Abstract

   Hypobaric hypoxia causes altitude sickness and significantly affects
   human health. As of now, focusing on rats different proteomic and
   metabolic changes exposed to different hypoxic times at extreme
   altitude is blank. Our study integrated in vivo experiments with tandem
   mass tag (TMT)- and gas chromatography time-of-flight (GC-TOF)-based
   proteomic and metabolomic assessments, respectively. Male
   Sprague-Dawley rats were exposed to long-term constant hypoxia for 40
   days or short-term constant hypoxia for three days, and their responses
   were compared with those of a normal control group. Post-hypoxia, serum
   marker assays related to lipid metabolism revealed significant
   increases in the levels of low-density lipoprotein (LDL), triglycerides
   (TG), and total cholesterol (TC) in the liver. In contrast,
   high-density lipoprotein (HDL) levels were upregulated in the long-term
   constant hypoxia cohorts and were significantly reduced in the
   short-term constant hypoxia cohorts. Furthermore, metabolic pathway
   analysis indicated that glycerolipid and glycerophospholipid
   metabolisms were the most significantly affected pathways in long-term
   hypoxia group. Subsequently, RT-qPCR analyses were performed to
   corroborate the key regulatory elements, including macrophage
   galactose-type lectin (MGL) and Fatty Acid Desaturase 2 (FADS2). The
   results of this study provide new information for understanding the
   effects of different hypobaric hypoxia exposure protocols on protein
   expression and metabolism in low-altitude animals.

   Keywords: Hypobaric hypoxia, Proteomics, Extreme altitude, Metabolomics

Highlights

     * •
       Proteomic and metabolomic analysis of rat liver under hypoxia
       conducted.
     * •
       Differential metabolites and proteins were found at different
       hypoxic time.
     * •
       Rats under two hypoxic protocols based on two omics were compared.

1. Introduction

   Hypoxia, predominantly occurring in high-altitude regions, is commonly
   induced by a deficient oxygen supply, leading to widespread alterations
   in protein expression across various human tissues [[47]1].
   Unacclimatized people traveling to altitudes >2500 m commonly suffer
   from acute altitude sickness, with symptoms often manifested as
   follows: headache, gastrointestinal symptoms, fatigue and/or weakness,
   and dizziness/light-headedness [[48]2]. According to estimates, almost
   over 500 million humans live at ≥1500 m, 14.4 million at ≥3500 m,
   although only 0.31 million at ≥5000 m [[49]3]. There are variations in
   the muscles of professional climbers returning from extremely high
   altitudes (>5500 m) in terms of mitochondrial volume density,
   subcellular distribution, oxidative enzyme activity, and protein
   expression [[50]4].

   Gut microbiota-metabolites-neuroendocrine changes in male rats
   short-term constant exposure to simulated altitude of 5500 m [[51]5],
   which is closely related to the body's metabolism. At different time
   (1–12 weeks) of exposure to hypobaric hypoxia at an altitude of 5500 m,
   the secretion of atrial natriuretic peptide and brain natriuretic
   peptide in rats will change, which is involved in cardiovascular
   homeostasis [[52]6]. Long-term hypoxia may lead to metabolic
   alterations including diminished mitochondrial function and enhanced
   glycolysis. Hypoxia-inducible factor 1α (HIF-1α) plays a central role
   in regulating hepatic lipid metabolism. The expression of HIF-1α in the
   liver promotes lipid uptake and synthesis while decreasing lipid
   oxidation, consequently ultimately contributing to hepatic steatosis
   [[53]7]. Increased expression of HIF-1α in the liver and serum under
   hypoxic conditions has been found [[54]8,[55]9]. In individuals with
   cardiorespiratory illnesses, short-term constant hypoxia diminishes the
   synthesis of the cytochrome P450 subfamily members, thereby increasing
   the risk of heart failure. Short-term hypoxia exposure may hinder the
   P450 subfamilies in the liver from metabolising medications, thus,
   potentially posing a risk to patients.

   The liver plays a pivotal role in most physiological processes and is
   essential for maintaining homeostasis. Recognised for its critical
   function in metabolising glucose, glycogen, and lipids, the liver
   provides energy-generating fuels to the adjacent tissues, thereby
   ensuring stability under extreme altitude conditions.

   Maladaptation caused by long-term constant hypoxia is closely linked to
   the liver metabolic status [[56]10]. For instance, long-term constant
   hypoxia is associated with the pathogenesis of nonalcoholic
   steatohepatitis (NASH), which is a prevalent liver disease.
   Nonalcoholic fatty liver disease (NAFLD) is thought to contribute to
   the progression of NASH, with alterations in lipid metabolism induced
   by short-term constant hypoxia linked to the activation of key enzymes
   and metabolic pathways. This suggests that oxidative stress and lipid
   peroxidation are critical factors that modulate NAFLD.

   Tandem mass tag (TMT)-based proteomic analysis involves labelling all
   proteins expressed in a genome or complex mixture. For TMT-based
   proteomics investigations on adaptation to high-altitude hypobaric
   hypoxia exposure, plasma proteome profiles and differentially expressed
   proteins have been reported in people at high and low altitudes
   [[57]8], and plasma proteomic changes have also been found in people
   exposed to different hypoxia time [[58]11]. Wang et al. identified 19
   differential proteins that play an important role in the occurrence and
   development of high-altitude polycythemia (HAPC) in plasma samples from
   HAPC patients and healthy controls [[59]12]. In addition,
   hypoxia-related changes in the proteins expression of rats for 35 days
   have been observed in our previous study [[60]13].

