Beta-amyloid stimulation of inducible nitric-oxide synthase in astrocytes is interleukin-1beta- and tumor necrosis factor-alpha (TNFalpha)-dependent, and involves a TNFalpha receptor-associated factor- and NFkappaB-inducing kinase-dependent signaling mechanism.

In Alzheimer's disease, beta-amyloid (Abeta) plaques are surrounded by activated astrocytes and microglia. A growing body of evidence suggests that these activated glia contribute to neurotoxicity through the induction of inflammatory cytokines such as interleukin (IL)-1beta and tumor necrosis factor-alpha (TNFalpha) and the production of neurotoxic free radicals, mediated in part by the expression of inducible nitric-oxide synthase (iNOS). Here, we address the possibility that Abeta-stimulated iNOS expression might result from an initial induction of IL-1beta and TNFalpha. We find that in Abeta-stimulated astrocyte cultures, IL-1beta and TNFalpha production occur before iNOS production, new protein synthesis is required for increased iNOS mRNA levels, and the IL-1 receptor antagonist IL-1ra can inhibit nitrite accumulation. Likewise, dominant-negative mutants of tumor necrosis factor-alpha receptor-associated factor (TRAF) 6, TRAF2, and NFkappaB-inducing kinase (NIK), intracellular proteins involved in IL-1 and TNFalpha receptor signaling cascades, inhibit Abeta-stimulated iNOS promoter activity. Our data suggest that Abeta stimulation of astrocyte iNOS is mediated in part by IL-1beta and TNFalpha, and involves a TRAF6-, TRAF2-, and NIK-dependent signaling mechanism.

Astrocytes and microglia are glial cells that play a major role in the inflammation observed in Alzheimer's disease (AD) 1 as well as many other neurodegenerative diseases (1)(2)(3)(4)(5). Upon stimulation from various agents or insults such as cytokines or trauma, these cells adopt a reactive phenotype, a morphological hallmark in AD pathology, during which they themselves may produce still more inflammatory cytokines and neurotoxic free radicals. One potentially detrimental free radical is the highly reactive nitrogen species peroxynitrite, which is a derivative of nitric-oxide (NO) production (6). In the brain, astrocytes are able to produce NO during inflammatory conditions by the enzyme inducible nitric-oxide synthase (iNOS or NOS2) (7,8).
Increased levels of iNOS production in astrocytes surrounding amyloid plaques (9), and significant peroxynitrite damage to neurons have been observed in AD brain (10). Peroxynitrite can also induce alterations in synaptosomal membranes of neurons, and is thus a potent source for oxidative stress in AD brain (11). Several lines of similar in vitro evidence also show that ␤-amyloid, the primary plaque component in AD brain, can stimulate NO production from astrocytes (12)(13)(14)(15), and that activated astrocytes can produce NO levels that are detrimental to neurons (12,16).
While the 42-amino acid ␤-amyloid peptide (A␤42) is capable of activating astrocytes in vitro and in vivo, resulting in the activation of NFB and the induction of iNOS (17), the molecular pathway(s) and participants governing astrocyte iNOS induction by A␤42 have not yet been described. In addition to activation of astrocytes, A␤42 can also activate microglia, leading to the production of interleukin (IL)-1␤ (15,18) and tumor necrosis factor (TNF) ␣ (12,19,20). Because IL-1␤ and TNF␣ can themselves stimulate iNOS production in astrocytes (21,22), A␤42 stimulation of iNOS in astrocytes could be a direct effect on the astrocyte or an indirect result of A␤42 stimulation of IL-1␤ and TNF␣ induction in microglia.
IL-1␤ is a critical inflammatory cytokine in AD. Microglia localized to amyloid plaques stain positively for IL-1, and increased numbers of IL-1␤-expressing microglia are associated with AD progression (23). IL-1␤ is also capable of stimulating astrocytes to produce additional pro-inflammatory cytokines such as IL-6, another inflammation marker associated with neurodegeneration (24 -26).
