Abstract

   Exposure to trace-level heavy metals and antibiotics may elicit
   metabolic disorder, alter protein expression, and then induce
   pathological changes in zebrafish embryos, despite negligible
   physiological and developmental toxicity. This study investigated the
   single and combined developmental toxicity of fluoroquinolones
   (enrofloxacin [ENR] and ciprofloxacin [CIP]) (≤0.5 μM) and heavy metals
   (Cu and Cd) (≤0.5 μM) to zebrafish embryos, and molecular responses of
   zebrafish larvae upon exposure to the single pollutant (0.2 μM) or a
   binary metal-fluoroquinolone mixture (0.2 μM). In all single and
   mixture exposure groups, no developmental toxicity was observed, but
   oxidative stress, inflammation, and lipid depletion were found in
   zebrafish embryos, which was more severe in the mixture exposure groups
   than in the single exposure groups, probably due to increased metal
   bioaccumulation in the presence of ENR or CIP. Metabolomics analysis
   revealed the up-regulation of amino acids and down-regulation of fatty
   acids, corresponding to an active response to oxidative stress and the
   occurrence of inflammation. The up-regulation of antioxidase and immune
   proteins revealed by proteomics analysis further confirmed the
   occurrence of oxidative stress and inflammation. Furthermore, the KEGG
   pathway enrichment analysis showed a significant disturbance of
   pathways related to immunity and tumor, indicating the potential risk
   of tumorigenesis in zebrafish larvae. The findings provide
   molecular-level insights into the adverse effects of heavy metals and
   antibiotics (especially in chemical mixtures) on zebrafish embryos, and
   highlight the potential ecotoxicological risks of trace-level heavy
   metals and antibiotics in the environment.

   Keywords: Fluoroquinolone, Metal, Inflammation, Tumor, Oxidative stress

Graphical abstract

   [33]Image 1
   [34]Open in a new tab

Highlights

     * •
       Trace-level FQs or/and metals had no developmental toxicity to
       zebrafish embryos.
     * •
       FQs and metals induced oxidative stress, inflammation, and lipid
       depletion.
     * •
       FQs promoted metal uptake and induced more severe inflammation and
       lipid depletion.
     * •
       Metabolomics confirmed oxidative stress, inflammation, and lipid
       depletion.
     * •
       FQs and metals increased potential risk of tumorigenesis revealed
       by proteomics.

1. Introduction

   Fluoroquinolones (FQs) such as enrofloxacin (ENR) and ciprofloxacin
   (CIP) are mainly administered to food animals (e.g., fish, pigs, and
   chicken) for the prevention and treatment of bacterial infections
   [[35]1,[36]2]. Because of their large uses worldwide, CIP and ENR were
   detected in environmental water samples at concentrations from several
   nM to several tens of μM [[37][3], [38][4], [39][5]]. For example, the
   ENR concentrations range from 0.016 to 0.068 nM in surface waters
   [[40]3], are up to hundreds of nM in treated wastewater effluents from
   hospitals and livestock farms [[41]4,[42]6], and even as high as
   2.50 μM in effluents of pharmaceutical wastewaters [[43]7]. More
   alarmingly, the CIP concentration was up to 93.6 μM, 2.72 μM, 0.794 μM,
   and 1.04 nM in the effluents of pharmaceutical wastewater, hospitals,
   livestock farms, and surface waters, respectively
   [[44]3,[45]4,[46]6,[47]7]. Thus, it is important to study the toxicity
   of ENR and CIP to aquatic organisms.

   Studies have reported that ENR and CIP could induce growth inhibition
   (8.63 μM ENR and 90.5 μM CIP), reproduction inhibition (17.5 μM ENR and
   90.5 μM CIP), oxidative stress (0.278 μM ENR and 0.302 μM CIP), lipid
   metabolic disturbance (0.003 μM ENR and CIP), and cardiotoxicity
   (4246 μM CIP) to non-target aquatic organisms [[48][8], [49][9],
   [50][10]]. Antibiotic exposure at environmentally-relevant levels may
   not inhibit the development, growth, and reproduction of aquatic
   organisms, but instead cause oxidative stress, inflammation, and
   metabolic disturbance [[51][11], [52][12], [53][13], [54][14]]. Indeed,
   the exposure of zebrafish embryos to 0.028 and 1.39 μM ENR for 3 days
   increased the catalytic activity of glutathione peroxidase but
   decreased the catalytic activity of glutathione-S-transferase [[55]14].
   Similarly, the exposure of zebrafish embryos to sulfamethoxazole
   (0.395 μM) increased the total antioxidant capacity, the gene
   expression of key proinflammatory cytokines, and the production of
   corresponding proteins in zebrafish embryos [[56]13]. The exposure of
   Nile tilapia to 0.001 μM oxytetracycline and 0.001 μM sulfamethoxazole
   increased the antioxidase activity, suppressed innate immunity, and
   stimulated inflammatory and detoxification responses in fish [[57]12].
   Nonetheless, previous studies mainly focused on the exposure to the
   individual antibiotic or the mixtures of antibiotics [[58]9] and rarely
   studied the toxicity of the mixtures of antibiotics and heavy metals to
   aquatic organisms.

