Pro-inflammatory gene expression and neurotoxic effects of activated microglia are attenuated by absence of CCAAT/enhancer binding protein β

A colony of C/EBPβ^+/- [28] mice on a C57BL/6-129S6/SvEv background was maintained. Animals from this colony showed no serological evidence of pathological infection. The animals were group-housed (5-6) in solid floor cages and received a commercial pelleted diet and water ad libitum. Experiments were carried out in accordance with the Guidelines of the European Union Council (86/609/EU) and following the Spanish regulations (BOE 67/8509-12, 1988) for the use of laboratory animals, and were approved by the Ethics and Scientific Committees from the Hospital Clínic de Barcelona.

Genomic DNA was isolated from 2 mg liver samples using Extract-N-AmpTissue PCR Kit (Sigma-Aldrich, XNAT2) following kit instructions. PCR amplification was performed in 20 μl total volume, using 1 μl of tissue extract, 0.8 μM C/EBPβ-1s forward primer (AAgACggTggACAAgCTgAg), 0.4 μM C/EBPβ-NeoAs (CATCAgAgCAgCCgATTgTC) and 0.4 μM C/EBPβ-4As (ggCAgCTgCTTgAACAAg TTC) reverse primers. Samples were run for 35 cycles (94°C for 30 s, 59°C for 30 s, 72°C for 90 s).

C/EBPβ+/- mice were crossed and pregnant females were sacrificed on the 19th day of gestation by cervical dislocation. Embryos (E19) were surgically extracted from the peritoneal cavity. Their livers were dissected and used to genotype the animal, whereas their brains were dissected and processed as previously described [29] with minor modifications. Cultures reached confluence after 16 ± 3 days in vitro (DIV) and were then subcultured.

Each flask was washed in serum-free medium and was digested with 0.25% trypsin-EDTA solution for 5 min at 37°C. Trypsinization was stopped by adding an equal volume of culture medium with FBS 10%. Cells were pelleted (7 min, 180 g), resuspended in 1 mL culture medium, and brought to a single cell suspension by repeated pipetting. Cells were seeded at 166000 cells/mL. These were therefore secondary cultures and they were used at 12 ± 3 DIV. Astrocytes were the most abundant cell type and microglial cells were approximately 20%.

Microglial cultures were prepared by mild trypsinization from mouse mixed glial culture as previously described [30].

Cortical neuronal cultures were prepared from C57BL/6 mice at embryonic day 16 as described [31]. Neuronal cultures were used at 5 DIV.

Microglial cultures were obtained as described [31]. After astrocyte removal, microglial cells were incubated with 0.25% trypsin for 10 min at 37°C. Trypsinization was stopped by adding the same volume of culture medium with 10% FBS. Cells were gently scraped and centrifuged for 5 min at 200 g. Pellets were resuspended in neuronal culture medium and aliquots of the cell suspension (10 μL/well) were seeded on top of 5 DIV primary neuronal cultures at a final density of 4 × 10⁵ cells/mL (1.3 × 10⁵ cells/cm²).

Mixed glial cultures: The culture medium was replaced 24 h prior to treatment. Mixed glial cultures were treated with 100 ng/mL lipopolysaccharide (LPS, Sigma-Aldrich, L-2654, E. coli serotype 026:B6) and 0.1 ng/mL recombinant mouse interferon-γ (IFNγ, Sigma-Aldrich, I4777) prepared from x10 solutions.

Neuronal-primary microglia co-cultures: 100 ng/mL LPS and 30 ng/mL IFNγ were added to the culture medium one day after seeding primary microglial cells on top of neuronal cultures.

NO production was assessed by the Griess reaction. Briefly, 50 μL aliquots of culture supernatants were collected 48 h after LPS+IFNγ treatment, and incubated with equal volumes of Griess reagent (1% sulphanilamide, 0.1% N-(1-naphthyl)ethylendiamine dihydrochloride, and 5% phosphoric acid) for 10 min at room temperature (RT). Optical density at 540 nm was determined using a microplate reader (Multiskan spectrum, Thermo Electron Corporation). Nitrite concentration was determined from a sodium nitrite standard curve.

