Research Article Received: 13 September 2013,
Revised: 8 November 2013,
Accepted: 8 November 2013
Published online in Wiley Online Library: 30 January 2014
(wileyonlinelibrary.com) DOI 10.1002/jat.2970
Characterization of the biochemical effects of naphthalene on the mouse respiratory system using NMR-based metabolomics Jia-Huei Honga, Wen-Ching Leeb, Yu-Ming Hsuc, Hao-Jan Lianga, Cho-Hua Wand, Chung-Liang Chienb and Ching-Yu Lina* ABSTRACT: Naphthalene is a ubiquitous environmental pollutant to which humans are exposed. Previous studies have demonstrated that naphthalene causes bronchiolar epithelial necrosis in the mouse distal airway, after parenteral administration. In this study, metabolic variations in the bronchoalveolar lavage fluid (BALF) and the lung tissues of naphthalene-treated mice and controls were examined using nuclear magnetic resonance (NMR)-based metabolomics to identify the toxic mechanism. Male ICR mice were treated with naphthalene [0, 50, 100 and 200 mg kg–1, intraperitoneally (i.p.)]. After 24 h, BALF and lung tissues were collected and prepared for 1H and J-resolved (JRES) NMR analysis after principal component analysis (PCA). PCA modeling of p-JRES spectra from the BALF, as well as hydrophilic and hydrophobic lung metabolites, enabled the high-dose group to be discriminated from the control group; increased levels of isopropanol, ethane, and acetone and lower levels of ethanol, acetate, formate, and glycerophosphocholine were detected in the BALF of mice treated with higher doses of naphthalene. Furthermore, increased isopropanol and phosphorylcholine-containing lipid levels and decreased succinate and glutamine levels were discovered in the lungs of naphthalene-exposed mice. These metabolic changes may be related to lipid peroxidation, disruptions of membrane components and imbalanced energy supply, and these results may partially explain the loss of cell membrane integrity in the airway epithelial cells of naphthalene-treated mice. We conclude that NMR-based metabolomic studies on BALF and lung tissues are a powerful tool to understand the mechanisms underlying respiratory toxicity. Copyright © 2014 John Wiley & Sons, Ltd. Additional supporting information may be found in the online version of this article at the publisher’s web-site. Keywords: metabolomics; nuclear magnetic resonance; naphthalene; respiratory toxicity; bronchoalveolar lavage fluid
Introduction
J. Appl. Toxicol. 2014; 34: 1379–1388
*Correspondence to: Ching-Yu Lin, Rm. 723, 7 F., No.17, Xuzhou Rd., Zhongzheng Dist., Taipei City 10055, Taiwan, Republic of China. E-mail: [email protected] a Institute of Environmental Health, College of Public Health, National Taiwan University, Taipei 10055, Taiwan, Republic of China b
Graduate Institute of Anatomy and Cell Biology, College of Medicine, National Taiwan University, Taipei 100, Taiwan, Republic of China
c
Laboratory Animal Center, College of Medicine, National Taiwan University, Taipei 100, Taiwan, Republic of China
d
Graduate Institute of Molecular and Comparative Pathobiology, School of Veterinary Medicine, National Taiwan University, Taipei 106, Taiwan, Republic of China
Copyright © 2014 John Wiley & Sons, Ltd.
1379
Lung disease is one of the most prevalent and fatal diseases and is among the top 10 causes of death worldwide. Epithelial non-ciliated (Clara) cells are a prominent cell type (Plopper et al., 1991) in the airway and are the principal site for cytochrome P450-dependent metabolism in the murine airway (Devereux et al., 1989; Plopper et al., 1987). Clara cells are susceptible to environmental toxicants [e.g. polycyclic aromatic hydrocarbons (PAHs)] and their reactive derivatives, which are formed via biological activation. PAHs are a family of carcinogenic and mutagenic compounds, 16 of which are listed by the US Environmental Protection Agency as highly potent or suspected carcinogens. Naphthalene, the most common PAH, is present in industrial waste, vehicle emissions, cigarette smoke, etc., and naphthalene accounts for 95% of total volatile indoor air PAHs derived from naphthalene-containing moth repellents, open fires or tobacco smoking (Ohura et al., 2004). Previous studies have demonstrated that naphthalene can be detected in human fat tissues (Stanley, 1986), breast milk (Pellizzari et al., 1982) and liver tissue (Tingle et al., 1993), illustrating the high frequency of exposure. Naphthalene treatment has been shown to produce dosedependent cell injury to the respiratory systems of mice [0, 50, 100, 200 and 300 mg kg–1, intraperitoneally (i.p.)] (Plopper et al., 1992). Murine distal airways are the most susceptible to
naphthalene (i.p.), demonstrating site- and species-selective toxicity. Naphthalene toxicity has been correlated with the rate and stereoselectivity of naphthalene metabolism (Buckpitt et al., 1995), glutathione depletion (West et al., 2000) and protein adduct formation (Cho et al., 1995; Lin et al., 2006a). The rate of naphthalene metabolism in the mouse airways is significantly higher than that in the airways of rats or hamsters. Moreover, preferential cytochrome P450-catalyzed formation of naphthalene 1R, 2S-oxide is associated with species- and siteselective naphthalene toxicity (Buckpitt et al., 1992).
J.–H. Hong et al. Reactive naphthalene metabolites that are covalently bound to proteins have been associated with naphthalene toxicity in the respiratory system (Lin et al., 2005; Lin et al., 2006a), and a wide variety of proteins have been found to be adducted with reactive naphthalene metabolites in different species. However, researchers studying proteins often require additional information on low-molecular-weight metabolites to understand a given biological response, as metabolites are crucial to cellular regulatory processes (Lin et al., 2006b). The aim of the present study was to understand the final biological steps involved in the cellular effects of naphthalene toxicity using a metabolomic approach in the mouse respiratory system. Metabolomics comprises the measurement of endogenous metabolites, including amino acids, nucleic acid precursors, lipids, and degradation products of chemical intermediates in catabolism and biosynthesis. The advantage of metabolomics is that it provides the most functional measure of cellular status and can help to describe an organism’s phenotype. Previous studies have successfully illustrated the use of metabolomic approaches to characterize the effects of 1-nitronaphthalene on the respiratory system of rats (Azmi et al., 2005). In the present study, the metabolic effects of naphthalene treatment on the respiratory tract were screened in the BALF and dissected lung tissues of treated mice. Moreover, Clara cell histopathology results were used to record naphthalene-induced cellular injury and correlate this injury with metabolic changes. We sought to correlate metabolic changes to the observed phenotypes and also aimed to identify the potential mechanisms underlying the pathogenesis of naphthalene exposure.
