Contents lists available at ScienceDirect

Ecotoxicology and Environmental Safety

journal homepage: www.elsevier.com/locate/ecoenv

Airborne polycyclic aromatic compounds contribute to the induction of thetumour-suppressing P53 pathway in wild double-crested cormorants

S.J. Wallacea, S.R. de Sollab, P.J. Thomasc, T. Harnerd, A. Engd, V.S. Langloisa,e,⁎

a Department of Chemistry and Chemical Engineering, Royal Military College of Canada, Kingston, ON, Canadab Ecotoxicology and Wildlife Health Division, Environment and Climate Change Canada, Burlington, ON, Canadac Ecotoxicology and Wildlife Health Division, Environment and Climate Change Canada, Ottawa, ON, Canadad Air Quality Processes Research Section, Environment and Climate Change Canada, Toronto, ON, Canadae Institut national de la recherche scientifique – Centre Eau Terre Environnement (INRS), Quebec City, QC, Canada

A R T I C L E I N F O

Keywords:BirdsPolycyclic aromatic compoundsToxicityGene expressionProtein expressionDNA methylation

A B S T R A C T

Polycyclic aromatic compounds (PACs), including polycyclic aromatic hydrocarbons (PAHs) and PAH-likecompounds are known or probable environmental carcinogens released into the environment as a by-product ofincomplete combustion of fossil fuels and other organic materials. Studies have shown that exposure to PACs inthe environment can induce both genotoxicity and epigenetic toxicity, but few studies have related PAC ex-posure to molecular changes in free ranging wildlife. Previous work has suggested that double-crested cor-morants (Phalacrocorax auritus; DCCO) exhibited a higher incidence of genetic mutations when their breedingsites were located in heavily industrialized areas (e.g., Hamilton Harbour, Hamilton, ON, Canada) as comparedto sites located in more pristine environments, such as in Lake Erie. The aim of this study was to determine ifairborne PACs from Hamilton Harbour alter the tumour-suppressing P53 pathway and/or global DNA methy-lation in DCCOs. Airborne PACs were measured using passive air samplers in the Hamilton Harbour area andlow-resolution mass spectrometry analysis detected PACs in livers of DCCOs living in Hamilton Harbour. Furtherhepatic and lung transcriptional analysis demonstrated that the expression of the genes involved in the DNArepair and cellular apoptosis pathway were up-regulated in both tissues of DCCOs exposed to PACs, while genesinvolved in p53 regulation were down-regulated. However, global methylation levels did not differ betweenreference- and PAC-exposed DCCOs. Altogether, data suggest that PACs activate the P53 pathway in free-rangingDCCOs living nearby PAC-contaminated areas.

1. Introduction

Polycyclic aromatic compounds (PACs), including parent polycyclicaromatic hydrocarbons (PAHs), alkylated PAHs, naphthalene, and di-benzothiophenes (Achten and Andersson, 2015), are widespread en-vironmental contaminants that are formed by the incomplete combus-tion of organic matter through natural (e.g., forest fires) andanthropogenic sources (e.g., burning of fossil fuels), or as componentsof petroleum products. Airborne PACs are generally associated withparticulates in the air (including PM2.5; IARC, 2010) with deposition asthe main route of entry of PAHs into the terrestrial and aquatic en-vironment (Kim et al., 2013). Breathing in PAHs from ambient air,smoking, and diet are the main routes of PAH entry into the humanbody (ACGIH, 2005; reviewed in Kim et al., 2013). The InternationalAgency for Research on Cancer classifies some PAHs as carcinogenic tohumans and animals (Class 1), while other PAHs are classified as

possibly carcinogenic (Class 2A, 2B; IARC, 2010).Many studies have demonstrated that PACs and their metabolites

are carcinogenic causing acute and chronic health effects in vertebrates.In human epidemiology studies, quantifiable links have been in-vestigated between ambient exposure through occupation (e.g., cokeoven workers) and the increased risk of lung cancer (reviewed inBoström et al., 2002). Mice exposed orally to benzo[a]pyrene hadhigher mRNA levels of cancer-related genes in lung tissue and most ofthese altered genes were involved in the tumour-suppressing P53pathway, including cell cycle arrest/senescence (CDKN1A or P21),apoptosis (BAX), and negative regulation of P53 (MDM2; Labib et al.,2012). Moreover, P53R2 (ribonucleotide reductase small subunit 2; alsoknown as RRM2B) and GADD45α (growth arrest and DNA damage)have been validated as biomarkers of benzo[a]pyrene exposure tohuman cells (Ohno et al., 2008; Xin et al., 2015). In addition, reactiveoxygen species formation due to PAH metabolite exposure can activate

https://doi.org/10.1016/j.ecoenv.2017.12.028Received 1 August 2017; Received in revised form 8 December 2017; Accepted 12 December 2017

⁎ Correspondence to: Institut national de la Recherche Scientifique – Centre Eau Terre Environnement (INRS-ETE), 490, de la Couronne, Québec, Canada G1K 9A9.E-mail address: valerie.langlois@inrs.ca (V.S. Langlois).

Ecotoxicology and Environmental Safety 150 (2018) 176–189

Available online 22 December 20170147-6513/ © 2017 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).

T

P53, which in turn, will augment OGG1 (8-oxoguanine DNA glycosy-lase), a gene that repairs DNA damaged via base excision repairs(Abedin et al., 2013). Of note, in over half of human cancers, P53 ismutated losing the ability to activate the tumour-suppressing pathways,which can lead to carcinogenesis (reviewed in Robles et al., 2002).

Furthermore, environmental contaminants can alter the stability ofthe genome through epigenetic modifications (Baccarelli and Bollati,2009). DNA methylation occurs by DNA methyltransferases (DNMTs)adding a methyl group to cytosine followed by a guanine (CpG),creating 5-methylcytosine (5-mC; Pfeifer et al., 2014). Ten-eleventranslocation (TET) proteins oxidize 5-mC to 5-hydroxymethylcytosine(5-hmC), an intermediate but stable state in the demethylation pathway(Pfeifer et al., 2014). A decrease in global 5-mC levels can lead togreater genomic instability and correlates with tumour progressionthrough carcinogenesis (Kisseljova and Kisseljov, 2005). Alternatively,hypermethylation can occur, by increasing DNMT1 activity inducing 5-mC production, or by impairing the demethylation system throughdecreasing TET activity, leading to some cancers in humans (Pfeiferet al., 2014). Alteration of methylation of gene promotors has been usedas a biomarker of lung carcinogenesis in humans (reviewed in Vineisand Husgafvel-Pursiainen, 2005). Other studies have clearly shown thatenvironmental contaminants, including PACs, can alter DNA methyla-tion levels in an experimental setting, including benzo[a]pyrenecausing global hypomethylation in mice (Wilson and Jones, 1983) andzebrafish (Corrales et al., 2014; Fang et al., 2013). However, PACs existin complex mixtures in the environment and can cause both hypo-methylation and hypermethylation in humans (reviewed in Ruiz-Hernandez et al., 2015), but little is known on how environmentalcontaminants affect wildlife DNA methylation levels in natural popu-lations (Head, 2014).

Air pollution, including airborne PACs has been shown to alter thehealth of wild birds. Cruz-Martinez et al. (2015) found that oil sandsrelated air emissions resulted in induction of detoxifying enzymes andaltered immune function in exposed tree swallows (Tachycineta bicolor).Concentrations of protoporphyrins, which are precursors to heme andto cytochrome P450 enzymes, from rock dove (Columba livia) excretawere proportional to airborne PAHs in the city of Milan, Italy (Sicoloet al., 2009). Although PAC emissions have been declining since the1950s, the highest levels of PACs occur in industrial areas (Kim et al.,2013). Canada's two largest steel mills are located in Hamilton Harbour,Ontario and produce emissions containing complex mixtures of PACs.These compounds have been found at high concentrations in harboursediment (1.6 – 1470 µg/g total PAH; CEPA, 1994) above the interimsediment quality guidelines ranging 0.006 – 0.1 µg/g dw for individualPAHs (Canadian Council of Ministers of the Environment, 1999). Inaddition, PACs have been detected at higher concentrations in harbourair (0.6–22.12 ng/m3 sum of 8 PACs, average 1.22 ng/m3 benzo[a]pyrene in 2013; HAMN, 2016) than the Ontario Ambient Air QualityCriteria guideline of 0.22 ng/m3 benzo[a]pyrene (HAMN, 2016). Thedouble-crested cormorant (Phalacrocorax auritus; DCCO) is an abundantpiscivorous waterbird that breeds and lives in large colonies in Ha-milton Harbour, ON. This bird population is continuously exposed toenvironmental PACs, including from the various steel mills and otherlarge industry that dot the landscape and from vehicular exhaust fromthe two major highways adjacent to the harbour. A previous study doneby King et al. (2014) detected PAHs in DCCO chick liver and bile fromsites in Hamilton Harbour, including the mutagenic phase I metabolitebenzo[a]pyrene-7,8-diol. In addition, microsatellite mutation rateswere 11-fold higher in DCCOs sampled from a site (Pier 27) locateddownwind of the coking ovens of the steel mills than those from thereference site (King et al., 2014). Earlier work on herring gulls (Larusargentatus) indicated that germline DNA mutation rates were higher atHamilton Harbour relative to cleaner reference groups, using minisa-tellites (Yauk and Quinn, 1996). Additionally, breathable particulatematter from Hamilton Harbour was found to be the casual factor behindthe induction of germline mutations in caged mice (Mus musculus,

Somers et al., 2004). These studies indicated that the increased rate ofmutations may be a signal of DNA damage as a result of PAC exposure.

