Methyltransferase SMYD3 impairs hypoxia tolerance by augmenting hypoxia signaling independent of its enzymatic activity

Hypoxia-inducible factor (HIF)1α, a main transcriptional regulator of the cellular response to hypoxia, also plays important roles in oxygen homeostasis of aerobic organisms, which is regulated by multiple mechanisms. However, the full cellular response to hypoxia has not been elucidated. In this study, we found that expression of SMYD3, a methyltransferase, augments hypoxia signaling independent of its enzymatic activity. We demonstrated SMYD3 binds to and stabilizes HIF1α via co-immunoprecipitation and Western blot assays, leading to the enhancement of HIF1α transcriptional activity under hypoxia conditions. In addition, the stabilization of HIF1α by SMYD3 is independent of HIF1α hydroxylation by prolyl hydroxylases and the intactness of the von Hippel-Lindau ubiquitin ligase complex. Furthermore, we showed SMYD3 induces reactive oxygen species accumulation and promotes hypoxia-induced cell apoptosis. Consistent with these results, we found smyd3-null zebrafish exhibit higher hypoxia tolerance compared to their wildtype siblings. Together, these findings define a novel role of SMYD3 in affecting hypoxia signaling and demonstrate that SMYD3-mediated HIF1α stabilization augments hypoxia signaling, leading to the impairment of hypoxia tolerance.

Hypoxia-inducible factor (HIF)1α, a main transcriptional regulator of the cellular response to hypoxia, also plays important roles in oxygen homeostasis of aerobic organisms, which is regulated by multiple mechanisms. However, the full cellular response to hypoxia has not been elucidated. In this study, we found that expression of SMYD3, a methyltransferase, augments hypoxia signaling independent of its enzymatic activity. We demonstrated SMYD3 binds to and stabilizes HIF1α via co-immunoprecipitation and Western blot assays, leading to the enhancement of HIF1α transcriptional activity under hypoxia conditions. In addition, the stabilization of HIF1α by SMYD3 is independent of HIF1α hydroxylation by prolyl hydroxylases and the intactness of the von Hippel-Lindau ubiquitin ligase complex. Furthermore, we showed SMYD3 induces reactive oxygen species accumulation and promotes hypoxia-induced cell apoptosis. Consistent with these results, we found smyd3-null zebrafish exhibit higher hypoxia tolerance compared to their wildtype siblings. Together, these findings define a novel role of SMYD3 in affecting hypoxia signaling and demonstrate that SMYD3mediated HIF1α stabilization augments hypoxia signaling, leading to the impairment of hypoxia tolerance.
It is well-known that oxygen profoundly affects physiology of aerobic organisms through multiple mechanisms. Molecular oxygen not only acts as the terminal electron acceptor at complex IV of the respiratory chain that yields energy during aerobic respiration and builds metabolites but also promotes to change the configuration and function of nucleic acids, sugars, lipids, proteins, and metabolites. Inadequate oxygen availability can lead to cellular dysfunction and even cell death. Under low oxygen (hypoxic) conditions, aerobic organisms utilize their cardiovascular system and respiratory system to ensure adequate oxygen delivery to cells and tissues. In addition, cells undergo adaptive changes to initiate gene expression that either enhance oxygen delivery or promote survival (1). In addition, hypoxic conditions can also trigger oxidative stress by generating uncontrolled reactive oxygen species (ROS) in mitochondria, which may pose a threat to cell survival. ROS, a generic term for a large family of oxidants derived from molecular oxygen, can be neutralized by catalase, peroxidase, and superoxide dismutase. However, under hypoxic conditions, disturbances in electron transport are associated with electron leakage from the respiratory chain, giving rise to increased ROS, which may be toxic to cells if ROS levels are not attenuated (2)(3)(4)(5).
In the process of hypoxia adaptation, the hypoxia signaling pathway mediated by hypoxia-inducible factor (HIF) plays a pivotal role (6)(7)(8)(9)(10)(11). As a key modulator of the transcriptional response to hypoxic stress, HIF is a heterodimer of bHLH-PAS proteins consisting of an O 2 -labile alpha subunit (HIFα) and a stable beta subunit (HIF1β)/(ARNT) that binds hypoxia response elements. Aerobic organisms possess three HIFα proteins, of which HIF1α and HIF2α are the most structurally similar containing two transactivation domain (N-terminal transactivation domain and C-terminal transactivation domain) (6,10).
Under well oxygenated (normoxic) conditions, HIFα subunit is hydroxylated at two highly conserved prolyl residues by the prolyl hydroxylases (PHDs) (also called EglNs), whose activity is regulated by O 2 availability (6,12,13). Hydroxylated HIFα generates a binding site for being recognized by the von Hippel-Lindau (pVHL) tumor suppressor protein complex, which is an ubiquitin ligase complex. As a result, HIFα is polyubiquitinated and subjected to proteasomal degradation. Under hypoxic conditions, PHDs activity is diminished, leading to stabilization and accumulation of HIFα proteins. Stabilized HIFα proteins dimerize with HIF1β, translocate to the nucleus, and induce transcription of genes involved in hypoxia adaptation or tolerance (6,7). The factors affecting hypoxia signaling pathway mainly impact on HIFα protein stability (14)(15)(16). In addition, FIH (factor that inhibits HIF)-mediated asparagine hydroxylation impairs the transcriptional activity of HIF by interrupting the interaction between HIF and the transcriptional cofactor CBP/p300 (17).
SMYD3 is a member of the SMYD lysine methylase family containing two conserved structural domains: the catalytic Su (var) 3-9, Enhancer-of-zeste, and N-terminal Trithorax (SET) domain, which is split by a Myeloid-Nervy-DEAF1 domain (33). The SET domain of SMYD3 is comprised of two sections: the S-sequence, which may function as a cofactor binder as well as for protein-protein interactions, and the core SET domain, which functions as the primary catalytic location domain, and the C-terminal domain (33)(34)(35). SMYD3 plays an important role in the methylation of various histone and nonhistone targets involved in tumorigenesis and affecting transcriptional regulation (36)(37)(38)(39)(40)(41)(42). In addition, it was reported previously that the oncogenic function of SMYD3 is partially independent on its methyltransferase activity (43,44).
Whether or not SMYD3 involved in hypoxia signaling is still not understood. In this study, we show that SMYD3 interacts with HIF1α and stabilizes HIF1α independent of its methyltransferase activity, leading to the augment of the hypoxia signaling, the accumulation of ROS, and the enhancement of hypoxia-induced cell apoptosis. By zebrafish model, we found that disruption of smyd3 facilities zebrafish hypoxia tolerance, which might be resulted from the impact of smyd3 on hypoxia signaling.

