Introduction

The brain is the organ with the highest energy demand in mammalian organisms. Among neural cells, astrocytes play a central role in regulating cerebral energy metabolism in dependence on neuronal activity (Vernadakis 1996; Pellerin et al. 2007). There are mainly two substrates to fulfill this energetic requirement, i.e. glucose and oxygen. The latter also acts as substrate of cytochrome c oxidase (COX), the terminal enzyme of the mitochondrial respiratory chain which is engaged in oxidative energy metabolism. COX is located in the inner mitochondrial membrane, where it catalyzes the terminal electron transfer from ferrocytochrome c to oxygen. In parallel to this process, protons are pumped across the inner mitochondrial membrane. Subsequently, this electrochemical gradient is used by the ATP synthase to produce ATP, the energy “currency” in cells. The catalytic core of the enzyme consists of three mitochondria-encoded subunits which form the catalytic center surrounded by ten nucleus-encoded subunits which fulfill regulatory functions (Tsukihara et al. 1996). For the regulatory subunits, isoforms have been identified including COX subunit IV. These isoforms are expressed in a tissue-specific and development-regulated manner (Grossman and Lomax 1997). Recently, we demonstrated that COX subunit IV isoforms (COX IV-1 and COX IV-2) are expressed depending on the brain cell type, the oxygen concentration, and toxic substances (Horvat et al. 2006; Singh et al. 2009). COX isoform IV-1 is expressed in all mammalian tissues including brain, whereas COX IV-2 shows higher transcription levels in neurons compared with astrocytes, and is up-regulated in astrocytes under hypoxic and toxic conditions. Treatment of astrocytes with 3-nitropropionic acid (NPA) induced the transcription of COX IV-2 in astrocytes from striatum (Singh et al. 2009). COX subunit IV is a key regulatory subunit of the COX complex. At high cellular energy levels, ATP acts as an allosteric inhibitor of the enzyme by binding to the matrix domain of subunit IV (Arnold and Kadenbach 1997, 1999). Thus, COX enzyme is enabled to sense the ATP/ADP ratio and to adjust the enzyme activity to cellular energy requirements. Except for the regulation of ATP production, COX subunit IV plays also an important role in the production of reactive oxygen species (ROS). We could recently show that a switch from COX IV-1 to COX IV-2 expression caused an increase in mitochondrial ROS production which was accompanied by increased necrotic cell death (Singh et al. 2009). For these reasons, COX subunit IV represents an interesting COX subunit candidate to be studied in the context with the correlation of enzyme structure, function, and regulation. A switch in the expression of COX subunit IV isoforms is of major importance for the regulation of enzyme activity. At physiological oxygen concentrations, COX subunit isoform IV-1 is expressed, whereas during oxygen deprivation, cortical astrocytes react with an elevation of COX IV-2 (Horvat et al. 2006). Another possibility of inducing hypoxia without affecting the oxygen concentration is the application of respiratory chain toxins. Here, we investigated the effect of so called inducers of chemical hypoxia, such as 10 mM sodium azide (NaN3), 10 mM potassium cyanide (KCN), and 50 μM cobalt chloride (CoCl2), on COX subunit IV isoform transcription and viability of astrocytes from different brain regions of female and male mice. We have chosen to study female and male cells separately with respect to a number of degenerative and toxic processes in the CNS showing a gender specificity that could result from a gender-specific regulation of COX. In this study, we observed an up-regulation of COX IV-2 in astrocytes under chemical hypoxia depending on the applied toxin and the gender and brain region.

Materials and Methods

Materials

Chemicals and reagents for molecular biological and cell culture techniques were purchased from Roth (Karlsruhe, Germany) or Invitrogen (Karlsruhe, Germany), unless noted otherwise.

Animals

BALB/c mice were obtained from Harlan Winkelmann GmbH (Borchen, Germany). Animal handling was performed in strict accordance with the published welfare rules for the care and use of laboratory animals at the University Clinic Aachen and the Government of the State of Nordrhein-Westfalen, Germany.

Preparation of Primary Astrocyte Cultures

Astroglial cultures were prepared from cortex and mesencephalon of postnatal day 1 female and male BALB/c mice in parallel in the exact same way and as previously documented (Horvat et al. 2006). The gender was determined by visual inspection of the anogenital distance that is larger in males (Beyer et al. 1991). The correctness of determining anatomically the gender of the mice was confirmed for the astrocyte cell populations by quantitative real time-PCR (qRT-PCR) checking for female- (Xist) and male-specific (SRY) gene transcription. Brains from decapitated mice were removed and transferred in preparation buffer consisting of 10 mM HEPES, 154 mM NaCl, 10 mM glucose, 2 mM KCl, and 15 μM BSA. Brain cortex and mesencephalon were dissected and after removal of the meninges, tissue pieces were incubated in PBS containing 0.1% (v/v) trypsin and 0.02% (w/v) EDTA for 15 min. Subsequently, it was minced with a Pasteur pipette and filtered through a 50 μm nylon mesh. The cell suspension was centrifuged at 300g (Eppendorf, Hamburg, Germany) for 4 min. The cell pellet was re-suspended and the cell suspension plated onto poly-l-ornithine-coated (Sigma-Aldrich, Munich, Germany) culture dishes and incubated in Dulbecco’s modified Eagle’s medium (DMEM; PAA, Coelbe, Germany) supplemented with 16% (v/v) fetal calf serum (PAA, Coelbe, Germany), 50 U/ml penicillin, 50 μg/ml streptomycin, 0.25 μg/ml amphotericin B (Fungizone®), 2 mM l-glutamine (Glutamax®). Cultures were maintained at 37°C in a humidified atmosphere of 95% air/5% CO2. Before reaching confluence, female and male astrocytes were in parallel trypsinized and plated at lower density. As female were not distinct from male astrocytes in their growth behavior, sub-confluent female and male astrocytes were taken on the same day in vitro and were in parallel incubated in neurobasal medium (NBM) supplemented with 0.2% (v/v) B27, penicillin, streptomycin, Fungizone®, and l-glutamine for 48 h and subsequently used for experiments.

