Abstract Background Growing evidence suggests a complex bidirectional interaction between gut microbes, gut-derived microbial metabolites, and diabetic kidney disease (DKD), known as the “gut-kidney axis” theory. The present study aimed to characterize the role of microbial metabolites in DKD. Methods Six-week-old db/db and littermate db/m mice were raised to 20 weeks old. The serum, urine, feces, liver, perinephric fat, and kidney were analyzed using liquid chromatography with tandem mass spectrometry (LC-MS/MS)-based metabolomic analyses. Results The db/db mice showed obvious pathological changes and worse renal functions than db/m mice. Indoleacetaldehyde (IAld) and 5-hydroxy-l-tryptophan (5-HTP) in kidney samples, and serotonin (5-HT) in fecal samples were increased in the db/db group. Phosphatidylcholine (PC), phosphatidate (PA), and 1-acylglycerophosphocholine (lysoPC) were decreased in liver and serum samples of the db/db group, while PC and lysoPC were decreased in kidney and perinephric fat samples. Suggested metabolomic homeostasis was disrupted in DKD mice, especially glycerophospholipid and tryptophan metabolism, which are closely related to the gut microbiome. Conclusions Our findings reveal the perturbation of gut microbial metabolism in db/db mice with DKD, which may be useful for building a bridge between the gut microbiota and the progression of DKD and provide a theoretical basis for the intestinal treatment of DKD. Keywords: Diabetic kidney disease, Gut microbial metabolism, Glycerophospholipid metabolism, Tryptophan metabolism 1. Introduction Diabetes mellitus (DM) affects approximately 10% of the global population [[43]1], and its morbidity and mortality are mainly related to complications of multiple organ systems, such as the kidneys. Moreover, approximately 40% of the diabetes cases progress to diabetic nephropathy (DKD), which is the prime reason of chronic kidney disease (CKD) globally and leads to kidney failure [[44]2]. Despite the tight control of blood sugar and lipids in patients with DKD, renal function often remains poor, partly due to the limited cognition of the complex metabolic processes of DKD [[45]3,[46]4]. DKD is related to many metabolic pathways and complex metabolic pathology [[47]5,[48]6]. Intestinal bacteria produced by the host organism produce metabolites, such as short-chain fatty acids [[49]7]and uremic toxins [[50]8]. These metabolites act as a bridge between DKD and gut microbiota, and can more directly reveal the mutual effect between disease and gut microbiota. The gut microbiota is considered as a key regulator of DKD development in patients with DM [[51]9,[52]10]. Intestinal dysbiosis leads to intestinal mucosal barrier impairment and entry of gut microbial uremic toxins into the circulatory system, which leads to systemic microinflammation, insulin resistance, and kidney damage [[53]11]. Gut microbiota and microbial metabolites, such as tryptophan and polyamine metabolism, can mediate renal fibrosis in CKD rats [[54]12,[55]13]. Increasing evidence indicates a complex correlation among intestinal microbiota, gut-derived microbial metabolites, and DKD [[56][14], [57][15], [58][16]]. However, there are few studies on the effects of specific metabolites from gut microbes and exogenous substances on DKD [[59]17]. The pathophysiological correlation between gut-derived microbial metabolites and DKD is unclear. Metabolites are the intermediate or final products of biochemical reactions, respectively, which can match the disease phenotype and reflect the biological information of genetics, drugs, food, environment, gut microbes, and host [[60]18]. Identifying the microbial metabolites derived from the gut of DKD and exploring their effects on multi-organ metabolism will help to unearth the precise pathogenesis of DKD and provide a theoretical basis for the intestinal treatment of DKD. In this study, we explore the possible mechanisms between the gut-serum-liver-perinephric fat-kidney metabolic axis and microbial metabolites in mice with DKD. 2. Materials & methods 2.1. Animals and research design Male 6-week-old C57BLKS/J-leprdb/leprdb mice (db/db mice, n = 15) and matched littermate C57BLKS/J-leprdb/leprm mice (db/m mice, n = 15, as normal controls) were purchased from the Nanjing University Experimental Animal Center. All mice were fed with 10 mL/kg sterile water for 12 weeks after adaptive feeding. All mice were maintained in SPF environment (22 ± 2 °C, 55 ± 10% humidity, with a 12-h light/dark cycle) and received food/water without restriction. From the 3rd to 14th week, we recorded body weights weekly, and fasting blood glucose (FBG) every four weeks. The 24-h urine samples, whole blood samples, feces, kidney tissues, perinephric fat, and liver tissues were obtained from each mouse at the completion of the animal experiment. The whole blood samples were allowed to stand for about 1 h at room temperature and centrifuged at 3000 rpm for 10 min. After centrifugation of the whole blood samples, the