The first two authors contributed equally to this work.
Abstract : In recent years inflammatory mechanisms have become increasingly appreciated as important steps in the Alzheimer's pathogenic pathway. There is accumulating evidence that amyloid β-peptide (Aβ), the peptide product of the cleavage of amyloid precursor protein, may promote or exacerbate local inflammation by stimulating glial cells to release immune mediators. In addition, clinical studies using nonsteroidal antiinflammatory drugs have found a reduced risk for Alzheimer's disease with their use. Here we show that the neurotoxic Aβ, a major plaque component, and lipopolysaccharides (LPS), an immune reaction-triggering portion of bacterial membranes, are both potent activators of the nuclear transcription factor NF-κB in primary rat astroglial cells. The activation was found to be concentration- and time-dependent and could be attenuated in the presence of NF-κB decoy nucleotides. The pretreatment by either 17β-estradiol (1-10 μg) or sodium salicylate (3-30 mM) reduced the Aβ (LPS)-induced activation of NF-κB by 48 (50%) and 60% (50%) of activated levels, respectively. In addition, 17β-estradiol (10 μM) and sodium salicylate (10 mM) were able to attenuate the increase in interleukin-1β levels following exposure to 25 μM Aβ. Our data suggest that the aberrant gene expression is at least in part due to Aβ-induced activation of NF-κB, a potent immediatearly transcriptional regulator of numerous proinflammatory genes ; this event takes place in astroglial cells. The results of our experiments provide a further understanding of the effects of estrogen and aspirin on astroglial cells exposed to Aβ and LPS.
The nuclear transcription factor NF-κB was originally identified as a DNA-binding protein interacting with the consensus sequence GGGACTTTCC in the mouse Ig κ light-chain enhancer, thereby inducing transcription of the κ light-chain gene (Sen and Baltimore, 1986). NF-κB resides in the cytoplasm as homo- or heterodimers of a family of structurally related proteins, p65, c-Rel, RELB, p50/p105, and p52/p100 (May and Ghosh, 1998). Although transcriptionally active homodimers of both p50 and p65 can form, the p50/p65 heterodimer is preferentially formed in most cell types. In the absence of stimulatory signals, the NF-κB heterodimer is retained in the cytoplasm by its physical association with an inhibitory phosphoprotein, IκB (Stancovski and Baltimore, 1997). Signals that induce NF-κB activity cause the dissociation and subsequent degradation of IκB proteins (Stancovski and Baltimore, 1997), allowing NF-κB dimers to enther the nucleus and induce gene expression (May and Ghosh, 1998). The function of NF-κB can therefore be explained by that of a “third messenger” molecule that transduces upstream signals from the cytoplasm into the nucleus in activated cells.
NF-κB is widely expressed in the nervous system and exists in both neurons and glial cells (Kaltschmidt et al., 1994 ; Lukasiuk et al., 1995 ; Bales et al., 1998). NF-κB has a pivotal role to the expression of many genes involved in mammalian immune and inflammatory responses. It is thought that NF-κB induces rapidly the expression of defense genes in threatening conditions such as viral and bacterial infections or physical stress. In this line, activated NF-κB is a positive regulator of genes whose products mediate the acute-phase response, lymphoproliferation, leukocyte adhesion, chemoattraction of macrophages, B- and T-cell activation, and antiviral response (May and Ghosh, 1998). Indeed, we and others were able to show that mixed glial cells use NF-κB to express cytokines like interleukin (IL)-1, IL-6, IL-8, or other inflammatory or immune-related processes, e.g., VCAM-1, inducible nitric oxide, and M-CSF (Gitter et al., 1995 ; Du Yan et al., 1997 ; Bales et al., 1998). Although the downstream events following NF-κB in the brain are poorly understood, many mostly pathogenic stimuli can activate NF-κB, including amyloid β-peptide (Aβ) (Dodel et al., 1996 ; Kaltschmidt et al., 