Endogenous glucocorticoids protect against TNF-alpha-induced increases in anxiety-like behavior in virally infected mice

Abstract

Endogenous glucocorticoids restrain proinflammatory cytokine responses to immune challenges such as viral infection. In addition, proinflammatory cytokines induce behavioral alterations including changes in locomotor/exploratory activity. Accordingly, we examined proinflammatory cytokines and open-field behavior in virally infected mice rendered glucocorticoid deficient by adrenalectomy (ADX). Mice were infected with murine cytomegalovirus (MCMV), and open-field behavior (36 h post-infection) and plasma concentrations of tumor necrosis factor (TNF)-alpha and interleukin (IL)-6 (42 h post-infection) were assessed. Compared to sham-ADX-MCMV-infected animals, ADX-MCMV-infected mice exhibited significant reductions in total distance moved, number of center entries, and time spent in center. These behavioral alterations were accompanied by significantly higher plasma concentrations of TNF-alpha and IL-6, both of which were correlated with degree of behavioral change. To examine the role of TNF-alpha in these behavioral alterations, open-field behavior was compared in wild-type (WT) and TNF-R1-knockout (KO), ADX-MCMV-infected mice. TNF-R1-KO mice exhibited significantly attenuated decreases in number of rearings, number of center entries and time spent in center, but not distance moved, which correlated with plasma IL-6. Given the potential role of brain cytokines in these findings, mRNA expression of TNF-alpha, IL-1 and IL-6 was assessed in various brain regions. Although MCMV induced increases in proinflammatory cytokine mRNA throughout the brain (especially in ADX animals), no remarkable differences were found between WT and TNF-R1-KO mice. These results demonstrate that endogenous glucocorticoids restrain proinflammatory cytokine responses to viral infection and their impact on locomotor/exploratory activity. Moreover, TNF-alpha appears to mediate cytokine-induced changes in open-field behaviors, especially those believed to reflect anxiety.

Introduction

Proinflammatory cytokines, including interleukin (IL)-1, IL-6, and tumor necrosis factor (TNF)-alpha, released during an immune challenge, such as viral infection, are well known to interact with multiple levels of the hypothalamic–pituitary–adrenal (HPA) axis, leading to the release of glucocorticoids (for review, see Silverman et al.¹ and Turnbull and Rivier²). These effects appear to be primarily driven by effects of proinflammatory cytokines on corticotropin releasing hormone (CRH) in the paraventricular nucleus of the hypothalamus, although a number of studies have shown that these cytokines can interact directly with the pituitary and adrenal glands, leading to glucocorticoid release.^1,3 Glucocorticoids, in turn, have been shown to restrain inflammatory responses in the periphery and central nervous system through antagonism of inflammatory signaling pathways, thereby inhibiting further production of inflammatory mediators.^4–12

Previous work by our group and others has demonstrated that removal of glucocorticoids (via adrenalectomy (ADX)) during viral infection is associated with elevations in plasma and tissue concentrations of proinflammatory cytokines and renders animals more susceptible to septic shock and death.^13–15 In addition, in the case of infection with murine cytomegalovirus (MCMV), a cytopathic virus that induces an early natural killer cell-mediated, anti-viral defense accompanied by elevations in plasma TNF-alpha, IL-1 and IL-6,¹⁵ the increased death rate in adrenalectomized, MCMV-infected, animals can be reversed by glucocorticoid replacement.¹⁵ Previous studies also have shown that the MCMV-induced corticosterone response, which parallels peak plasma cytokine concentrations and occurs around 36 h after infection, is dependent on IL-6.¹⁶ Interestingly, MCMV-induced sepsis and death in ADX animals, which occur later in infection, is mediated by TNF-alpha.¹⁵ Therefore, the IL-6-induced glucocorticoid response during infection with MCMV is essential for protection against TNF-alpha-mediated lethality. Whether excessive TNF-alpha release in glucocorticoid-deficient, virally infected mice leads to other consequences, including changes in behavior, has yet to be examined.

