ReseaRch aRticle

Liver disease can be categorized by cause, which may include infection, injury, exposure to drugs or toxic compounds, autoimmune processes or a genetic defect that leads to the deposition and build-up of damaging substances, such as iron or copper [1–4]. To understand the many signs of hepatotoxicity, such as inflammation, scarring, obstructions, clotting abnormalities and liver failure, it is important to understand the role of the liver in the metabolism of xenobiotics. The liver is the key organ in the metabolism and detoxification of foreign compounds. By using metabolomics technology, metabolites and the end products of metabolism can be quantified, and, then provide diagnostic biomarkers and information to interpret the mechanisms of hepatotoxicity.

Hepatic injury by xenobiotics can result in morphologic changes and increases in serum aspartate aminotransferase (AST) and ala-nine aminotransferase (ALT) levels, which are commonly used as clinical markers for liver damage [5,6]. AST and ALT are located in the cytoplasm of hepatocytes, and the levels of these enzymes increase with hepatocellular injury [7]. Morphologic changes are usually detected by liver biopsy, while AST and ALT levels are mea-sured in the blood. A drop in aminotransferase level might indicate a depletion of viable hepa-tocytes and acute massive (fulminant) hepatitis, which often has a poor prognosis. Conversely, exceedingly high (>1000 IU/l) aminotransferase

levels are typically associated with acute viral hepatitis, severe toxic or drug reactions or isch-emic hepatitis (i.e., an insufficient flow of blood to the liver). Less dramatic aminotransferase elevation is observed in many types of hepatic injury [8].

However, these diagnostic tests have lim-its; both biopsy and blood tests are considered invasive, and they have the potential for sam-pling error. Although elevated ALT can reflect even minimal amounts of active hepatocyte damage, AST and ALT level tests often result in false-negative or -positive outcomes depend-ing on injury time or the causative compound [9]. Moreover, there are no set guidelines currently linking particular drugs or toxic compounds to specific ranges of AST and ALT values.

According to Parkes et al., fibrosis stage in chronic hepatitis C could not be adequately determined using ten different panels of serum markers of hepatic fibrosis; therefore, more accu-rate biomarkers are needed to diagnose chronic liver damage [10].

Most recent metabolomic toxicology studies have been performed using several toxicants, and these experiments have generally aimed to iden-tify biomarkers of severe liver damage induced by high doses of an acute toxicant [1,11,12]. How-ever, a metabolomics-based biomarker study correlating enzyme values and endogenous metabolites has not been conducted, especially in the context of chronic toxicant exposure.

Investigation of metabolite alteration in dimethylnitrosamine-induced liver fibrosis by GC–MS

Background: A metabolomic study of biomarkers associated with dimethylnitrosamine (DMN)-induced hepatic fibrosis in Sprague-Dawley rats was performed using GC–MS. The clinical chemistry of the collected blood and the histopathology of excised liver samples were examined, and urine samples were prepared by solvent extraction. Results: Through pattern ana lysis, the DMN-treated group was divided into two subgroups based on the aspartate aminotransferase (AST) levels compared with the control, a moderately higher group (DMN subgroup A) and a significantly higher group (DMN subgroup B). Uric acid, orotic acid, N-phenylacetylglycine and glutaric acid were biomarkers for DMN subgroup A, aminomalonic acid was a biomarker for DMN subgroup B, and arabitol level distinguished control versus DMN treatment regardless of AST level. Conclusion: This study suggests that the identification and profiling of AST level-related metabolites may be useful as a diagnostic tool and for the study of the mechanism of liver fibrosis induced by DMN.

Hyun Kyoung Ju1, Ha Wook Chung1, Hee-Seung Lee2, Johan Lim3, Jeong Hill Park1, Sung Cil Lim4, Joon Mee Kim5, Soon-Sun Hong†2 & Sung Won Kwon*†1

1College of Pharmacy & Research Institute of Pharmaceutical Sciences, Seoul National University, Seoul 151-742, Korea 2College of Medicine (BK21), Inha University, Incheon 400-712, Korea 3Department of Statistics, Seoul National University, Seoul 151-742, Korea 4College of Pharmacy, Chungbuk National University, Cheongju 361-763, Korea 5Department of Pathology, Inha University Hospital, College of Medicine, Inha University, Incheon 400-712, Korea *Author for correspondence: Tel.: +82 2 880 7844 Fax: +82 2 886 7844 E-mail: swkwon@snu.ac.kr †Authors contributed equally.

