Gene expression, proteome and calcium signaling alterations in immortalized hippocampal astrocytes from an Alzheimer’s disease mouse model

Evidence is rapidly growing regarding a role of astroglial cells in the pathogenesis of Alzheimer’s disease (AD), and the hippocampus is one of the important brain regions affected in AD. While primary astroglial cultures, both from wild-type mice and from rodent models of AD, have been useful for studying astrocyte-specific alterations, the limited cell number and short primary culture lifetime have limited the use of primary hippocampal astrocytes. To overcome these limitations, we have now established immortalized astroglial cell lines from the hippocampus of 3xTg-AD and wild-type control mice (3Tg-iAstro and WT-iAstro, respectively). Both 3Tg-iAstro and WT-iAstro maintain an astroglial phenotype and markers (glutamine synthetase, aldehyde dehydrogenase 1 family member L1 and aquaporin-4) but display proliferative potential until at least passage 25. Furthermore, these cell lines maintain the potassium inward rectifying (Kir) current and present transcriptional and proteomic profiles compatible with primary astrocytes. Importantly, differences between the 3Tg-iAstro and WT-iAstro cell lines in terms of calcium signaling and in terms of transcriptional changes can be re-conducted to the changes previously reported in primary astroglial cells. To illustrate the versatility of this model we performed shotgun mass spectrometry proteomic analysis and found that proteins related to RNA binding and ribosome are differentially expressed in 3Tg-iAstro vs WT-iAstro. In summary, we present here immortalized hippocampal astrocytes from WT and 3xTg-AD mice that might be a useful model to speed up research on the role of astrocytes in AD.


Introduction
While in Alzheimer's disease (AD) astrocytes have been historically associated with reactive gliosis and neuroinflammation, a growing body of evidence suggests that astroglial alterations occur in the early stages of AD, compromising their housekeeping and homeostatic functions that in turn may result in synaptic and neuronal malfunction 1,2 .
Most of the information about the role of astrocytes in brain pathology has been collected in in vitro experiments on primary cultures. Easy to prepare and handle, astroglial primary cultures are made from different brain areas 3,4 and different animal species [5][6][7][8] . However, primary cultures present several limitations such as inter-culture variability, short culture lifetime and limited number of cells if cultures are prepared from specific brain areas, such as the hippocampus. To overcome these limitations, immortalized astroglial lines have been proposed. The first attempts to generate a permanent astroglial cell line, not deriving from brain tumors, date back to early eighties 9 . Since then, a number of immortalized astroglial lines have been established [10][11][12][13][14][15][16][17][18][19][20][21][22][23] . Most of them derive from a highly heterogeneous population of cortical primary astrocytes; however, immortalization of astrocytes from brain regions other than cortex have also been reported, e.g., from the cerebellum 9 or from the midbrain 21 . Surprisingly, few attempts have been reported to immortalize astrocytes from hippocampus and also from animal models of AD. In this regard, Morikawa et al. 18 have generated immortalized astrocytes from ApoE2, ApoE3 and ApoE4 knock-in mice.
In this report we introduce immortalized astroglial lines from the hippocampus of a well-characterized AD mouse model, 3xTg-AD, and from wild-type (WT) control mice, from now on referred to as 3Tg-iAstro and WT-iAstro, respectively. WT-iAstro cell lines show features of primary hippocampal astrocytes such as basic electrophysiological properties, and a similar transcriptional and proteomic profile. More importantly, 3Tg-iAstro show alterations in transcription and deregulation of Ca 2+ signaling as it was reported for its primary counterparts. To illustrate the versatility of this model we also performed shotgun mass spectrometry proteomic analysis and found that proteins related to RNA binding and ribosome are differentially expressed in 3Tg-iAstro vs WT-iAstro. These data demonstrate that iAstro lines represent a versatile and useful cellular model to investigate astroglial AD-related pathobiology.