A signal transduction pathway activated by IL-1␤ that ultimately leads to NFB activation has been described (27)(28)(29) and involves the TNF␣ receptor-associated factor-6 (TRAF6). Similarly, TNF␣ activates NFB via TRAF2 (28,30). Both TRAF2 and TRAF6 participation can then lead to the activation of NFB-inducing kinase (NIK), the common mediator of IL-1␤ and TNF␣ activation of NFB (31), and NIK can complex with and activate the IB kinase signalsome complex (␣,-␤,-␥) (32). Activated IB kinase signalsome complex phosphorylates IB, the inhibitor of the transcription factor NFB, directing the inhibitor for proteasome-mediated degradation and allowing NFB to translocate to the nucleus, where it binds to specific promoter response element sequences and stimulates gene transcription (for reviews, see Refs. 33 and 34). Activated NFB has been observed in AD brain (35,36), and we have shown previously that NFB activation is necessary for A␤42stimulated iNOS induction in astrocytes (17).
The molecular events that coordinate the neurotoxic inflammatory response in neurodegenerative diseases such as AD are crucial to understand. To determine if IL-1␤ induction is required for the A␤42 stimulation of iNOS in astrocytes, we used a series of approaches. We examined the temporal pattern of A␤42 stimulation of IL-1␤ and iNOS production in mixed glial cultures, and whether new protein synthesis was required for iNOS induction. The participation of IL-1␤ signal transduction pathways in iNOS activity was blocked at the ligand/receptor interface with the IL-1␤ receptor antagonist (IL-1ra). Finally, we inhibited either IL-1␤-mediated activation of NFB by using the dominant-negative truncated form of TRAF6, or TNF␣mediated activation of NFB by using the dominant-negative truncated form of TRAF2, or both TNF␣-and IL-1␤-mediated activation of NFB by the kinase-inactive form of NIK. We report here that inhibition of IL-1␤ signal transduction (and, to a lesser extent, TNF␣ signal transduction) resulted in decreased A␤42-stimulated nitrite production and iNOS promoter activity, demonstrating that A␤42-stimulated iNOS induction is mediated at least in part by inflammatory cytokine production.

MATERIALS AND METHODS
Cell Culture-Rat primary cortical glial cultures were prepared and maintained as described previously (16). Briefly, cells were grown in ␣MEM supplemented with 10% fetal calf serum (HyClone) and 1% penicillin/streptomycin (Life Technologies, Inc.). At least 24 h prior to stimulation, the serum-containing medium was removed, and cells were washed once with PBS and then incubated in serum-free ␣MEM containing N2 media supplements (Life Technologies, Inc.). Experiments were conducted on no later than tertiary passages of cells. From immunohistochemical analysis, the cell population was determined to be approximately 90 -95% astrocytes and 5-10% microglia. For cycloheximide and IL-1 receptor antagonist studies, inhibitors were added to medium immediately prior to stimulation.
␤-Amyloid (A␤42) Preparation-The 42-amino acid peptide form of ␤-amyloid (A␤42) was obtained from Dr. C. Glabe (University of California Irvine, CA) and was stored as lyophilized powder at Ϫ20°C. Aggregates of A␤42 were prepared at room temperature as described previously (14). A␤42 aggregated under these conditions is a mixture of globular and fibrillar peptide which activates astrocytes and is toxic to neurons (14,37). The absence of any contaminating endotoxin in the A␤42 preps was verified by direct measurement on Pyroplate LAL assay microplates (Associates of Cape Cod, Inc.).
Plasmids-The 7-kilobase pair human iNOS promoter-luciferase reporter construct (iNOS-Luc) has been described previously (38). The NFB reporter construct pNFB-SEAP (CLONTECH) contains four tandem NFB response element repeats upstream of the thymidine kinase basal promoter from the herpes simplex virus, allowing for the expression of the secreted alkaline phosphatase gene (SEAP).
The TRAF expression constructs were made as follows; full-length murine TRAF2 and murine TRAF6 cDNAs cloned into an SR␣ (SV40derived) promoter-driven vector (pME18S-FLAG) were kindly provided by Dr. J. Inoue (University of Tokyo, Tokyo, Japan) and have been described previously (39). These cDNAs were used as templates for subcloning by polymerase chain reaction into the CMV promoter-driven vector pCMV2-FLAG (Sigma). The following primers were used: for dominant-negative TRAF2 (containing the C-terminal TRAF2 region, amino acids 87-501; termed dnTRAF2-C), 5Ј-AAG CTT GCG GCC GCG AAT AGT AGT TCG GCC TTT CCA GAT AAC-3Ј and 5Ј-GGA TCC TCT AGA TGG CTA GAG TCC TGT TAG GTC CAC AAT AGC-3Ј; for dominant-negative TRAF6 (containing the C-terminal TRAF6 region, amino acids 301-531; termed dnTRAF6-C), pME18S-FLAG-TRAF6 was used as a template with the following primers: 5Ј-AAG CTT GCG GCC GCG AAT CCA AAT TAT GAG GAA ACT ATC AAA CAG-3Ј and 5Ј-GGA TCC TCT AGA TGG CTA CAC CCC CGC ATC AGT ACT TCG-3Ј. All polymerase chain reactions used the DNA polymerase PfuTurbo (Stratagene). Amplified products were digested with NotI and XbaI and then subcloned into the NotI/XbaI sites of pCMV2-FLAG.
