Metabolomic assessment of mechanisms underlying anti-renal fibrosis properties of petroleum ether extract from Amygdalus mongolica

Abstract Context The petroleum ether extract (PET) of Amygdalus mongolica (Maxim.) Ricker (Rosaceae) has an ameliorative effect on renal fibrosis (RF). Objective To evaluate the antifibrotic effects of A. mongolica seeds PET on RF by serum metabolomics, biochemical and histopathological analyses. Materials and methods Sixty male Sprague-Dawley rats were randomly divided into the sham-operated, RF model, benazepril hydrochloride-treated model (1.5 mg/kg) and PET-treated (1.75, 1.25, 0.75 g/kg) groups, and the respective drugs were administered intragastrically for 21 days. Biochemical indicators including BUN, Scr, HYP, SOD, and MDA were measured. Haematoxylin and eosin and Masson staining were used for histological examination. The serum metabolomic profiles were determined by UPLC-Q-TOF/MS and metabolism network analysis. Acute toxicity test was performed to validate biosafety. Results The PET LD50 was >23.9 g/kg in rats. PET significantly alleviated fibrosis by reducing the levels of Scr (from 34.02 to 32.02), HYP (from 403.67 to 303.17) and MDA (from 1.84 to 1.73), and increasing that of SOD (from 256.42 to 271.85). Metabolomic profiling identified 10 potential biomarkers, of which three key markers were significantly associated with RF-related pathways including phenylalanine, tyrosine and tryptophan biosynthesis, amino sugar and nucleotide sugar metabolism and tyrosine metabolism. In addition, three key biomarkers were restored to baseline levels following PET treatment, with the medium dose showing optimal effect. Conclusions These findings revealed the mechanism of A. mongolica PET antifibrotic effects for RF rats on metabolic activity and provided the experimental basis for the clinical application.


Introduction
Renal fibrosis (RF) and chronic kidney disease (CKD) affect 50% of the adults over 70 years of age, and 10% of the global population (Boor et al. 2010;Schmitt 2012). RF is a set of various physiological and pathological changes that accompany CKD of different aetiologies and progress to end-stage renal failure (Chen et al. 2016). Therefore, delaying or reversing the fibrotic process in renal tissues is a promising therapeutic strategy against CKD (Nogueira et al. 2017). Currently, RF is mainly treated with angiotensin converting enzyme inhibitors (ACEI), angiotensin receptor antagonists, transforming growth factor b 1 (TGF-b 1 ) neutralising antibodies or natural antagonists. In addition, gene therapy approaches have also been tested in clinical trials. Traditional Chinese medicine (TCM) formulations have the advantages of pleiotropic action and minimal side effects, and are a promising alternative for the prevention and treatment of RF (Ellison 2017;Ji and He 2018). According to the TCM theory, the treatment of RF involves clearing and purging the turbidity. The mechanistic basis of TCM action is increasingly being elucidated through metabolomics and transcriptomics, and several promising biomarkers have been identified for the early diagnosis and treatment of fibrosis (Qi et al. 2014;. Amygdalus mongolica (Maxim.) Ricker (Rosaceae) of the flat peach family is a deciduous, drought-tolerant shrub endemic to the Mongolian plateau (Ma 1989;Zhao 1995). According to the Records of Inner Mongolia Plant Medicine, the seeds have medicinal properties and are used in the 'Yu Li Ren' (Pruni semen) TCM formulation (Si 2003) to relieve cough, resolve phlegm, moisten dry intestines and dry throat, and treat Yin deficiency constipation and edoema through its diuretic effects. In addition, Pruni semen is used as an auxiliary treatment for kidney disease and nephritis (Zhou & Zheng 1960). The chemical constituents of A. mongolica include unsaturated fatty acids, amygdalin, organic acids, proteins, flavonoids, polysaccharides, alkaloids, a-VE and other pharmaco-active compounds Su et al. 2013;Zheng DH et al. 2013;Bai et al. 2015;Zheng et al. 2018). The petroleum ether extract (PET) of A. mongolica is rich in fatty acids such as oleic acid, palmitic acid, palmitic acid, linoleic acid, 8,11-octadecenedioic acid and arachidonic acid that can inhibit liver fibrosis by reversing lipid peroxidation (Zheng et al. 2016(Zheng et al. , 2018. Furthermore, A. mongolica extracts can also prevent pulmonary and renal fibrosis Hao et al. 2020;Jia et al. 2020), and the reno-protective effects have been attributed to PET and n-butanol. However, the mechanisms underlying the antifibrotic effects of A. mongolica PET are still unclear. A previous study showed that the effective dose range of A. mongolica PET for hypolipidemic and antioxidant action is 0.5426-1.5 g/kg (Zheng et al. 2016(Zheng et al. , 2018. Furthermore, oral administration of 21.5 g/kg A. mongolica oil for 14 days did not induce any toxic reactions in a rat model . A recent study showed that linoleic acid constitutes 67.56% of A. mongolica PET (Zheng et al. 2018). Since polyunsaturated fatty acids can effectively combat renal fibrosis in obstructive nephropathy by improving dyslipidemia, reducing lipid peroxidation and inflammation and inhibiting myofibroblast proliferation (Zhao et al. 2013;Han 2014), it is rational to surmise that linoleic acid is the major pharmaco-active component of A. mongolica. This study further investigates the therapeutic effects and potential mechanisms of A. mongolica PET in RF.
Metabolomics is the study of the metabolite profiles of cells, biological fluids and tissues, and their interactions. It is a highthroughput approach for screening novel biomarkers, and particularly useful for elucidating the multiple targets and metabolic pathways of TCM (Buriani et al. 2012;Zhou et al. 2014;Lyu et al. 2018). In this study, we established a rat model of RF by unilateral ureteral obstruction (UUO), and used metabolomics to explore the mechanism of PET action against RF. Our findings provide an experimental basis for the pharmacological analysis of other TCM formulations.

