The impact of IL-1 modulation on the development of lipopolysaccharide-induced cognitive dysfunction

All the experiments were conducted under the UK Home Office approved license. Wild type C57BL/6 male mice pathogen free, 12 to 14 weeks of age, weighing 25 to 30 g were housed in standard cages with no environmental enrichment in groups of five in a 12 hours light 12 hours dark cycle with controlled temperature and humidity, free access to water and standard rodent chow. IL-1R^-/- (kindly provided by Professor Dame Nancy Rothwell, University of Manchester [13]) were bred in-house on a C57BL/6 background and age-matched to wild type counterparts. Seven days of acclimatization were allowed before starting any experiment. All the animals were checked on a daily basis and those with evidence of poor grooming, huddling, piloerection, weight loss, back arching and abnormal activity were excluded in the experiments.

LPS derived from Escherichia Coli endotoxin (0111:B4, InvivoGen, San Diego, CA, USA, 1 mg/kg) was dissolved in normal saline and injected intraperitoneally. IL-1Ra (Amgen, Anakinra 100 mg/kg, Thousand Oaks, CA, USA) was given subcutaneously immediately before LPS administration. Dose response curve from LPS or IL-1Ra was obtained from our pilot studies to provoke or to suppress, respectively, a moderate degree of microglia activation. Control animals were injected with equivalent volumes (0.1 ml) of saline. Mice from each treatment group were randomly assigned for assessment of either cytokine response or cognitive behavior, in order to obviate possible confounding effects of behavioral testing on inflammatory markers [14].

Blood was sampled transcardially after thoracotomy under terminal anesthesia 30 minutes, 2, 6, and 12 hours and 1, 3, and 7 days after experiments in the different cohorts and centrifuged at 3,600 rpm for 7 minutes at 4°C. Blood samples taken from animals without any interventions served as controls. Plasma samples were stored at -20°C for further analysis. Plasma cytokine and HMGB-1 were measured using commercially available ELISA kits from Biosource (Camarillo, CA, USA) and Shino-test Corporation (Kanagawa 229-0011, Japan), respectively. The sensitivities of the assays were less than 3 pg/ml for TNFα, less than 7 pg/ml for IL-1β, less than 3 pg/ml for IL-6 and 1 ng/ml for HMGB-1.

The hippocampus was rapidly extracted under a dissecting microscope, placed in RNAlater solution (Applied Biosystems, Ambion, Austin, TX, USA) and stored at 4°C. Total RNA was extracted using RNeasy Kit (Qiagen, Austin, TX, USA) and quantified. The one-step quantitative (q) PCR was performed on a Rotor-Gene 6000 (Corbett Life Science, Austin, TX, USA), using Assay-On-Demand premixed Taqman probe master mixes (Applied Biosystems, Foster City, CA, USA). Each RNA sample was run in triplicate, and relative gene expression was calculated using the comparative threshold cycle ΔΔC_t and normalized to beta-actin. Results are expressed as fold-increases relative to controls.

Mice were euthanized and perfused transcardially with ice-cold heparinized 0.1 M PBS followed by 4% paraformaldehyde in 0.1 M PBS at pH 7.4 (VWR International, Lutterworth, Leicester, UK). The brains were harvested and post-fixed in 4% paraformaldehyde in 0.1 M PBS at 4°C and cryoprotected in 0.1 M PBS solutions containing 15% sucrose for 24 hours (VWR International, Lutterworth, Leicester, UK) and then 30% sucrose for a further 48 hours. Brain tissue was freeze-mounted in optimal cutting temperature (OCT) embedding medium (VWR International, Lutterworth, Leicester, UK). The 25 μm thick coronal sections of the hippocampus were cut sequentially in groups of six and mounted on Superfrost^® plus slides (Menzel-Glaser, Braunschweig, Germany). The rat anti-mouse monoclonal antibody, anti-CD11b (low endotoxin, clone M1/70.15) at a concentration of 1:200 (Serotec, Oxford, UK) was used to label microglia. Visualization of immunoreactivity for CD11b was achieved using the avidin-biotin technique (Vector Labs, Cambridge, UK) and a goat anti-rat secondary antibody (Chemicon International, Temecula, CA, USA) at a concentration of 1:200. A negative control omitting the primary antibody was performed in all experiments. Immunohistochemical photomicrographs were obtained with an Olympus BX-60 microscope (Olympus Corp., Tokyo, Japan) and captured with a Zeiss KS-300 colour 3CCD camera (Carl Zeiss AG, Tokyo, Japan). The assessment of staining, by an observer that was blinded to the interventional group, was based upon a four-point categorical scale [15].

