Plasma metabonomic analysis reveals the effects of salvianic acid on alleviating acute alcoholic liver damage

Abstract

Acute alcoholic liver damage is a common illness and poses a potential health risk for humans presently. Salvianic acid (SA) has been found to be effective in liver protection. However, the mammalian systems responses to acute alcohol exposure and the underlying biochemical mechanism of SA treatment are not clear. In this study, we systematically analysed the acute alcohol-induced metabonomic changes and the therapeutic effect of SA by using a ¹H NMR-based metabonomics approach together with histopathological and clinical biochemistry assessments. The rats in the treatment and model groups were gavaged with 5 g kg⁻¹ BW edible alcohol once every 12 h three times to establish the acute alcoholic liver damage model. SA-treated rats were gavaged with 20 mg kg⁻¹ SA for five days before alcohol administration. The model rats presented acute alcoholic injury with centrilobular inflammation and necrosis. SA treatment not only alleviated the hepatic damage but also promoted the recovery of liver function. We found that acute alcohol exposure induced significant elevation of lactate, glycerol, acetate, creatine and ketone bodies but reduction of glycine and TMAO/betaine. SA reversed the metabolic changes in multiple metabolic pathways, including anaerobic glycolysis, fatty acid oxidation, lipolysis, oxidative stress, creatinine and methylation metabolism. These findings provide an overview of the biochemical consequences of acute alcohol intake and new insights into the SA effects on acute alcoholic liver injury, demonstrating metabonomics as a powerful approach for examining the molecular mechanisms of Traditional Chinese Medicine.

---

Introduction

Wine holds a unique position in traditional Chinese culture. Alcoholic beverages, as a dietary component, have been an important part of people's lives. However, excessive drinking of alcohol causes serious health problems¹ ranging from fatty liver or liver inflammation to focal necrosis and liver sclerosis, which ultimately develop into cirrhosis and even liver cancer.^2,3 Furthermore, alcoholism is also a leading cause of morbidity and mortality worldwide.⁴ In the past nearly 20 years, a striking increase in alcohol consumption and related problems has occurred in China, and even in the forth coming years, there is an evidence of growing tendency, with the potential for a major impact on human health.^5,6

It is well known that alcoholism is often accompanied with acute alcoholic hepatic injury, which can induce hepatitis and liver steatosis, the early form of alcoholic liver disease. Abenavoli et al. reported that severe alcoholic hepatitis (AH) was an acute form of alcohol induced liver disease that was usually observed in patients who consumed large quantities of alcohol.⁷ A large number of studies have also reported that acute alcoholic liver injury was closely related to an imbalance in the antioxidant system in the body, and continuous oxidative stress led to liver steatosis, hepatitis, hepatic fibrosis, and even liver cirrhosis and liver cancer, which represent the later stages of alcoholic liver diseases (ALD).^8,9 In addition, acute alcoholic liver injury is correlated with hypothermia and hypotension,¹⁰ and it can cause dysfunctions in oesophageal, gastric, and duodenal motility.¹¹ Acute alcoholic liver injury is also a risk factor for developing colorectal adenomas and colorectal cancer.¹² In recent years, most works have focused on acute liver damage, but understanding of the metabolic disorders induced by acute alcohol exposure is lacking.

The optimal pharmacological treatment of acute liver injury is controversial and is one of the major challenges for ALD treatment. Currently, the medications commonly used for the treatment of acute alcoholic liver damage in clinic are glucocorticoids, pentoxifylline, metadoxine, and etanercept, and these drugs primarily act as anti-inflammatorys, anti-oxidants and protectors against fibrosis.^13,14 However, they also have some side effects, such as a toxic reaction or tolerance. For the treatment of liver damage, traditional Chinese medicine has advantages, such as fewer side effects and lower toxicity; thus, an increasing number of studies are focused on the treatment effects of traditional Chinese medicine. At present, silymarin is an herbal product that is widely used for its hepatoprotective potential.¹⁵ Moreover, soyasaponins-rich extract from soybean,¹ Gentiana manshurica Kitagawa,⁵ Antrodia camphorata¹⁶ and green tea extract¹⁷ have therapeutic effects on acute alcoholic liver damage. Salvianic acid [2-(3,4-dihydroxyphenyl)-2-hydroxy-propanoic acid, SA], one of the main active components of miltiorrhiza, has been reported to reduce alcohol absorption from the gut,¹⁸ and Wang et al. demonstrated that SA had a protective effect for acute hepatic injury.¹⁹ So far, there has been little research regarding the mechanism of the protective effect of SA on acute liver injury.

Metabonomics is a part of systems biology that reflects the overall response of an organism to external stimuli. Metabonomics employs analytical techniques, such as nuclear magnetic resonance (NMR) spectroscopy, mass spectrometry, and multivariate statistical analysis of spectroscopic data.²⁰ Since the emergence of this approach, it has been widely applied to understand physiological alterations during complex biological processes and disease states.²¹ At present, it has been widely applied to the mechanism study of traditional Chinese medicine, including Chinese herbal compound prescriptions and monomer. For instance, Chao et al. suggested that a ¹H NMR-based metabolomics approach was a useful platform for natural product functional evaluation, and they used that approach to reveal the mechanism of the effect of GA in mice with hepatic steatosis.²² To understand the biochemical alterations that occur throughout the organism during SA ingestion in acute hepatic injury models, we also adopted the metabonomic approach. The present study was undertaken to evaluate whether SA elicits protective action against acute hepatic injury induced by edible alcohol and to elucidate the metabolic regulatory mechanisms of SA using ¹H NMR spectroscopy and multivariate data analysis.