Materials and Methods Animal Handling
1380
Lung pathogen-free male ICR mice (7–8 weeks old, 33–35 g; BioLASCO Co., Taiwan, China) were grouped randomly and acclimatized for at least 5 days prior to use in the campus animal center (controlled 55 ± 10% relative humidity, 25 °C). The animals were maintained on an artificial 12-h light:12-h dark photoperiod with food and water provided ad libitum. To ensure an accurate dose of naphthalene treatment and be able to compare our findings to previous publications, mice were treated via i.p. injection. Mice for use in the metabolomic study received naphthalene (i.p.) in olive oil vehicle (50, 100 or 200 mg kg–1) or olive oil only at a concentration of 10 ml kg–1 body weight. There were six replicates in each group. To examine morphological changes at a possible transitional dose, a 75-mg kg–1 dose was added to the study; therefore, naphthalene at doses of 0, 50, 75, 100 and 200 mg kg–1 were administered (i.p.) in the histopathology studies. At 24 h post-treatment, the animals were sacrificed via cervical dislocation (metabolomics, to avoid chemical effects on the tracheal epithelium) or via overdose of sodium pentobarbital followed by exsanguination (histopathology). BALF and lung tissues were collected for nuclear magnetic resonance (NMR) analysis (metabolomics), while the chests of the mice were opened to infuse a fixative infusion into the lungs (histopathology). All procedures using animals were reviewed and approved by the National Taiwan University’s Institutional Animal Care and Use Committee.
wileyonlinelibrary.com/journal/jat
Sample Collection and Preparation for NMR-based Metabolomics To obtain BALF for use in the metabolomic study, mouse tracheas were cannulated with intravenous (i.v.) catheters in situ. One half of 1 ml of phosphate-buffered saline (PBS, room temperature) was gently introduced through the catheter to fill the lungs and was then withdrawn steadily. Fluids obtained from two repeated washes were combined, snap frozen in liquid nitrogen, freeze dried and stored at 80 °C until use in the metabolomic analysis. After BALF collection, intact lung tissues were immediately collected, rinsed with PBS, flash frozen in liquid nitrogen and stored at 80 °C. The frozen tissue samples were then manually pulverized with a liquid nitrogen-cooled mortar and pestle and then lyophilized. The homogeneous dry tissue powder was weighed, extracted with 1.16 ml g–1 (dry mass) methanol–chloroform–water (2:2:1.8) and vortexed (Bligh and Dyer, 1959; Lin et al., 2007). The hydrophilic (0.4 ml) and hydrophobic (0.2 ml) phases were collected separately and then dried using a speed vacuum for NMR analysis.
NMR Spectroscopy Prior to NMR analysis, BALF and hydrophilic lung metabolites were resuspended in 600 μl of D2O containing sodium phosphate buffer (0.1 M, pH 7.4) and the internal standard 3-trimethylsilyl-2,2,3, 3-d4-propionate (TMSP). Hydrophobic lung metabolites were resuspended in 600 μl of d-chloroform containing 0.03% (vol/vol) tetramethylsilane (TMS). NMR spectra were acquired on Bruker Avance-600 (for BALF) and Avance-500 (for lung extracts) spectrometers equipped with a 5-mm TXI CryoProbe (Bruker BioSpin) and operated at a 1H frequency (600.13 MHz for BALF; 500.13 MHz for lung extracts) at a temperature of 300 K at the Core Facility for Protein Structural Analysis, which is supported by the National Core Facility Program for Biotechnology in Academia Sinica, Taiwan. Six spectral information sets were acquired, including 1H and J-resolved (JRES) NMR spectra for the (i) BALF, (ii) lung hydrophilic extracts and (iii) lung hydrophobic extracts. 1 H NMR spectra for the BALF and the hydrophilic lung metabolites were acquired with presaturation of water resonance using the pulse sequence (relaxation delay: -90°-t-90°-tm-90°acquired-free induction decay). For each sample, 32 k data points were collected into a spectrum with 20 ppm width, a relaxation delay of 2.0 s and numbers of scan of 128. The short delay (t), mixing time (tm), and acquisition times are ~3 us, 150 ms, 1.36 s for the BALF; and ~6.5 us, 100–150 ms, 1.63 s for hydrophilic lung metabolites (Beckonert et al., 2007; Viant et al., 2009). Lung hydrophobic metabolites were acquired into 1H NMR spectra using the pulse program zg without water peak presaturation. Other parameters were measured in the same manner as that used to analyze the hydrophilic metabolite extracts, with the exception that a 12-ppm spectral width and 2.72-s acquisition time were used. JRES NMR spectra were acquired using the pulse sequence (relaxation delay-90°-t1-180°-t1-acquire free induction delay, with water suppression during relaxation delay), with 16 k data points on the F2 frequency axis and 40 data points on the F1 frequency axis and a relaxation delay of 2 s. The number of scans was 8, the number of dummy scans was 16 and the spectral width on the F2 axis was 6000 Hz. The spectral widths on the F1 axis for the BALF and the lung metabolites were 78 and 65 Hz, respectively (Lin et al., 2007; Viant, 2003).
Copyright © 2014 John Wiley & Sons, Ltd.
