Findings

Interleukin-1β (IL-1β) is a pro-inflammatory cytokine that is involved in the pathogenesis of a number of neurological disorders, possibly as a modulator of glutamatergic response [1]. This suggestion arises from the observation that IL-1β is often over-produced in injured tissues in which there are high levels of glutamate [2–4], and this over-production has been related to the exacerbation of glutamate-driven pathological conditions [4–6]. Various mechanisms have been identified that may explain the convergence between the IL-1β and glutamatergic systems [1], including hyperactivation of the NMDA receptor (NMDAR). IL-1β increases the activity of hippocampal neuronal NMDARs by phosphorylating the GluN2B subunit and thus enhancing NMDA-induced neuronal death [7]. The same mechanism is recruited in neurons as a consequence of the IL-1β released from glia by the HIV-virus glycoprotein gp120 [8], and underlies the pro-convulsive effect of IL-1β [9]. Whatever mechanism may be recruited by IL-1β, the involvement of IL-1RI is suggested by the uncontested neuroprotective effect of the IL-1 receptor antagonist (IL-1ra) [4, 10].

The binding of IL-1β to IL-1RI in the immune system leads to its association with the IL-1R accessory protein (IL-1RAcP) [11] and the myeloid differentiation primary response protein 88 (MyD88) [12] to form the core of the IL-1β/IL-1R signalling complex. However, little information is currently available concerning the molecular composition of the members of the IL-1R complex, or their subcellular distribution and functional cross-talk with NMDARs in neuronal cells [13–15]. This is a major gap in our knowledge of the pathological mechanisms involving IL-1β/IL-1RI in neurons that may be relevant to therapeutic interventions in the central nervous system (CNS).

The distribution of IL-1RI, IL-1RAcP and MyD88, together with the pre- and post-synaptic markers synaptophysin and PSD-95, was investigated in different subcellular compartments purified from adult rat hippocampi by means of western blotting [16], and by means of confocal microscopy of primary hippocampal neurons.

Subcellular fractionation showed that IL-1RI, MyD88 and IL-1RAcP were present in all of the tested fractions but, although IL-1RI and MyD88 were particularly enriched in the postsynaptic density (PSD) fraction (Figure 1A, left panel), together with PSD-95 and the GluN1 subunit of the NMDA receptor (Figure 1A, right panel), only traces of IL-1RAcP were present in the postsynaptic Triton-insoluble fraction (TIF) and PSD (Figure 1A, left panel). Confocal imaging showed that IL-1RI is distributed along dendrites and enriched in the post-synaptic compartment, as shown by the high degree of co-localisation with PSD-95 (34.3% ± 3.7%; Figure 1B, left panels). MyD88 was uniformly distributed along the neurons and moderately co-localised with PSD-95 (15.6% ± 2.8%; Figure 1B, right panels). IL-1RAcP labelling was intense and diffuse in the somatic cytoplasm of cultured neurons, and low and diffuse along the dendrites, and hardly co-localised with PSD-95 (4.1% ± 1.9%; Figure 1B, central panels). Overall, these data suggest that there is a different subcellular distribution of the members of the IL-1R complex protein in neurons, with IL-1RI (and, to a lesser extent, MyD88) being enriched at the post-synaptic sites.

Figure 1
figure 1

Characterisation of IL-1RI, IL-1RAcP and MyD88 subcellular distributions in rat hippocampi and primary hippocampal neurons. A: IL-1RI, IL-1RAcP and MyD88, together with markers of the pre-synaptic compartment (synaptophysin) and post-synaptic side (PSD-95, GluN1) were analysed in various rat hippocampus subcellular compartments by means of western blot. H = homogenate; S1 = low-speed supernatant; P1 = nuclei-associated membranes; S2 = high-speed supernatant; P2 = crude membrane fraction; Syn = synaptosomes; TIF = Triton-insoluble postsynaptic fraction; PSD = post-synaptic density. B: DIV14 hippocampal neurons were immunolabelled for IL-1RI, IL-1RAcP and MyD88 (upper panels), and PSD-95 as a post-synaptic marker (middle panels). The bottom panels show the merged images. Scale bar: 10 μM. High-magnification images are shown at the top of each panel. The antibody specific for IL-1RI was from Santa Cruz Biotechnology, Inc (Santa Cruz, CA) (M20) and its specificity has been tested by pre-absorption with the blocking peptide (data not shown).

