Characterization of Mouse Cell Line Ima 2.1 as a Potential Model System to Study Astrocyte Functions

ASlrocytes are activated in most chronic neurodegenerative diseases associated with inj/ammatory evenls such as Parkinsoll 's disease or Alzheimer's disease, but also ill stroke. Due to all aging populatioll worldwide, research efforls ill these areas are likely 10 expalld ill thefuture. This will entail an increased demand for appropriate experimental models. We introduce here the new immortalized mouse astrocyte cellline IMA 2.1 as an alternative to currently used primary astrocyte cultures .IMA 2./ were directly compared with primary mouse astrocytes with respect to fheir response to proillj/ammatory stimuli, expression of typical astrocyte markers, and 10 the ceillille's capacity to metabolize the parkinsonian toxin MPTP to its toxic metabolite MPP+. Under injlammatory conditions, mimicked with the addition ofa cytoleine mix, IMA 2.1 responded similarly to primary astrocytes with mRNA upregulation, expressioll of iNOS and COX-2, alld the release ofvarious injlammatory mediators. Analysis o.f astrocytic markers indicated that IMA 2.1 represent a relatively early, GFAP-negative stage of astrocyte development. Moreover, cOllversion of MPTP by monoamine oxidase-B proceeded in IMA at least as quickly as in primary cells. For all endpoints investigated, the cellline IMA 2./, derivedfrom a single clone, delivered reproducible results over aperiod of several years and allowed upscaling of experiments due 10 its easy handling compared with primary cells. 1 Intraduction Astrocytes are highly differentiated cells of the central nervous system that serve numerous functions , such as nu trient supply of neurons , regulation of cerebral blood flow , orchestration of neuronal growth and differentiation , maintenance of extracel-lular glutamate levels , and ion and liquid balance (Kettenmann and Ransom , 2005). They are furthermore involved in infl am-matory processes in the central nervous system that are observed not only following the invasion of pathogens but also in association with chronic neurodegenerative diseases such as Alzheimer's disease (AD) or Parkinson 's disease (PD) (Falsig et al., 2008). The primary effector cells of the brain in response to bacteria are microglia (Lund et al., 2005, 2006). They release proinflammatory cytokines that trigger the subsequent activation of astrocytes as a second line of defense (Lee et al., 1993; Henn et al., 2011; Falsig et al., 2006). Following stimulation, both cell types become rich sources of cytokines and pros ta-noids (Murphy et al., 1988; Meeuwsen et al., 2003). Moreover, they produce the free radicals nitric oxide (ONO) and superoxide (°02-) , which can lead finally to the formation of seve ral cyto-toxic …