The purity of cell culture has been proven by immunocytochemical staining of astrocytes with an antibody against glial fibrillary acidic protein (GFAP), a marker protein for astrocytes. As >95% of the cells were stained for GFAP, the astrocyte cultures were considered as virtually free of neurons, oligodendrocytes, and microglia.

Cell Treatment

Primary female and male astrocyte cell cultures placed on culture dishes at 37°C in a humidified atmosphere of 95% air/5% CO2 in the presence of NBM were treated on the same day in vitro and at the same density for chemical hypoxia with different inhibitors of mitochondrial respiration, i.e. 10 mM sodium azide, 10 mM potassium cyanide, and 50 μM cobalt chloride for 6 h (Sigma-Aldrich, Munich, Germany). Cells of the same preparation maintained under the same conditions except for the toxic treatment served as controls.

Cell Viability, Apoptosis and Necrosis Assays

Determination of cell viability was performed using the trypan blue exclusion. Treated and untreated astrocytes were washed with PBS, trypsinized, and immediately stained with 0.4% (w/v) Trypan Blue solution (Fluka, Steinheim, Germany) and counted. Cell viability was calculated as percentage of the ratio of unstained (viable) cells to the total number of cells (stained and unstained cells). Cells were counted from three distinct areas of three independent cell preparations. Viability of control cells was set 100%.

To distinguish apoptotic from necrotic astrocytes, toxin- and un-treated control cells were cultured on poly-l-ornithine coated coverlips and stained with 3 μM YO-PRO®-1 (Invitrogen, Karlsruhe, Germany) for 30 min followed by addition of 2 μg/ml Hoechst 33342 Trihydrochlorid (Hoechst; Invitrogen) for 10 min under culturing conditions (Horvat et al. 2006). Subsequently, cells were washed twice with PBS. Fixation was performed with 100% (v/v) ice-cold methanol (Merck) for 10 min at -20°C. Afterwards cells were rinsed with PBS and incubated in 2× SSC buffer consisting of 0.3 M NaCl, 0.03 M sodium citrate, pH 7.0, for 20 min. Subsequently, cells were treated with 1 μM PI (Invitrogen) in 2× SSC buffer for 5 min followed by washing with PBS and mounting. Viable (Hoechst-positive), early apoptotic (Hoechst- and PI-positive), late apoptotic and necrotic (Hoechst-, YO-PRO®-1-, and PI-positive) cell nuclei were scored under a fluorescence Axiophot microscope (Carl Zeiss, Oberkochern, Germany) at excitation wavelengths of 365 (Hoechst), 485 (YO-PRO®-1), and 530 nm (PI). Early apoptotic cells showed a YO-PRO®-1-/PI-positive cytoplasm, whereas late apoptotic were distinguished from necrotic cells by their additional YO-PRO®-1- and PI-positive cytoplasm. Between 100 and 500 cells were quantified in three separate experiments by counting cells in three distinct randomly selected areas per coverslip using the NIS-elements AR 3.0® software (Nikon, Düsseldorf, Germany).

Reverse Transcription

Total RNA was isolated from treated cells and untreated controls using PeqGold RNA pure (PeqLab, Erlangen, Germany) according to manufacturer’s protocol. RNA concentration was measured photometrically using BioPhotometer (Eppendorf). RNA integrity was tested randomly by 2% (w/v) agarose denaturing gel electrophoresis and ethidium bromide staining and visualized under UV-illumination. First strand complementary DNA (cDNA) was synthesized from 1 μg total RNA. In brief, total RNA dissolved in 11 μl diethyl pyrocarbonate- (DEPC-) H2O was pre-incubated at 70°C for 5 min and placed immediately on ice. Subsequently, the reaction buffer consisting of 8 U/μl SuperScript™ III Reverse Transcriptase, 4 mM dithiotreitol, 40 mM Tris–HCl, 60 mM KCl, and 2.4 mM MgCl2, and 0.4 mM each dNTP (Roti-mix® PCR3, Roth, Karlsruhe, Germany) was added to RNA giving a final volume of 25 μl. After incubation for 60 min at 37°C, reverse transcription was stopped by heat-inactivating the enzyme at 70°C for 15 min. Addition of water instead of RNA served as negative control. Transcripts of 18S ribosomal RNA (18S rRNA) served for normalization of samples, whereas the housekeeping gene hypoxanthine guanine phosphoribosyl transferase (HPRT), spanning over intron–exon borders, served as control for RNA purity.

