Amyloid β‐induced astrogliosis is mediated by β1‐integrin via NADPH oxidase 2 in Alzheimer's disease

Summary Astrogliosis is a hallmark of Alzheimer′s disease (AD) and may constitute a primary pathogenic component of that disorder. Elucidation of signaling cascades inducing astrogliosis should help characterizing the function of astrocytes and identifying novel molecular targets to modulate AD progression. Here, we describe a novel mechanism by which soluble amyloid‐β modulates β1‐integrin activity and triggers NADPH oxidase (NOX)‐dependent astrogliosis in vitro and in vivo. Amyloid‐β oligomers activate a PI3K/classical PKC/Rac1/NOX pathway which is initiated by β1‐integrin in cultured astrocytes. This mechanism promotes β1‐integrin maturation, upregulation of NOX2 and of the glial fibrillary acidic protein (GFAP) in astrocytes in vitro and in hippocampal astrocytes in vivo. Notably, immunochemical analysis of the hippocampi of a triple‐transgenic AD mouse model shows increased levels of GFAP, NOX2, and β1‐integrin in reactive astrocytes which correlates with the amyloid β‐oligomer load. Finally, analysis of these proteins in postmortem frontal cortex from different stages of AD (II to V/VI) and matched controls confirmed elevated expression of NOX2 and β1‐integrin in that cortical region and specifically in reactive astrocytes, which was most prominent at advanced AD stages. Importantly, protein levels of NOX2 and β1‐integrin were significantly associated with increased amyloid‐β load in human samples. These data strongly suggest that astrogliosis in AD is caused by direct interaction of amyloid β oligomers with β1‐integrin which in turn leads to enhancing β1‐integrin and NOX2 activity via NOX‐dependent mechanisms. These observations may be relevant to AD pathophysiology.


Introduction
Alzheimer's disease (AD) is a neurodegenerative disorder characterized by progressive loss of memory and cognitive function. The mechanisms underlying AD include an interplay between direct neurotoxicity, cerebrovascular pathology, inflammation, and cortical network dysfunction (Iadecola, 2004;Haass & Selkoe, 2007;Palop & Mucke, 2010). Whereas neuronal death has taken center stage in AD pathology, the putative contribution of glia to this disease has not been closely examined.
Astrocytes are the glial cells responsible for brain homeostasis and contribute to its protection in pathology (Kettenmann & Ransom, 2013). In neurological diseases, astrocytes show morpho-functional changes to adopt a phenotype called astrogliosis that can affect the course of the disorder. AD patient brains have shown hypertrophic astrocytes clustering around amyloid plaques (Duffy et al., 1980), upregulation of astrogliotic marker GFAP in brain homogenates (Meda et al., 2001), and elevated levels of GFAP in the CSF of AD patients (Fukuyama et al., 2001). Overall, these data suggest a correlation between increased levels of these proteins in reactive astrocytes and disease severity.
Amyloid-b (Ab) species and their accumulation in plaques are key steps in the pathogenesis of AD (Haass & Selkoe, 2007). Astrocyte pathology might be a consequence of a primary Ab-induced glioreactivity over progression of the disease. Indeed, reactive astrocytes near Ab plaques show impaired functions concerning Ca 2+ network hyperactivity (Delekate et al., 2014) and an aberrant production of gliotransmitters (Jo et al., 2014) as well as disrupted astrocyte-toastrocyte topology (Galea et al., 2015). In astrocytes, Ab-oligomers cause dysregulation of Ca 2+ homeostasis (Abramov et al., 2003;Alberdi et al., 2013) and abnormal production of NADPH oxidase (NOX)-derived reactive oxygen species (ROS) (Abeti et al., 2011) which leads to astrogliosis  and failure of astrocytic metabolism that may inflict neuronal cell death (Abramov et al., 2003;Abeti et al., 2011). Several Ab-oligomer receptors have been described, but no single candidate has been shown to be necessary and sufficient to account for all aspects of Ab-oligomer binding, glioreactivity, and toxicity (Viola & Klein, 2015). Among them, b1integrin was proposed as a partner in Ab-signaling (Sabo et al., 1995). Indeed, Ab-oligomers bind directly to low-intermediate activation conformers of b1-integrin to induce mitochondrial and synaptic dysfunction in AD (Woo et al., 2015). In addition, Ab-peptide induces apoptosis and downregulation of a1b1 integrin in neuronal cells, indicating a relationship between Ab-neurotoxicity and modulation of integrin expression (Bozzo et al., 2004(Bozzo et al., , 2010. Moreover, Aboligomers promote a high-affinity state of LFA1 integrin to provoke a rapid neutrophil adhesion, trafficking and extravasation into the CNS in AD models which is a key to the progression of cognitive decline and gliosis (Zenaro et al., 2015).
In this study, we explored the mechanistic relationships among Aboligomers, b1-integrin, NOX activities, and GFAP overexpression in vitro, in vivo, and in AD brain samples. We found that Ab-oligomers activate b1-integrin/PI3K/PKC/Rac/NOX pathway to upregulate NOX2 expression and to promote b1-integrin maturation, GFAP overexpression, and astrogliosis. We provide evidence showing that GFAP and NOX2 upregulation in Ab-injected mouse brain occurs through b1-integrin signaling. Finally, analysis of a mouse model of AD and human AD samples shows that b1-integrin and NOX2 levels are significantly higher in reactive astrocytes and correlate with Ab-load.

