Introduction

Chronic kidney disease (CKD) is a progressive condition that often leads to end-stage kidney disease (ESKD) and remains a major global health concern. Among the hereditary causes of CKD, Alport syndrome (AS) is a genetic disorder affecting the basement membrane; it is characterized by progressive glomerulonephropathy and sensorineural hearing impairment, which is caused by mutations in the genes encoding one of the three type IV collagen alpha chains1. Clinically, AS presents with early hematuria, followed by progressive proteinuria, and eventually renal failure, accounting for a substantial proportion of approximately 0.5% of new-onset end-stage kidney disease (ESKD) cases in adults and 12.9% in children2. Although renin-angiotensin system inhibitors help slow down renal decline, a substantial number of patients still progress to ESKD at a young age; therefore, early biomarkers and targeted interventions should be developed3,4,5.

Omic techniques have considerably advanced in the last decade, thereby revealing biomarkers for diagnosis and determining therapeutic targets for various diseases. In kidney diseases, various omic studies have been conducted using biofluids, such as blood, urine, and biopsy tissue6,7. Metabolomics has emerged as a cutting-edge platform for detecting changes in metabolite levels in biofluids and tissues in pathological states. It has also been used in the treatment of chronic kidney disease (CKD) in search for diagnostic and predictive biomarkers8,9. However, the altered metabolomic profile of AS has yet to be explored.

Thus, to enhance the understanding of the altered metabolomic profile of AS, we performed metabolomic profiling on Col4a3 knockout mice (Col4a3−/−) as a widely recognized AS model10,11 and examined the changes in organic acids (OAs), TCA cycle metabolites, free fatty acids (FFAs), amino acids (AAs), kynurenine pathway metabolites (KYNs), and nucleosides (NS). We aimed to characterize the metabolic alterations in the plasma and urine of Col4a3−/− AS mouse models at 4 weeks (4 w; early disease stage) and 7 weeks (7 w; advanced disease stage) compared with those of age-matched wild-type (WT) controls. We selected these ages based on previous studies showing that Col4a3−/− mice develop initial renal pathology by ~ 4 weeks and reach severe kidney disease by ~ 7 weeks10,11. Thus, we sought to identify the metabolomic changes that could serve as biomarkers for the early diagnosis and monitoring of AS progression.​.

Results

Altered metabolic profiles and univariate analysis

Plasma

Among the 205 metabolites included in our in-house database established using standard solutions, 90 metabolites comprising 19 OAs, 18 FFAs, 32 AAs, 9 KYNs, and 12 NSs were determined in the plasma samples from the WT and AS mice at 4 and 7 w of age by using GC-MS/MS and LC-MS/MS in the MRM mode (Sup. Table 1). The metabolite levels in the AS groups (4 and 7 w) were normalized to those in the corresponding WT groups (4 and 7 w), respectively, to evaluate metabolic alterations. Star symbol pattern analysis was performed to visually observe metabolic differences and distinguish between the WT and AS groups across the five metabolite classes (Fig. 1). The Wilcoxon rank-sum test revealed that 14 metabolites, including palmitic acid, N2,N2-dimethylguanosine, and N2-methylguanosine, showed significant changes (p < 0.05) between the 4 w WT and AS groups; furthermore, 20 metabolites, including 1-methylhistidine, erucic acid, and behenic acid, were significantly altered (p < 0.05) between the 7 w WT and AS groups (Sup. Table 1). The median values of plasma results and the normalized values are provided in Sup. Table 3.

Fig. 1
figure 1

Star symbol plots depicting the changes in plasma metabolite profiles between the WT and AS mice at 4 weeks (A–E) and 7 weeks (F–J). Each plot represents a specific metabolite class: (A,F) OAs, (B,G) FFAs, (C,H) AAs, (D,I) KYNs, and (E,J) NSs. The numbers on the rays correspond to those in Supplementary Table S1. WT, wild type; AS, Alport syndrome. *p < 0.05.

Full size image

Urine

A total of 110 metabolites, including 52 OAs, 30 AAs, 11 KYNs, and 17 NSs, were determined in the urine samples collected from the WT and AS mice at 4 and 7 w of age via GC-MS/MS and LC-MS/MS in the MRM mode (Sup. Table 2). The metabolite levels in the AS groups (4 and 7 w) were normalized to those in the corresponding WT groups (4 and 7 w), respectively, to assess metabolic alterations. Star symbol pattern analysis was conducted to visually monitor metabolic variations and discriminate between the WT and AS groups based on urinary profiles (Fig. 2). The Wilcoxon rank-sum test indicated that 19 metabolites, including β-alanine, 3-indoleglyoxylic acid, and 2-hydroxyhippuric acid, exhibited significant changes (p < 0.05) between the 4 w WT and AS groups; moreover, 35 metabolites, including alanine, acetoacetic acid, and hexanoylglycine, were significantly altered (p < 0.05) between the 7 w WT and AS groups (Sup. Table 2). The median values of urine results and the normalized values are presented in Sup. Table 4.

