Target Organ Toxicity in Rats After Subchronic Oral Exposure to Soil Extracts Containing a Complex Mixture of Contaminants

Complex mixtures of unknown contaminants present a challenge to identify toxicological risks without using large numbers of animals and labor-intensive screens of all organs. This study examined soil extracts from a legacy-contaminated pesticide packaging and blending site. HepG2 cytotoxicity was used as an initial screen of 18 soil samples; then, three extracts (A, B and C) from different locations at the study site were used for testing in animals. The first two extracts were identified as the most toxic in vitro, and the latter extract obtained from a location further from these two toxic sampling sites. Then, target organ toxicities were identified following biweekly oral gavage for one month of three soil extracts (0.1% in polyethylene glycol or PEG) compared to vehicle control in male Sprague–Dawley rats (n = 9–10/group). Exposure to extract A significantly increased neutrophils and lymphocytes compared to control. In contrast, all extracts increased plasma α−2 macroglobulin and caused mild-to-moderate lymphocytic proliferation within the spleen white pulp, all indicative of inflammation. Rats exposed to all soil extracts exhibited acute tubular necrosis. Cholinesterase activity was significantly reduced in plasma, but not brain, after exposure to extract A compared to control. Increased hepatic ethoxyresorufin-o-deethylase activity compared to control was observed following exposure to extracts A and B. Exposure to soil extract C in rats showed a prolonged QTc interval in electrocardiography as well as increased brain lipid peroxidation. Candidate contaminants are organochlorine, organophosphate/carbamate pesticides or metabolites. Overall, HepG2 cytotoxicity did not successfully predict the neurotoxicity and cardiotoxicity observed with extract C but was more successful with suspected hydrocarbon toxicities in extracts A and B. Caution should be taken when extrapolating the observation of no effects from in vitro cell culture to in vivo toxicity, and better cell culture lines or assays should be explored.

assessment of contaminants is usually determined through the component-based risk (CBR) assessment approach, where individual chemicals are considered (Coors et al. 2018;Posthuma et al. 2019). However, CBR assessments fail to consider that contaminants almost always occur in mixtures. In a CBR assessment, interactions among chemicals are not considered, which results in uncertainties pertaining to complex mixture assessments (Borgert et al. 2001;Heys et al. 2016;Chen et al. 2020). Further serious complication occurs when unknown compounds from legacy contamination are present in the complex mixture. Moreover, there is the possibility of partial chemical degradation to unknown products with similar or different toxicity from the parent contaminant. In response to these challenges, the bottomup approach has been used in risk assessments for complex mixtures (Tian and Bilec 2018). A variation of this approach was taken in this study, namely to use the observed toxic effects pattern in an animal model to identify contaminants 1 3 of potential concern rather than predicting effects based on chemical composition (Xia et al. 2019).
The current study site is a major industrial site with historical contamination dating back to the 1950s. Some contaminants are known (pesticides, petroleum and chemicals used in pesticide packaging), while others likely include unknown contaminants (from possible spills or improper dumping/burial by prior property owners or unknown breakdown products of known parent contaminants) (Gainer, Peters, Siciliano, unpublished). These contaminants are known to move in soil plumes, despite capping of almost the entire site with buildings, concrete or asphalt. The site continues to be an active industrial pesticide packaging site located within a Canadian urban area, leading to a need for assessing environmental and toxicological risk posed by the site. After extensive chemical analyses of soil samples throughout the site on multiple occasions over three decades, an initial human health risk assessment was performed based on the few known contaminants that were identified to be above concentrations of concern (see Table 1; Gainer, Peters, Siciliano, unpublished). Petroleum hydrocarbons are designated as fractions (F1, F2, F3 and F4) based on the molecular weight, which in turn determines their environmental compartmentalization and route of exposure (Turle et al. 2007). From this chemical screen, the exposure route of greatest toxicological risk was determined to be restricted to oral since the predominant fractions of hydrocarbons were the nonvolatile F3 and F4 (Gainer, Peters, Siciliano, unpublished). The contaminants identified in this initial chemical screen included pesticides (organochlorine, carbamates and organochlorine) and petrochemicals (medium chain), which have been shown to induce a wide range of toxic effects at various levels including molecular, cellular, tissues, and organs (Mesnage et al. 2014). These toxicities have been well-documented after exposure to individual chemicals in humans and animals, but difficulties arise in identifying specific candidates when found in a mixture (Das et al. 2017;Polanco Rodríguez et al. 2017). Further complications arise from the potential for unknown parent contaminants or active breakdown products that were not measured and thus not included in the risk assessment. Therefore, the initial chemical risk assessment was deemed to be suspect or at least incomplete. As a result, the current occupant pursued other methods to assess toxicity and potential human health and ecological risk of this site.
This complex problem needed to be tackled while minimizing animal usage in order conform to current animal ethics guidelines such as the European framework of REACH (Registration, Evaluation, Authorisation and Restriction of Chemicals (Wielen 2008)). The approach used an initial cell culture screen using HepG2 liver cell cytotoxicity to identify the three most toxic soil samples. The two most toxic soil samples were chosen plus one sample from a location furthest from these toxic samples. The three extracts were then used for oral exposures in a rat model to screen for toxicity pattern across multiple Table 1 Soil contaminants of potential concern (COPC) that exceeded generic guidelines identified in soil samples taken from the study site over the past three decades The year of analyses are indicated in the far-right column Contaminants listed exceed specified guidelines, CCME Canadian Council of Ministers of the Environment, AEP Alberta Environment and Parks, OME Ontario Ministry of the Environment, AMEC-F1 and F2 represents fraction of petroleum hydrocarbons, 2,4-D 2,4-dichlorophenoxyacetic acid, 2,4-DB-(4-(2,4-dichlorophenoxy) butyric acid), BTEX benzene, toluene, ethylbenzene and xylene complex and MCPA-2-methyl-4-chlorophenoxyacetic acid

