Metabolomics study of fasudil on cisplatin-induced kidney injury

Abstract Fasudil is a derivative of 5-isoquinoline sulfonamide, which is a Rho kinase inhibitor, a wide range of pharmacological effects. Fasudil has been shown to attenuate kidney injury caused by certain substances. In the present study, metabolomic analysis of mouse kidney tissues ultra-performance liquid chromatography/quadrupole time-of-flight mass spectrometry was used to determine the metabolomic changes in cisplatin-induced kidney injury and the fasudil-induced attenuation of cisplatin-induced kidney injury. Metabolomic profiling of kidney tissues revealed significant differences in metabolites between the control group and the cisplatin group and between the cisplatin group and the fasudil-intervention group. With metabolomic approach, 68 endogenous differential metabolites were found, and multivariate statistical analysis, accurate molecular weights, isotope tracers, mass-spectrometry secondary-fragment information, and standard-reference comparisons were used to identify these substances. Based on these differential metabolites, a metabolic-pathway network was constructed and revealed that fasudil primarily attenuated cisplatin-induced renal injury by modulating lipid and amino-acid metabolism. These results further demonstrate that kidney injury can be induced by cisplatin and, moreover, suggest that fasudil can be used to reduce kidney injury at early stages in patients treated with cisplatin.


Introduction
Cisplatin (CDDP) is a platinum-based chemotherapy drug that inhibits cellular division and exhibits broad-spectrum anti-tumor activity. CDDP can be used to treat solid tumors, such as those caused by head and neck cancer, testicular cancer, ovarian cancer, and lung cancer [1]. Cisplatin can bind to DNA bases and interfere with DNA synthesis to kill cells [2]. However, its side effects-which include severe ototoxicity, nephrotoxicity, and myelosuppression [3]-limit the use of CDDP [4]. Nephrotoxicity is the primary adverse reaction of CDDP. The increase in risk factors for nephrotoxicity can be related to the dose and frequency of CDDP use, as well as factors such as female sex, old age, and smoking. The most important and severe effect of CDDP in the kidney is acute kidney injury (AKI), and the incidence of AKI caused by CDDP is relatively high [5]. CDDP-induced kidney injury is a complex process involving multiple pathways that can cause changes at the molecular level [6,7].
AKI refers to acute injury to kidney function, including acute mild renal dysfunction [8]. Two main phenomena highlight the severity of AKI. First, AKI has serious consequences, and repeated AKI can induce chronic kidney disease [4]. For some critically ill patients, AKI is considerably dangerous and the mortality rate has been reported to be approximately 60% in such cases. [9]. Second, the incidence of AKI is markedly high, reaching 1-7% in inpatients and 1-25% in intensive-care-unit patients [10]. Additionally, the incidence of AKI is increasing over the years [11]. Efforts toward the prevention, early diagnosis, and early treatment of AKI have therefore drawn extensive attention. In recent years, several

Examination of biochemical indicators
The collected blood samples of the mice were examined in the laboratory, and the kidney function of each mouse was determined by assessing the contents of creatinine and urea nitrogen (BUN) in the blood.

Detection of serum kidney injury molecule-1 (KIM-1)
The collected blood samples of the mice were subjected to laboratory examination, and KIM-1 levels were determined using an enzyme-linked immunosorbent assay (ELISA) kit (Uscn, Wuhan, China) according to the manufacturer's instructions.

