The rat-derived oligodendrocyte progenitor cell line OLN-93 [34] was cultured as described previously [35].

The rat-derived immature astroglial cell line C6 [36, 37] was purchased from the American Type Culture Collection (ATTC) cultured as recommended.

Both cell lines were cultured to stable replication for up to ten passages after thawing in a humidified atmosphere at 37 ${}^{\circ }$C and 5% CO${}_2$ (21% O${}_2$, normoxia; Binder, CB150).

To investigate the causes of the protective effects, concentrations were selected which exhibit a long-lasting protective effect. Only by this approach, further experiments can be conducted to investigate the underlying mechanisms. Therefore, different concentrations were used for the different substances, which exhibited a long-lasting protective effect in earlier experiments. Therefore, different concentrations were used for the different substances or cell lines. Stock solutions (1000$\times$) of 17$\beta$-estradiol (E2; 1 $\mu$M for OLN-93 cells and 10 $\mu$M for C6 cells; Sigma-Aldrich) and dehydroepiandrosterone (DHEA; 100 nM; Sigma-Aldrich) were prepared in dimethylsulfoxide (DMSO). The DMSO was added to the control sample corresponding to the DMSO concentration in the treatment sample.

After 48 h of recovery in culture medium, OLN-93 and C6 media were exchanged to experimental media as previously described [20] containing steroids or DMSO. After 2 h of preincubation, the medium was renewed, and the cells were exposed to hyperoxia (humidified atmosphere at 37 ${}^{\circ }$C with 5% CO${}_2$ and 80% O${}_2$; Binder, CB50) or normoxia for the indicated time.

C6 and OLN-93 cells were preincubated for 2 h with 1 $\mu$M and 10 $\mu$M of E2 in DMSO, respectively, and subsequently damaged by hyperoxia for 48 h. A lactate dehydrogenase (LDH) cytotoxicity assay (Abcam) was used according to the manufacturer’s instructions to assess cell death in OLN-93 cells. LDH is released from damaged cells and can be measured in the supernatant. LDH oxidizes lactate and thereby produces NADH, which reacts with a red water-soluble tetrazolium to a yellow color. The color was quantified by a plate reader (Thermo Fisher, Multiscan Ex) at OD 450 nm. The OD 650 nm was measured as a reference and subtracted from OD 450 nm. Results are given as % cell death compared to DMSO-treated cells under hyperoxia.

A 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Promega) was used according to the manufacturer’s instructions to measure the viability of C6 cells. In this assay, MTT, a yellow tetrazole, is reduced to purple formazan in living cells. The conversion was measured by a plate reader at OD 595 nm and OD 650 nm as a reference and subtracted from OD 595 nm. For higher comparability among the survival assays, the viability results were processed to cell death by the formula: % cell death = 100- (% viability -100). Results are given as % cell death compared to DMSO-treated cells under hyperoxia.

To investigate the influence of steroid treatment and hyperoxia on the ER-$\alpha$, ER-$\beta$, GPER and AR receptor expression, C6 and OLN-93 cells were pretreated 2 h with DMSO, E2, DHEA or E2 and DHEA and subsequently exposed to 48 h of hyperoxia. A normoxia DMSO control was included. Afterwards, proteins were isolated using TRIzol Reagent (Life Technologies), followed by quantification using Roti-Quant protein quantitation assay (Carl Roth) according to each manufacturer’s instruction. Proteins were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride membranes (PVDF; Bio-Rad Laboratories) in a wet transfer system (all devices purchased from Bio-Rad Laboratories). The membranes were blocked 2 h in tris (hydroxymethyl)-aminomethan (Tris)-buffered saline (Carl Roth) containing 0.15% Tween 20 (Sigma-Aldrich) and 2.5% non-fat milk (Carl Roth). Afterward, membranes were incubated overnight with primary antibodies and the next day for 1 h with horseradish peroxidase-conjugated secondary antibodies. The following primary antibodies were used: anti-ER-$\alpha$ rabbit polyclonal antibody (1 : 200; Santa Cruz), anti-ER-$\beta$ rabbit polyclonal antibody (1 : 1,000; Affinity BioReagent, Thermo Fisher), anti-GPER rabbit polyclonal antibody (1 : 250; Abcam), anti-AR mouse monoclonal antibody (1 : 200; Abcam), and anti-$\beta$-actin rabbit polyclonal antibody (1 : 5,000; Sigma-Aldrich) (Table 1). Peroxidase-conjugated goat anti-rabbit and goat anti-mouse IgG (H + L) secondary antibodies were purchased from Jackson ImmunoResearch (1 : 5,000). Specific proteins were visualized with enhanced chemiluminescence detection reagent according to the manufacturer’s instructions (Thermo Fisher) and developed by an imaging system (Bio-Rad Laboratories, ChemiDoc XRS). Receptor proteins were quantified as a ratio to $\beta$-actin with densitometry using Image Lab software (Bio-Rad Laboratories). In ER-$\alpha$, not the full-length protein (66 kDa) but the 50 kDa splice variant was investigated since it was the most dominant variant in both cell lines. By calculating the ratio to $\beta$-actin expression, the concentration of receptor proteins is presented as semi-quantitative data compared to DMSO-treated cells under normoxia.

