Proteomic Response of the Brain to Hypoxic Stress in Marine Medaka Fish (Oryzias melastigma)

Hypoxia is a serious issue that affects the marine environment, with a growing number of hypoxic “dead zones” occurring each year. Reports have indicated that hypoxia is detrimental to the reproductive function and sexual development of fish via the disruption of endocrine signaling in organs involved in the hypothalamus–pituitary–gonad axis, including the brain. While we previously reported that hypoxia induces transcriptome-wide alterations in the brain of marine medaka (Oryzias melastigma), whether these effects were reflected at the protein level remains unclear. Therefore, the present study used high-throughput proteomic sequencing along with bioinformatics analysis to assess the short-term and multi-generational effects of hypoxia on the brain proteome of O. melastigma. We identified 36,567 peptides and 7,599 proteins (1% false discovery rate in brain samples), with functions involved in cellular and metabolic processes such as signaling and reproductive processes as well as energy production and conversion. Furthermore, we determined that hypoxia resulted in the significant differential expressions of 33 upregulated and 69 downregulated proteins in the short-term exposure group and 24 upregulated and 52 downregulated proteins in the multi-generational exposure group. Pathway enrichment analysis of the deregulated proteins indicated that hypoxia could impair brain function by altering arachidonic acid metabolism, tight junctions, and adrenergic signaling under short-term hypoxic exposure and by altering p53 and PI3K–Akt signaling under multi-generational hypoxic exposure, which may lead to the onset of neurodegenerative disorders including Alzheimer’s disease and amyotrophic lateral sclerosis. Ingenuity pathway analysis of the deregulated proteins showed that hypoxia affected common signaling pathways in the brain (e.g., integrin, paxillin, and epithelial adherens junction signaling) under both short-term and multi-generational exposures. Hypoxia also deregulated pathways specific to short-term exposure (including integrin-linked kinase, calcium, and integrin signaling) and multi-generational exposure (including sphingosine-1-phosphate signaling, endocannabinoid neuronal synapse pathway, and endoplasmic reticulum stress pathway). Overall, our results provide additional insights into the mechanisms of hypoxia disrupting neuronal function at the protein level in marine medaka.

Hypoxia is a serious issue that affects the marine environment, with a growing number of hypoxic "dead zones" occurring each year. Reports have indicated that hypoxia is detrimental to the reproductive function and sexual development of fish via the disruption of endocrine signaling in organs involved in the hypothalamus-pituitarygonad axis, including the brain. While we previously reported that hypoxia induces transcriptome-wide alterations in the brain of marine medaka (Oryzias melastigma), whether these effects were reflected at the protein level remains unclear. Therefore, the present study used high-throughput proteomic sequencing along with bioinformatics analysis to assess the short-term and multi-generational effects of hypoxia on the brain proteome of O. melastigma. We identified 36,567 peptides and 7,599 proteins (1% false discovery rate in brain samples), with functions involved in cellular and metabolic processes such as signaling and reproductive processes as well as energy production and conversion. Furthermore, we determined that hypoxia resulted in the significant differential expressions of 33 upregulated and 69 downregulated proteins in the short-term exposure group and 24 upregulated and 52 downregulated proteins in the multi-generational exposure group. Pathway enrichment analysis of the deregulated proteins indicated that hypoxia could impair brain function by altering arachidonic acid metabolism, tight junctions, and adrenergic signaling under short-term hypoxic exposure and by altering p53 and PI3K-Akt signaling under multi-generational hypoxic exposure, which may lead to the onset of neurodegenerative disorders including Alzheimer's disease and amyotrophic lateral sclerosis. Ingenuity pathway analysis of the deregulated proteins showed that hypoxia affected common signaling pathways in the brain (e.g., integrin, paxillin, and epithelial adherens junction signaling) under both short-term and multi-generational exposures. Hypoxia also deregulated pathways specific to short-term exposure (including integrin-linked kinase, calcium, and

