3.1 Cortical biochemical indexes from the HSD and CON groups
To validate the metabolic alteration effects of HSD on the brain tissues, biochemical parameters, including the energy metabolism-related metabolic enzymatic activities of Na-KATP, AChE, and LDH, the neurotransmitters of DA, E, NE, 5HT, and GABA, and the oxidative stress-related proteins of SOD, MDA, and Gpx in the ipsilateral cortex tissues of HSD and CON rats were also determined (Table 1). Compared with the control group, the levels of DA, E, NE, and MDA were increased significantly, whereas the contents of Na-KATP, AChE, LDH, 5HT, GABA, SOD, and Gpx were decreased in the HSD model.
Table 1
The quantification of biochemical parameters measured in cortex tissue CON and HSD animals by ELISA with the assay kits. (mean ± S.D.). * indicates statistical significance.
Groups
|
Metabolic enzymes
|
Na-KATP
(µmol/mg protein)
|
AChE
(U/g protein)
|
LDH4
(U/g protein)
|
CON
|
1.70 ± 0.224
|
539.00 ± 33.542
|
9250.43 + 425.394
|
HSD
|
116 ± 0.13*
|
235.86 ± 15.00*
|
6544.29 ± 873.84*
|
Groups
|
Neurotransmitters
|
dopamine (DA, nmo/g protein)
|
Epinephrine
(E, nmo/g protein)
|
Norepinephrine
(NE, nmol/g protein)
|
5-hydroxytryptamine
(5HT, nmol/g p rotein)
|
GABA
(nmol/g protein)
|
CON
|
65.64 ± 4.82
|
8.52 ± 0.58
|
7.64 ± 1.262
|
2.81 ± 1.00
|
6.49 ± 0.72
|
HSD
|
96.88 ± 7.74*
|
11.42 ± 119*
|
12.02 ± 1.78*
|
1.00 ± 0.17*
|
3.93 ± 1.08*
|
Groups
|
Oxidative stress indicators
|
SOD
(U/mg protein)
|
MDA
(nmol/mg protein)
|
GPX+
(U/g protein)
|
SOD
(U/mg protein)
|
MDA
(nmol/mg protein)
|
CON
|
153.09 + 37.62
|
2.88 ± 1.44
|
1042.27 ± 248.32
|
153.09 + 37.62
|
2.88 ± 1.44
|
HSD
|
114.97 ± 23.38*
|
6.23 ± 1.96*
|
746.77 + 188.13*
|
114.97 ± 23.38*
|
6.23 ± 1.96*
|
3.2 Metabolites identified in 1H NMR spectra of brain tissue samples
Figure 3 displays three representative 1H NMR spectra from the cortex (Fig. 3A), hippocampus (Fig. 3B), and striatum (Fig. 3C) of an HSD rat. A wide range of prominent metabolites were identified according to the 1H NMR data of brain tissue samples, and they are organic acid anions (lactate (Lac), malonate (Maln), succinate (Suc), fumarate (FMA), 2-hydroxybutyrate (2-HB), acetate (Ace), formate (For), taurine (Tau), ascorbate(Asc), Isobutyrate (IB), 3-hydroxyisobutyrate (3HIB)), amino acids (leucine (Leu), isoleucine (Ile), valine (Val), alanine (Ala), glycine (Gly), tyrosine (Tyr), phenylalanine (Phe), aspartate (Asp), glutamine (Gln), glutamate (Glu), threonine (Thr), serine(Ser), histidine(His)), neurotransmitters (\(\gamma\)-aminobutyrate (GABA)), energy-related metabolites (creatine (Cre), adenosine diphosphate (ADP), adenosine monophosphate (AMP)), phospholipid related metabolites (O-phosphocholine (Pcho), sn-glycero-3-phosphocholine (GPC)), and others (myo-inositol (MI), nicotinamide adenine dinucleotide (NAD+), niacinamide (NA), N-acetylaspartate (NAA), nicotinamide-adenine-dinucleotide phosphate (NADP+), UDP-N-acetylglucosamine, (UDPG), uridine (Uri), uracil (Ura), uridine 5'-monophosphate (UMP), inosine 5'-monophosphate (IMP), inosine (Ino), choline (Cho), carnitine (Car), glutathione (GSH), dimethylglycine (DMG), adenosine(Ade). Table S1 shows detailed information on the metabolite assignments.
