Abstract
Detecting the toxic effects of chemicals on reproduction and development without using mammalian animal models is crucial in the exploitation of pharmaceuticals for human use. Zebrafish are a promising animal model for investigating pharmacological effects and toxicity during vertebrate development. Several studies have suggested the use of zebrafish embryos for the assessment of malformations or embryo-fetal lethality (MEFL). However, a reproducible protocol as a standard for the zebrafish MEFL test method that fulfills global requests has not been established based on the International Council of Harmonisation (ICH) S5 (R3) guidelines. To establish such a toxicity test method, we developed a new and easy protocol to detect MEFL caused by chemicals, especially those with teratogenic potential, using fertilized zebrafish eggs (embryos) within 5 days of development. Our toxicity test trials using the same protocol in two to four different laboratories corroborated the high inter-laboratory reproducibility. Our test method enabled the detection of 18 out of 22 test compounds that induced rat MEFL. Thus, the prediction rate of our zebrafish test method for MEFL was almost 82% compared with that of rat MEFL. Collectively, our study proposes the establishment of an easy and reproducible protocol for the zebrafish MEFL test method for reproductive and developmental toxicity that meets ICH guideline S5 (R3), which can be further considered in combination with information from other sources for regulatory use.
INTRODUCTION
Since the teratogenicity of thalidomide was recognized in the 1960s, it has been acknowledged as one of the worst man-made medical disasters in history, and has led to the development of better drug regulation. Therefore, teratogenicity is one of the main concerns of pharmaceutical companies and may influence drug development decisions. The safety of pharmaceuticals, in terms of reproductive and developmental toxicity, should be carefully tested using multiple mammalian animal species prior to administrative approval for market circulation (ICH_S5(R3), 2020). In addition, pharmaceutical companies should inform patients fully of their risks. In accordance with recent developments regarding animal welfare, prospective toxicity tests are supposed to adopt the 3R principles (Replacement, Reduction, and Refinement) of animal use. Although embryo-fetal development studies using rats and rabbits under the 3R restriction are still necessary for the assessment of new therapeutic chemicals, alternative methods to detect chemical teratogenicity have been explored in the last two decades. The major alternative methods that do not use intact mammals are the in vitro assay using embryonic stem (ES) cells and whole embryo culture tests (Genschow et al., 2004). Despite technical advances in the directional differentiation of ES/induced pluripotent stem cells and the successful formation of organoids, macroscopic physiological effects across tissues are hardly detectable using such in vitro systems (Piersma et al., 2022). The use of invertebrate animals, which are evolutionarily distant and anatomically different from humans, is not logical for detecting toxicity in human reproduction and development. To establish a global standard for malformation or embryo-fetal lethality (MEFL) tests that accompany the 3R principles, the International Council for Harmonisation (ICH) suggested the development of alternative test methods to detect chemical-mediated biological hazards, such as lethality, malformation, and behavioral impairments. The ICH also published a guideline and its revisions on reproductive toxicity tests for human pharmaceuticals, with the latest version ICH S5 (R3) in 2020 (ICH_S5(R3), 2020).
Among non-mammalian animals, the zebrafish (Danio rerio) is the most promising animal species for reproductive and developmental toxicity tests for several reasons. First, the zebrafish is a vertebrate animal that has been extensively studied after mice and rats. Second, zebrafish husbandry can be achieved at a lower cost and in smaller spaces than those required for the maintenance of rodents. Third, a number of animals (~100 fertilized eggs) can be easily obtained from the mating of a single pair of fish. Fourth, embryogenesis in zebrafish occurs outside the mother’s body and is very rapid, with most organs forming within 5 days of development. Finally, zebrafish live in freshwater and are easily exposed to chemicals at any stage of their life. Taking these advantages of zebrafish into account, the Organization for Economic Co-operation and Development (OECD) generated Test Guideline 236 (TG236), which provides the requirements for acute toxicity tests using fish embryos (OECD, 2013). Although several studies have shown the usefulness of zebrafish embryos in reproductive and developmental toxicology (Ball et al., 2014; Brannen et al., 2010; Gustafson et al., 2012; Song et al., 2021), an easy and reproducible protocol for zebrafish testing has not yet been established. For instance, the selection of fertilized zebrafish eggs (embryos) and the conditions of chemical exposure that ensure reproducibility by different researchers in different laboratories have not been thoroughly assessed.
In this study, we introduce a new and easy protocol to detect MEFL caused by chemicals, especially those with teratogenic potentials, using zebrafish embryos, which was validated for its reproducibility and assessed for its correlation with rat or rabbit MEFL or human teratogenicity.