Results
Generation of immortalized hippocampal astrocytes (iAstro) from WT and 3xTg-AD mice Six immortalized cell lines from WT (WT-iAstro#1-6) and from 3xTg-AD (3Tg-iAstro#1-6) mice were generated, from separate primary astrocyte cell cultures. For immortalization, primary astroglial cultures were first depleted of microglial cells by magnetic-assisted cell sorting (MACS) using anti-CD11b-conjugated microbeads in order to obtain a population of highly purified astrocytes. Astrocytes were then transduced using retrovirus expressing SV40 large T antigen. Transformed cells were selected in G418, amplified and stabilized for 12 passages prior to characterization. No clonal selection was performed to maintain the natural hererogeneity of the cultures.
For logistic and experimental setting convenience, four lines for each strain were characterized for morphology and astroglial marker expression. The other two cell lines were confirmed for morphological identity to other lines, but were maintained as backups and have not been characterized (#1 and #5 for WT-iAstro and #1 and #4 for 3Tg-iAstro lines).
iAstro cells show a morphology similar and virtually indistinguishable to those of primary hippocampal astrocytes in bright field microscopy (Fig. 1a). We also evaluated the expression of the astroglial markers aquaporin-4 (AQP4), glutamine synthetase (GS) and aldehyde dehydrogenase 1 family member l1 (Aldh1l1) (Fig. 1b) by immunocytochemistry. Importantly, we found expression of all three markers in all cell lines. Immunocytochemical analysis for glial fibrillary acidic protein (GFAP) showed, instead, that only a small proportion of cells were positive in the established immortalized cell lines (15.4 ± 5.3 % in WT-iAstro vs 16.7 ± 5.9 % in 3Tg-iAstro cells, p > 0.05) (Fig. 1c), while 100% of cells were GFAP positive in WTand 3Tg-primary astrocytes.
We also decided to have a quantitative measure of the expression of the three markers by evaluating protein levels in western blotting on three of the four cell lines. The three markers were mostly decreased compared to primary cell lines, albeit at different levels. AQP4 was the least decreased, while Aldh1l1 was the most decreased, with qualitative consistency in the immortalized cell lines tested. Importantly, not only were all the markers detectable, but they were also represented at similar levels between the WT-and 3Tg-iAstro cell lines (Fig. 2).
We also evaluated the ability of the iAstro cell lines to be passaged in culture. iAstro lines did not change their morphology or marker expression significantly at least up to 20th passage (not shown).
Kir currents are present but are not different in WT-iAstro and 3Tg-iAstro Maintenance of ionic balance is one of the housekeeping functions of astrocytes and a feature which might have been lost during immortalization. Potassium buffering by inwardly rectifying K + (Kir) channels during neuronal activity is fundamental to maintain adequate synaptic transmission and neuronal excitability 24 . We have previously shown that Kir channels are functionally expressed in primary hippocampal astrocytes 25 . Therefore, we performed patch-clamp experiments in WT-iAstro#2 and 3Tg-iAstro#2 lines. To confirm that Kir function was not affected by the immortalization process, control hippocampal mouse astrocytes were also patched in order to record Kir current. Cells were exposed to a step protocol of increasing voltage (20 mV increments) from −180 mV to +60 mV to record current elicited by Kir channels. Before each voltage step increment, cells were kept at 0 mV for 300 ms in order to block outward potassium flow 26 . Application of this protocol triggered Kir current in the cell lines tested (Fig. 3a). Current (I), measured at each step of recording protocol, was plotted over corresponding applied voltage (V) to determine I/V curve for Kir channels in both cell lines. Raw current values were normalized to maximum I obtained at +60 mV for each cell (Fig. 3b). No differences were found in iAstro Kir currents through at least five passages. No significant differences were observed in Kir currents between control primary astrocytes and WTand 3Tg-iAstro lines.

iAstro cells are capable of glutamate uptake
Another fundamental function of astrocytes in the regulation of synaptic transmission is the uptake of glutamate through the action of sodium-dependent glutamate transporters 27 . In the hippocampus, GLT-1 (excitatory amino acid transporter 2 (EAAT2)) is the major glutamate transporter 28 . Therefore, we investigated the expression of GLT-1 and glutamate uptake in iAstro lines. As shown in Supplementary Figure 1A, anti-GLT-1 staining revealed membrane-localized expression of GLT-1 in both primary astrocytes and in iAstro lines. Supplementary Figure 1B shows that WT-iAstro cells are capable of glutamate uptake from the medium with a rate (10.73 ± 1.19 μmol/g protein, n = 3) comparable to that of primary astrocytes (19.38 ± 3.74 μmol/g protein, n = 3; p = 0.092). No differences were found in glutamate uptake between WT-iAstro and 3Tg-iAstro lines (10.77 ± 2.14 μmol/g protein, n = 3; p = 0.12).
ATP-induced Ca 2+ signaling is altered in immortalized hippocampal astrocytes from 3xTg-AD mice Astrocytes respond to external stimuli primarily by generating intracellular Ca 2+ signals, employing a combination of release of Ca 2+ from the internal stores via metabotropic mechanisms and Ca 2+ entry from the extracellular space through the plasma membrane via store-operated Ca 2+ entry [29][30][31] . We have previously shown that astrocytic Ca 2+ signaling in AD is altered using a variety of protocols/models usually employed for neuronal evaluation, including Aβ oligomer treatment 32 , an uncoupling peptide of ADAM10 and SAP97 33 , or triple transgenic mice 34 . Astrocytic Ca 2+ signaling has been shown to be altered in AD also by others in vivo and in vitro [35][36][37][38] .
In the current experimental setting the astroglial Ca 2+ signals are likely to be mediated by two types of purinergic receptors, P2Y1 and P2Y2. Relative quantification of messenger RNA (mRNA) of P2ry1 and P2ry2 in WT and 3Tg-iAstro lines using real-time PCR showed a higher levels of P2ry2 mRNA compared with P2ry1; moreover, P2ry2 was significantly more expressed in 3Tg-iAstro compared with WT, while levels of P2ry1 were not different. These data suggest that P2Y2 but not P2Y1 may mediate enhanced sensitivity of 3Tg-iAstro to ATP (Fig. 4d).
Surprisingly, and contrary to our previous observations on primary astrocytes 34 , we found that iAstro cell lines did not respond to DHPG.
Our aim, though, was to validate, independently of the expression levels, whether the differences in expression between primary WT and 3Tg astrocytes could be recapitulated in the immortalized cell lines. As shown in Fig. 6a, all 14 genes followed the changes found previously in primary astrocytes 40 . A plot in Fig. 6b shows high degree of correlation between primary and immortalized astrocytes. Taken together, these results show that after the immortalization hippocampal astrocytes retain the transcriptional changes found in primary 3xTg-AD vs WT astrocytes for, at least, a selected group of genes.

