SIRT1 Protects against Microglia-dependent Amyloid-β Toxicity through Inhibiting NF-κB Signaling*

Accumulating evidence suggests that neurodegeneration induced by pathogenic proteins depends on contributions from surrounding glia. Here we demonstrate that NF-κB signaling in microglia is critically involved in neuronal death induced by amyloid-β (Aβ) peptides, which are widely presumed to cause Alzheimer disease. Constitutive inhibition of NF-κB signaling in microglia by expression of the nondegradable IκBα superrepressor blocked neurotoxicity, indicating a pivotal role for microglial NF-κB signaling in mediating Aβ toxicity. Stimulation of microglia with Aβ increased acetylation of RelA/p65 at lysine 310, which regulates the NF-κB pathway. Overexpression of SIRT1 deacetylase and the addition of the SIRT1 agonist resveratrol markedly reduced NF-κB signaling stimulated by Aβ and had strong neuroprotective effects. Our results support a glial loop hypothesis by demonstrating a critical role for microglial NF-κB signaling in Aβ-dependent neurodegeneration. They also implicate SIRT1 in this pathway and highlight the therapeutic potential of resveratrol and other sirtuin-activating compounds in Alzheimer disease.

Neurodegenerative diseases appear to be caused by pathogenic proteins that affect neurons directly or contribute to neuronal death by engaging neurotoxic pathways in surrounding glia (1)(2)(3). In Alzheimer disease (AD), 3 neurodegeneration may be exacerbated by chronic inflammatory reactions of cells surrounding neuritic plaques, including microglia and astrocytes (4,5). High concentrations of fibrillar A␤ can activate microglia, resulting in tumor necrosis factor-␣-dependent expression of inducible nitric-oxide synthase (iNOS) and neuronal apoptosis (6). Nonfibrillar A␤, which may be the major pathogenic form of A␤ in the early stages of AD, also stimulates microglia to induce neurodegeneration. Dimeric and trimeric assemblies of A␤-(1-42) isolated from amyloid deposits elicited profound neurotoxicity in hippocampal neurons but only in the presence of microglia (7). Stimulation with soluble A␤ caused microglia to secrete toxic factors, including cathepsin B, and mediated neurodegeneration (8). Inhibiting the induction of long term potentiation with soluble A␤ involves activation of microglia and stimulation of iNOS and superoxide (9).
We hypothesized that the pathogenic engagement of microglia by A␤ involves activation of NF-B, a transcription factor that mediates immune and inflammatory responses (10) and controls the expression of both iNOS and cathepsin B (11,12). In AD brains, RelA/p65 immunoreactivity is greater in neurons and astrocytes surrounding amyloid plaques, raising the possibility of a role for NF-B in AD pathogenesis (13). In cultured neurons and glia, A␤ stimulation led to NF-B activation (12)(13)(14)(15). However, it remains unclear whether NF-B signaling actually contributes to AD-related neurodegeneration.
To test our hypothesis, we took advantage of the fact that NF-B activation is tightly regulated by inhibitory proteins, such as IB␣ (16). In response to stimuli, IB␣ is degraded to release the NF-B p50/RelA heterodimer, which rapidly translocates to the nucleus and activates the transcription of NF-B-inducible target genes (17,18). Replacing two critical N-terminal serines (Ser 32 and Ser 36 ) with alanines results in the IB␣ superrepressor (IB␣-SR), which effectively suppresses NF-B signaling (19). We used this superrepressor to suppress NF-B signaling specifically in microglia.
NF-B signaling is also modulated by post-translational modifications, including reversible acetylation of the RelA/p65 subunit (20). For example, full transcriptional activity of RelA/p65 requires acetylation of Lys 310 , which can be deacetylated by SIRT1, a class III histone deacetyltransferase (21). Activation of SIRT1 by resveratrol, a potent small molecule agonist (22), inhibits NF-B signaling by promoting deacetylation of Lys 310 of RelA/p65 (21).