MATERIALS AND METHODS
Animals
Zebrafish (Danio rerio) were reared and maintained in 3 L tanks in a recirculating system at 27 ± 1°C under a 14 hr light and 10 hr dark or a 16 hr light and 8 hr dark photoperiod according to the standard protocol (Westerfield, 2007). They were fed dry flakes and live brine shrimp. The wild-type zebrafish AB strain was purchased from the Zebrafish International Resource Center (ZIRC).
Test compounds
Thirty-two test compounds were used in this study. Among them, 29 are known to induce MEFL in non-clinical studies and/or human teratogens and are listed in the ICH Reference Compound List (ICH_S5(R3), 2020). These compounds were used as positive controls. The remaining three chemicals (cetirizine, saxagliptin, and vildagliptin) showed no toxic effects in rats or rabbits after exposure to 25 times the maximum recommended human dose; thus, they are known as non-MEFL in mammals (ICH_S5(R3), 2020). These compounds were used as negative controls. The information on the CAS number, supplier, primary solvent, and concentration range tested in this study are presented in Table 1 and Table S1. The test compounds were dissolved in dimethyl sulfoxide (DMSO) or water to prepare the master solutions immediately before each assay. These master solutions were diluted using 0.3x Danieau’s solution (17.4 mM NaCl, 0.21 mM KCl, 0.18 mM Ca(NO3)2, 0.12 mM MgSO4, and 1.5 mM Hepes, pH 7.2–7.5), which is referred to as fish water in this study. In the case where we used DMSO as a solvent, the final DMSO concentration was set at 0.5% for zebrafish embryo exposure. For each test compound, we tested 7 to 8 concentrations with a constant spacing factor of 2 to 3 to determine the 100% lethal concentration. In this study, we set the 100% lethal concentration as the highest dose for exposure. We also explored the concentrations of non-malformations and malformations to determine the range of effective concentrations. Otherwise, the concentration range was determined based on the maximum solubility of the test compounds.
Table 1. Test compounds used in this study.
| Test compounds |
CAS-NR |
Supplier |
| Acitretin |
55079-83-9 |
TCI |
| Aspirin |
50-78-2 |
Fujifilm Wako |
| Bosentan |
147536-97-8 |
MedChemExpress |
| Busulfan |
55-98-1 |
TCI |
| Carbamazepine |
298-46-4 |
Sigma |
| Cisplatin |
15663-27-1 |
Fujifilm Wako |
| Cyclophosphamide |
50-18-0(6055-19-2) |
Sigma |
| Cytarabine |
147-94-4 |
TCI |
| Dabrafenib |
1195765-45-7 |
Chemscene |
| Dasatinib |
302962-49-8 |
MedChemExpress |
| Fluconazole |
86386-73-4 |
TCI |
| 5-Fluorouracil |
51-21-8 |
Fujifilm Wako |
| Hydroxyurea |
127-07-1 |
Fujifilm Wako |
| Ibrutinib |
936563-96-1 |
MedChemExpress |
| Ibuprofen |
15687-27-1 |
Fujifilm Wako |
| Imatinib |
152459-95-5 |
TCI |
| Isotretinoin (13-cis-retinoic acid) |
4759-48-2 |
Fujifilm Wako |
| Methotrexate |
59-05-2 |
Fujifilm Wako |
| Pazopanib |
444731-52-6 |
MedChemExpress |
| Phenytoin (Diphenylhydantoin) |
57-41-0 |
Sigma |
| Pomalidomide |
19171-19-8 |
Fujifilm Wako (Fluorochem) |
| Ribavirin |
36791-04-5 |
Fujifilm Wako |
| Tacrolimus |
104987-11-3 |
MedChemExpress |
| Thalidomide |
50-35-1 |
TCI |
| Topiramate |
97240-79-4 |
TCI |
| Tretinoin (all-trans-retinoic acid) |
302-79-4 |
Fujifilm Wako |
| Trimethadione |
127-48-0 |
Sigma |
| Valproic acid |
99-66-1 |
Fujifilm Wako |
| Vismodegib |
879085-55-9 |
Chemscene |
| Cetirizine dihydrochloride |
83881-52-1 |
Sigma |
| Saxagliptin |
361442-04-8 |
BioVision |
| Vildagliptin |
274901-16-5 |
Cayman Chemical |
Preparation of zebrafish embryos
Adult female and male zebrafish were housed overnight in a breeding tank with a divider prior to mating. The following morning, the divider was removed to initiate mating after the light of the fish room turned on. After 20 min, the fish were removed to stop mating, and this time was defined as 0 hr post-fertilization (hpf). Fertilized eggs (embryos) were collected at 1–3 hpf and incubated in fish water at 28°C in 90–100 mm Petri dishes. We discarded abnormal embryos and selected healthy embryos within the chorion during the 1000-cell to early epiboly stages at 3–5 hpf by optical observation under a stereomicroscope. Typical characteristics of the abnormal embryos are as follows:
(i) Uneven blastoderm size and density,
(ii) Irregular blastoderm shape,
(iii) Presence of bubble-like structures in the blastoderm or yolk,
(iv) Irregular boundary between the blastoderm and yolk, and
(v) A diffuse black zone extending from the border between the blastocytes and yolk.