upper serum was collected. All left kidneys were taken for pathological section, and the right ones were taken for metabolomics analysis. Serum and other specimens were stored at −80 °C for biochemical assays and untargeted metabolomic sequencing. The experiments in this study were carried out in line with the requirements of the Laboratory Animal Experimentation law, and were allowed by the Experimental Animal Ethics of Jinan University (approval No. 202069-04). A detailed flowchart of experiments is shown in [61]Fig. 1a. Fig. 1. [62]Fig. 1 [63]Open in a new tab Basic characteristics of db/m (n = 6) and db/db (n = 6) mice. (a) The workflow of the animal experiment. (b) Weight changes in mice during the intervention. (C) FBG levels of mice during the intervention. (d–g) The level of (d) BUN, (e) serum Cys-C, (f) UCr, and (g) urine microalbumin in the two groups. (h) The average glomerular perimeters and (i) glomerular area detected by H&E staining in the two groups. (j) The percentage of glomerular and (k) interstitial fibrosis detected by Masson staining in the two groups. (l) Representative renal tissue pathology (200×) of mice in each group (H&E staining, PAS staining, and Masson staining). *P < 0.05 (db/m vs. db/db); “ns”, not significant. Fasting-blood glucose, FBG; blood urea nitrogen, BUN; serum cystatin C, Cys-C; urine creatinine, UCr. 2.2. Biochemical analysis and histopathological examination FBG levels were measured by a portable blood glucose meter (BAYER, Germany) and blood glucose test strips (BAYER, Germany). The levels of serum cystosin C (Cys-C) were detected using a mouse Cys-C ELISA Kit (E-EL-M3024, Elabscience, Wuhan, China). Urine creatinine (Ucr) and urine microalbumin were measured using Ucr enzyme-linked immunosorbent (ELISA) kit (MM-44289M1, Elabscience, Wuhan, China) and MAU/ALB ELISA kit (MM-0705M1, Elabscience, Wuhan, China), respectively. Blood urea nitrogen (BUN) levels were quantified using an automated biochemical analyzer (Hitachi High-Tecgnoologies, Japan). Student's t-test was used for comparison between the two groups. The formalin-fixed tissue was embedded in paraffin and cut into 4 μm thick sections, and further used for hematoxylin-eosin (H&E), Masson, and Periodic Acid Schiff (PAS) staining. The average glomerular perimeter, glomerular area, and the percentage of the fibrotic area were detected using the ImageJ software. Histological analysis was operated by two independent investigators using a blinded method. 2.3. Metabolomic analysis 2.3.1. Sample preparation 50 mg of each tissue was added in 1000 μL of internal standard mixture (methanol:acetonitrile:water in a 2:2:1 ratio). Next, the tissues were ground for 4 min at 35 Hz, and then sonicated thrice on ice bath for 5 min each. Each serum sample (50 μL) was added to internal standard mixture (acetonitrile:methanol in a 1:1 ratio, 200 μL). The serum was vortexed for 30 s and then sonicated on an ice bath for 10 min. After standing at −40 °C for 1 h, all tissues and serum samples were cryogenically centrifuged for 15 min at 4 °C and 12000r. The supernatants were collected for further testing. 2.4. LC-MS/MS analysis The UHPLC system (Vanquish, Thermo Fisher Scientific) with UPLC BEH Amide (2.1 mm × 100 mm, 1.7 μm) combined with a Q-Exactive HFX mass spectrometer (Orbitrap MS, Thermo) was used for LC-MS/MS analysis. Mobile phase A was composed of 25 mmol/L ammonium acetate and 25 ammonia hydroxide in water, while mobile phase B was composed of acetonitrile. The detailed parameters of MS conditions can be retrieved from our published article [[64]19]. 2.5. Data processing The MS data was changed into the mzXML format through the ProteoWizard software. Further raw data processing, such as peak detection, extraction, alignment, and integration, was performed using an in-house R package. The metabolites were annotated by the online database, namely Human Metabolome Database (HMDB) and Kyoto Encyclopedia of Genes and Genomes (KEGG), and a secondary MS database (BiotreeDB V2.1) with 0.3 as the cut-off. The quality control results show that the sample quality, experimental method, and system stability were reliable and suitable for subsequent metabolomic analysis. For details, see Supplementary Materials and Methods and [65]Figs. S1–S5. The relative intensity of each metabolite was normalized using the total ion current (TIC) of each metabolite. Metabolites with >50% missing values were removed from the data, and half the minimum value superseded the remaining null values. 