Amyloid-β 25-35 Induces Neurotoxicity through the Up-Regulation of Astrocytic System Xc−

Amyloid-β (Aβ) deposition, a hallmark of Alzheimer’s disease, is known to induce free radical production and oxidative stress, leading to neuronal damage. During oxidative stress, several cell types (including astrocytes) can activate the nuclear factor erythroid 2-related factor 2 (Nrf2), a regulator of several phase II detoxifying and antioxidant genes, such as the System Xc− subunit xCT. Here, we studied (i) the effect of the Aβ fragment 25-35 (Aβ25-35) on Nrf2-dependent System Xc− expression in U373 human astroglial cells and (ii) the effect of Aβ25-35-induced astrocytic response on neuronal cell viability using an in vitro co-culture system. We found that Aβ25-35 was able to activate an antioxidant response in astrocytes, by inducing both Nrf2 activation and System Xc− up-regulation. However, this astrocytic response caused an enhanced cell mortality of co-cultured SH-SY5Y cells, taken as a neuronal model. Consistently, the specific System Xc− inhibitor sulfasalazine prevented the increase of both neuronal mortality and extracellular glutamate levels, thus indicating that the neurotoxic effect was due to an augmented release of glutamate through the transporter. The involvement of NMDA receptor activation in this pathway was also demonstrated using the specific inhibitor MK801 that completely restored neuronal viability at the control levels. The present study sheds light on the Nrf2/system Xc− pathway in the toxicity induced by Aβ25-35 and may help to better understand the involvement of astrocytes in neuronal death during Alzheimer’s disease.


Introduction
Alzheimer's disease (AD) is recognized by World Health Organization as a global public health priority, being the leading cause of dementia, responsible for 50-75% of all world cases and results in the deterioration of selective cognitive performance, including memory and mental processing [1,2]. AD is characterized by deposition of amyloid-β peptide (Aβ) in senile plaques, intracellular neurofibrillary tangles consisting of hyper-phosphorylated tau, synaptic dysfunction and neuronal death. Some or all of these hallmarks are causally linked to the cognitive and behavioral deficits that denote this disease [1,3,4].
Most of the known genetic, medical, environmental, and lifestyle-related risk factors for AD are associated with increased oxidative stress [10,11]. Indeed, AD brain is under strong oxidative stress, manifested by increased protein and DNA oxidation, lipid peroxidation, free radical formation, nitro-tyrosine levels, and advanced glycation end products [12][13][14]. Noteworthy, we previously demonstrated that both Aβ [25][26][27][28][29][30][31][32][33][34][35] and Aβ  are able to induce oxidative stress in endothelial cells, by producing superoxide and hydroxyl radicals [15]. Moreover, Aβ can form pore in astrocytes membranes and allow the influx of calcium from the extracellular space [16]. This modulation of calcium levels can induce the activation of NADPH oxidase and the subsequent reactive oxygen species (ROS) production, thus leading to Aβ-induced oxidative stress in astrocytes [17,18]. Chronic oxidative stress conditions arise because of an imbalance between the production of prooxidant molecules (e.g., ROS) and antioxidant system (e.g., intracellular glutathione (GSH) production), in favor of pro-oxidant molecules [19].
During oxidative stress, a variety of cell types are able to up-regulate the activity of nuclear factor erythroid 2-related factor 2 (Nrf2), the main regulator of the antioxidant response, thus counteracting intracellular ROS accumulation and GSH depletion [20,21]. Unlike neurons, astrocytes can strongly up-regulate Nrf2-mediated gene expression, thus leading to a major resistance to oxidative injury than isolated neurons [22]. Upon changes in cellular redox state, Nrf2 migrates to the nucleus and successively binds to promoter regions, known as antioxidant responsive element (ARE), of many phase II detoxifying and antioxidant genes, such as catalase (CAT), γ-glutamyl-cysteine ligase (GCL), superoxide dismutase (SOD), heme-oxygenase-1 (HO-1), glutathione peroxidase (GPX), and System X c − subunit xCT [23,24]. System X c − is an amino acid antiporter and mediates the exchange of intracellular L-glutamate and extracellular L-cystine across the plasma membrane [25]. In astrocytes, L-cystine import through System X c − is crucial to glutathione production and protection from oxidative stress.
On the other side, however, glutamate export is a further route of release through which this neurotransmitter may provoke excitotoxicity [26,27]. Thus, System X c − has currently been related to both pathological and physiological processes in the central nervous system (CNS) [28]. Even though the induction of Nrf2-dependent gene expression has been commonly reported as a protective mechanism to withstand the effects of oxidative stress in astrocytes, the up-regulation of xCT induced by Nrf2 could be a possible source for excitotoxicity due to excessive release of glutamate [29].
Despite many clinical and experimental data suggesting astrocytes as the cell population liable for most of the neuronal death in several neurodegenerative disorders, the specific cellular mechanisms are not yet clearly defined.
Here, we studied the effect of the Aβ 25-35 on the induction of astroglial antioxidant response, focusing on the activation of Nrf2 transcription factor and on System X c − expression. We also analyzed the effect of astrocytic System X c − upregulation on the viability of neuronal cells in a co-culture system.