SMYD3 augments hypoxia signaling
We have previously identified that the monomethyltransferase, SET7, represses hypoxia signaling by catalyzing HIF-α methylation (30). To determine whether other methyltransferases also involved in hypoxia signaling, initially, we examined expression of a series of methyltransferases in HEK293T cells under hypoxia. As shown in Fig. 1A, the typical hypoxia responsive genes, including GLUT1, BNIP3, PDK, PGK1, and VEGF (30,31,45), were greatly induced under hypoxia, suggesting the hypoxic condition was achieved expectedly. Among the methyltransferase genes tested, SMYD2, SMYD4, SETD1A, EZH1, EZH2, and SUV420H1 were upregulated under hypoxia, but only SMYD3 was significantly suppressed (Fig. 1A), which provoked us to further test the impact of SMYD3 in affecting hypoxia signaling. Subsequently, we examined whether the effect of hypoxia on SMYD3 expression is dependent of HIF signaling. In H1299 cells, the expression of SMYD3 was significantly suppressed under hypoxia (Fig. S1A). However, in ARNT-deficient H1299 cells (ARNT −/− ) (Fig. S1B), hypoxia failed to induce expression of PGK1, a typical HIF1α target gene ( Fig. S1C) but could still suppress expression of SMYD3 (Fig. S1D). In addition, we added PX478 to inhibit HIF1α activity and then checked the effect of hypoxia on SMYD3 expression (46). When PX478 (100 μM) was added, hypoxia failed to induce expression of PGK1 ( Fig. S1E) but could still suppress expression of SMYD3 (Fig. S1F). These results suggest that the effect of hypoxia on SMYD3 is independent of HIF signaling.
To determine the effect of SMYD3 on hypoxia signaling, we overexpressed SMYD3 in HEK293T cells and examined expression of hypoxia responsive genes under normoxia or hypoxia. Ectopic expression of SMYD3 promoted expression of typical hypoxia responsive genes, including GLUT1, PGK1, and VEGF, under hypoxia ( Fig. 1, B-D). To further confirm these observations, we changed direct-hypoxia treatment to the addition of deferoxamine mesylate salt (DFX) or CoCl 2 , two widely used hypoxia-mimic conditions (47,48) and then examined the effect of SMYD3 on hypoxia responsive gene expression. Consistently, overexpression of SMYD3 also enhanced expression of GLUT1, PDK1, PGK1, and BNIP3 ( Fig. 1, E-J). SMYD3 is reported to downregulate the protein level of p53 (49), and p53 plays vital roles in hypoxia signaling (50). To exclude whether the effect of SMYD3 on hypoxia signaling was mediated by p53, we examined the effect of SMYD3 on hypoxia signaling in p53-deficient H1299 cells. Similar results were obtained by H1299 cells (Fig. S1, G-I). In contrast, knockout of SMYD3 in HEK293T cell resulted in a reduction of expression of GLUT1, PGK1, PDK1, or BNIP3 under hypoxia or CoCl 2 treatment (Fig. 2 SMYD3 binds to and stabilizes HIF1α, leading to an increase of nuclear HIF1α and enhancement of HIF1α-mediated target genes expression Given that HIF1α is one of the master regulators of hypoxia signaling, the enhancement of SMYD3 on hypoxia responsive gene expression promoted us to test whether this effect is mediated by HIF1α. Co-expression of SMYD3 together with HIF1α caused an induction of expression of GLUT1, PGK1, and VEGF mediated by ectopic expression of HIF1α in HEK293T cells (Fig. 3, A-C). HIF1α expression was confirmed by Western blot analysis (Fig. S4A). We next examined whether SMYD3 interacted with HIF1α. Co-immunoprecipitation assays indicated that ectopicexpressed HA-SMYD3 interacted with ectopic-expressed 3D). Semiendogenous coimmunoprecipitation assays indicated that ectopic-expressed HA-SMYD3 interacted with endogenous HIF1α under   together with pCMV-SMYD3 or pCMV empty vector (EV) (control) for 24 h. Data show mean ± SD; Student's two-tailed t test. *p < 0.05, **p < 0.01, ***p < 0.001. Data from three independent experiments. D, co-immunoprecipitation of HA-SMYD3 with Myc-HIF1α. HEK293T cells were co-transfected with indicated plasmids for 24 h. Anti-HA antibody-conjugated agarose beads were used for immunoprecipitation, and the interaction was detected by immunoblotting with the indicated antibodies. E, endogenous interaction between Smyd3 and Hif1α. Smyd3-deficient or wildtype MEF cells (Smyd3 −/− or Smyd3 +/+ ) under hypoxia for 4 h and anti-HIF1α antibody was used for immunoprecipitation. F, immunoblotting of exogenous Myc-HIF1α expression in H1299 cells transfected with an increasing amount of HA-SMYD3 expression plasmid (HA empty vector [-] was used as a control). G, immunoblotting of hypoxia (Fig. S4B). Endogenous interaction between SMYD3 and HIF1α was further confirmed in HEK293T cells under hypoxia (Fig. S4C). In Smyd3 +/+ MEF cells, but not in Smyd3 −/ − MEF cells, endogenous Smyd3 interacted with endogenous HIF1α (Fig. 3E). Furthermore, we examined whether the protein stability of HIF1α is affected by SMYD3. Coexpression of SMYD3 together with HIF1α caused HIF1α protein level was increased steadily (Fig. 3F). Overexpression of SMYD3 upregulated endogenous HIF1α protein level under hypoxia (Fig. S4D). By contrast, the endogenous Hif1α protein level was lower in Smyd3-null MEF cells (Smyd3 −/− ) compared to that in Smyd3-intact MEF cells (Smyd3 +/+ ) under hypoxia (Fig. 3G). Consistently, reconstitution of Smyd3 in Smyd3 −/− MEF cells caused an increase of Hif1α protein under hypoxia (Fig. 3H).
Since stabilized HIF1α needs to translocate into the nucleus to function as a transcription factor; therefore, we investigated the effect of SMYD3 on the nuclear HIF1α levels. Notably, overexpression of SMYD3 enhanced HIF1α protein in the nuclei of HEK293T cells (Fig. S4E). In agreement, Hif1α protein level was higher in the nuclei of Smyd3 +/+ MEF cells compared to the nuclei of Smyd3 −/− MEF cells, which was further confirmed by confocal microscopy (Fig. 3, I and J). Consistently, in cycloheximide pulse chase assay, overexpression of SMYD3 attenuated degradation of co-expressed HIF1α in HEK293T cells (Fig. S4F).
These data suggest that SMYD3 interacts with and stabilizes HIF1α, leading to an increase of nuclear HIF1α and enhanced HIF1α-mediated expression of target genes.
The induction of HIF1α target gene expression and stabilization of HIF1α by SMYD3 are independent of HIF1α hydroxylation and pVHL intactness Hydroxylation of HIF1α and subsequent proteasomal degradation mediated by pVHL E3 ubiquitin ligase complex plays a central role in HIF1α regulation. We further investigated whether regulation of HIF1α by SMYD3 relies on this way. Ectopic expression of SMYD3 enhanced HIF1α protein level (Fig. S5A) and expression of GLUT1, PGK1, and PDK1 induced by addition of FG4592, a specific inhibitor of PHDs (Fig. 4, A-C) (51). These data suggest that the induction of HIF1α target genes expression by SMYD3 might not be dependent of HIF1α hydroxylation. Furthermore, we knocked out VHL in HEK293T cells and then examined the effect of SMYD3 on hypoxia signaling (Fig. S5B). As expected, in VHL -/-HEK293T cells, the hypoxia responsive genes, including GLUT1, PGK1, PDK1, LDHA, BNIP3, PHD3, and PKM2, were increased compared to those in VHL +/+ HEK293T cells (Fig. S5C), indicating that VHL was disrupted in HEK293T cells efficiently. Ectopic expression of SMYD3 in VHL -/-HEK293T cells enhanced HIF1α protein level (Fig. S5D) and hypoxia responsive gene expression (Fig. 4, D-F) in a dosedependent manner. These data suggest that the induction of HIF1α target genes expression by SMYD3 is independent of pVHL intactness.
In addition, co-expression of SMYD3 together with HIF1α caused HIF1α protein level to increase steadily, which was not affected when the two prolyl residues (P402/P564) were mutated (HA-HIF1α-DM) (P402A/P564A) (Fig. S5, E-F). Furthermore, when FG4592 was added either in an increase of dose or in an extended time course, the protein level of endogenous Hif1α in Smyd3 +/+ MEF cells kept higher than that in Smyd3 -/-MEF cells (Fig. 4, G-J).
Taken together, these data suggest that the induction of HIF1α target gene expression and stabilization of HIF1α by SMYD3 is independent of HIF1α hydroxylation and pVHL intactness.
The stabilization and activation HIF1α by SMYD3 are independent of its methyltransferase activity Given that SMYD3 serves as a methyltransferase, we sought to determine whether the modulation of HIF1α by SMYD3 was mediated by the enzymatic activity of SMYD3. Under hypoxia, ectopic expression of enzymatic-inactive mutant of SMYD3 (SMYD3-F183A) still enhanced expression of PGK1 and PDK1 in HEK293T cells, similar to its wildtype form (Fig. 5, A and B).
Taken together, these data suggest that SMYD3 stabilizes and activates HIF1α independent of its methyltransferase activity.

