Third Military Medical University, an experimental animal center, provided 8–10weeks (18±2g) C57BL/6 male mice. They were kept at 25°C, with 45% humidity, and exposed to a 12-h light −12-h dark cycle. The hypoxia group was exposed to an altitude of 6,000m for 7days. The normoxic control group was kept at the normal atmosphere of about 300m for 7days. All procedures followed the standards of experimental animal ethics.

After the hypoxia treatment, a small amount of chloral hydrate (10%, w/v) was used to inject the mice to narcosis, which were then perfused with 400mL of normal saline (0.9%, v/v) intracardially; the hippocampus was harvested for western blot and quantitative real-time PCR (qRT-PCR). This was followed by perfusing with 400mL of normal saline (0.9%, v/v) and 350mL of paraformaldehyde (4%, v/v). Finally, we decapitated the mice and fixed their brains in paraformaldehyde (4%, v/v) overnight. The brains were dewatered using 30% sucrose in the same solution; sections of brain serial coronal (30μm) were collected with a cryostat (CM1900, Leica Microsystems, Germany) at −20°C and used further for immunofluorescence (IF).

The cell repository of the Chinese Academy of Sciences offered the PC12 cell line (high differentiation). The mouse hippocampal neurons HT22 cell line was purchased from iCell (Shanghai, China). We cultured PC12 cells in the RPMI-1640 medium (HyClone) and HT22 cells in the DMEM high glucose medium (Gibco), supplemented with 10% (v/v) fetal bovine serum (FBS; Gibco) and 1% (v/v) penicillin-streptomycin solution at a temperature of 37°C plus 5% CO₂.

For the experiments, the cells were placed in a hypoxic chamber with 1% O₂, 5% CO₂, and 94% N₂ for the design time. The normoxic control group incubated cells for the same duration with 5% CO₂ and 95% air.

Annexin V-FITC/PI double-staining and flow cytometry analyses were utilized to detect the apoptotic cells. Briefly, the PC12 cells were subjected to incubation at 1% oxygen concentration for 24 or 48h, and the cells were harvested and resuspended in phosphate-buffered saline (PBS) at 1×10⁵cells/mL. The PC12 apoptosis was detected using the Annexin V staining Protocols assay kit (Affymetrix eBioscience, 88-8005-72, United States) based on the manufacturer’s protocol. We added 5μL of annexin V-FITC and 195μL of annexin V-FITC binding buffer into the mixture after a 5-min centrifugation at 500×g at a temperature of 25°C. The solution was gently vortex-mixed, and the mixture was incubated for 10min in the dark at room temperature. Finally, after another centrifugation, 10μL PI and 190μL of FITC-conjugated annexin V binding buffer were supplemented. The samples were lightly vortex-mixed and assayed with a FACSCalibur Flow Cytometer (Becton Dickinson, San Jose, CA, United States) within half an hour.

The CCK-8 assay was carried out as reported in previous literature (Heng et al., 2010). The PC12 and HT22 cells were incubated in an oxygen-deficient chamber for 0, 24, or 48h and rinsed twice with PBS before 10μL of CCK-8 solution (CK04, Dojindo Laboratories, Japan) and 100μL of the substrate were added to it. The absorbance was detected at 450nm by a Multiskan Go Microplate Absorbance Reader (Thermo, Massachusetts, United States) by incubating the microplate at a temperature of 37°C for 2h. Cell viability was defined as the proportion of the viable cells compared to the control cells (100%). We repeated all experiments thrice, and the results were expressed as mean±SD.

The cytokines (IL-6, IL-1β, and IL-10), which contributed to inflammation, were assessed using tetramethylbenzidine (TMB) ELISA kit (M6000B, MLB00C, M1000B, R&D, United States) based on the manufacturer’s protocol. A microplate reader (Thermo, Massachusetts, United States) was utilized to detect the absorbance at 450nm.

The samples were blocked by QuickBlock™ blocking buffer for 1h (Beyotime Biotech Inc., P0260, Shanghai, China) for IF staining. The brain sections were afterward subjected to incubation with rabbit anti-GRP78/BIP (1:500, Proteintech, 66,574-1-Ig, United States) or rabbit anti-CHOP (1:500, Proteintech, 15,204-1-AP, United States) overnight at 4°C. Cy3-labeled Goat Anti-Rabbit IgG (H+L; 1:200, Beyotime Biotech Inc., A0516, Shanghai, China) was used to visualize the binding, while the antifade mounting medium with DAPI (Beyotime Biotech Inc., P0131, Shanghai, China) was applied to stain the nuclei. We imaged the immunolabeled brain slices with a CLSM (IX81, Olympus, Tokyo, Japan) and then analyzed them with a three-dimensional (3D) constructor, ImageJ software (National Institutes of Health, Bethesda, MD, United States).