   Assessing the toxicity of chemical mixtures to aquatic organisms is
   challenging due to their potential synergistic, additive, or
   antagonistic effects [[59]15]. For example, the co-exposure to 5 μM
   1-H-benzotriazole and 1 μM Cd decreased the acute toxicity of Cd and
   alleviated the Cd-induced liver atrophy in zebrafish [[60]16]. However,
   no-effect-concentrations of dibenzothiophene (1 μM) and Cd (0.2 μM) in
   mixtures caused morphological defects of thyroid and developmental
   toxicity in zebrafish embryos [[61]17]. Despite these progresses, the
   studies on the joint toxicity of FQs and heavy metals at
   environmentally relevant concentrations to aquatic organisms are still
   limited, which is imperative because FQs and heavy metals often
   coexisted in natural waters [[62]18,[63]19].

   Therefore, this study investigated the development and metabolic and
   proteomic responses of zebrafish embryos upon exposure to individual
   FQs (ENR or CIP), heavy metals (Cu or Cd), or binary mixtures of one
   fluoroquinolone and one heavy metal. The changes of reactive oxygen
   species (ROS), macrophage, and lipids in zebrafish larvae were examined
   using in vivo fluorescence imaging and in vitro fluorescence
   quantification. The inflammation and tumorigenesis potential of
   zebrafish embryos were further explored with metabolomics and
   proteomics analyses. This study provides novel insights into the
   potential risks of antibiotics and heavy metals to aquatic organisms.

2. Materials and methods

2.1. Zebrafish and chemicals

   Adult zebrafish (Tübingen line, 3-month-old, Danio rerio) was supplied
   by Nanjing EzeRinka Biotech Company. CIP (purity >98.0%),
   Cd(NO[3])[2]·4H[2]O (purity >99.0%), and
   2′-7′-dichlorodihydrofluorescein diacetate (H[2]DCFDA) were purchased
   from Macklin Biochemical Co., Ltd (China), ENR (purity >98.0%) from
   Aladdin Industrial Corporation, CuSO[4]·5H[2]O (purity >99.0%) from
   Sangon Biotech (China) Co., Ltd., 1-phenyl-2-thiourea (PTU) from Yuanye
   Bio-Technology Co., Ltd. (China), neutral red and Nile red from Eon
   Chemical Technology Co., Ltd (China), and tricaine from Tixiai Co.,
   Ltd. (China).

2.2. Zebrafish husbandry, egg collection, and exposure tests

   Adult zebrafish were maintained in an automatic circulation system at
   28 °C and a 14 h/10 h light/dark cycle. Details on zebrafish husbandry
   and egg collection are given in Text S1. The developmental toxicity
   tests were conducted in 12-well plates with 10 embryos per well in a
   constant temperature incubator at 28 °C according to OECD236 [[64]20].
   The embryos were exposed to the single pollutant or mixtures from 12 to
   96 h post-fertilization (hpf). The exposure tests were conducted in
   triplicate (i.e., a total of 30 embryos for each treatment). The
   spontaneous embryo movement was counted by an inverted biomicroscope
   (JSZ6S, Olympus, Japan) in 1-min intervals at 24 hpf. The heart rate
   was determined by an inverted microscope (BX51, Olympus, Japan) via
   counting heartbeats of embryos in 30-s intervals at 48 hpf. The
   mortality rate and hatching rate at 96 hpf were recorded.

   To investigate the developmental toxicity of metals and antibiotics to
   zebrafish embryos, the stock solution of Cu (50 mM), Cd (50 mM), CIP
   (50 μM), or ENR (50 μM) was added to the E3 medium (2 mM CaCl[2],
   0.5 mM MgSO[4], 0.75 mM NaHCO[3] and 0.08 mM KCl, pH 7.0 ± 0.5, and
   electrical conductivity 289 ± 5 μS cm^−1) according to OECD236 [[65]20]
   to reach the metal or antibiotic concentration range of 0.0005–0.5 μM.
   For the toxicity tests of metal-antibiotic mixtures, the concentration
   of Cu or Cd was fixed at 0.2 μM, and the concentration ratio of metal
   to antibiotic was set as 0, 0.25, 0.5, 1, and 2. Control groups were
   free of Cu, Cd, CIP, or ENR. Detailed exposure conditions are shown in
   [66]Table S1. The method for detecting the interactions between ENR/CIP
   and Cu/Cd is described in Text S2.

2.3. In vivo fluorescence imaging and quantification of ROS, macrophage, and
lipid

   During the exposure tests, 200 μM PTU was added into the exposure
   solution to inhibit pigmentation at 24 hpf. The 96 hpf embryos exposed
   to 0.2 μM of metal or/and 0.2 μM of antibiotic were immersed in 25 μM
   of H[2]DCFDA for 40 min, 2.5 μg mL^−1 of neutral red for 25 min, and
   0.5 μg mL^−1 of Nile red for 30 min at 28 °C in the dark to determine
   the ROS production, the macrophages accumulation, and lipid content,
   respectively. After incubation, the embryos were rinsed with culture
   media and phosphate-buffered solution, and then were anesthetized with
   0.01% tricaine. The stained embryos (3 embryos for each group) were
   observed at excitation/emission wavelengths of 485/530 nm, 450/610 nm,
   and 530/635 nm, respectively, using an inverted fluorescence microscope
   (Nikon Eclipse Ti, Japan).