Healthy embryos (3–5 hpf) were transferred to 24-well plates (IWAKI, 1820-024) at one embryo per well in fish water.
Exposure of zebrafish embryos to the test compounds
At 4–6 hpf, fish water was added to 1 mL of chemical-containing fish water with or without 0.5% DMSO. In one assay, 10 individual zebrafish embryos (10 wells, 4–6 hpf) were tested at each concentration. They were placed in a 28°C incubator with no medium change for up to 5 days post-fertilization (dpf) and subjected to the assessment of viability and morphological abnormality. The 0.5% DMSO alone caused neither lethal nor malformative effects in the zebrafish embryos (data not shown). The fish water, Petri dishes, and 24-well plates were prewarmed at 28°C with all procedures monitored to avoid temperature drops.
Assessment of viability and morphological abnormality
Zebrafish embryos show cardiac contractions at the onset of 22 hpf, hatch out of the chorion at 3 dpf, and are referred to as larvae (Gierten et al., 2020; Kimmel et al., 1995). At 5 dpf, the zebrafish larvae were assessed for viability and malformation. Larvae that fulfilled either of the following criteria were evaluated as dead:
Lethality 1: Coagulation of embryos.
Lethality 2: Lack of heartbeat within 10 sec.
Typically, zebrafish that were unhatched and died were not assigned for further morphological assessments. The live larvae were assessed for malformations. Malformations were evaluated based on morphological differences, that is, cardiac edema, small eyes, body curvature, and short or bent fins, in comparison with normal healthy larvae. Larvae showing at least one morphological abnormality were considered malformed. Larvae exposed to cisplatin remained alive in the chorion and exhibited cardiac edema. In such cases, we evaluated the unhatched samples for morphological assessment. The larval images were captured using a CMOS camera (ATZ, Kennis).
Classification of malformation
To assess teratogenicity, we classified the MEFL into four categories: “Malformation,” “Equivocal,” “Non-malformation,” and “Inconclusive.” The classification scheme is illustrated in Fig. 3, and the categories are described as follows:
Malformation
A chemical is strongly suspected to cause rat or rabbit MEFL and exhibit human teratogenic potentials.
Equivocal
The chemical is likely to cause rat or rabbit MEFL and exhibit human teratogenic potentials; however, at least, it probably has some embryo-fetal toxic potentials.
Non-malformation
There are negative results of rat or rabbit MEFL and human teratogenicity.
Inconclusive
There is no conclusive data regarding malformations or lethality.
For the classification process, quantitative or qualitative criteria were used based on malformation and lethality data.
Quantitative assessments were performed using a teratogenic index. Based on the 50% lethality rate (LC50) and 50% malformation effective rate (EC50), the teratogenic index was defined as the ratio of LC50 to EC50. If this index was 2.0 or more, the test compound was classified as “Malformation” (Fig. 3A). If the index was between 1.0 and 2.0, the test compound was classified as “Equivocal” (Fig. 3B).
Qualitative assessments are effective if neither the LC50 nor EC50 values are calculated. The lowest lethal concentration (LCmin) and lowest malformation effective concentration (ECmin) were determined, considering that the determination of the no observed adverse effect level (NOAEL) involved assessing the concentration-response relationship. The test compound was “Malformation” (Fig. 3A) if the ECmin was determined and the LCmin was equal to or greater than the highest concentration tested. If the LCmin was equal to the ECmin, the test compound was “Equivocal” (Fig. 3B). If the ECmin was larger than the LCmin, the test compound was “Non-malformation,” but only when the LCmin was determined (Fig. 3C). If neither lethality nor malformation was observed in the zebrafish larvae at 5 dpf, this was “Inconclusive” (Fig. 3D).
Statistics
Calculations of the LC50 and EC50 were performed using the graphics and statistics software GraphPad Prism, version 2.01 (GraphPad Software).
