Abstract
Certain cytokines activate the hypothalamic-pituitary-adrenal axis for glucocorticoid release, and these hormones can protect against cytokine-mediated pathologies. However, endogenous activation of such a pathway has not been established during infections. A prominent glucocorticoid response peaks 36 h following murine CMV (MCMV) infection, coincident with circulating levels of the cytokines IL-12, IFN-γ, TNF, and IL-6, and dependent on IL-6 for maximal release. These studies examined functions of the hormone induction. Mice rendered glucocorticoid deficient by adrenalectomy were more susceptible than intact mice to MCMV-induced lethality, and the increased sensitivity was reversed by hormone replacement. Lack of endogenous glucocorticoids resulted in increases in IL-12, IFN-γ, TNF, and IL-6 production, as well as in mRNA expression for a wider range of cytokines, also including IL-1α and IL-1β. Viral burdens did not increase, and actually decreased, in the livers of glucocorticoid-deficient mice. TNF, but not IFN-γ, was required for increased lethality in the absence of endogenous hormone. These results conclusively demonstrate the importance of induced endogenous glucocorticoids in protection against life-threatening effects resulting from infection-elicited cytokine responses. Taken together with the dependence on induced IL-6, they document existence of an immune system-hypothalamic-pituitary-adrenal axis pathway for regulating endogenous responses to viral infections.
The neuroendocrine and immune systems are interconnected with pathways available for reciprocal regulation (reviewed in Refs. 1 and 2). Following exposure to proinflammatory cytokines, the hypothalamic-pituitary-adrenal (HPA)3 axis is stimulated to release glucocorticoids from adrenal glands. These hormones can, in turn, down-regulate immune responses. Such interactions have been postulated to act as a regulatory feedback loop protecting the host against damaging immune responses 1, 2 . Studies to date with administered cytokines, glucocorticoids, or isolated microbial products such as bacterial LPS, i.e., endotoxin, indicate that cytokines, including IL-12, IFN-γ, TNF, and IL-1, can cause significant pathology with mortality if present at high levels and/or induced in the absence of glucocorticoid-mediated regulation and that glucocorticoids can protect against this pathology by inhibiting cytokine expression (Refs. 3–10; reviewed in 11 . As these cytokines also contribute to defense against microbial pathogens, their expression must be tightly regulated to limit detrimental effects as well as to access protective effects. However, endogenous immune response activation of the HPA axis as a pathway for regulating the detrimental cytokine responses resulting in disease and death remain to be proven during infectious challenge.
Previous studies in our laboratories have characterized a prominent glucocorticoid response during murine CMV (MCMV) infection with a sharp peak of corticosterone release occurring 36 h following infection and reaching levels of 200 ng/ml, or 30-fold above basal levels at the diurnal nadir 12 . Coincident with this response, high circulating levels of multiple proinflammatory cytokines are produced, including IFN-γ, TNF-α, IL-6, the p40 chain of IL-12, and biologically active IL-12 p70 heterodimer, but IL-1β induction is not detectable, and IL-1α is only modestly elevated 12, 13 . Although a range of cytokines is induced, IL-6 is the principal cytokine responsible for the glucocorticoid induction 12 . As many of the observed cytokines have potential both to mediate antiviral functions and to contribute to disease parameters, including mortality 14, 15, 16 , we investigated the effects of endogenous glucocorticoids on virus-induced cytokine expression and function. Our results conclusively demonstrate that glucocorticoids are essential for protection against the lethal effects of TNF during MCMV infection and that they regulate virus-elicited cytokines at both the protein and mRNA levels. These studies prove that endogenous glucocorticoid responses play an important role in protecting against immunity-associated disease during infection.
Materials and Methods
Mice
Male 5- to 10-wk-old C57BL/6 (C57BL/6NTacfBR, Taconic Laboratory Animals and Services, Germantown, NY) or immunocompetent and IFN-γ-deficient C57BL/6J (The Jackson Laboratory, Bar Harbor, ME) mice were kept on a 12-h (6:00 a.m. to 6:00 p.m.) light/dark cycle with lights on at 6:00 a.m. Mice were housed in the Brown University Animal Care Facility for at least 1 wk before experimental manipulation and used in accordance with institutional guidelines for animal care and use.