Immortalized astrocytes up-regulate iNOS in response to pro-inflammatory stimuli
The role of astrocytes in neuroinflammation is well ascertained 31,41 , and therefore we have investigated if iAstro lines responded to pro-inflammatory stimuli like bacterial lipopolysaccharide (LPS) or tumor necrosis factor-α (TNFα). As shown in Supplementary Figure 2, inducible nitric oxide synthase (iNOS) was induced upon treatment with both LPS (100 ng/ml for 3 h) and TNFα treatment (20 ng/ml for 6 h). No differences were found between WT and 3Tg-primary astrocytes or iAstro lines.

Mass spectrometry proteomics revealed alterations in translation and ribosome in 3Tg-iAstro compared to WT-iAstro
To further evaluate the iAstro lines, we performed a shotgun mass spectrometry proteomic analysis using the four WT and four 3Tg-iAstro lines. In total, 1119 and 1045 proteins were identified, common to the four analyzed WT-and 3Tg-iAstro lines, respectively (Supplementary Table 1), of which 856 proteins were present in both WT-and 3Tg-iAstro lines (Fig. 7a).
To demonstrate that WT-iAstro cells retain the astrocytic phenotype we compared their proteomic profile with two published datasets obtained through matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry. The first one, provided by Hanrieder et al. 43 , featured 130 unambiguous identified proteins from cultured rat cortical neuroglia. Our dataset included 84 out of 130 (64.6%) of their entries (see Supplementary  Table 2A). Among them, cytochrome C oxidase (COX5A), cytochrome C (CYC), ubiquitin (RS27A), chaperonin 10 (CH10), macrophage inhibitor factor (MIF), acetyl co-A binding protein (ACBP), thioredoxin (THIO), calmodulin (CALM), thymosin beta-10 (TYB4) and ribosomal protein S28 (RS28), reported by Hanrieder et al. 43 as the most astrocyte-specific proteins compared to both oligodendrocytes and microglia, were notably all present in our list. Secondly, we compared iAstro proteins with the proteomic profile provided by Yang et al. 44 , consisting of 178 different proteins from mouse cortical cultured astrocytes. Our list covered 131 out of 178 proteins (73.6%) from the dataset provided by Yang et al. 44 (see Fig. 7a and Supplementary Table 2B). Note a relatively small number of detected proteins in Hanrieder et al. 43 and Yang et al. 44 datasets compare to our list of proteins. This may be due to differences in the workflow of the proteome analysis.  . 6 The differences between primary cultured astrocytes from WT and 3xTg mice are retained in the iAstro cell lines. a RT-PCR performed on samples prepared from WT-iAstro#2, #3, #5, #6 and 3Tg-iAstro#2, #3, #4, #6. Data are expressed as mean ± SD ΔCt of gene/S18 of runs performed in triplicate. Unpaired two-tailed Student's t-test was used for statistical analysis. b Correlation plot of differential gene expression between primary astrocytes and iAstro lines. Abscissa axis shows log2 (fold change) of 3xTg-AD primary cultured hippocampal astrocytes vs WT astrocytes. Ordinate axis shows log2(fold change) of 3Tg-iAstro vs WT-iAstro lines Next, we performed differential gene expression analysis and identified 73 proteins (p < 0.05; cut-off 30% fold change) that were significantly changed between WT-iAstro and 3Tg-iAstro, of which 23 were up-regulated and 50 were down-regulated (Fig. 7b, Table 1 and Supplementary Table 3). Functional classification Gene Ontology (GO) analysis of all 73 differentially expressed proteins using DAVID (Database for Annotation, Visualization and Integrated Discovery) online GO tool returned two groups with significant enrichment score: the first group (enrichment score 12.8) contained 11 proteins all of which were components of ribosome; the In total, 1119 and 1045 proteins were detected in WTand 3Tg-iAstro lines, respectively, among which 856 were expressed in both types of samples. The list of 1119 proteins common for all four WT-iAstro lines was compared for presence of common entries with two astroglial proteomics datasets contributed by Hanreider et al. 43 and Yang et al. 44 . b Top 20 of up-and down-regulated proteins emerged from mass spectrometry proteomics second group (enrichment score 8.9) was composed of 5 proteins related to RNA binding and to the formation of ribonucleoprotein complex (Supplementary Table 4A). To explore functional significance of up-regulated vs down-regulated proteins, we analyzed separately 23 upregulated and 50 down-regulated proteins. Analysis of up-regulated proteins did not return significantly overrepresented GO terms, while functional annotation GO analysis of the 50 down-regulated proteins returned 10 significantly overrepresented GO terms which were related to RNA binding, ribonucleoprotein complex, ribosome and nucleus (Table 2 and Supplementary  Table 4B).
We then used STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) database that allows prediction of protein-protein interacting networks and clustering. For this we used a list of 73 differentially expressed proteins which included both down-and up-regulated hits. As shown in Fig. 8, STRING software found significantly more interactions than may be expected by . To search for possible interacting clusters we used STRING k-means clustering function followed by GO analysis. Figure 8 shows three clusters found by STRING. GO analysis revealed that cluster #1 (red in Fig. 8) returned two GO terms, myelin sheath and extracellular exosome. Cluster #2 (green in Fig. 8) returned 13 GO terms related to translation, ribosome and RNA binding, while cluster #3 (blue in Fig. 8) returned 15 GO terms related to nuclear ribonucleoprotein complex and splicing (Supplementary Table 5).
Altogether, this analysis suggests that in 3Tg-iAstro lines, the protein synthesis machinery may be impaired.