3.3 Metabolic profile alterations revealed by metabolomic analysis
PCA, an exploratory and unbiased analysis approach of the 1H NMR spectra from all brain extracts across different brain regions, was first made to reveal the main metabolic trends driven by hyperbaric exposure to a 400 msw heliox saturation environment. A PC1 vs. PC2 scatting plot obtained from PCA (Supplementary Information: Figure S1A, S1B, S1C) of integral bucket data revealed a certain discrimination with some overlap between the two classifications. Supervised investigations of PLS-DA (Figure S1A’, S1B’, S1C’) and OPLS-DA (Fig. 4A, 4B, 4C) models exhibited clear class discriminations of the metabolic profiles between groups. Figure 4 shows the score plots of the OPLS-DA model for the cortex region (A), the hippocampus region (B), and the striatum region (C), showing clear discrimination between the HSD groups and the controls. The high explained variation and the goodness of the prediction reflected by the values of R2X and Q2 (Figure S1A’, S1B’, S1C’, Fig. 4A, 4B, 4C) and permutation test plots (Fig. 4A’, 4B’, 4C’) indicated the robustness of the generated supervised models. The correlation coefficient (r) extracted from the S-line plots, variable importance in projection (VIP), and the p value from the nonparametric univariate tests were collected to assess the significant metabolites responsible for the class-discriminating patterns. Therefore, meeting any one of the three criteria (the absolute value of r greater than 0.50, the value of VIP greater than 1.0, together with p value less than 0.05), we selected a panel of statistically significant metabolites (Table 1) responsible for the separation between the CON and HSD groups. In Table 1, fold-change values greater than 1 indicate an increased level in the HSD group, while fold-change values less than 1 indicate a decreased level in the HSD group. The mean SD values of discriminative metabolites are listed in Table S2. Using the hierarchical cluster analysis of metabolites and the average linkage method, the generated heatmap (Fig. 5A, 5B, 5C) with dendrograms allows for a better visualization of three brain region metabolic alterations caused by hyperbaric exposure in a helium oxygen-saturated environment.
3.4 Metabolic disorders observed in different brain regions of HSD rats
The abovementioned multivariate analysis and univariate analysis using the bucket height of metabolites in the different groups provided a great work tube to identify discriminative metabolites revealing the potential neurologic metabolic alterations associated with HSD events. Elevated AMP, FMA, NA, and Phe and a decrease in Ala, Asn, Car, Cho, Cyt, GABA, GSH, Ino, Lac, Pcho, Phe, Tyr, Ura, and Uri were found in the cortex tissue of the HSD group compared with the CON group (Fig. 5A, Table 1). Meanwhile, in the hippocampus, Ala, GSH, Lac, Uri, Cyt, GABA, Tyr, and Ura also decreased and AMP increased in the HSD group, as they did in the cortex. Moreover, the upregulated Thr and downregulated Gly, ATP, Tau, Imp, Suc, Asc, and DMA were expressed in hippocampus extracts of the HSD group relative to the controls (Fig. 5B, Table 1, Fig. 5D). Compared with the CONS group, the contents of Cyt, GABA, Ura, Cho, and Thr also decreased as they did in the cortex and hippocampus. Additionally, increased levels of branched-chain amino acids (BCAAs, including Leu, Ile, Val), and Lys and decreased levels of Gln, NAA, NAD+, NADP+, Asp, a series of metabolites of ATP, Tau, IMP, and Suc, which also decreased in the hippocampus, and another two metabolites, Ino and Pcho, which also decreased in the cortex, were observed in the striatum tissues of HSD group rats (Fig. 5C, Table 1, Fig. 5D). Such many metabolites always indicate implicated molecular pathways with complexity and diversity. The integral bucket values of discriminant metabolites were quantified, and statistically significant fold-changes of their concentrations between the heliox-saturation-hyperbaric-exposed and control groups are summarized in Table S2 and Table 2 for the three brain regions.