2.6. Statistical analysis MetaboAnalyst 5.0 ([66]www.metaboanalyst.ca) was used for metabolomic data analysis. Global metabolic changes between the two groups were determined using the Orthogonal partial least squares discriminate analysis (OPLS-DA) model. The model parameters R2 and Q2 served as indicators for evaluating the interpretability and predictability of the model, respectively. R2 and Q2 > 0.05 indicate that the model was robust and reliable. The OPLS-DA model produced variable importance in projection (VIP) values. Metabolites satisfying conditions (P < 0.05, VIP >1, and |fold change| > 1.5) were selected as differentially expressed metabolites (DEMs). The metabolic pathway enrichment analysis was performed using the pathway analysis module of Metaboanalyst 5.0. Pathways with p < 0.05 and impact >0.1 were considered statistically significant. The identified metabolites were classified according to the HMDB database. The DEMs were classified using MetOrigin ([67]http://metorigin.met-bioinformatics.cn/) [[68]20]. The DEMs from microbiota and co-metabolites were considered as gut-derived microbial metabolites. And we performed a correlation analysis of microbial metabolites and co-metabolites from feces and the serum, liver, kidney, and perinephric fat ([69]Tables S17–S20). The levels of fecal microbial metabolites and co-metabolites were associated with the levels of microbial metabolites and co-metabolites from the serum, liver, kidney, and perinephric fat. In a correlation analysis network diagram, the variations of some microbial metabolites in feces are correlated with their variations in serum and peripheral organs ([70]Fig. S6). Volcano plots and heat maps were generated using the R packages ‘ggplot’ and ‘pheatmap’, respectively. Spearrelations between gut-derived microbial DEMs were performed in R using the Hmisc package. Data with a correlation coefficient >0.6 were visualized as chord diagrams in R using circlize packages. 3. Results 3.1. The baseline data of db/db mice Baseline characteristics have been described in our previously published paper [[71]21]. The body weight, FBG, renal function index (blood urea nitrogen, Cystatin-C, urine microalbumin) were higher in db/db than db/m mice from 8th - 20th week of feeding (P < 0.05). The db/db mice exhibited more severe renal pathology than the db/m mice. These results suggest the development of DKD in db/db mice. The baseline data were shown in [72]Fig. 1b–l and [73]Table S1. 3.2. Fecal metabolite profiling in db/db mice In all, 1255 metabolites (positive mode: 983 and negative mode: 272) were identified from fecal samples. OPLS-DA models indicate noticeable metabolic differences between the db/db and db/m feces samples (R2 = 0.995, Q2 = 0.759) ([74]Fig. 2a). We identified that 133 metabolites were upregulated, while 20 were downregulated ([75]Fig. 2b, [76]Table S2). Moreover, 153 DEMs in the feces were significantly enriched in six pathways, involving three metabolic abnormalities: energy metabolism (lipid and amino acid metabolism), and cofactor and vitamin metabolism ([77]Fig. 2c, [78]Table S3). Based on the origin of metabolites, the DEMs were further classified into five groups as follows: host (3), microbiota (13), co-metabolite (18), exogenous (93), and unknown (26) ([79]Fig. 2d, [80]Table S2). Co-metabolites refer to the DEMs derived from the microbiota and host. To better understand the underlying role of the gut microbiome, we analyzed the correlations and biological functions of 31 DEMs from microbiota and co-metabolites. In metabolite-metabolite correlations, most gut-derived microbial DEMs in the network were lipids ([81]Fig. 2e, [82]Table S4). The pathway analysis revealed that amino acid metabolism, including tryptophan metabolism, tyrosine metabolism, and phenylalanine metabolism and glycerophospholipid metabolism were the main pathways in the db/db group ([83]Fig. 2f, [84]Table S3). The expression and classification of these 31 microbe-derived metabolites are shown in [85]Fig. 2g ([86]Table S2), and we found that lipid metabolites accounted for the most. These findings suggest that amino acid and lipid metabolism disturbances characterize gut-derived microbial metabolites in fecal metabolic profiles. Fig. 2. [87]Fig. 2 [88]Open in a new tab Fecal metabolic profiling in db/db mice. (a) OPLS-DA score showing the conspicuous differential metabolic features of feces between db/m and db/db groups. (b) Volcano plot showing 153 DEMs in feces between the db/m and db/db groups. (c) KEGG pathway analysis of 153 DEMs in feces between the db/m and db/db groups. (d) Column graph showing the number of fecal DEMs from different sources. Co-metabolites refer to the DEMs derived from microbiota and host. (e) The chord diagram showing significant correlations between fecal DEMs of different classes or within the same class. Metabolite class is shown as a color bar around the circumference. Each line indicates a significant correlation (Spearman's correlation, r > 0.6, p < 0.05). Pink, positive correlation; Cyan, negative correlation. (f) KEGG pathway analysis of DEMs derived from microbiota and co-metabolites. (g) Heatmap of 31 DEMs derived from microbiota and co-metabolites of fecal samples. Metabolite classes and origins are shown on the left of the heatmap. The red font represents DEMs in the tryptophan metabolism and the glycerophospholipid metabolism pathway. Phosphatidylcholine, PC. (For interpretation of the references to color in this figure legend, the