SMYD3 induces ROS accumulation and enhances hypoxiainduced cell apoptosis
Many studies have reported that reduction of the cytotoxic ROS level is associated with cell survival during hypoxia adaptation (52) and that aberrant control of mitochondrial ROS levels is a major factor resulting in cell apoptosis with long-term exposure to hypoxic environments (53). We examined the effect of SMYD3 on ROS accumulation. Hypoxia treatment significantly induced ROS accumulation, while   much lower levels of intracellular and mitochondrial ROS were detected in Smyd3 −/− MEF cells compared to Smyd3 +/+ MEF cells by flow cytometry assay (Fig. 6, A-D).
To determine the biological consequences of the transcriptional activity enhancement of HIF1α by SMYD3, we compared cell apoptosis between Smyd3 +/+ and Smyd3 −/− MEF cells under hypoxia. More apoptotic cells were detected in Smyd3 +/+ MEF cells compared to Smyd3 −/− MEF cells by flow cytometry assay, which was further confirmed by confocal microscopy (Fig. 7, A and B).
Subsequently, we examined the effect of overexpression of Smyd3 on cell apoptosis. In contrast, overexpression of Smyd3 enhanced cell apoptosis under hypoxia as detected by flow cytometry assay, which was further confirmed by confocal microscopy (Fig. 8, A and B).
These data suggest that Smyd3 enhanced hypoxia-induced apoptosis, which might be mediated by HIF1α.

