C6-glioma cells derived from N-nitrosomethylurea induced rat glioblastoma (purchased from the ATTC-LGC bank) were plated at a density of 2.5 × 10⁴ cells/cm² in 10-cm diameter Petri dishes in DMEM (Gibco-BRL) and were supplemented with penicillin (100 U/ml), streptomycin (0.1 mg/ml), fungizone (0.5 mg/ml), sodium bicarbonate (2 mg/ml), and 5% fetal calf serum (FCS) (Gibco-BRL). Cultures were carried out during 5 DIV in a water-saturated incubator at 37°C with 5% CO₂ in air. The cell medium was changed every 2 days. At the end of the 5 DIV of growth, the cells were confluent (5 × 10⁵ cells/cm²).

The human microglia cell line (CHME-5, given by Professor Tardieu, Kremlin Bicêtre, France) was plated at a density of 1.5 × 10⁴ cells/cm² in 10-cm diameter Petri dishes with MEM (Gibco-BRL) supplemented with penicillin (100 μ/ml), streptomycin (0.1 mg/ml), sodium bicarbonate (2 g/l), glutamine (584 mg/ml), glucose (4 g/l), and FCS (10%). Cultures were performed in a water-saturated incubator at 37°C with 5% CO₂ in air. Microglia reached confluence after 4 DIV.

The C6-microglia co-cultures were performed by adding microglia (80% of confluence), growing either on coverslips or on beads, to C6 monolayer cultures (in C6 cell medium). Forty-eight hours before the co-culture, the monolayer of microglia was dissociated and the cell suspension was plated either on 12-mm diameter coverslips (5–10 × 10⁵ cells) or on collagen beads (3–5 × 10⁶ cells per ml of beads). The microglia medium was replaced by C6 cell medium. This change does not affect microglia growth and phenotype.

Flow cytometry analysis was performed with the fluorescent probes propidium iodide (PI), DIOC6, calcein-AM, Nonyl acridine orange (NAO), and cytodex fluorescent beads. All these products were from Molecular Probes. For all these experiments, microglia and C6 cells were cultured either in isolated conditions or in co-culture for 1, 3, 6, and 24 h. Microglia were cultured either on collagen beads or on coverslips, the C6 cells being cultured as monolayers on Petri dishes.

C6 cells were grown as monolayer in Petri dishes and microglia on collagen beads. For co-culture, the cell populations on their respective supports were mixed and at the indicated times, the beads were discarded and the C6 were scraped off. Oxygen uptake analysis was performed polarographically with a Clark type electrode (final volume 1 ml) with either the microglia on beads or the C6 cells in suspension. Respiration rates were calculated with respect to calibration of the oxygen electrode using air-saturated medium containing 220 μM O₂ at 37°C. The effects of the addition of oligomycin (1.25 μg/ml), CCCP (0.1 mM), and KCN (2 mM) (Sigma) were analyzed. After the respiration tests, the viability of the cells was assessed by their ability to exclude Trypan blue directly from the C6 cell suspension and from the microglial suspension after trypsin digestion of the collagen beads.

C6 cells were grown as monolayer in Petri dishes and microglia were grown either in isolated culture condition or in co-culture. Then cells were lysed with Tris–HCl (50 mM) Triton X-100 (0.3%) at pH 7.5. Citrate synthase activity was monitored spectrophotometrically at 412 nm by recording the kinetics of the oxidation of acetyl-coenzyme A with 5.5′-dithiobis-(2-nitro-benzoate) (DTNB, Sigma). Enzyme activity was calculated using the value 13600 M⁻¹ cm⁻¹ for the DTNB extinction coefficient at 412 nm. Results were expressed as nmol/10⁶ cells/min.

Simultaneously to oxygen uptake measurement, dehydrogenase activity was evaluated using the MTT Formazan (1-(4,5-dimethylthiazol-2-yl)-3,5-diphenylformazan) test (Sigma). Monolayer C6 cells or microglia grown on beads were incubated with MTT (0.5 mg/ml) at 37°C for 45 min. Then, cells were washed with PBS, spun down and the insoluble formazan salt accumulated inside the cells was solubilized with 1 ml of DMSO and quantified spectrophotometrically at 570 nm.