41ISSN 1757-618010.4155/BIO.12.296 © 2013 Future Science Ltd Bioanalysis (2013) 5(1), 41–51

This metabolomics study was performed using GC–MS on urine samples from rats treated with N,N-dimethyl nitrous amide, also known as dimethylnitrosamine (DMN). DMN is a well-studied alkylating agent that causes liver dis-ease by alkylating hepatocyte macromolecules [13–17]. DMN causes hemorrhagic zone 3 necro-sis, steatosis, progressive fibrogenesis, cirrhosis and fatty changes when repeatedly adminis-tered at doses near the LD

50 [14,18]. In contrast

to the events of acute hepatotoxicity, the liver progresses from fibrotic to cirrhotic in chronic hepatic fibrosis; this may cause fluctuations in the serological and biochemical data in chronic toxicology studies [11,19]. Clayton et al. has shown that the liver toxicity induced by paracetamol was related with its metabolic fate, and labora-tory animals with the same genetic backgrounds can be predicted with urine metabolite profile [20]. Additionally, in metabolomics studies, the laboratory animals could be classified as either responders or nonresponders due to the inter-subject variation induced by pharmacological or nutritional effect [21]. In the present study, the fluctuation of AST values was observed and the DMN-treated group was divided as two sub-groups, focusing on the correlations between AST level, ALT level and metabolite changes during liver damage.

This study was expected to identify biomark-ers altered by DMN, which could then be used to interpret the biochemistry of the disease.

Experimental�� Chemicals

All organic solvents, DMN, and N,O- bis(trimethylsilyl)trifluoroacetamide with 1% trimethylchlorosilane (BSTFA 1% TMCS) were purchased from Sigma-Aldrich (MO, USA).

�� Animals & treatmentsMale Sprague-Dawley rats were purchased from Orient Bio Laboratory Animal Research Center Co. Ltd (Gyeonggi-do, Korea). Animals were kept in a humidity- and temperature-controlled facility under 12-h day–night cycles, and stan-dard rat chow and free access to drinking water was provided. Rats were divided into two treat-ment arms, the DMN and control arms. The two groups were treated as described in our pre-vious report [22]. At the end of the fourth week, urine samples were collected in bottles that had been pre-cooled for 24 h and contained 0.5 ml of sodium azide solution (100 mg/ml), and all rats were euthanized. Blood samples and liver tissue

were collected for biomedical and histochemical analyses, and the collected samples were stored at -80°C until further ana lysis. All experiments were approved and reviewed by Inha University (Approval ID: 090601–9).

�� Sample preparationThe collected urine samples were centrifuged at 14,000 g for 5 min to remove macroparticles, and 200 µl of the supernatant was diluted in 800 µl of cold methanol. After vortexing for 10 s, the extracts were centrifuged at 14,000 g for 5 min and the supernatant obtained from centrifugation was transferred to a new glass vial. For derivatization of the sample extracts, samples were dried completely under nitrogen and then agitated with a mixture of BSTFA 1% TMCS and pyridine (1:2) at 60°C for 15 min. The reaction mixture was transferred to a 2-ml glass vial equipped with glass micro-inserts (Agilent Technologies, CA, USA) and capped immediately.

�� Histopathological dataLiver specimens were sliced into 5-µm thick sec-tions and fixed in formaldehyde for hematoxy-lin and eosin, and Masson’s trichrome staining. Liver section images were analyzed under a light microscope. The degree of hepatic fibrosis was measured as previously described [22].

�� Biomedical analysis of serumThe liver damage indicators, serum AST and ALT, were analyzed using the standard meth-ods (7600 Analyzer, Hitachi, Tokyo, Japan Wako Diagnostics reagents, Wako Pure Chemi-cal Industries Ltd, Osaka, Japan) by the Inha University Hospital (Incheon, Korea).

�� GC–MSGC–MS ana lysis was performed using a 6890 gas chromatograph (Agilent Technologies) coupled with a JMS-GC mate (Jeol Ltd, Tokyo, Japan). Chromatographic separation was per-formed on a DB-5 column (60 m × 0.25 mm ID, 0.25 µm film thickness, Hewlett-Packard, CA, USA). The GC oven temperature was held at 60°C for 5 min, increased to 320°C at a rate of 10°C/min and held at the final temperature for 10 min. A 1-µl sample was injected in split mode (10:1) and helium (99.9999% He) was used as a carrier gas at a constant flow of 1 ml/min. The transfer line and ion source temperatures were set at 280°C and 250°C, respectively. After a 300-s solvent delay, mass spectra were obtained

Key Terms

Xenobiotics: Foreign chemicals that are found in organisms.

Aspartate aminotransferase: Pyridoxal phosphate-dependent transaminase enzyme and is an important enzyme in amino acid metabolism.

Alanine aminotransferase: Transaminase enzyme that is found primarily in the liver and kidney.

Principal component analysis: Technique for finding a transformation that transforms an original set of correlated variables to a new set of uncorrelated variables.

Orthogonal partial least squares-discriminant analysis: Separates the partial least-squares components into two groups: components that are related to Y, called predictive; and components that are orthogonal to Y, called orthogonal.

Variable importance in projection: Values reflect the importance of terms in the model both with respect to Y.

Bioanalysis (2013) 5(1)42 future science group

ReseaRch aRticle | Ju, Chung, Lee et al.

at 20-scans per second with a mass range of m/z 55–500. The ionization energy was 70 eV in electron impact mode.