Discussion
Here we report the generation and characterization of novel immortalized astroglial lines from the hippocampus of a common AD mouse model, 3xTg-AD mice 45 and from its WT counterpart. For immortalization, we used transduction with SV40 large T antigen, a protocol that has been extensively used previously 14,46,47 . During exploitation of the protocol we did not proceed with clonal selection, but instead continued growing and expanding a population of transduced cells. This allowed us to avoid the inter-clonal heterogeneity that is a characteristic of clonal selection [48][49][50] . Validity of this approach was efficiently demonstrated in gene expression and proteomics analyses, in which four independently generated iAstro lines for both WT and 3Tg-AD genotypes were analyzed and gave consistent results, and also demonstrated that 3Tg-iAstro retains the differences as compared with WT-iAstro found previously in primary cultures 40 .
To validate our model we used a number of methods including immunocytochemistry, real-time PCR (RT-PCR), electrophysiology and Ca 2+ measurements. We show that the WT-iAstro cell lines express the astrocytic markers AQP4, GS and Aldh1l1, although at lower levels compared to primary cells and have electrophysiological properties comparable to primary astrocytes, at least as determined by evaluation of the Kir current. Similarly, real-time PCR of selected genes as well as mass spectrometry for proteins shows that the expression profile is similar to primary astrocytes. For a quantitative estimation of the astrocytic phenotype of iAstro lines, we have compared our proteomics data with similar data obtained on primary astroglial cultures. For this we used two lists reporting data from primary cultured rat 43 and mouse 44 cortical astrocytes. When our list was compared with the above-mentioned datasets, relatively high percentage of common proteins was found (65% of those reported by Hanrieder et al. 43 and 74% of those reported by Yang et al. 44 ), indicating that our list efficiently covers both analyzed datasets. This further confirms that after immortalization procedure, iAstro retains an astroglial phenotype. The only difference of note observed was a small proportion of GFAP-positive cells.
Having established that iAstro cell lines replicate primary culture, we next evaluated whether the changes observed in AD models could be replicated by the 3xTg-iAstro cells. For this, we capitalized on our previous data on Ca 2+ signaling and on transcriptional profiling. Deregulations of astroglial Ca 2+ signaling in AD have been reported 30,[51][52][53] . Our group has demonstrated that primary hippocampal astrocytes from 3xTg-AD mice exhibit enhanced ATP-induced Ca 2+ signals 34 and SOCE 39 . In line with results on primary cultures, 3Tg-iAstro exhibited significantly higher amplitude of Ca 2+ signals in response to ATP stimulation which is accompanied by enhanced SOCE. Analyzing components of the purinergic Ca 2+ signaling we noticed that P2ry2 ATPsensitive receptor is significantly up-regulated in both primary 3Tg astrocytes 40 and in 3Tg-iAstro lines, corroborating the Ca 2+ imaging data presented in Fig. 4. Notably, none of iAstro lines responded to DHPG or glutamate (not shown), suggesting that function or/and expression of mGluR5 may be hampered by the immortalization process.
We used the iAstro model for proteomics to demonstrate its usefulness and to add information on the role of astrocytes in AD. GO analysis of proteins down-regulated in 3Tg-iAstro as compared to WT-iAstro lines suggests that translation may be impaired in 3Tg-iAstro lines in two manners: (1) formation of nuclear ribonucleoprotein complex; and (2) formation of ribosomal multi-protein complex. Protein synthesis has already been suggested to be impaired in AD [54][55][56] and recent findings suggest that it may occur early in AD pathogenesis 57 . In AD hippocampi, down-regulation of several proteins involved in chromatin compacting and regulation of rRNA transcription has been reported 58 including nucleolin, which is also present in our list. Ribosomes