SH-SY5Y Cell Differentiation
SH-SY5Y cells were seeded in a confluent monolayer in culture dishes established for the experiments. For neuronal differentiation, cells were cultured for a week in Neurobasal medium (Gibco; Milan, Italy) supplemented with 2 mM L-glutamine, 10 µM Retinoic Acid (Sigma, Milan, Italy) and 1X B-27 supplement (Gibco). The medium was changed every two days.

MTT Assay
To test neuronal viability, differentiated SH-SY5Y cells were grown alone and in co-cultures with U373 using a Transwell culture system as previously reported [29]. For each sample in co-cultures, 1.5 × 10 4 neuronal cells were seeded in Transwell insert and 3 × 10 4 astroglial cells were plated in the lower compartment of a 6-well plate and allowed to grow for 24 h.
MTT assay was performed at the end of the incubation period as indicated by manufacturer's instructions and as reported elsewhere [29].

Preparation of Nuclear and Total Extracts
After treatments, the cells were mechanically detached with a scraper in cold PBS. Nuclear extracts were prepared as reported elsewhere [29]. The protein content of nuclear extracts was measured according to Bradford method [31]. The quality of fraction separation was verified by blotting nuclear and cytosolic fractions for the specific markers, lamin A and actin, respectively. Total extracts were prepared by mechanically detaching the cells with a scraper in cold PBS. Extracts were then prepared as reported in [29]. The total protein content was determined according to Bradford method [31].

Evaluation of Nrf2 Activation and System X c − Expression by Western Blotting
To measure Nrf2 nuclear levels, equal amounts of nuclear extracts (20 µg proteins/sample) were loaded in an 8% polyacrylamide gel, subjected to electrophoresis and transferred to nitrocellulose. After incubation with 5% non-fat dry milk for 1 h, membranes were incubated at 4 • C overnight with the polyclonal anti-Nrf2 antibody (1:1000) or with a polyclonal anti-lamin A (1:1000). To analyze System X c − expression, equal amounts of total extracts (15 µg proteins/sample) were subjected to SDS-PAGE. Electrophoresis was performed using a 10% polyacrylamide gel. Membranes were blotted with anti-xCT polyclonal antibody (1:5000) or polyclonal anti-actin antibody (1:1000).
Actin and lamin A were used as reference proteins for total and nuclear extracts, respectively. Anti-rabbit secondary antibody labeled with peroxidase was used at 1:10,000 dilution. ECL Western blotting detection reagents was used to detect immunoreactive bands that were captured by Chemi Doc TM XRS 2015 (Bio-Rad Laboratories, Hercules, CA, USA). Densitometric analysis was carried out using Image Lab software (Version 5.2.1; © Bio-Rad Laboratories).