RESULTS
Inter-laboratory assessments of the zebrafish MEFL test method
To establish a reproducible protocol for reproductive and developmental toxicity that can be validated by the ICH S5 (R3) guidelines, we explored the application of zebrafish embryos in MEFL tests in four different laboratories. Prior to the actual exposure to chemicals, we checked whether we could routinely collect healthy zebrafish embryos that showed normal development by at least 5 dpf. This was done because some spawned zebrafish eggs usually show stochastic developmental failure or are unfertilized eggs that undergo impaired early cleavage. We observed the embryos at 3–5 hpf and collected healthy embryos by removing morphologically abnormal ones (see details in Materials and Methods). The percentages of embryos that exhibited normal development up to 5 dpf out of the collected embryos that exhibited normal early embryogenesis at 3–5 hpf were 92 ± 3%, 88 ± 8%, 92 ± 2%, and 85 ± 7% in four different laboratories (three trials each). According to OECD TG236, for the test validity criteria, the overall fertilization rate of all eggs collected should be ≥ 70% in the batch tested. Our selection method for healthy embryos was approximately 90% reliable and may be regarded as being of a higher quality. Therefore, we applied this method to investigate the dose-dependent MEFL effects of the four test chemicals in two to four different laboratories. Healthy zebrafish embryos were exposed to the chemicals at the onset of 4–6 hpf (Fig. 1). One embryo in 1 mL of the test compound-containing fish water was placed in each well of a 24-well plate. Ten embryos were assayed for each concentration of each test compound. They were incubated at 28°C without solution changes and subsequently assessed to determine whether they had died or exhibited malformations at 5 dpf (Table S2).
The results of the same test (four positive test compounds: valproic acid, methotrexate, topiramate, and tretinoin) conducted by different laboratories were evaluated (Fig. 2A-D; Table S3). The exposure of zebrafish embryos to valproic acid caused lethal effects at high concentrations, and malformations such as cardiac edema and short body length at low concentrations (Fig. 2A; Fig. S1). The median lethal concentration (LC50) and median effective concentration (EC50) of valproic acid were 100–200 µM and 30–60 µM, respectively, providing comparable results in four different laboratories. The NOAEL and lowest observed adverse effect level (LOAEL) of valproic acid was 25 µM and 50 µM, respectively, in all four laboratories. Similarly, the NOAEL, LOAEL, and LC50 of methotrexate were comparable between two laboratories (Fig. 2B). Regarding topiramate, the rise of the malformation curve in the plotted graph was similar in two laboratories; however, the rate of malformation did not reach 100% at the highest concentration (400 µM) in one laboratory (Fig. 2C). Topiramate’s lethality was hardly detected at the highest concentration in both laboratories. In such cases, we could not calculate the LC50 or EC50. The NOAELs of topiramate and tretinoin were different in two laboratories, with one laboratory showing an anomalous dose-dependent effect (Fig. 2D). Such errors may have been caused by stochastic embryonic death due to developmental failure, raising the importance of selecting healthy embryos (see the Discussion section). Overall, we demonstrated the reproducibility of the zebrafish MEFL test method using four inter-laboratory assessments.
To address the toxicity of the other test compounds, we employed an additional 25 MEFL and three non-MEFL test compounds, for which the in vivo study yielded positive and negative results, respectively. We collected data on lethality and malformation at different concentrations and assessed the LC50 and EC50. Among the 32 compounds tested, sixteen (15 positive controls and one negative control) were associated with more than 50% lethality and malformations at the highest concentration, enabling the calculation of the LC50 and EC50 (Fig. 2; Table S3; Table S4). We estimated the LCmin and ECmin of the remaining 16 test compounds. Nine test compounds exhibited lethal and/or malformative effects at the highest concentrations, which enabled the estimation of the LCmin and/or ECmin. In contrast, the remaining seven compounds exhibited neither lethal nor malformative effects on zebrafish development.
Evaluation of the zebrafish MEFL test method
Our zebrafish MEFL test method was validated by comparison with data generated from in vivo studies. We evaluated 29 test compounds that were suggested to be rat MEFL in mammals, indicating that there were 29 positive controls. Our classification results were 18 “Malformation,” 3 “Equivocal,” 1 “Non-malformation,” and 7 “Inconclusive” (Table S3; Table S4). Exposure of zebrafish embryos to acitretin, bosentan, carbamazepine, cisplatin, cyclophosphamide, dabrafenib, dasatinib, 5-fluorouracil, hydroxyurea, ibuprofen, isotretinoin, pazopanib, ribavirin, tacrolimus, topiramate, tretinoin, trimethadione, and valproic acid caused malformations and lethality in a dose-dependent manner. Thus, we identified 18 test compounds as causing malformations in zebrafish (Table 2). The ratio of LC50 to EC50 is referred to as the teratogenic index and is used to determine the potency. Among the 18 malformation compounds, 13 showed dose-dependent teratogenicity and lethality with the teratogenic index 2 or more. Although we failed to determine the teratogenic index of bosentan, dasatinib, 5-fluorouracil, pazopanib, and topiramate because of low lethality at the highest concentration and the subsequent impossibility of LC50 calculation, exposure to these five test compounds clearly increased malformations in a dose-dependent manner, leading to the conclusion that they caused malformations. Exposure to aspirin, ibrutinib, or methotrexate also caused both malformations and death in a dose-dependent manner; however, the teratogenic index ranged from 1.1 to 1.7, indicating that the concentrations for the onset of lethality and malformations were comparable. Thus, we defined these three test compounds as equivocal cases that were suspected to have teratogenic potentials (Table 2). The application of busulfan at high concentrations induced death during zebrafish embryogenesis, whereas busulfan at lower concentrations did not cause morphological abnormalities. We then determined busulfan to be associated with non-malformation (Table 2). Exposure to cytarabine, fluconazole, imatinib, phenytoin, pomalidomide, thalidomide, or vismodegib did not cause either malformations or lethal effects in developing zebrafish embryos. Therefore, these tests were not conclusive, and seven tests were excluded from assay qualification.
