Developmental toxicity of alkylated PAHs and substituted phenanthrenes: Structural nuances drive diverse toxicity and AHR activation

Abstract

Polycyclic aromatic hydrocarbons (PAHs) are a diverse class of chemicals that occur in complex mixtures including parent and substituted PAHs. To understand the hazard posed by complex environmental PAH mixtures, we must first understand the structural drivers of activity and mode of action of individual PAHs. Understanding the toxicity of alkylated PAHs is important as they often occur in higher abundance in environmental matrices and can be more biologically active than their parent compounds. 104 alkylated PAHs were screened from 11 different parent compounds with emphasis on substituted phenanthrenes and their structurally dependent toxicity differences. Using a high-throughput early life stage zebrafish assay, embryos were exposed to concentrations between 0.1 and 100 μM and assessed for morphological and behavioral outcomes. The aryl hydrocarbon receptor (AHR) is often implicated in the toxicity of PAHs and the induction of cytochrome P4501A (cyp1a) is an excellent biomarker of Ahr activation. Embryos were evaluated for cyp1a induction using a fluorescence reporter line. Alkyl and polar phenanthrene derivatives were further assessed for spatial cyp1a expression and Ahr dependence of morphological effects. In the alkyl PAH screen 35 (33.7%) elicited a morphological or behavioral response and of those 23 (65%) also induced cyp1a. 31 (29.8%) of the chemicals only induced cyp1a. Toxicity varied substantially in response to substitution location, the amount of ring substitutions and alkyl chain length. Cyp1a induction varied by parent compound group and was a poor indicator of morphological or behavioral outcomes. Polar phenanthrenes were more biologically active than alkylated phenanthrene derivatives and their toxicity was not dependent upon the Ahr2, Ahr1a or Ahr1b when tested individually, despite cyp1a induction by 50% of polar phenanthrenes. Our results demonstrated that induction of cyp1a did not always correlate with PAH toxicity or Ahr dependence and that the type and location of phenanthrene substitution determined potency.

GRAPHICAL ABSTRACT

1. Introduction

Polycyclic aromatic hydrocarbons (PAHs) are a diverse class of chemicals defined by the presence of two or more fused aromatic rings. Human exposure to PAHs is widespread and largely attributed to inhalation of atmospheric PAHs, smoking, cooking fumes and ingestion of food containing PAHs (Abdel-Shafy and Mansour, 2016; Mallah et al., 2022). PAH exposure in humans has been linked to cancer, cardiovascular disease, and lung impairment with the majority of research focused on carcinogenicity (Mallah et al., 2022; Sun et al., 2021; Blumer et al., 1977; Mastrangelo et al., 1996; Polycyclic Aromatic Hydrocarbons (PAHs) Factsheet). PAHs released into the atmosphere through anthropogenic combustion activities can be deposited to soil, sediment, marine, and freshwater environments through dry and wet deposition or oil spills and road runoff (Wallace et al., 2020; Krzyszczak and Czech, 2021). PAHs have been detected in a variety of aquatic, terrestrial and marine species in the environment, with evidence that they bioaccumulate over time (Wallace et al., 2020). Experimental exposure of developing fish to PAHs can induce mortality, pericardial edema and other cardiotoxicity endpoints (Incardona et al., 2006; Scott et al., 2011; Donald et al., 2023; Geier et al., 2018a). A variety of PAHs also induce abnormal behaviors in fish resulting from sub-teratogenic exposures (Geier et al., 2018a; Shankar et al., 2019, 2022).

Desipite the large number of different PAH structures detected in the environment, the vast majority of research and monitoring has focused primarily on 16 “priority PAHs”, established by the US Environmental Protection Agency (EPA) in the 1970s, all of which were unsubstituted parent chemicals (Keith, 2015; Andersson and Achten, 2015; da Silva Junior et al., 2021). Although many have called for expansion of this list, most studies still focus on the original 16 “priority PAHs” (Keith, 2015; Andersson and Achten, 2015; da Silva Junior et al., 2021; Zhang et al., 2022). Studies have shown that parent PAHs are often less abundant in the environment than their substituted counterparts, and less toxic (Geier et al., 2018a; Qiao et al., 2017, 2020; Peng et al., 2023). Some substituted PAHs can be formed through abiotic and biotic degradation. For example, bio-remediation processes often rely on oxidation reactions, resulting in the creation of oxygenated metabolites more toxic than the parent PAHs (Huizenga and Semprini, 2023; Cooper et al., 2010; Huizenga et al., 2024; Wang et al., 2021). Alkyl substituted PAHs are not commonly degradation products but instead occur as co-contaminants with parent PAHs often in higher abundance (Moradi et al., 2022; Wnorowski et al., 2022; Golzadeh et al., 2021). Focusing current PAH monitoring and research efforts on the priority PAH list significantly underestimates total PAH contamination and the true hazard.

The canonical mechanism of PAH toxicity, especially in the context of human health, is based on our understanding of the 16 “priority PAHs”, and the most well studied of these, benzo(a)pyrene (BaP) (Mallah et al., 2022; Mastrangelo et al., 1996; da Silva Junior et al., 2021). Carcinogenicity and mutagenicity have been the most well studied endpoints for PAH toxicity (Blumer et al., 1977; Mastrangelo et al., 1996; da Silva Junior et al., 2021). The canonical mechanism for these endpoints relies on aryl hydrocarbon receptor (Ahr) agonism which induces cytochrome P450 (Cyp) enzymes and the production of toxic metabolites; an adaptive response gone awry (da Silva Junior et al., 2021; Nebert et al., 2004). As we learn more about variability in the toxicity and mechanisms amongst PAHs, it is increasingly clear that not all PAHs act through the same pathways (Scott et al., 2011; Incardona et al., 2005). Thus, it is not sufficiently descriptive to group all PAHs together for regulatory purposes. Instead, the Frank R. Lautenberg Chemical Safety for the 21st Century Act (2016) prioritizes chemical classification based on mode-of-action (MOA), defined by whole organism, cellular, or organ response or receptor level mechanism, among other toxicity testing modernization actions (Barron et al., 2015; Kienzler et al., 2019; Krewski et al., 2010; US EPA, 2016). In alignment with these priorities, it is important to understand the toxicity and MOAs of the individual components of PAH mixtures, with greater emphasis placed on substituted derivatives.