INTRODUCTION
Hypoxia, at levels of dissolved oxygen (DO) <2.0 mg L −1 , is becoming increasingly prevalent in aquatic environments. Notably, the global number of hypoxic "dead zones" has been increasing exponentially since the 1960s (Diaz and Rosenberg, 2008;Galic et al., 2018). The occurrence of aquatic hypoxia is partly exacerbated by water pollution in the form of excess nutrients from agricultural and urban waste, which leads to eutrophication (Conley et al., 2009;Jenny et al., 2016). Decreased oxygen levels generally cause adverse effects on the physiology and behavior of marine biota (Pollock et al., 2007;Galic et al., 2018). For example, hypoxia reduced the swimming speed in fish such as the Atlantic cod and dogfish (Metcalfe and Butler, 1984;Schurmann and Steffensen, 1994). Hypoxia also delayed development and growth in zebrafish embryos (Kajimura et al., 2005) and reduced the spawning, fertilization, and hatching success rates in common carp .
Hypoxia and its induced protein, hypoxia-inducible factor 1 (HIF-1), have been linked with several neurodegenerative diseases (Zhang et al., 2011). For example, amyotrophic lateral sclerosis (ALS) is characterized by damaged motor neurons, and hypoxia may cause ALS pathogenesis by inducing motor neuron death through impairment of HIF-1-regulated vascular endothelial growth factor (VEGF) signaling (Skene and Cleveland, 2001). While VEGF regulates blood supply and angiogenesis and is normally upregulated in response to hypoxia, ALS-like symptoms develop upon loss of the hypoxia response element in the VEGF promoter, and people with impaired VEGF activity have higher susceptibility to ALS when working in environments with a high probability of hypoxic exposure (Oosthuyse et al., 2001;Skene and Cleveland, 2001;Vanacore et al., 2010). Another example is Alzheimer's disease, which is caused by neuron damage due to the aggregation of amyloid β (Aβ) peptides (Peers et al., 2009). Since hypoxia has been shown to increase the amyloid precursor protein (APP) levels, which are then converted to Aβ peptides (Peers et al., 2009;Zhang et al., 2011), it may be a driver of Alzheimer's disease. Hypoxia was also reported to influence the progression of Parkinson's disease, which is characterized by the loss of dopaminergic neurons (Zhang et al., 2011). For adult rat models of Parkinson's disease, exposure to intermittent hypoxic episodes was found to partially restore dopamine synthesis in the right hemisphere of the brain and increase serum antioxidant activity; therefore hypoxic conditioning may delay neurodegeneration in Parkinson's disease (Belikova et al., 2012).
Notably, multi-generational exposure to mild hypoxia can induce antioxidant responses in various neuronal cell types in the brain to promote survival and protection against oxidative stresses, such as through increased PI3K-Akt signaling and erythropoietin expression in astrocytes, increased MAPK signaling, and brain-derived neurotrophic factor expression in cortical neurons (Terraneo and Samaja, 2017;. Cerebral blood and oxygen circulation is also affected by multi-generational hypoxia through increased angiogenesis and capillary density as well as elevated cerebral blood flow associated with increased nitric oxide (NO) synthesis and metabolism (Xu and LaManna, 2006;Terraneo and Samaja, 2017). In addition to antioxidant responses, multigenerational hypoxia simultaneously increases reactive oxygen species (ROS) production in the brain through the mitochondrial electron transport chain, thereby resulting in redox imbalance (Terraneo and Samaja, 2017;). An overall increase in the ROS levels leads to brain damage, with neuron apoptosis occurring through the release of mitochondrial apoptotic proteins into the cytosol (Turrens, 2003;Coimbra-Costa et al., 2017). The brain forms part of the hypothalamus-pituitary-gonad (HPG) axis, which is responsible for the regulation and production of hormones involved in reproductive function (Thomas and Rahman, 2009). Hypoxia has been reported to affect the hormone levels, reproductive development, and sex differentiation in fish Yu and Wu, 2006). Moreover, hypoxia alters the proteomic profile of Oryzias melastigma testes in a transgenerational manner . We previously demonstrated that the transcriptomic changes in the brains of O. melastigma upon hypoxia exposure were related to brain function and development, including synaptic transmission and potassium ion transport . However, it remains unknown whether changes at the transcriptome level are reflected at the protein level. Therefore, in the present study, we identified changes to the brain proteome of O. melastigma in response to short-term and multi-generational hypoxia exposure.