Glucose and lactate concentrations were determined from the cell culture medium of both the C6 monolayer cultures and the microglia on beads in isolated culture condition or in co-culture.

At different times, aliquots of the medium (175 μl) were removed and immediately added to 25 μl of a solution containing 35% (V/V) perchloric acid (PCA). Then, a solution of 2 N KOH/0.3 M MOPS was added until pH 6.5–7 performed neutralization. Enzymatic assays including glucose oxidase and peroxidase for glucose and lactate dehydrogenase for lactate were performed according to the supplier's recommendations (Sigma).

C6 cells and microglia were grown as for glucose and lactate determinations. At different times, cells in monolayer or on beads were collected, rinsed and frozen in liquid nitrogen. Then, proteins were precipitated in 0.9 M HClO₄ containing 50 mM EDTA and eliminated by centrifugation (9000 g, 4 min, 4°C). The supernatant was neutralized with 2 N KOH/0.3 M MOPS and the precipitate eliminated by centrifugation. ATP concentrations were determined with the luciferine–luciferase methods (Lemasters and Hackenbrock, 1979).

NMR measurements were carried out on a Bruker AM400 spectrometer operating at 9.4T with a 5-mm cell probe-head. ³¹P-NMR spectra were acquired at 162 MHz. Cell PCA extracts were prepared as described in Pianet et al. (1991). After removal of the culture medium, cells were rapidly rinsed twice with 0.9% NaCl and the dishes were put into liquid nitrogen. The cells were scraped off in ice-cold 0.9 M HClO₄ and the resulting suspension was centrifuged. The supernatant was neutralized by addition of 9 M KOH. After centrifugation, the KClO₄ precipitate was discarded and the extract filtered through a Chelex 100 column (H⁺ form). The pH of the solution was then adjusted to 7.4 with Tris-base before freeze-drying. The lyophilized powder was dissolved in 50 μl (100 mM) EDTA pH 7.4 and 500 μl D₂O. Each PCA extract spectrum was acquired overnight.

The acquisition parameters were as follows: 12 μs pulse; 0.819 s acquisition time; 0.5 s relaxation delay. Spectra were acquired with proton decoupling. Chemical shifts were expressed as ppm relative to the resonance at 17.4 ppm of the methylene diphosphonic acid (MPDA, Sigma) at pH 7.4. To evaluate the phosphorylated metabolite contents of the cells from spectra acquired under conditions of partial saturation and proton decoupling, we determined a correction factor calculated from spectra of samples containing known amounts of standard of each metabolite and MPDA submitted to the same treatment as PCA extracts (Merle et al., 1992).

Data are represented as the mean ± standard deviation, resulting of at least three independent experiments. Data were analyzed by Student's t test or ANOVA. The level of significance was set at p < 0.05.

C6 cells/microglia co-culture were prepared by growing microglia on beads that were subsequently added on top of C6 cell monolayers. This allowed separating the C6 cells and microglia at the end of the experiment. First of all, the ability of microglia to leave their support was checked by using DIO labeling and flow cytometry. Results indicated that microglia were slightly mobile and could leave the beads after 3 h of co-cultures (data not shown). However, the population did not exceed 3% at this time. After 6 and 24 h of co-culture, the proportion of microglia on the C6 monolayer did not exceed 6%.

Cell membrane integrity and microglial viability were analyzed during the first 24 h of co-culture by flow cytometry (Figure 1). Calcein methyl ester was used to check membrane integrity. This non-fluorescent molecule is membrane-permeant. Once inside the cells, calcein-AM is hydrolyzed by endogenous esterase into the highly green fluorescent calcein, which is retained in the cytoplasm of living cells. Figure 1A shows fluorescence of microglia in isolated culture conditions and various co-culture times. Results indicate a slight but non-significant decrease from 3 to 24 h of co-culture. Moreover, Figure 1B shows that propidium iodide was not significantly taken up by the microglia either in isolated culture or in co-culture. Taken together, these results indicate that the proximity of the C6 glioma did not seem to affect the viability of the microglia. The same results were observed on C6 glioma cells in co-culture (data not shown).