   Another fraction of the incubated embryos was homogenized by sonicating
   for 5 min (25 kHz). The suspensions were centrifuged at 5000 rpm for
   10 min, and then the supernatants were collected. The contents of ROS,
   macrophage, and lipid were determined by measuring the fluorescence of
   the supernatants by a fluorescence spectrometer (Cary Eclipse, Varian,
   USA) at excitation/emission of 485/530 nm, 450/610 nm, and 530/635 nm,
   respectively.

2.4. Determination of Cu and Cd concentrations in exposure solutions and
embryos

   The embryos were exposed to single Cu or Cd (0–0.5 μM) and
   metal-antibiotic mixtures in a 120-mm glass Petri dish with 100 embryos
   from 12 hpf to 96 hpf, the concentration of Cu or Cd in mixtures was
   fixed at 0.2 μM, and the concentration ratio of metal to antibiotic was
   set as 0, 0.25, 0.5, 1, and 2. Detailed exposure conditions are shown
   in [67]Table S1. Three replicates were independently performed for each
   treatment. Exposure solutions were replaced every day, and 10 mL
   exposure solutions prior to and after daily exposure were collected to
   quantify the concentration of metals in the solution before and after
   exposure. After exposure, the 96 hpf zebrafish embryos from the control
   and treatment groups were washed three times with ultrapure water, and
   the embryos were collected in a 10-mL centrifuge tube with 5 mL
   ultrapure water. The embryo suspensions were sonicated for 5 min
   (25 kHz), and then digested using 5 mL HNO[3] at 150 °C for 45 min on
   an electric heating board. After digestion, the residue was dissolved
   in 5 mL 2% HNO[3]. Cu and Cd concentrations in pre- and post-exposure
   solutions and digested solutions were quantified by inductively coupled
   plasma mass spectrometry (ICP-MS, X Series II, Thermofisher, USA).

2.5. Metabolomics analysis

   The 96 hpf embryos exposed to metals (0.2 μM), antibiotics (0.2 μM),
   metal (0.2 μM)-antibiotic (0.2 μM) mixtures were collected for
   metabolomics analysis, and the non-exposed 96 hpf embryos were used as
   the control. Zebrafish embryos were washed with ultrapure water, and a
   total of 350 (7 × 50) embryos were collected for each exposure group
   (i.e., 50 embryos in a 1.5 mL centrifuge tube per replicate, and 7
   replicates per treatment). Then, the metabolite extraction was
   performed. More detailed information on sample preparation and
   metabolomic analysis are described in Text S3, following our previous
   study [[68]21]. The differentially expressed metabolites (DEMs)
   (variable importance in projection [VIP] value of orthogonal partial
   least squares discriminant analysis [OPLS-DA] > 1 and the p value
   [Student's t-test] < 0.05) were selected.

2.6. Proteomics analysis

   The 96 hpf embryos exposed to metals (0.2 μM), antibiotics (0.2 μM),
   and metal (0.2 μM)-antibiotic (0.2 μM) mixtures were collected for
   proteomics analysis, with the non-exposure embryos as the control. For
   each treatment, 350 embryos were pooled as one sample and washed with
   ultrapure water, and the protein extraction and trypsin digestion were
   then performed. Differentially expressed proteins (DEPs) (fold change
   >1.50 or < 0.67) were selected. The DEPs data were processed using Gene
   Ontology (GO) annotations, Kyoto Encyclopedia of Genes and Genomes
   (KEGG) annotation, and subcellular localization. More detailed
   information on sample preparation and proteomic analysis is described
   in Text S4 [[69]22].

2.7. Statistical analysis

   SPSS 18.0 (PASW Statistics, IBM Company) was used for the statistical
   analysis. The data were presented as mean ± standard deviation (SD).
   One-way ANOVA was performed to compare the means of different exposure
   groups at p < 0.05. OPLS-DA was performed using SIMCA (V16.0.2)
   software. The Venn plots of DEPs were generated in Gephi 0.9.2. KEGG
   pathway classification of proteomics was performed according to the
   instruction of the database ([70]http://www.cusabio.cn/pathway.html).
   The protein-protein interaction (PPI) was drawn online using the STRING
   protein interaction network database ([71]https://www.string-db.org/).
   Figures were plotted using Origin 2020b.