In addition to airborne exposure, diet is potentially another im-portant route of PAC exposure. Indeed, DCCOs living in HamiltonHarbour are also being exposed from eating fish that are known tocontain high levels of PACs coming from the Randle reef coal tar de-posit; a site awaiting remediation and considered to be among the mostpolluted area in Canada (Hall and O'Connor, 2016). However, previouswork has demonstrated that the diet of the different DCCO coloniesliving in Hamilton Harbour were virtually identical as estimated byregurgitates, fatty acid, and stable isotope profiles (King et al., 2014,2017). The foraging range of DCCO is typically several km from theirhome colony (on average 2 km with a maximum of 40 km; Custer andBunck, 1992) and they often forage outside of the harbour proper on aregular basis. This suggests that adult DCCO from Hamilton Harbourcolonies, which are in close proximity to one another (< 3 km), likelyforage for the same resources in the Lake Ontario area. Therefore, anydifferential biomarker effects in DCCOs from Hamilton Harbour co-lonies would likely be due to PAC differences in airborne exposurerather than diet.

The aim of this study was to determine if environmental exposure tocomplex mixtures of PACs alter the tumour-suppressing P53 pathwayand/or global DNA methylation rates in wild DCCOs. To do this, wedetermined mRNA, protein, and global methylation levels in liver andlung tissue of DCCOs living at sites with differing PAC airborne ex-posures. By estimating the expression of genes associated with P53, wecan assess if PAC airborne exposure alters the tumour-suppressing P53pathway, which potentially could lead to carcinogenesis. We predictedthat DCCO living in Hamilton Harbour and downwind to the source ofPACs would have the most alterations to the P53 pathway and DNAmethylation levels than DCCO living on the reference site in Lake Erie.

2. Methods

2.1. Sample collection

Ten double-crested cormorant (DCCO) chicks were sampled at eachlocation: Mohawk Island in eastern Lake Erie (reference site;42.8345° N, 79.5226° W), Centre Island (43.3046° N, 79.8028° W) andPier 27 (43.2833° N, 79.7937° W) in Hamilton Harbour. DCCO colonieswere visited in mid-June 2014. All sampling was restricted to pre-fledgling DCCO chicks, thus ensuring that all PAC exposure was fromlocal sources either through the air or through diet via the parents.DCCO chicks were removed from the nest by hand and placed in a dark,covered box until processed. Chicks were decapitated following stan-dardized operating protocols and animal care standards (Environmentand Climate Change Canada ACC SOP-07). The bodies were im-mediately dissected to obtain lung and liver tissues that were targetedfor assessing the P53 pathway. The liver is generally considered theprimary organ responsible for metabolism of xenobiotics because itcontains the largest concentration of cytochrome P450 compared toother tissues. Lung tissue was collected, as it is the first point of contactwith airborne exposures. All samples were flash frozen in liquid ni-trogen and stored at −80 °C until genetic analysis.

2.2. Characterization of exposure

Exposure of DCCO to PACs were estimated in three ways: i) airborneexposure was assessed by measuring PACs using passive air samplersnear the colonies in Hamilton Harbour, ii) air speed and velocity weremeasured up- and down-wind of the major putative PAC sources inHamilton Harbour, relative to the location of the DCCO colonies, andiii) hepatic burdens of PACs were measured. Pier 27 was considered,putatively, downwind of steel mills and other heavy industry moreoften than Centre Island. Pictures of the two Hamilton Harbour co-lonies, with the terrain immediately upwind (WSW) of each can be seen

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

177

in Fig. A.1.

i) PAC analysis in airPACs were measured in air samples collected from three sites inHamilton in 2014 from May 12 to June 30. Two sites (Pier 15E and15W) in Hamilton were located on industrial properties near thePAC sources, and one site was upwind (Bayfront Park). PAC airsamples were collected using polyurethane foam (PUF) disk passiveair samplers, which have been shown to capture both gas-phase andparticle-phase PACs (Harner et al., 2013). Total PACs (ΣPACs) re-presented the sum of 75 analytes, including 17 parent PAHs, 20 C1-C4–alkylated PAHs, dibenzothiophenes, and four of their C1-C4analogs, and naphthalene (for details refer to the Appendix).

ii) Wind directions and PAC sourcesIn King et al. (2014), the expected exposure period for chicks fromhatching to fledging was estimated to be March 20 to June 14.Using this same time period, we collected wind direction for thisexposure period in 2014 from the Hamilton Air Monitoring Net-work (HAMN: hamnair.ca) and Environment and Climate ChangeCanada. We downloaded hourly average wind direction at 10 mheight from the ground at Station 29167 (2088 h), approximately600 m southwest of the Pier 27 colony, and the Burlington Pierstation (1800 h), approximately 1 km ESE from Centre Island. Wecalculated wind roses to estimate the direction of prevailing winds,relative to the putative PAC sources by calculating the proportion ofobservations in each cardinal (N,E,S,W) and intercardinal (NE, NW,SE, SW) directions. Thus, we were able to calculate the proportionof hours that each colony was downwind of the heavy industry foreach colony. Given the proportions, p1 and p2, (i.e., proportion oftime the Pier 27 colony was downwind of industry, and proportionof time the Centre Island colony was downwind of industry), wecalculated the odds ratio that the Pier 27 colony vs Centre Islandwas downwind of the heavy industry, given: =

−

−

OR p pp p

/(1 )/(1 )

2 2

1 1, plus

the 95% CI around the odds ratio. Pier 27 was considered down-wind with wind directions from 292.4° (WNW) to 112.5° (ESE), andCentre Island from 292.4° (WNW) to 67.4° (ENE).

iii) PAC analysis in liverPACs in DCCO liver (n = 7) were analyzed by SGS AXYS, using amodification of US EPA's Methods 1625B and 8270 C/D (SGS AXYSmethod MLA-021; for details refer to the Appendix). Total PACs(ΣPACs) represented the sum of 75 analytes, including 19 parentPAHs, 54 C1-C4–alkylated PAHs, and biphenyl and naphthalene(for details refer to the Appendix).

2.3. Gene expression

DCCO liver and lung tissue (total n = 10 per tissue, including fromthe same individuals analyzed for PAC concentration) were preparedfor gene expression analysis following MIQE guidelines (Bustin et al.,2009). Total RNA was extracted using the e.Z.N.A Total RNA Kit(Omega Bio-Tek, Norcross, GA, USA) and was treated with RQ1 RNase-free DNase (Promega, Madison, WI, USA). Concentration and puritywere measured using the Nanodrop 2000 spectrophotometer (Ther-mofisher, Ottawa, ON, CAN). Complementary DNA (cDNA) was syn-thesized using GoScript Reverse Transcription System (Promega, Ma-dison, WI, USA) according to manufacturer's protocol with randomprimers and RNA input to achieve 1 µg cDNA per sample. Each qPCRreaction consisted of 1x GoTaq qPCR Master Mix (Promega, Madison,WI, USA) containing Bryt green fluorescent dye, varying concentrationsof primers (Table A.2), nuclease-free water, and template cDNA. ACFX96 Real-Time PCR System (Bio-Rad Laboratories, Mississauga, ON,CA) was used with the following thermocycler profile: 3 min at 95 °Cfor enzyme activation, 40 cycles of denaturation for 15 s at 95 °C andannealing for 60 s at optimal temperature (Table A.2), and denaturationfor 10 s at 95 °C. A dissociation curve was generated by increasing the

temperature by 0.5 °C per cycle. Each sample was run in duplicate andeach plate was run with a standard curve containing six points of aserial 1:4 dilution from 50 ng DNA pooled from each treatment, a notemplate control and a no reverse transcriptase control all in duplicate.A successful run was determined by the standard curve having an ef-ficiency of 90–110% and a coefficient of determination (r2) valueof> 0.985 assessed in CFX Manager (version 3.1, Bio-Rad Laboratories,Mississauga, ON, CA). The relative standard curve method was used tocalculate mRNA levels of genes of interest involved in the P53 pathway(Table A.2). First, expression was averaged across duplicates and nor-malized to the mean expression of the reference genes (beta-actin,ACTβ; elongation factor 1, EEF1α1; ribosomal protein L8, RPL8; andribosomal protein L4, RPL4; Table A.2), which did not change withtreatment. Next, results were presented as fold changes± standarddeviation relative to the average normalized mRNA level in the samplesfrom the reference site. All products were cloned and sequenced toverify the amplification products were from the correct genes (for de-tails refer to the Appendix).