Results
Ab-Oligomers induce GFAP expression in response to NOXdependent redox regulation First, we found by real-time fluorescence measurements that Aboligomers (5 lM) induced a rapid increase in fluorescence that reached plateau after 30-60 min (Fig 1A). This increase (142 AE 9%, n = 20 compared with untreated cells after 60 min) was blocked by compounds that prevent the assembly of NOX, apocynin (APO; 111 AE 13%, n = 8 compared to cells treated only with APO; Cavaliere et al., 2012), and DPI (96 AE 5%, n = 8 compared to cells treated only with DPI). In addition, astrocytes treated with a peptide that inhibits p47 (phox) association with NOX2 (gp91ds-tat) showed a robust attenuated oligomeric Ab-induced signal (105 AE 9%, n = 4) ( Fig 1A). We also analyzed the role of Ab in the regulation of mRNA transcription and protein expression of NOX catalytic forms in cultured astrocytes. Here, we found that, among the NOX family, the mRNA levels of NOX2 were increased (1.6-fold), whereas NOX1 decreased (0.3-fold) ( Fig 1B). Western blot analysis further confirmed the increase in NOX2 (129 AE 9% and 122 AE 9%, n = 6) and the decrease in NOX1 (55 AE 17% and 39 AE 9%, n = 6) protein levels after Ab treatment for 6 and 24 h. DPI treatment abolished Ab-induced expression changes of NOX2 and NOX1 of astrocytes (Fig 1C-E). Overall, these data show that Ab-oligomers activate NOX enzymes and modulate NOX expression in a ROS-dependent manner. In addition, we observed that GFAP overexpression in astrocytes treated with Ab-oligomers was reduced by NOX inhibition with DPI compound (Fig 1F,G). This result revealed that ROS production in astrocytes by Ab-oligomers could be a key signaling in the expression changes of GFAP levels.
To examine whether Rac1 activates NOX enzymes upon Abstimulation, we transfected astrocytes with expression constructs carrying full length of wild-type Rac1, a dominant negative form of Rac1 (RacT17N) and a constitutively active form of Rac1 (Rac1Q61L). As shown in Fig 2F, Ab increased ROS level in astrocytes expressing the wild-type form of Rac1 (137 AE 2% and 130 AE 2%, n = 3), whereas failed in astrocytes expressing the dominant negative form of Rac1 (88 AE 2% and 90 AE 10%, n = 3). However, expression of RacQ61L resulted in a significant increase in ROS production (135 AE 2%) without Ab-stimulation compared to control cells ( Fig 2F). These results suggest an active role of classical PKCs in Ab-induced Rac1 activation of NOXdependent ROS signaling in astrocytes.
b1-Integrin-PI3K signaling is required for Ab-oligomerinduced PKC and NOX activation Here, we investigated the putative involvement of b1-integrin and PI3K in the molecular mechanisms triggered by Ab to activate NOX in astrocytes. First, integrin-mediated response to Ab-oligomers was examined in acutely isolated astrocytes from rat hippocampus. Fluorometric measurements of [Ca 2+ ] i showed that Ab-oligomers induced intracellular Ca 2+ increases (100%, n = 20 cells obtained from 4 rats, Fig 3A,B) that were blocked by co-incubation of Ab with RGDS peptide (36.6 AE 6.7%, n = 24 cells obtained from 4 rats, Fig 3A,B), which contains the integrin binding sequence and inhibits its function (Wright & Meyer, 1985), and with an antibody (Mendrick & Kelly, 1993) that specifically binds to the 130-kDa b1-integrin chain (aCD29, 35.6 AE 9.6%, n = 20 cells obtained from 4 rats, Fig 3A,B). Moreover, in cultured astrocytes, AKT phosphorylation by Ab-oligomers was reduced by RGDS peptide, indicating that Ab-peptide activates PI3K and PDK1 through integrin activity modulation (Fig 3C,D). The contribution of b1-integrin/PI3K/PDK1 activity to downstream PKC/NOX signaling was examined in Ab-treated astrocytes, together with aCD29 antibody, RGDS peptide, the PI3K inhibitor wortmannin, and the PDK1 inhibitor OSU03012. Western blot analysis of cell extracts and ROS measurements showed that phosphorylation of PKC/PKD and ROS levels was reduced by each specific inhibitor (Fig 3E,G). Thus, our data provide evidence that Ab-oligomers activate integrin-associated signaling pathways that regulate NOX-dependent redox signaling in astrocytes.