Intraduction
Astrocytes are highly differentiated cells of the central nervous system that serve numerous functions , such as nu trient supply of neurons , regulation of cerebral blood flow , orchestration of neuronal growth and differentiation , maintenance of extracellular glutamate levels , and ion and liquid balance (Kettenmann and Ransom , 2005).They are furthermore involved in infl ammatory processes in the central nervous system that are observed not only following the invasion of pathogens but also in association with chronic neurodegenerative diseases such as Alzheimer's disease (AD) or Parkinson 's disease (PD) (Falsig et al., 2008).The primary effector cells of the brain in response to bacteria are microglia (Lund et al., 2005(Lund et al., , 2006)).They release proinflammatory cytokines that trigger the subsequent activation of astrocytes as a second line of defense (Lee et al., 1993;Henn et al., 2011;Falsig et al., 2006).Following stimulation, both cell types become rich sources of cytokines and pros tanoids (Murphy et al., 1988;Meeuwsen et al., 2003).Moreover, they produce the free radicals nitric oxide (ONO) and superoxide (°02-) , which can lead finally to the formation of seve ral cytotoxic species such as peroxynitrite (ONOO') or hydroxyl radicals (OOH) (Brown and Neher, 2010;Beckman and Koppenol, 1996).These molecules can attack and modify cellular proteins, lipids, and DNA in neighboring neurons, particularly in association with chronic brain diseases, as illustrated in numerous reports (Schildknecht et al., 2011).
In a widely used in vivo model of PD, astrocytes are responsible for the conversion of the parkinsonian toxin l-methyl-4phenyl-I,2,3,6-tetrahydropyridine (MPTP) into its active metabolite l -methyl-4-phenylpyridinium (MPP+) (Ransom et al., 1987;Di Monte et al., 1991) .The responsible enzyme has been identified as monoamine oxidase-B (MAO-B) (Langston et al., 1984) .After release from astrocytes, MPP+ is selectively taken up by dopaminergic neurons leading to their death (Pöltl et al., 2011;Schildknecht et al., 2009).The resultant symptoms and pathology recapitulate key features observed in PD patients.The active toxin MPP+ is widely used in cellular cultures of neuronal cells .However, the presence of glial cells as such can have a profound impact on the degeneration process of neurons exposed to MPP+ or other toxins.To model the in viva situation more faithfully, co-culture models composed of astrocytes and neurons are desirable.They would allow studies of the process of MPTP conversion by astrocytes, as weil as the subsequent uptake and degeneration of neurons in a defined and easil y accessible in vitra model.Primary astrocytes are used in most currently established co-culture systems (Roque et al., 2011) .Availability and use of a defined and well -characterized astrocyte cell line that can be applied in such experirnental models would avoid the disadvantages of primary cultures.
Primary rodent astrocytes have been used for many in vitra disease models and to obtain biochemical insight into astrocyte function (Kim and Magrane, 2011;Falsig et al., 2004b;Christiansen et al., 2011).Although highly sophisticated isolation procedures have been developed, batch-to-batch variations and contamination with microglia represent considerable shortcomings in the use of primary cultures (Hansson, 1986).As primary astrocytes hardly proliferate, the high demand for animals puts an additional burden on this type of cellular model.For example, highly purified primary astrocyte preparations , as used for gene expression profiling (Falsig et al. , 2004a(Falsig et al. , , 2006) ) may yield less than I million cells per mouse and a single study may require the sacrifice of several hundreds of mice.In order to find alternative approaches, IlUmerous attempts have been undertaken to generate astrocyte cell lines by retroviral transduction of immortalizing oncogenes .The first approach dates back to 1990, when progenitor cells isolated from mouse brains were transduced with the myc-oncogene that allowed the generation of celliines with neuronal and glial phenotypes (Ryder et al., 1990).In a more sophisticated approach, mouse cortical astrocytes of varying differentiation stages were immortalized by SV40 T antigen, and this allowed the generation of mature and immature mouse astrocyte lines (Frisa et al., 1994).In an attempt to generate an immortalized human astrocyte cell line, astrocytes obtained from human fetal brain have been purified and transfected with the SV40 T antigen.However, it was noted that the high proliferation capacity of this newly generated A735 line was associated with downregulation of GFAP expression and subsequent loss of other astrocyte specific markers (Price et al., 1999) and thus that cell line does not represent an adequate substitute for primary cells .In a new attempt to provide an alternative to primary astrocyte cultures, we irnmortalized cortical astrocytes of wildtype BALB/c mice with the SV40 large T antigen, created single cell clones, and selected the clone IMA 2.1 based Oll its ability to respond to proinflammatory cytokines and its expression of functional monoamine oxidase-B (MAO-B).
During more than three years of use ofIMA 2.1 in our laboratory, its phenotype and response pattern towards cytokines remained stable .This makes the IMA 2.1 a reliable new model for certain astrocyte studies (Leist et al., 2010).In the present manuscript, we provide data on the characterization of this new cell line and suggest the IMA 2.1 cell line as a useful tool for a variety of in vitra studies of astrocyte functions including astrocyte-neuron crosstalk.It is intended as an overview for interes ted researchers to enable a firstjudgment on whether IMA 2 .1 may be useful for their planned investigations.262 2 Animals, materials, and methods 2.1 Immortalization of primary astrocytes from BALB/c mice Cerebral cortices were isolated from one-or two-day-old mice.After removal of the meninges, cortices were chopped into small pieces with a razor blade and further homogenized by enzymatic digestion in PBS/0.5% trypsin/EDTA (Gibco/Invitrogen, Karlsruhe, Germany), 0.5% DNase I (Sigma-Aldrich, Steinheim, Germany) for 15 min at 37°C.Trypsin was inactivated by addition of fetal 'calf serum (FCS; PAN Biotech, Aidenhach, Germany) and tissue was triturated using a Pasteur pipette.The homogenates were seeded on poly-D-Iysine hydrobromide (PDL; 10 jlg/ml; Sigma-Aldrich, MW 30-70 kDA) coated culture f1asks at a density of approximately 40 ,000 cells/cm 2 in DMEM (Gibco/Invitrogen) supplemented with 10% FCS, 2 mM L-Glutamin (Invitrogen) and 1% penicillin/ streptomycin (Invitrogen).After 8 days in culture, an astrocyte layer, overgrown by numerous microglial cells, oligodendrocyte precursor cells, and oligodendrocytes can be observed.To prepare purified astrocyte cultures, oli godendrocytes and surface microglia were f1ushed away from the astrocyte layer.The remaining microglia were removed by trypsinization of the primary cultures and differential adhesion for 30 min on uncoated culture dishes.The astrocytes remaining in the supernatant were collected, and approximately 0.5 x 10 6 cells were plated in PDL-coated 6-well plates.Cells were grown to confluency in culture medium without antibiotics.On the day of transfection, medium was changed to Opti-MEM I (Gibco).Lipofectamine 2000 (Invitrogen) was diluted 1 :50 in Opti-MEM land incubated for 5 min at room temperature.The plasmid psV3neo (Chang et al., 1986), coding for the large T antigen from SV40 (500 ng) was diluted in Opti-MEM I and mixed in a ratio of I: 1 with the lipofectamine dilution.The resulting transfection complexes were preincubated at room temperature for an additional 20 min, the transfection mixture (500 jll/well) was added carefully to the cells and then incubated at 37°C.After 6 h, the transfection mixture was replaced by standard cell culture medium and the cells were incubated for an additional 24 h.Stably transfected cells were selected with 800 jlg/ml geneticin (G418; Invitrogen).The resu ltant immortalized mouse astrocytes (lMA) were cultured at 37°C (5% C02) in DMEM Glutamax (high glucose) (GIB CO) with 5% FCS (fetal calf serum, from PAA) without antibiotics .They were trypsinized for 2 min with 0.5% trypsinlDMEM every 2-3 days and reseeded at a ratio of 1:5 or I: 10.Experiments were performed in DMEM containing 2% FCS.

Primary astrocyte cultures
Primary cortical astrocytes were generated from BALB/c mice bred at the animal facility of the University of Konstanz, Germany.All mice were housed at 22°C and 55% relative humidity in a 12 h day/night rhythm with free access to food and water according to national regulations and EU Directive 2010/63/EU.Primary cortical astrocytes were prepared from mouse pups at 24-48 h past partum , as described earlier (Henn et al. , 2011) .In brief, brains were removed, the cortices were dissected out, and hippocampi and meninges were carefully removed before digestion with trypsin and DNase .After trituration, the cell suspension was carefully layered over a 30% Percoll solution and centrifuged at 150 x g for 10 min .The astrocyte fraction was recovered, washed, and resuspended in DMEM (high glucose) , 20% FCS, 100 U/ml penicillin and 100 Jlg/ml streptomycin .Cells were seeded at a density of2 million cells (corresponding t04 brains) perTI75 flask , and cultured at 37°C in a 5% C02 atmosphere .The medium was changed after 4 days, and subsequently every second day.After 14 days in primary culture, cells were trypsinized and incubated in DMEM for 45 min at room temperature in a T75 flask for differential adhesion of any residual microglia.Non-adherentcells (astrocytes) were reseeded in DMEM plus 10% FCS, containing 4.5 g/I D-glucose and 2 mM Glutamax, and cultured in multiweIl dishes for 7-9 days before use in experiments.This protocol was shown to result in microglia or oligodendrocyte contaminations below our detection limit « 1%) by FACS, PCR, or immunostaining .