Fish Maintenance and Hypoxia Exposure
Approval for the experimental setup involving fish was obtained from the Committee on the Use of Live Animals in Teaching and Research (CULATR, no. 2714-12) at the University of Hong Kong. O. melastigma were kept at 5.8 mg O 2 L −1 , 28 ± 2 • C, pH 7.2, in a daily 14:10-h light/dark cycle, as previously described , until the start of the hypoxic/normoxic exposure experiment. For short-term exposure, sexually mature male O. melastigma were exposed to normoxic (5.8 ± 0.2 mg O 2 L −1 ) or hypoxic (1.5 ± 0.2 mg O 2 L −1 ) conditions, as  Frontiers in Marine Science | www.frontiersin.org described in Lai et al. (2016), for a period of 1 month. For multigenerational exposure, sexually mature male medaka fish were continuously exposed to either normoxic or hypoxic conditions for three generations. The progeny of every generation mated and continued to be exposed to hypoxia or normoxia, and the mate pairs (both males and females) were also kept in either normoxic or hypoxic conditions. The desired DO levels were achieved by continuously passing a constant ratio of premixed air and nitrogen through a 4-cm-wide gas-stripping column into a 300-L reservoir tank. A DO meter (YSI model 580) was used to monitor the DO levels twice per day. The temperature, salinity, pH, and ammonia levels were also kept constant and monitored daily throughout the exposure period.

Protein Extraction From Brain Tissues for iTRAQ
Total protein was extracted from the brain tissues of nine fish from each treatment group. Three brain tissues were pooled together into a single biological replicate, and three replicates were used for iTRAQ (isobaric tags for relative and absolute quantitation) analysis. Proteins were extracted using lysis buffer 3 [8 M urea, 1 mM phenylmethylsulfonyl fluoride (PMSF), 40 mM triethylammonium bicarbonate (TEAB)/Tris-HCl, 2 mM EDTA, and 10 mM dithiothreitol (DTT), pH 8.5] and two 5-mm magnetic beads. Proteins were released from the mixture using a TissueLyser (2 min, 50 Hz). The mixture was centrifuged at 25,000 × g, 4 • C, for 20 min. The supernatant was transferred to a new tube, reduced using DTT (10 mM) at 56 • C for 1 h, and then alkylated in the dark with iodoacetamide (IAM, 55 mM) at 25 • C for 45 min. The supernatant was centrifuged at 25,000 × g, 4 • C, for 20 min, and then the supernatant containing the proteins was collected. The Bradford assay was used to quantify the total protein concentration of the supernatant.

Protein Digestion and Peptide Labeling
The solution containing 100 µg protein in 8 M urea was diluted four times using TEAB (100 mM). The proteins were digested with Trypsin Gold (Promega, Madison, WI, United States) at a 40:1 protein/trypsin ratio at 37 • C overnight. The digested peptides were desalted using a Strata-X C18 column (Phenomenex) and then vacuum-dried. The peptides were reconstituted in 30 µl of TEAB (0.5 M) and vortexed. The iTRAQ Reagent 8Plex Kit was used for iTRAQ labeling of the peptides following the manufacturer's instructions. The peptides were then pooled and passed through a Strata-X C18 column (Phenomenex) for desalting and afterward vacuum-dried.

Peptide Fractionation
The peptides were dissolved in 2 ml of buffer A [95% H 2 O, 5% acetonitrile (ACN), pH adjusted to 9.8 with ammonia] and then loaded onto a high-pH RP column (particle size, 5 µm; Phenomenex). The peptides were passed through the column using a Shimadzu LC-20AB HPLC Pump system at a flow rate of 1 ml min −1 using the following gradient of buffer B (5% H 2 O, 95% ACN, pH adjusted to 9.8 with ammonia): 5% for 10 min, then 5-35% for 40 min, followed by 35-95% for 1 min and then 95% for 3 min. The gradient of buffer B was then decreased to 5% within 1 min and then maintained at 5% for 10 min. The peptides were eluted in fractions separated at 1-min intervals while monitoring absorbance at 214 nm. A total of 20 fractions were collected and vacuum-dried.

