Anal Bioanal Chem DOI 10.1007/s00216-016-9368-4
RESEARCH PAPER
The study on serum and urine of renal interstitial fibrosis rats induced by unilateral ureteral obstruction based on metabonomics and network analysis methods Zheng Xiang 1,2 & Hao Sun 2 & Xiaojun Cai 2 & Dahui Chen 2
Received: 10 October 2015 / Revised: 18 January 2016 / Accepted: 27 January 2016 # Springer-Verlag Berlin Heidelberg 2016
Abstract Transmission of biological information is a biochemical process of multistep cascade from genes/proteins to metabolites. However, because most metabolites reflect the terminal information of the biochemical process, it is difficult to describe the transmission process of disease information in terms of the metabolomics strategy. In this paper, by incorporating network and metabolomics methods, an integrated approach was proposed to systematically investigate and explain the molecular mechanism of renal interstitial fibrosis. Through analysis of the network, the cascade transmission process of disease information starting from genes/ proteins to metabolites was putatively identified and uncovered. The results indicated that renal fibrosis was involved in metabolic pathways of glycerophospholipid metabolism, biosynthesis of unsaturated fatty acids and arachidonic acid metabolism, riboflavin metabolism, tyrosine metabolism, and sphingolipid metabolism. These pathways involve kidney disease genes such as TGF-β1 and P2RX7. Our results showed that combining metabolomics and network analysis can provide new strategies and ideas for the interpretation of pathogenesis of disease with full consideration of Bgeneprotein-metabolite.^
Electronic supplementary material The online version of this article (doi:10.1007/s00216-016-9368-4) contains supplementary material, which is available to authorized users. * Zheng Xiang [email protected]
1
School of Traditional Chinese Medicine, Guangdong Pharmaceutical University, Guangzhou 510006, China
2
School of Pharmaceutical Sciences, Wenzhou Medical University, Wenzhou 325035, China
Keywords Metabonomics . Network analysis . Renal interstitial fibrosis . UPLC-Q-TOF-MS
Introduction Chronic kidney disease (CKD) represents a major challenge to public healthcare. Approximately 10 % of the current world population suffers from CKD, which gradually leads to loss of renal function if not treated in time. CKD eventually leads to uremia and renal fibrosis, necessitating dialysis to sustain the patients’ life and causing a huge economic burden to patients and society [1]. Renal interstitial fibrosis (RIF) is a common CKD complication that develops to end-stage kidney disease. RIF is caused by the loss of normal renal units, the proliferation of fibroblasts and myofibroblasts, and the production and accumulation of extracellular matrix [2]. Recent studies have indicated that the progression of RIF involves various molecular signaling pathways, such as TGF-β/Smad [3], p38 MAPK [4], extracellular signal regulated kinase 1/2, and cJun N-terminal kinase [5]. For decades, considerable efforts have been devoted to studying the pathogenesis of RIF, but its molecular mechanism remains unresolved to date [6]. Therefore, studying the pathogenesis of RIF is important to develop anti-renal fibrosis drugs and alleviate chronic renal failure. Metabolomics can describe metabolic profile changes of endogenous substances in an organism and characterize the different physiological and pathological states of organisms under external physical, chemical, and environmental stimuli [7]. Metabolomics has been used to discover the biomarkers, mechanisms, and pathological stages of CKD [8, 9]. Boelaert and Kobayashi et al. conducted a metabolomics discovery study on blood serum samples of patients in different stages of CKD by using UPLC-Q-TOF and GC-MS. The obtained
Z. Xiang et al.
equations from biomarkers were used to predict CKD stages with 81.3 % accuracy [10]. Hankemeier et al. applied GC-MS and liquid chromatography–mass spectrometry (LC–MS) to detect potential early indicators of pathological changes, found significant biomarkers, and suggested tryptophan metabolism for further biological studies [11]. Zhang et al. have recently investigated the serum metabolic profile of RIF and determined that RIF pathology involves the metabolic pathways of lipid and ketone body syntheses and the metabolism of energy [12]. These studies indicate that the expression of biomarkers reflects the balanced state of the organism and that metabolic profiles can accurately characterize the different physiological and pathological states of organisms, thereby opening a possibility to elucidate the mechanism of CKD development. However, the relationship among the genes, proteins, and metabolites of RIF has rarely been systematically analyzed. In general, transmission of biological information involves a biochemical process of multistep cascade starting from genes [13, 14]. However, because most of metabolites reflect the terminal information of the biochemical process, it is difficult to describe the transmission process of disease information based on metabolomics method [15]. The complex network method has become a powerful tool for research on network medicine [16], network biology [17], and network pharmacology [18]. The network analysis method can visually describe this cascade of biochemical events by multi-graph and has the potential to provide new strategies for systematic study of drug-disease mechanism [19]. Here, network analysis and metabolomics methods were combined to investigate pathogenesis of RIF from the systematic perspective of disease gene-disease protein-metabolite.