J. Appl. Toxicol. 2014; 34: 1379–1388
NMR-based metabolomics to study naphthalene-induced lung toxicity Spectral Pre-processing A line-broadening function (0.3 Hz), a sine function multiplication and one level of zero filling were applied to all 1H spectra prior to Fourier transformation. The spectra were phased and baseline-corrected by referencing them to TMSP/TMS (δ = 0 ppm). JRES data were zero filled to 16 k data points on the F2 axis and up to 128 increments on the F1 axis. Apodization of the JRES data was conducted using an unshifted sine function with 0.3 Hz of line broadening in both dimensions prior to Fourier transformation. Data were tilted (45°), symmetrized according to the center of the F1 axis and skyline-projected (Fonville et al., 2010). The JRES projections (p-JRES) were referenced and baseline-corrected as outlined above. The 1H and p-JRES spectral files were input using a custom-written ProMetab software (Version 1) within MATLAB (Version 7.8; MathWorks, Natick, MA, USA) for spectral pre-processing (Viant, 2003). Each 1H spectrum was binned from δ 0.2 to δ 10.0 ppm (to δ 9.0 for BALF spectra) in steps of 0.005 ppm, which maintained the data resolution while increasing the accuracy in chemical shift referencing. The peak area within each bin was then integrated. After removal of the TMSP/TMS, water or chloroform regions, the remaining spectral data were subjected to total spectra area normalization. Then, the normalized data were log transformed, and the variables were mean-centered prior to multivariate analysis (Beckonert et al., 2007; Dieterle et al., 2006; Fonville et al., 2010; Ludwig and Viant, 2010; Viant et al., 2003, 2006). There were six datasets in total, including 1H and p-JRES spectra for the BALF and the lung hydrophilic and hydrophobic metabolomes.
Results NMR Spectroscopy of Metabolites from Mouse BALF and Lung 1
Multivariate Statistical Analyses The processed 1H and p-JRES data were subjected to PCA using PLS_Toolbox (Version 3.5; Eigenvector Research, Manson, WA, USA) within MATLAB. Two principal components (PCs) dominant for data separation were chosen for the axes of score plots. Complementary loading plots generated by the axes of the PC and other variables were used for biomarker discovery by referencing the variables to the chemical shifts. Metabolite Identification and Statistical Analyses Corresponding loading plots from score plots were taken to identify the metabolites that were perturbed among the pattern recognition models. Peaks from 1H and p-JRES NMR spectra were referenced to previously published papers (Nicholson et al., 1995; Oostendorp et al., 2006), software (Chenomx NMR Suite, Professional Edition, Version 6.1; Chenomx Inc., NW Edmonton, AB, Canada) and databases (The human metabolome database: http://www.hmdb.ca). Normalized peak areas from all well-resolved peaks and identified metabolites were further analyzed using one-way analysis of variance (ANOVA) to confirm the significance of changes observed after naphthalene exposure. When statistical significance (P < 0.05) was reached among the groups, Dunnett’s test was then used to determine the statistical significance (P < 0.05) of all pairwise comparisons between groups. Lung Fixation for Histopathology
H, 2D JRES and p-JRES NMR spectra of the BALF (Supplementary Figs 1–3) and lung hydrophilic (Supplementary Figs 4–6) and hydrophobic metabolomes (Supplementary Figs 7–9) were obtained. Advantageous low spectral congested p-JRES peaks with flat baselines were taken for peak integration and statistical data analyses. Major hydrophilic metabolites were assigned by referring to tabulated chemical shifts (1H and p-JRES) and peak multiplicity (1H) (Azmi et al., 2005; Nicholson et al., 1995; Wei et al., 2009) and were confirmed using Chenomx NMR Suite software. A significantly elevated BALF metabolite with a singlet peak in the 1H spectra (δ 1.25 ppm) was further confirmed to be ethane, the only two-carbon alkane, using selected TOtal Correlation SpectroscopY (TOCSY) analysis. Major hydrophobic metabolites were assigned by referring to a tabulated chemical shift (1H and p-JRES) and peak multiplicity (1H) (Ong et al., 2009; Oostendorp et al., 2006; Waters et al., 2002). Effects of Naphthalene on the BALF Metabolome p-JRES NMR spectral data of BALF metabolomes obtained from the control and naphthalene-treated mice (50, 100 and 200 mg kg–1) were analyzed using unsupervised PCA. PCA reduces dimensionality, summarizes the similarities and differences of the dataset, and yields a simple graphical output called the scores plot (Fig. 1). As shown in Fig. 1, the scores plot was composed of PC 1 (42.60%) and 2 (15.35%), accounting for 57.95 % of the total variance within the BALF metabolome. In a PCA scores plot, each data point denotes a sample. The lower dose data (sample) points (olive oil only and 50 mg kg–1 naphthalene) clearly clustered and were separated from the higher dose data points (100 and 200 mg kg–1 naphthalene) based on PC1, reflecting dosedependent effects on the BALF metabolome. The complementary
Copyright © 2014 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jat
1381
The animals were sacrificed 24 h after the naphthalene treatments. The tracheas were then cannulated with an i.v. catheter in situ, and
J. Appl. Toxicol. 2014; 34: 1379–1388
the thoraxes were opened by diaphragmatic incision (Plopper et al., 1992). Tracheal infusion without removing the lungs from the chest was used to maintain their in situ shape and the relationship between the lobes. The collapsed lungs were prefixed with a glutaraldehyde (0.9 vol./vol. %)/paraformaldehyde (1.0 wt./vol. %) mixture in 0.1 mM cacodylate buffer (330 mOsmol, pH 7.4) via tracheal infusion and were maintained for 24 h at 30 cm height fluid pressure (Dungworth et al., 1976; Plopper et al., 1992; Van Winkle et al., 1999). The lungs were removed from the chest 24 h later, and the terminal bronchioles were sectioned for inclusion in a 1-mm2 block post-fixation. Tissue blocks were handled according to the large block embedding method (Plopper, 1990), which enables highresolution light microscopy and transmission electron microscopy to be conducted on the same specimen. Tissues were post-fixed in 1% osmium tetroxide, dehydrated using ethanol, infiltrated with propylene oxide (1,2-propylene oxide; Merck, Darmstadt, Germany) and embedded in Araldite 502 (modified bisphenol A epoxy; Electron Microscopy Science, Hatfield, PA, USA). The blocks were then sectioned using a Diatome (Diatome diamond knife; Diatome Co., The Patch, Victoria Australia) and an Ultracut (Reicher Ultracult; Leica Microsystems, Milton Keynes, UK) into 1- to 2-micron thick sections, and then stained with toluidine blue for light and electron microscopy (Van Winkle et al., 1999).