Surgical procedures and in vivo treatments
Adrenalectomy (surgical removal of adrenal glands; ADX) and sham (SHAM) operations were performed as described 17 . For ADX mice, 0.9% saline drinking water was supplemented with 50 μg/ml corticosterone for 3 days following surgery. Except where indicated, surgical recovery lasted a total of 5 days. Corticosterone administration to infected, ADX mice was performed by adding 30 or 300 μg/ml (Sigma, St. Louis, MO) to drinking water at infection initiation. Rat anti-mouse IFN-γ (ammonium sulfate-precipitated ascites of the XMG1.2 hybridoma (anti-IFN-γ)), chimeric hamster/mouse anti-TNF mAb with Fab of the hamster TN3.19–12 and a mouse IgG1 Fc region (Celltech, Plough, U.K.; a generous gift from Steven M. Opal, Department of Medicine, Brown University, Providence, RI), or control IgG, rat or mouse (Sigma), was injected i.p. 12 h before infection at 1 mg/mouse (anti IFN-γ) or 500 μg/mouse (anti-TNF). We have shown these protocols to be up to 99% effective at serum cytokine neutralization 12, 15 .
Virus
Stocks of Smith strain MCMV were generated in salivary glands as described 13 . Mice were i.p. infected with 100 μl of indicated virus dose or vehicle alone (1× medium 199 (Life Technologies, Grand Island, NY) supplemented with 3% heat-inactivated FBS (HyClone Laboratories, Logan, UT)) between 7:00 and 8:00 p.m. Titers in infected organs were assessed as plaque-forming units (PFU)/g of tissue using viral plaque assays as described 12 .
Serum and organ collection
Mice were retroorbitally bled under methoxyflurane anesthesia (Metofane, Pitman-Moore, Washington Crossing, NJ) into heparinized tubes (5 μl of 10,000 U/ml). Some clotting occurred. Samples were centrifuged, and supernatant fluids were collected and stored at −80°C. Lateral lobes of livers were placed into 1× DMEM (Life Technologies) supplemented with 10% FBS (HyClone Laboratories), 2 mM glutamine, 100 U of penicillin, and 100 mg of streptomycin (Life Technologies) (complete DMEM) and stored at −80°C for use in viral plaque assays. Spleens were placed into complete DMEM and stored at −80°C for plaque assays, used to isolate leukocytes for generation of conditioned media as described 13 , or immediately frozen in liquid nitrogen for later RNA isolation. Body weights were determined on a digital scale (Ohaus, Florham Park, NJ) following injection and again at the time of sacrifice.
Cytokine and hormone measurements
Cytokine protein levels were determined by sandwich ELISAs 12, 13, 14 . IL-12 p70 was measured using Ab capture with a biological assay for IFN-γ induction (product measured by ELISA), as described previously 13 . Limits of detection in serum for IL-1β, IL-12 p40, IL-12 p70, IFN-γ, TNF-α, and IL-6 were 15, 40, 4, 80, 40, and 80 pg/ml, respectively. (Limits of detection in conditioned media were 2- to 4-fold lower.) Simultaneous detection of multiple cytokine mRNAs was performed using RNA extracted from frozen spleens with TRIzol (Life Technologies) and Riboquant probe sets from PharMingen (San Diego, CA). The mCK-3b set was used to detect TNF-β, lymphotoxin (LT)-β, TNF-α, IL-6, and IFN-γ. The mCK-2b set was used to detect IL-12 p35, IL-12 p40, IL-10, IL-1α, and IL-1β. Briefly, 32P-labeled cRNA probes were hybridized overnight with 10 μg of total sample RNA at 56°C, digested with RNase A/T1 mixtures, extracted with phenol/chloroform, ethanol precipitated, and separated on 4% polyacrylamide/7 M urea gels. Probes were visualized by exposure to autoradiographic film. Probe sets included cRNA for L-32 (a ribosomal protein) to control for loading, hybridization, and extraction in final analyses, and assays included a tRNA sample to document complete digestion of unhybridized probes. Identity of protected bands was confirmed by plotting migration relative to undigested probe standards and by comparison with a 32P-labeled 100-bp RNA ladder (Ambion, Austin, TX). Densitometric analyses of autoradiograms were performed on scanned images using National Institutes of Health image software and a gel-plotting algorithm integrating band width and density. Commercially available assay kits, with limits of detection at 5 ng/ml, were used for corticosterone determinations (Immunochem, ICN Biomedicals, Costa Mesa, CA).
