Plasma Metabolomic Pro�ling of Individuals Susceptible to High Altitude through Gas Chromatograph-mass Spectrometry

Objective We aimed to characterize metabolic alterations of people ascending to high altitude and susceptible to acute mountain sickness (AMS). Methods Peripheral blood samples were collected from 36 healthy volunteers on the 3 rd day ascending to high altitude (4300m). AMS status was assessed using the Lake Louise Questionnaire. Plasma samples were characterized by gas chromatography-mass spectrometry (GC-MS) and principal component analysis (PCA) was used to discriminate the metabolite changes between sea level and high altitude status and between AMS group and non-AMS group. Results High altitude hypobaric hypoxia caused signi�cant and comprehensive metabolic changes in plasma, including 18 metabolites between sea level and high altitude, 6 metabolites between AMS group and non-AMS group. By using MetaboAnalyst 3.0, several key metabolic pathways were found to be involved, including cysteine and methionine metabolism, alanine, aspartate and glutamate metabolism. Conclusion The GC-MS pro�ling was a useful approach to analyze metabolites variances and provides potential targets for further investigation to understand the pathophysiological mechanism of hypobaric hypoxia and susceptibility to high altitude.


Introduction
In high-altitude plateau environments, the main physiological challenge is hypoxia.Upon rapid ascent to altitude above 3000 m, non-acclimatized healthy individuals are highly prone to contracting acute mountain sickness (AMS), high-altitude cerebral edema (HACE), or high-altitude pulmonary edema (HAPE) due to hypoxia 1,2 .With increasing number of people entering the plateau, AMS has become a health-related issue not only impairs human activity but also enhances the costs of health care.AMS is a transient condition and the routine diagnosis method is Lake Louise Score, which evaluates symptoms such as headache, weakness, sleeplessness, etc 3 .Despite the detailed clinical characterization, the pathophysiology of AMS remains unclear and this scoring system may not be suitable for early diagnosis, thus leading to the delay of treatment and worsening of disease 4 .Therefore, accurate and prompt biomarkers are urgently required to identify people who are sensitive to hypoxia at high altitude, but none of which are currently available.
Metabolomics platform is a newly developed tool for system biology, which can identify and quantify all small molecular metabolites within a biological system 5 .Metabolic pro ling is used to de ne metabolic changes in response to genetic differences, environmental in uences and disease or drug perturbations at global scale 6 .Equipped with advanced high throughput platforms, bio statistical analysis software and bioinformatics tool, metabolomics has become an irreplaceable approach to obtain and compare the full complement of metabolites in tissue samples between groups to discover slight changes to external stimulation 7 .It has developed rapidly and become an effective tool for disease diagnosis, biomarker screening, and characterization of biological pathways.Nuclear magnetic resonance (NMR), liquid chromatography-mass spectrometry (LC-MS) and gas chromatography-mass spectrometry (GC-MS) are currently prevailing techniques used to explore metabolic changes in urine or plasma 7 .Single analytical approach of NMR or LC-MS may not be able to provide full coverage of even the simplest metabolites with diverse concentration and mass range 8 .GC-MS, on the other hand, has been increasingly applied in laboratories because of its easier metabolites identi cation approach based on by comparing their mass spectrum and retention index with authentic reference standards or commercial libraries 9 .So far, studies focusing on using "omics" have provided large amount of background information about high-altitude illness 10 , among which metabolomics is a newly developed one.Johnson CH came up with the idea that proteomic variation can cause metabolomics variation 11 and this furthers the use of metabolomics in the research of altitude illness.
Since some individuals are more susceptible to high altitude diseases than others, the incidence of highaltitude diseases is variable among different population.Given that the pathophysiology of the susceptibility of AMS remains unresolved, the broader view provided by metabolomics may provide novel insights into the etiology of AMS and identify novel biomarkers for the prompt diagnosis of the condition.
Herein, a metabolomics method based on GC-MS and random forest (RF) models were applied to analyze the serum samples of people rapidly ascending to high altitude and to comprehensively investigate the metabolic pro ling changes of AMS.

Subject recruitment
Subjects recruited for analysis were male health volunteers living at sea level.Inclusion criteria were as follows: absence of acute infection during the past 6 months, absence of chronic in ammatory diseases, with no organic diseases, with the age >18 years, and >6 months stay at sea level.Volunteers were excluded if they had any health problems, abnormal complete blood count, chemistry panel or liver function results or those who had been to high altitude during the past 6 months.All subjects' general health status eligibility were assessed and determined prior to participation including medical history, physical check, and standard blood and urine analysis.They all ascended by train to Golmud district (altitude 4300m) in Qinghai, China.All subjects signed informed consent to all the risks involved in this study and they did not receive any preventative drugs before or during ascending to high altitude.Our study has been approved by the Ethics Committee of the Chinese PLA General Hospital.

