Introduction

Hypoxia-inducible factor 1 (HIF-1) is a crucial important transcriptional factor that can restore oxygen balance in the body by regulating the production of red blood cells, blood vessels, glycolysis, and apoptosis (Schofield and Ratcliffe 2004; Semenza 2014). Being a heterodimer composed of a specific α subunit and a ubiquitous β subunit, the transcriptional activity of HIF-1 predominantly depends on the stability of the α-subunit, which is functionally inhibited and degraded by prolyl hydroxylase enzymes under normoxia (Schofield and Ratcliffe 2004; Webb et al. 2009). Invertebrates possess only one HIF-α, while most jawed vertebrates possess the other two functional HIF-α isoforms, HIF-2α and HIF-3α, after having undergone two rounds of whole genome duplication (WGD) during evolution (Jiang et al. 2001; Loenarz et al. 2011; Schofield and Ratcliffe 2004). Having experienced a specific third round of WGD early in their evolution, teleosts generated two copies of each HIF-α isoform (HIF-αA/B). This was followed by the loss of one copy of each paralogous pair in most teleosts, except for cyprinids. Therefore, two HIF-1αs, HIF-1αA and HIF-1αB were retained in cyprinid fish (Rytkonen et al. 2011). (Rytkonen et al. 2013) The relaxation of negative selection on the oxygen sensing properties of the cyprinid hif-1αA coding sequence, coupled with the loss of developmental regulatory motifs in B paralogs, suggests that cyprinid paralogs may have undergone subfunctionalization, potentially serving distinct functions both in development and in responding to hypoxia (Rytkonen et al. 2013), including regulation of cellular functions like apoptosis and mitophagy.

Mitochondria play a central role in energy supply, reactive oxygen species (ROS) generation, and apoptosis and are thus a pivotal part of the cellular response to low oxygen. ROS is highly active and has irreversible effects on intracellular proteins, lipids, and DNA, causing cell damage that leads to more ROS production. Excessive ROS can inhibit the activity of PHD, an enzyme that targets and triggers degradation of HIF-1α, which leads to HIF-1α stabilization and activation, promoting HIF-mediated gene expression (Majmundar et al. 2010; Nathan and Cunningham-Bussel 2013; Sangbin et al. 2016; Zepeda et al. 2013). Meanwhile, HIF-1 can regulate the generation of ROS by unbalancing the electron flow in the Electron Transport Chain (Han et al. 2017; Kung-Chun Chiu et al. 2019). Accumulated ROS may not be quickly eliminated under hypoxia, which would exacerbate signals for autophagy (Ruth and Zvulun 2011).

Mitophagy, a cellular process that selectively removes damaged or superfluous mitochondria, is an essential process for mitochondrial quality control and dynamics (Lemasters 2005). Hypoxia is one of the key environmental stressors that can lead to mitophagy. Mitochondrial protein NIX (also known as BNIP3L) is a selective autophagy receptor that can be strongly induced by hypoxia (Zhang and Ney 2009). NIX promotes the loss of mitochondrial membrane potential (MMP) and enables the entry of mitochondria into autophagosomes for clearance (Sandoval et al. 2008). This process is also firmly established in hypoxia-induced cell death alongside BNIP3, which plays a critical role in mitochondrial depolarization and autophagy in numerous cell lines (Daido et al. 2004; Hamacher-Brady et al. 2007; Kanzawa et al. 2005). The transcriptional and translational mechanism of both NIX and BNIP3 are regulated by hypoxia and HIF-1α, and their expression is thus associated with tumor cell death (Sowter et al. 2001). The PINK1-Parkin pathway also mediates mitophagy to eliminate those mitochondria which have lost membrane potential in response to acute stress (Vincow et al. 2013). Parkin is a ubiquitin enzyme that labels damaged mitochondria by ubiquitinating outer mitochondrial membrane proteins (Tanaka et al. 2010; Yoshii et al. 2011), which are targeted by autophagy proteins ATG8 and LC3 to form autophagosomes and then subsequently degraded by lysosomes (Sylvie et al. 2017). The activation of the PINK1-Parkin pathway is hypoxia independent, but it is strongly affected by hypoxia and ROS (Liu et al. 2021; Wu et al. 2021; Xu et al. 2021; Zeb et al. 2021). FUNDC1, an outer mitochondrial membrane protein, is another mitochondrial receptor that mediates hypoxia-induced mitophagy. After being dephosphorylated, FUNDC1 induces the formation of autophagic vesicles by interacting with LC3 in response to hypoxia (Chen et al. 2014; Huifang et al. 2015; Liu et al. 2012b).

