GC-MS metabolomics on PPARα-dependent exacerbation of colitis

Abstract

Fenofibrate, a peroxisome proliferator-activated receptor α (PPARα) agonist, was found to exacerbate inflammation and tissue injury in experimental acute colitis mice. Through lipidomics analysis, bioactive sphingolipids were significantly up-regulated in the colitis group. In this study, to provide further insight into the PPARα-dependent exacerbation of colitis, gas chromatography-mass spectrometry (GC/MS) based metabolomics was employed to investigate the serum and colon of dextran sulfate sodium (DSS)-induced colitis mice treated with fenofibrate, with particular emphasis on changes in low-molecular-weight metabolites. With the aid of multivariate analysis and metabolic pathway analysis, potential metabolite markers in the amino acid metabolism, urea cycle, purine metabolism, and citrate cycle were highlighted, such as glycine, serine, threonine, malic acid, isocitric acid, uric acid, and urea. The level changes of these metabolites in either serum or colons of colitis mice were further potentiated following fenofibrate treatment. Accordingly, the expression of threonine aldolase and phosphoserine aminotransferase 1 was significantly up-regulated in colitis mice and further potentiated in fenofibrate/DSS-treated mice. It was revealed that beyond the control of lipid metabolism, PPARα also shows effects on the above pathways, resulting in enhanced protein catabolism and energy expenditure, increased bioactive sphingolipid metabolism and proinflammatory state, which were possibly related to the exacerbated colitis.

Introduction

Inflammatory bowel disease (IBD), including ulcerative colitis (UC) and Crohn’s disease (CD), is a group of chronic inflammatory conditions affecting the gastrointestinal tract.¹ Epidemiological studies revealed that the incidence rate of IBD has markedly increased worldwide, with the highest reported incidence in North America and Europe.^2,3 Previous studies suggested that IBD is associated with immunological disorders caused by genetic, environmental, and microbiological factors.^4,5 However, the accurate etiology of IBD still remains to be understood.

Peroxisome proliferator-activated receptor α (PPARα) is a nuclear hormone receptor that transcriptionally regulates lipid metabolism, atherosclerosis, and inflammation.^6–8 In a previous study using PPARα-null mice, fenofibrate, a clinically used hypolipidemic drug (a PPARα agonist), was found to exacerbate inflammation and tissue injury in experimental acute colitis mice under three distinct protocols, dextran sulfate sodium (DSS), trinitrobenzenesulfonic acid (TNBS), and Salmonella Typhi.⁹ Lipidomic analysis revealed that bioactive sphingolipids, including sphingomyelins (SM) and ceramides, were significantly increased in the colitis group compared to the control group, which was further potentiated following fenofibrate treatment. This study indicated that decreased hydrolysis and increased synthesis of SM, upregulated RIPK3-dependent necrosis, and elevated mitochondrial fatty acid β-oxidation were possibly related to the exacerbated colitis.⁹

In the above lipidomics study, lipids such as sphingolipids were expectedly highlighted since they have emerged as important signaling molecules regulating inflammatory responses; also, a primary action of PPARα activation is the transcriptional regulation of genes involved in lipid metabolism.¹⁰ Besides lipids, other low-molecular-weight metabolites, such as amino acids, bile acids, and fatty acids, were also closely correlated with IBD.^11–13 For example, both UC and CD were found to have an impact on amino acid metabolism, and levels of certain amino acids and citrate acid (TCA) cycle-related molecules were changed in colitis mice or UC patients;^14–16 malabsorbed bile acids in the colon cause diarrhea in pediatric IBD patients;¹⁷ and alterations in the expression of genes involved in intestinal fatty acid metabolism in IBD patients were identified.¹⁸ Beyond lipids, PPARα was also revealed to regulate amino acid metabolism by influencing the expression of several genes involved in trans- and deamination of amino acids and urea synthesis.¹⁹ However, little was reported concerning the comprehensive perturbation and interaction of PPARα activation and colitis on the low-molecular-weight metabolites such as amino acids. Hence, the aim of the present study was to provide further insight into effects of the PPARα agonist (fenofibrate) in colitis mice, with particular emphasis on changes in low-molecular-weight metabolites, to further decipher the PPARα-dependent exacerbation of colitis.

