Background
Infectious illnesses and chronic diseases elicit a constellation of metabolic and behavioral responses, including anorexia, weight loss, fever, lethargy, and disrupted sleep [
1,
2]. These sickness responses evolved as part of a highly coordinated disease-fighting strategy and over a short term confer adaptive benefits by conserving energy and diverting it to the immune system. However, if the underlying illness does not resolve in a timely manner and these sickness responses persist, they become maladaptive and can lead to the development of cachexia, a wasting syndrome that develops in patients with chronic illnesses (e.g., cancer, AIDS, chronic obstructive pulmonary disease, congestive heart failure, and dementia) [
3,
4].
Systemic inflammation is a common feature of the disparate infections and illnesses that cause sickness behaviors and cachexia [
4]. Activated immune cells release pro-inflammatory cytokines that act upon target tissues in an autocrine, paracrine, or endocrine manner and play an integral role in coordinating the host’s immune response. The brain is one such target for pro-inflammatory cytokine signaling. In response to peripheral inflammatory insults, the hypothalamus generates its own local inflammatory response as a means to amplify and propagate the inflammatory signal within the central nervous system (CNS) [
5]. This central inflammatory response involves many of the same cytokines that are released in the periphery (e.g., interleukin-1β (IL-1β), interleukin-6 (IL-6), and tumor necrosis factor-α (TNFα)) as well as chemokines that recruit leukocytes into the brain parenchyma (e.g., C-X-C motif chemokine 10 (CXCL10)). These inflammatory mediators modulate the activity of neural circuits controlling feeding, metabolism, body composition, arousal, and neuroendocrine function via direct and indirect mechanisms.
IL-1β plays a predominant role in centrally mediated sickness responses. Intracerebroventricular (icv) injection of IL-1β induces rapid and robust sickness behaviors in rodents [
6‐
10], and blocking central IL-1β signaling attenuates sickness behaviors in response to peripheral injection of the bacterial endotoxin lipopolysaccharide (LPS) [
11]. IL-1β signals through the type I interleukin-1 receptor (IL-1R1). In the rodent brain,
Il1r1 mRNA is primarily expressed by blood vessels, meninges, choroid plexus, and ependymal cells lining the cerebroventricles, but has also been reported in glia and discrete neuronal populations [
12‐
16]. When IL-1β engages the IL-1R1, the adaptor protein myeloid differentiation factor 88 (MyD88) is recruited to the activated receptor complex. This triggers an intracellular signaling cascade that causes the transcription factor nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) to translocate to the nucleus, where it binds to promoter regulatory elements and initiates transcription of inflammatory cytokine and chemokine genes [
17]. Although most IL-1β-induced inflammatory genes are regulated by NF-κB signaling, IL-1β can also activate MAPK pathways [
18]. MyD88 is requisite for many pro-inflammatory actions of IL-1β in the CNS, but IL-1β can signal via a MyD88-independent pathway in hypothalamic neurons [
19].
It is unknown which cell population(s) in the brain is/are the proximal targets for IL-1β with respect to the generation of sickness responses. MyD88 knockout (MyD88KO) mice are resistant to IL-1β-induced sickness behaviors [
6,
20]. Although populations of hypothalamic neurons that regulate feeding and metabolism express IL-1R1 and are activated or inhibited by IL-1β [
14,
15], these neurons do not appear to be the exclusive targets for IL-1β-induced sickness behaviors, because mice in which MyD88 is selectively deleted from neurons and astrocytes exhibit normal sickness behaviors in response to icv IL-1β [
6]. In contrast, conditional deletion of MyD88 from endothelial and myeloid cells (including microglia) driven by the Tie2 promoter confers resistance to anorexia, weight loss, reduced locomotor activity, and fever in response to icv IL-1β [
8].