Experimental section Reagents and materials HPLC-grade methanol and acetonitrile were purchased from Merck (Darmstadt, Germany). Formic acid and leucine enkephalin were obtained from Sigma-Aldrich (St. Louis, MO, USA). Ultrapure water was prepared with a Milli-Q water purification system (Millipore, France). All other reagents were of analytical grade. The assay kits for creatinine (Cre) and blood urea nitrogen (BUN) were purchased from the Nanjing Jiancheng Biotech Company (Nanjing, China). The antibodies of α-SMA, Col-I, and CTGF were obtain from Abcam company (London, UK). Animal handling Sixteen of Sprague–Dawley rats (200 ± 20 g) were supplied by the Centre of Laboratory Animals, Wenzhou Medical
University (Wenzhou, China). They are cared in the SPF laboratory. The room temperature was regulated at 23 ± 2 °C with 50 ± 5 % humidity. A 12-h light/dark cycle was set, with free access to standard diet and water. All rats were randomly divided into two groups of eight rats each as follows: model group and sham-operated group. All animals were allowed to acclimatize in metabolism cages for 1 week prior to treatment. All experiments were performed in accordance with the approved animal protocols and guidelines established by Medicine Ethics Review Committee for animal experiments of Wenzhou Medical University. Establishment of animal renal fibrosis models Unilateral ureteral obstruction (UUO) [20] is the common model of RIF. Rats were anaesthetized by abdominal injection of 10 % chloral hydrate (0.3 mL/100 g) and partially shaved. After routine sterilization and draping, a longitudinal incision was made on the left side of the abdomen. The skin was cut layer by layer until the left kidney was exposed. After dissociating the ureter, the end close to the renal pelvis was ligatured twice with 4–0 strings. The middle of this ligatured part was cut. Then all the organs were relocated. The skin was sutured layer by layer. Operation in the sham operation group was conducted as done in UUO group. After reaching the abdominal cavity, the left ureter was dissociated but not ligatured. Biochemical and immunohistochemical analyses Blood samples were collected and coagulated for 30 min and centrifuged at 3500 rpm for 10 min at 4 °C to obtain the rat serum. The contents of Cre and BUN in both of two groups were tested using biochemical assay kits. Kidney tissues were fixed in 4 % paraformaldehyde. Then, they were made into 4 μm sections through routine dehydration and paraffin embedding. Five to ten non-overlapping visual fields were selected randomly for each immunohistochemical section under ×200 microscope and the proportion of area with positive expression to total visual field area was calculated using Image Pro Plus software. Lastly, the values were averaged to obtain the average percentage of positively expressed area. Kidney tissues were fixed in 4 % paraformaldehyde, then sectioned (3 μm each) after routine dehydration and paraffin embedding. After these sections were subjected to hematoxylin-eosin staining (HE) and Masson’s trichrome staining, pathological changes in kidney tissues were observed under light microscope. Five to ten non-overlapping visual fields were randomly selected for every Masson’s trichrome-stained section under a ×200 microscope and the proportion of area with positive expression to the total visual field area was calculated considering green collagen deposition as the positive signal. Then, the average value was chosen
The study on serum and urine of renal interstitial fibrosis
as relative interstitium volume. The scoring standard adopted for semi-quantitative analysis of Masson's trichrome-stained sections was listed in Table 1 [21].