Accumulating data demonstrate that proinflammatory cytokines, including TNF-alpha, released during immune challenge contribute to behavioral alterations collectively referred to as ‘sickness behavior’. These behavioral changes include anhedonia, ano-rexia, altered sleep patterns, impaired cognition, altered social behavior, hyperalgesia and reduced locomotor activity and exploratory behavior.^17,18 Administration of proinflammatory cytokines alone or in combination to laboratory animals has been shown to reliably produce sickness behavior.^17–20 Moreover, antagonism (or genetic deletion) of cytokine activity has been found to ameliorate or reverse various symptoms of sickness behavior in cytokine or immune-challenged laboratory animals.^{17,19,21–31}

Given the relationship among glucocorticoids, proinflammatory cytokines and behavior, we hypothesized that virally (MCMV)-infected mice rendered glucocorticoid deficient by ADX would exhibit exaggerated proinflammatory cytokine release accompanied by evidence of increased behavioral change as assessed by open-field behavior. Behavioral observation of mice in an open field is a standard paradigm for the study of locomotor activity and exploration.^32,33 Moreover, open-field behavior has been shown to be altered in cytokine-treated and infected animals as well as animals with autoimmune disease.^{21,28,34–42} In addition, since TNF-alpha has been demonstrated to mediate MCMV-induced consequences on survival, we examined the possibility that TNF-alpha might also contribute to behavioral alterations in MCMV-infected animals. To abrogate TNF-alpha activity, TNF-alpha receptor 1 (TNFRp55) knockout (TNF-R1-KO) mice were employed. Genetic deletion of TNF-R1 (p55) has been shown to largely abolish TNF-alpha receptor signaling through nuclear factor kappa B, leading to significant impairment in host defenses.⁴³

Materials and methods

Subjects

C57BL/6 male mice (ages 6–10 weeks) were housed in the Emory University Animal Care Facility for at least 1 week before use, and all animals were kept on a 12 h light/dark cycle with lights on at 0700. TNF-R1-KO animals were purchased from Jackson Laboratories (Bar Harbor, ME, USA) (B6.129-TNFrsf1a^tmMak) and bred at Emory University. Normal (wild-type (WT)) C57BL/6 mice were used as controls in experiments using TNF-R1-KO animals (Jackson Laboratories). All protocols were approved by the Emory Institutional Animal Care and Use Committee.

Surgical procedure

Adrenalectomy (surgical removal of adrenal glands) and sham-ADX operations were performed as previously described.¹⁵ For ADX mice, 0.9% saline drinking water was supplemented with 50 mg/ml corticosterone (Sigma, St Louis, MO, USA) for 3 days following surgery, after which ADX mice were maintained on 0.9% saline drinking water without corticosterone supplementation. Recovery from surgery lasted a total of 5 days. Plasma ACTH levels were measured at the time of sacrifice to ensure complete ADX. Mice with ACTH values < 1000 pg/ml were considered incompletely ADX’d and were excluded from data analysis.

Virus

Stocks of salivary gland-extracted MCMV, Smith strain, were generated by Christine Biron (Brown University, Providence, RI, USA). Mice underwent intraperitoneal (i.p.) injection with MCMV (1 × 10⁵ plaque forming units (PFU)/mouse) in 100 μl of vehicle or with 100 μl of vehicle alone (1× media 199 (Invitrogen Life Technologies, Grand Island, NY, USA) with 3% fetal bovine serum (HyClone, Logan, UT, USA)) between 2000 and 2100 Infections took place 7 days post-ADX/sham surgery. Plasma IL-6 was measured (enzyme-linked immunosorbent assay (ELISA)) in all experiments to determine evidence of successful infection. MCMV-injected mice with IL-6 < 100 pg/ml were not included in data analysis.

Open field

A baseline open-field test was performed 5 days after surgery, (2 days before MCMV infection) and once again at 36 h post-MCMV infection. Each mouse was weighed and placed in a holding cage for no more than 3 min. Each mouse was removed from the holding cage and placed in the corner of an open-field apparatus (40 × 40 × 30 cm³, opaque plastic walls). After release into the open-field, data acquisition ran for 5 min. Total distance moved in the entire arena during the 5 min was recorded for each mouse and analyzed using Ethovision by Noldus (Leesburg, VA, USA). Distance was tracked using the color subtraction method, detecting objects darker than the background. Ethovision was also used to automatically count rearings and total time spent in the center as well as entries into the center of the field. The center was defined as any place in the box not within 8 cm of a wall. After testing, each mouse was placed back into its cage.