Ab-Oligomers regulate b1-integrin maturation and GFAP overexpression by controlling ROS generation
Sequential glycosylation of b1-integrin subunit (85 kDa) to form b1integrin precursor (105 kDa) and its mature form (125 kDa) represents a very important mechanism for its function modulation (Zheng et al., 1994). Next, we further investigated the role of Ab-oligomers in controlling the conversion of the b1-integrin precursor to its mature form in astrocytes. Analysis revealed that, after 6 h of treatment, Ab increased the levels of precursor b1 integrin (145 AE 24%, n = 6; Fig 4A, C), whereas 24 h of treatment increased the levels of the mature form (189 AE 22%, n = 4; Fig 4A,B). DPI inhibitor, RGDS peptide, and antibody aCD29 markedly decreased the levels of precursor and mature form levels (Fig 4A-C), suggesting that the intracellular ROS-produced after binding of Ab-to b1-integrin may modulate functional levels of b1integrin in astrocytes. In addition, we observed that GFAP overexpression in astrocytes treated with Ab-oligomers was reduced by b1-integrin activity inhibition with RGDS peptide and antibody aCD29 (Fig 4D,E).
Taken together, these data show that Ab-oligomers promote b1integrin maturation and GFAP overexpression by mechanisms that involve the Ab-binding to b1-integrin receptors and the subsequent oxidative signaling in astrocytes.