Quantitative Real Time-PCR Analysis

Quantitative real time-PCR (qRT-PCR) analysis of COX subunit IV isoforms was performed using SYBR® Green technology and carried out on the iQ5 detection system (Bio-Rad, Munich, Germany). Forward and reverse primers for specific amplification of Cox4i1 (5′-TATGCTTTCCCCACTTACGC-3′ and 5′-GCCCACAACTGTCTTCCATT-3′) and Cox4i2 (5′-AGATGAACCATCGCTCCAAC-3′ and 5′-ATGGGGTTGCTCTTCATGTC-3′), Hif1a (5′-TCAAGTCAGCAACGTGGAAG-3′ and 5′-TATCGAGGCTGTGTCGACTG-3′), and Hif2a (5′-AGCCAAACACGGAGGATATG-3′ and 5′-GTGTGGCTTGAACAGGGATT-3′), were designed eliminating the possibility of amplifying genomic DNA. For each set of primers, a basic local alignment search tool (BLAST, NCBI) search revealed that sequence homology was obtained specifically for the target gene. Standard and sample cDNA obtained after reverse transcription were diluted 1:10 and added to a solution containing 5 μM primers and IQ SYBR Green Supermix (Bio-Rad, Munich, Germany) consisting of 25 U/ml iTaq polymerase, 50 mM KCl, 20 mM Tris–HCl, 0.2 mM each dNTP, 3 mM MgCl2, SYBR Green I and stabilizers. The RT-PCR protocol was composed of an initial denaturation step for 3 min at 95°C followed by 40 cycles consisting of 10 s at 95°C, 30 s at the appropriate target gene annealing temperature (60°C—Cox4i1, Hif1a, Hif2a, 18S rRNA; 61°C—Hprt; 65°C—Cox4i2), 30 s at 72°C, and 10 s at 78°C. To obtain melting curves for the resulting PCR products, a final step was added to the RT-PCR consisting of 81 cycles of increasing temperature from 55°C to 81°C by 0.5°C for 10 s each step. The PCR products were quantified using the relative ΔC t method. Relative quantification relates the PCR signal of the target transcript to that of 18S rRNA in treated with respect to untreated cells. A test for an approximately equal efficiency of target amplification was performed by looking at ΔC t value variations with template dilutions. 18S rRNA and Hprt served as endogenous control in the validation experiments. The absolute value of the slope of log input amounts versus ΔC t should be approximately -3.3 and the efficiency approximately 100%. The validation experiments passed this test. The results are expressed as an average of triplicate samples of at least three independent experiments for control and treated cells.

Western Blot Analysis

For immunoblotting, primary cortical and mesencephalic cultures of astrocytes were rinsed in PBS and subsequently lysed in ice-cold hypotonic RIPA buffer consisting of 50 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1% (v/v) Nonidet P-40 (Sigma, Igepal, CA, USA), and protease inhibitor cocktail (Complete Mini, Roche, Mannheim, Germany). After centrifugation for 20 min at 13,000g and 4°C, the supernatant was collected and protein concentration was determined by applying the BCA™ Protein Assay Kit (Pierce, Bonn, Germany) according to manufacturer’s protocol, where BSA served as standard. Protein samples (30 μg per lane) were loaded onto and separated by 12.5% (v/v) discontinuous sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto a Hybond™ ECL™ nitrocellulose membrane (Amersham Biosciences, Buckinghamshire, UK) using Trans-Blot® SD Semi-Dry Electrophoretic Transfer Cell (Bio-Rad). An equal loading of samples and complete protein transfer were checked by staining of gels and blots with Coomassie blue and Ponceau-S, respectively. Unspecific binding sites on the membrane were blocked for 1 h with 5% (w/v) non-fat dry milk in TBST buffer consisting of 20 mM Tris, pH 7.6, 150 mM NaCl, and 0.1% (v/v) Tween 20. The nitrocellulose membrane was incubated overnight at 4°C with polyclonal anti-HIF-1α antibody (Novus Biologicals, LLC, Littleton, CO, USA) or polyclonal anti-β-actin antibody (Sigma-Aldrich) in TBST buffer containing 5% (w/v) non-fat dry milk. Protein detection was performed with peroxidase-conjugated goat anti-rabbit (Sigma-Aldrich) secondary antibody (Abcam). Peroxidase activity was visualized using enhanced chemiluminescence, ECL™ method (Amersham Biosciences).

Measurement of Intracellular Levels of Reactive Oxygen Species

Primary astrocytes were cultured in 96-well plates at a density of 5 × 104 cells/well. After treatment with or without the chemical toxins for 6 h, cells were loaded with 10 μM 6-carboxy-2′,7′-dichlorodihydrofluorescein diacetate, di(acetoxymethyl ester) (H2DCFDA AM; Invitrogen) diluted in PBS buffer and incubated for 45 min at 37°C. H2DCFDA AM is able to penetrate cells due to the acetoxymethyl ester which is hydrolyzed by intracellular esterases to form 2′,7′-dichlorofluorescin (DCFH). Oxidation of DCFH by hydrogen peroxide and hydroxyl radicals yields a highly fluorescent product, 2′,7′-dichlorofluorescein (DCF). The fluorescence intensity of DCF after excitation of the samples at a wavelength of 485 nm was measured at an emission wavelength of 535 nm using a fluorescence microplate reader (Tecan GENios, Crailsheim, Germany). The fluorescence intensities were normalized to the number of viable cells calculated by applying the Trypan Blue method and compared to control samples set 100%.

Statistical Analysis

Data are presented as mean ± SEM of at least three independent experiments. For statistical analysis, data were analyzed by applying ANOVA followed by a post hoc two-tailed, independent Student’s t test applying the software SPSS (Chicago, IL, USA). Values were regarded as statistically significant at ***p < 0.001, **p < 0.01, *p < 0.05.