AMS assessment and physiological examination
To evaluate AMS status, subjects recruited for this clinical study completed the self-reported questionnaire on the 3 rd day arriving at high altitude.The diagnosis of AMS was based on the Lake Louise Score (LLS) system 3,12 .Five symptoms including headache, gastrointestinal symptoms, fatigue/weakness, dizziness/light-headedness, and di culty in sleeping were assessed.Participants scored the severity of each symptom from 0 (no symptom) to 3 (maximal severity).Individuals whose total score exceeded 3 points and were accompanied with headache were diagnosed as AMS.And subjects whose LLS scores were lower than 3 points were selected as control group.Blood pressure, heart rate and oxygen saturation were also measured both at sea level and on the 3 rd morning after ascending to high altitude.Arm blood pressure (BP) monitor (KD-5903; Andon Health Co., Ltd.Tianjin) was used to determine brachial artery BP of each volunteer for three times and average value was acquired.SpO2 of ngertip and pulse rate were measured with pulse oximeter (YX301; Yuwell Medical Equipment & Supply Co., Ltd., Jiangsu), both of which were performed on right index ngers in the morning before breakfast.

Blood sample collection, conservation and preparation
Venous blood (4ml) was collected in EDTA-K 2 tubes from each overnight fasting individual in the morning at sea level and on the third day after ascending to Golmud.Plasma was separated through centrifugation at 3000rpm for 5 minutes, stored at -20°C and then transported to Beijing and stored at -80°C until further analysis.
The preprocessing procedures before GC-MS analysis included deproteinization, drying and chemically derivatization.Plasma was thawed at 4°C for 20 min and centrifuged for 15 min (15,800 rpm, 4°C).To the 100 μl supernatant, 400 μl methanol was added for deproteinization and centrifuged for another 15 min.Then ribitol was added as internal standards at the concentration of 1 mg/ml and mixed on a vortex for 15s.The above samples were put on ice in ventilated cabinet.After methanol was volatilized, they were put at -70°C for over 4h until dried.One microliter of each supernatant was then transferred to the sample for the following GC-MS procures.The remaining was combined to generate a pooled quality control (QC) sample.Then dried samples were derivatized using freshly prepared methoxylamine/pyridine (50 mg/ml) and incubated for 90min (30°C).Next, 80 μl of N-methyl-N-(trimethylsilyl) tri uoroacetamide (MSTFA) with 1% trimethylchlorosilane (TMCS) was added to the samples which were then incubated for 2h at 37°C.

Gas chromatography-mass spectrometry conditions
Samples were analyzed on the Agilent 7890A gas chromatograph with a 5975C time-of-ight mass spectrometer (Agilent Technologies, America) that equipped with an auto-sampler 7693 and electron ionization (EI) source.A 30m×0.25mm×0.25 μm deactivated fused silica capillary column (Agilent Technologies) was used for the GC-MS analysis.The primary temperature was maintained at 60°C for 4 min, programmed to 180°C at a rate of 10°C/ min, and slowed down to the rate of 3°C/min to 260°C, then rise the rate again at 10°C/min until 300°C and held for 10 min.The temperatures of the front injection port, interface and ion source were set at 270°C, 280°C and 230°C, respectively.Helium was used as the carrier gas with a ow rate of 1.0 ml/min.One microliter of sample was injected in the 10:1 split mode.
The MS quadrupole temperature was 150°C.The mass spectrometer was operated under electron impact (EI) mode at ionization energy of 70 eV, with a scanning extent of 85-500m/z and was set for ve scans per second.

Quality control
A quality control (QC) sample theoretically identical to the biological samples, with a metabolic and sample matrix composition similar to those of the biological samples under study, is used in our study 13 .A pooled QC strategy deriving from all the subject population was applied.Equal quantities were taken from every sample and mixed together as the QC sample.The serum samples and the QC samples were labeled.Before the experimental samples were injected, QC samples were used to equilibrate the GC-MS.All the samples were then analyzed in GC-MS at random in order to avoid the run order effect.

