MAPK-activated Protein Kinase 2 Deficiency in Microglia Inhibits Pro-inflammatory Mediator Release and Resultant Neurotoxicity

MAPK-activated protein kinase 2 (MAPKAP K2 or MK2) is one of several kinases directly regulated by p38 MAPK. A role for p38 MAPK in the pathology of Alzheimer disease (AD) has previously been suggested. Here, we provide evidence to suggest that MK2 also plays a role in neuroinflammatory and neurodegenerative pathology of relevance to AD. MK2 activation and expression were increased in lipopolysaccharide (LPS) + interferon γ-stimulated microglial cells, implicating a role for MK2 in eliciting a pro-inflammatory response. Microglia cultured ex vivo from MK2-deficient (MK2–/–) mice demonstrated significant inhibition in release of tumor necrosis factor α, KC (mouse chemokine with highest sequence identity to human GROs and interleukin-8), and macrophage inflammatory protein 1α on stimulation with LPS + interferon γ or amyloid-β peptide (1–42) compared with MK2+/+ wild-type microglia. Consistent with an inhibition in pro-inflammatory mediator release, cortical neurons co-cultured with LPS + interferon γ-stimulated or amyloid-β peptide (1–42)-stimulated MK2–/– microglia were protected from microglial-mediated neuronal cell toxicity. In a transgenic mouse model of AD in which amyloid precursor protein and presenilin-1 harboring familial AD mutations are overexpressed in specific regions of the brain, elevated activation and expression of MK2 correlated with β-amyloid deposition, microglial activation, and up-regulation of tumor necrosis factor α, macrophage inflammatory protein 1α, and KC gene expression in the same brain regions. Our data propose a role for MK2 in AD brain pathology, for which neuroinflammation involving cytokines and chemokines and overt neuronal loss have been documented.

The role of p38 mitogen-activated protein kinase (MAPK) 2 in the regulation of cytokine biosynthesis is now well estab-lished (1,2). MAPK-activated protein kinase 2 (MK2) is one of several kinases that are regulated through direct phosphorylation by p38 MAPK, and MK2 has therefore been a candidate for an effector role in p38 action in the inflammatory response. Studies using MK2 Ϫ/Ϫ mice support this hypothesis, because mice lacking MK2 are resistant to endotoxic shock, largely due to a reduction in serum tumor necrosis factor-␣ (TNF-␣) levels (3). Furthermore, macrophages and spleen cells taken from MK2-deficient animals show inhibition in the release of a range of pro-inflammatory mediators (3,4).
Alzheimer disease (AD) is a neurodegenerative disorder that affects primarily hippocampal and neocortical brain regions resulting in a progressive loss of cognitive and memory function and ultimately dementia. Post-mortem diagnosis of AD is facilitated by the presence of extracellular plaques comprising ␤-amyloid protein (A␤), and neuroinflammation is a persistent pathological hallmark. Microglia, the resident macrophages of the brain, are responsible for eliciting such an immune response in the CNS. These immune cells of monocytic lineage are activated by a wide range of stimuli and release pro-inflammatory mediators that, in turn, induce microglial autoactivation, thereby amplifying the inflammatory response. It has been proposed that elevated levels of ␤-amyloid in AD brain induces microglial activation and consequent release of pro-inflammatory cytokines, chemokines, and other potentially neurotoxic substances (5)(6)(7)(8)(9). Indeed, activated microglia can cause neuronal cell death in vitro (10 -13). The neuroinflammatory changes and microglial activation observed in AD pathology are therefore hypothesized to contribute to neuronal cell loss and associated dementia in this disease, as well as neuronal injury and neurodegeneration in other CNS disorders resulting from trauma, ischemia, or inflammation (14 -16). Modulation of the inflammatory response may retard the progression of AD through a reduction in neurodegeneration caused by chronic activation of microglia (17)(18)(19). Identification of regulators of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed. Tel.: 44-1279-622649; Fax: 44-1279-6222555; E-mail: ainsley.a.culbert@gsk.com. 2 The abbreviations used are: MAPK, mitogen-activated protein kinase; MK2, MAPK-activated protein kinase 2; AD, Alzheimer disease; CNS, central nervous system; LPS, lipopolysaccharide; IFN␥, interferon ␥; TNF-␣, tumor necrosis factor-␣; MIP-1␣, macrophage inflammatory protein 1␣; WT, wild type; A␤, ␤-amyloid protein; FCS, fetal calf serum; LDH, lactate dehydrogenase; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; ELISA, enzyme-linked immunosorbent assay; PBS, phosphatebuffered saline; RT, reverse transcriptase; IL, interleukin; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.
inflammation relevant to AD pathology is therefore of great therapeutic importance. Activation of the p38 MAPK pathway has been linked to inflammatory pathology both in AD and in mouse models of the disease (20 -23). In addition, A␤ activates p38 MAPK in cultured microglial cells (24). Because MK2 is an immediate downstream kinase of p38 MAPK, we hypothesized that MK2 itself might play a role in neuroinflammation and neurodegeneration relevant to AD. Here we have investigated the role of MK2 in microglial cell activation, and the resultant effects on microglial-mediated neuronal cell toxicity. In addition, we have correlated our findings in vitro with a potential role for MK2 in a transgenic mouse model of AD.