J.–H. Hong et al.
Figure 1. PCA scores plot from the NMR spectra of BALF metabolomes from mice treated with various doses of naphthalene. Samples separated in the first component (PC1, 42.60%) are illustrated for the control group (circle) and the groups treated with 50 (square), 100 (rhombus) and 200 –1 (triangle) mg kg naphthalene.
PC1 loadings plot revealed the metabolites that contributed to clustering (Fig. 2). Peaks pointing up (positive values in the PC1 axis) in the loadings plot indicate that the corresponding metabolites from the samples located on the positive side of the PC1 axis of the scores plot are higher in concentration than the samples on the negative side of the same axis. The peaks contributing to clustering in the scores plot were further analyzed using a traditional method (ANOVA) (Table 1). Increased amounts (P < 0.05) of isopropanol, ethane, and acetone and decreased amounts (P < 0.05) of ethanol, acetate, glycerophosphocholine (GPC), and formate (Table 1) were detected in the BALF of mice treated with 100 and 200 mg kg–1 naphthalene. Effects of Naphthalene on the Lung Metabolome The total numbers of lung samples analyzed by NMR were 6, 6, 5 and 5 for the 0, 50, 100 and 200 mg kg–1 naphthalene-treated groups, respectively. Missing samples were a result of instrumental problems during sample preparation. The metabolomes of hydrophilic extracts from lung tissues were separated by PCA based on similarities in the p-JRES NMR spectra. The scores plot from the first two PCs explained 43.58% of the variance in
the lung hydrophilic metabolome (Fig. 3) and demonstrated a trend of shifting for the four groups treated with various doses of naphthalene. The metabolites contributing to the shifting in the corresponding loadings plot (Fig. 4) were identified and further analyzed using a traditional statistical method (ANOVA) (Table 2). Increased amounts (P < 0.05) of isopropanol and decreased amounts (P < 0.05) of glutamine and succinate were observed in the groups exposed to 100 and 200 mg kg–1 naphthalene. An unassigned singlet at δ 8.27 was found to be significantly different (P < 0.05) among the groups. However, these differences were not found to be significant in the Dunnett’s test (Table 2). The p-JRES NMR spectra of lung hydrophobic metabolomes were also analyzed using PCA. The first two PCs covered 62.61 % of the total variance (Fig. 5). Samples from the control and higher dose groups (100 and 200 mg naphthalene) that had clustered were partially separated along PC1 (37.70%). The complementary loadings plot identified the metabolites contributing to the trend (Fig. 6); significantly increased amounts of (CH2)n from fatty acyl chains and N(CH3)3 from phosphorylcholinecontaining lipids were found in mice treated with 200 mg kg–1 naphthalene (P < 0.05), whereas cholesterol levels were decreased in mice treated with 50, 100 and 200 mg kg–1 naphthalene (P < 0.001) (Table 3). Effects of Naphthalene on the Morphology of Mouse Distal Airways Light microscope images provided good resolution of ciliated and non-ciliated (Clara) epithelial cells (Supplementary Fig. 10). Swollen and vacuolated bronchiolar epithelial cells were observed after naphthalene exposure in a dose-dependent manner. The control mouse lung specimens did not demonstrate any altered cells in the terminal bronchioles. In addition, no swollen and vacuolated Clara cells were found in the terminal bronchioles of mice treated with 50 mg kg–1 naphthalene, although there were a few swollen, vacuolated Clara cells in the terminal bronchioles of mice treated with 75 mg kg–1 naphthalene. In the 100 mg kg–1 naphthalene-treated mouse specimens, the majority of the epithelial cells, including ciliated and Clara cells, were swollen and vacuolated (Supplementary Fig. 10).
1382
Figure 2. PCA loadings plot of PC1 (42.60%) from the NMR spectra of BALF metabolomes from mice treated with various doses of naphthalene. Key: 1: Isopropanol, 2: Ethanol, 3: Ethane, 4: Lactate, 5: Acetate, 6: Unassigned, 7: Acetone, 8: Succinate, 9: Creatine, 10: Choline, 11: Taurine, 12: Glycerophosphocholine, 13: Glycerol, 14: Unassigned singlet, 15: Creatine phosphate, and 16: Formate.
wileyonlinelibrary.com/journal/jat
Copyright © 2014 John Wiley & Sons, Ltd.
J. Appl. Toxicol. 2014; 34: 1379–1388
NMR-based metabolomics to study naphthalene-induced lung toxicity Table 1. The resonance assignments of metabolites from mouse BALF and changes of those from mice treated with various doses of naphthalene (i.p.) after 24 h Changes relative to control Key
Metabolite
ppm (multiplicity)
1 2 3 4 5 6 7 8 9 10 12 14 15 16
Isopropanol Ethanol Ethane Lactate Acetate Unassigned Acetone Succinate Creatine Choline Glycerophosphocholine Unassigned Creatine phosphate Formate
1.18(d)c, 4.01(m) 1.19(t)c, 3.66(q) 1.25(s)c 1.33(d)c, 4.11(q) 1.92(s)c 1.98c 2.23(s)c 2.41(s)c 3.04(s)c 3.20(s)c 3.36(s)c 3.60(s)c 3.94(s)c, 3.03(s) 8.46(s)c
a
NA50
NA100
NA200
0.10%↓ 17.30%↑ 11.34%↓ 27.37%↑ 11.76%↑ 59.80%↓ 28.45%↓ 10.91%↑ 10.26%↓ 9.47%↑ 3.64%↑ 9.80%↑ 8.19%↑ 20.02%↑
70.73%↑ 80.63%↓ 422.07%↑ 1.95%↑ 73.06%↓ 85.06%↓ 436%↑ 47.28%↓ 35.12%↓ 11.43%↓ 87.37%↓ 96.41%↓ 68.04%↓ 83.03%↓
60.07%↑ 71.41%↓ 412.51%↑ 26.06%↑ 76.04%↓ 99.37%↓ 742%↑ 26.12%↓ 16.13%↑ 1.49%↑ 84.62%↓ 92.15%↓ 21.48%↓ 81.71%↓
ANOVAb (P-value)
Dunnett’s test (P < 0.05)
0.0006** 0.0002**
Revised: 8 November 2013,
Accepted: 8 November 2013
Published online in Wiley Online Library: 30 January 2014
(wileyonlinelibrary.com) DOI 10.1002/jat.2970
