Development of a High-throughput in vivo Screening Platform for Particulate Matter Exposures

Abstract

Particulate matter (PM) exposure is a public health burden with poorly understood health effect mechanisms and lacking an efficient model to compare the vast diversity of PM exposures. Zebrafish (Danio rerio) are amenable to high-throughput screening (HTS), but few studies have investigated PM toxicity in zebrafish, despite the multitude of advantages. To develop standardized exposure procedures, the urban PM standard reference material (SRM) 1649b was used to systematically determine sample preparation methods, design experimental controls, determine concentration ranges and evaluation procedures. Embryos (n=32/treatment) were dechorionated and placed into 96-well plates containing SRM1649b (0 – 200 μg/mL) at 6 hours post fertilization (hpf). Developmental toxicity was assessed at 24 and 120 hpf by evaluating morphological changes, embryonic/larval photomotor behavior, and mortality. Differences from blank medium and particle controls were observed for all biological responses measured. Differences due to SRM1649b concentration and preparation method were also observed. Exposure to SRM1649b from DMSO extraction was associated with changes in morphology and mortality and hypoactivity in photomotor responses compared to the DMSO control for the whole particle suspension (76, 68 %) and soluble fraction (59, 54 %) during the embryonic and larval stages, respectively. Changes in behavioral responses were not observed following exposure to the insoluble fraction of SRM1649b from DMSO extraction. The toxicity bias from PM preparation provided further impetus to select a single HTS exposure method. Based on the biological activity results, the soluble fraction of SRM1649b from DMSO extraction was selected and shown to have concentration dependent cyp1a/GFP expression. This rapid, sensitive and consistently scalable model is a potentially cost-effective vertebrate approach to study the toxicology of PM from diverse locations, and provides a path to identifying the toxic material(s) in these samples, and discover the mechanisms of toxicity.

1. Introduction

Exposure to ambient particulate matter (PM) is a global public health concern as the systemic effects of chronic exposures are continuing to be elucidated. Beyond respiratory outcomes, there are growing associations between PM and cardiovascular (Brook et al. 2010; Martinelli et al. 2013), digestive (Pedersen et al. 2017; Salim et al. 2014), and developmental effects (DeFranco et al. 2015; Rappazzo et al. 2014; Volk et al. 2013). Importantly, health outcomes associated with PM exposures vary based on differences in PM concentration (van Donkelaar et al. 2010; West et al. 2016) and composition (Philip et al. 2014), but the mechanisms of these variations on health outcomes is poorly understood. Comprehensive vertebrate toxicology studies are necessary to complement PM epidemiology research and mechanistically understand the impacts of variable PM exposures on health.

Currently, PM-induced toxicity is studied using in vitro and in vivo using a few select ambient PM samples. In vitro models lack the complexity of in vivo systems and cannot easily address metabolism or systemic effects, which are of growing concern with human PM exposures (Dominici et al. 2006). Rodent models (McGee et al. 2015; Wang et al. 2013; Wegesser et al. 2009) address systemic effects but make for slow and costly experiments. They also require large mass loadings of PM thus, limiting the number of samples that can be tested. Most importantly, rodent models of PM-induced toxicity are not conducive to high-throughput screening (HTS) and therefore, incapable of assessing the high diversity of global PM exposures. There is a clear need for a rapid, sensitive, and consistently scalable surrogate vertebrate model for PM-induced developmental toxicity.

Zebrafish (Danio rerio) provide a platform with numerous advantages such as optical clarity during development, rapid development, high fecundity, small size, and high genetic homology to humans (Howe et al. 2013). The major organ systems of zebrafish develop within 5 days post fertilization (dpf) and are easily visualized either by bright field microscopy or via numerous tissue specific fluorescent reporter lines of zebrafish available (Garcia et al. 2016; He et al. 2014). Neurodevelopment research is readily assessable in the model as there are well-defined behavior phenotypes and established assessment tools (Truong et al. 2016). The zebrafish molecular toolkit allows for rapid genetic and mechanistic research (Chlebowski et al. 2017a). Costs for zebrafish handling and housing are minimal compared to other rodent models and the amount of test chemical required for relevant dosing is typically an order of magnitude less than a comparable design in rodents. Finally, zebrafish are highly amenable to HTS with the necessary automation infrastructure already in place in a growing number of labs (Kessler et al. 2015; MacRae and Peterson 2015; Mandrell et al. 2012a).