The constructs pcDNA3-HA and pcDNA3-HA-NIK were kindly provided by Dr. M. Levrero (Università degli Studi di Roma La Sapienza, Rome, Italy). The pcDNA3-HA contains an HA epitope for N-terminal tagging in the expression vector pcDNA3 (Invitrogen); the pcDNA3-HA-NIK contains the full-length human NIK cDNA cloned into pcDNA3-HA (40). Dominant-negative (kinase-inactive) NIK was prepared by site-directed mutagenesis of pcDNA3-HA-NIK to change amino acid residues Lys 429 -Lys 430 to Ala-Ala (QuikChange, Stratagene) using the following primers: 5Ј-GCT TCC AGT GCG CTG TCG CCG CCG TGC GGC TGG AAG TAT TTC-3Ј and 5Ј-GGA ATA CTT CCA GCC GCA CGG CGG CGA CAG CGC ACT GGA AGC-3Ј. The resultant construct, termed dnNIK, was verified by DNA sequence across the mutated site and by restriction analysis.
Plasmids containing cloned partial cDNA fragments of rat IL-1␤, rat iNOS, and rat GAPDH have been described previously (14).
Transfections and Reporter Assays-Plasmid DNA was prepared by endotoxin-free Maxi-Preps (Qiagen) before transfections. Transfection of astrocyte cultures has been described previously (17). Briefly, for all transfections, the cationic lipid reagent Tfx-50 (Promega) at a 1:1 (charge:g of DNA) ratio was used. For transfections using the iNOS-Luc reporter construct, 4 ϫ 10 5 cells/well in a six-well tissue culture plate were transfected with 3.6 total g of DNA/well (2.0 g of expression construct and 1.6 g of reporter construct; in cases of increasing amounts of expression construct, the total amount of DNA transfected was kept constant by using the appropriate control cloning vector). For all transfections using the pNFB-SEAP reporter construct, 5 ϫ 10 4 cells/well in a 24-well tissue-culture plate were transfected with a total of 1 g of DNA/well (0.5 g of expression construct and 0.5 g of reporter construct, total amount of DNA kept constant as described above).
Luciferase assays were performed with the LucLite reagent (Packard) and read on a LumiCount luminometer (Packard) as described previously (17). Chemiluminescent SEAP assays were performed with the Phospha-Light assay reagents (Tropix) by the manufacturer's protocol and read on the same luminometer. Relative light units for both assays were calculated as a percentage of appropriate controls. At least eight replicates were conducted per experiment, and statistical significance (p Ͻ 0.05) was determined by Student's t test.
Northern Blotting-Northerns blots were performed as described previously (14). Briefly, total RNA was isolated from cells using RNeasy spin columns (Qiagen). Equal amounts of total RNA (800 ng/sample) were slot blot-loaded onto Durulon-UV membrane (Stratagene) and immediately UV cross-linked. Partial cDNA probes were 32 P-labeled using the RediPrime labeling kit (Amersham Pharmacia Biotech). Prehybridization and probe hybridization were performed at 68°C using QuikHyb solution (Stratagene) and by the manufacturer's protocol. Washed blots were exposed to Storm phosphorimaging plates (Molecular Dynamics) and densitometry was measured using ImageQuant version 1.1 software (Molecular Dynamics). To verify equal amounts of total RNA per slot, each blot was later stripped and re-probed for rat GAPDH mRNA. IL-1␤ and iNOS mRNA levels were normalized to the GAPDH mRNA levels, and relative increases were calculated from five separate experiments.