Although it has been demonstrated in mammals that HIF-1α is involved in ROS generation and hypoxia-induced mitophagy, how the specific “divergent collaboration” of HIF-1αA and HIF-1αB operates in regulating mitophagy induced by hypoxic stress in cyprinids remains unknown. To address this, we transfected zebrafish HIF-1αA and HIF-1αB into the EPC cells and investigated how they function in mitophagy regulation under induced hypoxic stress.

Material and Methods

Cell Culture and Viability Determination

Epithelioma papulosum cyprini (EPC) cells (obtained from the Center of Aquatic Cell Research and Preservation, Huazhong Agricultural University) were cultured in M199 medium (Gibco, Suzhou, China) with 10% fetal bovine serum (Hyclone, Logan, USA) and 1% Penicillin–Streptomycin (Invitrogen, Carlsbad, USA) at 28 °C. The EPC cells were passaged after reaching a minimum of 70% confluence to fresh ∼ 5 mL medium in 25 cm2 cell culture flasks (Sigma-Aldrich) with 3 replicates by a 1:2 ratio. Cell viability was detected using the CCK-8 cell counting kit (Beyotime, Shanghai, China). Cells were inoculated into 96-well plates and incubated under 20% O2, 10% O2, 3% O2, 1% O2, and 0.3% O2, respectively (InvivO2 300, Baker Ruskinn, Britain). Cell viability was measured every 6 h according to the manufacturer's instruction. The absorbance at 450 nm was detected by a multifunctional microplate reader SpectraMax M5 (Molecular Devices, Sunnyvale, CA, USA).

Plasmid Construction and Transfection

The indicated Myc-tagged Hif-1α constructs, pTK-Renilla luciferase reporter, HRE (hypoxia-responsive elements)-luciferase reporter, and pCMV-Myc vector were obtained from the Institute of Hydrobiology, Chinese Academy of Sciences. The full length of cDNA sequence of zebrafish HIF-1αA (NM_001308559.1) and HIF-1αB (NM_001310042.1) were subcloned into the pCMV-Myc vector (Clontech. Palo Alto, CA), and all recombinant plasmids were verified by DNA sequencing. We adopted the dual luciferase reporter assay system for transfection efficiency confirmation. The transfected HIF-1αA/B binds to the HRE, subsequently promoting the expression of firefly luciferase gene. Simultaneously, the constitutively expressing Renilla luciferase gene was used as an internal control for normalization. Eighteen hours before transfection, EPCs were seeded in 12-well plates with M199 medium (10% FBS medium, no antibiotics), then co-transfected with the indicated Myc-tagged HIF-1α constructs (HIF-1αA, HIF-1αB or empty vector), HRE-luciferase reporter, and pTK-Renilla luciferase reporter by using Lipofectamine 2000 transfection reagent (Invitrogen, USA). After transfection, the culture medium was removed and replaced with serum-free medium for 4 h, then replaced with 10% FBS medium, and incubated under 10% O2 for 24 h. Luciferase activity was detected by using a Dual Luciferase Reporter Assay System (Promega, Madison, WI, USA). The fraction of positively transfected cells was determined by counting. The calculation of the firefly-to-Renilla luciferase activity ratio for each sample serves as a reflection of the transfection efficiency of our target genes.

Measurement of Intracellular ROS

The reactive oxygen species (ROS) detection kit (Beyotime, Shanghai, China) was used to detect mitochondrial ROS generation. Cells were collected in a hypoxia operating system after 24 h incubation, washed twice with serum-free medium, added with diluted DCFH-DA working solution, and incubated for 20 min in an incubator at 28 °C in the dark. The cells were washed with serum-free medium for three times and then analyzed using a multifunctional microplate reader SpectraMax M5 (Molecular Devices, Sunnyvale, CA, USA). The ROS generation was also analyzed by Guava easyCyte 8HT flow cytometry (Merck Millipore, Billerica, MA, USA) at the same time.