The goal of these experiments was to examine the inflammatory responses of endothelial cells, microglia and astrocytes to IL-1β. While others have previously reported the effects of IL-1β on cellular activation and inflammatory gene expression in vivo and in isolated brain cell populations in vitro, less effort has been devoted to examining the interactions between different IL-1β-responsive brain cell populations, the directionality of signaling, or the potential for synergistic cellular actions. To this end, we took a systematic in vitro approach and measured inflammatory gene expression and NF-κB activity in primary mouse brain endothelial and glial cells, as well as in a recently described spontaneously transformed murine microglia cell line (SIM-A9) [
21]. We demonstrate that in response to IL-1β, microglia exhibit minimal inflammatory responses in isolation, but generate more robust responses when co-cultured with astrocytes and/or endothelial cells. We also find that the endothelial response to IL-1β stimulation is polarized, because application of IL-1β to the abluminal endothelial surface produces a more complex microglial response than that which occurs after the luminal endothelial membrane is exposed to IL-1β.
Methods
Animals
Adult male and female C57BL/6J (wild-type; WT), MyD88 knockout (MyD88KO), and CX3CR1-EYFP-Cre mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Mice were housed in a light- and temperature-controlled room and were provided with food and water ad libitum. All experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Animal Care and Use Committee of Oregon Health & Science University.
Drugs
Murine IL-1β (R&D Systems, Minneapolis, MN), murine TNF-α (R&D Systems), and l-leucine methyl ester hydrochloride (l-LME; Sigma, St. Louis, MO) were dissolved in PBS. LPS (Sigma) was dissolved in PBS + 0.1% bovine serum albumin (BSA). Nω-Nitro-l-arginine methyl ester hydrochloride (l-NAME; Sigma) was dissolved in phenol red-free DMEM (#31053, Life Technologies, Carlsbad, CA) supplemented with l-glutamine (2 mM) and gentamicin (50 μg/mL).
Primary brain microvessel endothelial cell cultures
Primary brain microvessel endothelial cell (BMEC) cultures were generated as previously described, with a few modifications [
22]. For each culture, 5 to 10 adult WT mice were decapitated under isoflurane anesthesia. Forebrains were isolated and separated from the meninges, minced, and then transferred to a tube containing collagenase CLS2 (1 mg/mL; Worthington, Lakewood, NJ) and DNase I (10 μg/mL; Sigma) in ggDMEM [DMEM (#11965) supplemented with
l-glutamine (2 mM) and gentamicin (50 μg/mL)] for 45 min at 37 °C in a shaking water bath. Cells were resuspended in endothelial culture medium containing 20% BSA (pH 7.4) and were spun at 1000×
g for 20 min at 4 °C. After aspirating the myelin-rich layer and supernatant, the cells were mixed with collagenase/dispase (1 mg/mL; Sigma) and DNase I (10 μg/mL) in ggDMEM and were incubated in a shaking 37 °C water bath for 30 min. Cells were then layered on a 33% Percoll (GE Healthcare Life Sciences, Pittsburgh, PA) gradient and spun at 1000×
g for 20 min at 4 °C. The microvessel layer was removed, mixed with ggDMEM, and spun for 8 min at 700×
g. At the end of the isolation, pellets were resuspended in endothelial culture medium [ggDMEM + endothelial cell growth supplement (100 μg/mL; Sigma) + 20% fetal bovine serum (FBS; Hyclone, Logan, UT)) + heparin (100 μg/mL; Sigma)] supplemented with puromycin (4 μg/mL; Sigma). Seventy-two h after seeding fragments into collagen-coated flasks, cells were washed with PBS and switched to puromycin-free culture medium. Medium was changed every 2–3 days until the cells reached confluence. Cells were then trypsinized and re-plated into 6-well plates for RNA analysis or 24-well plates containing poly-
d-lysine/laminin-coated coverslips (Corning, Corning, NY) for immunocytochemistry (ICC).