Disruption of smyd3 in zebrafish facilitates hypoxia tolerance
SMYD3 is evolutionary conserved among human, mouse, and zebrafish (Fig. 9A). In zebrafish liver cells, ectopic expression of zebrafish smyd3 caused an increase of expression of hypoxia responsive genes under hypoxia, including pdk1, vegf, and phd3 (Fig. 9, B-D), suggesting that the function of SMYD3 might be conserved between mammals and zebrafish. To determine the physiological role of the transcriptional activity enhancement of HIF1α by SMYD3, we took advantage of zebrafish model. We knocked out smyd3 in zebrafish via CRISPR/Cas9 and obtained one mutant line (Fig. 10A). Heteroduplex mobility assay (HMA) and quantitative RT-PCR (qPCR) assay indicated that smyd3 was disrupted efficiently in zebrafish (Fig. 10, B and C). One predicted peptide with 176 amino acids might be produced in smyd3-null zebrafish (Fig. 10D). By crossing smyd3 +/− (♀) × smyd3 +/− (♂), the offspring with smyd3 +/+ , smyd3 +/− , and smyd3 −/− genetic backgrounds were born at a Mendelian ratio (1:2:1), and no obvious defects in growth rate and reproduction capability were detected in smyd3 −/− zebrafish under normal conditions.
These data suggest that smyd3 impairs hypoxia tolerance, which might be mediated by its enhancement role on HIF1α transcriptional activity.