Operating conditions for mass spectrometry
Mass spectrometry analysis was performed using an electrospray ion source in positive and negative ion modes. Positive ion mode parameters were as follows: capillary voltage, 4000 V; drying gas flow rate, 11 l/min; drying gas temperature, 350 • C; nebulizing gas pressure, 45 psig, fragmentor voltage, 120 V; skimmer voltage 60 V, data acquisition range m/z 100-1100; the internal standard ions of m/z 121.0509 and m/z 922.0098 were selected for real-time mass calibration. Negative ion mode parameters were as follows: capillary voltage, 3500 V; drying gas flow rate,11 l/min;dry gas temperature, 350 • C; nebulizing gas pressure, 45 psig; fragmentor voltage 120 V, skimmer voltage 60 V; and data acquisition range m/z 100-1100; The internal standard ions of m/z 112.985587 and m/z 1033.988109 were selected for real-time mass calibration. The potential biomarker ions were further subjected to MS/MS analysis and the collision energy was adjusted between 10 and 50 V depending on the ionic conditions. The QC samples were randomly inserted in the sequence to validate the stability of the system. PCA was employed to assess the clustering of the QC samples in the PCA score plot from all the tested samples in HILIC and RP. Serum creatinine, KIM-1, and BUN levels in CDDP group were significantly higher than those in the NS group, CDDP+High Fas, and CDDP+Low Fas groups were lower than those in the CDDP group. We chose a high dose of Fas as the intervention group in subsequent studies. Compared with that in the NS group, the renal tubular structure in the CDDP group was disordered and vacuolated, indicating that the kidney injury was severe, while the renal injury in the CDDP+High Fas group was alleviated, and semi-quantitative scores of renal injury and counts of F4/80 positive cells also reflected the same trend (* statistically significant, P < 0.05) .

Data preprocessing
Raw data were converted to a universal format via Agilent MassHunter Qualitative software prior to pattern recognition. The converted data were further subjected to peak calibration and peak integration by XCMS (http://metlin. scripps.edu/download/). Finally, a 3D data matrix of retention time, mass-to-charge ratio, and peak intensity was produced. The modified 80% rule was used to remove missing values (i.e. to remove MS ions with a frequency [non-zero value] below 80% in a certain group). Data were centralized and normalized using MATLAB.

Statistical analysis
One-way analysis of variance (ANOVA) was performed using SPSS 11.0 software (IBM). The statistically significant differences among the three groups were compared. A P < 0.05 was considered indicative of statistical significance. Centralized and normalized data were imported into SIMCA-P V11.0 (Umetrics, Sweden) for principal component analysis (PCA) and partial least squares-discriminant analysis (PLS-DA), and the model was evaluated based on related R 2 and Q 2 values.

Biochemical and histological measurement
To confirmed the successful construction of mice models that showed the nephrotoxicity of cisplatin and screened the optimally protective dosage of the fasudil. We detected the two indicators commonly used in renal function: creatinine and urea nitrogen (BUN), and the level of AKI markers: the serum kidney injury molecule 1 (KIM-1). Just as shown in Figure 1A-C, Serum creatinine, KIM-1, and BUN levels in CDDP group were significantly higher than NS group, which indicated the kidney injury of nephrotoxicity. In addition, CDDP+High Fas and CDDP+Low Fas groups were lower than CDDP group, which proved the intervention of the fasudil. Although, there were no statistical difference between CDDP+High Fas and CDDP+Low Fas groups, CDDP+High Fas group had a lower trend. In thus, a high dose of fasudil at 40 mg/kg was defined as the optimal dose for the follow-up drug intervention group and used in next histological measurement and metabonomics analysis. Just as shown in Figure 1D-G, compared with the NS group, the renal tubular structure in the CDDP group was disordered and vacuolated, indicating that the kidney injury was severe, while the renal injury in the CDDP+High Fas group was alleviated, and semi-quantitative scores of renal tissue injury reflect the same trend ( Figure 1H-K). It has been reported that inflammation plays a major role in the pathophysiology of AKI [30]. Meanwhile, inflammation was evaluated by protein expression of the F4/80 macrophage marker, the counts showed the same trend, which further confirmed CDDP-induced kidney injury and the protective effects of Fas.