Co-culture model
Primary astrocytes or IMA were seeded into 24-well plates (Nunclon, Nunc, Roskilde , Denmark), and when confluency was reached, the cells were maintained for 7 additional days .Then, predifferentiated (2-days) human mesencephalic cells (LUHMES) were seeded on top of the glial cell layer at a density of 180,OOO/cm 2 in Advanced DMEM (Gibco), containing 2 mM L-glutamine, I x N2 supplement, and I ]lg/ml tetracycline for 4 more days.LUHMES cells were previously characterized with respect to their applicability as an in vitra model of terminally differentiated dopaminergic neurons (Schildknecht et al., 2009;Scholz et al., 2011;Lotharius et al., 2005) .To selectively visualize the neurons in these co-cultures, cells were fixed with 4% paraformaldehyde (PFA) for 15 min and permeabilized with 0.2% Triton X-100 in PBS for 10 min .Following blocking with 1 % bovine serum albumin (BSA) in PBS for 1 h, LUHMES were stained with an antibody against the neuronal class of ß-I1I-tubulin (TUJ-I; Convance , mouse , 1: 1000) and fluorescence-Iabeled secondary antibodies.

Translocation of NF-KB
Cells were plated in 96-well plates at 15 ,000 cells/well and cultured in DMEM plus 2% FCS .The cells were stimulated as indicated , then fixed with 4 % PFA for 15 min, permeabilized with 0.1% Triton X-100 for 10 min, blocked with 10% FCS for I h, immunostained with purified monoclonal mouse anti-NF-KB p65 (clone 20/NF-KB/p65 ; final dilution 1:300) antibody (BD Transduction Laboratories) over night, and detected with anti-mouse IgG Alexa-488 secondary antibody.Images were recorded with a high-resolution CCD camera on a Cellomics ArrayScan"t , wh ich is based on an automated Zeiss Axiovert-IOO microscope equipped with a 20x objective and a FITC/H-33342 filter set (Acx-l = 365 nm , Acx-2 = 475 nm , Acm = 535 nm) .Nuclei, stained with H-33342, were imaged first (channel I) for automated focusing and identification of valid objects.Subsequently, the stained antigens were imaged in the corresponding field s (chan ne I 2) .Based on these data sets, the nuclear translocation of NF-KB was quantified for each cell with the predefined algorithm " nuclear translocation ," essentially as described earlier (Henn et al., 2009(Henn et al., , 2011) ) .

Flow cytometry
Cells from 6-well plates were trypsinized and washed twice with PBS , containing 10% FCS .They were centrifuged for 5 min at 500 x g.FITC-conjugated anti-TLR2 antibody (clone 6C2; eBioscience; Alasdair Stewart, U.K) diluted in 2% FCS/PBS to a concentration of 1 Jlg per million cells (in 200 ]11) was added for 50 min at 4°C.An appropriate isotype control was used in a similar way.Cells were washed twice and fixed with 0 .5% PFA .For each sampie, a minimum of 10,000 cells within the gated cell population was analyzed using a flow cytometer (Accuri Cytometers, Cambs, U.K.) and the corresponding software CFlow.The percentage of positively stained cells relative to isotype control (maximum 1 % positive staining) was determined.

Measureme' nts of prostanoids and nitrite in cell culture supernatants
ProstagIandin E2 (PGE2) was determined by using commercially available EIA-kits (Assay Designs, MI, USA) according to the manufacturer's instructions.Nitrite (N02-), the stable autoxidation product of "NO , was measured by the Griess assay.Briefly, 30 #1 12.5 #M sulfanilamide (Sigma) and 30 #16 M HCI were mixed with 200 #1 cell culture supernatant at room temperature and incubated for 5 min.Absorbance was measured befOTe and after the addition of 25 #1 N-(I-naphthyl)ethylenediamide (12.5 #M) (Sigma) at 560 nm using a microtiter plate reader.Nitrite concentrations were calculated from a NaN02 standard curve in the range ofO.5-1O #M .

HPLC analysis
Detection of MPTP, MPP+, and MPDP+ was performed on a Kontron system (Goebel Analytic , Au/Hallertau, Germany) composed of a model 520 pump, model 560 autosampIer, models 535 and 430 diode array detectors, set at 245 nm for MPTP, 295 nm for MPP+, and 345 nm for MPDP+.Sam pIes were acidified with 9,tl perchloric aeid (70%) per ml volume of sampIe and centrifuged at 10,000 x g for 15 min.The supernatant was filtered through a Chromaphil PET-20/J5MSfilter with 0.2 pm pore size from Macherey Nagel (Düren , Germany) .
Separation was carried out on a CI8 nucleosil column (250 x 4.6 mrn; 5 #m particle size) from Macherey Nagel (Düren , Germany) at room temperature.The mobile phase consisted of acetonitrile : distilled water : triethylamine : sulfuric acid (12.50 : 86.18 : 1.04: 0 .28,v/v, pH 2.3) .The mobile phase was degassed with an online vacuum degasser and delivered isocratically at a ftow rate of I mllmin at an average pressure of 145 bar.Data analysis was performed with Geminyx II software (Goebel Analytic) .

