Acute Hypoxia Stress-Induced Apoptosis in Gill of Japanese Flounder (Paralichthys olivaceus) by Modulating the Epas1/Bad Pathway

Simple Summary Hypoxia is perhaps of the most pressing global environmental challenge in aquatic systems. Hypoxic stress can cause physiological changes in fish, resulting in injury and even death to them. Japanese flounder is a crucial benthic economic fish, but the mechanisms of its response to hypoxia are poorly understood. Here, we show that Japanese flounder can alter gill morphology, and induce apoptosis to weaken the risk caused by hypoxic stress. The mechanism of the EPAS1/Bad signal pathway in response to hypoxia was investigated from the perspective of transcription and epigenetics. Therefore, our results would enrich the theory of hypoxic stress on Japanese flounder and provide a reference for its healthy cultivation. Abstract The physiological responses and molecular mechanisms of apoptosis in Japanese flounder under hypoxic stress remain unclear. In the present study, we performed acute hypoxia stress on Japanese flounder (2.39 ± 0.84 mg/L) and detected gills responses in histomorphology and molecular mechanisms. The results showed that the volume of the interlamellar cell mass decreased and the gill lamellae prolonged, indicating the expansion of the respiratory surface area. Additionally, the fluorescence signal of apoptosis increased under hypoxic stress. In addition, the expression of two genes (EPAS1 and Bad) related to apoptosis increased about four-fold and two-fold, respectively, at 6 h of hypoxia. Meanwhile, the result of the dual-luciferase reporter assay showed that EPAS1 is a transcription factor, which could regulate (p < 0.05) the expression of the Bad gene, and we identified the binding site of EPAS1 was the AATGGAAAC sequence located near −766. DNA methylation assay showed that hypoxia affected the methylation status of CpG islands of EPAS1 and Bad genes. All results indicated that hypoxia could activate the EPAS1/Bad signal pathway to induce gill apoptosis of Japanese flounder. Our study provides new light on understanding the molecular mechanism of hypoxia-induced apoptosis in Japanese flounder.


Introduction
Hypoxia influences an ever-increasing number of marine organisms as a result of the combined effects of increased global warming and higher nutrient enrichment of coastal waters [1]. Organisms in the marine ecosystem living in anoxic environments, especially fish, are facing unprecedented challenges in development, physiology, and pathology [2]. Fish gills acted as an important role in physiological functions including gaseous exchange, osmoregulation, acid-base regulation, excretion of nitrogenous waste, and hormone production [3,4]. Gills are also the main organs that respond to the change in dissolved oxygen in the aquatic environment [5]. It was reported that the morphological structures of gills in many fish showed significant changes during hypoxia [6]. Hypoxiainduced gill remodeling may be related to changes in the expression levels of apoptosisrelated genes [7,8]. Therefore, hypoxia can induce gill morphology changes and cell apoptosis in fish gills.