Table 1 The scoring standard for semi-quantitative analysis of Masson’s trichrome-stained sections
Animal sample collection and handling
0
30 %
Extremely severe
Samples of 24-h urine were collected on the 21st day. The fresh urine samples were immediately centrifuged at 3500 rpm for 10 min at room temperature to remove particle contaminants. Urine (200 μL) was dissolved with H2O at 1:1 ratio and subjected to vortexing for 1 min. Then, the sample was centrifuged at 12,000×g for 10 min at 4 °C. The supernatant was obtained and 0.22 micron-filtered for UPLC-Q-TOF test. Samples of blood were collected on the 22th day, 100 μL rat serum was added with 200 μL methanol and subjected to vortexing for 2 min. Then, the sample was centrifuged at 12, 000×g for 10 min at 4 °C. The supernatant was dried with nitrogen gas and re-dissolved with 100 μL aqueous acetonitrile solution (acetonitrile/water = 1:2). After vortexing for 1 min, the solution was centrifuged at 12,000×g for 10 min, and the supernatant was subjected to UPLC-Q-TOF test. Conditions of chromatography and mass spectrometry UPLC-Q-TOF was purchased from American Waters company (Waters Corp., Milford, USA) for liquid chromatography. Mobile phase A was 0.1 % (V/V) formic acid solution while mobile phase B was acetonitrile. Urine was analyzed using HSS T3 chromatographic column. Serum samples were analyzed using BEH C18 chromatographic column. Injection volume for all samples was 2 μL. Column temperature was 35 °C and the autosampler temperature was maintained at 4 °C. The gradient conditions of UPLC for urine and serum were shown in Table 2. Electro spray ionization (ESI) was chosen as the ion source and ionization mode was negative. Temperature of the ion source was 110 °C, desolventizing gas temperature was 320 °C. Desolventizing N2 flow rate was 650 L/h, and cone N2 flow rate was 50 L/h. In the negative ion mode, the capillary ionization voltage was 2.8 kV. Cone voltage for sampling was 35 V and collision energy was 6 eV. Quadrupole scan range was m/z 50–1000. Data processing and analysis The mass data acquired was imported to MarkerLynx (MassLynx software, version 4.1) for peak detection and alignment. The retention time and m/z data for each peak were determined using MarkerLynx. The MarkerLynx parameters were set as follows: mass tolerance of 0.1 Da; noise elimination level of 5; full scan mode at amass range of 50–1000 amu; and initial and final retention times of 0 and 12 (11.5) min for urine (serum) data collection, respectively. All data were
Rank
Positively expressed area
Lesion
normalized to the summed total ion intensity per chromatogram, and the resultant data matrices were introduced to SIMCA-P 12.0 software for principal component analysis (PCA) and orthogonal projection to latent structures (OPLS) analysis. Prior to multivariate statistical analysis, all variables obtained from UPLC-Q-TOF data sets were mean-centered and scaled to Pareto variance. PCA, an unsupervised multivariate statistical approach, is used for variable reduction and separation into classes. The OPLS method is designed to handle variation in X that is orthogonal to Y. The variation in X which is orthogonal to Y is modeled by the orthogonal components. To maximize class discrimination and biomarkers, the data were further analyzed using OPLS-DA. The S-plot is an easy way to visualize an OPLS/O2PLS discriminant analysis model of two classes. It has mainly been used to filter out putative biomarkers from Bomics^ data. S-plots were calculated to visualize the relationship between covariance and correlation. Variables that significantly contributed to discrimination between groups were considered as potential biomarkers and subjected to further identification of the molecular formula. Other statistical analyses used include one-way analysis of variance and independent samples t test using R project. Statistically significant differences were set at P < 0.05.
Table 2
The gradient conditions of UPLC for urine and serum
Urine
Serum
Time (min)
A (%)
B (%)
Time (min)
A (%)
B (%)
