Metabonomic Variations Associated with AOM-Induced Precancerous

Apr 20, 2012 - (3, 29)Comprehensive analysis of the precancerous lesions will undoubtedly delineate characteristic metabolic alterations during the ...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/jpr

Metabonomic Variations Associated with AOM-Induced Precancerous Colorectal Lesions and Resveratrol Treatment Wen Liao,† Hai Wei,† Xiaoyan Wang,∥ Yunping Qiu,‡ Xiaojun Gou,† Xiaolei Zhang,† Mingmei Zhou,† Jianbing Wu,† Tao Wu,† Fang Kou,† Yongyu Zhang,† Zhaoxiang Bian,§ Guoxiang Xie,*,‡ and Wei Jia*,‡,† †

Center for Chinese Medical Therapy and Systems Biology, Shanghai University of Traditional Chinese Medicine, Shanghai 201204, China ‡ Department of Nutrition, University of North Carolina at Greensboro, North Carolina Research Campus, Kannapolis, North Carolina 28081, United States ∥ Ministry of Education Key Laboratory of Systems Biomedicine, Shanghai Center for Systems Biomedicine, Shanghai Jiao Tong University, Shanghai 200240, China § School of Chinese Medicine, Hong Kong Baptist University, Hong Kong SAR, China S Supporting Information *

ABSTRACT: Resveratrol (Res), 3,5,4′-trihydroxy-trans-stilbene, is an antioxidant found in the skin of red grapes and in several other plants. This phenolic compound has been recently reported to possess cancer chemopreventive activity that inhibits the process of carcinogenesis. However, the mechanisms underlying its anticancer effects remain largely unresolved. In this study, we investigated the chemoprotective effects of dietary Res in an azoxymethane (AOM) induced precancerous colorectal lesion model in male Wistar rats. The metabolic alterations in urine, sera, and colonic tissues of experimental rats perturbed by AOM intervention as well as the Res treatment were measured by a gas chromatography time-of-flight mass spectrometry (GC-TOFMS) analysis. Significant alterations of metabolites were observed in AOM group in urine, sera, and colonic tissues, which were attenuated by Res treatment and concurrent with the histopathological improvement with significantly decreased aberrant crypt foci (ACF) incidence. Representative metabolites include depleted glucose, β-hydroxybutyrate (ketone body), hypoxanthine, and elevated branched chain amino acids (isoleucine and valine) and tryptophan in colonic tissue, as well as elevated serum aminooxyacetate and urinary 4-hydroxyphenylacetate and xanthurenate. These metabolic changes suggest that the preventive effect of Res is associated with attenuation of impaired glucose and lipid metabolism and elevated protein breakdown in colonic tissues from AOM-exposed rats. It also appears that Res induced significant metabolic alterations independent of the AOM-induced metabolic changes. The significantly altered metabolites identified in Res-AOM group relative to AOM group include arachidonate, linoleate, glutamate, docosahexaenoate, palmitelaidate, 2-aminobutyrate, pyroglutamate, and threonate, all of which are involved in inflammation and oxidation processes. This suggests that Res exerts the chemopreventive effects on ACF formation by anti-inflammatory and antioxidant mechanisms in addition to amelioration of AOM-induced mitochondrial disruption. KEYWORDS: colorectal cancer, azoxymethane, AOM, resveratrol, gas chromatography time-of-flight mass spectrometry, GC-TOFMS, metabonomics



at approximately 8% for stage IV patients.4 Such a low survival rate is primarily due to the fact that a significant proportion of cancers initially are asymptomatic lesions and are not diagnosed or treated until they reach an advanced stage. Early detection and early intervention options are of great importance to reduce the cancer incidence and improve patient outcomes. Colorectal carcinogenesis is a complex multistep process that involves changes in both histomorphological appearances of the

INTRODUCTION Colorectal cancer (CRC) is the third leading cause of cancer deaths in the United States and an estimated 141 210 new cases and 49 380 deaths from CRC were reported in 2011.1 With the changes of life style and nutritional habits, many Asian countries have experienced an increase of 2−4 times in the incidence of CRC during the past few decades.2 In Shanghai, China, the incidence rates of CRC increased steadily from 6.09 and 5.70 to 14.70 and 14.35 per 100 000 in males and females, respectively, during 1973−2005.3 CRC patients at late stages suffer from a poor prognostic outcome with a low survival rate © 2012 American Chemical Society