Animal Housing and Hypoxia Stress
The Japanese flounder used in this study were obtained from Qingdao HaoRuiYuan aquaculture Co., Ltd., China. The fish were reared and sampled at Lab of Fish Breeding Physiology and Seed Engineering (Qingdao, China). Firstly, 60 individuals were temporarily housed in three tanks (1.5 m × 1.5 m × 0.6 m) in a recirculating seawater system for 7 days. The water temperature was nearly 16.09 ± 0.89 • C, pH was 7.0-7.5, and dissolved oxygen was 9.48 ± 0.42 mg/L. The photoperiod was 14 h light and 10 h dark. All fish were fed on Microbound Diet (SHENGSUO, Yantai, China) twice a day (8:30-17:30). The contents of crude protein, crude fat, crude ash, crude fiber, calcium, total phosphorus, and lysine in the Microbound Diet are 50%, 7%, 17%, 4%, 5%, 1.2%, and 2.5%, respectively. We removed residual feed and feces 30 minutes after feeding. All fish fasted for three days before the hypoxic treatment.
After all fish fasted for three days, we regulated the water hypoxia (2.39 ± 0.84 mg/L) by pumping nitrogen into the water using nitrogen cylinders. During this period, dissolved oxygen meter was used to monitor dissolved oxygen in water. In the hypoxic treatment, temperature, pH, and photoperiod were the same as in the temporary culture stage. We collected the control group under normal oxygen and the treatment group under hypoxia. The concentration of dissolved oxygen in the treatment group is 2 mg/L, which is more than half the dissolved oxygen concentration [29]. Then, gills of 6 fish were collected after 0 h (control group), 1 h, 3 h, 6 h, 12 h, and 24 h of hypoxia treatment. In detail, firstly, we used MS222 (100 mg/L) to anesthetize Japanese flounder. Secondly, we measured the length (14.92 ± 0.98 cm) and weight (66.59 ± 1.50 g) of Japanese flounder. Then, we dissected the fish immediately, and cut the first pair of gills on the left for branchial histological structure analysis of gills, and the second pair of gills on the left for apoptosis analysis. Both samples were placed in 4% paraformaldehyde. Finally, we collected the remaining two pairs of gills on the left of the fish as samples for DNA extraction, and all four pairs of gills on the right were collected as samples for RNA extraction. The samples for DNA and RNA extraction were placed in liquid nitrogen for quick freezing and then stored in −80 • C freezer.

Apoptosis Detection by TUNEL
For the detection of apoptosis, TUNEL staining was conducted using the TransDetect ® In Situ Fluorescein TUNEL Cell Apoptosis Detection Kit (Trans, Beijing, China). The gills were sectioned as described in the 2.2 branchial histological structure. Following the instruction, the paraffin sections were dehydrated in an ethanol series (95%, 90%, 85%, 75%, and 50%), permeabilized, labeled, and photographed under a fluorescence microscope (ECHO RVL-100-G, San Diego, CA, USA).

Phylogenetic Analysis
The amino acid sequences of EPAS1 and Bad gene were obtained from the National Center for Biotechnology Information (NCBI) (https://www.ncbi.nlm.nih.gov/, accessed on 28 August 2022). We used the neighbor-joining method (based on 1000 bootstrap replicates) to construct the phylogenetic trees of EPAS1 and Bad genes. Evolutionary distances were estimated according to the model of Poisson correction method. In the Biology 2022, 11, 1656 4 of 16 evolutionary tree of EPAS1, we used two vertebrates, Gallus gallus and Homo sapiens, as outgroups. In the evolutionary tree of Bad, we chose Homo sapiens, Capra hircus, Sus scrofa, Chelonia mydas, and Xenopus tropicalis as the outgroup.

RNA Extraction and Gene Expression
Total RNA of gill samples was obtained using Total RNA Extraction Reagent (Vazyme, Nanjing, China). The integrity of total RNA was determined by agarose gel electrophoresis (1.2%), and the concentration of total RNA was assessed with spectrophotometric analysis (A260:280 nm ratio). The cDNA synthesis was performed by HiScript III RT SuperMix for qPCR (+gDNA wiper) (Vazyme, Nanjing, China). Three biological replicates were performed at each time point (0 h, 1 h, 3 h, 6 h, 12 h, and 24 h) under three technical replicates. For qPCR, we used ChamQ™ SYBR ® Color qPCR Master Mix (Vazyme, Nanjing, China). Moreover, qPCR used an Applied Biosystems StepOne Plus Real-Time PCR System (Applied Biosystems, Waltham, MA, USA) to complete quantification. The specific primers of EPAS1 and Bad genes were designed by Primer Premier 5 and listed in Table S1. The primer of 18S (as a reference gene) refers to Huang et al., 2019 [30]. The reaction system and procedure of quantitative real-time PCR are listed in Tables S2 and S3. The melt curves of primers for three genes EPAS1, Bad, and 18S in qPCR were shown in Figure S1. The target gene expression was calculated with the 2 −∆∆Ct method [31].