Primary mixed glia cultures
The brains were harvested from neonatal mouse pups on P1-P5. Cortices were dissected, separated from the meninges, rinsed in dissecting buffer (350 mg/L NaHCO3, 6 g/L d-glucose, 300 mg/L BSA, 1.44 g/L MgSO4*7H2O, and 10 mM HEPES in Hank’s Balanced Salt Solution) and digested with papain (Worthington) for 5 min. The enzymatic digestion was terminated by the addition of culture medium [DMEM (#11885; Life Technologies) + 10% FBS + 1% penicillin/streptomycin (Life Technologies)]. Cell suspensions were filtered and spun at 150×g for 10 min. Cells were seeded into T-75 flasks and placed in a 37 °C incubator with 5% CO2/95% O2. Culture medium was changed 24 h later, and then every 3–5 days thereafter. Between days in vitro (DIV)10–DIV20, mixed glia cultures were dissociated with 0.05% trypsin-EDTA and seeded into 6-well plates (for RNA analysis), 24-well plates containing poly-d-lysine/laminin-coated coverslips for ICC, or the upper inserts of transwells for luciferase assays.
Astrocyte-enriched cultures
Neonatal mouse cortices were processed as described above, and cells were seeded into 6-well (1.5 × 10
6 cells/well) or 24-well (1.5 × 10
5 cells/well) plates. Seventy-two h later, 1 mM
l-LME was added to each well to selectively deplete microglia [
23].
l-LME treatment was repeated again 3 days later on DIV6. On DIV9, astrocytes were treated with 0.05% trypsin-EDTA, spun at 150×
g for 10 min, and then seeded into 6-well plates for RNA analysis.
Microglia isolation
Microglia-enriched cultures were derived from mixed glia cultures using a mild trypsinization technique [
24]. Compared to the more commonly used shaking method of isolating microglia, mild trypsinization has been reported to generate higher and purer yields of microglia, and the isolated microglia are less activated at baseline [
25]. Between DIV16 and DIV19, mixed glia cultures were incubated in 0.25% trypsin-EDTA diluted 1:3 in serum-free DMEM for 30 min at 37 °C, and then the detached astrocyte layer was aspirated. The underlying adherent microglia were dissociated with 0.25% trypsin-EDTA and seeded into 24-well plates containing poly-
d-lysine/laminin-coated coverslips for ICC, or were returned to the incubator for subsequent gene expression analyses.
SIM-A9 microglia cell line
SIM-A9 cells were kindly provided by Dr. Kumi Nagamoto-Combs [
21]. DNA was extracted with a DNeasy kit (Qiagen, Valencia, CA), and the SRY gene was amplified by PCR (forward primer: 5′-AGGCGCCCCATGAATGCATT-3′; reverse primer: 5′-TCCGATGAGGCTGATATTTATAG-3′). The lack of a band in SIM-A9 cells revealed that these cells are of female origin (data not shown). Cells were grown in DME/F-12 (Hyclone, #SH-30023) + 10% FBS + 10% donor horse serum (Serum Source International, Charlotte, NC) + 1% penicillin/streptomycin. Cells were detached with splitting medium (1 mM EDTA, 1 mM EGTA, and 1 mg/mL glucose in PBS) and seeded into 6- or 12-well plates.
NF-kB luciferase SIM-A9 cells
Human embryonic kidney cells (HEK293T) were seeded at a density of 1.5 × 10
7 cells/15-cm tissue culture dish (Corning) after coating the dishes with 0.01% poly-
l-lysine (Sigma). The HIV-1-based lentiviral vector stocks were produced by co-transfecting the helper constructs pLP1, pLP2, and pLP/VSVG [
26] and the transducing plasmid pHAGE NFkB-TA-LUC-UBC-GFP-W (a gift from Darrell Kotton (Addgene Plasmid #49343)) [
27]. SIM-A9 cells were transduced with the harvested vector pHAGE NFkB-TA-LUC-UBC-GFP-W in the presence of 8 μg/ml protamine sulfate (MP Biomedicals, Santa Ana, CA) and incubated at 37° C overnight [
28]. Cells were washed, spun for 5 min at 1000 rpm, and used for FACS analysis followed by sorting the GFP-positive vector transduced cells (Canto-II, BD InFlux cell sorter; BD Biosciences, San Jose, CA). Sorted cells were washed twice prior to expansion.