Received: March 22, 2012 Published: April 20, 2012 3436

dx.doi.org/10.1021/pr300284h | J. Proteome Res. 2012, 11, 3436−3448

Journal of Proteome Research

Article

colonic mucosa and at molecular level.5 A series of morphologically identifiable stages include: initial localized colonic epithelial hyperplasia, followed by formation of adenomas that progressively enlarge and ultimately develop into invasive cancers, that is, the adenoma−carcinoma sequence.5 The progression of CRC is usually classified by the TNM stages: Stage 0 (cancer is found only in the innermost lining of the colon and/or rectum), Stage I (cancer has grown through the innermost lining but has not spread beyond the colon wall or rectum), Stage II (cancer has spread to deeper layers of the wall of the colon or rectum, but not the lymph nodes), Stage III (cancer has metastasized to nearby lymph nodes but not to other parts of the body), and Stage IV (cancer has metastasized to other parts of the body, such as the liver and lungs) or grades.6 Early intervention of colorectal premalignant lesion with natural products, such as resveratrol (Res), is now one of the most promising approaches to prevent colorectal cancer.7,8 Res (3,5,4′-trihydroxy-trans-stilbene) is a polyphenolic phytoalexin, found in 72 plant species distributed in 31 genera and 12 families.9 It has been demonstrated to attenuate colitis and colon tumor multiplicity in mice exposed to azoxymethane (AOM) or dextran sodium sulfate (DSS),10,11 and is being evaluated as a potential cancer chemopreventive agent in humans.12 Extensive studies have indicated the inhibitory effect of Res on proinflammatory gene expression including toll-like receptor 413 and cyclooxygenase 2,14 and decreased proliferation of colonocytes.15 However, mechanistic studies aiming to understand the metabolic alteration in cancer cells by Res leading to cancer prevention have not been reported, to date. Recently, metabolic profiling approach has been introduced into cancer research and shown great potential for prognostic or predicative interpretation of cancer status. Metabonomic studies on urine, serum, and tissue of CRC patients have been widely launched and made great progress,16,17 which provided the possibility of early diagnosis of CRC cancer. Here we report a metabonomic study on a classical AOM-induced precancerous colorectal lesion in Wistar rats to mimic human subjects with precancerous colorectal lesions.18,19 In this study, we used a gas chromatography time-of-flight mass spectrometry (GCTOFMS) based metabonomic approach coupled with histopathology and biochemical analysis to obtain tissue, serum, and urine metabolite markers of AOM-induced precancerous colorectal lesions. We also used Res as a dietary intervention in the same animal model to gain mechanistic insights into its chemopreventive effects against colorectal carcinogenesis.



97.3%) to homogeneity at a ratio of 2 g Res/kg diet. Diets were prepared under yellow light and stored at 20 °C to minimize degradation of Res. Animal Treatment and Sampling

The animal study was conducted in accordance with Chinese national legislation and local guidelines, and performed at the Centre of Laboratory Animals, Shanghai University of Traditional Chinese Medicine, China. A total of 45 male Wistar rats of body weight 120−150 g were obtained from Shanghai Laboratory Animal Co. Ltd. (SLAC, China). Each rat was housed in one metabolic cage and kept in a barrier system with regulated temperature (23−24 °C) and humidity (60 ± 10%), on a 12/12 h light/dark cycle (lights on at 8:00 a.m.). All experimental rats were fed with rat chow at a dose of 30 g/day and received water ad libitum throughout the experiment; the average weight gain and food consumption were monitored. After one week acclimatization, the 45 rats were randomly divided into three groups: a AOM group (n = 15), injected subcutaneously with AOM (prepared in 0.9% saline before use and adjusted to pH 6.4, 15 mg/kg bodyweight) once a week for 2 weeks; the control group (n = 15), injected with the same volume of saline as processed in AOM group; Res-AOM group (n = 15), injected subcutaneously with AOM once a week for 2 weeks as processed in AOM group and receiving experimental diet containing Res from week 5 to week 12. The 7-week treatment period (from week 5 to week 12) was selected based on the published studies20 as well as our preliminary data demonstrating that the incidence of aberrant crypt foci (ACF) was significantly reduced as a result of Res exposure within this time frame. At the end of 12th week, urine and sera samples were collected and stored at −80 °C pending GC-MS analysis. All the rats were anesthetized and the tissues were dissected and weighed. Colonic tissue samples were used for ACF counting and metabonomic analysis. Sera samples were analyzed using a Hitachi 7600 automated analyzer to determine total cholesterol, low density lipoprotein (LDL), high density lipoprotein (HDL), and triglyceride (TG). Tissue Sample Preparation for ACF Counting