�� Data processing & statisticsAll obtained GC–MS data were converted to ASCII format and reduced into 0.1 min buck-ets, which were then normalized by creatinine levels and imported into SIMCA-P 12.0 (Umet-rics, Umeå, Sweden). Additionally, all variables were scaled using a Pareto scaling function to enhance the weight of medium features with-out inflating baseline noise. Principal compo-nent ana lysis was performed with all variables, and two principal components (PCs), PC1 and PC2, were plotted to identify trends between groups. The missing values tolerance was set at 60%, and outliers within the group were identified by Hotelling’s T2 distribution ana-lysis, which is a multivariate generalization of the Student’s t-distribution. Orthogonal partial least squares-discriminant ana lysis (orthogo-nal PLS-DA) and PLS-DA were employed for the supervised pattern ana lysis. The S-plot and variable importance in projection (VIP) list were obtained from these analyses, respectively. The S-plot was utilized to filter out putative bio-markers from the data, and VIPs were used to indicate key buckets for interpretation of the relationship between groups of variables. The combined compounds in the key buckets were divided manually by mass to charge ratio (m/z) and retention time, and their differences were evaluated by one-way ANOVA (student t-test p-values <0.05). Metabolites with a p-value <0.05 were classified as potential biomarkers to discriminate between the disease and control group. The compounds were identified using authentic standards and the NIST library.

Results & discussion�� AST, ALT & histological ana lysis

To score the degree of liver fibrosis, AST and ALT activities were assessed based on the histo-logical analysis of livers stained with hematoxy-lin and eosin and Masson’s trichrome. As shown in Figure 1, hematoxylin and eosin, and Masson’s trichrome staining reflected the degree of liver fibrosis in each group. Liver fibrosis, which is defined as the disruption of cell architecture and collagen growth, was increased by DMN treat-ment. Portal widening with occasional bridg-ing fibrosis and mild fatty changes was observed from the eighth to tenth DMN samples. In the seventh DMN sample, macrovesicular fatty

Central veinSteatosis

Steatosis

Nucleophilicinfiltrate

Fibrosis

Fibrosis

Figure 1. Masson’s trichrome-stained, and hematoxylin and eosin-stained liver specimens from control and dimethylnitrosamine treatment subgroup A and B samples. Masson’s trichrome-staining of (A) control, (B) dimethylnitrosamine treatment subgroup A and (C) subgroup B. Hematoxylin and eosin staining of (D) control, (E) dimethylnitrosamine subgroup A and (F) subgroup B. Magnification: 400-times for Masson’s trichrome staining and 200-times for hematoxylin and eosin staining.

changes without fibrosis were noted (Figure 2). A comparison of biochemical and histological analyses indicated that less elevated AST and ALT seemed to be related to occasional bridg-ing fibrosis accompanied by mild fatty changes (steatohepatitis), which are usually present with abdominal fullness or pain as their only complaints.

Mean AST levels were increased, albeit not significantly, from 109.6 ± 8.1 IU/l in the con-trol group (n = 5) to 498.6 ± 409.2 IU/l in the DMN-treated group (n = 10) (p > 0.05), while the mean ALT levels increased significantly from 46.6 ± 6.9 IU/l in the control group to 287.4 ± 221.9 IU/l in the DMN-treated group(0.01 < p < 0.05; Table 1). These results demonstrate the change in AST levels after DMN treatment.

According to this observation, AST seems to be correlated with the degree of liver fibrosis in the context of chronic damage, while ALT indi-cates acute liver damage. ALT was not a specific marker for hepatic fibrosis (chronic liver dam-age), even though ALT and AST both indicate liver damage.

�� Pattern ana lysis First, to identify the metabolites that are altered by treatment, the chromatograms of

Metabolite alteration in dimethylnitrosamine-induced liver fibrosis by GC–MS | ReseaRch aRticle

www.future-science.com 43future science group

all analyzed samples were compared in a 1D projection. However, it was difficult to inter-pret the changes between samples because the chromatograms ref lected several variables, including retention time, mass fragment pat-tern and intensity (Figure 3). To simplify the variables and improve the comparison, princi-pal component ana lysis, which is often used in metabolomics studies, was performed. Principal components were computed by the SIMCA-P+ program. The two major principal components, PC1 and PC2, are plotted in Fig-ure 4a & b, which shows the trends and outliers within each group. The outliers, which were identified by drawing a tolerance ellipse based on the Hotelling’s T2 distribution (0.95), did not appear within the groups. The principal component ana lysis score plot performed using

two groups show that the control and DMN groups were not discriminated by PC1 or PC2 (Figure 4C). In the supervised multivariate pat-tern ana lysis, the control versus DMN group was separated by PC1, and DMN samples were distinguished within group by PC1 and PC3, as shown in Figure 4D. Interestingly, this result seems to be affected by the levels of AST. Additionally, with respect to PC1, the pattern separated the control group, DMN one to six and DMN seven to ten. Samples seven to ten had lower AST levels than other DMN samples, and these samples are relatively similar to the control group with respect to PC3. This result suggests that the metabolites altered by changes in AST contribute to the PLS-DA model.