Analysis of Glutamate Concentration in Cell Supernatants
The release of glutamate in co-culture supernatants was evaluated using Glutamate Assay kit (BioVision; Florence, Italy), as reported elsewhere [29]. The concentration of glutamate in each sample was estimated using glutamate standard curve.
2.9. Immunofluorescence Analysis 1.5 × 10 5 U373 cells were seeded on 6-well dishes containing poly-L-lysine-treated glass coverslip. After treatments, cells were washed in PBS and fixed with 4% paraformaldehyde for 10 min at room temperature (RT) and permeabilized in methanol for 10 min at −20 • C. Cells were then washed and incubated for 1 h at RT with a blocking solution (5% FBS, 1% BSA in PBS). Next, coverslips were incubated overnight at 4 • C with polyclonal anti-System X c − (1:100). The secondary antibody anti-Rabbit IgG conjugated with Alexa Fluor 488 was diluted 1:500 and incubated at RT for 1 h. The nuclei were counterstained using Hoechst 33342 (Cod. H3570, Invitrogen; Milan, Italy, Thermofisher Scientific).

Statistical Analysis
Values are expressed as the mean ± standard error of the mean (SEM) of n observations. Statistical analysis was carried out by one-way ANOVA and subsequently by Bonferroni post-test. Differences are considered statistically significant at p ≤ 0.05.

Results and Discussion
Free radical production and oxidative stress play crucial roles in many neurodegenerative diseases, including Alzheimer's disease [12]. In many cell types, including astrocytes, ROS can activate a protective antioxidant response through Nrf2-mediated induction of antioxidant and phase II detoxifying genes (i.e., ARE genes).

Aβ 25-35 Activates Nrf2 in Astroglial Cells
Firstly, we investigated whether Aβ 25-35 could activate Nrf2 in astroglial cells. To this aim, U373 cells were treated with Aβ 25-35 (50 µM) for 2, 4 and 24 h and Nrf2 levels were measured in nuclear extracts by Western blot analysis. The results shown in Figure 1 indicate that Aβ 25-35 induced a 2.3-fold increase of the nuclear Nrf2 levels already at 2 h post-treatment and a 1.9-fold increase after 4 h of treatment.
Firstly, we investigated whether Aβ25-35 could activate Nrf2 in astroglial cells. To this aim, U373 cells were treated with Aβ25-35 (50 μM) for 2, 4 and 24 h and Nrf2 levels were measured in nuclear extracts by Western blot analysis. The results shown in Figure 1 indicate that Aβ25-35 induced a 2.3-fold increase of the nuclear Nrf2 levels already at 2 h posttreatment and a 1.9-fold increase after 4 h of treatment. Data are calculated relative to the housekeeping gene (i.e., nuclear lamin A) content and are the means ± SEM from three separate experiments, each performed in duplicate. One-way ANOVA, followed by Bonferroni's test, was used to define significant differences. * p ≤ 0.05 vs. CTRL; ** p ≤ 0.01 vs. CTRL.

Aβ25-35 Induces ARE Gene Expression in Astroglial Cells
Secondly, we verified whether Aβ-induced Nrf2 was able to regulate antioxidant ARE genes. In this respect, we found that the treatment of human U373 astroglial cells with Aβ25-35 (50 μM) for 4, 8, and 16 h was able to increase the mRNA expression of enzymes involved in maintenance of redox state, such as SOD1, SOD2, CAT, HO-1, GPX3, and GCLC ( Figure 2). The peak was reached at 8 h for SOD1, SOD2, GPX3, and GCLC, whereas CAT and HO-1 expression peaked at 4 h after Aβ25-35 treatment. This time frame was compatible with the earlier transcriptional activation of Nrf2. Data are calculated relative to the housekeeping gene (i.e., nuclear lamin A) content and are the means ± SEM from three separate experiments, each performed in duplicate. One-way ANOVA, followed by Bonferroni's test, was used to define significant differences. * p ≤ 0.05 vs. CTRL; ** p ≤ 0.01 vs. CTRL.