Metabolomics aims to identify and quantify the metabolites that serve as substrates and products in metabolic pathways in response to physiologic perturbations.^20,21 Mass spectrometry (MS) and nuclear magnetic resonance (NMR) are commonly applied in the metabolomics study.²² Among them, gas chromatography coupled to mass spectrometry (GC/MS) is widely used for profiling low-molecular-weight metabolites such as amino acids and fatty acids in various biofluids, outperforming other technical platforms due to its high sensitivity and reproducibility, as well as the availability of electron impact (EI) spectral libraries for structural identification.^23–26 In this study, GC-MS was used to characterize the metabolic profiles of the experimental colitis mice subjected to DSS and treated with fenofibrate. With the aid of multivariate analysis methods and metabolic pathway analysis, new hints on the effects of fenofibrate on colitis were explored and revealed.

Materials and methods

Chemicals and reagents

Methoxylamine hydrochloride, N-methyl-(trimethylsilyl)trifluoracetamide (MSTFA), chlorotrimethylsilane (TMCS), pyridine and ribitol (used as internal standard), and all the authentic standards were purchased from Sigma-Aldrich (St Louis, MO, US). Methanol and acetonitrile were purchased from TEDIA (Fairfield, OH, USA). Dextran sulfate sodium (DSS) was purchased from MP Biomedicals (Solon, OH, USA) (M_W = 36000–50000).

Mice and treatments

All animal treatments in this study were approved by the National Cancer Institute of National Institute of Health, USA. Wild-type and PPARα-null, 6-to-8-week-old male C57BL/6J mice were fed either standard diet or modified diet containing 0.1% fenofibrate ad libitum (Bioserv, Frenchtown, NJ) three days before the experiment and through the end of the study. Colitis was induced by the addition of drinking water containing 3.0% (wt/vol) DSS for 7 days. The presence of fenofibrate in the diet did not affect food or water intake as compared to the control groups not administered with the drug. The mice were killed 7 days after DSS administration and after 8 h of overnight fasting. Serum samples were collected by retro-orbital bleeding, and tissue samples were harvested and stored at –80 °C before analysis.

Colitis evaluation

Body weights of the mice were recorded daily. Diarrhea, rectal bleeding, and bloody stool daily were assessed and reported as a score from 0 to 4. The histological examination of the colon tissue was performed by blinded analysis and the severity of colon damage was determined by a routine hematoxylin and eosin (H&E)-stained section according to the morphological criteria described previously.^27,28

Metabolomics

For serum metabolomics analysis 30 μl serum were mixed with 4-fold acetonitrile solution containing 10 μl ribitol (0.04 mg mL⁻¹) as internal standard. The solution was vortexed for 30 s and sonicated for 10 min, and the resultant supernatant collected after being centrifuged at 10000 rpm for 10 min at 4 °C. The supernatant was dried using a vacuum concentrator (ThermoFisher, USA). For oximation, 30 μl methoxyamine hydrochloride (15 mg mL⁻¹) dissolved in pyridine were mixed with the dried sample before being heated for 1 h at 60 °C. Next, 90 μl MSTFA (containing 1% TMCS) were added for derivatization and the mixture was also heated for 1 h at 60 °C. The mixture was then centrifuged at 10 000 rpm for 10 min after being cooled and the resultant supernatant was subjected to GC/MS analysis. For colon tissue metabolomics analysis, about 20 mg accurately weighted tissues were homogenized with 50-fold methanol solution containing ribitol (0.04 mg mL⁻¹) as internal standard. The mixture was then centrifuged at 10 000 rpm for 10 min after 10 min of ultrasound. 800 μl of the supernatant was collected and then was dried. Oximation and the subsequent derivatization were performed using 30 μl methoxyamine hydrochloride (15 mg mL⁻¹) and 90 μl MSTFA (containing 1% TMCS) as described above and the resultant supernatant was subjected to GC/MS analysis.