3. Results and discussion

3.1. Developmental toxicity of metals or/and FQs

   In the single exposure groups ([72]Fig. 1), environmental
   concentrations of Cu, Cd, ENR, or CIP had no significant effects on the
   death rate, hatching rate, spontaneous movement, and heart rate of
   zebrafish embryos (p > 0.05). Similar to the present study, exposure to
   0.134 μM clarithromycin, 0.279 μM florfenicol, or 0.359 μM
   sulfamethazine did not induce the developmental inhibition of embryos
   [[73]11]. The heart rate of 48 hpf zebrafish embryos were not
   significantly affected after exposure to 1 μM of Cd [[74]23]. In
   addition, Johnson et al. [[75]24] found that exposure to 0.828 μM Cu
   did not induce the developmental inhibition (death rate, hatching rate,
   body length, and heart rate) of zebrafish embryos, while ≥1.45 μM of Cu
   could induce a significantly higher heart rate for zebrafish embryos.
   However, the heart rate of zebrafish embryos exposed to 4.25 mM CIP for
   4 h was significantly reduced due to the extremely high CIP
   concentration [[76]10].

Fig. 1.

   [77]Fig. 1
   [78]Open in a new tab

   Death rate (a and e) and hatching rate (b and f) of 96 hpf larvae,
   spontaneous embryo movement (Embryo mov.) of 24 hpf embryos (c and g),
   and heart rate of 48 hpf embryos (d and h) in the single exposure
   groups of metals (0–0.5 mM Cu or Cd) or FQs (0–0.5 mM ENR or CIP).

   Similarly, no significant effects on the death rate, hatching rate,
   spontaneous movement, and heart rate of zebrafish embryos were observed
   in the binary mixtures of Cu + ENR, Cu + CIP, Cd + ENR, and Cd + CIP
   ([79]Fig. 2). Therefore, both single pollutant and mixture exposures
   had insignificant developmental toxicity to zebrafish embryos, probably
   due to the low exposure concentrations in this study (≤0.6 μM). Willi
   et al. [[80]25] also observed that exposure to 0.056 nM and 0.561 nM
   mixtures of androgens and progestins had no observed effects
   (spontaneous muscle contractions, heart rate, edema, and paralysis) on
   zebrafish embryos, but altered the transcriptional expressions of genes
   (cyp19b, sult2st3, and cyp2k22). Thus, it is possible that molecular
   responses may be triggered by the low concentrations of Cu, Cd, ENR,
   and CIP individually or in mixtures, despite the insignificant
   developmental toxicity to zebrafish embryos.

Fig. 2.

   [81]Fig. 2
   [82]Open in a new tab

   Death rate (a and e) and hatching rate (b and f) of 96 hpf larvae,
   spontaneous embryo movement (Embryo mov.) of 24 hpf embryos (c and g),
   and heart rate of 48 hpf embryos (d and h) in the control group (Ctr)
   (without metal and FQs) and co-exposure groups of metals (0.2 mM Cu or
   Cd) and FQs (0–0.4 mM ENR or CIP).

3.2. Oxidative stress, inflammation, and lipid depletion in zebrafish larvae

   The in vivo fluorescence images of zebrafish larvae showed that the
   production of ROS was promoted when exposed to Cu, Cd, ENR, or CIP
   individually, compared with the control group ([83]Fig. 3a), and the
   increase of ROS depends on the concentration of the chemicals
   ([84]Figs. S3a and S3b). Previously, it was reported that the oxidative
   stress was induced when zebrafish embryos were exposed to 0.1 μM Cu or
   0.08 μM Cd [[85]26], and erythrocytes to 0.139 μM ENR or 0.151 μM CIP
   [[86]27]. In the single exposure groups, Cd induced greater ROS
   production than Cu ([87]Fig. 3d). Moreover, in general, exposure to
   metals resulted in higher ROS production than exposure to FQs
   ([88]Fig. 3d). The increase in ROS production indicates that oxidative
   stress occurred in zebrafish larvae. Moreover, ROS can target DNA,
   lipids, and proteins, and consequently affect cell physiological
   pathways, signaling cascades, and membrane properties, which ultimately
   cause inflammation and cell death [[89]28].

Fig. 3.

   [90]Fig. 3
   [91]Open in a new tab

   The in vivo fluorescence images of radical oxygen species (ROS) (a),
   macrophage (b), lipid (c) and the contents of ROS (d), macrophages (e),
   and lipid (f) in zebrafish larvae in the single and combined exposure
   groups of metals (0.2 mM) and fluoroquinolones (FQs) (0.2 mM).
   (Different letters suggest significant differences between different
   exposure groups, one-way ANOVA: p < 0.05).

   In the combined exposure groups, the addition of ENR or CIP decreased
   the ROS content in zebrafish larvae, compared with the larvae exposed
   to Cu or Cd individually ([92]Fig. 3d). This antagonistic effect of ENR
   and CIP on the oxidative stress induced by Cu or Cd in zebrafish was
   intriguing and probably resulted from the complexation of Cu and Cd
   with ENR and CIP (Text S5) and subsequent masking to ROS generation.