Transwell BMEC experiments
Primary BMEC were isolated as described above, with a few deviations. After spinning the endothelial fragments in the Percoll gradient, cells were resuspended in culture medium containing endothelial cell growth supplement (60 μg/mL), 20% plasma-derived platelet-free serum (Alfa Aesar, Ward Hill, MA), heparin (100 μg/mL), Glutamax (2 mM, Life Technologies), and puromycin (4 μg/mL). Endothelial fragments were seeded onto collagen-coated transwell inserts (catalog #3460, Corning Costar, Corning, NY). After 72 h, BMEC were switched to puromycin-free culture medium and were co-cultured with WT primary mixed glia in the lower wells. BMEC were confluent after 4 days in co-culture with mixed glia, at which time the transwell inserts were moved to wells containing SIM-A9 or NF-kB Luc SIM-A9 cells. We used SIM-A9 cells as surrogates for primary microglia because we had difficulty isolating the large numbers of primary microglia that would be necessary for these experiments. BMEC were treated with PBS or IL-1β (50 ng/mL) added to either the luminal surface (upper chamber) or abluminal surface (lower chamber) for 6 or 8 h. Cells were lysed and then assayed for luciferase activity or frozen at −80 °C for gene expression analysis. Transwell inserts were pooled (two inserts pooled together) for measuring BMEC gene expression. In a separate experiment, once the BMEC reached confluence, the transwell inserts were moved to wells that contained CX3CR1-EYFP-Cre primary mixed glia on poly-d-lysine/laminin-coated coverslips. PBS or IL-1β (50 ng/mL) was added to both the upper and lower chambers for 30 min before proceeding with immunostaining.
Nitric oxide synthesis blockade
BMEC and SIM-A9 cells were seeded into transwells as described above. Cells were switched to phenol red-free DMEM supplemented with 2 mM l-glutamine and 50 μg/mL gentamicin (phenol red-free ggDMEM). l-NAME (1 mM) or phenol red-free ggDMEM was added to both the upper and lower chambers for 1 h. Then, PBS or IL-1β (50 ng/mL) was added to the upper chambers (i.e., the luminal endothelial surface) for an additional 8 h. SIM-A9 cells were lysed and frozen at −80 °C for gene expression analysis. Supernatants from the upper chambers were analyzed for nitrate/nitrite levels using a nitrate/nitrite colorimetric assay kit (Cayman Chemical, Ann Arbor, MI) following the manufacturer’s instructions.
WT astrocyte-enriched cultures were treated with PBS or IL-1β (50 ng/mL) for 24 h, after which the astrocyte-conditioned media was removed and added to wells containing WT primary microglia or SIM-A9 cells for an additional 4 h. Gene expression was measured in the astrocytes, primary microglia, and SIM-A9 cells.
Gene expression analyses
All gene expression studies were conducted 24–48 h after cells were seeded into 6- or 12-well plates. Cells were washed with PBS and then switched to serum-free media for 30 min. Cells were then treated with vehicle (PBS or PBS + 0.1% BSA), IL-1β (50 ng/mL), TNF-α (50 ng/mL), or LPS (10 ng/mL) for 4 h prior to lysis and stored at −80 °C. SIM-A9 cells were treated with PBS or IL-1β (50 ng/mL) for 1, 2, 4, 8, or 24 h. Total cellular RNA was extracted using RNeasy kits (Qiagen). cDNA was generated using Taqman reverse transcription reagents as previously described [
29]. Real-time PCR was performed with Taqman reagents using an ABI 7300 system (Life Technologies). Each sample was run in triplicate, with 18S or β-actin as endogenous controls. Gene expression is presented in terms of relative quantity, or foldchange relative to the vehicle (control) group, and was calculated using the 2
−ΔΔCt method. Statistical analyses were performed on the ΔC
t values for each gene.
Luciferase assay
NF-kB Luc SIM-A9 cells were seeded into empty 12-well or 48-well plates, 12-well plates containing WT primary mixed glia, or in the lower chambers of transwell plates with WT primary mixed glia in the upper chambers. Forty-eight h later, cells were washed in PBS and incubated in serum-free media for 30 min prior to treatment. Cells were treated with vehicle (PBS or PBS + 0.1% BSA), IL-1β (50 or 100 ng/mL), or LPS (10 or 100 ng/mL) for 6 h and then lysed in Glo Lysis buffer (Promega Corporation, Madison, WI). Luciferase activity in the lysates was measured using the Bright-Glo Luciferase Assay System (Promega) and a BioTek Gen5 microplate reader (Winooski, VT). Luminescence was normalized to total protein content, which was measured using a Pierce BCA protein assay kit (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer’s instructions.
