Inhibition of Neural Crest Cell Migration by Strobilurin Fungicides and Other Mitochondrial Toxicants

Abstract

Cell-based test methods with a phenotypic readout are frequently used for toxicity screening. However, guidance on how to validate the hits and how to integrate this information with other data for purposes of risk assessment is missing. We present here such a procedure and exemplify it with a case study on neural crest cell (NCC)-based developmental toxicity of picoxystrobin. A library of potential environmental toxicants was screened in the UKN2 assay, which simultaneously measures migration and cytotoxicity in NCC. Several strobilurin fungicides, known as inhibitors of the mitochondrial respiratory chain complex III, emerged as specific hits. From these, picoxystrobin was chosen to exemplify a roadmap leading from cell-based testing towards toxicological predictions. Following a stringent confirmatory testing, an adverse outcome pathway was developed to provide a testable toxicity hypothesis. Mechanistic studies showed that the oxygen consumption rate was inhibited at sub-µM picoxystrobin concentrations after a 24 h pre-exposure. Migration was inhibited in the 100 nM range, under assay conditions forcing cells to rely on mitochondria. Biokinetic modeling was used to predict intracellular concentrations. Assuming an oral intake of picoxystrobin, consistent with the acceptable daily intake level, physiologically based kinetic modeling suggested that brain concentrations of 0.1-1 µM may be reached. Using this broad array of hazard and toxicokinetics data, we calculated a margin of exposure ≥ 80 between the lowest in vitro point of departure and the highest predicted tissue concentration. Thus, our study exemplifies a hit follow-up strategy and contributes to paving the way to next-generation risk assessment.

Keywords: adverse outcome pathway; data integration; developmental toxicity; hit confirmation; mitochondria; neural crest cells; next-generation risk assessment; strobilurin fungicides; toxicokinetics.

Figure 1

Outline of the screen process and follow-up studies. (A) A tiered testing strategy was applied to identify compounds that inhibit neural crest migration in the cMINC assay. The decision boxes indicate subfigures with exemplary details. (B) Exemplification of data resulting from cMINC pre-screen 1 on “blinded compounds” BC1, BC2 and BC3 (blinded at this stage, only compound IDs given). All shown compounds advanced to the next tier. Data for pre-screen 1 of selected compounds are given in Figures S2 and S3 (1N, 4n). (C) Exemplification of data resulting from cMINC pre-screen 2 for compounds shown in B. At this stage, 3 concentrations were tested, and compounds were classified based on the rules shown in A. Data for pre-screen 2 of selected compounds are given in Figure S4 (≥2N, 3n). (D) Full concentration–response curve for compound BC2 obtained in the primary screen. A ratio of BMC₂₅ (M)/BMC₁₀ (V) was calculated, and resulted in a hit call (≥3N, 3n). (E) After testing completion of all tiers, data were deposited at the NIEHS database. Subsequently, compounds were unblinded (e.g., BC2 was picoxystrobin). The hits were followed up in an orthogonal assay. * an offset of BMC₂₀ (M) vs. BMC₂₀ (V) of 2 was considered as an alert; ** 21 compounds were DNT hit calls. Four additional compounds were categorized as “borderline compounds”. BC: blinded compound.

Figure 2

Synopsis of screen data on mitochondria-related hits. In total, 115 compounds were screened in the cMINC assay. After completion of the primary screen, i.e., the last tier of testing, 21 compounds were classified as hits. According to the published literature, 12 out of 21 specific hits from the cMINC screen targeted mitochondrial respiration. (A) Complexes (roman numbers) of the electron transfer chain are shown. The green ellipse symbolizes the effect of uncouplers. The assumed targets of 12 screen hits are indicated. (B) Concentration–response curve of fenpyroximate, an example of a complex I (cI) inhibitor. (C) Concentration–response curve of fluazinam, an example of an uncoupler. (D) Concentration–response curves of azoxystrobin and picoxystrobin, two examples of complex III (cIII) inhibitors. Data of other mitochondrial inhibitors are given in Figure S5. All data are from ≥3 biological replicates. The data in the insert boxes are derived from curve fitting of the data. (E) Tabular overview of the 12 specific mitochondrial hit compounds and their respective BMC₁₀ (V) and BMC₂₅ (M). BMC₂₅ (M) was considered as the relevant threshold concentration for migration impairment. BMC₁₀ (V) was assumed to be the highest non-cytotoxic concentration. It was used as a reference point for follow-up testing in an orthogonal assay. *: no effect could be observed even at the highest tested concentration (HTC). To calculate the ratio, the HTC is used.