Monoamine oxidase (MAO) activity assay
The assay is based on the oxidation ofkynuramine to 4-hydroxyquinoline by MAO .Cells were homogenized in PBS, and total protein content was adjusted to I mg/mI .Then , 180 #1 of cell homogenate were supplemented with 20 #1 substrate kynuramine (Sigma) in PBS to yield a final concentration of 100 ~ M. In parallel, a standard curve of 4-hydroxyquinoline (0-100 JlM) was prepared .After 2 h of incubation, the reaction was terminated by the addition of 80 #1 perchloric acid (400 mM) .SampIes were centrifuged at 12,000 x g for 1 min, 200 #1 supernatant was transferred into a new tube and mixed with 400 JlI NaOH (I M).SampIes of the standard curve were treated accordingly and fluorescence was detected at A.:x 315 nm / A.:m 380 nm.

Cyclooxygenase (COX) activity
COX activity was determined by the conversion of the 14C_ labeled substrate arachidonic acid ( 14 C_AA) to prostanoids (Schildknecht et al. , 2008) .Cells were washed twice, collected in cold PBS and centrifuged at 1000 x g for 3 min .The pellet was dissolved in lysis buffer (20 mM HEPES, 1 % Triton X-100, 1 % aprotinin, 10% glycerol, pH 7.5) for 30 min on ice.Following centrifugation at 12,000 x g for 1 min, the supernatant was incubated with reaction buffer (80 mM Tris-HCI, 0.1 mM phenol , 5 #g/ml hematin, 5 #M 14C_AA, pH 8 .0)for 20 min .The reaction was terminated by the addition of ethyl acetate/ 2 M citric acid (30 : I) .After vortexing for at least Imin, the organic phase was collected, evaporated and used for analysis on silica TLC plates (Silica 60, Merck, Darmstadt, Germany) by chromatography.The mobile phase consisted of ethyl acetate : 2,2,4-trimethylpentane : acetic acid : water (110:50:20:100) .Plates were dried and used to expose a PhosphorImagerT>t screen overnight.For reading the screen, a PhosphorImagerT>t system from Molecular Dynamics , USA was used.Quantitieation was performed by the detection of total prostanoid formation utilizing ImageQuantT>t software.

13 Immu nocytoche mistry
Cells were grown on 10 mm glass cover slips (Men zel, Braunschweig, Germany) in 24-well plastic cell culture plates (Nun-c1 on T .II ) .Foll owin g treatm ent, cell s were fi xed with 4% PFA fo r 20 min at 37°C and washed with PBS .After blocking with 1% BSA (Calbiochem, San Diego, CA) for I h, primary antibodies (anti-CD44, I :500 , rat monoclonal, Caltag Laboratories; antinestin , 1:500 , rat monoclonal, Millipore) were added in PBS-Tween (0 .1 %) at 4°C over night.Secondary antibodies (anti-rat IgG-Alexa 488, Invitrogen, Darmstadt, Germany) were added for 45 min at RT.For vi sualization, an Olympus IX 81 microscope (Hamburg , Germany) equipped with an F-view CCD camera was used .Nuclei were stained with H-33342 CI jlg/ml) for 15 min pri or to the fin al washin g step.For image process in g, Cell P software (Olympus) was used .

. 14 Statistics
NF-KB translocation data are based on at least 200 valid cells per weil and are indicated as the means ±SEM of at least three wells fro m independent experiments.Flow-cytometry analysis was done in duplicates.Nitrite and PGE2 detection , enzyme acti vity assays and mRNA experiments were all performed in triplicates using independent cell preparations.Statistical analysis was performed with GraphPad Prism softw are (Version 4 .03),using t-test or one-way ANOVA with Bonferro ni 's post-hoc test, as appropriate.