C6 and OLN-93 cells were pretreated for 2 h with DMSO, E2, DHEA, or E2 and DHEA for signaling array experiments. Cells were collected after 6 h and 24 h of hyperoxia, according to the manufacturer’s instruction. As recommended by the manufacturer, the lysis buffer was supplement with Halt Protease and Phosphatase Inhibitor Single-Use Cocktail (100$\times$, Thermo Fisher). According to the manufacturer’s instruction, the cell lysates were quantified by Roti-Quant protein quantitation assay (Carl Roth). Afterward, cell lysates were analyzed using an intracellular signaling array kit (Cell Signaling Technology) to detect activated signaling molecules. For this actual human-specific array, 11 of 18 antibodies were species cross-reactive to rat-derived samples, as confirmed by the manufacturer-only 9 of those 11 antibodies delivered detectable signal.

According to the manufacturer’s instructions, cell lysates were added to the nitrocellulose-coated glass slides pre-coated with primary antibodies for 2 h. Afterward, the slides were incubated for 1 h with biotinylated antibodies, followed by a horseradish peroxidase-conjugated secondary antibody for 30 min. The slides were exposed to a substrate, and chemiluminescence was measured. Each array was processed with the “Protein Array Analyzer” macro in ImageJ (NIH) for background subtraction and grid measurement [38]. Normalization between arrays was calculated by scaling the individual intensities so that the mean intensities were the same across all compared arrays. The results show only proteins for which signal intensities were stronger than the background after 1 min of exposure. The signal intensities are given as the ratio to DMSO-treated cells, calculated for each experiment.

Statistical significance was assessed with GraphPad Prism 8.3.1 software (GraphPad Software, LLC) by Two-way ANOVA or Mixed Model depending on missing values. We corrected for multiple comparisons using Tukey’s post hoc test. Matched values were factored in as appropriate. Friedman Test was used for data of Fig. 2. The significance levels were denoted as *P 0.05, **P 0.01, and ***P 0.001. A level of P 0.05 was considered statistically significant. Data in the figures are represented as mean $\pm$ SD and result from at least four independent experiments performed on different days. Replicate wells were used and pooled for analysis. Due to the limited amount of data points per cell culture condition, some of the data must be considered explorative.

Fig. 2.

Effects of E2 and DHEA co-treatment on hyperoxia-mediated cell death. Cell survival was measured (A) in C6 cells by MTT assays and (B) in OLN-93 cells by LDH assays after treatment with E2 alone (10 or 1 $\mu$M respectively), DHEA (100 nM) alone or co-treatment under hyperoxia. DMSO hyperoxia control was set to 100% to exclude inter-assay variability. Significant differences to DMSO hyperoxia control are indicated by *; significant differences between steroid treatments are indicated by #. */# P $\le$ 0.05, **/## P $\le$ 0.01; ***/### P $\le$ 0.001.

Under hyperoxia, the effect of treatment on cell survival with DHEA alone, co-treatment with E2 or E2 alone was compared to DMSO control (set to 100%) (Fig. 2A, B). While all hormone conditions reduced cell death significantly in both cell types, co-treatment significantly increased cell survival compared to DHEA or E2 single treatment in OLN-93 cells (P $\le$ 0.001; Fig. 2B).

To investigate the influence of steroid treatment and hyperoxia on the steroid receptor availability, the expression of ER-$\alpha$, ER-$\beta$, GPER and AR was analyzed. For these experiments, C6 and OLN-93 cells were pretreated for 2 h with DMSO, E2, DHEA, or E2 and DHEA and afterward subjected to 48 h of hyperoxia.

