Background
The association between depression and inflammation has been recognized for some time [
1,
2]. Indeed, clinical trials have reported antidepressant treatment effects for anti-inflammatory agents such as non-steroidal anti-inflammatory drugs (NSAIDs), and pro-inflammatory cytokine inhibitors have also shown antidepressant treatment effects compared to placebo. Tumour necrosis factor alpha (TNFα) blockade, for example, improved depressive symptoms in patients with treatment-resistant depression, but only in patients with high baseline CRP levels [
3], suggesting that the anti-inflammatory therapy targets processes independent of the etiological mechanisms underlying major depressive disorder (MDD). However, the additive nature of inflammation-induced depressive-like behaviours when combined with MDD highlights that inflammation is likely to be clinically relevant tractable target in many clinical forms of depression. However, while the impact of inflammatory challenges on the negative affect component of depression has been examined, the impact of inflammation on other accompanying behaviours has often been overlooked. Aberrant social behaviours, particularly aggression, as well as psychomotor agitation, often accompany depression and stress-related conditions in man and rodents [
4‐
6]. Indeed, aggressive behaviour during major depression is associated with an enhanced risk of suicide [
7]. Altered neuroimmune responses are also known to contribute to the neurobiology of aggression [
8], and pro-inflammatory cytokine production, in particular, has been implicated in the mechanisms underpinning the stress response [
5,
9] as well as aggressive behaviour [
10‐
13].
Human and animal studies have linked aggression and impulsivity to the increased production of certain inflammatory mediators [
11,
14]. In particular, aggressive traits in humans have been associated with increased serum TNFα [
12], C-reactive protein [
15] and other cytokines [
16]. Indeed, patients in whom cytokines have been therapeutically administered often display signs of aggression [
17,
18]. Furthermore, systemic expression of inflammatory mediators, such as increased systemic interleukin-1 beta (IL-1β) and interleukin-6 (IL-6), are associated with locomotor agitation during aging [
11]. Conversely, mice selectively bred for high levels of aggression also display increased cytokine levels [
19] and knockout of both TNFα-receptor-1 and TNFα-receptor-2 abrogates aggressive behaviours [
20] suggesting that overall, cytokines and aggressive behaviours are linked. The finding that stress is associated with the induction of inflammation [
21‐
23] could be interpreted in evolutionary terms, as a coherent mechanism to enhance survival. Stressors, such as predation, could potentially lead to injury and infection. Thus, pre-activation of the immune system would theoretically enhance survival and recovery [
24].
In humans, parallels to the sickness behaviour observed upon systemic infection in rodents clearly exist. Interferon (IFN) therapy is known to induce transient signs of depression or malaise [
25], and systemic inflammatory diseases are known to be accompanied by depressive-like signs [
26,
27]. In rodents, CNS expression of pro-inflammatory cytokines IL-1β and TNFα contribute to anhedonia and behavioural measures of helplessness after chronic stress [
28,
29]. Pro-inflammatory changes are associated with altered serotonergic function [
30], over-expression of the 5-HT
2A receptor and over-expression of the serotonin transporter (SERT) [
29], which together with other 5-HT-related elements underlie mechanisms of depressive symptoms and social dominancy [
31]. Despite this, it remains unclear at what level and to what extent sickness behaviours and depression converge and how similar the underlying molecular profile is. For example, the impact of inflammation-induced depressive-like behaviour compared with chronic stress on measures of aggression and impulsivity or hyperactivity has been largely overlooked. Irrespective of whether the pathways leading to such aberrant behaviours are distinct, it is clear that a ‘double-hit’ of stress and infection impacts on the pathogenesis of depression [
32,
33].
In the current study, we sought to determine the degree to which the behaviours associated with chronic mild stress (CMS) may be influenced by a mild, low-dose lipopolysaccharide (LPS) challenge that does not normally give rise to anything other than transient and subtle changes in behaviour that persist for no more than a couple of hours. MDD is a disease that is characterized by a recurrent episode of depression, but it is often unclear what factors might have precipitated relapse. Here, we were interested to discover how the single LPS challenge would impact on behaviour in a pre-stressed animal at a time when the effect of LPS had resolved. In this way, it is possible to evaluate the residual effects of acute inflammation on stress-induced behavioural changes. We examined behavioural parameters of aggression and impulsivity/hyperactivity, anhedonia and helplessness, as well as the expression of inflammatory and serotonergic markers of the periphery and specific brain areas, including the medial pre-frontal cortex and hippocampus as these sites are well recognized to play a crucial role in the stress response [
34], and we have previously found that 5-HT
2A and SERT expression levels change in response to systemic inflammation [
35] and chronic stress paradigms [
29] in these regions.
