The Carcinogenome Project: In Vitro Gene Expression Profiling of Chemical Perturbations to Predict Long-Term Carcinogenicity

Prediction of molecular-specific mutagenic alerts and related mechanisms of chemicals by a convolutional neural network (CNN) model based on SMILES split

Structural alerts (SAs) are essential to identify chemicals for toxicity evaluation and health risk assessment. We constructed a novel SMILES split-based deep learning model (SSDL) that was trained and verified with 5850 chemicals from the ISSSTY database and 384 external test chemicals from published papers. The training accuracy was above 0.90 and the evaluation metrics (precision, recall and F1-score) all reached 0.78 or above on both internal and external test chemicals. In this model, the molecular-specific fragment importance of chemicals was first quantified independently. Then, the SA identification method based on the importance of these fragments was statistically analyzed and verified with the ISSSTY test and external test chemicals containing one of 28 typical SAs, and most of the performances were better than that of expert rules. Furthermore, a mutagenicity mechanism prediction method was developed using 237 chemicals with four known mutagenic mechanisms based on molecular similarity calibrated by the SSDL method and fragment importance, which significantly improved accuracy in three mechanisms and had comparable accuracy in the other one compared to traditional methods. Overall, the SSDL model quantifying fragment toxicity within molecules would be a novel potentially powerful tool in the determination and visualization of molecular-specific SAs and the prediction of mutagenicity mechanisms for environmental or industrial compounds and drugs.

Deciphering exogenous chemical carcinogenicity through interpretable deep learning: A novel approach for evaluating atmospheric pollutant hazards

Cancer remains a significant global health concern, with millions of deaths attributed to it annually. Environmental pollutants play a pivotal role in cancer etiology and contribute to the growing prevalence of this disease. The carcinogenic assessment of these pollutants is crucial for chemical health evaluation and environmental risk assessments. Traditional experimental methods are expensive and time-consuming, prompting the development of alternative approaches such as in silico methods. In this regard, deep learning (DL) has shown potential but lacks optimal performance and interpretability. This study introduces an interpretable DL model called CarcGC for chemical carcinogenicity prediction, utilizing a graph convolutional neural network (GCN) that employs molecular structural graphs as inputs. Compared to existing models, CarcGC demonstrated enhanced performance, with the area under the receiver operating characteristic curve (AUCROC) reaching 0.808 on the test set. Due to air pollution is closely related to the incidence of lung cancers, we applied the CarcGC to predict the potential carcinogenicity of chemicals listed in the United States Environmental Protection Agency's Hazardous Air Pollutants (HAPs) inventory, offering a foundation for environmental carcinogenicity screening. This study highlights the potential of artificially intelligent methods in carcinogenicity prediction and underscores the value of CarcGC interpretability in revealing the structural basis and molecular mechanisms underlying chemical carcinogenicity.

Exogenous Chemicals Impact Virus Receptor Gene Transcription: Insights from Deep Learning

Despite the fact that coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has been disrupting human life and health worldwide since the outbreak in late 2019, the impact of exogenous substance exposure on the viral infection remains unclear. It is well-known that, during viral infection, organism receptors play a significant role in mediating the entry of viruses to enter host cells. A major receptor of SARS-CoV-2 is the angiotensin-converting enzyme 2 (ACE2). This study proposes a deep learning model based on the graph convolutional network (GCN) that enables, for the first time, the prediction of exogenous substances that affect the transcriptional expression of the ACE2 gene. It outperforms other machine learning models, achieving an area under receiver operating characteristic curve (AUROC) of 0.712 and 0.703 on the validation and internal test set, respectively. In addition, quantitative polymerase chain reaction (qPCR) experiments provided additional supporting evidence for indoor air pollutants identified by the GCN model. More broadly, the proposed methodology can be applied to predict the effect of environmental chemicals on the gene transcription of other virus receptors as well. In contrast to typical deep learning models that are of black box nature, we further highlight the interpretability of the proposed GCN model and how it facilitates deeper understanding of gene change at the structural level.

Advancing chemical carcinogenicity prediction modeling: opportunities and challenges

Carcinogenicity assessment of any compound is a laborious and expensive exercise with several associated ethical and practical concerns. While artificial intelligence (AI) offers promising solutions, unfortunately, it is contingent on several challenges concerning the inadequacy of available experimentally validated (non)carcinogen datasets and variabilities within bioassays, which contribute to the compromised model training. Existing AI solutions that leverage classical chemistry-driven descriptors do not provide adequate biological interpretability involved in imparting carcinogenicity. This highlights the urgency to devise alternative AI strategies. We propose multiple strategies, including implementing data-driven (integrated databases) and known carcinogen-characteristic-derived features to overcome these apparent shortcomings. In summary, these next-generation approaches will continue facilitating robust chemical carcinogenicity prediction, concomitant with deeper mechanistic insights.