Dual-Luciferase Reporter Assay
We used TF (transcription factors) prediction website HumanTFDB (http://bioinfo. life.hust.edu.cn/, accessed on 28 August 2022) to analyze the promoter of Bad gene and found that EPAS1 is one of its transcription factors ( Figure S2). The expression plasmid (pc3.1~EPAS1) and reporter plasmid (pGL~Bad) were constructed by dual-luciferase reporter assay, and we verified the targeting relationship of EPAS1 and Bad. The gills' genomic DNA was extracted and a 2013-bp fragment of the 5' flanking region of Bad gene was isolated by PCR using 2 × Phanta Max Master Mix (Dye Plus) (Vazyme, Nanjing, China). The PCR product was purified and double digested by KpnI (NEB, Ipswich, MA, USA) and HindIII (NEB, Ipswich, MA, USA) before being inserted into pGL3-Basic plasmid by homologous recombination and this plasmid (pGL~Bad) was transferred into DH5α cells (Vazyme, Nanjing, China). Similarly, the EPAS1 coding sequence (CDS) was obtained by high-fidelity PCR. Then, double digested by BamHI (NEB, Ipswich, MA, USA) and Xhol (NEB, Ipswich, MA, USA) before being inserted into pcDNA 3.1(+) plasmid by homologous recombination, and the plasmid (pc3.1~EPAS1) was transferred into DH5α cells. The reaction system and procedure of the double digestion are listed in Table S4. After EPAS1 and Bad sequences were successfully cloned and sequenced (Sangon Biotech, Shanghai, China), plasmids were extracted by using EndoFree Mini Plasmid Kit II (TIANGEN, Beijing, China). Then, we transfected the expression plasmid (pc3.1~EPAS1) and reporter plasmid (pGL~Bad) into human embryonic kidney 293T (HEK293T) cells by Lipofectamine™ 3000 Reagent (Invitrogen, Waltham, MA, USA). The Renilla luciferase plasmid (pRL-TK) (Promega, Madison, WI, USA) could express Ranilla luciferase. So, pRL-TK plasmid (as a control) was used to quantify the transfection efficiency. Transfection steps were as follows: First, seeded cells were 50-70% confluent at transfection. Second, Lipofectamine™ 3000 was diluted using DMEM/High Glucose culture medium (Servicebio, Wuhan, China). Third, the plasmids were diluted using DMEM/High Glucose culture medium, and the mixture was mixed by adding P3000™ Reagent. Fourth, the second and third products were mixed, and the mixture was incubated for 15 minutes. Fifth, the fourth mixture was added to the cells. Sixth, the transfected cells were analyzed after incubation at 37 • C for 2 days.
HEK293T cells were cultured in DMEM/High Glucose culture medium containing 10% fetal bovine serum (FBS) (absin, Shanghai, China) at 37 • C. HEK293T cells were seeded in 24 well plates (Corning, New York, NY, USA) at a density of 1 × 10 5 cells per well and incubated at 37 • C. When the confluence of cells reached 50% to 70%, plasmids were transfected into HEK293T cells. Three duplicate wells were set for each group. In the end, double-fluorescence values were detected using SYNERGY HTX multi-mode reader (BioTek, Winooski, VT, USA) and Dual-Luciferase kit (Promega, Madison, WI, USA) after transfection for 2 days (48 h).
To determine the specific binding site of EPAS1, we used fragment deletion to narrow the range of binding sites, and then use bases mutation to determine the binding sites. The binding sites of EPAS1 transcription factor on Bad promoter sequence were predicted by the transcription factor prediction website. The method of constructing plasmids is similar to that of constructing pGL~Bad plasmids. Cell culture, transfection, and determination of double-fluorescence values are the same as the above methods. Fragment deletion plasmids were constructed including pGL~Bf1, pGL~Bf2, pGL~Bf3, pGL~Bf4, and pGL~Bf5. Then, pGL~Bmp3, pGL~Bm3, pGL~Bmp5, and pGL~Bm5 were directed bases mutation reporter plasmids. Fragment deletion and bases mutation details were found in 3.5. EPAS1 transcriptionally regulated Bad. The specific primers were designed by Primer Premier 5 and listed in Table S5. The reaction system and procedure of the PCR are listed in Tables S6 and S7. In particular, inverse PCR was used for base mutant plasmids.