Statistical analysis
Unless otherwise indicated, values are presented as means ± SE. For cytokine and viral titer analyses, p values were obtained using Microsoft Excel 4.0 by comparing treatment groups using a two-tailed Student’s t test. For survival curve analyses, p values were obtained with Statview SE using Mantel-Cox rank log or Breslow-Gehan-Wilcoxon tests.
Results
Role of endogenous glucocorticoids in survival
To evaluate the effects of endogenous glucocorticoids during viral infections, ADX or SHAM mice were uninfected or infected with MCMV. All endocrine-intact SHAM mice infected with up to 2 × 105 PFU MCMV or uninfected ADX mice had long term survivals. However, mortality of up to 80% occurred between 36 and 72 h after infection of ADX mice with virus doses of 2 × 105 or 1 × 105 PFU (Fig. 1,A). Thus, absence of adrenal glands resulted in increased susceptibility to MCMV-induced lethality. To determine whether endogenous glucocorticoids, as opposed to other adrenal hormones, were responsible for protection against mortality, MCMV-infected ADX mice were given low or high concentrations of corticosterone (30 or 300 μg/ml, respectively) in drinking water. Serum levels of corticosterone ± SE during normal diurnal cycles were 5 ± 3 and 134 ± 7 ng/ml at morning and evening times, respectively, and during MCMV infection they reached 203 ± 18 ng/ml. Administration of corticosterone to ADX mice resulted in levels of serum corticosterone reaching 2389 ± 902 and 54 ± 19 ng/ml for the 300 and 30 μg/ml doses, respectively, taken during morning time points at the end of experiments. The high dose completely protected, and the low dose significantly protected, against MCMV-induced death (Fig. 1 B). Hence, glucocorticoid replacement restored resistance. These results demonstrate that endogenous glucocorticoids mediate protection against infection-induced death.
Contribution of endogenous glucocorticoids to other pathologies and effects on viral burden
As MCMV is a cytopathic virus, it was possible that its lethal consequences were related to other virus-induced pathologies, or that viral loads were increased, in the absence of glucocorticoids. Because body weight loss is a general indication of sickness, this parameter was examined. In normal endocrine-intact mice, MCMV infections resulted in body weight loss. Over the time of experiments following infections, respective percentage total body weight changes ± SE in uninfected compared with infected mice were 2.7 ± 0.5 and −0.9 ± 1.0 at 36 h, and 3.4 ± 1.2 and −5.9 ± 1.1 (p < 0.01) at 60 h. These variations were not significantly altered in the absence of glucocorticoids; percentage body weight changes in SHAM and ADX mice at 60 h following infection were, respectively, −3.0 ± 2.9 and −2.8 ± 3.6. Thus, endogenous glucocorticoids did not protect against or contribute to weight loss.
To determine whether increased mortality was associated with increased viral replication, splenic and hepatic viral titers were quantitated. There were no differences in splenic viral burdens between ADX and SHAM mice at any time they were examined following MCMV infection (Table I). Hepatic MCMV titers also were similar between SHAM and ADX mice at early times; however, significant reductions were observed at 60 h following infection of ADX mice (Table I). These results dissociated viral burden and lethality. Taken together, the studies demonstrate that the MCMV-induced mortality in ADX mice is independent of viral loads and certain other virus-induced pathologies.
