Comparative analysis between zebrafish and an automated live-cell assay to classify developmental neurotoxicant chemicals.

Modern toxicology's throughput has dramatically increased due to alternative models, laboratory automation, and machine learning. This has enabled comparative studies across species and assays to prioritize chemical hazard potential and to understand how different model systems might complement one another. However, such comparative studies of high-throughput data are still in their infancy, with more groundwork needed to firmly establish the approach. Therefore, this study aimed to compare the bioactivity of the NIEHS Division of Translational Toxicology's (DTT) 87-compound developmental neurotoxicant (DNT) library in zebrafish and an in vitro high-throughput cell culture system. The early life-stage zebrafish provided a whole animal approach to developmental toxicity assessment. Chemical hits for abnormalities in embryonic zebrafish morphology, mortality, and behavior (ZBEscreen™) were compared with chemicals classified as high-risk by the Cell Health Index (CHI™), which is an outcome class probability from a machine learning classifier using 12 parameters from the SYSTEMETRIC® Cell Health Screen (CHS). The CHS was developed to assess human toxicity risk using supervised machine learning to classify acute cell stress phenotypes in a human leukemia cell line (HL60 cells) following a 4-h exposure to a chemical of interest. Due to the design of the screen, the zebrafish assays were more exhaustive, yielding 86 total bioactive hits, whereas the SYSTEMETRIC® CHS focusing on acute toxicity identified 20 chemicals as potentially toxic. The zebrafish embryonic and larval photomotor response assays (EPR and LPR, respectively) detected 40 of the 47 chemicals not found by the zebrafish morphological screen and CHS. Collectively, these results illustrate the advantages of using two alternative models in tandem for rapid hazard assessment and chemical prioritization and the effectiveness of CHI™ in identifying toxicity within a single multiparametric assay.

Utilization of Supervised Machine Learning to Understand Kinase Inhibitor Toxophore Profiles.

There have been more than 70 FDA-approved drugs to target the ATP binding site of kinases, mainly in the field of oncology. These compounds are usually developed to target specific kinases, but in practice, most of these drugs are multi-kinase inhibitors that leverage the conserved nature of the ATP pocket across multiple kinases to increase their clinical efficacy. To utilize kinase inhibitors in targeted therapy and outside of oncology, a narrower kinome profile and an understanding of the toxicity profile is imperative. This is essential when considering treating chronic diseases with kinase targets, including neurodegeneration and inflammation. This will require the exploration of inhibitor chemical space and an in-depth understanding of off-target interactions. We have developed an early pipeline toxicity screening platform that uses supervised machine learning (ML) to classify test compounds' cell stress phenotypes relative to a training set of on-market and withdrawn drugs. Here, we apply it to better understand the toxophores of some literature kinase inhibitor scaffolds, looking specifically at a series of 4-anilinoquinoline and 4-anilinoquinazoline model libraries.

Optimization of the 4-anilinoquin(az)oline scaffold as epidermal growth factor receptor (EGFR) inhibitors for chordoma utilizing a toxicology profiling assay platform.

The 4-anilinoquin(az)oline is a well-known kinase inhibitor scaffold incorporated in clinical inhibitors including gefitinib, erlotinib, afatinib, and lapatinib, all of which have previously demonstrated activity against chordoma cell lines in vitro. We screened a focused array of compounds based on the 4-anilinoquin(az)oline scaffold against both U-CH1 and the epidermal growth factor receptor (EGFR) inhibitor resistant U-CH2. To prioritize the hit compounds for further development, we screened the compound set in a multiparameter cell health toxicity assay. The de-risked compounds were then screened against a wider panel of patient derived cell lines and demonstrated low micromolar efficacy in cells. We also investigated the properties that gave rise to the toxophore markers, including the structural and electronic features, while optimizing for EGFR in-cell target engagement. These de-risked leads present a potential new therapeutic avenue for treatment of chordomas and new chemical tools and probe compound 45 (UNC-CA359) to interrogate EGFR mediated disease phenotypes.

Acute cell stress screen with supervised machine learning predicts cytotoxicity of excipients.

Excipients serve as vehicles, preservatives, solubilizers, and colorants for drugs, food, and cosmetics. They are considered to be inert at biological targets; however, several reports suggest that some could interact with human targets and cause unwanted effects. We investigated 40 commonly used drug excipients for cellular stress in the AsedaSciences® SYSTEMETRIC® Cell Health Screen, which was developed to estimate toxicity risk of small molecular entities (SMEs). The screen uses supervised machine learning (ML) to classify test compound cell stress phenotypes relative to a training set of on-market and withdrawn drugs. While 80% (n = 32) of the excipients did not show elevated risk in a broad, but pharmacologically relevant, concentration range (5 nM to 100 µM), we identified 20% (n = 8) with elevated risk. This group included two mercury containing preservatives, propyl gallate, methylene blue, benzethonium chloride, and cetylpyridinium chloride, all known for previously reported safety issues. All compounds were tested in parallel in an in vitro assay panel regularly used to investigate off-target effects of drug candidates. Target engagement in this assay panel confirmed risk-indicative biological activity for the same excipients, except propyl gallate, which may have a separate, interesting mechanism. We conclude that the SYSTEMETRIC Cell Health Screen, in conjunction with in vitro pharmacological profiling, can provide a fast and cost effective methodology for first line testing of SMEs, including excipients, to avoid cellular damage, particularly in the GI, where they are represented in high concentrations.