Contaminants of potential concern
Type of contaminants of potential concern Maximum concentration (mg/kg) organs. Oral exposure to the petrochemicals and pesticides already identified have been shown to induce toxic effects by causing oxidative stress, inducing detoxifying cytochrome P450 enzymes (CYP1A1), and inhibiting cholinesterase enzymes (Chan et al. 2009;Rekha et al. 2013;Parasuraman et al. 2014). These effects can translate into adverse outcomes in systemic organs such as kidney, liver, heart and immune organs. Thus, in addition to basic gross toxicological effects (survival, body weight, food/ water intake and organ weights), these organs of concern were examined using functional and histopathological approaches. We hypothesized that soil extracts identified as toxic through inhibition of HepG2 cell viability will elicit toxic effects in rat liver, kidney, brain and heart following oral exposure. To investigate this, a biweekly oral gavage of the two most toxic soil extracts identified in vitro (A, B) plus a third extract from a sample taken farthest from the toxic sample sites (C; all used at 0.1%) as well as a vehicle control (0.1% dimethyl sulfoxide or extract in polyethylene glycol) were administered twice weekly to male Sprague-Dawley rats (n = 10/group) for 4 weeks. The parameters of oxidative stress, inflammation, cholinesterase activity, CYP1A1 activity, cardiac arrhythmia and histopathology of target organs were analyzed.

Soil Sample Collection from Site, Initial Testing and Extraction
Soil with known contamination was collected from 19 boreholes (see Fig. 1 for an aerial picture of the site with borehole locations), at depths from 1 to 12 m, using incremental sampling methodology for a total of 18 different soil samples. Of the 18 soil samples from the pesticide manufacturing site, 17 were taken near known plumes of contamination running along the northern perimeter along the railway line or in the northwest corner ( Fig. 1), at different locations and soil depths. These various locations on the site ensured an optimum spatial coverage to ensure good representation of the samples was tested. In contrast, the one remaining soil sample that was screened from a borehole at the southeast edge of the property was furthest from the known areas of contamination but had the potential to be contaminated with other unknown historical contaminants ( Fig. 1). Thus, in total, 18 soil samples were extracted using an in-house extraction protocol. For this in-house protocol, 25 mL of the methanol/1 N HCl (90:10) extraction solution was added to the 10 g of each soil sample. The mixture was Fig. 1 The pesticide manufacturing site in this study, located within an urban area in a city in Canada (border of property indicated by blue demarcation). Legacy contamination is a major issue at this site, leading to a complex unknown mixture of contaminants in the soil and groundwater. This study examined three soil extracts taken from areas (A, B and C; indicated by circles on this map) 1 3 vortexed and centrifuged at 2000 × g for 5 min. The extract solution was decanted into centrifuged tubes. Twenty-five milliliters of the methanol/1 N HCl (90:10) extraction solution was added to each sample for a second extraction. The mixture was vortexed and centrifuged at 2000 × g for 5 min. A third extraction was done, repeating the steps for the second extraction. Twenty milliliters of each extracted solution was pipetted into glass tubes and 100 μL of 80:20 methanol/ glycerol solution added to each sample. Evaporation of the extract solution to approximately 3 mL using a Turbo-vap evaporator set at 40 °C with a nitrogen flow rate of approximately 15 psi was done. The concentrates were then resuspended in 6 mL dimethyl sulfoxide (DMSO) to produce an extract that was approximately equivalent to 1.11 g soil per mL of extract.

Screening of Soil Extracts for Cytotoxicity
HepG2 cells (ATCC, HB-8065) were grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% FBS and 1% penicillin-streptomycin. Cells were maintained in an incubator at 37 °C in a humidified atmosphere with 5% CO 2 . HepG2 cells were seeded at 5 × 10 4 cells/mL into 96-well culture plates and allowed to adhere for 24 h before exposure. The cells (in triplicate wells) were then incubated for 48 h with 0.5 mg of soil equivalent/mL of soil extracts. DMSO was used as a solvent control at a final concentration of 0.1% v/v in the culture media equivalent to 1 μL DMSO/mL of media. Each 96-well plate (with 200 μL of media) would have 0.09 μL of soil extracts added to achieve an exposure of 0.5 mg soil equivalent/mL in the assay. Both positive (1% Triton X-100) and negative (cell culture medium) controls were also applied. The WST-1 bioassay (Roche Applied Science, Indianapolis, IN) was used for an initial cytotoxicity screen of all 18 soil extracts on HepG2 cells. The assay was performed according to manufacturer's instructions (using a 1-h incubation with WST-1 solution), and the absorbance was measured at 450 nm using a POLARStar OPTIMA microplate reader (BMG Labtech). Data were expressed as a percentage of cytotoxicity with respect to negative control.

