Abstract

   Derived from polycyclic aromatic hydrocarbons (PAHs), oxygenated-PAHs
   (oxy-PAHs) may pose hazards to aquatic organisms, which remain largely
   unknown. Takifugu obscurus is an important anadromous fish species of
   high economic and ecological values. In the present study, T. obscurus
   was acutely exposed to 44.29 µg l^−1 9,10-phenanthrenequione (9,10-PQ)
   for 96 h. Changes of antioxidant indices and metabolite profiles in
   plasma were compared between 9,10-PQ treatment and the control. The
   results showed that 9,10-PQ treatment significantly increased
   malondialdehyde (MDA) content during 6 to 96 h, increased superoxide
   dismutase (SOD) and catalase (CAT) activities at 6 h, but decreased
   them at 96 h. These results indicated that 9,10-PQ induced oxidative
   stress to fish. Ultra-performance liquid chromatography-mass
   spectrometry (UPLC-MS) analysis revealed that four metabolic pathways
   were influenced in response to treatment with 9,10-PQ, including
   glycerophospholipid metabolism, phenylalanine, tyrosine and tryptophan
   biosynthesis, purine metabolism and sulfur metabolism. These pathways
   are associated with antioxidant mechanisms, biosynthesis of
   neurotransmitters and innate immune functions. Thus, the as-obtained
   results confirmed that 9,10-PQ induced oxidative stress and raised
   concerns of neurotoxicity and immunotoxicity to fish. Overall, the
   present study posed a high environmental risk of oxy-PAHs to aquatic
   ecosystems.

   Keywords: metabolomics, polycyclic aromatic hydrocarbons, toxicity,
   biomarker, oxidative stress

1. Introduction

   Polycyclic aromatic hydrocarbons (PAHs) are a group of typical
   persistent organic pollutants widespread in natural environments and
   have drawn great attention due to their mutagenic and carcinogenic
   toxicity [[28]1,[29]2]. This family of compounds consists of two or
   more fused benzene rings and contain derivatives with alkyl, nitrogen
   or oxygen substitutions [[30]3]. Among them, 16 PAHs have been
   stipulated as the prior pollutants by the Environmental Protection
   Agency of the USA (US EPA). PAHs are produced during incomplete
   combustion of organic materials and fossil fuels. More importantly,
   accidental oil spill, leakages from drilling operations and industrial
   wastewater release plenty of PAHs to aquatic environments [[31]4].
   Following the Gulf War oil spill event in 1991, PAHs could be detected
   in local seawater even after three years, with the concentration of
   total PAHs ranging from 21.14 to 320.5 µg l^−1 [[32]5]. During the oil
   spill of the Deepwater Horizon accident, severe PAH pollution was
   observed and the concentration of total PAHs reached 151 mg l^−1
   [[33]6]. Such high levels of PAHs would undoubtedly affect aquatic
   organisms.

   During the past decades, most studies have focused on the toxicity of
   parent-PAHs to fish, displaying increased deformities, lesions, tumours
   and other toxic effects [[34]7]. After exposure to PAHs mixture,
   zebrafish embryos exhibited morbid syndromes, including cardiac
   dysfunction, oedema, spinal curvature and abnormal development of
   craniofacial structures [[35]8]. Cultivation with PAH-contaminated
   sediments resulted in hepatic lesions in mummichogs (Fundulus
   heteroclitus) [[36]9]. Prevalence of liver tumour in brown bullheads
   (Ameiurus nebulosus) was even used as an indicator of PAHs
   contamination [[37]10]. During the emission of total PAHs,
   oxygenated-PAHs (oxy-PAHs) may also be released, which are derivatives
   of parent-PAHs. Moreover, after releasing in the natural environments,
   parent-PAHs can be transformed to oxy-PAHs [[38]11]. However, toxicity
   and environmental risks of oxy-PAHs to fish have not been well
   investigated.

