Although abundant reactive microglia are found associated with β-amyloid (Aβ) plaques in Alzheimer's disease (AD) brains, their contribution to cell loss remains speculative. A variety of studies have documented the ability of Aβ fibrils to directly stimulate microglia in vitro to assume a neurotoxic phenotype characterized by secretion of a plethora of proinflammatory molecules. Collectively, these data suggest that activated microglia play a direct role in contributing to neuron death in AD rather than simply a role in clearance after plaque deposition. Although it is clear the Aβ-stimulated microglia acutely secrete toxic oxidizing species, the identity of longer-lived neurotoxic agents remains less defined. We used Aβ-stimulated conditioned media from primary mouse microglia to identify more stable neurotoxic secretions. The NMDA receptor antagonists memantine and 2-amino-5-phosphopetanoic acid as well as soluble tumor necrosis factor α (TNFα) receptor protect neurons from microglial-conditioned media-dependent death, implicating the excitatory neurotransmitter glutamate and the proinflammatory cytokine TNFα as effectors of microglial-stimulated death. Neuron death occurs in an oxidative damage-dependent manner, requiring activity of inducible nitric oxide synthase. Toxicity results from coincident stimulation of the TNFα and NMDA receptors, because stimulations of either alone are insufficient to initiate cell death. These findings suggest the hypothesis that AD brains provide the appropriate microglial-mediated inflammatory environment for TNFα and glutamate to synergistically stimulate toxic activation of their respective signaling pathways in neurons as a contributing mechanism of cell death.
Alzheimer's disease (AD) is characterized by an accumulation of extracellular deposits of β-amyloid (Aβ) peptide containing senile plaques and abundant neurofibrillary tangles in the brain (Braak and Braak, 1997). There is extensive evidence that amyloid deposition provokes a microglial-mediated inflammatory response that contributes significantly to the cell loss and cognitive decline that is characteristic of this disease (Akiyama et al., 2000). Amyloid-dependent activation of microglia in vitro results in acquisition of a reactive phenotype and secretion of proinflammatory products, including reactive oxygen species, cytokines, and neurotoxins. However, the identity of the neurotoxic agent(s) generated by reactive microglia experimentally or in AD has not been resolved (Giulian et al., 1995; Ii et al., 1996; Bianca et al., 1999; Combs et al., 2001; Kingham and Pocock, 2001; Xie et al., 2002).
A number of in vitro studies have demonstrated that Aβ fibril stimulation increases microglial tumor necrosis factor α (TNFα) production (Klegeris et al., 1997; Galimberti et al., 1999; Tan et al., 2000; Combs et al., 2001). Although TNFα has been described as a neuronal anti-apoptotic agent (Barger et al., 1995; Gary et al., 1998; Glazner and Mattson, 2000), it is also capable of inducing neuronal death when presented to neurons as part of a glial-derived inflammatory milieu (Gelbard et al., 1993; Chao and Hu, 1994; Venters et al., 1999; Combs et al., 2001). The excitatory neurotransmitter, glutamate, is another proinflammatory product secreted by microglia in response to Aβ fibril stimulation (Klegeris et al., 1997; Noda et al., 1999; Ikezu et al., 2003). It is known that NMDA receptor (NMDAR)-expressing neurons are vulnerable to AD-associated loss, supporting a hypothesis of excitotoxic NMDA receptor activity-mediated death (Greenamyre and Young, 1989; Francis et al., 1993). Encouraging clinical trial data from Alzheimer's patients treated with the noncompetitive NMDA receptor antagonist memantine also supports the idea that NMDA receptor activity is required for disease progression (Jain, 2000; Marx, 2000; Reisberg et al., 2003; Rive et al., 2004; Tariot et al., 2004). Although the effectors of neuron loss remain unclear, data suggest that death may include oxidative damage-associated mechanisms as evidenced by increased neuronal immunoreactivity for inducible nitric oxide synthase (iNOS) and the peroxynitrite marker 3-nitrotyrosine (Vodovotz et al., 1996; Smith et al., 1997; Hensley et al., 1998; Heneka et al., 2001).