Results

Impairment of the Viability of Astrocytes from Cortex and Mesencephalon by Treatment with Inducers of Chemical Hypoxia

Primary astrocytic cultures mainly consisted of GFAP-immunoreactive cells (>95%) and were virtually free of neurons, oligodendrocytes, and microglial cells. To assess gender- and brain region-specific effects of inducers of chemical hypoxia on astrocytes, we treated cells from cortex and mesencephalon of female and male mice with NaN3, KCN, and CoCl2 for 6 h at different concentrations. Staining of astrocytes from both brain regions and both genders with Trypan Blue (Figs. 1a, b, 2a, b), Hoechst (Figs. 1c–f, 2c–f), and propidium iodide (data not shown) revealed optimal concentrations for NaN3, KCN, and CoCl2 to yield half-maximal effects on cell death rates of 10 mM, 10 mM, and 50 μM, respectively. Cortical and mesencephalic astrocytes showed maximum cell death of 5% under control conditions, whereas treatment of cells with the chemical toxins caused an elevation up to approximately 40% after 6 h (Figs. 1a, b, 2a, b). Strongest effects were seen with CoCl2 (Figs. 1, 2). Distinguishing apoptotic from necrotic cell death revealed similar results except for female cortical astrocytes showing the highest number of necrotic cells (5%) after KCN treatment (Fig. 1e). The least potent inducer of cortical astrocyte death was azide causing approximately 20 and 30% death rates in female and male astrocytes, respectively (Fig. 1a, b). Mesencephalic astrocytes survived the treatment with KCN and NaN3 better showing 79/72% and 68/70% viability of female and male cells, respectively (Fig. 2a, b). Accordingly, apoptotic and necrotic cell death was lowest for female/male cortical astrocytes (21/28 and 1/2%, respectively) treated with azide (Figs. 1c–f) and for mesencephalic female/male astrocytes (21/29 and 0/1%, respectively) treated with KCN (Fig. 2c–f). Brain region-specific differences were mainly seen for necrotic cell death exhibiting predominance in males (Figs. 1e, f, 2e, f). Similar gender-specific differences were observed for azide and cyanide (Figs. 1, 2). Apoptosis was more prominent than necrosis in cortical and mesencephalic cell populations (Figs. 1c–f, 2c–f). In general, female astroglia appeared to be less vulnerable to chemical hypoxia than male. Brain region-specific differences apparently concerned the different levels of effectiveness of the three toxins on cell viability, apoptosis and necrosis rates which could be based on different mechanisms of action of the applied toxins.

Fig. 1
figure 1

Viability of primary cortical astrocytes is impaired after treatment with azide (NaN3), cyanide (KCN), and cobalt (CoCl2). a, b The percentage of viable primary cortical astrocytes from female (a) and male (b) brains was decreased after 6 h treatment with 10 mM NaN3, 10 mM KCN, 50 μM CoCl2 when compared to untreated astrocytes. Viability was assessed by staining astrocytes with Trypan Blue and Hoechst 33342, PI and YO-PRO®-1. The number of untreated and treated viable cells was counted and expressed as % of total cell number undergoing the experiment. c–f Hoechst-, PI- and YO-PRO®-1-stained astrocytes from females (c, e) and males (d, f) were analyzed for apoptosis (c, d) and necrosis (e, f). Both, apoptosis and necrosis, were increased after toxin treatment with apoptosis showing a higher percentage of cell death. Significant gender-specific differences (#, indicated in graphs for male samples) were observed for viable cell number, apoptotic and necrotic cell death rates. Data represent mean ± SEM (# p < 0.05, ## p < 0.01, *** p < 0.001, n = 8). gk Astrocytes stained with fluorescent viability indicator dyes are presented in a merged picture (g) of Hoechst- (h, blue), PI- (i, red), and YO-PRO®-1-stained cells (k, green). Viable (arrow; Hoechst-positive), early apoptotic (star; Hoechst-, PI-positive), late apoptotic (arrowhead) and necrotic (block arrow; Hoechst-, PI-, YO-PRO®-1-positive) fluorescent cell nuclei are indicated. Early apoptotic cells showed a YO-PRO®-1-/PI-positive cytoplasm. Late apoptotic were distinguished from necrotic cells by their additional PI- and YO-PRO®-1-positive cytoplasm. (Color figure online)

Fig. 2
figure 2

Viability of primary mesencephalic astrocytes is impaired after treatment with azide (NaN3), cyanide (KCN), and cobalt (CoCl2). a, b The percentage of viable primary mesencephalic astrocytes from female (a) and male (b) brains was decreased after 6 h treatment with 10 mM NaN3, 10 mM KCN, 50 μM CoCl2 when compared to untreated astrocytes. Viability was assessed by staining astrocytes with Trypan Blue and Hoechst 33342, PI and YO-PRO®-1. The number of untreated and treated viable cells was counted and expressed as % of total cell number undergoing the experiment. cf Hoechst-, PI- and YO-PRO®-1-stained astrocytes from females (c, e) and males (d, f) were analyzed for apoptosis (c, d) and necrosis (e, f). Both, apoptosis and necrosis, were increased after toxin treatment with apoptosis showing a higher percentage of cell death. Significant gender-specific differences (#, indicated in graphs for male samples) were observed for viable cell number, apoptotic and necrotic cell death rates. Data represent mean ± SEM (# p < 0.05, ## p < 0.01, *** p < 0.001, n = 8)