Firstly, ED1 labeling shows that microglia in co-culture changed their morphology, being more retracted and ovoid. ED1 immunohistochemistry was performed to detect Cluster of Differentiation 68 (CD68). This protein is found on activated macrophages and its expression is correlated with phagocytic activity (Damoiseaux et al., 1994). Microglia expressed the ED1 antigen both in isolated (data not shown) and in co-culture (Figure 2) conditions. The ED1 labeling of microglia grown during 48 h and then shifted or not on Petri dishes without C6 cells gave at the t = 0 time point a similar labeling than that at t = 0 h when in co-culture. Labeling was located in the cytoplasm as well as in the processes. However, it appeared more intense in ovoid microglia in co-culture. This observation was mainly due to an increase in ED1 concentration, a consequence of the decrease in the cell volume in microglia. This ED1 labeling indicates that microglia were in an activated state with a potential phagocytic activity.

This phagocytic activity was monitored using fluorescent carboxylate beads and flow cytometry. Results are presented in Figure 3 and show at least two major populations with different uptake capabilities in view of the heterogeneous population observed under microscopy (with regard to the cellular morphology, Figure 2). Microglia cultured in isolated condition, either on Petri dishes or on collagen beads, show similar pattern to that at t = 0 h. Both in isolated culture conditions and in co-culture at t = 0 (Figures 3A,D), 45% of the microglia appeared to be in a low phagocytic state (M1) and 55% of the cells had a high capacity for phagocytosis (M2) as shown by an average fluorescence 300 times higher than that found in the M1 population. During the first 3 h of co-culture, the low phagocytic population (M1) decreased whereas the highest one increased (M2) (Figures 3B,D). However, after 6 h of co-culture, this shift reversed, showing at 24 h a large increase in the M1 population and a drastic decrease in the M2 phagocytic one (Figures 3C,D). The uptake capacity of microglia can be evaluated by considering the relation “percentage of M2 cells” × “fluorescence average” (Figure 3E). The M2 population with the highest fluorescence increased by a factor 3 between 0 and 3 h. After 24 h, the microglia uptake capacities were poorly efficient due to the decrease in both the population concerned and the quantity of beads where uptake was detected. The contribution of the M1 population remained low and insignificant.

So, in the C6 cell vicinity, after a transitory phagocytic activated phenotype, microglia reversed to a non-phagocytic one, although the ED1 labeling and the morphological changes were maintained.

In order to evaluate if changes in the cytoskeleton were involved in microglial activation, F-actin, tubulin and vimentin were stained by immunocytochemistry.

Microglia in isolated culture conditions were mostly flat cells. F-actin appeared around the stress fibers, the border of the cytoplasm membrane and the external membrane outgrowths (Figure 4A). During co-culture and particularly after 3 h, F-actin density increased, with labeling localized at the plasma membrane amassed in patches corresponding to the external outgrowths looking like kimono sleeves (Figure 4B). This F-actin distribution did not change between 3 and 24 h (data not shown).

Tubulin (Figures 4C,D) and vimentin (data not shown) in the microglia appeared as networks of continuous fibers spreading homogeneously throughout the cytoplasm. After 3 h of co-cultures, tubulin and vimentin labeling changed, the networks becoming unstructured and more diffuse. After 24 h of co-culture, the network recovered its fibrous perinuclear frame similar to that observed at t = 0 (data not shown).

Expression of Hsc70 and Hsp60 was assessed by immunocytochemistry. These two proteins are constitutively expressed by microglia at the cytoplasmic and mitochondrial levels, respectively. Hsp60 is present at the inner membrane level of mitochondria. The goal was to assess the possible induction of a cellular stress by C6 cells, indicated by observing a modification of their mitochondrial location.

Hsc70 and Hsp60 labeling of microglia grown in isolated culture conditions gave similar pattern to those shown in co-culture at time point 0 (Figures 5A,C, respectively).