.1 Characte ri zatio n of inflammatory response s
Initi ally, we used a pool of IMA , obtained by selection of the transfected cells by antibi otics , and then expanded the cells by passag ing in normal culture medium .After stable growth conditi ons were reached , such cells were tested for thei r capac ity to reaet to infl ammatory eytokines.Altogether, six eell pools from di ffe rent transfecti ons were used (lMA 1-6) to assess NF-KB translocati on from the cytosol to the nucleus.This signalin g response is usually observed after stimulation of astroeytes with vari ous cytokines as weil as toll -like receptor (TLR) li gands .Here , it was quantitati vely assessed 30 min aft er sti mul ati on with IL-I ß , TNF-a , IFN-y , or a combin ati on of all three cytokin es (eompl ete eytokin e mi x = CCM).T NF-a or IL-I ß stimul ati on alone was suffic ient to evoke max imal NF-KB translocation (Fig .I A) .Tn th ese ex periments, TM A 2 showed the best responses , and they were used thereafter.In order to obtain a homogeneous eell popul ation, IMA 2 were dil uted to th e single cell level and different clo nes ori gin ating fro m thi s procedure were generated and compared .Five TMA 2 sub- COX-2 = cyclooxygenase-2; NOS-2 = nitric oxide synthase-2; TYKi = inducible thymidylate kinase ; TNF-a = tumor necrosis factor 0; IL-1 ß = interleukin-1 ß; IL-6 = interleukin 6; IFN-ß = interferon-ß; IL-23 = interleukin-23; MCP-1 = monocyte chemotactic protein 1; TLR 3 = tOIl-like receptor 3; NOD 2 = nucleotide binding oligomerization domain 2; TLR 2 = toll-like receptor 2. Data are expressed as means ±SD of triplicate sampies.Statistical significance was determined by one-way ANOVA followed by Bonferroni's post-hoc test.•P<O .05 given alone (not shown) .Combination of the three cytokines (stimulation with CCM) resulted in a two orders of magnitude hi gher mRNA expression that exceeded by far the sum ofTNF-a and IL-I ß signal intensities (Fig. I B) .A similar pattern was observed for the induction of NOS-2 and the release of nitrite from primary astrocytes (Falsig et al., 2004a(Falsig et al., , 2006) ) .This indicates that IMA behaved similarly to murine astrocytes with respect to the cytokine response and, in particular, to the sy nergie activity ofTNF-a and TL-I ß with TFN-y .
In a CCM stimulation time course, NOS-2 mRNA displayed a rapid rise after 4 h , followed by a subsequent decline (Fig. 2A).Induction of mRNA coding for toll-like receptor-2 (TLR-2) was qualitatively similar to the si tuation observed with NOS-2.However, when compared quantitatively, the induction was far 266 less pronounced compared with NOS -2 (Fig. 2B).In contrast to TLR-2, TLR-4 expression was not regulated by CCM stimulation (not shown).
For a more comprehensive comparison of the IMA subclone IMA 2.1 with primary mouse astrocytes, both cell populations were activated with CCM for 4 h and mRNA expression levels of a set of astrocyte-relevant marker genes were measured.The results indicated that the IMA 2 .1 response resembles the response observed in primary astrocy tes (Fig .2C) .
Following th e analysis of a set of intl ammati on-related target genes on the mRNA level, we selected NOS-2 and COX-2 for a more detailed analysis on the protein and activity level.IMA 2 .1 were stimulated with CCM for various time intervals.Both NOS-2 and COX-2 protein became detectable after 4 h of CCM  stimulation (Fig. 3A) .Interestingly, NOS-2 mRNA peaked at 4 h, while NOS-2 protein expression displayed a peak after 16 h.Similar observations have been made in other inflammatory cells, such as lipopolysaccharide-stimulated primary rat alveolar macrophages (Schildknecht, unpublished) .
Accumulation of prostagiandin E2 (PGE2) in the supernatant served as an indirect indicator of COX activity.Unlike the gradual and continuing increase of nitrite as a marker of NOS activity, PGE2 demonstrated an increase up to 16 h after CCM stimulation .From then on, no further elevation was observed , despite a pronounced COX-2 protein expression (Fig. 3A,B).To test whether inducible NOS-2 was solely responsible for the nitrite accumulation, selective inhibitors of NOS-2, i.e., AMT and L-NMMA, were applied in different concentrations together with CCM for 16 h.The complete block of nitric oxide synthesis by these compounds suggests that no other NOS-isoform contributed to the observed "NO synthesis (Fig. 3C,D).
For a further investigation of the lack of PGE2 sy nthesis after 16 h of CCM activation, COX activity in homogenates of CCM-treated IMA was directly assessed by addition of the 14C-labeled substrate arachidonic acid, followed by separation and quantification of the prostanoids formed.In close correlation to the Western blot data in Fig. 3A, a constant increase in activity up to 12 h of CCM .stimulation was observed, which was followed by a rapid decline in COX activity to almost baseline values (Fig. 4A) .Similar observations were made in RAW 264.7 cells (Schildknecht et al. , 2006) and primary rat alveolar macrophages (unpublished) , in which a nitration and inactivation of COX-2 was observed.Although not investigated in detail in IMA 2.1 so far, it is Iikely that a similar mechanism is responsible for the observed effects, since we found that inactivation of COX-2 could be prevented with NOS-2 inhibitors (not shown) .
We examined TLR-2 protein expression on the surface of CCM-stimulated TMA 2.1 by FACS analysis (Fig .4B) in additi on to the mRNA profiling (Fi g. 2B).The surface expression continuously increased for up to 12 h.The delay between the time course of mRNA and protei n surface expression likely originated from a time period required for the synthesis and transport of newly formed TLR-2 from the Golgi-to the plasma-membrane.To test the importance of translocation, the secretory vesicle assembly blocker brefeldin A (BFA) was used.Under these conditions , a signifi cant reduction in TLR-2 surface expression was detected.This inhibitory effect also was seen for the baseline TLR-2 surface expression ofunstimulated cells (Fig .4C).Together, these data indicate that TMA showed inflammatory responses similar to those of primary astrocytes or other inflammatory cells.Moreover, the data suggest the importance of activity measllrements and protein quantification in addition to the mRNA profiling.

