1. Introduction
Zearalenone (ZEA) is a major fungal toxin produced by diverse
Fusarium species and is a significant global hazard to human and animal health [
1], [
2]. Due to its high stability, ZEA can be detected in a wide range of sources worldwide. In particular, cereal crops, such as barley, oats, corn, wheat, and rice, are highly vulnerable to ZEA contamination during the drying, processing, transportation, and storage stages, with contamination levels ranging from a few μg∙kg
−1 to several hundred mg∙kg
−1 and beyond [
3], [
4], [
5]. Additionally, ZEA can be found in spices, dried fruits, dried vegetables, and beverages such as milk and beer [
5]. Environmental samples, such as river water and wastewater, have also been shown to contain ZEA [
6], [
7]. ZEA contamination from diverse sources underscores the multiple exposure pathways through which animals can be exposed to this mycotoxin.
Animals are typically exposed to ZEA through the consumption of contaminated grains. However, ingested ZEA has been linked to various toxic effects, including hepatotoxicity, immunotoxicity, genotoxicity, reproductive toxicity, and potential carcinogenicity [
3], [
5], [
8]. ZEA causes liver damage in mice following exposure by affecting liver enzyme activity and triggering lipid peroxidation and oxidative stress [
9], [
10]. In immune cells such as T lymphocytes, ZEA overactivates the mitogen-activated protein kinase (MAPK) signaling pathway, leading to apoptosis and compromised immune function [
11]. Furthermore, ZEA inhibits proliferation, disrupts the cell cycle, and affects the transcriptional profile of intestinal porcine epithelial cell line-J2 (IPEC-J2) cells [
12]. In mouse Sertoli cells, ZEA triggers the accumulation of reactive oxygen species, resulting in endoplasmic reticulum stress, inhibition of cell proliferation, cell cycle arrest, and apoptosis [
13]. The mechanisms underlying ZEA toxicity are complex and involve various cellular processes. These toxic effects can occur individually or in combination and cause significant harm to various animal tissues and organs. In view of this, many countries set limits on the amount of ZEA in cereals and cereal-derived foods [
3], [
4], [
5]. In addition, the correlation between ZEA and human health remains unclear, as highlighted by the European Food Safety Authority. The International Agency for Research on Cancer classifies ZEA as a Class 3 carcinogen [
8].
Given the adverse effects of ZEA on humans and animals, various methods have been implemented to minimize contamination by removing ZEA from food and raw feed materials [
5]. However, no developed approach can entirely eliminate ZEA from crops and the uncleared toxins gradually accumulate in the body following consumption, posing a threat to human and animal health. Natural products (NPs) possess diverse functional activities that make them valuable for mitigating the adverse effects of mycotoxins, including ZEA. Consequently, many researchers have explored the potential of NPs to protect against the toxic effects of ZEA and minimize its detrimental effects [
14]. A growing number of studies have provided evidence for the protective effects of nutritional interventions using NPs [
10], [
14]. Incorporating these NPs into animal diets shows promise in mitigating the toxicities associated with ZEA. However, the discovery of effective NPs for detoxification purposes has often been accidental. These substances exhibit varying potencies and specific roles in regulating animals and cells. Furthermore, the identification of new potential detoxifying agents has been hindered by the prevailing practice of conducting only a single experimental attempt during NP development. Efforts have often focused on testing only a small number of substances, which hampers the comprehensive evaluation of the detoxification potential of a broader range of compounds [
14]. Therefore, a systematic and comprehensive approach is required to identify and evaluate a broader range of compounds with potential detoxification properties.
Zebrafish are a valuable model for conducting phenotype-based small-molecule screening assessments that can be translated into mammalian pharmacology [
15], [
16], [
17]. Their utilization in such screening has proven to be highly impactful and has the potential to discover more effective compounds [
17], [
18]. In contrast to traditional approaches that rely on single attempts, taking advantage of the suitability of zebrafish for throughput and small-molecule screening, researchers can screen a small-molecule library to discover compounds that fulfill their criteria [
18]. More importantly, substances screened on the same platform are more likely to establish evaluation criteria for potency and efficacy, thereby accelerating the development and utilization of these substances. Phenotype-based screening techniques in whole organisms have proven effective for discovering powerful new compounds with unique activities that target unexpected
in vivo targets [
17], [
19].
As the demand for novel antidotes continues to grow, we aimed to establish a systematic strategy for NP discovery. Here, we describe a phenotype-based screening approach to identify compounds that mitigate or counteract the adverse effects of ZEA exposure in animals. Using this strategy, we first identified 96 NPs and assessed the potency and efficacy of several effective candidate compounds based on the embryonic phenotype and locomotor activity using a scoring system and TCMacro script. Subsequently, we performed transcriptome and protein-protein interaction (PPI) network analyses to extract specific mRNA signatures for querying predicted compounds in the Connectivity Map (CMap) database. These predicted compounds, which could potentially reverse the gene expression profiles associated with ZEA toxicity, were further analyzed using our model to identify more effective NPs. Finally, the most promising NPs exhibiting high potency were experimentally validated to demonstrate the potential of the current strategy. This multistep approach leverages phenotypic assessment, transcriptomics, PPI networks, CMap analysis, and experimental validation to increase the likelihood of finding highly efficient and reliable NPs with detoxification potential.