Table 2. Comparison of MEFL of 22 test compounds in rat, rabbit, human and zebrafish.
| Test compounds |
Rat
MEFL |
Rabbit
MEFL |
Human
Teratogen |
|
Outcomes
in Zebrafish |
| Acitretin |
X |
X |
X |
|
Yes (Malformation) |
| Aspirin |
X |
- |
X |
|
Equivocal |
| Bosentan |
X |
- |
- |
|
Yes (Malformation) |
| Busulfan |
X |
X |
X |
|
No (Non-malformation) |
| Carbamazepine |
X |
X |
X |
|
Yes (Malformation) |
| Cisplatin |
X |
- |
- |
|
Yes (Malformation) |
| Cyclophosphamide |
X |
X |
X |
|
Yes (Malformation) |
| Dabrafenib |
X |
- |
- |
|
Yes (Malformation) |
| Dasatinib |
X |
- |
- |
|
Yes (Malformation) |
| 5-Fluorouracil |
X |
X |
X |
|
Yes (Malformation) |
| Hydroxyurea |
X |
X |
X |
|
Yes (Malformation) |
| Ibrutinib |
X |
X |
- |
|
Equivocal |
| Ibuprofen |
X |
- |
X |
|
Yes (Malformation) |
| Isotretinoin |
X |
X |
X |
|
Yes (Malformation) |
| Methotrexate |
X |
X |
X |
|
Equivocal |
| Pazopanib |
X |
X |
- |
|
Yes (Malformation) |
| Ribavirin |
X |
X |
- |
|
Yes (Malformation) |
| Tacrolimus |
X |
X |
- |
|
Yes (Malformation) |
| Topiramate |
X |
X |
X |
|
Yes (Malformation) |
| Tretinoin |
X |
X |
X |
|
Yes (Malformation) |
| Trimethadione |
X |
- |
X |
|
Yes (Malformation) |
| Valproic acid |
X |
X |
X |
|
Yes (Malformation) |
| True positive |
18 / 22 (82%) |
12 /15 (80%) |
11 /14 (78.6%) |
|
|
| Equivocal |
3 / 22 (14%) |
2 / 15 (13.3%) |
2 / 14 (14.3%) |
|
|
| False negative |
1 / 22 (4.5%) |
1 / 15 (6.7%) |
1 / 14 (7.1%) |
|
|
Cetirizine dihydrochloride, saxagliptin, and vildagliptin were used as negative controls in the zebrafish MEFL test. Our classification results were 1 “Malformation” and 2 “Non-malformation” (Table S4). Although these test compounds were unrelated to MEFL in mammals, they were supplied as negative controls, and all three test compounds caused embryonic death in zebrafish at the highest concentration tested. An effect that was negative in this assay was assigned the negative result for rat MEFL. Exposure to saxagliptin at a semi-lethal concentration induced slight cardiac edema. These three test compounds were highly soluble in water and could be applied to zebrafish embryos at much higher concentrations than the NOAEL in rats and rabbits. We did not observe malformations in zebrafish embryos when applied at concentrations equivalent to the pharmacologically active doses in humans.
Taken together, our zebrafish MEFL test method demonstrated 82% (18/22 [18 malformations + 3 equivocals + 1 non-malformation]) correspondence to the MEFL data in mammals (Table 2). The omission of malformations in zebrafish, which was the case with busulfan treatment, was only 4.5% (1/22). This indicates that our method is highly sensitive (few false negatives). Here, we declare the establishment of a zebrafish MEFL test method that meets the ICH S5 (R3) guidelines.
DISCUSSION
In this study, we developed a new test method using zebrafish embryos. Four laboratories belonging to different corporations tested the same protocol and independently obtained similar results for the MEFL. Our zebrafish MEFL test method could detect test compounds that cause malformations in mammals, including human teratogens, with a prediction rate of 79–82% (Table 2). Collectively, we established a zebrafish model that is adaptable to the ICH S5 (R3) guidelines.