   Metabolomics is also widely used in the study of high altitude.
   Metabolomics has played an important role in the study of urine
   biomarkers of rats under hypobaric hypoxia [[61]14]. Further, urinary
   metabolites of humans that may be used to identify acute mountain
   sickness susceptible subjects have been reported through the use of
   metabolomics [[62]15]. Differential metabolites in serum that can be
   used as early warning indicators of acute injury are found under
   conditions of high altitude hypoxic exposure [[63]16]. Based on the
   above studies, we hypothesized that there may be differences in
   expressed proteins and metabolites in rats after hypoxia.

   Compared to our previous study using only proteomics and a single
   long-term hypoxia protocol (a consecutive hypoxia for 35 days), in this
   study, we combined animal experiments under different hypoxia protocols
   with TMT-based proteomic analysis and metabolomic profiling to
   elucidate the metabolic alterations caused by oxygen condition at
   extreme altitude. Therefore, this study will be more comprehensive and
   our findings revealed that hypoxia induced metabolic pathway
   alterations in the liver based on changes in metabolites and expressed
   proteins. This will provide a scientific basis for confirming
   biomarkers for hypobaric hypoxia-induced diseases and understanding of
   the effects of hypoxia on metabolic pathways alterations in future
   human studies, although there are some differences between these animal
   data and human metabolic results.

2. Results

2.1. Plasma lipid metabolism profiles and basic information

   A Comprehensive analysis of metabolic markers and serum lipids in the
   short-term and long-term hypoxia groups was performed on Day 3 and Day
   40, respectively. As widely used liver enzyme, alanine aminotransferase
   (ALT) is a direct indication of liver health. On the other hand, the
   blood glucose (GLU) test, reflects the glucose level at the moment of
   sampling and is mainly used for the diagnosis of diabetes mellitus and
   to rule out the possibility of diabetes. Under different hypoxic
   conditions, our analysis using instruments chemie (Delfzijl, the
   Netherlands) kits found no significant changes in serum ALT or GLU
   levels. Moreover, TG, TC, and LDL levels were higher in the hypoxic
   group than in the normoxic group (P < 0.05) ([64]Fig. 1). Conversely,
   the HDL levels in the short-term hypoxia group showed a different
   pattern from those in the long-term constant hypoxia group ([65]Fig.
   1A,B). In particular, LDL levels were approximately doubled in the
   long-term constant hypoxia group compared to the normoxic group
   ([66]Fig. 1B). The effects of 40-day hypobaric hypoxia on rat well
   being, food and water intake, body weight, and ALT are shown in
   [67]Table 1. These findings suggest that hypoxia may significantly
   influence lipid and glucose metabolism.

Fig. 1.

   [68]Fig. 1
   [69]Open in a new tab

   Plasma lipid metabolism profiles. Serum levels of LDL, HDL, TG, CHO,
   GLU (n = 8). (A)Short-term hypoxia group vs normoxia. (B)Long-term
   hypoxia group vs normoxia ∗ P < 0.05.

Table 1.

   Comparison of rats under long-term hypoxia versus normoxia on Day 40.
       Variables     Normoxia (n = 8) Long-term hypoxia (n = 8) p-values
   Well-being[70]^a  +                –
   Food intake (g)   49.38 ± 1.98     47.98 ± 1.18              0.25
   Water intake (mL) 37.0 ± 1.5       8.5 ± 0.7                 <0.001
   Body weight (g)   388.98 ± 15.53   189.39 ± 6.87             <0.001
   ALT (U/L)         53.48 ± 1.82     54.57 ± 3.95              0.34
   [71]Open in a new tab
   ^a

   Liver histological evaluation was used to assess the health of the
   rats. +: healthy, -: obvious histological changes.

2.2. HE staining of liver tissue

   HE staining of right lobe of rat liver tissue in each group was
   observed under a microscope (400 × magnification) on Day 3 and Day 40
   ([72]Fig. 2). In the control group, normal liver cell morphology was
   observed, with no morphological changes of the central vein or hepatic
   sinusoids. In the hypoxic group, the volume of hepatocytes increased
   and many vacuoles of different sizes appeared in the cytoplasm under
   the use of a micrometer; however, the nuclear structure remained normal
   and neither the central veins nor the hepatic sinus were dilated or
   contracted.

Fig. 2.

   [73]Fig. 2
   [74]Open in a new tab

   HE staining of liver tissue (magnification 400×). This analysis
   revealed changes in the volume of hepatocytes, the presence of vacuoles
   in the cytoplasm. (A) Normoxia (B) Short-term hypoxia (C) Long-term
   hypoxia.

2.3. Rat liver metabolomics analysis

2.3.1. Rat liver metabolomics profiles from GC-TOF/MS data analysis

   Principal component analysis (PCA) was conducted on the GC-TOF spectra
   of liver samples obtained from animals subjected to hypoxic and
   normoxic conditions ([75]Fig. 3A and B). The PC score graphs exhibited
   a distinct segregation from those of the short-term and long-term
   constant hypoxic, and normoxic groups. Orthogonal Partial Least Squares
   Discriminant Analysis (OPLS-DA) ([76]Fig. 3C,D) and OPLS-DA permutation
   plots were used to identify the most distinctive variables defining
   grouping patterns. In the short-term constant hypoxia group, R2Y was
   found to be 0.89 with a Q2 intercept of −0.45, and the majority of R2
   and Q2 values on the right side exceeded those on the left side.
   Furthermore, Q2 intersects with the negative y-axis, thereby indicating
   that the model is reliable and free of overfitting. When R2 exceeds 0
   and Q2 is below 0, the model is not subject to overfitting, and is
   deemed reliable [[77]10]. The quality parameters of the OPLS-DA model
   suggested that the first seven components constituted the most
   effective cluster classification ([78]Fig. S1A). Additionally, the
   long-term constant hypoxia groups exhibited favourable model quality
   characteristics in the OPLS-DA analysis, with R2 and Q2 values of 0.88
   and −0.48, respectively. These values indicated that the first seven
   components were the most effective in classifying the clusters
   ([79]Fig. S1B). Volcano plots were constructed to identify
   significantly upregulated and downregulated metabolites that defined
   the clustering patterns ([80]Fig. 3E and F).