The PC12 cells were subjected to incubation in rabbit anti-GRP78/BIP (1:500, Proteintech, 66574-1-Ig, United States) or rabbit anti-ATF6 (1:500, Proteintech, 66563-1-Ig, United States) or rabbit anti-CHOP (1:500, Proteintech, 15204-1-AP, United States) overnight at a temperature of 4°C. The other steps were the same as the brain sections.

The PC12 cells were transfected siRNA targeted STAT3 to knockdown the STAT3 expression. Briefly, to prepare the lipid–siRNA complex, siRNA (GenePharma; Table 1) and Lipofectamine 2000 (Invitrogen) were diluted in Opti-MEM (Invitrogen) for 5min. Next, we mixed siRNA and Lipofectamine 2000 solution and incubated them for an extra 20min. After gently layering the siRNA-Lipofectamine 2000 complex, we placed it in the cell cultures for 4h. Finally, the substrate was replaced completely. The cells were thus transfected and used for further processes.

The cells were rinsed twice with ice-cold PBS, and the cell lysates were extracted using the radioimmunoprecipitation (RIPA) lysis buffer (Beyotime Biotech Inc. P0013B, Shanghai China) containing 1mM PMSF. The BCA Protein Assay Kit (Beyotime Biotech Inc., P0009, Shanghai, China) was used to determine the lysate’s protein concentration. Equivalent amounts of protein from each sample were extracted using 10% sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to a polyvinylidene difluoride (PVDF) membrane. The membrane, after blocking with 5% milk for 1h, was incubated with the primary antibody in the blocking buffer overnight at 4°C. The membrane was then rinsed three times with the blocking buffer and supplemented with the Goat Anti-Rabbit IgG(H+L) secondary antibody (Beyotime Biotech Inc., A0208, Shanghai, China) labeled with horseradish peroxidase (HRP) at a 1:2000 dilution in blocking buffer, and the membrane was then incubated for 1h at room temperature. Enhanced chemiluminescence (ECL) detection kit (BG0001, Bioground, Chongqing, China) was adopted to visualize antibody binding, and the images were captured using the Quality One software (Bio-Rad) scanner.

The primary antibodies and dilutions were as follows: anti-GRP78/BIP (1:200, Abcam, ab108615, United States), anti-ATF6 antibody (1:1000, Santa Cruz Biotechnology, SC-22799, United States), anti-GADD153/CHOP antibody (1:1000, Santa Cruz Biotechnology, SC-575, United States), anti-Phospho-STAT3 (Tyr705) antibody (1:1000, Cell Signaling Technology, 9145, United States), and anti-β-actin antibody (1:2000, BestBio, BBA1107-2, Shanghai, China).

We extracted total RNA with RNAprep pure Micro Kit (Tiangen, DP420, Beijing, China); it was reverse transcribed to cDNA by the RNA PCR Kit (Takara, RR036A, Shiga, Japan) based on the manufacturer’s guidance. Subsequently, we amplified the cDNA product with the PCR amplification kit (Takara, RR820A, Shiga, Japan). Taking the β-actin gene as an internal control, we determined the relative gene expression with the comparative CT method. Table 1 illustrates the primer sequences. The PCR settings were set below: 95°C for 30s, 40cycles at 95°C for 5s, and 60°C for 30s. The 2^−∆∆Ct methods were utilized to calculate the relative expression fold change of mRNAs.

The chromatin immunoprecipitation (ChIP) assay was done on HT22 cells cultured in IL-6 (40ng/mL) to promote the STAT3 activation by SeqHealth (Wuhan, China). Following the above-mentioned processes, the cell protein-DNA complexes were cross-linked using 1% formalin in PBS on a rocking device at room temperature for 15min. The formaldehyde was quenched by adding 125mM glycine and rocked for 5min at room temperature. Afterward, the cells were collected and treated with lysis buffer in 500μL per 5×10⁶ cells on ice for 10min. The chromatin was sheared to an average length of about 1kb using sonication. After ultracentrifugation (10min) with a refrigerator at 12,000×g, we collected the supernatants and performed ChIP using the EZ-Magna ChIP A/G Assay kit (Millipore, 17-10086, United States). The chromatin mixture was immunoprecipitated with anti-phospho-STAT3 (Tyr705) antibody (1:100, Cell Signaling Technology, 9,145, United States). The DNA was purified and sequenced with ChIP-seq.

Results are represented as mean±standard deviation (SD). The SPSS 18.0 statistical software program was used to make statistical analyses. The unpaired t test was utilized to compare two groups’ data, while we compared data from 3 or more groups with ANOVA. The differences were considered significant at p<0.05.