Nuclear extracts were prepared as described [32] with a few modifications. Nuclear protein was extracted from mixed glial cultures after 2 h LPS or LPS+IFNγ treatment. Cells from two wells of 6-well plate were scrapped into cold 0.01 M phosphate-buffered saline (PBS, pH 7.4) and centrifuged for 4 min, 4500 g at +4°C. The resulting pellet was resuspended in 400 μL of buffer A: 10 mM HEPES pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 0.5 mM phenylmethylsulphonyl fluoride (PMSF) and 1 mM dithiothreitol (DTT) and cells were swollen on ice for 15 min. After addition of 25 μL of 10% Igepal CA-630 (Sigma-Aldrich, I8896), cells were vigorously vortexed for 10 s and incubated for 10 min on ice, then a 10-min centrifugation at 13200 g was performed and the pellets were resuspended in 50 μL of buffer C consisting of 20 mM HEPES pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM PMSF and 1 mM DTT. Solutions A, B, C and PBS were supplemented with protease inhibitor cocktail Complete^® (Roche, 1836145). After 2 h of shaking at 4°C, nuclei were pelleted by a 5 min spin at 2000 g. The supernatant containing nuclear proteins was collected and protein amount was determined by the Lowry assay (Total Protein kit micro-Lowry, Sigma-Aldrich, TP0300). Oligonucleotides containing C/EBP consensus sequences (Santa Cruz Biotechnology, sc-2525) were labelled at their 3'-end using [α-³³P]dATP (3000 Ci/mmol; Dupont-NEN, NEG-612H) and terminal deoxynucleotidyltransferase (TdT; Oncogene Research Products, PF060), and purified using illustra MicroSpin G-50 Columns (GE, 27-5330-01). Five micrograms of nuclear proteins were incubated for 30 min at RT with the labelled oligonucleotides (25000 cpm/reaction assay) in binding buffer (20% glycerol, 5 mM MgCl₂, 2.5 mM EDTA, 2.5 mM DTT, 50 mM Tris-HCl, 250 mM NaCl and 0.2 mg/mL Poly(dI:dC)). After the addition of Hi-Density TBE buffer to samples (15% Ficoll type 400, 1x TBE, 0.1% Bromophenol Blue, 0.1% Xylene Cyanol), proteins were separated by electrophoresis on a 6% DNA retardation gel (Invitrogen, EC6365BOX) at 4°C, 90 min at 100 V in 0.5x TBE buffer. In supershift assay, 0.5 μg of rabbit anti-mouse C/EBPβ (Santa Cruz Biotechnology, sc-150) or IgG (Santa Cruz Biotechnology, No.sc-2027) were added 10 min before oligonucleotide incubation.

Protein levels were determined in primary mixed glial cells 16 h after treatments. For isolation of total proteins, two wells from 6-well plates were used per condition. After a cold PBS wash, cells were scrapped and recovered in 100 μL per well of RIPA buffer (1% Igepal CA-630, 5 mg/mL sodium deoxycholate, 1 mg/mL sodium dodecyl phosphate (SDS) and protease inhibitor cocktail Complete^® in PBS). The content of the wells was pooled, sonicated, centrifuged for 5 min at 10400 g and stored at -20°C. Protein amount was determined by the Lowry assay.

Fifty micrograms of denatured (2.5 mM DTT, 100°C for 5 min) total protein extracts were subjected to 10% SDS-PAGE and transferred to a PVDF membrane (Millipore, IPVH00010) for 90 min at 1 mA/cm². After washing in Tris-buffered saline (TBS: 20 mM Tris, 0.15 M NaCl, pH 7.5) for 5 min, dipping in methanol for 10 s and air drying, the membranes were incubated with primary antibodies overnight at 4°C: polyclonal rabbit anti-C/EBPβ (1:500, Santa Cruz Biotechnology, sc-150), monoclonal mouse anti-NO synthase-2 (NOS2; 1:200, BD Transduction Laboratories, 610431), monoclonal mouse anti-βactin (1:100000, Sigma-Aldrich, A1978) and polyclonal rabbit anti-GFAP (1:10000, DakoCytomation, Z0334) diluted in immunoblot buffer (TBS containing 0.05% Tween-20 and 5% no-fat dry milk). Then, the membranes were washed twice in 0.05% Tween-20 in TBS for 15 s and incubated in horseradish peroxidase (HRP)-labelled secondary antibodies for 1 h at RT: donkey anti-rabbit (1:5000, GE, NA934) or goat anti-mouse (1:5000, Santa Cruz Biotechnology, sc-2055). After extensive washes in 0.05% Tween-20 in TBS, they were incubated in ECL-Plus (GE, RPN2132) for 5 min. Membranes were then exposed to the camera of a VersaDoc System (Bio-Rad), and pixel intensities of the immunoreactive bands were quantified using the percentage adjusted volume feature of Quantity One 5.4.1 software (Bio-Rad). Data are expressed as the ratio between the intensity of the protein of interest band and the loading control protein band (β-actin).