Nitrite Determination-NO production was quantitated by measuring the levels of nitrite in the conditioned media using the Griess assay as described previously (16), except that nitrate in the conditioned medium was first reduced to nitrite with nitrate reductase and NADPH (Sigma) for 1 h at 37°C before the assay.
Cytokine ELISAs-For IL-1␤ measurements, 7.5 ϫ 10 4 cells/well were plated in a 48-well tissue culture plate. For TNF␣ measurements, 5 ϫ 10 4 cells/well were plated in a 48-well tissue culture plate. After 36 -48 h of incubation in serum-free ␣MEM containing N2 media supplement, cells were stimulated with either control diluent or 10 M A␤42. Conditioned media from duplicate wells per sample were col-lected at the indicated time points and then stored at Ϫ80°C until the ELISA. Immediately after the last time point was collected, cytokine levels in the conditioned media were measured by ELISA (rat IL-1␤ or rat TNF␣ assay kit, R&D Systems) and performed by the manufacturer's protocol.
Immunofluorescence-Identification of cell types expressing either IL-1␤ or iNOS was done by immunofluorescence as described previously (41) using the same antibodies as for the Western blots. Polyclonal goat anti-rat IL-1␤ antibody and monoclonal mouse anti-macNOS were used at 1/200 dilution. Monoclonal mouse anti-pig GFAP was used at 1/1000 dilution. Secondary fluorochrome-conjugated antibodies donkey antigoat IgG and donkey anti-mouse IgG (Jackson ImmunoResearch) were used at 1/400 dilutions. Images were digitally captured from a Zeiss Axiophot2 microscope using Spot camera (Diagnostic Instruments) and Metamorph software.

A␤42-stimulated Cytokine Production Occurs before iNOS
Production in Vitro-To determine the temporal sequence of events during A␤42 activation of glial cultures, a time-course Western blot profile was used to demonstrate that IL-1␤ protein production (proIL-1␤ form associated with cellular content prior to secretion) occurs before iNOS protein production ( Fig.  1). A␤42-stimulated IL-1␤ production was detected by 3 h, reached a peak by 6 h, and continued to be present through 36 h after stimulation. However, protein production of iNOS was not detected until 6 h after A␤42 stimulation, and iNOS levels steadily increased through 36 h. In cells treated with diluent alone, there was a delayed, minor IL-1␤ production detected at 6 -12 h, but no iNOS protein detected. GFAP levels did not increase at any time point examined, consistent with previous findings in A␤42 stimulation of astrocytes (14). To determine if A␤42 stimulates IL-1␤ (active form) secretion from these cells, an IL-1␤ ELISA was performed on conditioned medium ( Fig. 2A). Similar to the Western blot time-course profile of A␤42-stimulated proIL-1␤ production, active IL-1␤ was secreted into the conditioned medium and detectable by 3 h, and active IL-1␤ levels accumulated through 12 h.
Another cytokine rapidly induced in glia by A␤42 is TNF␣. Because TNF␣ is readily secreted from the cells, an ELISA was performed on cell conditioned media to measure the time course of TNF␣ production from A␤42-stimulated cells. As shown in Fig. 2B, there was a rapid increase in cytokine production upon addition of A␤42, with TNF␣ protein detected as early as 1 h after stimulation and high levels of TNF␣ sustained through 12 h. These time courses demonstrate that IL-1␤ and TNF␣ responses are early, rapid inflammatory re-sponses to A␤42 with protein production occurring within the first hr of exposure to A␤42, and support the possibility that A␤42-stimulated iNOS production may be a response to an initial IL-1␤ and/or TNF␣ production.
Protein Synthesis Inhibitor Blocks A␤42-stimulated iNOS mRNA Levels-If cytokine production is necessary for A␤42stimulated iNOS expression, then we would expect that cells pre-treated with cycloheximide, a protein synthesis inhibitor, should fail to express iNOS when treated with A␤42. Northern blot analysis (Fig. 3) of cells stimulated with 10 M A␤42 for 12 h in the presence of 1 M cycloheximide demonstrates that IL-1␤ mRNA can still be detected strongly, but iNOS mRNA levels are reduced. The Northern blots were stripped and then re-probed with GAPDH for normalization (data not shown). Quantitation of several experiments by phosphorimaging densitometry showed that the A␤42-stimulated iNOS mRNA levels were reduced in the presence of cycloheximide by approximately 50% compared with the mRNA levels seen in the A␤42stimulated cells in the absence of cycloheximide (Fig. 3). In cells treated with cycloheximide and stimulated with 1 mM Bt 2 cAMP, which can directly induce iNOS expression, there was no decrease in iNOS mRNA levels, but in fact a dramatic increase in iNOS mRNA levels (data not shown). These data indicate that the reduction of A␤42-stimulated iNOS mRNA in the presence of cycloheximide is not due to cell toxicity, and suggest that new protein synthesis is important for iNOS induction.