2.4. Protein quantification

Proteins were extracted by homogenizing liver and lung tissue froma subset of the same DCCO (n = 6 individuals per site) in lysis bufferwith phenylmethylsulfonyl fluoride crystals as a protease inhibitor. ThePierce 660 nm Protein Assay (ThermoFisher Scientific, Mississauga,ON, CAN) was used to determine protein concentration. Total proteinswere separated using SDS-PAGE and transferred to a PVDF membrane.Membranes were blocked with 5% milk for 1 h, incubated with theappropriate primary antibody (1:200 for P53 and 1:500 for P53R2;Santa Cruz Biotechnology, Inc., Dallas, TX, USA) overnight at 4 °C andincubated with the appropriate secondary antibody (anti-rabbit 1:2000for P53 and anti-goat 1:2000 for P53R2; BioShop Canada Inc.,Burlington, ON, CAN) for 45 min. Protein levels were visualized using aClarity Western ECL Substrate (Bio-Rad Laboratories, Mississauga, ON,CAN) for 5 min and the image was captured through chemiluminesenceusing the ChemiDoc XRS+ Imaging System (Bio-Rad Laboratories,Mississauga, ON, CAN). A linear dynamic range was determined foreach antibody by loading a range of 5–80 µg total protein and choosingthe amount to maximize the signal without being saturated (40 µg forP53 in the liver and 20 µg (liver), 40 µg (lung) for P53R2; for detailsrefer to the Appendix). A positive control (P53 recombinant protein; orMCF7 cells for P53R2; Santa Cruz Biotechnology, Inc., Dallas, TX, USA)and three samples used as between membrane controls were includedon each membrane to account for the variation within and amongmembranes. Intensity was determined using Lane and Bands tool andadjusted to local background in ImageLab (version 5.1, Bio-Rad La-boratories, Mississauga, ON, CAN) and protein levels were normalizedto the total protein loaded for that sample and the across membranecontrols. Protein levels are expressed as fold changes± standard de-viation relative to the protein level in the samples from the referencesite.

2.5. Global DNA methylation

DNA was extracted using DNeasy Blood and Tissue kits (Qiagen Inc.,Toronto, ON, CAN) from the same liver and lung tissue samples used inthe previous gene expression and protein assays (n = 10 individuals persite). To test for DNA methylation level, a dot blot analysis was used fordetection of global 5-methylcytosine (5-mC) and global 5-hydro-xymethylcytosine (5-hmC) levels. A linear dynamic range was estab-lished by blotting 150–400 ng DNA (5-mC) and 25–200 ng DNA (5-hmC) and comparing to the intensity of un-methylated and 100% me-thylated DNA (for details refer to the Appendix). The optimal amount of200 ng DNA was blotted and cross-linked onto a positively chargednylon membrane. Membranes were blocked with 10% milk diluted in2x SSCT (0.05% Tween-20) for 60 min, incubated with 5-mC primary

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

178

antibody (1:2000; Active Motif, Carlsbad, CA, USA) overnight at 4 °C,and incubated with a secondary antibody (peroxidase conjugated affi-nity purified anti-Mouse IgG, 1:14000; BioShop Canada Inc., Bur-lington, ON, CAN) for 45 min. To detect 5-hmC levels, the same pro-tocol was used except that the membrane was blocked with 5% milkdiluted in 2x SSCT for 40 min, incubated with 5-hmC primary antibody(1:7500; Active Motif, Carlsbad, CA, USA) overnight at 4 °C, and in-cubated with a secondary antibody (peroxidase conjugated affinitypurified anti-Rabbit IgG, 1:12000, BioShop Canada Inc., Burlington,ON, CAN) for 40 min. The membrane was treated with Clarity WesternECL Substrate (Bio-Rad Laboratories, Mississauga, ON, CAN) for 5 minand the image was captured through chemiluminesence using theChemiDoc XRS+ Imaging System (Bio-Rad Laboratories, Mississauga,ON, CAN). Methylation level was determined through intensity leveladjusted to local background using the volume tools in ImageLab(version 5.1, Bio-Rad Laboratories, Mississauga, ON, CAN). Sampleswere blotted in duplicate and a negative control (without DNA) wasblotted on each membrane. Membranes were treated with a 0.02%methylene blue solution and subsequently de-stained with 20% ethanolto normalize equal DNA loading. Methylation levels are presented asfold changes± standard deviation relative to the DNA methylationlevel in the samples from the reference site. A standard curve of 100%,75%, 50%, 25%, and 0% methylation was used to validate 5-mC levelsand confirm that a maximum of 3.5 fold change (75% increase) can bedetected with this method (more details in Appendix, Fig. A5).

2.6. Data analysis

PAC concentrations were log-transformed and compared using aone-way ANOVA. Gene expression, protein, and global methylationlevels were tested for normality using the Shapiro-Wilk goodness of fittest by treatment and for unequal variances among treatments using theBartlett's test (for details refer to the Appendix). Outlier individualswere removed from the gene expression analysis based on the ROUTstatistical method (Q = 1%; 0 − 3 per gene per site). Individuals werenot removed for the methylation assay, as it was highly variable. One-way ANOVAs were performed on mRNA, protein, and global methyla-tion levels. Data are presented as fold change relative to Mohawk Island(reference site). Tukey Honest Significant Difference post-hoc tests wereperformed on significant ANOVA values. All statistical analyses wereperformed in JMP (version 12.2.0; SAS Institute Inc.) or Statistica 7(Statsoft).

3. Results

3.1. Airborne PAH exposure

Concentrations of PACs were higher at both Pier 15E and 15Wcompared to Bayfront Park, with the aggregate sum ranging be-tween 355–534 ng/m3 at Pier 15, and 121 ng/m3 at Bayfront Park(Fig. 1a). Parent PAHs and alkylated PAHs each contributed aboutequally to the aggregate sum in air (41.6% to 55.1%, and 39.5–55.7%,respectively) at the Hamilton Harbour sites, with naphthalene and di-benzothiophene each contributing 4.3% or less. Prevailing winds inHamilton Harbour near Centre Island and Pier 27 indicate that theyprimarily originated W (range of NW to SW) from the direction of heavyindustry, or E or NE, with little N or S direction (Fig. 1b). We estimatedthat the proportion of the time Centre Island was downwind of theindustrial areas was 0.42 (95% CI 0.40–0.44), and for Pier 27 was 0.59(95% CI 0.57–0.61); the odds ratio of DCCO at Pier 27 being downwindwas 1.99 1.74–2.27) compared to those from Centre Island. Based uponthe predominant wind directions, the DCCO colony at Pier 27 wasdownwind of the putative airborne PAC sources approximately twice asmore frequently than the colony from Centre Island.

3.2. PAC burdens in the liver

The arithmetic means were 2.5, 4.3, and 9.6 ng/g PAHs in liver, forMohawk Island (reference site), Pier 27, and Centre Island, respectively.One DCCO chick from Mohawk Island had an unusually high con-centration (13.0 ng/g), indicating that there was substantial variationamongst some birds, which could be due to a difference in dietary ex-posure at that specific site. Mean concentrations of log-transformed sumPACs in liver differed among the three colonies (F[2,18] = 7.48,p<0.0043; Fig. 2)), and concentrations were higher in livers of birdsfrom the two colonies in Hamilton compared to Mohawk Island(p<0.0332; 0.004). Generally, alkylated PAHs were more abundantthan parent PAHs in DCCO liver (Fig. A.1) and were dominated byphenanthrenes/anthracenes, retene, or fluorenes.

3.3. Gene expression related to the tumour-suppressing P53 pathway

In liver and lung tissue, the levels of P53R2 mRNA were statisticallysignificant between sites (F[2,27] = 5.15, p = 0.014 and F[2,27] = 5.59,p = 0.011, respectively) with 1.5-fold increases for the birds from Pier27 compared to birds from Mohawk Island (Tukey HSD; p = 0.013 andp = 0.019; respectively; Fig. 3e and f). Similarly, BAX mRNA levelswere different among sites in the liver (F[2,28] = 15.14, p = 0.0001)and the lung (F[2,28] = 5.09, p = 0.020) with a 2-fold increase of BAXmRNA levels measured in birds from Pier 27 (Tukey HSD; p<0.0001and p = 0.015; respectively, Fig. 3m, n). In addition, a 0.5-fold changeincrease in MDM2 mRNA levels in the liver was observed in individuals

Fig. 1. a) Concentrations of PACs in air derived from polyurethane foam (PUF) disksamplers, representing airborne PAC exposure, from one site upwind (Bayfront Park) andtwo sites within heavy industry in Hamilton Harbour (Pier 15 W, Pier 15 E). The locationsof the Centre Island and Pier 27 colonies are also given. The blue column represents thesum of parent and alkylated naphthalenes, while the purple column represents the sum ofparent and alkylated dibenzothiophenes. b) Wind roses indicating prevalent wind di-rections and speeds, at stations nearest to the two DCCO colonies.

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

179

from Centre Island (F[2,25] = 3.52, p = 0.037; Fig. 3c); while a 0.3-foldchange decrease in MDM2 mRNA levels was observed in lung tissue ofindividuals from Pier 27 (F[2,25] = 4.13, p = 0.032; Fig. 3d). GADD45αtranscript levels were two-fold change higher in lung tissue from in-dividuals from Pier 27 compared to individuals from Centre Island(F[2,27] = 3.61, p= 0.043; Fig. 3j). No mRNA level changes were foundfor P53, P21, and OGG1.

3.4. Expression of P53 and P53R2 proteins

Although there was a decreasing trend from the reference site toPier 27, P53 protein levels were not different between sites in livertissue (p= 0.63). Similarly, P53R2 protein levels did not differ betweensites in liver (p = 0.87) or lung tissue (p = 0.70; data not shown; fordetails refer to the Appendix).

