The vital roles of red blood cells (RBCs) underscore the importance of functional and homeostatic cellular processes in erythroid precursors. Iron acquisition and trafficking by erythroid cells is a key physiological process for their survival, proliferation and differentiation. In contrast, exposure to hematotoxicants can deplete erythroid precursors or disrupt erythroid differentiation. Complete elucidation of mechanisms relevant to a biological process or response to an injury in erythroid cells can improve our understanding of the pathophysiology and etiology of RBC disorders. Interrogating the genome in the context of a cellular process is a key strategy to understand underlying molecular mechanisms. Comprehensive identification of pathways governing physiological as well as injurious cellular processes has become more feasible due to the emergence of genome-wide functional genomic approaches. Recently, the CRISPR-Cas9 system has revolutionized the field of functional genomics as it enabled simple, efficient and cost-effective platforms for large-scale genetic screening. In this series of studies, CRISPR-based genome-wide loss-of function screening was employed to identify genes and pathways contributing to erythroid key processes including iron uptake, heme trafficking and erythroid differentiation, as well as response to hematotoxicants.

Chapter 1 reviews the various CRISPR-Cas9 screening platforms and their applications in functional genomics. Different screening strategies using large-scale loss-of-function and gain-of-function approaches are compared in terms of their effectiveness, feasibility, and suitability to study particular cellular processes. Steps of pooled screening, data analysis methods and hit validation strategies are described and critical technical considerations are addressed. In addition, the power of CRISPR-based genetic screening in different research fields including basic cell biology, cancer, drug discovery, pharmacogenomics and toxicogenomics is demonstrated. Finally, limitations of the current CRISPR screening platforms are revised and strategies for improvement are proposed.

In chapter 2, a genome-wide CRISPR-based screen designed to study iron uptake in erythroid cells is described. The physiological iron form, transferrin-bound iron (TBI), is taken up by erythroid cells through the transferrin receptor (TfR) by endocytosis. Non-transferrin bound iron (NTBI), which is present in the circulation during iron overload, can also be acquired by erythroid cells. Uncontrolled cellular NTBI influx can result in iron toxicity. To identify molecular determinants of TBI and NTBI uptake, a genome-wide loss-of-function screen was performed in human K562 erythroleukemic cells which can utilize both TBI and NTBI to grow and proliferate. The screen revealed multiple genes whose disruption resulted in defective growth when either TBI or NTBI is the exclusive source of iron. Unsurprisingly, TBI uptake candidates included the transferrin receptor (TfR1) in addition to several components of the endocytic pathway. Follow up studies on one candidate, CCDC115, confirmed its role in cellular TBI uptake. CCDC115 is a V-ATPase assembly factor and its role in endosomal acidification likely underlies its function in transferrin iron influx.

In chapter 3, a genome-wide knockout screen investigating cellular mechanisms of heme trafficking and heme-induced erythroid differentiation is documented. Under physiological conditions, heme is synthesized in the mitochondria of erythroid cells where it plays structural as well as functional roles in erythroid differentiation. Additionally, erythroid cells are capable of taking up extracellular heme which might be utilized in pathological forms of erythropoiesis or could induce toxicity in the absence of other erythropietic signals. Heme treatment of K562 erythroleukemic cells induces erythroid differentiation resulting in a proliferation block. The screen was designed to identify genes whose inactivation alleviates the heme-induced proliferation block. The identified candidates include components of the clathrin-mediated endocytosis and vesicle acidification pathways in addition to epigenetic and RNA processing regulators. Intriguingly, the V-ATPase assembly factor CCDC115 was among the top candidates. Analysis of CCDC115 deficient cells unveiled an unprecedented role of CCDC115 in cellular heme uptake. Gene products whose loss results in heme-induced toxicity were also identified and included the heme oxygenase HMOX2, other detoxification enzymes and members of the ABC transporter family.

Chapter 4 reports identification of cellular mechanisms that modulate erythroid cell sensitivity to arsenic trioxide (ATO), a potent hematotoxicant and an effective anti-leukemic agent. A genome-wide loss-of-function CRISPR-based screen revealed novel molecular components influencing susceptibility of K562 erythroleukemic cells to ATO. Many of the candidates identified in the primary screen were simultaneously validated in a secondary screening approach. Functional enrichment analysis of validated genes revealed multiple pathways/processes implicated in the cellular response to ATO. The most significant pathway controls biosynthesis of selenocysteine, the 21st amino acid, and its incorporation into selenoproteins. Inactivation of components of this pathway resulted in notable cellular resistance to ATO. Based on the screening results, two models explaining the role of selenocysteine metabolism in ATO toxicity were proposed: the compromised thioredoxin reductase model and the arsenic-selenium-glutathione export model. Intriguingly, selenium pre-treatment of cells exhibited a protective effect against ATO cytotoxicity.