mRNA expression was determined in mouse mixed glial cells 6 h after treatments. For isolation of total RNA, 2 wells of 24-well plates were used per experimental condition. Total RNA was isolated using an Absolutely RNA Miniprep kit (Agilent Technologies-Stratagene 400.800) and 100 ng of RNA for each condition was reverse-transcribed with random primers using Sensiscript RT kit (Qiagen, 205213). cDNA was diluted 1/25 and 3 μL were used to perform qPCR. The primers (Roche) were used at a final concentration of 300 nM (Table 1). β-Actin and Rn18s mRNAs levels are not altered by treatments (data not shown). qPCR was carried out with IQ SYBR Green SuperMix (Bio-Rad, 170-8882) in 15 μL of final volume using iCycler MyIQ equipment (Bio-Rad). Primer efficiency was estimated from standard curves generated by dilution of a cDNA pool. Samples were run for 40 cycles (95°C for 30 s, 60°C for 1 min, 72°C for 30 s). Amplification specificity was confirmed by analysis of melting curves. Relative gene expression values were calculated with the comparative Ct or ΔΔCt method [33] using iQ5 2.0 software (Bio-Rad). Ct values were corrected by the amplification efficiency of the respective primer pair which was estimated from standard curves generated by dilution of a cDNA pool.

qChIP was performed as previously described [34] with modifications. Briefly, primary mixed glial cultures were cross-linked in 1% formaldehyde for 10 min at RT, quenched with 125 mM glycine for 5 min a RT. Cells were washed in PBS with 1 mM PMSF and protease inhibitor mix, then the cells were resuspended with 150 mM NaCl, 50 mM Tris-HCL pH7.5, 5 mM EDTA, 0.5% vol/vol NP-40, 1% vol/vol Triton X-100, 1% wt/vol SDS, 1 mM PMSF, protease inhibitor mix (IP Buffer). Chromatin shearing was obtained from 2 × 10⁵ cells using Labsonic M sonicator (7 × 30 s on and 30 s off; cycle 0.8; 100% amplitude). In parallel, an aliquot of chromatin sheared from each sample was separated as a loading control for the experiment (input). The protocol for chromatin immunoprecipitation (ChIP) was as follows: first, 10 μL of Dynabeads^® protein A (Invitrogen, 100.01D) were washed twice with 22 μL of cold IP Buffer (without SDS). Then the beads were resuspended in 11 μL of IP Buffer. Next, 90 μL of IP Buffer was added to a PCR tube with 10 μL of pre-washed protein A-beads. Two micrograms of polyclonal rabbit C/EBPβ antibody (Santa Cruz Biotechnology, sc-150X) or with 2 μg of rabbit IgG (Santa Cruz Biotechnology, sc-2027) as negative control were added and the mixture was incubated at 40 rpm on a rotating wheel for at least 2 h at 4°C. Then, the tube was placed on a magnetic rack for 1 min. The supernatant was discarded and 100 μL of sheared chromatin was added. Samples were incubated overnight at 40 rpm rotation at 4°C. Finally, the tube was placed on the magnetic rack for 1 min. The supernatant was discarded and the immunoprecipitation complex was washed three times with 100 μL of IP Buffer for 4 min on a rotating wheel and placed in the magnetic rack again for 1 min to discard the supernatant. The fourth wash was done with 10 mM Tris-HCl pH 8.0 and 10 mM EDTA buffer. Protein was degraded by a 2-h incubation at 68°C in 200 μL of IP Buffer complemented with 50 μg/mL of proteinase K. DNA was isolated with phenol-chloroform-isoamylalcohol 25:24:1 (Sigma-Aldrich, 25666 and P4556) extraction. Input and ChIP samples were analyzed with qPCR using SYBR green (Bio-Rad). Three microliters of input DNA (diluted 1/50) and ChIP were amplified in triplicate in 96-well optical plates using a MyIQ Bio-Rad Real Time Detection System. The C/EBPβ binding site in the IL-10 promoter was used as a positive control [35]. MatInspector was used to identify the proximal C/EBPβ consensus sequence in each analyzed promoter. The sequences for each amplified locus are indicated in the table 2. Samples were run for 45 cycles (95°C for 30 s, 62°C for 1 min, 72°C for 30 s), for further details see qPCR methods.