Fig. 3.

   [81]Fig. 3
   [82]Open in a new tab

   Effect of hypobaric hypoxia on differential metabolites in rat livers
   (n = 8)

   (A) PCA score scatter plot of normoxia vs short-term hypoxia. (B) PCA
   score scatter plot of normoxia vs long-term hypoxia (C) The OPLS-DA
   statistical analysis of normoxia vs short-term hypoxia (D) The OPLS-DA
   statistical analysis of normoxia vs long-term hypoxia (E) Volcano plot
   of significant metabolites between short-term constant hypoxia and
   normal groups. (F) Volcano plot of significant metabolites between
   long-term constant hypoxia and normal groups. (G) The top 34
   significantly up/down regulated metabolites in heatmap of short-term
   hypoxia group. (H) The top 23 significantly up/down regulated
   metabolites in heatmap of long-term hypoxia.

2.3.2. Identification of the metabolites

   A number of differentially expressed liver metabolites were selected
   using the criteria of VIP >1.0 and P-value<0.05. In the short-term
   hypoxia group, 34 significantly altered metabolites were found, of
   which 17 had elevating levels, such as cortisone, pantothenic acid, and
   3-hydroxybutanoic acid, and 17 had decreasing levels, such as L-threose
   2, oxalacetic acid, and sorbose 2 ([83]Table S1A). In the long-term
   hypoxia group, 23 significantly altered metabolites were found, of
   which behenic acid, trans-sinapinic acid, alanine 1, and glucose 1
   levels were downregulated and 19 metabolites, such as tartaric acid,
   glutaric acid, and 2-hydroxybutanoic acid were upregulated ([84]Table
   S1B). The heatmap distinctly illustrates the top 34 and 23
   significantly upregulated and downregulated metabolites in the
   short-term ([85]Fig. 3G) and long-term hypoxia groups ([86]Fig. 3H),
   respectively.

2.4. Rat liver proteomics analysis

2.4.1. Differentially expressed proteins

   To study the differences in liver proteins between the hypoxic and
   normoxia groups, we performed the analysis by TMT. The number of
   identified proteins was 5740 ([87]Table S2A). To screen differential
   proteins, the standard was set as fold change≥1.5 or fold change≤2/3. A
   total of 130 and 183 differentially expressed proteins were identified
   in the short-term and long-term hypoxia group, respectively([88]Tables
   S2B and 2C). Among the 130 differential proteins, 77 proteins were
   significantly up-regulated and 53 proteins were significantly
   down-regulated. Among the 183 differential proteins, 128 were
   significantly up-regulated and 55 were significantly down-regulated.
   These results suggest that hypobaric hypoxia has an effect on liver
   protein expression.

2.4.2. Gene ontology (GO) enrichment analysis of differentially expressed
proteins

   We performed a GO analysis to explore the potential roles of the
   differentially expressed proteins following exposure to constant
   hypoxic conditions. The analysis revealed the presence of several
   cellular components, including extracellular regions, membrane-enclosed
   lumens, macromolecular complexes, and cell junctions ([89]Fig. 4).
   Concerning molecular functions, our analysis identified enriched
   categories including catalytic activity, binding, transporter activity,
   and antioxidant activity ([90]Fig. 4). The identified enriched
   biological processes included cellular processes, single-organism
   processes, biological regulation, metabolic processes, and
   developmental processes ([91]Fig. 4). These findings suggest that
   constant hypoxia significantly affects metabolic processes and elicits
   responses to hypoxic stimuli in the rat liver.

Fig. 4.

   [92]Fig. 4
   [93]Open in a new tab

   GO-Standard. GO analysis revealed the presence of genes significantly
   enriched in specific cell components, molecular functions or biological
   processes during hypoxia. (A) Between short-term hypoxia and normoxia
   groups. (B) Between long-term hypoxia and normoxia groups.

2.4.3. Co-expression network analysis of differential expressed proteins

   Through weighted gene co-expression network analysis of bioinformatics
   analysis, we plotted gene co-expression network and core genes.
   [94]Fig. S2 illustrates the co-expression network of proteins in rat
   livers that exhibited significant differential expression following
   hypoxia. The primary proteins identified include solute carrier family
   24 member 1 (SLC24A1), aldehyde dehydrogenase 2 (ALDH2), ATP-citrate
   lyase (ACLY), acetyl-Coenzyme A carboxylase beta (ACACB), aldo-keto
   reductase 1C3 (AKR1C3), cytochrome P450 family 1 subfamily A member 1
   (CYP1A1), cytochrome P450 family 1 subfamily A member 2 (CYP1A2),
   cytochrome P450, family 4, subfamily A, polypeptide 8 (CYP4A8),
   acyl-CoA synthetase medium-chain family member 2 (ACSM2),
   ras-associated protein 1 (RAP1A), interferon-stimulated Gene 15
   (ISG15), phosphoglycerate dehydrogenase (PHGDH), 17β-hydroxysteroid
   dehydrogenase 2 (HSD17B2), and arachidonic acid 15-lipoxygenase-1
   (ALOX15). These proteins play critical roles in fatty acid metabolism.