Transgenic Mice-TASTPM mice, which are transgenic mice overexpressing the 695-amino acid isoform of human amyloid precursor protein (APP695) harboring the Swedish double familial Alzheimer disease mutation (K670N,M671L) and human presenilin-1 harboring the familial mutation M146V were generated and maintained at GlaxoSmithKline as described in detail previously (25,26). In this double transgenic line, both transgenes are under the control of the Thy-1 regulatory cassette, which directs expression to the cerebral cortex, amygdala, dentate gyrus, CA1-CA3, and the cerebellum. Expression of transgenes results in progressive amyloid deposition and pathology in the brain. For ex vivo tissue analyses, animals were humanely sacrificed by intraperitoneal injection of a lethal dose of pentobarbital sodium (Euthatal, Rhone Merieux, Harlow, UK).
MK2 Ϫ/Ϫ mice were generated in the laboratory of Professor Matthias Gaestel (Germany) and provided to GSK on a mixed 129v ϫ C57BL/6 background as described previously (3). Microglia were cultured ex vivo from wild-type (WT) and MK2 Ϫ/Ϫ mice as described below.
Culture of NTW8 Microglial Cells-Mouse microglial NTW8 cells were cultured as previously described (27). Cells were switched to serum-free medium 2 h prior to cell stimulation with LPS and IFN␥, or A␤  peptide.
In Vitro Kinase Assay-Lysates prepared either from NTW8 microglial cells or cortical CNS tissues from WT and TASTPM mice were assayed for MK2 activity using the MK2 immunoprecipitation kinase assay kit (Upstate Cell Signaling) according to the manufacturer's instructions. In brief, lysates were prepared from NTW8 microglial cells by washing cell monolayers in ice-cold PBS, followed by extraction in the lysis buffer supplied. For mouse cortical tissue, dissected tissue was powdered under liquid nitrogen and protein extracts were prepared as described previously (28). All lysates were clarified by centrifugation, and MK2 was immunoprecipitated from samples containing 1 mg of total cellular protein. Kinase reactions were performed by incubating immunoprecipitated MK2 at 30°C for 45 min with 10 ng of recombinant Hsp27 in the kinase assay buffer provided. Phosphorylation of Hsp27 was used as a measure of MK2 activity; phospho-Hsp27 and immunoprecipitated MK2 was resolved by SDS-PAGE followed by Western blot analysis. Blots were developed using ECL and quantified where shown by densitometry.
Quantitative RT-PCR-Total RNA was isolated from CNS tissues from WT and TASTPM mice, rat primary CNS cells, and NTW8 microglial cells using TRIzol reagent according to the manufacturer's instructions. First strand cDNA syntheses from equal quantities of RNA and aliquoting of the resulting cDNA products for subsequent parallel TaqMan PCR were all performed as described in detail previously (29). Additional reactions were performed using genomic DNA to produce a standard curve relating threshold cycle to template copy number. Primer (F and R) and probe (P) sets were designed from and the quencher dye, 6-carboxytetramethylrhodamine at the 3Ј end.
Immunohistochemistry Staining for ␤-Amyloid Protein and Activated Microglia-Brains from TASTPM mice were hemisected in the sagittal plane and immersion-fixed in 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS), pH 7.4, for 48 h. The brain was then processed into paraffin wax and 7-m sections were cut and subjected to standard immunohistochemical techniques. Briefly, the sections were de-waxed, re-hydrated through a series of graded alcohols, and washed in distilled water followed by PBS. To enhance antigenicity for phosphotyrosine and A␤, sections were treated by microwaving in 0.01 M citrate buffer, pH 6.0 (2 ϫ 5 min at 300 watts), followed by immersion in 80% formic acid for 8 min. Sections were then incubated in 0.3% H 2 O 2 in PBS for 30 min at room temperature to quench endogenous peroxidase activity. To enable labeling of both A␤ plaques and microglia, sections were incubated overnight at 4°C with both the rabbit polyclonal G30 (3 g/ml) and the mouse monoclonal to phosphotyrosine (1:1000) diluted in primary layer diluent (0.3% Triton X-100, 0.01% sodium azide, and 2% normal serum in PBS). Immunohistochemistry was completed with appropriate secondary biotinylated antibodies (Vector Laboratories Ltd., Peterborough, UK) diluted 1:500 in secondary layer diluent (0.3% Triton X-100 in PBS), followed by avidin-biotin complexation (Vector ABC, Vector Laboratories Ltd., Peterborough, UK) and visualized using diaminobenzidine (for phosphotyrosine) and VIP (for A␤40) according to the manufacturer's data sheets (Vector Laboratories Ltd.).
Isolation of Primary Microglia-Rat microglia were prepared from cerebral cortices of 1-2-day-old rat pups of either sex (strain: CD, Charles River, Margate, UK) (30). Animals were euthanaized in accordance with the 1986 Animals (Scientific Procedures) Act. The final cell pellet from cortex was resuspended in Dulbecco's modified Eagle's medium (high glucose Dulbecco's modified Eagle's medium with L-glutamine) supplemented with 10% FCS and 100 units/ml penicillin ϩ 50 g/ml streptomycin, and plated in 75-cm 2 poly-L-lysine-coated tissue culture flasks (Corning, Corning, NY) at a density of 1.5 brains per flask. Culture medium was changed after 24 h and then twice per week. Microglia were isolated on day 14 by shaking the flasks on an orbital shaker (New Brunswick Scientific) at 200 rpm for 2 h (37°C). The attached cell monolayers were highly enriched in astrocytes. The culture supernatant was transferred to plastic Petri dishes (Sterilin) and incubated for 45 min at 37°C (5% CO 2 , 95% air) to allow differential adhesion of microglia. The adherent microglial cells were maintained in growth medium until harvested for addition to neuronal cell cultures (3-5 days later).