Discussion
The modulation of HIF1α activity by its binding partners has been widely recognized, particularly, the most of these binding partners with enzymatic activity can regulate HIF1α activity through multiple posttranslational modifications, leading to the impacts on HIF1α activity in hypoxia signaling pathway (16,30,31,45,(54)(55)(56)(57)(58). Among them, lysine methylation of HIF1α have been widely investigated. SET7-mediated monomethylation and LSD1-mediated demethylation of HIF1α at lysine 32 synergistically regulates the stability and activity of HIF1α (30,59,60), while monomethylation and dimethylation of HIF1α at lysine 674 by G9a/GLP inhibits its transcriptional activity and expression of its downstream target genes (61). However, whether other methyltransferases also involved in hypoxia signaling remains largely unknown. SMYD3 is a well-defined methyltransferase (34)(35)(36). Here, we identify that SMYD3 binds to and enhances HIF1α activity, leading to the impairment of hypoxia tolerance, which is independent of its enzymatic activity. Of note, some binding partners with enzymatic activity also can affect HIF1α function independent of their enzymatic activity (31,(62)(63)(64). Therefore, it might be a common phenomenon that the proteins can affect HIF1α activity only through protein-protein interaction. However, due to the lack of structure data about the interaction between SMYD3 and HIF1α, we cannot provide more information for understanding the process and the underlying mechanisms of HIF1α activity enhancement by SMYD3.
SMYD3 contains two conserved structural domains: the Myeloid-Nervy-DEAF1 domain and the SET domain; the SET domain is consisted of the S-sequence, the core SET domain, and the C-terminus domain. The S-sequence is responsible for cofactor binding, while the core SET domain is responsible for the catalytic activity of the methyltransferase (33). Here, we find that SMYD3 binds and stabilizes HIF1α, leading to enhanced hypoxic signaling independent of its enzymatic activity. To further identify which structural domain of SMYD3 interacts HIF1α might give insights into the detailed mechanisms of SMYD3 for acting its roles in hypoxic signaling.
Given an importance of hypoxia signaling in tumor progression and cell metabolism, the present studies are mainly focused on investigating the effects of HIF1α binding partners in affecting these processes (65) (19,58,(66)(67)(68)(69)(70)(71)(72)(73)(74)(75)(76). In fact, the roles of hypoxia signaling in hypoxia adaptation and tolerance have been noticed, particularly for high-altitude adaptation (77)(78)(79)(80)(81). High altitude is defined as areas over 2500 m above sea level, in which the ambient oxygen is much lower than low altitude area. Humans living in these areas often face great challenges due to low oxygen. Genetic evidences indicate that some human genes have gone through adaptive mutation for high altitude adaptation, and the most of them are the core SMYD3 augments hypoxia signaling components of hypoxia signaling pathway (78). In this study, by cell culture system and zebrafish model, we found that disruption of Smyd3 impairs hypoxia-induced cell apoptosis, leading to the facilitation of hypoxia tolerance. These observations not only support an important contribution of HIF1α in hypoxia tolerance but also provide a practical research model for testing hypoxia tolerance by zebrafish model. To further use zebrafish as a model to investigate the factors involved in the regulation of hypoxia signaling as well as their impacts on hypoxia tolerance might open a new window for understanding the mechanisms of high-altitude adaptation.
In this study, we show that SMYD3 enhances hypoxiainduced cell apoptosis, resulting in the impairment of hypoxia tolerance. However, the multiple functions of HIF1α have been identified, and SMYD3 may also affect HIF1α functions other than hypoxia tolerance, such as tumorigenesis, cell metabolism, etc. To further figure out the other effects of SMYD3 mediated through HIF1α will help us to fully understand the physiological role of SMYD3 in hypoxia signaling and the underlying mechanisms.