3.5. Global methylation and hydroxymethylation levels

Global 5-mC and 5-hmC levels were highly variable in liver andlungs from DCCO chicks from the same site (Fig. 4), and thus, outlierswere not identified. A 0.5-fold increase in 5-mC levels was measured inthe lungs of DCCO chicks from Pier 27 compared to Centre Island (p =0.0005); however, methylation levels in DCCO from Pier 27 and CentreIsland were not significantly different from the individuals from Mo-hawk Island (Fig. 4b). Global 5-mC and 5-hmC levels were not alteredin DCCO liver tissue from any of the sites (Fig. 4a,c).

4. Discussion

Many studies have shown that PACs are genotoxic and carcinogenic,especially in a laboratory setting, but less is known about how airbornePAC exposure affect natural populations of wildlife (Smith et al., 2007).Sanderfoot and Holloway (2017) indicated that birds exposed to PAC inair pollutants exhibited adverse health responses. Goldfinches (Car-duelis carduelis), for example, exposed to both gases and particulatesfrom coal-fired plants had induced pulmonary morphological andphysiological responses to air pollution (Llacuna et al., 1993, 1996).However, the molecular mechanisms of action by which airborne PACsare altering bird health are unclear. In this study, we showed thatDCCOs exposed to airborne PACs, are responding to treatments by ac-tivating their DNA repair and apoptotic gene machinery.

Our estimates of PAC exposure were complex. We found that thePier 27 colony was downwind more often of PAC sources from heavyindustry, including steel mills, more than the colony at Centre Island. Inaddition, the high PAC concentrations in air previously measured0.6 km from Pier 27 (King et al., 2014), suggest that DCCOs from Pier27 appear to have greater exposure to airborne PACs. However, we didnot find a significant difference in the hepatic burdens between the twoHamilton Harbour colonies despite the putative difference in airbornePAH exposure. PAC body burdens were, as expected, lowest in DCCOsfrom Mohawk Island in Lake Erie, supporting the classification of thissite as a reference site. Although environmental concentrations of PAHshave recently declined in Hamilton Harbour (Burniston et al., 2016),concentrations of PAHs in bile were positively correlated with ethox-yresorufin O-deethylase (EROD) activity in DCCOs from 1989 and 1990from Lake Ontario, including Hamilton (Bishop et al., 2016). However,exposure to complex mixtures of PACs often results in toxicity in a non-additive manner, which may be due to individual PAC as aryl hydro-carbon receptor (AHR) agonists or CYP1A inhibitors (reviewed inBilliard et al., 2008). In another study, CYP activity in DCCOs did notdiffer whether they were orally exposed to 5 or 10 mL Deep WaterHorizon crude oil/kg food containing a PAH mixture, but activity inboth treatments was significantly higher than control (Alexander et al.,2017). Therefore, an increase in airborne PAC exposure (from low tohigh) may not mean an equal increase in genetic response. Further,vehicular exhaust from an adjacent highway, despite being roughlyequidistant to both Hamilton colonies, may contribute to PAH airborneexposure that may mask PAH sources from industry. Other compoundsmay also be contributing to the induction of cytochrome P450s, such aspolychlorinated biphenyls (PCBs) or other dioxin-like compounds, al-though concentrations of these compounds in fish eating wildlife inHamilton Harbour have dropped significantly since the early 1980s to2014 (de Solla et al., 2016). The use of passive samplers at the eachcolony, or the use of bird-borne passive samplers (e.g., Sorais et al.,2017), would better characterize airborne exposures for either coloniesor even individual birds. Future work will entail the characterization ofexposure at these sites.

Nonetheless, diet may be a source of PAC exposure to DCCOs, andcould contribute to observable differences in the PAC-mediated meta-bolic pathways, and to body burdens. The reactivity of PACs allows thehepatic mixed function oxidase system (including cytochrome P450s) toeffectively metabolize and excrete these compounds, slowing the ac-cumulation in tissues of organisms higher in the food chain (Bromanet al., 1990). Bioaccumulation factors of individual parent PAHs forfish-eating birds have been estimated to range between 0.02 and 1.3(Kwok et al., 2013) and hence parent PAHs tend to undergo trophicdilution (Wan et al., 2007). However, alkylated PAHs have higher ac-cumulation potential than parent PAHs (Brandt et al., 2002), whichwould have a more significant dietary component to exposure. How-ever, diet is likely similar between DCCOs from the two HamiltonHarbour colonies. In a previous study of both highly bioaccumulativecompounds (perfluorooctane sulfonate; PFOS) and biodilutive sub-stances (e.g., perfluorophosphinates; PFPIAs) in the plasma of juvenile

Fig. 2. a) Mean sum PACs; b) Proportion contribution of dibenzothiophenes, alkylatedPAHs, parental PAHs, and naphthalene to ∑PACs in livers of DCCO from three colonies.The blue column represents the sum of parent and alkylated naphthalenes while thepurple column represents the sum of parent and alkylated dibenzothiophenes. MohawkIsland (reference site) had one outlier concentration.

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

180

DCCOs, the concentrations were identical between those from Pier 27and Farr Island (which is only 800 m from Centre Island; De Silva et al.,2016). Unlike PACs, where the concentrations of individual congeners

in Hamilton in 2014 were as high as 87.1 ng/m3 (this study), thehighest concentration of PFOS in a worldwide study using PUF diskswas 0.15 ng/m3 (Genualdi et al., 2010), indicating that the main source

Fig. 3. mRNA levels of genes in DCCO liver and lung tissue (n =7–9 per site) involved in P53 regulation (a-d), DNA repair (e-j),cell cycle arrest (i-l), and apoptosis (m-n). Data were normalizedwith the average expression of the reference genes and are re-presented as fold change± standard deviation relative toMohawk Island (reference site). Means without letters are notsignificantly different.

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

181

of PFOS to wildlife is dietary, not inhalation. The similarity in dietaryexposure to perfluorinated compounds suggests that DCCOs from thetwo Hamilton colonies have similar dietary exposure to PACs.

Expression patterns of genes of interest were consistent with theestimated airborne PAC exposure results. The expression level ofP53R2, involved in DNA repair, was higher in both liver and lung tissuefrom DCCO chicks at Pier 27 compared to the reference site. OGG1 didnot differ among sites however; and GADD45a only differed in lungtissue. Previously, high PAC concentrations were measured in the airnearby the Pier 27 DCCO colony that had a higher number of mutationsthan birds from a reference site (King et al., 2014), thus DCCOs fromPier 27 in this study may have a need for P53R2 for DNA repair. Si-milarly, P53R2 was over-expressed in breast cancer cells in response tobenzo[a]pyrene exposure (Ohno et al., 2008) and in lymphoblast cellsin response to exposure to known genotoxins, including PAHs (Mizotaet al., 2011). In addition, P53R2 was expressed in response to DNAdamage in non-small cell lung cancer tissues suggesting that P53R2could be a biomarker for cancerous tumour development after exposureto various types of environmental contaminants (Uramoto et al., 2006).PAH-DNA adducts can interrupt replication machinery and increase thenumber of mutations, requiring the cell to increase P53-induced P53R2levels to provide more nucleotides to repair the damaged DNA (Ohnoet al., 2008). In addition, the increased BAX mRNA levels in DCCO liverfrom Pier 27, suggests that the apoptotic pathway is activated(Robertson and Orrenius, 2000; Vogelstein et al., 2000) in individualswith higher airborne PAC exposure. PACs work through AhR/ARNTtoxicity and thus could induce apoptosis through a similar mechanism,such as BAX (a pro-apoptopic protein) promoting the release of cyto-chrome c (Robertson and Orrenius, 2000). Similarly, BAX transcriptlevels increased in human hepatic cells exposed to genotoxic carcino-gens (Lee et al., 2014; Nakanishi et al., 2000) and in mice exposed tobenzo[a]pyrene (Labib et al., 2012).

MDM2 is an ubiquitin protein that is in a tightly regulated negativefeedback loop where increased levels of P53 induce MDM2 expression

forming a complex that promotes the degradation of P53 (Meek, 2015).MDM2 expression increased in DCCO liver tissue from Centre Island,whereas MDM2 expression decreased in DCCO lung tissue from Pier 27compared to Mohawk Island showing that the response of the MDM2-P53 feedback loop may be tissue- and cell-type dependent (reviewed inAppella and Anderson, 2001; Vogelstein et al., 2000). In addition, thismay be a result of DCCOs from Centre Island having higher PAC con-centration in the liver on average thus increasing MDM2 levels. MDM2expression was upregulated in human lung cancer cells exposed tobenzo[a]pyrene and 1-nitropyrene (Nakanishi et al., 2000), in humanhepatic cells exposed to genotoxic carcinogens including benzo[a]pyrene (Lee et al., 2014), and in mice exposed to benzo[a]pyrene (Labibet al., 2012). However, even a slight decrease in MDM2 levels, as seenin lung tissue from DCCOs from Pier 27, can inhibit tumour develop-ment (Eischen and Boyd, 2012) by decoupling of the MDM2-P53 com-plex allowing for P53 to be transcriptionally active (Meek, 2015). De-spite the difference in MDM2 mRNA levels, P53 mRNA and proteinlevels did not differ in either the liver or lung between sites. In DCCOsin this study, the decrease in expression of MDM2 as a negative reg-ulator of P53 combined with the upregulation of genes in the DNArepair (P53R2) and apoptopic (BAX) pathway suggests that P53 is ac-tive. However, P53 degrades rapidly in the cell and the protein level canbe difficult to assess accurately (Vogelstein et al., 2000). P53 is knownto have many post-translational modifications that can help eitherstabilize the level of P53 protein in the cell or promote its use indownstream pathways (Appella and Anderson, 2001).