Expression of immature astroeyte markers
For an initial comparison of the marker expression and morphology of primary mouse astrocy tes and IMA 2.1, cells were stained with antibodies selective for nestin and CD44 (Fig. 5A) .These two prote ins with cytoskeletal (nestin) and membraneous (CD44) localization allow the characterization of different morphological features.According to this, IMA 2.1 took the typical star-shape of astrocytes .Both markers were expressed in primary astrocytes and TMA 2.1, but the astrocytic extensions of TMA 2 .1 were slightly shorter compared with primary astrocytes.Both , nestin and CD44, are not considered typical astrocyte markers (Kuegler et al., 20 10 , 2012), but are rather typical for immature precursors.However, they are often found in primary neonatal astrocyte cultures (which are commonly used for many experiments).The strong expression in TMA suggests their rather early-stage astrocyte precursor phenotype.For the assessment of general cell morphology, IMA 2. 1 and primary astrocytes were stained with f1uorescent ph alloidin that selectively binds to F-actin.Conflllent IMA 2.1 were additionally stained by calcein-AM (1 JlM; 30 min) and the same region was photographed in the phase contrast mode (Fig .SB).In order to test the functionality of the mature astrocyte marker glutamine synthase in TMA 2.1, cells were maintained under different culture conditions.In the presence of 2 mM glutamine, astrocytes had optimal growth conditions and they proliferated largely independent of the FCS concentrations in the culture.At lower glutamine concentrations, IMA increased their doubling time and became more dependent on high concentrations of FCS.Addition of glutamate compensated for the lack of glutamine supply, at least at high FeS concentrations.These data indicate that TMA were capable of enzymatic conversion of glutamate by glutamine synthase (Fig .SC).Such a conversion of glutamate to glutamine is a typical function of astrocytes in the brain.
For further characterization of the enzymes involved in the glutamine/glutamate uptake and synthesis as weil as other glial-related targets, mRNA expression was directly compared between primary mouse astrocytes, IMA 2.1, and the astrocyteunrelated , murine control cell types 3T3 or BV-2 .In the selected set of markers, expression in primary astrocytes was in general much higher compared with IMA 2.1 (Fig. 6A,B) .In most cases, the expression in IMA was similar to the one of a non-astrocytic cell (3T3-fibroblasts or BV-2 microglia) (Henn et al. , 2009) .The RNA data were confirm ed also immunocytoc hemi cally, e.g. , by an absence of staining for glial fibrillary acidic protein (GFAP) .Four notable exceptions were found: aquaporin 4, a typical astrocyte marker was expressed 200-fold higher in IMA than in 3T3 .Nestin was expressed in IMA to a similar extent as in primary astrocytes, which is consistent with immunostaining res ults .SI OOß and Cnx43 , typical markers expressed early in astrogenesis, had an elevated expression relative to non-astrocy tic markers .In summary, these data suggest that IMA resemble the GFAP-negative astrocytes that can be derived from stern cells (Kuegler et al., 2012) and are at a relatively immature stage.

Metabolie eapacity of IMA 2.1
In addition to the descriptive phenotype markers, we also ex amined the actual metabolie competence ofIMA.MAO-B activity was chosen because of its prime import an ce in neurotoxicology.IMA 2. 1 were compared with HEK293, HeLa, the neuronal cell line LUHMES, and SH-SYSY neuroblastoma cells with respect to MAO activity.Addition of the MAO-substrate kynuramine to (A) IMA 2.1 or primary murine astrocytes were fixed, permeabilized, and stained with antibodies against the cell surface glycoprotein CD44 and the intermediate filament protein nestin (green).Nuciei were labeled with H-33342 (red).Alternatively, fluorescence-conjugated phalloidin, selectively binding to F-actin, was used for visualization of general cell morphology of IMA 2.1 and primary astrocytes.
(9) AdditionallY' living IMA 2.1 were labeled with calcein-AM (1 pM ; 30 min), the same region was also photographed in the phase contrast mode.(C) IMA 2.1 were seeded at a density of 10,000 ce ll s/cm 2 (2 % FCS) and 5,000 cell s/cm 2 (5% and 10% FCS) in the presence of different serum concentrations (2%, 5%, 10%).For each serum concentration, variable glutamine/glutamate levels were added, cells were allowed to grow for a total of 10 days and the number of cells was determined manually every 12 h.The doubling time was calcu lated from the growth curves .Doubling time for cells grown with 2% serum and 0.1 mM glutamine cou ld not be determined as cells did not tolerate these growth conditions.The value was set to 270 h for graphical representation .Data are expressed as means ±SD of quadruplicate sampies.Statistical significance was determined by one-way ANOVA followed by 90nferroni's post-hoc test.•P<0 .05 homogenized cell s indicated a signifi cantly hi gher MAO activity in IMA compared with the other cell types investigated (Fig. 7A).Thus , as in the ca se ofNOS-2 and TLR-2, the protein data did not correlate with the mRNA data .For a potenti al application of IMA 2.l in co-culture models together with dopaminergi c neurons in the fi eld of Parkinson's di sease research , conversion of the parkinsonian toxin MPTP into the dopaminergic neurotoxin MPP+ (Fig. 7B) was investigated in IMA 2.l .Addition of MPTP to IMA 2. 1 led to a time-dependent conversion into its active metabolite MPP+ .The reaction intermediate MPDP+ was also detectable in the cell culture supernatant (Fig. 7C).The same experiment was then run with primary mouse astrocytes that displayed a similar MPTP conversion rate compared with IMA 2.1 (Fig .7D) .To identify the MAOiso form responsible for the observed MPTP conversion by IMA 2.l, cells were pretreated with the MAO-A-selective inhibitor moc\obemide, the MAO-B-selective inhibitor (-)-deprenyl, or the isoform-unselective inhibitor pargyline for 30 min be fore MPTP was added.These experiments indicated an almost exclusive contribution of MAO-B to the observed conversion of MPTP (Fig. 7E) .For a direct comparison of the influence of proinflammatory conditi ons on MAO-B activity, IMA 2 .1 were pre-treated with CCM for 6 h for full activation.When MPTP was added as MAO-B substrate, MPP+ was detected as readout in the supernatant and revealed no sign ificant difference between the control and the CCM group (Fig .7F) .These findin gs correlate weil with a large body of literature, examining MPTP in the in vivo mouse model.

Compatibility with neuronal co-cultures
Fi nally, we tested whether IMA 2. 1 can be used in a co-cu 1ture model together with the human dopaminergic neuronal cellline LUHMES .For direct comparison, primary astrocytes or IMA 2 .1 were grown until confluency was reached .Then, predifferentiated LUHMES were added and differenti ated for four days.The morphology of LUHMES was analyzed by staining of th e ne uronal marker protein ß-lIT-tubulin .As illustrated in F igure 8, differentiated LUHMES grown alone, 01' on top of either primary astrocytes or IMA 2.1 di splayed simil ar neurite structures and cell morphology.These observations indicate that IMA 2. 1 can be applied in co-culture models e ither in direct contact with neurons as illustrated herein , 01' spati all y separated in a transweIl co-culture system .