In chapter 5, susceptibility of erythroid cells to another hematotoxicant, acetaldehyde, was studied. Acetaldehyde, the primary product of alcohol metabolism and an endogenous metabolite, is a potential carcinogen and its well-established genotoxicity is thought to underlie bone marrow failure in Fanconi Anemia. A genome-wide knockout screen in erythroleukemic K562 cells identified several determinants of sensitivity to acetaldehyde. Consistent with the documented role of aldehydes in DNA damage, multiple identified candidate genes encode DNA repair enzymes. The top candidate gene encodes the tumor suppressor OVCA2, whose function is unknown. The uncovered role of OVCA2 in detoxifying acetaldehyde was validated in a secondary screen and by individual disruption of the OVCA2 gene. Interestingly, OVCA2 deficient cells displayed increased accumulation of the acetaldehyde-derived DNA adduct N2-ethylidene-2G.

Overall, these studies demonstrated the effectiveness of CRISPR/Cas9 loss-of-function genetic screening in deciphering mechanisms relevant to erythroid cell survival, proliferation, differentiation and response to injury.

Chemical contaminants are being detected with increasing frequency in the environment, both as a result of better analytical monitoring techniques and from increasing contaminant concentrations. It is important to understand if un-monitored "emerging contaminant" chemicals can affect or cause toxicity in organisms of ecological importance or affect ecological health. Acute toxicity testing is a usual first step to determine contaminant effects on model organisms. In this work, the nominal acute LC₅₀ (concentration that kills fifty percent of animals tested) was established for chemicals from two classes of emerging contaminants, silver nanowires and chemical flame-retardants, on freshwater crustacean, Daphnia magna. Gene expression studies were then done with microarray technology at sub-acute 1/10 nominal LC₅₀ concentrations to determine genes differentially expressed in exposed animals as compared to control. Results were computationally analyzed with KEGG gene ontology pathway analysis to determine biological pathways affected, and with HOPACH clustering to determine similarities between different gene expression profiles. In each class of contaminant, each different exposure elicited a largely unique gene expression profile.

Further studies on characterization of the nanowires in Daphnia growth media and on nanowire uptake were conducted, and high-throughput, short-read sequencing was done on Daphnia magna RNA samples. A de novo transcriptome was assembled and used for differential gene expression analysis of sequenced RNA from control and silver nanowire-exposed daphnids. The silver nanowires investigated in this study were long and short polyvinylpyrrolidone-coated silver nanowires (PVP AgNW) and long and short silica-coated silver nanowires (SiO₂ AgNW). Ionic silver (Ag⁺) was also tested.

Silver nanowires were not as toxic as ionic silver, and toxicity varied as a function of size and of coating and in different Daphnia growth media. Toxicity could not be attributed to the concentration of dissolved silver in media, but might be caused by internal silver dosing and by a nanowire-specific effect, perhaps related to nanowire shape. The short SiO₂ coated AgNWs caused the most similar response to Ag⁺, and were also the most toxic of the nanowires. The long PVP AgNWs were the least toxic material under most conditions, and caused the most unique gene expression profile. This is the first study to show what appears to be uptake and in vivo modification of nanowires into an animal and the first to show toxic effects of a silver nanomaterial both caused by and independent of silver.

De novo assembly of short-read sequencing data resulted in a robust Daphnia magna transcriptome. Over 101,000 unique transcripts were identified, which illustrated the power of new sequencing technologies and computation algorithms to identify alternate splicing of RNA transcripts. The generated transcriptome was used as a scaffold on which short-read sequencing data were aligned to analyze differential gene expression in control versus AgNW-exposed animals. Preliminary results showed that gene expression analysis with sequencing data resulted in different numbers of differentially expressed genes than were determined with microarray techniques. A combination of four align-and-count methods were subsequently used, which all resulted in different sets of differentially expressed genes. Further work on confirming which method is best for this dataset will need to be done before the sequencing results can be further compared to microarray results. To date, it is unclear whether gene expression studies using sequencing data are more robust or informative than studies with microarray data. However, the assembled transcriptome represents a significant Daphnia magna genomic contribution.

