Citation:

Byrd, G., M. Culbreth, T. Shafer, F. Harris, AND J. Harrill. Towards an Animal Component-Free System for High Throughput Phenotypic Profiling of Human Neural Progenitor Cells. SOT Conference 2024: A New Approach Method (NAM) to Screen for the Impact of Endogenous Stress on Chemical Toxicity, Salt Lake City, UT, March 10 - 14, 2024. https://doi.org/10.23645/epacomptox.25352776

Impact/Purpose:

Presentation to SOT Conference 2024: A New Approach Method (NAM) to Screen for the Impact of Endogenous Stress on Chemical Toxicity 

Description:

Background and Purpose       High Throughput Phenotypic Profiling (HTPP) is an in vitro screening method that combines cell viability and “Cell Painting” assays to derive chemical-specific phenotypic profiles and phenotype altering concentrations (PACs) to inform hazard assessment. Human neural progenitor (hNP1) cells have been utilized in this method to inform developmental neurotoxicity (DNT) hazard assessment but, to this point, have been cultured on a mouse-derived laminin substrate. To facilitate transition to an animal component-free system, previous work identified LN111 as a human recombinant laminin substrate that supports the growth of hNP1 cells with high phenotypic similarity to hNP1 cells conventionally cultured on mouse-derived laminin. These findings were corroborated by further work wherein hNP1 cells cultured on either LN111 or mouse-derived laminin (MDL) were treated with a set of 23 phenotypic reference chemicals. Comparing the parallel screens revealed significant (p < .001) levels of phenotypic profile correlation at a minimum of one non-cytotoxic dose-level for every chemical tested and PACs within one order of magnitude for 87% of chemicals tested. In the present experiments, a set of 282 DNT-relevant compounds were screened in hNP1 cells cultured on LN111 or MDL and comparative analyses were performed to confirm LN111 as a suitable replacement for MDL in HTPP of hNP1 cells. Methods               This set of 282 compounds contained DNT reference compounds that either had reported DNT outcomes in mammals or were without evidence of DNT in mammals. These were tested alongside 3 phenotypic reference chemicals – aphidicolin, bafilomycin A1, and cucurbitacin I – that produce well-characterized phenotypic effects in hNP1 cells grown on either laminin type. Staurosporine was included as an assay positive control that decreases cell viability. Each test plate included 16 dimethyl sulfoxide vehicle control wells. Both screens were conducted independently using our established HTPP laboratory and data analysis pipelines before comparative analysis of screening results was performed. Results               As part of the established data analysis pipeline, cell viability and Cell Painting benchmark concentrations and hitcalls were determined via tcplfit2-based concentration response modeling. By these metrics, inactive and active chemicals were remarkably concordant, with approximately 80% of chemicals being consistently inactive or active across both screens. Of the chemicals that were active across both screens only 4 had benchmark concentrations that differed by more than an order of magnitude, with hNP1 cells seeded on LN111 being more sensitive in every case. Approximately 20% of chemicals were inactive on one laminin type and active on the other[CK1] , but activity was observed on LN111 and not MDL in 70% of these cases. The tcplfit2 curve fitting results were then utilized to calculate signed-scaled area under the curve (ssAUC) values for each of the 1300 phenotypic features measured for each chemical. Phenotypic profiles were then developed by calculating ssAUC values for a reduced feature set determined using a stepwise feature elimination procedure to remove redundant or non-informative features. Kendall correlations for profiles of like-chemicals across laminin types were higher than those calculated for mismatched pairs. This shift was particularly apparent for chemicals that were active across both laminin types. Of those chemicals that were concordantly active, 95% were at least moderately positively correlated (> 0.25), and two-thirds were strongly positively correlated (> 0.5). Kendall correlation-based clustering was then used to cluster the ssAUC profiles of all of the chemicals within each  . . .    

URLs/Downloads:

Record Details: