Wistar rats were handled in accordance with institutional guidelines and with permission of the Landesamt für Gesundheit und Soziales (LAGeSo), Berlin. Brains of day 17 Wistar rat embryos (E17) were isolated and cortices were dissected and seeded with a density of 175,000 cells per cm² and cultured for 9 days as described (Mergenthaler et al., 2012) unless otherwise stated. Briefly, primary cortical neurons were seeded in Neurobasal medium (NBM; Invitrogen, Thermo Fisher Scientific) supplemented with 25 μM glutamate, 0.5 mM L-glutamine, and a serum-free supplement (B27, N2, Invitrogen, Thermo Fisher Scientific, or GS21, MTI-GlobalStem). B27, N2, and GS21 are based on published formulations (Bottenstein and Sato, 1979; Brewer et al., 1993; Chen et al., 2008); see Table 1 for details on their composition. The medium was partially replaced every 4 days with NBM supplemented with 0.5 mM L-glutamine and serum-free supplement. Alternatively, neurons were cultured in BrainPhys medium instead of NBM which was prepared in house based on the published formulation (Bardy et al., 2015) prior to its commercial availability (see Table 2 for the exact formulation used herein). The bicarbonate concentration was slightly decreased to 26 mM to adjust the pH to 7.40. Primary cortical neurons were seeded in BrainPhys supplemented with 25 μM glutamate and a serum-free neuronal supplement (B27, N2, or GS21), and the medium was partially replaced during the cultivation period as described above.

On day 9 of cultivation, neuronal cultures were washed twice with PBS and then incubated under anoxic conditions (0% O₂, 37°C, 5% CO₂) in a Concept-400 hypoxia workstation (Ruskinn Technologies) or under normoxic conditions (21% O₂, 37°C, 5% CO₂) for 8 h in BSS₀ (116 mM NaCl, 5.4 mM KCl, 0.8 mM MgSO4, 1 mM NaH₂PO₄, 26.2 mM NaHCO₃, 10 μM glycine, 1.8 mM CaCl₂, 10 mM HEPES pH 7.4) as described (Mergenthaler et al., 2012). After the experimental procedures, the supernatant was analyzed directly or medium was added to the cells for 24 h until neuronal cell death was quantified by measuring lactate dehydrogenase (LDH) release.

Cell death was evaluated 24 h after metabolic deprivation by measuring LDH release from primary neurons in a coupled spectrophotometric assay as previously described (Niu et al., 2017). Briefly, 50 μl supernatant or 25 μl LDH standard (500 U/l, DiaSys Greiner) for data normalization were mixed with 200 μl of 212 μM β-NADH in 33.3 mM KH₂PO₄ and 66.7 mM K₂HPO₄ (pH 7.4), and then 25 μl of 22.7 mM pyruvate in 33.3 mM KH₂PO₄ and 66.7 mM K₂HPO₄ (pH 7.4) to start the reaction. The reduction of β-NADH to NAD+ is proportional to the LDH activity and was measured by absorbance at 340 nm on an MRX revelation (Dynex Technologies) plate reader at room temperature (RT). Total LDH release was measured after incubating neurons with Triton X-100 for 30 min at 37°C (final concentration: 0.5% v/v) and a second measurement with 25 μl supernatant or 25 μl LDH standard was performed. All data were normalized to the total LDH release measurements.

The colorimetric L-Lactic Acid Assay Kit (AAT Bioquest) was used according to the manufacturer’s instructions to measure lactate levels in the media of rat brain cortical neurons cultured for 9 days. Briefly, 50 μl sample was mixed with 50 μl assay buffer and absorbance was measured after an incubation period of 90 (Figure 2A) or 30 min (Figure 2B) at 550 nm on an MRX revelation (Dynex Technologies) plate reader at RT. Measurements of each well were normalized to its total protein content as quantified with a BCA Assay Kit (Thermo Fisher Scientific).