In an effort to link Aβ fibril deposition with microglial activation and subsequent neuron death, we used an in vitro approach that used Aβ fibril-stimulated primary mouse microglia and cortical neuron cultures. Our results demonstrate that Aβ-activated microglia secrete TNFα and glutamate to synergistically stimulate neuronal iNOS expression and subsequent peroxynitrite production leading to cell death. Cell death requires stimulation with both effectors, because neither alone was able to induce neuron loss. These findings suggest that nontoxic activation of the NMDA receptor can lead to neuronal death in the presence of the appropriate, potentiating, inflammatory stimuli.
Materials and Methods
Materials. The anti-iNOS (used for Western blot analysis), anti-NMDAR1, anti-TNF receptor I (TNFRI), and anti-TNFRII antibodies and protein A-agarose were from Santa Cruz Biotechnology (Santa Cruz, CA). The anti-βIII tubulin antibody was purchased from Upstate Biotechnology (Charlottesville, VA). Affinity-purified horseradish peroxidase-conjugated secondary antibodies were obtained from Santa Cruz Biotechnology. TNFα and soluble TNFR1 were purchased from R & D Systems (Minneapolis, MN). Memantine, anti-iNOS antibody (used for immunostaining), anti-nitrotyrosine antibody, 3-morpholino-sydnonimine N-ethylcarbamide (SIN-1) hydrochloride, the selective non-NMDA receptor antagonist 2,3-dihydroxy-6-nitro-7-sulfonyl-benzo[f]quinoxaline (NBQX), and 1400W.2HCl were from Alexis Biochemicals (Carlsbad, CA). Mn(III)tetrakis(4-benzoic acid) porphyrin (MnTBAP) was purchased from Biomol (Plymouth, PA). The 2-amino-5-phosphopetanoic acid (APV), NMDA, glutamate, lipopolysaccharide (LPS), and all other materials not specified were purchased from Sigma (St. Louis, MO).
Aβ fibril preparation. Peptides corresponding to amino acids 1-42 of human Aβ protein were purchased from American Peptide (Sunnyvale, CA). To fibrillize, Aβ peptides were resuspended in sterile dH20 (final concentration, 1 mm) followed by incubation for 1 week at 37°C.
Tissue culture. Neurons were cultured from cortices of embryonic day 16 (E16) mice (C57BL/6). Briefly, meninges-free cortices were isolated, trypsinized, and plated onto poly-l-lysine (0.05 mg/ml)-coated tissue culture wells (260 cells/mm2) for 7 d in vitro before use. Neurons were grown in Neurobasal media with l-glutamine and B27 supplements (Invitrogen, Rockville, MD) to consistently provide neuronal cultures >95% pure and to survive for at least 1 month in vitro. Culture purity was determined by cell counting after staining with the astrocytic marker anti-glial fibrillary acidic protein and the neuronal marker antimicrotubule-associated protein 2 (MAP2). Microglia were derived from postnatal day 1 (P1) mouse brains (C57BL/6). Briefly, meninges-free cortices from P1 mice were isolated and trypsinized. Cells were plated onto tissue culture plastic in DMEM-F-12 with l-glutamine (Invitrogen) containing 10% heat-inactivated FBS and 5% heat-inactivated horse serum and fed every third day. After ∼14 d, the cultures were shaken vigorously (30 min; 120 rpm on a rotary shaker) to remove microglia.
Cell stimulation. Microglia were stimulated by removing them from their respective media and plating onto immobilized Aβ1-42 (400 cells/mm2; 48 h) fibrils (48 pmol/mm2). Immobilized peptides were prepared by coating tissue culture wells with nitrocellulose and allowing fibrillar peptides to dry onto the wells. The use of immobilized peptides prevents their subsequent collection in the conditioned medium. After 48 h of stimulation, microglial-conditioned media were collected, clarified by centrifugation, and added to neuronal cultures with or without vehicle or specific drugs/reagents. Media from wells containing immobilized Aβ peptide alone were collected as a negative control. Neurons were stimulated for 0-72 h. Neurons were fixed and stained using a neuron-specific anti-MAP2 antibody, and a counting grid was placed over the wells to count neuron numbers from four identical fields for each condition. The average number of neurons (±SEM) was calculated for each condition. As neurons die, they retract processes and detach from the dish and are thus removed from viable counts. MAP2-positive cells with visible nuclei and immunostained processes, at least one cell diameter in length, are counted as neurons. Counting-based cell-survival quantitation reliably mirrors other viability measures such as lactate dehydrogenase release or MTT reduction in our hands. To eliminate the confound of contaminating cells contributing to total lactate dehydrogenase release or MTT reduction, we routinely assessed cell survival via counting-based quantitation. Each experiment was performed in quadruplicate three to four times. Values statistically different from controls were calculated using one-way ANOVA. The Tukey-Kramer multiple-comparisons posttest was used to determine p values.