Gender- and Brain Region-Specific Regulation of COX Subunit IV Isoform Transcription by Chemical Hypoxia

Quantitative real time-PCR (qRT-PCR) analysis revealed higher basal transcription level of COX IV-1 compared to COX IV-2 under normal culturing conditions of female and male astrocytes from both brain regions (data not shown). Significant changes in the transcription of COX isoforms IV-1 and IV-2 occurred after treatment with the chemical toxins (Figs. 3a–d, 4a–d). COX IV-1 transcript levels in female cortical astrocytes were increased by 120 and 170% after treatment with azide and cyanide, respectively (Fig. 3a), and no changes were observed in male astrocytes (Fig. 3b). In contrast, treatment with cobalt caused no effect in female but a decrease by approximately 40% in male astrocytes (Fig. 3a, b). A comparison of cortical with mesencephalic astrocytes also revealed significant differences. Azide and cyanide increased the COX IV-1 transcript levels in male mesencephalic astrocytes, whereas azide had no effect and cyanide provoked a decrease by approximately 40% in female astrocytes (Fig. 4a, b). Cobalt decreased COX IV-1 by 20–30% in both genders (Fig. 4a, b).

Fig. 3
figure 3

Inducers of chemical hypoxia caused a toxin-dependent change of COX IV-1 and COX IV-2 transcript levels. ad COX IV-1 and IV-2 transcript levels were quantified by qRT-PCR after treatment of cortical astrocytes from females (a, c) and males (b, d) for 6 h with 10 mM NaN3, 10 mM KCN, and 50 μM CoCl2. Transcription levels were normalized to 18S rRNA and compared with untreated control cells (* p < 0.05, *** p < 0.001, n = 5). a, b Astrocytes showed a maximal increase of COX IV-1 transcription after KCN and NaN3 treatment of female (a) and a decrease after CoCl2 treatment. c, d COX IV-2 transcription was increased after KCN application in female astrocytes (c) and decreased after KCN and CoCl2 treatment in males (d). e, f COX IV-1/COX IV-2 ratio showed a decline after treatment with NaN3 > KCN > CoCl2 in female (e) and male astrocytes (f). Significant gender-specific differences (#, indicated in graphs for male samples) are shown (# p < 0.05, ## p < 0.01)

Fig. 4
figure 4

Inducers of chemical hypoxia caused a toxin-dependent change of COX IV-1 and COX IV-2 transcript levels. ad COX IV-1 and IV-2 transcript levels were quantified by qRT-PCR after treatment of mesencephalic astrocytes from females (a, c) and males (b, d) for 6 h with 10 mM NaN3, 10 mM KCN, and 50 μM CoCl2. Transcription levels were normalized to 18S rRNA and compared with untreated control cells (* p < 0.05, *** p < 0.001, n = 5). a, b Astrocytes showed a decrease of COX IV-1 transcription after KCN and CoCl2 treatment in female cells (a) and an increase after NaN3 and KCN treatment but a decrease after CoCl2 application in males (b). c, d COX IV-2 transcription was decreased after KCN and increased after CoCl2 application in female astrocytes (c) and increased after NaN3 and KCN treatment of male cells (d). e, f COX IV-1/COX IV-2 ratio showed a decrease upon NaN3 and CoCl2 in females (e) and an increase after NaN3, KCN and a decrease after CoCl2 in male astrocytes (f). Significant gender-specific differences (#, indicated in graphs for male samples) are shown (# p < 0.05, ## p < 0.01)

COX IV-2 transcription exhibited clear gender- and brain region-specific differences. COX IV-2 was up-regulated (by approximately 130%) in female cortical astrocytes treated with KCN (Fig. 3c). Azide and cobalt showed no influence on female astrocytes (Fig. 3c), whereas all three toxins exerted a decreasing effect by approximately 50% on COX IV-2 transcription in male cortical cells (Fig. 3d). In mesencephalic female cultures, COX IV-2 transcription levels were increased by approximately 80% after cobalt, decreased by approximately 30% after cyanide, and not affected by azide (Fig. 4c). An opposing trend was detected in male astrocytes, where azide and cyanide induced COX IV-2 transcription by approximately 90 and 50%, respectively, and where cobalt showed no effect (Fig. 4d).

Accordingly, the ratios of COX IV-1/COX IV-2 also exhibited gender- and brain region-specific characteristics (Figs. 3e, f, 4e, f). The COX IV-1/COX IV-2 ratio was higher than one after azide and cyanide treatment in female and male cortical astrocytes, whereas cobalt caused a decrease below one in female and a ratio of approximately one in male astrocytes (Fig. 3e, f). Generally, the ratios were shifted to higher values in male compared to female cells. In male mesencephalic astrocytes, cyanide and azide induced a ratio higher than one, whereas cobalt showed values below one (Fig. 4f). In female mesencephalic astrocytes, azide and cobalt caused COX IV-1/COX IV-2 ratios below one and cyanide equal to one (Fig. 4e). Apparently, the COX IV-1/COX IV-2 ratios correlated reverse proportionally with apoptotic cell death rates according to gender and brain region.

