Germ cell-2 spermatids (GC-2spd) were purchased from the Conservation Genetics CAS Cell Bank (Shanghai, China) and cultured in DMEM (HyClone, Fisher Scientific International Inc., Logan, UT, United States) supplemented with 10% fetal bovine serum (Gibco, Thermo Fisher Scientific, Inc., Waltham, MA, United States) and 1% antibiotics (100 U/ml penicillin and 100 U/ml streptomycin, Invitrogen; Thermo Fisher, Waltham, MA, United States). GC-2spd cells were seeded into six-well plates at a concentration of 1 × 10⁵ cells/well (NEST, Wuxi, China) and incubated in an incubator containing 21% O₂ and 5% CO₂ (Thermo Fisher Scientific Inc., Waltham, MA, United States). The CoCl₂-induced hypoxia cell model is commonly used in research to evaluate the effects of hypoxia on cell proliferation and apoptosis as well as other deleterious effects of hypoxia (Yu et al., 2018; Zhang and Chen 2018; Zhang et al., 2019). Here, after 24 h of culturing, cells to be subjected to hypoxia treatment were either cultivated in a medium supplemented with 150 μM CoCl₂ (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) or transferred to a hypoxia workstation (Thermo Fisher Scientific Inc., Waltham, MA, United States) containing 1% O₂, 5% CO₂, and 94% N₂ at 37°C. Experiments were routinely performed in triplicate (n = 3).

GC-2spd cells were prepared as described previously under normoxic, hypoxic, and CoCl₂ conditions. After 12, 24, 48, and 72 h of culturing, each set of cells was digested with trypsin and washed twice with phosphate-buffered saline (PBS, HyClone, Fisher Scientific International Inc., Logan, UT, United States) at room temperature. Cell counting was performed using an improved Neubauer hemocytometer (Qiujing Industrial Co., Ltd., Shanghai, China).

GC-2spd cells subjected to hypoxic treatment for 48 h or treated with CoCl₂ for 24 h were stained using the PE Annexin V Apoptosis Detection Kit (BD Pharmingen; BD Biosciences, San Jose, CA, United States) according to the manufacturer’s protocol and analyzed using a flow cytometer and Kaluza software (Gallios; Beckman Coulter, Inc. Boulevard Brea, CA, United States).

GC-2spd cells subjected to hypoxic treatment for 48 h or treated with CoCl₂ for 24 h were washed twice with ice-cold PBS and then lysed in ice-cold radio immunoprecipitation assay lysis buffer (RIPA; Beyotime, Shanghai, China) supplemented with 1 mM phenylmethylsulfonyl fluoride (PMSF; Beyotime, Shanghai, China) for 0.5 h to extract total cellular proteins. Protein concentrations were determined using an enhanced BCA protein assay kit (Beyotime, Shanghai, China). A volume of 20 micrograms of protein was separated by 10% SDS-PAGE (Beyotime, Shanghai, China) and transferred onto nitrocellulose membranes (Millipore, Merck KGaA, Darmstadt, Germany). Next, the membranes were blocked with 5% non-fat dry milk in TBS containing 0.1% Tween (TBST) for 1 h at room temperature and then incubated overnight at 4°C with rabbit anti-HIF-1α antibodies (1:200; no. ab1; Abcam, Cambridge, United Kingdom) or mouse anti-β-actin antibodies (1:5,000; no. AF0003; Beyotime, Shanghai, China). Then, the membranes were washed with TBST and incubated with horseradish peroxidase-labeled goat anti-rabbit IgG (1:10,000, no. A0208, Beyotime, Shanghai, China) and horseradish peroxidase-labeled goat anti-mouse IgG (1:10,000, no. A0216; Beyotime) for 1 h at room temperature. The blots were visualized using an ultrasensitive enhanced chemiluminescence device (ECL, Thermo Fisher Scientific, Waltham, MA, United States). Band intensities were quantified using FluorChemHD2 software (ProteinSimple, San Jose, CA, United States).

Total RNA was extracted using TRIzol reagent (No. 15596026; Invitrogen; Thermo Fisher, Waltham, MA, United States) according to the manufacturer’s protocol. TRIzol (1 ml) was added to a microtube containing GC-2spd cells and vortexed vigorously. Then, the samples were incubated at room temperature for 5 min. Next, 200 μL of chloroform was added to the samples, vortexed for 15 s, incubated for 10 min at 20–22°C, and then centrifuged at 12,000 × g for 15 min at 4°C. The aqueous phase was collected in a microtube and mixed with an equal volume of isopropanol. After gentle mixing and incubation at −80°C for 30 min, the mixture was centrifuged at 12,000 × g for 15 min at 4°C. The final RNA pellets were resuspended in RNase-free water for the determination of RNA concentrations using a UV spectrometer.