The entire colon (from cecum to anus) was removed and washed thoroughly with 0.9% saline, cut longitudinally, laid flat on a polystyrene board, and fixed with 10% buffered formaldehyde solution overnight. The colon was then stained with 0.2% methylene blue for 3−5 min in saline in order to identify ACF, which can be visualized on a background of normal crypts since aberrant crypts have larger, often elongated openings and a thicker lining of epithelial cells compared with normal crypts. For topographic assessment of the colon, mucosal ACF was counted using a light microscope.21 The ACF were classified as small (1−3 crypts/focus) and large (>3 crypts/focus) by the number of crypts per foci.

EXPERIMENTAL SECTION

Chemicals

HPLC grade methanol was purchased from Merck Chemicals (Darmstadt, Germany). Pyridine was analytical grade and purchased from China National Pharmaceutical Group Corporation (Shanghai, China). L-2-Chlorophenylalanine was purchased from Intechem Tech. Co. Ltd. (Shanghai, China). Azoxymethane, BSTFA (1% TMCS), heptadecanoic acid, methoxyamine, and formaldehyde solution were purchased from Sigma Aldrich (St. Louis, MO). All aqueous solutions were prepared with ultrapure water produced by a Milli-Q system (18.2 MΩ, Millipore, Bedford, MA).

Sample Preparation for GC-TOFMS Analysis

Tissue samples were prepared according to our previous published method with minor modification.22 Briefly, each tissue sample (50 mg) was pulverized after being frozen in liquid nitrogen with the addition of 250 μL of mixed solvent (chloroform/methanol/water = 1:2.5:1, v/v/v).The lysate was ultrasonicated for 1 min and stored at −20 °C for 20 min, then centrifuged at 15 700g (4 °C) for 10 min. A total of 150 μL of aqueous supernatant was transferred to a GC vial containing two internal standards, L-2-chlorophenylalanine (10 μL, 0.3 mg/mL) and heptadecanoic acid (10 μL, 1.0 mg/mL). The deposit was

Diets Design

Experimental diets were prepared by thoroughly mixing basal diet with Res (Shanghai Ronge Pharmaceutical Co., Ltd., 3437

dx.doi.org/10.1021/pr300284h | J. Proteome Res. 2012, 11, 3436−3448

Journal of Proteome Research

Article

metabolic profiles of AOM induced precancerous lesion rats and other two groups could be obtained. In parallel, the metabolites identified by the PLS-DA model were validated at a univariate level using the Wilcoxon−Mann−Whitney test (p < 0.05). The resultant p-values for all metabolites were subsequently adjusted to account for multiple testing. The false discovery rate (FDR) method of Pike24 was used to perform the adjustment. The corresponding fold change shows how these selected differential metabolites varied between groups. Additionally, compound identification was performed by comparing the mass fragments with NIST 05 Standard mass spectral databases in NIST MS search 2.0 (NIST, Gaithersburg, MD) software with a similarity of more than 70% and finally verified by available reference standards.

rehomogenized with a T10 basic homogenizer (IKA, Staufen, Germany) for 30 s at 0 °C after adding 250 μL of methanol. After a second centrifugation, another 150 μL aliquot of supernatant was added to the mixture in the GC vial and vacuum-dried. The residue was derivatized using a two-step procedure. First, 80 μL of methoxyamine (15 mg/mL in pyridine) was added to the vial and kept at 30 °C for 90 min, followed by 80 μL of BSTFA (1% TMCS) at 70 °C for 60 min. Serum samples were treated with chemical derivatization strictly following our previously published procedure.17 Urine samples were prepared with minor modification to our previously published method of urine samples.23 Briefly, each 200 μL aliquot of urine samples was added into a 1.5 mL tube and centrifuged at 15 700 g (4 °C) for 10 min. A 50 μL supernatant was transferred to a new tube followed by the addition of urease (30 U) and incubation at 37 °C for 15 min. The metabolite extraction procedure was carried out after adding two internal standard solutions (10 μL of L-2chlorophenylalanine in water, 0.3 mg/mL; 10 μL of heptadecanoic acid in methanol, 1 mg/mL) and 170 μL of methanol. After vortexing for 30 s, the mixture was centrifuged at 15 700g (4 °C) for 5 min. A 200 μL supernatant was transferred to a 500 μL glass tube and dried under vacuum. The derivatization process was the same as serum sample preparation.17