Gene Expression of HIF-1α and HIF-2α

Hypoxia-inducible factor (HIF) is known to regulate COX IV isoform expression under hypoxic conditions. By studying HIF expression, we found no correlation between HIF and COX subunit IV isoform transcription upon chemical hypoxia. However, we detected a gender-specific and to some extent region-specific regulation of HIF isoforms by the toxins (Figs. 5, 6). HIF transcripts remained unchanged except for HIF-2α, which was down-regulated after cyanide treatment of female cortical astrocytes (Fig. 5c), whereas HIF-1α protein levels were increased to a similar extent after treatment with all three toxins (Fig. 5e). Male cortical astrocytes revealed an approximately twofold increase in HIF-1α transcript level upon azide and cyanide treatment (Fig. 5b) but no changes of HIF-2α (Fig. 5d). HIF-1α protein levels were highest after azide treatment followed by cyanide-induced increase (Fig. 5f), thereby reflecting the observed transcription data. Cobalt was effective in down-regulating HIF-1α transcription in male astrocytes (Fig. 5b), an observation that has not been made on the protein level where cobalt caused a higher HIF-1α protein level than in control cells (Fig. 5f).

Fig. 5
figure 5

Inducers of chemical hypoxia affected HIF-1α and HIF-2α transcripts in cortical astrocytes. HIF-1α (a, b) and HIF-2α (c, d) transcript levels were quantified by qRT-PCR and are shown as Western Blot data for HIF-1α normalized to β-actin (e, f) after treatment of cortical astrocytes from females (a, c, e) and males (b, d, f) for 6 h with 10 mM NaN3, 10 mM KCN, and 50 μM CoCl2. Transcription levels were normalized to 18S rRNA and compared with untreated control cells (* p < 0.05, n = 5). (A/B) Astrocytes showed an increase of HIF-1α transcription after NaN3 and KCN and a decrease after CoCl2 treatment of males (b), but no changes in females (a). c, d HIF-2α revealed no changes except for KCN-treated female astrocytes (c). Significant gender-specific differences (#, indicated in graphs for male samples) were observed for HIF-1α (# p < 0.05, ## p < 0.01). e, f HIF-1α protein levels were up-regulated after treatment with all three toxins in female and male astrocytes. Highest protein levels were detected after treatment with NaN3 and KCN in male astrocytes

Fig. 6
figure 6

Inducers of chemical hypoxia affected HIF-1α and HIF-2α transcripts in mesencephalic astrocytes. HIF-1α (a, b) and HIF-2α (c, d) transcript levels were quantified by qRT-PCR and are shown as Western Blot data for HIF-1α normalized to β-actin (e, f) after treatment of mesencephalic astrocytes from females (a, c, e) and males (b, d, f) for 6 h with 10 mM NaN3, 10 mM KCN, and 50 μM CoCl2. Transcription levels were normalized to 18S rRNA and compared with untreated control cells (* p < 0.05, n = 5). a, b Astrocytes showed no changes of the transcription of HIF-1α in females (a), but an increase after treatment with NaN3, KCN, and CoCl2 in males (b). c, d HIF-2α revealed a decrease in CoCl2-treated female astrocytes (c) and an increase in NaN3-treated male astrocytes (d). Significant gender-specific differences (#, indicated in graphs for male samples) were mainly observed for HIF-1α (# p < 0.05, ## p < 0.01). e, f HIF-1α protein levels were up-regulated after treatment with all three toxins and showed higher levels in male than female astrocytes

In female mesencephalic astrocytes, only HIF-2α was decreased after cobalt (Fig. 6c). HIF-1α protein levels, however, showed toxin-induced increases when compared with protein levels in untreated control cells. In males, all three toxins increased HIF-1α transcript and protein levels (Fig. 6b, f), whereas only azide increased HIF-2α transcription (Fig. 6d).

Toxin-Induced Changes of Intracellular Levels of Reactive Oxygen Species

In addition to COX and HIF transcription, we investigated the effect of inducers of chemical hypoxia on intracellular levels of reactive oxygen species (ROS; Figs. 7, 8). Treatment of cortical and mesencephalic astrocytes with all three toxins yielded elevated ROS levels. Cobalt-treated cells showed a two- to threefold elevation of ROS levels. No obvious gender-specific effects were found. In general, the effects of the three toxins resembled each other with respect to gender and brain region (Figs. 7, 8).

Fig. 7
figure 7

Intracellular peroxide production was elevated after treatment of cortical astrocytes with all three toxins. a, b Intracellular peroxide levels were approximately threefold increased in CoCl2-treated female (a) and male (b) astrocytes. An approximately twofold increase was observed in KCN-treated female cells (a). Significant gender-specific differences (#, indicated in graphs for male samples) are shown. Data were compared to untreated control samples set 100% (dotted lines, * ,# p < 0.05, ## p < 0.01, *** p < 0.001, n = 4)

Fig. 8
figure 8

Intracellular peroxide production was elevated after treatment of mesencephalic astrocytes with all three toxins. a, b Intracellular peroxide levels were increased two- to threefold in CoCl2-treated female (a) and male (b) astrocytes, whereas the other toxins increased the peroxide levels to a lower extent. No significant gender-specific differences were observed. Data were compared to untreated control samples set 100% (dotted lines; * p < 0.05, *** p < 0.001, n = 4)