   The pufferfish Takifugu obscurus is a commercially important species
   [[39]12], since it is a traditional and popular food in Asian
   countries. However, the natural resource of this species has been
   declining rapidly owing to overexploitation and environmental pollution
   [[40]13]. As an anadromous fish species, T. obscurus may swim across
   various freshwater and seawater environments, thus exhibiting a high
   risk of being exposed to PAHs [[41]14], since PAHs have been broadly
   detected in aquatic environments. To the best of our knowledge, only
   Wang [[42]15] compared the hepatic toxicity of three PAHs to T.
   obscurus. The results showed that treatments with 400 µg l^−1
   phenanthrene, 400 µg l^−1 3-methyl phenanthrene and 100 µg l^−1
   9,10-phenanthrenequinone significantly increased glutathione
   transferase (GST) activity. Treatments with 100 µg l^−1 phenanthrene,
   100 µg l^−1 3-methyl phenanthrene and 12.5 µg l^−1
   9,10-phenanthrenequinone significantly elevated malondialdehyde (MDA)
   contents. These results preliminarily revealed that PAHs induced
   significant oxidative stress to T. obscurus. More investigations are
   still required to comprehensively understand toxicity of PAHs and their
   derivatives to T. obscurus, which may contribute to the protection of
   natural resource and food safety of T. obscurus.

   Activities of superoxide dismutase (SOD), catalase (CAT) and content of
   MDA are traditional oxidative/antioxidant indices to evaluate harmful
   effects of xenobiotics on organisms [[43]16,[44]17]. Treatments with
   the water-soluble fraction of Arabian crude oil which mainly contained
   PAHs mixture, significantly (p < 0.05) lowered SOD activity in the
   European sea bass Dicentrarchus labrax [[45]18] and increased MDA
   content in Lateolabrax japonicus livers [[46]19]. At the
   transcriptional level, numbers of researches have been implemented to
   explore oxidative stress induced by PAHs in zebrafish and other aquatic
   organisms [[47]20–[48]23]. However, the effects of oxy-PAHs on changes
   of these antioxidant indices in T. obscurus have not been reported.
   Endogenous metabolites are the most proximal indicators, which can
   reflect metabolic alterations in response to exogenous stimuli
   [[49]24,[50]25]. The metabolomics technology, which detects the changes
   of all metabolites in organisms, has been widely applied to investigate
   metabolic alterations in response to xenobiotics [[51]26]. Among the
   methods used for metabolomics, liquid chromatography–mass spectrometry
   (LC-MS) is the most compatible technique for detection of small
   metabolites and displays robust reliability and reproducibility
   [[52]27,[53]28]. The LC-MS-based metabolomics has been successfully
   used to explore effects of two PAHs (benz[a]anthracene and
   benz[a]anthracene-7,12-dione) on metabolisms of zebrafish,
   demonstrating toxic influences on protein biosynthesis, mitochondrial
   function, neural development, vascular development and cardiac function
   [[54]29]. Thus, metabolomics analyses might be useful to explore
   influences of oxy-PAHs on fish at the metabolic level.

   9,10-phenanthrenequione (9,10-PQ) is an oxygenated derivate of the
   typical PAH phenanthrene. Plasma is a pool of metabolites and can
   systematically reflect metabolic changes in the whole body [[55]30].
   Thus, plasma is a suitable subject to investigate physiological changes
   of organisms in response to pollutants. In the present study, to assess
   the environmental risk of 9,10-PQ, T. obscurus juveniles were exposed
   to 9,10-PQ and changes of antioxidant indices and metabolite profiles
   in plasma were investigated. These results may provide basic
   information to understand the environmental effects of PAHs on aquatic
   ecosystems.

2. Material and methods

2.1. Animals

   Juvenile T. obscurus (24.96 ± 0.97 g, 10.19 ± 0.45 cm) were obtained
   from the Freshwater Fisheries Research Centre, Chinese Academy of
   Fishery Science (Wuxi, China). Before exposure, fish were acclimated in
   glass aquaria containing 50 l of dechlorinated tap water for two weeks.
   The water was aerated constantly. The pH, temperature and dissolved
   oxygen content (DO) of water were controlled at 7.8 ± 0.05, 30 ± 1°C
   and 6.9 ± 0.1 mg l^−1, respectively. Fish were fed twice a day under a
   photoperiod of 14 h : 10 h (light : dark). During this period, no fish
   died.

