Introduction
IL-33 is a member of the interleukin-1 (IL-1) cytokine family that plays important roles in various disorders including allergy, autoimmune, or cardiovascular diseases through its receptor ST2 and co-receptor IL-1 accessory protein (IL-1RAcP) [
1]. Recently, IL-33 has also been involved in the pathogenesis of central nervous system (CNS) diseases such as neurodegenerative diseases, stroke, or infectious diseases. Broadly and highly expressed in the CNS in physiological conditions, IL-33 is described as a key regulator of neuroinflammation [
2‐
4].
In experimental autoimmune encephalomyelitis (EAE), a model of multiple sclerosis disease (MS), a systemic administration of recombinant IL-33, from the day of immunization until day 18, induces a protective effect [
5]. However, the intraperitoneal administration of anti-IL-33 neutralizing antibodies also delayed the onset and the severity of EAE [
6]. These apparently opposite findings highlight the dual function of IL-33. Moreover, this dual function of IL-33 has also been observed in Alzheimer’s disease (AD). IL-33 is highly expressed in the vicinity of amyloid plaques and in glial cells in brain sections from AD patients suggesting that a prolonged IL-33 production may induce inflammatory molecule release and contribute to the AD pathogenesis with neuronal damage [
7]. However, more recently, Saresella et al. [
8] demonstrated a decrease of IL-33 in the serum of AD patients as compared with healthy controls. These clinical data highlight a complex pro- and anti-inflammatory properties of IL-33 in AD patients acting both at the central and systemic level. IL-33 dual functions have also been observed in CNS infectious diseases. We previously reported the essential role of the IL-33 receptor ST2 in the pathogenesis of experimental cerebral malaria (ECM) caused by
Plasmodium berghei Anka (
PbA)-infection in mice. We showed that ST2-deficient mice were resistant to
PbA-induced neuropathology [
9] and demonstrated a deleterious role of CNS endogenous IL-33 in the neuropathogenesis associated with cognitive disorders [
10]. Surprisingly, IL-33 deficient mice were not resistant to ECM [
11] and IL-33 systemic administration improved antimalarial drug treatment of ECM via Treg cells [
12,
13]. Thus, IL-33 has dual effects on infection, inflammation, and diseases of the CNS [
1] raising the question of the cellular and immunomodulators involved.
Immunohistological analyses and IL-33/citrine reporter mice showed that astrocytes [
14,
15] and oligodendrocytes [
10,
16] are the main cellular sources of IL-33 within the CNS. Moreover, ST2 receptor is overexpressed by astrocytes and microglial cells under pathophysiological conditions [
14]. Microglia could be the first glial cells to respond to IL-33 stimulation through the ST2/IL-1RAcP receptor complex [
15]. We previously showed a deleterious effect of CNS endogenous IL-33 through the activation of microglia leading to IL-1β release in ECM [
10]. IL-33 is not only involved [
15,
16] but essential for the microglial activation [
17]. Given the importance of microglia in the neurotoxic or neuroprotective inflammatory responses, CNS IL-33 may be a key factor in the neuroinflammatory processes and associated with cognitive impairments.
In this study, we show that recombinant mouse IL-33 administration in the hippocampus led to microglial cell activation and increased IL-1 production associated with cognitive disturbance.
Materials and methods
Mice and ethics statement
C57BL/6 J (wild-type; WT) male mice under specific pathogen-free (SPF) condition at 8 weeks of age were purchased from Janvier Labs (Le Genest Saint Isle, France). Mice deficient for both IL-1α and IL-1β were bred in the Transgenose Institute animal facility (CNRS UPS44, Orleans, France). They were issued from an intercross between IL-1α ΚΟ and IL-1β ΚΟ mice [
18]. As they were backcrossed 10-fold on C57BL/6 J background, C57BL/6 J control was used. Mice were housed at four per propylene cage with woodchip bedding, and kept under controlled conditions of temperature (20–22 °C), humidity (50%), and bright cycle (12/12-h dark/light), with free access to chow pellets and water. The animals were previously habituated to our animal facility at 4 weeks and used in experimental settings at 8 weeks of age. All animal experimental protocols complied with the French ethical and animal experiments regulations (see Charte Nationale, Code Rural R 214–122, 214–124 and European Union Directive 86/609/EEC) and were approved by the “Ethics Committee for Animal Experimentation of CNRS Campus Orleans” (CCO), registered (N°3) by the French National Committee of Ethical Reflexion for Animal Experimentation, under N° CLE CCO 2015-1084 and by the French “Ministère de l’enseignement supérieur, de la recherche et de l’innovation”, under number APAFIS #19264.