Animals and Exposures
Forty male Sprague-Dawley rats of weights 150-200 g were acquired from the supplier, Charles River Laboratories (Senneville, Quebec, Canada). Upon receipt of the Sprague-Dawley rats at the Animal Care Unit at the Western College of Veterinary Medicine (Saskatoon, SK), the rats were divided into four groups of ten animals each and allowed to acclimatize to their new environment for seven days. The rats were provided ad libitum access to water and a standard rat maintenance diet (Prolab RMH 3000, Missouri, USA). The rats were kept at a temperature of 22 °C and a 12-h light/dark cycle. All experiments were conducted according to protocol #20,170,109, as approved by the Animal Research Ethics Board at the University of Saskatchewan according to the guidelines of the Canadian Council on Animal Care.
The three soil samples used as extracts in the rat study (see Fig. 1 for the map of the study site and location where soil samples were obtained) are referred to as extracts A, B and C. In the control group, dimethyl sulfoxide (DMSO) was diluted to 0.1% in polyethylene glycol (PEG). For experimental groups, the soil extracts were stored and resuspended in DMSO and then diluted in PEG to 0.1% for use in rat experiments. The rats were orally gavaged twice weekly at a dose of 5 ml/kg on days 0, 3, 7, 10, 14, 17, 21, 24 and 28. On experiment day 30, the rats were euthanized with an overdose of isoflurane.

Basic Toxicological Measurements
Body weight and feed/water intake were measured and recorded weekly, on days 0, 7, 14, 21 and 28. Organ weights were measured post-euthanasia (kidney, liver, spleen, lungs and heart).

Echocardiogram Analyses
On day 30, an electrocardiogram (ECG) measurement was taken in rats under isoflurane anesthesia (5% induction, 1-3% maintenance with 100% oxygen) to assess the cardiac function using the PowerLab system and LabChart7Pro (Wildemann et al. 2015), with a minimum of 10-min ECG recorded once the rat was stabilized at maintenance anesthesia. The settings on the LabChart7Pro software were set to that specific for rats before analyses. For ECG analyses, nine (9) stable readings were used in analyses from the ECG at 30-s intervals. The parameters of interest (QTc interval, PR duration, QRS complex and heart rate) were measured in at least nine cardiac cycles and an average value then used for each rat. After ECG recordings were obtained, rats were euthanized with an overdose of isoflurane.

Blood Clinical Chemistry and Hematology
Blood was collected via cardiac puncture after an overdose of isoflurane before euthanasia and placed in serum and EDTA tubes. Immediately after, the heart was removed for analyses and to ensure death. The blood samples were analyzed at Prairie Diagnostic Services (Saskatoon, SK) for a small animal standard biochemistry panel including aspartate aminotransaminase (AST), alanine transaminase (ALT), alkaline phosphate (ALP), renal function tests including blood urea, creatinine, uric acid, glucose and electrolytes, complete blood cell count and white blood cell differentials including neutrophils, lymphocytes, eosinophils and monocytes.

Urine Chemistry
Immediately after induction of anesthesia, and before ECG recordings, anesthetized rats were injected intraperitoneal with 0.9% sodium chloride solution to ensure adequate urine was available for collection by the time of euthanasia and dissection. Urine chemistry was done using a Chem-stix multi-test strip (Roche Diagnostics, Indianapolis, USA) to screen for glomerular and tubular dysfunctions. Parameters measured were glucose, bilirubin, leucocytes, blood, protein and nitrite.

Assay of Oxidative Stress Biomarker
Oxidative stress was measured in the liver, lung, kidney and brain tissues using thiobarbituric acid reactive substances (TBARS) assay kit (Bioassays systems, Hayward, USA). Each tissue sample (liver, lung, kidney and brain) of 100 mg was weighed and homogenized in 200 μL of phosphate buffer saline (pH = 7.4). Subsequently, 75 μL of each sample was added to 150 μL of trichloroacetic acid and centrifuged for 24,000 × g for 5 min at 4 °C. The supernatant was used for the biochemical analyses in the method previously described (Ohkawa et al. 1979), but modified for animal tissues.

Assay of Cholinesterase Enzyme Activity
Both plasma and brain cholinesterase enzyme activities were measured using the cholinesterase kit (Sigma-Aldrich, St Louis, USA). For plasma samples, 10 μL was added to 390 μL of phosphate buffer. Brain tissue weighing 50 mg was added to 0.1 M phosphate buffer at pH = 7.5. Samples were homogenized on ice and centrifuged at 24,000 × g for 5 min at 4 °C. Cholinesterase enzyme activity was determined using absorbance at 412 nm.

Assessment of Plasma Inflammation
Inflammation was measured using an alpha-2 macroglobulin ELISA kit (Cell Bio labs, San Diego, USA) in plasma. The samples and standards were added per well (50 μL) and incubated for 2 h. Washing of plate was done, followed by the addition of 50 μL of biotinylated antibody per well. The samples in the wells were incubated for 1 h. After washing, 50 μL of chromogen substrate was added per well. Incubation was done for 10 min at room temperature. Finally, 50 μL of stop solution was added and read at 450 nm using mass spectrometry.