Despite the attributes of the zebrafish model, limited research has been conducted using PM exposures (Duan et al. 2017; Mesquita et al. 2016; Stevens et al. 2017) and the HTS capabilities have not yet been leveraged by the PM toxicology field. We set out to develop standardized protocols that would enable future high-throughput developmental toxicity screening of PM in zebrafish with the goal of efficiently detecting bioactivity from diverse PM samples.

2. Materials and Methods

2.1 Chemicals

Urban dust standard reference material (SRM)1649b was purchased from the National Institute of Standards and Technology (NIST) and stored away from light at 4°C. Carbon black was purchased from the Cabot Corporation and stored away from sunlight. The synthetic lung fluid (SLF) selected was Gamble’s Solution prepared fresh following previously described protocols (Marques et al. 2011) using magnesium chloride, sodium chloride, potassium chloride, disodium hydrogen phosphate, sodium sulfate, calcium chloride dihydrate, sodium acetate, sodium hydrogen carbonate, and sodium citrate dihydrate, all purchased from Sigma Aldrich.

2.2 Zebrafish Husbandry

Standard procedures for fish care followed at Sinnhuber Aquatic Research Laboratory (SARL) were utilized with adult fish for a wildtype (Tropical 5D) and a transgenic reporter line (Tg(cyp1a:nls-egfp)) (Kim et al. 2013b) that were maintained at 28±1 °C on a recirculating system, with a 14 h light/10 h dark cycle (Truong et al. 2016). Embryos were collected from group spawns of adult zebrafish (Varga 2011) and enzymatically dechorionated at 4 hours post fertilization (hpf) (Mandrell et al. 2012b). Embryos were then mechanically distributed into individual wells of a 96-well plate that contained 90 μl of embryo medium (Mandrell et al. 2012b; Noyes et al. 2015) and PM preparations or control solutions (10 μl) were added at 6 hpf. Treatment and control groups (n=32 embryos/group) were equally distributed across multiple 96-well plates to control for intra- and inter-plate variability. All experiments were conducted with fertilized embryos according to Oregon State University Animal Care and Use Protocols.

2.3 Developmental Toxicity Screen

Following embryo exposure at 6 hpf, the 96-well plates were sealed with Parafilm to prevent evaporation, wrapped in aluminum foil to prevent photodegradation, and placed on an orbital shaker at 235 rpm overnight to ensure gentle mixing after the exposure; plates were stored at 28 °C (Truong et al. 2016). Embryos were maintained in these static conditions throughout the exposure period (6–120 hpf). Developmental toxicity was assessed at 24 and 120 hpf for morphological and behavioral changes as well as mortality in all treatments and controls (n=32 embryos/group). Mortality and morphological outcomes were visually assessed using a dissecting microscope as previously described (Truong et al. 2011) and consisted of 22 endpoints including malformations (i.e. notocord, heart, brain, eye, caudal and pectoral fins), edema (i.e. yolk sac, pericardial), and changes in pigmentation. Behavior was assessed through photomotor responses in embryos (28 hpf) and larvae (120 hpf) as previously described in detail (Truong et al. 2016). Zebrafish embryos acclimated to dark environments exhibit spontaneous contralateral axis bending normally potentiated by a single flash of light; the embryonic photomotor response (EPR). The EPR is driven by light perception by non-visual photoreceptors in the early hindbrain (Kokel et al. 2013). To assess EPR in all treated and control embryos, 28 hpf embryos housed in a dark environment were recorded under infrared illumination before and following two pulses of bright light with a 9 s interval between pulses. All movement responses were digitally recorded with 850 frames at 17 frames s⁻¹ from underneath the custom 96-well plate mount (Reif et al. 2016). At 120 hpf, larval photomotor responses (LPR) were assessed using Zebrabox behavior chambers (ViewPoint Life Sciences) which track swimming distances with HD video at 15 frames s⁻¹ during light to dark transitions with 4 cycles of 3 min visible light and 3 min dark. In normal 120 hpf zebrafish larvae movement decreases when larvae are exposed to bright visible light and immediately increases upon the transition to darkness (IR light) with a steady tapering off of movement over the dark phase. In both EPR and LPR behavioral assays any dead or malformed animals were excluded from analysis with all experimental data being processed using custom R scripts executed from an in house LIMS interface called the Zebrafish Acquisition and Analysis Program (ZAAP). Statistical comparisons were as previously reported (Truong et al. 2016). In all analyses, controls were medium and preparation specific.