Mitochondrial Membrane Potential Assay

The MMP was detected using a JC-1 kit (Beyotime, Shanghai, China). The transfected cells were incubated under 10% O2 for 24 h and were then collected and incubated with 500 μl JC-1 staining solution for 30 min in the dark at 28 °C and monitored under a Leica TCS SP confocal microscope. The wavelength of excitation/emission light was set to 490/530 nm for JC-1 monomer detection and was set to 525/590 nm for JC-1 polymer detection.

Apoptosis Detection

The Annexin V-FITC/PI apoptosis kit (Yeasen, Shanghai, China) was used for apoptosis detection. Cells were incubated under 10% O2 for 24 h after transfection and then approximately 5 × 105 cells were collected. After resuspension, cells were incubated with 5 μL Annexin V-FITC and 5 μL PI Staining Solution for 10 min at room temperature in the dark and analyzed by Guava easyCyte 8HT flow cytometry (Merck Millipore, Billerica, MA, USA).

Mitophagy Detection by Mitochondria and Lysosomes Co-localization Assay

EPC cells were seeded at a density 1 × 105 on the 6 micro-well glass bottom plate (n = 3) after transfection and incubated under 10% O2 for 24 h. Cells were stained with 50 nm Lyso-Tracker Red (Beyotime, Shanghai, China) at 28 °C for 2 h. This was followed by the addition of 100 nm Mito-Tracker Green (Beyotime) and an incubation at 28 °C for 30 min. Cells were then fixed with 4% paraformaldehyde. After fixation, cells were treated with DAPI (Beyotime, Shanghai, China) for 3–5 min in room temperature, then washed with PBS for 2–3 times, each time 3–5 min. After treatment, cells were monitored under a Leica TCS SP confocal microscope.

Measurement of Intracellular ATP Content

The generation of ATP in cells was measured using an enhanced ATP assay kit (Beyotime, Jiangsu, China) according to the manufacturer’s instructions. ATP is measured using a luciferase-luciferin reaction assay. Cells were treated with 10% O2 for 24 h after transfection, then split by the lysis reagent, centrifugated at 12,000 g for 5 min to obtain supernatant. The protein concentration of samples was detected by BCA protein assay kit (Beyotime, Jiangsu, China). The chemical fluorescence intensity of samples was measured by a multifunctional microplate reader SpectraMax M5 (Molecular Devices, Sunnyvale, CA, USA) and the ATP concentration was calculated according to the standard curve.

Total RNA Isolation and Real-Time Quantitative PCR

Total RNA was extracted by Trizol reagent (Biosharp, Guangzhou, China). The quality and concentration of RNA were determined by NanoDrop 2000 (Thermo Scientific, Waltham, MA, USA) and agarose gel electrophoresis. An amount of 1 μg total RNA was reverse transcribed into cDNA using a cDNA synthesis kit (Vazyme, Nanjing, China). The steady-state mRNA levels were detected by Real-Time quantitative PCR based on SYBR green detection (Vazyme, Nanjing, China) with quantitative thermal cycler (MyiQTM 2 Two Color Quantity PCR Detection System, Bio-Rad, Hercules, CA, USA). Primers are listed in Table S1, supplementary files. Calculations for relative expression were performed using the 2−∆∆Ct method.

Statistical Analysis

All the experiments were presented at least three times with a minimum of three repeats. Statistical analysis was conducted with SPSS 16.0 software. Results were performed as means ± standard error (SE). The data were presented by one-way analysis of variance (ANOVA) and followed by Tukey’s multiple range tests. In each analysis, a value of P less than 0.05 was considered to indicate the statistical significance.