Identi cation and quanti cation of metabolites
Data from GC-MS was processed by GC/MSD Chem Station Software (Agilent) for auto-acquisition of GC total ion chromatograms (TICs) and fragmentation patterns.Then all transformed data was processed sequentially with the assistance of Qualitative Analysis B.04.Since there was a serious of split molecular ions for each compound fragmentation pattern, the National Institute of Standards and Technology (NIST) mass spectra library by the Chem Station Software was used to make the comparison.The standard mass chromatogram from NIST, which includes the mass charge ratios and the abundance of compound fragmentation, can search the following parameters including the compound's name, constitutional formula, retention time, relative molecular weight ect.For each peak, the software generated a list of similarities with those within the NIST library.Peaks with matching factor over 85% were assigned compound names, those having < 85% similarity were listed as unknown metabolites.Internal standard was used to obtain the relative quanti cation of these compounds.The results were exported in .cefles, which included the above parameters.

Multivariate data processing
Feature extraction and pre-procession of the obtained data were performed using R software, and then normalized and edited into two-dimensional data matrix by Excel 2010 software.Parameters including Retention time (RT), Mass-to-charge ratio (MZ), Observations (samples) and peak intensity were collected.Multiple variables statistical procedures of principal component analysis (PCA) and orthogonal partial least squares-discriminant analysis (OPLS-DA) were applied to observe the plasma pro les from different groups using SIMCA-P 13.0 software (Umetrics AB, Umea, Sweden).The quality of the models was evaluated by the relevant R 2 X (explainable index) and Q 2 (predictable index).PCA models were regarded as valid only if R 2 (X) > 0.4.PLS-DA models were regarded as valid only if Q 2 > 0.4.All of the data from the differentially expressed compounds were used in calculating PCA models (an unsupervised statistical procedure that uses orthogonal transformation to convert a set of observations of possibly correlated variables into a set of values of linearly uncorrelated variables called principal components).
Signi cantly characteristic differential metabolites or metabolic features between sea level and high altitude groups, AMS and non-AMS groups were screened using the OPLS-DA model.The Paired t-test was selected to measure the signi cance of each metabolite in separating sea level from high altitude group and AMS from non-AMS groups.Fold change is calculated through the Logarithmic value (2 as the base value) of ratio between AMS group and non-AMS group.A positive value indicates a relatively higher concentration present in AMS group, whereas a negative value indicates a relatively higher concentration in non-AMS group.The differences were considered signi cant when p <0.05.
Commercial databases, including the Kyoto Encyclopedia of Genes and Genomes (KEGG; http://www.genome.jp/kegg/)databases were used to search for metabolite pathways.More detailed analysis of the relevant pathways and networks of AMS is performed by MetaboAnalyst 3.0, which revealed that these differential metabolites are important for the organism response to high altitude or AMS and are responsible for multiple pathways.MetaboAnalyst is a free, web-based tool that combines results from a powerful pathway enrichment analysis concerning the conditions under study 14 .

General characteristics of enrolled subjects
In our current study, 36 volunteers met the inclusion and exclusion criteria and thus were included.Based on the Lake Louise Score System, 17 of the included subjects were divided into AMS group while the rest 19 were de ned as non-AMS.The clinical characteristics of these subjects were illustrated in Table 1.On the third day after arriving at high altitude, the physiological responses could be re ected through signi cantly increased heart rate (66.3 ± 5.6 at sea level, 81.3 ± 11.2 for AMS group and 82.3 ± 8.2 for non-AMS group) and blood pressure (75.3 ± 5.3 mmHg at sea level, 85.4 ± 8.9 mmHg for AMS group and 82.4 ± 7.2 mmHg for non-AMS group) and signi cantly decreased oxygen saturation (98.7 ± 1.8% at sea level, 86.9 ± 6.7% for AMS group and 89.3 ± 6.5% for non-AMS group).None of these subjects progressed to the more severely fatal forms of high altitude pulmonary edema or high altitude cerebral edema.Multivariate statistical analysis of metabolites 1892 features were collected in this experiment.To determine whether the metabolite ngerprints in plasma differed between non-AMS and AMS subjects, subjects at sea level and high altitude, we rst evaluated separation using unsupervised PCA and supervised OPLS-DA.The obvious separation was achieved between sea level group and high altitude group (Fig. 1A).For the two groups, there were 17 principal components.The PCA model score plots were characterized by the following parameters: R 2 X = 0.894 and Q 2 = 0.682.Based on the OPLS-DA model (Fig. 1B), sea level group and high altitude group were discriminated with an R 2 X = 0.508,R 2 Y = 0.78 Q 2 = 0.496.The OPLS-DA model is regarded as valid due to Q 2 > 0.4.