A 10-day stress procedure was selected in the present study because it has been previously shown to induce a depressive-like syndrome in mice, which is accompanied by changes in CNS serotonergic and pro-inflammatory genes [
29,
36]. Previous work in rats has shown that repeated LPS challenges, sufficient to induce sickness behaviour, when combined with chronic mild stress can induce additive increases in plasma corticosterone and TNFα in rats will enhance depressive-like behaviour [
37]. In contrast to these findings in rats, our investigations have established, using a single low-dose LPS (0.1 mg/kg) in CMS mice and a broader set of behavioural tests, that there is no simple additive effect when inflammation and stress are combined but highlight selective independent effects on a number of stress-related behaviours and on the underlying molecular biology.
Discussion
The studies reported here show that at a time when the effects of an intraperitoneal injection of LPS are no longer detectable in naïve animals, the combination of LPS with CMS increases depressive-like behaviours and inhibits the aggression and impulsivity induced by CMS. The aggressive and impulsive behaviours were accompanied by SERT induction in the hippocampus, which was ameliorated by the LPS treatment. The double-hit combination had no effect on LPS-induced TNFα expression but did suppress LPS-induced IL-1β mRNA expression. Overall, SERT upregulation, rather than 5-HT
2A or the pro-inflammatory cytokines, appears to correlate with the stress-induced aggressive and impulsive behaviours. A similar independent increase in SERT was previously reported in stressed animals that become anhedonic [
29]. Here, hepatic TNFα and IL-1β mRNA levels differed between stressed mice injected with LPS compared to LPS alone in a surprising manner revealing a dissociation between the regulation of TNFα and IL-1β mRNA expression. Moreover, these changes in hepatic cytokine expression appeared to be independent of corticosterone induction. These results are discussed in more detail below.
Using a low-dose LPS challenge after stress in both the sucrose preference test and the forced swim test, we showed that the downstream sequelae of a peripheral inflammatory response appeared to exacerbate the anhedonia and helplessness induced by stress. Indeed, there was significant synergy for a reduction in sucrose intake. Non-stressed mice exhibit polydipsia, and this in known to be reduced by stress [
57] and is further reduced by the LPS challenge, indicating that low levels of systemic inflammation that may not generate overt clinical signs per se can synergize with a stress-induced depressive illness and provoke a worsening phenotype. Thus, the diagnosis and treatment of low-grade inflammatory disease in patients may reduce some select depressive signs by mechanisms that are independent of those that are associated with major depression and those targeted by traditional antidepressants. Others have shown that the combination of endotoxin with stress in mice can result in increased mortality [
58] but such severe experiments (40 mg/kg LPS compared to 0.1 mg/kg in our studies) did not set out to explore the subtle relationship between low-level infection and stress. In rats, lower levels of endotoxin were previously used to discover how inescapable shock-induced stress would be altered [
59,
60]. In these experiments, the febrile response associated with inescapable shock and LPS was increased and this was associated with enhanced pro-inflammatory cytokine responses. As in our experiments, Johnson et al. [
60] found the relationship between cytokine expression and the double-hit of stress and inflammation was not a straightforward relationship; enhanced pro-inflammatory cytokine responses where not necessary to observe enhanced HPA or fever responses after LPS and inescapable tailshock.
Work studying the immune response after a stressful event has suggested that stress ‘primes’ the inflammatory response for an immune challenge, making it more sensitive [
15]. The depressive-like behaviours associated with an LPS challenge have also been shown to be ameliorated by imipramine and fluoxetine given prior to LPS administration [
61], and our results suggest that while antidepressants might target the post-infection component of the combination, anti-inflammatory therapy might also be beneficial. Indeed, celecoxib administered as an adjunctive non-steroidal anti-inflammatory drug (NSAID) appears to produce a positive therapeutic outcome in the treatment of depression [
62].
In this study, although chronically stressed mice exhibit anhedonia, they also display increased rates of aggressive behaviour in the resident-intruder test where attacking and crawl over behaviours were markedly increased. Crawl overs have been investigated in rats and form part of juvenile play fighting. However, such behaviour has also been observed in aggressive encounters. In rats, crawl overs occur when the rats are unfamiliar with one another and seem to be important in establishing dominance [
45]. Such stress-induced changes in attack frequency have been previously described using the same CMS regime as employed here [
36]. A paradoxical ‘anxiolytic-like profile’, manifest as increased impulsivity in the elevated O-maze, was also observed in response to stress, in line with previously reported findings [
63]. In contrast, stressed mice subjected to a low-level LPS challenge displayed a reduction in aggressive behaviour in the resident-intruder test and no signs of impulsivity/hyperlocomotion in the elevated O-maze. In studies of aggression and impulsivity, the combination of stress and low-level inflammation therefore appears to counteract, rather than exacerbate, the negative effects of stress on behaviour.