Microglial cell cultures from genotypically matched normal and homozygous MK2 Ϫ/Ϫ mice were also prepared. Mouse microglia were isolated from mixed glial cell cultures as described above for rat, except that animals were used at postnatal days 7-8 (to obtain greater cell yields). Harvested mouse microglia were used both for cytokine release assays, and in co-cultures of rat cortical neurons and mouse microglia. Cocultures utilizing mouse microglia were established in the same manner as microglia derived from rat cortex.
Cytokine/Chemokine Release Assays from Primary Mouse Microglia and NTW8 Mouse Microglial Cells-Microglia were plated in wells of a 96-well plate (poly-D-lysine, BD Biosciences) at a density of 100,000 cells per well and allowed to adhere overnight. Cells were stimulated to release pro-inflammatory mediators in medium containing 100 ng/ml LPS and 20 units/ml mouse IFN␥, or 30 M A␤  peptide. Prior to cell stimulations, A␤ peptide was prepared for this purpose by dissolving the peptide in 1,1,1,3,3,3-hexafluoro-2-propanol to a concentration of 1 mM. The 1,1,1,3,3,3-hexafluoro-2-propanol solvent was then evaporated, and the peptide film was re-dissolved in Me 2 SO to a concentration of 5 mM. Serum-free medium was pre-warmed to 37°C and added to dilute the peptide to 100 M, before adding to cells to give a final concentration of 30 M. Cell supernatants were harvested after 24 h and cytokine and chemokine release was assayed by ELISA according to the manufacturer's instructions. For experiments employing either LPS ϩ IFN␥ or A␤(1-42) treatment, cell viability was measured by MTT or LDH assay, respectively, according to the manufacturer's instructions.
Isolation of Cortical Neurons-Primary cerebral cortical cell cultures were prepared from gestational day 18 fetuses of timemated CD rats (Charles River) (31), with minor modifications. Briefly, after removal of meninges, cerebral cortices were dissected and dissociated with a papain tissue dissociation kit (Worthington, Lakewood, NJ) following the manufacturer's instructions. Cells were resuspended in Neurobasal medium containing 2% B27 supplements, 1 mM sodium pyruvate, 2 mM L-glutamine and penicillin (100 units/ml)/streptomycin (50 g/ml), and plated in poly-D-lysine-coated 48-well plates, 1 ϫ 10 5 cells/well in 0.35 ml. Poly-D-lysine-coated culture wells were exposed overnight to medium containing 10% FCS prior to cell seeding. After 2 days, cultures received 0.35 ml/well of plating medium, but containing B27 supplement without antioxidants. Cultures were maintained at 37°C in a 5% CO 2 humidified atmosphere. Neuronal cells were used on days 5-6 for experiments.
Co-cultures of Microglia and Cortical Neurons-Microglia were collected by mechanically scraping into culture medium. After centrifugation (200 ϫ g for 5 min) cells were resuspended in Neurobasal medium containing 2% B27 supplements (without antioxidants), 1 mM sodium pyruvate, 2 mM L-glutamine and penicillin/streptomycin. Co-cultures were prepared by plating a suspension of isolated microglia (rat or mouse) on neurons maintained for 5 days in vitro in a 1:1 ratio. Cells were incubated for 30 min and then washed once with plating medium to remove non-adherent contaminating macroglia and debris, and then refed with fresh plating medium.
Injury Induction in LPS ϩ IFN␥-stimulated Co-cultures-Neuron-microglia co-cultures were treated the day following addition of microglia to the neuronal cell monolayers. Culture medium was replaced with an equal volume of fresh medium containing LPS (0.5 g/ml) and rat or mouse IFN␥ (500 units/ ml). Analogous treatments were carried out for cultures composed of only cortical neurons or microglia. All cultures were incubated at 37°C for 3 days, after which time cell viability was determined by LDH release.
Neurotoxicity Assay in LPS ϩ IFN␥-stimulated Co-cultures-LDH release into the culture medium was used as a measure of cell death, and has been utilized previously to quantify neuronal cell injury in microglia-neuron co-cultures (12,13). LDH activity in the cell culture supernatants was determined after 72 h, using the CytoTox non-radioactive cytotoxicity assay kit following the manufacturer's instructions. Values of LDH release are expressed as milliunits/ml, unless indicated otherwise. Microglia and cortical neurons were also independently cultured for 72 h in the presence of stimuli, and LDH release from the microglia alone Ϯ stimuli were subtracted out from the values obtained from the combination of microglia and cortical neurons.
Quantitative Analysis of Neuronal Cell Injury by Arrayscan-All incubations were carried out at room temperature unless otherwise stated. Cells were fixed in PBS containing 4% (w/v) paraformaldehyde, 10% (w/v) sucrose, and 15 g/ml Hoechst 33342. The fixative was removed and cells were washed once with 200 l of PBS. Cells were then blocked in PBS, 1% (w/v) bovine serum albumin, 0.1% (w/v) Triton X-100 (1 h), followed by incubation for 1 h with anti-␤III-tubulin primary antibody (1:1000 in blocking buffer). Cells were next washed 3 times with blocking buffer and then incubated 1 h with Alexa 488-labeled secondary antibody (1:300 in blocking buffer). Cells were subsequently washed 3 times with PBS prior to imaging.