0 0.5 6 7 9 10.5 12 14 16
98 98 80 65 30 2 2 98 98
2 2 20 35 70 98 98 2 2
0 0.5 2 3.5 4 7.5 10.5 11.5 14 16
98 98 80 72.5 30 25 2 2 98 98
2 2 20 27.5 70 75 98 98 2 2
Z. Xiang et al.
Metabolites were identified with available biochemical databases, such as HMDB (http://www.hmdb.ca) [22], METLIN, (http://metlin.scripps.edu) [23], and MassBank (http://www.massbank.jp) [24]. MetaboAnalyst 3.0 was performed for pathway enrichment analysis of biomarkers and visualization of metabolomics [25]. Enriched metabolic pathways were imported into Cytoscape 2.8 via CytoKegg [26]. The proteins (metabolic enzymes) in these enriched pathways, which were combined and processed together, were extracted and mapped onto the Human Protein Reference Database (HPRD) using BioGenet plug-in. Adjacent proteins around these extracted proteins were utilized to extend the pathway network. Meanwhile, 86 target genes of kidney disease (see Electronic Supplementary Material (ESM) Table S1) were searched from DisGeNet [27], Drugbank [28], GRAC [29], and TTD [30] (ESM Table S1) with keyword Brenal or kidney.^ Among the genes obtained, 18 encoded proteins related to renal fibrosis. The extended pathway networks where disease genes and biomarkers were enriched, were integrated to establish disease protein-adjacent protein-pathway protein-metabolite network. In this network, kidney disease genes were chosen as the starting point while biomarkers were regarded as end point in order to measure the shortest distance between the two points. Thus, the shortest biochemical process from disease proteins to metabolites was obtained.
Results and discussion The contents of Cre and BUN were detected using biochemical assay kits. As seen in Table 3, there was a significant difference in Cre and BUN contents between the sham operation and UUO groups. The Cre and BUN contents in rats in UUO group increased significantly (P < 0.05). Semiquantitative analysis of immunohistochemical sections showed that the average proportions of areas with positive expression of alpha-smooth muscular actin (α-SMA), connective tissue growth factor (CTGF), and type I collagen (Col-I) in rat kidney tissues in UUO group were all significantly higher than those in sham operation group (Table 3). These results indicated that lesion had occurred in kidney tissues, leading to fibrosis. The structures of glomerulus and renal tubules were found to be normal under light microscope. The renal parenchyma in UUO group became obviously thinner, with obvious diffuse inflammatory cell infiltration and collagen in the renal interstitium. HE staining results for both of two groups are shown in Fig. S1 A and S1 B in the ESM. By observing Masson’s trichrome-stained sections under light microscope, the stained collagens were found to be mainly located in the tubular basement membrane and its surrounding environment in sham operation group. Rat collagenous fiber in UUO group
increased compared with sham operation group, and the collagen deposited in renal interstitium obviously increased. The status of Masson’s trichrome staining in both of two groups was shown in ESM Fig. S1 C and S1 D). By semi-quantitative analysis of collagens in Masson trichrome-stained sections, it was found that the relative volume of renal interstitium obviously increased in UUO group (P < 0.05) (Table 3). Immunohistochemical results of α-SMA, CTGF, and Col-I in rat kidney tissues in both groups are shown in Fig. 1. Expressions of Col-I (Fig. 1A), α-SMA (Fig. 1B), and CTGF (Fig. 1C) in rat kidney tissues were very low in sham operation group. Expressions of Col-I (Fig. 1D), α-SMA (Fig. 1E), and CTGF (Fig. 1F) in rat kidney tissues in UUO group were significantly higher. The above results further indicated that kidney tissues had been damage in terms of pathology and immunology. Pre-investigation was conducted before the full study to optimize the experimental conditions. Fingerprints of a small batch of test urinary and serum samples were acquired in positive and negative modes. We observed considerable noise and matrix effect in ESI positive mode. This condition resulted in a high baseline that would lead to the disregard of some low-abundance metabolites and the coexistence of multiple adduction ions. By contrast, adequate information on metabolites was detected and dominant quasi-molecular ions [M– H]− with a high signal-to-noise ratio formed in ESI negative ion mode. Therefore, full-scan detection was eventually set to ESI negative ion mode to maximize the number of detectable metabolites and the quality of data acquired. After optimization of the flow rate and column temperature for the chromatography and capillary voltage, flow, and temperature of desolvation gas for the mass spectrometry detector, the optimal parameters were fixed. Under the optimal conditions, a representative base peak intensity chromatogram of the rat urine and serum obtained in ESI negative ion mode is shown in Fig. 2. After processing as mentioned above, more than 2000 compounds were exported for each sample. The precision and repeatability of UPLC-Q-TOF were validated by the reduplicate analysis of six injections of the same quality control samples and six parallel samples prepared using the same preparation protocol. The relative standard deviations of the peak retention time and area value were 2 threshold of VIP. These differential metabolites were validated using Student’s t test to screen for potential biomarkers. The P value was set to 0.05 for significantly differential variables. On the basis of the above-mentioned criterion, 33 and 58 significantly different endogenous metabolites (ESM Table S2) from urine and serum were selected for further study, respectively. The precise molecular mass of endogenous metabolites was determined within a reasonable degree of measurement error (