2.2. Exposure and sampling

   9,10-PQ (purity > 98%) was purchased from ChemService (Worms, Germany).
   A stock solution of 44.29 mg l^−1 9,10-PQ was prepared in dimethyl
   sulfoxide (DMSO) and stored at 4°C in dark. The previous acute toxicity
   test in our laboratory discovered that the 96 h-LC[50] of 9,10-PQ to T.
   obscurus was approximately 88.59 µg l^−1. In the present study, the
   exposure concentration was set at 50% 96 h-LC[50] (44.29 µg l^−1) as
   high exposure concentration to test the acute toxicity of 9,10-PQ,
   which could maximize the acute toxic effect as well as not cause death.
   About 1‰ DMSO was used as the control. Fish were randomly distributed
   into treatment and control in three biological replicates. Each
   replicate contained 50 fish. The exposure lasted for 96 h and 75% of
   the culture media were renewed daily.

   During the exposure period, three fish were randomly sampled from each
   aquarium after 0, 2, 6, 12, 24, 48 and 96 h to determine antioxidant
   indices. After 96 h, four fish were randomly sampled from each aquarium
   for metabolome analysis. Approximately 0.5 ml of blood was collected
   from the caudal vein and immediately transferred into
   ethylenediaminetetraacetic acid (EDTA) centrifuge tube. After
   centrifugation at 5250g at 4°C for 10 min, plasmas were snap-frozen in
   liquid nitrogen and preserved at −80°C.

2.3. Determination of oxidative stress biomarkers

   Contents of MDA, SOD and CAT activities were measured following the
   methods of Placer et al. [[56]31], Sun et al. [[57]32] and Beer & Sizer
   [[58]33], respectively. One unit of SOD activity was defined as that
   enzyme quantity which causes 50% suppression of nitroblue tetrazolium
   (NBT) reduction. One unit of CAT activity was defined as the activity
   required to destroy 1 nmol of H[2]O[2] at 25°C in 0.2 M phosphate
   buffer for 1 min.

2.4. Metabolomics analysis

   For each sample, equal amount of plasmas from four fish in the same
   replicate were pooled. Then, 100 µl of pooled plasmas were mixed with
   300 µl of methanol to precipitate proteins, and 10 µl of 2.9 mg ml^−1
   DL-O-chlorophenylalanine was added as the internal standard. The
   mixture was vortexed for 30 s and centrifuged at 18 000g and 4°C for 15
   min. Finally, 200 µl of supernatant was transferred to new vials. After
   evaporation of solvent, the residues were dissolved in 100 µl of 5%
   acetonitrile solution containing 0.1% formic acid.

   Four microlitres of samples were injected into an UPLC-QTOF/MS system
   (Agilent 1290 Infinity LC System, Agilent 6530 UHD and Accurate-Mass
   Q-TOF, Agilent Technologies, Santa Clara, CA, USA) equipped with an
   electrospray ionization (ESI) source and a C18 column (2.1 × 100 mm,
   1.8 µm, Agilent Technologies). The mobile phase was 0.1% formic acid in
   water (mixture A), and 0.1% formic acid in acetonitrile (mixture B).
   The flow rate was 0.35 ml min^−1. The column temperature was 40°C and
   the automatic injector temperature was 4°C. The samples were eluted
   following the procedure: 0%–5% at 1 min, 5%–20% from 1 to 6 min,
   20%–50% from 6 to 9 min, 50%–95% from 9 to 13 min, and 95% from 13 to
   15 min.