Image acquisition within the 48-well plate and subsequent neurite outgrowth measurements were performed using Cellomics Arrayscan technology (Swallowfield, UK). Ultraviolet light was used to illuminate Hoechst 33342-labeled nuclei, allowing automated focusing upon the cells. A second excitation wavelength of 488 nm was used to obtain pictures of specifically labeled neurons. These images were then analyzed using Cellomics neurite outgrowth software to identify and measure neurites that fit defined criteria. Data are expressed as perimeter squared divided by 4 ϫ ϫ area ("P2/A"). Exposure times for each wavelength were determined empirically by the user.
Injury Induction in A␤(1-42)-stimulated Cortical Neuron/ Microglia Co-cultures-Microglia isolated from the cerebral cortex of normal and MK2 Ϫ/Ϫ mice were added to cell culture inserts (PET membrane, 0.4 m pore size, BD Biosciences) at 7.5 ϫ 10 4 cells per insert in Dulbecco's modified Eagle's medium with 10% FCS (0.4 ml/insert), and placed in a 24-well plate (notched for inserts) in Dulbecco's modified Eagle's medium with 10% FCS (0.8 ml/well). The following day, inserts were washed once with Neurobasal medium containing 2% B27 supplements (without antioxidants), 1 mM sodium pyruvate, 2 mM L-glutamine and penicillin/streptomycin, and transferred to a 24-well plate of rat cortical neurons (6 days in vitro, 2 ϫ 10 5 cells per well). The porous membrane allows free diffusion of molecules. The distance between the neuron monolayer and microglia on the insert membrane is 1 mm, according to the manufacturer's description. A␤  was then added to the microglia-containing inserts (15 M in the above Neurobasal medium); controls received an equal volume of medium with 0.2% Me 2 SO. The co-cultures were returned to the incubator for another 3 days.
Statistics-Data are given as mean Ϯ S.D. or S.E. Statistical analyses to determine group differences were performed either by two-sample equal variance Student's t test, or by one-way analysis of variance, followed by Dunnett's post-hoc test for comparisons involving more than two data groups.

MK2 Expression Is Enriched in Primary Microglial Cells
Compared with Primary Astrocytes and Neurons-To investigate the relative expression of MK2 in cell types within the CNS, we compared expression of MK2 in rat primary cultures of cortical neurons, microglia, and astrocytes by quantitative RT-PCR (Fig. 1). GFAP, CD11b, and MAP2 expression as markers of astrocytes, microglia, and neurons, respectively, confirmed culture purity (data not shown). MK2 expression was found to be highly enriched in microglial cells, supporting a role for MK2 function in these cells.
MK2 Expression, p38 MAPK Phosphorylation, and MK2 Activation Are Induced in NTW8 Microglial Cells Treated with LPS ϩ IFN␥-To investigate expression of MK2 in microglial cells and its response to a pro-inflammatory stimulus, NTW8 microglial cells were treated with LPS ϩ IFN␥ and MK2 expression was analyzed by quantitative RT-PCR (limiting numbers of primary microglia precluded such biochemical analyses). Up-regulation (ϳ2-fold) of MK2 expression was observed after 8 and 16 h of LPS ϩ IFN␥ treatment ( Fig. 2A), but not at earlier times (data not shown).
Up-regulation in MK2 gene expression was preceded by an increase in phosphorylation of p38 MAPK and an elevation in MK2 activity (assessed by phosphorylation of recombinant Hsp27 in vitro by MK2 immunoprecipitated from cellular lysates). LPS ϩ IFN␥-mediated induction of p38 MAPK phosphorylation (Fig. 2B) and MK2 activity (Fig. 2C) was observed not only at time points where MK2 expression was up-regulated, but also at 1 h post-treatment with LPS ϩ IFN␥.

MK2 Deficiency in Microglia Down-regulates LPS ϩ IFN␥stimulated and A␤(1-42)-stimulated Cytokine and Chemokine
Release-Stimulation of both NTW8 microglial cells and primary cortical microglia with LPS ϩ IFN␥ released an overlap- ping set of pro-inflammatory mediators detectable by ELISA (Table 1). However, treatment of either microglial culture type with A␤(1-42) peptide resulted in the detectable release of MIP-1␣ only (Table 1); A␤(1-42) reverse peptide from the same supplier as the A␤(1-42) peptide routinely failed to upregulate MIP-1␣ release. To investigate a role for MK2 in proinflammatory mediator release from microglia, ex vivo cultures of microglia from cortex of MK2 Ϫ/Ϫ and wild-type mice were established, and the cells were treated with either LPS ϩ IFN␥ or A␤(1-42) peptide. MK2 Ϫ/Ϫ microglia demonstrated significant reductions in MIP-1␣, TNF-␣, and KC release on treatment with LPS ϩ IFN␥, and a significant reduction in MIP-1␣ release on treatment with A␤(1-42) peptide (Fig. 3). Cell viability did not differ between wild-type and MK2 Ϫ/Ϫ cells.
