Machine Learning Models for Estrogen Receptor Bioactivity and Endocrine Disruption Prediction.

The U.S. Environmental Protection Agency (EPA) periodically releases in vitro data across a variety of targets, including the estrogen receptor (ER). In 2015, the EPA used these data to construct mathematical models of ER agonist and antagonist pathways to prioritize chemicals for endocrine disruption testing. However, mathematical models require in vitro data prior to predicting estrogenic activity, but machine learning methods are capable of prospective prediction from the molecular structure alone. The current study describes the generation and evaluation of Bayesian machine learning models grouped by the EPA's ER agonist pathway model using multiple data types with proprietary software, Assay Central. External predictions with three test sets of in vitro and in vivo reference chemicals with agonist activity classifications were compared to previous mathematical model publications. Training data sets were subjected to additional machine learning algorithms and compared with rank normalized scores of internal five-fold cross-validation statistics. External predictions were found to be comparable or superior to previous studies published by the EPA. When assessing six additional algorithms for the training data sets, Assay Central performed similarly at a reduced computational cost. This study demonstrates that machine learning can prioritize chemicals for future in vitro and in vivo testing of ER agonism.

Comparing Machine Learning Models for Aromatase (P450 19A1).

Aromatase, or cytochrome P450 19A1, catalyzes the aromatization of androgens to estrogens within the body. Changes in the activity of this enzyme can produce hormonal imbalances that can be detrimental to sexual and skeletal development. Inhibition of this enzyme can occur with drugs and natural products as well as environmental chemicals. Therefore, predicting potential endocrine disruption via exogenous chemicals requires that aromatase inhibition be considered in addition to androgen and estrogen pathway interference. Bayesian machine learning methods can be used for prospective prediction from the molecular structure without the need for experimental data. Herein, the generation and evaluation of multiple machine learning models utilizing different sources of aromatase inhibition data are described. These models are applied to two test sets for external validation with molecules relevant to drug discovery from the public domain. In addition, the performance of multiple machine learning algorithms was evaluated by comparing internal five-fold cross-validation statistics of the training data. These methods to predict aromatase inhibition from molecular structure, when used in concert with estrogen and androgen machine learning models, allow for a more holistic assessment of endocrine-disrupting potential of chemicals with limited empirical data and enable the reduction of the use of hazardous substances.

Comparison of Machine Learning Models for the Androgen Receptor.

The androgen receptor (AR) is a target of interest for endocrine disruption research, as altered signaling can affect normal reproductive and neurological development for generations. In an effort to prioritize compounds with alternative methodologies, the U.S. Environmental Protection Agency (EPA) used in vitro data from 11 assays to construct models of AR agonist and antagonist signaling pathways. While these EPA ToxCast AR models require in vitro data to assign a bioactivity score, Bayesian machine learning methods can be used for prospective prediction from molecule structure alone. This approach was applied to multiple types of data corresponding to the EPA's AR signaling pathway with proprietary software, Assay Central. The training performance of all machine learning models, including six other algorithms, was evaluated by internal 5-fold cross-validation statistics. Bayesian machine learning models were also evaluated with external predictions of reference chemicals to compare prediction accuracies to published results from the EPA. The machine learning model group selected for further studies of endocrine disruption consisted of continuous AC50 data from the February 2019 release of ToxCast/Tox21. These efforts demonstrate how machine learning can be used to predict AR-mediated bioactivity and can also be applied to other targets of endocrine disruption.

Estrogen receptor α and aryl hydrocarbon receptor independent growth inhibitory effects of aminoflavone in breast cancer cells.

BACKGROUND: Numerous studies have implicated the aryl hydrocarbon receptor (AhR) as a potential therapeutic target for several human diseases, including estrogen receptor alpha (ERα) positive breast cancer. Aminoflavone (AF), an activator of AhR signaling, is currently undergoing clinical evaluation for the treatment of solid tumors. Of particular interest is the potential treatment of triple negative breast cancers (TNBC), which are typically more aggressive and characterized by poorer outcomes. Here, we examined AF's effects on two TNBC cell lines and the role of AhR signaling in AF sensitivity in these model cell lines. METHODS: AF sensitivity in MDA-MB-468 and Cal51 was examined using cell counting assays to determine growth inhibition (GI50) values. Luciferase assays and qPCR of AhR target genes cytochrome P450 (CYP) 1A1 and 1B1 were used to confirm AF-mediated AhR signaling. The requirement of endogenous levels of AhR and AhR signaling for AF sensitivity was examined in MDA-MB-468 and Cal51 cells stably harboring inducible shRNA for AhR. The mechanism of AF-mediated growth inhibition was explored using flow cytometry for markers of DNA damage and apoptosis, cell cycle analysis, and ß-galactosidase staining for senescence. Luciferase data was analyzed using Student's T test. Three-parameter nonlinear regression was performed for cell counting assays. RESULTS: Here, we report that ERα-negative TNBC cell lines MDA-MB-468 and Cal51 are sensitive to AF. Further, we presented evidence suggesting that neither endogenous AhR expression levels nor downstream induction of AhR target genes CYP1A1 and CYP1B1 is required for AF-mediated growth inhibition in these cells. Between these two ERα negative cell lines, we showed that the mechanism of AF action differs slightly. Low dose AF mediated DNA damage, S-phase arrest and apoptosis in MDA-MB-468 cells, while it resulted in DNA damage, S-phase arrest and cellular senescence in Cal51 cells. CONCLUSIONS: Overall, this work provides evidence against the simplified view of AF sensitivity, and suggests that AF could mediate growth inhibitory effects in ERα-positive and negative breast cancer cells, as well as cells with impaired AhR expression and signaling. While AF could have therapeutic effects on broader subtypes of breast cancer, the mechanism of cytotoxicity is complex, and likely, cell line- and tumor-specific.

Aminoflavone-loaded EGFR-targeted unimolecular micelle nanoparticles exhibit anti-cancer effects in triple negative breast cancer.

Triple negative breast cancer (TNBC) is an aggressive subtype of breast cancer for which there is no available targeted therapy. TNBC cases contribute disproportionately to breast cancer-related mortality, thus the need for novel and effective therapeutic methods is urgent. We have previously shown that a National Cancer Institute (NCI) investigational drug aminoflavone (AF) exhibits strong growth inhibitory effects in TNBC cells. However, in vivo pulmonary toxicity resulted in withdrawal or termination of several human clinical trials for AF. Herein we report the in vivo efficacy of a nanoformulation of AF that enhances the therapeutic index of AF in TNBC. We engineered a unique unimolecular micelle nanoparticle (NP) loaded with AF and conjugated with GE11, a 12 amino acid peptide targeting epidermal growth factor receptor (EGFR), since EGFR amplification is frequently observed in TNBC tumors. These unimolecular micelles possessed excellent stability and preferentially released drug payload at endosomal pH levels rather than blood pH levels. Use of the GE11 targeting peptide resulted in enhanced cellular uptake and strong growth inhibitory effects in TNBC cells. Further, AF-loaded, GE11-conjugated (targeted) unimolecular micelle NPs significantly inhibit orthotopic TNBC tumor growth in a xenograft model, compared to treatment with AF-loaded, GE11-lacking (non-targeted) unimolecular micelle NPs or free AF. Interestingly, the animals treated with AF-loaded, targeted NPs had the highest plasma and tumor level of AF among different treatment groups yet exhibited no increase in plasma aspartate aminotransferase (AST) activity level or observable tissue damage at the time of sacrifice. Together, these results highlight AF-loaded, EGFR-targeted unimolecular micelle NPs as an effective therapeutic option for EGFR-overexpressing TNBC.