For metabolomics discovery, the derivatized samples were analyzed on a Thermo Scientific ITQ 1100™ GC/MSⁿ (Thermo-Fisher Electron Corporation, USA) with a GL-5MS column (30 m×0.25 mm i.d.; film thickness 0.25 μm) (GL Sciences, Inc. Japan) under the following conditions: the initial oven temperature was set at 70 °C, ramped to 150 °C by 5 °C min⁻¹, then risen from 150 °C to 200 °C by 3 °C min⁻¹ and held for 2 min, finally, ramped to 280 °C by 10 °C min⁻¹ and held for 2 min. 1.0 μl of sample solution was injected with split mode (the split ratio 30:1), with helium as the carrier gas at a flow of 1.0 mL min⁻¹. The temperature of the injector, the ion source and the transfer line was 250 °C, 220 °C, 250 °C, respectively. The electron energy was 70 eV. The mass spectrometer was operated in full scan mode from 50 to 1000 m/z and the solvent delay was set at 5 min.

For metabolite identification, the GC-MS spectra were subjected to searching using the NIST (National Institute of Standards and Technology) database installed in an ITQ 1100™ GC/MSⁿ system. To confirm the identities of the putative markers, mass spectra of the metabolites were compared with those of the available authentic standards. To ensure the stability of the GC/MS system, an equal volume of each sample was pooled together to generate a pooled quality control (QC) sample. This QC sample was processed in the same way as the real samples and then run randomly through the analytical batch. Concentrations of the metabolites in the samples were determined based on standard curves using authentic standards.

Multivariate data analysis and metabolic pathway analysis

The acquired GC-MS data were first converted into the CDF format. XCMS (https://xcmsonline.scripps.edu/) was used for nonlinear alignment of the data in the time domain and automatic integration and extraction of the peak intensities, using default GC/Single Quad parameters. The XCMS output data were first normalized by dividing by the internal standard peak using the Microsoft Excel software, and then loaded to SIMCA-P software (Umetrics, Kinnelon, NJ) and transformed by mean-centering and pareto scaling, the technique that increases the importance of low abundance ions without significant amplification of noise. Statistical models including principal components analysis (PCA) and partial least squares-discriminant analysis (PLS-DA) were established to represent the major latent variables in the data matrix. The discriminating metabolites were obtained using a statistically significant threshold of variable influence on projection (VIP) values obtained from the PLS-DA model where the metabolites with VIP values higher than 1 were selected.

Metabolic pathway analysis was performed by MBRole²⁹ based on the database sources including KEGG (http://www.genome.jp/kegg/), the Human Metabolome Database (http://www.hmdb.ca/), and PubChem (http://www.ncbi.nlm.nih.gov/pccompound/) to identify the affected metabolic pathways and facilitate further biological interpretation.

RNA analysis

RNA was extracted using the TRIzol reagent (Invitrogen). Quantitative real-time PCR (qPCR) was performed using cDNA generated from 1 μg total RNA with the SuperScript II Reverse Transcriptase kit (Invitrogen). Primers were designed for qPCR using Primer Express software (Applied Biosystems, Foster City, CA) and sequences are available upon request. qPCRs were carried out using SYBR green PCR master mix (Applied Biosystems) in an ABI Prism 7900HT Sequence Detection System (Applied Biosystems). Values were quantified using the comparative threshold cycle method, and samples were normalized to GAPDH.

Data analysis

Experimental values were expressed as mean ± S.D. Statistical analysis was performed using Prism 6.0 (GraphPad Software, Inc., San Diego, CA). Repeated measures analysis of variance (ANOVA) with the post hoc test, two-way ANOVA, and the Mann–Whitney test were used to evaluate the significance of differences between the groups where necessary. A p-value below 0.05 was considered statistically significant.