are composed of the ribosomal RNAs and the ribosomal proteins that form the small subunit (40S) which binds to mRNA and the large subunit (60S) which binds to transfer RNAs and amino acids. Small subunit contains 33 proteins of which 3 (9%) are present in our list of down-regulated proteins, while large subunit contains 46 proteins of which 8 (17.3%) are present in our list, altogether suggesting that alterations in translation may represent an early astrocyte-specific event in AD pathogenesis 57 . Finally, we acknowledge that the straightforward translation of the results obtained on Fig. 8 Analysis of possible protein-protein interaction network using STRING tool. List of differentially expressed proteins was subjected to STRING analysis which found significantly more associations between proteins that would have occurred by chance (p = 1.0e−16). Three clusters detected by k-means algorithm are colored as follows: cluster 1 (46 proteins, red), proteins related to myelin sheet and extracellular exosomes; cluster 2 (19 proteins, green), proteins related to translation and ribosome; cluster 3 (8 proteins, blue), proteins related to nuclear ribonucleoprotein complex and splicing. For related lists of proteins see also Supplementary Table 5 immortalized astrocytes in in vitro experiments to human AD pathogenesis is a rather speculative oversimplification. Further experiments, aiming at investigating the mechanistic aspects and confirmation of these results in vivo, are necessary to validate the presented data.
Transcriptome analysis of astrocytes freshly isolated from the brain of APPswe/PS1d9ex AD model mice at a symptomatic stage 59 , as well as of hippocampal cultured astrocytes from 3xTg-AD mice 40 , shows significant enrichment of genes in GO terms related to Ca 2+ . Proteomics data, however, do not reveal an overrepresentation of GO terms related to Ca 2+ . This may, at least in part, be explained by the fact that the shotgun mass spectrometry, used in this work, allows detection of only the most abundant proteins in iAstro lines. None of the proteins of the classical Ca 2+ signaling toolkit, like Ca 2+ transporters, channels or receptors, including P2Y2 purinergic receptor, has been detected. It is worth noting that the highest up-regulated protein in our analysis is calmodulin (CaM) ( Table 1), whose role is to sense fluctuation of Ca 2+ concentrations 60,61 . Downstream events include either transmission of the information to Ca 2+ -regulated signaling hubs, like Ca 2 + /CaM-activated kinases (CaMKs) of phosphatase calcineurin, or direct modulation of the activity of proteins, e.g., plasma membrane Ca 2+ ATPase. Therefore, it is reasonable to suggest that 3.6-fold CaM overexpression in 3Tg-iAstro as compared to WT-iAstro (Table 1) may alter the entire Ca 2+ -dependent cellular homeostasis. While characterization of CaM-dependent processes in iAstro lines lies beyond the scope of present work, it is worth noting that in AD-related research, CaMactivated enzymes like CaMKII or calcineurin have been recurrently mentioned as central elements of disease pathogenesis and represent possible pharmacological targets 41,62,63 . Last, it has been recently suggested that CaM binds with high affinity to Aβ, thus representing a direct target for toxic Aβ peptide 64 . In light of our observations, our data warrant further detailed examination of the role of CaM in AD.
In conclusion, we have immortalized and characterized hippocampal astrocytes from WT and 3xTg-AD mouse pups. Using complementary methodologies we show that iAstro lines retain astroglial pattern of protein expression, retain fundamental astroglial housekeeping functions and show transcriptional and functional alterations found previously in primary astrocytes from 3xTg-AD mice compared to WT mice. In addition, proteomic analysis suggests that protein synthesis, due to impaired nuclear RNA binding and alterations in ribosome composition, may be specifically impaired in AD astrocytes. Altogether, our results suggest that 3Tg-iAstro may be a useful tool to study astrocyte-related alterations in AD.