   Mass spectrometry was performed at both positive (ESI^+) and negative
   ion modes (ESI^−), separately. Source temperature was 100°C and gas
   flow rate was 50 l h^−1. The desolvation temperature was 350°C and the
   flow rate was 600 l h^−1. The capillary voltage and cone voltage were 4
   and 35 kV in positive ion mode, respectively, while 3.5 and 50 kV in
   negative ion mode, respectively. Data from 50 to 1000 m/z were
   collected with the scan time of 0.03 s and the interval time of 0.02 s.
   Leu-enkephalin was used as the lock mass ([M^+H]^+ of 556.2771 Da in
   ESI^+ and [M^−H]^− of 554.2615 Da in ESI^−). The pooled sample was used
   as quality control (QC) to validate the reproducibility and stability
   of the method and equipment.

2.5. Data processing and pattern recognition

   Raw UPLC-MS data was first processed using Mass Profiler software
   (Agilent), including peak detection, peak alignment and peak
   integration. Subsequently, peak areas of all samples were normalized to
   the total area. Afterwards, a two-dimensional data matrix was
   generated, including retention time (RT), compound molecular weight
   (mass) and peak intensity. The normalized data were subjected to
   SIMCA-P 14.1 (Umetrics AB, Umea, Sweden) for orthogonal partial
   least-squares discriminant analysis (OPLS-DA) to compare metabolic
   differences between treatment and control. Important variables on the
   projection (VIP) of the OPLS-DA model were used to discover the
   potential variables contributing to the differentiation. Differentially
   expressed metabolites (DEM) were defined as VIP > 1 and p < 0.05. The
   METLIN web-based metabolomics database ([59]https://metlin.scripps.edu)
   was used for tentative identification of significant features,
   including accurate mass and MS/MS spectral matching [[60]34].
   Verification of the identified metabolites was conducted using an
   in-house library of standards based on the accurate mass and retention
   time of metabolite standards provided by IROA Technologies.

2.6. Analysis of metabolic pathways

   Pathway analyses were performed based on the metabolome of Danio rerio
   and the Kyoto Encyclopedia of Gene and Genomes (KEGG) database [[61]35]
   using the MetaboAnalyst 2.0 [[62]36]. The KEGG enrichment analyses were
   statistically conducted using the Fisher's exact test.

2.7. Statistical analysis

   The variation (upregulation and downregulation) of DEMs were expressed
   as log[2](fold change). Antioxidant indices are presented as mean ±
   standard deviation (s.d.). Student's t-test was applied to compare
   significant differences between treatment and control. p < 0.05 was
   considered statistically significant.

3. Results

3.1. Changes of oxidative stress biomarkers

   Along with exposure time, SOD, CAT activities and MDA content did not
   significantly change in the control. However, in 9,10-PQ treatment, SOD
   and CAT activities increased first but then decreased, MDA contents
   increased at early time but then remained constant. Compared with the
   control, treatment with 9,10-PQ significantly increased MDA content at
   72.97%, 91.21%, 85.84%, 93.37% and 97.44% respectively from 6 h to 96 h
   (p < 0.05); SOD and CAT activities in 9,10-PQ treatment were
   significantly higher than those in the control at 6 h with the increase
   of 45.18% and 44.54%, respectively, but significantly lower than those
   in the control at 96 h with the decrease of 29.86% and 32.62%
   respectively (p < 0.05, figures [63]1–[64]3).

Figure 2.

   [65]Figure 2.
   [66]Open in a new tab

   SOD activity in plasma of T. obscurus under sublethal exposure to
   9,10-PQ. The data indicate mean (nine fish samples from three aquaria)
   and s.d. Significant differences (p < 0.05) between control and 9,10-PQ
   treatment are marked by asterisks (Student's t-test).

Figure 1.

   [67]Figure 1.
   [68]Open in a new tab

   MDA content in plasma of T. obscurus under sublethal exposure to
   9,10-PQ. The data indicate mean (nine fish samples from three aquaria)
   and s.d. Significant differences (p < 0.05) between control and 9,10-PQ
   treatment are marked by asterisks (Student's t-test).

Figure 3.