Hours of Infection . | . | MCMV Titers (log PFU/g tissue) . | . | |
---|---|---|---|---|
. | . | Splenic . | Hepatic . | |
36 | SHAM | 2.75 ± 1.03 | 4.85 ± 0.19 | |
ADX | 3.64 ± 0.47 | 4.74 ± 0.05 | ||
48 | SHAM | 1.50 ± 0.86 | 4.56 ± 0.16 | |
ADX | 1.17 ± 0.70 | 4.16 ± 0.91 | ||
60a | SHAM | 2.73 ± 0.33 | 4.78 ± 0.11 | |
ADX | 1.66 ± 2.34 | 3.85 ± 0.02b |
Hours of Infection . | . | MCMV Titers (log PFU/g tissue) . | . | |
---|---|---|---|---|
. | . | Splenic . | Hepatic . | |
36 | SHAM | 2.75 ± 1.03 | 4.85 ± 0.19 | |
ADX | 3.64 ± 0.47 | 4.74 ± 0.05 | ||
48 | SHAM | 1.50 ± 0.86 | 4.56 ± 0.16 | |
ADX | 1.17 ± 0.70 | 4.16 ± 0.91 | ||
60a | SHAM | 2.73 ± 0.33 | 4.78 ± 0.11 | |
ADX | 1.66 ± 2.34 | 3.85 ± 0.02b |
Infections were initiated 7 days instead of 5 days following surgeries.
p < 0.01.
Cytokine responses during infections of intact and ADX mice
As many of the cytokines induced early during MCMV infection have antiviral activities, the reductions in viral titers may have been due to increased cytokine production in ADX mice. Reciprocally, the lethal consequences of MCMV infection after ADX could have been caused by elevated cytokine responses. To extend our previous studies of cytokine responses 12 to examination of expression at the level of mRNA and to evaluate additional cytokines, RNase protection assays were conducted with probes for the two chains of IL-12, p35 and p40, IL-10, IL-1α, IL-1β, TNF-β, LT-β, TNF-α, IL-6, and IFN-γ. For these studies, RNA was prepared from the spleens of endocrine-intact mice that had been uninfected or infected with MCMV for 36 h. Consistent with our earlier studies of circulating protein levels, mRNA for the cytokines IL-12 p40, TNF-α, IL-6, and IFN-γ were induced by 3.5- to >10-fold following infection (Fig. 2,A). Interestingly, even though there was little to no detectable IL-1α and IL-1β protein induction in serum, mRNA levels for these cytokines were elevated in response to MCMV challenge (Fig. 2 A). Effects on cytokines were selective because mRNA levels for IL-10, TNF-β, and LT-β were not detectably or only modestly increased. Thus, this viral infection induces early mRNA expression of certain but not all cytokines.
To examine effects of endogenous glucocorticoids on cytokine expression, cytokine serum protein, splenic leukocyte conditioned media protein, and splenic mRNA levels were determined after infections of ADX and SHAM mice. Consistent with our earlier studies using 5 × 104 PFU MCMV for infections 12 , IL-1β protein was undetectable in serum samples isolated from infected SHAM mice, but detectable low levels were revealed in samples from infected ADX mice, although not uniformly (data not shown). Serum IL-12 p40, IFN-γ, TNF-α, and IL-6 protein levels all were detectable in infected SHAM-operated mice and reproducibly induced to higher levels in infected ADX mice with statistically significant increases (p < 0.05 or p < 0.01) for both TNF and IL-6 at 36 h, for all but IL-12 p40 at 48 h and for all of these cytokines at 60 h after infection (Table II). Levels of biologically active IL-12 p70 heterodimer were not detectably different in samples from ADX and SHAM mice. Examination of splenic mRNA levels did not reveal any statistically significant increases in cytokine expression by ADX as compared with SHAM mice infected with this viral dose, i.e., 5 × 104 PFU (Table III). However, cytokine production levels in media conditioned by splenic leukocytes isolated from infected ADX as compared with SHAM mice were increased in magnitudes similar to those observed in serum (data not shown). Thus, under these conditions of infection, endogenous glucocorticoids appear to regulate a variety of cytokines at the level of protein expression.