   Macrophages are an important part of the innate immune system, which
   are recruited to the inflammatory sites to participate in nonspecific
   (innate immunity) and specific defense (cellular immunity)
   [[93]29,[94]30]. Macrophages in zebrafish larvae were apparently
   elevated in all exposure groups compared with the control group
   ([95]Fig. 3b and e). Moreover, the increase of macrophages in zebrafish
   larvae was more pronounced in the combined exposure groups than in the
   single exposure groups. The results suggest that exposure to
   environmental concentrations of Cu, Cd, ENR, and CIP individually can
   cause inflammation in zebrafish larvae, and the inflammation response
   was aggravated when exposed to the metal-antibiotic mixtures.

   The fluorescence intensity of Nile red in zebrafish larvae decreased in
   all the exposure groups ([96]Fig. 3c and f), indicating that exposure
   to single or mixed metals and FQs reduced lipid content (or increased
   the lipid metabolism) in zebrafish larvae. Moreover, the decrease in
   lipid content in the combined exposure groups was greater than that in
   the single exposure groups. Similarly, previous studies also reported
   that exposure to Cd decreased the lipid content in crab, eel, and
   aquatic birds [[97][31], [98][32], [99][33]]. This may be due to the
   enhanced mitochondrial metabolism and the decreased synthesis of fatty
   acids [[100]31].

3.3. Bioaccumulation of metals in zebrafish embryos

   The reduced Cu and Cd in exposure solutions increased with increasing
   initial metal concentrations ([101]Fig. 4a and b), correspondingly
   resulting in an increased amount of Cu and Cd in zebrafish larvae
   ([102]Fig. 4e and f). Notably, in the combined exposure groups, the
   amounts of Cu and Cd both reduced in exposure solutions and accumulated
   in zebrafish larvae increased with increasing concentration of ENR or
   CIP ([103]Fig. 4c, d, g, and h). Thus, the addition of ENR or CIP
   promoted the uptake of metals by zebrafish embryos. This might be due
   to the uptake of metal–FQs complexes in addition to uptake of
   uncomplexed metals and/or a change in uptake and elimination kinetics
   [[104]34]. Indeed, the complexation of Cu and FQs occurred in the
   present study (Text S5, [105]Fig. S1, and [106]Fig. S2). Similarly, it
   was reported that the presence of ENR increased the absorption of Cd by
   Eisenia fetida due to the complexation between them [[107]34].
   Chlortetracycline facilitated the intracellular uptake of Cu and Cd by
   increasing the membrane permeability of Synechocystis sp [[108]35].
   Moreover, the number of reduced metals in the exposure solution was
   much more than the metals in zebrafish larvae, suggesting that most of
   the metals (73–95%) may be accumulated in the embryo's chorion. Wang
   et al. [[109]36] also found that more than 95% of Cu was accumulated in
   the chorion rather than in the embryo due to its strong adsorption to
   metals.

Fig. 4.

   [110]Fig. 4
   [111]Open in a new tab

   The total mass of metals reduced in exposure solutions (a–d) and
   accumulated in zebrafish larvae (e–h) in the single exposure groups of
   metals (0-0.5 mM) (a, b, e, and f) and co-exposure groups of metals
   (0.2 mM) and FQs (0–0.4 mM) (c, d, g, and h) (Different letters suggest
   significant differences between different exposure groups, one-way
   ANOVA: p < 0.05).

3.4. Metabolite responses to the exposure of metals or/and FQs

   In the metabolomics analysis, a total of 10644 peaks were detected in
   all the exposure groups, and 4587 were selected as DEMs. Volcano plots
   showed that the numbers of DEMs in the combined exposure groups
   ([112]Fig. 5 f-i) were higher than those in the single exposure groups
   ([113]Fig. 5a–d and j). Thus, exposure to the mixture of metals and FQs
   could more strongly disturb intracellular metabolites in zebrafish
   larvae than their single exposure. The stronger interference in the
   combined exposure groups was further confirmed by OPLS-DA
   ([114]Fig. 5e), which showed a farther distance between the binary
   exposure groups and the control group than those between the single
   exposure groups and the control group. This could be explained by the
   increased bioaccumulation of Cu and Cd in zebrafish larvae in the
   binary exposure groups ([115]Fig. 4g and h).

Fig. 5.

   [116]Fig. 5
   [117]Open in a new tab

   Volcano plots (a–d and f–i), orthogonal partial least squares
   discriminant analysis (OPLS-DA) (e), and sunflower plot of
   differentially expressed metabolites (DEMs) (j) in different exposure
   groups (Ctr: control; Cu/Cd concentration: 0.2 mM; ENR/CIP
   concentration: 0.2 mM).

   Among 4587 DEMs, only 385 DEMs were qualitatively identified. The
   identified DEMs mainly include organic acids and their derivatives
   (29.6–66.0%), lipids and lipid-like molecules (18.1–42.0%),
   organoheterocyclic compounds (6.3–13.0%), and nucleosides, nucleotides
   and analogs (4.0–6.6%) ([118]Fig. S4). Among them, organic acids and
   derivatives and lipids and lipid-like molecules were the most detected
   in all exposure groups. Furthermore, the majority of the organic acids
   and derivatives were up-regulated ([119]Fig. S5), whereas most of the
   lipids and lipid-like molecules were down-regulated in various exposure
   groups ([120]Fig. S6).