Table 2
Summary of a panel of discriminant metabolites identified to be significantly changed between heliox saturation-hyperbaric-exposed and control rats in the three brain regions of cerebral cortex tissues, hippocampus, and striatum. The cutoff value of |r| is 0.50. The cutoff value of VIP was 1.0. The value of p defining statistically significant differences was less than 0.05. The metabolites meeting one of three criteria can be judged as discriminant metabolites.
metabolites
|
FCa (rb, VIPc, pd) in cortex
|
FC (r, VIP, p) in hippocampus
|
FC (r, VIP, p) in striatum
|
ATP
|
/
|
0.72(-0.72, 1.58, 0.03)
|
0.62(-0.68, 1.38, 0.04)
|
Ala
|
0.93(-0.54, 1.15, 0.13)
|
0.89(-0.53, 1.4, 0.06)
|
/
|
AMP
|
1.31(0.79, 1.59, 0.00)
|
1.1(0.62, 1.34, 0.05)
|
/
|
Asc
|
/
|
0.87(-0.53, 1.13, 0.04)
|
/
|
Asn
|
0.85(-0.55, 1.12, 0.07)
|
/
|
/
|
Asp
|
/
|
/
|
0.93(-0.66, 1.34, 0.06)
|
Car
|
0.82(-0.59, 1.22, 0.02)
|
/
|
/
|
Cho
|
0.64(-0.79, 1.66, 0.01)
|
0.8(-0.62, 1.35, 0.1)
|
0.63(-0.92, 1.86, 0.00)
|
Cyt
|
0.8(-0.59, 1.26, 0.04)
|
0.69(-0.73, 1.56, 0.01)
|
0.61(-0.74, 1.49, 0.00)
|
DMA
|
/
|
0.39(-0.56, 1.25, 0)
|
/
|
FMA
|
1.26(0.51, 1.09, 0.08)
|
/
|
/
|
GABA
|
0.89(-0.54, 1.18, 0.05)
|
0.92(-0.82, 1.74, 0)
|
0.86(-0.53, 1.07, 0.04)
|
Gln
|
/
|
/
|
0.9(-0.5, 1.23, 0.06)
|
Gly
|
/
|
0.89(-0.6, 1.4, 0.02)
|
/
|
GSH
|
0.75(-0.75, 1.52, 0.01)
|
0.82(-0.57, 1.21, 0.05)
|
/
|
Ile
|
/
|
/
|
1.11(0.62, 1.33, 0.03)
|
IMP
|
/
|
0.79(-0.72, 1.56, 0)
|
0.82(-0.55, 1.13, 0.09)
|
Ino
|
0.53(-0.92, 1.86, 0)
|
/
|
0.68(-0.65, 1.3, 0.06)
|
Lac
|
0.88(-0.57, 1.15, 0.02)
|
0.87(-0.65, 1.6, 0.02)
|
/
|
Leu
|
/
|
/
|
1.09(0.64, 1.35, 0.02)
|
Lys
|
/
|
/
|
1.05(0.55, 1.27, 0.11)
|
NAA
|
/
|
/
|
0.87(-0.51, 1.23, 0.09)
|
NAD+
|
/
|
/
|
0.84(-0.58, 1.25, 0.03)
|
NADP+
|
/
|
/
|
0.57(-0.67, 1.37, 0.02)
|
NA
|
1.33(0.67, 1.44, 0)
|
/
|
/
|
Pcho
|
0.88(-0.48, 1.2, 0.07)
|
/
|
0.87(-0.54, 1.1, 0.08)
|
Phe
|
1.2(0.68, 1.48, 0.02)
|
/
|
/
|
Suc
|
/
|
0.78(-0.5, 1.34, 0.02)
|
0.83(-0.56, 1.28, 0.08)
|
Tau
|
/
|
0.92(-0.39, 1.17, 0.21)
|
0.79(-0.72, 1.5, 0.01)
|
Thr
|
/
|
1.09(0.67, 1.57, 0.01)
|
/
|
Tyr
|
0.78(-0.7, 1.43, 0.01)
|
0.9(-0.5, 1.26, 0.18)
|
0.74(-0.53, 1.19, 0.04)
|
Ura
|
0.83(-0.43, 1.07, 0.21)
|
0.37(-0.72, 1.55, 0.01)
|
0.69(-0.52, 1.07, 0.07)
|
Uri
|
0.68(-0.88, 1.8, 0)
|
0.78(-0.58, 1.35, 0.11)
|
/
|
Val
|
/
|
/
|
1.09(0.59, 1.35, 0.04)
|
aFold change (FC) between HSD-exposed rats and controls. b r with a positive value indicates a relatively higher concentration present in HSD-exposed rats, while negative values indicate a relatively lower concentration compared to the normal control. c VIP, Variable importance in projection. d p values indicate statistically significant changes between the two groups using a t test with a nonparametric test. The abbreviations of metabolites are shown in Table S1. The value of correlation value r, VIP and p from Student’s t test labeled in the brackets. The cutoff value of |r| is 0.50. The cutoff value of VIP was 1.0. The value of p defining statistically significant differences was less than 0.05. The metabolites meeting one of three criteria can be judged as discriminant metabolites. |