Hours of Infection . | . | Cytokines (pg/ml ± SE) . | . | . | . | . | Mortality (live/total) . | ||||
---|---|---|---|---|---|---|---|---|---|---|---|
. | . | IL-12 p40 . | IL-12 p70 . | IFN-γ . | TNF . | IL-6 . | . | ||||
36b | SHAM | 168 ± 44 | 38 ± 10 | 4,919 ± 784 | 112 ± 15 | 4,694 ± 756 | 16/16 | ||||
ADX | 324 ± 73 | 30 ± 3 | 14,295 ± 4,997 | 277 ± 55* | 34,002 ± 12,168* | 21/21 | |||||
48b | SHAM | BLD | BLD | 1,279 ± 50 | 68 ± 13 | 476 ± 87 | 13/13 | ||||
ADX | 42 ± 27 | BLD | 3,090 ± 522*** | 165 ± 25** | 2,182 ± 522*** | 10/15 | |||||
60c | SHAM | 44 ± 20 | NT | 149 ± 3 | 49 ± 8 | 310 ± 75 | 4/4 | ||||
ADX | 165 ± 8** | NT | 360 ± 34** | 75 ± 2* | 733 ± 41** | 3/4 |
Hours of Infection . | . | Cytokines (pg/ml ± SE) . | . | . | . | . | Mortality (live/total) . | ||||
---|---|---|---|---|---|---|---|---|---|---|---|
. | . | IL-12 p40 . | IL-12 p70 . | IFN-γ . | TNF . | IL-6 . | . | ||||
36b | SHAM | 168 ± 44 | 38 ± 10 | 4,919 ± 784 | 112 ± 15 | 4,694 ± 756 | 16/16 | ||||
ADX | 324 ± 73 | 30 ± 3 | 14,295 ± 4,997 | 277 ± 55* | 34,002 ± 12,168* | 21/21 | |||||
48b | SHAM | BLD | BLD | 1,279 ± 50 | 68 ± 13 | 476 ± 87 | 13/13 | ||||
ADX | 42 ± 27 | BLD | 3,090 ± 522*** | 165 ± 25** | 2,182 ± 522*** | 10/15 | |||||
60c | SHAM | 44 ± 20 | NT | 149 ± 3 | 49 ± 8 | 310 ± 75 | 4/4 | ||||
ADX | 165 ± 8** | NT | 360 ± 34** | 75 ± 2* | 733 ± 41** | 3/4 |
BLD, below limit of detection; NT, not tested.
Values presented are data combined from multiple experiments.
Infections for this experiment were initiated 7 days instead of 5 days following surgeries.
*p < 0.05, **p < 0.01, ***p < 0.001 compared to SHAM-operated.
Cytokine mRNA . | 5 × 104 PFU . | . | 1 × 104 PFU . | . | ||
---|---|---|---|---|---|---|
. | SHAM . | ADX . | SHAM . | ADX . | ||
IL-12 p35 | 7 ± 7 | 3 ± 1 | 15 ± 4 | 34 ± 6* | ||
IL-12 p40 | 107 ± 52 | 143 ± 25 | 6 ± 5 | 98 ± 16*** | ||
IL-10 | 19 ± 18 | 9 ± 2 | 3 ± 3 | 4 ± 1 | ||
IL-1α | 265 ± 109 | 278 ± 55 | 11 ± 4 | 65 ± 11** | ||
IL-1β | 543 ± 129 | 590 ± 91 | 45 ± 9 | 182 ± 38** | ||
TNF-β | 19 ± 5 | 21 ± 10 | 22 ± 11 | 54 ± 20 | ||
LT-β | 201 ± 37 | 195 ± 55 | 257 ± 57 | 261 ± 40 | ||
TNF-α | 184 ± 21 | 208 ± 52 | 68 ± 20 | 147 ± 21* | ||
IL-6 | 83 ± 13 | 210 ± 81 | BLD | 18 ± 6 | ||
IFN-γ | 485 ± 40 | 1,080 ± 234 | 75 ± 11 | 271 ± 61* |
Cytokine mRNA . | 5 × 104 PFU . | . | 1 × 104 PFU . | . | ||
---|---|---|---|---|---|---|
. | SHAM . | ADX . | SHAM . | ADX . | ||
IL-12 p35 | 7 ± 7 | 3 ± 1 | 15 ± 4 | 34 ± 6* | ||
IL-12 p40 | 107 ± 52 | 143 ± 25 | 6 ± 5 | 98 ± 16*** | ||
IL-10 | 19 ± 18 | 9 ± 2 | 3 ± 3 | 4 ± 1 | ||
IL-1α | 265 ± 109 | 278 ± 55 | 11 ± 4 | 65 ± 11** | ||
IL-1β | 543 ± 129 | 590 ± 91 | 45 ± 9 | 182 ± 38** | ||
TNF-β | 19 ± 5 | 21 ± 10 | 22 ± 11 | 54 ± 20 | ||
LT-β | 201 ± 37 | 195 ± 55 | 257 ± 57 | 261 ± 40 | ||