Detection of DNA Methylation Level
Due to the EPAS1 and Bad gene expression, four groups (0 h, 1 h, 6 h, and 24 h) were chosen for their promoter methylation analysis. The genomic DNA of the gill was isolated using FastPure ® Cell/Tissue DNA Isolation Mini Kit (Vazyme, Nanjing, China). An amount of 1µg of genomic DNA was modified by EpiArt™ DNA Methylation Bisulfite Kit (Vazyme, Nanjing, China). Methylation-specific PCR was conducted by using 2 × EpiArt™ HS Taq Master Mix Kit (Vazyme, Nanjing, China). The prediction of CpG islands and the design of primers were completed by MethPrimer design program (http://www.urogene.org/ methprimer/, accessed on 28 August 2022). MS-PCR products of EPAS1 and Bad were cloned into the pCE2 TA/Blunt-Zero Vector (Vazyme, Nanjing, China) to construct plasmids and then the plasmids were transferred into DH5α cells. Three biological replicates were performed in each group and the sequence of 10 clones for each individual has been performed using M13 forward and reverse primers. The formula [30] used to calculate the bisulfite modification efficiency was listed here: (the number of converted cytosines (except cytosines of CpG dinucleotides)/total number of cytosines (except cytosines of CpG dinucleotides) × 100). The specific primers were designed by Primer Premier 5 and listed in Table S8.

Statistical Analysis
All data were represented as means ± standard error. All data were analyzed by one-way analysis of variance (ANOVA) using SPSS 22.0 (SPSS, Chicago, IL, USA), followed by Duncan's post hoc test. Statistical significance was chosen at p < 0.05.

Effects of Hypoxia Stress on Gill Morphology
In the hypoxia treatment, most Japanese flounder were inactive and had difficulty in breathing. Some fish swam to the surface at intervals to take in oxygen and slowly swam back to the bottom of the tank. Thus, the gills, as respiratory organs, were first examined for morphological changes. HE staining of gill results are illustrated in Figure 1. The gill was neatly arranged at 0 h. With the prolongation of hypoxia time, the volume of the interlamellar cell mass (ILCM) decreased and the gill lamellae elongated, which increased the area of gill lamellae contact with water. Additionally, the gills became curved after 6 h and 24 h of hypoxia. The exposure to hypoxia triggered morphology change in the gill.
breathing. Some fish swam to the surface at intervals to take in oxygen and slowly swam back to the bottom of the tank. Thus, the gills, as respiratory organs, were first examined for morphological changes. HE staining of gill results are illustrated in Figure 1. The gill was neatly arranged at 0 h. With the prolongation of hypoxia time, the volume of the interlamellar cell mass (ILCM) decreased and the gill lamellae elongated, which increased the area of gill lamellae contact with water. Additionally, the gills became curved after 6 h and 24 h of hypoxia. The exposure to hypoxia triggered morphology change in the gill.

Effects of Hypoxia Stress on Gill Apoptosis
There was almost no TUNEL apoptotic signal in the 0 h group. However, the TUNEL apoptotic signals increased with the extension of hypoxia time. In detail, the apoptotic signals began to appear at 1 h of hypoxia and were very dense at 6 h of hypoxia ( Figure  2).

Effects of Hypoxia Stress on Gill Apoptosis
There was almost no TUNEL apoptotic signal in the 0 h group. However, the TUNEL apoptotic signals increased with the extension of hypoxia time. In detail, the apoptotic signals began to appear at 1 h of hypoxia and were very dense at 6 h of hypoxia ( Figure 2).

Phylogenetic Analysis
The evolutionary trees of Bad and EPAS1 were constructed to understand the evolutionary relationship. The results showed that the phylogenetic tree of Bad was divided into two branches, fish and other advanced vertebrates ( Figure 3A). Similarly, the phylogenetic tree of EPAS1 was mainly divided into two branches ( Figure 3B).