Metabolomics profiling
PCA score plot including all the test and QC samples shows that the QC sample features were tightly clustered in Supplementary Figures S1 and S2. The results demonstrated that the stability of the proposed method was satisfying. According to previous UPLC-MS conditions, kidney tissues samples of the mice from the NS group, CDDP group, and the CDDP+High Fas group were analyzed in hydrophilic interaction liquid chromatography (HILIC) and Reversed-phase chromatography (RPLC) modes. To investigate the CDDP-induced nephrotoxicity and changes in urinary metabolic profiles in mice following Fas intervention, unsupervised PCA and PLS-DA was used to determine differences in metabolites among the three groups, as shown in Figures 2 and 3 in different modes. There was The samples from different groups showed differences in PCA score plots and PLS-DA score plots from the NS, CDDP, and CDDP+ High Fas Groups clustered together and were clearly separated. a clear separation trend in the NS group, CDDP group, and CDDP+Fas group, indicating a certain degree of difference among the three groups. In HILIC modes, when three components were calculated in the positive mode, the cumulative R 2 X, R 2 Y, and Q 2 were 0.432, 0.975, and 0.487, respectively, while the cumulative R 2 X, R 2 Y, and Q 2 in the negative mode were 0.773, 0.946, and 0.684. Meanwhile, the cumulative R 2 X, R 2 Y, and Q 2 were 0.612, 0.965, and 0.592 in RPLC positive modes and 0.587, 0.957, and 0.415 in RPLC negative modes. No over-fitting was observed in either ESI positive or ESI negative according to the results of the permutation test.

Screening potential biomarkers
To identify biomarkers associated with CDDP-induced nephrotoxicity and the renal protection effect of Fas, we used a PLS-DA method to produce the score plot and scatter plot (S-Plot) of the PLS-DA for both HILIC and RPLC modes in the positive and negative ion modes. As shown in Figure 4, there was an apparent separation among the NS group, CDDP group, and CDDP+ High Fas group. In the S-plot, a greater distance from the origin indicated a greater contribution to the difference among the groups. In this experiment, we considered metabolites with a variable importance in projection (VIP) > 1.0 to be representative of differential metabolites. We further compared the differential variables among the NS group, CDDP group, and CDDP+ High Fas group using one-way ANOVA. The variables that differed significantly among the three groups were considered differential variables. A preliminary assignment of the mass-spectral peaks was performed based on the element-matching function provided by high-resolution mass spectrometry data, as well as via Metlin (http://metlin.scripps.edu/), HMDB (http://www.hmdb.ca/), and Mass Bank (http://www.massbank. The jp/) database searches. Further high-resolution MS/MS analysis of the samples was performed to obtain mass-spectrometry fragment information of differential metabolites, and the differential metabolites were structurally confirmed by network database searches and comparisons with standards. Finally, 68 differential metabolites associated with the cisplatin nephrotoxicity and the kidney protection of fasudil were identified, shown in Table 1. The results indicate that the metabolic pattern in the kidney of mice with cisplatin-induced renal injury was reversed to normal levels after fasudil treatment.

Possible metabolic pathways of differential metabolites
According to the network databases MetaboAnalyst (http://www.metaboanalyst.ca) and KEGG (http://www.genome. jp/kegg), the 68 differential metabolites were constructed into a metabolic-pathway map ( Figure 5). The tricarboxylic acid cycle in the energy metabolism was used as the center in this metabolic map, which involved multiple metabolic pathways that mainly included lipid metabolism and amino-acid metabolism. In terms of lipids, three phospholipids-lysophosphatidylcholine (LysoPC), phosphatidylcholine (PC), and phosphatidylethanolamine (PE)-were the main metabolites of the pathway. In the amino-acid metabolism, the metabolism of three aromatic amino acids-tryptophan, tyrosine, and phenylalanine-was dominant, which indicates that these metabolism pathways may play key roles in mediating the protective effect of Fas on CDDP-induced kidney injury and this metabolic pathway map may be related to the mechanism underlying the nephrotoxicity caused by CDDP and the renal protection of Fas, Deeper investigation of these pathways should be helpful in elucidating the therapeutic mechanism of Fas.