At the epigenetic level, global hypomethylation occurred in lungtissue of DCCOs from Centre Island compared to Pier 27, but did notdiffer from DCCOs from Mohawk Island. A decrease in methylationgenerally results in an increase in transcription, as seen in birds (Head,2014). Alteration of methylation states may lead to genomic instabilityand increased disease risk in humans (reviewed in Hou et al., 2012).Exposure to environmental contaminants has also altered methylationin birds; for example, hens exposed to cadmium through diet had higher

Fig. 4. Global 5-methylcytosine (5-mC) levels in a) liver and b)lung tissue and global 5-hydroxymethylcytosine (5-hmC) levels inc) liver and d) lung tissue of DCCO (n = 6 per site) exposed todifferent airborne PAC levels. Data are represented as foldchange± standard deviation relative to Mohawk Island. Globalmethylation level intensities of two representative individualsfrom each colony location are shown. Means with no letters arenot significantly different.

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

182

global methylation levels and higher DNMT mRNA levels in the liverand kidney compared to control birds (Zhang et al., 2009). Ad-ditionally, hypomethylation was correlated with the presence of benzo[a]pyrene-DNA adducts interfering with methyltransferases in bluegillsunfish (Shugart, 1990). The higher methylation levels in DCCO lung atPier 27 compared to Centre Island and the absence of a change in hy-droxymethylation levels could mean airborne PAHs in Hamilton Har-bour are interfering with DNMT's activity, but other potential epige-netic mechanisms could produce the same response (Hou et al., 2012).Although immunostaining is an advantageous method to understandthe overall methylation signature, information on the methylation levelof specific genes is lost which could be useful for explaining the mRNAchanges seen (Kamstra et al., 2015).

Overall, evidence suggests that both the DNA repair and apoptotic-related gene expression levels were altered in birds from sites with highairborne PACs (534–355 ng/m3). It is acknowledged that free-rangingwildlife are exposed to complex mixtures of contaminants so our resultsalso include a response from exposure to a variety of other agonists andinhibitor agents. However, our study still provides evidence that

natural populations of DCCOs living near highly industrialized areas areactivating a molecular response to mutagens.

Acknowledgments

Dr. Jing Zhang helped with the DNA methylation dot blot protocol.Glenn Barrett, Laura King, Liam Graham, Sarah Verbaan, ShannonSimpkins, Kyna Intini, and Kimberly O’Hare helped with sample col-lection. This work was supported by the Strategic TechnologyApplications of Genomics in the Environment (STAGE) fromEnvironmental and Climate Change Canada to SdS, PJT, and VSL andby the Canada Research Chair in Ecotoxicogenomics and EndocrineDisruption to VSL.

Declaration of interest

There were no financial, professional, or personal conflicts of in-terest with any author listed on this manuscript.

Appendix

PAC analysis in the air

Pictures of the two colonies in Hamilton Harbour, ON relative to the industry in the area are shown in Fig. A.1. Prior to their deployment, PUFdisks were pre-cleaned by sonication in deionized water, and then placed into pre-cleaned Accelerated Solvent Extraction (ASE) cells for solventextraction according to the method described in Harner et al. (2013). Field blanks were collected to assess potential contamination, and consisted ofa clean PUF disk that is deployed in the sampler housing for approximately one minute, and then subsequently treated the same as the rest of the PUFdisks. Prior to the extraction with ASE, the PUF disks were spiked with labelled standards d10-Acenaphthene, d10-Anthracene, d12-Benz(a)anthra-cene, d12 Chrysene, d12-Benzo(b)fluoranthrene, d12-Benzo(e)pyrene, d12-Indeno(123-cd)pyrene, d14-Dibenza(ah)anthracene, d10-Benzo(b)naphtho(2,1-d)-thiophene, d12−2,6-Dimethylnaphthalene in order to determine surrogate recoveries.

The sample extracts were spiked with isooctane, concentrated using rotary evaporation, and were cleaned and fractionated on a column (ID0.9 cm) with 4 g of activated silica, 2.5 g of activated alumina and 2 g of sodium sulfate. Fraction 1 was eluted with 20 mL petroleum ether (PE),fraction 2 with 20 mL PE/acetone (50/50, v/v), and fraction 3 with 20 mL methanol. Only fraction 2 was used for further sample analysis, whichcontains the target analytes. Fraction 2 was reduced down to 1 mL using a gentle stream of nitrogen, and C13-Phenanthrene (10 ng/µL) was added toall samples as an internal standard prior to its analysis. Sample extracts were analyzed using a gas chromatography (GC) system (Agilent 6890) witha mass selective (MS) detector (Agilent 5975) on a DB-XLB column (30 m, i.d. of 250 µm, 0.25 µm film thickness, J&W Scientific). The GC/MS wasoperated using electron impact (EI) in selected ion monitoring (SIM) mode.

Individual calibration curves were generated for quantifying parent PAHs and dibenzothiophene, and for the alkylated PAHs and alkylated

DCCO

DCCO

A

B

Fig. A.1. Locations of the DCCO colonies at CentreIsland (A) and Pier 27 (B), in Hamilton Harbour,Ontario. The DCCO icon shows the location of therespective colonies, and the arrows in both picturesshow the secondary-intercardinal direction WSW(~240°). Both photos were taken within 90 m ofeach other; at a 27 and 36 mm focal length, respec-tively (hence B is at a 1.3 x zoom compared to A).The relative proximity of the DCCO colony to theheavily industrialized area is apparent in (B). Notethat (B) is a composite of two adjacent pictures.

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

183

Table A.1Mean Method Detection Limits (MDL) for PACs analyzed in DCCO livers (w.w.).

Compound Mean MDL (ng/g)

Biphenyl 0.0420Naphthalene 0.0730Acenaphthene 0.0599Acenaphthylene 0.0682Anthracene 0.0189Benz[a]anthracene 0.0118Benzo[a]pyrene 0.0328Benzo[b]fluoranthene 0.0222Benzo[e]pyrene 0.0305Benzo[ghi]perylene 0.0198Benzo[j,k]fluoranthenes 0.0257Chrysene 0.0122Dibenz[a,h]anthracene 0.0215Dibenzothiophene 0.0192Fluoranthene 0.0153Fluorene 0.0216Indeno[1,2,3-cd]pyrene 0.0210Perylene 0.0337Phenanthrene 0.0174Pyrene 0.0151Retene 0.05401,2,6-Trimethylphenanthrene 0.02591,2-Dimethylnaphthalene 0.08611,4,6,7-Tetramethylnaphthalene 0.04411,7-Dimethylfluorene 0.03741,7-Dimethylphenanthrene 0.01781,8-Dimethylphenanthrene 0.01791-Methylchrysene 0.01141-Methylnaphthalene 0.05401-Methylphenanthrene 0.02452,3,5-Trimethylnaphthalene 0.04362,3,6-Trimethylnaphthalene 0.04412,4-Dimethyldibenzothiophene 0.02342,6-Dimethylnaphthalene 0.07002,6-Dimethylphenanthrene 0.01792/3-Methyldibenzothiophenes 0.03282-Methylanthracene 0.02562-Methylfluorene 0.05112-Methylnaphthalene 0.05142-Methylphenanthrene 0.02483,6-Dimethylphenanthrene 0.01813-Methylfluoranthene/Benzo[a]fluorene 0.03413-Methylphenanthrene 0.02505,9-Dimethylchrysene 0.01935/6-Methylchrysene 0.01237-Methylbenzo[a]pyrene 0.03459/4-Methylphenanthrene 0.0250C1 Phenanthrenes/Anthracenes 0.0245C1-Acenaphthenes 0.0364C1-Benzo[a]anthracenes/Chrysenes 0.0118C1-Benzofluoranthenes/Benzopyrenes 0.0345C1-Biphenyls 0.0260C1-Dibenzothiophenes 0.0328C1-Fluoranthenes/Pyrenes 0.0341C1-Fluorenes 0.0511C1-Naphthalenes 0.0514C2 Phenanthrenes/Anthracenes 0.0179C2-Benzo[a]anthracenes/Chrysenes 0.0193C2-Benzofluoranthenes/Benzopyrenes 0.0314C2-Biphenyls 0.0352C2-Dibenzothiophenes 0.0234C2-Fluoranthenes/Pyrenes 0.0320C2-Fluorenes 0.0374C2-Naphthalenes 0.0861C3-Benzo[a]anthracenes/Chrysenes 0.0184C3-Dibenzothiophenes 0.0253C3-Fluoranthenes/Pyrenes 0.0327C3-Fluorenes 0.0473C3-Naphthalenes 0.0439C3-Phenanthrenes/Anthracenes 0.0259C4-Benzo[a]anthracenes/Chrysenes 0.0189C4-Dibenzothiophenes 0.0241C4-Fluoranthenes/Pyrenes 0.0307C4-Naphthalenes 0.0441C4-Phenanthrenes/Anthracenes 0.0540

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

184

dibenzothiophenes. One to three individual standards were used to semi-quantify each homolog group (C1, C2, C3, C4). The responses and con-centrations of the individual alkyl homologs were summed and used to generate a calibration curve for each respective homolog group. When thetotal response area of the homolog group was above the highest concentration in the calibration curve, the corresponding concentration wasextrapolated from the linear region.