Results

Hypoxia Reduces Cell Viability and Induces ROS Production

The ability to tolerate hypoxia varies greatly among different cell types. To determine the viability of cells under different hypoxia stress, a CCK-8 assay was performed after the EPCs were treated with different oxygen levels (0.3%, 1%, 3%, 10%, 20%). Morphologically, the shape and density of EPCs were severely affected after being treated in acute hypoxia (0.3%, 1%, 3% O2) for 24 h, and over 50% of cells died in 0.3% O2 by the end of 48 h, whereas cells under 10% O2 were not inhibited (Fig. 1A). We also examined mitochondrial ROS generation following a 24 h exposure of cells to varying oxygen concentrations. In comparison to normoxia, ROS production in cells exhibited a significant increase under 0.3%, 1%, and 3% hypoxia. However, ROS production remained unchanged at 10% O2 (Fig. 1B). The CCK-8 assay results (Fig. 1C) were in accordance with the cell morphology. The cell viability was not affected by 3% and 1% O2 after 12 h but sharply declined in 0.3% O2 group. After 24 h, the viability of all the cells suffered from acute hypoxia and viability was inhibited. However, cell viability under 10% O2 was relatively high, reaching a peak at 24 h. The viability remained high over 48 h, which may have been due to the stress response to the hypoxia. These findings indicate that the viability of EPCs is influenced by hypoxia in a manner dependent on both time and oxygen concentration. In light of these results, we opted for the 10% O2 treatment for our subsequent experiments.

Fig. 1
figure 1

Cell state and ROS generation under various degree of hypoxic stress after 24 h. A Severe morphological changes in EPCs treated with 3% O2, 1% O2, and 0.3% O2 were observed. Conversely, shape and density were well maintained under 10% O2. B ROS generation varied in EPCs treated with different oxygen concentrations for 24 h. Higher fluorescence represents increased ROS production. C Cell viability declined sharply in the 0.3% O2 group after 12 h and was affected under 3% and 1% O2 after 24 h. The cells in the 10% O2 group did not show compromised viability (n = 3 per group)

Full size image

HIF-1αA Promotes ROS Production and Affects Mitochondrial Membrane Potential of EPCs

We next investigated how the two HIF-1α paralogs affect the stability of EPCs. Prior to that, we assessed the transfection efficiency of HIF-1αA and HIF-1αB. In our experiment, we observed that the fraction of positively transfected cells ranges from 40 to 50%, as determined by counting (Table S2, supplementary files). The Dual Luciferase Reporter Assay demonstrated significantly higher luciferase activity in cells transfected with either HIF-1αA or HIF-1αB comparing to control groups (Figure S1A). Notably, as the luciferase activity of the negative control (Myc vector) exceeded that of the blank control, we further conducted an examination of the mRNA levels of HIF-1αA and HIF-1αB in each experimental group to address the concerns. The results revealed a pronounced increase in the expression levels of HIF-1αA and HIF-1αB in the cells transfected with their respective counterparts (Figure S1B). ROS generation in cells transfected with HIF-1αA increased significantly, while the ROS level in cells transfected with HIF-1αB was not significantly affected, indicating that HIF-1αA and HIF-1αB were functionally different in regulating ROS production (Fig. 2A, B). The promotion of ROS can induce further damage to mitochondria; therefore, the MMPs of the transfected EPCs were investigated. The JC-1 staining showed that the MMP was downregulated only by HIF-1αA (Fig. 2C, D), indicating that the HIF-1αA induces depolarization of the MMP by promoting ROS production, which can lead to decreased stability of mitochondria.

Fig. 2
figure 2

ROS production and mitochondrial membrane potential regulated by HIF-1αA and HIF-1αB. ROS production and mitochondrial membrane potential were measured after 24 h incubation under 10% O2 (n = 3 per group). A ROS production measured by flow cytometry. B ROS production following transfection with either HIF-1αA or HIF-1αB. Significant differences between treatments are marked with different letters (P < 0.05). C Mitochondrial membrane potential was assessed under a Leica TCS SP confocal microscope using JC-1 Green and JC-1 Red. D Red/green florescence ratio assessed mitochondrial membrane potential in cells with HIF-1αA and HIF-1αB. Significant differences between treatments are marked with different letters (P < 0.05)