Materials and methods

Chemicals

The purity of SA was 98%, and it was purchased from Nanjing Ze Lang Pharmaceutical Technology Co., LTD (Nanjing, China). The 98% SA was configured to an aqueous solution (2.5 mg ml⁻¹), and 95% edible alcohol was purchased from Guangzhou Wenrui Scientific Instruments Co., LTD (Guangzhou, China). The 95% edible alcohol was diluted with distilled water to a concentration of 50% (w/v). Sodium azide (NaN₃), sodium dihydrogen phosphate (NaH₂PO4), disodium hydrogen phosphate (Na₂HPO₄) and heparin ((C₂₆H₄₁NO₃₄S₄)_n) are domestic analytical reagents. Heavy water (D₂O) containing TSP was purchased from Qingdao Teng Long Technology Co., LTD (Qingdao, China).

Animal experiments and sample collection

Sixty male Sprague-Dawley (SD) rats (150 ± 10 g) were obtained from the Laboratory Animal Institute of Guangzhou University of Chinese Medicine. This study was reviewed and approved by the Ethics Committee of Guangdong Pharmaceutical University (no. SPF20120125). The rats were kept in an air-conditioned room, with a 12 h light/dark cycle and an ambient temperature of 25 ± 1 °C. After acclimatisation for 10 days, the rats were randomly assigned to three groups. The control group (n = 20) was given a normal diet. The SA treatment group (n = 20) was given 20 mg kg⁻¹ (ref. 19) BW salvianic acid (SA) by intragastric administration once a day for five days. The rats in the control and model groups received distilled water with an equal volume to the SA solution for five days. One hour after the last SA treatment, the rats in the treatment and model groups were gavaged with 5 g kg⁻¹ (ref. 1 and 5) BW edible alcohol (50% w/v) once every 12 h for three times to establish acute alcoholic liver damage model. At the same time, the control rats were gavaged with distilled water.

For each group, blood samples were obtained for plasma and serum collection by orbital venous plexus at 0 h (pre alcohol administration). Half of the rats in the three groups were killed by cervical dislocation after ether anesthesia at 6 h and 24 h (post the last time of alcohol administration) for the collection of liver tissues for histopathological examination and blood samples for plasma and serum collection. The plasma samples were collected for metabolite determinations, and the serum samples were collected for biochemical assays. The liver tissues were fixed in 10% formalin solution for haematoxylin and eosin (H & E) staining.

Histopathology and clinical biochemistry

At 0 h, 6 h and 24 h, the body weight of each remaining rat was measured. All liver tissues were embedded in paraffin wax, and 4 μm sections were stained with HE for assessment by the Department of Pathology, Guangdong Pharmaceutical University.

Clinical biochemical analyses of the serum samples, including measurements of alanine aminotransferase (ALT), aspartate aminotransferase (AST), serum creatinine (Scr) and blood urea nitrogen (BUN) levels were performed by the Clinical Laboratory of the First Affiliated Hospital of Guangdong Pharmaceutical University using a Beckman DXC 800 automatic analyser (Beckman, Los Angeles, CA, USA). The acquired data were statistically analysed using the SPSS 16.0 software.

Sample preparation for NMR spectroscopy

The plasma samples were unfrozen at room temperature and prepared by mixing 300 μL of the sample with 200 μL of phosphate buffer (0.2 M Na₂HPO₄/NaH₂PO₄, pH 7.4) to minimise chemical shift variations, and 80 μL of D₂O containing 0.05% TSP as internal standard for chemical shif reference (δ 0.00 ppm) was pipetted into a 5 mm NMR tube.

Data acquisition of NMR

¹H NMR spectra of plasma were acquired at 298 K on a Bruker Avance III 500 MHz spectrometer using a one-dimensional Carr–Purcell–Meiboom–Gill (CPMG) pulse sequence (recycle delay −90°–(τ–180°–τ)_n acquisition) with water suppression which can eliminate the peak overlap. A total of 128 transients were collected into 32k data points using a spectral width of 10 kHz, with an additional relaxation delay of 3 s and a mixing time of 100 ms. The exponential function corresponding to a line-broadening factor of 0.3 Hz was applied to all acquired free-induction decays (FIDs) before Fourier transformation. To assist NMR signal assignments, as described previously,^23,24 a series of 2D NMR spectra with water presaturation were acquired for selected samples, including ¹H–¹H correlation spectroscopy (COSY), ¹H–¹H total correlation spectroscopy (TOCSY) and ¹H–¹³C heteronuclear single quantum correlation (HSQC). For COSY and TOCSY experiments, the spectral width was 10.5 ppm, 80 transients for each of 128 increments were acquired into 2k data points for both dimensions. TOCSY 2D NMR spectra used a MLEV-17 spin-lock scheme with a mixing time of 80 ms. For HSQC, 240 scans per increment and 120 increments were acquired into 2k data points corresponding to the spectral widths of 10.0 ppm and 150 ppm for ¹H and ¹³C respectively.