Full size image

Co-immunoprecipitation experiments involving the components of the IL-1R complex and the AMPA and NMDA receptor subunits were performed to assess the localisation and interactions of IL-1RI within distinct subdomains of the PSD structure [17]. Protein homogenates (200 μg) from rat hippocampi were immunoprecipitated [18] with antibodies specific for IL-1RI, for the GluA1 subunit of AMPA receptors, or for the GluN2B subunit of the NMDA receptor. Each sample was then evaluated for the presence of: i) IL-1RI, IL-1RAcP and MyD88; ii) the GluN2B subunit: and iii) PSD-95. Figure 2A shows that, in hippocampal lysates, IL-1RI not only co-precipitated with IL-1RAcP and MyD88, but also with GluN2B; the absence of any PSD-95 signal in the IL-1RI co-immunoprecipitates excludes the possibility that, under our experimental conditions, the co-precipitation of GluN2B reflected nonspecific immunoprecipitation of insoluble synaptic proteins. In line with this, GluN2B co-precipitated with IL-1RI, thus confirming the association between these components (Figure 2A, right lane). Finally, none of the members of the IL-1R complex was detectable in the immunocomplex of the GluA1 subunit of AMPA receptors, which suggests a specific interaction between the GluN2B subunit of the NMDA receptor and IL-1R complexes (Figure 2A).

Figure 2
figure 2

Interaction between IL-1RI and the GluN2B subunit of NMDA receptors. A: Total homogenate was immunoprecipitated (i.p.) with antibodies against IL-1RI, GluA1 or GluN2B, and the presence of GluN2B, IL-1RI, PSD-95, IL-1RAcP and MyD88 in the immunocomplex was evaluated by means of western blot. IL-1RI, IL-1RAcP and MyD88 co-precipitated with GluN2B but not with GluA1. (*) Nonspecific bands were detected in the No IgG lane. B: GST-IL-1R(CD) and GST-PSD-95(PDZ1-2) fusion proteins, and GST alone were incubated in a pull-down assay with total homogenate from rat hippocampus. The western blot analysis was performed using the GluN2B antibody. C: Hippocampal cultures were exposed in the absence or the presence of IL-1β (30 minutes, 0.05 ng/ml) or NMDA (10 minutes, 50 μM). Neuronal lystes were immunoprecipitated with anti-IL-1RI, and the presence of GluN2B and IL-1RI in the immunocomplex was evaluated by means of western blot. Treatment with NMDA but not with IL-1β led to a significant increase in the IL-1β/GluN2B complex (p < 0.05 NMDA vs control).

Full size image

The association between IL-1RI and GluN2B was confirmed by a pull-down assay based on a fusion protein of the cytoplasmic domain of IL-1RI with GST (GST-IL-1Rcd) (Figure 2B), which contained the C-terminal 369-569 aa domain of IL-1RI. As a positive control, we used a GST-PSD-95 (PDZ1-2) fusion protein that has been previously shown to bind the GluN2B subunit of NMDA receptors [18]. Lysates from rat hippocampal neurons were applied to affinity beads and extensively washed, after which the bound material was resolved by SDS-PAGE and underwent immunoblotting analysis using an antibody raised against GluN2B. Figure 2B shows that both IL-1Rcd and PSD-95 (PDZ1-2) associated with the GluN2B subunit, thus confirming a specific association between IL-1RI and GluN2B.

As it is well known that the synaptic localisation of receptors and ion channels, together with their protein-protein interactions, are modulated in response to various stimuli, and that they undergo dynamic changes under physiological and pathological conditions [19, 20], we investigated the possibility that IL-1RI distribution and interaction with GluN2B may be dynamically modulated. Given the relationship between the IL-1β receptor complex and NMDAR, we treated primary hippocampal neurons with IL-1β, 0.05 ng/ml, for 30 min (a concentration that also enhances NMDAR activity) [7] or NMDA, 50 μM, in ACSF buffer [7]: the NMDA was applied to the neurons for 10 min, after which the cells were washed and incubated for a further 20 min in ACSF buffer. We first tested whether IL1-β and/or NMDA modulated the interaction between IL-1RI and the GluN2B subunit of the NMDA receptor (Figure 2C). IL-1RI was immunoprecipitated from total lysates of primary hippocampal neurons treated or not with NMDA, 50 μM, or IL-1β, 0.05 ng/ml, and assayed for GluN2B by means of western blotting (Figure 2C). The results show that only NMDA significantly increased the interaction between IL-1RI and GluN2B (Figure 2C; p < 0.05 NMDA vs control).