Metabolic flux was measured using the Seahorse XFe96 Extracellular Flux Analyzer (Seahorse Bioscience, Agilent Technologies). Appropriate cell numbers and compound concentrations were titrated in preliminary experiments (data not shown). The cell seeding density and compound concentrations given below represent conditions with the optimal dynamic range in our preliminary assays. Neurons were seeded at a density of 20,000 cells/well, and cultured for 9 days in NBM or BrainPhys as described above in the presence of the different neuronal supplements (B27, N2, or GS21). Two types of metabolic analyses were performed: the cell mito stress test to assess respiratory activity, and the glycolysis stress test to assess glycolysis (Figure 3). Both assays were performed in DMEM-D5030 (Sigma–Aldrich) containing 143 mM sodium chloride and 2 mM L-glutamine (pH 7.35–7.40). The medium was supplemented with 10 mM glucose for the cell mito stress test. Prior to the assays, cultures were washed with the respective medium. To characterize the effects of B27, N2, and GS21 on metabolic flux, neurons were cultured with the respective supplements in the media indicated in the figure legends. The cell mito stress test and glycolysis stress test were performed in the absence or presence of the different supplements. For the latter, the supplements were individually added to the assay medium as indicated in the figure legends so that this incubation was performed in the presence of the same supplement as the cultivation period. Neurons were pre-incubated in this medium for 1 h at 37°C (atmospheric CO₂) before the start of the measurements. For the cell mito stress test, the sensor cartridge was loaded with 0.75 μM oligomycin (Sigma–Aldrich), 0.75 μM carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP, Sigma–Aldrich), 1 μM rotenone (Sigma–Aldrich), and 1 mM antimycin A (Sigma–Aldrich), and for the glycolysis stress test with 10 mM glucose,1 μM oligomycin, and 100 mM 2-deoxy-D-glucose (2-DG; Carl Roth). These reagents were consecutively injected into each well in this order. Measurements were taken 20 min after pre-incubation and 6, 12, and 18 min after each injection with 3 min mixing intervals in between prior to the next injection. For analysis and depiction in the figures, we used the following measurements after each injection: for the cell mito stress test basal 3, oligomycin 3, FCCP 1, and rotenone/antimycin 2; for the glycolysis stress test basal 3, glucose 3, oligomycin 3, and 2-DG 2. After each experiment, total protein concentration of each well was determined with a BCA Protein Assay Kit (Thermo Fisher Scientific) to normalize oxygen consumption rates (OCR) and extracellular acidification rates (ECAR) measurements to the total protein content of each well.

Neuronal cultures from different embryos were considered as independent observations. Statistical graphing was performed by using GraphPad Prism 5.0. One-way ANOVA and Turkey’s HSD post hoc tests were calculated for all experiments with n ≥ 3 in GraphPad Prism 5.0 or IBM SPSS Statistics 24.0. A p < 0.05 was considered statistically significant. In all figures, this is represented by ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, and ^∗∗∗p ≤ 0.001. Figures display single observations as single data points as well as mean ± SD with a whisker plot.

Neurons that were depleted of glucose (i.e., no glucose present during incubation) for 8 h were rescued from cell death when B27 was added to the deprivation buffer (BSS₀), indicating that B27 might be able to support neuronal survival for a limited time under these conditions (Figure 1A). This effect was not seen when neurons were incubated in NBM or BSS (BSS₂₅) containing 25 mM glucose with or without B27 for 8 h. Furthermore, decreasing glucose concentrations, thereby eliciting glucose deprivation (GD), with or without addition of B27 to BSS did not have any effect on neuronal survival under normoxic conditions (Figure 1B). However, as before, neurons depleted of glucose were protected from cell death when incubated in the presence of B27. On the contrary, we found that B27 decreased neuronal cell death under oxygen deprivation (OD) at low glucose concentrations, although this effect did not reach statistical significance at 0.5 mM (p = 0.49), 1 mM (p = 0.197), and higher glucose concentrations (Figure 1C). Furthermore, B27 protected neurons from cell death after oxygen–glucose deprivation (OGD), a model of hypoxia–ischemia, when no glucose was present in the deprivation buffer (Figure 1C). In summary, these data indicate that B27 is able to protect neurons from cell death triggered by glucose depletion under normoxia and hypoxia, as well as at low glucose levels under hypoxia.