NMR data processing and multivariate data analysis

The phase and baseline of all the spectra were corrected manually, and referenced to the chemical shift of TSP (δ 0.00). Then, the spectra were bucketed and automatically integrated with the software package AMIX. The spectra with the region δ 0.5–9.0 were segmented into integral regions with equal width of 0.004 ppm, and in order to eliminate the influences of water suppression, the region of δ 4.7–5.2 was excluded. The integrals of these buckets were normalised to the total sum of the spectral integrals.

All the normalized integral values were subjected to PCA and OPLS-DA using the software Simca-P⁺ 12.0 (Umetrics, Sweden). Principal components analysis (PCA), as a non-supervised modeling, was firstly carried out to examine inherent variation in the data set. The PCA models were further tested by external test sets. Subsequently, OPLS-DA models were established in order to screen biomarkers for each group by using ¹H NMR data and class information as the X matrix and Y matrix respectively. The scale was par in OPLS-DA. The results were presented with scores and loadings plots. The scores plots showed the group clusters. The loadings plot provided potential biomarkers in correlation coefficient-coded form through using a MATLAB script programmed by Mr Lei Zhang. For the correlation coefficient-coded loadings, the color-coded variables indicate the significant contribution of the metabolite to the class separation. The significance is reduced gradually ranging from the “hot” color (e.g., red) to a “cold” color (e.g., blue). The OPLS-DA model quality is indicated by the values of R² and Q², which represent the quality of fit and predictability of the model respectively.²³ For further ensuring the validity of all the models, the response permutation testing (RPT) was performed. In current study, a cutoff value of |r| > 0.632 (r > 0.632 and r < −0.632) was chosen for correlation coefficient as significant based on the test p value (p < 0.05).

Results

Histopathology and clinical biochemistry

The histopathological examination results are shown in Fig. 1. In the normal group (Fig. 1A), liver slices showed typical hepatic cells with well-preserved cytoplasm and a clear nucleus. However, a large area of edema, centrilobular inflammation and necrosis were observed in the model group at 6 h post alcohol administration (Fig. 1B). As shown in Fig. 1C, a regeneration of hepatocytes and a decrease of both inflamed and necrotic areas were observed in the SA treatment group. This indicated that the severe toxic insult occurred at 6 h post alcohol administration and suggested that SA could protect the liver from acute alcohol-induced hepatic damage.

Fig. 1 Representative histological haematoxylin–eosin staining graphs of the liver centrilobular area at 6 h. A: control group. B: acute alcohol administration model group. C: salvianic acid treatment group. A1 and A2, B1 and B2, and C1 and C2 are the same area at different magnifications.

In addition, acute alcohol-induced liver dysfunction was calibrated by elevated serum levels of ALT and AST. As shown in Table 1, serum ALT and AST dramatically increased at 6 h after alcohol administration compared to the controls, and they remained at higher levels at 24 h. SA pretreatment decreased the levels of ALT and AST compared with the model group. There was no significant difference between the SA-treated and control groups at 24 h. These results indicate that SA pretreatment has an efficient therapeutic effect on liver dysfunction induced by acute alcohol administration. Moreover, statistically significant increased levels of Scr and BUN were found in models compared to the controls at 6 h, and remaining higher levels at 24 h. However, SA treatment reversed the acute alcohol-induced changes with decreased levels of Scr and BUN. The results indicated that SA was also effective in alleviating renal dysfunction induced by acute alcohol administration.

Table 1 Summary of weight, ALT, AST, Scr and BUN levels among the three groups at 6 h and 24 h^a

Group	6 h					24 h
	Weight (g)	ALT (U/L)	AST (U/L)	Scr (μmol L^−1)	BUN (mmol L^−1)	Weight (g)	ALT (U/L)	AST (U/L)	Scr (μmol L^−1)	BUN (mmol L^−1)
a Values are presented as mean ± SD.^*: to compare with controls ^*p < 0.05; ^**p < 0.01; ^△: to compare with models ^△p < 0.05; ^△△p < 0.01.
Control	257.75 ± 8.72	35.3 ± 3.3	102.3 ± 8.0	15.61 ± 1.84	5.53 ± 1.30	254.70 ± 5.43	38.5 ± 8.5	104.5 ± 7.5	15.43 ± 2.15	5.42 ± 1.34
Model	256.15 ± 14.51	80.0 ± 4.2^**	200.0 ± 3.0^**	28.83 ± 3.42^**	14.15 ± 1.25^**	245.17 ± 11.63	65 ± 4.1^*	176.6 ± 13.5^*	20.89 ± 3.75^*	10.69 ± 1.35^*
SA	246.50 ± 17.48	40.0 ± 3.6^△	149.0 ± 10.5^*△	16.02 ± 1.92^△△	5.76 ± 1.65^△△	248.53 ± 12.63	37.0 ± 8.0^△	115.0 ± 4.0^△	15.87 ± 2.05^△	6.01 ± 1.03^△