Quantitation of secreted TNFα. Conditioned media were taken from microglia after 48 h of Aβ (48 pmol/mm2) stimulation, and concentrations of secreted TNFα were determined using commercially available mouse TNFα colorimetric sandwich ELISA plates (R & D Systems).
Western blotting and immunoprecipitation. Neurons were stimulated as described above and then lysed in ice-cold radioimmunoprecipitation assay (RIPA) buffer (1% Triton X-100, 0.1% SDS, 0.5% deoxycholate, 20 mm Tris, pH 7.4, 150 mm NaCl, 10 mm NaF, 1 mm Na3VO4, 1 mm EDTA, 1 mm EGTA, and 0.2 m PMSF) and sonicated, and insoluble material was removed by centrifugation (14,000 rpm; 10 min; 4°C). Human brain samples were frozen in liquid nitrogen, pulverized and lysed in RIPA, sonicated, and spun as above. Protein concentrations were quantitated using the method by Bradford (1976). Proteins were resolved by 10% SDS-PAGE and incubated in primary antibody overnight at 4°C, followed by incubation in 2° horseradish peroxidase-conjugated antibody. Antibody binding was detected via enhanced chemiluminescence (Pierce, Rockford, IL).
Immunoprecipitations were performed by incubation of aliquots of the cellular lysates or brain homogenates with the primary antibody (1 μg of antibody/mg of protein lysate) overnight at 4°C and with protein A-agarose for 2 h at 4°C. The immunoprecipitates were washed three times in RIPA buffer and then resolved by 10% SDS-PAGE and Western blotted as described.
Immunohistochemistry. To perform culture immunohistochemistry, neurons were stimulated as described above and fixed in 4% paraformaldehyde at 37°C for 30 min at select times. To assess survival, neurons were stained with an anti-MAP2 antibody (Sigma) using Vector VIP (Vector Laboratories, Burlingame, CA) as the chromagen. Culture immunohistochemistry visualizing NMDAR1, TNFRII, and TNFRI was performed using commercial primary antibodies listed above. Antibody binding was visualized using FITC or Texas Red-conjugated 2° antibodies (Santa Cruz Biotechnology) or Vector VIP. Double-labeling immunocytochemistry visualizing neuronal iNOS was performed using a Zeiss (Thornwood, NY) LSM 510 Meta confocal microscope. Neurons were double labeled using anti-MAP2 antibody and anti-iNOS antibody (Alexis Biochemicals). Sections were mounted in ProLong Gold antifade reagent with 4′,6′-diamidino-2-phenylindole (DAPI) (Molecular Probes, Eugene, OR) for imaging.
Nitrite/nitrate assay. To quantitate nitric oxide synthase activity, neurons were removed to antioxidant-free Neurobasal media with B27 supplements (Invitrogen) during stimulations. Media were collected after 72 h of stimulation, and nitrite/nitrate concentrations were calculated according to manufacturer specifications (Stressgen Biotechnologies, San Diego, CA). To account for cell death in TNFα/NMDA-treated cultures, nitrite/nitrate values were normalized against counted, immunostained neuron number per well. The experiments were performed three times with six repeats per condition and then averaged (±SEM). Values statistically different from controls were calculated using one-way ANOVA. The Tukey-Kramer multiple-comparisons posttest was used to determine p values.
Glutamate quantitation assay. To quantitate glutamate secreted by Aβ-stimulated microglia, conditioned media were generated as described above (Cell stimulation). Glutamate concentrations in the collected, clarified media were determined according to manufacturer specification using the commercial glutamate quantitation reagents, Amplex Red Glutamic Acid/Glutamate Oxidase Assay kit (Molecular Probes). The experiments were performed three times with eight repeats per condition and then averaged (±SEM). Values statistically different from controls were calculated using one-way ANOVA. The Tukey-Kramer multiple-comparisons posttest was used to determine p values.