High-Performance Liquid Chromatography and Mass Spectrometry Detection
For each peptide fraction, buffer A [0.1% formic acid (FA) and 2% ACN in H 2 O] was added and the reconstituted fraction was centrifuged for 10 min at 20,000 × g. The supernatant was then loaded using the autosampler onto the C18 trap column of the LC-20AD nano-HPLC instrument (Shimadzu, Kyoto, Japan) at a rate of 5 µl min −1 for 8 min. The fraction was passed through the trap column and the peptide separation was performed on an self-packed 75-µm-wide analytical C18 column at a rate of 300 nl min −1 using a buffer B (0.1% FA and 2% H 2 O in ACN) gradient going from 8 to 35% in 35 min, then up to 60% in 5 min. The gradient was kept at 80% for 5 min and then decreased to 5% within 1 min, and the column was then equilibrated for 10 min. Mass spectrometry was done using the TripleTOF 5600 System (SCIEX, Framingham, MA, United States) fitted with a Nanospray III source (SCIEX, Framingham, MA, United States), with a pulled-quartz tip emitter (New Objectives, Woburn, MA, United States), with the setting of the ion spray voltage at 2,300 V, curtain gas and nebulizer gas of 30 and 15, respectively, and the interface heater temperature set at 150 • C. Data were collected using the Analyst 1.6 software (AB SCIEX, Concord, ON, United States) on high-sensitivity mode. With a MS1 accumulation time of 250 ms, the mass range for scanning set at 350-1,500 Da, and filtering for MS scans exceeding 150 counts per second with a charge state between 2 + and 5 + and dynamic exclusion of half the peak width (12 s), up to 30 product ion scans were selected for collision-induced dissociation (CID). The collision energy was adjusted for precursor ions for CID, and a Q2 transmission window at 100% for 100 Da was used for iTRAQ data collection.

Bioinformatics Analysis and Protein Quantification
Using tandem mass spectrometry (MS/MS) data in MGF format as input, the peptides labeled with isobaric tags were quantified using IQuant software (Wen et al., 2014). The Mascot Percolator software package was used to rescore the peptide-spectrum matches (PSMs). PSMs were pre-filtered at a false discovery rate (FDR) of 1%, and then used for protein set assembly. Proteinlevel FDRs for the identified set of proteins were estimated using the picked protein FDR strategy (Savitski et al., 2015), and the identified proteins at a protein-level FDR < 0.01 were used for pathway, Clusters of Orthologous Groups (COG)/Eukaryotic Orthologous Group (KOG), and Gene Ontology (GO) analyses. Differentially expressed proteins (DEPs) were identified based on the filtering criteria of relative fold change > 1.2 and q-value < 0.05. The list of DEPs were used for additional GO, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway, and COG/KOG analysis.

Protein Identification in the Brain of O. melastigma
By using the iTRAQ technology, a total of 1,235,781 spectra were generated, and 36,567 peptides and 7,599 proteins were identified at a FDR cutoff < 0.01 in the brain samples (Table 1).
Our results demonstrated that 3,229 of the identified proteins originated from a single peptide ( Figure 1A) and over 20% of the proteins had a molecular mass over 100 kDa ( Figure 1B and Supplementary Table 1).