Results

GC-MS metabolomics analysis of experimental colitis induced by DSS

DSS treated mice had significantly decreased body weights and other signs of intestinal injury including rectal bleeding, diarrhea, shorter colon length and inflammation and epithelial degeneration. Upon treatment with a fenofibrate-supplemented diet, wild-type (WT) mice treated with DSS displayed a more severe colitis, whereas fenofibrate-treated PPARα-null mice were less susceptible to colitis⁹ (body weight data of the mice were shown in Table S1 (ESI^†). Body weight figures, colon lengths, rectal bleeding and diarrhea scores, and images of hematoxylin and eosin-stained colon tissues of the mice can be found in our recent publication at: Am. J. Physiol. Gastrointest. Liver Physiol., 2014, 307, G564–G573). These results indicated that fenofibrate treatment exacerbates tissue injury and inflammation in colitis mice, possibly depending on PPARα activation.

In this study, GC-MS based metabolomics was performed to discover other metabolite markers than lipids to gain mechanistic insight into the exacerbation of colitis by fenofibrate treatment. In serum, the PCA scores plot showed distinct clustering for the four groups, i.e., control-treated WT mice, DSS-treated WT mice, fenofibrate/DSS-treated WT and PPARα-null mice, with the cumulative R²X 0.659 and Q² 0.543 for this model (Fig. 1A). Similarly, the PLS-DA scores plot exhibited clear separation among the four groups with the cumulative R²X 0.771, R²Y 0.841, and Q² 0.442 (Fig. 1B). In colon, both PCA(R²X 0.602 and Q² 0.196) (Fig. 1C) and PLS-DA scores plot (R²X 0.818, R²Y 0.99, and Q² 0.763) (Fig. 1D) showed distinct clustering and clear separation for the four groups, too. These results indicated that from the view of metabolomics, the phenotypes of the four groups could be clearly distinguished, in accordance with the previous pathological findings.⁹

Fig. 1.

(A) PCA scores plot (with the cumulative R²X 0.659 and Q² 0.543) and (B) PLS-DA scores plot (with the cumulative R²X 0.771, R²Y 0.841, and Q² 0.442) of serum samples from all the four examined groups, i.e. control (WT-CTR/water), colitis (WT-CTR/DSS), fenofibrate-treated colitis (WT-FF/DSS on WT mice and KO-FF/DSS on PPARα-null (KO) mice). (C) PCA scores plot (with the cumulative R²X 0.602 and Q² 0.196) and (D) PLS-DA scores plot (with the cumulative R²X 0.818, R²Y 0.99, and Q² 0.763) of colon samples for the four examined groups.

Identification and quantitation of the potential metabolite markers

The common way to screen the potential markers related to diseases is to compare the control and disease groups and then screen the differentiating metabolites. Based on the PLS-DA models (for serum and colon, respectively) discriminating the control and colitis groups, a bunch of significantly altered ions were revealed (Fig. S1, ESI^†), and were later identified as metabolites listed in Table 1. It could be observed that the altered metabolites in serum and colon were quite different.

Table 1.