Full size image

Diverged Roles of HIF-1αA and HIF-1αB in Apoptosis and Mitophagy Regulation

The ROS production and depolarization of mitochondrial membrane are closely related to mitophagy and cell apoptosis. We therefore next investigated the roles of HIF-1αA and HIF-1αB in regulating mitophagy and apoptosis under hypoxia. The AnnexinV-FITC/PI staining revealed a significant promotion of apoptosis when HIF-1αA was over-expressed, while no remarkable changes were observed with HIF-1αB over-expression (Fig. 3A, B). For mitophagy experiments, the co-localization of mitochondria and lysosomes were investigated 24 h after the EPCs were transfected with HIF-1αA or HIF-1αB, respectively. We observed a noteworthy rise in the co-localization of mitochondria and lysosomes in cells transfected with HIF-1αB (Fig. 3C), indicating that the over-expression of HIF-1αB enhances mitophagy. A slight degree of co-localization was also noted in other groups, potentially attributable to the presence of endogenous HIF-1αB.

Fig. 3
figure 3

Apoptosis and mitophagy of the EPCs induced by HIF-1αA and HIF-1αB. A Apoptosis in EPCs using flow cytometry. B Cell apoptosis following HIF-1αA or HIF-1αB transfection. C Mitophagy following HIF-1αA and HIF-1αB transfection was detected by mitochondria and lysosomes co-localization assays. Mitophagy was induced in EPCs transfected with HIF-1αB. Significant differences between treatments are marked with different letters (P < 0.05)

Full size image

HIF-1αB Mediates the Expression of Key Genes Related to Mitochondrial Autophagy

Since HIF-1αB was the dominant paralog that regulated mitophagy, we next investigated how over-expression of HIF-1αA and HIF-1αB regulated mitophagy-related genes. We first measured the expression of genes involved in the classic PINK1-Parkin mitophagy pathway, including pink1, parkin, vdac1, optn, ambra1, and sqstm1. Both pink1 and optn were not affected by HIF-1αA or HIF-1αB over-expression. Conversely, the expression of parkin was strongly up-regulated by HIF-1αB. The transcript level of vdac1, a crucial substrate of parkin that regulates mitophagy, was also up-regulated together with sqstm1, an autophagic adaptor that is essential for mitochondrial clearance. Intriguingly, a second autophagic adaptor, ambra1, was downregulated by HIF-1αB over-expression only. These results indicate that the PINK1-Parkin mitophagy pathway was regulated by HIF-1αB under hypoxic stress (Fig. 4).

Fig. 4
figure 4

Transcripts of the PINK-Parkin pathway. Significant differences between treatments are marked with different letters (mean ± SE, n = 3, P < 0.05)

Full size image

We also investigated how HIF-1αA and HIF-1αB regulated bnip3 and nix expression, as each of these transcripts are key to cell apoptosis and mitophagy. Our results revealed divergent regulation by HIF-1αA and HIF-1αB. Transcript levels for bnip3, together with beclin1, were significantly up-regulated by HIF-1αA, while nix was strongly up-regulated by HIF-1αB. Transcripts for bcl-2, an apoptosis inhibitor, were also up-regulated, but only by HIF-1αB. Over-expression of HIF-1αA promoted the expression of foxo3, while both HIF-1αA and HIF-1αB had suppressive effects on mTOR regulation (Fig. 5). Both Ulk1 and Ulk2 are two important initiators to regulate the induction of autophagy, and the expression level of these genes was strongly affected by HIF-1αA. HIF-1αB also exerted the effects on the two initiators, and in cells transfected with HIF-1αB, transcript levels of ulk1 were remarkably increased; however, this was not observed for ulk2.

Fig. 5
figure 5

Transcripts involved in mitophagy and apoptosis. Significant differences between treatments are marked with different letters (mean ± SE, n = 3, P < 0.05)

Full size image

Intriguingly, a crucial hypoxia-sensitive mitophagy-involved gene, fundc1, was up-regulated by HIF-1αA but not HIF-1αB, while lc3, a key protein contributing to major steps of autophagy, was only regulated by HIF-1αB under hypoxic stress (Fig. 5). Taken together, although the mitophagy process was mainly regulated by HIF-1αB, HIF-1αA also contributed to the regulation of genes important for mitophagy under hypoxia.

