to vessel walls and increase the production of tissue factor while simultaneously preventing the formation of PGI2 that typically inhibit proinflammatory cytokine production (44, 64) as well as platelet activation (7) and aggregation (4, 9, 29, 33, 42).

Perspectives and Significance

Our studies demonstrate for the first time a complete time course of the possible mediators involved in HPA activation following peripheral E. coli challenge. The data indicate that early PGE2 synthesis is not only responsible for early corticosterone responses following an immune challenge but is also responsible for controlling inflammatory cytokine responses that ultimately themselves fine tune the HPA response. It should not be surprising that extremely rapid mechanisms, even ACTH-independent mechanisms, have evolved to stimulate glucocorticoid production during bacterial infection given their importance in reducing the risk of septic shock and modulating immune responses to increase the chances of survival during infection.

GRANTS

This work was supported by National Institute of Mental Health Grant R21MH-77004 and the Ohio Board of Regents Research Initiative.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: Z.R.Z. and J.D.J. conception and design of research; Z.R.Z., V.M.P., R.M.C., and J.D.J. performed experiments; Z.R.Z., V.M.P., and J.D.J. analyzed data; Z.R.Z., V.M.P., and J.D.J. interpreted results of experiments; Z.R.Z. prepared figures; Z.R.Z. drafted manuscript; J.D.J. edited and revised manuscript; J.D.J. approved final version of manuscript.

REFERENCES

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
View Abstract