Fig. 2
figure 2

Star symbol plots of the urinary metabolite profiles comparing the WT and AS groups at 4 weeks (A–D) and 7 weeks (E–H). Each radar plot shows different metabolite classes: (A,E) OAs, (B,F) AAs, (C,G) KYNs, and (D,H) NSs. The numbers on the rays correspond to those in Supplementary Table S2. WT, wild type; AS, Alport syndrome. *p < 0.05.

Full size image

Multivariate analysis

Plasma

Based on the criteria of p value (< 0.05) and variable importance in the projection (VIP) score (> 1.0) from partial least squares discriminant analysis (PLS-DA; Sup. Table 1), 28 metabolites, including palmitic acid, 5-methylcytidine, and MTA, were considered key candidates that distinguished between the WT and AS groups in the plasma of 4 and 7 w mouse models. Any metabolite that satisfied these criteria at either 4 or 7 w was included in the selection. Additional PLS-DA was performed using these metabolites, which distinguished the WT and AS groups at 4 and 7 w (Fig. 3A and D), to further validate the discriminative capability of the 28 selected metabolites; this distinction observed at 7 w (Sup. Figure 1B), as compared to 4 w (Sup. Figure 1 A), was further confirmed by heatmap analysis.

Fig. 3
figure 3

Multivariate and pathway analysis of plasma metabolites in the WT and AS mice at 4 weeks (A–C) and 7 weeks (D–F) by using 28 selected discriminatory metabolites. (A,D) PLS-DA score plots showing the separation between the WT and AS groups. (B,E) VIP scores from PLS-DA analysis, highlighting the top metabolites contributing to group separation. (C,F) Metabolic pathway enrichment plots based on the identified discriminatory metabolites. Circle size reflects pathway impact; color intensity reflects − log10 (p value), where red corresponds to more significant pathways.

Full size image

Urine

Based on the same selection criteria used for plasma (Sup. Table 2), 42 metabolites, including lactic acid, isovalerylglycine, and glutamic acid, were selected as key candidates that distinguished between the WT and AS groups in the urine of 4 and 7 w mouse models. Metabolites were included if they met these criteria at either time point (4 or 7 w). Further PLS-DA using the 42 selected metabolites did not separate the WT and AS groups at 4 w (Fig. 4A), but a clear separation was observed at 7 w (Fig. 4D and Sup. Figure 1D) compared to 4 w (Fig. 4C and Sup. Figure 1 C) which was also supported by heatmap analysis.

Fig. 4
figure 4

Multivariate and pathway analysis of urinary metabolites in WT and AS mice at 4 weeks (A–C) and 7 weeks (D–F) by using 42 selected discriminatory metabolites. (A,D) PLS-DA score plots showing the separation between the WT and AS groups. (B,E) VIP scores from PLS-DA analysis, highlighting the top metabolites contributing to group separation. (C,F) Metabolic pathway enrichment plots based on the identified discriminatory metabolites. Circle size reflects pathway impact; color intensity reflects − log10 (p value), where red corresponds to more significant pathways.

Full size image

Identification of differential metabolites

Plasma

Twenty-eight metabolites were examined via univariate ROC analysis, which is widely used to evaluate the diagnostic performance of individual metabolites (Table 1). The results showed that 13 metabolites, including 1-methylguanosine, N2-methylguanosine, and N2,N2-dimethylguanosine, in the plasma of the 4 w AS group were significantly altered (p < 0.05, VIP score > 1.0, and AUC > 0.8) compared with those in the plasma of the 4 w WT group (Fig. 3B). Furthermore, 14 metabolites, including 1-methylhistidine, erucic acid, and behenic acid, in the plasma of the 7 w AS group were significantly altered (p < 0.05, VIP score > 1.0, and AUC > 0.8) compared with those in the plasma of the 7 w WT group (Fig. 3E). Notably, palmitic acid, 5-methylcytidine, and MTA showed consistent trends (p < 0.05, VIP score > 1.0, and AUC > 0.8) in the same direction at both time points, while creatinine displayed opposite trends (p < 0.05, VIP score > 1.0, and AUC > 0.8): decreased at 4 w and increased at 7 w.