Classification of Identified Proteins in the Brain of O. melastigma
The proteins identified in the brain of O. melastigma were subjected to GO and KOG annotation to determine their functional roles in the brain. In GO annotation, the molecular functions, cellular components, and biological processes of the proteins were classified (Figure 2A). Our results indicated that the majority of these proteins were involved in binding and catalytic activity and were also responsible for structural molecule activity and signal transducer activity (Figure 2A). Many of the proteins were localized to the cell membrane, organelle, macromolecular complex, as well as the synapse (Figure 2A).
In the analysis of biological processes, our results identified enrichment in terms of cellular and metabolic processes. Furthermore, these proteins also served roles in response to stimuli, developmental process, signaling, cell proliferation, and reproductive process (Figure 2A). Then, KOG annotation was conducted by matching with similar protein sequences of multiple reference genomes for functional and evolutionary genome analysis to cover three categories: (1) metabolism, (2) information storage and processing, and (3) cellular processing and signaling ( Figure 2B). We identified the involvement of these proteins in the transport and metabolism of a large number of molecules, including lipids, inorganic ions, amino acids, carbohydrates, and coenzymes. Additionally, energy production and conversion were also highlighted ( Figure 2B).

Effect of Hypoxia on Differential Protein Expression
A comparative proteomic analysis was conducted to determine the differential protein expression in hypoxic brains of O. melastigma. Our results indicate that short-term hypoxia exposure could lead to the deregulated expression of 33 upregulated and 69 downregulated proteins in the F0 generation ( Figure 3A and Table 2). In the multi-generational exposure group, we observed the upregulation of 24 proteins and the downregulation of 52 proteins relative to the corresponding normoxic group ( Figure 3B and Table 3). Then, we compared the identified differentially expressed proteins with our previous report of transcriptome data . We found five and seven common deregulated genes-proteins in short-term and multi-generational hypoxia exposures, respectively ( Table 4).

Effects of Hypoxia on Neural Functions
To understand the neurotoxicological effects of hypoxia, the hypoxia-induced DEPs were subjected to GO and pathway enrichment analysis. Through GO enrichment analysis, we determined that short-term ( Figure 4A and Supplementary Table 2) and multi-generational ( Figure 4B and Supplementary  Table 3) hypoxia exposures could lead to significantly different alterations of the biological processes and molecular functions (p < 0.05). For biological processes, short-term hypoxia exposure affected a wide range of lipid metabolic processes such as phospholipids, glycerolipids, glycerophospholipids,   and phosphatidylinositols. Additionally, phosphate-containing compound metabolic processes, including phosphorus and organophosphate, were interrupted by short-term hypoxia exposure ( Figure 4A and Supplementary Table 2). Interestingly, multi-generational hypoxia exposure led to the alteration of the different acid biosynthetic processes such as carboxylic acids, organic acids, and cellular amino acids. Additionally, multi-generational hypoxia exposure could trigger a response to toxic substances and result in cell death and apoptosis. More importantly, multi-generational hypoxia exposure can dysregulate the immune system ( Figure 4B and Supplementary  Table 3). Through pathway enrichment analysis, we determined that short-term hypoxia exposure altered the arachidonic acid metabolism and tight junctions in the brain tissue (p < 0.05) (Figures 4C,D). Furthermore, this could lead to hypertrophic cardiomyopathy through the dysregulation of adrenergic signaling and to Alzheimer's disease (p < 0.05) (Figures 4C,D). On the other hand, multi-generational hypoxia exposure could lead to a larger number of alterations in the brain of O. melastigma (p < 0.05). This could trigger the development of neuronal diseases such as Alzheimer's disease and amyotrophic lateral sclerosis (p < 0.05) (Figures 4E,F). Additionally, multi-generational exposure can also lead to tuberculosis, atherosclerosis, and various cancers, which possibly occur through the dysregulation of certain signaling pathways, such as the p53 signaling pathway and the PI3K-Akt signaling pathway (p < 0.05) (Figures 4E,F). Taken together, our results suggest that short-term and multi-generational exposures could lead to different neural responses in the brains of fish.