The Superfund Research Program Analytics Portal: linking environmental chemical exposure to biological phenotypes

The OSU/PNNL Superfund Research Program (SRP) represents a longstanding collaboration to quantify Polycyclic Aromatic Hydrocarbons (PAHs) at various superfund sites in the Pacific Northwest and assess their potential impact on human health. To link the chemical measurements to biological activity, we describe the use of the zebrafish as a high-throughput developmental toxicity model that provides quantitative measurements of the exposure to chemicals. Toward this end, we have linked over 150 PAHs found at Superfund sites to the effect of these same chemicals in zebrafish, creating a rich dataset that links environmental exposure to biological response. To quantify this response, we have implemented a dose-response modelling pipeline to calculate benchmark dose parameters which enable potency comparison across over 500 chemicals and 12 of the phenotypes measured in zebrafish. We provide a rich dataset for download and analysis as well as a web portal that provides public access to this dataset via an interactive web site designed to support exploration and re-use of these data by the scientific community at http://srp.pnnl.gov .

Toxicogenomics scoring system: TGSS, a novel integrated risk assessment model for chemical carcinogenicity prediction

BACKGROUND

Given the increasing exposure of humans to environmental chemicals and the limitations of conventional toxicity test, there is an urgent need to develop next-generation risk assessment methods.

OBJECTIVES

This study aims to establish a novel computational system named Toxicogenomics Scoring System (TGSS) to predict the carcinogenicity of chemicals coupling chemical-gene interactions with multiple cancer transcriptomic datasets.

METHODS

Chemical-related gene signatures were derived from chemical-gene interaction data from the Comparative Toxicogenomics Database (CTD). For each cancer type in TCGA, genes were ranked by their effects on tumorigenesis, which is based on the differential expression between tumor and normal samples. Next, we developed carcinogenicity scores (C-scores) using pre-ranked GSEA to quantify the correlation between chemical-related gene signatures and ranked gene lists. Then we established TGSS by systematically evaluating the C-scores in multiple chemical-tumor pairs. Furthermore, we examined the performance of our approach by ROC curves or prognostic analyses in TCGA and multiple independent cancer cohorts.

RESULTS

Forty-six environmental chemicals were finally included in the study. C-score was calculated for each chemical-tumor pair. The C-scores of IARC Group 3 chemicals were significantly lower than those of chemicals in Group 1 (P-value = 0.02) and Group 2 (P-values = 7.49 ×10^-5). ROC curves analysis indicated that C-score could distinguish "high-risk chemicals" from the other compounds (AUC = 0.67) with a specificity and sensitivity of 0.86 and 0.57. The results of survival analysis were also in line with the assessed carcinogenicity in TGSS for the chemicals in Group 1. Finally, consistent results were further validated in independent cancer cohorts.

CONCLUSION

TGSS highlighted the great potential of integrating chemical-gene interactions with gene-cancer relationships to predict the carcinogenic risk of chemicals, which would be valuable for systems toxicology.

Understanding exposures and latent disease risk within the National Institute of Environmental Health Sciences Superfund Research Program

Understanding the health effects of exposures when there is a lag between exposure and the onset of disease is an important and challenging topic in environmental health research. The National Institute of Environmental Health Sciences (NIEHS) Superfund Basic Research and Training Program (SRP) is a National Institutes of Health (NIH) grant program that uses a multidisciplinary approach to support biomedical and environmental science and engineering research. Because of the multidisciplinary nature of the program, SRP grantees are well-positioned to study exposure and latent disease risk across humans, animal models, and various life stages. SRP-funded scientists are working to address the challenge of connecting exposures that occur early in life and prior to conception with diseases that manifest much later, including developing new tools and approaches to predict how chemicals may affect long-term health. Here, we highlight research from the SRP focused on understanding the health effects of exposures with a lag between exposure and the onset of the disease as well as provide future directions for addressing knowledge gaps for this highly complex and challenging topic. Advancing the knowledge of latency to disease will require a multidisciplinary approach to research, the need for data sharing and integration, and new tools and computation approaches to make better predications about the timing of disease onset. A better understanding of exposures that may contribute to later-life diseases is essential to supporting the implementation of prevention and intervention strategies to reduce or modulate exposures to reduce disease burden.