Materials
The seeds of A. mongolica were collected from Alashanyabrai Gobi Inner Mongolia in 2018. The dry mature seeds were identified by Professor Songli Shi of Inner Mongolia University of Science and Technology Medical College of Baotou (Supplementary Figures S1 and S2). Benazepril hydrochloride was purchased from Beijing Novartis Pharmaceutical Co. Ltd., pentobarbital sodium from Merck (Germany) and sodium carboxymethyl cellulose from Tianjin Kaitong Chemical Reagent Co. Ltd. The penicillin sodium injection solution was from North China Pharmaceutical Co. Ltd. (batch number: F7116323). HE and Masson staining kits were purchased from Nanjing Jiancheng Technology Co. Ltd. LC-MS grade methanol (67-56-1) and acetonitrile (75-05-8) were obtained from Honeywell, and formic acid from SIGMA (64-18-6).

Preparation of petroleum ether extract from Amygdalus mongolica
The seeds of A. mongolica were peeled, deshelled and crushed, and extracted with 95% and 70% ethanol. The extraction conditions were as follows: temperature 70 C, solid-to-liquid ratio 1:10, and time 2 h. The extracts were combined and concentrated under reduced pressure, and the total ethanol extract was uniformly dispersed in distilled water, extracted with petroleum ether, and concentrated under reduced pressure to obtain PET (28.5% yield). PET composition was analysed as previously described (Zheng et al. 2018), and the main fatty compounds were unsaturated fatty acids (73.2%) such as oleic acid (28.87%) and linoleic acid (38.69%), and saturated fatty acids like palmitic acid (19.66%).

Acute toxicity test of A. mongolica PET
Ten male and female SD rats each (weighting 180-200 g) were acclimatised for 3-4 days and fasted overnight. The animals were given 23.9 g/kg PET once via the oral route and observed for 14 days for the clinical signs of toxicity .