Ethoxyresorufin-O-Deethylase (EROD) Assay
The S9 fraction of hepatic tissues were assessed using EROD activity. A modified protocol (Kennedy and Jones, 1994) was used where the protein concentration in the S9 fraction was determined following EROD analysis in the same 96-well plate. Each sample (100 mg) was homogenized in 0.05 M Tris buffer at pH = 7.5 on ice. The lysates were centrifuged at 40,000 × g for 10 min at 4 °C. The supernatant was decanted and 10 μL added to the 96-well plate already filled with HEPES buffer, bovine serum albumin and resorufin. Afterward, 30 μL of 7-ethoxyresorufin was added to all wells. The plate was then incubated at room temperature for 10 min and 30 μL of NADPH added to all wells except blanks. A second incubation was done at 30 min in the dark at 25 °C. Fluorescamine (60 μL) was added to all wells and then incubated at room temperature for 15 min. Plates were read for resorufin and fluorescamine signals at the same time with excitation at 570/630 nm and emission at 365/480 nm using fluorescence.

Histopathology
The liver, heart, lung, spleen, and kidney tissues were initially fixed in 10% neutral buffered formalin for 72 h and then stored in 70% ethanol. The tissues were processed in a tissue processor (RVG/1 Histology Vacuum Tissue Processor) and embedded in paraplast Xtra paraffin wax (Leica Biosystems). Tissue blocks were sectioned at a thickness of 5-micron using an automated microtome (Microm-HM350) and a hematoxylin-eosin (H&E) staining protocol. Assessments were conducted by examining each organ in three zones (left lateral, median and right lateral) using a light microscope image analysis software (Massachusetts, USA). All tissues were scanned for histopathology by a board-certified veterinary pathologist. Heart, lung and liver samples did not reveal any differences and thus were not evaluated further (data not shown). However, spleen and kidney sections showed evidence of pathology. Each slide was demarcated randomly into four sections and counts of specific lesions of interest made. Histopathological lesions in sections were viewed at a magnification of 200 × for the kidney and at 40 × for the spleen.
A qualitative estimation of the lesions in specific organ structures was done (kidney; acute tubular necrosis, spleen; white pulp lymphoid proliferation). The qualitative estimation of acute tubular necrosis in the kidney was assessed as: 0 = no lesion found in the kidney tubules in view, 1 = lesion found in 25% of kidney tubules in view, 2 = lesion found in 50% of kidney tubules in view, 3 = lesion found in 75% of kidney tubules in view and, 4 = lesion found in 100% or throughout all kidney tubules in view. Similarly, in the spleen: 0 = no proliferation found within the white pulp units in view, 1 = proliferation found in 25% of white pulp units in view, 2 = proliferation found in 50% of white pulp units in view, 3 = proliferation found in 75% of white pulp in view and 4 = proliferation found within 100% of white pulp units in view. For each of the tissues (spleen and kidney), four different views were examined for each tissue (spleen or kidney) and an average determined for that animal. An incidence of the lesions in kidney (acute tubular necrosis) and spleen (white pulp proliferation) was graded as: mild = 0-25%, moderate = 25-75%, and severe = more than 75%.

Data Evaluation and Statistical Analyses
Results were expressed in mean ± standard error of the mean (SEM). The data sets were screened for their normality using the Kolmogorov-Smirnov test and homogeneity of variance using the Brown-Forsythe test. Outliers were checked and removed using the ROUT method (Q = 1%), after which datasets met the parametric assumptions. The ROUT method is based on the false discovery method using a robust regression and outlier removal (Motulsky and Brown 2006). Datasets that did not meet parametric assumptions were normalized using natural log transformation. Differences between groups (experimental and control) were determined using the one-way analysis of variance (ANOVA) followed by a Tukey post hoc analysis using GraphPad Prism version 8 (GraphPad Software, San Diego, USA). Statistical significance was set at p ≤ 0.05.

Cytotoxicity of Soil Extracts
From the HepG2 cytotoxicity data (Fig. 2) where 18 soil extracts were tested, three extracts were chosen for use in rat experiments. Of all the soil extracts tested only extract Th17-12 (2-3) significantly decreased cell viability after 48-h exposure of HepG2 cells. Similarly, a second extract (Th17-20 (4-5)) resulted in almost 20% decrease in viability ( Fig. 2) but was not significantly different from control. These two extracts (referred to as extracts A and B, respectively) were chosen for the rat experiment as potentially toxic samples. A third extract, from Th17-07 (8-9) is referred to extract C and chosen as a soil sample from a location further from known areas of contamination.