2. Materials and methods
2.1. Chemicals
ZEA was purchased from Sigma (USA) and dissolved in dimethyl sulfoxide (DMSO) to obtain stock solutions of 20 mg∙mL−1 before being stored at −20 °C. Prior to each experiment, test solutions of ZEA were made using a culture medium consisting of CaCl2∙2H2O (294.0 mg∙L−1), MgSO4∙7H2O (123.3 mg∙L−1), NaHCO3 (63.0 mg∙L−1), and KCl (5.5 mg∙L−1). The chemicals in this part used were of analytical grade and obtained from Sinopharm Chemical Reagent Co. Ltd. (China). The screening process involved two stages: random screening (first screening) and CMap analysis screening (second screening). A commercially available NP library comprising 3948 small molecules (HY-L021; MCE, USA) was selected for its high bioactivity and functional roles. Stock solutions of all NPs were prepared in either 100% DMSO or H2O at a concentration of 10 mmol∙L−1 and stored at −80 °C.
2.2. Zebrafish husbandry
Adult zebrafish of the AB strain were fed brine shrimp (
Artemia nauplii) twice daily. They were kept in an aquarium system with flowing water and a 14-h light cycle followed by a 10-h dark period at a constant temperature of (28 ± 1) °C. Zebrafish were placed in spawning cages in the afternoon before the experiment, and eggs were synchronously produced the following day. After spawning for half an hour, zebrafish embryos were collected, washed with culture medium, and then incubated in Petri dishes filled with culture medium in an incubator at (28 ± 1) °C. Within 1.25 hours post-fertilization (hpf), a stereomicroscope (Olympus SZ61; Olympus Corporation, Japan) was used to assess embryos; unfertilized or defective embryos were excluded from the study. For further details on the process, refer to the toxicity test guidelines provided by the Organization for Economic Co-operation and Development (OECD, 2013) [
20]. All experiments were approved by the Institutional Animal Care and Use Committee of the Sichuan Agricultural University.
2.3. Exposure experiment design
2.3.1. Gradient concentration experiments
Zebrafish embryos at 6 hpf were used for ZEA-induced gradient toxicity experiments. The tests were performed in 96-well plates, with one embryo per well or in six-well plates containing 20 embryos per well. The wells were filled with test solutions containing different concentrations of ZEA, ranging from 0 to 2500 μg∙L−1. Incubation occurred at a temperature of (28 ± 1) °C. In the 96-well plates, the final liquid volume was 200 μL per well, while in the six-well plates, it was 4 mL per well. In addition, each exposure group, including the control group, was administered 0.025% DMSO. The exposure solutions were replaced every 24 h, and all embryos and larvae were monitored simultaneously to ensure that dead individuals were promptly removed.
2.3.2. Initial screen
At 5 hpf, individual zebrafish embryo was transferred to the separate wells of a 96-well plate. The first round of screening was performed at 6 hpf. Specifically, 10, 20, 40, or 80 µmol∙L−1 of each NP was co-dosed with 2000 μg∙L−1 of ZEA in three replicate wells. The alleviating effect was evaluated at 24 and 48 hpf through manual assessment of toxic phenotypes, such as shortened tail, smaller head/eyes, and opaque tissues, using light microscopy. To be considered rescued from ZEA-induced toxicity, zebrafish embryos were required to show alleviation of all three phenotypes. The number of embryos recovered from the toxic phenotype for each compound was calculated, and a hit was declared if a compound rescued at least 2/3 of the fish from the ZEA-induced phenotypes. The test solution did not change during the culture. All compounds that qualified as hits were tested individually during the second screening round.
2.3.3. Secondary screen
Compounds that showed activity in the initial screening test were subjected to concentration-response experiments to create concentration-response curves, which were then verified. In the secondary screen experiments, test solutions of 2000 μg∙L−1 ZEA and different concentrations of hit compounds were added to each well of the six-well plates containing 20 embryos. The control and exposure groups were treated with 0.05% DMSO and all other settings were the same as those used in the gradient concentration experiment. The potency of each compound in the concentration-response experiment was calculated as the 50% effective concentration (EC50).
2.4. Embryo/larvae toxicity assessments
2.4.1. Malformation, mortality, and hatching rates
The exposed embryos and larvae were assessed daily using a stereomicroscope (Olympus SZ61; Olympus Corporation). Malformation and mortality rates were determined at 24, 48, 72, 96, and 120 hpf, whereas hatching rates were recorded at 72, 96, and 120 hpf. Embryonic coagulation, missing heartbeat, lack of somite formation, and failure of tailbud detachment from the yolk sac were used as indicators of mortality. The detailed procedure can be found in the toxicity test guidelines provided by OECD (2013) [
20].
2.4.2. Morphological characteristics
At 72 and 120 hpf, zebrafish larvae were anesthetized with 0.016% MS-222 (Sigma) and imaged using a stereomicroscope (Olympus SZ61; Olympus Corporation). The morphological features of the larvae were evaluated according to previously described criteria [
21]. All measurements, such as those for body length (BL), distance from mouth to anus (DMA), and swimming bladder area (SBA), were conducted using ImageJ software (ImageJ, USA).
2.4.3. Toxicity score
Larvae were evaluated for malformation and hatching status according to previously described methods, with slight modifications [
22]. The detailed methods are presented in Appendix A. The scoring rules are listed in Appendix A Table S1.
2.5. Potency analysis
To determine EC
50, we followed previously described methods with minor modifications [
23]. The detailed methods are presented in Appendix A.
2.6. Efficacy analysis
Efficacious compounds from the secondary screening were selected for further analysis. A spontaneous movement-based efficacy analysis was conducted for each compound of interest to investigate its potential to counteract ZEA-induced toxicity. Spontaneous movement was quantified using the TCMacro method with ImageJ software (ImageJ), with slight modifications to previous methods [
24]. The detailed methods are presented in Appendix A.
2.7. RNA-seq data collection and analysis
The analysis workflow is described in Appendix A.
2.8. RNA extraction and quantitative real-time PCR
The detailed methods are presented in Appendix A. The specific primer sequences are listed in Appendix A Table S2, with β-actin as the reference gene.