Measurement of supernatant IL-1β
WT BMEC were seeded into the upper inserts of 16 transwells. In eight of the transwells, SIM-A9 cells were seeded into the lower chambers. The other eight transwells contained empty lower chambers. PBS or IL-1β (50 ng/mL) was added to the upper chambers (n = 4 each for SIM-A9 + PBS, SIM-A9 + IL-1β, empty lower chambers + PBS, and empty lower chambers + IL-1β). After 8 h, the supernatant was removed from both the upper and lower chambers and frozen at −80 °C. Supernatant IL-1β was measured by ELISA (Thermo Fisher Scientific, Mouse IL1β ELISA Ready SET Go) according to the manufacturer’s instructions. Sensitivity of the assay was 8 pg/mL.
ICC
Mixed glia, primary microglia, and BMEC were processed for ICC 24–48 h after they were seeded onto poly-d-lysine/laminin-coated coverslips. Cells were washed in PBS and incubated in serum-free media for 30 min. Cells were then treated with vehicle (PBS or PBS + 0.1% BSA), IL-1β (50 ng/mL), or LPS (10 ng/mL) for 30 min. Cells were then fixed in 4% paraformaldehyde for 30 min and blocked for 30 min in 0.3% Triton-X 100 + 1% BSA in PBS. Cells were incubated with primary antibodies diluted in PBS + 0.3% Triton-X 100 + 5% normal serum at 4 °C overnight. Primary antibodies were used at the following concentrations: rabbit anti-p65 NF-κB (1:1000; Cell Signaling, Danvers, MA), mouse anti-GFAP (1:2000; Millipore, Billerica, MA), rat anti-PE-CAM (1:100; BD Pharmingen, San Jose, CA), and chicken anti-GFP (1:1000; Abcam, Cambridge, MA). The next day, cells were incubated in Alexa Fluor secondary antibody (diluted in PBS + 0.3% Triton-X 100 + 1% normal serum) for 2 h at room temperature. The secondary antibodies were donkey anti-rabbit 555, goat anti-mouse 633, donkey anti-rat 488, and goat anti-chicken 488 (Life Technologies). Cells were stained with DAPI and mounted onto gelatin-coated slides using Aqua-Poly/Mount (PolySciences, Inc., Warrington, PA). Cells were imaged with a Nikon Ti Eclipse inverted microscope and NIS Elements software (Nikon Instruments Inc, Melville, NY). Grayscale images were merged and pseudo-colored using Adobe Photoshop CS6 (Adobe Systems, San Jose, CA). To quantify microglial NF-κB nuclear localization, 10 random fields from each coverslip (2–3 coverslips per treatment group) were imaged. An observer who was blinded to the treatment groups tabulated the percentage of YFP-positive cells with concentrated nuclear NF-kB labeling. To quantify BMEC NF-κB nuclear localization, three random fields from each coverslip or transwell (two to three coverslips or transwells per treatment group) were imaged. The percentage of DAPI-labeled nuclei that were co-labeled with NF-κB was tabulated by a blinded observer. For occludin immunostaining, transwell inserts containing BMEC were fixed in ice cold 100% ethanol for 30 min, blocked for 30 min in PBS + 3% FBS, and then incubated in rabbit anti-occludin antibody (Thermo Fisher Scientific; 1:200) at 4 °C overnight. The next day, cells were incubated in goat anti-rabbit 488 (Life Technologies) for 2 h at room temperature and then mounted with Fluoromount-G with DAPI. Cells were imaged with a DM4000 B fluorescent microscope (Leica Microsystems, Buffalo Grove, IL) equipped with a DFC340 FX camera (Leica).