Requirements
To develop an easy and reproducible protocol for the test method using zebrafish embryos, we followed two standard conditions for the fish embryo acute toxicity test, in accordance with standard conditions (Westerfield, 2007) or OECD TG236 (OECD, 2013). The temperature for chemical exposure was set at 28°C, which is the standard temperature for zebrafish development. The concentration range for chemical exposure was set from zero to the lethal concentration. While the onset and endpoint of chemical exposure in the OECD TG236 were set at 0 and 96 hpf, respectively, those in our test were set at 4–6 hpf and 5 dpf, which respectively corresponded to the beginning of gastrulation and the end of embryogenesis (Kimmel et al., 1995).
In addition to these basic conditions, we employed three zebrafish-specific tips to promote the ease and reliability of the zebrafish MEFL test method. We first attempted to select suitable embryos. Generally, some of the clutches of spawned zebrafish eggs die because of fertilization failure or stochastic developmental defects. According to OECD TG236, for the test validity criteria, the overall fertilization rate of all eggs collected should be ≥ 70% in the batch tested, in which a low fertilization rate was reported. We observed that the selection of morphologically normal embryos at 3–5 hpf was sufficient to obtain healthy embryos that displayed normal development up to at least 5 dpf. In this study, we performed embryo selection in four different laboratories that performed well in most cases. In one laboratory, zebrafish embryos exposed to topiramate showed dose-independent malformations and mortality. This unusual result can be attributed to the careless selection of zebrafish embryos. This means the selection of healthy embryos in that laboratory could have been improved. Second, spawned zebrafish eggs were collected at 1–3 hpf. We found that the chorion of zebrafish embryos is soft and sometimes fragile soon after spawning but becomes firm by 1 hpf. The collection of spawned eggs at 1–3 hpf, rather than at 0–1 hpf, appeared to be better for obtaining healthy embryos. Third, we created a list of malformation patterns that may fulfill almost all the morphological abnormalities. We tested 32 compounds and collected many different images that covered various developmental defects, such as cardiac edema, small head, small eye, bent pectoral fin, and body curvature (Fig. S1). By sharing these diagnostic criteria, we could improve the objective judgments that worked in four different laboratories. However, this judgment may require actual experience of observation rather than a glance at the images.
Validation of the zebrafish MEFL test method
Our zebrafish MEFL test method was validated by comparing it to tests performed by four different laboratories and data generated from in vivo studies using the ICH Reference Compound List. The observational parameters of this test method were mortality and morphological assessment. We evaluated the effects of the test compounds on malformations and lethality based on quantitative and qualitative assessments. Assessments were performed using LC50, EC50, LCmin, and ECmin.
Valproic acid-induced MEFL, which was independently tested in four different laboratories, was highly correlated with quantitative data, such as LC50, EC50, LOAEL, and NOAEL. Similarly, in most cases (three out of four test compounds), similar results were independently obtained in different laboratories. In contrast, the dose curves for the LOAEL and LC50 values of tretinoin appeared to differ slightly between the two laboratories. However, the teratogenic index (LC50/EC50) of tretinoin was comparable between the two laboratories. The morphological patterns of malformations, such as cardiac edema and lack of a lower jaw, were commonly observed in tretinoin-treated embryos. As stochastic malformations and death in the absence of tretinoin were observed in one laboratory, the discrepancy in this tretinoin case is attributable to a selection error of zebrafish embryos for chemical exposure. This problem can be eliminated by improving the selection of healthy embryos. Overall, our inter-laboratory assessments were reproducible and reasonable.
We established an MEFL classification intended to detect teratogenic potency, which was expressed as 4 categories: “Malformation,” “Equivocal,” “Non-malformation,” and “Inconclusive.” “Non-malformation” was regarded as a negative outcome, and “malformation” was a positive outcome in the assay. Among the 29 test compounds that induced rat MEFL, seven were “inconclusive,” whereas 22 were conclusively validated. Eighteen out of the 22 test compounds (sensitivity: 82%) were regarded positive in zebrafish, as in rat MEFL, three (14%: aspirin, ibrutinib, and methotrexate) were regarded equivocal, and the remaining one (4.5%: busulfan) was negative. Regarding rabbit MEFL and human teratogens, the sensitivity was almost 80% compared with zebrafish. As in rabbit MEFL, two test compounds (iburutinib and methotrexate) were equivocal, and the remaining one (busulfan) was false negative. As in human teratogens, two (aspirin and methotrexate) were equivocal, and the remaining one (busulfan) was false negative.