Compared with all pattern ana lysis data, the maximized group separation was well designed by the orthogonal PLS-DA model (Figure 5). The results of goodness of fit and the predict-ability of the pattern ana lysis models were calcu-lated as R2 and Q2, respectively, and the results are shown in Table 2. The high R2 and Q2 values demonstrate that the model was clearly classified and predictable.

�� Biomarkers of liver fibrosis induced by DMNMetabolites were analyzed by GC–MS, and the first annotation of each metabolite was performed using the NIST search library. To confirm endogenous metabolites detected by GC–MS, a built-in library consisting of the retention time and mass fragments produced at 70 eV energy was developed with authentic standards of amino acids, fatty acids and sugars (Table 3). All identified metabolites were inte-grated according to thier specific m/z, which were normalized to the creatinine levels in each sample. The creatinine levels were determined with manual peak integration of creatinine peak, which was confirmed by creatinine stan-dard. The first step of the biomarker search was to find metabolites that were altered between the control and treatment groups. In the present study, to identify biomarkers that might indi-cate DMN exposure, key buckets with small jackknife uncertainty bars were first selected in the VIPs list. Among the selected metabo-lites, arabitol was identified as a discriminator between the control and DMN-treated groups. Uric acid, orotic acid, N-phenylacetylglycine (PAG) and glutaric acid were identified as metabolites that were significantly different

Figure 2. Masson’s trichrome-stained liver specimens from the dimethylnitrosamine treatment samples. (A) Sixth, (B) seventh, (C) eighth, (D) ninth and (E) tenth dimethylnitrosamine treatments. Early bridging fibrosis was observed in the sixth sample, and macrovesicular fatty changes without fibrosis were observed in the seventh sample. Portal widening with occasional bridging fibrosis and mild fatty changes were observed in the eighth and tenth samples. Magnification: 100-times.

ReseaRch aRticle | Ju, Chung, Lee et al.

Bioanalysis (2013) 5(1)44 future science group

between controls and subgroup A (Figure 6). The level of aminomalonic acid decreased when AST and ALT values were significantly elevated compared with controls, and it was

slightly (but not significantly) increased at low levels of AST, even though the histo pathology demonstrated liver fibrosis. Meanwhile, 5-oxo-proline was identified as the biomarker that best

Retention time (min)

Inte

nsi

ty

18:03.90.E+00

5.E+06

1.E+07

2.E+07

2.E+07

19:22.3 20:40.7 21:59.1 23:17.5 24:35.9 25:54.3

Retention time (min)

Inte

nsi

ty

18:03.90.E+00

5.E+06

1.E+07

2.E+07

2.E+07

19:22.3 20:40.7 21:59.1 23:17.5 24:35.9 25:54.3

Retention time (min)

Inte

nsi

ty

18:03.90.E+00

5.E+06

1.E+07

2.E+07

2.E+07

19:22.3 20:40.7 21:59.1 23:17.5 24:35.9 25:54.3

Figure 3. GC–MS total ion chromatograms of dimethylnitrosamine-treated and control samples. (A) Control group, (B) dimethylnitrosamine-treated subgroup A with low aspartate aminotransferase and moderate alanine aminotransferase values, (C) dimethylnitrosamine-treated subgroup B with high aspartate aminotransferase and high alanine aminotransferase activities.

Table 1. Alanine aminotransferase and aspartate aminotransferase activities before (control group) and after dimethylnitrosamine administration.

Number Control Dimethylnitrosamine

AST activity (IU/l) ALT activity (IU/l) AST activity (IU/l) ALT activity (IU/l)

1 118 43 1261 778

2 108 57 1000 466

3 97 50 926 468

4 110 40 330 191

5 115 43 404 328

6 - - 405 129

7 - - 170 144

8 - - 158 130

9 - - 179 134

10 - - 153 106

Average 109.6 46.6 498.6 287.4

SD 8.1 6.9 409.2 221.9ALT: Alanine aminotransferase; AST: Aspartate aminotransferase.

Metabolite alteration in dimethylnitrosamine-induced liver fibrosis by GC–MS | ReseaRch aRticle

www.future-science.com 45future science group

separated subgroups A and B. These metabolites were shown in the coefficient plot of X-variables with Y. Figure 7 shows that the retention times of glutaric acid, 5-oxoproline, arabitol, orotic acid, PAG and uric acid were strongly correlated with intensity (Y) in the established PLS-DA model.