   [69]Figure 3.
   [70]Open in a new tab

   CAT activity in plasma of T. obscurus under sublethal exposure to
   9,10-PQ. The data indicate mean (nine fish samples from three aquaria)
   and s.d. Significant differences (p < 0.05) between control and 9,10-PQ
   treatment are marked by asterisks (Student's t-test).

3.2. Total ion current chromatograms

   The total ion current (TIC) chromatograms of QC samples tested before
   and after experiments showed reproducible patterns in both ESI^+ and
   ESI^− models ([71]figure 4a,b), demonstrating high stability and
   repeatability of UPLC-MS analyses. The TIC chromatograms of plasma
   samples differed between treatment and control in either ESI^+ or ESI^−
   models. For instance, the peak of the control at 3.6 min disappeared in
   9,10-PQ treatment in the ESI^+ model. The differences in TIC
   chromatograms indicated that treatment with 9,10-PQ influenced the
   metabolite profiles in fish plasma ([72]figure 4).

Figure 4.

   [73]Figure 4.
   [74]Open in a new tab

   Typical UPLC-MS TIC chromatograms of (a) QC in positive ion mode; (b)
   QC in negative ion mode; (c) the control sample in positive ion mode;
   (c) the control sample in negative ion mode; (e) 9,10-PQ sample in
   positive ion mode; (f) 9,10-PQ sample in negative ion mode.

3.3. Orthogonal partial least-squares discriminant analysis analyses and
differentially expressed metabolites

   OPLS-DA analysis clearly discriminated treatment with 9,10-PQ and the
   control in plots ([75]figure 5). All samples were located within the
   95% confident intervals (black circle in [76]figure 5). R^2Y values
   indicate the contribution rate and Q^2 values represent the predictive
   ability of the supervision model. In ESI^+ model, R^2Y and Q^2 were
   0.988 and 0.625, respectively; in ESI^− model, R^2Y and Q^2 were 0.949
   and 0.601, respectively. The high R^2Y and Q^2 values in the present
   study indicated reliability of the OPLS-DA models.

Figure 5.

   [77]Figure 5.
   [78]Open in a new tab

   OPLS-DA score plot of metabolic profiles in T. obscurus plasma. (a)
   Positive ion mode; (b) negative ion mode. Brown triangles indicate the
   plasma samples of the control while yellow triangles indicate the
   plasma samples of 9,10-PQ group.

   Based on the OPLS-DA results, DEM were identified. Overall, 21
   metabolites were differentially expressed between 9,10-PQ treatment and
   the control in ESI^+ model and 10 DEMs in ESI^− model. Compared with
   the control, LPA(0:0/16:0), L-tryptophan, dehydrojuvabione,
   PS(15:1(9Z)/22:6(4Z,7Z,10Z,13Z,16Z,19Z)) and cinnamaldehyde were
   upregulated greatly with log[2](fold change) higher than 2;
   Nα-acetyl-L-glutamine and guanosine were downregulated remarkably with
   log[2](fold change) lower than −2 ([79]table 1).

Table 1.