To this end, previous work from our lab has covered zebrafish developmental toxicity screening of over 200 substituted PAHs, including hydroxy-, quinone-, alkyl-, nitro-, oxy-, and amino-PAHs (Geier et al., 2018a; Shankar et al., 2019; Knecht et al., 2013). The present study expands upon this work to include 104 alkylated PAH derivatives from 11 different parent chemicals with emphasis on substituted phenanthrenes and their structurally dependent toxicity differences. Understanding the toxicity of alkylated PAHs is of particular interest as they often dominate PAH contaminated sites across various matrices and are predicted to be more persistent (Wnorowski et al., 2022; An et al., 2022; Wassenaar and Verbruggen, 2021; Saha et al., 2012). In this study, we focused further on alkyl phenanthrenes to identify potential differences in modes of action by examining Ahr dependence and determining spatial cytochrome P4501A (cyp1a) induction profiles of each phenanthrene derivative. To our knowledge, this is the largest in vivo toxicity and cyp1a induction screen of alkylated PAHs. Building our knowledge of individual PAH activity will move the field closer to understanding the relevant structure activity relationships and inform hazard-based chemical prioritization.

2. Methods

2.1. Chemicals

Analytical grade standards were verified by the Chemical Mixtures Core for the Oregon State University Superfund Center. Verified standards were dissolved in dimethyl sulfoxide (DMSO) if obtained neat or solvent exchanged to DMSO. Based on commercial availability and PAH solubility, the prepared stock solutions varied in concentration from 0.18 to 10 mM. A detailed chemical list with CAS, purity, and vendors is provided in Supplemental Information1-Table 1 (SI1-T1).

2.2. Zebrafish (Danio rerio) husbandry

Adult wildtype tropical 5D fish, cyp1a reporter fish(Tg(cyp1a:nlsegfp)) (Kim et al., 2013), and ahr2^hu3335, ahr1a^osu6 and ahr1b^wh39 mutants used in this study were maintained under Institutional Animal Care and Use Committee protocol 2021–0227 at the Oregon State University Sinnhuber Aquatic Research Laboratory (SARL, Corvallis OR). Fish were kept in densities of ~500 fish/35-gallon tank (5D) and 40 fish/6-L tank (mutant lines) at 28 ^◦C with a 14 h:10-h light:dark cycles, tank water was recirculated, filtered water was supplemented with Instant Ocean salts. Adult fish were fed twice a day with Sparos Zebrafeed 300 μm. Spawning initiated upon first light in the morning. The resulting embryos were collected, screened for malformed or unfertilized individuals, and selected for the same developmental stage (Kimmel et al., 1995). At 4 h post fertilization (hpf), embryos were enzymatically dechorionated and at 6 hpf, were singulated into Falcon (Corning) 96-well round bottom tissue culture treated plates prefilled with 100 μL E2 embryo medium using our automated embryo placement systems (Mandrell et al., 2012).

2.3. Toxicity testing

At 6 hpf, embryos were statically exposed to chemical stocks dispensed into the 96 well plates using a Hewlett Packard D300e digital dispenser (Truong et al., 2016). Wells were normalized to 1% DMSO and gently shaken while dispensing using custom parameters. Plates were sealed with ThermalSeal RTS (Excel Scientific) to prevent evaporation. Plates were kept for 18 h on a shaker at 225 revolutions per minute at 28 ^◦C in the dark. At 28 hpf, embryos were evaluated for embryonic photomotor response (section 2.6.1). For the remainder of the exposure, plates were kept at 28 ^◦C in the dark without shaking, until 120 hpf when larval photomotor response and morphology screening were conducted (sections 2.6.2, respectively and 2.5). The solutions were not exchanged over the course of the experiment. 96-well plates were labeled with a unique barcode and all data collected for the following endpoints were captured with a custom laboratory information management system (Zebrafish Acquisition and Analysis Program) (Truong et al., 2016).

2.4. Exposure concentrations

Each chemical’s concentration was chosen following an initial range finding (data not shown). Each chemical was tested at 12 concentrations in a linear curve from 0.05 to 100 μM, treatment n = 14 and control n = 24, with two replicate plates. The maximum test concentration was set to 1% of the stock concentration as developing zebrafish can only tolerate up to 1% DMSO. A total of 8 polar phenanthrene derivatives were tested using the described protocol in a logarithmic curve from 0.1 to 100 μM to maximize data in the linear region of the modeled curve (Geier et al., 2018a). Full details on concentrations used for all chemicals can be found in SI1-T2. All reported concentrations are nominal water concentrations.

2.5. Morphological assessments

At 28 hpf, embryos were assessed for mortality and spontaneous movement. At 120 hpf, embryos were assessed for 10 morphological endpoints: mortality, craniofacial, bent axis, edemas, muscular/cardiovascular, lower trunk/caudal fin, brain, skin/pigmentation, wavy notochord and lack of touch response. These endpoints are more thoroughly described in the supplemental material (SI1-T3). Each endpoint was scored as binary presence/absence data and scores were compiled into a summary endpoint: “any effect”, the cumulative incidence of all endpoints per animal. The “any effect” endpoint was used to calculate the lowest effect level (LEL) and the benchmark dose of 50% (BMD50), 50% change relative to the background. The LEL is the lowest concentration for which the percentage affected exceeded a significance threshold over control (p < 0.05). Significance was calculated using a binomial test as previously described in Truong et al. (2014). LEL values are reported in SI1-T4, but not used for further analysis. BMD50 values were calculated as previously described in Truong et al. (2022). Briefly, an unrestricted 3-parameter log-logistic model for dichotomous data with “extra risk” was fit following guidelines described in the EPA BMDS v3.2 manual (US Environmental Protection Agency (EPA), 2022). These curves were used to calculate the reported BMD50 values (SI1-T5). Raw binary scores per animal are available in Supplemental Information 3 (SI3).

2.6. Behavioral assessments

2.6.1. Embryonic photomotor response

Embryos were kept in the dark before the embryonic photomotor response (EPR) assay was conducted at 28 hpf. During this developmental stage, zebrafish exhibit spontaneous contralateral tail contractions as the muscles in the tail are innervated (Kimmel et al., 1995). This movement is initiated by light detected by photoreceptors in the hind brain (Kokel et al., 2013).