---

Metabolites identification of ¹H CPMG spectra of plasma

Typical ¹H CPMG NMR spectra of plasma from control, model and SA-treated rats at 6 h are displayed in Fig. 2. The endogenous metabolites involved in the spectra were assigned based on the literatures^25–27 and were further confirmed by 2D NMR spectra. The dominant metabolites presented in the plasma included glycerophocholine (GPC); membrane metabolites, including choline, phosphocholine (PC); ketone bodies (acetone acetoacetate); a series of amino acids; organic acids, such as lactate, creatine, etc.; glycoproteins and glucose (Table S1†). Visual inspection of these spectra revealed that the alcohol-treated rats had markedly higher lactate than the control rats, and SA treatment reduced the level of lactate compared with the model (Fig. 2). To obtain detailed information regarding SA-induced metabolic alterations in the rats fed with alcohol, a multivariate data analysis including PCA and OPLS-DA was performed.

Fig. 2 Representative plasma ¹H NMR spectra. A: control group, B: model group and C: salvianic acid treatment group at 6 h. Keys: (1) lipid; (2) isoleucine; (3) leucine; (4) valine; (5) lactate; (6) alanine; (7) acetate; (8) N-acetyl glycoprotein; (9) O-acetyl glycoprotein; (10) acetone; (11) acetoacetate; (12) pyruvate; (13) glutamine; (14) creatine; (15) choline; (16) PC; (17) GPC; (18) taurine; (19) TMAO; (20) betaine; (21) glycine; (22) glycerol; (23) α-glucose; (24) β-glucose; (25) unsaturated lipid; (26) tyrosine; (27) 1-methylhistidine; (28) phenylalanine; (29) formate.

Multivariate analysis and statistical analysis of plasma spectra data

Fig. 3A shows the scores plot of PCA representing the distribution of all plasma samples at 0 h (●), 6 h (▲) and 24 h (♦) in the control group, which showed no obvious metabolic differences over the entire period. However, the PCA scores plot for the model group (Fig. 3B) showed an obvious classification trajectory from 0 h to 24 h, with a maximum shift at 6 h. This indicates the great metabolic disorders that are induced by acute alcohol exposure occurred at 6 h. For the SA treatment group, it was noted that the spectra data of plasma samples at three time points were clustered more closely, and the samples at 24 h were similar to those at 0 h. At 6 h (Fig. 3D), SA-treated samples were located between the control and model groups, which demonstrated that SA treatment efficiently reduced liver damage. There were no obvious classifications between the control and SA treatment groups at 24 h suggesting a good recovery of liver injury at 24 h (Fig. 3E). The models in Fig. 3 were further tested by splitting the samples into training and external test sets (Fig. S1†). The prediction pattern suggested the models had good predicted performance. The PCA results demonstrated that SA treatment group shared the similar metabolic characteristics with control group, which revealed that SA treatment had an effect on regulating the metabolic variations in model rats.

Fig. 3 Principal component analyses of plasma ¹H NMR spectra data at different time points from three groups. A, B, C: PCA scores plots of control (PC1, 29.1% vs. PC2, 16.1%, R² = 45.2%, Q² = 40.8%), model (PC1, 57.0% vs. PC2, 22.3%, R² = 79.3%, Q² = 57.5%) and SA (PC1, 50.7% vs. PC2, 20.0%, R² = 70.7%, Q² = 56.1%) treatment rats, respectively. D, E: PCA scores plots of the control, model and SA treatment rats at 6 h (PC1, 65.7% vs. PC2 23.1%, R² = 88.8%, Q² = 50%) and 24 h (PC1, 60.8% vs. PC2, 24.9%, R² = 85.7%, Q² = 56.1%) respectively.

After the crude screening by PCA, OPLS-DA models were constructed to enlarge the separation among the groups. The model quality is indicated by the values of R²X and Q² (Tables S2 and S3†), which represent the quality of fit and predictability of the model, respectively.²³ The dominant metabolites associated with rats treated with alcohol and SA are annotated in the OPLS-DA coefficient plots (Fig. 4 and 5) and summarised in Tables S2 and S3.† The results of the permutation tests are shown in Fig. S2 and S3.†

Fig. 4 O-PLS-DA scores (A1, B1, C1) and coefficient-coded loadings plots (A2, B2, C2) from plasma ¹H spectra data at 6 h. controls (○), acute alcohol administration model rats (■) and SA-treated rats (◇). Correlation coefficients of the metabolites labelled in the figures are shown in Table S2.†

Fig. 5 O-PLS-DA scores (A1, B1, C1) and coefficient-coded loadings plots (A2, B2, C2) from plasma ¹H spectra data at 24 h. Controls (○), acute alcohol administration model rats (■) and SA-treated rats (◇). Correlation coefficients of the metabolites labelled in the figures are shown in Table S3.†

At 6 h, alcohol exposure induced marked elevation in the levels of lactate, glycerol, acetate, creatine and ketone bodies including acetone and acetoacetate compared to control rats. There was also a reduction in the levels of glycine and TMAO/betaine (Fig. 4A2). The results revealed that great plasma metabolic disorders appeared at 6 h post alcohol feeding, which is consistent with the severe acute liver damage and liver dysfunction observed by HE staining and clinical assessments. SA treatment reversed some of the alcohol-induced changes, especially for acetone, acetate, creatine, lactate, glycine and TMAO/betaine (Fig. 4B2). It was noted that the differentiation between the SA treatment and control groups was smaller than the classification between the model and control groups. Further analysis revealed that SA treatment completely restored the alcohol-induced changes of acetone, acetate, creatine, glycine and TMAO/betaine to normality, while the levels of lactate, alanine, acetoacetate and glycerol remained higher than those in controls (Fig. 4C2). These comparisons suggest that SA treatment alleviates the negative symptoms of acute liver damage induced by alcohol, which is confirmed by HE staining findings and clinical assessments.