Aminomalonic acid is generated when errors in protein synthesis or oxidative damage to amino acid residues in proteins occur; therefore, an elevated aminomalonic acid level after DMN treatment was expected because DMN induces oxidative damage and impairs protein synthesis [23]. Contrary to this expectation, a decreased

level of aminomalonic was observed after DMN treatment. In a previous study, Xue et al. showed that aminomalonic acid is a biomarker of hepa-tocellular carcinoma due to the degradation and ubiquitinylation of IRP2, which is related to the conversion of cysteine into aminomalonic acid [24]. Additionally, iron deposition accompa-nied with fibrosis induced by DMN was reported by He et al. [25]. These reports suggest that an increase in aminomalonic acid may be related to iron metabolism. However, while further study of this relationship is needed, our results suggest that aminomalonic acid could be a biomarker

0t[2]

t[1]0

10,000

-10,000

20,000

-20,000

40,00020,000-20,000 60,000-40,000-60,000

30,000

-30,000

0t[2]

t[1]0

10,000

-10,000

20,000

-20,000

40,00020,000-20,000 60,000-40,000-60,000

30,000

-30,000

0t[3]

t[1]0

10,000

-10,000

20,000

-20,000

40,00020,000-20,000 60,000-40,000-60,000

30,000

-30,000

0t[2]

t[1]0

10,000

-10,000

20,000

-20,000

40,00020,000-20,000 60,000-40,000-60,000

30,000

-30,000

Figure 4. Multivariate ana lysis by principal component ana lysis of control and dimethylnitrosamine treatment samples and the partial least-squares discriminant ana lysis. (A) Control, (B) dimethylnitrosamine group, (C) control and dimethylnitrosamine and (D) partial least-squares discriminant ana lysis. The scatter plots were plotted by t[1], and t[2], for principal component ana lysis, and t[1] and t[3] for partial least-squares discriminant ana lysis. The tolerance ellipses of all models were based on Hotelling’s T2 and drawn as circles. t[1]: 1st component; t[2]: 2nd component; t[3]: 3rd component.

ReseaRch aRticle | Ju, Chung, Lee et al.

Bioanalysis (2013) 5(1)46 future science group

of hepatic fibrosis when serum AST and ALT levels are high.

Uric acid, orotic acid, glutaric acid and PAG increased in subgroup A but were not changed in subgroup B. Uric acid is a final product of purine metabolism and is produced after hypo-xanthine is oxidized to xanthine; there was a significant increase in uric acid in DMN sub-group A compared with the control group. This result is consistent with a previous report and the observation of elevated hypoxanthine guanine phospho ribosyl transferase levels after DMN injury [26,27]. However, as shown in Figure 6, uric acid was increased when the AST level is low, not high. Abnormalities in AST, that is, during liver dysfunction, may promote aspartate accumula-tion and induce nucleic acid synthesis in the pres-ence of 5-phosphoribosyl 1-pyrophosphate, lim-iting the use of uric acid for recycling purposes [101]. Additionally, significantly elevated uric acid seems to be correlated with fatty changes rather than liver fibrosis. This suggests that an increase or decrease of uric acid after chronic DMN treat-ments due to nucleic acid metabolism; uric acid can reflect the AST level, but it cannot be used as a biomarker of hepatic fibrosis.

The increased level of orotic acid is well documented in relation to ammonia detoxifi-cation and arginine-deficient diets [28].More-over, according to a study of gene regulation changes promoted by DMN, arginosuccinate synthetase, which is involved in the urea cycle, was decreased by DMN treatment, resulting in increased urinary orotic acid levels [27]. How-ever, in subgroup B, the expected result was not observed. Orotic acid is generated by pyrimidine synthesis, in which carbamyl phosphate and aspartate are condensed in a reaction catalyzed by AST [29]. Therefore, the level of orotic acid could be affected by AST activities, and high levels of AST in the serum indicate that neither pyrimidine synthesis nor the urea cycle is acti-vated. Such as orotic acid and uric acid, metabo-lites affected by AST levels should be considered as biomarkers, but these were not identified as discriminators of hepatic fibrosis.

5-oxoproline, an uncommon amino acid derivative in which the free amino group of glu-tamic acid is cyclized to form a lactam, plays a role in glutathione synthesis and has been used as an indicator for glutathione synthetase deficiency [30]. Based on the results of this study, 5-oxop-roline is a discriminator between subgroup A and B (Figures 6 & 7) even though the level of 5-oxoproline was not altered significantly after

DMN treatment. This result contrasts with a previous report that glutathione synthetase gene expression was induced by DMN; liver regen-eration induced by 4 weeks of DMN treatment might have caused this fluctuation, even though glutathione synthetase participates in the liver fibrosis process [27].

Table 3. Altered metabolites in urine samples using databases based on NIST and built-in libraries.

Compunds TMS m/z Retention time (min)

Representative m/z

Glutaric acid† 276 (2-TMS) 18.46 261

Aminomalonic acid‡ 335 (3-TMS) 19.46 320

5-oxoproline† 273 (2-TMS) 20.28 258

Arabitol‡ 512 (5-TMS) 22.42 319

Orotic acid† 357 (3-TMS) 22.58 357

N-phenylacetylglycine† 265 (1-TMS) 24.30 265

Uric acid† 456 (4-TMS) 26.39 456†Biomarkers to subdivide the dimethylnitrosamine-treated group based on the aspartate aminotransferase levels. Metabolites with a p-value <0.05 were identified using authentic standards; others were identified by NIST library search. ‡Biomarkers to distinguish between control and dimethylnitrosamine-treated groups.TMS: Trimethylsilylation.