2.9. Western blotting
Western blotting was performed as previously described [
25]. Additional details are provided in Appendix A.
2.10. Analysis of cell apoptosis
Detailed methods are listed in Appendix A.
2.11. Analysis of cell cycle
Propidium iodide (PI) staining and flow cytometry were used to analyze cell cycle progression. Staining was performed as previously described [
26]. The detailed methods are presented in Appendix A.
2.12. Analysis of cell senescence
Detailed methods are listed in Appendix A.
2.13. Birefringence analysis
Muscle birefringence was analyzed as described previously, with slight modifications [
27]. The detailed methods are presented in Appendix A.
2.14. Statistical analyses
Statistical analyses were performed using GraphPad Prism Version 9.5.0 (Graph Pad Software Inc., USA). Data were assessed for normal distribution using the D’Agostino−Pearson omnibus and Shapiro-Wilk tests. The statistical methods are stated in the figure legends. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 denote significant differences. Box plot whisker limits indicate each cohort’s minimum and maximum values, whereas error bars in all lines and bar graphs show the standard deviation.
3. Results
3.1. ZEA induces a clear concentration toxicity response in zebrafish
We analyzed the outcomes obtained from each concentration in the gradient concentration test to construct an appropriate screening model. The assessment involved the morphological defects and the lowest observed effect concentration (LOEC) in zebrafish embryos or larvae after exposure to ZEA for 24, 48, 72, 96, and 120 h. Results indicated that at 24 hpf, 80% of embryos treated with 1500 μg∙L
−1 displayed altered head and tail morphology (
Figs. 1(a) and
(b), and Appendix A Fig. S1). Additionally, from 48 to 120 hpf, embryos exposed to concentrations equal to or greater than 1500 μg∙L
−1 exhibited yolk sac or pericardial edema. However, ZEA concentrations that were equivalent to or less than 1000 μg∙L
−1 did not elicit any abnormalities under the established time points, and the morphology of embryos was comparable to that of the controls. Altogether, 1500 μg∙L
−1 was the lowest concentration causing extensive morphological abnormalities (Appendix A Table S3).
Next, we examined the hatching and mortality rates. The hatching delay was observed at ZEA concentrations as low as 500 μg∙L
−1 (
Fig. 1(c)), confirming that the ZEA toxicity effect is present even at the lowest concentrations. Specifically, all embryos hatched by 96 hpf when the concentration of ZEA was equal to or lower than 1000 μg∙L
−1. However, results varied at higher concentrations, with embryos exposed to 2000 μg∙L
−1 and above failing to hatch and exhibiting severe growth abnormalities throughout the 5-d experimental period. Likewise, at high concentrations (2000 μg∙L
−1 and above), ZEA toxicity triggered embryonic lethality (
Fig. 1(d)) due to a more severe defect, whereas low concentrations (equal to or below 1000 μg∙L
−1) did not. It is important to note that ZEA toxicity is dose-dependent. The results can be better summarized in terms of the concentration that caused lethality in 50% of larvae (LC
50), concentration that caused any malformation in 50% of larvae (MC
50), and concentration that interfered with hatching in 50% of larvae (IC
50) (Appendix A Table S4). The LC
50 value was initially 2380 μg∙L
−1, whereas the MC
50 value was initially 1347 μg∙L
−1. Over time, these values decreased, reaching approximately 1694 and 1241 μg∙L
−1, respectively. The IC
50 ranged from 1143 to 1531 μg∙L
−1. These results support the conclusion that the combined effects of ZEA toxicity are concentration-dependent, varying according to the time of exposure and concentration tested.
Lower concentrations of ZEA (equal to or below 1000 μg∙L
−1) had fewer associated details compared to higher concentrations. Given that ZEA elicited a hatching delay, it is possible that the ZEA toxicity effect appears at low concentrations. To explore this further and to capture more information on the concentration gradient effect of ZEA, we separately quantified the DMA, eye area, heart area, SBA, head height, and yolk sac area of zebrafish at 72 or 120 hpf (
Figs. 1(e)-(h), and Appendix A Fig. S2). Exposure to 1000 μg∙L
−1 led to a significantly decreased DMA (
P < 0.05) at 72 hpf compared with that of the control (
Fig. 1(e)). In addition, exposure to 1000 μg∙L
−1 resulted in a significantly reduced area of the eye and heart (
P < 0.05) at 72 hpf (
Figs. 1(f) and
(g)). However, these effects were largely absent after 120 hpf. These results suggest that, unlike other malformations observed at 1500 μg∙L
−1 ZEA, the adverse effects of 1000 μg∙L
−1 ZEA may be derived from the inhibition of growth, some of which were able to recover. Notably, 50% of embryos treated with 1000 μg∙L
−1 ZEA exhibited a narrow or disappeared SBA at 120 hpf (
Fig. 1(h)), suggesting persistent toxic effects of ZEA. In contrast, embryos treated with 1500 μg∙L
−1 ZEA showed sustained and more severe morphological changes over time compared to controls (
Figs. 1(e)-(h)). In addition, exposure to 1500 μg∙L
−1 ZEA also altered the head height and yolk sac area (Figs. S2(a) and (b)). These results support the observation that a lower concentration (1000 μg∙L
−1) suppressed the growth of zebrafish.
Based on these observations, the toxic effects were observed at all of the tested ZEA concentrations. ZEA showed more gradual transitions between no effect and lethality, where the intermediate concentrations between the lethal and no effect concentrations caused hatching delay or produced growth inhibition/ or significant malformations. These results are summarized in Appendix A Table S5.