Table 1 Levels of the 28 selected metabolites, p values, VIP score, and AUC values in plasma of 4 w and 7 w AS mouse model.
Full size table

Urine

Forty-two metabolites were assessed via univariate ROC analysis (Table 2). Six metabolites, including β-alanine, 3-indoleglyoxylic acid, and 2-hydroxyhippuric acid, in the 4 w urine samples and 31 metabolites, including acetoacetic acid, hexanoylglycine, and N2,N2-dimethylguanosine, in the 7 w urine samples (Fig. 4E) showed significant alterations (p < 0.05, VIP score > 1.0, and AUC > 0.8) compared to 4 w (Fig. 4B) between the WT and AS groups. Among them, seven metabolites were altered (p < 0.05, VIP score > 1.0, and AUC > 0.75) in the same direction at both time points. Conversely, suberic acid and 5-hydroxyindoleacetic acid exhibited opposite trends (p < 0.05, VIP score > 1.0, and AUC > 0.75) between 4 and 7 w.

Table 2 Levels of the 42 selected metabolites, p values, VIP score, and AUC values in urine of 4 w and 7 w AS mouse model.
Full size table

Metabolic pathway analysis

Plasma

Pathway analysis was performed to investigate the biological relevance of the 28 differential metabolites. In the 4 w plasma, the key pathways included pyruvate metabolism, TCA cycle, and glycolysis/gluconeogenesis (p < 0.05, impact score > 0.1; Fig. 3C). Similarly, in the 7 w plasma, the main altered pathways were alanine, aspartate and glutamate metabolism, D-glutamine and D-glutamate metabolism, and pyruvate metabolism (p < 0.05, impact score > 0.1; Fig. 3F). A complete list of significantly altered pathways is provided in the corresponding figures.

Urine

Pathway analysis was performed to confirm the association of the 42 differential metabolites with relevant metabolic pathways. In the 4 w urine, the main altered pathways were as follows: alanine, aspartate, and glutamate metabolism; β-alanine metabolism; and pyruvate metabolism (p < 0.05, impact score > 0.1; Fig. 4F). In the 7 w urine, the key pathways were ketone body synthesis and degradation, tryptophan metabolism, and alanine, aspartate, and glutamate metabolism (p < 0.05, impact score > 0.1; Fig. 4F). A full list of significantly enriched pathways is presented in the corresponding figures.

Discussion

AS, caused by mutations in type IV collagen, is a hereditary cause of progressive CKD12. In the present study, we leveraged AS as a representative model of CKD progression. Notably, the metabolic derangements identified in our AS model, such as impaired fatty acid oxidation and shifts in energy metabolism, were similar to those reported in other forms of CKD13. Therefore, the metabolic reprogramming observed in AS is likely not unique to this disease; instead, it reflects common pathways in CKD progression14. By using AS as a model, our findings could be compared with and were relevant to broader CKD contexts, such as diabetic and hypertensive nephropathies, strengthening the general relevance of the metabolic alterations identified. In this study, we investigated the metabolic profiles in plasma and urine from AS and WT mice through GC-MS/MS and LC-MS/MS. We performed multivariate analysis to identify differential metabolites. We selected 28 key AS-related metabolites that were different from those found in the plasma of WT mice (p < 0.05, VIP score > 1.0). Specifically, 13 metabolites were altered at 4 w, and 14 metabolites were changed at 7 w (p < 0.05, VIP score > 1.0, AUC > 0.8). In addition, 42 metabolites were different in the urine of mice with AS from those of WT mice (p < 0.05, VIP score > 1.0). Specifically, 6 metabolites were altered at 4 w, and 31 metabolites were altered at 7 w (p < 0.05, VIP score > 1.0, AUC > 0.8). Pathway analysis revealed that several metabolic pathways, including alanine, aspartate, and glutamate metabolism, arginine biosynthesis, arginine and proline metabolism, D-glutamine and D-glutamate metabolism, pyruvate metabolism, TCA cycle, and glycolysis/gluconeogenesis, in the plasma of AS mice were affected. Notably, the consistent involvement of arginine biosynthesis and alanine/aspartate/glutamate metabolism at early and later stages suggests that disruptions in amino acid metabolism are a core feature of CKD progression. Alterations in arginine metabolism may influence nitric oxide availability and impair renal vascular function, while disturbances in alanine, aspartate, and glutamate metabolism indicate the broad dysregulation of amino acid catabolism and energy production, which are changes often observed in CKD15.