2.5. Pathway analysis

   We conducted Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis,
   pathway analysis, KEGG-pathway enrichment, and GO enrichment of
   bioinformatics analysis to enhance our understanding of the
   intracellular pathways affected by hypobaric hypoxia. In the
   KEGG-pathway enrichment analysis of proteomics, both steroid hormone
   biosynthesis, linoleic acid metabolism, and arachidonic acid metabolism
   of the two hypoxic groups were confirmed ([95]Fig. 5A and B). This is
   partially consistent with the results of pathway analysis ([96]Table
   S3A) and KEGG analysis ([97]Table S4A) of metabolomics in the
   short-term hypoxia group. In the short-term hypoxia group, ascorbate
   and aldarate metabolism confirmed([98]Fig. 5A) were also identified in
   KEGG analysis of metabolomics ([99]Table S4A). Unsaturated fatty acid
   metabolic process was determined in GO enrichment analysis of
   proteomics ([100]Fig. S3A) and pathway analysis of metabolomics
   ([101]Table S3A). In addition, the regulation of lipolysis in
   adipocytes and ABC transporters confirmed in the long-term hypoxia
   ([102]Fig. 5B) were reconfirmed in the KEGG analysis of the
   metabolomics ([103]Table S4B).

Fig. 5.

   [104]Fig. 5
   [105]Open in a new tab

   Pathway analysis. (A)KEGG-pathway analysis of short-term hypoxia vs
   normoxia. (B) KEGG-pathway analysis of long-term hypoxia vs normoxia.
   (C) Metabolic pathways with red(up-regulated)/blue (down-regulated)
   dots representing the differentially expressed compounds of short-term
   hypoxia vs normoxia. (D) Metabolic pathways with red(up-regulated)/blue
   (down-regulated) dots representing the differentially expressed
   compounds of long-term hypoxia vs normoxia. (E) Metabolic pathway
   analysis bubble plot of short-term hypoxia vs normoxia. (F) Metabolic
   pathway analysis bubble plot of long-term hypoxia vs normoxia.

   By KEGG analysis of metabolomics, differential metabolites in the
   short-term and long-term hypoxia groups were labeled in the metabolic
   pathway, especially retinol metabolism and chemical carcinogenesis were
   confirmed in both groups ([106]Fig. 5C and D). Based on the metabolic
   pathway analysis, in short-term constant hypoxia group, linoleic acid
   metabolism was most affected by metabolic pathway analysis ([107]Fig.
   5E).

   Correspondingly, it was evident that the glycerolipid and
   glycerophospholipid metabolism pathways were the most significantly
   affected in long-term hypoxia group ([108]Fig. 5F).

   Through GO enrichment analysis of proteomics, we identified the top 30
   biological processes in the short-term hypoxia group and the long-term
   hypoxia group, respectively ([109]Fig. S3). Through our analysis of the
   proteomics data, we identified protein precursors that play significant
   roles in many metabolic pathways, including fatty acid degradation and
   histidine metabolism in short-term hypoxia group ([110]Fig. 5A) and
   fatty acid metabolism in long-term hypoxia group ([111]Fig. 5B). These
   findings indicate that hypobaric hypoxia has a major impact on pathways
   in rat livers.

2.6. Verification of differentially expressed proteins between control and
hypoxic samples using real-time quantitative polymerase chain reaction
(RT-qPCR)

   To verify the quantitative data obtained by TMT proteomics, RT-qPCR was
   performed. The selection criteria of genes were: a) up- or
   down-regulation in all replicate analyses of proteomics; and b)
   associated with metabolic pathways; and c) potential as a biomarker
   after hypoxic exposure at high altitude. In the short-term constant
   group, [112]Fig. 6A illustrates the gene expression levels of Fatty
   Acid Desaturase 2 (FADS2), Angiopoetin-like 4 (ANGPTL4), Solute Carrier
   Family 24 Member 1 (SLC24A1), Acyl-CoA Synthetase Medium Chain Family
   Member 2 (ACSM2), Cytochrome P450 Family 1 Subfamily A Member 1
   (CYP1A1), and Cytochrome P450 Family 1 Subfamily A Member 2 (CYP1A2).
   In the long-term constant hypoxia group, adipose triglyceride lipase
   (ATGL), macrophage galactose-type lectin (MGL), long-chain acyl-CoA
   synthetase (ACSL), glycerol-3-phosphate acyltransferases (GPAT), and
   purple acid phosphatase (PAP) were selected to validate the proteomics
   results ([113]Fig. 6B). These findings are consistent with those of
   proteomic analysis. The reference gene against which expression was
   measured is GAPDH.

Fig. 6.

   [114]Fig. 6
   [115]Open in a new tab

   Verification of differentially expressed proteins by RT-qPCR (n=8). (A)
   The control vs short-term hypoxia group. (B) The control vs long-term
   hypoxia group (∗P < 0.05).

3. Discussion

   Recently, a TMT-liquid chromatography-tandem mass spectrometry
   (LC-MS/MS) method was developed for the identification and
   quantification of proteins. Yu et al. [[116]17] revealed diabetic foot
   ulcer pathogenesis and potential therapeutic targets by TMT-based serum
   proteomic profiling. Xia et al. [[117]18] employed TMT label-based
   proteomics to identify protein differences in the cerebral cortex of
   TauT-knockout rats. Consequently, systematic screening of deregulated
   proteins could provide profound insights into the molecular mechanisms
   underlying the responses to stimuli.

   Frequent exposure to high altitudes often leads to hypoxia. Zhao's
   study on the adaptation of Tibetan sheep lungs to high-altitude hypoxia
   using proteomic sequencing revealed key proteins and pathways related
   to gas transport, oxidative stress, and angiogenesis [[118]19]. The
   results linked to these proteins and pathways concurred with those of
   the physiological studies, thereby corroborating the physiological
   findings reported by Zhao.

