The cMINC assay was challenged on the one hand with compounds of known mode of action and on the other hand with environmental compounds and other chemicals of interest.
Compounds with known mode of action
A broad set of compounds with known mode of action was tested in the cMINC assay to obtain information on which pathways are relevant for NCC migration and could be potential toxicological targets (Fig.
6.4). Most compounds interfering with the cytoskeleton disturbed NCC migration. This is not surprising, but only shows proof-of-concept. Proliferation-inhibition does not necessarily affect migration, nor do inflammatory processes play an important role. In contrast to this, protein synthesis and protein transport, but not protein degradation, seem to be important for NCC migration. Moreover, gap junctions could play a role, as both inhibitors disturbed NCC migration. Agonists and antagonists of several receptor
6.3 Compounds disturbing human NCC migration
systems were also tested. Mostly, interventions on the RA receptor systems disturbed migration, but other receptor systems do not seem to play an important role. In a last step, various cytokines were tested, many of them were shown to affect cell migration in other cellular systems. However, NCC did not respond to them, except for PDGF-AA that slightly increased migration (data not shown).
To summarize, these results show that (1) the system reacts to known migration-inhibitors and (2) is not prone to very unspecific influences. The fact that RA receptor agonists and antagonists both disturbed NCC migration suggests that compounds known to interfere with this system should be prioritized for cMINC testing. Future studies could also investigate whether synthesis of a particular protein is necessary for NCC migration or whether migration is disturbed by general cellular stress.
strong weak no alert
compounds with defined mode of action
Disruptor of cytoskeletonColchicin (microtubuli) cycloRGDfV 200 µM
CytoD (actin skeleton) EGCG (integrin binding) Taxol (microtubuli) other migration-related NSC23766
processes PP2
PTX SP600125
protein synthesis Brefeldin A (protein transport) Chloroquine (lysosome)
and homeostasis Cycloheximide (protein synthesis) MG-132 (proteasome)
Thapsigargin (ER stress)
Tunicamycin (glycoprotein synthesis) gap junctions 18a-Glycyrrhetinic acid
Carbenoxolone
cell proliferation AraC
Aphidicolin Etoposide
inflammation cytokine mix LPS 5 µg/ml
nuclear receptor 1-850 (THR) UVI 3003 (RXR) CH223191 (AhR) 100 µM
(agonist / antagonist) CD2665 (RAR) Am80 (RAR) 20 µM
ER50891 (RAR) T3 2 µM
Retinaldehyde Retinoic Acid
other receptor/signalling Staurosporin (protein kinase) Chlorantraniliprol (RyR) 100 µM
systems SU5402 (VEGFR, FGFR) Dantrolene (RyR)
JTV-519 (RyR) 10 µM
BDNF 40 ng/ml
EGF 40 ng/ml
Endothelin-1 400 nM
Endothelin-3 400 nM
FGF2 40 ng/ml
GDNF 40 ng/ml
SDF 500 ng/ml
TGF-b1 10 ng/ml
VEGF 400 ng/ml
Figure 6.4: Tested compounds part I: Compounds with defined mode of action. Compounds are grouped in columns according to the updated prediction model. The molecular target is indicated in brackets. The indicated concentration is the highest tested concentration for compounds that were not cyototoxic.
Organochlorines DDT Triclosan 2,3,7,8-TCDD 75 nM
Dieldrin Lindane 20 µM
Heptachlor dioxin-like PCB (3 tested, see ms 3)
Hexachlorophene
ortho-chlorinated PCB (25 tested, see ms 3)
Organobromines PBB18 PBDE-153
PBDE-47 PBDE-99
Tetrabromobisphenol A
Organophosphates BPDP ETA-DOPO DOPO 100 µM
EG-DOPO Tricresyl phosphate EDA-DOPO 100 µM
EHDP EMIM DEP 20 µM
other pesticides Chlorpyriphos Carbaryl IPBC 1.25 µM
Rotenone Aldicarb 20 µM
Abiraterone 6-Hydroxydopamine Acetylsalicylic acid 20 µM
Diethylstilbestrol Berberine chloride Acetaminophen 20 µM
LiCl Geldanamycin L-Ascorbic acid 20 µM
Valinomycin VPA MPP+ 20 µM
Phenobarbital sodium salt 20 µM Saccharin sodium salt 20 µM
MMT = Methylcyclopentadienyl manganese tricarbonyl IPBC= 3-Iodo-2-propynyl butylcarbamate
TCDD = 2,3,7,8-Tetrachlorodibenzo-p-dioxin
Figure 6.5: Tested compounds part II: Environmental, drug-like and other toxicants. Compounds are grouped in columns according to the updated prediction model. The indicated concentration is the highest tested concentration for compounds that were not cytotoxic. Abbreviations: EMIM DEP:
1-Ethyl-3-6.3 Compounds disturbing human NCC migration
Environmental and drug-like compounds
Testing of several classes of environmental compounds, industrial chemicals and drug-like compounds (Fig. 6.5) revealed, that many halogenated organic and organophosphorous compounds are strong NCC migration-inhibitors. Several metal-containing compounds are weak NCC migration-inhibitors, whereas the majority of non-POP pesticides, industrial and drug-like compounds as well as all PAH were not toxic to NCC at the tested concentrations.