All samples were blank-corrected by subtracting the field blank mean (n = 3) from each sample. Samples were also recovery corrected based onthe surrogate recoveries for each sample. The method detection limit (MDL) was estimated as the mean of the blanks + 3 SD. Total PACs (ΣPACs)represented the sum of 75 analytes, including 17 parent PAHs, 20 C1-C4–alkylated PAHs, dibenzothiophenes, and four of their C1-C4 analogs, andnaphthalene. PAC quantities accumulated in the PUF disk (ng/PUF) were converted to concentrations in air using effective air sample volumes asdescribed in Harner et al. (2013).

PAC analysis in the liver

For each chick, a portion of their liver was homogenized, dried with anhydrous Na2SO4, and spiked with 16 perdeuterated surrogate PAHstandards. Liver samples (5–6 g) were then Soxhlet extracted with dichloromethane (DCM) for 18 h, and the remainder was oven dried to determinemoisture content. The extract was then concentrated to ∼2 mL and solvent exchanged into hexane for cleanup. Two PAC fractions were extracted;the first alkane fraction was obtained after elution on a silica column with pentane and subsequently discarded. The aromatic fraction was obtainedafter elution with dichloromethane. This extract was analyzed, after concentration and addition of recovery standards, by low-resolution massspectrometry (LRMS) using an RTX-5 capillary GC column. The LRMS was operated at a unit mass resolution in the electron impact (EI) ionizationmode using multiple ion detection (MID) acquiring at least one characteristic ion for each target analyte and surrogate standard. Concentrations oftarget PACs were calculated using the isotope dilution method of quantification. Most of the alkylated PAHs were determined using response factorsbased on a multipoint calibration using, when available, individual alkylated PAH standards, whereas other alkylated PAHs were determined bysingle-point calibration. Compounds were quantified by comparing the area of the quantification ion to that of the corresponding deuterium-labelledstandard and correcting for response factors. For all samples where analytes were above the detection limit, concentrations of the analytes were first

Fig. A.2. Profile of PAHs and alkylated PAHs in DCCO liver from Pier 27 and Centre Island (Hamilton Harbour) and Mohawk Island (Lake Erie).

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

185

blank subtracted, and then if the value was less than 3 times the level of the sample blank, it was treated as below detection limits (Table A1). Theprofile of total PAHs and alkylated PAHs in livers from the three colonies can be seen in Fig. A.2.

Cloning and sequencing of genes

All primers were designed by aligning sequences of closely related species (National Centre for Biotechnology Information, NCBI, http://www.ncbi.nlm.nih.gov/) in BioEdit (Ibis Biosciences, Carlsbad, CA, USA), with the exception of both rpl8 and eef1a, which were designed by others(Crump et al., 2016). Primers were tested on DCCO liver samples using PCR on a Mastercycler Pro S (Eppendorf Canada, Mississauga, ON, CAN) andvisualized on a 1% agarose gel. The band of the correct size was excised from the gel and the product was extracted using e.Z.N.A. Gel Extraction Kit(Omega Bio-Tek, Norcross, GA, USA). The product was cloned using the pGEM-T Easy Vector System (Promega, Madison, WI, USA) for ligation andJM109 High Efficiency Competent Cells for transformation. The ligated product was plated on LB plates and incubated for 12–16 h. Colonies wereconfirmed through PCR and the plasmid was extracted using the e.Z.N.A. Plasmid Mini Kit (Omega Bio-Tek, Norcross, GA, USA). Products weresequenced at Robarts Research Institute (London, ON, CA) and sequences were confirmed to be the correct gene and species by using the Basic Local

Table A.2List of genes and their main biological function used to assess gene expression in DCCO liver and lung samples. Primers were designed for DCCO in this study except if indicated by anasterisk (*), which were designed by Crump et al. (2016).

Overallfunction

Target gene Main biological Function Sequence (5’−3’) Annealingtemperature (°C)

Amplicon size(bp)

[Primer] (nM)

Tumour P53 Tumour suppressor F GAGCACGTGGCCGAAGTG 64 143 250regulation Tumour-suppressing protein R CGCTTGGTGGTCTCGTCGTC 250

MDM2 Targets P53 forproteasomal

F CCTAGATGCTGGACCCTTCG 60 240 150

E3 ubiquitin-protein ligase degradation R GCTAGATGTTGATGGCTGAG 150DNA repair P53R2 Produces nucleotides for

DNAF CAGACTGGGCTCTGAAGTGG 58 150 400

ribonucleotide reductase repair R GAGTCAGTCCAGGCATCAGG 400OGG1 DNA repair enzyme F TGGTACACGGTGTACGGC 58 118 6508-oxoguanine glycosylase R CAGTCAGTCCCACGTCCA 550

DNA repair/ GADD45α Growth arrest and DNA-damage

F AACGACATCAACATCCTGCG 62 105 325

Cell cycle arrest growth arrest and DNA damage-inducible protein

induced R GACCAGGACGCAGTGGAG 325

Cell cycle P21 Regulator of cell cycleprogression

F GCGCTGGAACTTCAACTTTG 58 160 350

arrest cyclin-dependent kinase inhibitor1

at G1 R CTTAGGCAGGACCTTGTGGG 350

Apoptosis BAX Pro-apoptopic regulator F CCATGGAGTACATGCGGGAG 58 148 350Bcl−2-associated X protein R TTCCAGATGGTAAGCGAGGC 350

Reference ACTβ Structural framework ofcells

F GGCCATCAGGGAGTTCATAG 58 146 350

genes beta-actin R CTACAGCTTCACCACCACAG 350RPL8 Structural constituent of

ribosomeF GTTTACTGTGGCAAGAAAGC 58 147 350

ribosomal protein L8 R GACAGTGGCATAGTTTCCAG 350RPL4* Structural constituent of

ribosomeF AGAAGCGTTACGCCATCTGT 60 208 400

ribosomal protein L4 R CGCTGGGAGGCATAAACCTT 400EEF1α1* Delivery of aminoacyl

tRNAs toF CTCTGCGTCTGCCTCTTCAA 60 102 400

elongation factor 1 ribosomes R TAACCACCATGCCAGGCTTC 400

Fig. A.3. P53 protein level in DCCO a) liver; and P53R2 protein level in DCCO b) liver; and c) lung tissue from sites differing in airborne PAC exposure (n = 6 per site). Data arerepresented as fold change± standard deviation relative to Mohawk Island. Protein level intensities of two individuals from each colony location are shown.

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

186

Alignment Search Tool (BLAST) from NCBI. DCCO specific primers were designed using these sequences with an optimal product size of 90–110 bpfor qPCR (Table A.2).

Protein quantification

Proteins were extracted by homogenizing the liver and lung tissue in lysis buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% Triton X-100,0.5% sodium deoxycholate, 0.1% SDS, 1 mM sodium orthovanadate, 1 mM NaF, 1 mM EDTA) with the addition of PMSF crystals as a proteaseinhibitor using the Retsch Mixer Mill MM400 (Fisher Scientific, Mississauga, ON, CAN) for two minutes at 20 Hz. Samples were centrifuged at 4 °Cfor 15 min at 10,000 ×g and total proteins were collected from the supernatant. The Pierce 660 nm Protein Assay (ThermoFisher Scientific,Mississauga, ON, CAN) was used to determine protein concentration. Loading buffer (4x Laemmli sample buffer, BioRad and 10% β-mercap-toethanol) was added to each sample at a 1:4 ratio and samples were boiled at 95 °C for five minutes. Samples containing 20 −40 µg total proteinwere loaded in Mini-PROTEAN TGX 4–15% Stain-Free gels (Bio-Rad Laboratories, Mississauga, ON, CAN) and separated for 30 min at 200 V in aTris/Glycine/SDS buffer (Bio-Rad Laboratories, Mississauga, ON, CAN). Proteins were transferred to a PVDF membrane using the Trans-Blot Turbosystem (Bio-Rad Laboratories, Mississauga, ON, CAN). Membranes were blocked with 5% milk for 1 h, incubated with the appropriate 1° antibody(1:200 P53, 1:500 P53R2, Santa Cruz Biotechnology, Inc., Dallas, TX, USA) overnight at 4 °C, and incubated with the appropriate 2° antibody (anti-rabbit 1:2000 P53, anti-goat 1:2000 P53R2; BioShop Canada Inc., Burlington, ON, CAN) for 45 min on a rocking platform. Membranes were washed3 x five min with TBST buffer (20 mM Tris pH7.5, 150 mM NaCl, 0.1% Tween 20) in between each step. Protein levels were visualized using ClarityWestern ECL Substrate (Bio-Rad Laboratories, Mississauga, ON, CAN) for 5 min and the image was captured through chemiluminesence using theChemiDoc XRS+ Imaging System (Bio-Rad Laboratories, Mississauga, ON, CAN). Protein levels are expressed as fold changes± SD relative to theprotein level in the samples from the reference site (Fig. A.3).