Aβ 25-35 Induces ARE Gene Expression in Astroglial Cells
Secondly, we verified whether Aβ-induced Nrf2 was able to regulate antioxidant ARE genes. In this respect, we found that the treatment of human U373 astroglial cells with

Aβ 25-35 Induces System X c − in Astroglial Cells
Among ARE genes, we further focused our attention on System X c − , which is involved in the maintenance of GSH intracellular levels and the redox state. In the same experimental conditions described above, we found that Aβ [25][26][27][28][29][30][31][32][33][34][35] was able to increase the mRNA expression of xCT, the catalytic subunit of System X c − . As shown in Figure 3, we observed a maximum reached at 8 h post-treatment.
Consistently, we observed that Aβ 25-35 was able to up-regulate System X c − also at protein level. In particular, a 24 h treatment of U373 cells with Aβ 25-35 (50 µM) caused a two-fold increase of System X c − protein levels in whole cell extracts when compared to controls, as verified by Western blot analyses (see Figure 4).
Furthermore, we used the confocal microscopy to also evaluate the expression and localization of System X c − in U373 cells treated with Aβ 25-35 (50 µM) for 24 h. Figure 5 shows an increased levels of System X c − , the latter being mainly localized on plasma membrane of treated cells compared to controls at 24 h post-treatment.

Aβ25-35 Induces System Xc − in Astroglial Cells
Among ARE genes, we further focused our attention on System Xc − , which is involved in the maintenance of GSH intracellular levels and the redox state. In the same experimental conditions described above, we found that Aβ25-35 was able to increase the mRNA  After incubation at 37 °C, cells were homogenized, and total RNA has been purified to evaluate mRNA content of System Xc − by RT-qPCR. Results are computed relative to GAPDH content, taken as a housekeeping gene. Bars are the means ± SEM from three separate experiments, each performed in duplicate. One-way ANOVA, followed by Bonferroni's test, was used to define significant differences. * p ≤ 0.05 vs. CTRL.
Consistently, we observed that Aβ25-35 was able to up-regulate System Xc − also at protein level. In particular, a 24 h treatment of U373 cells with Aβ25-35 (50 μM) caused a twofold increase of System Xc − protein levels in whole cell extracts when compared to controls, as verified by Western blot analyses (see Figure 4). The graph shows the densitometric analysis of the western blots for each sample. Data are computed relative to the internal housekeeping gene (actin) and are the means ± SEM from three separate experiments, each carried out in duplicate. One-way ANOVA, followed by Bonferroni's test, was used to define significant differences. ** p ≤ 0.01 vs. CTRL. Furthermore, we used the confocal microscopy to also evaluate the expression and localization of System Xc − in U373 cells treated with Aβ25-35 (50 μM) for 24 h. Figure 5 shows  Consistently, we observed that Aβ25-35 was able to up-regulate System Xc − also at protein level. In particular, a 24 h treatment of U373 cells with Aβ25-35 (50 μM) caused a twofold increase of System Xc − protein levels in whole cell extracts when compared to controls, as verified by Western blot analyses (see Figure 4). The graph shows the densitometric analysis of the western blots for each sample. Data are computed relative to the internal housekeeping gene (actin) and are the means ± SEM from three separate experiments, each carried out in duplicate. One-way ANOVA, followed by Bonferroni's test, was used to define significant differences. ** p ≤ 0.01 vs. CTRL. Furthermore, we used the confocal microscopy to also evaluate the expression and localization of System Xc − in U373 cells treated with Aβ25-35 (50 μM) for 24 h. Figure 5 shows

Aβ25-35 Induces Glutamate Release through System Xc −
The above results would seem to be in agreement with the concept that astrocytes play an important role in providing antioxidant support to neighboring neurons. In fact, post-mitotic neurons are thought to survive for many decades despite their relatively low The above results would seem to be in agreement with the concept that astrocytes play an important role in providing antioxidant support to neighboring neurons. In fact, post-mitotic neurons are thought to survive for many decades despite their relatively low intrinsic antioxidant defenses [22,[32][33][34]. Given the role of System X c − in providing the cell with cystine, its augmented expression and/or activity increases intracellular levels of cysteine. That is the rate-limiting substrate for the synthesis of GSH, thereby endowing astrocytes with an effective antioxidant response.
Nevertheless, the up-regulation of System X c − can increase extracellular glutamate release and potentially cause excitotoxicity. To verify whether the treatment of astrocytes with Aβ [25][26][27][28][29][30][31][32][33][34][35] can enhance the release of glutamate in the extracellular space, we quantified the levels of glutamate in the supernatant of U373 co-cultured with differentiated SH-SY5Y cells in the presence of Aβ 25-35 (50 µM) for 24 h. As shown in Figure 6, Aβ 25-35 -treated cells released about 50% more glutamate with respect to untreated cells. As a control, we found that glutamate was also released from mono-cultured U373 cells treated with Aβ 25-35 for 24 h, thus proving its astroglial origin (data not shown).
24 h and subjected to immunofluorescence staining using anti-System Xc − 1:100 (OriGene, green) antibodies as reported in Materials and Methods. Nuclei (blue) are stained with Hoechst 33342.