Figure 5. Metabolic pathway network in kidney tissues and the effect of Fas on CDDP-induced kidney injury
The levels of related differential metabolites identified in the NS, CDDP, and CDDP+ High Fas Groups are showed in the histogram.
The different metabolites in the colored square frame refer to the same species of LysoPC, PC, and PE. The red squares represent metabolites found in the study, whereas the gray squares represent metabolites not found in the study, which were predictable metabolites in the pathway network. The solid arrows indicate the direct production of downstream or upstream metabolites and dotted arrows indicated the indirect production of downstream or upstream metabolites, * means statistic differences (P < 0.05) between CDDP Group versus NS Group, # means statistic differences (P < 0.05) between CDD+ High Fas Group versus CDDP Group.

Discussion
As a traditional anti-tumor drug, CDDP has been demonstrated to exhibit potent nephrotoxicity. There are four main pathophysiological mechanisms of AKI, namely proximal tubule injury, oxidative stress, inflammation, and renal vascular injury. We successfully constructed a mouse model of CDDP-induced AKI via intraperitoneal injection of CDDP. On this basis, we used Fas as a therapeutic intervention in CDDP-induced AKI mice and demonstrated that Fas had a protective effect on CDDP-induced AKI. Additionally, we used metabolomeic methods to analyze the metabolites in mouse blood samples and searched for potential biomarkers of AKI at the level of endogenous metabolites, attempting to elucidate the mechanisms of CDDP-induced AKI and Fas-mediated renal protection. Among the 68 differential metabolites found, there were 19 phospholipids, of which PC, LysoPC, and PE were dominant. The fold change of PC in the CDDP group was less than 1 relative to the control group, while the fold change of PC in the Fas group was more than 1 relative to the CDDP group. These findings indicated that the PC level in the mouse blood decreased after the injection of CDDP but increased in mouse blood after the administration of Fas. It has been reported in the literature that PC has a protective effect on CDDP-induced nephrotoxicity in rats [31]. Similarly, in the study, the PC level was substantially reduced after injection of CDDP in mice, which may be due to PC reducing the degree of CDDP-induced kidney damage through its antioxidative activity [32] and a large amount of PC being consumed. In the blood of mice given Fas, PC was significantly higher than that in mice with CDDP-induced AKI that were not given Fas, suggesting that Fas has a protective effect on CDDP-induced AKI, reduces the oxidative stress induced by CDDP, and prevents excessive consumption of PC. As a metabolite of PC, the LPC level was also reduced in the CDDP group, but the difference between the Fas group and the CDDP group was not significant. LPC participates in inflammatory responses through various pathways in the body [33], thereby affecting CDDP-induced AKI. Because a large amount of PC is used to reduce the damage caused by CDDP in the body, a greater amount of PE needs to be converted to PC. Therefore, the PE level in the CDDP group was significantly higher than that in the control group, while the PE level in the Fas group was higher than that in the CDDP group. We hypothesize that Fas may reduce tissue damage from CDDP by increasing PE levels. Sphinganine-1-phosphate is also a metabolite of PE, and its content in the blood of the CDDP group also showed an upward trend. LysoPA (18:2/0:0) can be produced by LysoPE (18:2/0:0), and it has a great influence on neutrophil chemotaxis. In the present study, the LysoPA blood level during CDDP-induced AKI was decreased. In addition, phosphatidylinositol (PI) (20:4/18:0), sphingomyelin (SM; d18:1/18:0), and their products are related to signal transduction. The use of CDDP can increase the blood contents of these molecules, suggesting that CDDP may induce related signaling transductions in the body to become abnormal. Sphingosine 1-phosphate is also a metabolite of SM (d18:1/18:0). Sphingosine 1-phosphate can increase the level of prostaglandin E2 (PGE2) by inducing the expression of cyclooxygenase-2 (COX-2) [34], thereby affecting inflammatory responses in the of kidney. Additionally, studies have shown that sphingosine 1-phosphate can protect the kidney with AKI by activating sphingosine 1-phosphate receptors [35], and the magnitude