IL-1␤ Receptor Antagonist Decreases the Levels of A␤42-stimulated iNOS and Nitrite Production-As a more specific means of inhibiting the effects of IL-1␤, we stimulated cells with A␤42 for 18 h in the absence or presence of the IL-1 receptor antagonist (IL-1ra). IL-1ra attenuated A␤42-stimulated levels of iNOS protein as determined by Western blot analysis (Fig. 4A), but did not affect GFAP protein levels. Similarly, we used increasing concentrations of IL-1ra and measured the levels of nitrite, the stable metabolite of NO, that accumulated in the conditioned medium. Cells pre-treated for approximately 5 min with increasing concentrations of IL-1ra showed a dose-dependent decrease in nitrite levels in both A␤42-stimulated cells and IL-1␤-stimulated cells (Fig. 4B). Complete inhibition of IL-1␤-stimulated nitrite production was not observed as the concentration of receptor antagonist necessary for such an effect was not reached (IL-1ra concentrations were initially optimized for inhibition of A␤42-stimulated nitrite levels, which require less receptor antagonist than direct IL-1␤ administration). IL-1ra did not inhibit the levels of nitrite produced by 1 mM Bt 2 cAMP-stimulated astrocytes, demonstrating the specificity of the receptor antagonist as cAMP can stimulate iNOS transcription via molecular mechanisms different from IL-1␤ (42,43).

Dominant-negative TRAF Proteins Can Block NFB Activation in
Astrocytes-In many cell types, the IL-1␤ signaling pathway stimulates the activation of NFB via TRAF6, and the TNF␣ signaling pathway stimulates NFB activation via TRAF2. Dominant-negative TRAF6 (N-terminal truncation, leaving only the amino acids from 301 to 531; termed dnTRAF6-C) has been reported to be able to inhibit NFB activation by IL-1␤; similarly, dominant-negative TRAF2 (amino acids 87-501; termed dnTRAF2-C) can inhibit NFB activation by TNF␣ (27). We tested the effects of these dominantnegative constructs on A␤42 stimulation of astrocytes. First, we overexpressed dnTRAF6-C and dnTRAF2-C in the cells and verified that each construct can inhibit NFB activity, as measured by a co-transfected NFB SEAP reporter construct. As expected, dnTRAF2-C was able to inhibit TNF␣ but not IL-1␤ activation of the NFB reporter in astrocytes (Fig. 5A), and conversely, dnTRAF6-C was able to inhibit IL-1␤ but not TNF␣ activation of the NFB reporter (Fig. 5B). Interestingly, it has recently been reported that dnTRAF6-C but not dnTRAF2-C can inhibit LPS-stimulated NFB activity in an endothelial cell line and a monocyte cell line (44), suggesting the importance of TRAF6 in LPS stimulation of NFB.
Dominant-negative TRAF6 Inhibits iNOS Promoter Activation by A␤42-While the participation of TRAF2 and TRAF6 in NFB activation by specific cytokines is now well identified, the involvement of TRAF2 and TRAF6 in stimulation of gene expression is less well studied. Only recently has TRAF6 been implicated in IL-8 promoter activity in macrophages (45). To address the possibility that iNOS promoter activation by A␤42 is mediated by TNF␣ or IL-1␤, we co-transfected cells with the iNOS promoter-luciferase reporter construct and increasing amounts of dnTRAF2-C or dnTRAF6-C. As shown in Fig. 6, increasing amounts of dnTRAF6-C reduced A␤42-stimulated iNOS promoter activity by approximately 80%, and increasing amounts of dnTRAF2-C reduced A␤42-stimulated iNOS promoter activity by approximately 60%, suggesting that IL-1␤, and to some extent TNF␣, are critical mediators of A␤42stimulated iNOS activation. In contrast, dnTRAF2-C and dnTRAF6-C decreased Bt 2 cAMP-stimulated iNOS promoter activity by only ϳ25% (Fig. 6), suggesting only a minor role for IL-1␤ and TNF␣ in Bt 2 cAMP-stimulated iNOS activity.