Total RNA (3 μg) was separated by 7 M urea-denatured 15% PAGE and stained with the SYBR Gold solution (Invitrogen; Thermo Fisher, Waltham, MA, United States). Small RNAs with sizes of 17–50 nt and 80 nt were excised from the gel and extracted as previously described (Chen et al., 2016) for the quantification of RNA modification and for RNA-seq analysis.

Purified RNA (50–100 ng) was transferred to a microtube and incubated with 5 U/ul alkaline phosphatase (no. P5521, Sigma Chemica; Merck KGaA, Darmstadt, Germany), 0.01 U/ul phosphodiesterase I (no. P3134; Sigma; Merck KGaA, Darmstadt, Germany), and 5 U/ul benzonase (No. E8263; Sigma; Merck KGaA, Darmstadt, Germany) at 37°C for 3 h. Then, 3 K devices with an omega membrane (Pall Corporation, New York, NY, United States) were used to remove the enzymes from the mixture by centrifugation. Mass spectrometric analysis was performed using a Xevo-TQ-S mass spectrometer connected to an Acquity-UPLC I-class system (Waters Corporation, Milford, MA, United States) equipped with an electrospray ionization source. Except for pseudouridine which was analyzed in the negative ion mode, for the analysis of the other molecules, the MS system was operated in the positive ion mode using a multiple reaction monitoring (MRM) scan model. LC-MS/MS data were processed using MassLynx software (version 4.1). The percentage of each modified ribonucleoside was normalized to the total amount of quantified ribonucleosides with the same nucleobase. For example, the percentage of m⁶A was calculated as the molar concentration of m⁶A/the molar concentrations of m⁶A+ Am + A+ m¹A+ I + Im + m¹I.

Purified tRNA molecules (80 nt RNA fragments) were incubated at 37°C for 15 min in RPMI 1640-base systems: 15 ng of RNA in 9 μL of RPMI 1640 (HyClone, Fisher Scientific International Inc., Logan, UT, United States), 9 μL of RPMI 1640 containing 0.8% serum (FBS, Clark Bioscience, Virginia, United States), or 9 μL of RPMI 1640 containing 0.0003 μL of RNase A/T1 (2 mg/ml of RNase A and 5,000 U/ml of RNase T1). After incubation, the samples were immediately placed on ice, and RNA loading dye (2×, New England Biolabs Inc., Ipswich, United Kingdom) was added to them. A volume of four microliters of the samples was run on a 15% native PAGE gel at 4°C. The gels were stained with SYBR Gold before imaging.

Northern blotting was performed as previously described (Chen et al., 2016; Zhang J. et al., 2018). In brief, RNA extracted from GC-2spd cells was separated by 15% urea-PAGE, transferred onto nylon membranes (Roche, Basel, Switzerland), and cross-linked by UV at an energy level of 0.12 J. Then, the membranes were pre-hybridized with DIG pre-hybridization reagent (DIG Easy Hyb Granules, Roche, Basel, Switzerland) for at least 1 h at 42°C. For tRNA-Gly, tRNA-Ala, and 5.8S-rRNA detection, the membranes were incubated overnight at 42°C with 20 µM DIG-labeled oligonucleotide probes synthesized by Takara (Takara Bio Inc., Dalian, China), listed as follows:

tRNA-Gly: 5′-AAT​TCT​ACC​ACT​GAA​CCA​CCC​ATG​C-3’;

The membranes were sequentially washed with low-stringent buffer (2 × SSC with 0.1% (w/v) SDS), high-stringent buffer (0.1×SSC with 0.1% (w/v) SDS), and washing buffer (1×SSC) at 42°C. After washing, the membranes were incubated in 1 × blocking buffer (Roche, Basel, Switzerland) at room temperature for 3 h, and then anti-DIG antibodies (Anti-Digoxigenin-APFab fragments, Roche, Basel, Switzerland) were added to the blocking buffer at a ratio of 1:10,000 and incubated for an additional 1 h. The membranes were washed four times with DIG washing buffer (1× maleic acid buffer and 0.3% Tween-20) for 15 min each time, rinsed with DIG detection buffer (0.1 M Tris–HCl and 0.1M NaCl, PH:9.5) for 5 min, and then incubated with the CSPD reagent (Roche, Basel, Switzerland) in the dark for 20 min at 37°C before being imaged using a ChemiDoc imaging system (Bio-Rad, Hercules, CA, United States).