We used Col4a3−/− mice as an experimental model of AS, one of the well-known genetic causes of CKD. In patients with AS, renal problems begin with asymptomatic hematuria, followed by progressive proteinuria, and eventually progress to CKD. In Col4a3−/− mice, microscopic hematuria and proteinuria occur; then, CKD develops and progresses to end-stage kidney disease; eventually, the mice die in the mean age of 14 weeks10. In previous studies, Col4a3−/− mice show focal multilaminated thickening and thinning beginning in the external capillary loops at 4 weeks and renal interstitial fibrosis and inflammation at 7 weeks of age10,11. Therefore, we considered 4 and 7 weeks of age to represent early renal disease and CKD, respectively. We checked the metabolites in the blood and urine and compared them with WT at the corresponding time periods. To date, only one metabolomic study on AS has been published to investigate metabolite distribution and altered metabolism in kidney tissues from AS by the sodium-glucose cotransporter-2 inhibitor or dipeptidyl peptidase-4 inhibitor16. For the first time, we demonstrated the dysregulation of certain metabolites in the plasma and urine of AS mice. Not surprisingly, similar metabolic disturbances have been observed in other kidney diseases. For example, metabolomic studies in diabetic kidney disease have reported an increase in saturated fatty acids such as palmitic acid and disruptions in amino acid metabolism, which are consistent with our observations17,18. Therefore, many of the metabolites we identified are likely part of common CKD pathways rather than unique to AS.

In the plasma of 4 w mice with AS, 13 metabolites were significantly altered (p < 0.05, VIP score > 1.0, AUC > 0.8) compared with those of 4 w WT mice. Among them, the level of pyruvic acid increased prominently, leading to an enrichment of the pyruvate metabolism pathway. Pyruvate metabolism is a key event in mitochondria, which are the primary intracellular organelles that generate energy, and mitochondrial dysfunction is associated with various metabolic disorders19. Previous studies showed altered mitochondrial energy metabolism in kidney diseases. Li et al.20 revealed that functional glycolysis increases, but mitochondrial respiration decreases in the kidney of subtotal nephrectomized mice. They also demonstrated an increased amount of glycolytic enzyme and the enrichment of the glycolysis/gluconeogenesis pathway20. These results were consistent with our study; indeed, the altered mitochondrial function might be implicated in kidney disease progression. The changes in glycolysis/gluconeogenesis, pyruvate metabolism, and TCA cycle observed in the plasma of 4 w AS mice in our study were related to energy metabolism, and similar changes occurred in the plasma of 7 w AS mice. These observations were consistent with the known association; that is, it shows alterations in the disturbed energy metabolism in patients with CKD and suggests that metabolites related to these changes can be biomarker candidates for diagnosis and treatment targets. The statistical significance of these metabolic pathways that were significantly altered in the plasma of 4 w AS mice was relatively reduced in the plasma of 7 w AS. Instead, additional metabolic pathways such as alanine, aspartate, and glutamate metabolism were perturbed in the plasma of 7 w AS mice. Arginine metabolism was highlighted in our findings because L-arginine serves as a precursor of nitric oxide, polyamines, and other metabolites vital to renal function21. Therefore, disruptions in arginine availability can impair vascular homeostasis and promote fibrosis in CKD21. Likewise, perturbations in glutamate metabolism can indicate impaired ammonia handling and acid–base imbalance in a diseased kidney22. Changes in metabolic pathways were more prominent in 7 w AS than in 4 w AS and might be a metabolic signature of CKD.