   Differentially expressed metabolites derived from OPLS-DA in treatment
   with 9,10-PQ compared with the control. FC: fold change.
   VIP name detected mass RT (min) p-value (t-test) log[2](FC)
   ESI^+
   2.417 dodecanedioic acid 230.152 10.226 0.000 0.932
   2.410 guanosine 283.0918 1.397 0.000 −1.433
   2.386 LPA(0:0/16:0) 410.2288 4.034 0.001 4.527
   2.229 coenzyme Q4 454.3268 9.834 0.002 1.385
   2.203 L-tryptophan 204.0788 11.990 0.002 2.621
   2.065 spermidine 145.158 0.565 0.004 1.557
   2.051 LysoPC(20:4(5Z,8Z,11Z,14Z)) 543.3323 10.188 0.004 −1.029
   1.900 L-phenylalanine 165.0791 2.174 0.009 0.778
   1.899 dehydrojuvabione 264.1733 11.427 0.010 2.123
   1.893 malic acid 134.0229 1.394 0.010 −1.846
   1.888 ectoine 142.0742 0.820 0.010 1.208
   1.814 PS(15:1(9Z)/22:6(4Z,7Z,10Z,13Z,16Z,19Z)) 791.4415 4.931 0.014
   2.839
   1.770 dodecylphosphocholine 351.2411 11.950 0.017 1.059
   1.716 LysoPC(22:5(4Z,7Z,10Z,13Z,16Z)) 569.348 10.755 0.021 1.372
   1.594 homoglutamine 160.0887 11.658 0.034 0.257
   1.592 LysoPC(22:6(4Z,7Z,10Z,13Z,16Z,19Z)) 567.3328 10.175 0.034 −0.907
   1.556 octylamine 129.1516 4.293 0.039 −1.423
   1.548 LysoPC(15:0) 481.3166 11.598 0.040 −0.589
   1.544 PC(13:0/0:0) 453.2854 10.360 0.040 −0.775
   1.532 propionyl-L-carnitine 217.132 1.410 0.042 1.035
   1.501 cinnamaldehyde 132.0625 0.915 0.047 2.159
   ESI^−
   2.143 2-acetolactic acid 132.0437 38.989 0.007 1.337
   2.009 Nα-acetyl-L-glutamine 188.0799 50.817 0.013 −3.370
   2.008 LysoPC(15:0) 481.3167 697.672 0.013 −0.925
   1.928 glutathione, oxidized 612.1526 47.434 0.017 1.370
   1.924 guanosine 283.0918 83.546 0.018 −2.062
   1.848 3-furoic acid 112.0166 46.445 0.023 0.756
   1.777 dehydrochorismic acid 224.0357 516.608 0.030 0.442
   1.760 inosine 267.0273 239.943 0.032 0.615
   1.744 dehydroascorbic acid 174.0166 71.069 0.034 1.245
   1.715 sulfuric acid 97.9677 50.209 0.037 1.016
   [80]Open in a new tab

3.4. Enrichment of Kyoto Encyclopedia of Gene and Genomes pathways

   In the ESI^+ model, two pathways, glycerophospholipid metabolism, and
   phenylalanine, tyrosine and tryptophan biosynthesis were significantly
   enriched. In the ESI^− model, purine metabolism and sulfur metabolism
   were significantly enriched. The purine metabolism involved three DEMs
   (inosine, guanosine and sulfuric acid) and four
   lysophosphatidylcholines (LysoPCs) contributed to the
   glycerophospholipid metabolism ([81]figure 6 and [82]table 2).

Figure 6.

   [83]Figure 6.
   [84]Open in a new tab

   Enrichment of metabolic pathways based on differentially expressed
   metabolites. (a) Positive ion mode; (b) negative ion mode. All matched
   pathways are depicted based on –log(p) value from pathway enrichment
   analysis and pathway impact score from pathway topology analysis.
   Colour gradient indicates the significance of the pathway ranked by
   –log(p) value (red: higher –log(p) values and yellow: lower –log(p)
   values). Circle size indicates pathway impact score (the larger the
   circle the higher the influence score).

Table 2.

   Enrichment of KEGG pathways. DEM: differentially expressed metabolites.
   pathway Name DEM −log(p) impact
   ESI^+
   glycerophospholipid metabolism LysoPC(22:6(4Z,7Z,10Z,13Z,16Z,19Z))
   3.880 0.264
   LysoPC(20:4(5Z,8Z,11Z,14Z))
   LysoPC(22:5(4Z,7Z,10Z,13Z,16Z))
   LysoPC(15:0)
   phenylalanine, tyrosine and tryptophan biosynthesis L-phenylalanine
   3.426 0.500
   L-tryptophan
   ESI^−
   purine metabolism inosine 6.843 0.005
   guanosine
   sulfuric acid
   sulfur metabolism sulfuric acid 3.405 0.25
   [85]Open in a new tab