For each RNA library, more than 10 million reads (raw data) were generated using an Illumina HiSeq 2000 device (Bgi Technology Co., Ltd., Shenzhen, China). Small RNA library construction, library quality validation, and RNA sequencing were performed by Personalbio small RNA service (Shanghai, China), as described in our previous article (Zhang Y. et al., 2018). Clean reads were obtained and analyzed using SPOTS1.1 (small non-coding RNA annotation pipeline optimized for rRNA- and tRNA-derived small RNAs: https://github.com/junchaoshi/sports1.1) (Shi et al., 2018). Expression levels were normalized to reads per million (RPM).

Transcriptome sequencing was performed by Personalbio (Shanghai, China). In total RNA, mRNA with a poly-A structure was enriched using oligo (DT) magnetic beads. Then, the mRNA was randomly interrupted, connectors were added to it, and cDNA fragments were synthesized by reverse transcription, using RNA as a template for library construction. Finally, PCR amplification was performed to enrich library fragments. After RNA extraction, purification, and library construction, the samples were subjected to double pair-end (PE) sequencing using the Illumina NovaSeq sequencing platform.

The raw sequencing data were filtered using Cutadapt (von Hoyningen-Huene et al., 2019); then, high-quality sequence fragments were spliced using HISAT2 (Gangaraj and Rajesh 2020) and quantified using StringTie (Pertea et al., 2015). Differentially expressed genes were screened using the edgeR function in the R package (Yang et al., 2019); GO enrichment analysis was performed using the clusterProfiler function in the R package (Yu et al., 2012); GSEA enrichment analysis was performed using GSEA software (Subramanian et al., 2005). Gene sets were considered statistically enriched if the normalized p-value was less than 0.05.

Data are presented as mean ± SEM, and each experiment was repeated at least three times. Statistical analyses were performed using GraphPad Prism 8 (GraphPad Software, San Diego, CA, United States). Data on proliferation, apoptosis, and RNA modifications in GC-2spd cells were analyzed by ordinary one-way ANOVA multiple comparisons or two-way ANOVA multiple comparisons; multiple comparisons were performed using Sidak’s or Dunnett’s test. Differences were considered statistically significant at p < 0.05.

In our hypoxic GC-2spd cell model (Figure 1A), as compared to the normoxic group, there was a significant increase in HIF-1α expression levels in both the hypoxia and CoCl₂ treatment groups (Figure 1B), indicating that the cells were in a hypoxic state. There was a decrease in the number of cells counted at 12, 24, 48, and 72 h following seeding (Figure 1C). At 72 h, the number of cells in the hypoxia and CoCl₂ treatment groups was less than half of that in the normoxic group (Figure 1D). The percentages of apoptotic and non-apoptotic dead cells in the normoxic, hypoxia-treated, and CoCl₂-treated groups were 5.96, 12.02, and 6.63, and 4.61, 9.15, and 30.63%, respectively. The high cell death rate in the CoCl₂-treated group indicated the potential chemical toxicity of CoCl₂ in GC-2spd cells (Figures 1E,F). To determine whether RNA modification was involved in cellular responses to hypoxic exposure, the total RNA modification profiles of GC-2spd cells were determined using our previously developed high-throughput LC-MS/MS-based RNA modification detection platform (Chen et al., 2016; Zhang J. et al., 2018) (Figure 1G). In total, 19 types of RNA modifications were detected and quantified, of which the levels of two (Im and m² ₂ ⁷G) were significantly increased in GC-2spd cells (Figures 1H,I); the levels of the majority of RNA modifications were unchanged (Figures 1J-Z).

