Prevention of the β-amyloid peptide-induced inflammatory process by inhibition of double-stranded RNA-dependent protein kinase in primary murine mixed co-cultures

Sodium fluoride (NaF), phenylmethylsulfonyl fluoride (PMSF), protease and phosphatase inhibitor cocktails, dithiothreitol (DTT), 0.01% poly-L-lysine solution, Percoll^®, sterile filtered dimethylsulfoxide Hybri-Max^® (DMSO), Triton X-100, paraformaldehyde (PFA), annexinV-fluorescein isothiocyanate (FITC) apoptosis detection kit and all reagent-grade chemicals for buffers were purchased from Sigma (St Quentin Fallavier, France); DMEM (1 g/L), MEM and Neurobasal media, B-27 Supplement, 200 mM L-glutamine, 5,000 units of penicillin (base) and 5,000 μg of streptomycin (base)/mL (PS) mixture, 0.5 g/L Trypsin/0.2 g/L EDTA 4Na, Fetal Bovine Serum, Certified (FBS), Horse Serum, NuPAGE^® Novex^® Bis-Tris Mini Gels, NuPAGE^® LDS 4X LDS Sample Buffer, NuPAGE^® Sample Reducing Agent (10X), NuPAGE^® MES SDS Running Buffer and NuPAGE^® Antioxidant, iBlot^® Gel Transfer Device (EU), the Prolong Gold antifade reagent with 4',6-diamidino-2-phenylindole (DAPI) and the Zenon mouse IgG labelling kit from Gibco-Invitrogen (Fisher Bioblock Scientific distributor, Illkirch, France); the imidazolo-oxindole compound C16 from Merck Chemicals Calbiochem^® (Nottingham, UK). For western blot, primary antibodies and secondary anti-rabbit IgG antibody conjugated with horseradish peroxydase were purchased from Cell Signalling (Ozyme, St Quentin Yvelines, France) excepted anti-P_T451-PKR from Eurogentec (Seraing, Belgium), anti-β tubulin and anti-β actin from Sigma (St Quentin Fallavier, France), anti-amyloid peptide (clone WO2, recognizes amino acids residues 4-10 of Aβ) from Millipore (St Quentin-Yvelines, France), peroxidase-conjugated anti-mouse IgG from Amersham Biosciences (Orsay, France). For immunofluorescence, anti-glial fibrillary acidic protein (GFAP) antibodies were purchased from Cell Signalling (Ozyme, St Quentin Yvelines, France), microtubule associated protein 2 (MAP2) from Abcam (Paris, France), macrosialin or murine homologue of the human CD68 from AbD Serotec (Düsseldorf, Germany), anti-P_T451-PKR from Biosource (Nivelles, Belgium), secondary antibodies from DakoCytomation, (Trappes, France) and IgG- and protease-free bovine serum albumin (BSA) from Jackson ImmunoResearch Europe Ltd (Interchim distributor, Montluçon, France).

First, primary glial cultures were prepared from C57BL/6J mouse embryos of 18 days. Brains were quickly removed, and cerebral cortico-hippocampal regions were dissected in ice-cold and sterile 1X PBS (154 mM NaCl, 1.54 mM KH₂PO₄, 2.7 mM Na₂HPO₄ 7H₂O, pH 7.20 ± 0.05) containing 18 mM glucose and 1% PS as previously described [38]. Cells were then dissociated mechanically using a pipette into DMEM/1% PS, transferred into tubes containing FBS at the bottom (1 mL/30 mL cell suspension) and centrifuged at 300 × g for 10 min at 4°C. The cell pellet was suspended into DMEM/1% PS and centrifuged again. This step was repeated once. After the centrifugation, cells were suspended into DMEM/10% FBS/1% PS, seeded at a density of 4 × 10⁵ cells/mL in Nunc EasYFlask™ (75 cm²) coated with 0.001% poly-L-lysine and then incubated at 37°C in a humidified 5% CO₂ atmosphere. Medium was replaced every five days. These cells were cultured until day 14, the day of microglia purification.