In this study, we assessed the role of NF-B signaling in microgliamediated A␤ toxicity in mixed cortical neuronal/glial cultures. We also determined whether agents that enhance RelA/p65 deacetylation protect against A␤-dependent neurodegeneration.

EXPERIMENTAL PROCEDURES
Primary Neuronal Culture-Cortices were isolated from Sprague-Dawley rat pups (Charles River Laboratories, Wilmington, MA) on postnatal day 0 or 1. To establish mixed cortical cultures, cells were plated at 160,000 cells/ml in plating medium containing Dulbecco's modified Eagle's medium, 10% fetal bovine serum, 0.5 mM Glutamax, 100 units/ml penicillin, and 100 g/ml streptomycin. After 6 days, the medium was replaced with Neurobasal A medium supplemented with N2 (NBA/N2). All treatments were conducted on day 7 in NBA/N2. Tissue culture supplies were purchased from Invitrogen unless stated otherwise.
Lyophilized A␤ powder was reconstituted immediately before use in dry dimethyl sulfoxide at 2.2 M and diluted in Dulbecco's modified Eagle's medium/F-12 medium at 40 M. The nonfibrillar A␤ solutions were further diluted in NBA/N2 medium to 10 -20 M. Atomic force microscopy revealed no fibrillar structures in this type of A␤ preparation (data not shown). To eliminate microglia in mixed cortical cultures, cultures were treated with either Ac-LDL-SAP (Advanced Targeting Systems, San Diego, CA) at 3 g/ml for 18 h or 5 mM leucine methyl ester (LME) (Sigma) for 12 h. These agents selectively kill microglia in mixed cultures. To assess the effects of resveratrol (EMD Biosciences, San Diego, CA) on NF-B activation and A␤ toxicity, primary cultures were treated initially with resveratrol for 60 min and then with resveratrolcontaining A␤ solution (final concentrations: resveratrol, 30 -50 M; A␤, 10 M).
To inhibit NF-B signaling, we generated lentiviral vectors expressing an IB␣-SR-EGFP fusion protein under the control of the ubiquitin-C promoter (Lenti-IB␣-SR) or IB␣-SR-His under the control of the promoter of the macrophage colony-stimulating factor (MCSF) gene (Lenti-MCSF-IB␣-SR). The IB␣-SR-EGFP construct (47) was generated by cloning IB␣ (SS/AA) cDNA (HindIII/XbaI) into pEGFP-C3 (HindIII/XbaI). After isolation by PCR using a high fidelity polymerase (Roche Applied Science), the IB␣-SR-EGFP construct was inserted into the AscI and BstEII sites of the modified Lenti-EGFP vector, also known as FUGW (48), to generate Lenti-IB␣-SR. To generate Lenti-MCSF-EGFP, the MCSF promoter was isolated from pSK-cFms (a gift from Dr. Toru Miyazaki, Friedrich Miescher Institute, Basel, Switzerland, and Dr. Stella E. Tsirka, SUNY, Stony Brook, NY) (49) to replace the ubiquitin promoter of Lenti-EGFP. To generate Lenti-MCSF-IB␣-SR, IB␣-SR-His cDNA was isolated by PCR with high fidelity polymerase to replace the EGFP gene (BamHI/EcoRI) of Lenti-MCSF-EGFP. Appropriate construction was verified by sequencing the ligation junctions. The expression of the fusion protein was confirmed by Western blots with anti-IB␣ and anti-EGFP antibodies (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). The FUGW vector was a kind gift from Dr. David Baltimore (California Institute of Technology, Pasadena, CA).
For expression of SIRT1, a lentiviral vector expressing human SIRT1 was generated (Lenti-SIRT1). The human SIRT1 cDNA (American Type Culture Collection, Manassas, VA) was isolated by PCR using a high fidelity polymerase (Roche Applied Science) and cloned into the AscI and BstEII sites of the modified FUGW vector to generate Lenti-SIRT1.