2.4 PM Exposure Methods

Appropriate procedures for developmental toxicity screening of PM in zebrafish were established using NIST SRM1649b. This SRM was selected as a well-characterized ambient PM mixture from the Washington, DC area (NIST 2016) with adequate mass to determine exposure methods for future ambient PM screening studies in zebrafish. Several aspects were considered in developing the exposure paradigm including: sample preparation, experimental controls, and concentration ranges.

2.4.1 Sample preparation

Whole particle suspensions of dried, weighed SRM1649b were prepared for exposures of zebrafish. To determine the appropriate medium for re-suspension, three solvents were tested: ultrapure water, 100 % dimethyl sulfoxide (DMSO), and synthetic lung fluid (SLF, Gamble’s solution) (Marques et al. 2011). These solvents were selected for their ability to dissolve a variety of PM components that vary in solubility and, in the case of SLF, to replicate the conditions in the human lung. For all preparations, medium was added to the SRM1649b and left at room temperature for 30 min prior to re-suspension. To establish the potential impacts of re-suspension methods, both sonication and vortexing were tested for each of the medium preparations. Samples were waterbath sonicated at 40 kHz for 30 min (Branson, model 3510) or vortexed for 30 s. Following re-suspension, embryo medium was added to the preparations resulting in a final concentration of 200 μg/mL of SRM1649b in 1 % resuspension medium in each well containing a single embryo.

To assess the developmental impact of the soluble and insoluble fractions of SRM1649b, preparations were created for each of the re-suspension medium and distribution methods used. Aliquots of each whole particle suspension were centrifuged for 5 min at 13,000 g and the supernatant was collected as the soluble fraction. Two additional washes in the corresponding medium, with identical procedures, were performed to ultimately recover washed particles that were designated as the insoluble fraction of each sample. Soluble (2 re-suspension methods in 3 types of medium, resulting in n=6) and insoluble (n=6) fractions were prepared for comparison to the whole particle suspensions previously described. In total, 18 methods that varied by medium, re-suspension method, and fraction of SRM1649b were tested for developmental toxicity in zebrafish.

2.4.2 Experimental controls

Vehicle controls were used in all developmental toxicity screens corresponding to the medium used for re-suspension of SRM1649b. Final concentration of the vehicle was 1 % of the total volume in all treatments and controls. Additionally, to account for the potential developmental outcomes resulting from the physical presence of particles, carbon black (Cabot Corporation) was used as a particle control. The concentration and preparation of carbon black was identical to the corresponding SRM1649b treatments.

2.4.3 Concentration ranges

Based upon the aforementioned sample preparation and experimental control studies, a subset of the preparation methods was further assessed for developmental toxicity at a range of concentrations through serial dilution of a 200 μg/ml stock of SRM1649b (200, 100, 50, 25, 12.5 μg/ml). Preparations of SRM1649b were sonicated in all three solvents (water, DMSO, SLF) with the whole particle suspension, soluble, and insoluble fractions being tested for developmental toxicity as described above.

2.5 Cyp1a/GFP Reporter Fish

Following selection of a standardized PM exposure protocol, the Tg(cyp1a:nls-egfp) transgenic reporter fish (Kim et al. 2013b) was used for developmental toxicity testing with the DMSO soluble fraction of SRM1649b (0, 12.5, 25, 50, 100, and 200 μg/mL), at a final concentration of 1 % DMSO in embryo medium. The only deviation from the above developmental toxicity testing procedures (Section 2.3) was that, prior to euthanasia at 120 hpf, larvae were evaluated for cyp1a/GFP expression using a Keyence BX-X700 fluorescence microscope (Keyence North America, Itasca, IL). This cyp1a/GFP reporter line has previously been shown to induce cyp1a expression with high sensitivity and harbors a fosmid construct with GFP under control of a recombinant zebrafish cyp1a promoter (Chlebowski et al. 2017b; Kim et al. 2013b).