Discussion

There is emerging evidence that astrocytes play a crucial role for neuronal survival under toxic and degenerative conditions in the CNS (Kipp et al. 2006; Maragakis and Rothstein 2006). Essential functions of astrocytes which are important in this context include the maintenance of the “neuronal milieu” involving energy homeostasis (Barres 1991). Neuronal loss can result from direct action of the toxin on neurons or indirectly involving a disturbance of astroglial function. In the absence of astrocytes, neurons are more prone to the effects of mitochondrial toxins and excitotoxic damage (Dugan et al. 1995). This indicates the importance of astrocytes for neuronal function and survival. Chemical toxins, such as azide, cyanide and cobalt, are widely used as inducers of chemical hypoxia and/or neurodegenerative processes based on all three toxins acting as inhibitors of the mitochondrial respiratory chain (Rose et al. 1998; Zhang et al. 2007; Cooper and Brown 2008). For instance, cyanide acts as a potent inhibitor of mitochondrial oxidative metabolism and produces mitochondria-mediated death of dopaminergic neurons associated with a Parkinson-like syndrome (Zhang et al. 2009). However, the mechanisms of their action beyond that and with respect to different potencies of these toxins occurring in a gender- and brain region-specific manner are unknown. Therefore, we aimed to correlate the different levels of toxicity of azide, cyanide, and cobalt with the transcription levels of cytochrome c oxidase subunit IV (COX IV) isoforms, the production of reactive oxygen species (ROS), and the percentage of apoptotic/necrotic death of female and male astrocytes from cortex and mesencephalon.

Besides low necrosis rates, all three toxins provoked massively apoptosis in astroglia which has been previously described to represent the major death pathway in other cell types (Prabhakaran et al. 2007; Zhang et al. 2007, 2009; Fang et al. 2008). Although this holds true for female and male primary astrocytes from both brain regions, the extent to which apoptosis and necrosis occurred differed with the applied toxin. Depending on the toxin, we observed gender- and brain region-specific differences and a different potency for each toxin. With respect to a higher vulnerability of primary brain cells compared with that of widely used cell lines we have chosen a shorter toxin treatment of six instead of 24 h (Rose et al. 1998; Fukuda et al. 2007; Zhang et al. 2007). At 6 h, we found significantly elevated apoptotic and necrotic cell death rates, but only to an extent that allowed structural and functional studies. Cobalt proved to be the most potent toxin. It has already been described as a mitochondrial toxin inducing chemical hypoxia (Vengellur and LaPres 2004; Fukuda et al. 2007). In comparison with azide and cyanide, cobalt induced maximal necrotic and apoptotic cell death in both genders and brain regions, except for female cortical astrocytes, where cyanide was most powerful in promoting necrosis.

All three toxins act as inducers of chemical hypoxia and function as inhibitors of COX, the terminal enzyme of the respiratory chain. Likewise, oxygen deprivation represents an inhibitor of COX by depriving the enzyme of its catalytic substrate. However, the mechanism of inhibiting an enzyme by depriving its substrate or by adding an inhibitor could be different. Nevertheless, a structurally and functionally intact COX is essential for proper cell function and survival. On the contrary, a dysfunctional COX can cause increased cell death rates based on mainly two mechanisms, one of them represents an impairment of mitochondrial energy metabolism that is essential to provide cells with ATP on their demand. The other mechanism involves the respiratory chain as a producer of ROS, thereby increasing cellular oxidative stress (Kadenbach et al. 2004; Horvat et al. 2006; Singh et al. 2009). Here, we demonstrate that azide, cyanide, and cobalt caused increased intracellular ROS levels being highest after treatment of astrocytes with cobalt, which also provoked the highest percentage of cell death. In addition, increased ROS levels could have been provoked by a toxin-mediated impairment of the antioxidant defense system, such as an inhibition of catalase by azide (Thurman and Chance, 1969), cyanide (Ardelt et al. 1989), and cobalt (Yasukochi et al. 1974).

For a better understanding of the involvement of COX in ROS-mediated cell death, we analyzed the transcription levels of COX subunit IV isoforms. Recently, we demonstrated an important role of COX for both these processes, i.e. energy and ROS production. These two mechanisms represent also two major mechanisms of neurodegeneration (Kadenbach et al. 2004; Horvat et al. 2006; Singh et al. 2009). We investigated the specific role of COX subunit IV isoforms for the regulation of ATP and ROS production and showed a decrease of ATP levels in astrocytes after oxygen deprivation for 6 h (Horvat et al. 2006). This regulatory mechanism involves the expression of a second isoform to COX subunit IV (COX IV-2). The ubiquitously expressed COX IV-1 is 1 out of 10 regulatory subunits of the enzyme complex which overall consists of 13 subunits (Tsukihara et al. 1996). COX IV-1 plays a pivotal role in adapting the mitochondrial energy production to the cellular energy demand. It does so by binding ATP to the matrix site of COX IV-1 at high intracellular energy levels causing an allosteric inhibition of the enzyme. At decreasing energy levels, this inhibition is relieved and the enzyme is maximally activated (Arnold and Kadenbach 1997, 1999; Kadenbach and Arnold 1999). This mechanism of regulating the mitochondrial energy production in mammalian cells is abolished when the COX isoform IV-2 is expressed. The switch of the COX subunit IV isoforms occurs in astrocytes under hypoxic conditions, when COX IV-1 is exchanged to COX IV-2 (Horvat et al. 2006). This was the first evidence for a COX subunit to be regulated in dependency on the oxygen concentration in mammals.

