We previously demonstrated that ICE KO mice are resistant to central LPS-induced reduction in food intake
 and feeding behavior
. However, ICE KO mice lack this protection following systemic LPS administration
 although they are protected against peripheral endotoxic shock induced by higher doses of LPS
. These findings suggested that there is a distinction between responses elicited by an activation of the central innate immune system versus the peripheral immune system. Given the more recent data showing that IL-1 is critically involved in the development of depression-like behaviors
[10, 16, 34, 35], these findings stimulated us to test the hypothesis that ICE KO mice might be protected from central but not peripheral LPS-induced depression-like behaviors. To our knowledge, we are now the first to report that ICE is required for development of depression-like behavior following a central, but not systemic, LPS challenge.
ICE is a cysteine protease and is the enzyme principally responsible for cleavage of two critical pro-inflammatory cytokines, pro-IL-1β and pro-IL18, from their inactive precursors to their mature active secreted forms. ICE is constitutively present within cells as an inactive precursor co-localizing with a group of proteins that collectively form the inflammasome. Inflammasomes contain nucleotide and oligomerization domain-like receptor (NLR) family proteins that act as intracellular receptors for pathogen-associated molecular patterns (PAMPs) or danger-associated molecular patterns (DAMPs) in a similar fashion as the widely recognized toll-like receptors
[22, 36, 37]. Following recognition of PAMPs and DAMPs by NLRs, there is a transient increase in ICE activity
. The precise mechanism by which LPS can directly activate inflammasomes and ICE activity has not been fully elucidated. However, LPS-induced upregulation of ICE expression has largely been shown to occur in cells of the myeloid lineage such as central microglia
 and peripheral monocytes/macrophages
. We confirmed that ICE expression in the brain was upregulated by both central and peripheral LPS in WT mice.
ICE activity has substantial influence over behavioral responses during inflammation, presumably through its role in processing inactive IL-1β
. Genetic deletion of ICE decreases LPS-induced IL-1β secretion
. It has been well established that IL-1β activates the hypothalamic-pituitary-adrenal axis and induces behavioral changes associated with sickness and depression, such as anhedonia, disruption of sleep, cognitive disturbances, temperature disregulation, and consumption of food
[9, 26, 27]. Our findings indicate a prominent role for central ICE during inflammation-induced depression. These new data are in in agreement with recent work that has demonstrated a role for IL-1ß in the development of depression-like behavior utilizing a variety of models. Chronic exposure to IL-1β diminished sucrose preference and social exploration; which are indicative of the depressive symptoms of anhedonia and social withdrawal, respectively
[10, 28]. Chronic mild stress (CMS) of mice not only elevates IL-1β levels but also results in depression-like symptoms, including a decrease in sucrose preference (to the point of aversion) and decreased social exploration. These symptoms are dependent on IL-1β as either type 1 IL-1 receptor (IL-1R) deficiency or the injection of IL-1 receptor antagonist (IL-1RA) block the effect of CMS
. Similarly, several other studies using models of CMS, chronic unpredictable stress (CUS), chronic pain, and ischemic stroke have demonstrated that reducing IL-1 signaling blocks depression-like behaviors including reduced sucrose preference and increased immobility time in the FST
[16, 34, 35]. Similar CUS causes mice to display decreased preference for sucrose solution compared to unstressed mice, and this anhedonic response is blocked by ICV administration of IL-1RA
. This later finding is important as it illustrates that IL-1β activity within the brain is required for the development of depression-like behavior. Chronic pain has also been linked to depression with a possible cause and effect relationship
[38, 39]. Moreover, IL-1 has been extensively examined for its role in mediating symptoms of neuropathic pain (reviewed in
). Utilizing the spared nerve injury model of chronic pain, increased time of immobility during the FST was blocked by central administration of IL-1RA into the lateral ventricle. Middle cerebral artery occlusion, which is a model of ischemic stroke, leads to a robust induction of brain IL-1β expression and IL-1 signaling in this model plays an important role in post-stroke depression as evidenced by a reduction in sucrose consumption that is blocked when mice are treated with an ICV injection of IL-1RA
. All of these studies established that brain IL-1 signaling is of significant importance for mediating depression-like behaviors since each model targeted IL-1 signaling in the brain. All of these results are in agreement with the new findings reported here that deletion of ICE blocks the LPS-induced increase in FST immobility and decrease in sucrose preference only when LPS is administered centrally. Clearly, non-ICE dependent mechanisms for depression-like behavior remain functional when LPS is administered peripherally.
Despite reduced IL-1β secretion, ICE KO mice have similar a sickness response compared to WT mice in terms of loss of body weight following both systemic and central LPS challenges. This lack of attenuation of the sickness response is not surprising since IL-1R1 knockout mice also lose body weight similar to WT mice when treated with ICV or IP LPS
. Further, inhibiting IL-1 signaling with ICV or IP administration of IL-1RA is not sufficient to block body weight loss in LPS treated rats
. It was postulated several years ago that LPS-induced sickness behaviors require the presence of either TNF-α or IL-1β, but when IL-1 signaling is absent, TNF-α assumes a more prominent role
. This postulate is consistent with the present findings that both WT and ICE KO mice displayed elevated TNF-α expression in the brain following LPS. Therefore, similar degrees of sickness behavior following LPS injection in WT and ICE KO mice is consistent with the known pleiotropic properties of both IL-1β and TNF-α.