4. Discussion

   Phenanthrene is one of the 16 prior PAH pollutants stipulated by US
   EPA, and its toxicity to aquatic organisms has been well studied.
   However, its oxygenated derivative 9,10-PQ has been seldom
   investigated. T. obscurus is an important migratory fish species and
   reveals a high risk of exposure to fossil oil, which contains high
   content of PAHs and 9,10-PQ. The high-throughput metabolomics can
   reveal metabolic alterations, which can directly reflect the toxic
   effects of exogenous pollutants. The present study applied the
   UPLC-MS-based metabolomics technology to investigate plasma responses
   to acute 9,10-PQ exposure in T. obscurus. The results showed that
   significant oxidative stress was induced and a total of four KEGG
   pathways were significantly influenced, which represented the major
   toxic effects of 9,10-PQ on physiological metabolism in T. obscurus.

   MDA content represents the level of lipid peroxidation (LPO). SOD and
   CAT are two typical antioxidant enzymes. These three indices are
   frequently used to assess oxidative stress induced by xenobiotics
   [[86]37]. In the present study, after exposure to 9,10-PQ, MDA content
   significantly increased, suggesting peroxidation of cell membranes.
   Similarly, Lin et al. [[87]19] reported that treatment with pyrene
   increased MDA contents in Lateolabrax japonicus. Exposure to 200 µg
   l^−1 phenanthrene significantly increased the LPO values in the muscle
   of estuarine guppy Poecilia vivipara at the level of Bayes factor > 24
   [[88]38]. Treatment with benzo(a)pyrene increased MDA contents in
   Ruditapes philippinarum significantly [[89]39]. To resist damages of
   oxidative stress, antioxidant mechanisms, such as SOD and CAT might be
   activated [[90]40]. As the most important and powerful antioxidant
   enzymes, SOD could detoxify superoxide anions while CAT could reduce
   H[2]O[2] [[91]39]. In the present study, both SOD and CAT activities
   significantly increased after exposure to 9,10-PQ for 6 h, which might
   be responses to oxidative stress. However, after 96 h, their activities
   were significantly lower than those in the control, probably because
   over-production of superoxide anions and H[2]O[2] might exceed the
   elimination capacity of SOD and CAT, and then damage the antioxidant
   defence system. This result was in accordance with previous research
   which discovered the same pattern of SOD and CAT activities in
   Carassius auratus in response to phenanthrene exposure [[92]41]. These
   results demonstrated that oxy-PAHs exhibited oxidative effects on fish,
   which was similar to parent-PAHs.

   Glycerophospholipid metabolism was significantly (−log(p) = 3.880)
   affected in response to 9,10-PQ exposure. Similarly, this pathway was
   also altered by phenanthrene in human keratinocytes HaCaT cells
   [[93]42]. Glycerophospholipids are the main lipids in cell membrane
   [[94]43,[95]44]. They regulate membrane potential, curvature and
   influence membrane fusion, fission and ion transport [[96]45]. Thus,
   altered glycerophospholipid metabolism might then influence cellular
   functions. LysoPCs are major lipid components of plasma derived from
   phosphatidylcholines (PC) by the enzyme lecithin-cholesterol
   acyltransferase (LACT) [[97]46,[98]47]. LysoPCs play important roles in
   regulating immunologic cellular functions, peculiarly in priming
   neutrophils, which are main cells of innate immunity [[99]48]. In the
   present study, four LysoPCs were involved in this pathway. Among them,
   three LysoPCs were downregulated and one LysoPC was upregulated in
   response to 9,10-PQ treatment. The overall downregulation of LysoPCs in
   the present study might negatively affect structure and functions of
   cellular membranes as well as the innate immunity.

