Low oxygen alters mitochondrial function and response to oxidative stress in human neural progenitor cells

Differentiation of human embryonic stem cells into NPCs

To generate NPCs, human embryonic stem cells BR-1 (Fraga et al., 2011) (kindly provided by Prof. Lygia Pereira, São Paulo University - USP) were grown on polystyrene plates (TPP, Switzerland) covered with Matrigel (BD Biosciences, Franklin Lakes, NJ, USA) in StemPro medium containing 8 ng/mL basic fibroblast growth factor and 0.1 mM β-mercaptoethanol (all from Thermo Fischer Scientific, Waltham, MA, USA). After propagation, the cells were differentiated as neural cells using inhibitors of bone morphogenetic protein (Noggin; R&D Systems, USA), and transforming growth factor-beta (SB431542; Tocris Bioscience, Bristol, UK) (Chambers et al., 2009). At this moment, cells showed morphology and expression of markers consistent to those of neural progenitor cells Figs. S2A and S2B.

Cultivation of NPCs in a physiological environment

NPCs were grown in an environment containing 3% oxygen (physioxia) in an oxygen control chamber (ProOx model C21; BioSpherix, Parish, NY, USA). This equipment was kept at 37 °C and 5% CO₂, and 3% pO₂ was established by a N₂-controlled injection and monitored by an external probe (Mettler Toledo, Colombus, OH, USA).

Growth kinetics

Cellular growth under different oxygen conditions was evaluated for 18 days (3 passages of 6 days). A total of 6 × 10⁵ NPCs/mL were plated into 24-well tissue culture plates. Each day, cells from two wells of each condition were detached with Accutase (Millipore, Darmstadt, Germany) and counted in a Neubauer chamber.

Glucose and lactate measurements

Glucose and lactate concentrations were determined using a YSI-2700 biochemistry analyzer (Yellow Springs Instruments, Yellow Springs, OH, USA). This measurement is based on quantification of hydrogen peroxide generated upon reaction of these organic molecules catalyzed by glucose or lactate oxidases immobilized on membranes.

Immunostaining assays

NPCs were seeded in 96-multiwell µClear dishes (Greiner, Austria) covered with 2.5 µg/mL laminin (Sigma-Aldrich, USA). After 6 days, these cells were fixed with 4% paraformaldehyde (Sigma-Aldrich, USA) in phosphate-buffered saline for 30 min.

Then, the cells were treated with 0.5% Triton X-100 (Sigma-Aldrich, St. Louis, MO, USA), blocked with 5% bovine serum albumin (Sigma-Aldrich, St. Louis, MO, USA), incubated with the following primary antibodies: rabbit anti-human-histone H2A (H2A.X) (1:100; Cell Signaling, USA), mouse anti-Nestin (1:100; Chemicon, Temecula, CA, USA), mouse anti-PSA-NCAM (1:100; Millipore, Germany) and rabbit anti-Tbr2 (1:100; Millipore, Darmstadt, Germany). Subsequently, samples were incubated with the following secondary antibodies: goat anti-rabbit AlexaFluor 488 IgG (1:400; Thermo Fischer Scientific, Waltham, MA, USA) and goat anti-mouse Alexa Fluor 594 IgG (1:400; Thermo Fischer Scientific, Waltham, MA, USA). Nuclei were stained with 0.5 µg/mL 4′-6-diamino-2-phenylindole (DAPI).

Regions of interest were visualized and identified, and the immunofluorescence emission of the cells was quantified using an Operetta high content analysis system and Harmony software (PerkinElmer, Waltham, MA, USA). In these experiments, three technical replicates of each biological replicate (N) were performed.

Twenty-five fields per well were chosen randomly. An average of 3,125 nuclei per well were analyzed. The high-content and high-throughput screening platforms used herein allow the evaluation of different samples in a large scale, with simultaneous identification of several parameters using automated fluorescence microscopy, using specific markers for different proteins (Li, 2014).

Mitochondrial content and ΔΨ_M quantification assays

Measurement of the mitochondrial mass of NPCs was performed using 0.3 µM Mitotracker DeepRed FM (Thermo Fischer Scientific, Waltham, MA, USA), a dye that integrates into active mitochondria (568-nm excitation and 675-nm emission). The ΔΨ_M was estimated by cationic staining with 1.6 µM JC-1 (Thermo Fischer Scientific, Waltham, MA, USA) (488-nm excitation). This dye exists as a monomer at low concentrations, with fluorescence emission at 525 nm (shown here in green). As it accumulates in the mitochondria, which is membrane potential-dependent, the dye forms aggregates that exhibit a maximum emission at 590 nm (shown here in yellow). The ratio of aggregate to monomer concentration can be used as a measurement of ΔΨ_M (Reers, Smith & Chen, 1991).