   Metabolomics, emerging from proteomics, enables the identification of
   small-molecule metabolites and reflects changes in endogenous
   metabolites during progression. Chen et al. [[119]20] demonstrated that
   prolonged intermittent (O[2] from 21 % to 6 %,for 8 h per day, lasted
   12 weeks) hypoxia induces significant changes in lipid metabolism in
   the mouse liver, thereby leading to extensive liver damage. Yan et al.
   [[120]21] discovered that obese rats exhibited a significant increase
   in serum free fatty acid (FFA) and myocardial TG levels upon exposure
   to long-term constant persistent (O[2] 14 %, 4 weeks) hypoxia at an
   altitude of 4300 m for eight weeks.

   The liver is crucial for maintaining homeostasis and regulating
   metabolism. These processes require a significant amount of oxygen,
   which makes the liver highly sensitive to hypoxic conditions. It can be
   inferred that hypoxic conditions prompt changes in hepatic metabolism,
   particularly affecting lipid metabolism. The findings and insights
   inspired the design and execution of the present investigation.

   In this study, our investigation revealed no significant changes in ALT
   or GLU levels under hypoxic conditions (P < 0.05). Similarly, Song et
   al. [[121]22] observed no significant differences in GLU levels between
   normoxic and long-term high-altitude exposure groups (at 4300 m with
   O[2] 14 % for eight weeks). Thomas et al. [[122]23] showed that
   whole-body glucose tolerance, initially triggered by long-term
   intermittent hypoxia (2 weeks, 1-min cycle, fraction of inspired oxygen
   21–5%, 8 h/day), was effectively counteracted.

   In this study, GC-TOF metabolomics was used to explore liver metabolism
   under hypoxic conditions that are common at high altitudes.
   Specifically, we utilised advanced pattern recognition techniques,
   including PCA and OPLS-DA analysis, which revealed a clear distinction
   between hypoxia-exposed and normoxic control groups. Essential
   metabolites were analysed to explore metabolic pathways, revealing
   several key pathways related to lipid metabolism: glycerolipid
   metabolism, glycerophospholipid metabolism, and the regulation of
   lipolysis. Krishnan et al. [[123]24] found that hypoxia induces FFA
   uptake and glycerin biosynthesis by activating peroxisome
   proliferator-activated receptor (PPAR), leading to cardiac steatosis.
   Consistent with our findings, Liao et al. [[124]25] reported that
   glycerophospholipid content was 7.03-fold higher in hypoxic conditions.
   Systemic metabolic homeostasis preservation is partly achieved through
   lipolysis modulation, which is controlled by a complex regulatory
   network [[125]26]. In groups exposed to long-term hypoxia, as an
   integral part of the glycerolipid/free fatty acid (GL/FFA) cycle
   [[126]27], examination of differentially expressed proteins and
   differential metabolites by KEGG analysis revealed that the regulation
   of lipolysis in adipocytes was enriched, which is consistent with
   enhanced lipolysis of 3T3-L1 cells under hypoxia [[127]28]. We
   hypothesized that these pathways contribute significantly to changes in
   liver metabolism, as exemplified by increased alanine levels in rats
   subjected to hypoxia.

   Oxaloacetate and malic acid are considered to be the end products of
   the tricarboxylic acid cycle [[128]29]. In the short-term hypoxia
   group, a significant down-regulation of oxaloacetate may imply a
   reduction in levels of tricarboxylic acid (TCA) cycle in rats to some
   extent. In the long-term hypoxia group, glutaric acid, a precursor of
   α-ketoglutarate, was significantly increased compared with the normoxia
   group, and no significant changes in oxaloacetate and malic acid were
   found. In addition, 2-hydroxybutyrate has been shown to stimulate the
   expression of lipid-degrading proteins in hepatocytes and inhibit oleic
   acid-induced triglyceride accumulation [[129]30]. In this study, we
   found that 3-hydroxybutyrate, which is precursor of 2-hydroxybutyrate,
   and 2-hydroxybutyrate, were identified to be significantly elevated in
   the short-term and long-term hypoxia groups, respectively. Therefore,
   we hypothesized that rats may be as resistant to long-term
   hypoxia-induced lipid metabolism disorders as possible through the
   transformation of differential metabolites.

   We validated pecific genes that play crucial roles in lipid metabolism
   under various hypoxic conditions. Based on the results of the
   proteomics, in the long-term hypoxia group ([130]Fig. 6B), we
   investigated the GL/FFA cycle by measuring MGL, ACSL, and GPAT levels,
   which in turn alter lipid metabolism [[131]25]. MGL, ACSL, and GPAT
   levels were downregulated in the long-term hypoxia exposure group. MGL
   encodes a serine hydrolase belonging to the AB hydrolase superfamily
   and its main function is to facilitate the transformation of
   monoacylglycerides into free fatty acids and glycerol. Although MGL
   regulates the levels of fatty acids that serve as signalling molecules,
   it can potentially be inhibited by hypoxic exposure [[132]31]. ACSL
   catalyse the activation of long-chain fatty acids, thereby enabling
   their utilisation in the production of cellular lipids and breakdown
   through beta-oxidation. This process is necessary for the integration
   of fatty acids into phosphatidylcholine. Triacylglycerol (TAG)
   synthesis is initiated by sequential acylation of glycerol-3-phosphate
   by GPAT, which results in the formation of lysophosphatidic acid.
   Subsequently, the lysophosphatidic acid undergoes further
   esterification and transformation to phosphatidic acid [[133]32].
   Overall, MGL, ACSL, and GPAT are all related to glycerolipid
   metabolism. Our unique observations indicated that these variables were
   all downregulated after 40 days of hypoxic exposure. We speculate that
   the GL/FFA cycle is partially inhibited after 40 days of long-term
   hypoxic exposure, which might be involved in lipometabolic disorders.