Second, primary cultures with neurons and astrocytes were prepared from cortex and hippocampus of C57BL/6J mouse embryos of 18 days as above. Cells were suspended in MEM/Neurobasal (1:1) supplied with 18 mM glucose, B-27 Supplement, 1% glutamine, 2.5% FBS, 2.5% horse serum and 1% PS, and seeded in 6-well plates (10⁶ cells per well) coated with 0.001% poly-L-lysine. Cultures were then maintained at 37°C in a humidified 5% CO₂ atmosphere. At day 5, neurons and astrocytes were cultured with microglia purified from the primary culture described above.

Third, microglia were purified from glial cultures on day 14 as previously described with some modifications [39]. Briefly, confluent glial cultures were dissociated with trypsin/EDTA and cell suspensions were suspended in 1 mL of 70% isotonic Percoll and transferred into a 5 ml glass tube. Two mL of 50% isotonic Percoll were gently layered on top of the 70% layer and then 1 mL of 1X PBS layered on top of 50% isotonic Percoll layer. Tubes were centrifuged at 1200 × g for 45 min at room temperature (RT) with a program including minimum acceleration and brake in a swinging bucket rotor.

Purified microglia occupied the interface between 70 and 50% isotonic Percoll. The top interface between 1X PBS and 50% isotonic Percoll containing all other central nervous system (CNS) elements was carefully removed and microglia layer was transferred into a new tube and washed twice by adding 1 mL PBS and centrifuged at 500 × g for 5 min at RT. Cells were counted and seeded at the density of 150,000 cells per well into 6-well plates containing the primary culture of neurons/astrocytes to 5 days old in order to obtain a density of microglia close to that already described. Indeed, the density of microglia in the CNS of the normal adult mouse brain is variable depending on the brain region and represents 5% in the cerebral cortex, according to Lawson et al. [40]. The mixed murine co-cultures with neurons, astrocytes and microglia were then used three days later for experiments and a fourfold confocal staining with cell and nucleus markers (DAPI, MAP-2, GFAP, CD68 for nuclei, neurons, astrocytes and microglia, respectively) was investigated in cells seeded on poly-L-lysine-coated glass coverslips to quantify neurons, astrocytes and microglia. In additional file 1, figure S1, we show that neurons, astrocytes and microglia represent about 36, 57 and 6% of total cells, respectively, i.e. close to what is physiologically observed in the cortex.

Co-cultures were treated with either C16 (specific inhibitor of PKR) at different concentrations (210 nM (IC50) and 1 μM) or DMSO (vehicle of C16) at less than 1%, in serum-free MEM:Neurobasal (1:1)/1% glutamine/1% PS medium 1 hour before 20 μM Aβ42 (or exactly 11 nmol in 550 μL of medium in each well receiving Aβ42) for 72 h at 37°C. Aβ42 was previously incubated 48 h at 37°C for aggregation as recommended by the Merck Chemical supplier [41]. The concentration of Aβ42 was chosen based on previous work in primary cultures [38, 42]. After treatment, media were conserved in order to analyse Aβ42 monomers and oligomers by immunoblotting and fibrillar form of Aβ42 by scanning electron microscopy in our experimental conditions (see the additional file 2, Figure S2). Results show the presence of a mix composed with monomers, oligomers (8 and 12 kDa) and a dense network of fibrils. As the specific toxicity of these different states of Aβ is not clearly demonstrated, we decided to incubate cells with this whole mixture.

Nuclear extracts were prepared as previously described [43]. Firstly, the cytoplasmic fraction was isolated and discarded, and the nuclear pellet was then lysed in nuclear lysis buffer (20 mM Hepes pH 7.9, 400 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 0.5 mM PMSF, 1% of protease and 1% phosphatase inhibitor cocktails) during 2 h at 4°C. Then, vials were centrifuged at 1,600 × g for 5 min at 4°C and the supernatant was isolated. The quantity of total protein was measured with a Biorad protein assay kit.