This assay was previously described in Reif et al. (2016). Embryo movements were recorded throughout the assay, 30 s (sec) of background, a 1 s pulse of visible light, 9 s of dark, a second 1 s pulse of light, 10 s of dark. The videos were analyzed using a custom C# program to compute the movement index for each frame by pixel ratioing between consecutive frames. Statistical analysis was performed using the 9 s before the first pulse (background), 8 s following the first pulse of light (excitatory), 7 s following the second pulse of light (refractory). The average movement values over these intervals were used as the response values in the behavioral data analysis (section 2.6.3).

2.6.2. Larval photomotor response

The larval photomotor response (LPR) assay was conducted at 120 hpf using a light-dark cycle and the video tracking software of the ZebraBox (ViewPoint Life Sciences, Montreal, Canada) behavior chamber. The 24-min assay consisted of 4 light/dark cycles of 3 min light, 3 min dark. Larval movement was tracked from 6 to 24 min using ZebraLab motion analysis software version 5.29.0.70. The area under the curve from the second light/dark cycle was calculated for the light and dark periods and used as the response values in the behavioral data analysis (section 2.6.3).

2.6.3. Behavioral data analysis

Response values were used to calculate both LEL and BMD50 values. To calculate LEL, response values were compared to the control using a Kolmogorov-Smirnov test (Bonferroni-corrected p value threshold = 0.05), EPR and LPR LEL results are reported in SI1-T6 and T7, respectively. To calculate BMD50 values, curves were fit following methods described in Thunga et al. (2021). Wells with observed morbidity or mortality were excluded from the analysis. Any treatment with >30% of the wells excluded was removed. LPR curves could not be generated for 8-methylbenz(a)pyrene and 7,12-dimethylbenz(a)anthracene due to insufficient morphologically normal larvae, these are indicated with NA for LPR in the following figures. Hyperactive and hypoactive curve fitting was done as reported in Truong et al. (2022). BMD50 values reported in the figures do not indicate whether the curve was hypo/-hyperactive. This information is available in SI1-T8 and -T9 for EPR and LPR, respectively, raw movement values are available in SI3. Manual quality control of the curves was conducted by visual inspection, and modeled BMD50 values were removed based on high variability between groups especially when they overlapped with control values, data included in Supplemental Information includes these removed values.

2.7. cyp1a reporter line expression

The Ahr is often implicated in the toxicity of PAHs and the induction of cytochrome P4501A (cyp1a) is highly responsive to Ahr agonism. The cyp1a reporter line expresses enhanced green fluorescence protein (EGFP) via the cyp1a promoter (Kim et al., 2013). Reporter fish were exposed to the concentration just below the LEL to maximize the probability of having enough morphologically normal fish for imaging. If no LEL was calculated, then cyp1a reporter fish were exposed to the highest concentration tested. If exposure at these concentrations resulted in greater than 6 malformed or dead fish, the next concentration down was used until there were enough fish for imaging (N = 12). Final concentrations used are included in SI1-T1. Six of the twelve fish exposed were imaged using the Keyence BZ-X710 imager with a GFP filter at 10X magnification at 1/3s, high sensitivity, and 20% excitation light. Images were compared to daily controls and a binary presence or absence of cyp1a induction for each chemical was noted. Representative images are included in Supplemental Information 2(SI2) Fig. 2.

2.8. Phenanthrene parent group

2.8.1. Chemical list

A total of 24 chemicals included in previous screens from the phenanthrene parent group (including alkyl substituted phenanthrenes, phenanthrene, and polar substituted phenanthrenes) underwent further analysis (Geier et al., 2018a; Shankar et al., 2019). To foster comparisons, chemicals in the phenanthrene parent group from the current screen were only pursued when tested up to 90 μM or if they elicited an abnormal morphology or behavior. Chemicals included in this list are indicated in SI1-T1.

2.8.2. Spatial cyp1a expression

Confocal images were acquired to localize cyp1a expression. Reporter fish were exposed to concentrations that maximized the probability of having enough morphologically normal fish for imaging (section 2.7). Six of the twelve fish exposed were imaged in 2 orientations: with their head facing right or left (3 per). Confocal images were acquired using the Oxford Andor BC43 at 20X using a 529 nm laser set at 10% and 100ms exposure over a Z-stack set using auto-step size for the width of the larvae at its widest point. Each chemical was then screened for the presence/absence of cyp1a expression in 5 locations (liver, vasculature, gall bladder, intestine, and myoseptum). Some locations were only visible from one side of the fish (liver, left, and gallbladder, right) example images are available in SI2 Fig. 3.

2.8.3. ahr1a^osu6 line development and validation

Using CRISPR-Cas9 targeting of exon 2 of the ahr1a gene, we generated founder embryos with a 4-basepair deletion in the Helix-loop-helix DNA-binding domain, which results in a frameshift mutation at amino acid (AA) 33 and a premature stop codon at AA39. The deletion was confirmed using Sanger sequencing and a TaqMan SNP genotype assay was developed to confirm homozygous individuals. Details for line generation and deletion confirmation can be found in SI2 Section 1. Validation of the ahr1a^osu6 line was done by exposure of wildtype 5D and ahr1a^osu6 lines to Ahr1a agonists followed by immunohistochemistry (IHC) for cyp1a at 120 hpf. IHC methods are reported elsewhere (Mathew et al., 2006). Briefly, at 4 hpf dechorionated 5D and ahr1a^osu6 embryos were exposed to PAH concentrations expected to induce cyp1a expression in the liver with minimal morphological effects, (Xanthone 20 μM, 9-methylphenanthrene 100 μM, and 5-nitronaphthene 10 μM; N = 16). 1% DMSO was used as the vehicle control. At 120 hpf, 8 larvae without morphological effects were euthanized using tricaine and fixed overnight in 4% paraformaldehyde (PFA) at 4 ^◦C. Embryos were permeabilized in 0.005% trypsin for 10 min on ice, followed by a rinse with phosphate buffered saline +0.1% Tween 20 (PBST) and subsequent fixation in 4% PFA for 10 min. Embryos were blocked for 1 h at room temperature with 10% normal goat serum (NGS) in PBS + 0.5% Triton X-100, and incubated overnight with the primary antibody mouse α fish Cyp1a monoclonal antibody (Biosense Laboratories Bergen, Norway) (1:500) in 1% NGS. The acetylated tubulin antibody (Sigma T6793) (1:4000) in 1% NGS was used as a positive control. Lastly, embryos were washed in PBST and incubated for 2 h in secondary antibody (Alexa-Fluor 594 goat anti-mouse IgG) before imaging using the Oxford Andor BC43 confocal microscope at 20X using a GFP laser at 10% and 100ms per frame.