In isolated culture conditions, microglia highly expressed the Hsc70 proteins in the cytoplasm (Figure 5A). External extensions resembling leaves and buds were also labeled. After 3 h (Figure 5B) and during 24 h of co-culture (longer times were also tested), neither the expression level of the Hsc70 protein nor its location were significantly modified. Moreover, the inducible Hsp70 proteins were undetectable (data not shown). Therefore, the transitory activation of the microglia in co-culture did not appear associated with a stress situation, as assessed by the expression of the Hsp70 protein family. Labeling of Hsp60 protein in isolated microglia showed a homogeneous distribution of the mitochondria both inside the cell body and throughout the processes with a typical dot line organization (data not shown). When microglia were cultured in the presence of C6 cells, the dot labeling of Hsp60 seemed clustered around the nucleus after 1 h. This phenomenon was greater after 3 h (Figure 5D) and still visible after 24 h of co-culture.

Comparing results obtained by immunocytochemistry on cytoskeleton modification and on Hsp60 location (a mitochondrial protein) we can conclude that the observed mitochondrial relocation does not reverse unlike the microtubule railways.

Endogenous O₂ consumption of the microglia in co-culture (Figure 6A) was monitored to investigate the potential involvement of mitochondria relocation on respiratory properties. This value strongly decreased after 3 h of co-culture to reach its lowest level (50% of the consumption measured at t = 0). Then it recovered to reach 65% of the basal level after 24 h of co-culture. The uncoupled maximal microglia respiration in the presence of CCCP was also affected. The ratio endo/CCCP, which is representative of the O₂ consumption yield (Martin et al., 1998; de Gannes et al., 2000), decreased during the first 3 h of co-culture but represented 80% of the normal yield after 24 h of co-culture.

C6 respiration was also affected during the first 3 h (Figure 6B). However, the endogenous respiration remained at 50% of the normal level, while the CCCP O₂ consumption had recovered at t = 6 h. Consequently, the endo/CCCP yield remained at 50% of the control capacity.

Changes in mitochondrial enzyme activities were assessed in order to monitor transient respiratory modifications. The transformation of the tetrazolium salt (MTT) in the purple insoluble formazan is partly due to mitochondrial dehydrogenase activities (Liu et al., 1997). MTT reduction and citrate synthase activity in the microglia showed a pattern similar to endogenous respiration (Figure 6C) although they were able to restore their activities at 24 h or even presented a higher activity at the end, which was not observed for the respiration. In C6 cell culture, MTT reduction was not affected by co-culture (Figure 6D). However, citrate synthase followed a similar pattern to that of endogenous respiration, reaching its lowest level after 3 h of co-culture (33% of the control value) with a slight recovery after 24 h of co-culture (60% of the control value).

The fluorochrome NAO binds with a high affinity to the mitochondrial cardio-lipins (Ratinaud et al., 1988; Lizard et al., 1990). Consequently, the binding of NAO appears independent of respiratory status and reflects the mitochondrial mass (Maftah et al., 1989). Cell staining with NAO confirmed the perinuclear location of the microglial mitochondria in co-culture (data not shown). Furthermore, flow cytometry analysis of the co-cultured microglia showed a unique homogeneous population with a 2.5-fold fluorescence increase as soon as 1 h of co-culture. This high level of fluorescence was maintained during the 24 h-time course of the co-culture (Figure 7A).

The same experiment was conducted on C6 cells. NAO binding also showed a homogeneous population with a similar increase in fluorescence after 3 h of co-culture (Figure 7A).

Fluorochrome DIOC6 is a cationic probe whose intracellular accumulation is dependent on the potential gradient between the inner and outer cell compartments (Nicholls and Ward, 2000). Its accumulation does not appear to be specific to the mitochondrial compartment. However, at the used DIOC6 concentration, the fluorescence labeling of the cells demonstrated a major contribution of the mitochondria versus the reticulum, as described by Rottenburg and Wu (1998). After incubation of microglia with DIOC6, fluorescence microscopy revealed fluorescent dots inside the cells, similar to those observed after Hsp60 and NAO labeling (data not shown). When microglia were also analyzed by flow cytometry, the results showed a homogeneous fluorescent microglial population. The fluorescence level decreased as soon as 1 h of co-culture (data not shown), reaching 70% of the control level and evolving toward the control value after 24 h of co-culture (Figure 7B). The DIOC6 accumulation in the C6 cells did not change after 1 h of co-culture (data not shown) but decreased after 3 h of co-culture (60% of the control value) (Figure 7B). The fluorescence level did not change thereafter.