Cultured cells were fixed with 4% paraformaldehyde in PBS for 20 min at RT. For immunocytochemistry using fluorescence labelling, cells were permeated with chilled methanol for 7 min, then washed with PBS. Cells were incubated overnight at 4°C with 7% normal goat serum (Vector, S-1000) in PBS containing 1% Thimerosal (Sigma-Aldrich, T5125) and primary antibodies: polyclonal rabbit anti-C/EBPβ (1:500, Santa Cruz Biotechnology, sc-150), monoclonal mouse anti-NOS2 (1:200, BD Transduction Laboratories, 610431), polyclonal rabbit anti-GFAP (1:1000, DakoCytomation, Z0334) and monoclonal rat anti-CD11b (1:300, Serotec, MCA711G, clone 5C6). After rinsing in PBS, cells were incubated for 1 h at RT with secondary antibodies: goat anti-mouse Alexa 546 (1:1000, Molecular Probes, A-11018), goat anti-rabbit Alexa 546 (1:1000, Molecular Probes A-11010), Alexa 488 (1:1000, Molecular Probes, A-11070) or goat anti-rat Alexa 488 (1:500, Molecular Probes, A-11006). After secondary antibody incubation, cells were stained with Hoechst 33258 for 7 min. For immunocytochemistry using peroxidase labelling, cells were permeated and endogenous peroxidase activity was blocked by incubation with 0.3% H₂O₂ in methanol for 10 min. Non-specific staining was blocked by incubating the cells with 10% normal goat serum in PBS containing 1% BSA for 20 min at RT. The cells were then incubated with monoclonal mouse anti-MAP2 primary antibody (1:2000, Sigma-Aldrich, M1406) overnight at 4°C. In MAP2 staining, biotinylated horse anti-mouse secondary antibody (1:200, Vector, BA-2000) for 1 h at RT. Following incubation with ExtrAvidin^®-Peroxidase (1:500, Sigma-Aldrich, E2886) for 1 h at RT, colour was developed with diaminobenzidine (Sigma-Aldrich, D5637). The antibodies were diluted in PBS containing 1% BSA and 10% normal horse serum (Vector, S-2000). Microscopy images were obtained with an Olympus IX70 microscope and a digital camera (CC-12, Soft Imaging System GmbH).

Neuronal viability was evaluated by MAP2 immunostaining using ABTS (2, 3'-azinobisethylbenzothiazoline-6-sulphonic acid) and absorbance analysis [31]. Neuronal viability was expressed as a percentage of control levels.

Hoechst-33258- and CD11b-positive cells were semi-automatically counted from 20x photomicrographs using ImageJ 1.42I NIH software. For each experiment (n = 4), three wells per condition were used and four fields per well were counted in a blind manner. NOS2-positive cells were counted manually from 20x photomicrographs. For each experiment (n = 11), two wells per condition were used and two fields per well were counted.

Data were analyzed using GraphPad 4.02. Two-way analysis of variance (ANOVA) followed by Bonferroni post-test was used when the effect of genotype on treatment was studied and vice versa. One-way ANOVA was used followed by Dunnet's post-test when comparing versus control or Bonferroni's post-test when comparing versus different experimental conditions. Values of p < 0.05 were considered statistically significant. Error bars are presented in all graphs as standard error of the mean (SEM).