Mortalities, Body Weight, Organ Weight, Water and Food Intake
In terms of mortality, one rat death was reported, which was not a result of the toxicity, but due to difficulties in gavaging on the experimental days, leading to aspiration and subsequent death the following day. Therefore, no mortalities as a result of toxicity were observed in the treatment or control groups. Moreover, no overt sublethal toxicities (gross behavioral toxicities such as grooming, stress or aggression) were seen in either control or experimental groups. Oral gavage with soil extracts did not result in a significant change in final body weight (Table 2). There were no significant Fig. 2 WST-1 cytotoxicity assay of HepG2 cells treated with 0.1% of soil extracts from the industrial site for 48 h. Bars are mean ± SEM from four independent assays. Data were analyzed by ANOVA with Dunnett's post hoc (* = p < 0.05 compared to the media control). Sample names refer to the well location on site and depth where soil was taken (see map Fig. 1 in main body of paper where Th17 refers to the entire study site, the 5th and 6th numerals after the dash refer to the well site and the numerals in bracket refer to the depth (feet) within the well where the sample was taken) changes in feeding and drinking water between experimental and control groups (Table 2). Post-euthanasia, there were no significant weight changes in most organs (heart, kidney, lungs and spleen) after the 4-week exposure. However, the weight of the liver was significantly reduced after the exposure to soil extract A compared to controls (Table 2).

Clinical Blood and Urine Chemistry, Hematology
The complete blood count showed a statistically significant increase in plasma lymphocyte (7.83 ± 1.56 × 10 9 /L) of twofold change and plasma neutrophil (1.28 ± 0.73 × 10 9 /L) of threefold change in the treatment group exposed to the soil extract A compared to the control group. The counts of the other blood cells remained statistically insignificant after exposure to other soil extracts (Table 3). Biochemical markers in blood and urine depicting liver and kidney organ functions showed no significant change after exposure to soil extracts for 4 weeks compared to the control group (see Supplemental Data Tables S1 and S2).

Cholinesterase Assay
A significant reduction (1.3-fold) in plasma cholinesterase activity compared to control following exposure to extract A was observed (Fig. 3). However, oral exposure to soil extracts B and C did not show any significant change in the plasma cholinesterase enzyme. Conversely, no significant change was observed in brain cholinesterase among treatment groups (Fig. 3).

Ethoxyresorufin-O-Deethylase Assay
A significant increase in EROD activity was found in rats gavaged with soil extract B compared to controls, while rats exposed to soil extract A showed a significantly higher EROD activity than all other groups (Fig. 4). Thus, extracts A and B showed a significant 30-and 10-fold increment in EROD activity compared to the controls, respectively (Fig. 4). Full chemical analyses were not performed on these soil extracts, but limited analyses to Table 2 Effects of a complex mixture of solvent extracts from contaminated soil on changes in the body and relative organ weights in male rats after twiceweekly gavage for 4 weeks Data are presented as mean ± SE (n = 9-10 rats/group). Different letters represent significant differences (P < 0.05; Tukey's test after one-way ANOVA). Whereas no letter superscripts are shown, no significant treatment differences were observed

Determination of Systemic Inflammation
Histopathological analyses of spleen sections revealed splenic white lymphoid proliferation evidenced by enlarged geminal zones (GZ: inner deep purple zone within white pulp) and marginal zones (MZ: outer region in light purple around white pulp) (Fig. 5A, B, C and D). White pulp proliferation was present in the spleens of all experimental groups but in variable proportions. They are ranked in the following order from highest to lowest among treatments: extract C (66%) > extract B (45%) > extract A (34%) > Control. No changes in red pulp histology were observed among groups, and all histological features of red pulp were normal (Fig. 5). Plasma alpha-2 macroglobulin levels were significantly elevated in rats orally exposed to all three soil extracts compared to the control (Fig. 5E). The level of inflammation was significantly elevated among all groups, with the following order from highest fold-change from the control: extract A (3.8) > B (2.9) > C (2.4) (Fig. 5E).

Determination of Oxidative Stress in Tissues
Oxidative stress, indicated by lipid oxidation and increased TBARS, was significantly higher in brain homogenates from rats exposed to soil extract C compared to control (Fig. 6C). In contrast, homogenates of the kidney from rats exposed to soil extract B caused a significant increase in TBARS concentration compared to control (Fig. 6D). These significant increments corresponded to a 1.2-fold and 1.9-fold of TBARS in kidney and brain, respectively. No significant differences among Plasma and brain acetylcholinesterase activity in male Sprague-Dawley rats gavaged twice weekly for 4 weeks with contaminated soil extracts or control vehicle. Data are shown as mean ± standard error of the mean (n = 9-10/group). Different letters indicate significant differences among groups (P < 0.05; Tukey's tests after one-way ANOVA)

Fig. 4
Cytochrome P450 1A1 activity was measured using ethoxyresorufin-o-deethylase (EROD) activity in liver S9 fractions from male Sprague-Dawley rats gavaged twice weekly for 4 weeks with contaminated soil extracts or control vehicle. Data are shown as mean ± standard error of the mean (n = 9-10/group). Different letters indicate significant differences among groups (P < 0.05; Tukey's tests after one-way ANOVA) groups were detected in liver TBARS (Fig. 6A) and lung TBARS (Fig. 6B).

Electrocardiography
Electrocardiography (ECG) assessed changes in cardiac function, using heart rate, QTc (corrected QT) segment duration, QRS duration and PR interval. A significantly prolonged QTc (1.25-fold) was observed in the group of rats exposed orally to a high dose of the soil extract C after the 4 weeks compared to control (Fig. 7A). However, there were no significant changes in the PR interval, heart rate, and QRS duration among groups (Fig. 7AB, C and D).