To further confirm the presence of alterations in tRNA and sncRNA modification in GC-2spd cells subjected to hypoxic treatment, the RNA modification profiles of RNA fragments of different sizes were determined by LC-MS/MS. Total RNA was first extracted and isolated by denatured PAGE, and then 80 nt-sized RNA molecules (most of them were tRNAs) and 17–50 nt-sized RNA molecules (sncRNA) were excised and purified for the identification of RNA modification signatures (Figure 2A). The levels of seven types of RNA modifications were altered in 80 nt-sized RNA fragments, with Um, m⁵C, m¹I, and m⁵U being significantly upregulated, and the levels of m³C, m¹G, and m² ₂G being significantly decreased following hypoxia and CoCl₂ treatment (Figures 2B–H). Aside from these seven altered RNA modifications, the levels of 10 more types of RNA modifications in 80 nt-sized RNA molecules were unchanged (Figures 2I-R). It has been established that RNA modifications are closely associated with the structure and stability of tRNAs; therefore, tRNA stability was further evaluated by evaluating its resilience to RNase degradation in native PAGE gels. tRNA stability in the hypoxia and CoCl₂ treatment groups was found to be significantly different from that in the normoxic group, as determined using RNase A/T1 and 20% FBS containing a unique combination of RNases (Figure 2S). Furthermore, the expression levels of representative tRNA, tRNA-Gly, and tRNA-Ala molecules were found to significantly decrease under hypoxic and CoCl₂ treatment conditions (Figures 2T–V).

The RNA modification signatures of 80 nt-sized RNA molecules (most of which were tRNAs) were previously found to be altered and this affected tRNA stability in GC-2spd cells subjected to hypoxic stress. We thought it necessary to further investigate modifications and the expression levels of small RNAs (17–50 nt). The levels of nine types of RNA modifications (Cm, m³C, m⁵C, Gm, m¹G, m²G, m² ₂G, m⁷G, and m¹A) were significantly increased in treated GC-2spd cells (Figures 3A–I), while those of the other five types were unchanged (Figure 3J-N). High-throughput small RNA sequencing revealed alterations in the expression levels of 138 sncRNAs in cells subjected to hypoxic treatment (Figure 3O). More specifically, the expression levels of 18 miRNAs, including Let-7d, mir-145a, and mir-210, significantly increased, and the levels of 29 miRNAs, including mir-125b, mir-25, and mir-365, significantly decreased (Figure 3P). The levels of 15 tsRNAs (tRNA-derived small RNAs) were upregulated and those of 30 tsRNAs were downregulated, and this may partly have been due to RNA modification-induced alterations in tRNA stability (Figure 3Q). rRNA-derived small non-coding RNAs (rsRNAs) are newly identified small non-coding RNAs (Chu et al., 2017) that play important roles in paternally acquired epigenetic inheritance, sperm rapid response to metabolic alterations (Natt et al., 2019; Natt and Ost 2020), and organ inflammation (Chu et al., 2017). In this study, we also identified a large number of rsRNAs from 4.5, 5, 5.8, 12, 16, 18, 28, and 45S rRNAs. The most enriched rsRNAs were derived from 18, 28, and 45S rRNAs (Figure 3R). Under hypoxic conditions, there was an increase in the levels of most rsRNAs derived from 28, 5.8, and 5S rRNA, while the levels of those derived from 4.5S rRNA decreased, indicating that rRNA might be sensitive to hypoxic stress (Figure 3S). Moreover, the levels of YRNA-derived small RNAs (ysRNAs) were found to increase under cellular hypoxic stress conditions, especially those of small RNAs derived from RNY-1. The locations of ysRNAs on RNY-1 were further evaluated, and most ysRNAs^RNY1 were found to be derived from the 3′ end of RNY1. However, the levels of ysRNAs^RNY1 derived from both the 5′ and 3’ ends of RNY1 increased under hypoxic conditions (Figures 3T–U).

As previously reported (Guzzi et al., 2018; Shi et al., 2019), modifications in sncRNAs, such as tsRNAs, are involved in the regulation of gene transcription. To evaluate the effects of RNA modifications on GC-2spd cell growth under hypoxic conditions via the regulation of gene transcription, transcriptomic sequencing of GC-2spd cells was conducted. A total of 1,130 genes were found to be upregulated (353 genes) and downregulated (777 genes) in the hypoxia-treated group (Figure 4A). Biological pathway analysis revealed that the dysregulated genes were enriched in RNA and nucleotide metabolic pathways. Cellular component analysis revealed that the differentially expressed genes were mainly located in the mitochondrial outer membrane, organelle membrane, and the transferase complex; molecular pathway analysis revealed that these altered genes were involved in RNA binding and acted on RNA and other molecular processes (Figure 4B). Gene set enrichment analysis showed that these differentially expressed genes were mainly enriched in aerobic respiratory pathways, including hypoxic, glycolytic, and apoptotic pathways (Figure 4C). Furthermore, RNA modification writers and erasers were also found to be regulated in the treatment groups as compared to the normal group, and this might be the direct reason why RNA modification levels were affected (Figure 4D).