Materials and methods
Animals 3xTg-AD mice used in this work were introduced by Frank LaFerla, Salvatore Oddo and colleagues in 2003 45 . These mice were developed on the mixed 129/C57BL6 background bearing knock-in mutation PS1 M146V 65 and in which APP swe and Tau P301L transgenes were introduced. The 3xTg-AD animals show major histological hallmarks of AD represented by senile plaques and neurofibrillary tangles. These mice show progressive learning and memory deficit beginning from 4 months of age 66 . The 3xTg-AD mice and their respective non-transgenic controls (WT) 45 were housed in the animal facility of the Università del Piemonte Orientale, were kept at three to four per cage and had unlimited access to water and food. Animals were managed in accordance with the European directive 2010/63/UE and with Italian law D.l. 26/2014. The procedures were approved by the local animal-health and ethical committee (Università del Piemonte Orientale) and were authorized by the national authority (Istituto Superiore di Sanità; authorization number N. 22/ 2013). All efforts were made to reduce the number of animals by following the 3R (replacement, reduction and refinement) rule.

Primary astroglial cultures preparation and astrocyte purification
For primary astroglial cultures, WT and 3xTg-AD P0-P2 pups were killed by decapitation. Hippocampi were rapidly dissected and minced with a scalpel blade in cold calcium-and magnesium-free Hank's Balance Salt Solution (Sigma-Aldrich). The tissues were then digested with trypsin (Sigma-Aldrich; 0.25%, 37°C) and triturated with 30 strokes of an automatic pipette. Non-dissociated tissue was allowed to sediment for 2 min and cell suspension was centrifuged for 5 min (200 × g), resuspended in complete culture medium (Dulbecco's modified Eagle's medium (DMEM; Sigma-Aldrich, Cat. No. D5671) supplemented with 10% fetal bovine serum (Gibco, Cat. No. 10270), 2 mM L-glutamine (Sigma-Aldrich), and 1% penicillin/streptomycin solution (Sigma-Aldrich) and plated in 60 mm Petri dishes (Falcon) (hippocampi from 3-6 pups per dish). Cells were maintained in a 5% CO 2 37°C incubator. At~90% of confluence, cells were detached and microglial cells were removed by MACS using anti-CD11b-conjugated beads (Miltenyi Biotech, Cat. No. 130-093-634). Purified astrocytes were collected, resuspended in complete culture medium and plated in a 35 mm dish for immortalization.

Production of SV40-containing replication-defective retroviral vectors
Phoenix cells producing a replication-defective retrovirus 67 , grown in 100 mm culture dishes (Falcon, 1.5 × 10 6 cells per dish), were transfected with pBABE-neo encoding neomycin phosphotransferase and SV40 large T antigen (Addgene, plasmid ID 1780 68 ), using Lipofectamine 2000 reagent (Life Technologies, Segrate, Italy), according to the manufacturer's instructions. At 48 h after transfection, cell medium containing the retroviral vectors was collected and filtered through 0.4 μm filters. The retroviral particles were precipitated by polyethylene glycol (O/N, 4°C) prepared as described elsewhere 69 . The precipitate was concentrated by centrifugation (3500 × g, 30 min, 4°C), the supernatant was discarded and the pellet was resuspended in complete culture medium (200 μl per 8 ml of initial medium), divided in 100 μl aliquots and stored at −80°C.

Astrocyte immortalization
MACS-purified hippocampal astrocytes were plated in 35 mm culture dishes (2.5 × 10 5 cells per dish) and 24 h later were transduced with retrovirus expressing SV40 large T antigen. To increase the infection efficiency, dishes were shaken (80 rpm) overnight in a cell incubator. After 24 h, fresh medium was added for 72 h and then replaced with medium containing 0.4 mg/ml G418 disulfate salt solution (Sigma-Aldrich) 70 . At 3 to 4 weeks after cell selection, surviving cells were first sub-cultured in 100 mm dishes, expanded and maintained in complete culture medium supplemented with 0.4 mg/ml G418 until passage 10. After thawing from cryopreservation, iAstro lines were maintained in complete culture medium without G418 and used for experiments between passages 12 and 20.

Total RNA extraction and real-time PCR
Total RNA was extracted from 1 × 10 6 cells using Absolutely RNA miRNA kit (Agilent, Santa Clara, CA) according to the manufacturer's instructions. Total RNA (0.5-1 μg) was retro-transcribed using random hexamers and ImProm-II RT system (Promega, Milan, Italy). Realtime PCR was performed using iTaq qPCR master mix according to the manufacturer's instructions (Bio-Rad, Segrate, Italy) on a SFX96 Real-time system (Bio-Rad). To normalize raw real-time PCR data, S18 ribosomal subunit was used. Sequences of oligonucleotide primers are provided in Supplementary Materials. The real-time PCR data are expressed as delta-C(t) of gene of interest to S18 allowing appreciation of the relative expression level of a single gene.

Electrophysiological recordings
To perform patch-clamp experiments in whole-cell configuration, both WT-iAstro and 3Tg-iAstro were plated separately in 35 mm dishes 24 h prior experiments. Cells were plated at low density to allow recordings from isolated astrocytes. They were transferred from culture medium to an extracellular solution containing (in mM): 138 NaCl, 4 KCl, 2 CaCl 2 , 1 MgCl 2 , 10 glucose and 10 HEPES at pH 7.25 adjusted with NaOH. Borosilicate patch pipettes were pulled with a P-1000 puller (Sutter Instruments, USA) and were filled with a solution containing (in mM): 140 KCl, 2 NaCl, 5 EGTA, 0.5 CaCl 2 and 10 HEPES at pH 7.25 adjusted with KOH. Pipette tip resistance containing this solution was between 3 and 5 MΩ. Experiments were performed using an EPC7 Plus amplifier (HEKA Elektronik, Germany) in voltage-clamp configuration (holding potential, -80 mV). Access resistance (8-12 MΩ) was compensated (80-90%) and experiments were performed at RT and in a static bath. Data were acquired at 5 kHz and filtered at 1 kHz using a 7-pole Bessel filter and digitized with a low noise data acquisition system, Digidata 1440A (Molecular Devices, USA). Data were recorded and analyzed in pClamp 10 (Molecular Devices, Crisel Instruments, Italy). Data were initially processed with Microsoft Excel. Plots, bar diagrams and figure preparations were finalized with GraphPad Prism (GraphPad Software, La Jolla, CA).