Identified metabolite markers in the serum (colon) of colitis mice^a

No.	Retention time (min)	m/z	ID	Identification results	B vs. A	C vs. B	D vs. C	Pathway
1	17.42	147.1116	C00149	Malic acid	↑*	↑	↓	Citrate cycle
2	10.80	188.9971	C00086	Urea	↑**(↑)	↑(↑)	↓(↓)	Purine metabolism
3	8.42	146.9984	C00209	Oxalic acid	↓	↓	↑	Glyoxylate and dicarboxylate metabolism
4	30.05	147.1917	C00031	Glucose	↓*	↓**	↑	Glycolysis/gluconeogenesis
5	27.27	273.1575	C00311	Isocitric acid	↑	↓	↑	Citrate cycle
6	35.81	305.1802	C00137	Myoinositol	↑(↓*)	↓(↓)	↑(↑)	Inositol phosphate metabolism
7	7.87	147.1257	C00037	Glycine	(↑*)	(↑)	(↓)	Glycine, serine and threonine metabolism
8	14.12	204.1879	C00065	Serine	(↑)	(↑)	(↓)	Glycine, serine and threonine metabolism
9	14.82	218.1231	C00188	Threonine	(↑)	(↑)	(↓)	Glycine, serine and threonine metabolism
10	8.88	147.1195	C00246	Butanoic acid	(↑**)	(↑*)	(↓)	Butanoate metabolism
11	35.92	147.1106	C00366	Uric acid	(↑)	(↑**)	(↓)	Purine metabolism
12	26.10	265.2826	C15587	Purine	(↓**)	(↓)	(↑)	Purine metabolism

^aA, B, C, and D represent control mice, DSS-treated WT mice, fenofibrate/DSS-treated WT and fenofibrate/DSS-treatedPpara-null mice, respectively; ↑ or ↓ represents the up- or down-regulation of the metabolites in serum; ↑ or ↓ in brackets represent the up-or downregulation of the metabolite levels in colon.

^*p < 0.05,

^**p < 0.01.

In the QC samples, for all the metabolites listed in Table 1, the relative standard deviation (RSD) ranged from 0.06% to 0.42% for the retention times and ranged from 2.39% to 5.21% for the peak areas, which demonstrated the robustness of the method. Relative levels of the potential metabolite markers in serum and colon of the four groups were shown in Fig. 2 (fold changes compared to the control group). In serum of DSS-treated WT mice, malic acid and urea levels were significantly increased but oxalic acid and glucose levels were decreased compared to the control mice. In colon, compared to the control mice, amino acids including glycine, serine, and threonine, as well as butanoic acid, uric acid, and urea were increased while purine and myoinositol were decreased in colitis mice. All the above trends of the metabolites in serum or colon were further potentiated in fenofibrate/DSS-treated WT mice other than in PPARα-null mice. The level fluctuations of these potential metabolite markers, especially those apparently correlated with fenofibrate treatment, may provide clues for investigating the exacerbation of fenofibrate on experimental colitis.

Fig. 2.

Target quantitation of significantly changed metabolites in serum and colon based on standard curves using authentic standards. (A) Serum metabolite levels in the examined groups. (B) Colon metabolite levels in the examined groups. Data were expressed as mean ± SD. *p < 0.05 and **p < 0.01.

Metabolic pathway analysis

A functional enrichment analysis facilitating further biological interpretation was subsequently performed using MBRole to reveal the most relevant pathways. It was shown in Table S2 (ESI^†) that the revealed metabolite markers were mainly involved in the following pathways: purine metabolism (uric acid, glycine, and urea), glyoxylate and dicarboxylate metabolism (malic acid, isocitric acid, and oxalic acid), aminoacyl-tRNA biosynthesis (threonine, serine, and glycine), glycine, serine and threonine metabolism (glycine, threonine, and serine), galactose metabolism (glucose, myoinositol), and citrate cycle (isocitric acid, malic acid).

Analysis of gene expression

According to Table 1 and Table S2 (ESI^†), some metabolites together with the involved pathways were revealed to be probably associated with colitis and the PPARα-dependent exacerbation. Afterwards, expression of some key genes in these pathways was evaluated in colon samples of various groups using qPCR, among which threonine aldolase (THA) and phosphoserine aminotransferase 1 (PSAT1) were significantly up-regulated in colitis mice compared to control mice, and further potentiated in fenofibrate/DSS-treated WT mice (Fig. 3). These observations supported our metabolomics findings as will be discussed below.^30,31

Fig. 3.