The inflammatory factors were observed to determine whether the hypobaric hypoxia-induced ER stress is related to inflammation in the hippocampus. The ELISA assay displayed a remarkable increase in IL-6, a dramatic drop in IL-10 in the hypoxia group, and no difference was seen in IL-1β between the hypoxia and normoxia groups (Figure 1).

The inflammatory factor (IL-6) ligand-receptor interaction causes JAK activation, which in turn activates STAT3 by tyrosine (Tyr705) phosphorylation (Wrighting and Andrews, 2006). Our data showed elevated mRNA of STAT3 in the hypoxia group (Figure 2D). Meanwhile, IL-6 significantly increased the protein levels of p-STAT3 (Tyr705; Figures 2A,B). The above results indicated an enhanced inflammatory response to hypobaric hypoxia in the hippocampus of mice.

The CCK8 assay and apoptosis analysis by flow cytometry were performed to investigate the hypoxia effect on the neuron cells’ viability and apoptosis. Exposure to 1% O₂ for 48h markedly decreased the cell viability (Figure 3A). In apoptosis analysis by flow cytometry, viable cells were not stained by annexin V-FITC or PI (lower left quadrant, Q4), early apoptotic cells were bound with annexin V-FITC (lower right quadrant, Q3), and late apoptotic cells were with V-FITC and PI (upper right quadrant, Q2). Flow cytometry with PI staining and Annexin V displayed a remarkably higher percentage of apoptotic PC12 cell count subjected to 1% O₂ exposure for 48h than the control group (Figures 3B,C). Among the increased apoptotic cells, the percentage of late apoptosis cells were significantly higher than that of early apoptosis cells (3.58±0.58 vs. 1.82±0.37, p<0.05). Therefore, the findings indicate that hypoxia reduced viability and augmented apoptosis in neuron cells.

ER stress is significant in HI-induced apoptosis in vitro (Chen et al., 2008; Roussel et al., 2013) and in vivo (Galehdar et al., 2010; Lopez-Hernandez et al., 2015). The GRP78, ATF6, and CHOP levels were observed in hypoxia animal and cellular models to understand the mechanism of ER stress. The expression of GRP78 and CHOP in the hippocampus showed increased ER stress in the hypobaric hypoxia exposure groups by IF (Figures 4A,B). The proteins related to ER stress, ATF6 and GRP78, and the apoptosis-related protein, CHOP, increased in the hypobaric hypoxia exposure groups (Figures 4C,D). Hypobaric hypoxia mice showed augmented mRNA levels of STAT3, GRP78, ATF6, and CHOP (Figure 4E). The IF staining of ATF6, GRP78, and CHOP in PC12 cells showed increased ER stress in the normal baric hypoxia (NH) exposure groups (Figures 4F,G). Thus, hypoxia increased ER stress in vivo and in vitro.

The STAT3 activation by Tyr705 phosphorylation plays a role in ER stress-induced neuron apoptosis. However, the genomic binding pattern of p-STAT3 (Tyr705) in neurons is not yet clear. We used the ChIP-seq to identify genome-wide target sites of p-STAT3 (Tyr705) in HT22 cells. A total of 2676 p-STAT3 (Tyr705)-enrichment genes were identified. The binding sites located in the promoter TSS made up 2.73% of the total readings of p-STAT3 (Tyr705)-enrichment genes (Figure 5A). The gene ontology (GO) analysis of the genes related to the peak suggested that the p-STAT3 (Tyr705) target genes took part in many biological activities, like single-multicellular organism process, protein binding, and biological regulation (Figure 5B). The Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis implied that the p-STAT3 (Tyr705) genes related to the peak were remarkably enriched in the pathways of metabolism, which included glutamatergic synapse, insulin secretion, dopaminergic synapse, calcium signaling pathway, and cAMP signaling pathway (Figure 5C). From the peak distribution in upstream and downstream transcription start site (TSS), we observed the enrichment of p-STAT3 (Tyr705) binding sites close to TSS (Figures 5D,E), indicating the binding of p-STAT3 (Tyr705) with ATF6 promoter region (Table 2).

We examined the ATF6 levels where STAT3 was either overexpressed or silenced in PC12 to ascertain the causal relationship between ATF6 and STAT3 activation. The protein levels and mRNA levels of ATF6 declined in activated p-STAT3 (Tyr705) by IL-6 but upregulated in inhibited STAT3 by siRNA (Figures 2A–D). However, the protein and mRNA expressions of CHOP augmented in activated p-STAT3 (Tyr705) by IL-6 and decreased in inhibited STAT3 by siRNA (Figures 2A–D).

Although this study confirmed the injury effect of STAT3 on neuron apoptosis through downregulated ATF6 under hypoxia in cells, there are some limitations, such as the application of gene knockout mice. Further research on the cognitive impairment of STAT3 should be verified in animal behavior assays.