1997). Aβ, the peptide product of the cleavage of amyloid precursor protein, plays a pivotal role in Alzheimer's disease (Mattson, 1997). Among the broad variety of actions reported following exposure to Aβ, there is accumulating evidence that Aβ may promote or exacerbate local inflammation by stimulating glial cells to release immune mediators (Barger and Harmon, 1997 ; Du Yan et al., 1997 ; Wyss-Coray et al., 1997 a,b). Aβ has been shown to induce release of IL-1, IL-6, IL-8, basic fibroblast growth factor, transforming growth factor-β1 (Wyss-Coray et al., 1997a,b), and complement proteins (Eikelenboom et al., 1989), all substances that have been localized to senile plaques (Rogers et al., 1992a,b ; Gitter et al., 1995). Moreover, in brains of Alzheimer's disease patients, there is a close association of inflammatory mediators and activated glial cells with compact neuritic plaques (Rogers et al., 1992a,b ; McGeer and McGeer, 1995 ; Lue et al., 1996 ; Nilsson et al., 1998). Recently, epidemiological studies on the use of nonsteroidal antiinflammatory drugs (NSAIDs) provide further evidence for inflammatory mechanisms as being important for Alzheimer's disease. These studies have concluded that the intake of NSAIDs significantly decreases the risk of developing Alzheimer's disease (Stewart et al., 1997). Independently, in epidemiological studies where postmenopausal women received an estrogen replacement therapy (Kawas et al., 1997 ; Baldereschi et al., 1998), a similar reduction in the risk to develop Alzheimer's disease was estimated, arguing for a steroid-related pathway comparable to that of NSAIDs. Despite those important observations, the mechanisms by which these chemically and physiologically distinctly acting drugs provide neuroprotection are far from being understood.
In the light of the immune-mediated process following exposure to Aβ, we investigated whether one of these two drugs, NSAIDs or estrogen, is able to reduce activation of NF-κB, a prerequisite for the inflammatory process. Moreover, if this activation is part of a general immunmodulatory pattern, other inflammatory responses should be blocked by 17β-estradiol and/or sodium salicylate. Therefore, we further tested the effects of 17β-estradiol and sodium salicylate following exposure to lipopolysaccharides (LPS), which are known as strong promoters of immune responses in macrophages (Hambleton et al., 1995) and astrocytes (Schumann et al., 1998). In our studies using primary rat mixed glial cultures, we found a concentration- and time-dependent activation of NF-κB following exposure to Aβ and LPS. This activation could be blocked in a concentration-dependent fashion by 17β-estradiol as well as sodium salicylate. In addition, both compounds were able to block IL-1β protein levels in astroglial cells. We conclude that the protective effect by these two distinct compounds may be in part mediated by reducing NF-κB activation and subsequently abolishing harmful gene expression in glial cells.
MATERIALS AND METHODS
Primary fetal rat cortical astroglial cultures were collected and cultured as previously described (Condorelli et al., 1989 ; Bales et al., 1998). In brief, cerebral cortices from embryonic day 18 rat fetuses were dissected and dissociated with trypsin-EDTA (GibcoBRL) for 15 min at 37°C. A single-cell suspension was made by trituration of the cell pellet in DNase I (GibcoBRL). Cells were plated at a density of 0.55 × 106/ml in Neurobasal medium (GibcoBRL) supplemented with 0.5 mM glutamine, 10% dialyzed fetal bovine serum (GibcoBRL), and penicillin-streptomycin. After 48 h, the medium was changed to Dulbecco's modified Eagle's medium high-glucose (GibcoBRL) supplemented with penicillin-streptomycin and 10% dialyzed fetal bovine serum. Primary astrocyte cultures were used after reaching confluence, usually after 14 days in vitro, or following one or two passages. Immunostaining with the astrocyte-specific marker glial fibrillary acidic protein showed that these cultures contained >90% astrocytes, indicating their homogeneity.