0

t0[1

]

t[1]0

10,000

-10,000

20,000

-20,000

20,00010,000-10,000 30,000-20,000-30,000

30,000

40,000

-30,000

-40,000

Figure 5. Multivariate ana lysis by orthogonal partial least-squares discriminant ana lysis. Control and dimethylnitrosamine-treated groups were well separated by orthogonal partial least-squares discriminant ana lysis modeling. t[1]: 1st component; to[1]: The score that summarizes X variation orthogonal to Y.

Table 2. Goodness of fit and predictability of pattern ana lysis models.

Model Goodness of fit (R2) Predictability (Q2)

Principal component ana lysis 0.946 0.899

Partial least-squares discriminant ana lysis

0.940 0.845

Orthogonal partial least-squares discriminant ana lysis

0.940 0.872

Metabolite alteration in dimethylnitrosamine-induced liver fibrosis by GC–MS | ReseaRch aRticle

www.future-science.com 47future science group

Figure 6. Box-whisker plots of endogenous metabolites and biomarkers altered by dimethylnitrosamine treatment. (A) Glutaric acid, (B) N-phenylacetylglycine, (C) orotic acid, (D) uric acid, (E) arabitol, (F) 5-oxoproline, (G) aminomalonic acid. The y-axis plot represents the intensity ratio normalized to creatinine level, and box plots were generated with a stacked column chart using bottom (quartile 1), 2Q (median - quartile 1), and 3Q (quartile 3 - median) values. Control: n = 5; dimethylnitrosamine subgroup A: n = 4; dimethylnitrosamine subgroup B: n = 6. Error bars on the box plot are ± whisker value. *p < 0.05.

Control DMN subgroup A DMN subgroup B0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Inte

nsi

ty r

atio

Control DMN subgroup A DMN subgroup B0.0

0.1

0.2

0.3

0.4

0.5

Inte

nsi

ty r

atio

0.0

0.1

0.2

0.3

Inte

nsi

ty r

atio

Control DMN subgroup A DMN subgroup B0

1

2

3

4

5

6

7

Inte

nsi

ty r

atio

Control DMN subgroup A DMN subgroup B

0.0

0.2

0.4

0.6

0.8

Control DMN subgroup A

Inte

nsi

ty r

atio

DMN subgroup B

0.0

0.1

0.2

0.3

0.4

0.5

Inte

nsi

ty r

atio

Control DMN subgroup A DMN subgroup B

0.00

0.01

0.02

0.03

0.04

0.05

0.06

Inte

nsi

ty r

atio

Control DMN subgroup A DMN subgroup B

* *

*

*

*

*

**

ReseaRch aRticle | Ju, Chung, Lee et al.

Bioanalysis (2013) 5(1)48 future science group

Glutaric acid is an intermediate breakdown product of lysine in collagen synthesis, and PAG, as an acyl glycine, is a well-known bio-marker for liver phospholipidosis [31]. The pat-tern of these two biomarkers was also depen-dent on the change in AST level, as shown in Figure 6, indicating their correlation. There-fore, glutaric acid and PAG were designated as biomarkers related to mitochondrial fatty acid b-oxidation; fatty changes caused by DMN exposure were observed in DMN subgroup A.

Arabitol, which has not previously been reported to be related to DMN, was identi-fied as a discriminator between control and DMN-treated groups (both subgroup A and subgroup B). Although it was recently reported that congenital liver cirrhosis was correlated with elevated urine levels of arabitol, this study is the first to demonstrate that arabitol is a biomarker of DMN exposure [32]. Arabitol could be used as a potential discriminator of hepatic damage induced by DMN rather than other hepatotoxins.

ConclusionA metabolomics study using GC–MS was per-formed to identify biomarkers of the diverse

pattern of liver f ibrosis induced by DMN. After DMN pretreatment, arabitol, 5-oxop-roline, aminomalonic acid, uric acid, orotic acid, PAG and glutaric acid levels were altered, and arabitol was confirmed as a biomarker of DMN-related damage, independent of ele-vated AST and ALT. Based on the results of the multivariate ana l ysis, the DMN groups were categorized into two groups, and uric acid, orotic acid, PAG and glutaric acid showed a significant increase when both lev-els of AST and ALT were significantly raised. Aminomalonic acid was decreased when AST level was slightly elevated. Overall, the pattern ana lysis of the data from this study, including numerous variables, showed that aminoma-lonic acid accurately ref lects the status of hepatic damage, and it may be a very useful tool when accompanied with biochemical and histological data. Consequently, liver damage induced by DMN can be represented using biomarkers obtained from a metabolomics study and metabolic biomarkers will be use-ful for studying the mechanism and diagnosis of liver disease with serum biomarkers such as AST and ALT.