Plasma and brain collection

At 42 h post-infection (between 1430 and 1500), animals were killed, and trunk blood was collected under low stress conditions (within 3 min of exposure to isofluorane anesthesia). Blood was collected into EDTA-tubes, kept on ice and centrifuged at 4°C. Plasma was aliquoted and stored at −80°C until assayed for hormone/cytokine levels. Whole mouse brains were flash-frozen in dry ice at the time of killing, stored at −80°C, and then later were partially thawed in ice-cold TRIzol reagent (Invitrogen Life Technologies, Carlsbad, CA, USA) before dissection.

Plasma hormone and cytokine assays

Plasma ACTH was measured by radioimmunoassay (RIA) (Allegro HS-ACTH, Nichols Institute Diagnostics, San Juan Capistrano, CA, USA; limit of detection = 5 pg/ml; intra-assay CV = 3.1%; inter-assay CV = 7.3%). Plasma IL-6 was measured by sandwich ELISA (Quantikine murine kit, R&D Systems, Minneapolis, MN, USA; limit of detection = 3 pg/ml; intra-assay CV = 4.5%; inter-assay CV = 7.1%). Plasma TNF-alpha was also measured by sandwich ELISA (Quantikine murine kit, R&D Systems, Minneapolis, MN, USA; limit of detection = 5.1 pg/ml; CV = 6.6%; inter-assay CV = 7.5%).

RNase protection assays

Simultaneous detection of cytokine mRNAs (TNF-alpha, IL-1 alpha, IL-1 beta, IFN-alpha, and IL-6) was performed using RNA extracts obtained from crude prefrontal cortical, hippocampal and hypothalamic dissections along with a riboquant custom probe set from PharMingen (San Diego, CA, USA). Following dissection, brain tissue samples were immediately homogenized with TRIzol reagent to yield total RNA, and were centrifuged to remove cellular debris. RNA pellets for each brain section were obtained from three to five mice, and were resuspended in nuclease-free water for group/batch processing. 32P-uridine triphosphate-labeled cRNA probes were hybridized overnight with 20 μg of total sample RNA at 56°C, digested with RNase A/T1 mixtures, extracted with phenol/chloroform, ethanol precipitated and then separated on 5% polyacrylamide/8 m urea gels. The gels were dried and then subjected to autoradiography to observe protected bands. Specific bands were identified on the basis of their individual migration patterns in comparison with the undigested probes. The bands were quantitated by densitometric analysis (National Institutes of Health Image, Bethesda, MD, USA) and were expressed as arbitrary pixel units normalized according to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) density measures in each sample. All samples were analyzed in triplicate.

Statistical analysis

Values are presented as means±s.e.m. Two-way analysis of variance (ANOVA) was employed to assess for main effects of glucocorticoid status (sham vs ADX) and infection (media vs MCMV) or their interaction on dependent variables within a given genotype (WT vs TNF-R1-KO). A similar strategy was used to evaluate the main effect of genotype and infection as well as their interaction on dependent variables within the ADX condition. For analysis of brain cytokine mRNAs, two-way ANOVA was used to assess for main effects of treatment condition (sham-media vs sham-MCMV vs ADX-MCMV or WT-ADX-MCMV vs TNFR1-KO-ADX-MCMV) and brain region (prefronatal cortex vs hippocampus vs hypothalamus) and their interaction. The Student–Newman–Keuls method was used for post hoc comparison of specific group means. All tests of significance were two-tailed with the alpha level set at 0.05.

Results

Plasma cytokine concentrations and cytokine mRNA expression in the brain of intact and ADX, MCMV-infected WT mice

As shown in Figure 1, MCMV infection had a significant main effect on peripheral blood TNF-alpha and IL-6 concentrations (F [1,26] = 27.14, P < 0.001 and F [1,25] = 56.56, P < 0.001, respectively) with MCMV-infected mice exhibiting higher plasma concentrations of TNF-alpha and IL-6 compared to non-infected animals at the time of killing (42 h post-infection). Glucocorticoid status (ADX vs sham-operated) also exerted a significant main effect on peripheral cytokine concentrations (TNF: F [1,26] = 8.10, P < 0.01; IL-6: F [1,25] = 25.05, P < 0.001), and there was a significant interaction between MCMV infection and glucocorticoid status for both cytokines (TNF: F [1,26] = 6.76, P < 0.05; IL-6: F [1,25] = 25.34, P < 0.001). Post hoc comparisons revealed that MCMV-infected mice rendered glucocorticoid deficient by ADX had significantly higher plasma TNF-alpha and IL-6 concentrations than MCMV-infected, sham-operated mice as well as non-infected, ADX, control animals.