Cultures were exposed to freshly prepared Aβ1-40 (lot ZM 605 ; Bachem). Freshly prepared Aβ1-40 was made as a 1 mM stock solution in water and used immediately. The sequences of phosphorothioated decoy oligonucleotides for NF-κB were as follows : antisense, 5′-GGT GGA ATC TCC TGG GTG GAA TCT CCT GGG TGG AAT CTC CTG G-3′ and sense 5′-CCA GGA GAT TCC ACC CAG GAG ATT CCA CCC AGG AGA TTC CAC C-3′ (Lin et al., 1995). Cultures were treated with NF-κB decoy (Research Genetics), 17α-estradiol (Sigma), 17 β-estradiol (Sigma), and sodium salicylate (Sigma) for 16 h before exposure to Aβ1-40 or LPS for the indicated intervals. Cells were pretreated with ICI 164,384 (1 μM ; Eli Lilly) 16 h before the treatment with 17 β-estradiol.
Electrophoretic mobility shift assay (EMSA)
Nuclear extracts from astrocyte cultures were prepared using high-salt buffer (Schreiber et al., 1989). In brief, cultures were rinsed with cold phosphate-buffered saline and removed by scraping in fresh phosphate-buffered saline. Following centrifugation, the cell pellet was disrupted with 400 μ1 of cold hypotonic buffer [10 mM HEPES (pH 7.9), 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, and 0.5 M phenylmethylsulfonyl fluoride] for 15 min on ice and incubated an additional 5 min after addition of 25 μl of 10% Nonidet P-40. The cytoplasmic protein is contained in the supernatant following centrifugation (2,500 g, 10 min). The nuclear pallet was resuspended in 50 μl of cold buffer [20% glycerol, 20 mM HEPES (pH 7.9), 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride] and rocked at 4°C for 1 h. Protein content determinations were made using a BCA Protein Assay Kit (Pierce). An oligonucleotide corresponding to the DNA binding consensus site for NF-κB (agt tga ggg gac ttt cc agg c ; Trevigen) and that for p-53 (tac aga aca tgt cta agc atg ctg ggg ; Santa Cruz) were endlabeled using [γ32P]ATP and T4 polynucleotide kinase (GibcoBRL). Binding reactions were performed by incubating 2.5-10 μg of nuclear protein with buffer containing 10 mM Tris-Cl (pH 7.9), 50 mM NaCl, 0.5 mM EDTA, 1 mM dithiothreitol, 5% glycerol, 2 μg of poly(dI-dC), and 100,000 cpm of radiolabeled oligonucleotide. Incubation was carried out at room temperature for 30 min. The specificity of binding was demonstrated by addition of unlabeled competitor, a mutant oligonucleotide sequence (agt tga ggC gac ttt ccc agg c), or antibody specific to the carboxy terminus of NF-κB, p65 (SC-372X ; Santa Cruz Biotechnology). Unlabeled competitor and mutant oligonucleotides were added to the above binding reaction, whereas antibody to p65 was added after the 30-min binding reaction and allowed to incubate for an additional 15 min before binding products were separated by gel electrophoresis (5% polyacrylamide and 0.5× TBE). The gel was dried and processed for autoradiography with preflashed film. Autoradiographs were digitized and analyzed using Image-quant version 1.1 for the Macintosh (Molecular Dynamics).
Western blot analysis
Western blot analysis was performed on 10 μg of whole-cell extracts from treated and untreated cultures. Whole-cell extracts were prepared by lysing cells in a buffer containing 1% Nonidet P-40, 0.1% sodium dodecyl sulfate, 50 mM Tris (pH 8.0), 50 mM NaCl, 0.05% deoxycholate, and protease inhibitors (Boehringer Mannheim). Proteins were size-fractionated on a 4-12% polyacrylamide gradient gel (NuPAGE ; NOVEX) and transferred onto nitrocellulose (Hybond N ; Amersham). The blots were then probed with polyclonal antibodies for IL-1β (Chemicon).