Based on the statistical analysis results of the normalised integrals of metabolites, Table 2 summarized the discriminating metabolites screened out in Fig. 4, accounting for the differentiation among three groups at 6 h.

Table 2 Statistical analysis results of the main metabolite changes in plasma at 6 h^a

Metabolites	Chemical shift	6 h variations
		T1	T2	T3
a T1, model vs. control; T2, SA vs. model; T3, SA vs. control; ^*p < 0.05, ^**p < 0.01, ^***p < 0.001.
Lactate	1.33 (d)	↑^***	↓^**	↑^*
Alanine	1.48 (d)	↑^*	—	↑^*
Acetone	2.23 (s)	↑^**	↓^**	—
Acetoacetate	2.28 (s)	↑^**	—	↑^*
Acetate	1.92 (s)	↑^*	↓^*	—
Creatine	3.04 (s)	↑^**	↓^**	—
TMAO, betaine	3.27 (s)	↓^*	↑^*	—
Glycine	3.54 (s)	↓^*	↑^*	—
Glycerol	3.56 (dd)	↑^*	—	↑^*

---

The levels of lactate, TMAO/betaine, glycine and glycerol remained greatly changed in the model group after 24 h recovery (Fig. 5A2). However, it was noted that glycerol had a higher concentration at 24 h than at 6 h, which suggests that the reduced gluconeogenesis has not recovered at 24 h. Compared with model rats, SA treatment regulated the changes of all the aforementioned metabolites (Fig. 5B2). Therefore, the metabolic profiles in SA-treated rats shared similarities with those in controls at 24 h (Fig. 5C2), revealing the positive effects of SA on the recovery of acute liver injury.

Table 3 summarized the variation of the normalized integrals of plasma metabolites screened out in Fig. 5, accounting for the differentiation among three groups at 24 h.

Table 3 Statistical analysis results of the main metabolite changes in plasma at 24 h^a

Metabolites	Chemical shift	24 h variations
		T1	T2	T3
a T1, model vs. control; T2, SA vs. model; T3, SA vs. control; ^*p < 0.05, ^**p < 0.01.
Lactate	1.33 (d)	↑^*	↓^*	—
TMAO, betaine	3.27 (s)	↓^*	↑^*	—
Glycine	3.54 (s)	↓^*	↑^*	—
Glycerol	3.56 (dd)	↑^**	↓^**	—

---

Discussion

The current study analysed the metabolic consequences of SA treatment on alcohol-induced acute liver damage by employing a metabonomic strategy. Regarding the models, our results showed that acute alcohol administration caused acute liver injury with inflammation and necrosis and liver dysfunction with elevated levels of ALT and AST at 6 h after the last dose. These results are consistent with previous reports.^28–30 The NMR metabolomic analysis demonstrated that acute alcohol exposure caused comprehensive metabolic alterations in the plasma, including marked elevation in the levels of lactate, glycerol, acetate, creatine and ketone bodies, and reduction in the levels of glycine and TMAO/betaine. SA was found to be effective in the treatment of alcohol-induced acute liver damage by the results of histopathological examination and clinical biochemistry assessments at 6 h. Furthermore, it was found that SA could reverse some of the alcohol-induced metabolic changes, such as the levels of acetone, acetate, creatine, lactate, glycerol, glycine and TMAO/betaine. SA affected multiple metabolic pathways, including anaerobic glycolysis, creatinine metabolism, fatty acid oxidation, lipolysis, oxidative stress and methylation metabolism (Fig. 6). It was noted that the metabolic profiles in the control and SA treatment groups shared similar characteristics at 24 h. All of the above results indicated that SA had a significant effect on ameliorating acute liver damage and in promoting recovery.

Fig. 6 Effect of SA treatment on the plasma metabolic pathways involved in acute alcoholic liver damage rats. Metabolites in red or blue represent a higher or lower level in SA-treated rats compared with the model rats, respectively.