Human tissue. Age-matched midtemporal gyrus AD and control brain tissue was obtained from Dr. Joseph Rogers (Alzheimer's Disease Core Center, Sun Health Research Institute, Sun City, AZ). Human tissue use was reviewed and approved by the University of North Dakota Institutional Review Board (project IRB-200403-255).
Animal care and use. All procedures were reviewed and approved by the University of North Dakota Institutional Animal Care and Use Committee (protocol 0012-3). Mice were housed at the Center for Biomedical Research on a 12 h light/dark cycle and were allowed food and water ad libitum. Pregnant mice were killed via CO2 asphyxiation, and embryos were killed via decapitation.
Proinflammatory secretions from Aβ-stimulated microglia stimulate neurotoxicity via stimulation of TNFα and NMDA receptors
We and others have documented that primary mouse microglia and the human THP-1 monocyte cell line respond to fibrillar Aβ peptides by activation of a complex tyrosine kinase-based signaling cascade leading to production of neurotoxic secretory products (McDonald et al., 1997; Combs et al., 1999; Yates et al., 2000). Previous efforts identified the proinflammatory cytokine TNFα as one factor necessary for microglial-conditioned media-dependent death. To begin identifying additional toxic components of the conditioned media, we first considered the possibility that activated microglia secrete glutamate or other NMDA receptor ligands (Giulian et al., 1995; Noda et al., 1999; Guillemin and Brew, 2002; Ikezu et al., 2003). For confirmation, we demonstrated that conditioned media from Aβ-stimulated microglia were potently toxic to mouse cortical neuron cultures (Fig. 1A). As expected, Aβ stimulated microglial TNFα secretion (Fig. 1B), and addition of soluble TNFRI (p55) to conditioned media provided complete neuroprotection from Aβ-stimulated microglial-conditioned media (Fig. 1C). These results extend our previous conclusions using TNFα neutralizing antibodies in the conditioned media (Combs et al., 2001). Importantly, the toxic action of the Aβ-stimulated microglial-conditioned media is dependent on microglial secretory products and not contaminating β-amyloid fibrils (Fig. 1D).
To determine whether NMDA receptor ligands were an additional conditioned media component required for toxicity, we assessed the ability of the activated microglia to secrete glutamate. A number of previous reports have documented the ability of activated microglia to secrete this excitatory neurotransmitter (Barger and Basile, 2001; Ikezu et al., 2003; Nakamura et al., 2003). Aβ stimulation resulted in a significant increase of glutamate secretion into the conditioned media by microglia (Fig. 2A). To verify the contribution of secreted glutamate to condtioned media-stimulated death, the noncompetitive NMDA receptor antagonist memantine or the competitive NMDA receptor antagonist APV was added to the media after addition to the neurons (Davies et al., 1981; Bormann, 1989). Both reagents provided complete protection demonstrating that NMDA receptor activation is a component of the conditioned media-stimulated death (Fig. 2B). Consistent with this observation, the competitive quisqualate receptor antagonist NBQX offered no protection from the conditioned media-mediated toxicity (Sheardown et al., 1990) (Fig. 2C). Nevertheless, to avoid activation of multiple glutamate receptor subtypes, subsequent experiments typically used NMDA instead of glutamate.
Recombinant TNFα and NMDA induce neurotoxicity similar to conditioned media from Aβ-stimulated microglia
The conditioned media experiments identified both glutamate and TNFα as components required for neurotoxicity. To determine whether these stimuli alone were sufficient to induce conditioned media-dependent type death, neurons were treated with mouse recombinant TNFα and NMDA or glutamate. Both NMDA (Fig. 3A) and glutamate (Fig. 3B) induced neuron death in a dose-dependent manner when added to neurons in the presence of TNFα. Importantly, stimulation with TNFα alone had no affect on viability (Fig. 3A). Doses of glutamate and NMDA needed to induce the synergistic death with TNFα were sufficiently low, to prevent direct excitotoxicity, and consistent with our quantitated values from conditioned media. Neuron loss induced by microglial-conditioned media or TNFα and NMDA displayed a prolonged time course occurring between 24 and 72 h, suggesting a similar mechanism (Fig. 4A). Importantly, a transient 1 h stimulation with conditioned media or TNFα and NMDA was sufficient to induce neuron loss by 72 h, demonstrating that neuronal death was rapidly initiated despite the delayed time course of loss (Fig. 4B).