Unlike C6 cells, microglia recovered their mitochondrial respiratory properties. Therefore, the question rises if energetic metabolism of microglia and C6 cells was altered.

During the first 24 h of co-culture, the microglia and the C6 cell densities did not change significantly, being similar to the densities observed in isolated culture conditions. In addition, the proportions of cell populations were also unchanged: 73% of C6 cells and 27% of microglia.

In isolated culture conditions, the microglia and the C6 cells consumed glucose and released lactate (Figures 8A,B). Initial rates of glucose consumption were similar to those of lactate release for each cell type and under similar conditions (Table 1). C6 cells showed initial rates four times higher than those of the microglia, so glucose in the C6 medium was totally depleted after 24 h of culture. The glucose consumption and lactate release profiles in the co-cultures resembled those of the isolated C6 cultures. These results revealed the large metabolic contribution of the C6 cells (73%) and the fact that the proximity of the microglia had apparently no effect on the C6-cell glucose and lactate metabolisms. Moreover, the contribution of C6 cells to the metabolism of the microglia (27%) was not detectable.

For both cell types, ATP contents decreased during the first hours of co-culture, reaching a plateau between 6 and 10 h (Figure 8C). Whether we quantified ATP by luminometry (Figure 8C) or by ³¹ P-NMR spectroscopy (Figure 9, Table 2), the microglial ATP levels appeared more affected than those of C6 cells. Indeed, after 24 h of co-culture, the microglial ATP level was only 30% of the control value, whereas the C6-cell ATP level was 45% of the control value. NMR spectroscopy analyses showed a 60% decrease in the ATP/ADP ratio after 3 h of co-culture, which possibly reflects a new steady state when both cells are cultured together.

Analysis of the other resonances (Figure 9) clearly identified variations in phosphocreatine (PCr), inorganic phosphate (Pi), phosphorylcholine (PC), and glycerophosphorylcholine (GPC). The glycerophosphorylethanolamine (GPE) content was low in both cell types and the variations observed during the co-cultures were not significant (data not shown). The microglial and C6-cell PCr levels decreased in a similar proportion during the first 24 h of co-culture. In parallel, the Pi levels increased, with a higher increase for microglia in co-culture. Both PC and GPC levels decreased (Figures 9B,D). After 3 h and during the 24 h of co-culture, PC levels represented 75% and 50% of the control values for microglia and C6 cells, respectively. However, at 24 h, the ratio PC/GPC, which reflects membrane turnover, had increased (3.7 and 1.8 times for microglia and C6 cells, respectively).

In the C6 environment, the microglia changed their morphology. Their constant high ability to hydrolyze calcein-AM, to retain calcein and to present a low uptake of propidium iodide suggests that the viability of these two cell types was not affected. Microglia did not develop apoptotic processes, unlike the case when macrophages are overstimulated and involved in inflammatory reactions (Smith et al., 1998). Furthermore, the activated microglia did not induce the death of C6 cells. When co-cultures were kept for several days in vitro, the microglia colonized the C6 area without developing macrophagic behavior (data not shown).

This cellular co-existence raises the question whether short transient changes occur in the microglia phenotype when they grow in a tumoral environment and if correlations with energy metabolism can be made.

During co-culture, the cell numbers of each population did not change significantly. This lack of increase is in agreement with the average 35-h half-life of these cell types in isolated culture conditions (data not shown). Moreover, the protein content of both cell types was not modified.

During the short-term of co-cultures, an increasing part of the microglia population is connected to a positive ED1 labeling and a higher ability of phagocytosis. After 3 h of co-culture, cells appeared engaged in a macrophagic state. Nevertheless, this activated microglia state seems to be more subtle since the uptake properties were silenced after 24 h of co-culture even if the ED1 expression was still maintained. This suggests another activated state, intermediate, but not macrophagic. Microglia reached this state after 3 h of co-culture. In the same way, the location of F-actin at the level of the cytoplasmic membrane, as well as the external roughness resembling leaves and buds, reflected also an activated microglial status (Bauer et al., 1994; Damoiseaux et al., 1994; Taupenot et al., 1996). These changes in F-actin distribution are consistent with F-actin fragmentation in activated macrophages (Badaut et al., 2008). Such fragmentation was shown to alter phagosome movements toward the juxtanuclear region. However, it was not caused only by phagocytic uptake and/or the presence of phagosomes. Similarly, in the C6 cell vicinity, the microglia F-actin distribution that we observed did not seem clearly related to the properties of the phagocytes.