Mechanisms Underlying the Neurotoxicological Effect of Hypoxia
To elucidate the possible molecular mechanisms behind the effects of hypoxia on brain toxicology in fish, ingenuity pathway analysis (IPA) was conducted. The canonical pathway analysis in IPA highlighted both similar and different responses of the brain to short-term and multi-generational hypoxia exposures. The results indicate that both short-term and multi-generational hypoxia exposures could lead to the alteration of integrin signaling, paxillin signaling, Parkinson's signaling, epithelial adherens junction signaling, and D-myo-inositol (1,4,5)-trisphosphate degradation and metabolism (Table 5). However, only integrin subunit alpha 11 (ITGA11) and inositol polyphosphate-5-phosphatase J (INPP5J) were found to be commonly deregulated under these two hypoxia treatments. Furthermore, certain pathways that serve important roles in brain function were observed to be unique to either short-term or multi-generational hypoxia exposure. Under short-term hypoxia exposure, ILK signaling, tight junction signaling, calcium signaling, integrin signaling, RhoA signaling, CXCR4 signaling, thrombin signaling, mTOR signaling, calcium transport I, protein kinase A signaling, and PAK signaling were associated with neurological diseases or functions and were highlighted in the analysis. These signaling pathways were controlled by a cluster of proteins including Actn3, MYH6, MYL2, MYL4, PPP2R3A, TNNT3, TTN, PRKD1, PRKD1, RPS27, and ANXA5 (Table 6 and Supplementary  Table 4). Under multi-generational hypoxia exposure, another group of pathways including anandamide degradation, GABA receptor signaling, the sumoylation pathway, pentose phosphate pathway, sphingosine-1-phosphate signaling, endocannabinoid neuronal synapse pathway, and endoplasmic reticulum stress pathway were altered. The DEPs involved in these pathways included FAAH, CACNA1C, SLC6A13, RFC5, RHOA, TKT, and CASP3 (Table 7 and Supplementary Table 5). These data suggest that short-term and multi-generational hypoxia exposures could lead to both common and unique mechanisms of neurotoxicity in marine fish.