2.6 Statistical Analyses

All experimental data was processed as described in 2.3 with statistical significance computed as previously reported (Truong et al. 2016). Barplots and additional statistical significance calculations were completed with SigmaPlot 13.0 (San Jose, CA). Heatmaps were generated using RStudio (Boston, MA) with the gplots, heatmap. 2 package. Medium controls were included for all experiments with 1 % of the corresponding medium (water, SLF, or DMSO) in embryo medium serving as the medium control. For experiments using particle controls (200 μg/mL carbon black in corresponding medium), statistical significance was determined compared to both medium controls (1 % water, SLF, or DMSO in embryo medium) and SRM1649b treatments of the same medium and preparation (whole particle suspension, soluble, or insoluble fraction). Significance for all experiments was determined using two or three way analysis of variance (ANOVA) tests and pairwise multiple comparison procedures (Hom-Sidak method) with significance set at p<0.05. In all analyses, controls for statistical comparison were medium and preparation specific to the SRM1649b treatment preparation (i.e. the soluble fraction of SRM1649b from water extraction had a medium control of 1 % water and a particle control of the soluble fraction of carbon black from water extraction).

3. Results and Discussion

3.1 Sample Preparation and Experimental Controls

Developmental toxicity was assessed in zebrafish treated with SRM1649b prepared using various re-suspension medium and methods and for each of these preparations the whole particle suspensions, soluble, and insoluble fractions were tested. Table 1 depicts the percent incidence of mortality and morphological outcomes following treatment of zebrafish embryos with SRM1649b (200 μg/mL), prepared by the methods outlined. Total mortality, any effect except mortality (includes all morphological endpoints assessed excluding mortality), and any effect (all morphological endpoints and mortality) are reported with statistical significance assessed from both medium and particle (carbon black) controls. There were no significant differences between the medium controls for any of the responses, eliminating the concern that the medium itself results in significant differences in mortality or morphological changes. Survival rates in medium controls were over 95 % at 24 hpf and approximately 90 % at 120 hpf.

Table 1.

Percent incidence of mortality and morphological changes following particle control or SRM1649b treatment from various preparation methods

		Mortality		Any Effect Except Mortality		Any Effect
		Vortex	Sonicate	Vortex	Sonicate	Vortex	Sonicate
Whole Particle Suspension
Water	Particle Control	21.88	12.50	10.00	6.67	31.25	18.75
	SRM1649b	18.75	9.38	34.48	58.06*	50.00	65.62*
SLF	Particle Control	12.50	18.75	9.68	17.24	21.88	34.38
	SRM1649b	46.88*	53.12*	24.14	37.04	68.75*	84.38*
DMSO	Particle Control	31.25	12.50	0.00	20.00	31.25	31.25
	SRM1649b	75.00*	56.25*	17.24	40.74	90.62*	90.62*
Soluble Fraction
Water	Particle Control	12.50	6.25	6.45	12.90	18.75	18.75
	SRM1649b	9.38	28.12	26.67	37.04	34.38	59.38*
SLF	Particle Control	6.25	40.62	12.90	13.04	18.75	50.00
	SRM1649b	25.00	25.00	23.33	55.17*	46.88	75.00*
DMSO	Particle Control	18.75	18.75	6.45	16.67	25.00	34.38
	SRM1649b	62.50*	84.38*	33.33	16.67	90.62*	96.88*
Insoluble Fraction
Water	Particle Control	18.75	6.25	6.25	6.25	25.00	12.50
	SRM1649b	3.12	21.88	32.26	48.39*	34.38	68.75*
SLF	Particle Control	12.50	6.25	6.45	19.35	18.75	25.00
	SRM1649b	28.12	15.62	30.00	48.39*	56.25*	62.50*
DMSO	Particle Control	3.12	9.38	9.38	10.34	12.50	18.75
	SRM1649b	15.62	31.25	43.33*	45.16*	56.25*	75.00*

Percent incidence recorded at 120 hpf for all particle control and SRM1649b treatment preparations at 200 μg/mL (n=32/treatment). All preparations were at a final volume of 1 % medium (water – ultrapure water; SLF – synthetic lung fluid; DMSO – dimethyl sulfoxide) in embryo medium with particle controls in the appropriate medium at 200 μg/mL of carbon black. Significance of SRM1649b treated zebrafish from controls was defined as incidences that were statistically significant (above threshold) from medium (final concentration of 1 % water, SLF, or DMSO in embryo medium) or particle (200 μg/mL of carbon black in appropriate medium) controls and were indicated by bold text or *, respectively. A Fisher’s Exact test was used to identify the significance threshold.