3.2. Delineation of toxic phenotype
Above the concentration of 1500 μg∙L−1, we observed a previously unidentified toxic phenotype characterized by abnormalities such as a shortened tail, smaller head/eyes, and opaque tissues (Fig. S1). The severity of these abnormalities was influenced by both exposure time and ZEA dose. Specifically, increasing concentration gradually darkened head brightness and reduced embryo tail length in all embryos present at 24 hpf under dechorionation conditions. Importantly, the embryos from different concentrations (1000, 1500, 2000, and 2500 μg∙L−1) were clearly distinguished based on the presence, absence, or severity of these phenotypes.
The observed phenotypes can be roughly divided into mild, moderate, and severe (Fig. S1). Mild abnormalities (ZEA1500) were characterized by a partially opaque head and tail, with less well-defined structures than in controls. Moderate abnormalities (ZEA2000) were characterized by mostly opaque head/eye, trunk, and tail structures, with some structures disappearing completely. Activity was completely abolished in embryos with severe abnormalities (ZEA2500), accompanied by markedly reduced head and eye sizes and tail lengths. We observed that the probability of these abnormalities at 2000 μg∙L−1 was 100% (mainly moderate abnormalities) in all 24 hpf embryos, and the phenotype assay could be scaled to a 96-well plate format without further dechorionation conditions. This suggests that the toxic phenotype can be appropriately used for rapid-throughput screening.
3.3. Toxic phenotype-based chemical screening
Given that phenotypes at 24 hpf can serve as indicators of toxicity levels, we hypothesized that such phenotypes could facilitate rapid-throughput screening of substances with detoxification potential. To test this hypothesis, we first tested whether different NPs could mitigate the severity of these abnormalities. A total of 96 NPs were tested at four different concentrations. Appendix A Table S6 contains a list of all NPs tested, CAS numbers and the results from the screening concentrations of 10, 20, 40, and 80 μmol∙L
−1. Three percent of the hits (three different NPs) exhibited clear improvements in phenotype and enhanced embryo activity. One of the NPs, baicalin, has been shown to attenuate ZEA toxicity in animal models [
28]. Similarly, baicalin was determined as a potential mitigation molecule in the zebrafish phenotype assay based on its alleviating effect. The second NP, hydroxytyrosol, can effectively relieve the toxicities of multiple mycotoxins (such as deoxynivalenol and ochratoxin-A) [
29], [
30]. However, its effect on ZEA toxicity remains unclear. The final NP, fraxetin, was newly identified in this study. Notably, 10 μmol∙L
−1 fraxetin was sufficient to produce obvious toxicity-alleviating effects against ZEA exposure (Appendix A Fig. S3(a)), whereas baicalin and hydroxytyrosol required 40 μmol∙L
−1 for similar effects. These three NPs were used for further analyses.
To further validate whether the three NPs were potential detoxification substances for ZEA, we performed concentration-response experiments in zebrafish after 72 h of exposure. Malformation, mortality, and hatching rates were used to describe remission and improvement outcomes. We then established a concentration-response curve according to the malformation, mortality, and hatching results. Based on these results (
Fig. 2), two NPs (fraxetin and hydroxytyrosol) were observed to have alleviating effects on ZEA toxicity and one (baicalin) had unclear effects (with apparent morphological abnormalities or other toxicities; Appendix A Fig. S3(b)). However, there were some differences between the potencies of the two NPs. Fraxetin, the most potent NP (
Fig. 2(e)), had EC
50 values below 5 μmol∙L
−1 (4.57 μmol∙L
−1), whereas hydroxytyrosol was approximately 20-fold less potent, with an EC
50 of 92.08 μmol∙L
−1 (
Fig. 2(j)). Surprisingly, fraxetin was sufficient to fully restore the malformation and mortality rates in the concentration range tested (
Figs. 2(a)-(c)). Although comparing toxicity-alleviating effects across experiments is difficult, no NPs with high potency, such as fraxetin, have been reported [
14]. These examples demonstrate that an
in vivo zebrafish model combined with phenotypic analysis can be an efficient screening tool for identifying and developing substances with more efficient detoxification activity.
3.4. Spontaneous movement-based efficacy analysis
We attempted to develop an assay that could better compare the detoxification capabilities of these NPs against typical mycotoxins like ZEA. Our data suggest that exposure to varying ZEA concentrations can lead to changes in zebrafish fate, including hatching delays and growth inhibition, which have garnered considerable attention. Based on the toxic phenotype, we speculated that ZEA toxicity likely influenced the pattern of spontaneous movement. In this case, behavioral assays may help evaluate the toxicity induced by ZEA. To test this possibility, we characterized the spontaneous movement parameters in 19-22 hpf embryos using the TCMacro method. Tail coiling during spontaneous movement is typically observed at 17 hpf, with frequency peaking and then declining thereafter [
31]. Our results indicated significant differences in tail coiling frequency distribution at all tested ZEA concentrations. For instance, at 19 hpf, the control group exhibited a predominant tail coiling frequency distribution of 50%-100%, whereas the ZEA-exposed group predominantly showed a distribution of 0-50% (
Fig. 3(a)). The magnitude of this effect was concentration-dependent, suggesting that ZEA affected spontaneous movement in a concentration-dependent manner. Remarkably, the changes in the coil frequency were not purely inhibitory. From the frequency-time curves, the frequency peak in the control group occurred at 21 hpf and began to decrease thereafter (
Fig. 3(b)). The frequency peak observed at 500 or 1000 μg∙L
−1 lagged behind the control peak by approximately 1-2 h. During the tested period from 19 to 22 hpf, the frequency peak in the 500 μg∙L
−1 continuously shifted from 37% to 77%, eventually achieving levels comparable to the control (77.40% vs 77.29%). Conversely, higher concentrations of ZEA (1500 μg∙L
−1 and above) further hindered spontaneous movement function, resulting in lower frequency and decreased movement intensity (
Fig. 3(c)). Achieving the maximum frequency peak at these concentrations became challenging. Additionally, we generated a coiling frequency and intensity barcode matrix (
Fig. 3(d)), demonstrating the efficacy of this approach in efficiently distinguishing between groups, even at the lowest concentration tested. Overall, these findings confirmed that the spontaneous movement assay yields more detailed information than anticipated.