of this protective effect was reduced when the sphingosine 1-phosphate levels in the CDDP group were decreased. There are also reports indicating that sphingosine 1-phosphate can activate Rho-kinase [36], which can damage renal tubules [16]. The body reduces the level of Rho-kinase-induced damage to the renal tubules by reducing the level of sphingosine 1-phosphate, and Fas protects the kidneys because it is a Rho-kinase inhibitor. In the present study, the level of sphingosine 1-phosphate in the Fas group was lower than that in the CDDP group. We hypothesize that Fas inhibits Rho-kinase by reducing the level of sphingosine 1-phosphate. Unsaturated fatty acids are another type of important PC metabolite. In the present study, we found arachidonic acid. Arachidonic acid can be produced from linoleic acid. Arachidonic acid can directly and indirectly mediate inflammatory response and it is a precursor of prostaglandins, thromboxane, and leukotriene. Arachidonic acid can also generate hydroxyeicosatetraenoic acid (HETE) under the action of cytochrome P450 (CYP450), and we detected 16R-HETE in the present study, which can inhibit the adhesion and aggregation of neutrophils. HETE regulates disease processes by affecting intravascular balance, inflammation, cell growth, apoptosis, and oxidative stress [37]. The most important mechanisms underlying kidney injury induced by CDDP are inflammation and oxidative stress. In the present study, the level of 16R-HETE was considerably elevated in the blood of mice injected with CDDP, and the body inhibited inflammatory responses and oxidative stress by increasing 16R-HETE levels.
In the present study, we found that the levels of eight amino acids showed dramatic differences in the mouse sera among the three groups, including six essential amino acids (L-valine, L-lysine, L-proline, L-tryptophan, L-phenylalanine, and L-tyrosine) and two non-essential amino acids (L-homocysteine and L-kynurenine). L-proline, L-tryptophan, L-phenylalanine, and L-tyrosine showed similar trends in the mice from all three groups, and their concentrations were lower in the CDDP group than in the control group. The trends of L-valine, L-lysine, L-homocysteine, and L-kynurenine were opposite; their concentrations in the CDDP group were higher than those in the control group, and their concentrations in the Fas group were lower than those in the CDDP group. It has been shown in the literature that L-tryptophan can inhibit the early stage of renal failure caused by adriamycin, [38], so a large amount of L-tryptophan may have been consumed in the mice from the CDDP group in our present study. It can be seen on the metabolic pathways map that L-tyrosine can be converted to L-tryptophan. Among the differential metabolites found in the present study, Indolelactic acid and L-kynurenine were the products of L-tryptophan. The receptors of tyrosine kinases (Eph family) serve as key modulators of various cellular functions [39]: it has been reported that overexpression of EphA1 in the kidney attenuates renal fibrosis and improves renal function by inhibiting Rho and MAPK [40]. As the inhibitor of Rho kinases, Fas may alleviate kidney injury by promoting the expression of EphA1, which in turn activates tyrosine kinases. Higher activation of tyrosine kinases consumes more tyrosine, which may indicate why the level of tyrosine was lower after the administration of Fas than in the CDDP group in our present study. Among the metabolites of L-phenylalanine, caffeic acid and phenylacetylglycine were the differential metabolites found in the present study. Caffeic acid phenethyl ester has been shown to have protective effects against nephrotoxicity and oxidative kidney damage. [41]. The decrease in the level of caffeic acid in the present study may have been due to the synthesis of caffeic acid phenethyl ester to attenuate CDDP-induced nephrotoxicity. It has also been reported that caffeic acid inhibits organic anion transporters (OAT1 and OAT3) in rat kidneys. [42] However, there has been no report on the relationship between caffeic acid and organic cation transporter (OCT), and OCTs have a close relationship with the entry of CDDP into kidney cells [43]. Phenylacetylglycine may be a suitable biomarker associated with mitochondrial toxicity [44]. Pantothenic acid is produced by L-valine metabolism, and it is the raw material of coenzyme A (CoA). In the present study, the pantothenic acid level of