Changes in a few individual metabolites were noteworthy in the plasma of 7 w AS mice. Specifically, palmitic acid increased significantly in the plasma of 4 w and 7 w AS mice. Palmitic acid (16:0) is the most common saturated fatty acid found in the human body, accounting for 20–30% of total fatty acids23. The accumulation of palmitic acid and its metabolites in non-adipose tissues leads to lipotoxicity, ultimately resulting in cellular dysfunction and damage24. Palmitic acid contributes to tubulointerstitial damage by inducing inflammation, insulin resistance, and endoplasmic reticulum stress in proximal tubular cells or podocytes25,26. In this respect, the increased palmitic acid in the plasma of 4 w and 7 w is closely related to the progression of renal failure. Therefore, we could consider plasma palmitic acid as a biomarker for the early CKD. In addition to lipids, the methylated nucleoside 5-methylcytidine was altered. In our study, 5-methylcytidine levels were significantly lower in the AS mice than in the WT mice at 4 and 7 w. 5-Methylcytidine is a marker of nucleic acid methylation turnover, and increased urinary 5-methylcytidine levels are associated with the declining renal function in patients with CKD27. Its consistent reduction in the AS mice may reflect enhanced nucleic acid retention or altered epigenetic modifications because of kidney injury. However, studies have yet to clarify whether these metabolite changes are pathology drivers or consequences. For instance, palmitic acid accumulation can actively worsen kidney damage through lipotoxic effects on cells, suggesting a causal role. Conversely, changes in metabolites such as 5-methylcytidine or ketone bodies may be secondary responses to the disease, indicating cell breakdown or a compensatory shift in metabolism. As such, further studies should distinguish the cause from the effect; however, our findings implicated these metabolites early in AS and throughout its progression.

The WT and AS mice were definitively classified by the altered urinary metabolomic profile, especially by the altered metabolites and metabolic pathways in 7 w AS mice. Among them, the synthesis and degradation of ketone bodies and the metabolism of tryptophan were the most prominent. Ketones are one of the important energy sources for the kidney, which also plays an important role in ketone homeostasis through tubular reabsorption28,29,30. In particular, 3-hydroxybutyrate (beta-hydroxybutyrate), one of the common ketone bodies, elicits a renal protective effect on various kidney diseases, as indicated by its anti-inflammatory effect31. In our study, 3-hydroxybutyrate concentrations increased in the plasma of 7 w AS. The metabolic pathway involving ketone bodies was altered, and acetoacetic acid concentrations in 7 w urine increased. These observations may explain the significance of ketone body metabolism in chronic kidney disease, and it increases compensatorily in CKD. The kynurenine pathway is one of the major downstream metabolic pathways of tryptophan, accounting for approximately 95% of tryptophan metabolism32. The kynurenine pathway and its metabolites are known as key modulators of the inflammatory response. Proinflammatory cytokines such as TNF-α and IFN-γ induce indoleamine 2,3-dioxygenase, which is the rate-limiting enzyme of the kynurenic pathway; as a result, the production of kynurenine pathway metabolites such as kynurenine, kynurenic acid, and quinolinic acid, which exhibit anti-inflammatory properties, increases33,34. This enhancement of kynurenine metabolism not only alleviates inflammatory responses but also promotes the production of NAD+, which is necessary for cellular respiration35. Several studies have demonstrated that kynurenic pathway metabolites are altered in animal models and in patients with AKI and CKD36. Our data also revealed that the tryptophan metabolism pathway and the metabolites of the kynurenine pathway, such as picolinic acid levels in urine of 7 w AS, were altered. This altered metabolism might be a metabolic signature of CKD and could be used as a biomarker36. Specifically, kynurenine pathway metabolites such as kynurenine, kynurenic acid, and quinolinic acid are included in the list as protein-bound uremic toxins by the European Uremic Toxins Work Group37,38.

Our findings suggested that certain metabolites could be useful at different stages of disease management. When altered at the early stage (e.g., at 4 weeks in our AS model), metabolites may serve as diagnostic biomarkers, signaling the onset of a disease before an irreversible damage occurs. For instance, an early increase in plasma palmitic acid or pyruvate may help in the early diagnosis of AS or other CKDs if observed in patients. When progressively altered by 7 weeks (the advanced stage), metabolites can function as prognostic biomarkers, indicating disease progression or severity. For example, the extensive disruption of amino acid and ketone pathways at 7 w indicates advanced pathology; correspondingly, the metabolites in these pathways (such as 3-hydroxybutyrate or certain uremic toxins) may predict rapid progression to renal failure. In addition, metabolites intimately involved in disease mechanisms may be therapeutic targets or monitoring markers. For example, since palmitic acid likely contributes to renal injury, interventions that reduce palmitic acid levels or its toxic effects may slow disease progression17. If such therapies are used, plasma palmitic acid or related lipid metabolites can be tracked as a biomarker to monitor treatment response or efficacy.