Subsequently, we investigated whether the most effective dose of fraxetin (20 and 40 μmol∙L
−1) or hydroxytyrosol (160 and 320 μmol∙L
−1) could fully restore ZEA toxicity (1500 μg∙L
−1) by conducting a coiling movement assay. We found that two concentrations of fraxetin, 20 and 40 μmol∙L
−1, counteracted the toxicity effects of ZEA in zebrafish, rescuing coiling frequency and intensity to almost normal levels (
Fig. 3(e)). The frequency/intensity
-time curves of fraxetin were very similar to those of the controls (
Fig. 3(f)), indicating that fraxetin was sufficient to recover zebrafish growth. Hydroxytyrosol at 160 μmol∙L
−1 partially counteracted the toxicity effects of ZEA, restoring a low toxicity barcode (approximately 1000 μg∙L
−1) to ZEA-treated animals (
Fig. 3(e)). As the concentration of hydroxytyrosol increased, there was a slight increase in the alleviation effect. However, the coiling frequency and intensity were still weaker than those of the control or fraxetin-treated embryos under the same conditions (
Figs. 3(f) and
(g)), indicating a weak ability of hydroxytyrosol to detoxify ZEA. This was also observed in the malformation rate results (
Figs. 2(a), (b), (f), and
(g)). Thus, it may be possible to establish behavioral assays for ZEA in zebrafish to compare the detoxifying efficacies of similar NPs. This provides a more convenient method for visualizing the alleviation effect.
In addition to the efficacy evaluation based on behavioral assays, we validated the role of fraxetin and hydroxytyrosol on growth by separately quantifying features in the head, eye, heart, yolk sac, and swimming bladder of 72 or 120 hpf zebrafish. Two concentrations of fraxetin, 20 and 40 μmol∙L
−1, significantly reversed growth inhibition and morphological abnormalities induced by ZEA (
Figs. 4(a)-(g)). DMA and BL levels in the fraxetin group were higher than those in the controls (
Fig. 4(b) and Appendix A Fig. S4(a)), indicating a remarkable alleviation. Furthermore, fraxetin may have promoted the absorption of the yolk sac (
P = 0.09) to keep in pace with zebrafish growth (
Fig. 4(d)). The principal component analysis (PCA)/clustering analysis of all morphological parameters showed no significant phenotypic differences among the three groups (
Fig. 4(h)). Both 160 and 320 μmol∙L
−1 concentrations of hydroxytyrosol partially counteracted the toxicity effects of ZEA (
Figs. 4(i)-(l), Appendix A Figs. S4(b) and (c)). A small head and disappearing swimming bladder were still visible at 72 and 120 hpf (
Figs. 4(i) and
(k)). Cluster differences were also apparent in PCA (
Fig. 4(j)). All morphological parameters were weaker than those of the control, indicating that the ability of hydroxytyrosol to detoxify ZEA was substantially weak. This finding was consistent with the predictions of the ZEA behavioral assays.
3.5. Identification of hub nodes and subnetworks in gene network analysis of ZEA toxicity test
The alleviating effect of fraxetin was highly effective, which prompted us to further analyze the effectiveness of functionally related molecules. We first studied the change in gene expression profiles of 24 hpf zebrafish embryos treated with 1500 μg∙L
−1 ZEA compared with those of DMSO-treated controls (see Section 2). To visualize the biological network relationships among these differentially expressed genes (DEGs), we performed a gene network analysis in Cytoscape using STRING
† (Appendix A Fig. S5). The cytoHubba plugin was used to identify densely connected subnetworks with differential expression and to assess the network hubs that could indicate vital regulatory nodes in the network. We identified 10 key nodes using the MCC, DMNC, MNC, Degree, and EPC algorithms (Appendix A Table S7). We used Cytoscape to remap the interaction layout and narrowed our search until we identified a defined subnetwork consisting of 79 DEGs (
Fig. 5(a)). In this subnetwork,
p53 emerged as the largest node in the zebrafish toxicity test, with a degree of 56. Previous studies in other cell lines have indicated that
p53 is critical for ZEA toxicity [
32], [
33]. This verifies the existence of comparable gene networks in zebrafish and other cells that respond to ZEA toxicity. To confirm the core location of the entire subnetwork, we used a clustering algorithm (MCODE) to generate three clusters: clusters 1
-3.
3.6. Fraxetin reverses the alteration of the signaling pathway induced by ZEA
Next, we focused on the pathway of the subnetwork and the role of the cluster nodes. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was used to examine the association with known pathways. Functional enrichment analysis revealed that the entire subnetwork was mainly involved in cell growth and death (Appendix A Fig. S6(a)), consistent with observed toxic effects in ZEA-exposed zebrafish, such as opacity and reduced body size. These genes were predominantly involved in protein processing in the endoplasmic reticulum, p53 signaling pathway, MAPK signaling pathway, apoptosis, FoxO signaling pathway, cell cycle, and cell senescence (Fig. S6(b)). Notably, apoptosis, cell cycle, and cell senescence are known to be downstream of regulatory networks. Furthermore, a few induced genes were identified via KEGG pathway enrichment analysis (
Figs. 5(b) and
(c)), which participate in various signaling pathways and could be simultaneously required by multiple pathways. Many of these genes also belonged to the cluster nodes. To determine the impact of fraxetin, we analyzed its ability to reverse specific gene effects. Fraxetin treatment significantly reduced gene expression, with 40 μmol∙L
−1 recovering nine test genes to control levels compared to five at 20 μmol∙L
−1, potentially mitigating ZEA toxicity (
Fig. 5(d)). Based on this analysis, we examined the expression levels of the key node protein p53 (Appendix A Fig. S7). As expected, fraxetin significantly decreased ZEA-induced p53 expression. These results support the role of fraxetin in reversing ZEA toxicity.