Figure 3

Effect of mitochondrial toxicants on neural crest cell ATP levels and production. (A) Effect of four mitochondrial toxicants on NCC ATP levels. ATP levels were measured at 1 h, 6 h and 24 h after addition to NCC cultures. A complete data set on other compounds is displayed in Figure S6. Data are expressed as means ± SEM from three independent biological replicates and are shown relative to the solvent control. (B,C) The effects of toxicants on ATP production rates are shown. Cells were treated with single concentrations corresponding to the BMC₁₀ (V) of the cMINC screening (see Figure 2). Data on oxygen consumption rates under different metabolic conditions were used to calculate “glycoATP” as measure of the glycolytic ATP production rate and “mitoATP” as measure of mitochondrial ATP production rate. Dotted lines in (B) indicate the ATP production rate of cells exposed to solvent (0.1% DMSO). Data are expressed as means ± SD from two independent biological experiments.

Figure 4

Effect of mitochondrial toxicants on neural crest cell oxygen consumption. The oxygen consumption rate (OCR) of NCCs was recorded. After baseline measurements for 20 min, cells were exposed to mitotoxicants at a concentration corresponding to the BMC₁₀ (V) of the cMINC Screen (see Figure 2). Then, oligomycin, FCCP and rotenone/antimycin A were added sequentially, as indicated by dotted vertical lines. OCR data are normalized to the cell count and expressed as means ± SD from two independent biological experiments. (A) strobilurins/complex III inhibitors, (B) complex I inhibitors, (C) uncouplers.

Figure 5

Hypothetical AOP linking mitochondrial inhibition of neural crest cells to developmental toxicity. A putative AOP was constructed. Below the AOP, we indicated potential assays to test KEs and their linkage. We picked the complex III inhibitor picoxystrobin as an exemplifying compound. Thus, the respective picoxystrobin assay exposure times used in this study are shown. MIE: molecular initiating event; KE: key event; KER: key event relationship; darker blue boxes indicate assays used to establish the AOP; lighter blue boxes indicate assays that can confirm the AOP; AO: adverse outcome; TEP: toxicity endophenotype; cIII: mitochondrial complex III; OCR: oxygen consumption rate; Glu vs. Gal: glucose vs. galactose medium conditions; biomarker: could also be a modifying factor of KER2, but needs more research.

Figure 6

Setup and performance of the neural crest transwell migration assay. (A) Schematic illustration of the transwell migration assay. In the beginning, the NCCs are plated into the transwell inserts. The difference in FBS concentration between the upper and lower compartment stimulates NCCs to migrate through the membrane pores. Toxicants were applied in both compartments. After 6 h, the number of cells that reached the downward surface of the membrane was quantified. (B) Results of compound testing in the transwell assay: For calibration of the assay, cytochalasin D (CytoD) was used as positive control. Omission of FBS (no FBS) was used as second control for “inhibited” migration (shown in purple); pink: hit compounds of cMINC screen known to affect mitochondrial respiration; blue: negative controls of cMINC screen. All compounds were tested at a single concentration corresponding to the BMC₁₀ (V) from the cMINC screen (see Figure 2E). Transwell migration is measured as the ratio of “migrated cells in the presence of toxicants to the number of migrated cells in the absence of toxicant”. The dotted line at 75% indicates the threshold for classification of compounds as specific migration inhibitors in the transwell assay. The black line in the violin plots represents the median. The black dots represent data from individual experiments. Data are from ≥2 independent biological experiments. FBS: fetal bovine serum.

