miR-125a-5p inhibits glycolysis by targeting hexokinase-II to improve pulmonary arterial hypertension

Differences in metabolic patterns between the normal group and MCT-PH group

The attribution ranges of the signal-dependent fragments of the rat arterial serum samples were 0 ppm-3.0 ppm, 3.0 ppm-5.0 ppm and 5.0 ppm-10.0 ppm (Figure 1A–1C). Nuclear magnetic resonance (NMR) spectra were acquired and the samples were analyzed using Chenomx NMR Suite 8.0 to determine the type and concentration of the metabolites. The sample data was normalized using the Pareto Scaling method, and the corresponding score map and load map were obtained through PCA analysis. In the PLS-DA score map, different colors represent different sample groups. It was found that the metabolic profile of the normal group samples were distinguishable from MCT-PH group. The load map corresponding to the score map showed that the farther away from the center point the metabolite, such as glucose, lactate and alanine, the greater their contribution to the distinction between samples (Figure 1D–1F).

Figure 1. There are differences in metabolic patterns in the pulmonary hypertension model. Preparation of rat serum was seen in previous studies [23], and the serum were preserved in -80°C for further metabolomics analysis. Metabolite attribution spectrum of signal fragment of rat arterial blood serum samples: (A) attribution range (0.0ppm-3.0ppm interval); (B) attribution range (3.0ppm-5.0ppm interval); (C) attribution range (5.0ppm-10.0ppm interval). (Associated metabolite number identification: 1: 2-Hydroxybutyrate; 2: 2-Hydroxyisobutyrate; 3; 2-Hydroxyisovalerate; 4: 2-Hydroxyvalerate; 5: 2-Oxoglutarate; 6: 3-Hydroxybutyrate; 7: Acetate; 8: Acetoacetate; 9: Acetone; Alanine; 11: Arginine; 12: Ascorbate; 13: Aspartate; 14: Betaine; 15: Butyrate; 16: Carnitine; 17: Choline; 18: Citrate; 19: Citrulline; 20: Creatine; 21: Creatinine; 23: Dimethylamine; 24: Ethanol; 25: Formate; 26: Fucose; 27: Fumarate); 28: Glucose; 29: Glutamate; 30: Glutamine; Glycerol; 31: Glycine; 32: Guanidoacetate; 33: Isobutyrate; 34: Isoleucine 35: Lactate; 36: Leucine; 37: Lysine; 38: Malonate; 39: Mannose; 40: Methanol; 41: Methionine; 42: ethylsuccinate; 43: Phenylalanine; 44: Proline; 45: Propylene glycol; 46: Pyruvate; 47: Succinate; 48: Trigonelline; 49: Tryptophan; 50: Tyrosine; 51: sn-Glycero-3- phosphocholine; 52: Urea; 53: Valine; 54: myo-Inositol.) Changes of metabolic patterns in rats with pulmonary hypertension: (D) PCA analysis, statistical analysis was performed on rat arterial serum samples, NMR spectra were acquired, and samples were analyzed by Chenomx NMR Suite 8.0 to determine the type and concentration of metabolites in the samples. The sample data was normalized using the Pareto Scaling method, and the corresponding score map and load map were obtained by using PCA analysis. (E) PLS-DA analysis: In the PLS-DA score map, different colors represent different sample groups. It is found from the figure that the metabolic profile of the NC group samples is distinguishable from the PH group. The load map corresponding to the score map shows that the farther away the metabolites such as Glucose, Lactate, and Alanine are from the center point, the greater the contribution to the distinction between samples. (F) VIP is a sort of variable weight importance to provide the most important variables and how important they are in each group. The larger the VIP value, the greater the contribution they make in the differentiation of the sample. Generally, the variable with a default VIP greater than 1 has a significant difference.

Using NMR metabolomics analysis, it was observed that the increase of glucose uptake in MCT-PH rats, the accumulation of various glycolysis products, such as lactic acid, and the activation of the glycolysis pathway, indicate abnormal vascular cell proliferation and energy metabolism in MCT-PH rats (Figure 2).