RESEARCH PAPER
The study on serum and urine of renal interstitial fibrosis rats induced by unilateral ureteral obstruction based on metabonomics and network analysis methods Zheng Xiang 1,2 & Hao Sun 2 & Xiaojun Cai 2 & Dahui Chen 2
Received: 10 October 2015 / Revised: 18 January 2016 / Accepted: 27 January 2016 # Springer-Verlag Berlin Heidelberg 2016
Abstract Transmission of biological information is a biochemical process of multistep cascade from genes/proteins to metabolites. However, because most metabolites reflect the terminal information of the biochemical process, it is difficult to describe the transmission process of disease information in terms of the metabolomics strategy. In this paper, by incorporating network and metabolomics methods, an integrated approach was proposed to systematically investigate and explain the molecular mechanism of renal interstitial fibrosis. Through analysis of the network, the cascade transmission process of disease information starting from genes/ proteins to metabolites was putatively identified and uncovered. The results indicated that renal fibrosis was involved in metabolic pathways of glycerophospholipid metabolism, biosynthesis of unsaturated fatty acids and arachidonic acid metabolism, riboflavin metabolism, tyrosine metabolism, and sphingolipid metabolism. These pathways involve kidney disease genes such as TGF-β1 and P2RX7. Our results showed that combining metabolomics and network analysis can provide new strategies and ideas for the interpretation of pathogenesis of disease with full consideration of Bgeneprotein-metabolite.^
Electronic supplementary material The online version of this article (doi:10.1007/s00216-016-9368-4) contains supplementary material, which is available to authorized users. * Zheng Xiang [email protected]
1
School of Traditional Chinese Medicine, Guangdong Pharmaceutical University, Guangzhou 510006, China
2
School of Pharmaceutical Sciences, Wenzhou Medical University, Wenzhou 325035, China
Keywords Metabonomics . Network analysis . Renal interstitial fibrosis . UPLC-Q-TOF-MS
Introduction Chronic kidney disease (CKD) represents a major challenge to public healthcare. Approximately 10 % of the current world population suffers from CKD, which gradually leads to loss of renal function if not treated in time. CKD eventually leads to uremia and renal fibrosis, necessitating dialysis to sustain the patients’ life and causing a huge economic burden to patients and society [1]. Renal interstitial fibrosis (RIF) is a common CKD complication that develops to end-stage kidney disease. RIF is caused by the loss of normal renal units, the proliferation of fibroblasts and myofibroblasts, and the production and accumulation of extracellular matrix [2]. Recent studies have indicated that the progression of RIF involves various molecular signaling pathways, such as TGF-β/Smad [3], p38 MAPK [4], extracellular signal regulated kinase 1/2, and cJun N-terminal kinase [5]. For decades, considerable efforts have been devoted to studying the pathogenesis of RIF, but its molecular mechanism remains unresolved to date [6]. Therefore, studying the pathogenesis of RIF is important to develop anti-renal fibrosis drugs and alleviate chronic renal failure. Metabolomics can describe metabolic profile changes of endogenous substances in an organism and characterize the different physiological and pathological states of organisms under external physical, chemical, and environmental stimuli [7]. Metabolomics has been used to discover the biomarkers, mechanisms, and pathological stages of CKD [8, 9]. Boelaert and Kobayashi et al. conducted a metabolomics discovery study on blood serum samples of patients in different stages of CKD by using UPLC-Q-TOF and GC-MS. The obtained
Z. Xiang et al.
equations from biomarkers were used to predict CKD stages with 81.3 % accuracy [10]. Hankemeier et al. applied GC-MS and liquid chromatography–mass spectrometry (LC–MS) to detect potential early indicators of pathological changes, found significant biomarkers, and suggested tryptophan metabolism for further biological studies [11]. Zhang et al. have recently investigated the serum metabolic profile of RIF and determined that RIF pathology involves the metabolic pathways of lipid and ketone body syntheses and the metabolism of energy [12]. These studies indicate that the expression of biomarkers reflects the balanced state of the organism and that metabolic profiles can accurately characterize the different physiological and pathological states of organisms, thereby opening a possibility to elucidate the mechanism of CKD development. However, the relationship among the genes, proteins, and metabolites of RIF has rarely been systematically analyzed. In general, transmission of biological information involves a biochemical process of multistep cascade starting from genes [13, 14]. However, because most of metabolites reflect the terminal information of the biochemical process, it is difficult to describe the transmission process of disease information based on metabolomics method [15]. The complex network method has become a powerful tool for research on network medicine [16], network biology [17], and network pharmacology [18]. The network analysis method can visually describe this cascade of biochemical events by multi-graph and has the potential to provide new strategies for systematic study of drug-disease mechanism [19]. Here, network analysis and metabolomics methods were combined to investigate pathogenesis of RIF from the systematic perspective of disease gene-disease protein-metabolite.