b1-Integrin mediates the effects of soluble Ab-oligomers on hippocampal astrocytes in vivo
To extend the above characterization of biological effects of Aboligomers to in vivo, we injected Ab-oligomers (125 ng) or vehicle alone into hippocampus of C57 adult mice. Ab-injection led to astrocyte reactivity in the dentate gyrus compared with vehicle-injected mice, as revealed by quantification of immunolabeled area with astrocyte markers GFAP (  (B) Real-time PCR was performed using specific primers for the target genes using RNA isolated from cultured astrocytes treated with Ab oligomers. Bars represent fold change of gene expression normalized to values of untreated cells. Data represent the average of three independent cultures. (C-G) Astrocytes were treated with Ab (5 lM) or with Ab together with the NOX inhibitor DPI (5 lM), and protein expression (NOX1, NOX2, and GFAP) was analyzed by Western blot. Histograms represent the intensities of bands normalized to b-actin, or GAPDH levels displayed as a percentage of the nontreated cells or DPI-treated cells (100%). *P < 0.05, **P < 0.01, ***P < 0.001 compared to nontreated cells, # P < 0.05, ## P < 0.01 compared to Ab-treated cells.
hoc IgM+Ab vs CD29 + Ab, P < 0.05) in astrocytes compared with them of isotype IgM-injected mice (n = 6 animals in all instances). Unexpectedly, isotype IgM antibody enhanced significantly the NOX2 activation by Ab compared with Ab alone (Fig 4G,I ANOVA, P < 0.001; Tukey post hoc IgM+Ab vs Ab, P < 0.001). Although the nature of this enhancement is unknown, it could be due to the presence of autoantibodies to Astrocytes were treated with Ab-oligomers (5 lM) for 30 min, and cell extracts were used to measure Rac1 activation (loaded with GTP) by affinity precipitation assay and PAK phosphorylation. Rac 1 and total PAK show total loading controls. (C right, D) Effect of G€ o 6983 (100 nM) treatment in Ab-induced PAK phosphorylation as analyzed by Western blot. Graph bars represent the intensities of pPAK normalized to total PAK loading and expressed as a percentage of untreated cells or inhibitor-treated cells as controls. (E) Differential effects of PKC inhibitors on Ab-induced ROS levels. Astrocytes were pretreated with PKC inhibitors GF109203X, G€ o6983 and Rottlerin and subsequently with Ab 5 lM. (F) Astrocytes were transfected with plasmids coding for Rac1 wt , Rac1 T17N , Rac1 Q61L . In (E) and (F), ROS levels were measured in nontreated cells or cells treated with Ab-oligomers (5 lM, 30, 60 min). The results show the relative fluorescence normalized to untreated cells or PKC inhibitor-treated cells (100%). *P < 0.05, **P < 0.01, ***P < 0.001 compared to nontreated cells; # P < 0.05, ## P < 0.01, ### P < 0.001 compared to Ab-treated cells or to Rac1 wt cells. astrocyte proteins including GFAP or oxidative stress markers (El-Fawal & O'Callaghan, 2008).
Overall, these histological findings demonstrate that amyloid-b oligomers trigger NOX2 upregulation and astrogliosis through b1integrin signaling in the mouse hippocampus.
Next, we analyzed whether b1-integrin and NOX2 levels were elevated in reactive astrocytes in the vicinity of amyloid plaques or apart from them. Quantification of confocal images, as pixel sum of each antibody labeling in GFAP-positive pixels, showed an increased expression of these protein levels in hippocampal astrocytes of 18-month-old 3xTg-AD mice (integrin b1: 158 AE 32% and NOX2: 147 AE 27%), as well as GFAP protein levels (334 AE 36%, n = 3), compared to agematched non-Tg mice (100%). These increases were more pronounced in astrocytes surrounding the amyloid plaques (GFAP: 578 AE 62%, n = 3; b1-integrin: 142 AE 15%, n = 3; NOX2: 210 AE 38%, n = 3) compared to age-matched non-Tg mice (100%) (Fig 5G-I). Taken together, these data show that astrocytes from the hippocampus of 3xTg-AD mice expressed higher b1-integrin and NOX2 levels compared to non-Tg mice and suggest that the increased expression of these proteins correlates with the levels of soluble b-amyloid peptide.

Reactive astrocytes from prefrontal cortex of AD patients show higher levels of b1-integrin and NOX2
In order to assess the relevance to AD of dysregulated levels of b1-integrin and NOX2 in reactive astrocytes, we examined by Western blot, dot blot, and immunohistochemistry assays postmortem samples of prefrontal cortex brain from thirteen control and twenty patients with Alzheimer's disease classified as AD-II, III, IV and V-VI (Table 1) (Braak & Braak, 1995). Interestingly, the Western blot analysis revealed that b1integrin expression was upregulated during the progression of the disease (Fig 6A,C ANOVA, P < 0.01; Tukey post hoc Control vs AD-IV-VI, P < 0.001; Tukey post hoc ADII-II vs AD-IV-VI, P < 0.01). In addition, NOX2 and GFAP expression showed an increase in the latest stages (Fig 6D,E; ANOVA, P < 0.005; Tukey post hoc Control vs ADIV-VI, P < 0.05; Tukey post hoc AD-II-III vs ADIV-VI, P < 0.05, respectively).
Prefrontal cortex brain sections were also analyzed to study whether the aberrant increase in b1-integrin and NOX2 levels was observed in reactive astrocytes clustered around amyloid plaques. Quantification of confocal images from three representative cases of AD patients and matched controls (AD-V-VI; n = 100 astrocytes) showed an increased expression of b1-integrin (Fig 6J,K) and NOX2 levels ( Fig 6I,K) (147 AE 24%, and 146 AE 27%, respectively), as well as GFAP protein levels (179 AE 21%), compared to controls (100%) (Fig 6I-K). Overall these data show that astrocytes from the cortex of AD patients express higher b1-integrin and NOX2 levels compared to nondemented controls, suggesting a role for these proteins in the progression of Alzheimer's disease.