Flame retardants from several chemical classes (PBDE, organophosphates, DOPO-derivates), e.g. struc-turally not closely related compounds, were tested. Most of them disturbed NCC migration, but at different concentrations. For example, DOPO-derived flame retardants were toxic at 25-100µM, whereas organophosphorous and polybrominated flame retardants were toxic at 5-10µM. If they have the same flame retarding capacity, this would indicate that DOPO-derived flame retardants are less hazardous to NCC. Interestingely, there are some specific chemical structures that seem to be toxic to NCC:
Only organophosphorous flame retardants with aromatic side chains, but not aliphatic ones, and only non-planar, but not dioxin-like PCBs did interfere with NCC migration.
To summarize, these results indicate that the cMINC assay might be more suitable to prioritize environ-mental chemicals than drug-like compounds.
Comparison with NCC toxicants from literature
Only few compounds have been described in the literature as NCC toxicants (see part 1.3.4). They comprise ethanol, cyclopamine, RA, triazole-derived fungicides and VPA. Of these five, only the last three compounds have been tested in the cMINC assay.
RA strongly inhibited NCC migration and reduced cell speed. Also related compounds like retinaldehyde and RA receptor antagonists disturbed NCC migration. Several triazole-derived fungicides were tested in the cMINC assay (Fig. 6.5), but only triadimefon reproducibly disturbed NCC migration.
VPA had only a weak effect in the cMINC assay. It should be noted that for VPA only oneex vivo study found an effect using chicken neural tube explants (Fuller et al., 2002). In this study, the apparent inhibition of migration could also be due to inhibition of proliferation. Moreover, another study using rat neural tube explants could not confirm an effect on NCC migration (Usami et al., 2015).
Comparison of MINC and cMINC hits
With the previously established MINC assay, approximately 50 compounds have been tested for NCC migration-inhibition. Although the cMINC assay was performed in the same lab with cells from the same differentiation protocol, results could only partially be reproduced (Fig. 6.6). Surprisingly, some compounds were only effective at 10 times higher concentrations compared to the MINC assay. It could be that some compounds were not reproduced because the cMINC assay more strictly controls for cell proliferation and cell viability. Moreover, it might be that some compounds are only effective upon long (48 h) exposure but not in the shorter scenario.
Figure 6.6: Comparison of MINC and cMINC hits. Hit compounds from the MINC assay (Dreser et al., 2015; Zimmer et al., 2012, 2014) were compared to results of the cMINC assay.
Future directions
Figure 6.7: Combination of toxicity data with exposure information. To find out wether a given toxicity information is rele-vant, PBPK modelling should be performed to estimate the actual concentration. On the other hand, exposure data and risk assess-ment should be included to find out wether the in vitro effect is likely to be relevant in real life.
As it appeared that NCC are particularily sensitive to halo-genated and organophosphorus compounds, it will be of interest to find the mechanism. Moreover,in vivo studies with non-mammalian or mammalian models could help to an-swer the question whether this toxic effect is relevantin vivo.
However, information from risk assessment and exposure data should be taken into account (Fig. 6.7). Many NCC toxicants were effective only at relatively high concentrations (10-20µM) and it is questionable whether such concentra-tions would be reachedin vivo. On the other hand, it is at present also not known which fraction of the toxicants stick to the culture dish and are thus not available for the NCC.
Therefore, physiologically based pharmacokinetic (PBPK) modelling would help to estimate the correspondingin vivo concentration that is expected to induce NCC toxicity.