Commercially available ELISA kits were used for assessing TNFα (sensitivity: 2 pg/mL), IL-1β (sensitivity: 15 pg/mL) and IL-6 (sensitivity: 2 pg/mL) according to the manufacturers' instructions (BioLegend, Ozyme, St Quentin Yvelines, France). The range of analysis was between 7.8-6000 pg/mL. Cell lysates were diluted (1:2) with the assay diluents and all steps were performed at RT. The enzymatic reaction was stopped after 15 min incubation with tetramethylbenzidine (TMB) substrate by adding 2N H₂SO₄ and the optical density (OD) was read at 450 nm within 30 min, using the Multiskan^® spectrum spectrophotometer. The cytokine levels were then calculated by plotting the OD of each sample against the standard curve. The intra- and inter-assay reproducibility was > 90%. OD values obtained for duplicates that differed from the mean by greater than 10% were not considered for further analysis. For convenience all results are expressed in pg/mL and in pg/mg protein for culture medium and cell lysates, respectively.

Samples (30 μg proteins of cell lysates or nuclear extracts) were prepared for electrophoresis by adding NuPAGE^® LDS 4X LDS sample buffer and NuPAGE^® Sample Reducing Agent (10X). Samples were then heated up 100°C for 5 min and loaded into NuPAGE^® Novex^® Bis-Tris Mini Gels, and run at 200 V for 35 min in NuPAGE^® MES SDS running buffer containing 0.5% NuPAGE^® antioxidant. Gels were transferred to nitrocellulose membranes using the iBlot^® Dry blotting system set to program 20V for 7 min. Membranes were washed for 10 min in Tris-buffered saline/Tween (TBST: 20 mM Tris-HCl, 150 mM NaCl, pH 7.5, 0.05% Tween 20) and blocked 2 h in TBST containing 5% non fat milk or 5% bovine serum albumin (BSA).

Blots were incubated with primary antibody in blocking buffer overnight at 4°C. Antibodies used were rabbit anti-P_T451-PKR (1:100), mouse anti-P_S32/36-IκB (1:500), rabbit anti-IκB (1:500), rabbit anti-P_S536-NF-κBp65 (1:500), rabbit anti-NF-κBp65 (1:500) and rabbit anti-caspase3 8G10 (1:500) which detects endogenous levels of full-length and large fragments of caspase-3 resulting from cleavage at aspartic acid 175. Membranes were washed 2 times with TBST and then incubated with the peroxidase-conjugated secondary antibody either anti-rabbit or anti-mouse IgG (1:1000) according to the origin of primary antibody during 1 hour at RT. Membranes were washed again and exposed to the chemiluminescence ECL luminol plus western blotting system (Amersham Biosciences, Orsay, France) followed by signal capture with the Gbox system (GeneSnap software, Syngene, Ozyme distributor). After 2 washes in TBST, membranes were probed with mouse antibody against tubulin (1:10000) or actin (1:100000) overnight at 4°C. They were then washed with TBST, incubated with peroxidase-conjugated secondary antibody anti-mouse (1:1000) for 1 h, exposed to the chemiluminescence ECL luminol western blotting system and signals were captured. Automatic image analysis software was supplied with Gene Tools (Syngene, Ozyme distributor). Ratios protein/tubulin or actin were calculated and are shown in the corresponding figures.

After treatment, cells on coverslips were washed once with PBS and fixed with 4% PFA for 15 min at RT. After three washes with PBS, the permeabilizing and blocking PBS buffer (137 mM NaCl, 2.7 mM KCl, 1.7 mM KH₂PO₄, 10.14 mM Na₂HPO₄, pH 7.4 containing 0.3% triton X-100 and 5% of IgG- and protease-free BSA) was added during 1 h at RT.