Our study has several limitations. First, this study is a targeted metabolomic research in which we detected and quantified only predefined metabolites. Therefore, these metabolites represented AS and CKD because we did not detect the metabolites that we did not predefine. To supplement this discrepancy, we believe that additional untargeted metabolomic studies seem necessary. Second, we were unable to externally validate the metabolomic signatures detected in this study. Furthermore, because this study was conducted on an AS model, extending its application to all types of CKD is challenging. Third, relevant statistical techniques were used to discriminate the metabolites of plasma and urine between normal and diseased kidneys; however, a causal relationship was not established. Nonetheless, our study is the first to identify the extensive metabolomic profiles of plasma and urine in an AS model and to detect statistically significant metabolites that could distinguish renal failure from normal subjects. The plasma and urine metabolites identified in this study can be applied as biomarkers for diagnosing and predicting the prognosis of CKD or for estimating upper pathways that are therapeutic targets. They are vital as individual metabolites, but they can also be used collectively to constitute a metabolic signature of CKD. To further extend the results of this study, researchers should perform a metabolomic study involving plasma and urine from human patients with AS and CKD.

In conclusion, this systematic metabolomic study revealed that 28 metabolites in the plasma and 42 metabolites in the urine were associated with AS in mice at 4 w and 7 w. These altered metabolites were related to the altered mitochondrial function and kidney cell injury. Through such altered metabolomic profiles, we could obtain insight into developing biomarkers for diagnosing and predicting prognosis and therapeutic targets for CKD.

Materials and methods

Animals

Mice were maintained in a 12 h light/dark cycle and given free access to standard chow and tap water. WT and Col4a3−/− (AS) mice on a congenic 129 × 1/SvJ background were purchased from the Jackson Laboratory (Bar Harbor, ME, USA), and only male mice were used in this study. Tail tip genotyping was performed to verify the genotype of Col4a3−/− mice by using the following primers: common, 5′-CCA GGC TTA AAG GGA AAT CC-3′; WT reverse, 5′-TGC TCT CTC AAA TGC ACC AG-3′; and mutant reverse, 5′-GCT ATC AGG ACA TAG CGT TGG-3′. Then, 4 and 7 w Col4a3 knockout (AS) mice (n = 9 and 10 mice, respectively) were used. Age-matched WT mice without Col4a3 mutation were used as healthy controls (n = 14 and 10 mice, respectively). Urine samples were collected from metabolic cages 2 days prior to sacrifice, and the mice aged 4 or 7 weeks were euthanized for plasma collection and metabolomic analysis (Sup. Figure 2). Plasma was collected from the cardiac puncture and centrifuged at 845 × g for 10 min. Urine samples were collected from metabolic cages to examine the metabolites 2 days before the mice were sacrificed. Mice were anesthetized with an intraperitoneal injection of ketamine (80–100 mg/kg; Ketamine Hydrochloride, Yuhan Corp., Seoul, Korea) and xylazine (5–10 mg/kg; Rompun®, Bayer Korea, Seoul, Korea). Adequate anesthesia was confirmed by the loss of the pedal withdrawal reflex approximately 5 min after injection. Following anesthesia, kidneys were harvested under aseptic conditions. Mice were subsequently euthanized by cervical dislocation. All animal experiments were approved by the Institutional Animal Care and Use Committee of Chonnam National University Medical School and conducted in accordance with the institution guidelines for experimental animal care and use (CNU IACUC-H-2020-1). Animal experiments are reported according to the Animals in Research Reporting In Vivo Experiments 2.0 (ARRIVE 2.0) guidelines.

Chemicals and reagents

The following reagents and materials were used for the analysis: ammonium formate, trimethylamine (TEA), 205 standard solutions (68 OAs, 62 FFAs, 44 AAs, 13 KYNs, and 18 NSs), and internal standard (IS) solutions, which were purchased from Tokyo Chemical Industry, Tokyo, Japan, and from Sigma-Aldrich, St. Louis, MO, USA; acetic acid and formic acid were obtained from Wako Pure Chemical, Osaka, Japan; acetonitrile (ACN), methanol, and distilled water (DW) were sourced from J.T. Baker Inc., Phillipsburg, NJ, USA; a Spin-X centrifuge tube filter with cellulose acetate (0.45 μm pore size) was supplied by Costar, Corning, NY, USA; sodium chloride, dichloromethane (DCM), ethyl acetate (EA), and diethyl ether (DEE) were purchased from Kanto Chemical, Chuo-ku, Tokyo, Japan; tetrahydrofuran (THF) was obtained from Fisher Scientific Korea, Incheon, South Korea; sodium hydroxide (NaOH) and sulfuric acid were sourced from Daejung Reagents Chemicals, Siheung, South Korea; and N-methyl-N-(tert-butyldimethylsilyl) trifluoroacetamide (MTBSTFA) with 1% tert-butyldimethylchlorosilane (TBDMCS) and O-methoxyamine hydrochloride were obtained from Thermo Scientific, Bellefonte, PA, USA. All other reagents were of analytical grade.