Figure 1.

MCMV-induced plasma TNF-alpha and IL-6 in ADX and non-ADX Mice. Plasma TNF-alpha (a) and IL-6 (b) concentrations after i.p. injection of MCMV (1 × 10⁵ PFU/ mouse) or vehicle (media) were measured in WT mice having undergone a sham surgery or ADX. Trunk blood was collected from mice at 42 h post-infection (n = 5–10 animals per group). Data were analyzed by two-way ANOVA, and group differences were assessed using the Student–Newman–Keuls post hoc test of significance. *P < 0.05 compared to ADX/media group. ^#P < 0.05 compared to sham/MCMV group.

Messenger RNA expression of cytokines in the brain also differed as a function of MCMV infection and glucocorticoid status (TNF-alpha: F [2,18] = 280.26, P < 0.001; IL-1 alpha: F [2,18] = 423.11, P < 0.001; IL-1 beta: F [2,18] = 60.36, P < 0.001; IFN-alpha: F [2,18] = 16.81, P < 0.001; IL-6: F [2,18] = 46.88, P < 0.001) (Table 1, Figure 2), although there was no main effect of brain region on cytokine expression. Sham-operated, MCMV-infected mice reliably exhibited increases in TNF-alpha and IL-1 alpha across all brain regions, while ADX-MCMV-infected mice exhibited significantly greater brain cytokine mRNA for all cytokines in all brain regions compared to both media-treated and MCMV-infected mice.

Table 1.

Brain cytokine mRNA expression in adrenal intact and adrenalectomized, MCMV-infected and media-treated WT mice

	Prefrontal cortex			Hippocampus			Hypothalamus
	Sham- media	Sham- MCMV	ADX- MCMV	Sham- media	Sham- MCMV	ADX- MCMV	Sham- media	Sham- MCMV	ADX- MCMV
TNF-alpha	8±2	77±11*	358±20*#	9±2	69±17*	223±24*#	11±2	67±6*	340±41*#
IL-1 alpha	125±4	281±19*	765±37*#	114±7	225±42*	816±33*#	116±6	248±17*	1001±83*#
IL-1 beta	28±5	35±9	247±17*#	19±5	30±13	144±9*#	18±3	25±5	215±19*#
IFN-alpha	24±5	27±3	55±9*#	22±4	29±9	40±5	18±1	24±4	50±6*#
IL-6	25±3	5±5*	150±14*#	23±7	36±13	84±5*	22±1	35±8	116±20*#

Abbreviations: ADX, adrenalectomy; sham, sham adrenalectomy; MCMV, murine cytomegalovirus; WT, wild-type.

Results are expressed as arbitrary units±s.e.m. normalized to GAPDH in each sample (n = 3–5 per group).

^*P < 0.05 compared to sham-media within the same brain region.

^#P < 0.05 compared to sham-MCMV within the same brain region.

Figure 2.

Brain cytokine mRNA expression in adrenal intact and ADX WT and TNF-R1-KO mice treated with MCMV or media. TNF-R1-KO and WT mice received i.p. injections of MCMV (1 × 10⁵ PFU/mouse) or vehicle (media) after having undergone a sham surgery or ADX. Whole brains were collected at 42 h post-infection (n = 3–5 animals per group), and crude dissections of prefrontal cortex, hippocampus and hypothalamus were performed. Total RNA was extracted from brain tissues, and an RNase protection assay including cRNA probes for TNF-alpha, IL-1 alpha, IL-1 beta, IFN-alpha and IL-6 was conducted. A representative experiment is depicted. See Tables 1 and 2 for statistical analyses of autoradiographic results from multiple RNase protection assays.