The chemical flame-retardant nominal LC₅₀ values ranged from 58 μg/L (pentaBDE) to 3.96 mg/L (octaBDE). These chemicals are not very soluble in water, so the LC₅₀ values only represent the amount of chemical added to the exposure system, not the amount of chemical dissolved in water or the amount of chemical available to the animals. These nominal LC₅₀ values likely underestimate toxicity of FRs to Daphnia magna. Chemical flame-retardants tested were Firemaster® 550 (FM550), Firemaster® BZ54 (BZ54), pentabrominated diphenyl ether (pentaBDE), octabrominated diphenyl ether (octaBDE), triphenyl phosphate (TPP), tetrabromobenzoate (TBPH). The un-brominated analog to TBPH, phthalate di(2-ethylhexyl) phthalate (DEHP), was also tested. Gene expression profiles had little similarity between chemicals, indicating potentially unique biological effects instead of general toxicity, such as by narcosis. Additional lipidomic and metabolomic studies with FM550 and pentaBDE at 1/10 LC₅₀ showed distinct differences between effects of the two compounds. Genomic results indicate that FM550 exposure may affect transcription and translation of mRNA in Daphnia magna, but further studies are warranted to determine exact mode of toxicity.

Both chemical flame-retardants and silver nanowires are highly toxic to Daphnia magna and cause unique molecular responses. However, exposure concentrations used in this study are typically higher than concentrations detected in environmental samples. Further studies are needed at lower concentrations to confirm the lower limits of effect, and to further investigate specific modes of toxicity. Distinct biological processes were affected by exposure to each chemical and by exposure to different concentrations of one chemical, FM550. This work was limited by lack of annotation of the Daphnia genome and by limited characterization of the Daphnia proteome. Molecular studies in Daphnia were limited, as Daphnia cell lines do not exist. This work contributes, however, to basic knowledge about acute toxicity and molecular effects of environmental contaminants.

DISSERTATION ABSTRACT

The adverse outcome pathway (AOP) approach combines patterns of molecular events across levels of biological organization and complex endpoints such as reproduction, behavior, and toxicity, and provides regulatory decision criteria for Environmental Protection Agency (EPA). Useful empirical data for AOPs is derived from key indicators of toxicant exposure such as conserved transcriptional factors, nuclear receptors and their targets. One approach to establish changes in toxicant response is gene expression profiling, or ‘omics’ tools, which can lead to improved mechanistic models of toxicity in closely related species (fathead minnow and zebrafish), and divergent species (fish and crustaceans). Gene expression patterns have been used to characterize toxicant-dependent signaling for crustaceans exposed to metals, flame retardants, and narcotic toxicants. Gene expression studies can identify new assay development needs, and generate new hypotheses about conserved molecular pathways leading to direct evidence from knockout models.

In the first study, we focus on conserved nuclear receptors in the model crustacean Daphnia magna exposed to endocrine disrupting chemicals (EDCs). The Daphnia reproductive X-Y axis model involves key orthologous genes which drive metabolism and endocrine-toxicity. Previous work in Daphnia hypothesized a unified mechanism of endocrine disruption that leads to inappropriate production of male offspring through effects on a sex-differentiation gene (dsx-1) in the methyl farnesoate (MF) pathway. We carried out a microarray gene expression study with eight different EDCs in female Daphnia and found two distinct patterns of transcriptional response induced by male-inducing chemicals. Pyriproxifen and methoxychlor produced similar gene expression patterns overall and in key endocrine response genes, while methyl farnesoate and arochlor exposure shared a distinct response pattern. In particular, there were inverse responses at the ecdysteroid receptor (EcR), a PAS-binding protein that regulates key steroid biosynthesis metabolism between the two sets. We also noted differences in key molting genes (CYP450s) from the ecdysteroid pathway. Together our study demonstrates a dsx-1 dependent mechanism for male-differentiation (dsx-1) in the ovary, and a dsx-1 independent EDC mechanism for altered reproductive mode.

In the next study, we assessed Daphnia magna exposure to polycyclic aromatic hydrocarbons (PAHs) and pyrethroids, which may lead to neurotoxicity. Since PAH exposure activates the conserved AHR pathway in vertebrates, we tested the hypothesis that PAH exposure in Daphnia would induce AHR-related gene expression endpoints (cytochrome P450s, oxidative stress, metabolism). Despite the lack of a clear orthologous AhR gene, we observed induction of antioxidant pathways and CYP450-related oxidoreductase enzymes that suggest similar crustacean and vertebrate mechanisms of PAH toxicity.