Establishment of UUO model and treatment regimen
Sixty male SPF grade male Sprague-Dawley rats weighing 170-200 g were purchased from the Department of Medical Sciences of Peking University (Department of Experimental Animal Science; licence number SCXK (Beijing) 2011-0012). The animals were randomly divided into the sham-operated (SDG), model (MOD), benazepril-treated (BH), and the high (PET-H), medium (PET-M) and low (PET-L) dose PET groups (n ¼ 10 each). RF was induced by unilateral ureteral ligation as described previously . The animals in the SDG and MOD groups were given daily oral gavage of 4 mL normal saline. The dosage of BH was 1.5 mg/kg/day, and that of PET-H, PET-M and PET-L were 1.75, 1.25 and 0.75 g/kg/day, respectively (Zheng et al. 2016(Zheng et al. , 2018. The drugs were administered for 21 days. The experimental protocol was approved by Medical Ethics Committee of Baotou Medical College of Inner Mongolia University of Science and Technology.

Specimen collection
Twenty-four hours after the last drug administration, the animals were weighed and anaesthetised by intraperitoneal injection of 3% pentobarbital. Blood was drawn from the abdominal aorta and centrifuged at 3000 rpm for 10 min. The serum was collected for biochemical and metabolomic analyses. The left kidney lobe was cut into 1 cm 3 pieces and fixed in 10% paraformaldehyde solution for HE and Masson staining, and the right lobe was frozen at À80 C.

Biochemical and histological examination
Serum levels of serum creatinine (Scr), blood urea nitrogen (BUN) and serum albumin (ALB) were measured by an automated analyser. The content of superoxide dismutase (SOD), malondialdehyde (MDA) and hydroxyproline (HYP) in the renal tissues were determined using specific kits as per the manufacturer's instructions. For histological examination, the paraffinembedded tissues were cut into 3-4 lm-thick sections and stained with HE and Masson dyes as per standard protocols. The sections were observed under a light microscope for pathological changes, and fibrotic injury was scored as previously described .
The mass spectrometry (MS) conditions were as follows: ion source Gas1 À 60 PSI, ion source Gas2 À 60 PSI, interface heating temperature À650 C, ion spray voltage in positive ion mode À500 V, ion spray voltage in negative ion mode À4500 V. MS data were collected in the IDA mode. The TOF quality scan range was 60-1200 Da and was completed within 150 milliseconds with a total cycle time of 0.56 s. A 40 GHz four-anode/ channel multi-channel TDC detector was used to monitor the scans, and the four times of each scan were added when the pulse frequency was 11 kHz.

Data analysis
The raw MS data was converted to mzXML format using the MSConvert software. The XCMS program was then used for data processing, including peak selection, peak grouping, retention time correction, second peak grouping, and isotope and adduct extract annotation. The metabolites were annotated using online KEGG and HMDB databases by matching the exact molecular mass data (m/z) of samples with the standards. Finally, the identified metabolites were identified using an inhouse fragment spectrum library. The multivariable data matrix was analysed by SIMCA-P 14.1 software (Umetrics AB, Umeå, Sweden). Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were used to visually analyse the clustering results, and the potential biomarkers of RF were screened based on variable importance of projection (VIP) > 1 and p value (t-test) < 0.05. The metabolic pathways of differential metabolites were analysed by MetPA with MetaboAnalyst 4.0 software, and the key RF-related pathways were screened on the basis of impact > 0.02 and À log (P) > 2. The diagnostic power of the putative biomarkers was determined by the ROC test, the KEGG pathway database was used for network analysis. SPSS17.0 was used to analyse the experimental data that were expressed as x6s: The means of different groups were compared by single factor analysis of variance (p < 0.05).

Acute toxicity test
No apparent symptoms were detected following oral administration 23.9 g/kg PET for 14 days, indicating that it is non-toxic according to the 'Food Safety Toxicity Evaluation Procedures and Methods' .