3.1.1 Sample preparation

Significant mortality compared to corresponding medium and particle controls, was observed in a subset of the SRM1649b prepared by vortexing or sonication of the whole particle suspension and soluble fraction preparations, but not for any of the preparations in the insoluble fraction. Significant increases in mortality were also observed for the whole particle suspension (p=0.029) and soluble fraction (p=0.022) compared to the insoluble fraction. Thus, the insoluble fraction of SRM1649b is not responsible for the significant mortality observed and the soluble components initially bound to the particles were driving the observed toxicity. Additionally, there was a significant difference in mortality between DMSO and water (p=0.006), demonstrating the importance of medium selection when preparing PM samples. Sample preparation impacted mortality in developing zebrafish through both the medium and fraction of sample tested.

Significant morphological changes were observed in all fractions from sonication but not vortexing, highlighting a potential bioactivity bias created by PM preparation. Specifically, there were significant increases in morphological changes associated with medium and particle controls following treatment of SRM1649b that was vortexed in DMSO while sonication resulted in significant morphological changes associated with all three media tested. Based on this and the particle size distribution data (see Figure S1), it is apparent that particle size can lead to sonication-dependent morphological changes in developing zebrafish. Sonication results in a broader size range of particles and these particles are toxic regardless of the medium used in preparation. Consistent with the mortality results, sample preparation differentially affected morphology in developing zebrafish.

Regarding mortality and morphological outcomes, sonication of SRM1649b resulted in significant increases in all media and fractions. Vortexing effects were fraction and medium specific. Overall for any observed effect, there was a significant increase in biological responses associated with the suspension method (sonication vs. vortexing, p<0.001). This confirms the potential for a preparation method bias that should be eliminated by adopting sonication as standard practice when preparing PM for toxicological study.

3.1.2 Particle control

For all preparation methods, particle controls (200 μg/mL of carbon black in appropriate medium) did not significantly differ from medium controls (1 % water, SLF, or DMSO in embryo medium). Carbon black has previously been used in toxicology studies as a model particle to determine if the physical presence of particles causes toxicity (Kong et al. 2017) or if the compounds sorbed to the particles are responsible for toxicity. These findings indicate that the presence of particles in the embryo medium of developing zebrafish was not responsible for the mortality or morphological outcomes observed. However, significant differences were observed for morphological outcomes associated with exposure to the SRM1649b insoluble fraction suggesting that the particle size of SRM1649b may have been associated with developmental toxicity at 120 hpf (Fig. 1). The larger SRM particle size distribution was confirmed by comparing carbon black to SRM1649b samples prepared by vortexing and sonication through dynamic light scattering (see Figure S1). As the particle size distribution of carbon black did not replicate that of SRM1649b, carbon black was not a valid particle control for the putative SRM1649b particle size associated toxicity. An appropriate carbon black size distribution could be engineered to serve as a future control. But an alternative interpretation is that the SRM1649b needed further extraction with multiple solvents to remove bioactive compounds sorbed to the particles. Both of these options should be addressed in follow-up studies.

Figure 1.

Representative images at 120 hpf after exposures starting at 6 hpf in DMSO. Images displaying morphology of larvae following exposure to A) 1 % DMSO control (medium control), B) 200 μg/mL whole particle suspension of carbon black in DMSO (particle control), C) 200 μg/mL whole particle suspension of SRM1649b in DMSO, and D) 200 μg/mL insoluble fraction of SRM1649b in DMSO. With pericardial and yolk sac edema (C) and bent axis (C and D) indicated with arrows.

Our results thus far emphasize the importance of a standardized protocol for future zebrafish PM exposure studies to allow meaningful comparisons between studies and research groups. To this end, a refined set of sonicated preparations were studied in the concentration ranges described below, with vortexed SRM1649b preparations no longer considered.

3.2 Concentration Ranges

Sonication of SRM1649b in all re-suspension media and fractions were selected for further study at a range of concentrations (0, 12.5, 25, 50, 100, and 200 μg/mL) with comparisons made to medium controls (1 % water, SLF, or DMSO in embryo medium). The highest concentration was not increased above the initial testing because higher concentrations of SRM1649b exposure were found to result in mortality and morphological changes that prevented the behavioral assays from having adequate statistical power. These findings are consistent with previous mortality assessments of zebrafish exposed to concentrations over 200 μg/mL of ambient fine PM (Duan et al. 2017).