The behavioral study was conducted using a dedicated conditioning chamber (Med Associates Inc., St. Albans, VT, USA). Mice were trained and tested on separate days. LPS was injected within 30 minutes following training. The fear conditioning paradigm was used as previously described, with minor modifications [16]. Three days after training, mice were returned to the same chamber in which training occurred (context), and freezing behavior was recorded. Freezing was defined as lack of movement except that required for respiration. Approximately three hours later, freezing was recorded in a novel environment and in response to the cue (tone). The auditory cue was then presented for three minutes, and freezing scored again. Freezing scores for each subject were expressed as a percentage for each portion of the test. Memory for the context (contextual memory) for each subject was obtained by subtracting the percent freezing in the novel environment from that in the context. All assessments were performed in a blinded fashion.

Statistical analyses were performed using GraphPad Prism version 5.0a (GraphPad Software, San Diego, CA, USA). The results are expressed as mean ± standard error of the mean. Data were analyzed with analysis of variance followed by Newman-Keuls post hoc test wherever appropriate. For categorical data, non-parametric Kruskal-Wallis followed by Dunn's test was used. A P < 0.05 was considered to be statistical significance.

To investigate the effects of inflammation on cognitive function we measured systemic and central cytokines after LPS administration. TNFα release occurred very rapidly and transiently; after 30 minutes it was significantly increased (104.18 ± 7.36 pg/ml), peaking at two hours and returning to normal at six hours post-injection (Figure 1a; P < 0.01, P < 0.001 vs control). LPS evoked a robust systemic response that induced a stereotypical cytokine release. Both IL-1β and IL-6 were significantly up-regulated from two hours. IL-1β increased four-fold and plasma levels continued to steadily increase until 24 hours (Figure 1b; 73.49 ± 5.42 pg/ml, P < 0.001 vs control). IL-6 expression was markedly elevated at two hours, decreasing at six hours but still significantly detectable at 24 hours compared with naïve animals (Figure 1c; 134.37 ± 8.43 pg/ml, P < 0.01 vs control). During this time, animals showed classic symptoms of sickness behavior (reduced motility, poor grooming, huddling, piloerection, back arching). Levels of HMGB-1 at 2, 6, and 12 hours post LPS were no different from baseline levels; a 1.5-fold increase was observed from 24 hours after LPS and remained elevated up to day 3 (Figure 1d; 25.77 ± 4.2 pg/ml, P < 0.01, P < 0.001 vs control). The systemic inflammatory response resolved after day three and all cytokine levels returned to baseline by day seven. To assess the central inflammatory response to LPS we measured levels of IL-1β and IL-6 mRNA expression in the hippocampus. We noted a 6.5-fold increase in IL-1β mRNA expression and a 15-fold increase in IL-6 in the hippocampus at six hours after LPS injection (Figures 1e and 1f; P < 0.001 vs control). In both cases the increased transcription returned to normal values by 24 hours. The increase in IL-1β both in plasma and in the hippocampus led us to investigate whether blocking the IL-1 receptor could ameliorate the signs of LPS-associated cognitive dysfunction. A single preemptive dose of IL-1 Ra was able to significantly reduce plasma levels of IL-1β at 6 and 24 hours (Figure 2a, 32.7 ± 5.45 pg/ml, 6.2 ± 1.03 pg/ml, P < 0.01 and P < 0.001 vs LPS, respectively). Similarly, levels of IL-6 were also reduced at the same time-points (Figure 2b; 91.02 ± 15.17 pg/ml, 14.05 ± 2.34 pg/ml, P < 0.001, P < 0.001 vs LPS, respectively). Interestingly, IL-1Ra treatment had no effects on HMGB-1 levels, which maintained a similar pattern to that seen after LPS injection in the absence of IL-1Ra (Figure 2c).