Transcriptome-Wide Prediction and Measurement of Combined Effects Induced by Chemical Mixture Exposure in Zebrafish Embryos

BACKGROUND

Humans and environmental organisms are constantly exposed to complex mixtures of chemicals. Extending our knowledge about the combined effects of chemicals is thus essential for assessing the potential consequences of these exposures. In this context, comprehensive molecular readouts as retrieved by omics techniques are advancing our understanding of the diversity of effects upon chemical exposure. This is especially true for effects induced by chemical concentrations that do not instantaneously lead to mortality, as is commonly the case for environmental exposures. However, omics profiles induced by chemical exposures have rarely been systematically considered in mixture contexts.

OBJECTIVES

In this study, we aimed to investigate the predictability of chemical mixture effects on the whole-transcriptome scale.

METHODS

We predicted and measured the toxicogenomic effects of a synthetic mixture on zebrafish embryos. The mixture contained the compounds diuron, diclofenac, and naproxen. To predict concentration- and time-resolved whole-transcriptome responses to the mixture exposure, we adopted the mixture concept of concentration addition. Predictions were based on the transcriptome profiles obtained for the individual mixture components in a previous study. Finally, concentration- and time-resolved mixture exposures and subsequent toxicogenomic measurements were performed and the results were compared with the predictions.

RESULTS

This comparison of the predictions with the observations showed that the concept of concentration addition provided reasonable estimates for the effects induced by the mixture exposure on the whole transcriptome. Although nonadditive effects were observed only occasionally, combined, that is, multicomponent-driven, effects were found for mixture components with anticipated similar, as well as dissimilar, modes of action.

DISCUSSION

Overall, this study demonstrates that using a concentration- and time-resolved approach, the occurrence and size of combined effects of chemicals may be predicted at the whole-transcriptome scale. This allows improving effect assessment of mixture exposures on the molecular scale that might not only be of relevance in terms of risk assessment but also for pharmacological applications. https://doi.org/10.1289/EHP7773.

Rethinking Environmental Carcinogenesis

The 2010 report of the President's Cancer Panel concluded that the burden of cancer from chemical exposures is substantial, while the programs for testing and regulation of carcinogens remain inadequate. New research on the role of early life exposures and the ability of chemicals to act via multiple biological pathways, including immunosuppression, inflammation, and endocrine disruption as well as mutagenesis, further supports the potential for chemicals and chemical mixtures to influence disease. Epidemiologic observations, such as higher leukemia incidence in children living near roadways and industrial sources of air pollution, and new in vitro technologies that decode carcinogenesis at the molecular level, illustrate the diverse evidence that primary prevention of some cancers may be achieved by reducing harmful chemical exposures. The path forward requires cross-disciplinary approaches, increased environmental research investment, system-wide collaboration to develop safer economic alternatives, and community engagement to support evidence-informed action. Engagement by cancer researchers to integrate environmental risk factors into prevention initiatives holds tremendous promise for reducing the rates of disease.See all articles in this CEBP Focus section, "Environmental Carcinogenesis: Pathways to Prevention."

Drivers of and Obstacles to the Adoption of Toxicogenomics for Chemical Risk Assessment: Insights from Social Science Perspectives

BACKGROUND

Some 20 y ago, scientific and regulatory communities identified the potential of omics sciences (genomics, transcriptomics, proteomics, metabolomics) to improve chemical risk assessment through development of toxicogenomics. Recognizing that regulators adopt new scientific methods cautiously given accountability to diverse stakeholders, the scope and pace of adoption of toxicogenomics tools and data have nonetheless not met the ambitious, early expectations of omics proponents.

OBJECTIVE

Our objective was, therefore, to inventory, investigate, and derive insights into drivers of and obstacles to adoption of toxicogenomics in chemical risk assessment. By invoking established social science frameworks conceptualizing innovation adoption, we also aimed to develop recommendations for proponents of toxicogenomics and other new approach methodologies (NAMs).

METHODS

We report findings from an analysis of 56 scientific and regulatory publications from 1998 through 2017 that address the adoption of toxicogenomics for chemical risk assessment. From this purposeful sample of toxicogenomics discourse, we identified major categories of drivers of and obstacles to adoption of toxicogenomics tools and data sets. We then mapped these categories onto social science frameworks for conceptualizing innovation adoption to generate actionable insights for proponents of toxicogenomics.