Open-field behavior in intact and ADX, MCMV-infected WT mice

MCMV infection had a significant main effect on total distance moved, number of rearings, center entries and time spent in the center (F [1,23] = 43.44, P < 0.001; F [1,23] = 25.59, P < 0.001; F [1,23] = 19.19, P < 0.001; and F [1,23] = 12.28, P < 0.01, respectively) (Figure 3), whereas glucocorticoid status had a main effect on number of center entries and time spent in the center (F [1,23] = 8.19, P < 0.01 and F [1,23] = 5.23, P < 0.05, respectively). Significant interactions of MCMV infection and glucocorticoid status were found for distance moved and center entries (F [1,23] = 4.82, P < 0.05 and F [1,23] = 6.81, P < 0.05, respectively). Post hoc comparisons revealed that ADX, MCMV-infected mice exhibited more dramatic alterations in open-field behavior with significant reductions in distance moved, center entries and time spent in the center compared to all other groups (Figure 3). Sham, MCMV-infected mice also exhibited significantly decreased distance moved and number of rearings compared to non-infected animals, but other open-field behaviors were not significantly altered in these mice compared to non-infected animals.

Figure 3.

Effects of MCMV infection and adrenalectomy on open-field behavior. Total distance moved (a), number of rearings (b), number of entries into the center (c), and time spent in the center (d) at 36 h after i.p. injection of MCMV (1 × 10⁵ PFU/mouse) or vehicle (media) were measured over 5 min in WT mice having undergone a sham surgery or ADX (n = 5–10 animals per group). Data were analyzed by two-way ANOVA, and group differences were assessed using the Student–Newman–Keuls post hoc test of significance. *P < 0.05 compared to respective media-treated control group. ^#P < 0.05 compared to sham/MCMV group.

Relationship between plasma cytokines and open-field behavior in WT mice

Plasma concentrations of TNF-alpha in sham and ADX, MCMV-infected mice were negatively correlated with total distance moved (Pearson r = −0.72, P < 0.01), number of center entries (Pearson r = −0.65, P < 0.01) and time spent in the center (Pearson r = −0.56, P < 0.05), but not number of rearings (r = −0.18, P = NS) (Figure 4). Similar correlations were found with plasma IL-6 concentrations (data not shown). In addition, both number of center entries (Pearson r = 0.77, P < 0.001) and time spent in center (Pearson r = 0.60, P < 0.05) were positively correlated with total distance moved, and TNF-alpha and IL-6 were positively correlated with each other (Pearson r = 0.83, P < 0.001).

Figure 4.

MCMV-induced TNF-alpha and open-field behavior. MCMV-induced TNF-alpha levels in sham-operated and ADX, MCMV-infected animals were negatively correlated with total distance moved (a), number of entries into the center (c), and time spent in the center (d) of an open field, but not number of rearings (b). **P < 0.01, *P < 0.05.

Plasma cytokine concentrations and cytokine mRNA expression in the brains of ADX, MCMV-infected WT and TNF-R1-KO mice

To investigate the role of TNF-alpha in alterations of open-field behavior as well as plasma and CNS cytokines in infected mice rendered glucocorticoid deficient, ADX-MCMV-infected WT (C57Bl/6) and TNF-R1-KO mice were examined. MCMV infection again had a significant main effect on both TNF-alpha and IL-6 in ADX animals (F [1,29] = 50.23, P < 0.001 and F [1,26] = 91.51, P < 0.001, respectively), just as reported above (Figure 1). Nevertheless, there was no main effect of genotype (WT versus TNF-R1-KO) on TNF-alpha or IL-6 plasma concentrations (F [1,29] = 1.72, P = 0.2 and F [1,26] = 0.55, P = 0.46, respectively), nor any interaction between MCMV infection and genotype (TNF: F [1,29] = 1.99, P = 0.17; IL-6: F [1,26] = 0.56, P = 0.46). As shown in Figure 5a, ADX-TNF-R1-KO mice exhibited a robust plasma TNF-alpha response to MCMV infection, albeit the response was slightly but significantly lower than that observed in MCMV-infected, ADX, WT mice. Of note, however, the plasma TNF-alpha response in ADX, TNF-R1-KO mice to MCMV was significantly higher than that observed in MCMV-infected, sham-operated WT animals (see Figure 1) [168 (s.e.m. 22) pg/ml versus 80 (s.e.m. 8) pg/ml, t = 2.69, df = 13, P < 0.05]. MCMV-induced plasma IL-6 levels were also robustly elevated in ADX-TNF-R1-KO mice and were similar to those observed in MCMV-infected-ADX-WT mice (Figure 5b).

Figure 5.