Experimental section Reagents and materials HPLC-grade methanol and acetonitrile were purchased from Merck (Darmstadt, Germany). Formic acid and leucine enkephalin were obtained from Sigma-Aldrich (St. Louis, MO, USA). Ultrapure water was prepared with a Milli-Q water purification system (Millipore, France). All other reagents were of analytical grade. The assay kits for creatinine (Cre) and blood urea nitrogen (BUN) were purchased from the Nanjing Jiancheng Biotech Company (Nanjing, China). The antibodies of α-SMA, Col-I, and CTGF were obtain from Abcam company (London, UK). Animal handling Sixteen of Sprague–Dawley rats (200 ± 20 g) were supplied by the Centre of Laboratory Animals, Wenzhou Medical
University (Wenzhou, China). They are cared in the SPF laboratory. The room temperature was regulated at 23 ± 2 °C with 50 ± 5 % humidity. A 12-h light/dark cycle was set, with free access to standard diet and water. All rats were randomly divided into two groups of eight rats each as follows: model group and sham-operated group. All animals were allowed to acclimatize in metabolism cages for 1 week prior to treatment. All experiments were performed in accordance with the approved animal protocols and guidelines established by Medicine Ethics Review Committee for animal experiments of Wenzhou Medical University. Establishment of animal renal fibrosis models Unilateral ureteral obstruction (UUO) [20] is the common model of RIF. Rats were anaesthetized by abdominal injection of 10 % chloral hydrate (0.3 mL/100 g) and partially shaved. After routine sterilization and draping, a longitudinal incision was made on the left side of the abdomen. The skin was cut layer by layer until the left kidney was exposed. After dissociating the ureter, the end close to the renal pelvis was ligatured twice with 4–0 strings. The middle of this ligatured part was cut. Then all the organs were relocated. The skin was sutured layer by layer. Operation in the sham operation group was conducted as done in UUO group. After reaching the abdominal cavity, the left ureter was dissociated but not ligatured. Biochemical and immunohistochemical analyses Blood samples were collected and coagulated for 30 min and centrifuged at 3500 rpm for 10 min at 4 °C to obtain the rat serum. The contents of Cre and BUN in both of two groups were tested using biochemical assay kits. Kidney tissues were fixed in 4 % paraformaldehyde. Then, they were made into 4 μm sections through routine dehydration and paraffin embedding. Five to ten non-overlapping visual fields were selected randomly for each immunohistochemical section under ×200 microscope and the proportion of area with positive expression to total visual field area was calculated using Image Pro Plus software. Lastly, the values were averaged to obtain the average percentage of positively expressed area. Kidney tissues were fixed in 4 % paraformaldehyde, then sectioned (3 μm each) after routine dehydration and paraffin embedding. After these sections were subjected to hematoxylin-eosin staining (HE) and Masson’s trichrome staining, pathological changes in kidney tissues were observed under light microscope. Five to ten non-overlapping visual fields were randomly selected for every Masson’s trichrome-stained section under a ×200 microscope and the proportion of area with positive expression to the total visual field area was calculated considering green collagen deposition as the positive signal. Then, the average value was chosen
The study on serum and urine of renal interstitial fibrosis
as relative interstitium volume. The scoring standard adopted for semi-quantitative analysis of Masson's trichrome-stained sections was listed in Table 1 [21].