Because NIK is involved in mediating both IL-1␤ and TNF␣ signaling pathways, an over-expressed kinase-inactive NIK (dominant-negative NIK, termed dnNIK) should inhibit both IL-1␤-and TNF␣-stimulated NFB activation (46). When cells were co-transfected with the iNOS promoter reporter construct and increasing amounts of dnNIK, A␤42-stimulated iNOS activity was inhibited by approximately 80%, whereas the Bt 2 cAMP-stimulated iNOS promoter activity was not significantly reduced (Fig. 7). The dnNIK inhibition curve for A␤42stimulated iNOS promoter activity is similar to the dnTRAF6-C inhibition curve of the iNOS promoter activity, suggesting that TRAF6 (and thus IL-1␤) is the major mediator of A␤42-stimulated iNOS production in these rat glial cells, and that TRAF2 (and thus TNF␣) is a secondary mediator of A␤42-stimulated iNOS production.
IL-1␤ Localizes to Microglia and iNOS Localizes to Astrocytes in A␤42-stimulated Glial Culture-To identify the cell type(s) responsible for IL-1␤ and iNOS production in the rat glial cultures, cells were stimulated with A␤42 for 12 or 36 h and then IL-1␤ and iNOS localization monitored by FITC immunofluorescence. Double labeling for GFAP with Texas Red immunofluorescence was done to allow for astrocyte identification. As shown in Fig. 8A, IL-1␤ localizes to non-astrocytic cells (no overlapping FITC/Texas Red fluorescence). The IL-1␤-positive cells were primarily microglia, as determined by OX42 staining (data not shown). In contrast, as shown in Fig. 8B, iNOS was detected in astrocytes (overlapping FITC/Texas Red fluorescent cells). These data suggest that A␤42 stimulates IL-1␤ production by activated microglia, which in turn can participate in the activation of astrocytes to produce iNOS. DISCUSSION We demonstrate here that in rat glial cultures, the amyloid plaque component A␤42 stimulates IL-1␤ and TNF␣ cytokine production in microglia prior to astrocyte iNOS production, and that the A␤42-stimulated iNOS promoter activity and NO production occur in an IL-1␤-dependent (and to a lesser extent, TNF␣-dependent) manner via a signal transduction pathway that involves TRAF6 and NIK. We not only therefore define a clear linkage among A␤42, inflammatory cytokines, and putative peroxynitrite neurotoxicity, but we also demonstrate for the first time that TRAF6, TRAF2, and NIK can mediate an inflammatory response that is relevant to AD. Because there is an imperative need to understand inflammation in AD, identi- fying each component such as these in a signal transduction pathway that ultimately leads to deleterious effects in the central nervous system such as peroxynitrite-mediated damage to neurons is important from a therapeutic standpoint.
The pro-inflammatory cytokine IL-1 is a critical neurotoxic component to such an inflammatory pathway, and is found in activated microglia localized to amyloid plaques in AD (47). Evidence that IL-1␤ can mediate neuronal injury has been shown recently in transgenic mouse models of IL-1␤-converting enzyme (ICE). Significantly less ischemic brain injury occurs in ICE-knockout mice (48) as well as in dominant-negative ICE transgenic mice (49). Similarly, adenoviral-directed cerebral expression of IL-1ra in mouse brain also resulted in reduced ischemia-induced brain injury (50). A major consequence of IL-1␤ production in the brain is the stimulation of astrocytic iNOS activity. IL-1␤ stimulation of iNOS in rat, mouse, and human astrocytes as well as in other cell types that can participate in inflammation, such as endothelial cells, hepatocytes, and macrophages, has been well characterized (for reviews, see Refs. 51 and 52), and such IL-1␤-stimulated iNOS activity has been suggested to be detrimental in several neurodegenerative disorders including AIDS dementia complex, the murine multiple sclerosis model of experimental autoimmune encephalitis, and AD (53). This detrimental effect of iNOS activity is supported by evidence from knockout mice studies, which demonstrate that under ischemic conditions, there is a delayed reduction of brain injury and neurological deficits in iNOS-null mice (54). Previously, we showed (17) that A␤42, the primary amyloid plaque component in the brain, can activate glia in vitro in a dose-and time-dependent manner, as determined by morphological response and IL-1␤ expression, and that A␤42 stimulates iNOS activity in astrocytes via the transcription factor NFB. Whether this astrocyte iNOS activity is a direct result of A␤42 stimulation of astrocytes or an indirect result, requiring the participation of other glial factors or inflammation mediators such as IL-1␤ had not been defined previously. While studies consistent with ours (12,13,18) have shown that different forms and preparations of A␤42 can stimulate glial cells to produce cytokines or NO, there have been no previous reports to demonstrate and define the linkage between A␤42-stimulated microglial IL-1␤ (or TNF␣) and astrocytic iNOS.