In the current investigation, increased levels of lactate and alanine were observed in the models compared with the controls. The plasma metabolite changes in this model were similar to those changes in human and rats with chronic alcoholic liver damage,^31,32 it has been reported that elevated lactate and alanine are due to hypoxia and glycolysis. However, this is contrary to a previous report³³ that showed a decrease in lactate in both the liver and serum of a model involving a single dose of ethanol; the lactate decrease was a result of pyruvate futile cycling in the liver. This inconsistency may be due to differences in blood collection times, the degree of liver damage and the modelling methods.^34,35 It was noted that SA treatment partially reversed the change in lactate level at 6 h, and lactate returned to its normal level at 24 h. As is well known, lactate is an end product of anaerobic glycolysis.³⁶ It is typically interpreted as a marker of anaerobic metabolism, and its accumulation usually accounts for a higher energy demand in a biological system.³⁷ Therefore, we proposed that SA treatment intervened in anaerobic glycolysis and energy metabolism disorder, as a result of decreased lactate and alanine in the acute liver damage model.

In NMR ¹H spectra, the signal of creatinine at δ 4.03 is too small to be observed due to the trace content and limited detection sensitivity. However, biochemical analyses of serum creatinine showed the significant increase in model rats. Therefore, the combined metabonomics and biochemical analysis showed that both creatine and creatinine increased at 6 h in model rats, while the SA treatment reversed the change completely. As is well known, creatinine is a nonenzymatic breakdown product of creatine and phosphocreatine, and the creatine-phosphocreatine system is crucial for cellular energy transportation.³⁸ More importantly, it is generally known that creatinine is used as a routine detection index for renal dysfunction.³⁹ It was reported that the levels of creatinine markedly increased post chronic ethanol ingestion,⁴⁰ which indicated that ethanol causes kidney function disruption, not just liver disease alone. The significantly reduced creatine and creatinine in the SA-treated rats compared with the model rats suggested that SA had an obvious impact on cellular energy transportation and in relieving the renal function disruption induced by alcohol, which has been proved by the results of renal function tests (Table 1).

TMAO and betaine are overlapped at δ 3.27. We could not confirm they were both or one of them contributed to the changes at δ 3.27. However, we could conclude that ethanol exposure altered the methylation process because both of them were all involved in methylation metabolism.^30,41–43 Therefore, the reduced levels at δ 3.27 indicated that ethanol exposure altered the methylation process, and SA treatment regulated the methylation metabolism by up-regulating the level of TMAO or/and betaine at δ 3.27.

We observed a significant elevation in the levels of ketone bodies, such as acetoacetate and acetone, in the model rats at 6 h. Similarly, it was found that ethanol feeding impacted ketone body formation.³³ As is well known, acetoacetate is produced from acetyl-CoA during the breakdown of fatty acids⁴⁴ and acetoacetate can be further converted to acetone.^34,45 De Buck et al. indicated that ketone body accumulation is a result of enhanced fatty acid oxidation.⁴⁶ In our study, the observed increases in acetoacetate and acetone may be related to the up-regulation of fatty acid oxidation in the model rats. Compared with the model rats, the SA-treated rats showed normal levels of acetone and acetoacetate at 6 h, which suggested that the SA pretreatment inhibited fatty acid oxidation and alleviated the energy metabolism disorder. We also noted an elevated concentration of acetate in the model rats, which is a product not only of fatty acid oxidation in peroxisomes but also of alcohol metabolism. In agreement with relevant reports, alcohol is conversed to acetaldehyde by a key alcohol-metabolizing enzyme (ADH),⁴⁷ and acetaldehyde is then oxidised to acetate by ALDH; the activated acetate is then introduced into the Krebs cycle as acetyl-CoA.⁴⁸ The excessive accumulation of acetate in the model rats revealed the disordered conversion systems of acetate–acetyl-CoA, which further influenced the citric acid cycle.^48–50 In our study, we observed an obvious reduction of acetate in the SA-treated rats compared with the models, indicating that SA regulated the conversion systems of acetate–acetyl-CoA and fatty acid oxidation by regulating energy metabolism.

Glycerol is an important precursor for the synthesis of glucose via gluconeogenesis and is a product of lipolysis. When energy consumption is high and the body uses stored fat as a source of energy, lipolysis is promoted, releasing excessive glycerol and fatty acids into the bloodstream.⁵¹ In our study, we observed a slight increase in glycerol at 6 h in the model rats, and it remained at a high level until 24 h. This result implied that gluconeogenesis was possibly inhibited and lipolysis was enhanced. It is also supported by previously published data, which showed that an increase of glycerol in the NAFLD/NASH was a result of reduced gluconeogenesis.⁵² It was found that SA treatment reversed the change of glycerol induced by alcohol consumption to a normal level. This result suggested that SA up-regulated gluconeogenesis and inhibited lipolysis to alleviate the metabolic disorder induced by alcohol.

In our study, a decreased level of glycine, which is considered to be closely associated with oxidative stress, was observed in the model group compared with the control group both at 6 h and 24 h. This was consistent with the findings in hepatic fibrosis.¹ As is well known, glycine possesses many properties, such as cytoprotective, antiinflammatory, immunomodulatory and so on.⁵³ Especially, it was reported that the hepatoprotective effect of glycine was considered to be associated with oxidative stress.⁵⁴ The decreased level of glycine in the models suggested that alcohol caused serious liver damage, which can be found from the HE staining of the liver. Increased glycine was found in the SA treatment group, indicating the antioxidant effect of SA on acute liver damage.