RESULTS

AOM-Induced Precancerous Colorectal Lesions

As shown in Supporting Information Figure S1, the body weights of rats in AOM group were significantly increased relative to the control group, while in Res-AOM group, the body weights were decreased at the sixth week post AOM injection (Supporting Information Figure S1A) without significant difference in food intake (Supporting Information Figure S1B). The serum cholesterol levels together with the other three serum lipid markers, LDL, HDL, and TG, in AOM group were lower than those in controls (Figure 1), though the rats in AOM group showed a remarkable increase in body weight. However, these changes in total cholesterol, LDL, and HDL were attenuated in Res-AOM group (Figure 1). Typical histological ACF lesions were found in AOM group (Figure 2A,B), which confirms that the precancerous colorectal lesion rat model was successfully produced in the current experiment. A sharply increased ACF number was observed in the descending (Figure 2C) and proximal colon (Figure 2D) of AOM group, as compared with the controls (the number of ACF is 0) at the end of 12th week. The number of ACF in descending colon was greater than that in proximal colon as shown in Figure 2. Moreover, the number of ACF in Res-AOM group was significantly lower than in the AOM group (Figure 2C,D).

GC-TOFMS Spectral Acquisition

Each 1 μL aliquot of the derivatized solution was splitless injected into an Agilent 6890N gas chromatography coupled with a Pegasus HT time-of-flight mass spectrometer (Leco Corporation, St. Joseph, MI). The samples of AOM-group, Res-AOM group, and control group were run alternately, to minimize systematic analytical deviations. Separation was achieved on a DB-5 ms capillary column (30 m × 250 μm i.d., 0.25 μm film thickness; (5%-phenyl)-methylpolysiloxane bonded and cross-linked; Agilent J&W Scientific, Folsom, CA), with helium as the carrier gas at a constant flow rate of 1.0 mL/min. The temperature of injection, transfer interface, and ion source was set to 270, 260, and 200 °C, respectively. The GC temperature programming for the analysis of serum and tissue samples is the same as in our previous report.17 For urine samples, the instrument conditions, chromatographic and mass parameters, and GC temperature programming were referred to our previous report on urine metabolic profiling.23

Metabolite Variations in AOM Treated Rats

Among a total of 403, 451, and 455 metabolite features obtained from the GC-TOFMS spectra of tissue, sera, and urine samples, 85, 92, and 125 metabolites were identified with NIST 05 standard mass spectral databases with a similarity >70% and 59, 70, and 89 were further verified by available reference standards, respectively. Supporting Information Figure S2 illustrates the scores plots of PLS-DA model of the subjects from control group, AOM group, and Res-AOM group. ResAOM group and AOM group were all clearly separated from the controls. We selected the differentially expressed tissue, serum, and urine metabolites in the AOM-group relative to control group based on the VIP values (VIP > 1) by three-component PLSDA models (R2X = 0.384, R2Y = 0.997, Q2(cum) = 0.828; R2X = 0.403, R2Y = 0.932, Q2(cum) = 0.671; R2X = 0.474, R2Y = 0.976, Q2(cum) = 0.633), respectively (Figure 3A−C) constructed with the identified metabolites. The PLS-DA scores plot constructed with all the GC-TOFMS spectral features from colonic tissues, sera, and urine of control group, AOM group, and Res-AOM group were provided in Supporting Information