We then evaluated whether the members of the IL-1β receptor complex could be re-distributed in different neuronal compartments after stimulation with IL-1β or NMDA. Both NMDA and IL-1β significantly increased the amount of IL-1RI in the postsynaptic TIF fraction (Figure 3A; p < 0.05, IL-1β or NMDA vs control). The treatments did not affect the synaptic distribution of either IL-1RAcP or MyD88. These results were confirmed by confocal microscopy (Figure 3C), which showed an increase in the co-localisation of IL-1RI with PSD-95 as quantified in the graph (Figure 3C).

Figure 3
figure 3

Effect of NMDA and IL-1β on IL-1RI subcellular localisation. A: Western blot analysis of the TIF fraction obtained from control, IL-1β-treated (0.05 ng/ml) and NMDA-treated hippocampal cultures (50 μM). The same amount of proteins was loaded in each lane. IL-1β increases IL-1RI localization in the Triton-insoluble fraction (TIF) (*p <0.05) leaving unaffected IL-1RAcP and MyD88 levels. Values are means ± S.E of 4 independent experiments. B: Western blot of IL-1RI from control, IL-1β-treated (0.05 ng/ml) and NMDA-treated (50 μM) hippocampal cultures exposed (+BS3 lanes) or not (-BS3 lanes) to the cross-linking agent BS3. IL-1RI high-molecular-weight complexes that didn't enter the gel are not shown. C: Hippocampal neurons were either left untreated (control) or treated with IL-1β (0.05 ng/ml, 30 minutes) or NMDA (50 μM) fixed, and immunolabeled for IL-1RI (green) and PSD-95 (red) as a postsynaptic marker. Data are expressed as percentage of IL-1RI colocalization with PSD-95 (AIM4.2 software, Zeiss). White arrows indicate PSD-95 positive clusters in the merge panel. Scale bar: 5 μM.

Full size image

The increase in IL-1RI receptors at the postsynaptic site may be due to new synthesis and delivery of receptors from the endoplasmic reticulum, or to lateral diffusion from adjacent compartments [19, 21], and this was addressed by carrying out surface expression assays using the non-cleavable, membrane-impermeable crosslinking agent BS3 [22]. Primary hippocampal neurons were treated with IL-1β, 0.05 ng/ml, or NMDA, 50 μM, and then exposed to BS3, lysed and blotted for IL-1RI. The intracellular amount of IL-1RI was reduced by NMDA but not by IL-1β (p < 0.05, NMDA vs control; Figure 3B). The reduction in intracellular IL-1RI after NMDA exposure, together with its increase in the synaptic fraction, suggests that NMDAR activation favours the membrane insertion of new IL-1RI. Alternatively, the increase in IL-1RI in the synaptic membrane may be attributable to stabilisation of the complex with NMDAR (within the core of the PSD), which could prevent lateral movement and/or endocytosis. In either case, a new pool of receptors would be made available. On the contrary, IL-1β possibly enriches IL-1RI at post-synaptic sites, promoting its lateral translocation (i.e. membrane diffusion) from extra-synaptic sites; however, this probably does not occur within the core microdomain of the PSD, as suggested by the unchanged levels of IL-1RI associated with the NMDAR complex.

In conclusion, ours are the first findings showing a molecular interaction between IL-1RI and the GluN2B subunit of NMDAR, and suggest a new molecular mechanism by means of which IL-1β and NMDA may dynamically regulate IL-1RI at post-synaptic sites. Furthermore, NMDA-dependent activation increases the amount of IL-1RI inserted into the membrane capable of interacting with released IL-1β. This suggests a new molecular mechanism by means of which IL-1β may contribute to excitotoxicity, thus opening up new possibilities for targeted inhibition strategies that can be used in IL-1β/glutamate-driven CNS diseases.