Quantitative real time PCR analysis of colon THA and PSAT1 mRNA expression in control (WT-CTR/water), colitis (WT-CTR/DSS), fenofibrate-treated colitis (WT-FF/DSS on WT mice and KO-FF/DSS on PPARα-null (KO) mice). Data were expressed as mean ± SD. * indicates p < 0.05 and ** indicates p < 0.01. Abbreviations: THA, threonine aldolase; PSAT1, phosphoserine aminotransferase 1.

Discussion

In this study, a GC/MS-based metabolomics approach was performed on serum and colon tissues of DSS-induced colitis mice treated with fenofibrate. Levels of several metabolite markers were altered in control, colitis, and fenofibrate/DSS-treated WT or PPARα-null mice. Among them, malic acid, glycine, serine, threonine, uric acid, and urea were increased in either serum or colons of colitis mice, whereas oxalic acid, glucose, purine, and myoinositol were decreased. Upon treatment with a fenofibrate-supplemented diet, these trends were potentiated in WT mice other than PPARα-null mice. These metabolites with further potentiated levels after fenofibrate treatment (but not in KO-FF/DSS group) may possibly be related to PPARα activation, and hence could provide clues for investigating the exacerbation of fenofibrate on experimental colitis. Though some metabolites, such as serum urea, had similar levels in WT-CTR/DSS and KO-FF/DSS groups, these two groups could not be directly compared since their biological backgrounds were totally different. WT-CTR/DSS mice were in the basal physiological status whereas the KO-FF/DSS group mice were knockout mice. The similar levels of some metabolites in these two groups may partly because both groups had colitis. Through metabolic pathway analysis, glycine, serine and threonine metabolism, purine metabolism, citrate cycle metabolism, etc. were possibly related to the etiology of colitis as well as the exacerbation by fenofibrate.

It was reported that an active inflammatory state leads to highly elevated energy expenditure due to enhanced protein catabolism.^32,33 DSS-induced colitis mice suffered from diarrhea and muscle atrophy associated with enhanced protein catabolism, which may have caused the increases in amino acid levels.¹⁵ At the same time, PPARα activation could also increase protein degradation, possibly resulting from its lipid-lowering action.¹⁹ It was revealed that WY 14 643, another PPARα agonist, raised total amino acid concentration (38%) in fat-fed rats, largely explained by glycine, serine and threonine increases.³⁴ In our study, in compatible with Ericsson’s work,³⁴ with the dual regulation of colitis and PPARα activation, up-regulation of amino acids including serine, glycine and threonine was observed in colon of colitis mice, and was further potentiated after fenofibrate treatment.

Serine, glycine and threonine are three closely related amino acids that share common biochemical pathways. Threonine can be converted to glycine via THA. In this study, colon gene expression of THA was found to be significantly up-regulated in colitis mice and further potentiated in fenofibrate/DSS-treated WT mice, consistent with an increased colon glycine level in colitis mice and its further potentiation after fenofibrate treatment. This also indicated a possible correlation between PPARα activation and THA expression, which still awaits further study. Serine can be derived from four possible sources, and the pathway of serine utilization is a major source of one-carbon groups, in which glycine and 5,10-methylenetetrahydrofolate (5,10-MTHF) are generated.³⁵ In their work, Ericsson et al.³⁴ attributed the WY 14643-induced increase of serine and glycine to boosted de novo synthesis of serine through the phosphorylated intermediate pathway (where glycine was generated as a byproduct), which causes that genes including phosphoserine aminotransferase 1 (PSAT 1) were markedly up-regulated in liver and kidneys of the rats.^19,34 In this study, colon PSAT1 expression was significantly increased in the fenofibrate-treated colitis group, consistent with the up-regulated serine and glycine levels for this group, and was in accordance with the previous literature.³⁴ Furthermore, serine is an important precursor for the synthesis of sphingolipids. Our previous lipidomic analysis revealed significantly increased bioactive sphingolipids in the colitis mice.⁹ Therefore, the up-regulation of serine in colitis mice and its further potentiation after fenofibrate treatment may partly lead to the up-regulation of sphingolipids and are quite supportive to our previous lipidomics findings.