18:0

3.9

18:1

0.1

18:2

0.3

18:3

0.5

18:4

0.7

18:5

0.1

19:0

0.3

19:1

0.5

19:2

0.7

19:3

0.1

19:4

0.3

18:5

0.5

20:0

0.7

20:1

0.1

20:2

0.3

20:3

0.5

20:4

0.7

20:5

0.1

21:0

0.3

21:1

0.5

21:2

0.7

21:3

0.1

21:4

0.1

21:5

0.5

22:0

0.6

22:1

0.1

22:2

0.2

22:3

0.4

22:4

0.6

22:5

023

:00.

223

:10.

423

:20.

6

23:4

0.2

23:5

0.4

24:0

0.6

24:1

0.0

24:2

0.2

24:3

0.4

24:4

0.0

24:5

0.0

25:0

0.2

25:1

0.4

25:2

0.6

25:3

0.0

25:4

0.2

25:5

0.4

26:0

0.6

26:1

0.0

26:2

0.2

26:3

0.3

26:4

0.5

26:3

0.1

27:0

0.1

27:1

0.3

23:3

0

Glutaric acid 5-oxoproline Arabitol Orotic acid N-phenylacetylglycine

Co

effC

S

Uric acid

Var ID (primary)

-0.2

-0.1

0.0

0.1

0.2

Figure 7. Coefficient plot of the partial least-squares discriminant ana lysis model. These coefficients refer to centered and scaled X-data, with Y scaled but not centered. The coefficient is significant when the confidence interval does not cross zero, and nine buckets show significance. Among those buckets, the six, including the metabolites selected from variable importance in projection lists, were expressed as red boxes; these X variables are strongly correlated with Y.

Metabolite alteration in dimethylnitrosamine-induced liver fibrosis by GC–MS | ReseaRch aRticle

www.future-science.com 49future science group

Future perspectiveMetabolomics has been an emerging technol-ogy for the identification of the metabolites induced by xenobiotics or environmental fac-tors. In particular, it has many advantages over the traditional approach for the discovery of the biomarkers linked with diverse diseases. As with the study presented here, the biomarkers inter-related with the status of targeted organism will be useful for early diagnosis of disease as well as status of disease.

Financial & competing interests disclosureThis research was supported by the Basic Science Research Program through the National Research Foundation of Korea, funded by the Ministry of Education, Science and Technology (2012-0004080), and was supported by a

grant from the Next-Generation BioGreen 21 Program (No. PJ008202), Rural Development Administration, Republic of Korea. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the m anuscript apart from those disclosed.

Editing service was used to check grammatical errors, which was supported by the BK21 grant.

Ethical conduct of research The authors state that they have obtained appropriate insti-tutional review board approval or have followed the princi-ples outlined in the Declaration of Helsinki for all human or animal experimental investigations. In addition, for investi gations involving human subjects, informed consent has been obtained from the participants involved.

Executive summary

�� Endogenous metabolites profiling was performed by GC–MS, which is known as a very stable and robust instrument.

�� Based on the pattern ana lysis of the data, the dimethylnitrosamine (DMN) groups were categorized into two groups and biomarkers related with the diverse pattern of liver fibrosis were identified.

�� Aminomalonic acid in DMN subgroup A, uric acid, orotic acid, phenylacetylglycine and glutaric acid in DMN subgroup B, and arabitol for both of the subgroups, were determined as the biomarkers correlated with aspartate aminotransferase levels.

References1 Ross JS, Symmans WF, Pusztai L,

Hortobagyi GN. Pharmacogenomics and clinical biomarkers in drug discovery and development. Am. J. Clin. Pathol. 124(Suppl.), S29–S41 (2005).

2 Mitchell JR, Jollows DJ. Progress in hepatology. Metabolic activation of drugs to toxic substances.Gastroenterology 68(2), 392–410 (1975).

3 Parke DS. The Biochemistry of Foreign Compounds. Pergamon Press, Oxford, UK (1968).

4 Farber JL. Xenobiolics, drug metabolism and liver injury. Monogr. Pathol. 28, 43–53 (1987).

5 Abdel-Wahhab MA, Gamil K, El-Kady AA, El-Nekeety AA, Naguib KM. Therapeutic effects of korean red ginseng extract in egyptian patients with chronic liver diseases. J. Ginseng Res. 35(1), 69–79 (2011).

6 Zimmerman HJ. Hepatotoxicity: the Adverse Effects of Drugs and Other Chemicals on the Liver (Second edition). Lippincott Williams & Wilkins, Philadelphia, PA, USA (1999).

7 Kim YS, Kim YH, Noh JR, Cho ES, Park JH, Son HY. Protective effect of korean red ginseng against aflatoxin B

1-induced

hepatotoxicity in rat. J. Ginseng Res. 35(2), 243–249 (2011).

8 Mary L. Basic Skills in Interpreting Laboratory Data (fourth edition). American Society of Health-System Pharmacists, Bethesda, MD, USA (2009).

9 Carey E, Carey WD. Noninvasive tests for liver disease, fibrosis, and cirrhosis: is liver biopsy obsolete? Cleve. Clin. J. Med. 77(8), 519–527 (2010).