Figure 2. Pathway analysis found four significant differential metabolic pathways. Using nuclear magnetic resonance (NMR) metabolomics analysis, the increase of glucose uptake in PH rats, the accumulation of various glycolysis products such as lactic acid, and the activation of glycolytic pathway in PH rats were observed. This data indicated that there was a phenomenon in vascular cell proliferation and energy metabolism in PH rats. (1. Glycolytic pathway, 2. Arginine and proline metabolic pathway, 3. Butyric acid metabolic pathway, 4. Glycine, serine and threonine metabolism.)

mi-RNA is related with increased HK-II expression

According to the results, there was an increase in HK-II expression in MCT-PH (Figure 3A). Clustering analysis was used to detect the miRNAs and samples based on the signal values of each miRNA. In the clustering results of the miRNAs, it was found that miRNAs that are clustered have a similar expression in the samples and may perform similar functions.

Figure 3. (A) Increased expression of HK2 in PH induced by MCT. (B) Overexpress HK2 lentivirus and differential MICRORNA screening. PASMCs were transfected in vitro and screened for negatively correlated miRNA molecules with HK-II expression. The control group and PASMCs overexpressing HK2 were detected by miRNA gene expression profiling. (C) Regulated gene expression of cell proliferation was significantly active. Sequence-based miRNA target gene prediction method was applied. The number of target genes predicted by this differential miRNA using MIRANDA was 2432. Based on the number of genes contained in differentially significant function and their degree of enrichment in the database, a targeted map can be created for significant function according to enrichment order. In the figure, the ordinate is the function of differential gene, and the abscissa is enrichment. Each column on the graph represents a significant differential gene function. The greater the difference in gene function, the higher the ranking. (D) Tumor-associated metabolism was active through expression. Pathway analysis was based on the KEGG database. Fisher's exact test and chi-square test were used for differential genes. Based on the analysis results, significantly up-regulated gene pathway can be used to construct a map of differential gene pathways. The ordinate is the name of differential gene pathway, and the abscissa is negative logarithm of p value (-LgP). The larger the value, the smaller the p value.

In the clustering diagram, abscissa represents the sample name between the groups, while the ordinate represents the differential miRNA. Red indicates that the differentially expressed miRNA has a high expression value in the grouped sample, and green indicates that the differentially expressed miRNA has a low expression value (Figure 3B). Based on the analysis results, each differential gene that was found to be significantly upregulated can be used to construct a profile of significant differential gene function. In the figure, the ordinate is the difference gene function name, and the abscissa is the negative logarithm of p value (-LgP). The larger the value, the higher the functional significance level of the differentially expressed gene. Thus, the results show that gene expression inhibited cell proliferation was significantly active (Figure 3C). GO enrichment analyses were used by DAVID as a comprehensive analysis tool, and 3 types of gene function were classified by biological processes and molecular functions. Pathway analysis was based on the KEGG database. The pathways of target genes were analyzed using Fisher's exact test and chi-square test. A p value of <0.05 was used to screen significant pathways. The results show that tumor-associated metabolism was significantly active (Figure 3D).

miR-125a-5p decreases glycolysis induced by MCT-induced PASMCs by targeting HK-II and inhibiting its proliferation

Due to the increased expression of HK-II in MCT-PH rats, we selected the top 6 miRNAs based on the fold change value of miRNAs whose expression profile chip data was downregulated (miR-411-5p, miR-210-3p, miR-532-3p, miR-125a-5p, miR-322-3p and miR-708-5p). Primer information is shown in Table 1. RT-qPCR was used to detect the expression of the PASMCs. There was no significant difference found between the expression of miR-532-3p and miR-322-3p, while the expression of the other four microRNAs in PASMCs had significantly decreased. miR-125a-5p and miR-210-3p showed the most pronounced decreases (Figure 4A). The bioinformatics analysis targeting algorithms (TargetScan and http://microRNA.org) showed that miR-125a-5p contained candidate binding sites in the 3’-UTR of HK-II mRNA (Figure 4C). Combining these results with gene chip results, we selected miR-125a-5p for further validation.