Simultaneous TNFα and NMDA stimulation induces neuron death in a manner requiring inducible nitric oxide synthase activity and peroxynitrite formation
We previously demonstrated that microglial-conditioned media induce neuron death in a manner requiring increased expression and activity of iNOS (Combs et al., 2001). Increased iNOS activity can facilitate the reaction of nitric oxide with superoxide anion to form the strong oxidizing and protein tyrosine nitrating agent peroxynitrite (Beckman et al., 1992; Ischiropoulos et al., 1992). Our previous work demonstrated that increased iNOS activity in conditioned media-treated cultures stimulated increased peroxynitrite formation as determined by increased levels of protein nitration (Combs et al., 2001). To confirm that TNFα and NMDA receptor stimulation induced death via an identical mechanism, we compared the abilities of the iNOS-selective inhibitor 1400W.2HCl and the superoxide dismutase mimetic and peroxynitrite scavenger MnTBAP to protect neurons from conditioned media (Fig. 5A) and synergistic TNFα and NMDA (Fig. 5B) stimulation (Garvey et al., 1997; Zingarelli et al., 1997; Klann, 1998; Day et al., 1999). Addition of 1400W as well as MnT-BAP was sufficient to protect neurons from TNFα and NMDA-dependent death.
Correspondingly, TNFα and NMDA treatment stimulated increased iNOS protein levels and activity (Fig. 6A,B) as well as increased protein nitration (Fig. 7A) within treated neurons. Importantly, treatment of neuronal cultures with the peroxynitrite donor SIN-1 produced a qualitatively similar increase in protein nitrotyrosine levels (Fig. 7B) (Hogg et al., 1992). These data verified that TNFα and NMDA-mediated death shares mechanism with conditioned media-dependent death involving increased iNOS expression and activity. A mechanism of iNOS/peroxynitrite-mediated death in AD is correlatively supported by the observation of increased iNOS protein levels and protein tyrosine nitration in AD brains compared with age-matched controls (Figs. 6C, 7C).
Although neuronal iNOS immunoreactivity has been reported previously (Heneka et al., 2003; Grange-Messent et al., 2004; Rodrigo et al., 2004; Small et al., 2004), its expression is certainly well documented in glia as well (Xie et al., 2002; Dehmer et al., 2004; Iravani et al., 2004; Morale et al., 2004; Wang et al., 2004). To confirm that our iNOS-mediated death was occurring via neuronally generated nitric oxide, we performed immunohistochemistry to determine which cell types in our culture were expressing iNOS. It is difficult to eliminate completely all contaminating cell types from the cultures, and the presence of contaminating astrocytes or microglia could have been responsible for our observed iNOS effects. Double labeling of our cultures after treatment with TNFα and NMDA demonstrated a robust increase in iNOS immunolabeling that was localized within MAP2-expressing neuronal cells (Fig. 8). These data confirmed that the stimulation was responsible for increasing protein levels of neuronal iNOS.
Specific neurons coexpress NMDA and TNF receptors in vitro
NMDA receptors are heteromers comprised of multiple subunits, and receptor activity properties are determined primarily by the composition of specific NMDAR2 isoforms (Monyer et al., 1992; Zhang et al., 1994; Boeckman and Aizenman, 1996; Monaghan and Larsen, 1997). Similarly, the TNFα-dependent signaling response is dependent on the specific receptor isoform, TNFRI (p55) or TNFRII (p75), involved (Natoli et al., 1998). To characterize NMDA and TNFα receptor isoform expression in our neuron cultures, Western blot analysis was performed at 7 d in vitro before stimulation. Western blot analysis demonstrated neuronal expression of p55 TNFRI and low-level expression of p75 TNFRII along with the ubiquitous R1 subunit of the NMDA receptor (Fig. 9A). Immunohistochemical analysis verified that neurons display heterogeneous levels of NMDAR1 immunoreactivity that colocalizes with TNFR1 and TNFRII immunoreactivity (Fig. 9B). These results support our conclusion that synergistic stimulation of these receptors is responsible for a cross-talk response leading to neuron death.