Standard Preparation and calibration curve construction

Standard stock solutions of 68 OAs with 3,4-dimethoxybenzoic acid as the IS, 62 FFAs with pentadecanoic acid (PDA) as the IS, and 44 AAs with 13C2-phenylalanine as the IS were prepared at 10 mg/mL in methanol, toluene, and 0.1 M HCl, respectively. Additionally, stock solutions of 13 KYN standards were prepared at 10 mg/mL in 0.1 M HCl, and 18 NS standards were prepared at 10 mg/mL in DW. All stock solutions were stored at − 20 °C. For analysis, working solutions of OAs and FFAs were prepared at 10 and 100 ng/µL, respectively, with their corresponding ISs at 10 ng/µL, while AAs were diluted to 1, 10, and 100 ng/µL with 13C2-phenylalanine at 10 ng/µL. Working solutions of KYNs were prepared at 0.01, 0.1, and 1 ng/µL with 3,4-dimethoxybenzoic acid as the IS at 0.1 ng/µL, and NSs were prepared at 0.01, 0.1, and 1 ng/µL with 3-deazauridine as the IS at 0.5 ng/µL. Calibration curves were constructed using mixtures of standard solutions at 10 concentration points for OAs, FFAs, and AAs and at 9 concentration points for KYNs and NSs. A total of 68 OA and 62 FFA standards were used over a concentration range of 10–10,000 ng, with 3,4-dimethoxybenzoic acid (100 ng) as the IS for OAs and PDA (100 ng) as the IS for FFAs. For AAs, 44 standards were used over a range of 2.5–2,000 ng, with 13C2-phenylalanine (100 ng) as the IS. For KYNs, 13 standards were quantified over a range of 0.05–20 ng, with 3,4-dimethoxybenzoic acid (1 ng) as the IS. For NSs, 18 standards were quantified over a range of 0.01–20 ng, with 3-deazauridine (5 ng) as the IS.

Sample Preparation for OA and FA profiling analyses in plasma and urine through GC-MS/MS

Methoxime/tert-butyldimethylsilyl (MO/TBDMS) derivatization was applied to the plasma and urine samples from the AS mouse model for the simultaneous profiling of OA and FA via GC-MS/MS39,40,41,42. The plasma (40 µL) and urine (10 µL) samples were processed separately; each sample was placed in a 1.5 mL Eppendorf tube and spiked with 0.1 µg of ISs (3,4-dimethoxybenzoic acid and PDA). Then, 100 µL of ACN was added to each tube. The mixtures were vortexed and centrifuged at 12,300 × g for 3 min. The supernatants were transferred to DW in a glass vial, and the aqueous phase was adjusted to pH ≥ 12 with 5 M NaOH (100 µL). MO was derived by adding 1 mg of O-methoxyamine hydrochloride to each sample and incubating at 60 °C for 60 min. The solution was acidified to pH ≤ 2 by using 10% sulfuric acid (300 µL) and saturated with sodium chloride. Liquid–liquid extraction was conducted twice with 3 mL of DEE and 2 mL of EA. After 5 µL of TEA was added, the extracts were evaporated to dryness under a gentle nitrogen stream at 40 °C. For TBDMS derivatization, the dried extracts were reacted with 10 µL of toluene and 20 µL of MTBSTFA at 60 °C for 60 min. Lastly, 1 µL of the derivatized sample was injected into the GC-MS/MS system.

Sample Preparation for AA, KYN, and NS profiling analyses in plasma and urine via LC-MS/MS

ACN was added to the plasma (30 µL) and urine (10 µL) in a 1.5 mL Eppendorf tube and spiked with 0.2 µg of ISs (13C1-phenylalanine and 13C1-leucine for AA profiling, 3,4-dimethoxybenzoic acid for KYN profiling, and 3-deazauridine for NS profiling). They were mixed and centrifuged at 12,300 × g for 3 min. Afterward, the supernatant was transferred to a Spin-X centrifuge filter tube and centrifuged at 12,300 × g for 3 min. Lastly, 1 µL was introduced for LC-MS/MS analysis43.