NF-κB activation following exposure to Aβ1-40 in rat primary astroglial cultures
Primary rat cortical astroglial cultures were exposed to increasing concentrations of Aβ1-40 for 8 h. NF-κB activation was determined by EMSA. The specificity of NF-κB binding was confirmed by the ability of excess unlabeled κB-specific oligonucleotide to diminish autoradiographic detection of binding signal as well as the inability of a mutant oligonucleotide (G to C substitution) to attenuate binding (data not shown). The constitutive activity of NF-κB in astroglial cultures was low (Kaltschmidt et al., 1994 ; Bales et al., 1998). Following exposure to Aβ1-40, NF-κB was activated in a concentration- and time-dependent fashion. Activation of NF-κB as determined by EMSA was readily seen after 0.5 h and reached maximal levels after 6-8 h (data not shown). The activation of NF-κB in primary rat astroglial cultures was concentration-dependent within a concentration range of 5-25 μM Aβ1-40 when assayed 8 h after exposure (Fig. 1).
Sodium salicylate and 17β-estradiol were added to the culture 16 h before exposure to Aβ1-40. At the concentrations of 17β-estradiol and sodium salicylate used, no toxic effect to astroglial cultures could be observed [cell viability was quantified using fluorescein diacetate/propidium iodide (data not shown ; Du et al., 1997)]. The range of sodium salicylate concentration was similar to the amounts in plasma for optimal antiinflammatory effects in patients with rheumatic diseases [1-3 mM (Famaey and Paulus, 1992)], and the concentration of 17β-estradiol was chosen based on its ability to convey neuroprotection (Behl et al., 1997). A concentration-dependent inhibition against Aβ-induced NF-κB activation was observed in both drugs (Fig. 2) as well as following exposure to NF-κB decoy reduced Aβ-induced NF-κB activation to 42 and 33% at concentrations of 500 nM and 1 μM, respectively (Fig. 3). 17β-Estradiol at a concentration of 10 μM caused a decrease of Aβ-induced NF-κB activation by 46% ; sodium salicylate reduced Aβ-induced activation by 50% at ~10 mM (Fig. 3).
Intracellular IL-1β protein induced by Aβ (25 μM) can be significantly inhibited by 17β-estradiol (10 μM) and sodium salicylate (10 mM). The estrogen receptors were not involved because 17α-estradiol acted like 17β-estradiol with the same concentration and an estrogen receptor-specific antagonist, ICI 164, 384 was not able to block 17β-estradiol's action (Fig. 4A).
NF-κB activation following exposure to LPS in rat primary astroglial cultures
LPS, a highly conserved component of the outer membrane of gram-negative bacteria, are well known to stimulate peripheral macrophages and glial cells to release various cytokines and eicosanoid mediators of the immune response. In the next set of experiments, we investigated the effect of LPS on NF-κB in rat primary astroglial cultures. In peripheral macrophages, several reports have shown a strong induction of NF-κB binding following LPS. In our setting of cultured cortical glial cells, LPS induced NF-κB activation in a concentration-dependent pattern in the concentration range from 10 to 100 ng/ml (Fig. 1). At a concentration of 100 ng/ml LPS induced a marked NF-κB activation starting at 1 h and reaching ~3.5-fold over control levels at 6 h (Fig. 1). In the presence of sodium salicylate as well as 17β-estradiol a decrease of the NF-κB activation pattern could be observed that was concentration-dependent (Figs. 2 and 5). At a concentration of 20 mM sodium salicylate LPS-induced NF-κB activation was reduced to ~42% of that in astroglial cultures treated with LPS (100 ng/ml) alone. Similarly, 17β-estradiol at concentrations of 1 and 10 μM caused a decrease of NF-κB activation by ~20 and 50%, respectively. The effect was statistically significant at a test level of p < 0.05 compared with the control.
To confirm that the effects elicited by Aβ1-40 and LPS are specific for NF-κB (Fig. 3). In a concentration-dependent pattern NF-κB decoy decreased NF-κB binding in the EMSA following exposure to Aβ1-40 and LPS. NF-κB decoy with the antisense sequence was incapable of reducing Aβ1-40- or LPS-induced activation (data not shown).