   In the short-term hypoxia constant hypoxia group, the levels of CYP1A1
   and CYP1A2, both of which encode the cytochrome P450 superfamily of
   enzymes, were significantly downregulated. Short-term mild hypoxia
   lowers the overall content and activity of P450, as well as the in vivo
   expression of CYP1A1 and CYP1A2 [[134]33]. Cytochrome P450 proteins
   function as monooxygenases, and facilitate many processes involved in
   drug metabolism as well as the creation of cholesterol, steroids, and
   other lipids. These findings were confirmed by qPCR ([135]Fig. 6). The
   levels of FADS and ANGPTL4 were upregulated in the livers of the
   short-term hypoxia group compared to the control group. FADS2 belongs
   to the gene family of fatty acid desaturases, which encode delta-6
   desaturases (D6D), essential enzymes in the production of long chain
   polyunsaturated fatty acids (LC-PUFA) that regulate glucose and lipid
   metabolism in animals [[136]34]. ANGPTL4 is a secreted glycoprotein
   that is activated by peroxisome proliferator-activated receptors and
   plays a pivotal role in the physiological regulation of lipid and
   glucose metabolism [[137]35]. In summary, the genes mentioned above
   exhibited a significant correlation with metabolic pathways critical
   for oxidative stress, thus, providing invaluable insights and laying a
   solid groundwork for subsequent research.

   Disruption of hepatic oxygen homeostasis is associated with the
   aetiology of a broad range of liver diseases [[138]36].In line with our
   findings, hypoxia and HIF-1α strongly stimulate the production of
   hypoxia-inducible lipid droplet-associated (HILPDA) [[139]37], and in
   mice with severe liver disease such as NASH, HILPDA deficiency slightly
   reduces liver triglyceride storage [[140]38], suggesting hypoxia can
   affect lipid distribution in hepatocytes.

   It should be noted that the occurrence and severity of hypoxic
   responses in rats are known to vary significantly within populations,
   with two opposite extreme phenotypes–high resistance (HR) and low
   resistance (LR) to hypoxia [[141]39]. Mironova et al. [[142]40] found
   that in the cerebral cortex of LR rats, the number of mitochondria,
   which are tiny electron-dense, increased about threefold. Conversely,
   after this hypoxia exposure, the number of tiny mitochondria in the
   cerebral cortex of HR rats remained almost constant and remained
   significantly higher than that of control LR rats. Mironova et al.
   [[143]41] also analyze the mitochondrial ATP-dependent K^+ transport
   systems in rats with varying levels of short-term constant hypobaric
   hypoxia tolerance. Using NADH-dependent respiratory substrates, it was
   discovered that the HR rats had an ATP-dependent swelling rate 50–75 %
   higher than the LR animals. This suggests that the initial difference
   in hypoxic tolerance will result in different responses in rats under
   different hypoxia exposure treatments. At the same time, given the
   intricate mechanisms and pathways involved in liver metabolism, the
   precise regulatory mechanisms warrant extensive research for
   comprehensive elucidation.

   Our study demonstrated that hypoxia induces wide metabolic alterations
   in the liver. In response to hypoxia, rats maintained normal physical
   activity by downregulating biosynthetic processes, thereby reducing
   their energy expenditure. This will also provide a basis for research
   on specific issues that are common for the general population in plains
   to travel to extreme altitudes, such as the length of stay.

4. Methods

4.1. Animals and treatment

   Following a one-week acclimatization period during which the rats were
   fed a standard diet, they were subjected to constant hypoxic
   conditions. An in vivo experiment was performed according to previously
   reported methodologies [[144]21]. Male Sprague-Dawley rats were
   randomly assigned into three distinct groups (n = 8/group): normal
   group, the rats were exposed to normoxia; two hypoxia protocols were
   used, one group was subjected to long-term hypoxia for a continuous
   40-day period while the other group experienced acute hypoxia for a
   3-day period. Both hypoxic groups were subjected to hypobaric hypoxia
   at an altitude of 5500 m (380 mmHg). This was achieved using an
   FLYDWC50-1C low-pressure hypoxic experimental chamber (Guizhou Fenglei
   Air Ordinance LTD, Guizhou, China). Rats were housed under controlled
   environmental conditions: a stable temperature of 25 ± 2 °C, humidity
   of 50 ± 10 % and a consistent 12-h light/dark cycle. Rats are placed in
   rearing cages with regularly changed normal laboratory diet and
   sterilised water, while the cages are located in low-pressure oxygen
   chamber. On the final day, the rats were euthanised, and their blood
   and liver tissues were collected and subsequently stored at −80 °C.
   Sodium citrate was employed as anticoagulants. In this research, no
   deaths occurred in the rats. This study was reviewed and approved by
   the Ethics Committee of Medical Ethics Committee of Qinghai University
   School of Medicine with the approval number: 2019-ZJ-876. The work
   described has been carried out in accordance with the U.K. Animals
   (Scientific Procedures) Act, 1986. All animal studies comply with the
   ARRIVE guidelines.

4.2. Plasma lipid metabolism profiles

   The Vitros DT60 II Chemistry System (Johnson & Johnson, Minnesota, USA)
   was used to enzymatically measure total cholesterol levels. Lipoprotein
   particles were separated with a SMART LC system (Pharmacia Biotech,
   Uppsala, Sweden) using a Superose-6 column, which facilitated the
   measurement of LDL and HDL levels. Liver samples were homogenised with
   a Stir-Pak® (Barrington, IL). Triglyceride levels in the liver were
   quantified using a kit (Instruchemistry, Delfzijl, Netherlands).