Furthermore, we demonstrated that NPA, a mitochondrial toxin and inhibitor of the respiratory chain complex II (succinate dehydrogenase), which is often used to induce chemical hypoxia (Nishino et al. 1998), influences COX IV isoform expression (Singh et al. 2009). In striatal astrocytes treated with NPA, COX IV-2 transcription is increased, whereas COX IV-1 is decreased. However, under normoxic but toxic conditions, we found that elevated mitochondrial ROS production but not ATP deprivation was the cause for increased necrotic cell death of astrocytes. There is convincing evidence for a causal relationship between increased COX IV-2 isoform expression, increased COX activity, and elevated mitochondrial ROS production as demonstrated by the application of a siRNA system against this COX isoform (Singh et al. 2009). Apparently, toxin-treated cells exhibit a lower survivability when COX IV-2 instead of COX IV-1 is expressed. Interestingly, our present study revealed a similar mechanism, in particular in female astrocytes. After application of cyanide and cobalt to female cortical and mesencephalic astrocytes, the up-regulation of COX IV-2 was accompanied by increased ROS production and necrotic cell death. Thus, these results support our earlier findings, but emphasize a gender-specific aspect. For male cortical and mesencephalic astrocytes, another correlation appears plausible. The ratio of COX IV-1/COX IV-2 exhibited the lowest level after treatment with cobalt. This effect was paralleled by the highest level of ROS production and necrosis. Taking into account that a low COX IV-1/COX IV-2 ratio stands for elevated COX IV-2 in relation to COX IV-1 levels, irrespectively of a down- or up-regulation of either isoform, this ratio represents a valuable marker for increased COX IV-2 expression with the appropriate functional consequences for COX activity and ROS production.

With respect to a higher oxygen concentration when mitochondrial toxins are applied than at conditions of oxygen deprivation, we would suggest a higher ROS production under aerobic toxin-mediated inhibition of mitochondria (Gores et al. 1989). Preliminary results indicate elevated intracellular ROS levels induced by hypoxia although to a lesser extent (data not shown) than the levels observed after application of mitochondrial toxins, such as CoCl2. Furthermore, intracellular ROS levels reflect a balance between ROS-producing and ROS-eliminating pathways. Thus, increased ROS levels could also have been provoked by a toxin-mediated impairment of the antioxidant defense system, such as an inhibition of catalase by azide (Thurman and Chance 1969), cyanide (Ardelt et al. 1989), and cobalt (Yasukochi et al. 1974).

Fukuda et al. (2007) supported our observation of a hypoxia-mediated regulation of COX IV isoform expression (Horvat et al. 2006) and showed that HIF is involved in the reciprocal transcription of COX subunit IV isoforms. Although the mechanism of action of mitochondrial toxins is certainly not mediated by oxygen deprivation, toxins like cyanide, azide, and cobalt also appear to signal via hypoxia response element (HRE) promotor to activate HIF-1α (Zhang et al. 2007). This subsequently stimulates a cascade of gene transcription that might additionally be activated by ROS (Stowe and Camara 2009). We aimed to understand if chemical hypoxia involves a similar regulatory mechanism. Cobalt, which exerts the greatest effect on astrocyte viability, is known to activate hypoxic signaling by stabilizing and increasing HIF-1α expression (Vengellur et al. 2005; Fukuda et al. 2007; Yaung et al. 2008). This, in turn, was shown to cause a reduction of COX IV-1 and an elevation of COX IV-2 (Fukuda et al. 2007). Although cobalt-treated female astrocytes did not show any significant changes in HIF-1α transcription, protein expression was up-regulated by all three toxins irrespectively of the brain region and gender. The discrepancy between transcription and protein expression data is a well-known phenomenon and can be explained by life times of transcriptional products and post-transcriptional regulation of gene expression. Our data support a HIF-mediated transcriptional down-regulation of cox4i1 and up-regulation of cox4i2 for cobalt-treated female mesencephalic astrocytes and a more than twofold up-regulation of cox4i2 in cyanide-treated female cortical astrocytes. As a decreased cox4i1/cox4i2 transcript ratio can be considered as an indirect indicator of such a reciprocal transcription of COX subunit IV isoforms, such an effect can also be suggested for cobalt-treated female cortical/male mesencephalic astrocytes and azide-treated female mesencephalic astrocytes. This indicates a mediated predominantly by cobalt reciprocal transcription of COX IV isoforms as previously observed for hypoxia-treated cortical astrocytes (Horvat et al. 2006) and a different mechanism of transcriptional control in cobalt- versus azide-/cyanide-treated astrocytes. With respect to cobalt resembling most closely the effects of hypoxia on astrocyte viability, COX IV isoform transcription pattern, and ROS production after reoxygenation (data not shown), cobalt is apparently the toxin of choice to mimic hypoxia by application of chemical toxins.

This also suggests a careful consideration and comparison of data obtained after hypoxia induced by oxygen deprivation or application of mitochondrial/cellular toxins. In addition, the observed deviance of our data from those published by others could be due to differences between cell lines and primary cells or due to gender-specific regulatory aspects that were not been explicitly studied in this context.

In conclusion, there is convincing evidence pointing at common pathways of hypoxia and neurodegenerative processes in the CNS with mitochondrial toxins as inducers. This emphasizes the essential role of mitochondrial regulation of astrocytes for neuronal destruction and opens the possibility of mitochondria-based common strategies for prevention/therapy against cell death. The gender-specific regulation of COX IV isoforms, ROS production, and cell death might represent an important mechanism accounting for known differences in the sensitivity towards and severity of distinct neurodegenerative and acute disorders.