The results of our study demonstrate that cultured primary cortical mixed glial cells respond robustly to exposure of Aβ1-40 and LPS with a concentration-dependent increase in NF-κB activation. Aβ and LPS appear to be initiators for NF-κB activation. However, as the maximal activation of NF-κB is at 6 h, we hypothesize that other factors may be involved later. Further detailed investigation to address this issue needs to be performed. Pretreatment with 17β-estradiol as well as salicylic acid attenuated the activation of the nuclear transcription factor NF-κB in a concentration-dependent pattern, thus modulating the transcriptional sequelae initiated by NF-κB, which results in an abridged expression of genes involved in immune and inflammatory responses.
The observation that postmenopausal estrogen replacement therapy is associated with a decreased risk of Alzheimer's disease raises the issue through which mechanism(s) estrogen exerts its antineurodegenerative effect. Using neurons, large varieties of mechanisms by which estrogen promotes their survival against Aβ have been proposed (Green et al., 1997), including reduced oxidatives stress (Behl et al., 1995 ; Goodman et al., 1996Gridley et al., 1997), reduced toxicity of Aβ (Green et al., 1996), enhanced nerve cell outgrowth (Brinton et al., 1997), and reduced neuronal generation of Aβ species (Jaffe et al., 1994 ; Xu et al., 1998). In contrast with neurons, however, only minor attention has been drawn to the effect of estrogen on glial cells and their underlying mechanisms, although the glial network is known to be of pivotal importance for the survival of neurons. The results from our experiments introduce a new mechanism by which estrogen may mediate its neuroprotective effect. Our experiments clearly demonstrate that 17β-estradiol interferes in an important step of glial activation and inflammatory/immune response by attenuating NF-κB activation and its resultant IL-1β overexpression following exposure to an external stimulus.
The concentration of 17β-estradiol used by us is 1-10 μM, which indicates that 17β-estradiol may act as an antioxidant (Inestrosa et al., 1998). The data from 17α-estradiol and ICI 164,384 experiments appear to confirm this point. Our data suggest that micromolar concentrations of 17β-estradiol have an antiinflammatory effect. However, whether using such high concentrations of 17β-estradiol is relevant to the estrogen replacement therapy or not needs further investigation. In addition to an antioxidant function, reports showed 17β-estradiol also being neuroprotective in nanomolar concentrations by unknown estrogen receptor-dependent mechanisms (Pike, 1999 ; Singer et al., 1999).
Although only limited data from studies in the brain are available to indicate what kind of cytokines, e.g., II-1, IL-6, or IL-8, chemokines, e.g., ICAM-1, or inflammatory reactions, e.g., inducible NO•, underlie the control of NF-κB, it can be assumed, based on data from experiments with peripheral leukocytes and macrophages, that the release of these substances has a deleterious impact on neurons. In our experimental setting, IL-1β protein levels are increased following Aβ treatment and can be attenuated in the presence of sodium salicylate and 17β-estradiol.
Aspirin and sodium salicylate have recently been shown to have a neuroprotective effect against neurotoxicity elicited by the excitatory amino acid glutamate in rat primary neuronal cultures and hippocampal slices (Grilli et al., 1996). Similar to our results in Aβ-treated glial cells, aspirin or sodium salicylate causes a reduction of NF-κB level, which correlates with the survival of cerebellar granule neurons (Grilli et al., 1996). Although cerebellar granule cells consist mostly of neurons, it is still not clear whether the neuroprotective effect in the presence of aspirin was due to a reduction of NF-κB activation in neurons or glial cells
We thank Dr. Ann E. Kingston (Eli Lilly and Company) for supplying primary cultured glial cells and Dr. Na Yang (Eli Lilly and Company) for supplying ICI 164,384. This work was supported by a grant from Lilly Center for Women's Health.