Characterization of the biochemical effects of naphthalene on the mouse respiratory system using NMR-based metabolomics Jia-Huei Honga, Wen-Ching Leeb, Yu-Ming Hsuc, Hao-Jan Lianga, Cho-Hua Wand, Chung-Liang Chienb and Ching-Yu Lina* ABSTRACT: Naphthalene is a ubiquitous environmental pollutant to which humans are exposed. Previous studies have demonstrated that naphthalene causes bronchiolar epithelial necrosis in the mouse distal airway, after parenteral administration. In this study, metabolic variations in the bronchoalveolar lavage fluid (BALF) and the lung tissues of naphthalene-treated mice and controls were examined using nuclear magnetic resonance (NMR)-based metabolomics to identify the toxic mechanism. Male ICR mice were treated with naphthalene [0, 50, 100 and 200 mg kg–1, intraperitoneally (i.p.)]. After 24 h, BALF and lung tissues were collected and prepared for 1H and J-resolved (JRES) NMR analysis after principal component analysis (PCA). PCA modeling of p-JRES spectra from the BALF, as well as hydrophilic and hydrophobic lung metabolites, enabled the high-dose group to be discriminated from the control group; increased levels of isopropanol, ethane, and acetone and lower levels of ethanol, acetate, formate, and glycerophosphocholine were detected in the BALF of mice treated with higher doses of naphthalene. Furthermore, increased isopropanol and phosphorylcholine-containing lipid levels and decreased succinate and glutamine levels were discovered in the lungs of naphthalene-exposed mice. These metabolic changes may be related to lipid peroxidation, disruptions of membrane components and imbalanced energy supply, and these results may partially explain the loss of cell membrane integrity in the airway epithelial cells of naphthalene-treated mice. We conclude that NMR-based metabolomic studies on BALF and lung tissues are a powerful tool to understand the mechanisms underlying respiratory toxicity. Copyright © 2014 John Wiley & Sons, Ltd. Additional supporting information may be found in the online version of this article at the publisher’s web-site. Keywords: metabolomics; nuclear magnetic resonance; naphthalene; respiratory toxicity; bronchoalveolar lavage fluid
Introduction
J. Appl. Toxicol. 2014; 34: 1379–1388
*Correspondence to: Ching-Yu Lin, Rm. 723, 7 F., No.17, Xuzhou Rd., Zhongzheng Dist., Taipei City 10055, Taiwan, Republic of China. E-mail: [email protected] a Institute of Environmental Health, College of Public Health, National Taiwan University, Taipei 10055, Taiwan, Republic of China b
Graduate Institute of Anatomy and Cell Biology, College of Medicine, National Taiwan University, Taipei 100, Taiwan, Republic of China
c
Laboratory Animal Center, College of Medicine, National Taiwan University, Taipei 100, Taiwan, Republic of China
d
Graduate Institute of Molecular and Comparative Pathobiology, School of Veterinary Medicine, National Taiwan University, Taipei 106, Taiwan, Republic of China
Copyright © 2014 John Wiley & Sons, Ltd.
1379
Lung disease is one of the most prevalent and fatal diseases and is among the top 10 causes of death worldwide. Epithelial non-ciliated (Clara) cells are a prominent cell type (Plopper et al., 1991) in the airway and are the principal site for cytochrome P450-dependent metabolism in the murine airway (Devereux et al., 1989; Plopper et al., 1987). Clara cells are susceptible to environmental toxicants [e.g. polycyclic aromatic hydrocarbons (PAHs)] and their reactive derivatives, which are formed via biological activation. PAHs are a family of carcinogenic and mutagenic compounds, 16 of which are listed by the US Environmental Protection Agency as highly potent or suspected carcinogens. Naphthalene, the most common PAH, is present in industrial waste, vehicle emissions, cigarette smoke, etc., and naphthalene accounts for 95% of total volatile indoor air PAHs derived from naphthalene-containing moth repellents, open fires or tobacco smoking (Ohura et al., 2004). Previous studies have demonstrated that naphthalene can be detected in human fat tissues (Stanley, 1986), breast milk (Pellizzari et al., 1982) and liver tissue (Tingle et al., 1993), illustrating the high frequency of exposure. Naphthalene treatment has been shown to produce dosedependent cell injury to the respiratory systems of mice [0, 50, 100, 200 and 300 mg kg–1, intraperitoneally (i.p.)] (Plopper et al., 1992). Murine distal airways are the most susceptible to
naphthalene (i.p.), demonstrating site- and species-selective toxicity. Naphthalene toxicity has been correlated with the rate and stereoselectivity of naphthalene metabolism (Buckpitt et al., 1995), glutathione depletion (West et al., 2000) and protein adduct formation (Cho et al., 1995; Lin et al., 2006a). The rate of naphthalene metabolism in the mouse airways is significantly higher than that in the airways of rats or hamsters. Moreover, preferential cytochrome P450-catalyzed formation of naphthalene 1R, 2S-oxide is associated with species- and siteselective naphthalene toxicity (Buckpitt et al., 1992).
J.–H. Hong et al. Reactive naphthalene metabolites that are covalently bound to proteins have been associated with naphthalene toxicity in the respiratory system (Lin et al., 2005; Lin et al., 2006a), and a wide variety of proteins have been found to be adducted with reactive naphthalene metabolites in different species. However, researchers studying proteins often require additional information on low-molecular-weight metabolites to understand a given biological response, as metabolites are crucial to cellular regulatory processes (Lin et al., 2006b). The aim of the present study was to understand the final biological steps involved in the cellular effects of naphthalene toxicity using a metabolomic approach in the mouse respiratory system. Metabolomics comprises the measurement of endogenous metabolites, including amino acids, nucleic acid precursors, lipids, and degradation products of chemical intermediates in catabolism and biosynthesis. The advantage of metabolomics is that it provides the most functional measure of cellular status and can help to describe an organism’s phenotype. Previous studies have successfully illustrated the use of metabolomic approaches to characterize the effects of 1-nitronaphthalene on the respiratory system of rats (Azmi et al., 2005). In the present study, the metabolic effects of naphthalene treatment on the respiratory tract were screened in the BALF and dissected lung tissues of treated mice. Moreover, Clara cell histopathology results were used to record naphthalene-induced cellular injury and correlate this injury with metabolic changes. We sought to correlate metabolic changes to the observed phenotypes and also aimed to identify the potential mechanisms underlying the pathogenesis of naphthalene exposure.