Figure 4. The expression of each microRNA in primary PASMCs of PH rats induced by MCT was verified by RT-qPCR. (A) RT-qPCR was used to detect the expression of six microRNAs that may act on HK-II in PASMCs of PH: rno-miR-411-5p, rno-miR-210-3p, rno-miR-532-3p, rno-miR-125a-5p, rno-miR-322-3p, rno-miR-708-5p. (B) Dual luciferase assay confirmed that miR-210-3p and miR-125a-5p can inhibit luciferase activity. The 3'-UTRs containing normal HK-II mRNA were cloned into the MCS (Multiple cloning site) in the pmirGLO vector to construct the HK2-125/210-pmirGLO vector. This vector was then co-transfected into 293T cells with these two microRNA mimics to determine whether the microRNA acts on the target gene of plasmid by detecting luciferase activity. (C) The bioinformatics analysis targeting algorithms (TargetScan and http://microRNA.org) showed that miR-125a-5p contained candidate binding sites in the 3’-UTR of HK-II mRNA. (D) The results of luciferase reporter gene assay showed that miR-125a-5p mimic significantly inhibited firefly luciferase activity of wild-type plasmid vector, while firefly luciferase activity of mutant plasmid vector hardly changed. (E) Transfection with miR-125a-5p mimic significantly increased the mRNA expression of miR-125a-5p. (F) The expression of HK-II protein in PASMCs was significantly increased in PH group, and the expression of HK-II in PH+miR-125a-5p mimic group was significantly decreased. The results demonstrated that miR-125a-5p negatively regulates the expression of HK-II in PASMCs. (*P<0.05, compared with control group, #P<0.05, compared with WT-miR-125a-5p mimics group, ¥P<0.05, compared with PH group).

In order to identify whether the miR-125a-5p can bind to the 3'-UTRs of HK-II mRNA, 3'-UTRs containing normal HK-II mRNA were cloned into the pmirGLO multiple cloning site (MCS) to construct a HK-II-125 pmirGLO vector. This vector was then co-transfected into 293T cells with microRNA agomir to determine whether the microRNAs act on the target gene of the plasmid by detecting luciferase activity.

miR-125a-5p was selected to further verify if its regulation of HK-II protein is a direct effect. Luciferase reporter gene assay showed that the miR-125a-5p mimic significantly inhibited the firefly luciferase activity of the wild-type plasmid vector, while the mutant plasmid vector showed little change in its firefly luciferase activity (Figure 4D). The above results indicate that miR-125a-5p can directly bind to the 3'-UTRs of HK-II mRNA, suggesting that miR-125a-5p may directly act on HK-II mRNA to regulate protein expression. In order to verify the effect of miR-125a-5p on HK-II, miR-125a-5p was transfected into primary PASMCs and changes in HK-II protein expression were observed. The results of the RT-qPCR showed that miR-125a-5p expression significantly decreased in MCT-PH group, while miR-125a-5p mimic transfection could increase miR-125a-5p levels (Figure 4E). HK-II protein expression in PASMCs significantly increased in MCT-PH group, and the level of HK-II expression in the miR-125a-5p mimic group significantly decreased, which indicates that miR-125a-5p negatively regulated the expression of HK-II in PASMCs (Figure 4F).

In order to investigate the effect of miR-125a-5p on the proliferation of the PASMCs, miR-125a-5p was transfected into normal and MCT-PH group PASMCs. The results show that the proliferation level of PASMCs in the MCT-PH group was significantly higher than that of the normal group. The proliferation level of PASMCs transfected with the miR-125a-5p mimic was significantly lower than that of the MCT-PH group and MCT-PH + negative control group (Figure 5A and 5B). It was also shown that the transfection of exogenous miR-125a-5p could inhibit PASMC proliferation in MCT-PH.