Statistical analysis
Data are expressed as the mean ± SEM for each group. Statistical analyses were performed using GraphPad Prism 5 for Mac OS X. Groups were compared by Student’s t tests or ANOVA followed by Bonferroni-corrected t tests. For the SIM-A9 time course experiment, groups were compared by two way ANOVA (time × treatment) followed by Bonferroni-corrected t tests. Differences were considered significant when p < 0.05.
Discussion
IL-1β signaling in the CNS plays a critical role in innate immunity and cellular inflammatory responses. To address the question of which cells mediate IL-1β-induced disruptions in behavior, metabolism, and neuroendocrine function in vivo, many groups have genetically manipulated IL-1R1 or MyD88 expression in specific brain cell populations. For example, the Tie2 promoter is commonly exploited to target gene constructs to endothelial cells. We previously demonstrated that Tie2Cre-MyD88
Lox/Lox mice are completely resistant to anorexia, weight loss, fever, and reduced locomotor activity in response to icv IL-1β [
8]. Using the Tie2 promoter to knockdown endothelial IL-1R1 expression [
31] or to restore endothelial IL-1R1 expression in an IL-1R1 null background [
32], other investigators have concluded that endothelial IL-1R1 signaling is necessary and/or sufficient for fever, reduced locomotor activity, CNS leukocyte infiltration, and activation of microglia and hypothalamic neurons in response to icv IL-1β. Although the Tie2 promoter is often described as an endothelial-specific Cre-driver, the Tie2 lineage is present in all microglia [
33‐
35]. Thus, the relative contributions of endothelial vs. microglial IL-1β signaling in the generation of sickness responses cannot be distinguished in mice harboring genetic manipulations linked to the Tie2 promoter.
Endothelial cells are cellular targets for inflammatory mediators and play a role in generating tissue responses to systemic infections. In the non-inflamed brain, cerebrovascular cells are the principal sites of IL-1R1 expression [
12], and disruption of IL-1β signaling in endothelium attenuates or abolishes IL-1β-induced fever and reduced locomotor activity [
31,
36]. In the rodent brain, IL-1R1 mRNA and protein are expressed by parenchymal endothelial cells and perivascular cells in the choroid plexus and meninges [
32,
37]. We demonstrated that IL-1β induces NF-κB nuclear localization in primary BMEC from WT, but not MyD88KO mice. This is consistent with reports that IL-1β rapidly induces NF-κB nuclear translocation and
Nfkbia mRNA expression (a transcriptional marker of NF-κB activity) in the brain microvasculature following central or peripheral injection [
38‐
40], and that this response is absent in MyD88-deficient mice [
41]. We also observed that IL-1β elicits robust increases in mRNAs for inflammatory cytokines, adhesion molecules, the chemokine
Cxcl10, and synthetic enzymes for production of nitric oxide and prostanoids in WT BMEC. Similar transcriptional and secretory profiles in response to IL-1β have also been demonstrated in BMEC in vivo [
40] and in human and murine BMEC cell lines [
42,
43].
In response to pathogens or tissue damage, microglia alter their morphology and release pro-inflammatory cytokines and chemokines. Microglia are activated by various inflammatory stimuli, including pathogen-associated molecular patterns (e.g., LPS) and cytokines (e.g., TNFα) [
44,
45]. In vivo, microglia do not express IL-1R1 under basal conditions, although IL-1R1 is induced in hippocampal microglia following brain injury [
13,
32]. Although we detected very low basal levels of
Il1r1 mRNA in our primary microglia cultures, we cannot rule out the possibility that this was due to
Il1r1 expressed by residual astrocytes (and/or other CNS cells) that were not removed during the trypsinization process. Similarly, Pinteaux et al. (2002) reported a low level of
Il1r1 mRNA expression in primary microglia cultures, but could not rule out oligodendrocyte progenitor cell contamination [
46].