Aβ25-35 Induces Glutamate Release through System Xc −
The above results would seem to be in agreement with the concept that astrocytes play an important role in providing antioxidant support to neighboring neurons. In fact, post-mitotic neurons are thought to survive for many decades despite their relatively low intrinsic antioxidant defenses [22,[32][33][34]. Given the role of System Xc − in providing the cell with cystine, its augmented expression and/or activity increases intracellular levels of cysteine. That is the rate-limiting substrate for the synthesis of GSH, thereby endowing astrocytes with an effective antioxidant response.
Nevertheless, the up-regulation of System Xc − can increase extracellular glutamate release and potentially cause excitotoxicity. To verify whether the treatment of astrocytes with Aβ25-35 can enhance the release of glutamate in the extracellular space, we quantified the levels of glutamate in the supernatant of U373 co-cultured with differentiated SH-SY5Y cells in the presence of Aβ25-35 (50 μM) for 24 h. As shown in Figure 6, Aβ25-35-treated cells released about 50% more glutamate with respect to untreated cells. As a control, we found that glutamate was also released from mono-cultured U373 cells treated with Aβ25- 35 for 24 h, thus proving its astroglial origin (data not shown).  To confirm that Aβ 25-35 -elicited glutamate release occurred through System X c − activation, we analyzed the levels of glutamate in the supernatants of co-cultures in the presence of sulfasalazine (SSZ; 300 µM), a specific inhibitor of System X c − . We observed that SSZ treatment prevented Aβ 25-35 -induced glutamate release, reducing its levels in the extracellular space. These data clearly demonstrate that in astroglial cells the treatment with Aβ 25-35 increases the release of glutamate by eliciting System X c − up-regulation (see Figure 6).
ure 7a). Afterwards, we evaluated the viability of neuronal cells grown in co-cultures with astrocytes, in the presence of Aβ25-35. As shown in Figure 7b, a treatment with Aβ25-35 for 24 h caused a significant reduction of 50% less viability of differentiated neuronal SH-SY5Y cells co-cultured with U373 cells in comparison to untreated cells. To verify whether Aβ25-35-induced neurotoxicity was due to an enhancement of glutamate export through System Xc − , we carried out the MTT assay in co-cultures treated with Aβ25-35 for 24 h in the presence of SSZ (300 μM). Our results indicate that SSZ prevented neurotoxicity in SH-SY5Y cells co-cultured with U373 cells, thus restoring neuronal viability at the control level (see Figure 7b). These data suggest that Aβ25-35-induced neurotoxic effect is mediated by increased glutamate release due to System Xc − up-regulation in astroglial cells.  The release of glutamate via System X c − from both microglia and astrocytes has been found to increase excitotoxicity of cortical neurons [35][36][37][38][39][40]. Noteworthy, changes in glutamate transport have been demonstrated in a mouse model for Alzheimer's disease. Enhanced cortical expression of VGLUT3 and xCT along with a strong trend towards increased cortical extracellular glutamate levels have been reported elsewhere [41]. In neurons, excitotoxicity occurs via glutamate-induced overactivation of NMDA receptor and subsequent perturbed cellular calcium homeostasis and mitochondrial alterations [42,43].
To assess whether the reduced neuronal viability, as triggered by extracellular glutamate release through System X c − , was effectively due to the activation of NMDA receptor, we performed MTT assay on SH-SY5Y (grown in co-cultures with U373 cells) treated for 24 h with Aβ [25][26][27][28][29][30][31][32][33][34][35] in the presence of MK801 (10 µM), an NMDA receptor antagonist. As shown in Figure 7b, MK801 prevented neuronal toxicity restoring the percentage of neuronal living cells at the control level. These results clearly indicate that Aβ 25-35 -induced neurotoxicity is mediated by the activation of NMDA receptor elicited by System X c − -dependent glutamate release.
Note that during neuroinflammation, activated astrocytes and microglia have been reported to release and maintain high concentrations of extracellular glutamate [40]. Moreover, excitotoxic glutamate release and high levels of System X c − expression were observed in microglia, due to a prolonged need for oxidative protection [44]. Interestingly, neurons co-cultured with astrocytes were observed to be more susceptible than neurons alone to hypoxic cell death after treatment with IL-1β, an effect being mediated by enhanced glutamate efflux from astrocytes through System X c − [36]. Very recently, we have reported that HIV-1 Tat protein was able to induce neurotoxicity by eliciting Nrf2-mediated System X c − activation [29]. Previously, we reported that HIV-1 Tat can induce neuro-toxicity by eliciting the spermine oxidase-dependent ROS generation through NMDA receptor stimulation in SH-SY5Y cells, which in turn leads to GSH depletion and oxidative stress [45]. Finally, we have recently demonstrated that System X c − participated in the increase of glutamate excitotoxicity in the neocortex of a mouse model (Dach-SMOX), displaying a chronic oxidative stress [46]. Although Aβ 25-35 may have some limits in representing the whole Aβ peptide and is a scarce version in vivo, it is particularly worthy of attention in light of its great oxidative stress generation capacity and extreme toxicity in neuronal cells and synaptosomes [9]. Altogether, our data show how inflammatory pathways and oxidative stress may converge to an intersection point, represented by activation of System X c − , thereby suggesting a possible explanation for the mechanism involved in excitotoxicity induced by Aβ [25][26][27][28][29][30][31][32][33][34][35] . It should be pointed out, however, that Aβ 25-35 is just one of the fragments and this does not exclude the possibility that other parts of Aβ, than the 25-35 fragment, can be involved in the induction of Nrf2/SystemX c − pathway.