Cetirizine dihydrochloride, saxagliptin, and vildagliptin did not induce MEFL in rats or rabbits when applied at 25 times the maximum recommended human dose (ICH_S5(R3), 2020). Although we intended to use these three test compounds as negative controls in our zebrafish test method, exposure to these test compounds caused lethal effects at the highest concentration, which was 1,400 to 707,000 folds of the circulating level after human administration (Fig. 4). Furthermore, saxagliptin also induced slight cardiac defects at a semi-lethal concentration in zebrafish. However, the concentration was 141,000-fold higher than the circulating level of saxagliptin in human plasma. These unexpected results suggest the complementary usefulness of the zebrafish MEFL test method in which a wide range of chemical concentrations are applicable to zebrafish embryos.
Being compared to an effective concentration between lethality and malformation, we could not clearly separate the lethality and malformation for three “Equivocal” test compounds. Cardiac malformations in the presence of these test compounds were indistinguishable from heart defects in previous zebrafish embryo studies (Wang et al., 2021; Sun et al., 2009; Song et al., 2021). This correspondence suggests that our strict judgment “Equivocal” implies that the test compound is likely suspected of having teratogenic potential, and at least it probably has some embryo-fetal toxic potential. The administration of busulfan to pregnant animals causes skeletal malformations in fetal rats (ICH_S5(R3), 2020). Skeletal abnormalities were not assessed in this study. Moreover, alcian blue staining and labeling of the cartilages, which are formed within 3 days of zebrafish development, could have revealed busulfan-induced malformations in zebrafish. Collectively, our zebrafish test method could detect test compounds that cause rat MEFL with a prediction rate of 82%. In addition, 14% of the zebrafish equivocally exhibited MEFL. Overall, this indicates that our method is a highly sensitive (with few false negatives) and effective alternative to animal testing. It should be noted that this prediction rate based on our stand-alone test is merely suggestive, as the actual prediction rate should be further demonstrated in combination with the results obtained from other sources, along with the consideration of adverse outcome pathways, at least because unclear results of equivocal test compounds should be independently re-examined.
Seven “Inconclusive” test compounds (cytarabine, fluconazole, imatinib, phenytoin, pomalidomide, thalidomide, and vismodegib) appeared to neither cause malformations nor lethality in zebrafish while they caused MEFLs in mammals. Although we cannot exclude the possibility that they are harmless to zebrafish, this inconsistency can be attributed to the limitation of chemical solubility in water and chemical permeability into zebrafish embryos. The highest exposure concentrations tested for imatinib, phenytoin, and vismodegib in zebrafish embryos were low compared to the equivalent doses in toxicity tests using rats and rabbits (ICH_S5(R3), 2020) (Table 3). In cases where the test compounds are insoluble in water, there may be limitations to the use of the zebrafish test methods.
Table 3. Comparison of malformation and lethal dose in zebrafish, rat, rabbit and human.
| Test compounds |
(µM) |
| Zebrafish MFD |
Rat
Cmax
LOAEL/NOAEL |
Rabbit
Cmax
LOAEL/NOAEL |
Human
dose
Cmax |
| Imatinib |
125 |
25 / 7 |
Not identified / 107 |
ND |
| Phenytoin |
50 |
106 / 53 |
135 / 107 |
57 |
| Vismodegib |
50 |
17 / ND |
ND |
31 |
ND: Not determined
In cases where test compounds dissolve well in water, and the test results are not considered biologically meaningful. The interpretation of positive results using the zebrafish test method is more valuable than the classification process. Similarly, in cases where exposure concentrations are considered to be sufficient for the toxicological evaluation, expert judgment on the basis of the total weight of evidence may be useful to interpret the data as “Inconclusive.” Although the criteria are not provided, for example, by ensuring an adequate exposure margin, the “Inconclusive” classification may be determined to be a negative outcome supported by sufficient scientific evidence. It is important to consider the physical/chemical characteristics of substances and biologically meaningful interpretations of test data for toxicological evaluation.
In summary, we compared the LOAEL of 22 zebrafish positive controls and three negative controls in rabbit and rat MEFL, along with the equivalent doses of pharmaceuticals teratogenic to humans (Fig. 4). The comprehensive data showed that the LOAEL of MEFLs in zebrafish were closer to those in rabbits and rats in several cases (i.e., carbamazepine, cyclophosphamide, 5-fluorouracil, pazopanib, and topiramate). This supports our conclusion that toxicity tests using zebrafish embryos are an alternative and versatile animal test for screening malformations. Since the morphological patterns in mammals and zebrafish appear to be similar, the adverse outcome pathway of malformations is likely to be similar among vertebrates. Therefore, our zebrafish MEFL test method is not only useful for the prediction of the teratogenicity of test compounds but also for the identification of adverse outcome pathways of teratogenicity. We believe that our MEFL test method using zebrafish embryos that meet the ICH S5 (R3) guidelines will be verified by successive replications of the zebrafish test method. We also suggest that the test method should be considered in combination with information from other sources for regulatory use.