DISCUSSION
Hypoxia is a pressing global problem facing marine ecosystems, and it is believed that this problem cannot be solved in the near future due to global warming. Studies have demonstrated that hypoxia is an endocrine disruptor, affecting both male and female reproductive systems and interfering with the hormonal reproductive hormonal system through the HPG axis. As the core of the HPG axis, the brain was reported to be interrupted by hypoxia. We have previously shown that hypoxic stress induced transcriptomic changes in the brain of O. melastigma. However, studies of the hypoxic responses of proteomes in other organs and tissues of fish remain limited. For example, in the heart of goldfish, hypoxia deregulated five enzymes involved in glycolysis and gluconeogenesis, including fructose-1,6-bisphosphate aldolase (Imbrogno et al., 2019). Presently, we attempted to elucidate the proteomic changes of fish brains after short-term and multi-generational hypoxia exposures.
In the first part of our study, we identified proteins in the brain of O. melastigma for functional analysis, which demonstrated that these proteins contribute to important brain FIGURE 4 | Continued functions. These functions include signal transducer activity, which is important for the functioning of neurotransmitters and hormones in regulating neuronal functions (Cooper, 2000). One possible outcome is the onset of chronic neurodegenerative diseases. For example, Alzheimer's disease was highlighted in our results. More importantly, we identified some proteins that were associated with reproductive processes, including WBP2 N-terminal like, double-strand break repair protein,   male-enhanced antigen 1, spindlin, and synaptonemal complex protein 1. Of the identified proteins, WBP2 N-terminal like was reported to serve an important role in spermatozoa and oocyte activation (Hamilton et al., 2018;Gonzalez-Castro et al., 2019). The other identified reproductive protein, doublestrand break repair protein, is responsible for generating either non-crossovers or crossovers during the process of homologous recombination in meiosis (Gray and Cohen, 2016). We also identified the male-enhanced antigen 1 (mea1) in the brain of medaka. Notably, the mea1 protein was reported to be expressed specifically in elongated spermatids and residual bodies, which suggests a functional role in  Pan et al., 2014;Hussain and Macklin, 2017 Tight junction signaling MYH6, MYL2, MYL4, PPP2R3A Luissint et al., 2012;Tietz and Engelhardt, 2015 Calcium signaling MYH6, MYL2, MYL4, TNNT3 Heizmann, 1993;Kawamoto et al., 2012;Bading, 2013 RhoA signaling MYL2, MYL4, TTN Govek et al., 2005;Gross et al., 2007;Stankiewicz andLinseman, 2014 CXCR4 signaling MYL2, MYL4, PRKD1 Jiang et al., 2013;Cheng et al., 2017;Bonham et al., 2018 Thrombin signaling MYL2, MYL4, PRKD1 Wang and Reiser, 2003;Ben Shimon et al., 2015;Krenzlin et al., 2016 mTOR signaling PPP2R3A, PRKD1, RPS27 Graber et al., 2013;Lee, 2015;Ryskalin et al., 2017 Calcium transport I ANXA5 Lin and Way, 1984;Heizmann, 1993;Bading, 2013 Protein kinase A signaling MYL2, MYL4, PRKD1, TTN Gervasi et al., 2007;Huang et al., 2011;Dagda and Das Banerjee, 2015 PAK signaling MYL2, MYL4 Kreis and Barnier, 2009;Civiero and Greggio, 2018 Vogel et al., 1993;Weidenfeld et al., 1994;Scherma et al., 2019 GABA receptor signaling CACNA1C, SLC6A13 Owens and Kriegstein, 2002;Luján et al., 2005;Wu and Sun, 2015 Sumoylation pathway RFC5, RHOA Schorova and Martin, 2016;Peters et al., 2017;Bernstock et al., 2018 Pentose phosphate pathway TKT Brekke et al., 2014;Bouzier-Sore and Bolaños, 2015 Sphingosine-1-phosphate signaling CASP3, RHOA Prager et al., 2015;Grassi et al., 2019 Endocannabinoid neuronal synapse pathway CACNA1C, FAAH Castillo et al., 2012;Augustin and Lovinger, 2018 Endoplasmic reticulum stress pathway CASP3 Raghubir et al., 2011;Hashimoto and Saido, 2018;Tan et al., 2018 spermiogenesis (Ohinata et al., 2002). Spindlin, a maternal effect factor associated with the meiotic spindle during the meiosis of oocytes (Zhang et al., 2015), is responsible for the oocyte-to-embryo transition in carp fish (Sun et al., 2010). Synaptonemal complex protein 1 (SYP-1), which is localized to the short arms of chromosomes, serves a role in recruiting the chromosomal passenger complex and phosphorylated histone H3 to the short arm of chromosomes during meiosis I, thereby promoting the disjunction of chromosomes in meiosis I (Sato-Carlton et al., 2017). After elucidating the functions of the identified proteins, we investigated the adverse effects of both short-term and multigenerational hypoxia exposures. Upon comparing the results, we observed some common deregulation of the cell signaling pathways, including integrin signaling, paxillin signaling, and Parkinson's signaling, that are related to neural functions and diseases caused by both short-term and multi-generational hypoxia exposures. Integrins are a class of transmembrane receptors involved in cell adhesion to the extracellular matrix (ECM) (Giancotti, 1999) and are expressed in various brain sections, including the hippocampus and cortex (Pinkstaff et al., 1999). As receptors, integrins play a crucial role in the pathophysiology of many brain diseases via signal transmissions from the ECM to the cell. Notably, aberrations in integrin signaling within the brain may impair the synaptic activity in brain disorders such as schizophrenia and epilepsy (Wu and Reddy, 2012). Paxillin is an adaptor protein for focal adhesions that coordinates the integrin-mediated rearrangements of the cytoskeleton (Rashid et al., 2017). In human cancer research, paxillin was identified as an oncogene that promotes cell migration and invasion in gliomas (Chen et al., 2016).