Data Analysis

The acquired MS files from GC-TOFMS analysis were performed according to our previously published method with minor modification.17,23 Briefly, the MS files were exported in NetCDF format by ChromaTOF software (v3.30, Leco Co., CA). CDF files were extracted using custom scripts in the MATLAB 7.0 (The MathWorks, Inc.) for data pretreatment procedures such as baseline correction, denoising, smoothing, alignment, internal standard exclusion, and normalization. The resulting data was analyzed in the SIMCA-P 11.5 Software package (Umetrics, Umeå, Sweden). Multivariate statistical analysis, partial least-squares-discriminant analysis (PLS-DA), was performed. In this study, the default 7-round cross-validation in SIMCA-p software package was applied with one-seventh of the samples being excluded from the mathematical model in each round, in order to guard against overfitting. On the basis of a variable importance in the projection (VIP) threshold of 1 from the PLS-DA model, a number of metabolites responsible for the differentiation in the 3438

dx.doi.org/10.1021/pr300284h | J. Proteome Res. 2012, 11, 3436−3448

Journal of Proteome Research

Article

Figure 1. Quantitative measurement of rat serum lipid levels of total cholesterol, low-density lipoprotein (LDL), triglyeride (TG), and high density lipoprotein (HDL) in the control group, AOM group, and Res-AOM group (mM, mean ± SE). *p < 0.05 vs control group, **p < 0.01 vs control group.

Figure S3. Univariate statistical analysis, Wilcoxon−Mann− Whitney test, was performed on metabolites identified from GC-TOFMS analysis of tissues, sera, and urine samples to evaluate their significance. A total of 20, 11, and 17 differentially expressed metabolites in tissue, serum, and urine were selected and 19, 9, and 14 of them were verified by reference standards, respectively, with a p-value less than 0.05 (AOM group vs controls, Tables 1−3). All these metabolites remained statistically significant after multiple testing. The altered metabolites included decreased mannose, glucose, erythrose, hypoxanthine, β-hydroxybutyrate, cystine, pantothenate, and γ-aminobutyrate and elevated isoleucine, valine, ornithine, tryptophan, tyrosine, phenylalanine, and glycerol phosphate in the tissue; increased tryptophan, aminooxyacetate, and 4-hydroxy-L-proline and decreased succinate, fumarate, isoleucine, valine, docosahexaenoate, and phenylalanine in serum; and increased 3,4dihydroxybutanoate, 3-methyladipate, pseudo uridine, serine, putrescine, xanthurenate, adipate, 4-hydroxyphenylacetate and decreased alpha-hydroxyglutarate, hippurate, and 3-hydroxybenzoate in urine of AOM group rats as compared with controls.

group based on the VIP values (VIP > 1) by three-component PLS-DA models (R2X = 0.346, R2Y = 0.990, Q2(cum) = 0.676; R2X = 0.374, R2Y = 0.943, Q2(cum) = 0.513; R2X = 0.514, R2Y = 0.990, Q2(cum) = 0.913), respectively (Figure 3D,E and Supporting Information Figure S4). A list of differential metabolites (Tables 1−3) including arachidonate, linoleate, glutamate, docosahexaenoate, palmitelaidate, 2-aminobutyrate, pyroglutamate, and threonate was identified and most of them are different from those differential metabolites induced by AOM intervention.



DISCUSSION

A great number of studies in recent years conclude that malignant tumors were associated with altered metabolic pattern involving cell signaling pathway changes and the dysregulation of glycolysis and respiration, known as the Warburg effect.3,17,25,26 Previous metabonomic studies revealed systemic metabolic variation associated with gastrointestinal carcinoma including gastric cardia and colorectal cancer.3,17,27 Abnormal metabolic variation existed throughout the different pathological stages of colorectal cancer.3,17,28 The precancerous state, transitioning from normal cell to malignant cell, involves characteristic metabolic transformation which impacts cellular and global metabolic phenotype.3,29 Comprehensive analysis of the precancerous lesions will undoubtedly delineate characteristic metabolic alterations during the carcinogenesis, providing mechanistic insights into the effects of cancer treatment strategies. ACF is widely accepted as precancerous lesions in the colon of animals and humans.30 Although animals with a large number of ACF in their colon were still apparently healthy, the physiological and biochemical indices became significantly different post AOM exposure. A significant body weight increase was found in AOM group relative to the control group (without significant difference in food intake) at 6 week