Discussion
In an aging population , the incidence of chronic neurodegenerative diseases such as PD or AD is constantly increasing.This results in intensifi ed basic and c1in ical research in thi s field , as evidenced by the higher use of experimental an imals for basic research, and in particular, the widespread generation and use of new transgenic animal models (Leist et al., 2008).In the present manuscript, we characterized the mouse cell line IMA with respect to its potential as substitute for primary astrocyte cultures.Particular emphasis was laid on IMA's response to proinflammatory stimuli and its ability to serv e as model for th e conversion of the parkinsoni an tox in MPTP.In the fie ld of PD research, in vitro models that recapitulate these two events are desired .
T he standard method to obtain astrocytes is the establishment of primary cell cultures by isolation of cells from brain tissue ofnewborn animals.This costly and time-consuming procedure represents a clear logistic bottleneck in molecular mechanistic research on astrocyte function.Hence, replacement of primary astrocytes by an adequate cell line has a high potential to advance our know ledge about astrocyte biology and , at the same time, contributes to reduce the number of experimental animals used.For successful replacement, new in vitro models however    1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), the reaction intermediate 1-methyl-4-phenyl-2,3-dihydropyridinium (MPOP+) and the active toxin 1-methyl-4-phenylpyridinium (MPP+). (C +0) MPTP (20 11M) was added to IMA 2.1 or to primary mouse astrocyte cultures.At the time points indicated, supernatant was co llected and MPTP, MPDP+, and MPP+ were measured by HPLC.(E) To investigate which MAO isolorm accounts lor the observed activity, MPTP (20 11M) was added together with the MAO-A se lective inhibitor moclobemide, the MAO-B selective inhibitor (-)-deprenyl, or the isolorm-unselective inhibitor pargyline.After 24 h, MPP+ was evaluated in the supernatant by HPLC.(F) For a comparison on the impact 01 inllammation on MAO-B activity, IMA were stimulated with CCM lor Fig. 8: Appllcatlon oflMA 2.1 In co-culture models Neuronal LUHMES cells were either differentiated alone (control).or on top 01 conlluent layers of IMA 2.1 or primary rat glia cells.For a phenotypic assessment, LUHMES were stained for neuronspecilic ß-III-tubulin (green), nuclei were stained with Hoechst H-33342 (red).require aeeeptanee by the seientifie eommunity.Previous astroeyte eelliines failed to become widely distributed , and they failed to be adopted by the research community.
Primary cultures of astrocytes are undoubtedly closer than any immortalized celliine to an in vivo situation, however also primary eells have, aside from limited availability, clear cut limitations.For example, contamination with microglia ean signifieantly affeet the ollteome of an investigation (Saura , 2007) .It is also important to realize that brain region-dependent differenees in astroeytic phenotype ex ist that ean have a profound impact on the observations made.Moreover, it has been clearly demonstrated that the eommonly used cultures of neonatal astroeytes differ vastly from adult astroeytes in the brain (Zhang and Barres, 2010).
In the present study, we evaluated the mouse astrocyte eell line TMA with respeet to its applicability in inflammation stlldies and for its suitability as an MPTP eonversion model.In comparison to primary astroeytes, TMA are easy to handle, do not require laborious isolation methods that usually lead to bateh-to-bateh variations, and are stable in their phenotype over years, whieh is aprerequisite for the reproducibility of results and inter-Iaboratory comparability.This aspect also opens the field for potential applications in va lidation assays.We also observed that , in contrast to other ceH types, genetic overexpression by transfeetion or electroporation, 01' the siRNA-mediated knockdown of targets of interest can be performed in IMA (data not shown) .
The suitability of TMA as inflammation model was assessed by the detection ofNF-KB translocation into the nucleus, whieh represents one of the key events involved in the induetion of so-called immediate early genes.The deteetion method by an automated microscope system on the single cell level showed that all IMA reacted simultaneously to different cytokines.Such a homogeneous cell population is of great advantage for many biochemieal and signal transduction studies.For a general overview of the inftammatory competence, we have chosen ehanges of mRNA levels as readout, as suggested previously (Kuegler et al., 2010) .However, characterization based on transcript levels alone represented only a first overview.Further information on the actual expression, localization, and activity of a protein or enzyme was therefore presented here for NOS-2, MAO-B , or TLR-2, as examples.
The low GFAP mRNA expression, and the absence of GFAP staining (not shown) in IMA might raise concerns about the astrocytic origin of the cells.However, the original cell pool stained uniformly positive for GFAP directly after transduction .During the four passages thereafter, the expression of GFAP was lost, but the cells maintained the same morphologic features.It has previously been reported that GFAP expression depends highlyon the proliferation status of cells.Accordingly, GFAP was reported to be downregulated in a proliferating astrocyte cellline (Price et al., 1999) .This might explain the low abundance observed in IMA.It is also becoming increasingly clear that a large subpopulation of astrocytes does not express GFAP, also under in vivo conditions .Confluent eells kept for more than a week displayed elevated levels of GFAP that were however still signifieantly lower eompared with th e expression in pri-mary astrocytes.The established cell line IMA 2.1 obviously also lacks so me genuine astrocyte functions such as functional expression of glutamine synthase, as weil as specific uptake of glutamate .These observations suggest limitations of IMA 2.1 in the field of glutamate/EAAT-toxicity research.
For potential applications in parkinsonian MPTP models, we compared the MAO-dependent conversion of MPTP by IMA and primary mouse glial cells that indicated a comparable MAO-B activity in IMA when adjusted to equal total protein content.The IMA cell line therefore represents an easily accessible platform for mechanistic studies on MPTP conversion.
In order to c10ser simulate the situation in the brain, we then tested whether IMA can form co-cultures with the human dopaminergic celliine LUHMES.The LUHMES cellline that can be differentiated into post-mitotic neurons with a dopaminergic phenotype is very sensitive to culture conditions.Therefore, the successful co-cultivation was initially surprising, and these data indicate that IMA 2.1 can most Iikely also be used in combination with other neuronal cell types (Stiegleret al., 20 11;Zimmer et al., 2011).The rationale for using a human neuronal cell li ne is based .on the intended use of $uch co-cultures for human disease modeling .The relevant target cells for most diseases are neurons.To our current knowledge, LUHMES cells represent one of the most favorable in vitro models of dopaminergic neurons .We recently characterized this celliine (Scholz et al., 20 11;Schildknecht et al., 2009) extensively, and we believe that no mouse cellline with similar features is available at present.Vice versa, the use of human astrocytes is much less established than the one of murine astrocytes, and the cells are less accessible.Notably, the combination of cells from different species does not result in histoincompatibility reactions , as long as no T cells are present .Moreover, most metabolites that are exchanged, in addition to most neurotransmitters, lipid mediators , growth factors and xenobiotic metabolites relevant to the co-culture, are not species-specific .Therefore , combinin g cell types of different species created no difficulties .We are aware of the disadvantages of the two-species system for the study of certain immune medi ators (e.g ., IFN-y) that are species-specific.On the other hand , two species systems also have enormous advantages, as certain cytokines derived from one of the cell types may be blocked selectively, without inflllencing the same cytokine from the other cell type (e.g. , human or murine CCL-5), and especially the regulation of transcripts can be followed for each cell type selectively by choosing appropriate PCR primers.Such advantages can be very important for mechanistic studies of reciprocal ce li-type interactions.In preliminary results (not shown), we observed that MPTP addition to the co-culture model of IMA and LUHMES leads to a time-dependent conversion into the active toxin MPP+ that subsequently is taken up by LUHMES cells and leads to a selective neurite degeneration and ultimately cell death while viability of TMA, in which the conversion takes place , is not affected.These findings however require further investi gations and confirmation .
The suitability of IMA for co-culture models in combination with neurons opens a promising perspective for a new in vitro system to study celllllar interactions that may be relevant for disease.For instance, mechanistic and kinetic aspects of the toxicity of the Parkinsonism-inducing toxicant MPTP may be studied, wh ich is not possible in pure neuronal cultures.