Conclusions
Given the role played by astrocytes in maintaining the homeostasis of extracellular space in the brain, their response may affect neuronal function and provide antioxidant support to neighboring neurons both in normal and pathological conditions. Here, we show that Aβ 25-35 is able to trigger an antioxidant response in astrocytes, by inducing both Nrf2 and ARE-driven genes, including System X c − . However, the induction of System X c − in astrocytes seems to be also responsible for mortality of neuronal cells, due to sustained glutamate release and neuronal NMDA receptor activation. Although further studies are needed, it is tempting to speculate that neurodegeneration can be exacerbated by converting oxidative stress to excitotoxicity via System X c − (for a schematic model see Figure 8).  In astrocytes, Aβ25-35 elicits an antioxidant response by transcriptionally inducing Nrf2-driven ARE genes, such as CAT, HO-1, SOD1, SOD2, GPX3, GCLC, and System Xc − . While the L-cystine import through System Xc − is crucial to protection from oxidative stress (e.g., GSH production), the export of glutamate may cause neurodegeneration through the activation of NMDAr on neuronal cells. For more details see text.
In conclusion, the present study highlights the importance of the Nrf2/System Xc − pathway for a better understanding of the role of astrocytes as a cell population responsi- X c − is crucial to protection from oxidative stress (e.g., GSH production), the export of glutamate may cause neurodegeneration through the activation of NMDAr on neuronal cells. For more details see text.Abbreviations: Aβ, amyloid-β; ARE, antioxidant responsive element; CAT, catalase; GCLC, glutamate-cysteine ligase; GPX3, glutathione peroxidase; GSH, reduced glutathione; HO-1, heme-oxygenase-1; NMDAr, N-methyl-D-aspartate receptor; Nrf2, nuclear factor erythroid 2-related factor 2; SOD, superoxide dismutase.
In conclusion, the present study highlights the importance of the Nrf2/System X c − pathway for a better understanding of the role of astrocytes as a cell population responsible for the death of neurons in AD.