Preventive Effect of Res on AOM Induced Precancerous Colorectal Cancer

All the rats in Res-AOM group showed significantly less number of ACF compared to AOM group (Figure 2C,D), demonstrating the preventive effects of Res on the AOMinduced precancerous colorectal lesions. The variations of differential metabolites listed in Tables 1−3 upon AOM intervention were investigated in Res-AOM group, and as a result, not all the metabolites were attenuated or normalized after Res treatment. The heatmap in Figure 4 showed that significantly altered metabolites in AOM group were attenuated in ResAOM group (fold changes nonsignificant are in black). We further selected the differentially expressed tissue, serum, and urine metabolites in the Res-AOM group relative to AOM3439

dx.doi.org/10.1021/pr300284h | J. Proteome Res. 2012, 11, 3436−3448

Journal of Proteome Research

Article

Figure 2. Topographic assessment of the normal tissue (A) and AOM-induced colorectal precancerous cancer (B), and the number of ACF in descending (C) and proximal (D) colon. **p < 0.01 vs AOM group.

amino acids, carbohydrates, and organic acids including fatty acids and nucleic acids are listed in Tables 1−3. The metabolic profiling is able to reveal the global metabolic perturbation associated with AOM-induced carcinogenesis and Res treatment. The heatmap (Figure 4) generated with the differential metabolites contributing for the separation of AOM group and Res-AOM group from control group indicate less significant fluctuation of metabolite levels in the Res-AOM group, suggesting that Res could attenuate the AOM-induced metabolic perturbation in rats. The inhibition of metabolic alteration by Res is consistent with their chemopreventive effects of carcinogenesis as demonstrated by the significantly decreased number of ACF in Res-AOM group. However, not all the differentially expressed metabolites in AOM group were attenuated or normalized by Res intervention. Therefore, a list of differential metabolites (Tables 1−3) responsible for the separation between AOM group and Res-AOM group was identified. These “Res-induced” markers are different from those differential metabolites in AOM group relative to the controls. Many of these significantly altered metabolites in tissue, sera, and urine samples, including arachidonate, linoleate, glutamate, docosahexaenoate, palmitelaidate, 2-aminobutyrate, pyroglutamate, and threonate, are associated with inflammation and oxidation. It has been reported that Res has

postdose of AOM, along with the appearance of ACF, a visible morphological precancerous lesions marker.31,32 Res demonstrated considerable effects on lowering weight gain of AOM rats as well as ACF numbers in both descending and proximal colon, which is similar to the chemopreventive effects of many other dietary agents such as green tea,33 curcumin,34 cruciferous vegetables,35 folate,36 and raspberries.37 It was reported that AOM impact tumor endocrine system, causing a disordered metabolic phenotype similar to that of obesity and hyperglycemia,38,39 which may accelerate the cancerization.40 Low total cholesterol was associated with higher overall cancer incidence and mortality in human subjects.41 Furthermore, investigations published recently suggest that lower levels of serum cholesterol might be a marker of existing or undiagnosed malignancy.42,43 GC-MS based metabonomic study reveals significant variations in the AOM group, characterized by 20, 11, and 17 metabolites differentially expressed in colonic tissue, serum, and urine, respectively. The PLS-DA models derived from our metabonomic data were able to discriminate the AOM group and Res-AOM group from their healthy counterparts, highlighting the diagnostic potential of this noninvasive analytical approach. The differentially expressed metabolites such as 3440

dx.doi.org/10.1021/pr300284h | J. Proteome Res. 2012, 11, 3436−3448

Journal of Proteome Research

Article

Figure 3. Metabolic profiles depicted by PLS-DA scores plot of GC-TOFMS spectral data from the (A and D) colonic tissue, (B and E) serum, and (C and F) urine of control group, AOM group, and Res-AOM group.

potent antioxidant44 and anti-inflammatory45 properties and might play an important role in protecting the colon against

carcinogen-induced neoplasia. The significant alteration in inflammation and oxidation-related metabolites suggests that 3441

dx.doi.org/10.1021/pr300284h | J. Proteome Res. 2012, 11, 3436−3448

3442

16.82 17.1 15.64 6.97 21.72 18.22 16.23 10.13 9.37 21.79 12.02 9.51 9.79 8.88 15.88 5.97 21.11 17.49 13.25 14.96 18.78 7.15 23.1 6.14 19.59 11.44 21.13 23.12 17.66

Rt (min) 1.6 1.6 1.8 1.4 2 2 1.8 1.3 1.8 1.7 1.6 1.5 1.4 1.6 1.4 1.6 1.3 1.4 1.8 2.2

VIP

b

1.50 2.09 3.93 2.52 2.71 2.89 1.03 3.22 7.08 1.06 1.04 1.26 4.35 8.94 2.96 6.35 1.64 7.98 1.65 6.45