3.7. Fraxetin markedly represses apoptosis, G2 phase arrest, and cell senescence triggered by ZEA
Based on the above-described results, ZEA toxicity seems to be related to apoptosis, cell cycle regulation, or cell senescence, suggesting that it directly affects cell fate. To determine how fraxetin alters cell growth and death, we used acridine orange (AO) staining to examine the apoptotic patterns in zebrafish embryonic cells with or without fraxetin treatment. We observed that ZEA substantially increased apoptosis in these cells compared to those in control zebrafish (
Figs. 6(a)-(d)). However, co-treatment with fraxetin reversed the apoptosis induced by ZEA treatment alone. This effect was most evident on the dorsal side of the notochord (DSN). We confirmed these results using Hoechst staining, which showed that fraxetin reduced nuclear condensation and fragmentation (
Fig. 6(e)). These findings indicate that fraxetin has a detoxifying effect in zebrafish, which helps prevent ZEA-induced cell death and ultimately improves cell survival upon toxic damage.
Additionally, prior studies have demonstrated ZEA’s ability to induce G2 phase cell cycle arrest in mouse cells [
13], [
34], [
35]. Our pathway analysis also revealed the enrichment of cell cycle-related processes. To validate this in the zebrafish model, we analyzed the cell cycle using flow cytometry and observed an increase in the G2 fraction after ZEA treatment (
Figs. 6(f) and
(g)). Interestingly, fraxetin treatment reversed the ZEA-induced cell cycle arrest, suggesting that fraxetin could restore cell growth in zebrafish by modifying the cell cycle.
Next, we investigated whether fraxetin treatment prevents cell senescence. Senescent cells were detected through a senescence associated β-galactosidase (SA-β-gal) assay in zebrafish at 24 and 72 hpf. No significant change in β-galactosidase staining was observed at 24 hpf in zebrafish (
Figs. 6(h) and
(i)). However, at 72 hpf, ZEA treatment resulted in an apparent increase in the number of senescent cells (
Figs. 6(j) and
(k)). Consistent with apoptosis and cell cycle arrest, coexposure to fraxetin completely reversed this effect. Together, these findings reinforce the screening results, indicating that fraxetin is a scavenger of ZEA toxicity.
3.8. CMap analysis can be used to identify more potent and efficacious NPs
Alterations in signaling pathways may play a crucial role in ZEA toxicity. Identifying compounds that can reverse these gene expression profiles could be an effective strategy for screening detoxification substances for ZEA. To perform this search, we used CMap
†, a database of gene expression profiles generated from human cells treated with a wide range of bioactive, small molecules [
36]. Using a gene signature of interest makes it possible to identify compounds that can induce the desired changes in gene expression [
36], [
37]. Therefore, the first signature (all DEGs;
Fig. 7(a) and Appendix A Table S8) was queried against the CMap platform. Interestingly, preliminary results indicated that fraxetin gene expression was inversely correlated with ZEA in at least five cell lines (Appendix A Fig. S8(a)). This suggests the feasibility of the CMap analysis prediction. Next, we selected fraxetin as a control and extracted two sets of query results: ① a summary subset and ② an HA1E cell subset. Notably, we focused on the HA1E (kidney cell) subset because of its highly negative correlation with fraxetin, indicating a higher likelihood of discovering NPs similar to fraxetin. Based on this, multiple compounds were identified that negatively regulated gene expression patterns in response to ZEA toxicity, including fraxetin. Because we focused only on NPs, we performed an additional alignment analysis of the compounds obtained using the CMap databases and the existing NP library (MCE, HY-L021). Through a name duplicate search, we identified 120 shared NPs from the two databases, with 73 NPs originating from the summary subset and 47 from the HA1E subset. Notably, these NPs partially overlapped. The complete workflow is shown in
Fig. 7(b) for query 1.
To further elucidate the results of the CMap analysis, zebrafish embryos were treated with top-ranking NPs (query 1: summary and HA1E subsets). Subsequently, we assessed changes in toxic phenotype using the screening test. The substance names tested, along with their corresponding CAS numbers and the outcomes observed at concentrations of 10, 20, 40, and 80 μmol∙L
−1, are summarized in Appendix A Table S9. From these top-ranking NPs, we identified 11 candidate compounds that improved the embryonic phenotype (
Figs. 7(c)-(e)). Hit rates for the summary and HA1E subsets were 27.27% (3/11) and 47.37% (9/19, with one overlapping compound), respectively. These findings strongly suggest that CMap enhances the accuracy and success rate of screening.
Next, we attempted to determine whether the hub gene set could identify these candidate compounds by repeating this procedure (
Fig. 7(b)-query 2). To test this hypothesis, we queried CMap using 68 valid signatures induced or repressed by ZEA (Appendix A Fig. S8(b) and Table S10). The second result and our first search highly overlapped; both search strategies identified prednisolone, pepstatin, and paclitaxel as predicted detoxification compounds, along with similar top-ranking compounds (Appendix A Fig. S8(c),
Figs. 7(c) and
(d)). Screening tests for other top-ranking compounds identified 11 additional candidates, corresponding to a hit rate of 40% (18/45, with seven overlapping compounds) when screening the second library (Appendix A Table S11, Fig. S8(d), and Fig. S7(e)). Together, the hub gene set and all the differential gene sets can be used as query signatures for CMap.