3.2.1 Mortality and Morphological Changes

Visual assessments at 24 and 120 hpf were made to determine mortality and morphological changes following exposure to medium controls and PM treatments. Significant increases in any observed effect occurred at the highest concentration (200 μg/mL) of SRM1649b, in all three media, compared to appropriate media controls. There were no significant differences between the three medium controls. When comparing percent incidence of individual effects between different fractions and types of medium (Fig 2), in general the higher concentrations (100, 200 μg/mL) of SRM1649b had increased mortality and morphological outcomes compared to lower concentrations. Mortality was mainly observed in the water and DMSO preparations and absent in the SLF preparations suggesting unique bioactivity compared to extraction in the other medium. A majority of the measured endpoints were elevated following exposure to the whole particle suspension in DMSO, soluble fraction from DMSO extraction, and insoluble fraction from SLF extraction with 200 μg/mL of SRM1649b compared to the other preparations. Importantly, the trends observed in morphological endpoints were similar between the whole particle suspension in DMSO and soluble fraction from DMSO extraction of SRM1649b, but the incidences of aberrant morphology were not present in association with exposure to the insoluble fraction from DMSO extraction. This suggested that the SRM1649b soluble fraction from DMSO extraction was the most bioactive.

Figure 2.

Heatmap of Mortality and Morphological Changes following SRM1649b exposure starting at 6 hpf. Z-scores of each column (mortality or morphological endpoint) are reported for all media (water, SLF: synthetic lung fluid, and DMSO) and preparations (WPS: whole particle suspension, soluble, and insoluble fractions) at each concentration of SRM1649b. Only observed endpoints are displayed from the 22 screened endpoints. Z-scores reflect the number of standard deviations from the mean incidence of the endpoint (column). Increasing color intensity indicates an increased difference from the mean incidence of an outcome. Each preparation, originally n=32, was used for zebrafish exposures starting at 6 hpf with data reported for assessments taken at 24 and 120 hpf, unless noted the data reported was observed at 120 hpf.

3.2.2 Embryonic Photomotor Response (EPR)

To determine if exposure to SRM1649b starting at 6 hpf had effects below the threshold for gross toxicity (mortality/malformations) we also tested for subtle behavior changes in photomotor response assays. The EPR assay was conducted at 28 hpf, this assay does not directly identify neurotoxic chemicals, since non-neural targeting chemicals can have low dose effects manifest in behavior. However, we note that the zebrafish EPR is a robust predictor of future adverse outcomes of chemical exposure (Reif et al. 2016).

Significant hypoactivity was observed from the medium control (1 % DMSO in embryo medium) in embryos exposed to DMSO extraction of SRM1649b (200 μg/mL, Fig. 3), in the background (43, 46 %) and excitatory (76, 59 %) phases for the whole particle suspension and soluble fractions, respectively. The insoluble fractions of 200 μg/mL of SRM1649b extracted in water and SLF also showed significant hypoactivity with fold changes from controls of 44 and 54 %, respectively, in the excitatory phase (K-S test, p<.05, delta>40%). Thus, at the highest tested SRM1649b concentration, whole particle suspensions in DMSO and the soluble fraction from DMSO extraction impacted early photomotor responses. SRM1649b preparations from water and SLF extractions did not have significant effects, aside from in the insoluble fraction. These results suggested that the DMSO soluble components of SRM1649b contained most of the PM derived bioactivity but that significant bioactivity.

Figure 3.

Embryonic Photomotor Response (EPR) following SRM1649b exposure at 6 hpf. Embryo movement before, during, and after 2 pulses of light separated by 10 s measured at 28 hpf. Each group initially consisted of 32 embryos (any dead or malformed animals were excluded from analysis) that were exposed to a range of SRM1649b concentrations (0, 12.5, 25, 50, 100, or 200 μg/mL) in 1 % DMSO for the A) whole particle suspension in DMSO and B) soluble fraction from DMSO extraction. Significance from medium control (0 μg/mL SRM1649b, 1 % DMSO) during both the background (10 s prior to pulse of light 1) and excitatory (10 s interval between pulses of light) phases was determined by K-S test, p<.05, delta>40% and indicated by *. The number of animals (initially n = 32) for A) the whole particle suspension in DMSO was 29, 31, 28, 30, 29, and 27 and 31, 30, 29, 30, 29, and 30 for B) the soluble fraction from DMSO extraction at 0, 12.5, 25, 50, 100, 200 μg/mL of SRM1649b, respectively.