Histopathology of Tissue Sections
Microscopic sections of the liver, heart and lung tissues showed no differences between the treated animals and the controls after oral exposure to all three soil extracts (A, B and C). In contrast, significant histopathology was observed in kidney sections. The control group showed relatively healthy cortices, medulla, glomeruli and tubules, but acute tubular necrosis was seen in multiple regions in both cortex and medulla of all exposed groups (A, B and C; Fig. 8). Tubular necrosis was present in about 70% of all rats exposed to soil extracts. Rats exposed to soil extract C showed the highest prevalence of tubular necrosis of almost 75%. Loss of brush border epithelium in tubules was found in rats exposed to soil extract A of about 68%. Evidence of hydropic changes and vacuolation was observed in about 15% of all exposed rats. About 20% of rats in each exposed group showed sporadic evidence of tubular hemorrhage and glomerular shrinkage.

Discussion
The study used a combination of cell culture and traditional rat model methods to assess toxic effects from oral soil extract exposure from a legacy contaminated industrial site.
To minimize the numbers of animals needed, differential toxicities of complex soil extracts originating from a legacy contaminated site were prioritized using initial in vitro cytotoxicity screens. The in vitro cell lines were screened for the effects of complex mixtures (hydrocarbons and pesticides), but this cell-based assay is limited by its ability to discriminate among possible mixture interactions. Of the 18 soil extracts tested, two toxic soil sample extracts and one potentially less toxic extract were identified based on WST-1 assay in HepG2 cell culture and were then tested in a rat model. The liver is the primary biotransformation organ exposed to environmental contaminants and expected to be a target organ for toxicity and hence the selection of HepG2 cell lines. The HepG2 cell is metabolically active, comparable to hepatic parenchymal cells in vivo. Lipid peroxidation is a common mechanism of action involved in DNA damage mediating cytotoxicity within HepG2 cell lines expressed as reduced growth and proliferation, which is detected by the WST-1 assay. This mechanism of toxicity is targeted by the multiple identified contaminants (pesticides, hydrocarbons Fig. 6 Lipid oxidation was measured using thiobarbituric acid reactive substances (TBARS) assay in liver, kidney, brain and lung tissue from male Sprague-Dawley rats gavaged twice weekly for 4 weeks with control vehicle or contaminated soil extracts. Data are shown as mean ± standard error of the mean (n = 9-10/ group). Different letters indicate significant differences among groups (P < 0.05; Tukey's test after one-way ANOVA) and heavy metals) on site. However, toxicities involving either extrahepatic organ-specific toxicities or mechanisms of toxicity independent of oxidative stress might have been missed in this initial cell culture screen. Although contaminants from the soil samples were extracted using the methanol-HCl procedure, we acknowledge the possible absence of some chemicals in the extract that fall outside methanol's extraction range. The next step in the present study was to use male Sprague-Dawley rats to identify patterns of toxicity and target organs after oral exposure to the three soil extracts chosen. Since this site is currently primarily known for transport and storage of pesticides, it was expected that the unknown mixture of soil contamination would contain organochlorine, organophosphate and carbamate pesticides, as well as petrochemicals and their breakdown products. The chemical analysis data of compounds detected above scheduled quantities shown in Table 1 confirm that lindane, 2,4-D, chlordane and methoxychlor, among others, were detected at some point over three decades of study. However, a very large amount of petroleum and other hydrocarbons have also been consistently detected, from low molecular weight benzene, xylene and other F1 hydrocarbon chemicals to higher molecular weight fractions F2 and F3 as well as petroleum hydrocarbons. Moreover, the location of soil sample sites over the years was not consistent and the chemicals continued to degrade for decades. For the current study, new boreholes were made to take the soil samples, and it is not known whether the same chemicals detected previously are still present. In this study, the two sites shown to inhibit Data are shown as mean ± standard error of the mean (n = 9-10/ group). Different letters indicate significant differences among groups (P < 0.05; Tukey's test after one-way ANOVA). QTc calculated using the Bazzet's correction cell viability (extracts A and B) were assumed to be representative of most of the industrial site since these were from areas closest to the railway and the northwest corner. In contrast, extract C was expected to have little to no toxicity based on a HepG2 WST-1 assay and the fact that it was furthest from the active areas of the property and the railway. In fact, results from this study showed that all three extracts showed toxicity in the rats, albeit with extracts A and B being more similar compared to extract C. Based on the petroleum hydrocarbon assessment, all three extracts contained similar total petroleum hydrocarbon levels. However, the hydrocarbon fractions known to induce EROD, the F3 and F4 hydrocarbons, were highest in extract B, not extract A. More detailed chemical analyses would be needed to determine specific compounds that correlated to EROD activity. Kidney damage and systemic inflammation were the primary targets from these contaminants, but potential risk was also observed due to evidence of brain oxidative stress and cardiotoxicity from extract C.
In the rat toxicity study, there were no significant changes in the consumption of feed, water and other parameters such as body and organ weights, which are essential in assessing basic toxicity. Organochlorine, organophosphate and carbamate pesticides have been shown to modulate endocrine effects, causing an increase in body weight (Pelletier et al. 2002). Soil from this study site contains several toxicants above scheduled quantities: organochlorines such as the herbicide dicamba as well as insecticides lindane and methoxychlor, plus the organophosphate malathion. Despite the potential for these pesticides to cause gross toxicities, this study found no significant increase in body or organ weight between the control and treated groups. This lack of effect on body and organ weights may be ascribed to the low levels and short period of exposure, which was subchronic compared to previous studies which had chronic exposures (Meggs and Brewer 2007).
Lymphocytic and neutrophil counts were significantly elevated in the experimental group exposed to soil extract A at the end of the 28-day oral exposure. The primary role of lymphocytes is to partake in both cellular and humoral immunity (Kretschmer et al. 2005;Modaresi and Jalalizand 2011). Therefore, the significant increase in the plasma lymphocyte and neutrophil counts after exposure to soil extract A points toward immunostimulatory and pro-inflammatory effects. Alpha-2 macroglobulin, a biomarker of inflammation, was significantly increased within the plasma of rats exposed to all three soil extracts, which is indicative of systemic inflammation (Lasram et al. 2014). An additional finding pointing to immunostimulation and inflammation was detected within the splenic tissues of experimental groups that revealed a mild-to-moderate white pulp lymphoid proliferation. The white pulp of the spleen has been shown in previous studies to undergo proliferation upon exposure to inflammatory agents such as pesticides and petrochemicals (Igwebuike et al. 2007;Voloshin et al. 2014).