Our results indicate that Aβ-stimulated microglia secrete TNFα and glutamate to synergistically induce neuron death requiring iNOS activity and peroxynitrite production. Cell death required stimulation with both ligands, because neither alone at the doses used was sufficient to induce neuron loss. Based on our findings, we propose the hypothesis that Aβ-activated microglia secrete TNFα and glutamate to initiate a signaling cross-talk response capable of converting two innocuous stimuli into one promoting oxidative damage-associated neuron death. It is encouraging that other microglial-mediated inflammatory neurodegenerative paradigms, ventricular LPS infusion and hippocampal Aβ injection, have been shown previously to involve NMDA receptor-dependent neuron death (Willard et al., 2000; Miguel-Hidalgo et al., 2002). Additionally, it has been reported that microglial activation in vitro with secreted amyloid precursor protein or LPS also results in neuron death in a manner involving NMDA receptor activity as well as nitric oxide synthesis (Kim and Ko, 1998; Barger and Basile, 2001). Collectively, these results suggest that microglial control of extracellular glutamate concentrations is an important proinflammatory response in addition to classically described cytokine and reactive oxygen/nitrogen species secretion (Rimaniol et al., 2000).
Although we propose a role for TNFα potentiation of NMDA receptor-dependent death, previous studies have characterized an ability of TNFα to provide neuroprotection (Barger et al., 1995; Houzen et al., 1997; Mattson et al., 1997; Carlson et al., 1999; Kaltschmidt et al., 1999; Keller et al., 1999; Glazner et al., 2000; Diem et al., 2001; Viel et al., 2001; Bruce-Marchetti et al., 2004). Importantly, some of these neuroprotective paradigms have demonstrated TNFα protection against NMDA-mediated death in apparent contrast to our findings (Houzen et al., 1997; Furukawa and Mattson, 1998). One possibility for this discrepancy could be a result of differences in expression levels of NMDA or TNFα receptors. We and others have observed that expression of NMDA and TNFα receptors is dependent on culture age, culture density, and the presence of contaminating glia (Daniels and Brown, 2001; Church and Hewett, 2003) (data not shown). Previous culture protocols contained an apparently higher number of glia than ours, and the culture age and neuronal plating density used also differ (Houzen et al., 1997; Furukawa and Mattson, 1998). In this study, we sought to determine direct effects of TNFα and NMDA receptor stimulation on neurons rather than determine cell-cell mechanisms invoked in mixed cultures. For this reason, we used a nearly pure population of neurons at a defined age and density to make our observations. We have shown direct neuronal expression of NMDA receptors colocalizing with both TNFRI and TNFRII at 7 d in vitro, supporting our hypothesis that we are observing direct effects of TNFα and NMDA on neurons. Therefore, our system allows future opportunity to dissect neuronal intracellular signaling cross talk between the cytokine and NMDA receptors. Certainly, the in vitro nature of our system implies the possibility that both NMDA and TNFα receptors may operate differently in vivo. However, the concentrations of NMDA used are within typical in vitro experimental concentrations, and TNFα concentrations are in the physiologic range (Floyd and Krueger, 1997; Furukawa and Mattson, 1998).
We noted that subpopulations of neurons in our cultures colocalize TNFRI or TNFRII with the NMDAR1 subunit of the NMDA receptor. It will be important to determine whether a specific NMDA receptor subunit or TNFα receptor isoform expression/subcellular colocalization identifies neurons for death. NMDA receptor subunit composition appears to regulate susceptibility to NMDA-dependent death (Sinor et al., 2000). Additionally, previous in vivo work has demonstrated TNFRI and TNFRII mediating neurotoxic and neurotrophic stimulations, respectively (Yang et al., 2002). Identifying the TNFα receptor-mediating signaling cross talk with a specific subunit containing NMDA receptor may provide mechanisms to explain the well documented paradoxical role of this cytokine in vivo in contributing to both neurodegeneration and neurorepair during inflammation-associated conditions (Stoll et al., 2000; Saha and Pahan, 2003).