To link molecular endpoints to an ecological phenotype, we used an automated video-tracking system to assess Daphnia neuro-behavioral endpoints including swimming speed and tank location preference. Altered Daphnia magna swim behaviors were upward (top of tank) and downward (bottom of the tank), similar to phototactic behaviors recognized in zebrafish anxiety-related research. Pyrethroids reduced swim speed and induced a bottom swimming phenotype at doses well below acute toxicity. Swim responses differed within the PAH chemical class. Daphnia behaviors in light and dark environments are dependent upon circadian clock mechanisms, which may be sensitive to toxicant exposure. Thus, we hypothesized that altered circadian clock (CLK) mechanisms may be a nexus between toxicant exposure and altered behavior. We observed that exposure to phenanthrene (PHE), pyrene, and cyfluthrin influenced cryptochrome-2 expression, an upstream regulator of Daphnia circadian clock genes. We also identified a broad range of functional genes involved in PHE-altered neurological pathways, involving dopamine and the Mbln12 gene involved in photo-receptor cell growth. The behavioral and molecular studies suggest that these chemicals disrupt neurologic function which results in altered behavioral outcomes.

In the final study, we used a zebrafish developmental and behavioral chemical toxicity screening assay to assess the toxicity of a candidate biofuel (2,5-DMF) and related furan compounds. Because of structural similarity to aromatic hydrocarbons, we hypothesized that 2,5-DMF targets the AhR signaling pathway and produces characteristic zebrafish developmental toxicity. AhR agonists such as 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and beta-napthoflavone (BNF) cause characteristic pericardial edema and Cyp1a1 activation in zebrafish. However, we found that 2,5-DMF toxicity, as well as some polyaromatic hydrocarbons, did not produce morphological defects. Non-AhR mechanisms that include hypoxia and neuro-behavioral effects were observed. We assessed AHR-dependent mechanisms using morphological endpoints, AHR-XRE cell-based reporter assays, and in silico AhR-LBD binding models. Zebrafish were also assessed for embryonic and larval photomotor responses. We evaluated several furan derivatives and found that 2-ethyl furan and 2,3-dihydrofuran had the least biological effect in our assays. The highest aquatic risk to developing fish was from 2,5-dimethyl furan, and 2-methyl furan. Overall, we determined that zebrafish exposure to biofuel candidates did not induce AhR mediated toxicity, but produced significant behavioral toxicity from non-AhR mechanisms.

Despite their ubiquity in modern life, industrial chemicals are poorly regulated in the United States. Statutory law defines industrial chemicals as chemicals that are not foods, drugs, cosmetics, nor pesticides, but may be used in consumer products, and this distinction places them under the purview of the Toxic Substances Control Act (TSCA), which received a substantial update when the US congress passed a revision of the act in 2016. The revised law, the Frank R. Lautenberg Chemical Safety for the 21st Century Act addresses many but not all of TSCA’s failings, and rightfully emphasizes the development and adoption of high throughput screens, in vitro, and alternative assays to improve the process for registering new chemicals and to address the tens of thousands of untested chemicals currently in the TSCA inventory. As the discipline of toxicology gradually shifts from its’ history as a reactive science (responding to problems after they’ve occurred) to a proactive science (attempting to predict and circumvent dangers to human and environmental health), two things become clear: 1.) traditional low throughput toxicological testing methodologies are inadequate to address both the volume of chemicals of interest and the pace of research; and 2.) the modern industrial chemical ecosystem is complex and no single testing solution will be appropriate for all the actors that populate that ecosystem.

To address these challenges, three interventions are proposed, each of which targets a different population within the industrial chemical ecosystem. The first intervention is a suite of computational toxicology methods targeted towards chemists in the initial phases of chemical design and development. The second intervention is an alternative assay, the yeast functional toxicogenomic assay, permits industrial or government labs to rapidly investigate differences in cytotoxic mechanism between different chemicals – even if they are structurally very similar. The third intervention explored in this work is a method for enhancing the metabolic capacity of cell lines currently used by regulators for high throughput cytotoxicity testing. These interventions individually are not necessarily appropriate for all actors across the US industrial chemicals ecosystem, but as bespoke solutions they may be quite useful, and, it is hoped, support a larger exploration of where and how similar efforts may be spent most effectively to reduce industrial chemical hazard.

Background

Objective

Methods

Results

Conclusions

Background

Objectives

Methods

Results

Conclusions