In C6 cells, neither steroid treatment nor hyperoxia led to significant ER-$\alpha$ or $\beta$ or GPER expression changes compared to DMSO hyperoxia control (Fig. 3A, B, C). Treatment with E2 led to a significant increase of AR compared to DHEA (P $\le$ 0.05) (Fig. 3D).

Fig. 3.

Effects of steroid treatment and hyperoxia on receptor expression in C6 cells. The expression of (A) 50 kDa splice variant of ER-$\alpha$, (B) ER-$\beta$, (C) GPER and (D) AR in C6 cells were analyzed by western blot after treatment with E2 alone (10 $\mu$M), DHEA (100 nM) alone or co-treatment. Data were normalized to normoxia DMSO control (set to 100%). Significant differences between hyperoxia DMSO control and treatments are indicated by * P $\le$ 0.05, ** P $\le$ 0.01; *** P $\le$ 0.001.

In OLN-93 cells, treatment with DHEA plus E2 led to a significant increase of ER-$\alpha$ compared to DMSO under hyperoxia (P $\le$ 0.01). DHEA (P $\le$ 0.01) and its combination with E2 (P $\le$ 0.001) increased ER-a significantly compared to E2 alone (Fig. 4A). The relative quantification of ER-$\beta$ led to similar results (Fig. 4B).

There were no significant differences for GPER or AR observed under hyperoxia (Fig. 4C, D).

Fig. 4.

Effects of steroid treatment and hyperoxia on receptor expression in OLN-93 cells. The expression of (A) 50 kDa splice variant of ER-$\alpha$, (B) ER-$\beta$, (C) GPER and (D) AR in OLN-93 cells was analyzed by western blot after treatment with E2 alone (1 $\mu$M), DHEA (100 nM) alone or co-treatment. Data were normalized to normoxia DMSO control (set to 100%). Significant differences between hyperoxia DMSO control and treatments are indicated by * P $\le$ 0.05, ** P $\le$ 0.01; *** P $\le$ 0.001.

In C6 cells, ERK1/2 (Thr202/Tyr204), Akt (Thr308 and Ser473), S6 ribosomal protein (Ser235/236; S6RP), Bad (Ser112), ribosomal protein S6 kinase (Thr389; p70S6K), proline-rich Akt/PKB substrate 40 kDa (Thr246; PRAS40), and glycogen synthase kinase-3$\beta$ (Ser9; GSK-3$\beta$) were detectable while Stat3 (Tyr705) or p38 (Thr180/Tyr182) were not (Fig. 5A, B).

Fig. 5.

Intracellular signaling among hyperoxia and treatments in C6 cells. To explore intracellular signaling, a signaling array was used. After 2 h treatment with DMSO, E2, DHEA, or DHEA/E2 co-treatment, cells were exposed to 6 h and 24 h of hyperoxia. C6 cells were treated with DMSO, 10 $\mu$M E2, 100 nM DHEA, or DHEA and E2 combined and afterward exposed to (A) 6 h and (B) 24 h hyperoxia. Significant differences from hyperoxia/DMSO control, set to 1.0, are indicated by # P $\le$ 0.05. Significant differences between steroid treatments are indicated by * P $\le$ 0.05. White bars represent E2-treated cells, light grey bars represent DHEA-treated cells, and dark gray bars represent DHEA/E2-co-treated cells.

After 6 h of hyperoxia, Akt (Ser473) was significantly downregulated, whereas GSK-3$\beta$ was significantly upregulated in DHEA plus E2-treated cells compared to DMSO control (Fig. 5A). The activation of Akt seemed to be transient.

After 24 h of hyperoxia, there were no significant differences in E2- and DHEA-treated cells (Fig. 5B). Under DHEA + E2 co-treatment, Bad, PRAS40, and GSK-3$\beta$ were significantly upregulated compared to DMSO control. There were no differences between the different steroid treatments (Fig. 5A, B).

In OLN-93 cells, the signaling mentioned above molecules and AMP-activated protein kinase $\alpha$ (Thr172; AMPK$\alpha$) were detectable. Again, no distinct activation of Stat3 or p38 was measured (Fig. 6A, B).

Fig. 6.