Materials and Methods Animal Handling
1380
Lung pathogen-free male ICR mice (7–8 weeks old, 33–35 g; BioLASCO Co., Taiwan, China) were grouped randomly and acclimatized for at least 5 days prior to use in the campus animal center (controlled 55 ± 10% relative humidity, 25 °C). The animals were maintained on an artificial 12-h light:12-h dark photoperiod with food and water provided ad libitum. To ensure an accurate dose of naphthalene treatment and be able to compare our findings to previous publications, mice were treated via i.p. injection. Mice for use in the metabolomic study received naphthalene (i.p.) in olive oil vehicle (50, 100 or 200 mg kg–1) or olive oil only at a concentration of 10 ml kg–1 body weight. There were six replicates in each group. To examine morphological changes at a possible transitional dose, a 75-mg kg–1 dose was added to the study; therefore, naphthalene at doses of 0, 50, 75, 100 and 200 mg kg–1 were administered (i.p.) in the histopathology studies. At 24 h post-treatment, the animals were sacrificed via cervical dislocation (metabolomics, to avoid chemical effects on the tracheal epithelium) or via overdose of sodium pentobarbital followed by exsanguination (histopathology). BALF and lung tissues were collected for nuclear magnetic resonance (NMR) analysis (metabolomics), while the chests of the mice were opened to infuse a fixative infusion into the lungs (histopathology). All procedures using animals were reviewed and approved by the National Taiwan University’s Institutional Animal Care and Use Committee.
wileyonlinelibrary.com/journal/jat
Sample Collection and Preparation for NMR-based Metabolomics To obtain BALF for use in the metabolomic study, mouse tracheas were cannulated with intravenous (i.v.) catheters in situ. One half of 1 ml of phosphate-buffered saline (PBS, room temperature) was gently introduced through the catheter to fill the lungs and was then withdrawn steadily. Fluids obtained from two repeated washes were combined, snap frozen in liquid nitrogen, freeze dried and stored at 80 °C until use in the metabolomic analysis. After BALF collection, intact lung tissues were immediately collected, rinsed with PBS, flash frozen in liquid nitrogen and stored at 80 °C. The frozen tissue samples were then manually pulverized with a liquid nitrogen-cooled mortar and pestle and then lyophilized. The homogeneous dry tissue powder was weighed, extracted with 1.16 ml g–1 (dry mass) methanol–chloroform–water (2:2:1.8) and vortexed (Bligh and Dyer, 1959; Lin et al., 2007). The hydrophilic (0.4 ml) and hydrophobic (0.2 ml) phases were collected separately and then dried using a speed vacuum for NMR analysis.
NMR Spectroscopy Prior to NMR analysis, BALF and hydrophilic lung metabolites were resuspended in 600 μl of D2O containing sodium phosphate buffer (0.1 M, pH 7.4) and the internal standard 3-trimethylsilyl-2,2,3, 3-d4-propionate (TMSP). Hydrophobic lung metabolites were resuspended in 600 μl of d-chloroform containing 0.03% (vol/vol) tetramethylsilane (TMS). NMR spectra were acquired on Bruker Avance-600 (for BALF) and Avance-500 (for lung extracts) spectrometers equipped with a 5-mm TXI CryoProbe (Bruker BioSpin) and operated at a 1H frequency (600.13 MHz for BALF; 500.13 MHz for lung extracts) at a temperature of 300 K at the Core Facility for Protein Structural Analysis, which is supported by the National Core Facility Program for Biotechnology in Academia Sinica, Taiwan. Six spectral information sets were acquired, including 1H and J-resolved (JRES) NMR spectra for the (i) BALF, (ii) lung hydrophilic extracts and (iii) lung hydrophobic extracts. 1 H NMR spectra for the BALF and the hydrophilic lung metabolites were acquired with presaturation of water resonance using the pulse sequence (relaxation delay: -90°-t-90°-tm-90°acquired-free induction decay). For each sample, 32 k data points were collected into a spectrum with 20 ppm width, a relaxation delay of 2.0 s and numbers of scan of 128. The short delay (t), mixing time (tm), and acquisition times are ~3 us, 150 ms, 1.36 s for the BALF; and ~6.5 us, 100–150 ms, 1.63 s for hydrophilic lung metabolites (Beckonert et al., 2007; Viant et al., 2009). Lung hydrophobic metabolites were acquired into 1H NMR spectra using the pulse program zg without water peak presaturation. Other parameters were measured in the same manner as that used to analyze the hydrophilic metabolite extracts, with the exception that a 12-ppm spectral width and 2.72-s acquisition time were used. JRES NMR spectra were acquired using the pulse sequence (relaxation delay-90°-t1-180°-t1-acquire free induction delay, with water suppression during relaxation delay), with 16 k data points on the F2 frequency axis and 40 data points on the F1 frequency axis and a relaxation delay of 2 s. The number of scans was 8, the number of dummy scans was 16 and the spectral width on the F2 axis was 6000 Hz. The spectral widths on the F1 axis for the BALF and the lung metabolites were 78 and 65 Hz, respectively (Lin et al., 2007; Viant, 2003).
Copyright © 2014 John Wiley & Sons, Ltd.