Staining of neurons, astrocytes and microglia was performed by incubating coverslips overnight at 4°C with a mix containing rabbit anti-MAP2 (1:50), mouse anti-GFAP (1:100) and rat anti-CD68 (1:25) in PBS containing 0.3% triton X-100 and 1% of BSA. Cells were then rinsed twice with PBS before 1 h incubation at RT with the mix containing secondary antibodies: swine anti-rabbit FITC (1:20), goat anti-mouse AlexaFluor 647 (1:25) and goat anti-rat R-Phycoerythrin (RPE) (1:25) diluted in PBS/0.3% triton X-100/1%BSA. Finally, cells were washed twice in PBS and twice in distilled water before using the Prolong Gold antifade reagent with DAPI.

Staining of P_T451-PKR and cell marker (MAP2, GFAP or CD68) was performed in PBS/0.3% triton X-100/1% BSA overnight at 4°C by using rabbit anti-P_T451-PKR (1:25) with chicken anti-MAP2 (1:100) and mouse anti-GFAP (1:100). After incubation, cells were washed twice with PBS before incubated with swine anti-rabbit (1:30) conjugated with tetramethylrhodamine isomer R (TRITC), goat anti-chicken FITC (1:50) and goat anti-mouse AlexaFluor 647 (1:25) for 1 h at RT. A sequential labelling for P_T451-PKR and CD68 was performed. Firstly, cells were incubated with anti-CD68 antibodies overnight at 4°C, washed and incubated with goat anti-rat RPE. Secondly, cells were incubated with anti-P_T451-PKR overnight at 4°C, washed and incubated with swine anti-rabbit FITC (1:20). Finally, coverslips were washed and mounted as described above.

Annexin V-FITC labels phosphatidylserine sites on the membrane surface. The kit used also includes propidium iodide (PI) to label cellular DNA in necrotic cells where the cell membrane has been totally compromised. For this labelling, cells were incubated with annexinV-FITC (1:50) and PI (1:100) in 1X binding buffer for 10 min at RT. Cells were then fixed with 4% PFA for 15 min at RT. After three washes with PBS, cells were incubated in the permeabilizing and blocking PBS buffer for 1 h at RT and with anti-MAP2 and anti-GFAP or with anti-CD68 in the same experimental conditions as described for the previous staining of P_T451-PKR.

Multiply labelled samples were examined with a spectral confocal FV-1000 station installed on an inverted microscope IX-81 (Olympus, Tokyo, Japan) with Olympus UplanSapo x60 water, 1.2 NA, objective lens. Fluorescence signal collection, image construction, and scaling were performed using the control software (Fluoview FV-AS10, Olympus). Multiple fluorescence signals were acquired sequentially to avoid cross-talk between image channels. Fluorophores were excited with 405 nm line of a diode (for DAPI), 488 nm line of an argon laser (for Alexa 488 or FITC), 543 nm line of an HeNe laser (for TRITC and RPE) and the 633 nm line of an HeNe laser (for AlexaFluor 647). Emitted fluorescence was detected through spectral detection channels between 425-475 nm and 500-530 nm, for blue and green fluorescence, respectively and through a 560 nm and a 650 nm long pass filters for red and far red fluorescence, respectively. The images then were merged as an RGB image.

Cells were seeded on poly-L-lysine-coated glass coverslips at the same density described above. Treated primary co-cultures were rinsed briefly with PBS and fixed for 2 h at 4°C with 100 μM phosphate buffer (pH 7,4) containing 3% glutaraldehyde. After several rinses, they were post-fixed 1 h in 1% osmium tetroxide. Cells were washed again and dehydrated in acetone. Thereafter, samples were critical point-dried with a BAL-TEC CPD 030 using acetone and liquid carbon dioxide as the transition fluid. The dried specimens were coated with gold (25-35 nm thickness) using a sputtering device (BAL-TEC LCD 005). The samples were examined and photographed with a JEOL JSM-840 electron microscope.

Results are expressed as means ± SEM. Data for multiple variable comparisons were analysed by a one-way ANOVA followed by a Newman-Keuls' test as a post hoc test according to the statistical program GraphPad Instat (GraphPad Software, San Diego, CA, USA). The level of significance was p < 0.05.