LPS ϩ IFN␥-stimulated Microglia Are Neurotoxic in Co-culture with Cortical Neurons-We first co-cultured cortical neurons (5 days in vitro) and rat microglia to determine whether microglia stimulated with LPS ϩ IFN␥ directly cause neuronal cell death. The effect of LPS ϩ IFN␥ on microglia/neuronal co-cultures as well as individual neuronal and microglial cultures, as controls, was quantified by determining LDH release into the culture medium. Against this baseline, LDH release increased with time in the stimulated co-cultures becoming significant at 3 days, indicating enhanced neuronal cell death (Fig. 4A). Based upon these results, a time point of 3 days was chosen for all subsequent experiments. Treatment of co-cultures with either IFN␥ or LPS alone did not increase neurotoxicity (data not shown). Co-culture with non-stimulated microglia showed no increase in LDH release above that attributable to loss of neurons and microglia, demonstrating that neuronal cell death requires microglial stimulation (Fig. 4A). LPS ϩ GAPDH expression was unchanged in cells treated with LPS ϩ IFN␥ and was therefore used to normalize MK2 expression. The mean expression level for LPS ϩ IFN␥-stimulated samples was taken as 100%. Data are expressed as mean % of LPS ϩ IFN␥-stimulated cells Ϯ S.E. (n ϭ 4 from two independent experiments). ***, p Ͻ 0.001, and **, p Ͻ 0.01 compared with untreated cells. B and C, protein extracts were prepared from NTW8 microglial cells treated Ϯ LPS (100 ng/ml) ϩ IFN␥ (20 units/ml) for the times shown, and used either for Western blotting to assess p38 MAPK phosphorylation (B), or to immunoprecipitate MK2 for use in kinase activity assays (C). B, protein extracts for Western blotting were prepared from duplicate wells of cells, and 5 g of each sample was subjected to SDS-PAGE analyses, followed by Western blotting with either a phospho-p38 MAPK antibody (upper panel ), or an anti-p38 MAPK antibody that recognizes total p38 MAPK protein (lower panel ). C, MK2 was immunoprecipitated for use in kinase activity assays in vitro using recombinant Hsp27 as a substrate for MK2-directed phosphorylation. Protein extracts for use in kinase assays were prepared from duplicate wells of cells. Kinase assays were subjected to SDS-PAGE analysis followed by Western blotting with either a phospho-Hsp27 antibody (upper panel ), or an anti-MK2 antibody that recognizes total MK2 protein to confirm immunoprecipitation of MK2 for all samples (lower panel ).  IFN␥-stimulated cortical neuron/microglia co-cultures were also immunostained for the neuron-specific marker ␤III-tubulin, and analyzed using Cellomics Arrayscan. Neurons cultured in the presence of LPS ϩ IFN␥-treated microglia for 3 days underwent a striking degeneration of their neuritic fields, accompanied by debris and loss of cell bodies (Fig. 4B). These data are presented in quantitative fashion in Fig. 4C, and are consistent with an increase in LDH release that reflects neuronal cell death. LPS ϩ IFN␥ had no direct effect on LDH release from primary neuronal cell cultures, and medium conditioned for 1 or 3 days from activated or non-stimulated microglia had no effect on neuronal cell cultures when added at day 6 in vitro for 3 days (data not shown). Similar neurotoxic behaviors for LPS ϩ IFN␥-activated microglia have been reported by others (12). However, although medium conditioned from LPS ϩ IFN␥-stimulated microglia failed to induce neuronal cell injury, we have shown that physical contact between activated micro-glia and neurons is not required for neurotoxicity. This was demonstrated by culturing rat microglia on transwell inserts containing a porous membrane and these were then transferred to culture wells with an established monolayer of cortical neurons. The microglia were stimulated by LPS ϩ IFN␥, and neuronal cell vitality was assessed after 3 days by MTT assay (data not shown; detailed methods performed as for injury induction in A␤-stimulated cortical neuron/microglia, except that microglia were stimulated by treatment with 0.5 g/ml LPS and 500 units/ml IFN␥). Taken together, these data imply that soluble factors released from LPS ϩ IFN␥-stimulated microglia are responsible, at least in part, for neuronal cell injury, and that these factors may be short-lived species.

MK2 Deficiency Abolishes the Toxic Effect of LPS ϩ IFN␥stimulated Microglia on Cortical
Neurons-To investigate the involvement of MK2 in neuronal cell injury mediated by LPS ϩ IFN␥-stimulated microglia, rat cortical neurons were co-cultured with microglia derived from MK2-deficient and wildtype mice. Neuronal co-cultures established with wild-type microglia and treated with LPS ϩ IFN␥ developed significant neuronal cell death 3 days later (Fig. 5A), analogous to the effect of LPS ϩ IFN␥-stimulated rat microglia co-cultured with cortical neurons. In contrast, co-cultures composed of rat cortical neurons and microglia from MK2-deficient animals, incubated with LPS ϩ IFN␥ were markedly less affected (Fig. 5A). Phosphorylation of p38 MAPK in LPS ϩ IFN␥-stimulated MK2 Ϫ/Ϫ Cortical neurons co-cultured with microglia without any stimuli (untreated) survived well and formed extensive neuritic networks, whereas marked neuron-specific degeneration was observed in co-cultures treated with LPS and IFN␥ (LPS ϩ IFN␥). C, the extent of neuronal cell injury caused by co-culture with LPS ϩ IFN␥-stimulated microglia after 72 h was quantified by Arrayscan analysis of ␤III-tubulin labeling. The data are mean Ϯ S.D. from 8 replicates from the same culture. ***, p Ͻ 0.001 compared with untreated cells. microglia was comparable with that of wild-type microglia, demonstrating intact LPS ϩ IFN␥ stimulation of this kinase upstream of MK2 in MK2 Ϫ/Ϫ microglia (Fig. 5B).  are neurotoxic and if MK2 plays a role. The co-culture model was modified by plating microglia from wild-type and MK2 Ϫ/Ϫ mice on cell culture inserts containing a porous membrane, followed by transfer to culture wells with an established monolayer of cortical neurons. The microglia were then challenged with 15 M A␤(1-42), and neuronal cell vitality was assessed after 3 days by MTT. Initial experiments showed this concentration of A␤ , added to the inserts induced neuronal cell injury in the presence, but not absence of microglia. Under these conditions A␤(1-42)treated microglia produced a significant, 37.0 Ϯ 6.4%, loss of neuronal cell viability, which was abolished when MK2 Ϫ/Ϫ microglia were used instead (Fig. 6).