Previous studies have reported the effects of organophosphates and petrochemicals to induce oxidative stress and inflammation after subchronic oral exposure in rats (El-Bini et al. 2015;Elelaimy et al. 2012;Ita 2011). These effects are consistent with known effects of some of the chemicals detected above scheduled quantities at this study site, namely the petroleum hydrocarbons as well as organochlorines such as lindane and methoxychlor. Many of these chemicals are also agonists of the aryl hydrocarbon receptor (AhR), consistent with the observed induction of hepatic CYP1A1 after exposure to soil extract A. Previous studies have reported immune cell activation through activation of the AhR by chemicals such as petrochemical and organochlorine (Bankoti et al. 2010;Marshall et al. 2008). Considering these immune effects, the results of this study would be consistent with hydrocarbons in all of the soil extracts being linked to the observed plasma lymphocytosis and neutrophilia. Chronic inflammation is known to produce de novo diseases such as auto-immune disease and vascular leakiness but can also worsen chronic human diseases such as heart disease or atherosclerosis (Sanmarco et al. 2018).
Lipid peroxidation, indicative of oxidative stress, was found in the kidney and brain tissues of rats exposed to soil extracts B and C, respectively, compared to the control. The brain has an extensive metabolic activity with a low level of antioxidant mechanisms, which predisposes it to oxidative stress when exposed to toxicants (Possamai et al. 2007). Thus, the higher predisposition to effects by toxicants supports the observed results in the brain. Pesticides and petrochemicals such as lindane, methoxychlor, 2,4-D and F1-F2 hydrocarbon fractions found at this study site are metabolized in the liver, while the kidney remains the main excretory organ exposed to metabolites of these compounds. Therefore, this excretory pathway for most of these contaminants predisposes the kidney to toxicities such as oxidative stress.
There was a significant inhibition of plasma cholinesterase by 14.6% compared to the controls after exposure to soil extract A, but not the brain. This disparity in the site of cholinesterase enzyme inhibition can be attributed to the effect of an organophosphate or carbamate metabolite (Kopjar et al. 2018). The prominent organophosphate malathion, with its metabolite malaoxon, is highly active and hydrophilic with little ability to cross the blood-brain barrier, which makes plasma cholinesterase a major target in this regard (Baconi et al. 2013). While the chemical analyses of soil samples over several decades from this site did show elevated levels of malathion, present results in rats suggest that parent malathion may no longer be present in the current soil samples tested in this study. Instead, the active breakdown metabolite, malaoxon, which has never been actively measured, might be present in the soil to explain the plasma cholinesterase inhibition.
The histopathological assessment revealed moderate-tosevere nephrotoxicity within the experimental group after exposure to the soil extracts. This is in line with the kidney's function to eliminate both xenobiotics and environmental contaminants (Vaidya et al. 2010). Acute tubular necrosis was the main pattern of toxicity identified, and literature has shown reactive oxygen species to be one of the main underlying mechanisms (Kalender et al. 2007;Li et al. 2016). Reactive oxygen species (ROS) impair enzymatic and structural protein molecules, which lead to eventual cell death (Avdagić et al. 2008). Acute tubular necrosis was found mainly in the cortex, which was characterized by dilation of tubules filled with a pink cast (proteinaceous material), and the shedding of epithelial cells with or without nuclei into the lumen of tubules. This histopathological lesion was in line with a previous rat study after oral exposure to organophosphate pesticides (Rekha et al. 2013). The acute tubular necrosis was most severe in kidneys of rats exposed to soil extract C. In contrast to the significant histopathological lesions found in the kidneys of exposed rats to all soil extracts, normal serum biomarkers of kidney injury including urea, creatinine and electrolytes were recorded. This can be explained by the high functional reserve of kidneys in the presence of abnormal renal histopathological changes (Palsson and Waikar 2018). In this study, total serum albumin and protein with high molecular weight and low glomerular filterability did not change between the groups exposed to different soil extracts and the controls. Besides, there was no significant proteinuria associated with oral exposure to the different soil extracts (supplemental data Table S1). This could be attributed to the selective renal tubular toxicity, which spared the glomerulus, especially the integrity of the podocytes within its basement membrane. This agrees with a previous study where oral exposure to petrochemicals resulted in deranged serum renal function and tubular damage but normal serum protein and albumin levels (Azeez et al. 2013). Taken together, the soil from the southeast corner of the study site, corresponding to extract C, has toxic effects consistent with light weight hydrocarbons such as xylene and benzene that have previously been detected above scheduled quantities at this study site.
No significant histopathological liver lesions were found in the experimental groups exposed to the soil extracts. This is inconsistent with the literature where oral exposure to organochlorines, organophosphates or carbamates resulted in hepatic steatosis, necrosis and elevated malondialdehyde (Akbel et al. 2018;Karami-Mohajeri et al. 2017). The liver is normally the first target after oral exposure to toxicants once they are absorbed from the gastrointestinal tract and undergo first-pass metabolism (Pylayeva-Gupta 2011). Unexpectedly, the histopathology of the liver and blood clinical chemistry for liver function remained unaffected after oral exposure to the soil extracts. Similarly, there was no significant increase in oxidative stress within the liver after exposure to all three extracts compared to control, which is a measure of redox imbalance and subsequent toxicity. These findings were in contrast to previous studies that found elevated markers of liver injury and liver histopathological lesions (Abu El-Saad and Elgerbed 2010; Liu et al. 2017). An explanation could be the effective hepatic metabolism of these parent compounds and adequate antioxidant systems, which possibly resulted in no significant oxidative stress in the liver of experimental groups, as well as no elevation in serum liver enzymes and biomarkers. In addition to the HepG2 cell line, which represents the biotransforming and proximate target organ (liver), a kidney cell line depicting excretory and high internal burden of contaminant exposure will be included in future studies to predict the extra-hepatic toxic effects.
Prolongation of corrected QT duration (QTc) was observed in rats exposed to soil extract C, most likely due to neurotoxicity and may relate to the coincident increased brain oxidative stress in this group. While brain cholinesterase activity was unaffected in this group, neurotoxicity that involved cholinergic stimulation and subsequent nonselective nature of acetylcholine in binding to either muscarinic or nicotinic receptors is consistent with the observed cardiac effect. Binding of acetylcholine to nicotinic receptors causes a surge in the sympathetic drive while also stimulating the parasympathetic system, which is known to underlie the prolongation of the QTc that can lead to the progression into other fatal arrhythmias such as torsade de pointes and ventricular fibrillation (Abraham et al. 2001).