2.8.4. ahr2^hu3335, ahr1a^osu6 and ahr1b^wh36 mutant line exposures

Ahr mutant lines were used to investigate the Ahr dependence of toxicity. Due to gene duplication and loss events in the evolution of teleost fish, they typically have multiple Ahrs while mammals, including humans, usually have one (Postlethwait et al., 2004; Shankar et al., 2020). Zebrafish have 3 co-orthologs, Ahr1a, Ahr1b and Ahr2. Exposures using knock out lines for each of these genes were used to determine the individual Ahr dependence of morphological effects observed in 5D fish.

ahr2^hu3335 fish express a truncated, non-functional Ahr2 protein (Goodale et al., 2012). ahr1b^wh36, the Ahr1b knockout line expresses a truncated non-functional Ahr1b protein (Karchner et al., 2017). 5D, ahr2^hu3335, ahr1a^osu6 and ahr1b^wh36 lines were exposed to substituted phenanthrenes with a morphological effect (determined by a calculated BMD50), at the lowest concentration eliciting a 100% effect in the “any effect” endpoint, n = 16. Additionally, ahr2^hu3335 and 5D embryos were exposed to 7 concentrations of 1-methyl-7-isopropylphenanthrene (retene) (1, 5, 20, 35, 50, 65, 100 μM), n = 24.

2.9. Analysis

Figures were generated using R version 4.3.1 (R Core Team, 2023). Heatmaps were generated using the pheatmap package (version 1.0.12) (Kolde, 2019). Clustering in Fig. 3 was calculated using the clustering method “complete”. Chemical fingerprints, which are binary strings used to describe the structure of a chemical in a machine-readable manner, were used to generate the principal component analysis (PCA) in Fig. 4. Open babel was used to convert chemical SMILES into a mol file. Using R, standard fingerprints were generated using the rcdk package (Voicu et al., 2020) and features with near zero variance between chemicals were trimmed using the nearZeroVar() function in the caret package (Kuhn, 2008). This data was used as the input for the base R function prcomp() and principal components 1 and 2 are show in Fig. 4.

Fig. 3. Phenotypic effect clustering of polar and alkyl substituted phenanthrenes.

Heatmap showing BMD50 concentrations for effects in morphology, EPR, and LPR in red. Yellow indicates the cyp1a reporter fish had expression in the indicated location. The heatmap is clustered based on phenotypic effects. Polar phenanthrenes are indicated with bolded names.

Fig. 4. Chemical similarity clustering of substituted phenanthrenes.

PCA plot displaying chemical fingerprint information for each chemical. The closer chemicals are to one another, the more similar they are in structure. Chemical labels are colored based on their morphology BMD50 values.

3. Results and discussion

3.1. Summary of alkylated PAH activity

The developmental toxicity of 104 PAHs was assessed in the embryonic zebrafish model. These chemicals covered 11 parent PAH groups, with phenanthrene and naphthalenes making up 49% of the chemicals the breakdown of chemicals in each parent group can be find in SI2 Table2. 38 (36.5%) chemicals did not cause morphological or behavioral effects, nor did they induce cyp1a (Fig. 1). 63.5% of the chemicals had some activity in the screening measures and 35 (33.7%) resulted in an adverse outcome. Of the 35 chemicals that elicited morphology or behavioral effects, 23 (65.7%) had cyp1a induction at 120 hpf, demonstrating that cyp1a induction alone is insufficient as a biomarker of alkyl PAH biological activity. Cyp1a expression was the least discriminatory of the measures, with 31 (29.8%) chemicals inducing cyp1a expression without any detectable adverse outcome, over half (57%) of all the chemicals that induced cyp1a. If canonical Ahr activation followed by cyp1a induction led to the creation of toxic metabolites, we would expect to see a much stronger correlation than we did between cyp1a induction and abnormal phenotypes. We note that, of the 69 chemicals that elicited no abnormal outcomes, 52 (75.4%) had maximum achievable test concentrations below 6 μM or were from the naphthalene group which rapidly volatizes from the aqueous test medium (see section 3.2.1). It is possible that we were unable to detect bioactivity because of standard concentration availability or volatilization from our aqueous system.

Fig. 1. Binning of zebrafish response after exposure to 104 PAHs.

Venn diagram for the number of chemicals which had an effect in each of the endpoints or had no effect in any. Cyp1a induction indicates a chemical induced expression in the reporter line, morphology effect indicates a chemical had a BMD50 reported for the chemical in “any effect”, behavior effect indicates a chemical had a BMD50 reported in either LPR, EPR, or both, and no effect had no effect in any of the endpoints.

3.2. Comparing the bioactivity profiles of 67 alkylated PAHs by parent group

The heatmap in Fig. 2 and subsequent discussion includes chemicals with a maximum tested concentration of 90–100 μM or that caused an adverse outcome, 67 of the 104 tested chemicals. This approach was used to exclude chemicals with low tested concentrations, which made it difficult to determine their biological activity. The heatmap displays the BMD50 value for the “any effect” endpoint, EPR and LPR. Cyp1a expression and parent group are annotated along the left. Within the parent groups chemicals are organized by molecular weight from lowest to highest (Fig. 2). Discussion is grouped by parent group to limit comparisons to structurally similar chemicals, each discussion includes a brief review and comparison of our findings with the existing literature and previous testing in our system. It is important to note that many of the cited studies focused on carcinogenicity, genotoxicity or mutagenicity endpoints, especially through metabolic activation, oxidative stress and DNA adduct formation. The endpoints tested in this study were distinct from these and not directly comparable. However, it is possible for adverse developmental outcomes to result through these mechanisms and it is valuable to understand differences and similarities in their structure-activity relationships (Zou et al., 2024).

Fig. 2. Parent chemical grouping of biological activity of alkyl substituted PAHs.