Discussion
Astrogliosis is a neuropathologic hallmark in AD whose severity strongly correlates with the density of reactive astrocytes and the robust increase in GFAP in both brain and CSF (Muramori et al., 1998;Fukuyama et al., 2001). In the present study, we unveiled a new mechanistic pathway that drives astrogliosis in AD-like pathology. Thus, our data indicate that amyloid-b oligomers modulate integrin receptor activity in astrocytes which ultimately results in GFAP upregulation through NADPH oxidasemediated redox signaling (Fig S1, Supporting information). This idea is further supported by the presence of elevated expression of b1-integrin and NOX2 which correlate with amyloid b load in vivo models of AD and in the disease proper.
The primary goal of the current study was to unmask the molecular players that contribute to the dysregulation of astrocyte physiology during AD progression. It is well known in these cells that Ab promotes calcium imbalance by mechanisms that involve endoplasmic reticulum activation and stress , regulation of L-type calcium channel expression (Daschil et al., 2015), and calcium-sensing receptor activation (Chiarini et al., 2016). In addition to these activities, the glial cellular stress generated via NOX enzyme activities may contribute to leading gliamediated neurotoxicity in AD (Angelova & Abramov, 2014). However, the impact of NOX in signal transduction of astrocyte dysfunction has not been explored in detail in experimental paradigms relevant to AD disease. Here, we first observed that Ab oligomers enhance NOX2 activity and its expression in a ROS-dependent manner, which results in increased GFAP protein expression in astrocytes in vitro. Because GFAP induction is a key to astrocyte process extension and their thickening in reactive gliosis in AD (Yang & Wang, 2015), we studied the upstream mechanisms of NOX activation as to shedding light on the pathological remodeling of astroglia associated with AD progression. In particular, we explored the mechanistic relationships among b1-integrin, NOX2, and GFAP expression because amyloid-b peptides can bind b1-integrin (Woo et al., 2015) and in this manner activate NOX enzymes (Moraes et al., 2015). Thus, in neurons, the interaction of amyloid-b peptides with integrins changes cell-adhesion capacity (Sabo et al., 1995) and leads to ROS production, mitochondrial dysfunction, and apoptosis (Woo et al., 2015). In addition, a direct Fig. 5 Increased levels of b1-integrin, NOX2, GFAP in astrocytes of hippocampi of 3xTg-AD (A) Representative images of Western blots of b1-integrin, NOX2, GFAP, and oligomeric Ab-expression in hippocampi of 18-month-old 3xTg-AD and non-Tg mice. (B, C) Graphs showing relative protein levels to b-actin (n = 16 mice). Data are expressed as a percentage of non-Tg values (100%). (C) Protein levels of oligomeric Ab in 3xTg-AD mice were found to correlate with (D) b1-integrin, (E) NOX2, and (F) GFAP in 18-month-old 3xTg-AD mice. (G, H) Photomicrographs of triple immunofluorescence staining for GFAP (green), NOX2, and b1-integrin (red) and Ab (white) on coronal sections from non-Tg and 3xTg-AD mice. Scale bar: 20 lm. (H) Quantitative analysis was performed by measuring fluorescence intensity by pixel sum of b1-integrin and NOX2 localized in GFAP + cells. (I) Bars represent values in 3xTg-AD mice relative to those obtained from non-Tg mice. *P < 0.05, **P < 0.01, compared with non-Tg mice.
b1-Integrin and NOX2 mediate astrogliosis in AD, A. Wyssenbach et al. interaction between amyloid-b peptide and integrin occurs in inflammatory events in AD models, as the activation of LFA1 receptor in neutrophil accumulation on vascular inflammation (Zenaro et al., 2015). However, the contribution of integrin activity to astrogliosis mechanisms in AD has not been reported.
Various properties have been attributed to integrin activity in astrocytes, such as an important role in defining cellular properties of the blood brain barrier in the cerebral cortex (Venkatesan et al., 2015), and its ability to attenuate astrogliosis after spinal cord injury (Renault-Mihara et al., 2011). In addition to those properties, we have demonstrated here that b1-integrin activation by amyloid-b oligomers in cultured astrocytes elevates cytosolic Ca 2+ levels and triggers PI3K/PKC/ Rac/NOX2 signaling which results in NOX2 upregulation, a GFAP level increase, and b1-integrin maturation. These findings strongly suggest that soluble amyloid-b can activate and regulate b1-integrin availability to exacerbate intracellular signals leading to oxidative stress and astrogliosis.