MCMV-induced TNF-alpha and IL-6 in ADX, TNF-R1-KO mice. Plasma TNF-alpha (a) and IL-6 (b) levels after i.p. injection of MCMV (1 × 10⁵ PFU/mouse) or vehicle (media) were measured in ADX WT and TNF-R1-KO mice. Trunk blood was collected from mice at 42 h post-infection (n = 5–10 animals per group). Data were analyzed by a two-way ANOVA, and group differences were evaluated using the Student–Newman–Keuls post hoc test of significance. *P < 0.05 compared to respective media-treated control group. ^#P < 0.05 compared to WT-MCMV-infecetd group.

As shown in Figure 2, ADX-WT and TNF-R1-KO mice exhibited marked induction of all cytokine mRNAs in prefrontal cortex, hippocampus and hypothalamus following MCMV infection compared to sham-operated MCMV- or media-treated mice. Nevertheless, there was no main effect of genotype on cytokine mRNA expression, although there was a significant interaction between genotype and brain region for TNF-alpha (F [2, 18] = 5.24, P < 0.05), with TNF-alpha expression being significantly reduced in the hypothalamus of TNF-R1-KO mice compared to WT animals (see Table 2). No other differences in cytokine expression in the brain were found between TNF-R1-KO and WT mice.

Table 2.

Brain cytokine mRNA expression in adrenalectomized, MCMV-infected, WT and TNF-R1 KO mice

	Prefrontal cortex		Hippocampus		Hypothalamus
	WT	TNF-R1 KO	WT	TNF-R1 KO	WT	TNF-R1 KO
TNF-alpha	308±42	369±22	258±33	187±19*	292±18	175±29*#
IL-1 alpha	681±65	969±136	811±28	698±76	901±32	779±138
IL-1 beta	270±17	333±51	208±63	138±44	197±5	186±43
IFN-alpha	126±63	172±92	125±81	97±48	144±88	118±73
IL-6	236±83	344±60	182±95	151±58	205±108	164±54

Abbreviations: KO, knockout; MCMV, murine cytomegalovirus; WT, wild-type.

Results are expressed as arbitrary units±s.e.m. normalized to GAPDH in each sample (n = 3–5 per group).

^*P < 0.05 compared to the prefrontal cortex in TNF-R1 KO animals.

^#P < 0.05 compared to WT group in same brain region.

Open-field behavior in ADX, WT and TNF-R1-KO mice infected with MCMV

Regarding open-field behavior, MCMV infection had a significant main effect on total distance moved, number of rearings, number of center entries and time spent in center (F [1,29] = 58.15, P < 0.001; F [1,29] = 15.42, P < 0.001; F [1,29] = 28.15, P < 0.001; and F [1,29] = 9.52, P < 0.01, respectively), whereas genotype had a significant main effect on the number of center entries and time spent in center (F [1,29] = 5.22, P < 0.05 and F [1,29] = 8.00, P < 0.01, respectively). A significant interaction was found between infection and genotype for total distance moved (F [1,29] = 6.33, P < 0.05). Post hoc comparisons indicated that MCMV-infected, ADX, TNF-R1-KO mice exhibited significantly more rearings, time spent in the center, and number of entries into the center compared to ADX, MCMV-infected WT mice, however, no differences were observed between TNF-R1-KO mice and WT mice in total distance moved (Figure 6). Given the lack of reversal of reduced distance moved in TNF-R1-KO mice, a correlation was conducted between plasma IL-6 concentrations and total distance moved in ADX, MCMV-infected, TNF-R1-KO mice. Of note, a significant negative correlation was found between IL-6 levels and total distance moved in these animals (Pearson r = −0.70, P < 0.05). No such correlation was found between distance moved and plasma TNF-alpha concentrations (Pearson r = −0.36, P = 0.13).

Figure 6.

Effects of MCMV infection and adrenalectomy on open-field behavior in TNF-R1-KO Mice. Total distance moved (a), number of rears (b), number of entries into the center (c), and time spent in the center (d) at 36 h after i.p. injection of MCMV (1 × 10⁵ PFU/mouse) or vehicle (media) were measured over 5 min in ADX WT or TNF-R1-KO mice (n = 5–10 animals per group). Data were analyzed by a two-way ANOVA, and group differences were assessed using the Student–Newman–Keuls post hoc test of significance. *P < 0.05 compared to respective media-treated group. ^#P < 0.05 compared to WT-MCMV-infected group.