Heatmap displaying BMD50 values for EPR, LPR, and Any Effect endpoints for all alkyl chemicals tested. darker red indicates a lower BMD50 value, white indicates no BMD50 value was calculated, and grey indicates that there were less than four concentrations with ≥70% unaffected larva, it was not possible to calculate a BMD50 value for the behavior endpoint. Whether or not the chemical induced cyp1a expression in the cyp1a reporter line is noted with black (yes induction) and grey (no induction) annotations. Chemicals are ordered by parent grouping in order of molecular weight lowest to highest as indicated by the colored annotation. Within parent groupings chemicals are ordered by number of additional carbons. The maximum test concentration for each chemical is shown in the bar plot.

3.2.1. Naphthalene

17 alkyl naphthalenes were tested. None elicited morphological effects, however, 2,3-dimethylnaphthalene and 2-ispropylnaphthalene caused abnormal EPR and 2-isopropylnaphthalene also caused cyp1a induction. The minimal number of effects observed in the four assay endpoints in this screen can be at least partially explained by the fate of low molecular weight PAHs in our 96 well exposure system, as naphthalene is the lowest molecular weight PAH (Geier et al., 2018b). Geier et al. found that the smaller PAHs with lower log K_ow values tend to have lower concentration uptake rates, a ratio of the chemical concentration in embryo tissue at 48 hpf and the nominal media concentration (Geier et al., 2018b). The development of a more appropriate high throughput methodology for testing lower molecular weight and volatile PAHs in our system is currently ongoing.

3.2.2. Fluorene

Six alkyl fluorenes were tested. Two induced morphological and behavioral effects (1,8-DMFL and 2-EFL), one induced only morphological effect (9-MFL), and two induced only behavioral effects (1,7-DMFL and 2-MFL). Only 2-MFL, which produced an EPR effect but not a morphological effect, induced cyp1a. Similar to 2-MFL, 2-nitroFL was previously shown to induce Cyp1a expression (Geier et al., 2018a). Substitution at the 2 position may be important for fluorene Ahr agonism, however 2-EFL, the most potent morphologically active alkyl fluorene (BMD50 28.7 μM), did not induce cyp1a. Results from Geier et al. showed that fluorene elicited an abnormal LPR but was not morphologically active and did not induce Cyp1a expression (Geier et al., 2018a). Two other studies of fluorene in zebrafish showed some toxicity (Kim et al., 2020; Incardona et al., 2004). Kim et al. found a minimal increase in mortality and no cyp1a induction, while Incardona et al. found an increase in axis defects and mild paracardial edema when embryos were exposed with renewal (Kim et al., 2020; Incardona et al., 2004).

While other studies have shown some activity of fluorene in zebrafish, our study suggests that the methylated substituents of fluorene are more bioactive than the parent. Some MFLs are positive for mutagenic potential in bacterial assays where the parent is not and 1,9-DMFL was tumorigenic in newborn mice (Rice et al., 1987). Morphological/behavioral effects in the absence of cyp1a induction in many of the MFLs tested suggest that these chemicals may act through mechanisms other than canonical Ahr agonism (Coelho et al., 2022).

3.2.3. Anthracenes

Seven alkyl anthracenes were tested. The three highest molecular weight anthracenes elicited morphological effects. The two alkyl anthracenes with the lowest BMD50 for morphology also had EPR effects, but none elicited LPR effects. It is unlikely that the increase in potency at higher molecular weights was due to solubility differences as previous work showed that the aqueous solubility of alkyl anthracenes did not positively correlate with molecular weight. Rather, it was previously reported that solubility decreased with increased methylation and decreased with molecular symmetry, a measure of how many distinguishable positions a molecule can have in rigid space (Kang et al., 2016; Pinal, 2004). All but one alkyl anthracene with adverse effects, 2, 7-dimethylanthracene (2,7-DMAN), also induced cyp1a expression. Lack of cyp1a induction for 2,7-DMAN could be due to the limited maximum test concentration (20 μM). Two chemicals which were tested to 100 μM were inactive in all the endpoints (9,10-DMAN and 2-tert-butylanthracene).

The parent chemical anthracene can cause photoinduced toxicity, especially in fish, but this effect requires co-exposure to UV light (Oris et al., 1984; Bowling et al., 1983; Mujtaba et al., 2011).When previously tested in our lab (with no UV co-exposure), anthracene was not biologically active while all other substituted anthracenes were active independent of UV exposure (Geier et al., 2018a). Previous in vitro work has demonstrated that methyl anthracenes, in an isomer-specific manner, affected gap junction signaling between epithelial cells while anthracene did not (Upham et al., 1996). Our study found that most alkyl anthracenes were biologically active, with morphological effects at the highest molecular weights, and potency did not correspond to the number of alkyl groups.

3.2.4. Phenanthrene

Fourteen alkyl phenanthrenes were tested. Twelve of the fourteen induced cyp1a. Ten elicited morphological/behavioral effects, two of which did not induce cyp1a. The most common effect for alkylated phenanthrenes was EPR, with seven eliciting EPR effects, including all the singly methylated phenanthrenes. Four phenanthrenes across dimethyl, ethyl, trimethyl, and pentyl groups did not elicit abnormal morphological/behavioral phenotypes. Notably, 9-ethylphenanthrene (9-EPHE) and 9-pentylphenanthrene were inactive while 9-MPHE caused an EPR effect. Additionally, 3-MPHE and 3-EPHE had the lowest EPR BMD50 values (9.3 and 13.3 μM respectively), while 3,6-DMPHE, the only other alkyl phenanthrene with a substitution in the 3-position, was developmentally inactive. Despite the alkyl phenanthrenes having a high percentage of substituents with sub-teratogenic effects, only retene elicited morphological effects. Developmental cardiotoxicity of parent and 5 alkylated phenanthrenes was previously reported in two fish species, but only retene induced cyp1a at 48hpf (Incardona et al., 2024). The lack of concordance between our study and what has been reported may be attributed to the exposure method and timing. In this study, exposures were static, beginning at 6hpf, while the Incardona study used a passive dosing method, which maintains constant medium concentrations at maximum solubility 24–48 hpf (Incardona et al., 2024). Additionally, cyp1a expression was quantified using different methods: visible expression at 120 hpf in a cyp1a reporter line in this study and gene expression analysis at 48 hpf in the Incardona study (Incardona et al., 2024). In summary, while most alkylated phenanthrene structures induced cyp1a and elicited abnormal behavioral effects, morphological effects were only seen in one chemical (retene). We examine substituted phenanthrenes in more depth in 3.3.