3.5 Metabolite correlation analysis
The relationship between or among metabolites was so complex in Fig. 5A’, 5B’ and 5C’. For the energy metabolites, the positive correlations for Lac vs Ala in HSDC, AMP vs IMP in HSDH, ATP vs Pcho in HSDS, Suc vs GABA/Gln/Tyr/NAA in HSDS, and the negative correlations for Lac vs Thr and GSH vs Suc in HSDH, Suc vs Val in HSDS were present in the metabolite correlation plots. For the neurotransmitters, the positive correlations for GABA vs Car in HSDC, GABA vs Cho in CONH, and GABA vs Suc/Ino/Gln in HSDS were present in the metabolite correlation plots. The negative correlations for GABA vs FMA in the CONC and GABA vs NAA/Ura/NADP + in the CONS were present in the metabolite correlation plots. For the metabolites related to oxidative stress, the positive correlations for GSH vs Asn in the HSDC and the CONC, Tau vs Lac in the HSDH, Tau vs Suc/Ala in the CONH, the negative correlations for GSH vs Suc in the HSDH, Tau vs Thr in the HSDH were present in the metabolite correlation plots.
3.6 Metabolomics pathway analysis
Quantitative pathway analysis consisting of pathway enrichment analysis and pathway impact from pathway topology revealed highly statistically significant HSD-induced modulations to a series of metabolic pathways. Pathway impact scores, together with false discovery rate (FDR) and p values, are described in Fig. 6. Pathways were considered significantly enriched if p values were lower than 0.05; the profiled metabolites (hits) relative to the total metabolites of the pathway (match status) were higher than 1; and the impact scores (indicating the impact of significantly affected metabolites in the pathway based on network topology measure of relative betweenness centrality) were higher than 0. The pathways in the cortex (Fig. 6A) with the greatest metabolic impact value were phenylalanine, tyrosine, and tryptophan biosynthesis > phenylalanine metabolism > pyrimidine metabolism > alanine, aspartate, and glutamate metabolism. The pathways in the hippocampus (Fig. 6B) with the greatest metabolic impact were phenylalanine, tyrosine, and tryptophan biosynthesis > glutathione metabolism > glycine, serine, and threonine metabolism > purine metabolism > pyrimidine metabolism > alanine, aspartate, and glutamate metabolism > butanoate metabolism. The pathways in the striatum (Fig. 6C) with the greatest metabolic impact were alanine, aspartate, and glutamate metabolism > nicotinate and nicotinamide metabolism > purine metabolism.