× × × × × × × × × × × × × × × × × × × ×

P

c

10 10−2 10−2 10−2 10−2 10−2 10−3 10−2 10−4 10−2 10−2 10−2 10−2 10−3 10−2 10−3 10−3 10−4 10−3 10−3

−2

2.50 3.21 4.14 3.49 3.49 3.49 6.60 3.57 6.60 2.13 2.13 2.29 4.35 2.13 3.49 1.84 6.60 6.60 6.60 1.84

× × × × × × × × × × × × × × × × × × × × 10 10−2 10−2 10−2 10−2 10−2 10−3 10−2 10−3 10−2 10−2 10−2 10−2 10−2 10−2 10−2 10−3 10−3 10−3 10−2

−2

adjusted p

d

AOM group relative to controls 0.36 0.44 0.60 0.60 0.61 0.64 0.79 1.13 0.61 0.62 0.63 0.79 1.12 1.16 1.19 1.20 1.25 1.29 1.33 1.88

FC

e

2.51 6.12 2.00 4.66 1.32 4.07 4.03 7.11 2.06 3.42 4.15 1.62 5.13 8.53 1.59 1.11 7.34 3.63 3.56 5.09

× × × × × × × × × × × × × × × × × × × ×

P

c

10 10−1 10−1 10−1 10−1 10−1 10−7 10−1 10−3 10−2 10−3 10−2 10−1 10−2 10−1 10−1 10−2 10−2 10−2 10−4

−1

3.34 6.45 2.85 5.48 2.21 5.08 8.06 7.11 1.38 9.07 2.07 6.48 5.69 1.71 2.45 2.02 1.63 9.07 9.07 5.09

× × × × × × × × × × × × × × × × × × × × 10 10−1 10−1 10−1 10−1 10−1 10−6 10−1 10−2 10−2 10−2 10−2 10−1 10−1 10−1 10−1 10−1 10−2 10−2 10−3

−1

adjusted p

d

Res-AOM group relative to controls 0.73 0.89 0.76 0.91 0.78 0.90 0.51 1.02 0.65 0.76 0.58 0.81 1.05 1.15 1.15 1.14 1.19 1.25 1.28 2.02

FC

f

2.3 2.4 1.9 2.1 2.4 2.3 2.3 1.9 3.2

1.9 2.5 2.2 2.1 1.9 1.9 3 1.9

VIP

b

7.87 5.01 3.13 2.22 5.08 9.18 8.04 3.97 1.32

4.27 3.82 1.25 1.78 4.39 4.14 3.42 3.50

× × × × × × × × ×

× × × × × × × ×

Pc

10−3 10−3 10−2 10−2 10−3 10−3 10−3 10−2 10−6

10 10−3 10−2 10−2 10−2 10−2 10−5 10−2

−2

1.95 1.73 4.39 3.43 1.73 1.95 1.95 4.39 2.25

4.39 1.73 2.36 3.03 4.39 4.39 2.91 4.39

× × × × × × × × ×

× × × × × × × ×

10−2 10−2 10−2 10−2 10−2 10−2 10−2 10−2 10−5

10 10−2 10−2 10−2 10−2 10−2 10−4 10−2

−2

adjusted pd

Res-AOM group relative to AOM group

0.71 0.73 0.86 0.90 1.13 1.16 1.34 1.36 2.12

2.03 2.03 1.28 1.52 1.27 1.40 0.65 0.91

FCg

a Asterisks (*) indicate metabolites are verified by reference standards. bVariable importance in the projection (VIP) was obtained from PLS-DA model with a threshold of 1.0. cP-values were calculated from Wilcoxon−Mann−Whitney test. dAdjusted for multiple testing with FDR control.24 eFold change (FC) was obtained by comparing those metabolites in AOM group to control group. fFC was obtained by comparing those metabolites in Res-AOM group to control group. gFC was obtained by comparing those metabolites Res-AOM group to AOM group. FC with a value >1 indicates a relatively higher concentration present in AOM group or Res-AOM group while a value 1 also indicates a relatively higher concentration present in Res-AOM group while a value 1 indicates a relatively higher concentration present in AOM group or Res-AOM group while a value 1 also indicates a relatively higher concentration present in Res-AOM group while a value 1 indicates a relatively higher concentration present in AOM group or Res-AOM group while a value 1 also indicates a relatively higher concentration present in Res-AOM group while a value