4.3. HE staining

   The right lobe of rat liver was fixed in 4 % paraformaldehyde for 24 h,
   embedded in paraffin, and cut into 4 μm-thick slices for HE staining.
   The slices were baked at 60 °C overnight, then dewaxed with xylene and
   soaked in ethanol at various levels: xylene for 10 min; xylene for
   10 min; anhydrous ethanol for 5 min; 95 % ethanol for 5 min; 80 %
   ethanol for 5 min; 75 % ethanol for 5 min; distilled water for 2 min.
   Then the slices were stained with haematoxylin for 10 min, rinsed with
   tap water until there was no foam, fractionated with 1 % hydrochloric
   acid alcohol for 1 s, and rinsed with tap water for 5 min, rinsed with
   tap water for 5 min, and stained with eosin for 2 min at room
   temperature. Subsequently, gradient dehydration, transparency and
   sealing was carried out as follows: 50 % ethanol for 2 min; 75 %
   ethanol for 2 min; 95 % ethanol for 2 min; 95 % ethanol for 2 min;
   anhydrous ethanol for 2 min; anhydrous ethanol for 2 min; mixed
   solution of xylene and anhydrous ethanol (1:1) for 1 min; and xylene
   for 2 min. Neutral gum was used to seal the slices. Finally, 40 × and
   400 × ultra-structures of liver tissues were observed using light
   microscopy, and the results were analysed and photographed. Ten high
   magnification fields ( × 400) were observed for each animal.

4.4. The metabolomics analysis of rat liver samples

4.4.1. GC-TOF spectroscopic analysis

   In the case of groups exposed to long-term oxygen deprivation, liver
   samples were preserved by freezing and then combined with 200 lL of
   deuterated 0.2M sodium phosphate buffer (with a pH of 7.4) that
   included 1 mM TSP in D[2]O. TSP was used as a chemical shift reference
   (d = 0) and D[2]O was used as a lock solvent. The samples were
   centrifuged at 5000 rpm for 5 min to remove particulate matter. The
   supernatant was transferred to a sample tube. The GC-TOF spectra were
   obtained using a Bruker Avance III spectrometer (Bruker, Germany)
   equipped with a Bruker probe operating at 600 MHz and 298 K. A standard
   one-dimensional water pre-saturation pulse sequence (NOESYPR1D) was
   used to suppress the water signals.

   A total of 192 scans were conducted, each lasting 1.7 s, and split into
   64,000 data points across a9615.4 Hz spectral width. A 2-s relaxation
   delay was observed between the scans. Raw FIDs were processed with an
   exponential function by applying 0.3 Hz line broadening, followed by
   fast Fourier transformation. GC-TOF spectra from each sample were
   subjected to phase and baseline correction using a Bruker Topspin 2.1
   (Bruker, Germany). The TSP reference peak was set to d = 0. The spectra
   of normoxic and hypoxic groups were visually inspected.

4.4.2. Data reduction and pattern recognition (PR) analysis

   Following phase and baseline corrections, the GC-TOF spectra underwent
   additional adjustment to account for pH-dependent peak shifts using the
   icoshift (Interval Correlation Optimized Shifting) technique. Icoshift
   is a software tool that corrects spectral misalignments through an
   insertion/deletion model while adjusting the spectral intervals. It
   utilises fast Fourier transformation for the quick alignment of
   extensive GC-TOF datasets in seconds. Spectra were binned from
   d = 0.2–10 ppm using an optimized bucketing algorithm (OBA) with a
   0.01 ppm bucket width and 50 % slackness. The OBA adjusts the bucket
   sizes by finding the lowest data points to set the bucket borders based
   on the averaged GC-TOF spectra. Slackness allows border adjustments to
   determine the local minima in the spectrum. Regions containing urea
   (5.2–6.2 ppm) and water (4.2–5.2 ppm) were excluded from binning.

   The binned data were analysed using MetaboAnalyst 3.0. Normalisation
   and scaling used the sum of intensities and Pareto scaling,
   respectively, followed by a PR analysis. Liver metabolic profiles were
   analysed using PCA and OPLS-DA to identify similarities or differences.
   PCA is an unsupervised method that reduces complex data to principal
   components (PCs) with PCA score plots analysed for class-independent
   clusters. OPLS-DA, a supervised PR method, modelled spectral and class
   information links with its score plots examined for class-dependent
   clusters. The concentrations of polar metabolites (specifically, the
   top 25 significant buckets) in animal livers before and after exposure
   were calculated by manually integrating individual peaks and
   normalising them against the total spectra. These concentrations were
   visualised using box-and-whisker plots. Normalised concentrations were
   used for metabolic pathway analysis to evaluate the changes in response
   to hypoxia. Metabolic pathway analyses were performed using
   MetaboAnalyst 2.0 online. The metabolic pathway analysis employed a rat
   (Rattus norvegicus) pathway library, global testing, and out-degree
   centrality algorithms, thereby yielding fit factors and impact values
   via pathway enrichment and topology analysis.

4.5. The proteomics analysis of rat liver samples

4.5.1. Protein extraction and digestion

   In order to reduce the presence of lipids, proteins were removed
   utilizing a protein extraction kit (Promega). The extracted proteins
   were quantified using a BCA Protein Assay Kit (Bio-Rad, USA). Each
   sample, containing 1g of protein, was heated for 5 min with 24g of
   total protein, then subjected to electrophoresis on a 16 % stacking gel
   and an 8 % resolving gel, and stained with Coomassie Brilliant Blue.
   Overnight digestion with trypsin (Promega) in NH[4]HCO[3] at 37 °C
   followed. After the process of filtration, a total of approximately
   2400 μg of protein was collected.