Western blotting
Specific astroglial proteins were detected in WT-and 3Tg-iAstro lysates by western blotting (WB). Each sample, containing 100 µg of protein, was diluted 1:1 in 2× Laemmli sample buffer, heated at 95°C for 5 min, then resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Proteins were electrophoretically transferred to nitrocellulose membranes and the membranes were blocked for 1 h in 5% (w/v) non-fat dry milk in Tris-buffered saline containing Tween-20. After incubation with appropriate primary and secondary antibodies, signals were revealed using enhanced chemiluminescence ( Densitometric analysis was performed using Quantity One software and is expressed as mean ± SEM from at least 3 independent runs. Analysis of variance (ANOVA) test was used for statistical analysis. Note relatively big variability between technical replicates in WB analysis, which may be explained by low level of specific astroglial proteins expressed in iAstro lines.

Fura-2 calcium imaging
For Ca 2+ imaging, iAstro lines grown onto 24 mm round coverslips were loaded with Fura-2/AM (Life Technologies, Milan, Italy, Cat. No. F1201) in the presence of 0.005% Pluronic F-127 (Life Technologies, Cat. No. P6867) and 10 μM sulfinpyrazone (Sigma, Cat. No. S9509) in KRB solution (125 mM NaCl, 5 mM KCl, 1 mM Na 3 PO 4 , 1 mM MgSO 4 , 5.5 mM glucose, 20 mM HEPES, pH 7.4) supplemented with 2 mM CaCl 2 . After loading and 30 min of de-esterification, the coverslips were mounted in an acquisition chamber on the stage of a Leica epifluorescence microscope equipped with a S Fluor 40×/ 1.3 objective. Cells were alternatively excited at 340/380 nm by the monochromator Polichrome V (Till Photonics, Munich, Germany) and the fluorescent signal was collected by a CCD camera (Hamamatsu, Japan) through bandpass 510 nm filter; the experiments were controlled and images analyzed with MetaFluor (Molecular Devices, Sunnyvale, CA, USA) software. The cells were stimulated by 20 μM ATP. To quantify the difference in the amplitude of Ca 2+ transients, the ratio values were normalized according to the formula (ΔF)/F 0 (referred to as norm. Fura ratio).

Proteomic analysis In solution digestion
Cell lysates were digested using the following protocol: samples were prepared to have 100 μg of protein in a final volume of 25 μl of 100 mM NH 4 HCO 3 . Proteins were reduced using 2.5 μl of dithiothreitol (200 mM DTT stock solution) (Sigma) at 90°C for 20 min, and alkylated with 10 μl of Cysteine Blocking Reagent (iodoacetamide (IAM), 200 mM Sigma) for 1 h at room temperature in the dark. DTT stock solution was then added to destroy the excess of IAM. After dilution with 300 μl of water and 100 μl of NH 4 HCO 3 to raise pH 7.5-8.0, 5 μg of trypsin (Promega, Sequence Grade) was added and digestion was performed overnight at 37°C. Trypsin activity was stopped by adding 2 μl of neat formic acid and samples were dried by Speed Vacuum 71 .
The peptide digests were desalted on the Discovery® DSC-18 solid-phase extraction (SPE) 96-well Plate (25 mg/well) (Sigma-Aldrich Inc., St. Louis, MO, USA). The SPE plate was preconditioned with 1 ml of acetonitrile and 2 ml of water. After the sample loading, the SPE was washed with 1 ml of water. The adsorbed proteins were eluted with 800 μl of acetonitrile/water (80:20). After desalting, samples were vacuum evaporated and reconstituted with 20 μl of 0.05% formic acid in water. Then, 2 μl of stable-isotope-labeled peptide standard (DPEVRPTSAVAA, Val-13C5 15N1 at V10, Cellmano Biotech Limited, Anhui, China) was spiked into the samples before the liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis and used for instrument quality control.

Label-free proteomic analysis
LC-MS/MS analyses were performed using a micro-LC Eksigent Technologies (Dublin, USA) system with a stationary phase of a Halo Fused C18 column (0.5 × 100 mm, 2.7 μm; Eksigent Technologies, Dublin, USA). The injection volume was 4.0 μl and the oven temperature was set at 40°C. The mobile phase was a mixture of 0.1% (v/v) formic acid in water (A) and 0.1% (v/v) formic acid in acetonitrile (B), eluting at a flow rate of 15.0 μl/min at an increasing concentration of solvent B from 2 to 40% in 30 min. The LC system was interfaced with a 5600+Triple-TOF system (AB Sciex, Concord, Canada) equipped with a DuoSpray Ion Source and CDS (Calibrant Delivery System). Samples used to generate the SWATH-MS (sequential window acquisition of all theoretical mass spectra) spectral library were subjected to the traditional data-dependent acquisition (DDA): the mass spectrometer analysis was performed using a mass range of 100-1500 Da (TOF scan with an accumulation time of 0.25 s), followed by a MS/MS product ion scan from 200 to 1250 Da (accumulation time of 5.0 ms) with the abundance threshold set at 30 cps (35 candidate ions can be monitored during every cycle). Samples were then subjected to cyclic data-independent analysis (DIA) of the mass spectra, using a 25 Da window. A 50 ms survey scan (TOF-MS) was performed, followed by MS/MS experiments on all precursors. These MS/MS experiments were performed in a cyclic manner using an accumulation time of 40 ms per 25 Da swath (36 swaths in total) for a total cycle time of 1.5408 s. The ions were fragmented for each MS/MS experiment in the collision cell using the rolling collision energy. The MS data were acquired with Analyst TF 1.7 (SCIEX, Concord, Canada). Three instrumental replicates for each sample were subjected to the DIA analysis 72,73 .