Determination of biochemical indicators and histopathological analysis
The serum levels of BUN, Scr and ALB, and the content of MDA and HYP in the renal tissues were significantly higher in the MOD versus SDG groups, whereas the renal SOD activity was significantly reduced in the former (p < 0.01), indicating successful RF modelling. Treatment with BH and the low and medium doses of PET restored all indices to normal levels (p < 0.01; Figure 1; Table 1), thereby showing reno-protective and antioxidant effects.
As shown in the H&E and Masson-stained tissue images in Figure 2, there were no significant histopathological changes in the tubule-interstitium of the kidneys in the SDG rats. In contrast, UUO modelling resulted in markedly reduced glomeruli, significant atrophy of renal tubules with bare basement and interstitial widening, massive infiltration of inflammatory cells, and focal fibrosis. Treatment with BH, PET-L, PET-M and PET-H reduced interstitial inflammation and fibrosis, and mitigated the damage to some renal tubules as well. PET-M showed the most significant effect (p < 0.01). Taken together, both PET and BH can protect kidney tissues from UUO-induced fibrosis.

Multivariate data analysis
Serum samples of each group were assessed by principal component analysis (PCA) and partial least squares discrimination analysis (PLS-DA) in positive and negative ion mode ( Figure 3). As shown in the PCA score plots (Figure 3(A,B)), the samples of each group were well clustered in the unsupervised mode. In addition, the MOD and SDG showed greater distinction in the anion mode, while PET-L, PET-M and PET-H were distinct from MOD and closer to SDG. PLS-DA on the other hand showed further separation of MOD from SDG, PET-L, PET-M and PET-H, while the treatment groups were closer to SDG and PET-M was most similar to SDG (Figure 3(C,D)). The replacement test diagram was used to assess PLS-DA overfitting ( Figure  3(E,F)), which showed that R 2 and Q 2 from left to right was lower than the original value on the far right and Q 2 intersected with Y axis in the negative half axis, indicating good fitting and reliable results.

Screening and identification of potential biomarkers
As shown in the PLS-DA score plots (Figure 4(A,B)), MOD and SDG had good differentiation in positive and negative ion modes, which indicated distinct metabolomic profiles. In addition, the class permutation tests (Figure 4(C,D)) indicated good fitness and predictive values of these models. Ten differentially abundant metabolites (eight in positive ion mode and two in negative ion mode) were identified in the model group relative to the control based on VIP > 1 and p < 0.05 (Table 2). As shown in the heat maps in Figure 5(A), the metabolites that increased (red) or decreased (green) in the model group were restored to normal levels by the different doses of PET.
Moreover, we compared the concentrations of potential biomarkers among all groups and demonstrated that biomarkers were normalised or reversed by the treatment with PET, respectively. Venn-diagram ( Figure 5(C)) shows the results of number of potential biomarkers for significant callbacks in each group, PET-H, PET-M and PET-L could respectively callback 5, 9, and 8 potential biomarkers, and simultaneously act on 4 same potential biomarkers together. Taken together, RF significantly alters the abundance of several metabolites, which are potential diagnostic biomarkers.

Diagnostic potential and biological function of RF-related metabolites
ROC analysis was performed on the 10 potential biomarkers to determine their diagnostic potential. As shown in Figure 6(A), the AUC values of all biomarkers were greater than 0.9, indicating high diagnostic accuracy.
MetPA was next performed to identify the crucial biomarkers among the ten potential metabolites and the metabolic pathways most affected by RF. As shown in Figure 6(C), three metabolic pathways involved in amino sugar and nucleotide sugar metabolism, phenylalanine, tyrosine and tryptophan biosynthesis and tyrosine metabolism were markedly dysregulated (-Log(P) >2 and impact >0.02). These pathways are considered are closely associated with the occurrence and development of RF. Finally, three key serum metabolites were identified from these pathways. All values represent the mean ± SD. Ã p < 0.05 and ÃÃ p < 0.01 compared to the model group. # p < 0.05 and ## p < 0.01 compared to the sham group. Values represent the mean ± SD. Ã p < 0.05 and ÃÃ p < 0.01 compared to the model group. # p < 0.05 and ## p < 0.01 compared to the sham group.