Corroboration of these data was achieved by injecting IL-1R^-/- animals with the same dose of LPS and measuring cytokine expression in plasma. At 24 hours, after LPS in the IL-1R^-/-, a time at which there was markedly increased cytokines and clear evidence of sickness behavior in untreated wild type mice, levels of IL-1β and IL-6 were comparable with the wild type mice that received IL-1Ra treatment (Figures 2a and 2b; P < 0.0001, P < 0.001 vs LPS). Contrary to the cytokine changes, the LPS-induced elevation of HMGB-1 was not abrogated in the IL-1R^-/- or IL-1Ra-treated animals (Figure 2c).

The hippocampal transcriptome findings of the pro-inflammatory cytokines prompted interest for other possible markers of neuroinflammation. Normal controls, both from WT and IL-1R^-/-, showed no signs of microgliosis (Figures 3a, e and 3i). Minimal immunoreactivity was reported in naïve animals in which microglia maintained small cell bodies with thin and long ramified pseudopodia (Figure 3b). Resting microglia shifted to a 'reactive profile' after LPS exposure, acquiring an amoeboid morphology with hypertrophy of the cell body and retraction of the pseudopodia. Reactive microglia displayed morphological changes including increased cell body dimensions, shortened and clumpy processes with higher levels of CD11b immunoreactivity compared with naïve animals. Microglial activation was reported at days one and three post exposure (Figures 3c and d;P < 0.01, P < 0.05 vs control, respectively), returning to the baseline resting state by day seven. Pre-treatment with IL-1Ra effectively reduced the number of reactive microglia at days one and three (Figures 3f and 3g). In order to corroborate these findings, we repeated the experiment using IL-1R^-/- animals and exposing them to LPS. No microglial activation was noted in LPS treated IL-1R^-/- mice (Figures 3k and 3l).

To relate the inflammatory response to memory functioning, we used trace fear conditioning in which mice are trained to associate a tone with a noxious stimulation (foot shock). The brief gap between the tone termination and the shock onset allows assessment of hippocampal integrity [16]. The high level of freezing seen in the naïve animals is indicative of good learning and memory retention. Contextual fear response shows a reduced immobility (freezing) at day three, revealing and hippocampal-dependent memory impairment (Figure 4; P < 0.005 vs naïve trained). Pre-treatment with IL-1Ra significantly ameliorated this cognitive dysfunction, abolishing also the symptoms of sickness behaviour otherwise evident in LPS-treated animals (Figure 4; P < 0.05 vs LPS). During the initial 24 hours following LPS administration animals show classic sign of sickness behavior, in particular reduced motility, poor grooming, and back arching. Remarkably, animals treated with pre-emptive IL-1Ra had no signs of sickness behavior, which functionally reflected in better memory retention and no microgliosis. LPS administration caused a permanent retrograde amnesia at both days 3 and 7 (Figure 5).

Cytokines play an important role in mediating the inflammatory response after infection or aseptic traumatic injury. The innate immunity is rapidly triggered after LPS, primarily via activation of toll-like receptor 4 (TLR-4) [17]. Activation of TLR-4 induces a multitude of pro-inflammatory cytokines via activation of transcription factors, nuclear factor (NF)κB [18]. This prompt response provides a favorable environment for the synthesis and upregulation of both IL-1β and IL-6, which together contribute to the perpetuation of the inflammatory challenge. Also the rapid increase in TNFα following LPS, which is reported as already present after 30 minutes, promotes synthesis of other cytokines and the initiation of the acute-phase response, chemokine release and oxidative stress. Systemic cytokines, including IL-1β, can bind receptors and translocate through the intact blood-brain barrier (BBB) [19]. Neural afferents are known to be a fast and reliable pathway in the immune-to-brain signaling. Vagal-mediated signaling can rapidly induce brain cytokines and manifest the classic symptoms of the acute-phase response, including neuroinflammation [20]. As we have reported a significant increase in both IL-1β and IL-6 mRNA transcription at six hours in the hippocampus, the neuronal route may be the likely pathway to trigger the early activation of these genes and the initial changes in the CNS. Vagotomy was previously shown to partially attenuate sickness behavior both after LPS and IL-1β administration [21], but not in the context of hippocampal-dependent cognitive dysfunction.