Conclusion
This study describes the target organ toxicities following oral exposure to soil extracts from a contaminated site. Results support our hypothesis that the main targets after oral exposure to these contaminated soil extracts were liver, kidney and cardiac toxicities, as well as CYP1A1 and acetylcholinesterase enzyme modulation. However, the pattern of observed effects differed among extracts selected from different locations on the study site. Soil extract A was shown to have the highest impact on in vitro cytotoxicity as well as induction of in vivo hepatic CYP1A1, induction of plasma alpha-2 macroglobulin and inhibition of plasma acetylcholinesterase, while extract B showed similar but lesser toxic effects. Extracts A and B effects are suggestive of both petroleum hydrocarbon and metabolites or breakdown products of organophosphate/carbamate contamination. However, extract C was predicted to be non-toxic based on its location further from known spills and lack of cytotoxicity in the HepG2 cell culture screen. Contrary to this expectation, extract C was found to stimulate brain oxidative stress, inflammation, cardiotoxicity and nephrotoxicity, suggestive of presence of pesticides that could cross the blood-brain barrier as well as petrochemicals that were also metabolized in the kidney in this soil sample. However, actual risks to humans following oral exposure to the soil from the study site will require further assessments to confirm chemical identities of contaminants. While the previous chemical analyses are consistent with these classes of chemicals being present and elevated in the study site soils, analyses of the same borehole locations should be done and a more extensive measurement of pesticide metabolites should be included. Additionally, further experiments must be performed using different routes of exposure to ascertain possible differences and extent in toxicity. Overall, the cell culture method did not successfully predict the neurotoxicity and cardiotoxicity observed with extract C but was more successful with suspected hydrocarbon and intact pesticide compounds toxicities. This highlights the fact that seemingly similar end-points testing in vitro toxicity do not always translate into the same in vivo effects or vice versa. Specifically, caution should be taken when comparing lethality in cell lines to sublethal end-points in a whole animal where toxicant distribution, metabolism and multi-organ compensation can obscure effects. Conversely, sublethal cell culture effects should also be considered when using it as a screen for toxicity to better compare to sublethal in vivo effects.
Funding This study was funded by the Natural Sciences and Engineering Research Council (NSERC) and the industry sponsor.

Conflict of interest
The author (s) acknowledge the potential conflict of interest due to funding received from the industry partner and owner of the study site for this research project.
Ethical Approval Ethical approval granted by the University of Saskatchewan Animal Research Ethics Board (Protocol number 20170109).