Our results demonstrating that A␤42-stimulated astrocyte iNOS mRNA production requires new protein synthesis is consistent with previous studies using cycloheximide to inhibit LPS-stimulated iNOS mRNA production in fetal hepatocytes (55). Cycloheximide was partially able to reduce A␤42-stimulated iNOS mRNA production. The partial reduction of iNOS mRNA may reflect the fact that the 1 M cycloheximide concentration is not enough to completely inhibit protein synthesis. Higher concentrations of cycloheximide were not used because they proved to be toxic to the primary astrocyte cultures; the 1 M cycloheximide concentration did not promote a strong inflammatory response or toxic response in the astrocytes. A slight increase in iNOS mRNA was detected in cells treated with control buffer alone in the presence of cycloheximide, but this weak stimulation has been seen before in other systems with a similar concentration of cycloheximide (55). Similarly, the observation that A␤42-stimulated nitrite could not be inhibited completely with increasing concentrations of IL-1ra is consistent with previous studies using human fetal astrocytes stimulated with IL-1␤ (56), where the effective concentration of receptor antagonist needed to completely inhibit IL-1␤-stimulated iNOS mRNA production was several orders of magnitude higher than the concentration of stimulus.
The ability of dnTRAF2-C to inhibit TNF␣-but not IL-1␤stimulated NFB activation in cultured astrocytes, and the ability of dnTRAF6-C to inhibit IL-1␤-but not TNF␣-stimulated NFB activation in cultured astrocytes are consistent with previous studies with 293 cells, which demonstrate TRAFspecificity in cytokine stimulation of NFB (27). In addition, we have extended our investigations beyond the NFB response element reporter construct, and using the iNOS promoter reporter construct, we show, as has recently been shown for TRAF2 and the IL-8 promoter (45), that TRAF-and NIK-dependent signaling can also produce direct consequences in models of inflammation, emphasizing the need to characterize further how TRAFs and NIK interact in signal transduction mechanisms.
There are other potential signal transduction pathways activated during inflammation, which may also participate in A␤42-stimulated iNOS production in astrocytes. For example, the p38 MAPK and c-Jun N-terminal kinase/stress-activated protein kinase stress kinase pathways that target the AP-1 transcription factor could contribute to iNOS promoter activity. In mixed glial cultures, it has been reported (57) that p38 MAPK can mediate TNF␣-and IL-1␤-stimulated iNOS activation in astrocytes, based on the observation that the p38 inhibitor SB203580 is able to inhibit cytokine-stimulated iNOS activity and nitrite production. Whether or not p38 MAPK plays a role in A␤42-stimulated IL-1␤ production in microglia or regulates iNOS expression in astrocytes in cooperation with NFB has not been addressed. A recent study showed that p38 MAPK was important in LPS-stimulated IL-1␤ transcription in murine macrophages (58), so it is possible that in a mixed glial culture containing both microglia and astrocytes, suppressing p38 MAPK leads indirectly to iNOS inhibition via an IL-1␤ inhibition in microglia. Indeed, it has recently been demonstrated that activated p38 MAPK can be immunolocalized to microglia that are associated with amyloid plaques in AD brain (59), and amyloid fibrils can activate p38 MAPK in microglia in vitro (60).
Altogether, our data substantiate a model of inflammation observed in AD where A␤42 activates microglia to produce pro-inflammatory cytokines such as IL-1␤ and TNF␣. These cytokines in turn activate surrounding astrocytes, which exacerbate the inflammation with the production of neurotoxic mediators such as iNOS. The resultant peroxynitrite ultimately damages local neurons and contributes to the neurodegeneration observed in AD.