Figure 7

Comparison of internal exposure estimates and primary effect potency. (A) Schematic illustration of approaches to arrive at an estimate of a maximal (tolerable) exposure level of picoxystrobin. For picoxystrobin, no current data on consumption and food residues are available from EFSA. In an alternative approach, the lowest observed effect level (LOEL) of animal studies was used (9 mg/kg/day). By assuming a standard safety factor of 100, we estimated a human daily threshold dose of 0.09 mg/kg. In a second approach, we used the acceptable daily intakes (ADIs) suggested in a 2012 report of a joint meeting of FAO/WHO (REF: https://www.fao.org/3/i3111e/i3111e.pdf (accessed on 15 June 2024)). Both scenarios lead to the same upper exposure limit for picoxystrobin of 0.09 mg/kg (per day). (B) A physiologically based kinetic (PBK) model was established for picoxystrobin. The model was parametrized to reflect a population of pregnant subjects in gestational week 20, and their foetus, with a daily intake of 0.09 mg/kg (see (A)), was modelled. The predicted concentrations of picoxystrobin are shown. Data (green lines) are population averages of pregnant subjects (n = 100), aged between 18 and 45. The dashed lines indicate the 5th and 95th percentiles of the population. (C) The left graph shows the concentration–response curve for the oxygen consumption rate (OCR) of NCCs directly (20 min offset) after the picoxystrobin injection. Measurements were performed in glucose or galactose medium. Data are shown for two independent experiments. Each data point shown is the average of three technical replicates. The right graph shows the concentration–response curve of picoxystrobin in the transwell assay. The assay was performed either in glucose or galactose medium. Data are expressed as means ± SEM from three independent biological experiments. ^# LOEL of animal study is based on a 90-day dog study (REF: https://www.fao.org/3/i3111e/i3111e.pdf (accessed on 15 June 2024)). *^,§ There is also an old (no longer valid) value by EFSA of 0.043 mg/kg/day from the year 2004 [71].

Figure 8

Consideration of biokinetics for refined hazard (potency) estimates. (A) Concentration–response curve of the oxygen consumption rate (OCR) in NCCs cultured in galactose medium after a 24 h treatment with picoxystrobin. The highest tested concentration was 2.8 µM (highest non-cytotoxic exposure). Data are shown for two independent experiments (as in Figure 7B). (B) The cMINC assay was performed as in Figure 1, but the NCCs were cultured in galactose medium. The insert box gives picoxystrobin potency data for migration (M) and cytotoxicity (V), and their ratio. Data are expressed as means ± SEM from seven independent experiments. (C) Schematic illustration of the distribution of picoxystrobin in a cell culture well according to the in silico biokinetics prediction model. Data for each compartment are given either as percentage (left) or as concentrations (right) for a nominal concentration of 1 µM. (D) Tabular overview of the distribution of picoxystrobin in the different compartments at a nominal concentration of 1 µM. Medium_t: total medium; Medium_b: bound in medium; Medium_u: unbound in medium; Cells_t: total amount in cells; Cells_M: mitochondrial compartment; Cells_L: lysosomal compartment; Cells_R: “rest” of the cells. The correction factor indicates the change vs. the nominal concentration. * The enrichment factor is defined as the distribution ratio of the compound in the compartments vs. the medium. (E) Synoptic overview of predicted and measured concentrations of picoxystrobin. Data on internal exposure in humans (left) are from the PBK model (Figure 7). Right: the concentration ranges at which picoxystrobin showed adverse effects in the experiments (e.g., migration inhibition in NCCs). The margins of exposure (MoE) for the mother and the fetus were estimated from these data by forming the ratios of hazard concentrations and exposure concentrations. For the fetal hazard concentrations, we considered (i) an upper limit, defined by the results of (acutely) inhibited respiration (see Figure 7C) and (ii) a lower limit defined by the results of inhibited migration in Gal medium (see (B)). For the hazard concentration in an adult, the inhibited respiration after 24 h exposure was used (see (A)). Exposure data used here were the modelled fetal brain concentration (100 nM range) and the maternal plasma concentration (200–300 nM range) (see Figure 7B). The fetal brain concentration was also used for the biokinetics-corrected MoE; here, the modelled concentration in the cells was used instead of the nominal concentration (see (C)). Exposure[i]: internal exposure measure in concentration (molarity) units. MoE: ratio of “minimally toxic concentration” and exposure[i].