Table 1 The scoring standard for semi-quantitative analysis of Masson’s trichrome-stained sections
Animal sample collection and handling
0
30 %
Extremely severe
Samples of 24-h urine were collected on the 21st day. The fresh urine samples were immediately centrifuged at 3500 rpm for 10 min at room temperature to remove particle contaminants. Urine (200 μL) was dissolved with H2O at 1:1 ratio and subjected to vortexing for 1 min. Then, the sample was centrifuged at 12,000×g for 10 min at 4 °C. The supernatant was obtained and 0.22 micron-filtered for UPLC-Q-TOF test. Samples of blood were collected on the 22th day, 100 μL rat serum was added with 200 μL methanol and subjected to vortexing for 2 min. Then, the sample was centrifuged at 12, 000×g for 10 min at 4 °C. The supernatant was dried with nitrogen gas and re-dissolved with 100 μL aqueous acetonitrile solution (acetonitrile/water = 1:2). After vortexing for 1 min, the solution was centrifuged at 12,000×g for 10 min, and the supernatant was subjected to UPLC-Q-TOF test. Conditions of chromatography and mass spectrometry UPLC-Q-TOF was purchased from American Waters company (Waters Corp., Milford, USA) for liquid chromatography. Mobile phase A was 0.1 % (V/V) formic acid solution while mobile phase B was acetonitrile. Urine was analyzed using HSS T3 chromatographic column. Serum samples were analyzed using BEH C18 chromatographic column. Injection volume for all samples was 2 μL. Column temperature was 35 °C and the autosampler temperature was maintained at 4 °C. The gradient conditions of UPLC for urine and serum were shown in Table 2. Electro spray ionization (ESI) was chosen as the ion source and ionization mode was negative. Temperature of the ion source was 110 °C, desolventizing gas temperature was 320 °C. Desolventizing N2 flow rate was 650 L/h, and cone N2 flow rate was 50 L/h. In the negative ion mode, the capillary ionization voltage was 2.8 kV. Cone voltage for sampling was 35 V and collision energy was 6 eV. Quadrupole scan range was m/z 50–1000. Data processing and analysis The mass data acquired was imported to MarkerLynx (MassLynx software, version 4.1) for peak detection and alignment. The retention time and m/z data for each peak were determined using MarkerLynx. The MarkerLynx parameters were set as follows: mass tolerance of 0.1 Da; noise elimination level of 5; full scan mode at amass range of 50–1000 amu; and initial and final retention times of 0 and 12 (11.5) min for urine (serum) data collection, respectively. All data were
Rank
Positively expressed area
Lesion
normalized to the summed total ion intensity per chromatogram, and the resultant data matrices were introduced to SIMCA-P 12.0 software for principal component analysis (PCA) and orthogonal projection to latent structures (OPLS) analysis. Prior to multivariate statistical analysis, all variables obtained from UPLC-Q-TOF data sets were mean-centered and scaled to Pareto variance. PCA, an unsupervised multivariate statistical approach, is used for variable reduction and separation into classes. The OPLS method is designed to handle variation in X that is orthogonal to Y. The variation in X which is orthogonal to Y is modeled by the orthogonal components. To maximize class discrimination and biomarkers, the data were further analyzed using OPLS-DA. The S-plot is an easy way to visualize an OPLS/O2PLS discriminant analysis model of two classes. It has mainly been used to filter out putative biomarkers from Bomics^ data. S-plots were calculated to visualize the relationship between covariance and correlation. Variables that significantly contributed to discrimination between groups were considered as potential biomarkers and subjected to further identification of the molecular formula. Other statistical analyses used include one-way analysis of variance and independent samples t test using R project. Statistically significant differences were set at P < 0.05.
Table 2
The gradient conditions of UPLC for urine and serum
Urine
Serum
Time (min)
A (%)
B (%)
Time (min)
A (%)
B (%)