Intracellular signaling among hyperoxia and treatments in OLN-93 cells. To explore intracellular signaling, a signaling array was used. After 2 h treatment with DMSO, E2, DHEA, or DHEA/E2 co-treatment, cells were exposed to 6 h and 24 h of hyperoxia. OLN-93 cells were treated with DMSO, 1 $\mu$M E2, 100 nM DHEA, or DHEA and E2 combined and afterward exposed to (A) 6 h and (B) 24 h hyperoxia. Significant differences from hyperoxia/DMSO control, set to 1.0, are indicated by # P $\le$ 0.05. Significant differences between steroid treatments are indicated by * P $\le$ 0.05. White bars represent E2-treated cells, light grey bars represent DHEA-treated cells, and dark gray bars represent DHEA/E2-co-treated cells.

After 6 h of hyperoxia, there was a significant downregulation of GSK-3b under DHEA treatment (Fig. 6A). Of interest, there was a significant difference between DHEA single and its combination treatment with E2 in phosphorylation of Akt (Thr308). The activation of Akt in E2 and DHEA/E2 co-treated cells seems to be transient as in the C6 cells because there was a significant upregulation of the Akt pathway (Fig. 6A).

After 24 h of hyperoxia, no significant differences were observed (Fig. 6B).

The effects of hyperoxia and steroid hormones on receptor density have been investigated in only two studies to date. In one study, astrocytes were shown to downregulate the amount of progesterone receptor by hyperoxia [39]. The other study showed in cultured smooth muscle cells of the respiratory tract that hyperoxia (with 50% O${}_2$) does not change the receptor density of ER-$\alpha$ and ER-$\beta$ [40] in human fetuses.

Hyperoxia induced significant changes in receptor expression compared to DMSO control in OLN-93 cells. The effects on the C6 cells were the only minor. Since a similar trend in regulation could be observed, the adverse finding could be based on the limited number of experiments.

Since full-length ER-$\alpha$ was hardly detectable in both cell types and there was a relatively large amount of a ~50 kDa splice variant of ER-$\alpha$, this variant was quantified. The mRNA of ER-$\alpha$ can undergo alternative splicing, whereby one or more exons are deleted [41]. In the 50 kDa band for ER-$\alpha$ examined here, exon 7 is depleted [42]. Not much is known about this splice variant so far. However, it does not seem to modulate gene expression [43]. The function of this splice variant in receptor-mediated protection should be the subject of future investigations.

In C6 cells, E2 seems to induce AR via cross-regulation under hyperoxia. The induction of AR via E2 has already been described for other tissues [44]. However, it remains to be clarified whether the effect of DHEA is an amplification of autoinduction and/or a cross-regulation since DHEA can bind to classical ERs and AR.

In OLN-93 cells, ER-$\alpha$ and -$\beta$ were upregulated under the pathological situation. However, this effect was reversed for all three receptors by treatment with E2 by autorepression. This autorepression by E2 was previously also found in other estrogen-dependent tumor cells [45].

Receptor-mediated survival can be initiated via genomic and non-genomic pathways. The activation of signal transduction pathways can be temporary (transient activation) or persistent (permanent activation). Transient signals can be activated just a few minutes after adding factors and can disappear after minutes or hours [46, 47]. E2 and DHEA can induce non-genomic neuroprotective signaling pathways via ERK1/2 and Akt [8, 48, 49].

It was previously shown that PI3K/Akt/Bad signaling mediates protection of E2 against oxidative stress-induced apoptosis in C6 cells [50], and treatment with DHEA prevents cell death via PI3K/Akt/mTOR signaling in newborn neurons [49]. Even without sustained upregulation of Akt itself, we found downstream signaling proteins of the PI3K/Akt/Bad and mTOR pathways significantly upregulated in C6 cells after 24 h of hyperoxia in the co-treatment group. Of interest, it was previously reported that induction of cell death could induce ERK1/2 phosphorylation in astrocytes [51, 52], and E2 can reverse hyperoxia-induced phosphorylation of ERK1/2 in astrocytes [53].

Although it was shown that E2 activates ERK1/2 and Akt signaling to prevent cell death in oligodendrocyte precursor cells [8], we only found downregulation of GSK-3b under DHEA treatment after 6 h in OLN-93 cells. This might be due to differences in cell culture, experimental setting or even the used oligodendrocyte cell type.