The final effector of the iNOS-dependent death we describe is likely peroxynitrite formation and oxidation of target molecules, including DNA and proteins (Radi et al., 1991; King et al., 1993). Our demonstration of a typical peroxynitrite decomposition ratio, larger concentrations of nitrate (NO-3) than nitrite (NO-2), in the neuronal media suggests the generation of peroxynitrite (Pfeiffer et al., 1997). The oxidative capacity of this molecule is assumed mostly to be responsible for its neurotoxicity. Indeed, our neuroprotection provided by selective iNOS inhibitors as well as radical scavenging reagents supports this idea. We chose to limit our study to characterize toxicity because of relatively long-lived species in the microglial-conditioned media. This allowed us to describe neuronal production of oxidizing/nitrating species in response to a defined stimulation of microglial-secreted TNFα and glutamate. However, it is probable that several cell types as well as additional molecules contribute to neuronal death in vivo during similar inflammatory events. For example, astrocytes and microglia are also very effective at generating damaging oxidizing/nitrating species (Xie et al., 2002; Dehmer et al., 2004; Iravani et al., 2004; Morale et al., 2004; Wang et al., 2004). In vivo or during glia-neuron coculture paradigms, Aβ-activated glia can certainly generate labile oxidizing species that contribute acutely to the resultant neuronal death (Xie et al., 2002). Therefore, to completely determine specific cellular generation and contribution of oxidizing species to neuronal death, future experiments will require in vivo disease modeling.
The coincident stimulation-induced neuron death identified from our microglial-mediated inflammation paradigm can help to understand a potential mechanism of neuron death occurring in AD brains. It has been suggested from both in vivo histological studies as well as in vitro work that Aβ plaque-associated microglia directly contribute to neuron loss in AD via release of a plethora of neurotoxic products in a number of different paradigms (Akiyama et al., 2000). We identified a specific mechanism by which two secretory products from Aβ-activated microglia can directly lead to oxidative damage-dependent death. We predict that neurons that coexpress specific TNFα and NMDA receptors in human adult brain are susceptible to the particular inflammation-associated death we describe. More importantly, the specific cellular localization of TNFα receptor subtypes with certain NMDA receptor subunits may identify vulnerable neuronal populations in the AD brain. In correlative support of this hypothesis, it is known that levels of TNFα and neuronal TNFRI expression in the AD brain are elevated (Fillit et al., 1991; Bruunsgaard et al., 1999; Tarkowski et al., 1999; Zhao et al., 2003), and NMDA receptor-expressing neurons represent a population vulnerable for loss during disease (Greenamyre and Young, 1989; Francis et al., 1993). Additionally, neurons in the AD brain display increased immunoreactivity for both iNOS and tyrosine-nitrated proteins similar to our in vitro observations (Figs. 6, 8) (Vodovotz et al., 1996; Smith et al., 1997; Hensley et al., 1998). The microglial-mediated death mechanism we observed may provide specific molecular targets amenable to targeting for neuroprotective therapeutic design. For example, the success of the noncompetitive NMDA receptor antagonist, memantine, for treatment of AD patients supports the idea that NMDA receptor activity is required for disease progression (Jain, 2000; Marx, 2000; Reisberg et al., 2003; Rive et al., 2004; Tariot et al., 2004). Finally, it is conceivable that the inflammation-associated death we characterized may be mechanistically important for neuron loss in a variety of inflammation-associated neurodegenerative conditions besides AD.
This work was supported by National Institutes of Health (NIH)-National Institute on Aging Grants 1 RO3 AG20294-01 and 1 R15 AG16192-01 and by NIH-National Center for Research Resources Grant 1 P20 RR17699-01. We thank Tami Cassavan for expert technical assistance.
Correspondence should be addressed to Dr. Colin K. Combs, Assistant Professor, Department of Pharmacology, Physiology, and Therapeutics, University of North Dakota, School of Medicine and Health Sciences, 501 North Columbia Road, Room 3730, Grand Forks, ND 58202. E-mail:.
Copyright © 2005 Society for Neuroscience 0270-6474/05/252566-10$15.00/0