Recombinant lentivirus was generated by cotransfecting the transfer vector with two helper plasmids, delta8.9 (packaging vector) and VSV-G (envelope vector), into 293T cells (American Type Culture Collection) and purified by ultracentrifugation. Viral titers were deter-mined with p24 enzyme-linked immunosorbent assays (Perkin Elmer, Boston, MA) at the Gladstone-UCSF Laboratory of Clinical Virology.
Immunocytochemistry and Cell Death Quantification-Cultures were fixed in 4% paraformaldehyde in phosphate-buffered saline (135 mM NaCl, 2.7 mM KCl, 43 mM Na 2 HPO 4 , 14 mM KH 2 PO 4 , pH 7.4) for 30 min at room temperature. After permeabilization in phosphate-buffered saline with 0.1% Triton for 10 min, cells were placed in blocking buffer (phosphate-buffered saline with 10% fetal bovine serum and 0.01% Triton) for 30 min. The primary antibodies to MAP2 (1:500; Chemicon, Temecula, CA) and GFAP (1:1000; DAKO, Carpinteria, CA) were applied in this blocking buffer for 2 h at room temperature or overnight at 4°C and visualized with anti-mouse (MAP2) or anti-rabbit (GFAP) conjugated with fluorescein isothiocyanate or Texas Red (Vector Laboratories, Burlingame, CA). For immunostaining of microglia with CD11b, similar procedures were conducted under nonpermeabilizing conditions by omitting Triton in all solutions. In some cases, microglia were labeled with tomato lectin (1:500; Vector Laboratories) in blocking buffer (phosphate-buffered saline with 5% normal goat serum) overnight at 4°C and visualized with Texas Red NeutrAvidin BrdUrd Incorporation-Cell proliferation was measured with Cell Proliferation enzyme-linked immunosorbent assays (Roche Applied Science) as recommended by the manufacturer. Briefly, cells were prelabeled with 10 M BrdUrd for 12-18 h, fixed, and incubated with anti-BrdUrd-POD. Substrate was added, and proliferation was quantified with an automated plate reader (Molecular Devices, Sunnyvale, CA).
Statistical Analysis-Statistical analyses were carried out with Graphpad Prism (San Diego, CA). Unless otherwise indicated, differences among means were evaluated by analysis of variance, followed by the Tukey-Kramer post hoc test, as indicated. Unpaired two-tailed Student's t test was used to evaluate differences between two means, as indicated. The null hypothesis was rejected at the 0.05 level.

Microglia Mediate A␤ Toxicity in Mixed Primary Cortical Cultures-
but not A␤-(42-1), induced a significant increase in BrdUrd incorporation in mixed cultures. BrdUrd incorporation was quantified with a cell proliferation enzyme-linked immunosorbent assay (Roche Applied Science) and expressed relative to nontreated controls. The bars represent means Ϯ S.E. of three independent experiments conducted in triplicate. **, p Ͻ 0.01. *, p Ͻ 0.05 (unpaired Student's t test). F, A␤-induced proliferation of microglia quantified by counting CD11b-positive cells in 20 -40 random fields (ϫ400). ***, p Ͻ 0.001 (unpaired Student's t test). Similar results were obtained from two additional independent experiments (not shown). G, treatment with acetylated LDL-saporin (Ac-LDL-SAP) or LME increased the percentage of MAP2-positive neurons surviving the A␤ treatment. Values are means Ϯ S.E. of six (Ac-LDL-SAP) and two (LME) independent experiments. ***, p Ͻ 0.001 (Tukey-Kramer test).
high molecular mass species (Fig. 1B). Atomic force microscopy and electron microscopy revealed no fibril structures in the preparations (data not shown). After 24 and 48 h in culture, the fraction of high molecular mass A␤ increased slightly (Fig. 1B, arrow).