J. Appl. Toxicol. 2014; 34: 1379–1388
NMR-based metabolomics to study naphthalene-induced lung toxicity Spectral Pre-processing A line-broadening function (0.3 Hz), a sine function multiplication and one level of zero filling were applied to all 1H spectra prior to Fourier transformation. The spectra were phased and baseline-corrected by referencing them to TMSP/TMS (δ = 0 ppm). JRES data were zero filled to 16 k data points on the F2 axis and up to 128 increments on the F1 axis. Apodization of the JRES data was conducted using an unshifted sine function with 0.3 Hz of line broadening in both dimensions prior to Fourier transformation. Data were tilted (45°), symmetrized according to the center of the F1 axis and skyline-projected (Fonville et al., 2010). The JRES projections (p-JRES) were referenced and baseline-corrected as outlined above. The 1H and p-JRES spectral files were input using a custom-written ProMetab software (Version 1) within MATLAB (Version 7.8; MathWorks, Natick, MA, USA) for spectral pre-processing (Viant, 2003). Each 1H spectrum was binned from δ 0.2 to δ 10.0 ppm (to δ 9.0 for BALF spectra) in steps of 0.005 ppm, which maintained the data resolution while increasing the accuracy in chemical shift referencing. The peak area within each bin was then integrated. After removal of the TMSP/TMS, water or chloroform regions, the remaining spectral data were subjected to total spectra area normalization. Then, the normalized data were log transformed, and the variables were mean-centered prior to multivariate analysis (Beckonert et al., 2007; Dieterle et al., 2006; Fonville et al., 2010; Ludwig and Viant, 2010; Viant et al., 2003, 2006). There were six datasets in total, including 1H and p-JRES spectra for the BALF and the lung hydrophilic and hydrophobic metabolomes.
Results NMR Spectroscopy of Metabolites from Mouse BALF and Lung 1
Multivariate Statistical Analyses The processed 1H and p-JRES data were subjected to PCA using PLS_Toolbox (Version 3.5; Eigenvector Research, Manson, WA, USA) within MATLAB. Two principal components (PCs) dominant for data separation were chosen for the axes of score plots. Complementary loading plots generated by the axes of the PC and other variables were used for biomarker discovery by referencing the variables to the chemical shifts. Metabolite Identification and Statistical Analyses Corresponding loading plots from score plots were taken to identify the metabolites that were perturbed among the pattern recognition models. Peaks from 1H and p-JRES NMR spectra were referenced to previously published papers (Nicholson et al., 1995; Oostendorp et al., 2006), software (Chenomx NMR Suite, Professional Edition, Version 6.1; Chenomx Inc., NW Edmonton, AB, Canada) and databases (The human metabolome database: http://www.hmdb.ca). Normalized peak areas from all well-resolved peaks and identified metabolites were further analyzed using one-way analysis of variance (ANOVA) to confirm the significance of changes observed after naphthalene exposure. When statistical significance (P < 0.05) was reached among the groups, Dunnett’s test was then used to determine the statistical significance (P < 0.05) of all pairwise comparisons between groups. Lung Fixation for Histopathology
H, 2D JRES and p-JRES NMR spectra of the BALF (Supplementary Figs 1–3) and lung hydrophilic (Supplementary Figs 4–6) and hydrophobic metabolomes (Supplementary Figs 7–9) were obtained. Advantageous low spectral congested p-JRES peaks with flat baselines were taken for peak integration and statistical data analyses. Major hydrophilic metabolites were assigned by referring to tabulated chemical shifts (1H and p-JRES) and peak multiplicity (1H) (Azmi et al., 2005; Nicholson et al., 1995; Wei et al., 2009) and were confirmed using Chenomx NMR Suite software. A significantly elevated BALF metabolite with a singlet peak in the 1H spectra (δ 1.25 ppm) was further confirmed to be ethane, the only two-carbon alkane, using selected TOtal Correlation SpectroscopY (TOCSY) analysis. Major hydrophobic metabolites were assigned by referring to a tabulated chemical shift (1H and p-JRES) and peak multiplicity (1H) (Ong et al., 2009; Oostendorp et al., 2006; Waters et al., 2002). Effects of Naphthalene on the BALF Metabolome p-JRES NMR spectral data of BALF metabolomes obtained from the control and naphthalene-treated mice (50, 100 and 200 mg kg–1) were analyzed using unsupervised PCA. PCA reduces dimensionality, summarizes the similarities and differences of the dataset, and yields a simple graphical output called the scores plot (Fig. 1). As shown in Fig. 1, the scores plot was composed of PC 1 (42.60%) and 2 (15.35%), accounting for 57.95 % of the total variance within the BALF metabolome. In a PCA scores plot, each data point denotes a sample. The lower dose data (sample) points (olive oil only and 50 mg kg–1 naphthalene) clearly clustered and were separated from the higher dose data points (100 and 200 mg kg–1 naphthalene) based on PC1, reflecting dosedependent effects on the BALF metabolome. The complementary
Copyright © 2014 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/jat
1381
The animals were sacrificed 24 h after the naphthalene treatments. The tracheas were then cannulated with an i.v. catheter in situ, and
J. Appl. Toxicol. 2014; 34: 1379–1388
the thoraxes were opened by diaphragmatic incision (Plopper et al., 1992). Tracheal infusion without removing the lungs from the chest was used to maintain their in situ shape and the relationship between the lobes. The collapsed lungs were prefixed with a glutaraldehyde (0.9 vol./vol. %)/paraformaldehyde (1.0 wt./vol. %) mixture in 0.1 mM cacodylate buffer (330 mOsmol, pH 7.4) via tracheal infusion and were maintained for 24 h at 30 cm height fluid pressure (Dungworth et al., 1976; Plopper et al., 1992; Van Winkle et al., 1999). The lungs were removed from the chest 24 h later, and the terminal bronchioles were sectioned for inclusion in a 1-mm2 block post-fixation. Tissue blocks were handled according to the large block embedding method (Plopper, 1990), which enables highresolution light microscopy and transmission electron microscopy to be conducted on the same specimen. Tissues were post-fixed in 1% osmium tetroxide, dehydrated using ethanol, infiltrated with propylene oxide (1,2-propylene oxide; Merck, Darmstadt, Germany) and embedded in Araldite 502 (modified bisphenol A epoxy; Electron Microscopy Science, Hatfield, PA, USA). The blocks were then sectioned using a Diatome (Diatome diamond knife; Diatome Co., The Patch, Victoria Australia) and an Ultracut (Reicher Ultracult; Leica Microsystems, Milton Keynes, UK) into 1- to 2-micron thick sections, and then stained with toluidine blue for light and electron microscopy (Van Winkle et al., 1999).