GC-MS/MS

The metabolic profiles of OAs and FFAs were simultaneously analyzed using GCMS-TQ8040 (Shimadzu Corp., Kyoto, Japan) equipped with an Ultra-2 (5% phenyl–95% methylpolysiloxane bonded phase; 25 m × 0.20 mm i.d., 0.11 μm film thickness) cross-linked capillary column (Agilent Technologies, USA) and interfaced with triple quadrupole mass spectrometer. Samples were introduced at a split mode (10:1). The temperatures of ion source, interface, and injector were 230 °C, 300 °C, and 260 °C, respectively. The carrier gas was helium with a flow rate of 0.5 mL/min in a linear velocity flow mode, and the collision gas was argon. The oven temperature was programmed for the analysis of OA and FA profiles: 100 °C for 2 min and 100–300 °C at 10 °C/min with an 8 min holding time. In the multiple reaction monitoring (MRM) mode, one quantitative ion was used for quantification, and two qualitative ions were used for qualitative analysis in an electron impact ionization mode at 70 eV (Sup. Table 5). Collision energies for each compound were individually optimized within the range of 5–50 V in 5 V increments, and the final values used for each transition are summarized in Sup. Table 5.

LC-MS/MS

The metabolic profiles of AAs, KYNs, and NSs were simultaneously analyzed using a Shimadzu Nexera UPLC system (Shimadzu Corp., Kyoto, Japan) coupled with an LCMS-8050 triple quadrupole mass spectrometer (Shimadzu Corp., Kyoto, Japan). Different chromatographic columns and mobile phases were applied for each metabolite class. For AA profiling, an Intrada amino acid column (150 mm × 2 mm, 3 μm; Intakt, USA) was used. The mobile phases consisted of phase A (ACN/25 mM ammonium formate/THF/formic acid [90/160/750/3, v/v/v/v]) and phase B (ACN/100 mM ammonium formate in distilled water [20/80, v/v]). The flow rate was 0.6 mL/min, and the gradient program was as follows: 0% B (0–2.5 min), 17% B (2.5–6.5 min), 100% B (6.5–10 min, hold for 2 min), and re-equilibration at 0% B (12–15 min). For KYN profiling, separation was performed using a Zorbax Eclipse XDB-C18 column (100 mm × 4.6 mm, 3.5 μm; Agilent Technologies, USA). The mobile phases were 0.1% formic acid in distilled water (A) and 0.1% formic acid in ACN (B). The flow rate was set to 0.5 mL/min with the following gradient: 55% B (0–10 min), increased to 100% B (13–15 min), and returned to 0% B (15–20 min) for re-equilibration. For NS profiling, a Hydro-RP C18 column (150 mm × 4.6 mm, 4 μm; Phenomenex, USA) was used. The mobile phases were 10 mM ammonium acetate in distilled water with 3 mL of acetic acid (A) and methanol (B). The flow rate was 0.5 mL/min, and the gradient was set as follows: 0% B (0–1 min), 35% B (1–15 min), 100% B (15–18 min, hold for 2 min), followed by re-equilibration at 0% B (20–25 min). Nebulizing gas was set at a flow of 2.0 L/min. Drying and heating gas was set at a flow rate of 10.0 L/min. The temperatures of interface, desolvation line, and heat block were set at 300 °C, 250 °C, 400 °C, respectively. The MS system was operated in an electrospray ionization mode and a MRM mode. A collision-induced dissociation gas was used at 270 kPa. Collision energies and cone voltages for each compound were individually optimized and the final values used for each transition are summarized in Sup. Table 6.

Star symbol pattern recognition analysis

The concentrations of OAs, FFAs, AAs, KYNs, and NSs were calculated with their standard calibration curves. Metabolite levels in urine were normalized to their creatinine values. The concentration mean values of each metabolite of 4 w and 7 w AS groups were normalized to the corresponding mean values of the 4 w and 7 w WT groups, respectively. Each normalized value was plotted as a line emanating from the common central point. Star symbol patterns were drawn with normalized values in Microsoft Excel (Microsoft, Redmond, WA, USA)44,45,46.

Statistical analysis

Data from animal experiments were presented as mean ± standard deviation (SD). After concentration data were converted to log10-transformed data, statistical comparisons of univariate analysis were performed via the Wilcoxon rank-sum test to distinguish discriminatory features between the two groups. Significant comparison results were indicated as p values (< 0.05). Metabolomic data, including principal component analysis, partial least squares-discriminant analysis (PLS-DA), hierarchical clustering heatmap analysis, univariate receiver operating characteristic (ROC) curve analysis, and pathway analysis, were examined via multivariate statistical analyses in MetaboAnalyst 6.0 (https://www.metaboanalyst.ca)47,48,49.