Microglia are also known to be sensitive to stress and to give a characteristic response by the over-expression of the constitutive Hsc70 proteins, and the induction of Hsp70 forms located first in the cytoplasm and then in the nucleus (Voisin et al., 1996; de Gannes et al., 2000).

During the 24 h of co-culture and for longer times, neither the expression level of the Hsc70 proteins nor their locations were significantly modified (Figure 5), indicating that the C6 cell environment did not seem to generate any microglial stress, as assessed by unchanged expression of hsp70 protein family. Indeed, the inducible Hsp70 proteins were undetectable. In neurological inflammatory settings, Hsp70 induction of monocytes/macrophages could be involved in the immune response to pathogens and contribute to the processing and presentation of antigens (DeNagel and Pierce, 1992; Bachelet et al., 1998). Moreover, monocytes/macrophages enhance their antioxidant defenses by generating Hsp70, which interferes with the release of cytochrome c from the mitochondria (Kagaya et al., 2006) and thereby prevents them from undergoing apoptosis subsequent to oxidative stress. In addition, the lack of C6-induced oxidative stress could be related to the relocation of F-actin, a process that could prevent the induction of iNO-synthase if linked to the fragmentation of the stress fibers (Fernandes et al., 1996; Higuchi et al., 1996). Consequently, the iNO-synthase activity could be inhibited in activated macrophages. Such a property helps to avoid the triggering of apoptosis. This situation can be linked to the high esterase activities and the lack of propidium iodide permeability that we observed, suggesting that the microglia did not enter the apoptotic pathway despite the ATP readjustment. On the other hand, Hsc70 accumulation in the external membrane excrescences suggests a contribution of the constitutive heat shock protein in the regulation of endocytosis, since Hsc70 is involved in the inhibition of the assembly of clathrin-coated vesicles (in both the clathrin-dependent and clathrin-independent endocytosis pathway) (Chappell et al., 1986). This external Hsc70 location appeared concomitantly to the loss of bead uptake and to the fragmentation of F-actin.

During the short-term co-cultures, the vimentin and tubulin networks were altered, although these changes were very transitory. In the microglia, the mitochondria were previously described as organites which co-localized with either vimentin or tubulin (de Gannes et al., 2000). In our study, simultaneously with the cytoskeleton readjustments, the mitochondria clustered in a perinuclear location. However, this situation remained even though the vimentin and tubulin networks recovered their normal immunological aspect.

These short-term modifications demonstrate the sensitivity of microglia to a foreign cell environment. However, the microglia remained activated but with silenced macrophage properties. In the C6 glioma environment, the bead uptake and the intermediate filament network alteration were quick and transitory. The lack of Hsp70 expression, the Hsc70 and the F-actin locations, like the absence of any change in the microglia viability, could be closely associated with the down-regulated immunophenotype environment generated by the glioma cells, suggesting a symbiosis between tumor cells and macrophages (Badie and Schartner, 2001). Since most of the microglia on collagen beads did not have any direct contact with the C6 monolayer during the co-culture time course, the observed phenotypic adaptation was probably more dependent on the release of diffusible molecules which could be cytokines or growth factors that are now well-known in immunosuppressive tumor biology (Giulian et al., 1994a,b; Hussain and Heimberger, 2005).

Nevertheless, the location of the mitochondria, which were still perinuclear after 24 h of co-culture, raises the question of a putative microglial metabolic readjustment in a foreign cellular environment.