Estimating methylation levels

Similar to the protein quantification, a linear dynamic range was established by blotting 150–400 ng DNA (5-mC) and 25 – 200 ng DNA (5-hmC)and comparing to the intensity of un-methylated and 100% methylated DNA. 200 ng DNA for 5-mC had the highest chemiluminesence intensitydetected normalized by the DNA intensity without saturation of DNA on the membrane (i.e. decrease in DNA intensity at 400 ng; Fig. A.4). 200 ngwas also within the range of 0–100% methylated DNA (Fig. A.4B). To make a non-methylated DNA control, DNA from DCCOs from all three siteswere pooled and purified using the Illustra Ready-To-Go GenomiPhi V3 DNA amplification kit (GE Healthcare Life Sciences, Mississauga, ON, CAN)

y = -730.66x + 665561R² = 0.11760

200000

400000

600000

800000

1000000

1200000

0 100 200 300 400

DNA

load

ing

inte

nsity

Amount of DNA loaded (ng)

y = 0.0148x + 1.0857R² = 0.5213

0

2

4

6

8

10

12

0 100 200 300 400

Ra�o

of c

hem

ilum

ines

cent

inte

nsity

/ DN

A lo

adin

g in

tens

ity

Amount of DNA loaded (ng)

DCCO samples

unmethylated

100% methylated

Fig. A.4. Establishing the linear dynamic range for methylation levels by loading 150 – 400 ng DNA resulting in decreasing trends for A) DNA loading intensity and increasing trends forB) chemiluminescent intensity normalized to the DNA loading.

y = 0.0354x - 0.0261R² = 0.9879

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

0 20 40 60 80 100

Fold

chan

ge co

mpa

red

to 2

5%

met

hyla

ted

sam

ples

Percent of methylated DNA

DCCO samples

Fig. A.5. Standard curve of fold change of intensity with increasing percentage of methylated DNA. Since 0% methylated DNA did not give off a signal higher than the background, foldchange is relative to 25% methylated sample.

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

187

and Zymo's Genomic DNA Clean and Concentrator Kit (Zymo Research Corp., Irvine, CA, USA). To make the 100% methylated DNA control, 800 ngof the non-methylated DNA was artificially methylated by incubating at 30 oC for 4 h in a reaction containing 1x CpG buffer, 600 µM S-adeno-sylmethionine (SAM), and 4 units CpG methylase. The authors understand that the sample may not be completely 100% methylated but it was usedas the upper limit. A standard curve of 100%, 75%, 50%, 25%, and 0% methylation was used to validate 5-mC levels and confirm that a maximum of3.5 fold change (75% increase) can be detected with this method (Fig. A5).

Data analysis

To adhere to normality and equal variances before performing further analyses the following transformations were performed: mRNA levels of allgenes for the liver were log10 transformed except for BAX with a Box-Cox (λ = 1.8) transformation, BAX, GADD45α, and P53R2 mRNA levels in thelung were transformed using Box-Cox (λ = −0.6, −0.4, log(mRNA)*1.18, respectively), P53 protein levels in the liver were log10 transformed, and5-hmC levels in the lung were log10 transformed.

References

Abedin, Z., Louis-Juste, M., Stangl, M., Field, J., 2013. The role of base excision repairgenes OGG1, APN1 and APN2 in benzoâ¸pyrene-7,8-dione induced p53 mutagenesis.Mutat. Res. 750, 121–128.

ACGIH (American Conference of Governmental Industrial Hygienists), 2005. Polycyclicaromatic hydrocarbons (PAHs) biologic exposure indices (BEI) Cincinnati. OH:American Conference of Governmental Industrial Hygienists.

Achten, C., Andersson, J.T., 2015. Overview of polycyclic aromatic compounds (PAC).Polycycl. Aromat. Compd. 35, 177–185.

Alexander, C.R., Hooper, M.J., Cacela, D., Smelker, K.D., Calvin, C.S., Dean, K.M.,Bursian, S.J., Cunningham, F.L., Hanson-Dorr, K.C., Horak, K.E., Isanhart, J.P., Link,J., Shriner, S.A., Godard-Codding, C.A.J., 2017. CYP1A protein expression and cat-alytic activity in double-crested cormorants experimentally exposed to deepwaterHorizon Mississippi Canyon 252 oil. Ecotoxicol. Environ. Saf. 142, 79–86.

Appella, E., Anderson, C.W., 2001. Post-translational modifications and activation of p53by genotoxic stresses. Eur. J. Biochem. 268, 2764–2772.

Baccarelli, A., Bollati, V., 2009. Epigenetics and environmental chemicals. Curr. Opin.Pediatr. 21, 243–251.

Billiard, S.M., Meyer, J.N., Wassenberg, D.M., Hodson, P.V., Di Giulio, R.T., 2008.Nonadditive effects of PAHs on early vertebrate development: mechanisms and im-plications for risk assessment. Toxicol. Sci. 105, 5–23.

Bishop, C.A., Trudeau, S., Janz, D., Hughes, K.D., 2016. Contaminant concentrations andbiomarkers in 21-day old herring gulls (Larus argentatus) and double-crested cor-morants (Phalacrocorax auritus) from eastern Lake Ontario, and from HamiltonHarbour in western Lake Ontario in 1989 and 1990. Aquat. Ecosyst. Health Manag.19, 181–191.

Boström, C.-E., Gerde, P., Hanberg, A., Jernström, B., Johansson, C., Kyrklund, T.,Rannug, A., Törnqvist, M., Victorin, K., Westerholm, R., 2002. Cancer risk assess-ment, indicators, and guidelines for polycyclic aromatic hydrocarbons in the ambientair. Environ. Health Perspect. 110, 451–488.

Brandt, C.A., Becker, J.M., Porta, A., 2002. Distribution of polycyclic aromatic hydro-carbons in soils and terrestrial biota after a spill of crude oil in Trecate, Italy. Environ.Toxicol. Chem. 21, 1638–1643.

Broman, D., Näf, C., Lundbergh, I., Zebühr, Y., 1990. An in situ study on the distribution,biotransformation and flux of polycyclic aromatic hydrocarbons (PAHs) in an aquaticfood chain (Seston - Mytilus edulis L. – Somateria mollissima L.) from the Baltic: anecotoxicological perspective. Environ. Toxicol. Chem. 9, 429–442.

Burniston, D.A., Jia, J., Charlton, M.N., Thiessen, L., McCarry, B.E., Marvin, C.H., 2016.Trends in Hamilton Harbour suspended sediment quality. Aquat. Ecosyst. HealthManag. 19, 141–149.

Bustin, S.A., Benes, V., Garson, J.A., Hellemans, J., Huggett, J., Kubista, M., Mueller, R.,Nolan, T., Pfaffl, M.W., Shipley, G.L., Vandesompele, J., Wittwer, C.T., 2009. TheMIQE guidelines: minimum information for publication of quantitative real-time PCRexperiments. Clin. Chem. 55, 611–622.

Canadian Council of Ministers of the Environment, 1999. Canadian Sediment QualityGuidelines for the Protection of Aquatic Life: Polycyclic Aromatic Hydrocarbons(PAHs). Canadian Environmental Quality Guidelines, 1999. Canadian Council ofMinisters of the Environment, Winnipeg, MB, Canada.

Canadian Environmental Protection Act, 1994. Polycyclic aromatic hydrocarbons.Priority Substances List Assessment Report. En40-215/42E.

Corrales, J., Fang, X., Thornton, C., Mei, W., Barbazuk, W.B., Duke, M., Scheffler, B.E.,Willett, K.L., 2014. Effects on specific promoter DNA methylation in zebrafish em-bryos and larvae following benzo[a]pyrene exposure. Comp. Biochem. Physiol. C163, 37–46.

Crump, D., Farhat, A., Chiu, S., Williams, K.L., Jones, S.P., Langlois, V.S., 2016. Use of anovel double-crested cormorant ToxChip PCR array and the EROD assay to determineeffects of environmental contaminants in primary hepatocytes. Environ. Sci. Technol.50, 3265–3274.

Cruz-Martinez, L., Fernie, K.J., Soos, C., Harner, T., Getachew, F., Smits, J.E.G., 2015.Detoxification, endocrine, and immune responses of tree swallow nestlings naturallyexposed to air contaminants from the Alberta oil sands. Sci. Total Environ. 502, 8–15.

Custer, T.W., Bunck, C., 1992. Feeding flights of breeding double-crested cormorants at 2Wisconsin colonies. J. Field Ornithol. 63, 203–211.

De Silva, A.O., Spencer, C., Ho, K.C.D., Al Tarhuni, M., Houde, M., de Solla, S.R., Lavoie,R., King, L.E., Muir, D.C.G., Fair, P., Wells, R., Go, C., 2016. Perfluorophosphinates in

northern pike (Esox lucius), double-crested cormorants (Phalacrocorax auritus), andbottlenose dolphins (Tursiops truncatus) in relation to other perfluoroalkyl acids.Environ. Sci. Technol. 50, 10903–10913.

de Solla, S.R., Weseloh, D.V.C., Hughes, K.D., Moore, D.J., 2016. 40 year decline of or-ganic contaminants in eggs of herring gulls (Larus argentatus) from the great lakes,1974 to 2013. Waterbirds 39, 171–180.

Eischen, C.M., Boyd, K., 2012. Decreased Mdm2 expression inhibits tumor developmentand extends survival independent of arf and dependent on p53. Plos One 7, e46148.

Fang, X., Thornton, C., Scheffler, B.E., Willett, K.L., 2013. Benzo[a]pyrene decreasesglobal and gene specific DNA methylation during zebrafish development. Environ.Toxicol. Pharmacol. 36, 40–50.