MK2 Expression Is Up-regulated in TASTPM Mouse Cerebral Cortex: Correlation with Up-regulation in MIP-1␣, TNF-␣, and KC Expression and Microglial Activation-
The observed upregulation of MK2 expression in activated microglial cells in vitro encouraged an investigation of MK2 expression, proinflammatory gene expression, and microglial activation in the TASTPM transgenic mouse model of AD. TASTPM mice are a double transgenic line that overexpress mutant forms of human amyloid precursor protein and presenilin-1 specifically in the CNS, resulting in progressive amyloid deposition and pathology in the brain (25,26). Both transgenes are under the control of the Thy-1 regulatory cassette, which directs expression to the cerebral cortex, amygdala, dentate gyrus, CA1-CA3, and the cerebellum (25,26). Gene expression studies utilized cortical and cerebellar tissues from WT and TASTPM mice. Although transgenes are expressed in cerebellum as well as cortex, plaque pathology is evident in cortex but not in cerebellum of TASTPM mice. Gene expression changes occurring in the cortex of TASTPM mice, but not the cerebellum, can therefore be correlated to ␤-amyloid plaque pathology and not transgene overexpression per se. Gene expression was examined in hemisected cortex and cerebellum taken from 11-, 13-, and 15-month-old WT and TASTPM mice (Fig. 7A shows 15-month data only). MK2 expression was significantly up-regulated in the cortex but not the cerebellum, and was accompanied by an up-regulation in pro-inflammatory gene expression, including TNF-␣, MIP1-␣, and KC. Up-regulation of MK2 as well as these pro-inflammatory genes was also observed in the cortex but not cerebellum from 11-and 13-month-old TASTPM mice (data not shown).
Given the up-regulation in MK2 and pro-inflammatory gene expression observed, we investigated whether microglia were  , and used for reverse transcription and TaqMan PCR (4 -6 animals per group; gene expression for each animal analyzed separately). The levels of GAPDH expression did not differ significantly between transgenic and wild-type animal groups in either cortex or cerebellum, and GAPDH expression was therefore used to normalize data for MK2, TNF␣, KC, and MIP1␣ expression. Data are expressed as mean % GAPDH expression Ϯ S.E., and comparisons were drawn between transgenic and wild-type groups. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. B, a photomicrograph of an A␤ plaque with closely associated microglia in the cortex of a 12-month-old TASTPM mouse. The plaque was labeled with polyclonal antibody G30 that specifically recognizes the A␤40 C terminus; microglia were labeled with an antibody to phosphotyrosine. Labeling of the A␤ plaque and microglia are indicated by solid and open arrows, respectively. See "Experimental Procedures" for details. activated in the cortex of TASTPM mice, because microglial cells are the likely cells to elicit the pro-inflammatory changes observed. Microglial activation was assessed by anti-phosphotyrosine immunohistochemistry, because phosphotyrosine immunoreactivity is reported to be largely restricted to microglia in the CNS, and is highly up-regulated on conversion of microglial morphology from ramified (resting) to amoeboid (activated) forms (32). Microglial activation was observed surrounding and in close association with ␤-amyloid plaques in cortical tissue from 11-and 12-month-old TASTPM mice (Fig.  7B shows an example from a 12-month-old mouse). In addition to phosphotyrosine staining, microglial clustering around ␤-amyloid plaques was also confirmed in 11-and 12-month-old TASTPM mice by immunostaining for the microglial-specific antigens CD45 and CD11b (data not shown).
MK2 Activity Is Up-regulated in Cerebral Cortex of TASTPM Mice-To determine whether the increase in MK2 gene expression observed in TASTPM mouse cortex was matched by an up-regulation in MK2 activity, immunoprecipitation kinase assays were performed using protein extracts prepared from hemisected cortical tissue from 15-month-old TASTPM mice (tissue was derived from animals used for gene expression analyses). A significant increase in MK2 activity in the transgenic animal group was confirmed (Fig. 8).

DISCUSSION
The present study provides compelling evidence supporting a role of MK2 in microglial neuroinflammation and neurodegeneration. MK2 expression was enriched in primary cultures of microglial cells compared with neuronal and astrocytic cells, proposing a role for MK2 function in the former. Furthermore, MK2 expression was up-regulated in activated microglial cultures, and in the cerebral cortex of TASTPM mice, in which neuroinflammation and microglial activation are observed. Upregulation of MK2 expression both in vitro and in vivo was matched by a corresponding induction in MK2 activity, implicating MK2 as a key driver of inflammatory mediator release and pro-inflammatory gene expression. Moreover, ex vivo microglial cultures prepared from MK2 Ϫ/Ϫ mice displayed a diminished cytokine/chemokine release profile. When co-cultured with cortical neurons, MK2-deficient microglia were unable to cause neuronal cell death in the presence of inflammatory stimuli, including A␤ , in contrast to wild-type microglia.