To study the role of C/EBPβ in glial activation we used C/EBPβ-null mice. Because of the infertility of C/EBPβ-null females and a perinatal death rate of approximately 50% for C/EBPβ-null neonates, we have modified the standard procedures to prepare mixed glial cultures from CNS tissue pools of several mouse neonates and designed a protocol to prepare secondary mixed glial cultures from the cerebral cortex of one single E19-E20 mouse embryo (see Methods for details). Forty-one C/EBPβ-null mice and forty-one wild-type littermates were used during this study. To ensure that wild-type and C/EBPβ-null glial cultures were comparable, we first analyzed total cell density and abundance of their two main cell types, astrocytes and microglia, in both cultures. No differences were observed between wild-type and C/EBPβ-null cultures in total cell density as assessed by automatic counting of Hoechst 33258-stained nuclei (Figure 1A), but a moderate increase in total cell number was induced by LPS and LPS+IFNγ. C/EBPβ absence did not affect microglial density as assessed by CD11b-positive cell counting (Figure 1B). Estimation of astrocytes number in these cultures is not trivial. Astrocytes are densely packed, almost all nuclei are surrounded by GFAP-positive filaments, and it is often difficult to discern whether a given nucleus belongs to a GFAP-positive cell or, in fact, the GFAP signal belongs to a neighbor astrocyte. We therefore analyzed total GFAP content by western blot as an indirect estimation of astroglial number and no differences were observed between wild-type and C/EBPβ-null glial cultures (Figure 1D, E). Neither CD11b nor GFAP immunocytochemistry revealed differences between wild-type or C/EBPβ-null cultures in morphology of microglial cells or astrocytes, respectively (Figure 1C, F). These results indicate that wild-type and C/EBPβ-null mixed glial cultures do not differ in total cell density or in proportions or morphology of their two major cell types, astrocytes and microglia.

In this study, we have used LPS and LPS+IFNγ to study the role of C/EBPβ in glial activation in secondary cultures. The effects of both stimuli on C/EBPβ expression in glial cultures have not been compared before. As seen in Figure 2A-D, both LPS and LPS+IFNγ induced strong increases in C/EBPβ mRNA levels 6 h after treatment, and in nuclear levels of both activating (Full/LAP) and inhibitory (LIP) C/EBPβ isoforms 24 h after treatment. The increases in C/EBPβ mRNA and protein induced by LPS and LPS+IFNγ were of similar magnitude.

Since the mRNA or protein levels of a transcription factor are of relative importance to study its functionality, we studied the DNA binding activity of C/EBPβ in LPS- or LPS+IFNγ-treated glial cells. Electrophoretic mobility shift assays showed that binding of nuclear proteins to a DNA oligonucleotide containing the C/EBPs consensus sequence was increased by LPS and LPS+IFNγ treatments (Figure 3A, lanes 1-3). Supershift experiments showed the presence of C/EBPβ in shifted complexes I to III (Figure 3A lanes 4-6). The specificity of the supershift is demonstrated by the lack of supershift elicited by the same concentration of IgG (Figure 3A lanes 7-9). This indicates that C/EBPβ is a key component of C/EBPs DNA binding complexes during LPS- and LPS+IFNγ-induced glial activation.

Next, we estimated the binding of C/EBPβ to the promoters of four major pro-inflammatory genes: nitric oxide synthase 2 (NOS2), IL-1β, IL-6 and TNFα, in mixed glial cultures using a qChIP assay (Figure 3B). In untreated glial cultures, no specific binding of C/EBPβ was measurable in any of the four promoters analyzed. However, 2 h after LPS treatment, C/EBPβ binding was observed in the NOS2 promoter. Interestingly, in LPS+IFNγ-treated glial cultures C/EBPβ binding was observed in all four promoters analyzed and, in the case of the NOS2 promoter, C/EBPβ binding was significantly higher than in LPS-treated glial cultures (Figure 3B).