0 0.5 6 7 9 10.5 12 14 16
98 98 80 65 30 2 2 98 98
2 2 20 35 70 98 98 2 2
0 0.5 2 3.5 4 7.5 10.5 11.5 14 16
98 98 80 72.5 30 25 2 2 98 98
2 2 20 27.5 70 75 98 98 2 2
Z. Xiang et al.
Metabolites were identified with available biochemical databases, such as HMDB (http://www.hmdb.ca) [22], METLIN, (http://metlin.scripps.edu) [23], and MassBank (http://www.massbank.jp) [24]. MetaboAnalyst 3.0 was performed for pathway enrichment analysis of biomarkers and visualization of metabolomics [25]. Enriched metabolic pathways were imported into Cytoscape 2.8 via CytoKegg [26]. The proteins (metabolic enzymes) in these enriched pathways, which were combined and processed together, were extracted and mapped onto the Human Protein Reference Database (HPRD) using BioGenet plug-in. Adjacent proteins around these extracted proteins were utilized to extend the pathway network. Meanwhile, 86 target genes of kidney disease (see Electronic Supplementary Material (ESM) Table S1) were searched from DisGeNet [27], Drugbank [28], GRAC [29], and TTD [30] (ESM Table S1) with keyword Brenal or kidney.^ Among the genes obtained, 18 encoded proteins related to renal fibrosis. The extended pathway networks where disease genes and biomarkers were enriched, were integrated to establish disease protein-adjacent protein-pathway protein-metabolite network. In this network, kidney disease genes were chosen as the starting point while biomarkers were regarded as end point in order to measure the shortest distance between the two points. Thus, the shortest biochemical process from disease proteins to metabolites was obtained.
Results and discussion The contents of Cre and BUN were detected using biochemical assay kits. As seen in Table 3, there was a significant difference in Cre and BUN contents between the sham operation and UUO groups. The Cre and BUN contents in rats in UUO group increased significantly (P < 0.05). Semiquantitative analysis of immunohistochemical sections showed that the average proportions of areas with positive expression of alpha-smooth muscular actin (α-SMA), connective tissue growth factor (CTGF), and type I collagen (Col-I) in rat kidney tissues in UUO group were all significantly higher than those in sham operation group (Table 3). These results indicated that lesion had occurred in kidney tissues, leading to fibrosis. The structures of glomerulus and renal tubules were found to be normal under light microscope. The renal parenchyma in UUO group became obviously thinner, with obvious diffuse inflammatory cell infiltration and collagen in the renal interstitium. HE staining results for both of two groups are shown in Fig. S1 A and S1 B in the ESM. By observing Masson’s trichrome-stained sections under light microscope, the stained collagens were found to be mainly located in the tubular basement membrane and its surrounding environment in sham operation group. Rat collagenous fiber in UUO group
increased compared with sham operation group, and the collagen deposited in renal interstitium obviously increased. The status of Masson’s trichrome staining in both of two groups was shown in ESM Fig. S1 C and S1 D). By semi-quantitative analysis of collagens in Masson trichrome-stained sections, it was found that the relative volume of renal interstitium obviously increased in UUO group (P < 0.05) (Table 3). Immunohistochemical results of α-SMA, CTGF, and Col-I in rat kidney tissues in both groups are shown in Fig. 1. Expressions of Col-I (Fig. 1A), α-SMA (Fig. 1B), and CTGF (Fig. 1C) in rat kidney tissues were very low in sham operation group. Expressions of Col-I (Fig. 1D), α-SMA (Fig. 1E), and CTGF (Fig. 1F) in rat kidney tissues in UUO group were significantly higher. The above results further indicated that kidney tissues had been damage in terms of pathology and immunology. Pre-investigation was conducted before the full study to optimize the experimental conditions. Fingerprints of a small batch of test urinary and serum samples were acquired in positive and negative modes. We observed considerable noise and matrix effect in ESI positive mode. This condition resulted in a high baseline that would lead to the disregard of some low-abundance metabolites and the coexistence of multiple adduction ions. By contrast, adequate information on metabolites was detected and dominant quasi-molecular ions [M– H]− with a high signal-to-noise ratio formed in ESI negative ion mode. Therefore, full-scan detection was eventually set to ESI negative ion mode to maximize the number of detectable metabolites and the quality of data acquired. After optimization of the flow rate and column temperature for the chromatography and capillary voltage, flow, and temperature of desolvation gas for the mass spectrometry detector, the optimal parameters were fixed. Under the optimal conditions, a representative base peak intensity chromatogram of the rat urine and serum obtained in ESI negative ion mode is shown in Fig. 2. After processing as mentioned above, more than 2000 compounds were exported for each sample. The precision and repeatability of UPLC-Q-TOF were validated by the reduplicate analysis of six injections of the same quality control samples and six parallel samples prepared using the same preparation protocol. The relative standard deviations of the peak retention time and area value were 2 threshold of VIP. These differential metabolites were validated using Student’s t test to screen for potential biomarkers. The P value was set to 0.05 for significantly differential variables. On the basis of the above-mentioned criterion, 33 and 58 significantly different endogenous metabolites (ESM Table S2) from urine and serum were selected for further study, respectively. The precise molecular mass of endogenous metabolites was determined within a reasonable degree of measurement error (