Comparison of metabolites between AMS group and non-AMS group
The variable importance in the projection (VIP) parameter re ecting the importance of variables was applied to lter the important metabolites in the model.Variables with VIP values higher than 1 were selected in this study and variables without support of its con dence interval were rejected.The qualitative analysis of differential metabolites is based on the retention time (RT), m/z and the matched metabolites were researched in the NIST library.18 variables were identi ed by comparison between the sea level group and high altitude group (Table 2), among which alanine and cyclohexasiloxane signi cantly increased in high altitude group, while other 16 metabolites decreased compared with sea level group.In addition, 6 variables were identi ed between AMS group and non-AMS group, among which uric acid decreased in AMS group, while Propanoic acid, silane, talose, D-glucose and 1-Deoxyglucose increased compared with non-AMS group (Table 3).

Metabolic pathway analysis for differentially expressed metabolites
After searching the KEGG databases from the differential metabolites, we found 8 metabolomic pathways between AMS and non-AMS group and 27 pathways between sea level and high altitude group.
The 18 differential metabolites between sea level group and high altitude status were found to be primarily involved in the aminoacyl-tRNA biosynthesis metabolism, cysteine and methionine metabolism, taurine and hypotaurine metabolism, thiamine metabolism, alanine, aspartate and glutamate metabolism, arginine and proline metabolism, pantothenate and CoA biosynthesis, nitrogen metabolism, galactose metabolism, nicotinate and nicotinamide metabolism, glycine, serine and threonine metabolism (Table 4).The 6 metabolites between AMS group and non-AMS group were found to be primarily involved in the glycolysis or gluconeogenesis metabolism, pentose phosphate pathway, propanoate metabolism, galactose metabolism, nicotinate and nicotinamide metabolism, starch and sucrose metabolism (Table 5).Consequently, potential target metabolic pathway analysis with MetaboAnalyst revealed that metabolites, important for the host response to high altitude are the metabolism of aminoacyl-tRNA biosynthesis, cysteine and methionine, taurine and hypotaurine, thiamine, alanine, aspartate and glutamate, arginine and proline, pantothenate and CoA biosynthesis, nitrogen, galactose, nicotinate and nicotinamide, glycine, serine and threonine (Fig. 2A).Metabolism of glycolysis or gluconeogenesis, pentose phosphate, propanoate, galactose, nicotinate and nicotinamide, starch and sucrose were found to be distributed in the AMS group (Fig. 2B).Detailed related metabolic pathways were constructed using the reference map by searching KEGG (Fig. 3A and 3B).

The levels of key metabolites involved in the pathways of altered metabolites
To validate the levels of