DISCUSSION

We identify the most salient drivers and obstacles. From 1998 through 2017, adoption of toxicogenomics was understood to be helped by drivers such as those we labeled Superior scientific understanding, New applications, and Reduced cost & increased efficiency but hindered by obstacles such as those we labeled Insufficient validation, Complexity of interpretation, and Lack of standardization. Leveraging social science frameworks, we find that arguments for adoption that draw on the most salient drivers, which emphasize superior and novel functionality of omics as rationales, overlook potential adopters' key concerns: simplicity of use and compatibility with existing practices. We also identify two perspectives-innovation-centric and adopter-centric-on omics adoption and explain how overreliance on the former may be undermining efforts to promote toxicogenomics. https://doi.org/10.1289/EHP6500.

Investigating DNA adduct formation by flavor chemicals and tobacco byproducts in electronic nicotine delivery system (ENDS) using in silico approaches

The presence of flavors is one of the commonly cited reasons for use of e-cigarettes by youth; however, the potential harms from inhaling these chemicals and byproducts have not been extensively studied. One mechanism of interest is DNA adduct formation, which may lead to carcinogenesis. We identified two chemical classes of flavors found in tobacco products and byproducts, alkenylbenzenes and aldehydes, documented to form DNA adducts. Using in silico toxicology approaches, we identified structural analogs to these chemicals without DNA adduct information. We conducted a structural similarity analysis and also generated in silico model predictions of these chemicals for genotoxicity, mutagenicity, carcinogenicity, and skin sensitization. The empirical and in silico data were compared, and we identified strengths and limitations of these models. Good concordance (80-100%) was observed between DNA adduct formation and models predicting mammalian mutagenicity (mouse lymphoma sassy L5178Y) and skin sensitization for both chemical classes. On the other hand, different prediction profiles were observed for the two chemical classes for the modeled endpoints, unscheduled DNA synthesis and bacterial mutagenicity. These results are likely due to the different mode of action between the two chemical classes, as aldehydes are direct acting agents, while alkenylbenzenes require bioactivation to form electrophilic intermediates, which form DNA adducts. The results of this study suggest that an in silico prediction for the mouse lymphoma assay L5178Y, may serve as a surrogate endpoint to help predict DNA adduct formation for chemicals found in tobacco products such as flavors and byproducts.

Application of transcriptomic and microRNA profiling in the evaluation of potential liver carcinogens

Hepatocarcinogens are agents that increase the incidence of liver cancer in exposed animals or humans. It is now established that carcinogenic exposures have a widespread impact on the transcriptome, inducing both adaptive and adverse changes in the activities of genes and pathways. Chemical hepatocarcinogens have also been shown to affect expression of microRNA (miRNA), the evolutionarily conserved noncoding RNA that regulates gene expression posttranscriptionally. Considerable effort has been invested into examining the involvement of mRNA in chemical hepatocarcinogenesis and their potential usage for the classification and prediction of new chemical entities. For miRNA, there has been an increasing number of studies reported over the past decade, although not to the same degree as for transcriptomic studies. Current data suggest that it is unlikely that any gene or miRNA signature associated with short-term carcinogen exposure can replace the rodent bioassay. In this review, we discuss the application of transcriptomic and miRNA profiles to increase mechanistic understanding of chemical carcinogens and to aid in their classification.

Exploring the Use of Compound-Induced Transcriptomic Data Generated From Cell Lines to Predict Compound Activity Toward Molecular Targets

Pharmaceutical or phytopharmaceutical molecules rely on the interaction with one or more specific molecular targets to induce their anticipated biological responses. Nonetheless, these compounds are also prone to interact with many other non-intended biological targets, also known as off-targets. Unfortunately, off-target identification is difficult and expensive. Consequently, QSAR models predicting the activity on a target have gained importance in drug discovery or in the de-risking of chemicals. However, a restricted number of targets are well characterized and hold enough data to build such in silico models. A good alternative to individual target evaluations is to use integrative evaluations such as transcriptomics obtained from compound-induced gene expression measurements derived from cell cultures. The advantage of these particular experiments is to capture the consequences of the interaction of compounds on many possible molecular targets and biological pathways, without having any constraints concerning the chemical space. In this work, we assessed the value of a large public dataset of compound-induced transcriptomic data, to predict compound activity on a selection of 69 molecular targets. We compared such descriptors with other QSAR descriptors, namely the Morgan fingerprints (similar to extended-connectivity fingerprints). Depending on the target, active compounds could show similar signatures in one or multiple cell lines, whether these active compounds shared similar or different chemical structures. Random forest models using gene expression signatures were able to perform similarly or better than counterpart models built with Morgan fingerprints for 25% of the target prediction tasks. These performances occurred mostly using signatures produced in cell lines showing similar signatures for active compounds toward the considered target. We show that compound-induced transcriptomic data could represent a great opportunity for target prediction, allowing to overcome the chemical space limitation of QSAR models.