TNF-α | 184 ± 21 | 208 ± 52 | 68 ± 20 | 147 ± 21* | ||
IL-6 | 83 ± 13 | 210 ± 81 | BLD | 18 ± 6 | ||
IFN-γ | 485 ± 40 | 1,080 ± 234 | 75 ± 11 | 271 ± 61* |
MCMV infections were for 36 h with three to five mice per group. Results are expressed in arbitrary units ± SE normalized to L-32 mRNA in each sample. BLD, below limit of detection (statistical analysis not performed).
* p < 0.05, **p < 0.01, ***p < 0.001 compared to SHAM.
Because a proportion of the ADX mice were succumbing to MCMV infections with the 5 × 104 PFU dose (see Table II), cytokine values at the 48- and 60-h time points represented expression in only the surviving animals. Moreover, as high cytokine mRNA levels were induced in response to 5 × 104 PFU of MCMV (Fig. 2,A, Table III), these values may have been at maximal levels and limited detection of increases in ADX mice. Therefore, experiments were conducted with a lower dose of virus, i.e., 1 × 104 PFU. Under these conditions, only IFN-γ and IL-6 were induced to detectable levels in the serum of infected mice, but ADX resulted in increases in both of these; values (pg/ml ± SE) in SHAM and ADX mice, respectively, were 127 ± 81 and 546 ± 116 (p < 0.01) at 36 h and 1156 ± 182 and 2697 ± 451 (p < 0.01) at 48 h for IFN-γ and were 542 ± 89 and 680 ± 113 at 36 h and 215 ± 63 and 298 ± 240 at 48 h for IL-6. Quantitation of splenic mRNA expression under the conditions of 36-h lower dose infections revealed elevations in ADX as compared with SHAM-operated mice significant for IL-12 p35, IL-12 p40, IL-1α, IL-1β, TNF-α, and IFN-γ and detectable for IL-6 (Fig. 2,B and Table III). Thus, during this infection, absence of endogenous glucocorticoids is associated with increased expression and extended production of multiple cytokines at both mRNA and protein levels.
Effects of corticosterone administration on cytokine expression in ADX mice
To further investigate the role of glucocorticoids in regulating MCMV-induced cytokine responses, ADX mice were given corticosterone (30 μg/ml) in their drinking water and serum cytokines were examined 36 h following MCMV infection. Consistent with the modest enhancement of IL-12 p40 expression in infected ADX mice, levels of this cytokine were relatively insensitive to the effects of exogenously added corticosterone (Fig. 3,A). However, the elevated IFN-γ (Fig. 3,B), TNF (Fig. 3,C), and IL-6 (Fig. 3,D) levels following MCMV infection of vehicle-treated ADX mice, were dramatically reduced by corticosterone replacement (Fig. 3). These values were comparable with those in vehicle- or corticosterone-treated, SHAM-operated mice. Taken together, these data demonstrate potent glucocorticoid regulation of IFN-γ, TNF, and IL-6 protein production early during MCMV infection.