3.9. Riboflavin was used to verify the CMap results in zebrafish
The effects of the 22 candidate compounds are shown in
Fig. 7(e) and Appendix A Fig. S8(e), including several potential detoxifying or antagonistic compounds from the published reports [
38], [
39], [
40]. However, despite obtaining many candidate NPs, most exhibited relatively low potency. Thus, we selected the most potent compounds for further study because active NPs with low potency for ZEA may not be able to entirely reverse toxicity and would be less helpful in providing insights for subsequent studies.
Based on the NPs screen effect, we roughly categorized these candidates into two groups. The first group comprised only three NPs: hispidin, daphnetin, and riboflavin. We hypothesized that the effects of these three NPs would likely be closer to that of hydroxytyrosol (Appendix A Fig. S9). Notably, riboflavin, a vitamin B, captured our interest and was chosen for further verification of CMap results. We evaluated the potency and efficacy of riboflavin, and our results suggested that riboflavin could not fully restore ZEA-induced deficits in zebrafish across the range of concentrations tested (
Figs. 8(a)-(d)). Its potency (87.83 μmol∙L
−1) was indeed similar to that of hydroxytyrosol and was about 19-fold lower than that of fraxetin (
Fig. 8(e)). Additionally, based on the coiling movement assay, the frequency and intensity barcode matrix of 160 μmol∙L
−1 riboflavin was similar to, albeit weaker than, that of the controls (
Fig. 8(f)). Interestingly, the effect of 240 μmol∙L
−1 riboflavin produced a dramatic fall in coiling frequency and intensity, which indirectly suggested a decrease in detoxification efficacy (
Fig. 8(g)). Hence, riboflavin exhibits some benefits in reducing ZEA toxicity, although it did not surpass that of fraxetin in this regard.
3.10. Fraxetin alone promotes the growth of zebrafish larvae
In our experiments, fraxetin was the most effective NP at attenuating ZEA toxicity.
Figs. 9(a) and
(b) shows that 72 or 120 h of fraxetin treatment alone could also increase BL and DMA dose-dependent, with a peak effect at 40 or 80 μmol∙L
−1, respectively. We also examined the distance between the myotomes and the muscle birefringence strength (
Figs. 9(c)-(h)). The relative distance between the myotomes gradually increased with a progressive increase in fraxetin concentration. Similarly, there was an upward trend in the strength of muscle birefringence. These observations suggest that fraxetin induces a mild growth promotion.
To test whether fraxetin-treated zebrafish morphology was normal, the head, yolk sac, and swimming bladder parameters were quantified. The results indicated that fraxetin, at the concentrations used in the above experiment, had no adverse effects on 72 or 120 hpf zebrafish morphology (
Figs. 9(i)-(k)). In addition, no adverse effects of fraxetin on hatching rates were observed. Conversely, fraxetin treatment promoted the hatching of 60 hpf zebrafish embryos (
Fig. 9(l)). Based on these findings, no prominent toxicities were observed for fraxetin in zebrafish.
4. Discussion
The toxic effects of ZEA have been extensively studied in zebrafish [
41], [
42]. The results of these studies and our results revealed that ZEA can induce toxicity-related phenotypes during the embryonic and larval stages in zebrafish. Although the specific toxic phenotypes studied in our experiments have not been previously characterized, we utilized phenotype-based screening to identify NPs that counteract the toxic effects of ZEA in zebrafish. A similar approach may be used to efficiently detect the toxic effects mediated by other mycotoxins and identify compounds that antagonize these mycotoxins. Importantly, we developed an efficient evaluation process by combining a toxicity scoring system with automated measurement scripts (TCMacro method) that track and quantify spontaneous movement parameters. This methodology enables the testing of candidate compounds at specific concentrations and provides a convenient means to visualize the alleviating effects on the platform, facilitating comparative analysis of multiple compounds and the selection of those offering superior benefits. Consequently, the development and utilization of these substances will be significantly expedited. Similarly, this process was verified and utilized to compare the potency and efficacy of three representative compound candidates: fraxetin, hydroxytyrosol, and riboflavin. Furthermore, the discovery that new candidate compounds reverse the toxic effect in distinct ways suggests that a large-scale phenotype-based screening strategy provides a new perspective on the relationship between small-molecule NPs and ZEA toxicity.
ZEA induces widespread changes in the transcriptional profile of zebrafish cells, affecting the expression of over 500 mRNAs, which aligns with previously published findings [
12], [
34]. However, only a few of these mRNAs have been studied extensively in zebrafish and other animal models. Although we are still unsure of the exact mechanisms regulating these mRNA changes in response to ZEA-induced toxicity, we utilized these data to establish a CMap query signature (which formed the basis for our first CMap query). In addition, using the PPI networks constructed by hub genes, we remapped the interaction layout and narrowed our search until we identified a defined subnetwork consisting of 79 DEGs (which formed the basis for our second CMap query). We observed a significant overlap between these two sets of query signatures when predicting high-scoring compounds in two independent CMap queries, suggesting that the subnetwork represents global changes. In addition, CMap analysis identified dozens of compounds predicted to be antidotes against ZEA-induced toxic stress based on our model, indicating that our screening strategy could be combined with the CMap database to expand the screening scale and hit rate. Overall, further analysis of differentially expressed mRNAs and their associated biological functions may help unravel the complex cellular responses to ZEA and shed light on the specific pathways through which this toxin exerts its effects in zebrafish. This knowledge can aid the development of strategies to mitigate or counteract the adverse effects of ZEA exposure.