Il1r1 mRNA was undetectable in SIM-A9 cells, nor was it detected in two other microglial cell lines [
42]. Although microglia are a major source of IL-1β in the CNS, there is conflicting evidence in the literature regarding whether microglia themselves are direct targets for IL-1β signaling. IL-1β increased pro-inflammatory cytokine (
Il1b,
Tnf, and
Il6) gene expression and chemokine (MIP-1α and MIP-1β) secretion in cultured human fetal microglia [
47,
48]. In contrast, Pinteaux et al. (2002) did not observe changes in cytokine expression or secretion, NF-κB activation or MAPK activity in IL-1β-treated murine primary microglia [
46]. Likewise, IL-1β did not induce inflammatory gene expression in the murine microglia cell lines EOC2 or EOC20 [
42]. In our studies, IL-1β-treated primary microglia and SIM-A9 cells exhibited small increases in
Nfkbia mRNA. The functional significance of this is unclear, because we did not detect NF-κB nuclear localization in isolated primary microglia or increased luciferase activity in NF-κB Luc SIM-A9 cells. We also observed small (1.5- to 2-fold) yet statistically significant increases in other inflammatory genes in primary microglia and SIM-A9 cells. The most pronounced effect of IL-1β was on
Nos2 gene expression, which was increased 7.8-fold in WT primary microglia (but was only marginally elevated at the same 4 h time point in SIM-A9 cells).
Nos2 is transcriptionally regulated by diverse inflammatory stimuli and is widely accepted as a marker for classical (M1) microglial activation [
49,
50]. We conclude that isolated microglia do not exhibit significant inflammatory responses to IL-1β. However, it is possible that IL-1β pre-conditions microglia for the arrival of subsequent inflammatory stimuli (and potentially, a more robust inflammatory response) by shifting them toward an M1 phenotype.
Astrocytes express IL-1R1 [
13,
46,
51] and are direct targets for IL-1β signaling in vitro [
52]. We observed increased inflammatory gene expression in WT mixed glia and enriched astrocyte cultures in response to IL-1β, but not in corresponding cultures from MyD88KO mice. Furthermore, we observed abundant NF-κB nuclear localization in GFAP-positive astrocytes in IL-1β-treated WT mixed glia cultures. We next sought evidence of astrocyte-microglia crosstalk in response to IL-1β, similar to what other groups have demonstrated in response to LPS in vitro [
23,
53]. We observed nuclear translocation of NF-κB in approximately 50% of microglia in IL-1β-treated mixed glia cultures. Because we did not observe IL-1β-induced nuclear NF-κB in isolated primary microglia or enhanced luciferase activity in isolated NF-κB Luc SIM-A9 cells, it is likely that astrocytes play an intermediary role in relaying the IL-1β signal to microglia. We observed increased luminescence in response to IL-1β when NF-κB Luc SIM-A9 cells were seeded into the same wells as WT mixed glia, but not when NF-κB Luc SIM-A9 cells were physically separated from IL-1β-treated WT mixed glia in transwells. These results are consistent with the hypothesis that direct physical contact between astrocytes and microglia is necessary for IL-1β-induced NF-κB activation in microglia, whereas soluble astrocyte-derived factors activate microglia via NF-κB-independent mechanisms.
We observed that conditioned media from IL-1β-treated astrocytes increased
Nos2 gene expression in primary microglia. However,
Nos2 gene expression was significantly reduced in SIM-A9 cells that were exposed to astrocyte-conditioned media, regardless of whether the astrocytes were treated with PBS or IL-1β, suggesting that astrocytes release signals that dampen microglial nitric oxide production. A role for astrocytes in inhibiting microglial Nos2 production has previously been demonstrated in LPS-treated mixed glia cultures [
54]. The different
Nos2 responses of primary microglia and SIM-A9 cells could be indicative of a fundamental difference between the two cell types in their ability to respond to inflammatory signals. Alternatively, induction of
Nos2 mRNA could be due to residual astrocyte contamination in our primary microglia cultures.