The microglial endogenous and uncoupled O₂ consumptions between 1 and 3 h were affected. The ratio of these rates showed a decrease in the respiration yield associated with a decrease in dehydrogenase and citrate synthase activities. However, cell viability, as reflected by the esterase activity and the lack of PI permeability, was not affected. Since citrate synthase is often considered as an indicator of mitochondrial content (Walenta et al., 1989; Bredel-Geissler et al., 1992), the decrease in respiration could be due to a loss of mitochondrial content, although its expression and its activity are also subject to cell regulation (Martin et al., 1998). Nevertheless, experiments performed with NAO demonstrated the absence of mitochondrial loss. More surprising was the increase in NAO fluorescence after 3 h of co-culture. This anomaly was of too short a duration to suggest an increase in the mitochondrial population. The rise in the mean fluorescence level thus suggests the exposure of new cardio-lipin binding sites linked to an extension of the inner mitochondrial membrane surface or to the swelling of these organites. These changes in the mitochondrial aspects were concomitant with the clustering of these organites, which were visible after Hsp60 immunolabeling.

The decrease in mitochondrial potential through the alteration of the DIOC6 fluorescence could be due to the decrease in O₂ consumption during the 1–3 h of co-culture. This observation shows the inability of the respiratory chain to maintain the proton gradient between the external and the internal faces of the inner membrane. This could be the consequence of the decrease in electron donors at the different entries of the respiratory chain, which could be due to the loss of dehydrogenase activities and/or the loss of membrane integrity leading to proton leakage. This latter possibility, which can be evidenced by the decrease in O₂ consumption in the presence of the ATPsynthase inhibitor oligomycin, was not significant enough to validate this hypothesis (data not shown). However, through DIOC6 observations, the recovery of the membrane potential, like the citrate synthase and dehydrogenase activities, also appeared to prevent apoptosis despite ATP depletion (Richter et al., 1996). This is different from the apoptosis pathway which occurs after LPS induction of microglia in which NO-synthase is activated. The increase in NO generates membrane damages, a loss of glutathione and a decrease in the activity of the mitochondrial electron transport associated with a depletion of the cellular ATP content (Moss and Bates, 2001).

C6 cells are known to have high metabolic plasticity. Their oxidative phosphorylation capacities are dependent on their age. During the growth phase, C6 respiration decreases and its metabolism evolves toward a glycolytic status (Martin et al., 1998). To minimize the contribution of the glioma cell growth state, experiments were always conducted with C6 cells at three days in vitro. At this stage, the endogenous and the uncoupled O₂ consumption represented 30% and 90%, respectively, of the maximal corresponding O₂ consumption observed when C6 cells were in the lag phase (15 natom/min/10⁶ cells) (Pasdois et al., 2003). Moreover, all the results during the 24 h of co-culture were compared with isolated C6 cultures of the same age in vitro.

C6 respiration was also affected during the first co-culture step, although the mitochondrial content was not affected. Nevertheless, contrary to what was observed with microglia, the specific activities of the dehydrogenases were not modified. Moreover, the mitochondrial potential did not recover after the first 24 h of co-culture like the endogenous respiration, suggesting a defect in supplying reduced substrates at the points of entry to the electron transport chain, a phenomenon possibly due to the loss of citrate synthase activity.

The C6 cells used up glucose from the medium during the first 24 h of isolated culture, as described by other authors (Passonneau et al., 1978; Pianet et al., 1991; Bouzier et al., 1998a,b). Astrocytes in vitro were shown to take up glucose from the medium at a similar rate, converting it mainly into lactate (Pellerin and Magistretti, 1994). Microglia also consumed glucose but the rate was four times lower. In isolated C6 and microglia cultures, as in co-culture, the lactate release rates appeared similar to the glucose disappearance rates. However, we should keep in mind that one molecule of glucose gives rise to two molecules of lactate. Therefore, these results suggest that, whatever the culture type, microglia and C6 cells metabolize glucose not only through glycolysis pathway but also through other pathways needed for their energetic requirements. Previous studies (Bouzier et al., 1998a,b) demonstrated that lactate consumption is also performed by these cells after conversion into pyruvate, which is located at an important metabolic node. Indeed, C6 cells without glucose are able to maintain their energetic status by using pyruvate and glutamine as carbon donor (Pianet et al., 1991; Portais et al., 1996).

In co-cultures, glucose utilization and lactate release rates were not significantly different from those of the C6 cells in isolated culture conditions. Since 76% of the cells in co-culture were C6 cells, we cannot rule out a possible alteration of microglial metabolism, a phenomenon which could be hidden by the high C6 cell metabolism. Nevertheless, the microglial environment did not induce any significant readjustments of the C6 cell glycolytic pathway.