Figure 5. Effect of miR-125a-5p on PASMCS proliferation. Cell proliferation levels were determined by CCK-8 kit (A) and proliferating cell nuclear antigen (PCNA) (B). Glucose consumption (C) and lactic acid content (D) were measured by assay kits. (*P<0.05, compared with control group, #P<0.05, compared with PH group).

The glucose consumption of the PASMCs in MCT-PH group was significantly higher than that of the PASMCs in normal group (Figure 5C). The levels of lactic acid produced by PASMCs transfected with miR-125a-5p was significantly higher than that produced under normal conditions (Figure 5D). After miR-125a-5p mimic intervention, glucose consumption and lactic acid production of the PASMCs decreased significantly, suggesting that glycolysis in PASMCs increases under MCT conditions and that miR-125a-5p mimic intervention could decrease glycolysis in PASMCs.

miR-125a-5p agomir upregulates HK-II expression of PASMCs in vivo, which decreases mPAP, leading to the reversal of right ventricular hypertrophy and improvement of cardiac function

The mean pulmonary artery pressure (mPAP), RVHI and pulmonary vascular remodeling values measured in MCT-PH group were significantly higher than those of MCT-PH agomir group, (Figure 6A–6D), while the TAPSE, PAAT and PAAT/CL values of MCT-PH group were significantly lower than those of MCT-PH agomir group (p<0.05). Although the RVEDD and RVESD values of MCT-PH group were higher than that of MCT-PH agomir group, the difference was not significant (Figure 6E–6J).

Figure 6. (A) The mPAP data were collected and Rated by Powerlab-ML221 (AD Instruments, New South Wales, Australia). (B) RVHI was assessed by weighing the RV separately from the left ventricle (LV) with the septal wall (SW). (C) Paraffin-embedded sections were prepared and stained with hematoxylin and eosin (H&E). (D–J)Pulmonary arterioles between 50 and 200 μm in external diameter (ED) were chosen for morphological analysis. Wall thickness (WT) and ED of pulmonary arteries were measured using IPP 6.0 image analysis software (Media Cybernetics). The remodeling of pulmonary arterioles was calculated as WT% = 2 x WT/ED x100%. The probe of color Doppler ultrasound system GE vivid-E9 was placed next to the sternum and measured the pulmonary artery acceleration time (PAAT), right ventricular end thickness (RVWT), right ventricular end systolic diameter (RVESD), right ventricular end diastolic diameter (RVEED), tricuspid annular systolic excursion (TAPSE). The data measured by 3 cardiac cycles is generally taken and averaged. (*P<0.05, compared with control group, #P<0.05, compared with PH group)

The expression of miR-125a-5p in the PASMCs decreased significantly in MCT-PH group (Figure 7A), while miR-125a-5p expression in MCT-PH agomirgroup was higher than that of MCT-PH group (Figure 7B). The HK-II expression of the PASMCs of MCT-PH group significantly increased compared with that of MCT-PH agomir group, demonstrating that miR-125a-5p negatively regulates the expression of HK-II in PASMCs. These results are consistent with cell levels and results of the in vivo experiments.

Figure 7. Detection of pulmonary vascular miR-125a-5p and HK-II expression. (A) RT-PCR assay showed that the expression of miR-125a-5p in PASMCs was significantly lower in PH group. The level of miR-125a-5p expression in PH+ agomir group was higher than that in PH group. (B) WB results showed that the expression of HK-II was significantly increased in the PH group, thus demonstrating that miR-125a-5p negatively regulates the expression of HK-II in PASMCs. (*P<0.05, compared with control group, #P<0.05, compared with PH group).