The ability of integrins to bind to ligands is governed by integrin conformation, or activity, and this is an important route to the regulation of integrin function (Paul et al., 2015). In particular, amyloid-b oligomers exhibit direct high-affinity binding to b1-integrin in neutrophils (Zenaro et al., 2015) and in neurons (Woo et al., 2015), inducing a rapid adhesion of both human and mouse neutrophils to endothelial ligands and conformational alteration and loss of surface b1-integrin in neurons. In astrocytes, we observed that amyloid-b oligomers increase PI3K, Rac, and NOX activities, and GFAP levels, which in all instances were reverted by a b1-integrin function-blocking antibody and RGDS peptide. In addition, the conversion of precursor into mature b1-form was promoted by Ab-treatment, suggesting that, unlike in neurons, the cell surface levels and function of b1-integrin in astrocytes is upregulated by amyloid-b oligomers. Together, these findings support the idea that these peptides bind to b1-integrin in the astrocyte plasma membrane and serve as a key initiator of astrogliosis in AD.
The effects observed in cultured astrocytes were validated in vivo in the adult hippocampus by means of microinjection of amyloid-b oligomers allowing a dissection of early astrogliosis via NOX activation prior to frank neurodegeneration. Furthermore, Ab-oligomer-induced astrogliosis was neutralized by an antibody blocking b1-integrin function confirming the role of this cell-adhesion receptor. Correlation of b1integrin expression to astrocyte reactivity has been previously reported in a mouse model of spontaneous seizures. Astrocyte hypertrophy and upregulation of GFAP and vimentin were observed in mice with a conditional deletion of b1-integrin. Interestingly, this reactive gliosis appeared in the absence of cell death and blood-brain barrier disturbances (Robel et al., 2009), suggesting that alterations in b1integrin-mediated signaling may hence be a primary mechanism implicated in eliciting specific aspects of reactive gliosis and confirm the results observed in the Ab-injected mouse model.
Several transgenic mouse reproducing hallmarks of AD pathology have been developed based on the ability of expressing APP, presenilin, and tau mutations. In the current study, the 3xTg-AD was used because brains from these mice accumulate amyloid-b oligomers in an age-dependent manner (Oddo et al., 2006), as confirmed here by quantification of Ab-oligomers and immunohistochemistry in hippocampal samples of 6-, 12-and 18-month-old mice. In addition, we observed a striking correlation between the levels of Ab-load in 3xTg-AD and of astrogliosis as measured by GFAP levels, as well as of b1integrin and NOX2. Indeed, astrogliosis is more prominent near the amyloid plaques which suggest that b-amyloid drives the change in astrocyte phenotype by activating b1-integrin and NOX2 in 3xTg-AD. Finally and in agreement with the findings in this AD model, higher levels of b1-integrin, NOX2, and GFAP were found in samples of the prefrontal cortex and in reactive astrocytes at advanced stages of AD (IV-VI). Importantly, an increase in b1 integrin and NOX2 expression was associated with increased levels of amyloid beta peptide. In addition, immunohistochemical studies in the AD brains revealed prominent astroglia reaction surrounding amyloid b plaques. Overexpression of integrins on human astrocytes has been extensively described in neoplastic human brains. Specifically, b1 integrin dysfunction has been associated with mechanisms of cell migration and brain invasion of glioma and astrocytoma cells (Paulus et al., 1993); however, no data are available for the contribution of b1 integrin to astrocyte dysregulation in neurodegenerative diseases. Moreover, analysis of AD brains has demonstrated that both frontal and temporal regions exhibit a significant increase in NOX2 activity and in NOX2 cytosolic subunits  expression throughout the disease progression, most notably in the more advanced stages of the disease. Unlike our results, the NOX activation in AD brains was exclusively attributed to microglial cells (Ansari and Scheff, 2011) and not to astrocytes or neurons. Further studies are needed to elucidate the function of b1 integrin and NOX2 in astrocytes of human brains.
In conclusion, we have found a new mechanism underlying astrogliosis in AD, which highlights astrocytes as a primary target of amyloid-b via activation of b1-integrin and redox signaling through NOX2. In turn, our results suggest that intermediaries in the signaling cascade described in this study can be promising protein biomarkers in AD progression. Finally, we propose that targeting the mechanisms of b-amyloid-driven and integrin-dependent astrogliosis may render astrocytes functionally competent and possibly ameliorate the course of AD.