MitoTracker and JC-1 dyes, diluted in Dulbecco’s modified Eagle’s medium/F12 (Thermo Fischer Scientific, Waltham, MA, USA), were applied to NPCs for 40 min at 37 °C. Fluorescence emission readings were performed in a controlled 5% CO₂ and 37 °C environment. Hoechst 33342 (1 µM, Thermo Fischer Scientific, Waltham, MA, USA) was used for nuclear staining.

Thirty-three fields per well were captured randomly. An average of 825 cells were analyzed per well.

ROS measurement assay

Quantification of superoxides was performed using 10 µM dihydroethidium (DHE; Thermo Fischer Scientific, Waltham, MA, USA). This dye, when oxidized in the cytosol, intercalates with DNA and emits fluorescence at 605 nm. DHE was applied similarly to MitoTracker and JC-1. As a positive control, NPCs were induced to produce ROS by a 40 min pretreatment with 3.6 µM antimycin A, a mitochondrial complex III inhibitor that stabilizes semi-quinone radicals and favors the escape of electrons to oxygen, thus forming superoxide anions.

Twenty-five fields per well were captured randomly. An average of 6,800 nuclei were analyzed per well.

Oxygen consumption measurement

Oxygen consumption was measured by high-resolution respirometry using an Oroboros O2k Oxygraph at 37 °C. DataLab software (Oroboros Instruments, Innsbruck, Austria) was used for data acquisition and analysis. NPCs were enzymatically detached from the plate, diluted in culture medium, and seeded to the Oroboros at a concentration of 1 × 10⁶ cells/mL. The routine oxygen consumption of cells, measured before the addition of modulators of mitochondrial function, was determined after stabilization of the steady state of oxygen consumption for 10–15 min. Subsequently, ATP synthesis was inhibited with 2 µg/mL oligomycin. Oxygen consumption related to oxidative phosphorylation coupled to ATP synthesis was determined by the difference between routine respiration and oligomycin-insensitive respiration.

To uncouple oxidative phosphorylation, the protonophore carbonyl cyanide p-trifluoromethoxyphenylhydrazone was titrated. The resulting maximum oxygen consumption flux value was established as the maximum capacity of the mitochondrial electron transport system. Finally, the nonoxidative phosphorylation oxygen flux was determined by blocking the electron transporting system with 1 µg/mL antimycin A. The residual oxygen flux represents oxidases in the cell sample.

Antioxidant enzymatic activity assay

The activities of the antioxidant enzymes superoxide dismutase (SOD) and glutathione peroxidase (GPx) were measured using commercial kits (cat. 19160 and CGP1, respectively; Sigma-Aldrich, St. Louis, MO, USA). Briefly, the SOD colorimetric assay (440 nm) determines the presence of superoxide radicals in a tetrazolium-coupled reaction. The GPx assay measures NADPH depletion (340 nm). The SOD and GPx enzymatic activities were calculated according to the kit instructions.

To induce ROS production and a possible increase in antioxidant enzymatic activity, the cells were treated with antimycin A, as described previously. Antimycin A showed no effect on apoptosis during the time of treatment (Fig. S1).

Statistical analysis

The unpaired t test was used to compare average differences between two groups. In the case of multiple variable comparisons, one-way analysis of variance was used with the Bonferroni post-test. The null hypothesis of equality between averages was refuted if p < 0.05 (*), p < 0.01 (**), or p < 0.001 (***). Means and standard errors of the mean were plotted.

Growth and glycolytic metabolism of human NPCs are not altered in physioxia

The growth kinetics (Fig. 1) as well as glucose uptake and lactate production (Fig. 2) of NPCs in physioxia (3%, v/v) and normoxia (21%, v/v) were similar, indicating that impaired cell growth or a metabolic shift from oxidative respiration to glycolysis (Warburg Effect: Warburg, 1956) does not occur in the normoxia-physioxia transition.

Mitochondrial content and ΔΨ_M are altered in NPCs grown in physioxia

Mitochondrial labeling was performed using the MitoTracker probe, which is internalized by active mitochondria. Fluorescence intensity measurements indicated less mitochondrial content in physioxia-grown NPCs (Figs. 3A and 3B). As a first approach to evaluate mitochondrial function, we quantified ΔΨ_M as an indicator of the proton motive force. An increased difference of ΔΨ_M can be directly related to a strict commitment to ATP formation by FoF₁ATP-synthase and/or to decreased proton leakage to the mitochondrial matrix (Jastroch et al., 2010). The data showed increased ΔΨ_M in physioxia-grown NPCs (Figs. 3C and 3D).