Conclusion

Our results showed that this alcohol dosage induced obvious liver damage and significant metabolic disorder at 6 h after the alcohol administration. These results are similar to the metabolic changes observed in human alcoholic liver damage. SA treatment alleviated liver injury and reversed the alterations in several metabolic pathways; for example, SA caused the inhibition of anaerobic glycolysis, fatty acid oxidation and lipolysis, it alleviated oxidative stress, and it regulated creatinine and methylation metabolism (Fig. 6). Our work shed light on the molecular mechanisms of SA on acute alcoholic liver injury through the use of ¹H NMR-based metabonomics.

Conflict of interest

The authors declare that there are no conflicts of interest.

Acknowledgements

We acknowledge the financial supports from the National Natural Science Foundation of China (81274059, 81274060, 81473413).

References

1. X. S. Yang, C. Dong and G. X. Ren, J. Agric. Food Chem., 2011, 59, 1138–1144.
2. M. G. Neuman, Alcohol Res Health., 2003, 27, 307–316.
3. K. G. Ishak, H. J. Zimmerman and M. B. Ray, Alcohol.: Clin. Exp. Res., 1991, 15, 45–66.
4. H. Fernando, S. Kondraganti, K. K. Bhopale, D. E. Volk, M. Neerathilingam, B. S. Kaphalia, B. A. Luxon, P. J. Boor and G. A. Shakeel Ansari, Alcohol.: Clin. Exp. Res., 2010, 34, 1937–1947.
5. L. H. Lian, Y. L. Wu, S. Z. Song, Y. Wan, W. X. Xie, X. Li, T. Bai, B. Q. Ouyang and J. X. Nan, J. Agric. Food Chem., 2010, 58, 13013–13019.
6. H. Liang, J. B. Wang, H. J. Xiao, D. Wang, W. Q. Wei, Y. Qiao and P. Boffetta, BMC Public Health, 2010, 10, 730.
7. L. Abenavoli, N. Milic, S. Rouabhia and G. Addolorato, World J. Gastroenterol., 2014, 20, 2159–2167.
8. A. Dey and A. I. Cederbaum, Hepatology, 2006, 43, S63–S74.
9. G. Lengyel and Z. Tulassay, Orv. Hetil., 2006, 147, 1113–1118.
10. P. Loria, G. Marchesini, F. Nascimbeni, S. Ballestri, M. Maurantonio, F. Carubbi, V. Ratziu and A. Lonardo, Atherosclerosis, 2014, 23, 99–109.
11. G. Addolorato, E. Capristo, G. Gasbarrini and G. F. Stefanini, Gut, 1997, 41, 417–418.
12. M. Bardou, S. Montembault, V. Giraud, A. Balian, E. Borotto, C. Houdayer, F. Capron, J. C. Chaput and S. Naveau, Gut, 2002, 50, 38–42.
13. K. T. Suk, M. Y. Kim and S. K. Baik, World J. Gastroenterol., 2014, 20, 12934–12944.
14. T. R. Morgan, D. Chao and G. Botwin, Clin. Liver Dis., 2013, 2, 84–88.
15. W. Zhang, R. T. Hong and T. L. Tian, Biomol. Ther., 2013, 21, 264–269.
16. M. T. Wu, B. S. Tzang, Y. Y. Chang, C. H. Chiu, W. Y. Kang, C. H. Huang and Y. C. Chen, J. Agric. Food Chem., 2011, 59, 4248–4254.
17. J. H. Park, S. J. Kim, I. Hwang, K. C. Bae, J. H. Bae and D. K. Song, Alcohol Alcohol., 2013, 48, 59–67.
18. G. Colombo, R. Agabio, C. Lobina, R. Reali, P. Morazzoni, E. Bombardelli and G. L. Gessa, Alcohol, 1999, 18, 65–70.
19. C. Y. Wang, F. L. Ma, J. T. Liu, J. W. Tian and F. H. Fu, Biol. Pharm. Bull., 2007, 30, 44–47.
20. E. Holmes, H. Tang, Y. Wang and C. Seger, The assessment of plant metabolite profiles by NMR-based methodologies, Planta Med., 2006, 72, 771–785.
21. F. C. Dong, L. L. Zhang, F. H. Hao, H. Tang and Y. Wang, J. Proteome Res., 2013, 12, 2958–2966.
22. J. Chao, T. I. Huo, H. Y. Cheng, J. C. Tsai, J. W. Liao, M. S. Lee, X. M. Qin, M. T. Hsieh, L. H. Pao and W. H. Peng, PLoS One, 2014, 9, e96969.
23. L. M. Zhang, L. M. Wang, Y. L. Hu, Z. G. Liu, Y. Tian, X. Wu, Y. Zhao, H. Tang, C. Chen and Y. Wang, Biomaterials, 2013, 34, 7117–7126.
24. F. Li, Q. Xu, T. Zheng, F. Huang and L. Han, J. Pharm. Biomed. Anal., 2014, 88, 225–234.