4.5.2. TMT protein labelling and high pH reversed phase fractionation (HPRP)

   There were a total of 6 tubes used for all samples. According to the
   BCA test results, in each group, 100 μg of protein from 4 samples were
   randomly mixed and allocated to a 1.5 mL centrifuge tube. Thus, each
   group has two sets of biological replicates(n = 4∗2) [[145]42]. Sample
   preparation and labelling followed the protocol provided by the Pierce
   TMT® Mass Tagging Kits and Reagents Kit (catalogue number 90113). The
   high pH reverse-phase fractionation kit comprised a high pH buffer
   (0.1 % triethylamine) and twelve centrifugal columns with pH-resistant
   reverse-phase resin. Each column was used in a microcentrifuge to
   separate 10–100 μg of peptide samples, thereby simplifying sample
   complexity. Equal amounts of each sample were combined to yield 6
   samples. Subsequently, samples were collected and stored on ice.

4.5.3. LC-MS analysis

   Each sample received 6 injections for LC-MS/MS analysis,. Phase
   separation was performed by high-performance liquid chromatography
   (HPLC) system. Protein analysis was performed using Quality Engineering
   Plus (QE-Plus) software (Thermo Scientific, Waltham, USA). The initial
   MS settings included a resolution of 65,000 at m/z 200, a maximum
   injection time (IT) of 49 ms, and an automatic gain control (AGC) of
   1E6. Follow-up MS featured a resolution of 35,000.

4.6. Bioinformatic analysis

   Data were analysed using a comprehensive rat database. Bioinformatic
   data were processed using the Prisma software suite and R, an advanced
   statistical computing environment. Proteins were screened based on a
   fold-change ratio exceeding 1.30, with p-values below 0.05 considered
   statistically significant. Hierarchical clustering analysis facilitated
   the visualisation of protein expression levels. Enrichment analyses
   were conducted using R statistical software, targeting the KEGG and GO
   pathways (Cran Inc., USA). GO annotations were divided into three
   domains: biological processes (BP), cellular components (CC), and
   molecular functions (MF). KEGG links known molecular interactions
   including metabolic pathways, complexes, and biochemical reactions.
   Cluster analysis determined correlations among differentially expressed
   proteins based on GO categories, KEGG pathways, and protein domain
   enrichment.

4.7. RNA preparation, quality control and Real time PCR

   Rat tissues were subjected to RNA extraction and isolating using TRIzol
   reagent (Invitrogen, Beijing, China). Quality control of the extracted
   RNA was performed using a Thermo Nanodrop 3000 and an Agilent 2100
   Bioanalyzer, along with an Agilent RNA 6000 Nano Kit. High-quality RNA
   was analysed using microarray technology until it met the specified
   criteria; when using the Thermo Nanodrop 3000, the A260/A280 ratio was
   between 1.8 and 2.2. Furthermore, when using the Agilent 2100
   bioanalyzer, the RNA Integrity Number (RIN) should be 7 or higher, and
   the 28S/18S ratio should exceed 1.5. RNA was reverse-transcribed and
   converted into cDNA with RevertAid First Strand cDNA Synthesis Kit
   (Thermo Scientific™, K1622). PCR was performed by PCR Master Mix (2X)
   (Thermo Scientific™, K0171) and StepOne™ RT-PCR system (Applied
   Biosystems™, 4376357). List of gene-specific primers for RT-PCR
   analysis is presented in [146]Table S5.

4.8. Data analysis and statistics

   The findings were documented in a database and analysed using SPSS 22.0
   (SPSS Inc., Chicago, IL, USA). The median and quartile of each set of
   data were calculated. Pearson's correlations were calculated and for
   non-normally distributed variables, Kruskal-Wallis assessed differences
   in serum metabolite levels across time periods, compared to the
   baseline control before hypoxia exposure. The Dunn test and
   Bonferroni's multiple comparison post hoc tests were used for pairwise
   comparisons when the overall test was statistically significant, with
   the significance level set at P < 0.05.

Data availability statement

   The mass spectrometry proteomics data were deposited in Figshare with
   [147]https://doi.org/10.6084/m9.figshare.25376461
   ([148]https://figshare.com/s/b428f985a49dac3bc3a3).

   However, this study did not report the original codes. Further details
   necessary to reevaluate the data presented in this work can be obtained
   from the corresponding author upon enquiry.

Ethics and Consent Statement

   The animal study protocol was reviewed and approved by the Ethics
   Committee of Medical Ethics Committee of Qinghai University School of
   Medicine with the approval number: 2019-ZJ-876.

Funding

   Funding for this work was provided by the National Natural Science
   Foundation of China (Grant No. 82360333) and the Qinghai Provincial
   Department of Science and Technology, grant No.2023-ZJ-792.

Code availability

   No custom code has been used.

CRediT authorship contribution statement

   Jin Xu: Writing – original draft, Methodology, Conceptualization.
   Wen-jie Chen: Writing – original draft, Methodology. Han-bing Hu: Data
   curation. Zhi-wei Xie: Data curation. Dong-ge Zhang: Investigation. Jia
   Zhao: Data curation. Jing Xiang: Formal analysis. Qi-yu Wei:
   Investigation. Tawni Tidwell: Writing – review & editing,
   Investigation. Olivier Girard: Conceptualization, Writing – review &
   editing. Fu-hai Ma: Formal analysis. Zhao-wei Li: Methodology. Yan-ming
   Ren: Writing – review & editing.

Declaration of competing interest

   The authors declare that they have no known competing financial
   interests or personal relationships that could have appeared to
   influence the work reported in this paper.

Acknowledgements