Divergent Impact of HIF-1αA and HIF-1αB on Mitochondrial Oxidative Phosphorylation-Related Genes and ATP Production

We next investigated the impact of HIF-1αA and HIF-1αB over-expression of mitochondrial oxidative phosphorylation of EPCs 24 h after hypoxic stress. We chose representative genes that encode subunits from each mitochondrial complex, one mitochondrial gene, and one nuclear gene (except for Complex II, which is encoded by four nuclear genes).

For Complex I, the expression levels of mtnd1 and ndufs1 were measured, and the results showed a remarkable increase of ndufs1 in the cells transfected with HIF-1αB (Fig. 6). For Complex II, we measured sdha and sdhb expression levels following transfection. There was a significant increase of sdha in the cells with HIF-1αB over-expression (Fig. 6). For Complex III, the transcription levels of cytochrome b (mtcyb) and uqcrc1 were determined. The expression of mtcyb was up-regulated by HIF-1αA, while that of the uqcrc1 was significantly up-regulated by HIF-1αB (Fig. 6). For Complex IV, the expression levels of coxI and coxIV were measured, and it was revealed that HIF-1αB up-regulated the level of coxIV, but neither HIF-1αA nor HIF-1αB exerted effects on the mitochondrial gene coxI (Fig. 6). These results suggest that HIF-1αA and HIF-1αB have different roles in regulating mitochondrial complex synthesis.

Fig. 6
figure 6

Transcripts for mitochondrial complexes in the ETC. Significant differences between treatments are marked with different letters (mean ± SE, n = 3, P < 0.05)

Full size image

Lastly, the karyogenes from each complex were primarily affected by HIF-1αB, which may further alter ATP production. Therefore, ATP content was measured, and the results revealed that under hypoxia, the production of ATP was driven by HIF-1αB (Fig. 7).

Fig. 7
figure 7

ATP levels under hypoxia following HIF-1αA or HIF-1αB transfection. Significant differences between treatments are marked with different letters (mean ± SE, n = 3, P < 0.05)

Full size image

Discussion

For most cell types, hypoxia can lead to the decrease of proliferation. However, cell proliferation can be maintained in certain cell populations when subjected to hypoxic stress, which is one of the common pathological traits in tumor cells, and this adaptive response is also crucial for maintenance of stem cell populations in the face of oxygen insufficiency (Hubbi and Semenza 2015). How mammalian cells sense oxygen levels to coordinate diverse biological outputs during hypoxia has been well demonstrated in recent years. Perhaps the best studied mechanism-mediating responses in low oxygen environments are hypoxia-inducible factors (HIFs). HIFs are stabilized by hypoxia and control the expression of several genes. For fish, however, the exact roles of HIF paralogs under hypoxia are not fully characterized.

The epithelioma papulosum cyprini (EPC) cell line is derived from the epithelial cells found in the skin of the fathead minnow (Pimephales promelas), a freshwater fish belonging to the Cyprinidae family, same as zebrafish. HIF-1α exhibits a remarkable degree of conservation, as evidenced by prior research indicating that zebrafish HIF-1α can successfully recruit HIF-β and operate effectively even within human cells. (Chen et al. 2020; Guan et al. 2014). Consequently, it is reasonable to anticipate that zebrafish HIF-1α can readily recruit HIF-β and function effectively in EPCs. Zebrafish HIF-1αA consists of 717 amino acids, while HIF-1αB comprises 776 amino acids. These two paralogs exhibit a substantial sequence similarity within their binding domains, encompassing the basic helix-loop-helix (bHLH) and Per-Arnt-Sim (PAS) domains, the oxygen-dependent degradation domain (ODDD), and the C-terminal transactivation domain (C-TAD) (see Figure S2). Notably, a deletion is present within the conserved N-terminal ODDD hydroxylation motif (LXXLAP), and similar deletions have been observed in other cyprinid fish species (Guan et al. 2014). This deletion is believed to potentially reduce the protein's sensitivity to oxygen levels (Rytkonen et al. 2013). The N-TAD, working in conjunction with the C-TAD, plays a pivotal role in influencing the repertoire of target genes activated by HIF-1. The coexistence of both N-TAD and C-TAD empowers HIF-1 to selectively trigger a specific subset of genes in response to hypoxia. Consequently, variations in the amino acid sequences within the transactivation domains (TADs) of these two paralogs may contribute to the divergence in their target gene profiles.