Comparisons of EPR behavior in the background and excitatory phases were made between the high SRM1649b concentration (200 μg/mL) and the range of concentrations for each of the preparation methods (Fig. 4) to characterize the concentration response profile. Significant hyperactivity was observed in the background phase associated with the lower SRM1649b concentrations (12.5, 25, and 50 μg/mL) prepared in DMSO, compared to the 200 μg/mL whole particle suspension in DMSO and soluble fraction from DMSO extraction. Significant EPR effects in the background phase were not observed with SRM1649b extracted in water or SLF.

Figure 4.

Comparison of EPR between SRM1649b concentrations. Significant differences in movement for SRM1649b concentrations (12.5, 25, 50, 100 μg/mL) compared to the high SRM1649 treated group (200 μg/mL) were determined for each medium and preparation during the A) background and B) excitatory phases of the EPR assay at approximately 28 hpf. Each group initially consisted of 32 embryos (dead or malformed animals were removed from analysis, viable animal counts for each treatment are provided in Table S1) that were exposed, starting at 6 hpf, to a range of SRM1649b concentrations (12.5, 25, 50, 100, or 200 μg/mL) in 1 % medium (Water, SLF, or DMSO). The percent change in movement relative to the movement of the 200 μg/mL SRM1649b group of the same medium and fraction was reported. Significant differences in movement from the 200 μg/mL SRM1649b group in the appropriate medium and fraction was determined by K-S test, p<.05, delta>40% and indicated by *.

EPR in the range of SRM1649b concentrations were compared to EPR at the highest SRM1649b concentration (200 μg/mL) during the excitatory phase (Fig. 4B). There was significant hyperactivity associated with all concentrations (0–100 μg/mL) of SRM1649b prepared in DMSO compared to the 200 μg/mL whole particle suspension in DMSO and soluble fraction from DMSO extraction. Significant excitatory phase increases from the 200 μg/mL treatments were also observed in the whole particle suspensions in water and SLF but not the corresponding soluble fraction preparations.

3.2.3 Larval Photomotor Response (LPR)

Swimming behavior was tracked at 120 hpf in the zebrafish larvae during light/dark transitions to measure total movement during the photomotor assay. Normal LPR behavior in zebrafish is characterized by little movement in the light phases and pronounced increases in swim distance during the dark phases, tapering off over the duration of the phase. For all SRM1649b preparations and concentrations in DMSO the combined light/dark phase movement showed significant LPR changes relative to the medium control (1 % DMSO in embryo medium). At 200 μg/mL of SRM1649b the percent change in total swimming distance throughout the light/dark transitions was reduced (hypoactive) by 68 % in the whole particle suspension in DMSO (Fig. 5A) and 54 % in the soluble fraction from DMSO extraction (Fig. 5B). Hyperactivity relative to the medium control LPR was observed in the soluble fraction from water extraction (200 μg/mL, 43 %) and from SLF extraction (25 μg/mL, 41 % and 100 μg/mL, 42 %). None of the concentrations or preparations of the insoluble fraction resulted in significant changes in LPR from the appropriate medium controls.

Figure 5.

Larval photomotor response to alternating light and dark cycles following SRM1649b exposure starting at 6 hpf. Data are reported as mean ± standard deviation at each second time bin for total movement during each phase over time. The number of animals (initially n = 32) for A) the whole particle suspension in DMSO was 29, 27, 27, 30, 27, and 13 and 27, 28, 27, 27, 28, and 17 for B) the soluble fraction from DMSO extraction at 0, 12.5, 25, 50, 100, 200 μg/mL of SRM1649b, respectively.

As anticipated from with the gross toxicity and EPR findings above, the LPR was also most strongly affected by the whole particle suspension and soluble fraction from DMSO extraction preparations of SRM1649b. This trend was not observed in the water or SLF preparations.

3.2.4 Standardized Protocol Selection

While preparations of SRM1649b in the three media tested had observable developmental effects, DMSO was the only medium that resulted in concordant outcomes between the whole particle suspension and soluble fraction across morphological and behavioral assessments. The insoluble fraction from DMSO extraction had very modest effects in all biological responses measured, suggesting that the presence of particles alone was not responsible for the biological outcomes. As an organic solvent, DMSO dissolves many compounds present in ambient PM including transition metals, inorganic salts, and nitrates (Kim et al. 2015) and known organic hazards (e.g. polycyclic aromatic hydrocarbons, PAHs) (Kim et al. 2013a). DMSO provides the additional benefit of being a well-established vehicle in zebrafish HTS (Letamendia et al. 2012; Padilla et al. 2012), eliminating concerns of developing additional exposure methods. Selection of DMSO as the media for the standardized protocol enables use of a commonly available laboratory solvent that is routinely used for zebrafish chemical testing and results in robust biological responses following PM treatment. For purposes of a single HTS protocol, DMSO extraction was selected, however this does not diminish the importance of using alternative solvents for future extraction of ambient PM. Due to the variability in sources and meteorology, PM composition will vary (Philip et al. 2014b) and depending on the anticipated components (water or organic soluble) combinations of the medium tested in this research could be utilized to ensure this variability in PM composition is captured in the exposures.