3.4.1. Organic acids and derivatives

   Most amino acids, as the majority of organic acids and derivatives,
   were upregulated in all the exposure groups ([121]Fig. S5). Moreover,
   the upregulation of amino acids in the mixture exposure groups was
   stronger than in the single exposure groups. Thereinto, glycine,
   glutamic acid, and serine are utilized to generate glutathione (GSH),
   which has antioxidant activity and detoxification effects, and plays an
   important role in the immune system [[122]37,[123]38]. The
   up-regulation of these metabolites ([124]Fig. 6a) implies the active
   responses of zebrafish larvae to external stimuli [[125]11,[126]39].
   Moreover, the previous study has reported that metal ions are highly
   chelated with some amino acid (i.e., threonine, isoleucine, and serine)
   through strong covalent bonds [[127]40], thus controlling the
   accumulation and release of metal ions in organisms. For example, when
   Synechocystis sp. exposed to Cu, the sequestration proteins bound to
   the extracellular Cu, and thus inhibited the excessive Cu uptake
   [[128]35]. Cd stress induced the production of metallothionein protein
   by Eisenia fetida effectively, detoxifying Cd through binding it in the
   tissue [[129]34].

Fig. 6.

   [130]Fig. 6
   [131]Open in a new tab

   Fold changes (FC) of amino acids (a) and fatty acids (b) related to
   inflammation, FC of the ratio of kynurenine to tryptophan (Kyn/Trp)
   (c), and heatmap of FC for LysoPLs (d) in different exposure groups
   compared to the control group (Cu/Cd concentration: 0.2 mM; ENR/CIP
   concentration: 0.2 mM).

   Lactate has been used as a marker of inflammation [[132]41]. Glutamine
   plays an important role in reducing inflammation by providing fuel for
   T-cell [[133]42]. Moreover, glycine also has anti-inflammatory and
   immunomodulatory effects via suppressing the activation of
   transcription factors, the formation of free radicals, and the
   production of inflammatory cytokines [[134]43]. Therefore, the
   up-regulation of lactate, glutamine, and glycine ([135]Fig. 6a)
   indicates the occurrence of inflammation in zebrafish larvae. In
   addition, the ratio of kynurenine to tryptophan (Kyn/Trp) has been
   widely used as a biomarker indicating cytokines (interferon-gamma)
   mediated inflammatory activity [[136]44]. Previous studies have
   reported the increase of Kyn/Trp ratio was positively correlated with
   inflammation and inflammation-related diseases [[137]45,[138]46]. The
   Kyn/Trp in all exposure groups also significantly increased
   ([139]Fig. 6c), indicating the occurrence of inflammation and immune
   activation [[140]47].

3.4.2. Lipids and lipid-like molecules

   Eicosapentaenoic acid (EPA), docosahexaenoic acid (DHA),
   docosapentaenoic acid (DPA), and oleic acid were significantly
   down-regulated, and the degree of down-regulation in the combined
   exposure groups was greater than those in the single exposure groups
   ([141]Fig. 6b). As EPA, DHA, DPA, and oleic acid can be used to produce
   specialized pro-resolving lipid mediators (SPM) [[142]48], their
   down-regulation could decrease SPM production. Because SPM plays a role
   in hindering the production of pro-inflammatory cytokines and prevents
   chronicity of inflammation and/or autoimmunity [[143]49], the lower SPM
   production may facilitate inflammation. Moreover, recent evidence
   indicates that insufficient production of SPM can trigger chronic
   inflammation, which is associated with several chronic diseases, such
   as chronic obstructive pulmonary disease, type-2 diabetes, and
   rheumatoid arthritis [[144][49], [145][50], [146][51]]. In addition,
   EPA and DHA, mainly found in fatty fishes, have health benefits such as
   anti-arrhythmic, hypolipidemic, and antithrombotic effects
   [[147]52,[148]53]. Thus, our results suggest that exposure to metals
   or/and FQs may reduce the nutritional value of freshwater fish.

   Lysophospholipids (LysoPLs), including lysophosphatidylcholine (LysoPC)
   and lysophosphatidyl ethanolamine (LysoPE), were down-regulated in all
   the exposure groups ([149]Fig. 6d). LysoPC had anti-inflammatory
   effects under various pathological conditions [[150]54], and the
   decrease of LysoPC was associated with many chronic
   inflammation-related diseases such as rheumatoid arthritis [[151]55],
   type-2 diabetes [[152]56], and liver cirrhosis [[153]57]. Furthermore,
   the decrease in LysoPC levels is considered a potential biomarker of
   cancer [[154]58]. On the other hand, LysoPE can stimulate chemotactic
   migration and cellular invasion in cancer by activating the
   mitogen-activated protein kinase (MAPK) signaling [[155]59,[156]60].
   Therefore, the down-regulation of LysoPC and LysoPE may induce
   inflammation and even increase the risk of cancer.