EPCs are not hypoxia-tolerable, and their viability can be severely depleted under acute hypoxia. Our experiments showed that the survival rate of the cells under hypoxia is time- and oxygen-concentration-dependent. Nevertheless, cells can usually improve their ability to adapt to hypoxia to a certain degree and improve their viability through self-regulation, thus promoting survival in a short time (Lee et al. 2020). Given the functions of HIF-1α, the increased viability of cells under 10% O2 may be due to the function of endogenous HIF-1αs.

Under hypoxic stress, mitochondria can become uncoupled because there is not enough O2 to serve as electron receptors during oxidative phosphorylation. As a result, free electrons are increased in the form of ROS. The accumulation of ROS will oxidize and damage macromolecules in cells, which will lead to elevated mitochondrial membrane permeability, the release of cytochrome C from mitochondria, the activation of apoptosis-inducing factors, and subsequently apoptosis (Ruth and Zvulun 2011). In most physiological or pathological cases, a change of MMP is an early response of apoptosis. Studies show that there is decline of MMP is one of the important factors leading to apoptosis (Yang et al. 1997). Our results revealed that the ROS production in EPC cells is oxygen-concentration-dependent (Fig. 1B). Studies have shown that the increase in mitochondrial ROS caused by hypoxia could extend the lifespan of multiple types of cells in vitro (Bell et al. 2007; Liu et al. 2005; Schulz et al. 2007). This response, however, must be finely tuned as overproduction of ROS can eventually result in damage leading to cell death. HIF-1α regulates the generation of ROS by unbalancing the electron flow in the Electron Transport Chain. The specialized functions of HIF-αA and HIF-αB isoforms in cyprinids have been the subject of investigation regarding their distinct roles in the regulation of mitochondrial biogenesis and oxidative phosphorylation. Research findings indicate that, when stabilized by CoCl2, both HIF-αA and HIF-αB isoforms exhibit the capacity to enhance mitochondrial biogenesis. HIF-1αA and HIF-1αB have partially overlapped target gene pools. They have similar functions in promoting mitochondrial biogenesis, but different roles in other aspects like oxidative phosphorylation, the tricarboxylic acid cycle, and the maintenance of mitochondrial stability (Chen et al. 2020). HIF-α is unstable when oxygen is sufficient. The CoCl2 -cultivated experiment stabilizes HIF-α under normoxic conditions, but the cells do not experience hypoxia. As a result, it remains unclear how HIF-1αA and HIF-1αB function in response to hypoxia. Our results demonstrate that HIF-1αA maintained ancestral gene function of regulating ROS under hypoxia, which may be beneficial to both HIF-1αA and HIF-1αB because intracellular ROS can inhibit the activity of prolyl hydroxylase to prevent the degradation of HIF-1α (Li et al. 2014). Our study also illustrated that over-expression of HIF-1αA causes the collapse of MMP, an outcome of excessive accumulation of ROS, and increases apoptosis under hypoxic stress. These results suggested that HIF-1αA can induce apoptosis through the promotion of ROS and mitochondrial depolarization when cells are subjected to hypoxic stress. This ability appears to be lost for HIF-1αB after duplication of the ancestral gene.