Compound C16 is one of the most specific valuable imidazolo-oxindole inhibitors of PKR autophosphorylation that also rescues a PKR-induced translational block in a rabbit reticulocyte lysate system at micromolar concentrations [37]. Furthermore, previous data have shown that 1 μM C16 markedly reduces levels of P_T451-PKR and caspase-3 activity in Aβ42-treated SH-SY5Y cells [27, 43]. The T451 phosphorylated site in the PKR activation loop is required in vitro and in vivo for high-level kinase activity [44].

We first evaluated toxicity of compound C16 at 210 nM (IC50) and 1 μM compared to its DMSO vehicle ( < 1%). By using scanning electron microscopy, we showed that the majority of cultured cells were neurons and astrocytic glial cells (Figure 1). Amongst these were some round cellular elements ranging from 10 to 15 μm in diameter which were identified as microglia cells. In experimental conditions with DMSO or 210 nM C16, microglia looked like spherical smooth cells in contact with neurons and the astroglial layer. No reactive microglia were observed in these control conditions. However, 1 μM C16 greatly affected the integrity of cells in co-cultures, with neuronal death, disruption of axonal network and activated astrocytes. The microglia looked like macrophages (Figure 1). Based on these observations, further experiments were performed with the effective concentration 210 nM, corresponding to IC50 of compound C16 [37].

At its IC50, C16 significantly reduced by 33% the prominent activation of PKR induced by 20 μM Aβ42 over 72 h in the co-cultures as shown by immunoblotting from nuclear extracts (Figure 2). Confocal staining of P_T451-PKR confirmed the activation of PKR under Aβ42 exposure compared to DMSO-treated cells (Figure 3C, G and 3K, O versus 3A, E and 3I, M, respectively). Moreover, co-staining with the neuronal marker MAP2 indicated that P_T451-PKR was present in neurons, with intense perinuclear, nuclear and axonal staining, compared to DMSO-treated cells (Figure 3: C, G compared to A, E). Treatment with C16 decreased perinuclear and nuclear staining induced by Aβ42, but some axons remained stained (Figure 3D, H). The co-cultures incubated with compound C16 alone resembled those incubated with DMSO alone (Figure 3B, F, J, N versus 3A, E, I, M, respectively). In astrocytes labeled by antibodies against GFAP, a diffuse cytoplasmic staining of P_T451-PKR and a robust staining in spine-like structures of astrocytic processes with Aβ42 (Figure 3C and 3G) were observed and were well prevented by C16 treatment (Figure 3D, H). Microglia stained with anti-CD68 antibodies displayed a high level of activated PKR after 72 h of Aβ42 exposure (Figure 3K, O) compared to DMSO-treated cells (Figure 3I, M). There was also a change in cellular morphology; microglia were activated with appearance of thick processes and irregular shape with Aβ42 treatment (Figure 3K, O). C16 partially rescued this activation of PKR in microglia. Furthermore, we found only microglia with no thick processes around cell bodies as with C16 alone (Figure 3. merge images J and L).

The same experimental conditions were followed to study activation of the NF-κB/IκB signaling pathway. Results obtained by immunoblotting from cell lysates are presented as the ratio of phospho-protein/total protein in order to evaluate the activation of both proteins. The results show a significant increase in phosphorylation of IκB at serine 32-36 (79%) and NF-κB at serine 536 (629.8%) with Aβ42 exposure (Figure 4). p65-mediated transcription is regulated by S536 phosphorylation in the transactivation domain (TAD) by a variety of kinases (TRAF family member-associated NF-κB activator (TANK)-binding kinase (TBK), IKKα, and p38) through various signalling pathways. This phosphorylation enhances p65 transactivation potential [45‐47].