To assess whether microglia are required for A␤-(1-42)-induced neurotoxicity, we first eliminated microglia by treating the mixed cultures with 3 g/ml acetylated low density lipoprotein saporin (Ac-LDL-SAP), a toxin conjugated to a ligand (Ac-LDL) targeting a microgliaspecific scavenger receptor. This treatment eliminated over 80% of CD11b-positive microglia and had little effect on the viability of neurons and astrocytes in the absence of A␤ (data not shown), consistent with previous studies (7,23). However, this reduction in the number of microglia markedly attenuated neuronal death after exposure to A␤-(1-42) (Fig. 1G). In addition, microglia were depleted by treatment with 5 mM LME, a compound that causes lysosome-mediated cell death primarily of microglia (24,25). Incubation with LME for 12 h eliminated  DECEMBER 2, 2005 • VOLUME 280 • NUMBER 48 about 50% of CD11b-positive microglia without affecting other cell types (data not shown). This treatment also significantly decreased A␤-induced neurotoxicity (Fig. 1G). Therefore, the marked neurotoxicity induced by nonfibrillar A␤ in mixed cortical cultures is dependent on the presence of microglia.

NF-B Signaling in Microglia Mediates A␤ Toxicity
A␤ Activates NF-B Signaling in Glia-To investigate the role of NF-B signaling in microglia-dependent A␤ toxicity, we first determined whether nonfibrillar A␤ activates NF-B signaling in primary microglia. A lentiviral vector expressing dEGFP under the control of the 5ϫ B enhancer element (Lenti-B-dEGFP; Fig. 2A) was used to assess NF-B signaling in specific cell types. With its short half-life (1-2 h), the dEGFP gene is suitable for detecting the inducible activation of NF-B (26). BAY11-7082, a specific inhibitor of NF-B signaling, inhibited LPS-induced dEGFP expression in microglial BV2 cells infected with Lenti-B-dEGFP, confirming that the dEGFP expression under the control of B enhancer elements was NF-B-dependent (Fig. 2B). In pri- mary microglia infected with Lenti-B-dEGFP, low levels of EGFP expression were detected in the absence of stimuli. Exposure to A␤-(1-42) significantly increased EGFP expression, indicating activation of NF-B signaling by A␤ in primary microglia (Fig. 2C).
A␤-(1-42) also activated NF-B signaling in mixed cortical cultures infected with Lenti-B-dEGFP. In the absence of A␤ treatment, these cultures showed low levels of dEGFP expression and intact neurons with normal MAP2 immunostaining (Fig. 2D). Treatment with A␤ markedly increased EGFP expression in nonneuronal cells and significantly decreased the survival of MAP2-positive neurons (Fig. 2D). EGFP expression was colocalized with CD11b and GFAP immunoreactivity but not with MAP2 (Fig. 2, D and E), suggesting that nonfibrillar A␤ stimulates NF-B activation in both microglia and astrocytes but not in neurons. A␤-stimulated NF-B activation was further confirmed with the increased expression of NF-B target genes, iNOS (Fig. 2F) and IB␣ (Fig. 2G), quantified with real time RT-PCR.
Inhibition of NF-B Signaling Protects against Microglia-dependent A␤ Toxicity-To determine whether NF-B activation plays an important role in microglia-dependent A␤ toxicity, we inhibited NF-B signaling in cortical cultures by infecting them with lentiviral vectors expressing the IB␣ superrepressor (IB␣-SR) fused to the C terminus of EGFP under the control of the ubiquitin promoter (Lenti-IB␣-SR). Immunostaining with anti-EGFP and anti-MAP2 (Fig. 3A) revealed expression of the IB␣-SR-EGFP fusion protein in both neuronal (arrow) and nonneuronal cells. Compared with cultures infected with Lenti-B-dEGFP alone, cultures doubly infected with Lenti-B-dEGFP and Lenti-IB␣-SR had markedly suppressed A␤-stimulated dEGFP expression (Fig. 3, B and D). In addition, more MAP2-positive neurons survived A␤-  treatment in cultures expressing IB␣-SR (Fig. 3, C and E). These results indicate that NF-B activation plays a critical role in mediating A␤ toxicity in mixed cortical cultures.