Additionally, hypoxia could induce the invasive potential of glioblastomas through the HIF-2α-EPHB2-paxillin axis, thereby suggesting a possible interaction between hypoxia and paxillin signaling (Qiu et al., 2019). Parkinson's disease is a progressive neurodegenerative disorder (Beitz, 2014). Recently, it has been reported that oxidative stress may play a role in Parkinson's disease (Islam, 2016). Moreover, numerous studies on signaling in Parkinson's disease models have demonstrated a link between hypoxia-inducible factors and Parkinson's disease (Kim et al., 2020;Zhang et al., 2020). Therefore, the results of our study further suggest hypoxia as a possible cause of Parkinson's disease.
Apart from the common effects, both short-term and multi-generational hypoxia exposures could lead to unique neural alterations. Notably, the metabolisms of various classes of lipids (e.g., phospholipids, glycerophospholipids, and phosphatidylinositols) were found to be affected by shortterm hypoxia exposure. Phospholipids (glycerophospholipids and sphingolipids) are major structural components of the plasma membrane of various cell types of the brain, including neurons and glial cells (Adibhatla and Hatcher, 2007;Kosicek and Hecimovic, 2013). Neural membrane function can be affected by their glycerophospholipid composition (Farooqui et al., 2000). Imbalances of phospholipid metabolism have previously been reported in cases of Alzheimer's disease (Kosicek and Hecimovic, 2013). Furthermore, short-term hypoxia exposure also affected arachidonic acid metabolism in the fish brain. Adequate levels of arachidonic acid, a polyunsaturated fatty acid, are required for normal cerebral functions and structures, including synaptic signaling and neurotransmitter release (Ojeda et al., 1989;Kaufmann et al., 1996;Contreras and Rapoport, 2002). Altered brain arachidonic acid metabolism has previously been associated with several neurological, neurodegenerative, and brain diseases, including schizophrenia and Parkinson's and Alzheimer's disease (Müller et al., 2002;Teismann et al., 2003;Esposito et al., 2008). Unlike the responses to short-term hypoxia exposure, multigenerational hypoxia exposure could lead to the alteration of the carboxylic acid metabolism in the fish brain. Carboxylic acids are a class of organic acids characterized by a carboxyl (-COOH) functional group. Various carboxylic acids have been reported to play important roles in neural functions. For example, 5-methoxyindole-2-carboxylic acid can mitigate brain damage from stroke (Sumien et al., 2019), while losartan carboxylic acid, the potent AT1 blocker metabolite of losartan, improved motor coordination and enabled central nervous system protection from an increase in locomotor activity (Subudhi et al., 2018). Apart from the deregulation of organic acid metabolism, long-term hypoxia exposure could also alter certain cell signaling pathways in the brain, such as PI3K-Akt signaling and p53 signaling. PI3K-Akt is an important signaling pathway in cell proliferation and cell cycle control (Yu and Cui, 2016). It has been reported that aberrations in PI3K-Akt signaling may be involved in the pathogenesis of Alzheimer's disease (Gabbouj et al., 2019) and brain malformation (Jansen et al., 2015). Notably, an Alzheimer's disease mouse model demonstrated that the activation of PI3K-Akt signaling could mitigate neurodegeneration and memory impairment (Ali et al., 2017). For example, the transcription factor p53, a mediator of cellular response to stress, is required for normal brain development (Contreras et al., 2018). More importantly, p53 is the key mediator of autophagy and apoptosis, processes that are known to be dysregulated in ALS and Parkinson's and Alzheimer's diseases (Ghavami et al., 2014). Interestingly, some cancer studies reported that the p53 and Akt pathways were linked to the modulation of apoptosis under hypoxic conditions (Leszczynska et al., 2015;Yun and Glazer, 2015). In line with the above results, ALS, which is characterized by the deterioration of motor neurons (McCombe et al., 2020), was also highlighted in our analysis. Therefore, multi-generational hypoxia exposure could result in behavioral disorders in fish via neurodegeneration. Due to the harmful effects of this potential interaction, future studies should be conducted to address this issue.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: http://www.ebi.ac.uk/ pride/archive/projects/PXD022579.

ETHICS STATEMENT
The animal study was reviewed and approved by Committee on the Use of Live Animals in Teaching and Research (CULATR, #2714-12) at the University of Hong Kong.

AUTHOR CONTRIBUTIONS
KL and RK contributed to the conception, organization, and execution of the research project. SW and NT contributed to the experimental setup and sample preparation. YT, JZ, WT, XL, TC, and DA contributed to the bioinformatic analysis and statistical analysis. RW contributed to manuscript preparation. All authors contributed to the article and approved the submitted version.