EPAS1 and Bad Gene mRNA Relative Expression
The results of relative gene expression showed that EPAS1 and Bad mRNA first increased and then decreased ( Figure 4A,B). Specifically, EPAS1 expression was about twofold (p = 0.01) and four-fold (p = 0.00) higher at 1 h and 6 h of hypoxia than at 0 h, respectively, then decreased to near 0 h level. The mRNA expression level of Bad was similar to EPAS1, which also increased about two-fold (p = 0.00) at 6 h of hypoxia, and then decreased gradually.

EPAS1 Transcriptionally Regulated Bad
There is a positive correlation (R = 0.741, R 2 = 0.5495, Figure 5A) between the mRNA relative expressions of EPAS1 and Bad using the linear regression analysis. As shown in Figure 5B, the relative luciferase activities were upregulated about two-fold (p = 0.00) when the cotransfection pGL3~Bad and pc3.1~EPAS1 into HEK293T cells. The assay of luciferase activity showed that the transcription of pGL3~Bad was upregulated by pc3.1~EPAS1. The method of fragment deletion was used to detect the specific binding sites of transcription factor EPAS1 ( Figure 5C). The results showed that the first significant (p < 0.05) difference occurred between Bf3 and Bf4, and the fluorescence value of Bf3 is about 99-fold higher than that of the pGL group. The second significant (p < 0.05) difference occurred between Bf5 and pGL and the fluorescence value of Bf5 is about 70-fold higher than that of the pGL group. These results indicated that the EPAS1 binding sites were located between −830~−679 or −529~−1. Therefore, the two predicted binding sequences were mutated. The relative luciferase activity of the first mutant plasmid was about 21-fold lower (p = 0.00) than that of the Bf3 group ( Figure 5D). However, the relative luciferase activity of the second mutant plasmid was not obviously different from that of the control plasmid ( Figure 5E).

Phylogenetic Analysis
The evolutionary trees of Bad and EPAS1 were constructed to understand the evolutionary relationship. The results showed that the phylogenetic tree of Bad was divided into two branches, fish and other advanced vertebrates ( Figure 3A). Similarly, the phylogenetic tree of EPAS1 was mainly divided into two branches ( Figure 3B).    Table S9.
(B) EPAS1 phylogenetic analysis. The accession numbers of EPAS1 protein are listed in Table S10. The numbers in the branches represent the percentage of the associated taxa clustered together in the bootstrap test (1000 replicates).

EPAS1 and Bad Gene mRNA Relative Expression
The results of relative gene expression showed that EPAS1 and Bad mRNA first increased and then decreased ( Figure 4A,B). Specifically, EPAS1 expression was about twofold (p = 0.01) and four-fold (p = 0.00) higher at 1 h and 6 h of hypoxia than at 0 h, respectively, then decreased to near 0 h level. The mRNA expression level of Bad was similar to EPAS1, which also increased about two-fold (p = 0.00) at 6 h of hypoxia, and then decreased gradually.  Table S9. (B) EPAS1 phylogenetic analysis. The accession numbers of EPAS1 protein are listed in Table S10. The numbers in the branches represent the percentage of the associated taxa clustered together in the bootstrap test (1000 replicates).

DNA Methylation Levels of EPAS1 and Bad Gene
The methylation status of EPAS1 promoter and Bad promoter are shown in Figure S3. There are 8 CpG dinucleotides per 193 bp in EPAS1. The results showed that the total methylation levels were 87.01% ± 0.22% ( Figure 6A,B). Linear regression analysis was conducted between methylation levels and mRNA expression levels of EPAS1, but there was no relationship between them (R = 0.088, R 2 = 0.008; Figure 6C).