Discussion

The present study demonstrates that when the inhibitory influence of glucocorticoids on proinflammatory cytokines was removed during viral infection, significant increases in peripheral and central nervous system cytokines were observed in association with significant alterations in open-field behavior, including reduced locomotor activity (distance moved) and reduced exploratory behavior (number of rearings, center entries, time spent in center). Behavioral changes were correlated with plasma concentrations of TNF-alpha and IL-6. In mice with the TNF-R1 (p55) receptor gene ‘knocked out’, alterations in exploratory activity that were apparent in ADX, virally infected mice were largely reversed. Of note, however, alterations in locomotor activity (distance moved) persisted in ADX, MCMV-infected, TNF-R1-KO animals and correlated with increases in IL-6, suggesting that there may be differential association of these cytokines with relevant behavioral end points.

Previous studies have demonstrated the importance of glucocorticoids in restraining the inflammatory response to immune challenge. Indeed, animals rendered glucocorticoid deficient through surgery (ADX) or pharmacologic manipulation (e.g. administration of antagonists of the glucocorticoid receptor) have been shown to exhibit increased proinflammatory cytokine expression in both the periphery and brain following a number of immune challenges, including viral infection and administration of bacterial endotoxin, lipopolysaccharide (LPS).^12,15 Of note, although not examined in this study, replacement of glucocorticoids in glucocorticoid-deficient animals has been shown to reverse the detrimental consequences of these effects.^13,15

Consistent with the observed changes in the behavior of MCMV-infected mice, a number of studies have found alterations in locomotor and exploratory activity in immune-challenged animals. For example, rats or mice peripherally injected with LPS (i.p.),^{19,28,34,35,44} turpentine (subcutaneous),²⁴ and influenza (intranasal)^19,24,28 or centrally injected with human immunodeficiency virus (HIV)-1 gp120 (intracerebroventricular – i.c.v.)⁴⁵ or mycoplasma fermentans³⁶ (i.c.v.) have been shown to display reduced locomotor and exploratory behavior. In addition, in mouse models of autoimmune disease, autoimmune mice display less exploratory behavior than control animals.^37,46 Changes in open-field behavior have also been documented in animals administered cytokines in isolation, including TNF-alpha and IL-1 and in animals with relevant cytokine genes deleted or overexpressed.^{23,39–42,47–49} Taken together with our results, these findings support the notion that cytokines released during immune activation can influence open-field behavior, both in terms of gross locomotor activity as well as exploration.

Regarding the role of specific cytokines in behavioral changes, the data presented herein provide evidence that TNF-alpha may play an important role in anxiety-like behaviors within the context of a viral infection. Although virally infected animals rendered glucocorticoid deficient by ADX exhibited significant reductions in both overall locomotor activity (as measured by distance moved) and exploratory activity (as measured by number of rearings, time spent in center and center entries), only the effects of infection on time spent in center, center entries and number of rearings were significantly reversed in MCMV-infected, ADX, TNF-R1-KO animals. Time spent in center, center entries and number of rearings are generally believed to reflect anxiety-like behaviors and are responsive to anxiolytic medications including benzodiazepines and 5HT_1A agonists, but not antidepressants, such as serotonin reuptake inhibitors.³² These open-field behaviors are also sensitive to anxiogenic stimuli including inverse agonists of the benzodiazepine receptor as well as CRH receptor agonists.³² Thus, the impact of MCMV-induced proinflammatory cytokines in ADX animals on behaviors believed to reflect anxiety appears to be mediated by TNF-alpha or other relevant downstream cytokines.

In support of our findings that TNF-alpha plays a role in mediating virus-induced increases in anxiety-like behavior, systemic administration of TNF-alpha has been shown to provoke anxiogenic responses in rodents as assessed by the elevated plus maze test of anxiety.⁵⁰ Of note, IL-1 beta has been shown to induce a similar anxiogenic response, and therefore it remains possible that the behavioral effects of genetic deletion of TNF-R1 may be mediated by downstream influences of TNF-alpha on IL-1 expression.⁵⁰ However, in the current study, no decrease in CNS IL-1 beta expression was observed in TNF-R1-KO vs WT animals, suggesting that the effects of genetic deletion of TNF-R1 were not mediated by a decrease in IL-beta expression. Transgenic mice expressing high levels of brain TNF-alpha,^48,49 as well as mice with increased hippocampal uptake of TNF-alpha following traumatic brain injury,⁵¹ have also been shown to exhibit alterations in exploratory activity including a reduced number of rearings. Moreover, high levels of TNF-alpha have been associated with symptoms of anxiety in humans under conditions of psychological stress⁵² or following administration of LPS.⁵³ Interestingly, benzodiazepines have been shown to inhibit LPS-induced production of TNF-alpha from human microglial cells⁵⁴ and monocytes⁵⁵ as well as murine macrophages.⁵⁶ These findings suggest that inhibition of TNF-alpha production (or other proinflammatory cytokines), centrally or peripherally, could contribute to the mechanism of action of these anxiolytic agents.