ACKNOWLEDGMENTS
We thank Dr. Keiko Motoyama and Dr. Mutsumi Suzuki (Japan Pharmaceutical Manufacturers Association) for fruitful discussions and animal care.
Funding statement
This study was supported by a grant-in-aid from the Japan Agency for Medical Research and Development (AMED) under Grant Number JP20mk0101131 and the Ministry of Health, Labor, and Welfare (MHLW), Japan. H.H. was also supported by the Long-Range Research Initiatives of the Japan Chemical Industry Association.
Conflict of interest
The authors declare that there is no conflict of interest.
REFERENCES
- Ball, J.S., Stedman, D.B., Hillegass, J.M., Zhang, C.X., Panzica-Kelly, J., Coburn, A., Enright, B.P., Tornesi, B., Amouzadeh, H.R., Hetheridge, M., Gustafson, A.L. and Augustine-Rauch, K.A. (2014): Fishing for teratogens: a consortium effort for a harmonized zebrafish developmental toxicology assay. Toxicol. Sci., 139, 210-219.
- Brannen, K.C., Panzica-Kelly, J.M., Danberry, T.L. and Augustine-Rauch, K.A. (2010): Development of a zebrafish embryo teratogenicity assay and quantitative prediction model. Birth Defects Res. B Dev. Reprod. Toxicol., 89, 66-77.
- Genschow, E., Spielmann, H., Scholz, G., Pohl, I., Seiler, A., Clemann, N., Bremer, S. and Becker, K. (2004): Validation of the embryonic stem cell test in the international ECVAM validation study on three in vitro embryotoxicity tests. Altern. Lab. Anim., 32, 209-244.
- Gierten, J., Pylatiuk, C., Hammouda, O.T., Schock, C., Stegmaier, J., Wittbrodt, J., Gehrig, J. and Loosli, F. (2020): Automated high-throughput heartbeat quantification in medaka and zebrafish embryos under physiological conditions. Sci. Rep., 10, 2046.
- Gustafson, A.L., Stedman, D.B., Ball, J., Hillegass, J.M., Flood, A., Zhang, C.X., Panzica-Kelly, J., Cao, J., Coburn, A., Enright, B.P., Tornesi, M.B., Hetheridge, M. and Augustine-Rauch, K.A. (2012): Inter-laboratory assessment of a harmonized zebrafish developmental toxicology assay - progress report on phase I. Reprod. Toxicol., 33, 155-164.
- ICH S5 (R3) (2020): ICH S5 (R3) Guideline on detection of reproductive and developmental toxicity for human pharmaceuticals - Scientific guideline.
- Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Ullmann, B. and Schilling, T.F. (1995): Stages of embryonic development of the zebrafish. Dev. Dyn., 203, 253-310.
- OECD. (2013): OECD Guidelines for the Testing of Chemicals Section 2. Effects on Biotic Systems. Test No. 236: Fish Embryo Acute Toxicity (FET). Test.
- Piersma, A.H., Baker, N.C., Daston, G.P., Flick, B., Fujiwara, M., Knudsen, T.B., Spielmann, H., Suzuki, N., Tsaioun, K. and Kojima, H. (2022): Pluripotent stem cell assays: modalities and applications for predictive developmental toxicity. Curr. Res. Toxicol., 3, 100074.
- Song, Y.S., Dai, M.Z., Zhu, C.X., Huang, Y.F., Liu, J., Zhang, C.D., Xie, F., Peng, Y., Zhang, Y., Li, C.Q. and Zhang, L.J. (2021): Validation, Optimization, and Application of the Zebrafish Developmental Toxicity Assay for Pharmaceuticals Under the ICH S5(R3) Guideline. Front. Cell Dev. Biol., 9, 721130.
- Sun, S., Gui, Y., Wang, Y., Qian, L., Liu, X., Jiang, Q. and Song, H. (2009): Effects of methotrexate on the developments of heart and vessel in zebrafish. Acta Biochim. Biophys. Sin. (Shanghai), 41, 86-96.
- Wang, X., Liao, Q., Chen, H., Gong, G., Siu, S.W., Chen, Q., Kam, H., Ung, C.O., Cheung, K.K., Rádis-Baptista, G., Wong, C.T. and Lee, S.M. (2021): Toxic Peptide From Palythoa caribaeorum Acting on the TRPV1 Channel Prevents Pentylenetetrazol-Induced Epilepsy in Zebrafish Larvae. Front. Pharmacol., 12, 763089.
- Westerfield, M. (2007): The zebrafish book. A guide for the laboratory use of zebrafish (Danio rerio). University of Oregon Press.