Intrahippocampal microinjection
Mice divided into 4 groups, received intrahippocampal injections of either vehicle PBS containing 0.1% BSA as a carrier (PBS-BSA) or recombinant mouse (rm) IL-33 protein (R&D Systems, Abingdon, UK; 200 ng/μl in PBS-BSA), in the absence or in the presence of minocycline hydrochloride (MP Biomedicals, Illkirch, France) was administered daily (i.p, 50 mg/kg in NaCl 0.9%) during 10 days, including 7 days before surgery and 3 days post-surgery. Before intrahippocampal injections, mice anesthetized with ketamine/xylazine (100 μL/10 g i.p. of 29.4 mg/mL ketamine plus 3.05 mg/mL xylazine) were secured in the stereotaxic apparatus (KOPF instruments, Lidingö, Sweden). Burr holes were drilled bilaterally in the skull above the hippocampus at 2.0 mm posterior to bregma, and ±1.8 mm lateral to bregma. Then, mice received bilateral intrahippocampal injection of rmIL-33 protein at 400 ng in a total volume of 2 μL of PBS-BSA by side. Control animals received PBS-BSA vehicle. A 10-μL Hamilton syringe (Hamilton, Reno, NV, USA) controlled by a Stereotaxic Injector (KD Scientific, Holliston, USA) was used to inject the solution at a rate of 0.25 μL/min in the hippocampus at −1.80 mm to Bregma. After the surgery and to facilitate recovery, each mouse was placed alone per cage until the end of the experiments. Groups of sham animals were subjected to a similar hippocampal surgery, without PBS-BSA or rmIL-33 injection with or without minocycline pretreatment.
Spatial habituation test
Spatial habituation to a novel environment is commonly used for the exploration of non-associative learning and memory processes linked to hippocampal structures [
19‐
21]. As previously described [
22], to explore the learning component, 1 day after the surgical intervention, the animal was allowed to explore an open field (OF) (40 cm × 40 cm) for 10 min (trial session). After 24 h, the mouse was re-exposed for 10 min to the same OF (test session). During each session, the exploratory measures were quantified using the Ethovision tracking system (version 10, Noldus Technology, Wageningen, Netherlands). Locomotor activity was indexed by the distance traveled in the entire open-field arena. To explore intrasession habituation during the trial session, the distance traveled between the first and the last minute was compared. The intersession habituation was assessed by comparing the full distance traveled during both sessions. All sessions were performed at 10 lux to limit the anxiogenic component of the novel environment.
Real-time quantitative polymerase chain reaction (RT-qPCR)
At the indicated time, total RNA from the hippocampus was isolated using TRI-Reagent (Sigma-Aldrich, Saint-Quentin Fallavier, France) as previously described [
10] and reverse transcripted (Superscript III reverse transcriptase, Invitrogen, Carlsbad, CA). Quantitative real-time PCR reactions were performed using GoTaq qPCR-Master Mix (Promega, Charbonnières-les-Bains, France) and primers for
Nos2,
Il1b, Tnfa,
Ifng,
Arg1, Chil3, Il10, and Igf1 (Qiagen, Hilden, Germany). After normalization using
18S-RNA expression as a housekeeping gene, raw data were analyzed by the 2
ΔΔCt method [
23].
Tissue preparation and immunofluorescence
For immunostaining, mice were deeply anesthetized and transcardially perfused with ice-cold PBS followed by 4% paraformaldehyde (PFA). The brains were collected, post-fixed for 48 h in 4% PFA, and cryo-protected in a 30% sucrose solution for 1 week. Then, 14 μm brain cryo-sections mounted onto glass slides were incubated in citrate buffer (pH = 6) at 80 °C for 30 min, followed by incubation with blocking solution (TBS 1X; 1% BSA; 10% FCS; 0.3% Triton; 1% NaN3) during 45 min in a wet chamber at room temperature. After incubation overnight at 4 °C with anti-Iba-1 antibody (Abcam, Cambridge, England, ab5076; 1:500), the sections were washed in TBS and incubated with Alexa 488 secondary antibody (Abcam, ab150129, 1:1000) for 1 h. The slides were rinsed, then counter-stained with DAPI for 10 min, mounted with Fluoromount-G (SouthernBiotech, Birmingham, England), and dried before observation using ZEISS AXIOVERT 200 M/Apotome microscope (Zeiss, Oberkochen, Germany). Serial sections were collected at ×20 magnification to reconstruct each whole-hippocampal image software (ZEN2.1, Zeiss). The images were collected as Z-series of 18 optical slices to obtain a sufficient resolution to perform the morphological analysis of microglial cells. For each mouse, 3 representative stacks of images of the hippocampus were recorded. Positive cells for Iba-1 were counted (50–100 cells) and their morphology analyzed in each area e.g. the cornu ammonis (CA)1/CA2, CA3 and the dentate gyrus (DG). Image analysis and processing were performed with the software Image J -Fiji [
24] using the “concentric circles” plugin. For the Sholl analysis, the intersection number per radian was defined each 5 μm from the center of each cell (
n = 3 mice per treatment with 50-100 microglia analyzed per mouse). This analysis was performed by a blinded experimenter.