3.2.5. Dibenzothiophene (DBT)

Three alkyl DBTs were tested. The DBT heterocycle was the only parent group in our study from which no member induced cyp1a expression. 4-methyldibenzothiophene (4-MDBT) was the only active substituent, eliciting abnormal morphology and EPR, BMD50 34 μM and 10.2 μM, respectively. The lack of cyp1a induction by alkyl DBT exposure here agreed with previous in vitro studies that showed minimal AHR signaling associated with alkyl DBT or DBT (Lam et al., 2018; Marvanová et al., 2023). Lam et al. found a 78-fold increase in AHR activity associated with exposure to 2,8-DMDBT compared to DBT in the H4IIE-luc assay, but we did not see this cyp1a induction associated with a different DMDBT, 4,6-DMDBT, in this in vivo study (Lam et al., 2018). Previous work highlighted in a recent review found that, generally, heterocyclic PAHs including DBT are often highly persistent and toxic (Chlebowski et al., 2017; Ghosh and Mukherji, 2023; Çelik et al., 2023). Instances of adverse outcomes with no cyp1a induction could indicate Ahr agonism that activated non-canonical pathways, perhaps by dimerizing with proteins other than aryl hydrocarbon nuclear translocating protein or interactions with receptors that do not lead to cyp1a induction (Denison and Faber, 2017; Joshi et al., 2015). Lack of cyp1a induction in our study indicates that the toxicity of 4-MDBT could be through non-canonical Ahr agonism or was Ahr-independent (Coelho et al., 2022).

3.2.6. Pyrene

1-methylpyrene was the only alkyl pyrene tested and it caused morphological and EPR effects, and induced cyp1a. Previously, pyrene was shown to impact developmental morphology and LPR, but not EPR (Geier et al., 2018a). 1-methylpyrene is a genotoxic carcinogen hypothesized to act through the formation of 1-Sulfooxymethylpyrene and subsequent DNA adduction in mammals (Rice et al., 1987; Bendadani et al., 2014; Jiang et al., 2015). Studies have not yet demonstrated a similar mode of action in zebrafish.

3.2.7. Benz(a)anthracene (BAA)

Four alkyl BAAs were tested. Only 7,12-dimethylbenz(a)anthracene (7,12-DMBAA) caused morphological effects. It had a morphological BMD50 of 3.5 μM, and a EPR BMD50 of 18.9 μM. It was not possible to calculate an LPR BMD50 due to the high incidence of morphological effects at 120 hpf. 7,12-DMBAA is a well-characterized carcinogen linked to various cancer endpoints in vivo and in vitro (da Silva Junior et al., 2021). Studies suggest these effects occur through metabolic activation and DNA adduct formation dependent upon AHR and Cyp activity (RamaKrishna et al., 1992; Kinoshita and Gelboin, 1972). In fish, its toxicity is Ahr2 dependent but not Cyp1a dependent, similar to the mode of action of TCDD (Incardona et al., 2006; da Silva Junior et al., 2021; RamaKrishna et al., 1992; Huberman et al., 1979). 7, 12-DMBAA, 3-MBAA and 3-methylcholanthrene induced cyp1a. 3-MBAA did not cause any morphological effects but was only tested to 2.6 μM and induced cyp1a and an EPR effect. EPR is known to strongly correlate with morphological effects later in zebrafish development, suggesting that 3-MBAA may impact morphology at a later life stage or a higher concentration than we could achieve (Reif et al., 2016). 1-MBAA did not cause morphological/behavioral effects or induce cyp1a expression. Notably, 8,9,11-TMBAA, which was excluded from this analysis, was tested up to 50 μM and did not elicit morphological/behavioral effects but did induce cyp1a. Although it may cause other effects at higher concentrations than those tested, 8,9,11-TMBAA is less potent than 7,12-DMBAA and 3-MBAA.

A previous study showed similar morphological impacts of 7,12-MBAA, unsubstituted BAA and single-methylated BAA chemicals (8-MBAA and 4-MBAA) in zebrafish using yolk sac microinjection methodology (Dubiel et al., 2022). Previous work in our lab has shown some morphological effects of exposure to unsubstituted BAA, but this and the singly methylated BAAs included in the present study had substantially lower morphological potency than 7,12-DMBAA in our exposure system (Geier et al., 2018b). Another study found that 4-, 8-, and 9-MBAA induced morphological effects, but 9-MBAA was significantly less potent and did not cause mortality (Fang et al., 2022). Additionally, 4-, 8-, and 9-MBAA were able to activate the Ahr, treatment with an Ahr antagonist did not change the toxicity of 8- or 9-MBAA but decreased the toxicity of 4-MBAA (Fang et al., 2023). This suggests that the dependence of toxicity of singly methylated BAAs varies based on methylation position. Further studies are required to determine if the activity of 7, 12-DMBAA and 3-MBAA are acting in an Ahr and Cyp1a dependent manner in our system.

Our results are consistent with previous studies demonstrating that differential alkyl substitution greatly impacts the toxicity of BAAs, increasing or decreasing potency relative to parent BAA, and that while it was the most potent included in this study, 7,12-DMBAA may not be the only methylated BAA that induces morphological effects.

3.2.8. Benzo(c)phenanthrene (BCP)

Six alkyl BCPs were tested. 2- and 3-methylbenzo(c)phenanthrene (MBCP) elicited morphological effects. 3-MBCP was more potent with a BMD50 of 18.8 μM vs. 41.2 μM for 2-MBCP. The other two singly methylated BCPs, 4- and 5-MBCP, did not elicit morphological/behavioral effects. However, 4-MBCP was the only BCP to induce cyp1a expression. 1,12-DMBCP was the only alkyl BCP with a behavioral effect and had a BMD50 value of 13.1 μM for EPR. Notably, none of the three alkyl BCPs with morphological/behavioral effects induced cyp1a. Previous literature suggests that BCP likely acts through the canonical pathway of Cyp-dependent toxic metabolism resulting in DNA damage and oxidative stress in vivo and in vitro (Suzuki et al., 2022; Jacob et al., 1996; Stevenson and Von Haam, 1965; Baum et al., 2001). Lack of cyp1a induction in our study indicates that alkyl BCPs could also be acting as Ahr antagonists or through Ahr-independent mechanisms (Coelho et al., 2022).