the model group showed an upward trend compared with that of the control group, which may indicate that CDDP causes abnormal energy metabolism. N-acetyllysine and N-acetylalanine were produced by L-lysine and L-proline, respectively, and their levels were increased significantly in the CDDP group in the present study. N-acetyllysine is associated with histone acetylation. Low-dose histone deacetylase inhibitors can protect cells in vitro and attenuate CDDP-induced apoptosis. [45]. Both L-homocysteine and taurine are produced by cysteine. Among the differential metabolites found in the present study, L-homocysteine and taurine exhibited similar trends. Specifically, the L-homocysteine and taurine levels in the CDDP group showed an upward trend relative to those in the control group, and the levels in the Fas group were lower than those in the CDDP group. L-homocysteine and taurine have antioxidative effects in the body, and studies have shown that taurine can exert anti-inflammatory effects through its antioxidative effects to attenuate CDDP-induced nephrotoxicity. [46]. Urea is the main final catabolic product of amino acids, and the fold change of the CDDP group relative to the control group was 3.544 in the present study, while the fold change of the Fas group relative to the CDDP group was 0.796. Creatine is a metabolite of arginine and it is dehydrated to form creatinine. The fold change of the creatinine level in the CDDP group relative to the control group was 5.343 in the present study, while the corresponding fold change in the Fas group relative to the CDDP group was 0.731. Because urea and creatinine are mainly excreted through the urine, an increase in their blood levels may suggest damaged kidney function, which may have confirmed that the CDDP-induced kidney injury model was successfully established in our present study. Creatinine and urea nitrogen are also important indicators for evaluating kidney function. Collectively, these results demonstrate that Fas had a protective effect on kidney function in the present study.
Energy metabolism also showed dramatic changes in the present study. It is based on the tricarboxylic acid cycle and involves the energy metabolism of sugars, amino acids, and fatty acids. Citric acid, cis-aconitic acid, and succinic acid are three organic acids in the tricarboxylic acid cycle. In the present study, their blood levels in the mice from the three groups showed substantial changes, but their trends were not all consistent. The blood glucose level in the CDDP group was significantly higher than that in the control group, and the concentration of pyruvic acid-the main product of glucose-was considerably lower in the CDDP group in the present study. The concentration of gluconic acid, the other product of glucose, was increased by more than seven times in the CDDP group, which may be due to the pathway by which pyruvic acid is produced from glucose being blocked. The serum levels of the other two non-primary products of pyruvic acid-mesaconic acid and 2-C-methyl-D-erythritol 4-phosphate-did increase in the CDDP group, suggesting that the normal energy metabolism pathway had been inhibited. The increased levels of mesaconic acid and 2-C-methyl-D-erythritol 4-phosphate may also be associated with a decrease in pyruvic acid levels in the CDDP group in the present study. Long-chain fatty acids enter mitochondria by forming fatty acyl carnitine. The blood levels of four long-chain fatty acyl carnitines-palmitoylcarnitine, vaccenylcarnitine, linoleylcarnitine, and stearoylcarnitine-found in the present study were significantly higher in the CDDP group, and those of the Fas group were lower than those in the CDDP group. However, the level of propionylcarnitine-another short-chain fatty acyl carnitine-was lower in the mice from the CDDP group, and the short-chain carnitine is produced by β-oxidation of long-chain carnitine. The main source of proximal tubule energy is fatty acids, and CDDP can affect the oxidation of fatty acids by inhibiting kidney PPAR-α, [47], such that the four long-chain fatty acyl carnitines cannot be oxidized and continue to increase. Because short-chain fatty acyl carnitines cannot be normally generated and enter the tricarboxylic acid cycle, the short-chain fatty acyl carnitine, propionylcarnitine, is continuously consumed and reduced. At the same time, Fas may protect fatty acid metabolism in the kidney, thereby reducing the level of CDDP-induced kidney damage. We further explored whether Fas had an agonistic effect on peroxisome proliferator-activated receptor α (PPAR-α). The reason for the decrease in L-carnitine levels in the CDDP group may be due to CDDP hindering the synthesis of L-carnitine, or it may be because fatty acids constantly bind to L-carnitine, which cannot be released via oxidation.