Fluorescence-activated cell sorting
The hippocampus of 3 mice perfused with phosphate-buffer saline (PBS) was pooled and the cellular suspensions were prepared using the Neural Tissue Dissociation Kit (Miltenyi Biotec, Paris France), according to the manufacturer’s instructions. Cells were stained with extracellular conjugated antibodies: Fixable Viability Dye (eBiosciences™, 65-0865-14, 1/800), anti-CD45 V450 (BD Horizon™, 560501, 1:100), anti-CD11b PerCP/Cy5 (BD Pharmingen™, 560993, 1:100) and blocked with non-conjugated anti-CD16/32 (BD Pharmingen™, 553142, 1:100) for 20 min at 4 °C. Then, the cells were washed before fixation. Intracellular IL-1β pro-form stained with PE-conjugated specific antibody (eBioscience™, 12-7114-80, 1:20) was visualized after cell permeabilization for 20 min at 4 °C with Cytofix/Cytoperm Plus Kit (BD Biosciences, Paris, France). This antibody recognizes only the pro-form of mouse IL-1β and does not detect the cleaved and secreted mature IL-1β form. Cells were then washed and re-suspended in lysing solution (BD FACS™ Lysing Solution) before the acquisition. Data were acquired with a flow cytometer (BD FACSCanto II) and analyzed with FlowJo v7.6.5 software (Tree Star, Ashland, OR). Very low SSC and very low FSC were excluded to strictly define the populations of interest. IL-1β pro-form staining was measured using geometric mean fluorescence intensity (GMFI). For the analysis, live single cells were pre-gated. Then, CD11b+/CD45low cells were gated as microglia, while CD11b+/CD45high cells or CD11b−/CD45high cells were gated as infiltrating macrophage or lymphocyte cells, respectively. FMO controls were also included to define populations of Fixable Viability Dye cells and CD45, CD11b, and IL-1β-expressing cells.
Statistical
Statistical significance was determined with GraphPad Prism v6 (GraphPad Software, La Jolla, CA). Standard errors of the mean are reported as SEM. To analyze non-parametric data, Mann-Whitney test for 2 series was used or Kruskal-Wallis followed by Dunn’s multiple comparison for more series. P values ≤ 0.05 were considered statistically significant.
Discussion
The implications of IL-33 has been described in many neuropathologies [
1], not only as protective [
14,
16] but also as disruptor [
15,
17] of neuronal homeostasis. IL-33 exerts pleiotropic effects on the immune system, both on type 2 and type 1 immune responses, in the periphery but also at the CNS level. Despite the presence of IL-33 in a healthy brain [
4] and in CNS pathologies [
1], the multifold functions of IL-33 in CNS remain unclear. To elucidate the role of endogenous IL-33 in the CNS, the present study explored the consequences of intrahippocampal injection of recombinant IL-33 on cognitive function and neuroinflammatory processes.
Using spatial habituation tasks in an open field, allowing to address hippocampal non-associative learning and memory processes [
19‐
21,
26], we show that the habituation to a novel environment was intact in IL-33 hippocampal treated mice 1-day post-surgery. These results indicate that neither the micro-lesion induced by the injection nor the IL-33 treatment had a neurological impact on learning at this stage. However, 48 h after intrahippocampal injection, IL-33-treated mice displayed a complete impairment of spatial memory retrieval. Unlike control mice, they were not able to recognize the previously explored environment, indicating that long-term habituation was significantly affected after rmIL-33 administration. These results suggest that a massive IL-33 release might disturb neuronal function and affect the memory retrieval process. IL-33 has been previously involved in cognitive defects observed in neuropathological conditions such as reflected in Alzheimer’s disease, multiple sclerosis, and experimental cerebral malaria [
1,
10,
22]. Our data further show that injecting recombinant IL-33 directly in the hippocampus could mimic an acute exposure of IL-33 and its effects on cognitive processes.
To explore the link between the cognitive defect induced by IL-33 and the neuroinflammatory response, mice were pre-treated with minocycline. This antibiotic is able to cross the blood-brain barrier and exhibits anti-inflammatory properties preventing memory deficits in several neuropathologies [
25]. In the present study, chronic administration of minocycline alone before intrahippocampal injections in control mice did not affect learning and spatial memory processes. However, our data also reveal that pre-treatment with minocycline seems to prevent the spatial memory retrieval impairment induced by IL-33 administration. This rescue of the IL-33-induced phenotype suggests that the cognitive impairments induced by IL-33 involved a neuroinflammatory process.