Establishment of a PH rat animal model

All animal experimental procedures were performed in accordance with relevant guidelines and regulations and approved by the Institutional Animal Care and Use Committee of Fujian Medical University

A PH model was successfully established through intraperitoneal injection of monocrotaline (MCT) (MCT; Sigma-Aldrich, St. Louis, MO) into rats during the early stages of the research study [7]. Eighteen SD rats (Slac Laboratory Animals Co. Ltd) were randomly divided into 3 groups: normal control (Ctr), monocrotaline-induced pulmonary arterial hypertension (MCT-PH)+NC and MCT-PH+ agomiR-125a. Rats in the MCT-PH+NC and MCT-PH+agomiR-125a groups were intraperitoneally administered 40 mg/kg MCT, while those in the Ctr group were injected with an equal volume of saline. Two weeks later, miR-125a agomir (50nM; Huzhou Hippo Biotechnology Co., Ltd, China) or normal saline (0.9% NaCl) were injected every 4 days to rats and subjected to MCT until 5 weeks.

Nuclear magnetic resonance (NMR) spectrum processing and pattern recognition analysis

The serum metabolites were measured in each group 5 weeks after the model was established. Serum samples were collected from the abdominal aorta of the rats, and NMR spectra were mechanically obtained. Each serum acquisition parameter and its specific setting were set as follows: Temperature (K): 298.15, Magnet Frequency (MHz): 599.83, Transients/Scans: 64, Frequency Domain Size: 65536, Spectral Width: 7225.34, Time Domain Size: 28902 and Pulse Sequence: metnoesy.

The absolute concentration of the metabolites was obtained through formula conversion using known DSS internal standard concentrations and its peak area was combined with peak areas of different metabolites on the NMR spectrum. The variable matrix was used as source data for subsequent PCA and PLS-DA analyses. First, principal component analysis (PCA) was conducted to distinguish between different groups of metabolic patterns, followed by partial least squares discriminant analysis (PLS-DA). PCA and PLS-DA were performed by conducting a data analysis using the PCA methods Bioconductor package and PLS package of R language, respectively. Finally a visual map was constructed using the ggplot2 package.

Western blotting analysis

Whole cell lysates from the cells and pulmonary arterioles were prepared in a lysis buffer (Solarbio, China) with protease/phosphatase inhibitor cocktail (Cell Signaling Technology, USA). Then, the concentration of the extracted protein was determined using a BCA protein assay kit (Solarbio, China). After equal amounts of total proteins (20 ug) were separated using SDS-PAGE and then transferred onto a nitrocellulose membrane, the membrane was incubated with primary antibodies against β-actin (8H10D10, CST, USA), HK-II (2106, CST, USA) and proliferating cell nuclear antigen (PCNA) (13110, CST, USA), at a dilution of 1:2500, followed by horseradish peroxidase-conjugated goat anti-rabbit or anti-mouse secondary antibodies (Sigma, USA), at a dilution of 1:2500. Finally, the protein bands were visualized using an enhanced chemiluminescence substrate (Western Lightning VR Plus-ECL), and the relative abundances of the target proteins were measured using ImageQuant TL software. β-actin (CST, 1:2000) served as the internal reference in this study.

Isolation and culture of primary cells of PASMCs

Pulmonary arteries were rapidly obtained onto an ultra-clean platform. The pulmonary arteries obtained were placed in a digestive solution containing 5 mg/ml type I collagenase, 1 mg/ml papain and 1 mg/ml BSA and were digested at 37°C for 20 minutes. Each digested pulmonary artery was removed from the digestive solution using a nylon mesh and rinsed with D-Hanks solution 2-3 times. Then, the digested pulmonary artery was transferred into a small beaker with 1 ml of D-Hanks solution and pipetted slowly to disperse the cells. The cell suspension was transferred into a culture flask and was thereafter added into a DMEM low-sugar medium containing 15% fetal bovine serum and cultured at 37°C in a 5% CO₂ incubator, and cell fusion reached 70-80% after about one week. The PASMCs were isolated from the control and the MCT-PH rats and primary cultured PASMCs were used in all experiments.