DNA Methylation Levels of EPAS1 and Bad Gene
The methylation status of EPAS1 promoter and Bad promoter are shown in Figure  S3. There are 8 CpG dinucleotides per 193 bp in EPAS1. The results showed that the total methylation levels were 87.01% ± 0.22% ( Figure 6A,B). Linear regression analysis was conducted between methylation levels and mRNA expression levels of EPAS1, but there was no relationship between them (R =0.088, R 2 = 0.008; Figure 6C). There are 42 CpG dinucleotides per 400 bp in Bad. The total methylation level of the Bad gene displayed a lower level (2.62% ± 0.23%) compared to the EPAS1 gene. Compared with the 0 h group, the change trends first increased about two-fold (p = 0.01) at 1 h after hypoxia and then gradually decreased with the prolongation of hypoxia stress ( Figure 7A,B). This methylation change was mainly concentrated at the 19th (−816 site) and 30th (−718 site) CpG sites. Furthermore, linear regression analysis was conducted between methylation levels and mRNA relative expression levels of Bad, but there was no relationship between them (R = 0.055, R 2 = 0.003) ( Figure 7C). Bad gene displayed a lower level (2.62% ± 0.23%) compared to the EPAS1 gene. Compa with the 0 h group, the change trends first increased about two-fold (p = 0.01) at 1 h a hypoxia and then gradually decreased with the prolongation of hypoxia stress ( Fig  7A,B). This methylation change was mainly concentrated at the 19th (−816 site) and 3 (−718 site) CpG sites. Furthermore, linear regression analysis was conducted betw methylation levels and mRNA relative expression levels of Bad, but there was no relati ship between them (R = 0.055, R 2 = 0.003) ( Figure 7C).

Discussion
Hypoxia is a serious problem in aquaculture, which affects the survival of aquatic animals. Hypoxia exposure caused apoptosis of ovarian follicle cells in Atlantic croaker [32], the central system of the sturgeon of sturgeon (Acipenser shrenckii) [33], skeletal muscle in Japanese flounder(Paralichthys olivaceus) [27], and gills in blunt snout bream (Megalobrama amblycephala) [8]. In all these studies, hypoxia seriously affected fish survival and development and caused severe apoptosis. In this study, the gills of Japanese flounder were curved at 6 h and 24 h of hypoxia, and the ILCM volume decreased after hypoxia, which increased the contact area between gill lamellae and water. At the same time, the TUNEL apoptosis signals almost began to increase with hypoxia, and the signals were very dense at 6 h of hypoxia. However, there were few TUNEL apoptotic signals in the 0 h group. These results indicated that Japanese flounder would have a significant apoptotic response to acute hypoxia stress, with a reduction of the ILCM and elongation of gill lamellae, which indicates the expansion of the respiratory surface area to adapt to a more severe hypoxia environment.
Bad is very essential for apoptosis [34]. We analyzed the mRNA expression of the Bad gene in the gills of Japanese flounder under hypoxic stress, the results indicated that the expression level of Bad was upregulated about two-fold after 6 h of hypoxia compared with 0 h of control. Therefore, the upregulated expression of Bad may influence the apoptosis of Japanese flounder gills against hypoxic stress. Similar to our study, the expression of Bad in blunt snout bream was significantly upregulated after 4 and 7 days of hypoxia, and the apoptosis rate of gill cells was also significantly increased [8]. Especially, EPAS1 showed similar changes in gene expression that the expression level was first increased about four-fold at 6 h after hypoxia and then decreased. Meanwhile, the expression of EPAS1 and Bad genes was analyzed by linear regression, and the results showed that their expressions were positively correlated. Their similar gene expression suggests that they may be important in hypoxia-induced apoptosis. Studies have shown that the gills of crucian carp showed protruding lamella during hypoxia, which may be caused by the combination of reduced cell proliferation and increased apoptosis induction. Therefore, the increased expression of the Bad gene may lead to an increase in apoptosis, which has been confirmed in the TUNEL experiment. These results may be the reason for the increase in gill respiratory surface area.
Hypoxia may cause apoptosis. HIF-1α, a relative of HIF-2α/EPAS1, could activate the expression of Nip3 (member of the Bcl-2 family) in response to hypoxia because of the functional HIF-1-responsive element (HRE) in the Nip3 promoter [35]. It is speculated that EPAS1 may be directly related to the promoter of Bad. Here, the results of the dualluciferase reporter assay indicated that Bad is regulated by EPAS1 ( Figure 5B). EPAS1 could activate the Bad gene transcription. Dual-luciferase reporter assay was used to determine the specific binding sites of EPAS1 on the Bad gene promoter. According to the fragment deletion results in Figure 5C, the EPAS1 binding sites were likely around −766 or −507. When the above sites were mutated, the fluorescence value of the first site decreased about 21-fold, while the fluorescence value of the second site was almost unchanged ( Figure 5D,E). Based on the results of the dual-luciferase reporter assay, we concluded that the binding site of EPAS1 was a sequence (AATGGAAAC) located near −766.
DNA methylation is essential for mammalian epigenetics [36]. Hypoxia can affect gene DNA methylation status. In the tumor, hypoxia acts as a regulator of DNA methylation [37]. Therefore, methylation-specific PCR [38] was used to detect the methylation status of EPAS1 and Bad gene promoters. However, the promoter of the EPAS1 gene was highly methylated at each treatment time. However, there was no relationship between the methylation status of EPAS1 and its expression ( Figure 6C). In addition, no CpG island was found in the proximal promoter region of the Bad gene, but there was a CpG island with 42 CG sites in the 5 noncoding region. The results of Bad methylation showed that hypoxia increased about two-fold in 1 h ( Figure 7B), indicating that hypoxia could increase the methylation level of the Bad 5 non-coding region. However, there was no correlation between this methylation and Bad expression ( Figure 7C). We speculated that the region currently detected is not an important site for DNA methylation, and the sites regulating the expression of EPAS1, and Bad genes may exist in the undetected CG sites [39]. In other words, the upregulation of EPAS1 and Bad genes may be related to the methylation degree of other regions of its promoter. Secondly, EPAS1 and Bad genes may be regulated by other genes on hypoxia-related pathways. These results suggest that the methylation of EPAS1 and Bad might not directly regulate the expression of their genes.
Hypoxia, as a cell death stimulus, caused apoptosis of gill in Japanese flounder. In response to hypoxic stress, gill tissue structure rapidly changed, followed by rapid upregulation of EPAS1 and Bad gene expression in response. Meanwhile, the apoptosis cells in the gill were increased after 6 h of hypoxia compared to 0 h of control. This experiment identified the binding site of EPAS1 on the Bad promoter and attempted to solve the regulatory relationship of this signal pathway from the epigenetic modification. Collectively, EPAS1/Bad signaling pathways are important for hypoxia-induced apoptosis. These findings reported here shed new light on a direct link between EPAS1 and a proapoptotic member of the Bcl-2 family and offer a reasonable apoptosis mechanism for hypoxia-stimulated Japanese flounder.