Deamination of amino acids in the liver releases ammonia, which is efficiently converted to nontoxic urea by the urea cycle.³⁶ In this study, increased urea levels were observed in both serum and colon of colitis mice, indicating an increased urea synthesis in IBD.³⁷ It was previously demonstrated that patients with active IBD often present with negative nitrogen balance because of high urinary nitrogen excretion,³⁸ mostly urea-nitrogen,³⁹ resulting in an up-regulated urea synthesis. After fenofibrate treatment, the colitis mice had even higher urea level in serum/colon. This may be because PPARα mediated liver growth further increases the urea production capacity.¹⁹ At the same time, the colon uric acid level was increased in colitis mice. Uric acid is generated from the metabolism of purines,⁴⁰ and its positive correlation with inflammation was found in a small clinical series of heart failure patients,^41,42 and a possible association between the uric acid level and markers of inflammation in a population-based sample of participants was revealed, suggesting that uric acid might contribute to the proinflammatory state.⁴³

In this study, mice treated with DSS showed a decrease in body weight, and were subjected to rectal bleeding and diarrhea compared to controls. These observations suggest deficiencies of macronutrients, which is a common symptom observed in IBD.⁴⁴ Glucose serves as an energy source for normal intestinal mucosa.⁴⁵ In this study, a marked drop in serum glucose concentration mirrored the high energy deficit of the body and decreased energy availability due to decreased nutrient absorption through the impaired intestine,⁴⁶ consistent with previous investigation.⁴⁷ At the same time, increased serum levels of citrate cycle intermediates (such as malic acid) indicated the high demand and rapid utilization of energy, and was synchronized with the boosted glycolysis to facilitate the increased rate of the citrate cycle.⁴⁶

In addition, according to the correlation of serum and colon levels for the metabolite markers (Fig. S2, ESI^†), some metabolites, such as glycine and isocitric acid, seemed to be positively correlated in serum and colon in the colitis mice; after fenofibrate-treatment, the correlation of serum and colon levels for some metabolites such as malic acid and urea was obviously increased, showing the further disturbed metabolism pathways as discussed above. The perturbed metabolic pathways revealed in this study were summarized in Fig. 4.

Fig. 4.

Schematic representation of the metabolic pathways. Metabolites in bold black were revealed as potential metabolite markers in this study, and the column values in histograms were expressed as mean ± SD of their relative levels. Sphingomyelins (SM) and ceramides in sphingolipid metabolism were reported in our previous study. Metabolites in italic were not detected in this study.

Conclusion

In this study, GC-MS metabolomics was used to provide further insight into the PPARα-dependent exacerbation of colitis, with particular emphasis on changes in low-molecular-weight metabolites. With the aid of multivariate analysis and metabolic pathway analysis, potential metabolite markers in the amino acid metabolism, urea cycle, purine metabolism, and citrate cycle were highlighted. It was revealed that beyond the control of lipid metabolism, PPARα also shows effects on the above pathways, resulting in enhanced protein catabolism, elevated energy expenditure, increased bioactive sphingolipid metabolism and proinflammatory state, which were possibly related to the exacerbated colitis.

Abbreviations

DSS

Dextran sulfate sodium

IBD

Inflammatory bowel disease

PLS-DA

Partial least squares-discriminant analysis

VIP

Variable influence on projection

PCA

Principal components analysis

PPARα

Peroxisome proliferator-activated receptor α

SM

Sphingomyelin

TNBS

Trinitrobenzene sulfonic acid

MSTFA

N-Methyl-(trimethylsilyl)trifluoracetamide

TMCS

Chlorotrimethylsilane

GC/MS

Gas chromatography coupled to mass spectrometry

THA

Threonine aldolase

PSAT1

Phosphoserine aminotransferase 1