Unexpectedly, ICE KO mice had reduced locomotor activity as assessed by the number of line crossings and rearings compared to WT mice. A possible explanation for reduced exploratory behavior measured in ICE KO mice is increased anxiety in these mice. Exploratory behavior is often used as an important screening tool for anxiety-like behavior
. We cannot rule out the possibility that reduced locomotor activity exhibited by ICE KO mice is indicative of an anxiogenic phenotype but our experiments were not designed to properly test this possibility. However, it is important to note that locomotor activity of ICE KO mice 24 h after LPS was unaffected by either central or systemic LPS challenge. We interpret these data to suggest that ICE KO mice display a more rapid recovery from sickness behavior compared to WT mice. Again, testing this possibility was not the focus of our experiments. Importantly for the current body of work, the decrease in locomotor activity did not translate into an elevated time of immobility during the FST. The equal time of immobility of saline-treated WT and ICE KO mice during this test indicates that reduced locomotor activity was not due to depression-like behavior and that differences in performance during the FST did not result from a motor deficit. However, as ICE inhibitors may draw increasing interest as treatments for inflammation-associated diseases, it will be important to evaluate different alterations for other ICE-dependent behaviors.
Cytokine signals are propagated diffusely throughout the brain following an initial immune stimulation. This propagation occurs primarily by the induction of additional IL-1ß and other pro-inflammatory cytokines in brain resident cells including microglia
[44–48]. Despite having reduced IL-1β secretion in response to LPS
, ICE KO mice still develop depression-like behavior following systemic but not central LPS administration. This finding alludes to compensatory actions of other peripheral cytokines to induce inflammatory mediators within brain despite the deficiency in mature IL-1β, or a lack of peripheral IL-1ß involvement in peripherally induced depression-like behaviors. In our experiments, both WT and ICE KO mice displayed increased early (4 h) expression of brain IL-1β, TNF-α and IL-6 following both systemic and central LPS. At 24 h, brain IL-1β and TNF-α were no longer elevated in central LPS-treated ICE KO compared to saline-treated ICE KO mice even though expression of these genes remained elevated in LPS-treated WT mice. In our model, the reduced induction of IL-1β mRNA expression following LPS treatment likely results from a quicker extinguishing of the feed-forward cytokine signaling, in agreement with data demonstrating that IL-1β induces its own expression and the expression of TNF-α and IL-6 within brain
[49, 50]. This was in contrast to what was observed following systemic LPS because most inflammatory mediators that we measured were increased similarly in WT and ICE KO mice. In contrast, TNF-α expression remained elevated at 24 h following systemic LPS administration in both WT and ICE KO mice. This finding indicates that even in the absence of IL-1ß secretion, TNF-α may mediate depression-like behavior because its absence in the brain at 24 h in ICV-treated ICE KO mice corresponds to a lack of depression-like behavior and its continued presence at 24 h in ICV-LPS WT, IP-LPS WT, and IP-LPS ICE KO mice corresponds to the presence of depression-like behaviors. A role for TNF-α in depression-like behavior has been directly shown. Even extremely low doses of TNF-α administered ICV causes depression-like behavior as assessed as increased time of immobility during both the FST and tail suspension test
. In addition, TNF-R1 deficient mice and mice treated with a neutralizing antibody to TNF-α had a decreased time of immobility during the FST, indicating an anti-depressant response. This study supports work showing that TNF receptor deficient mice have lower immobility during the FST, again indicating an anti-depressant response. The TNF receptor deficient mice also have increased consumption of a sucrose solution, indicative of an anhedonic response mediated by TNF-α
. In further support of a role for TNF-α in depression, human patients afflicted with plaque psoriasis showed significant improvement in Beck Depression Inventory and Hamilton Rating Score for depression when treated with the TNF neutralizing drug Etanercept
. Patients treated with Etanercept showed significant improvement in sexual function, sleep, irritability, and other symptoms of depression that impacted quality of life compared to patients receiving placebo. These data indicate that TNF-α is probably involved in mediating depression-like behaviors. Together with our current data, we hypothesize that in the absence of IL-1ß, depression-like behavior is present only when central TNF-α expression is elevated following the LPS challenge.
Expression of IL-10 and IL-1RA in the brain remain elevated in ICE KO and WT mice following systemic LPS exposure at 24 h. This was not the case following central LPS, further supporting a role for IL-1β in a sustained brain inflammatory response. Compensatory actions of other cytokines such as TNF-α
 contribute to peripheral immune activation cascades when IL-1 action is lost. We also found that ICE KO mice have decreased mRNA expression of genes associated with microglia and astrocyte activation, CD11b and GFAP, following central LPS treatment. We interpret these findings as evidence that IL-1β is important for maintaining activation of glial cells in response to neuroinflammation. Indeed, reduced cytokine expression observed in ICE KO following central LPS is reflective of reduced glial cell activation. We are intrigued by our finding that brain MHCII expression was not different between ICE KO and WT mice following central LPS (data not shown) as this may indicate a less prominent role for IL-1β to induce an antigen presenting phenotype in microglia. Based on these data, our results add to evidence that ICE and subsequently IL-1ß signaling plays a necessary role for initiating and sustaining a full inflammatory response within the brain that manifests behaviors associated with depression.