To date, research on the detoxification of fungal toxins has been limited, typically focusing on only a few small molecules. We conducted tests on hundreds of NPs, all of which were rigorously evaluated under the same stringent control conditions. This extensive testing allowed a comprehensive analysis of the effects of these substances and ensured reliable and consistent results. Throughout our screening process, we successfully identified and validated several NPs that exhibit alleviating effects. Fraxetin, hydroxytyrosol, and baicalin were among the first antidotes discovered during initial compound screening. Among these, hydroxytyrosol, a polyphenolic compound, and baicalin, a flavone compound, have gained considerable attention owing to their antioxidant and anti-inflammatory properties. Interestingly, hydroxytyrosol can alleviate the toxicity caused by various mycotoxins (such as deoxynivalenol and ochratoxin-A) in the kidney and HepG2 cells [
29], [
30]. However, the specific mitigating effect on ZEA toxicity remains unclear [
38]. Similarly, baicalin can protect against ZEA-induced organ injury in chicks [
28], further confirming its role as a ZEA antidote. Furthermore, our study identified fraxetin as an effective NP capable of reversing the toxic effects induced by ZEA in zebrafish, adding to the breadth of potential antidotal substances identified using our screening strategy.
One important finding of our study is the remarkable effectiveness of fraxetin, as it demonstrated the ability to completely reverse the toxic effects triggered by ZEA on the indicators we tested. ZEA induces cell cycle arrest in the G2 phase and apoptosis in zebrafish. These responses have been consistently observed in various cell models [
13], [
35], [
43]. Fraxetin is a coumarin found in the traditional medicinal plant
Fraxinus rhynchophylla. It is known for its various biological activities, including antioxidant, anti-inflammatory, and anticancer properties. Fraxetin has long been studied for its potential use in cancer treatment, specifically for its ability to induce apoptosis and cell cycle arrest [
44], [
45]. Interestingly, our results suggest that fraxetin can reverse apoptosis and cell cycle arrest caused by ZEA. In addition, a senescent cellular state is an indefinite cell cycle arrest due to irreparable sublethal damage, such as DNA damage or oxidative stress, and the inability to enter apoptosis [
46]. We also observed many senescent cells in the zebrafish after prolonged exposure to ZEA. We believe that this is one of the primary cellular damage responses in ZEA toxicity. Recent studies suggested that ZEA induces senescence in cardiovascular cells
in vitro and
in vivo [
47]. Unexpectedly, fraxetin eliminated this effect. These novel findings highlight the potential of fraxetin to reverse ZEA-induced toxicity.
Mechanistically, fraxetin participates in multiple signaling pathways activated by ZEA, including protein processing in the endoplasmic reticulum, p53 signaling, and FoxO signaling. We demonstrated alterations in some critical genes of these signaling pathways, including
hsp70.1,
p53,
cdkn1a,
rbl2,
fosl1a,
gadd45ab,
baxa,
fos,
mdm2,
ccng1, and
parp3. Indeed, some of these gene are also the key downstream effectors of apoptosis, cell cycle, and senescence (e.g.,
p53,
cdkn1a,
rbl2,
gadd45ab,
baxa, and
parp3) [
48], [
49], [
50], [
51]. Notably, the effect of fraxetin on these downstream effectors coincided with the mitigation of the toxic effects of ZEA. KEGG pathways also highlighted that altered p53 and FoxO signaling pathways are upstream events leading to the observed effects. In other words, the mitigation of fraxetin could intrinsically encompass the inhibition of two signaling pathways, p53 and FoxO, that could merge downstream to effectively reverse the ultimate fate of cells [
52], [
53], [
54]. If so, an issue arises as to which potential mediators and upstream events lead to p53 or FoxO pathway inhibition. Similarly, the effects of fraxetin may involve additional pathways. It is likely that these signaling pathways, including protein processing in the endoplasmic reticulum or the MAPK pathway, interact with the p53 and FoxO pathways during ZEA-induced signaling [
55], [
56]. In other words, multiple pathways likely interact (crosstalk) to produce downstream effects. These questions are complex and require further investigation. Nevertheless, the discovery that fraxetin is an effective antidote against ZEA-induced changes in zebrafish gene expression suggests a promising avenue for future studies. It should be noted that although phenotypic approaches to mycotoxin antidote discovery using zebrafish address some of the limitations of traditional approaches several challenges remain, including insufficient pixelated resolution and experimental complexity. In addition, although we found that many NPs with detoxification effects in other animals also alleviate toxic phenotypes in zebrafish, some NPs failed to produce any detectable change in the phenotypic screen. Future research should investigate these issues.
5. Conclusions
In summary, we sought to establish an efficient and scalable phenotypic screening process for the identification of mycotoxin antidotes. By employing toxic phenotyping in living vertebrates, specifically zebrafish, our study offered a new perspective for the screening of valuable NPs. Harnessing the screening potential of zebrafish will enable us to expedite the discovery of valuable NPs while gaining a deeper understanding of how NPs can effectively mitigate the toxic effects induced by mycotoxins. This innovative strategy holds great promise for advancing compound discovery and enhancing our knowledge of NP-based interventions for mycotoxin-induced toxicity.
Acknowledgments
This research was financially supported by the National Natural Science Foundation of China for Outstanding Youth Science Foundation (31922086), the National Key Research and Development (R&D) Program of China (2018YFD0900400), and the Young Top-Notch Talent Support Program and Government Guidance for Local Scientific and Technological Development Projects (23ZYZYTS0513).
Compliance with ethics guidelines
Hong-Yun Zhang, Wei-Dan Jiang, Pei Wu, Yang Liu, Hong-Mei Ren, Xiao-Wan Jin, Xiao-Qiu Zhou, and Lin Feng declare that they have no conflict of interest or financial conflicts to disclose.
Appendix A. Supplementary material
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2024.03.016.
PDF
(7458KB)