Conclusions
Acute hypoxia stress can activate the EPAS1/Bad signal pathway, by raising EPAS1 expression to promote the expression of Bad, leading to apoptosis of Japanese flounder gill. This signal pathway may not be directly regulated by DNA methylation. In conclusion, hypoxia induced gill apoptosis in Japanese flounder by acting on the hypoxic signal pathway EPAS1/Bad in transcriptional regulation. This study enriched the mechanism of hypoxia-induced gill apoptosis in Japanese flounder and provided a new idea for healthy fish culture.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/biology11111656/s1, Table S1: Primers used for q-PCR. Table S2: System used for q-PCR. Table S3: Procedure used for q-PCR. Table S4: System and procedure used for double digestion. Table S5: Primers of plasmid construction for dual-luciferase reporter assay. Table S6: System used for dual-luciferase reporter assay. Table S7: Procedure used for dual-luciferase reporter assay. Table S8: Primers of methylation status detecting in EPAS1 and Bad promoter by MS-PCR. Table S9: The accession numbers of Bad protein. Table S10: The accession numbers of EPAS1 protein. Figure S1: The melt curves of primers for three genes EPAS1 (A), Bad (B), 18S (C) in qPCR. Figure S2: The predicted binding site of EPAS1 on the Bad promoter. Figure   Data Availability Statement: Data used to support the findings from this study are available from the corresponding author upon request.