Mitochondrial autophagy, or mitophagy, is a crucial strategy for cells to prolong viability under hypoxia or other environmental stressors. In this study, only HIF-1αB significantly promoted the co-localization of mitochondria and lysosomes, suggesting a prominent role in the induction of mitophagy. However, both HIF-1αA and HIF-1αB were functionally involved in different mitophagy pathways regulation at the gene level. The PINK-Parkin pathway, usually mediates mitophagy under cellular starvation. Its activation is not hypoxia dependent, but it can be strongly affected by hypoxia and ROS. In our study, the expression of Parkin was up-regulated by HIF-1αB. Two autophagic adaptors Vdac1 and Sqstm1, each important in the PINK-Parkin-mediated clearance of mitochondria (Geisler et al. 2010), were also increased by HIF-1αB. In addition, Ambra1 has emerged as a scaffold protein regulating autophagy in multiple ways. The down-regulation of Ambra1 by HIF-1αB may be due to its negative regulation to autophagy by promoting proteasomal degradation (Cianfanelli and Cecconi 2015). However, studies also demonstrate its positive role in inducing mitophagy regardless of Parkin and Sqstm1 (Antonioli et al. 2015; Strappazzon et al. 2015). These results suggested that HIF-1αB was predominantly responsible for mitophagy induction through the PINK-Parkin pathway under oxygen deficiency.

Hypoxia can activate mitophagy in different cell types through Bnip3L/Nix, Bcl-2/Bnip3, and Fundc1 (Liu et al. 2012a; Zhou et al. 2011). Our data revealed that over-expression of HIF-1αA induced expression of multiple genes involved in these pathways including Bnip3, Fundc1, Beclin1, and Foxo3 (Fig. 5); these proteins act as intermediates in autophagy and apoptosis. For instance, Beclin1, an ortholog of the Atg6 in yeast, is essential for localization of autophagic proteins to a pre-autophagosomal structure (He and Levine 2010), and this function is closely related to the function of Bcl-2, an anti-apoptotic protein and inhibitor of mitochondrial dysfunction during apoptosis, which is transcriptionally controlled by HIF-1αB. The dissociation of Beclin1 from Bcl-2 is essential for its autophagic activity (Maiuri et al. 2007). Foxo3 is a transcriptional factor responsible for rapid induction of autophagy in cells by transactivating multiple autophagy-related genes. Studies demonstrate that Foxo3 is a substrate and is consumed by basal autophagic degradation, which confers low sensitivity to apoptosis. If autophagy is inhibited, cells will activate pro-apoptotic genes such as Bbc3 via the accumulation of Foxo3 becoming sensitized to apoptotic signals (Fitzwalter and Thorburn 2018; Thorburn et al. 2014). Therefore, HIF-1αA possesses dual functions in regulating mitochondrial autophagy and cell death, and to some extent, it plays more of a role than HIF-1αB in promoting apoptosis. Conversely, HIF-1αB appears to be primarily responsible for the induction of mitophagy, preventing apoptosis via transactivating Bcl-2. Autophagy and apoptosis are two important cellular processes with complex and intersecting protein networks and the transition from “self-eating” to “self-killing” relies on the extent to which stress can be tolerated in terms of hypoxia and ROS overproduction. Therefore, our study shows that the timing of these processes is finely tuned by the divergent function of HIF-1αA and HIF-1αB.

Since the mitochondrial ETC is the largest consumer of intracellular O2 for the generation of ATP, it can be expected that changes in O2 supply will affect the activity mitochondrial complexes in the ETC. The results revealed functional divergence and target shifting of HIF-1αA and HIF-1αB (Fig. 6), suggesting that the protein from one copy targets mitochondrial genes while the other primarily targeted nuclear genes that encodes subunits of the complexes. Complex IV (also known as cytochrome c oxidase, COX) is the last complex within the ETC, providing 4 electrons to O2 to generate water. Since Complex IV possesses a high affinity for O2, the ETC can function at near anoxic levels (Cooper 1990), cells can maintain their ATP levels during hypoxia, which slows accumulation of intermediate ROS. Our study suggests that HIF-1αB inherited a role in promoting ATP generation from its ancestral gene. We propose this is the reason HIF-1αB does not facilitate ROS production.

Conclusions

Subfunctionalization results in an increase in the preservation of duplicated gene copies, which are essential to the generation of evolutionary novelty. The two HIF-1α duplicates were specifically retained in cyprinid fish. Our study unraveled their diverged function in regulating ROS and ATP generation, mitochondrial autophagy, and apoptosis. The distinct collaboration between the two HIF-1α copies has unveiled precise regulation of mitophagy and apoptosis in response to hypoxia exposure.