Regarding potential mechanisms that may be involved in TNF-alpha effects on anxiety-like behaviors, TNF-alpha has been shown to induce the expression of a number of anxiogenic neuropeptides including CRH.^57,58 In addition, as noted above, TNF-alpha effects may be mediated by induction of downstream cytokines such as IL-1, which has been shown to have a behavioral profile similar to TNF-alpha in open-field testing.^40–42 Finally, consideration should be given to the possibility that reduced host responses to viral infection in TNF-R1-KO mice may be responsible (in relatively non-specific fashion) for the reversal of anxiety-like behaviors in glucocorticoid-deficient, MCMV-infected animals. As mentioned previously, TNF-R1-KO mice have been shown to exhibit reduced host responses.⁴³ Nevertheless, plasma IL-6 responses in ADX, MCMV-infected-TNF-R1-KO animals were similar to those seen in ADX, MCMV-infected, WT mice, suggesting that at least induction of this inflammatory cytokine was not altered as a function of genetic manipulation of the TNF-alpha receptor gene.

Of note, the effects of MCMV infection in ADX animals on locomotor activity were not reversed in TNF-R1-KO animals and were correlated with IL-6. IL-6 has been shown to play a role in locomotor/ exploratory activity in several studies. For example, i.c.v. injection of IL-6 has been shown to reduce locomotor activity and IL-6-deficient animals have been shown to exhibit increased locomotor activity in open-field testing following i.p. injection of LPS or IL-1.^23,38,59 Interestingly, however, one study reported that administration of IL-6 only reduced locomotor activity in the presence of IL-1-beta, while having no effects on it own.⁶⁰ Taken together with the results from our study (where TNF-R1-KO animals exhibited unchanged CNS IL-beta responses), these data raise the possibility that additional pathways (such as IL-1) associated with IL-6 (but not primarily involving TNF-alpha) may be responsible for altered locomotor activity in virally infected mice. Indeed, as noted above, IL-1 has been shown to reduce locomotor activity both alone and in combination with IL-6.^40–42,60

Given the critical role played by glucocorticoids in regulating immune responses (especially restraining inflammatory responses), patient populations with altered glucocorticoid signaling as result of reduced hormone or reduced glucocorticoid receptor expression/function, including post-traumatic stress disorder (PTSD) and major depression, may be especially vulnerable to the pathological/behavioral consequences of prolonged exposure to elevated circulating levels of proinflammatory cytokines. Indeed, both of these disorders have been found to be associated with an exaggerated inflammatory responses including increases in plasma concentrations of TNF-alpha and IL-6, and both are associated with increased symptoms of anxiety and alterations in locomotor activity, including psychomotor slowing and fatigue.^61–64 It has also been suggested that proinflammatory cytokines may contribute to some of the behavioral features of other disorders including chronic fatigue syndrome and fibromyalgia, which are also characterized by reduced glucocorticoids and increased proinflammatory cytokines.^61,65,66

Several limitations of this study warrant further consideration. First, we did not replace glucocorticoids to substantiate that corticosterone and not some other adrenal product or factor regulated by glucocorticoids including catecholamines was the mediator of the observed behavioral changes. That we have previously demonstrated the role of corticosterone replacement in reversing increased lethality in ADX, MCMV-infected animals, however, lends support to our contention that similar results would be obtained in the current study.¹⁵ We also did not antagonize IL-6, and therefore, although correlational data in the presence of reduced TNF-alpha receptor signaling support the idea that IL-6 may have a specific relationship with reduced distance moved (locomotor activity), this notion remains purely conjectural without further experimentation.