Discussion
Renal fibrosis is a complex pathological process, and metabolomic techniques can help screen for sensitive biomarkers and analyse endogenous metabolites in vivo to elucidate its underlying mechanisms. We established a rat model of RF by UUO, and used pharmacodynamics and metabolomics to study the antifibrotic mechanisms of A. mongolica PET. Benazepril, an angiotensin II inhibitor that reduces plasma aldosterone levels, increases vasodilatation, lowers blood pressure, relieves oxidative stress, increases glomerular filtration rate and reduces proteinuria (Wang et al. 2006;Zhang et al. 2013), was used as the positive control. The reno-protective effects of the different drugs were analysed on the basis of biochemical indices. Serum BUN and Scr levels are indicators of the extent of renal damage and fibrosis (Orchard et al. 2001;Lijnen 2005). The pathological basis of renal fibrosis are oxidative stress and inflammation . MDA, the final product of free radical-induced lipid peroxidation, is elevated in fibrotic lesions ). In addition, the content of HYP is closely related to the severity of renal interstitial fibrosis (Xiao et al. 2015). We found that different doses of A. mongolica PET reduced the levels of serum BUN, Scr and ALB, and the in situ content of MDA and HYP, and increased SOD activity. Thus, PET can delay the fibrotic process by alleviating oxidative damage, which was also corroborated by the significant improvement in the histopathological indices of fibrosis (Xue et al. 2004). A previous study showed that the effective dose range of A. mongolica PET for its hypolipidemic and antioxidant action is 0.5426-1.5 g/kg. Since the high dose (1.75 g/kg) used in our study was higher than the maximum effective dose, the medium dose (1.25 g/kg) was more effective in delaying and mitigating renal fibrosis. Metabolomics is a powerful tool for evaluating drug efficacy and mechanism of action. Metabolomics analyses showed that PET mainly targets amino sugar and nucleotide sugar metabolism, phenylalanine, tyrosine and tryptophan biosynthesis and tyrosine metabolism, which are affected during renal fibrosis and lead to aberrant levels of L-tyrosine, dopamine and N-acetyl-D-glucosamine 6-phosphate (Figure 7). The medium PET dose was effective on all three metabolic pathways and restored the key biomarkers, and was therefore the optimal dose for antifibrotic effects.
L-Tyrosine is a non-essential amino acid and a precursor of various bioactive molecules, including brain catecholamine neurotransmitters including norepinephrine and dopamine (DA) (Wang et al. 2019). DA is an endogenous nitrogen-containing    (2007) showed that L-tyrosine scavenged free radicals like DPPH (1,1-diphenyl-2-picrylhydrazyl radical), ABTS, superoxide anion and H2O2, and also reduced ferrous ion chelation. Øvrehus et al. (2019) found that 11 amino acids, including tyrosine, phenylalanine, dopamine, homocysteine and serine, were affected in hypertensive nephrosclerosis patients. This disorder is characterised by a 30-70% decrease in urine output, endothelial dysfunction, atherosclerosis and renal fibrosis due to dysregulated dopamine biosynthesis in the kidney (Zhang 2013). N-Acetyl-D-glucosamine 6-phosphate (GLCNAC6P) is involved in the metabolism of amino sugars and nucleotides. It is produced following glycolysis of N-acetyl-D-glucosamine by N-acetyl-D-glucosaminidase (NAG). NAG is abundant in the lysosomes of renal proximal tubule cells, and a marker of albuminuria and microalbuminuria in patients with type 2 diabetes mellitus, as well as of renal injury, inflammation and oxidative stress . Taken together, A. mongolica PET alleviates oxidative stress and inflammation in the kidney, and delays renal fibrosis by restoring amino sugar and nucleotide sugar metabolism.

Conclusions
Amygdalus mongolica PET inhibited RF and improved renal function by restoring three key biomarkers and three metabolic pathways, which are likely involved in inhibiting release of inflammatory factors, reducing oxidative stress response, and regulating energy metabolism disorder. This study provides the experimental basis for the clinical application of A. mongolica PET.