Within the brain, cytokines interact with microglia cells. Pro-inflammatory cytokines can directly interact with many of the pattern recognition receptors expressed on the surface of these cells [22]. Upon activation, microglia exhibit discernible morphologic changes and secrete cytokines, reactive oxygen species, excitotoxins (such as calcium and glutamate) and neurotoxins such as amyloid-β [23]. Activated microglia also inhibit neurogenesis in the hippocampus following endotoxemia, thereby exacerbating the extent of injury on memory processing [24].

To assess memory retention we used trace fear conditioning in which mice are trained to associate a foot shock with a given environment or tone [25]. The extent to which an animal freezes to a context is largely dependent on the hippocampus [26]. Hippocampal-dependent memory impairment was evident after three days post-LPS. Residual inflammation, primarily via reactive microglia, is possibly associated with this second-phase behavioral abnormality. At these time points, levels of HMGB-1 were also elevated and prompted us to further investigate the role of these factors in the development of cognitive dysfunction. As the cognitive impairment was also present at day seven post LPS exposure (Figure 5) when inflammatory markers returned to baseline, this suggests that the initial acute-phase response may have interfered with processes of memory consolidation in the hippocampus.

IL-1β has a pivotal role in sustaining the neuroinflammatory response and closely interacts with memory processing and long-term potentiation [27, 28]. Self-regulation and inhibition of IL-1β is normally achieved with the neutralizing action of endogenous IL-1Ra, which directly competes for binding to the receptor [29, 30]. Transcription of endogenous IL-1Ra would normally occur temporally delayed from the synthesis of IL-1, thus following pharmacological intervention we aimed to block the receptor a priori impeding binding and limiting the damage mediated by the effector molecule. When the IL-1 receptor is disabled, either blocked pharmacologically (IL-1Ra) or by genetic intervention (IL-1R^-/-), the inflammatory response is not sustained as reflected by lower cytokine release and microglia activation, thus ameliorating the cognitive dysfunction as reported here. Treatment with IL-1Ra provides a significant improvement in cognitive dysfunction, confirming the crucial role of IL-1β in memory processes and behavior. However, as IL-1Ra exerts protective effects also by reducing apoptosis and ischemia [31], the behavioral improvement could also reflect a wider action of this treatment not only on the immune system. Although there was a temporal correlation between microglia activation and late-release of HMGB-1, neither IL-1Ra nor IL-1R^-/- changed levels of this cytokine. This evidence supports the notion that blocking IL-1 is sufficient to reduce the microglia activation and ameliorate the memory abnormality. Other receptors may be involved in sustaining this inflammatory challenge; for example, HMGB-1 has been shown to activate TLRs and receptor for advanced glycation end-products and it has been reported as a key late pro-inflammatory mediator in sepsis, with considerable pathological potential [11, 32].

Some limitations of our study must be pointed out. As IL-1Ra is able to translocate directly into the brain [33], we are unable to discriminate whether peripheral cytokines and/or de-novo production in the CNS account for this cognitive dysfunction. Also, recently it has been shown that peripheral monocytes can enter the brain causing sickness behavior. This process strongly relies on TNFα signaling, especially in activating microglia and recruiting active monocytes into the CNS [34]. By targeting microglia we have selected a robust marker to correlate local inflammation with the functional behavioral abnormality. However, in this study we cannot determine the nature of the microgliosis, whether they are infiltrated macrophages that crossed the BBB or actual microglia. Although our primary aim was to characterize the importance of inflammatory mediators in cognitive dysfunction and by using LPS this can be more easily defined, a septic model using cecal ligation and perforation would have been more clinically applicable in reproducing the complexity of the polymicrobial septic pathology.