Central to our findings are the observations that both MK2 expression and activity were induced in in vitro and in vivo models of neuroinflammation. In addition to p38 MAPK-directed phosphorylation-dependent activation of MK2, up-regulation of MK2 expression may itself contribute to induction of MK2 activity, suggesting a role for MK2 in chronic inflammation. Up-regulation of MK2 expression and activity in activated microglial cells correlated well with up-regulation in MK2 expression and activity in cortex of TASTPM mice. These data support a role for induction of MK2 activity in activated microglia in TASTPM mice, thereby driving pro-inflammatory mediator production. Although MK2 is more highly expressed in primary cultures of microglia than in neurons or astrocytes, induction of MK2 expression and activity in TASTPM cortex may not be limited to microglia, because MK2 expression and activity is likely to be induced in other CNS cell types as well. Whereas p38 MAPK-directed phosphorylation-dependent activation of MK2 is well documented, relatively few reports describe induced changes in MK2 mRNA. Regulation of MK2 expression is reported to be under the control of p38 MAPK (33), indicating that p38 MAPK may exert its regulation of MK2 activity not only by phosphorylation, but also by modulation of its expression. MK2 expression is induced also in response to toxic insults (34,35). Induction of MK2 expression and activity in CNS tissue of TASTPM mice may therefore be a consequence of amyloid pathology, such as amyloid plaque deposition, inflammatory cytokines, or reactive oxygen species. The increased activity of MK2 immunoprecipitated from cortical tissue extracts of TASTPM mice and from activated microglial cultures is thus likely to reflect both increased expression of MK2 and increased activation of MK2 due to p38 MAPK activity. Indeed, activation of the p38 MAPK pathway occurs in other in vitro and in vivo models relevant to AD, and in AD brain itself (20 -23, 36).
MK2 is capable of regulating cytokine expression at the posttranscriptional level (reviewed in Ref. 4). It has been proposed that MK2 influences mRNA stability by regulating the action of proteins that facilitate mRNA degradation, and may accomplish this by interfering with the ability of such proteins to bind AU-rich sequences in the mRNA 3Ј-untranslated region (37)(38)(39). Multiple AU-rich sequence-binding proteins have been identified in macrophages, one of which, heterogeneous nuclear ribonucleoprotein A0, has been reported to be a substrate for MK2 (39). However, despite robust p38 MAPK/MK2 pathway activation, we could find no evidence that heterogeneous nuclear ribonucleoprotein A0 was phosphorylated in LPS ϩ IFN␥-stimulated microglial cells (data not shown), implying that MK2 does not utilize heterogeneous nuclear ribonucleoprotein A0 to regulate cytokine mRNA stability in microglia. Identification of additional substrates of MK2 is therefore required to answer the question of precisely how MK2 regulates cytokine expression in microglial cells. MK2 activity was assayed in vitro using an immunoprecipitation kinase assay for HSP27 phosphorylation (4 -6 animals per group; MK2 activity in lysates generated from each animal was analyzed separately and in duplicate). HSP27 phosphorylation was quantified by densitometry (a representative immunoblot is shown), and the mean MK2 activity in the transgenic animal group was taken as 100%. Data are mean Ϯ S.E. and comparisons were drawn between transgenic and wild-type groups, *, p Ͻ 0.05.
A range of pro-inflammatory mediators including TNF-␣, interleukin-8 (IL-8), interleukin-6 (IL-6), interleukin-1␤ (IL-1␤), IFN␥, and cyclooxygenase 2 are regulated by MK2 activity in a variety of biological systems (3-4, 40 -43). Our data show consistent inhibition of production of TNF-␣, and of the chemokines KC (mouse chemokine with highest sequence identity to human GROs and IL-8) and MIP-1␣ from LPS ϩ IFN␥-stimulated MK2 Ϫ/Ϫ microglia, when compared with MK2 ϩ/ϩ microglia. In contrast, MK2 Ϫ/Ϫ deficiency failed to alter microglial cell output of IL-6, monocyte chemotactic protein 1, and nitric oxide in the presence of LPS ϩ IFN␥ (data not shown). The lack of consistent knockdown in IL-6 release from LPS ϩ IFN␥-stimulated microglia is in contradiction to reports of MK2 regulation of expression of this cytokine in other biological systems. Conceivably, the contribution that MK2 makes to the regulation of IL-6 expression in LPS ϩ IFN␥-activated microglia may be less than that of TNF-␣, KC, and MIP-1␣.
Our findings are the first to demonstrate a role of MK2 in regulating MIP-1␣ release. Importantly, MIP-1␣ release was repressed not only in LPS ϩ IFN␥-stimulated MK2 Ϫ/Ϫ microglia, but also in cells treated with A␤(1-42) peptide. It was not possible to determine the effects of MK2 deficiency on other A␤(1-42)-regulated pro-inflammatory mediators, as MIP-1␣ was the only pro-inflammatory mediator detectable by ELISA from A␤(1-42)-stimulated microglial cultures. Indeed, both we and others have shown that although stimulation of microglial cultures with A␤(1-42) peptide can induce expression of multiple cytokines and chemokines at the mRNA level (data not shown), MIP-1␣ remains the sole pro-inflammatory mediator measurable at the protein level in cell supernatants. These data suggest that MIP-1␣ expression is highly sensitive to A␤(1-42) stimulation, and that MK2 plays a role in A␤(1-42) induction of MIP-1␣ expression.