To further investigate the role of B27 in protecting neurons from cell death under metabolic deprivation, we incubated cultured neurons in the basal medium DMEM-D5030 for 98 min without glucose as described in the section “Materials and Methods” for metabolic flux analysis. As expected, addition of 10 mM glucose resulted in glucose turnover and accumulation of lactate over a period of 15 min (Figure 2A). However, consistent with a profound effect on neuronal glucose metabolism, addition of B27 together with 10 mM glucose completely abolished lactate accumulation in that timeframe (Figure 2A). Next, we added B27 to neurons undergoing hypoxia with different glucose concentrations in BSS or NBM containing 25 mM glucose (Figure 2B). Under all conditions, lactate accumulation was decreased by the addition of B27, although this effect did not reach statistical significance (BSS₁₀ p = 0.295, BSS₂₅ p = 0.055, and NBM p = 0.881). Together, these data suggest that B27 interferes with glycolysis.

To further investigate the metabolic changes elicited by neuronal culture conditions, we performed live cell metabolic flux analysis (Figure 3). This assay allows simultaneous measurements of the two fundamental pathways of glucose utilization by measuring ECAR (Figure 3A) as a measure of neuronal lactate production, and OCR (Figure 3B) as an indicator of neuronal respiratory chain activity. Addition of substrates and inhibitors in sequential order as indicated allows dynamic measurements of metabolic parameters of glycolysis and oxidative phosphorylation (see Figure 3 and “Materials and Methods” for details). In the following, neurons were cultured in the presence of B27 or the alternative neuronal culture supplements N2 and GS21, and we performed metabolic flux measurements in the presence (closed symbols in Figures 4, 5) or absence (open symbols in Figures 4, 5) of the respective culture supplement (Figure 3C).

When directly assaying glycolysis, we found that addition of B27 to the assay medium acutely inhibited glycolytic pathways by decreasing glycolysis as well as the glycolytic capacity (Figure 4A, circles) but did not change respiratory activity (Figure 4B, circles). When using the N2 supplement instead of B27 (Figures 4A,B, triangles), we also measured inhibition of metabolic function. Compared to the more modern neuronal culture supplement GS21 (squares), addition of B27 or N2 substantially decreased glycolytic capacity of neurons (Figure 4A). In addition, respiratory activity (OCR, Figure 4B) was increased after cultivation (open squares) with GS21 and after its addition (solid squares) to neurons undergoing metabolic flux analysis compared to cultivation with or addition of B27 (circles) or N2 (triangles). When we directly probed neuronal respiratory activity (Figures 4C,D), we found that maximal respiration was highest when neurons were cultured in the presence of GS21 but measurements were performed in the absence of the supplement (Figure 4C, open squares). Glycolysis was not affected under these experimental conditions (Figure 4D). Thus, our data indicate that the crucial supplements for culturing neurons may alter metabolic function immediately after their addition to the assay medium as well as when they are continuously present in the culture media. However, this effect is complex, as no single ingredient of B27 (Table 1) had a comparable effect on neuronal metabolism (Supplementary Figure 1). Importantly, culturing neurons in the presence of GS21 resulted in a denser neurite network compared to B27 or N2 (Supplementary Figure 2), thereby corroborating the results by Chen et al. (2008) that B27 can negatively affect neuronal synaptic plasticity and neurite integrity.

Finally, we investigated the metabolic function of neurons cultivated in the recently introduced BrainPhys medium (Figure 5), which is considered to be more physiological than previous neuronal culture media. In contrast to NBM (Figure 4), B27 and N2 did not inhibit glycolysis when using BrainPhys medium and the supplements were added for the measurements. However, adding GS21 during measurements still resulted in significantly higher glycolysis compared to B27 and significantly higher glycolytic capacity compared to B27 and N2 (Figures 5A,B). Furthermore, mitochondrial respiratory function analyzed by measuring maximal respiration was significantly higher when neurons were incubated with GS21 during OCR measurements. Finally, ATP-producing glycolysis after complete inhibition of electron transport with rotenone and antimycin A (increased ECAR, Figure 5D) only remained active under culture and treatment conditions using GS21.