3.5. Protein responses to the exposure of metals or/and FQs

   A total of 4404 proteins were qualitatively detected by the proteomics
   analysis, and 1035 DEPs were identified in all the exposure groups. The
   number of DEPs induced by Cd (464) was higher than by Cu (169). The
   number of DEPs induced by ENR (448) was higher than that by CIP (103)
   ([157]Fig. 7a–d). The results are in agreement with their ROS
   production in zebrafish larvae ([158]Fig. 3d and e). Exposure to Cd or
   ENR induced more DEPs than other single and combined exposure groups
   ([159]Fig. 7a–d). Additionally, when comparing the protein responses of
   the combined exposure groups with their corresponding single exposure
   groups, no consistent patterns could be found, which may be due to
   complex up- and down-regulation of the myriad of proteins in response
   to pollutant exposure as a function of exposure duration and
   physiological conditions (such as growth stage) [[160]61]. This
   ambiguity on the relationship of exposure and protein patterns points
   to a challenge in proteomics, which should be further investigated.

Fig. 7.

   [161]Fig. 7
   [162]Open in a new tab

   Venn plots of differentially expressed proteins (DEPs) (The number in
   the figure means the number of DEPs.) (a–d), the number of DEPs related
   to response to stimulus (e) and immune system process (f), and fold
   change (FC) of antioxidases (g) and immune proteins (h) in the single
   and combined exposure groups of metals and antibiotics (Asterisk means
   significant differences between exposure groups and control group: FC >
   1.50 or < 0.67; Cu/Cd concentration: 0.2 mM; ENR/CIP concentration: 0.2
   mM).

3.5.1. GO functional classification of DEPs

   According to the GO functional classification ([163]Fig. S7 and
   [164]Fig. 7e and f), many proteins related to response to stimulus and
   immune system process were dysregulated significantly in all the
   exposure groups. This corresponds to the occurrence of inflammation and
   possible tumorigenesis in zebrafish larvae, which was also corroborated
   by the metabolomics analysis ([165]Fig. 6). Moreover, corresponding to
   the increase of ROS content ([166]Fig. 3a and d), the typical
   antioxidases, such as superoxide dismutase (sod1), glutathione
   S-transferase (gstp1), and peroxiredoxin (prdx1 and prdx6), were
   up-regulated in most of the exposure groups ([167]Fig. 7g), indicating
   the active response of zebrafish larvae to oxidative stress [[168]22].

   Additionally, the majority of the typical immune proteins, such as
   lysozyme (lyz), tumor necrosis factor receptor-associated protein
   (trap1), two subtypes of heat shock protein 90 (hsp90), hsp90a.1 and
   hsp90ab1, and migration inhibitory factor (mif) were up-regulated in
   all the exposure groups ([169]Fig. 7h). Among them, lyz, produced by
   leukocytes, is an essential part of innate immunity in fish [[170]62].
   Trap1 could promote inflammation by regulating the activation of
   proinflammatory ligands [[171]63]. Hsp90 is vital to maintain cellular
   homeostasis [[172]64], and the overexpression of hsp90 leads to
   enhanced proliferation, migration, invasion, and tube cell-dependent
   tumor angiogenesis in vitro and in vivo [[173]65]. The protein of
   macrophage mif could induce the inflammatory reaction by inhibiting the
   migration of macrophages, sustaining macrophage viability, and
   stimulating the expression of the proinflammatory cytokines
   [[174]66,[175]67]. Moreover, mif potentially promotes tumorigenesis by
   inhibiting the classic tumor suppressor gene p53 [[176]68]. Therefore,
   the up-regulation of the typical immune proteins further confirms the
   occurrence of inflammation and may also suggest a potential risk of
   tumorigenesis.

3.5.2. KEGG pathway analysis

   KEGG pathway enrichment analysis showed that many pathways related to
   tumor and immunity were significantly altered ([177]Fig. 8a). Among the
   enriched pathways related to immunity, three chronic disease pathways
   (inflammatory bowel disease, systemic lupus erythematosus, and
   rheumatoid arthritis) were associated with chronic inflammation,
   suggesting the occurrence of inflammation. Among the enriched pathways
   related to the tumor, the PI3K-Akt pathway is vital to the growth and
   survival of tumor cells [[178]69]. Focal adhesion, associated with the
   development of malignancy perturbations, plays an important role in the
   proliferation, survival, and migration of cells [[179]70].
   Extracellular matrix (ECM)-receptor interaction is critical to the
   formation of breast tumor, pancreatic tumor, neural tumor, and liver
   tumor [[180][71], [181][72], [182][73], [183][74]] by regulating
   cellular survival, proliferation, differentiation, shape, polarity, and
   motility [[184]75]. These three pathways were significantly
   dysregulated in most exposure groups except for CIP. Taken together,
   the results suggest that exposure to FQs or/metals caused significant
   changes in the pathways related to immunity and even might increase the
   potential risk of tumorigenesis in zebrafish larvae.

Fig. 8.

   [185]Fig. 8
   [186]Open in a new tab

   a, KEGG pathway enrichment of differentially expressed proteins (DEPs)
   in different exposure groups. b, Protein-protein interaction (PPI)
   network of 55 DEPs that involved in the pathways related to the tumor.
   c, Fold changes (FC) of proteins related to the tumor (Asterisk
   suggests fold change >1.5 or < 0.67; Cu/Cd concentration: 0.2 mM;
   ENR/CIP concentration: 0.2 mM).