Previous studies demonstrated that IL-33/ST2 pathway modulated the production of cytokines and chemokines in neuropathological conditions [
1,
3,
17]. We assessed the direct effects of IL-33 role on the inflammatory context by gene expression analysis. We quantified mRNA expression levels in the hippocampus of molecular markers usually used to define pro-inflammatory or regulatory immune response [
27].
Nos2,
Il1b,
Tnfa, and
Ifng are mediators of pro-inflammatory responses whereas
Arg1,
Chil3,
Il10, and
Igf1 are associated with immunoregulatory mechanisms. In control mice, we observed a transient inflammatory response induced by the injection at 24 h and resolving at 48 h post-injection. This transient response to a slight trauma is correlated with the ability of the organism to return to a homeostasis state without adverse effects on behavior [
28]. However, at 48 h, the intrahippocampal IL-33 injection induced a neuroinflammatory environment with overexpression of pro-inflammatory and immunoregulatory markers mRNA. Minocycline administration reduced this inflammatory context in terms of
Il1b and
Ifng expression at 48 h, contributing to the resolution of inflammation. These results suggest that exogenous IL-33 induces a neuroinflammatory phenotype associated with long-term habituation disturbance.
To explore the cellular process involved in IL-33-induced immune response, we focused on microglia, the first active immune barrier in the CNS strongly expressing IL-33 receptor ST2 [
14]. We investigated the hippocampal microglia reaction by immunochemistry using Iba-1 staining. Indeed, in response to a neuroinflammatory context induced by LPS administration, resident microglia alter their shape in a specific way as compared with infiltrated peripheral cells with a rounder morphology [
29,
30]. Sholl analysis on Iba1 immunofluorescent staining revealed a significant increase of proximal intersections per radius in the CA1/CA2, CA3, and DG regions 48 h after IL-33 treatment. This reactive morphology associated with an increase of microglial cell number demonstrated maintenance of their activated state. Minocycline administration through its anti-inflammatory activity attenuated the microglia activation of IL-33 treated mice, in line with previous reports in cognitive disorders [
31,
32]. This result suggests that the deleterious function of IL-33 pathway on spatial memory retrieval processes requires microglia activation, especially in the hippocampal formation. Indeed, in healthy conditions, microglia regulate neuronal activity, synaptic plasticity, and adult neurogenesis required for learning and memory. In many neuropathologies, the microglia reactivity state has been characterized based on morphological modifications and the release of cytokines, chemokines, and growth factors, modulating neuronal and synaptic functions. This activated phenotype should be beneficial and associated with inflammatory changes to combat the injury and return to a homeostatic state. However, these defense processes could be over-stimulated and cause significant damage to behavior [
33], as we demonstrated hereafter intrahippocampal IL-33 injection.
To confirm the IL-33-induced microglial reactivity, cells from the hippocampi of IL-33-treated mice were analyzed by flow cytometry at 48 h post-surgery. Although neither macrophage nor lymphocyte recruitment was observed, microglia number was significantly increased in IL-33-treated mice, confirming our immunohistological data. This analysis revealed also an overexpression of the IL-1β immature form in hippocampal microglia 48 h after IL-33 injection. Thus, in vivo, IL-33 treatment promotes IL-1β microglia production, as previously demonstrated in vitro [
10], indicating that microglia contribute to the pro-inflammatory response. We then hypothesized that the cognitive impairment induced by exogenous IL-33 may be mediated in part by microglia derived IL-1. To test this point, we performed IL-33 intrahippocampal microinjections in IL-1αβ deficient mice. The absence of IL-1 cytokines prevented spatial memory retrieval impairment induced by IL-33 administration even if neuroinflammatory markers, except IL-1β and IL-1α, were upregulated. Thus, IL-1β-producing microglia are required for IL-33 neurotoxic effects on cognitive impairment. In our experimental conditions, IL-1 contribution to these cognitive defects impairment could involve its non-immunological activities. Indeed, this cytokine has been described as critical for learning and memory in a dose-dependent manner [
34]. Here, we verified that IL-1αβ deficient mice behave as WT mice in terms of habituation to spatial novelty in our experimental conditions. Prolonged up-regulation of pro-inflammatory cytokines, especially IL-1β, has been associated with a decrease in synaptic plasticity, as well as a deficit in spatial learning [
35,
36]. This IL-1β disruptive effect on cognitive functions could involve the inhibition of long-term potentiation generation at the neuronal level and/or defect of neurotrophic factors production [
34]. All these parameters should be further investigated in our futures studies.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.