Protein database search
The mass spectrometry files were searched using Protein Pilot (AB SCIEX, Concord, Canada) and Mascot (Matrix Science Inc., Boston, USA). Samples were input in the Protein Pilot software v. 4.2 (AB SCIEX, Concord, Canada), which employs the Paragon algorithm, with the following parameters: cysteine alkylation, digestion by trypsin, no special factors and false discovery rate (FDR) at 1%. The UniProt Swiss-Prot reviewed database containing mouse proteins (version 20july15, containing 23,304 sequence entries). The Mascot search was performed on Mascot v. 2.4, the digestion enzyme selected was trypsin, with 2 missed cleavages and a search tolerance of 50 ppm was specified for the peptide mass tolerance, and 0.1 Da for the MS/MS tolerance. The charges of the peptides to search for were set to 2+, 3+ and 4+, and the search was set on monoisotopic mass. The instrument was set to ESI-QUAD-TOF (electrospray ionization quadrupole time-of-flight) and the following modifications were specified for the search: carbamidomethyl cysteines as fixed modification and oxidized methionine as variable modification.

Protein quantification
The quantification was performed by integrating the extracted ion chromatogram of all the unique ions for a given peptide. The quantification was carried out with PeakView 2.0 and MarkerView 1.2. (Sciex, Concord, ON, Canada). Six peptides per protein and six transitions per peptide were extracted from the SWATH files. Shared peptides were excluded as well as peptides with modifications. Peptides with FDR lower than 1.0% were exported in MarkerView for the t-test.

Analysis of identified proteins
For identification of the number of detected proteins for each genotype, the intersection of four lists, corresponding to four analyzed lines per genotype, was performed, yielding lists of common proteins detected in four WTand four 3Tg-iAstro lines (1119 and 1045, respectively).
The intersection of the latter two lists gave proteins detected in both WT-and 3Tg-iAstro cells (856 proteins).

Gene ontology analysis
GO analysis was performed using DAVID v.6.8 tool (https://david.ncifcrf.gov/) 74 . For the analysis of the overrepresented GO terms, for the background, a list containing all proteins detected in WT-iAstro and 3Tg-iAstro lines (1308 proteins) was used. Over-represented GO terms which passed Benjamini correction (p < 0.05) were considered significant. For prediction of protein-protein interactions and clustering using kmeans algorithm, STRING v.10.5 online software was used (https://string-db.org/) 75 .

LPS and TNFα treatment
For treatment with bacterial LPS (lipopolysaccharides from Escherichia coli O111:B4, Sigma, Cat. No. L2630) and TNFα (Peprotech, London, UK, Cat. No. 300-01A), cells were plated in 12-well plates (1 × 10 5 cells per well) and, upon confluence, were treated with either LPS (for 3 h) or TNFα (for 6 h). The cells were then lysed in 500 μl Trizol Reagent (Life Technologies) and total RNA was extracted according to the manufacturer's instructions. First-strand complementary DNA and real-time PCR were performed as described above. Oligonucleotide primers for iNOS are listed in Supplementary Materials.

Experimental groups selection and statistical analysis
For the selection of experimental groups, the following criteria were adopted: in experiments in which entire populations of cells were analyzed (western blotting, realtime PCR and proteomic analysis), experimental group was composed of 4 independently generated iAstro lines for each genotype (WT-iAstro#2, #3, #5 and #6, and 3Tg-iAstro#2, #3, #4 and #6) representing biological replicates. Technical replicates consisted of at least three independent experiments of each iAstro line. In experiments, in which single-cell analysis was performed (immunocytochemistry, electrophysiology and Fura-2 Ca 2 + imaging), experimental groups were composed of independent experiments (at least three coverslip (technical replicates) from at least three different experiments) of one line per genotype were used (WT-iAstro#2 and 3Tg-iAstro#2).
Statistical analysis was performed using GraphPad Prism software v.7. For analysis of real-time PCR and Ca 2+ imaging data, a two-tailed unpaired Students's t-test was used. For western blot ANOVA on raw followed by Tukey's post-hoc test was used. Differences were considered significant at p < 0.05.