To determine whether NF-B signaling specifically in microglia is responsible for the toxicity of nonfibrillar A␤ in those cultures, we restricted expression of IB␣-SR to microglia with the microglia-specific MCSF promoter (Lenti-MCSF-IB␣-SR). In cultures infected with the control lentiviral vector expressing EGFP (Lenti-MCSF-EGFP), the MCSF promoter-directed EGFP expression was localized primarily in tomato lectin-labeled microglia and not in MAP2-positive neurons or GFAP-positive astrocytes (Fig. 4A). Expression of IB␣-SR in microglia significantly increased the survival of MAP2-positive neurons (Fig. 4B) in a dose-dependent manner (Fig. 4C). These results indicate that

NF-B signaling in microglia plays a pivotal role in A␤-dependent neurodegeneration in mixed cortical cultures.
SIRT1 Overexpression and Resveratrol Inhibit A␤-stimulated NF-B Signaling and Protect against A␤ Toxicity-Regulation of NF-B signaling involves modulation of the acetylation status of its RelA subunit by p300 and CREB-binding protein acetyltransferases or by HDAC3 and SIRT1 deacetylase (20,21,27). To further probe the molecular mechanisms underlying NF-B activation in A␤-stimulated mixed cortical cultures, agents promoting deacetylation of RelA/p65 were used to modulate NF-B signaling. The effects of SIRT1 overexpression on NF-B signaling were assessed by counting EGFP-expressing cells in cultures infected with Lenti-B-dEGFP in the presence or absence of Lenti-SIRT1. Western blots confirmed that infection of Lenti-SIRT1 resulted in elevated levels of SIRT1 (supplemental Fig. 1). In cultures infected with Lenti-B-dEGFP alone (control), A␤ induced a significant increase in the number of dEGFP-expressing cells, most of which were glia (Fig. 5, A and B). Overexpression of SIRT1, however, markedly reduced A␤-stimulated EGFP expression, suggesting that SIRT1 deacetylase blocked the up-regulation of NF-B signaling by A␤-(1-42) (Fig. 5, A and B).
The addition of resveratrol, a polyphenol compound that activates SIRT1, also blocked A␤-induced dEGFP expression, indicating that resveratrol strongly attenuates A␤-stimulated NF-B signaling in glial cells (Fig. 5, A and B). We further explored this effect in microglial BV2 cells; treatment with LPS for 3 or 6 h drastically increased the levels of iNOS mRNA, a well established NF-B target gene. In the presence of 30 M resveratrol, the levels of LPS-induced iNOS mRNA were markedly lower (Fig. 5C). This result confirms the inhibitory effect of resveratrol on NF-B signaling in microglia.
Full transcriptional activity of RelA/p65 requires acetylation of Lys 310 (20), a known target of the SIRT1 deacetylase (21). Indeed, using an antibody specific for acetylated Lys 310 (50), we found that A␤-(1-42)induced NF-B activation was associated with increased levels of acety-lated Lys 310 (Fig. 5D). Treatment with resveratrol diminished the levels of RelA/p65 with acetylated Lys 310 , but not of total RelA/p65 (Fig. 5D). Consistent with a previous study (21), these findings suggest that resveratrol inhibits NF-B signaling, possibly by promoting deacetylation of RelA/p65 on Lys 310 .
The profound inhibitory effects of SIRT1 overexpression on NF-B signaling were associated with effective protection against A␤ toxicity (Fig. 6A). Resveratrol also increased the number of surviving MAP2positive neurons in A␤-treated cultures in a dose-dependent manner (Fig. 6B). To exclude the possibility that resveratrol acted directly on neurons to exert its protective function, we examined the effects of resveratrol in cultures depleted of microglia by pretreatment with Ac-LDL-SAP. Resveratrol afforded no significant protection in these cultures (Fig. 6C), indicating that this effect depends on the presence of microglia. Indeed, in cortical neuronal cultures that contain less than 5% glial cells, resveratrol showed no protective effects against neurotoxicity induced by A␤-(1-42) (supplemental Fig. 2). Moreover, in cultures in which NF-B signaling is constitutively inactivated, resveratrol also had no additional effect on the survival of MAP2-positive neurons (Fig. 6D). Our results suggest that resveratrol protects against A␤ toxicity in mixed cortical cultures by inhibiting microglial NF-B signaling through activation of SIRT1.