   A total of 55 DEPs were involved in the pathways related to the tumor,
   and they can be classified into 3 major groups by the protein-protein
   interaction (PPI) network analysis ([187]Fig. 8b). Group 1 composed
   14-3-3 proteins (encoded by ywha-genes) and mapk. The 14-3-3 proteins
   are regulators of intracellular signaling pathways targeting oncogenic
   proteins [[188]76], and mapk is a complex interconnected signaling
   cascade with frequent involvement in the initiation and progression of
   the tumor [[189]77]. Moreover, the 14-3-3 proteins can activate mapk
   [[190]76]. Group 2 consisted of collagens (encoded by col-genes),
   fibronectin (encoded by fn1a-genes), and laminin (encoded by
   lama5a-genes), and they are the major components of the ECM [[191]78].
   Fibronectin and laminin provide the crucial connections for the
   interaction of other ECM components (collagens) and cells, which may
   mediate tumorigenesis [[192]75]. The major components of Group 3 were
   glutathione S transferase (encoded by gst-genes) and mapk. High levels
   of glutathione S transferase are closely related to oxidative stress
   induced by ROS [[193]22] and can promote cancer cell survival
   [[194]79]. Moreover, ROS could activate the downstream target mapk
   proteins [[195]80].

   Notably, PPI network analysis also revealed that Group 1 was closely
   related to Group 2 and Group 3, but no relation was observed between
   Group 2 and Group 3. The 14-3-3 proteins in Group 1 could participate
   in the adhesion of cells and ECM (fibronectin and laminin in Group 2),
   which were the key proteins linking Group 1 with Group 2
   ([196]Fig. S8). The adhesion of tumor cells may contribute to
   tumorigenesis [[197]81]. Group 3 was related to Group 1 through the
   interactions between different mapk proteins involved in these two
   groups. The mapk pathway is activated through the cascade of different
   mapk proteins [[198]77] in the order of map2k2a in Group1 and mapk8 and
   mapk10 in Group 3 ([199]Fig. S8), resulting in the close relationship
   between these two groups.

   In addition to the DEPs involved in the pathways related to the tumor,
   some ribosomal protein L (rpl), keratin (krt), and fatty acid-binding
   protein (fabp) also play important roles in tumorigenesis [[200][82],
   [201][83], [202][84]]. For example, the tumor suppressors, the proteins
   encoded by rpl36, rpl36a, rpl23, and rpl23a [[203]84], were
   down-regulated in the majority of the exposure groups ([204]Fig. 8c).
   This could enhance the tumor susceptibility [[205]85]. Krt4 and krt5
   are key genes that regulate lipoma and glioma, respectively
   [[206]82,[207]83], and the proteins encoded by these genes were
   significantly upregulated by the exposure to ENR and the Cd + ENR
   mixture, which may lead to the formation of lipoma and glioma
   ([208]Fig. 8c). Fabp2, fabp3, fabp7a, and fabp10a are vital to the
   development of hepatocellular carcinoma [[209]86], and the
   up-regulation of the proteins encoded by these genes implies the
   increased likelihood of hepatic tumorigenesis when exposed to
   environmental concentrations of metals and antibiotics ([210]Fig. 8c).

4. Conclusion

   This study investigated the developmental toxicity of FQs (ENR and CIP)
   and heavy metals (Cu and Cd), individually and in combination, toward
   zebrafish embryos. The molecular responses of zebrafish larvae to these
   pollutants were explored by the metabolomics and proteomics analysis.
   Exposure to single and binary mixture of FQs and metals did not cause
   any developmental toxicity to zebrafish embryos, but induced oxidative
   stress, inflammation, and lipid depletion. Moreover, the inflammation
   and lipid depletion were more severe in the mixture exposure groups
   than in the single exposure groups, probably due to the increased metal
   bioaccumulation in the presence of ENR or CIP. We observed inflammation
   and potential biomarkers of tumorigenesis in zebrafish larvae by using
   metabolomics and proteomics analysis, when the zebrafish embryos were
   exposed to a no-observed-effect concentration of FQs and metals. The
   findings highlight the potential toxicological risks of pollutants at
   environmentally relevant trace levels, which may be overlooked in the
   acute toxicity tests. Moreover, FQs and heavy metals are ubiquitous in
   natural waters, and thereby their potential ecological and human health
   risks should not be ignored. This study was limited to the early
   development stage of zebrafish (≤96 hpf). In a natural aquatic
   ecosystem, chronic exposure at low pollutant concentrations might
   further contribute to the complexity and uncertainty in the toxicity of
   pollutant mixtures. Therefore, future studies examining the long-term
   toxicological effects of environmental concentrations of pollutants are
   encouraged.

Declaration of competing interest

   The authors declare that they have no known competing financial
   interests or personal relationships that could have appeared to
   influence the work reported in this paper.

Acknowledgements