A Chemogenomic Screening Platform Used to Identify Chemotypes Perturbing HSP90 Pathways.

Compounds that modulate the heat shock protein (HSP) network have potential in a broad range of research applications and diseases. A yeast-based liquid culture assay that measured time-dependent turbidity enabled the high-throughput screening of different Saccharomyces cerevisae strains to identify HSP modulators with unique molecular mechanisms. A focused set of four strains, with differing sensitivities to Hsp90 inhibitors, was used to screen a compound library of 3680 compounds. Computed turbidity curve functions were used to classify strain responses and sensitivity to chemical effects across the compound library. Filtering based on single-strain selectivity identified nine compounds as potential heat shock modulators, including the known Hsp90 inhibitor macbecin. Haploid yeast deletion strains (360), mined from previous Hsp90 inhibitor yeast screens and heat shock protein interaction data, were screened for differential sensitivities to known N-terminal ATP site-directed Hsp90 inhibitors to reveal functional distinctions. Strains demonstrating differential sensitivity (13) to Hsp90 inhibitors were used to prioritize primary screen hit compounds, with NSC145366 emerging as the lead hit. Our follow-up biochemical and functional studies show that NSC145366 directly interacts and inhibits the C-terminus of Hsp90, validating the platform as a powerful approach for early-stage identification of bioactive modulators of heat shock-dependent pathways.

Fluorinated amino-derivatives of the sesquiterpene lactone, parthenolide, as (19)f NMR probes in deuterium-free environments.

The design, synthesis, and biological activity of fluorinated amino-derivatives of the sesquiterpene lactone, parthenolide, are described. A fluorinated aminoparthenolide analogue with biological activity similar to the parent natural product was discovered, and its X-ray structure was obtained. This lead compound was then studied using (19)F NMR in the presence and absence of glutathione to obtain additional mechanism of action data, and it was found that the aminoparthenolide eliminates amine faster in the presence of glutathione than in the absence of glutathione. The exact changes in concentrations of fluorinated compound and amine were quantified by a concentration-reference method using (19)F NMR; a major benefit of applying this strategy is that no deuterated solvents or internal standards are required to obtain accurate concentrations. These mechanistic data with glutathione may contribute to the conversion of the amino-derivative to parthenolide, the active pharmacological agent, in glutathione-rich cancer cells.

A synergistic antiproliferation effect of curcumin and docosahexaenoic acid in SK-BR-3 breast cancer cells: unique signaling not explained by the effects of either compound alone.

BACKGROUND: Breast cancer is a collection of diseases in which molecular phenotypes can act as both indicators and mediators of therapeutic strategy. Therefore, candidate therapeutics must be assessed in the context of multiple cell lines with known molecular phenotypes. Docosahexaenoic acid (DHA) and curcumin (CCM) are dietary compounds known to antagonize breast cancer cell proliferation. We report that these compounds in combination exert a variable antiproliferative effect across multiple breast cell lines, which is synergistic in SK-BR-3 cells and triggers cell signaling events not predicted by the activity of either compound alone. METHODS: Dose response curves for CCM and DHA were generated for five breast cell lines. Effects of the DHA+ CCM combination on cell proliferation were evaluated using varying concentrations, at a fixed ratio, of CCM and DHA based on their individual ED50. Detection of synergy was performed using nonlinear regression of a sigmoid dose response model and Combination Index approaches. Cell molecular network responses were investigated through whole genome microarray analysis of transcript level changes. Gene expression results were validated by RT-PCR, and western blot analysis was performed for potential signaling mediators. Cellular curcumin uptake, with and without DHA, was analyzed via flow cytometry and HPLC. RESULTS: CCM+DHA had an antiproliferative effect in SK-BR-3, MDA-MB-231, MDA-MB-361, MCF7 and MCF10AT cells. The effect was synergistic for SK-BR-3 (ER⁻ PR⁻ Her2⁺) relative to the two compounds individually. A whole genome microarray approach was used to investigate changes in gene expression for the synergistic effects of CCM+DHA in SK-BR-3 cells lines. CCM+DHA triggered transcript-level responses, in disease-relevant functional categories, that were largely non-overlapping with changes caused by CCM or DHA individually. Genes involved in cell cycle arrest, apoptosis, inhibition of metastasis, and cell adhesion were upregulated, whereas genes involved in cancer development and progression, metastasis, and cell cycle progression were downregulated. Cellular pools of PPARγ and phospho-p53 were increased by CCM+DHA relative to either compound alone. DHA enhanced cellular uptake of CCM in SK-BR-3 cells without significantly enhancing CCM uptake in other cell lines. CONCLUSIONS: The combination of DHA and CCM is potentially a dietary supplemental treatment for some breast cancers, likely dependent upon molecular phenotype. DHA enhancement of cellular curcumin uptake is one potential mechanism for observed synergy in SK-BR-3 cells; however, transcriptomic data show that the antiproliferation synergy accompanies many signaling events unique to the combined presence of the two compounds.