Discussion
In our current study, we performed an integrated analysis of metabolomics pro ling with the whole blood of individuals that exposed to high altitude hypoxia.Traveling to elevations above 3000 m is associated with the risk of developing AMS, HACE and HAPE 15 .Among these complications, AMS is a severe and common disease that occurs after a rapid ascent to high altitude.Identi cation of molecular markers has the potential to improve the understanding of pathophysiological mechanisms and the diagnosis and prognosis of a condition, as well as to identify the most e cacious therapy 16 .To date, clinical screening for AMS mainly relies on patient interviews, physician's examination and oxygen saturation tests.Speci c tests or biomarkers, which would be more reliable and accessible for early diagnosis, are under urgent requirement.The large amount of data generated by omics research may be effective for the diagnosis of high altitude illnesses and may be used to improve our understanding of the pathogenesis of illnesses 10 .
Among these omics methods, metabolomics studies have enormous potential to identify potential biomarkers.Gas chromatograph-mass spectrometry (GC-MS), which harvests the information from various spectra of metabolites and facilitates rapid metabolite identi cation and quanti cation, proved to be a powerful platform for identifying biomarkers and understanding biochemical pathways to improve diagnosis, prognosis, and treatment of disease 17 .Thus, in our current study, we performed GC-MS to investigate metabolic biomarkers associated with the pathophysiological response and susceptibility to high altitude hypoxia.Our results showed that the plasma metabolic pro les between the sea level and high altitude status were obviously different, while the separation was not that obvious between AMS group and non-AMS group.This indicates that the metabolic pro le in the plasma of subjects ascending to high altitude and of AMS patients is altered.
Metabolic pathway analysis was conducted to further understand molecular function of these plasma metabolites.The metabolic pathways are discussed in detail below.In the metabolic pro les, glucose, the predominant energy sources in most organisms, decreased in AMS group, indicating a possible upregulation of glycolysis.During high altitude acclimation, plasma lactate and pyruvate were increased more markedly in fed rats compared to fasted rats when rapidly exposed to an 8000 m altitude 18 , which utilizes blood glucose as the substrate preferentially during hypoxia.In addition, glycolysis is correlated with an increased a nity of glucose receptors for deoxyglucose, stimulated by some growth factors and cytokines 19 .
Uric acid, the end product of purine metabolism, is generated by the action of the enzyme XO, which catalyses the last two steps of uric acid conversion: hypoxanthine to xanthine and from xanthine to uric acid 14 .The increased level of uric acid in AMS group is consistent with previous reports 20 .
The levels of a wide variety of amino acids were altered by the pathogenesis of high altitude hypobaric hypoxia.Amino acid metabolism is so complex that a large number of metabolites were involved in the process 21 .Chang Liu et al. also reported that AA metabolism pathway was one of the most pivotal alterations after acute hypoxia exposure and may account for variations in response patterns to hypoxia stimuli 22 .The dysregulation of proteolysis, oxidative catabolism, and gluconeogenesis can lead to the metabolic disorder of amino acids.The decreased level of aspartic acid, a key metabolite in the pathway of alanine, aspartate and glutamate metabolism was observed in high altitude group subjects.In addition, the level of alanine increased at high altitude, tyrosine and proline, two branched-chain amino acids, may be an important alternative energy substrate.It seems that the reduction in ATP production due to the inhibition of citrate cycle induced by the hypobaric hypoxia of high altitude could lead to the utilization of branched-chain amino acids as energy compensation 23 .Branched-chain amino acids have been suggested as a useful supplementation in the treatment of lung disease and in trekking at high altitude 24 .
The signi cantly decreased level of L-cysteine was also observed in high altitude group.Cysteine is known to increase the intracellular stores of glutathione there by enhancing endogenous antioxidant levels 25 .The effects of NAC are mainly generated through cysteine and glutathione.It has bee reported that NAC supplemented rats showed less free radical production in comparison with hypoxic rats 26 .
Luo et al. studied the metabolomic variation in the plasma of HAPE patients using 1H NMR and found that HAPE patients had signi cant several metabolites increased while some decreased in HAPE patients compared with the control group.According to his research, the HAPE patients had signi cant increase in valine, lysine, leucine, isoleucine, glycerol phosphoryl choline, glycine, glutamine, glutamic acid, creatinine, citrate, and methyl histidine and decrease in -and β-glucose, trimethylamine, and the metabolic products of lipids 19 .
In our current research, serum was collected from normal control volunteers before and after rapidly ascending to high altitude.In addition, a GC-MS-based metabolomics platform coupled with the machine learning method random forest models were successfully applied to explore the serum metabolic differences and changes of acute mountain sickness patients.The results showed that random forest models revealed characteristic and advantages on the discrimination between AMS and normal controls.Components that differed in levels between non-AMS group compared to AMS patients may serve as potential biomarkers for scanning acute mountain sickness at an early stage.To summarize, in AMS patients, aerobic metabolism is inhibited, whereas anaerobic glycolysis is increased.
Changes in the levels of a number of related metabolites are obvious, but the speci c regulatory sites and regulatory mechanisms that are involved and the biologic signi cance of such factors require further study.

Conclusion
Despite the limitation of sample size, this study illustrates the useful application of metabolomics analysis, based on GC-MS of blood plasma samples, for the investigation of metabolic changes in subjects rapidly ascending to high altitude and are susceptible to AMS.Moreover, this study suggests that metabolite analysis could provide a new understanding of AMS and may be useful for the pathophysiological mechanism investigation and in the diagnosis of AMS.The green line represents the regression line for R2, and the blue line represents that for Q2.
The pathway impact of high altitude on plasma metabolites with MetaboAnalyst 3.0.A. Altered metabolic pathways between high altitude and sea level groups; a, aminoacyl-tRNA biosynthesis; b, cysteine and methionine metabolism; c, taurine and hypotaurine metabolism; d, thiamine metabolism; e, alanine, aspartate and glutamate metabolism; f, arginine and proline metabolism; g, pantothenate and CoA biosynthesis; h, nitrogen metabolism; i, galactose metabolism; j, nicotinate and nicotinamide metabolism; k, glycine, serine and threonine metabolism.B. Altered metabolic pathways between AMS group and non-AMS group; a, glycolysis or gluconeogenesis; b, pentose phosphate pathway; c, propanoate metabolism; d, galactose metabolism; e, nicotinate and nicotinamide metabolism; f, starch and sucrose metabolism.

Figures
Figures

Figure 2 The
Figure 2

Table 1
General physiological characteristics of enrolled subjects (Data are presented as mean ± SD)

Table 2
Signi cantly different metabolites between sea level and high altitude status