Effects of cytokine neutralization
Because survival in ADX mice infected with MCMV began declining immediately following peak induction of proinflammatory cytokines, and because IFN-γ and TNF are known to contribute to mortality in other systems, the contributions of these two cytokines to virus-induced lethality were examined. ADX mice treated with Ab-neutralizing IFN-γ (Fig. 4,A) or rendered completely deficient in IFN-γ by genetic mutation (Fig. 4,B) were not protected. Thus, there was absolutely no requirement for IFN-γ in mortality of ADX mice. In contrast, Ab-mediated neutralization of TNF completely protected a majority of the MCMV-infected ADX mice (Fig. 4 C). The increased survival in anti-TNF-treated mice was statistically significant and was reproduced in independent experiments. These results show that TNF is a critical cytokine for lethality during this infection of glucocorticoid-deficient mice.
Discussion
These studies demonstrate the roles of endogenous glucocorticoids in regulating early cytokine responses and disease during viral infections. The hormones were shown to protect against MCMV-induced lethality, because mice rendered glucocorticoid deficient by removal of adrenal glands (ADX) had increased sensitivity and because resistance was restored by corticosterone replacement. The glucocorticoid effects on sensitivity to virus-induced lethality did not correlate with changes in body weight or viral burdens, but rather with elevations in both mRNA and protein expression levels of the cytokines IL-12 p40, IFN-γ, TNF-α, and IL-6. The increased sensitivity to virus-induced mortality in the absence of glucocorticoids was IFN-γ independent but TNF dependent. Collectively, the results conclusively demonstrate the importance of endogenous glucocorticoids in mediating defense against disease resulting from a viral infection and define regulation of cytokine expression and/or responsiveness as a mechanism for protection.
There has been speculation about conditions for existence and importance of an immune-HPA pathway for endogenous induction of glucocorticoids to regulate cytokine responses 1, 2 . However, there is no direct evidence to support its activation and function during infections. Studies blocking glucocorticoid effects show increased susceptibility of stressed hosts, including infected hosts, to mortality and disease 1, 2, 11, 18, 19 , but none of these identifies an infection/immunity-induced endogenous glucocorticoid response as a mechanism for protection against elicited cytokines. Others establish the role of endogenous glucocorticoids in survival, and the role of TNF in death, after LPS challenge 7, 8 , but do not address conditions of infections. Others demonstrate the lethal effects of TNF induced during bacterial infections, but not the role of endogenous glucocorticoids 20 . Taken together with our earlier study demonstrating the infection-induced IL-6 dependence of the glucocorticoid response, the experiments presented here provide evidence establishing existence and importance of an immune-HPA axis pathway for regulating endogenous responses to viral infections.
Interestingly, although TNF levels were significantly elevated in ADX mice during MCMV infection, increases in serum and spleen were surprisingly subtle. It is possible that glucocorticoid deficiencies result in additional effects promoting TNF-mediated sensitivity or in higher TNF levels localized within certain vital organs. ADX has been demonstrated to increase sensitivity to exogenously administered TNF 7 , and there are several pathways by which glucocorticoids might modulate TNF responsiveness. The effects could be a result of regulating other cytokines and/or cytokine receptors. For example, as IL-12 has been shown to increase expression of TNF receptors 21, 22 , glucocorticoid-mediated inhibition of IL-12 expression could result in decreased TNF receptor expression. Alternatively or additionally, glucocorticoids may differentially regulate cytokine expression in particular organs. Examinations of different organs from ADX mice challenged with LPS have shown that TNF induction is significantly elevated in certain nervous system tissue, including the pituitary, hypothalamus, and hippocampus, but not in the spleen 9 . Therefore, increases in local cytokine production in different tissues may have greater contributions to lethality but not be reflected proportionally in the circulation.