Experimental procedures
Cell line and culture conditions HEK293T and H1299 cells originally obtained from American Type Culture Collection were cultured in Dulbeccos' modified Eagle medium (VivaCell Biosciences) with 10% fetal bovine serum (FBS) at 37 C in a humidified incubator containing 5% CO 2 . RCC4 cells were provided by Peter J. Ratcliffe and maintained as described previously (30). Zebrafish liver cells were provided by Dr Shun Li and maintained as described previously (82). Smyd3-deficient or wildtype MEF cells (Smyd3 −/− or Smyd3 +/+ ) were established as described previously (83) and cultured in Dulbeccos' modified Eagle medium supplemented with sodium pyruvate (110 mg/L), 10% FBS, 1× nonessential amino acids (Sigma), and 1% penicillinstreptomycin at 37 C in a humidified incubator containing 5% CO 2 . During hypoxia treatment, the cells were cultured under hypoxic condition (1% O 2 , 5% CO 2 , and balanced with N 2 ) by using the NBS Galaxy 48R incubator. The cells were transfected with various amounts of plasmids as indicated by VigoFect (Vigorous Biotech).

Quantitative real-time PCR assay
Total RNAs were extracted using RNAiso Plus (TaKaRa Bio). cDNAs were synthesized using the Revert Aid First Strand cDNA Synthesis Kit (Thermo Scientific). qPCR assays were conducted with MonAmp SYBR Green qPCR Mix (high Rox) (Monad Bio.). The procedure was done according to the protocol provided by the manufacturer. The primers for quantitative RT-PCR assays are listed in Table S1.

Immunoprecipitation and Western blot
Co-immunoprecipitation and Western blot analysis were performed as described previously (45). Anti-HA antibodyconjugated agarose beads (#A2095) were purchased from Sigma. Protein G Sepharose (#17-0618-01) was purchased from GE HealthCare Company. The blots were photographed with the Fuji Film LAS4000 mini-luminescent image analyzer. The protein levels were quantified with Image J software (National Institutes of Health) based on the band density obtained by Western blot analysis.

CRISPR-Cas9 knockout cell lines
To generate HEK293T knocked-out cell lines of indicated genes, sgRNA sequence were ligated into Lenti-CRISPRv2 plasmid and then co-transfected with viral packaging plasmids (psPAX2 and pMD2.G) into HEK293T cells. Six hours after transfection, medium was changed, and viral supernatant was collected and filtered through 0.45-μm strainer. Targeted cells were infected by viral supernatant and selected by 1 μg/ml puromycin for 2 weeks. The sgRNA sequence targeting VHL was described as previously (84). The sgRNA sequence targeting SMYD3 is 5 0 -CCAAGAAGTCGAACGGAGTC-3 0 . The sgRNA sequence targeting ARNT is GTCGCCGCTT AATAGCCCTC.