We also investigated the possibility that IL-1β-treated brain endothelial cells relay inflammatory signals to microglia. When all three cell types (BMEC, astrocytes, and microglia) were cultured together in transwells, IL-1β induced NF-κB nuclear localization in approximately 50% of microglia. This percentage of microglial activation is comparable to what we observed in IL-1β-treated mixed glia cultures. IL-1β did not induce luciferase activity in NF-κB Luc SIM-A9 cells that were co-cultured with BMEC. In one of our transwell experiments, IL-1β-treated SIM-A9 cells exhibited greater induction of
Nfkbia mRNA in the absence of BMEC than they did in the presence of BMEC, which suggests that endothelial cells release signals that dampen microglial NF-κB signaling. However, endothelial IL-1β treatment did increase inflammatory gene expression in SIM-A9 cells. When IL-1β was added to the luminal endothelial surface, only
Il1b gene expression was induced in the underlying SIM-A9 cells. Adding IL-1β to the abluminal endothelial surface also induced
Il1b mRNA expression in SIM-A9 cells, as well as
Il6 and
Ptgs2 mRNAs. The fact that SIM-A9 cells did not exhibit inflammatory responses to IL-1β in the absence of BMEC provides further confirmation that the SIM-A9 cells were not directly activated by IL-1β, but rather by an endothelial-derived signal. This is consistent with in vivo experiments demonstrating that after peripheral or central IL-1β injection, increased
Nfkbia mRNA expression in brain endothelial cells precedes microglial activation [
38,
40].
Given their polarization and position at the blood-brain interface, it is not surprising that brain endothelial cells have the ability to differentially respond to inflammatory signals arriving via the blood vs. locally generated within the CNS. IL-1R1 is expressed on both the luminal and abluminal endothelial cell membranes [
55]. In our transwell experiments, IL-1β induced similar patterns of inflammatory gene expression in BMEC regardless of whether it was added to the luminal or abluminal endothelial membrane. BMEC can respond to inflammatory signals arriving on one surface and subsequently secrete cytokines from the opposing membrane [
30]. Our finding that abluminal IL-1β treatment induced the expression of more diverse inflammatory genes in SIM-A9 cells than luminal IL-1β treatment is consistent with the observation that icv IL-1β injection elicits a broader profile of inflammatory gene expression in the brain (both in terms of number of genes and anatomical localization) than i.v. IL-1β treatment [
40]. Compared to luminal LPS treatment, BMEC exhibit greater IL-6 secretion when LPS is applied to the abluminal surface [
30]. It is possible that the abluminal endothelial surface is more responsive to inflammatory stimuli in general, perhaps due to different receptor expression profiles and/or densities on the luminal and abluminal membranes, membrane-specific activation of different intracellular signaling pathways, and/or crosstalk with adjacent inflammation-sensitive perivascular cells [
56,
57].
Future studies will be devoted to identifying the astrocyte- and endothelial-derived signaling molecules that activate microglia under inflammatory conditions. Our observation that only astrocyte-derived signals induced NF-κB activity in NF-κB Luc SIM-A9 cells is consistent with the hypothesis that astrocytes and endothelial cells use different signaling mediators to activate microglia. Furthermore, it is plausible that BMEC release different signaling molecules depending upon which endothelial membrane is exposed to IL-1β, because the SIM-A9 transcriptional profile differed in response to luminal vs. abluminal IL-1β treatment. Based upon the BMEC transcriptional response to IL-1β, likely candidate molecules for endothelial-microglia communication include IL-6, TNFα, and prostaglandins. With respect to the latter, prostaglandin signaling is essential for IL-1β-induced fever [
55,
58,
59], and pharmacological or genetic blockade of prostaglandin synthesis at least partly reverses IL-1β-induced anorexia [
60,
61]. However, there are conflicting reports in the literature about whether brain endothelial or perivascular cells are the source of vascular prostaglandin production following IL-1β stimulation [
55,
57,
62]. Nitric oxide is another possible mediator of endothelial-microglia signaling, because BMEC produce nitric oxide in response to inflammatory stimuli (including IL-1β), and treating primary microglia cultures with a nitric oxide donor increases IL-1β production [
63,
64]. However, blocking nitric oxide production did not prevent luminal BMEC IL-1β treatment from inducing SIM-A9
Il1b synthesis. Finally, examining the role of direct communication between endothelial cells and astrocytes, as well as the involvement of other brain cell populations (e.g., neurons, pericytes, and oligodendrocytes) in propagating local inflammatory responses in the CNS is a worthy subject for future investigation.