Construction of the HK-II overexpressing lentivirus and gene chip detection

The PASMCs were transfected in vitro and then used to screen for miRNA molecules that were negatively correlated with HK-II expression. The control group and PASMCs overexpressing HK-II were detected using miRNA gene expression profiling.

The TRIzol method was used to extract and purify total RNA from the PASMC samples and was tested using NanoDrop 2000 and Agilent Bioanalyzer 2100. The extracted total RNA was obtained as double-stranded cDNA from the RNA, and a FlashTag Biotin HSR RNA Labeling Kit reagent was used to connect fluorescent dyes, Cy5 and Cy3, to dUTP. Fragmented cRNA was eluted and stained on a GeneChip® Fluidics Station 450, in a GeneChip® Hybridization Oven 645. A GeneChip® Scanner 3000 7G was used to scan the hybridized chips and the data was extracted to be used in calculations.

microRNA target gene prediction

Differentially expressed microRNAs were screened using MIRANDA to predict genes regulated by these differentially expressed microRNAs, and DAVID (https://david.ncifcrf.gov/) was used to perform Gene Ontology analysis of molecular functions and biological processes on the proteins transcribed by the genes predicted.

Real-time quantitative polymerase (RT-qPCR)

The primers were designed based on the predicted microRNA maturation sequences. Primer information was used to verify the expression of the microRNAs in the primary PASMCs of MCT-PH rats. Normal control cells and MCT were transfected with microRNA agomir and transfected with a mimic negative control (NC). MicroRNA was extracted using QIAGEN's miRNeasy® Mini Kit, while RNA concentration and purity were detected using a ultra-micro UV-visible spectrophotometer.

Total RNA was extracted from the PASMCs using TRIzol reagent (Invitrogen, USA), as instructed by the manufacturer and treated using DNase I. Then 2 ug of total RNA was reverse-transcribed using random primers and a Reverse Transcription Kit (Promega, USA). The first-strand cDNA samples obtained were then torrent-amplified using SYBRVR Green I on a Rotor-Gene 3000 A (Corbett Robotics, Australia). All PCR cycling conditions were modified to 94°C for 2 min, followed by 40 cycles of 95°C for 15s, 60°C for 15s, and 72°C for 30s. The relative changes in mRNA expression were calculated using the 2-ΔΔCt method, which was normalized using β-actin. The experiment was repeated thrice on each group. The primers were designed based on mRNA sequences and were synthesized by Shanghai Generay Biotech Co., Ltd. (Table 1).

Verification of the relationship between microRNAs and HK-II gene targeting

The predicted wild-type (wt) miR-125a binding sequences of the 3’-UTR HK-II were synthesized and cloned downstream of the luciferase gene in the pmirGLO luciferase vector (Promega) to generate HK-II-wt. Their corresponding mutated sequences (mut) were created using a GeneTailor Site-Directed Mutagenesis System (Invitrogen) and were cloned into the same vector and named as HK-II-mut. Prior to transfection, 293T cells were seeded into 24-well plates (1 × 103 cells/well) and cultured for 24 h. Then, the 293T cells were transiently cotransfected with 100 ng of wt or mut of HK-II 3’-UTR and 10 nM of miR-125a or miR-NC, together with 20 ng of Renilla luciferase vector (Promega) using Lipofectamine 2000. The cells were collected 48 h post-transfection and their luciferase activity was measured through a luciferase reporter assay conducted using a Dual-Luciferase reporter assay system (Promega). Firefly luciferase activity was normalized to Renilla luciferase activity.

microRNA cell transfection

The microRNA agomir and mimics agomir NC were diluted with 250 μL of Opti-MEM, and 10 μL of the transfection reagent was diluted with 250 μL of Opti-MEM, then the two diluted solutions were mixed and left to stand for 15 min. The medium in the 35 mm culture dish was discarded and 1000 μL of fresh medium was added to the above mentioned 500 μL mixture. After transfection for 24 hours, the culture solution was discarded, washed once with PBS and replaced with 1500 μL of fresh culture solution.

miR-125a agomir design: NC control: sense, 5'-UUCUCCGAACGUGUCACGUTT- 3', antisense, 5'-ACGUGACACGUUCGGAGAATT-3'; miR-125a agomir: sense, 5'- UCCCUGAGACCCUUUAACCUGUGA-3', antisense, 5'-ACAGGUUAAAGGGUCUCAGGGAUU-3'.