Genualdi, S., Lee, S.C., Shoeib, M., Gawor, A., Ahrens, L., Harner, T., 2010. Global pilotstudy of legacy and emerging persistent organic pollutants using sorbent-im-pregnated polyurethane foam disk passive air samplers. Environ. Sci. Technol. 44,5534–5539.

Hall, J.D., O'Connor, K.M., 2016. Hamilton harbour remedial action plan process: con-necting science to management decisions. Aquat. Ecosyst. Health Manag. 19,107–113.

Hamilton Air Monitoring Network (HAMN), 2016. ⟨http://www.hamnair.ca/⟩.Harner, T., Su, K., Genualdi, S., Karpowicz, J., Ahrens, L., Mihele, C., Schuster, J.,

Charland, J.-P., Narayan, J., 2013. Calibration and application of PUF disk passive airsamplers for tracking polycyclic aromatic compounds (PACs). Atmos. Environ. 75,123–128.

Head, J.A., 2014. Patterns of DNA methylation in animals: an ecotoxicological perspec-tive. Int. Comp. Biol. 54, 77–86.

Hou, L., Zhang, X., Wang, D., Baccarelli, A., 2012. Environmental chemical exposures andhuman epigenetics. Int J. Epidemiol. 41, 79–105.

IARC, 2010. Some non-heterocyclic polycyclic aromatic hydrocarbons and some relatedexposures. IARC Monographs on the Evaluations of Carcinogenic Risks to Humans 92World Health Organization, Lyon, France.

Kamstra, J.H., Aleström, P., Kooter, J.M., Legler, J., 2015. Zebrafish as a model to studythe role of DNA methylation in environmental toxicology. Environ. Sci. Pollut. Res.22, 16262–16276.

Kim, K.-H., Jahan, S.A., Kabir, E., Brown, R.J.C., 2013. A review of airborne polycyclicaromatic hydrocarbons (PAHs) and their human health effects. Environ. Int. 60,71–80.

King, L.E., de Solla, S.R., Small, J.M., Sverko, E., Quinn, J.S., 2014. Microsatellite DNAmutations in double-crested cormorants (Phalacrocorax auritus) associated with ex-posure to PAH-containing industrial air pollution. Environ. Sci. Technol. 48,11637–11645.

King, L.E., de Solla, S.R., Marentette, J.R., Lavoie, R.A., Kyser, T.K., Campbell, L.M., Arts,M.T., Quinn, J.S., 2017. Fatty acids, stable isotopes, and regurgitate reveal diet dif-ferences between Lake Ontario and Lake Erie double-crested cormorants(Phalacrocorax auritus). J. Gt. Lakes Res. 43, 132–140.

Kisseljova, N.P., Kisseljov, F.L., 2005. DNA demethylation and carcinogenesis. Biochem70, 900–911.

Kwok, C.K., Liang, Y., Leung, S.Y., Wang, H., Dong, Y.H., Young, L., Giesy, J.P., Wong,M.H., 2013. Biota-sediment accumulation factor (BSAF), bioaccumulation factor(BAF), and contaminant levels in prey fish to indicate the extent of PAHs and OCPscontamination in eggs of waterbirds. Environ. Sci. Pollut. Res. 20, 8425–8434.

Labib, S., Yauk, C., Williams, A., Arlt, V.M., Phillips, D.H., White, P.A., Halappanavar, S.,2012. Subchronic oral exposure to benzo(a)pyrene leads to distinct transcriptomicchanges in the lungs that are related to carcinogenesis. Toxicol. Sci. 129, 213–224.

Lee, W.J., Kim, S.C., Lee, S.J., Lee, J., Park, J.H., Yu, K.-S., Lim, J., Kwon, S.W., 2014.Investigating the different mechanisms of genotoxic and non-genotoxic carcinogensby a gene set analysis. PLOS One 9 (e86700).

Llacuna, S., Gorriz, A., Durfort, M., Nadal, J., 1993. Effects of air pollution on passerinebirds and small mammals. Arch. Environ. Contam. Toxicol. 24, 59–66.

Llacuna, S., Gorriz, A., Riera, M., Nadal, J., 1996. Effects of air pollution on hematologicalparameters in passerine birds. Arch. Environ. Contam. Toxicol. 31, 148–152.

Meek, D.W., 2015. Regulation of the p53 response and its relationship to cancer.Biochem. J. 469, 325–346.

Mizota, T., Ohno, K., Yamada, T., 2011. Validation of a genotoxicity test based on p53R2gene expression in human lymphoblastoid cells. Mut. Res. /Gen. Toxicol. Environ.Mutagenesis 724, 76–85.

Nakanishi, Y., Pei, X.-H., Takayama, K., Bai, F., Izumi, M., Kimotsuki, K., Inoue, K.,Minami, T., Wataya, H., Hara, N., 2000. Polycyclic aromatic hydrocarbons

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

188

carcinogens increase ubiquitination of p21 protein after the stabilization of p53 andthe expression of p21. Am. J. Respir. Cell Mol. Biol. 22, 747–754.

Ohno, K., Ishihata, K., Tanaka-Azuma, Y., Yamada, T., 2008. A genotoxicity test systembased on p53R2 gene expression in human cells: assessment of its reactivity to var-ious classes of genotoxic chemicals. Mut. Res. 656, 27–35.

Pfeifer, G.P., Xiong, W., Hahn, M.A., Jin, S.-G., 2014. The role of 5-hydro-xymethylcytosine in human cancer. Cell Tissue Res. 356, 631–641.

Robertson, J.D., Orrenius, S., 2000. Molecular mechanisms of apoptosis induced by cy-totoxic chemicals. Crit. Rev. Toxicol. 30, 609–627.

Robles, A.I., Linke, S.P., Harris, C.C., 2002. The p53 network in lung carcinogenesis.Oncogene 21, 6898–6907.

Ruiz-Hernandez, A., Kuo, C.-C., Rentero-Garrido, P., Tang, W.-Y., Redon, J., Ordovas,J.M., Navas-Acien, A., Tellez-Plaza, M., 2015. Environmental chemicals and DNAmethylation in adults: a systematic review of the epidemiologic evidence. Clin.Epigenetics 7, 55.

Sanderfoot, O.V., Holloway, T., 2017. Air pollution impacts on avian species via in-halation exposure and associated outcomes. Environ. Res. Lett. 12, 083002.

Shugart, L.R., 1990. 5-methyl deoxycytidine content of DNA from bluegill sunfish(Lepomis macrochirus) exposed to benzo[a]pyrene. Environ. Toxicol. Chem. 9,205–208.

Sicolo, M., Tringali, M., Orsi, F., Santagostino, A., 2009. Porphyrin pattern and methe-moglobin levels in Columba livia applied to assess toxicological risk by air pollution inurban areas. Arch. Environ. Contam. Toxicol. 57, 732–740.

Smith, P.N., Cobb, G.P., Godard-Codding, C., Hoff, D., McMurry, S.T., Rainwater, T.R.,Reynolds, K.D., 2007. Contaminant exposure in terrestrial vertebrates. Environ.Pollut. 150, 41–64.

Somers, C.M., McCarry, B.E., Malek, F., Quinn, J.S., 2004. Reduction of particulate airpollution lowers the risk of heritable mutations in mice. Science 304, 1008–1010.

Sorais, M., Rezaei, A., Okeme, J.O., Diamond, M.L., Izquierdo, R., Giroux, J.F., Verreault,J., 2017. A miniature bird-borne passive air sampler for monitoring halogenatedflame retardants. Sci. Total Environ. 599–600, 1903–1911.

Uramoto, H., Sugio, K., Oyama, T., Hanagiri, T., Yasumoto, K., 2006. p53r2, p53 in-ducible ribonucleotide reductase gene, correlated with tumor progression of non-small cell lung cancer. Anticancer Res. 26, 983–988.

Vineis, P., Husgafvel-Pursiainen, K., 2005. Air pollution and cancer: biomarker studies inhuman populations. Carcinogenesis 26, 1846–1855.

Vogelstein, B., Lane, D., Levine, A.J., 2000. Surfing the p53 network. Nature 408,307–310.

Wan, Y., Jin, X., Hu, J., Jin, F., 2007. Trophic dilution of polycyclic aromatic hydro-carbons (PAHs) in a marine food web from Bohai Bay, North China. Environ. Sci.Technol. 41, 3109–3114.

Wilson, V.L., Jones, P.A., 1983. Inhibition of DNA methylation by chemical carcinogens invitro. Cell 32, 239–246.

Xin, L., Wang, J., Wu, Y., Guo, S., 2015. The development of GADD45α luciferase reporterassays in human cells for assessing the genotoxicity of environmental pollutants.Toxicol. Mech. Methods 25, 136–142.

Yauk, C.L., Quinn, J.S., 1996. Multilocus DNA fingerprinting reveals high rate of heritablegenetic mutation in herring gulls nesting in an industrialized urban site. Proc. Natl.Acad. Sci. USA 93, 12137–12141.

Zhang, J., Fu, Y., Li, J., Wang, J., He, B., Xu, S., 2009. Effects of subchronic cadmiumpoisoning on DNA methylation in hens. Environ. Toxicol. Pharm. 27, 345–349.

S.J. Wallace et al. Ecotoxicology and Environmental Safety 150 (2018) 176–189

189