   As previously reported, treatments with PAHs (benz[a]anthracene and
   benz[a]anthracene-7,12-dione) exhibited neurobehavioral toxicity to
   zebrafish (Danio rerio) by affecting the phenylalanine, tyrosine and
   tryptophan biosynthesis pathway and downregulating dopamine, serotonin,
   phenylalanine, tyrosine and tryptophan [[100]29]. Similarly,
   developmental neurotoxicity was observed in Japanese medaka embryos
   when exposed to benz[a]anthracene [[101]49]. Moreover, He et al.
   [[102]50] demonstrated that exposure to pyrene led to
   neurodevelopmental lesions in Sebastiscus marmoratus embryo. In the
   present study, neurotransmitters were detected, probably because these
   compounds are mainly localized to the neural tissues rather than
   plasma. However, treatment with 9,10-PQ also significantly affected the
   phenylalanine, tyrosine and tryptophan biosynthesis pathway (−log(p) =
   3.426), which contained two upregulated DEMs (L-phenylalanine and
   L-tryptophan). L-phenylalanine is an essential amino acid to animals
   and is also the precursor of tyrosine, catecholamine neurotransmitters,
   dopamine, norepinephrine and epinephrine [[103]51,[104]52]. These
   results were not conflicted with the above publications. Inhibition of
   the transformation process from L-phenylalanine to neurotransmitters
   would accumulate more L-phenylalanine and L-tryptophan. More
   investigations are required to validate this hypothesis.

   In response to 9,10-PQ exposure, purine metabolism was the most
   affected pathway (−log(p) = 6.843), involving upregulation of inosine
   and sulfuric acid as well as downregulation of guanosine. Inosine is
   formed by the deamination of adenosine [[105]53], displaying ability in
   reduction of oxidative stress and elevation of peroxidase activity
   [[106]54]. Besides, inosine can also improve immune response, for
   instance, bactericidal activity and lysozyme activity in Japanese
   flounder P. olivaceus [[107]55]. In the present study, content of
   inosine increased 1.53-fold in treatment with 9,10-PQ, compared with
   the control, indicating that T. obscurus may produce more inosine to
   resist oxidative stress and harmful effects on immunity caused by
   9,10-PQ. Downregulation of guanosine in the present study was similar
   to a previous cell culture experiment, in which exposure to 160 µg l^−1
   PAHs mixture significantly downregulated the level of guanosine due to
   inhibition of hypoxanthine guanine phosphoribosyl transferase (HGPRT)
   [[108]56,[109]57]. Oxidation of guanosine is a biomarker of oxidative
   stress [[110]58]. Considering the increased MDA level in the present
   study, decreased guanosine content might be a response to oxidative
   stress, which would negatively affect the repair of DNA damage
   [[111]57] and the non-specific immunity system [[112]59].

   Sulfuric acid is an important metabolite in both purine metabolism and
   sulfur metabolism pathways. It is an essential component for
   biosynthesis of antioxidant products, such as glutathione. Both
   previous publications and the present study demonstrated that oxy-PAHs
   induced production of reactive oxygen species (ROS) [[113]60]. Thus,
   the increased content of sulfuric acid in the 9,10-PQ treatment was
   probably a response of activation of antioxidant status. However, the
   effect of oxy-PAHs on sulfur metabolism has been poorly studied and
   further researches are still required. The natural resource of T.
   obscurus has decreased dramatically in the past few decades. Migration
   across various habitats increased the risk of being exposed to
   industrial wastewater, oil spill and leakage. Oxy-PAHs have higher
   toxicity and water solubility compared with parent-PAHs [[114]61], but
   have not received international supervision [[115]62]. The present
   study for the first time investigated effects of 9,10-PQ on plasma
   metabolites and antioxidant indices in T. obscurus. The results
   displayed that 9,10-PQ induced oxidative stress, neurotoxicity and
   harmed innate immunity, raising environmental concerns of oxy-PAHs on
   aquatic ecosystem.

5. Conclusion

   Acute exposure to 9,10-PQ induced oxidative stress and influenced
   phenylalanine, tyrosine and tryptophan biosynthesis,
   glycerophospholipid, purine and sulfur metabolisms in T. obscurus
   plasma. More investigations on the toxicity of oxy-PAHs to T. obscurus
   are required to profoundly understand the hazards of oxy-PAHs.

Supplementary Material

   Reviewer comments
   [116]rsos191356_review_history.pdf^ (414.8KB, pdf)

Acknowledgements