Fig. 1 :Fig. 2 :
Fig.1: Inflammatory response to different cytoklnes (A) IMA 2.1 were stimulated with IL-1ß (10 ng/ml), TNF-a (10 ng/ ml), or IFN -y (20 ng/ml), or by a complete cytokine mix (CCM) composed 01 all three cytokin es lor 30 min.Cells were then lixed and stained with an anti-NF-xB p65 antibody.Quantitative assessment 01 NF-KB translocation Irom the cytosol to the nucleus was perlormed with an automated microscope (Cellomics Array Scan).This method allows detection 01 NF-KB translocation on the single cell level.Data are displayed as percentage 01 cells positive for NF-KB translocation compared to all cells detected .(B) IM A 2.1 were stimulated with cytokines as indi cated for 16 h, nitric oxi de synthase-2 (NOS-2) mRNA wa s detected as representative lor an early immediate gene.Data are expressed as means ±SD 01 quadruplicate sampi es.Statisti cal signilicance was determin ed by one-way ANOVA lollowed by Bonferroni 's po st-hoc test.•P<0 .05 activated by CCM (IL-1ß, TNF-a, IFN -y) for the time periods investigated.(A) The amount 01 cyclooxygenase-2 (COX-2) and nitric oxide synthase-2 (NOS-2) was evaluated on the protein level as shown by Western blot analysis.(B) Prostagiandin E2 (PGE2) as indicator for COX activity and nitrite as indicator lor NOS activity were assessed in the respective cell culture supernatants.(C+D) To discriminate between NOS-2-dependent "NO lormation and "NO derived Irom constitutively expressed NOS proteins, the NOS-2 selective inhibitors AMT or L-NMMA were applied in different concentrations together with a cytokine mix lor 16 h.Nitrite served as indicator for NOS activity.Data are expressed as means ±SD 01 quadruplicate sampies.Statistical significance was determined by one-way ANOVA followed by Bonferroni's post-hoc test.'P <0.05 Regulation of Inflammatlon-relevant proteins (A) IM A 2.1 were stimulated with CCM for the time intervals indicated, enzymatic activity of COX was directly assessed in cell Iysates by the addition of 14C-labeled arachidonic acid.(B) IMA 2.1 were activated with a cytokine mix as indicated, cells were sta in ed with an anti-TLR-2 antibody and detected by flow cytometry.(C) IMA 2.1 were treated with CCM , brefeldin A (BFA), or a combination 01 CCM and BFA lor 16 h, TLR-2 surface expression .was analyzed by flow cytometry.Data are expressed as means ±SD of quadruplicate sampies.Statistical signili cance was determined by one-way ANOVA lollowed by Bonlerroni's post-hoc test.•P<0.05

(
A) For a direct comparison 01 monoamine oxidase (MAO) activity, homogenates (1 mg/mi protein) 01 IMA 2.1 , the neuronal ceillines LUHMES and SH-SY5Y, or HeLa and HEK 293 were incubated with the MAO substrate kynuramine (100 11M) lor 2 h.The reaction product hydroxychinoline was measured Iluorimetrically and the resulting concentrations were plotted.(B) Overview on the structures 01