DISCUSSION
This study supports the glial loop hypothesis by demonstrating a causal link between NF-B signaling in microglia and neurotoxicity of A␤. Targeted inhibition of NF-B signaling in microglia attenuated A␤ toxicity in mixed cortical cultures, suggesting that microglial NF-B signaling may be critical in mediating the toxic effects of AD-related inflammatory responses. Our results also showed that reversible acetylation is an important mechanism in modulating A␤-stimulated NF-B signaling. Inhibition of NF-B signaling by SIRT1 deacetylase and its agonist resveratrol was strongly neuroprotective. Thus, stimulating the deacetylation of NF-B with resveratrol or other sirtuin-activating compounds may have therapeutic value in AD.
In mixed cortical cultures, microglia-dependent A␤ toxicity was associated with strong activation of NF-B signaling in both microglia and astrocytes. However, targeted inactivation of NF-B signaling in microglia with IB␣-SR was sufficient to block A␤ toxicity, indicating that microglial NF-B signaling played a pivotal role in this pathological process. Our previous studies showed that cathepsin B, an NF-B target gene (11), is a critical factor in microglia-mediated A␤ toxicity (8). In the current study, we investigated whether cathespin B was also directly responsible for A␤ toxicity in mixed cultures. Inhibition of cathepsin B by CA074, a specific inhibitor of cathepsin B (8), reduced A␤ toxicity significantly (Fig. 7A). Moreover, CA074 exerted no additional protection in cultures in which microglia were eliminated (Fig. 7A). These results provide additional support for an essential role for cathepsin B in microglia-mediated A␤ toxicity. Other NF-B target genes, such as iNOS, have also been identified as mediators of microglia-dependent toxicity induced by nonfibrillar A␤ (9,12). Based on these findings, we hypothesize that nonfibrillar A␤ engages a pathogenic microglial loop in the early stages of AD. This loop involves the NF-B-mediated activation of target genes, including cathepsin B and iNOS, which are directly responsible for causing neuronal injury (Fig. 7B).
How might nonfibrillar A␤-(1-42) activate NF-B signaling in microglia? Previous studies showed that A␤-(1-40) interacts with tumor necrosis factor-␣ receptor I on the neuronal surface to induce NF-B nuclear translocation in these cells (28). Another receptor, the receptor for advanced glycation end products (51), also binds to A␤ peptides and activates NF-B signaling directly or indirectly (52). It remains to be determined which receptors nonfibrillar A␤-  interacts with on the surface of microglia to activate NF-B signaling. Microglial NF-B signaling could be further stimulated by factors released from injured neurons, resulting in a vicious cycle exacerbating microglia-dependent neurotoxicity. NF-B activation in astrocytes might help maintain or counteract this vicious cycle (Fig. 7B). Although several components of this hypothetical framework remain to be tested experimentally, our findings strongly suggest that inhibiting the activation of NF-B or of NF-B-dependent microglial gene products such as cathepsin B could block this pathogenic cascade and increase neuronal survival in AD.
Our results also suggest that neuronal NF-B signaling is unlikely to play an important role in microglia-mediated A␤ toxicity. In fact, no constitutive and A␤-stimulated NF-B signaling was observed in neurons in the culture conditions used in this study. In contrast, incubation of mixed cultures with a high concentration (50 mM) of KCl increased NF-B-dependent dEGFP expression in neurons (data not shown). In a similar vein, KCl and other factors that can promote NF-B activation may account for the NF-B activation observed in cerebellar granule neurons and hippocampal neurons in previous studies (28 -30). The presence of high concentrations of KCl might have enhanced NF-B activation in cerebellar granule cells (31), and retinoic acid, an NF-Bstimulating factor and a component of the commonly used B27 supplement, might have enhanced NF-B signaling in cultured hippocampal neurons (28 -30).