CCK-8 method for detecting cell proliferation

CCK-8 solution (10 μL) was added into the 96-wells plates, then the plates were incubated for 0.5 to 4 hours and OD at 450 nm was measured using a microplate reader. [(MCT OD value - blank OD value) - (Control OD value - blank OD value)] / Control OD value.

Detection of miRNA-125a-5p expression

The expression level of miRNA-125a-5p was measured using an Applied Biosystems miR-125 family detection kit (TaqMan MicroRNA Assays).

Detection of glucose consumption and lactic acid content

The ketone-sulfuric acid colorimetric method and lactic acid detection kit (Solarbio) were used to measure glucose consumption and lactic acid production, respectively, as a reflection of glycolysis.

Measurement of mPAP and right ventricular hypertrophy index (RVHI)

After treatment, the rats were anesthetized using 10% chloral hydrate (400 mg/kg). In brief, polyethylene micro-catheters (Chinese Peking Union Medical Physiology) were inserted into the pulmonary artery via the right external jugular vein and connected to a transducer. The mPAP data were collected and analyzed using Powerlab-ML221 (AD Instruments, New South Wales, Australia). RVHI was assessed by weighing the RV separately from the left ventricle (LV) with the septal wall (SW).

Pulmonary vascular remodeling analysis

Paraffin-embedded sections were prepared and stained with hematoxylin and eosin (H&E). Pulmonary arterioles with an external diameter (ED) of 50-200 μm were chosen for the morphological analysis. Wall thickness (WT) and ED of the pulmonary arteries were measured using IPP 6.0 image analysis software (Media Cybernetics). The remodeling of the pulmonary arterioles were calculated as WT% = 2 x WT/ED x100%.

Cardiac ultrasound detection

Color Doppler Ultrasound Diagnostic (GE gray-E9), with a probe frequency of 12 MHz, which maintains a frame rate of 180-220 s when acquiring images was used. SD rats were found to have a fast heart rate of about 400 beats per minute. Ultrasound measurement parameters were based on standards established by the American Society of Echocardiography [19], along with a setting of Doppler flow spectrum and a M-curve scan speed of > l00 mm/s. The SD rats were anesthetized using pentobarbital injection (120 mg/kg, intraperitoneal injection), placed on an examination bed and the coupling agent was applied to left chest.

During the ultrasonography, the II lead electrocardiogram was recorded synchronously and heart rate was measured. The apex of the R wave was the diastole end, and the end of the T wave was the systole end. Pulmonary artery acceleration time (PAAT) was measured by placing probe next to sternum. A M-mode ultrasound was used to measure right ventricular end thickness (RVWT), right ventricular end systolic diameter (RVESD), right ventricular end diastolic diameter (RVEED) and tricuspid annular systolic excursion (TAPSE) in the apical four-chamber view. The amplitude was measured from the end-diastolic phase to the end-systolic phase. In general, data was measured in 3 cardiac cycles and averaged.

Statistical analysis

PCA and PLS-DA statistical analyses were performed on NMR data using Chenomx NMR Suite 8.0 software. The quality of the PLS-DA model was evaluated using a permutation test. The SPSS 23.0 statistical software package was used to analyze specific metabolite concentration data. All measured data are presented as mean ±standard deviation. Statistical analysis for multiple group comparisons were performed by one-way analysis of variance (ANOVA), followed by post-hoc Dunnett’s test. A p value of less than 0.05 was considered to be significant. All experiments described above were performed at least three times.