To study the involvement of C/EBPβ in the regulation of pro-inflammatory gene expression, mRNA levels of NOS2, IL-1β, IL-6 and TNFα were analyzed by qPCR in wild-type and C/EBPβ-null cultures treated with LPS or LPS+IFNγ for 6 h. In wild-type cultures all four mRNAs were strongly upregulated by LPS. This effect was exacerbated by co-treatment with IFNγ in the case of NOS2 (+92.3%), but not in the case of IL-1β, IL-6 or TNFα (Figure 4). In C/EBPβ-null cultures LPS induced upregulation of IL-1β, IL-6 and TNFα mRNAs, which was similar to that observed in wild-type cultures. However, as expected from qChIP results, the LPS-induced increase in NOS2 mRNA levels was significantly lower in C/EBPβ-null than in wild-type glial cultures (-67.4%, p < 0.05). The pattern of gene expression induced by LPS+IFNγ was more affected by lack of C/EBPβ. Thus, LPS+IFNγ-induced mRNA levels of NOS2 and IL-1β were significantly lower in C/EBPβ-null than in wild-type cultures. TNFα and IL-6 mRNA levels did not differ statistically between the two genotypes (Figure 4). In contrast to the pro-inflammatory gene pattern, mRNA levels of the anti-inflammatory cytokines IL-4 and transforming growth factor β (TGFβ1) were not altered by LPS or LPS+IFNγ treatments and no significant changes in IL-4 or TGFβ1 mRNA levels were observed between wild-type and C/EBPβ-null glial cultures under any experimental condition (Figure 4).

The important reduction in NOS2 mRNA levels in activated C/EBPβ-null glial cultures prompted us to analyze NOS2 protein levels by western blot and immunocytochemistry, and generation of NO by colorimetric detection of nitrites (Griess assay). In wild-type cultures NOS2 protein expression was induced by LPS and more markedly by LPS+IFNγ. In C/EBPβ-null cultures LPS-induced NOS2 levels were not significantly different from wild-type whereas LPS+IFNγ-induced NOS2 protein levels were markedly reduced (-77.4%, p < 0.0001) (Figure 5A, B). NO levels correlated well with the NOS2 protein data and a strongly significant attenuation in NO production induced by LPS+IFNγ was seen in C/EBPβ-null cultures (Figure 5C).

The reduction in LPS+IFNγ-induced NOS2 expression in C/EBPβ-null glial cultures seen by western blot was confirmed by immunocytochemistry. We did not observe by immunocytochemistry any NOS2-positive cells in untreated cultures (not shown), whereas in LPS- (not shown) and LPS+IFNγ-treated wild-type cultures, NOS2 immunoreactivity was observed in 14.0 ± 3.6% of total cells (Figure 5D, E). The vast majority of NOS2-positive cells in LPS+IFNγ-treated wild type mixed glial cultures also expressed CD11b (99.3 ± 1.4%; n = 11) and very rarely NOS2-positive cells expressed GFAP (0.6 ± 1.2%; n = 11) indicating that in these conditions NOS2 expression in mouse cortical mixed glial cultures is predominantly microglial. In C/EBPβ-null cultures the number of NOS2 cells was dramatically reduced after either LPS (not shown) or LPS+IFNγ treatments (Figure 5D, E). As seen in Figure 5D, the reduction of NOS2-positive cells could not be attributed to a reduction in microglial density.

Activated microglia have strong neurotoxic potential [36]. The observations of reduced expression of pro-inflammatory mediators in LPS+IFNγ-activated C/EBPβ-null glial cells, particularly microglia, prompted us to analyze whether the neurotoxic effects of LPS+IFNγ-activated microglia could be attenuated by C/EBPβ absence. To this aim, wild-type and C/EBPβ-null microglial cells were isolated and co-cultured with wild-type neurons. No neuronal death was observed when neurons not co-cultured with microglia were treated with LPS+IFNγ or when neuron/wild-type microglia co-cultures were treated with LPS alone (data not shown). In contrast, LPS+IFNγ treatment of neuron/wild-type microglia co-cultures resulted in death of 51.2% of neurons, as estimated by MAP2/ABTS/ELISA (Figure 6). Interestingly, in neuron/C/EBPβ-null microglia co-cultures treated with LPS+IFNγ, MAP2 immunoreactivity levels were equal to control levels (Figure 6) indicating that the neurotoxicity induced by LPS+IFNγ-treated microglia was completely abolished in the absence of C/EBPβ. In this model, NO production plays a major role in the neurotoxicity elicited by activated microglia since the NOS2 inhibitor 1400W (10 μM) completely abolished neuronal death in LPS+IFNγ-treated neuron/microglia co-cultures (Gresa-Arribas et al, unpublished observations).