Besides the absence of such NF-B-stimulating additives in our cultures, the design of our reporter construct may have also helped focus our detection on A␤-induced NF-B activation. NF-B-independent expression was minimized by including only the B enhancer elements and the TATA box without any constitutive promoter, such as the SV40 promoter, which was used in another reporter construct (32). Furthermore, we used a short lived dEGFP to minimize the accumulation of reporter gene expression activated by base-line NF-B signaling. Last, the strong protective effect of microglia-specific IB␣-SR expression also argues against a significant role of neuronal NF-B signaling in our cultures.
In probing the molecular mechanisms underlying NF-B activation, we identified a neuroprotective approach that involves deacetylation of RelA/p65. A␤-stimulated NF-B signaling is modulated by acetylation of RelA/p65, where the acetylation status of specific lysine residues affects both the DNA binding ability and transcriptional activity of the protein (21,27). Acetylation at five lysine residues (Lys 122 , Lys 123 , Lys 218 , Lys 221 , and Lys 310 ) is associated with NF-B activation. SIRT1 deacetylates Lys 310 of RelA/p65 without affecting the acetylation status of other lysine residues (21). Consistent with this finding, we showed that acti- vation of NF-B signaling by A␤ was associated with acetylation of Lys 310 , which was deacetylated by treatment with resveratrol. However, it is not yet clear whether acetylation of the four other lysine residues is also involved in A␤-stimulated NF-B activation.
Our study indicates that SIRT1 and resveratrol protect neurons against microglia-dependent A␤ toxicity by inhibiting NF-B in microglia. However, SIRT1-dependent RelA/p65 deacetylation sensitized cancer cell lines to apoptosis induced by tumor necrosis factor-␣ (21). Whereas up-modulation of NF-B-regulated gene products in neurons may protect these cells against cytokines, excitotoxic insults, and oxidative stress (15,33,34), up-modulation of those gene products in microglia may result in the secretion of neurotoxic factors that mediate neurodegeneration. We hypothesize that microglial NF-B activation prevails in AD.
In considering SIRT1 stimulation as a potential therapeutic strategy in AD and related conditions, it is important to note that SIRT1 deacetylates other nonhistone substrates besides NF-B, including p53, FOXO, and others (35)(36)(37). SIRT1-dependent deacetylation of p53 and FOXO has been associated with the attenuation of p53-or FOXO-dependent transcription and apoptosis (35)(36)(37). SIRT1 is also involved in the attenuation of axonal degeneration (38) and polyglutamine repeat-induced neurodegeneration (39), although the exact molecular pathways remain undefined.
In our culture model, SIRT1/resveratrol clearly prevented microgliadependent A␤ toxicity. Although we cannot exclude the possibility that SIRT1/resveratrol had some direct effects on neurons, three lines of evidence suggest that such effects played only a minor, if any, role in the neuroprotection we observed. First, A␤ toxicity in mixed cultures depended specifically on microglial NF-B activation, which was inhibited by SIRT1/resveratrol. Second, resveratrol had no protective effects in pure neuronal cultures exposed to nonfibrillar A␤. Third, resveratrol had no protective effects in A␤-treated mixed cultures after microglia were depleted or NF-B signaling was inhibited, suggesting that the primary neuroprotective mechanisms of resveratrol involved both microglia and NF-B signaling.
In numerous epidemiological studies, anti-inflammatory drugs, such as nonsteroidal anti-inflammatory drugs, were associated with a lower risk of AD, suggesting a preventive role (40 -42). In clinical trials, however, some nonsteroidal anti-inflammatory drugs did not slow cognitive decline in patients with mild to moderate AD (43). Effective targeting of microglial activation in AD may require an even earlier or more specific intervention (42,44,45). Indeed, to preserve the beneficial functions of glial cells and to more selectively suppress their detrimental responses, it will probably be necessary to identify and block specific pathogenic pathways in specific glial cells. Our study identified NF-B signaling in microglia as a critical pathogenic pathway and SIRT1 as a potential target for blocking this pathway by drug treatment. Our findings highlight the potential therapeutic value of resveratrol and other sirtuinactivating compounds in protecting against neuronal loss in AD and related conditions.