The Promises and Challenges of Toxico-Epigenomics: Environmental Chemicals and Their Impacts on the Epigenome

BACKGROUND

It has been estimated that a substantial portion of chronic and noncommunicable diseases can be caused or exacerbated by exposure to environmental chemicals. Multiple lines of evidence indicate that early life exposure to environmental chemicals at relatively low concentrations could have lasting effects on individual and population health. Although the potential adverse effects of environmental chemicals are known to the scientific community, regulatory agencies, and the public, little is known about the mechanistic basis by which these chemicals can induce long-term or transgenerational effects. To address this question, epigenetic mechanisms have emerged as the potential link between genetic and environmental factors of health and disease.

OBJECTIVES

We present an overview of epigenetic regulation and a summary of reported evidence of environmental toxicants as epigenetic disruptors. We also discuss the advantages and challenges of using epigenetic biomarkers as an indicator of toxicant exposure, using measures that can be taken to improve risk assessment, and our perspectives on the future role of epigenetics in toxicology.

DISCUSSION

Until recently, efforts to apply epigenomic data in toxicology and risk assessment were restricted by an incomplete understanding of epigenomic variability across tissue types and populations. This is poised to change with the development of new tools and concerted efforts by researchers across disciplines that have led to a better understanding of epigenetic mechanisms and comprehensive maps of epigenomic variation. With the foundations now in place, we foresee that unprecedented advancements will take place in the field in the coming years. https://doi.org/10.1289/EHP6104.

Gene expression and cytosine DNA methylation alterations in induced pluripotent stem-cell-derived human hepatocytes treated with low doses of chemical carcinogens

The increasing number of man-made chemicals in the environment that may pose a carcinogenic risk emphasizes the need to develop reliable time- and cost-effective approaches for carcinogen detection. To address this issue, we have investigated the utility of human hepatocytes for the in vitro identification of genotoxic and non-genotoxic carcinogens. Induced pluripotent stem-cell (iPSC)-derived human hepatocytes were treated with the genotoxic carcinogens aflatoxin B₁ (AFB1) and benzo[a]pyrene (B[a]P), the non-genotoxic liver carcinogen methapyrilene, and the non-carcinogens aflatoxin B₂ (AFB2) and benzo[e]pyrene (B[e]P) at non-cytotoxic concentrations for 7 days, and transcriptomic and DNA methylation profiles were examined. 1569, 1693, and 2061 differentially expressed genes (DEGs) were detected in cells treated with AFB1, B[a]P, and methapyrilene, respectively, whereas no DEGs were found in cells treated with AFB2 or B[e]P. In contrast to the profound cellular transcriptomic responses, exposure of iPSC-derived hepatocytes to the test chemicals resulted in minor random alterations in global DNA methylome, most of which were not associated with changes in gene expression. Overall, our results demonstrate that the major non-genotoxic effect of exposure to carcinogens, regardless of their mode of action, is a profound global transcriptomic response rather than global DNA methylome alterations, indicating the significance of transcriptomic alterations as an informative endpoint in short-term in vitro carcinogen testing.

L1000 Viewer: A Search Engine and Web Interface for the LINCS Data Repository

The LINCS L1000 data repository contains almost two million gene expression profiles for thousands of small molecules and drugs. However, due to the complexity and the size of the data repository and a lack of an interoperable interface, the creation of pharmacologically meaningful workflows utilizing these data is severely hampered. In order to overcome this limitation, we developed the L1000 Viewer, a search engine and graphical web interface for the LINCS data repository. The web interface serves as an interactive platform allowing the user to select different forms of perturbation profiles, e.g., for specific cell lines, drugs, dosages, time points and combinations thereof. At its core, our method has a database we created from inferring and utilizing the intricate dependency graph structure among the data files. The L1000 Viewer is accessible via http://L1000viewer.bio-complexity.com/.