Physioxia-grown NPCs feature an increased mitochondrial respiration capacity

Our data suggest that changes in the mitochondrial physiology are induced by O₂ availability (Fig. 3). A possible outcome of improved ΔΨ_M during mitochondrial function could be a more efficient oxygen consumption directed to ATP production. To verify the full potential of the mitochondria to destine environmental oxygen to oxidative respiration efficiently, we tested the O₂ consumption in NPCs grown previously either in normoxia or physioxia in conditions in which plenty of O₂ was available, i.e., atmospheric concentrations, by high resolution respirometry. Our data show higher oxygen uptake rates for NPCs previously cultured at 3% O₂ (v/v). The oxygen flux coupled to ATP production in NPCs cultured at 3% O₂ (v/v) was approximately 70% higher than that of cells cultured at 20% O₂ (v/v) (Fig. 4). Additionally, an increased maximum respiratory capacity of physioxia-grown NPCs was observed, compared to those cultured in normoxia. This observation corroborates our previous results, showing that the NPCs did not suffer from restricted mitochondrial function due to lower oxygen availability. Instead, the physioxia conditions increased the potential of oxidative phosphorylation in NPCs cultured at 3% O₂ (v/v).

NPCs grown in physioxia produce less ROS

To verify whether increased oxygen consumption could cause physioxia-grown NPCs to produce more ROS, due to impartial oxygen reduction caused by electron leakage from the respiratory chain, we measured ROS production. We used the dye DHE, a permeable probe oxidized in the cytoplasm, which further intercalates DNA and emits fluorescence. As DHE fluorescence was measured in NPCs in their native oxygen environments (i.e., normoxia and physioxia), we observed that, under routine conditions, ROS production was equivalent in both situations (Fig. 5). However, after antimycin A addition, physioxia-grown cells showed lower ROS levels from mitochondrial complex III induced by semi-quinone radical stabilization, compared to normoxia conditions (Fig. 5).

These results suggest that NPCs grown under physiological oxygen are more resistant to redox imbalance. Moreover, this finding also suggests that their antioxidant defense system may be more efficient to scavenge the mitochondrial ROS generated.

GPx activity is lower in NPCs grown in physioxia

To assess whether the antioxidant defense system of physioxia-grown NPCs is more active under routine conditions, we measured the activities of SOD and GPx in these cells. SOD is responsible for catalyzing the dismutation of superoxide anions in hydrogen peroxide and water. Then, GPx oxidizes intracellular glutathione, reducing peroxide to alcohol and water.

Our data show that NPCs grown either in normoxia or physioxia have equivalent enzymatic activities; however, when treated with antimycin A, the NPCs cultured under lower oxygen concentrations showed a less-pronounced increase of GPx activity in response to the ROS increase, while the SOD activity level was sustained (Fig. 6). This finding indicates that, unlike what was hypothesized previously, NPCs grown in physioxia and treated with antimycin A do not show decreased ROS production due to greater antioxidant activity in the pathway analyzed herein. Instead, GPx, one of the core enzymes of the cellular antioxidant machinery, depicts a decreased response to the ROS increase in these NPCs.

Although the glutathione pathway is not the only one involved in superoxide anion detoxification, it seems to be extremely relevant in NPCs, even surpassing the activity of catalase and peroxiredoxin 3, as shown previously (Mitozo et al., 2011; Xi et al., 2014). Taken together, these data suggest that, in fact, the antioxidant defense enzymes may be downregulated in physiological oxygen concentrations, compared to atmospheric ones.

Physioxia-grown NPCs suffer increased DNA damage when exposed to hydrogen peroxide

Although physioxia-grown NPCs produced lower levels of ROS when treated with antimycin A (Fig. 5), our data show that these cells also had a decreased activity of GPx (Fig. 6), one of the main enzymes responsible for peroxide detoxification. Therefore, we investigated whether NPCs cultivated under low oxygen concentrations would be more susceptible to stress. To this end, we quantified DNA damage, an indication of oxidative stress (Konyalioglu et al., 2013), in NPCs grown under routine conditions and after hydrogen peroxide treatment.

To quantify the impact of these treatments on DNA damage, the antibody against H2A.X was used to measure altered DNA. H2A.X is phosphorylated when DNA strands break, which signals for DNA repair and cell cycle arrest. Staining of H2A.X appears within the nucleus and can be monitored by both the overall fluorescence intensity and the number of visibly detectable aggregated structures.

Under our routine conditions, we observed that NPCs grown in normoxia or physioxia had equivalent levels of DNA damage (Fig. 7). However, after exposure to hydrogen peroxide, cells cultured under lower oxygen concentrations showed increased DNA damage, which was observed by increased fluorescence intensity both in the whole nucleus and in H2A.X spots (Fig. 7C). Moreover, the population of nuclei containing more than 20 spots decreased; while the whole stained nuclei population, which depicted so many spots that they became indistinguishable from one another, was increased in physioxia-grown NPCs (Fig. 7D).

Increased H2A.X staining of NPCs grown in physioxia and treated with hydrogen peroxide suggests that these cells are more susceptible to exogenous ROS than those cultured in normoxia. Taken together, our results show that higher susceptibility to oxidative stress could be partly due to a lower response of antioxidant enzymes to stress agents in physioxia-grown NPCs, which could contribute to decreased resistance to oxidative stress.