Our data showing the IFN-γ independence of MCMV-induced lethality are conclusive, because effects of both neutralizing Ab treatments and genetic mutations resulting in complete IFN-γ deficiency were examined. The lack of a role of IFN-γ in MCMV-induced mortality is intrinsically different from the IFN-γ-modulated toxicity induced by LPS 4, 5 . As LPS is not a replicating organism and is rapidly cleared, MCMV replication may substitute for an IFN-γ requirement. Alternatively, this cytokine, through its antiviral functions, may mediate both beneficial and detrimental activities during infection. As a result, there could be no net observable effect in the absence of IFN-γ under infection conditions, whereas a net negative effect can be demonstrated in the context of a nonreplicating LPS challenge.
Although TNF contributes to mortality during infections in the absence of glucocorticoids, the cytokine also is important for mediating antiviral functions 15, 16 . In contrast to IFN-γ, the role of TNF in lethality may be demonstrable under the conditions of glucocorticoid deficiencies because detriments outweigh protective effects or because other enhanced immune responses compensate for lack of TNF-mediated protective effects. The first possibility is supported by the demonstrated increased mortality upon exogenous TNF administration during MCMV infections 16, 23 . However, the second possibility is supported by the results presented here showing that additional antiviral cytokines, including IFN-γ, are elevated to higher levels in ADX mice. Collectively, the results demonstrate the delicate balance of protective and detrimental cytokine effects.
The effects of alternatively stimulated glucocorticoids on host immune responses have been examined after influenza virus infection of mice. Glucocorticoid responses elicited by restraint stress have been found to decrease T cell-derived cytokines 24 and mononuclear cell accumulation in lungs and draining lymph nodes of infected mice 25 . As cytokine production, cellular infiltration, and lethality all appear at the later times associated with adaptive immune responses 19, 24, 25 , glucocorticoid-mediated protection during influenza virus infection may be through prevention of T cell-mediated disease rather than proinflammatory cytokine-mediated lethality. Thus, the contrasts between MCMV and influenza infection studies indicate that different mechanisms of endogenous glucocorticoid protection may operate during particular viral infections.
IL-1α and IL-1β are proinflammatory cytokines that are induced in response to LPS. Although IL-1α is present at early times during MCMV infection, we have previously been unable to detect circulating IL-1β protein 12 . The studies reported here show IL-1β and IL-1α mRNA induction at 36 h of MCMV infection. There are many examples of dissociation between mRNA and protein expression. However, the lack of detectable IL-1β protein in the presence of mRNA may be due to the multiple additional levels of control for protein release and availability. The pro-IL-1β molecule requires cleavage by enzymes such as the IL-1β-converting enzyme 26 , and these also can be differentially regulated in particular tissues and in response to different stimuli 27 . Moreover, as the type II IL-1 receptor can act as a “decoy” molecule for binding and sequestering IL-1β protein, cytokine availability also could be modified by regulation of this receptor 28 . Given the multiple levels for regulating IL-1β, there is likely to be minimal functional IL-1β protein detectable even in the presence of high mRNA expression.
In conclusion, these studies have definitively established a protective role of endogenous glucocorticoid responses during a viral infection and identified the mechanism for mortality resulting in the absence of these responses. Thus, they show that glucocorticoids are key factors in regulating the delicate balance between protective and detrimental consequences resulting from immune responses to infections. By dissecting the individual components of the regulatory interactions between endocrine and immune systems, they further our understanding of pathways for defense against both infection and infection-induced cytokine disease.
Acknowledgements
We thank Leslie Cousens, Khuong (Ken) Nguyen, Gary Pien, and Michael Primiano for help during surgeries and harvesting low stress samples, as well as Cecilia Po, Tracy Pisell, and Cindy Su for technical assistance with the RNase protection and corticosterone assays. We also thank Drs. S. Opal, R. Coffman, and P. Scott for reagents.
Footnotes
This work was supported by National Institutes of Health Grants RO1-MH47674, KO2-MH00680, T32-ES07272, and RO1-CA41268.
Abbreviations used in this paper: HPA, hypothalamic-pituitary-adrenal; M, murine; ADX, adrenalectomy; PFU, plaque forming unit; LT, lymphotoxin.