Regarding ATP measurements, results obtained with the luciferase method were consistent with ³¹P-NMR evaluation of phosphorylated metabolites. The ATP content of the C6 cells and microglia did not change significantly during the 24 h of isolated cultures. The evolution of ATP in the C6 cells is in agreement with the results of Pianet et al. (1991) despite the glucose depletion in the medium. In co-culture, the two cell types showed large ATP and ATP/ADP decreases that were still present after 24 h, suggesting an energetic readjustment to a lower ATP steady state. The changes in this metabolite content can be associated with the changes in respiratory properties, which drastically decreased during the first hours of co-culture in both cell types, and their stagnancy in the C6 cells. However, the maintenance of a decreased ATP level in the microglia did not appear to be dependent only on the endogenous and uncoupled respiration since they recovered. The possible energetic needs of the phagocytic processes remains an open question, since after 24 h of co-culture the different profiles of the fluorescent bead uptake were lower than the control culture ones. Furthermore, for both cell types, the PCr content decreased and the inorganic phosphate content increased during the first 24 h. These observations strongly suggest the establishment of a new metabolic steady state of these cells when in close vicinity with each other. In the case of the microglia, this metabolic behavior was not related to a stress situation. Indeed, in isolated conditions, microglia display a high PCr content compared to monocytes (de Gannes et al., 1998) in which the ATP and the PCr contents, like the respiratory properties, decrease after heat stress but recover after 24 h to a normal energetic status.

Like PC and phosphatidylethanolamine, phosphomonoesters (PME) are substrates of the first steps of phospholipid synthesis, which, with cholesterol, contribute to membrane synthesis. Phosphodiesters (PDE) like GPC and GPE are catabolic products of phospholipids. Increases in PME and in the PME/PDE ratio are indicators of a change in phospholipid turnover, the anabolism being faster than the catabolism. In numerous brain tumors, the PME and the PDE contents change depending on the tumoral type and grade (Heindel et al., 1988; Rutter et al., 1995; Lehnhardt et al., 2001; Tugnoli et al., 2001; Maintz et al., 2002). In meningiomas and glioblastomas, elevated PME and reduced PDE contents are also observed. This increase in the PME/PDE ratio has already been described in dividing cells in the neonatal brain (Younkin et al., 1984) and in the neoblastoma cell line (Pettegrew et al., 1987). Moreover, the PC increase seems linked to the early response of tumor cells to growth factors (Ruiz-Cabello and Cohen, 1992). Furthermore, the phospholipid and cholesterol metabolisms appear important since they contribute to resistance to necrosis (Bacurau et al., 1999).

In co-culture, both cell types decreased their PC and GPC contents and showed adaptive behavior at the level of the cellular membranes after 3 h of culture. This is reminiscent of the modification of NAO binding at the level of mitochondrial membranes. However, the GPC content decreased more than the PC one in both cell types and the resulting PC/GPC strongly increased, suggesting that the phospholipid anabolism status was maintained. This phenomenon seemed more pronounced in the microglia than in the C6 cells. These phospholipid changes reveal another form of metabolic adaptation which cannot be assimilated to a stress situation in which PC and GPC are known to increase (de Gannes et al., 1998).

In summary, this work provides new evidence that tumor cells exert a control on the microglial environment and induce metabolic consequences. The parameters monitored describe an original state of the microglia sharing some of the common features of activated macrophages, whereas other properties such as phagocytosis are silenced. Moreover, simultaneously metabolic and respiratory changes are observed displaying an incomplete recovery of the respiration and a potential new energetic steady state.

The relations between these transitory properties of phagocytosis, the phenotypic, and metabolic alterations that occur in parallel need to be better understood and correlated to the release of diffusible molecules like cytokines and growth factors. This would throw fresh light on the inflammatory status of microglial cells when in the vicinity of tumoral cells. The induction and control of this status could help to limit the immunity escape generated by tumor cells and could restore microglial properties to generate a pro-inflammatory environment in glioma. It could also be interesting to modify microglia metabolic steady state by drugs to see potential links and changes in immune properties.