10 Parkes J, Guha IN, Roderick P, Rosenberg W. Performance of serum marker panels for liver fibrosis in chronic hepatitis C. J. Hepatol. 44(3), 462–474 (2006).

11 Lindon JC, Holmes E, Nicholson JK. Metabonomics and its role in drug development and disease diagnosis. Expert Rev. Mol. Diagn. 4(2), 189–199 (2004).

12 Kim KB, Chung MW, Um SY et al. Metabolomics and biomarker discovery. NMR spectral data of urine and hepatotoxicity by carbon tetrachloride, acetaminophen, and d-galactosamine in rats. Metabolomics 4(4), 377–392 (2008).

13 Reynolds ES. Comparison of early injury to liver endoplasmic-reticulum by halomethanes, hexachloroethane, benzene, toluene, bromobenzene, ethionine, thioacetamide and dimethylnitrosamine. Biochem. Pharmacol. 21(19), 2555–2561 (1972).

14 Barnes JM, Magee PN. Some toxic properties of dimethylnitrosamine. Brit. J. Ind. Med. 11(3), 167–174 (1954).

15 Magee PN, Barnes JM. Carcinogenic nitroso compounds. Adv. Cancer Res. 10, 163–246 (1967).

16 Archer MC. Mechanisms of action of n-nitroso compounds. Cancer Surv. 8(2), 241–250 (1989).

17 Magee PN. Cellular and alkylation of cell components. In: CIBA Foundation Symposium – Cellular Injury. Wiley, Boston, MA, USA, 1–22 (1964).

18 Shu LM, Hollenberg PF. Alkylation of cellular macromolecules and target specificity of carcinogenic nitrosodialkylamines: metabolic activation by cytochromes P450 2B1 and 2E1. Carcinogenesis 18(4), 801–810 (1997).

19 Rouiller CH. Experimental Toxic Injury of the Liver. Academic press, New York, NY, USA (1964).

20 Clayton TA, Lindon JC, Cloarec O et al. Pharmaco-metabonomic phenotyping and personalized drug treatment. Nature 440(7087), 1073–1077 (2006).

21 Li H, Ni Y, Su M et al. Pharmaco metabo-nomic phenotyping reveals different responses to xenobiotic intervention in rats. J. Proteome Res. 6(4), 1364–1370 (2007).

ReseaRch aRticle | Ju, Chung, Lee et al.

Bioanalysis (2013) 5(1)50 future science group

22 Lee HS, Jung KH, Park IS, Kwon SW, Lee DH, Hong SS. Protective effect of morin on dimethylnitrosamine-induced hepatic fibrosis in rats. Dig. Dis. Sci. 54(4), 782–788 (2009).

23 Copley SD, Frank E, Kirsch WM, Koch TH. Detection and possible origins of aminomalonic acid in protein hydrolysates. Anal. Biochem. 201(1), 152–157 (1992).

24 Xue R, Lin Z, Deng C et al. A serum metabolomic investigation on hepatocellular carcinoma patients by chemical derivatization followed by gas chromatography–mass spectrometry. Rapid Commun. Mass Spectrom. 22(19), 3061–3068 (2008).

25 He JY, Ge WH, Chen Y. Iron deposition and fat accumulation in dimethylnitrosamine-induced liver fibrosis in rat. World J. Gastroentero. 13(14), 2061–2065 (2007).

26 George J, Chandrakasan G. Biochemical abnormalities during the progression of hepatic fibrosis induced by dimethylnitrosamine. Clin. Biochem. 33(7), 563–570 (2000).

27 Su LJ, Hsu SL, Yang JS, Tseng HH, Huang SF, Huang CY. Global gene expression profiling of dimethylnitrosamine-induced liver fibrosis: from pathological and biochemical data to microarray ana lysis. Gene Expr. 13(2), 107–132 (2006).

28 Fico ME, Motyl T, Milner JA. Species comparison of the influence of ammonia on orotic acid and urea biosynthesis in liver. J. Nutr. 114(3), 613–621 (1984).

29 Biller J. Interface of Neurology and Internal Medicine. Lippincott Williams & Wilkins, Philadelphia, PA, USA (2008).

30 Ristoff E, Larsson A. Glutathione Synthetase Deficiency. Orphanet, Paris, France (2003).

31 Delaney J, Neville WA, Swain A, Miles A, Leonard MS, Waterfield CJ. Phenyl acetyl-glycine, a putative biomarker of phospholipidosis: its origins and relevance to phospholipid accumulation using amiodarone treated rats as a model. Biomarkers 9(3), 271–290 (2004).

32 Verhoeven NM, Huck JH, Roos B et al. Transaldolase deficiency. liver cirrhosis associated with a new inborn error in the pentose phosphate pathway. Am. J. Hum. Genet. 68(5), 1086–1092 (2001).

�� Website101 Introduction to metabolic biochemistry,

nucleic acid metabolism. www.whatislife.com

Metabolite alteration in dimethylnitrosamine-induced liver fibrosis by GC–MS | ReseaRch aRticle

www.future-science.com 51future science group