25. A. Zira, S. Kostidisa, S. Theocharis, F. Sigala, S. B. Engelsen, I. Andreadou and E. Mikros, Toxicology, 2013, 303, 115–124.
26. B. Shi, J. Tian, H. Xiang, X. Guo, L. Zhang, G. Du and X. Qin, Behav. Brain Res., 2013, 241, 86–91.
27. J. K. Nicholson, P. J. Foxall, M. Spraul, R. D. Farrant and J. C. Lindon, Anal. Chem., 1995, 67, 793–811.
28. S. J. Kim, Y. S. Jung, D. Y. Kwon and Y. C. Kim, Biochem. Biophys. Res. Commun., 2008, 368, 893–898.
29. R. K. Yang, X. N. Han and R. L. Delude, J. Lab. Clin. Med., 2003, 142, 322–329.
30. S. Siegmund, S. Haas, A. Schneider and M. V. Singer, Best Pract. Res., Clin. Gastroenterol., 2003, 17, 519–542.
31. B. U. Bradford, T. M. O'Connell, J. Han, O. Kosyk, S. Shymonyak, P. K. Ross, J. Winnike, H. Kono and I. Rusyn, Toxicol. Appl. Pharmacol., 2008, 232, 236–243.
32. G. Ugarte, T. Pereda, M. E. Pinto, F. Lorca and F. Sepulveda, Rev. Med. Chile, 1967, 95, 67–72.
33. P. C. Nicholas, D. Kim, F. T. Crews and J. M. Macdonald, Chem. Res. Toxicol., 2008, 21, 408–420.
34. V. L. Massey and G. E. Arteel, Front. Physiol., 2012, 3, 193.
35. H. Q. Yin, M. Kim, J. K. Kim, G. Kong, K. S. Kang, H. L. Kim, B. I. Yoon, M. O. Lee and B. H. Lee, Toxicol. Appl. Pharmacol., 2007, 223, 225–233.
36. A. P. Zhou, J. J. Ni, Z. H. Xu, Y. Wang, S. Lu, W. Sha, P. C. Karakousis and Y. F. Yao, J. Proteome Res., 2013, 12, 4642–4649.
37. E. Kokushi, S. Uno, T. Harada and J. Koyama, Environ. Toxicol., 2012, 27, 404–414.
38. Y. Y. Zhao, Y. L. Feng, X. Bai, X. J. Tan, R. C. Lin and Q. Mei, PLoS One, 2013, 8, e59617.
39. N. Monteiro, M. Branco, S. Peres, F. Borges and K. Mansinho, J. Int. AIDS Soc., 2014, 17, 19565.
40. C. Latchoumycandane, L. E. Nagy and T. M. McIntyre, Free Radical Biol. Med., 2014, 69, 403–416.
41. J. Li, X. L. Zhao, Y. X. Liu, X. H. Peng, S. F. Zhu, H. Guo, Y. L. Liu, M. H. Wan and W. F. Tang, BMC Gastroenterol., 2014, 14, 115.
42. M. Lever, P. M. George, S. Slow, D. Bellamy and J. M. Young, PLoS One, 2014, 9, e114969.
43. A. J. Barak, H. C. Beckenhauer and D. J. Tuma, Alcohol, 1996, 13, 395–398.
44. L. L. Wang, Y. F. Tang, S. Liu, S. Mao, Y. Ling, D. Liu, X. He and X. Wang, PLoS One, 2013, 8, e65675.
45. S. Kumari, R. Tishel, M. Eisenbach and A. J. Wolfe, J. Bacteriol., 1995, 177, 2878–2886.
46. J. De Buck, R. Shaykhutdinov, H. W. Barkema and H. J. Vogel, PLoS One, 2014, 9, e111872.
47. X. Liu, A. Hakucho, J. Liu and T. Fujimiya, World J. Gastroenterol., 2014, 20, 18249–18259.
48. C. H. Liu, R. Raghu, S. H. Lin, S. Y. Wang, C. H. Kuo, Y. J. Tseng and L. Y. Sheen, J. Agric. Food Chem., 2013, 61, 11231–11240.
49. A. J. Wolfe, Microbiol. Mol. Biol. Rev., 2005, 69, 12–50.
50. S. C. Zhang, G. A. Nagana Gowda, V. Asiago, N. Shanaiah, C. Barbas and D. Raftery, Anal. Biochem., 2008, 383, 76–84.
51. Y. H. Sun, C. C. Xu, C. S. Li, C. Xia, C. Xu, L. Wu and H. Zhang, Vet. Q., 2014, 10, 1080.
52. D. Beyoğlu and J. R. Idle, J. Hepatol., 2013, 59, 842–858.
53. C. Y. Chen, B. T. Wang, Z. C. Wu, W. T. Yu, P. J. Lin, W. L. Tsai and S. C. Shiesh, Clin. Chim. Acta, 2013, 420, 150–154.
54. L. Yuan and N. Kaplowitz, Mol. Aspects Med., 2009, 30, 29–41.

---

Footnote

† Electronic supplementary information (ESI) available: Table S1, ¹H NMR data and assignments of the metabolites in rat plasma; Table S2, significant changes in plasma metabolites at 6 h; Table S3, significant changes in plasma metabolites at 24 h; Fig. S1, prediction tests for the models in Fig. 3; Fig. S2, permutation test plots (200 permutations) for plasma at 6 h; Fig. S3, permutation test plots (200 permutations) for plasma at 24 h. See DOI: 10.1039/c5ra00823a

---