The demonstration that MK2 plays a role in TNF-␣, KC, and MIP-1␣ release from LPS ϩ IFN␥-and A␤(1-42)-stimulated microglia correlated well with up-regulation of expression of these same cytokines in cortical tissue of TASTPM mice, in which up-regulation of MK2 expression and activity has also been demonstrated. In particular, MIP-1␣ was a particularly robust marker of inflammation in TASTPM mice; transcript levels were up-regulated ϳ20-fold in cortex from TASTPM mice compared with WT. These data propose a role for MK2 in regulating neuroinflammation resulting from ␤-amyloid pathology in TASTPM mice, and, possibly, AD brain pathology. A more direct proof may require crossing MK2-deficient mice with TASTPMs, which is beyond the scope of the present study. Indeed, up-regulation of chemokine and cytokine expression has been documented in AD brain (5,44,45).
LPS ϩ IFN␥-stimulated and, more importantly, A␤(1-42)stimulated MK2-deficient microglia were less able to induce neuronal death in co-cultures than LPS ϩ IFN␥or A␤(1-42)stimulated MK2 ϩ/ϩ microglia, thereby supporting a role for MK2 in the release of neurotoxic substances. Inhibition of p38 by pharmacological intervention is reported to protect mouse cortical neurons from inflammation-induced neurotoxicity when co-cultured with LPS-stimulated mixed glial cells (46). Here, we have shown that LPS ϩ IFN␥-stimulated p38 MAPK phosphorylation was comparable between wild-type and MK2-deficient microglia. The diminished output of pro-inflammatory mediators from MK2-deficient microglia compared with wild-type microglia, and the inability of MK2-deficient microglia to induce neuronal injury, is therefore not attributable to a defect in LPS ϩ IFN␥-stimulated p38 MAPK activation. Our data extend previous findings to implicate MK2 as a specific effector of p38 action not only in LPS-stimulated microglia, but also microglia stimulated by A␤(1-42) peptide. Such a role for MK2 in the generation of microglial neurotoxic substances may explain, at least in part, the in vivo neuroprotection reported in MK2 KO mice subjected to ischemic injury (41). The precise nature of the neurotoxic species released from activated microglia remains undefined. The consistent inhibition of TNF-␣, KC, and MIP-1␣ release from LPS ϩ IFN␥-stimulated MK2 Ϫ/Ϫ microglia reported here highlighted these pro-inflammatory mediators as potential candidates, but these proteins in recombinant form were not directly toxic to cortical neurons when applied either singly or in combination (data not shown). However, we cannot rule out the possibility that TNF-␣, KC, and MIP-1␣ may have neurotoxic action that is dependent on the presence of other necessary, but as yet unidentified, microglial factors. Furthermore, it is also possible that these cytokines may form members of a mixture of microglial factors that signal to microglia in a paracrine and/or autocrine fashion, thereby contributing to an output from microglia that is ultimately capable of inducing neurotoxicity. Our data suggest that short-lived species may form part of the output of soluble microglial factors responsible for neuronal cell injury. Intriguingly, microglial activation, MK2 activation, and up-regulation of pro-inflammatory gene expression occur in CNS tissue from TASTPM mice without overt neuronal loss in areas of amyloid pathology. One explanation for this lack of neurodegeneration is that transgenic mice overexpressing mutant amyloid precursor protein may have increased activation of cell survival pathways (47).
MK2-regulated pro-inflammatory mediators TNF-␣, MIP-1␣, and KC play key roles not only in neuroinflammation and neurodegeneration, but also in other components of AD pathophysiologies such as cognitive decline and Tau hyperphosphorylation. A large body of evidence describes neurotoxic roles for TNF-␣ within the CNS (48 -50), as well as its ability to affect long-term potentiation and learning (51)(52). For chemokines such as MIP-1␣ and KC, it is likely that their production by plaque-surrounding cells results in the recruitment and accumulation of astrocytes and microglia in an attempt to phagocytose the plaque, thereby creating a vicious cycle of inflammation resulting in a chronic glial inflammation and ultimately neurodegeneration. MIP-1␣ has been implicated as a key contributor to CNS pathology in the lysosomal storage disorder, Sandhoff disease, because deletion of MIP-1␣ decreased microglial activation and neuronal apoptosis in a mouse model of this disease (53). Aside from resultant effects on neurodegeneration, neuroinflammation has also been proposed to play a causative role in Tau pathology (54), and KC in particular may be a trigger for Tau hyperphosphorylation (55).
In conclusion, our study strengthens further the involvement of the p38 MAPK/MK2 pathway in neuroinflammation, and more specifically, in the release from microglia of pro-inflammatory mediators of relevance to neuroinflammatory pathol-ogy in a mouse model of ␤-amyloid deposition. The data, together with our demonstration that MK2 deficiency in microglia protects neurons from inflammation-induced neurotoxicity, make it tempting to speculate that MK2 may be a key modulating pathway of inflammatory mediators linked to neurodegeneration in conditions such as AD.