J.–H. Hong et al.
Figure 1. PCA scores plot from the NMR spectra of BALF metabolomes from mice treated with various doses of naphthalene. Samples separated in the first component (PC1, 42.60%) are illustrated for the control group (circle) and the groups treated with 50 (square), 100 (rhombus) and 200 –1 (triangle) mg kg naphthalene.
PC1 loadings plot revealed the metabolites that contributed to clustering (Fig. 2). Peaks pointing up (positive values in the PC1 axis) in the loadings plot indicate that the corresponding metabolites from the samples located on the positive side of the PC1 axis of the scores plot are higher in concentration than the samples on the negative side of the same axis. The peaks contributing to clustering in the scores plot were further analyzed using a traditional method (ANOVA) (Table 1). Increased amounts (P < 0.05) of isopropanol, ethane, and acetone and decreased amounts (P < 0.05) of ethanol, acetate, glycerophosphocholine (GPC), and formate (Table 1) were detected in the BALF of mice treated with 100 and 200 mg kg–1 naphthalene. Effects of Naphthalene on the Lung Metabolome The total numbers of lung samples analyzed by NMR were 6, 6, 5 and 5 for the 0, 50, 100 and 200 mg kg–1 naphthalene-treated groups, respectively. Missing samples were a result of instrumental problems during sample preparation. The metabolomes of hydrophilic extracts from lung tissues were separated by PCA based on similarities in the p-JRES NMR spectra. The scores plot from the first two PCs explained 43.58% of the variance in
the lung hydrophilic metabolome (Fig. 3) and demonstrated a trend of shifting for the four groups treated with various doses of naphthalene. The metabolites contributing to the shifting in the corresponding loadings plot (Fig. 4) were identified and further analyzed using a traditional statistical method (ANOVA) (Table 2). Increased amounts (P < 0.05) of isopropanol and decreased amounts (P < 0.05) of glutamine and succinate were observed in the groups exposed to 100 and 200 mg kg–1 naphthalene. An unassigned singlet at δ 8.27 was found to be significantly different (P < 0.05) among the groups. However, these differences were not found to be significant in the Dunnett’s test (Table 2). The p-JRES NMR spectra of lung hydrophobic metabolomes were also analyzed using PCA. The first two PCs covered 62.61 % of the total variance (Fig. 5). Samples from the control and higher dose groups (100 and 200 mg naphthalene) that had clustered were partially separated along PC1 (37.70%). The complementary loadings plot identified the metabolites contributing to the trend (Fig. 6); significantly increased amounts of (CH2)n from fatty acyl chains and N(CH3)3 from phosphorylcholinecontaining lipids were found in mice treated with 200 mg kg–1 naphthalene (P < 0.05), whereas cholesterol levels were decreased in mice treated with 50, 100 and 200 mg kg–1 naphthalene (P < 0.001) (Table 3). Effects of Naphthalene on the Morphology of Mouse Distal Airways Light microscope images provided good resolution of ciliated and non-ciliated (Clara) epithelial cells (Supplementary Fig. 10). Swollen and vacuolated bronchiolar epithelial cells were observed after naphthalene exposure in a dose-dependent manner. The control mouse lung specimens did not demonstrate any altered cells in the terminal bronchioles. In addition, no swollen and vacuolated Clara cells were found in the terminal bronchioles of mice treated with 50 mg kg–1 naphthalene, although there were a few swollen, vacuolated Clara cells in the terminal bronchioles of mice treated with 75 mg kg–1 naphthalene. In the 100 mg kg–1 naphthalene-treated mouse specimens, the majority of the epithelial cells, including ciliated and Clara cells, were swollen and vacuolated (Supplementary Fig. 10).
1382
Figure 2. PCA loadings plot of PC1 (42.60%) from the NMR spectra of BALF metabolomes from mice treated with various doses of naphthalene. Key: 1: Isopropanol, 2: Ethanol, 3: Ethane, 4: Lactate, 5: Acetate, 6: Unassigned, 7: Acetone, 8: Succinate, 9: Creatine, 10: Choline, 11: Taurine, 12: Glycerophosphocholine, 13: Glycerol, 14: Unassigned singlet, 15: Creatine phosphate, and 16: Formate.
wileyonlinelibrary.com/journal/jat
Copyright © 2014 John Wiley & Sons, Ltd.
J. Appl. Toxicol. 2014; 34: 1379–1388
NMR-based metabolomics to study naphthalene-induced lung toxicity Table 1. The resonance assignments of metabolites from mouse BALF and changes of those from mice treated with various doses of naphthalene (i.p.) after 24 h Changes relative to control Key
Metabolite
ppm (multiplicity)
1 2 3 4 5 6 7 8 9 10 12 14 15 16
Isopropanol Ethanol Ethane Lactate Acetate Unassigned Acetone Succinate Creatine Choline Glycerophosphocholine Unassigned Creatine phosphate Formate
1.18(d)c, 4.01(m) 1.19(t)c, 3.66(q) 1.25(s)c 1.33(d)c, 4.11(q) 1.92(s)c 1.98c 2.23(s)c 2.41(s)c 3.04(s)c 3.20(s)c 3.36(s)c 3.60(s)c 3.94(s)c, 3.03(s) 8.46(s)c
a
NA50
NA100
NA200
0.10%↓ 17.30%↑ 11.34%↓ 27.37%↑ 11.76%↑ 59.80%↓ 28.45%↓ 10.91%↑ 10.26%↓ 9.47%↑ 3.64%↑ 9.80%↑ 8.19%↑ 20.02%↑
70.73%↑ 80.63%↓ 422.07%↑ 1.95%↑ 73.06%↓ 85.06%↓ 436%↑ 47.28%↓ 35.12%↓ 11.43%↓ 87.37%↓ 96.41%↓ 68.04%↓ 83.03%↓
60.07%↑ 71.41%↓ 412.51%↑ 26.06%↑ 76.04%↓ 99.37%↓ 742%↑ 26.12%↓ 16.13%↑ 1.49%↑ 84.62%↓ 92.15%↓ 21.48%↓ 81.71%↓
ANOVAb (P-value)
Dunnett’s test (P < 0.05)
0.0006** 0.0002**