3.2.9. Chrysene

Five alkyl chrysenes were tested and all induced cyp1a expression in zebrafish. Singly methylated 3- and 5-methylchrysene (MC) had similar BMD50 values for morphological effects (60–70 μM) while 2-MC did not elicit morphological or behavioral effects. 5,12-DMC had a lower morphology BMD50 (44.2 μM) than 3- and 5-MC, and was the only alkyl chrysene to cause behavioral effects (LPR BMD50 8.2 μM), although chrysene and 6-MC have previously been shown to have LPR effects (Geier et al., 2018b). The inactivity of 2-MC compared to 3- and 5-MC shows that specific structural differences affect biological activity, even though all alkyl chrysenes tested can induce cyp1a. The higher potency of 5,12-DMC was likely not a simple function of higher molecular weight as 6-ethylchrysene has an identical weight and was inactive.

Consistent with our results, previous work has established that chrysene does not impact zebrafish developmental morphology while methylated chrysenes do (Incardona et al., 2005, 2006; Geier et al., 2018a; Lin et al., 2015; Alqassim et al., 2019). In vitro and Ames assays have shown methylated chrysenes to be more carcinogenic and mutagenic than the parent chemical (Machala et al., 2008; Cheung et al., 1993). However, only 5-MC compared to all other singly methylated chrysenes was shown to form DNA adducts in vitro (Machala et al., 2008). In mice, only 3-and 5-MC acted as tumor initiators and 5-MC acted as a tumor initiator and promoter (Hoffmann et al., 1974). Many 5-MC studies have demonstrated its genotoxic, mutagenic, and carcinogenic potential (da Silva Junior et al., 2021). Differences in carcinogenic potential appear to vary with methyl position perhaps due to steric hindrance of DNA adduct repair (Machala et al., 2008). Our results correspond well with the literature where both 3- and 5-MC were more active than 2-MC. We found that 5,12-DMC was more morphologically potent than 5-MC. This suggests the need for more methyl chrysene structure-activity research to understand their mechanisms and their range of hazard potential.

3.2.10. Benzo(a)pyrene (BaP)

Three alkyl BaPs were tested, and all induced cyp1a expression, 7-, 8-, and 9-methylbenzo(a)pyrene (MBaP). 8-MBaP was the only substituted BaP to elicit morphological effects and had a BMD50 of 0.3 μM, the most potent of all the structures tested in this screen. Exposure to 7-MBaP did not lead to morphological effects up to 100 μM. 9-MBaP had an EPR BMD50 of 1.2 μM, this may indicate a deficit that would manifest as a morphological effect at a later life stage or if tested at a higher concentration (Reif et al., 2016). BaP is widely understood to act as an AHR agonist to induce metabolic activation by Cyps into benzo(a)pyrene diol epoxide (BPDE) resulting in DNA adduct formation, DNA damage, and potentially carcinogenesis (Bukowska et al., 2022). While a variety of other mechanisms have been identified for BaP toxicity, most involve metabolic activation leading to carcinogenesis or oxidative stress. Previously BaP has been found to induce cardiac malformations in developing zebrafish, which are hypothesized to be at least in part due to diol-epoxide mediated DNA damage (Zou et al., 2024).

Literature review suggests that the mutagenic potential of methylated BaPs varies based on methylation position and metabolism pathway, however, nothing indicated that 8-MBaP might be more mutagenic than BaP (Wang et al., 2022). Generally, alkylation slows the intrinsic clearance of BaP and favors metabolic oxidation of the aliphatic side chain instead of aromatic ring oxidation which is key in the activation of BaP to BPDE and other toxic metabolites (Bukowska et al., 2022; Wang et al., 2022). The potent toxicity of 8-MBaP may not occur through BPDE formation. The proposed meso-methylation mechanism for the bioactivity of BaP and related PAHs may explain the toxicity of 8-MBaP. Meso-methylation via methyltransferase enzymes and modification of the parent chemical followed by subsequent substitution reactions results in a strong leaving group (Br, OH, OSO₃H, etc.) bound through a carbon to the PAH, resulting in a reactive metabolite. These subsequent substitution reactions are majorly performed by AHR induced Cyp P450 enzymes (Flesher and Lehner, 2016). In the case of BaP, the proposed methylated PAH with carcinogenicity is 6-MBaP (Surh et al., 1989). While 8-MBaP is not yet known to be part of the meso-methylation mechanism for BaP toxicity, substitution of 8-MBaP with a strong leaving group could explain its potency.

Our results concord with Fang et al. who found that 8-MBaP was the most potent methylated BaP tested and 7-MBaP did not elicit morphological effects in zebrafish (Fang et al., 2022). Subsequently, Fang et al. demonstrated that while the toxicity of BaP could be counteracted by the addition of an Ahr antagonist, this was not the case with 8-MBaP, despite its ability to activate Ahr responsive genes (Fang et al., 2023). Ahr independence of 8-MBaP would make the mechanisms previously mentioned unlikely explanations. Confirmation of the Ahr independence of 8-MBaP toxicity in our system is ongoing.

3.2.11. Dibenz(a,h)anthracene (AHA)

3-methyldibenzo(a,h)anthracene (MAHA) was the only readily procurable alkyl AHA. It induced cyp1a and elicited an abnormal EPR (BMD50 18.3 μM). Parent AHA can be bioactivated by P450 enzymes to dihydrodiols and diol epoxides. Chang et al. found that AHA and some, but not all, metabolites, are tumorigenic both in dermal mouse exposures and newborn mouse exposures, resulting in skin, pulmonary and hepatic tumors (RamaKrishna et al., 1992; Chang et al., 2013). The tumorigenic potential of AHA has been confirmed in a variety of other mammalian studies (Buening et al., 1979; Platt et al., 1990; Lubet et al., 1983). AHA previously elicited morphological and LPR effects but no EPR in embryonic zebrafish at concentrations similar to those used herein for 3-MAHA (Geier et al., 2018a). Emergent EPR effects from 3-MAHA and loss of morphological effects suggest it acts through a mechanism different from AHA.