To determine the effect of PKR inhibition on cytokine levels in our cell lysates and released into the medium, samples were assayed by ELISA to quantify TNFα, IL-1β and IL-6 levels. Intracellular levels of these three cytokines were significantly higher in cells treated with 20 μM Aβ42 for 72 h (increase of 86.2% TNFα, 84% IL-1β and 50.6% IL-6) compared to DMSO-treated cells (Figure 5A). Treatment with 210 nM C16 significantly decreased levels of TNFα- and IL-1β-induced Aβ42 (83.2% and 60.7% inhibition, respectively) but failed to prevent IL-6 production (Figure 5A). Cytokine levels (TNFα and IL-1β) in Aβ42-exposed cells pretreated with 210 nM C16 were comparable to those measured in the absence of Aβ42 (Figure 5A).

Levels of released TNFα and IL-1β were also significantly increased after Aβ42 exposure (58.3% and 60.7% respectively) compared to DMSO-treated cells (Figure 5B). As for produced cytokines, the Aβ42-induced release of TNFα and IL-1β was significantly prevented by 210 nM C16 (50.1% and 52.7% inhibition, respectively). No significant change was observed for released IL-6, but levels of produced and released IL-6 remained very low in our experimental conditions (Figure 5B).

As we have shown before, Aβ42 induced the NF-κB signaling pathway and cytokine production, which were prevented by the inhibitor of PKR, compound C16. The beneficial effect of C16 has also been analyzed by using scanning electron microscopy. In micrographs, 20 μM Aβ42 largely affected co-cultures, producing massive neuronal loss (Figure 6). Axonal and dendritic networks were also altered with many disruptions of axons and dendrites, which clearly appeared thinner than with DMSO or 210 nM C16 treatments. Microglia were activated and different morphological changes were observed: microglia cells displayed numerous spiny processes along their cell bodies and cytoplasmic projections, and some cells underwent transformations into multipolar cells or cells with at least one thin process extending a distance greater than three times the cell's body diameter, known as "process-bearing microglia". Some occasional short secondary branches were also observed (see inset in micrograph of Aβ42-treated co-culture in Figure 6). On the contrary, in C16/Aβ42 experimental conditions, microglia looked like smooth cells with few spines as with DMSO or C16 treatment without Aβ42 treatment. While some neurons were dead, compared to treatment with DMSO alone, the network of axons and dendrites was preserved and comparable to the network observed with DMSO or C16 treatments (Figure 6).

Caspase-3 is known to be a crucial mediator of apoptosis through its protease activity. Activation of caspase-3 requires proteolytic processing of its inactive zymogen into activated fragments after cleavage at aspartic acid 175. In order to evaluate apoptosis in cell co-cultures, we studied the activation of caspase-3 in cell lysates represented by the ratio of cleaved-caspase-3/total caspase-3 (Figure 7). Results show a great increase in activation of caspase-3 after Aβ42 exposure for 72 h (97%) compared to DMSO-treated cells. This activation was totally prevented by 210 nM C16, and the ratios were comparable to those obtained in DMSO-treated cells.

During apoptosis, phosphatidylserine is translocated from the inner to the outer plasma membrane leaflet. This externalization was analyzed with annexin V-FITC staining to examine apoptotic states of the different cell populations after treatment with 210 nM C16 and Aβ42 exposure for 72 h (Figure 8). Furthermore, the apoptosis detection kit includes propidium iodide (PI) to label the cellular DNA in necrotic cells. This combination allows the differentiation among early apoptotic cells (annexin V-FITC-positive, PI-negative), necrotic cells (annexin V-FITC-positive, PI-positive), and viable cells (annexin V-FITC-negative, PI-negative). In all conditions examined, no PI staining associated with annexin V-FITC staining was observed. The state of necrotic cells was probably at a maximum, with complete nucleus destruction, explaining the lack of PI staining. Thus, co-staining annexin V-FITC and cell markers excluded PI incubation in our protocol. The results show that prominent annexin V-FITC staining colocalizes with MAP2 staining after Aβ42 exposure, whereas GFAP-positive cells appeared unaffected (Figure 8A and 8C). We found also a diffuse and very weak staining of annexin V-FITC in CD68-positive cells (Figure 8E and 8G). Exposure to 210 nM C16 yielded no evidence of apoptosis either in neurons (Figure 8B and 8D) or in microglia (Figure 8F and 8H).