Immunofluorescence confocal microscopy
Immunofluorescence staining was conducted as previously described (83). Cells were seeded on glass coverslips and cultured as indicated. Then, the cells were fixed in 4% paraformaldehyde in PBS for 30 min at 25 C. After washing three times by PBS, the slides were blocked in the blocking buffer (5% goat serum, 2 mg/ml BSA, 0.1%Triton X-100 in PBS) for 1 h and incubated with primary antibodies overnight at 4 C, followed by incubation with Alexa Fluor 488 goat anti-rabbit IgG or Alexa Fluor 594 goat anti-mouse IgG for 1 h at 25 C. Subsequently, the slides were mounted with VECTA-SHIELD mounting medium containing DAPI and photographed with Leica SP8 laser scanning confocal fluorescence microscope.

Nucleus and cytoplasm separation
Nucleus and cytoplasm separation was conducted with the Nuclear and Cytoplasmic Extraction Kit (#78833, Thermo Scientific) according to the protocol provided by the manufacturer. The extracts were analyzed by Western blot analysis.
To ensure the efficiency of fraction separation, anti-α-tubulin antibody was employed to monitor cytoplasmic proteins, and anti-Histone H3 antibody was used to monitor nuclear proteins.

Measurement of intracellular ROS level
MEF cells were cultured under hypoxia as indicated. After treatment, MEF cells were collected and counted. Cells (1 × 10 6 ) were incubated in PBS solution containing 1 μM of CM-H 2 DCFDA (#C6827, Thermo Fisher) at 37 C for 60 min and then washed with PBS three times, followed by flowcytometric analysis.

Measurement of mitochondrial ROS level
MEF cells were cultured under hypoxia as indicated. After treatment, MEF cells were collected and washed with PBS. Then, the cells were incubated in PBS solution containing 5 μM of MitoSOX Red (# M36008, Thermo Fisher) for 10 min at 37 C and then washed gently three times with PBS, followed by flow-cytometric analysis.

Detection of apoptotic cells
MEF cells were cultured under hypoxia or treated with DFX as indicated. For flow cytometry analysis, the cells were harvested and stained with FITC-Annexin V and PI with FITC Annexin V Apoptosis Detection Kit I (#556547, BD Pharmingen) according to the manufacturer's instructions. Apoptotic cells were detected using Beckman CytoFLEXS, and the data were analyzed with CytExpert software. Besides, the cells were stained with Annexin V-FITC Apoptosis Detection Kit (#C1062, Beyotime) according to the manufacturer's instructions in 6-well plate and imaged under a florescent microscope Nikon TE2000-U.

Hypoxia treatments of zebrafish
Hypoxia treatments of zebrafish were conducted in the hypoxia workstation (Ruskinn INVIVO2 I-400) as described previously (85). For zebrafish larvae (3 days postfertilization [dpf]) experiment, two dish were filled with 10 ml of water. Smyd3-null larvae (3 dpf, n = 30) (smyd3 −/− ) were put into one dish, and their wildtype siblings (3 dpf, n = 30) (smyd3 +/+ ) were put in the second dish. The oxygen concentration in the hypoxia workstation was adjusted to 2% ahead of time. Then, two dishes were put into the hypoxia workstation simultaneously. Four hours later, the samples were harvested for qPCR analysis. This experiment was repeated three times. For the adult zebrafish (3-months postfertilization [mpf]) experiment, zebrafish of similar weight were chosen for further experiments. Two flasks were filled with 200 ml of water. Three smyd3-null zebrafish (smyd3 −/− ) were put into one flask, and three wildtype siblings (smyd3 +/+ ) were put into the second flask. The oxygen concentration in the hypoxia workstation was adjusted to 5% ahead of time. After putting the flasks containing zebrafish into the hypoxia workstation, the behavior of the zebrafish was closely monitored. All animal protocols were approved by the Institutional Animal Care and Use Committee at Institute of Hydrobiology, Chinese Academy of Science.

Statical analysis
GraphPad Prism software (7.0) was used for all statistical analysis. Results with error bars express mean ± SD. Statistical analysis was performed by using Student's two-tailed t test. A p value less than 0.05 was considered significant. Statistical significance is represented as follows: *p< 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Data availability
Further information and requests for resources and reagents should be directed to and will be fulfilled by Xing Liu and Wuhan Xiao.
Supporting information-This article contains supporting information.