Experimental procedures Animals
All experiments were conducted under the supervision and with the approval of our internal animal ethics committee (University of the Basque Country, UPV/EHU). Animals were handled in accordance with the European Communities Council Directive. All possible efforts were made to minimize animal suffering and the number of animals used.

Astrocyte cell culture
Primary cultures of cerebral cortical astrocytes were prepared from P0-P2 Sprague Dawley rats as described previously (McCarthy & de Vellis, 1980). After 8 days, cells were plated onto PDL-coated plates and maintained for 2 days. Culture medium was replaced with IMDM with 1% FBS 24 h before Ab-treatment.

Measurement of ROS generation
Astrocytes (1x10 4 ) were exposed to Ab-oligomers  alone or together with antagonists or inhibitors, as indicated. Cells were loaded with 10 lM CM-H2DCFDA, and ROS levels were assayed as is previously described .

Preparation of freshly isolated astrocytes
Rat brains (P8-P11) were coronally cut into 300-lm-thick slices and maintained in ice-cold aCSF containing 75 mM sucrose bubbled with carbogen (pH 7.4). Slices were incubated for 20 min in aCSF (pH 7.4) supplemented with 1 lM sulforhodamine 101. Acutely isolated cells were obtained from slices as described previously (Matthias et al., 2003). Cells were seeded into coverslips coated with PDL, and astrocytes were identified as red fluorescent labeled cells. Intracellular Ca 2+ levels were determined using Fura2 according to the previously described method .

RT-PCR
Total RNA was isolated from cultured astrocytes (1910 5 ) using RNA mini-prep kit (Agilent Technologies, Santa Clara, CA, USA). First-strand cDNA synthesis was carried out with reverse transcriptase Superscript TMIII (Invitrogen, Barcelona, Spain) using random primers. Specific primers for NOX1-NOX4 and DUOX1, DUOX2 were obtained from a previous study (Reinehr et al., 2007). Real-time quantitative PCRs were carried out with 25 ng of reverse-transcribed RNA and 300 nM of primers diluted in SYBRGreen PCR master mix reagent (Invitrogen, Barcelona, Spain). Relative levels of expression of target genes were calculated by means of a normalization factor, based on the geometric mean of multiple internal control genes and calculated by software.

Brain specimens
Formalin-fixed paraffin-embedded sections from the prefrontal cortex and frozen samples of 25 patients with AD as well as thirteen controls were obtained from the Neurological Tissue Bank Hospital Cl ınic-IDIBAPS Biobank (Table 1). AD samples were grouped by Braak and Braak classification (Braak & Braak, 1995), into four groups: AD-II, AD-III, AD-IV, AD-V-VI. All samples were matched by age and gender. performed as previously described . Fluorescencestained sections were observed and photographed with a Leica TCS SP8 confocal microscope, and the pixel sum of NOX2 and b1-integrin in GFAP-positive pixels were quantified with the LASAF software (Leica Microsystems, Mannheim, Germany).

Statistical analysis
One-way analyses of variance followed by Tukey post hoc tests, onetailed Student's t-tests, and the Pearson correlation coefficient were used unless otherwise indicated. All data are represented as mean AEs.e.m. Statistical significance was set at P ≤ 0.05.

Supporting Information
Additional Supporting Information may be found online in the supporting information tab for this article:  b1-Integrin and NOX2 mediate astrogliosis in AD, A. Wyssenbach et al.