We found inosine, xanthine, uric acid, and allantoin to be metabolites in purine metabolism. A large amount of xanthine dehydrogenase can be converted into xanthine oxidase under conditions of adenosine triphosphate (ATP) deficiency, and inosine can produce hypoxanthine, which is first oxidized to xanthine under the action of xanthine oxidase and then oxidized to uric acid. In this process, the levels of inosine, xanthine, and uric acid were significantly lower in the CDDP group than in the control group in the present study. The decrease in inosine and xanthine levels in the present study may have been be due to an increase in the level of xanthine oxidase. Under the action of CDDP, the body produces a large number of oxygen free radicals and causes oxidative stress responses. Moreover, oxidative stress contributes to the activation Rho-kinase, which causes target organ injury [48]. Uric acid has an antioxidative effect and a large amount of uric acid is oxidized to allantoin, which can be used as a marker of oxidative stress [49]. The level of uric acid is then greatly reduced and the content of allantoin increased. In the present study, the allantoin level in the CDDP group was significantly higher than that in the control group, while the allantoin level in the Fas group was lower than that in the CDDP group. This may be because Fas, an inhibitor of Rho-kinase, can reduce the content of oxygen free radicals caused by CDDP.
We also found several nephrotoxic substances and substances with large fold changes in these differential metabolites. Indoxyl sulfate and p-cresol sulfate are well-defined nephrotoxic substances. Under the action of intestinal flora, tryptophan and tyrosine produce indoxyl and p-cresol, which are transported to the liver to carry out sulfation to form indoxyl sulfate and p-cresol sulfate. We also found p-cresol glucuronide, which is the product of p-cresol glucuronidation. The blood levels of these three substances increased substantially after the injection of CDDP in mice in the present study. The highest fold change (15.114) of the CDDP group compared with that of the control group was observed in the indoxyl sulfate level in the RPLC mode in the present study, and indoxyl sulfate can act as a sensitive indicator to determine chemical-induced AKI. [50]. Indoxyl sulfate and p-cresol sulfate can induce inflammation in proximal tubule cells [51] and indoxyl sulfate can induce apoptosis in glomerular mesangial cells [52]. In the present study, there was no significant decrease in the blood levels of these two substances in the mice injected with Fas, indicating that Fas cannot attenuate CDDP-induced increases in the levels of these substances. The fold change of the CDDP group compared with that in the control group at the aminobenzamide level was 12.142. The increase in the aminobenzamide level enhanced the antioxidative activity [53] and reduced further oxidative damage to the kidney. Ascorbate 2-sulfate also has a strong antioxidative effect, and in RPLC mode, the fold change of the CDDP group was 9.434 compared with the control group.

Conclusions
In the present study, we analyzed the complex process of CDDP-induced kidney injury from the level of endogenous metabolites by metabolomics and explored the mechanisms underlying the early action of Fas to attenuate CDDP-induced kidney injury. We found 68 potential biomarkers involving phospholipid metabolism, amino-acid metabolism, energy metabolism, and purine metabolism. Additionally, we preliminarily elucidated possible related metabolic pathways and explored the mechanisms of CDDP-induced kidney injury and Fas-mediated renal protection. These findings may provide a reference for early diagnosis and intervention of AKI. However, the molecular mechanisms underlying CDDP-induced kidney injury and Fas-mediated renal protection require further investigation.

Author Contribution
Demeng Xia wrote the manuscript. Xueli Lai and Kaiwen Wu analyzed the data. Panyu Zhou and Lei Li performed the experiments. Zhiyong Guo and Shuogui Xu designed the study.