3.3. Comparison of biological activity, cyp1a expression patterns and Ahr dependence of polar and alkylated phenanthrenes

Phenanthrenes are often the most abundant parent PAH contaminant among foodstuffs, drinking water, and PM2.5 particle surfaces (England et al., 2024). This parent group was selected for further analysis as both parent and substituted structures are cardiotoxic in fish and across model systems (Donald et al., 2023; Incardona et al., 2004, 2024; England et al., 2024). Additionally, the phenanthrene parent group had the largest number of diverse chemical standards available (24) allowing us to investigate structure-activity relationships.

3.3.1. Polar substituted phenanthrenes are more active than alkyl substituted

Unsubstituted phenanthrene was inactive, while all substituted phenanthrenes tested had activity in at least one end point (Fig. 3). The seven hydroxy and quinone phenanthrenes clustered together based on phenotypic outcome as they all elicited both morphological and EPR effects, apart from 9-nitrophenanthrene which had no EPR effects. The two quinones elicited the most potent morphological effects, however all the polar phenanthrenes (1,4-phenanthrenedione>9,10-phenanthrenequinone>9-nitrophenanthrene>3-hydroxyphenanthrene (3-HPHE)>4-HPHE>1-HPHE>9-HPHE>9-aminophenanthrene) were more morphologically potent than any of the alkyl phenanthrenes. Polar phenanthrene toxicity may correlate to increased solubility and biological availability and should be explored further via body burden in the developing zebrafish. Cyp1a expression location did not cluster with phenotypic outcomes, suggesting that its expression could not predict phenotypic outcome. All the alkyl phenanthrenes induced cyp1a expression in contrast to the polar phenanthrenes where only 50% induced cyp1a while all were morphologically active. This suggested that canonical Ahr agonism was unrelated to polar phenanthrene toxicity. Hydroxy, nitro and amino phenanthrenes clustered together based on structure while alkyls clustered together and the two quinones were further from the rest of the chemicals (Fig. 4). Despite the close alkyl phenanthrene structural clustering, only retene had a significant impact on developmental morphology. The number of substitutions and their location was a stronger driver of teratogenicity among alkyl phenanthrenes than it was for polar phenanthrenes, where the presence of a polar substitution drove teratogenicity. Variable sensitivity to structural differences was likely due to a different toxic mechanism between the polar and alkyl phenanthrenes.

3.3.2. Polar substituted phenanthrenes act Ahr independently

While the cyp1a reporter showed the extent and location of cyp1a expression, it was unable to determine if Ahr activation is necessary for toxic outcomes. To this end, three independent Ahr knock out lines were used: ahr2^hu3335, ahr1a^osu6 and ahr1b^wh36. IHC analysis indicated a functional knock out of Ahr1a evidenced by the loss of Cyp1a expression in the liver in ahr1a^osu6, representative images are included in SI2 Fig 4.

The toxicity of polar substituted phenanthrenes was not dependent upon the three Ahrs (Fig. 5). It is important to note that the Ahrs were knocked out individually in these exposures so compensatory activity of the two functional homologs of the Ahr could not be ruled out. However, lack of any change in toxicity from 100% effect in combination with the inconsistency of cyp1a induction provided strong evidence that the observed toxicity was not entirely dependent on the Ahr. Retene, however, did show partial dependence upon the Ahr2 based on reduced incidence of “any effect” (100% in 5D and 67% in ahr2^hu3335). To confirm these results, 5D and hu3335 embryos were exposed to retene at 5 concentrations and the wildtype 5D fish exhibited a significant increase in morphological effects (>35 μM), while ahr2^hu3335 showed no significant bioactivity (SI2 Fig. 5), demonstrating retene toxicity was Ahr2 dependent in zebrafish as has been previously reported (Scott et al., 2011; Wilson et al., 2022).

Fig. 5. Ahr dependence of substituted phenanthrene morphological effects.

Bar plot of percent any effect observed in 5D, ahr2^hu3335, ahr1a^osu6 and ahr1b^wh36) lines (denoted by color) exposed at the EC100 or 100 μM for each chemical or control dimethyl sulfoxide (DMSO).

4. Conclusion

This study contributes to a growing body of work which demonstrates that substituted PAHs are often developmentally hazardous and that their toxicity is variable in potency and MOA. Further study to elucidate the toxicokinetics and molecular responses of these chemicals is warranted and would provide more thorough insights into the MOA of these chemicals and provide essential information for human and environmental health risk assessments. This work revealed that relatively subtle structural differences among substituents of a given parent class produced dramatically different toxic outcomes. This provides further evidence that chemical class or even parent chemical group hazard characterization is inadequate. Instead, a phenotype-anchored, mechanism of action-driven hazard classification would be a considerably more informed and accurate approach. Given their diversity and environmental prevalence, substituted PAHs warrant more robust environmental monitoring, careful hazard characterization, and refined structure-activity classification to fully understand the risk that their real-world mixtures present.

HIGHLIGHTS.

• 104 alkyl PAHs were tested for biological activity and cyp1a induction in zebrafish.

• Alkyl PAH toxicity varies by substitution location within and across parent groups.

• AHR dependence and spatial cyp1a induction examined for alkyl/polar phenanthrenes.

• Polar phenanthrenes were more active than alkyl and less likely to induce cyp1a.

• Cyp1a expression does not always indicate that observed toxicity is AHR dependent.

Glossary

PAH

polycyclic aromatic hydrocarbon

EPA

Environmental Protection Agency

BaP

Benzo(a)pyrene

Ahr

aryl hydrocarbon receptor

Cyp

cytochrome P450

MOA

mode of action

Cyp1a

cytochrome P4501A

DMSO

dimethyl sulfoxide

Hpf

hours post fertilization

LEL

lowest effect level

BMD50

benchmark concentration 50%

EPR

embryonic photomotor response

LPR

larval photomotor response

EGFP

enhanced green fluorescence protein

AA

amino acid

NGS

normal goat serum

PCA

principal component analysis

BAA

benz(a)anthracene

BCP

benzo(c)phenanthrene

C

crysene

AHA

dibenz(a,h)anthracene

DBT

dibenzothiophene

FL

fluorene

PHE

phenanthrene

M

methyl

DM

dimethyl

TM

trimethyl

E

ethyl

NGS

normal goat serum

Data availability

I have shared the raw data in a supplementary files