Changes in measures of aggressiveness and impulsivity/hyperactivity were accompanied by differential expression of SERT in the brain. In the hippocampus, mRNA levels of SERT were increased in chronically stressed mice. In stressed mice challenged with LPS, expression levels of SERT in the hippocampus did not change, but they did tend towards a decrease in the pre-frontal cortex. Chronically stressed mice without exposure to LPS displayed a non-significant increase in SERT expression in the pre-frontal cortex. These data are in accordance with our previous observations [
30]. Elevated SERT expression was previously reported in mice displaying aggressive behaviour induced by repeated social confrontation stress [
64]. The increase in SERT in the limbic structures of the brain is frequently found after stressors of various types [
65]. In contrast, a decrease in SERT expression in similar structures was shown to be a molecular correlate of clinical depression [
66] and of an experimentally induced depressive-like state in animals [
67]. These data, in combination with our own, suggest changes in molecular signals within specific brain regions may result in behaviourally distinct outcomes.
In vitro and in vivo studies have shown that pro-inflammatory cytokines, such as IL-1β and TNFα, can increase SERT activity via the p38 MAPK signalling pathway [
46]. Behavioural signs of helplessness resulting from circulating cytokines have been shown to be prevented by a blockade of SERT [
68]. Furthermore, SERT mutant rats show abnormal behaviour (including decreased sucrose preference, decreased spontaneous activity and increased anxiety [
69]) and CNS cytokine expression profiles in response to LPS [
70]. In humans, however, the reverse appears to be true. Clinical studies reveal that decreased SERT function, associated with the short variant of the SERT gene and lower SERT activity, correlates with an increased risk of developing depression during IFN-α treatment [
71]. Indeed, our own work has demonstrated that there is no change in the release of 5-HT in response to LPS, suggesting a post-synaptic mechanism may be more crucial to sickness behaviour [
72]. Thus, the relationship between SERT activity and responsiveness to pro-inflammatory factors in the regulation of depression pathogenesis appears to be complex and is liable to explain the differences we observed in aggressive behaviour associated with stress alone vs stress in combination with an inflammatory challenge.
The levels of 5-HT
2A mRNA were different in mice subjected to stress alone to those additionally challenged with LPS. Previously, elevation of 5-HT
2A in the limbic structures was documented as an important correlate of a depressive-like state, which represents a target for pharmacological treatment [
73]. In line with our previous observations [
29,
72], such changes were found in the pre-frontal cortex of stressed mice but not in naïve or stressed mice injected with LPS. However, a significant elevation of 5-HT
2A expression was detected in the hippocampus of the two latter groups, in line with similar findings elsewhere showing that inflammation significantly affects 5-HT
2A [
35,
74]. The similarities in receptor expression profiles regardless of stress exposure suggest that changes in the expression of the 5-HT
2A receptor are unlikely to mediate the exacerbated behavioural effects observed in the double-hit mice.
Importantly, our low-dose LPS challenge in naïve animals resulted in the over-expression of TNFα in several brain structures, including the pre-frontal cortex, but this was not associated with alteration in the behaviours tested. Such findings are in accord with previously published results, showing that cytokine over-expression exerts minimal effects on social behaviour in rodents [
56]. The expression of IL-1β in the dorsal raphe nucleus was significantly elevated in both naïve and stressed LPS-treated groups. However, this effect is also unlikely to underlie behavioural differences between chronically stressed mice, with or without LPS challenge, since naïve mice showed no obvious behavioural changes in aggression or depressive-like behaviours.
Stress is well known to increase circulating cortisol, and there is evidence linking cortisol levels and depression. Depressed patients frequently show dexamethasone non-suppression, suggesting hyperactivity of the Hypothalamic–pituitary–adrenal HPA axis [
75]. Corticosterone levels are similar in animals subjected to either CMS or LPS and thus could not explain the phenotypic differences observed between stressed and LPS-challenged animals. These data are in line with previously reported findings [
76] although oddly, the increase in corticosterone as a result of stress does not appear to reduce the hepatic inflammatory response. This data, and that in adrenalectomized animals, suggests that the pro-inflammatory profile during stress is independent of cortisol and may be the result of anti-inflammatory cytokines and downstream signalling pathways [
77].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
YC participated in the design of the experiments, carried out the molecular studies, performed the statistical analysis and drafted the manuscript. AT, NM and VN carried out the behavioural experiments and helped with the analysis. HWS, VC, CS and K-PL advised on the experimental design and helped to draft the manuscript. DCA and TS conceived of the study, participated in its design and coordination and edited the manuscript. All authors read, edited and approved the final manuscript.