Concordance between the whole particle suspension in DMSO and soluble fraction from DMSO extraction would suggest that the easier of the two to work with, the soluble fraction, is likely a sufficient representation of the PM hazard potential. Selecting a single preparation is cost effective and reduces the sample mass required for studies, extending the of use ambient collected samples. Additionally, use of a soluble fraction avoids the inherent concerns of agglomeration and sedimentation of particles when using a whole particle suspension (Knaapen et al. 2002) and the potential confounding effects on bioactivity. We therefore believe that the best approach is use of the soluble fraction from DMSO extraction as the standard preparation method for PM screening in zebrafish.

3.2.5 Cyp1a expression following exposure to the soluble fraction of SRM1649b from DMSO extraction

With selection of the soluble fraction as the best approach to anchor standardization, we sought to evaluate PM associated spatial induction of CYP1A expression. CYP1A induction is a well characterized biomarker for aryl hydrocarbon receptor (AHR)-mediated xenobiotic metabolism (Andreasen et al. 2002) and was recently associated with protection against the teratogenic effects of PM exposure in developing zebrafish (Massarsky et al. 2016). CYP1A induction by the soluble fraction of SRM1649b from DMSO extraction was evaluated in embryos of the Tg(cyp1a:nls-egfp) transgenic reporter line starting at 6 hpf (Figure 6). We note that developmental toxicity responses to the soluble fraction of SRM1649b from DMSO extraction did not differ between the reporter line and Tropical 5D strain used for initial methods development (raw data in SI). The 1% DMSO medium control group did not exhibit distinct cyp1a/GFP expression while all SRM1649b exposed animals exhibited concentration dependent expression in the gut (starting at 12.5 μg/mL), vasculature (starting at 50 μg/mL), and liver (starting at 100 μg/mL).

Figure 6.

Cyp1a Expression Following Exposure to SRM1649b. Representative images of cyp1a/GFP expression in Tg(cyp1a:nls-egfp) transgenic reporter larval zebrafish (120 hpf) following exposure to varying concentrations of the soluble fraction of SRM1649b from DMSO extraction (0–200 μg/mL) starting at 6 hpf. Brightfield overlay images with cyp1a/GFP expression are displayed in the left column with corresponding fluorescence images of cyp1a/GFP expression in the right column, all images were captured using a Keyence BZ-X700 fluorescence microscope.

Similar tissue specific expression has been previously observed in this reporter line at similar exposure time points for individual PAHs (Chlebowski et al. 2017a; Xu et al. 2015), including PAHs present in SRM1649b (NIST 2016): benzo[a]pyrene, 7-Nitrobenz[a]anthracene, 2-Nitrofluoranthene, 3-Nitrofluoranthene, 3-Nitrophenanthrene, 9-Nitrophenanthrene, and 1-Nitropyrene. While these individual PAHs were tested at concentrations orders of magnitude higher than found in SRM1649b, the potential for additive or synergistic effects of these and additional PAHs in the samples together with the presence of a variety of additional compounds (trace metals, nitrates, sulfates, organic compounds) found in PM may have led to the similar cyp1a expression patterns observed. Continued mechanistic research on complex ambient PM mixtures coupled with known developmental zebrafish responses to individual components will help identify the mechanisms driving adverse health effects from PM exposure.

3.2.6 Biological Significance and Human Health Relevance

Developmental exposure of zebrafish to SRM1649b resulted in significant changes in bioactivity and CYP1A expression. SRM1649b was selected for this research as it is a well-characterized, commercially available urban PM sample. Use of this mixture allowed for the avoidance of using compositionally different PM samples during methods development which we hypothesize will alter the biological response data. Gross morphological outcomes and mortality resulted from SRM1649b exposure and comparable endpoints were observed with single PAH exposures of compounds measured in SRM1649b including parent (i.e. benz[a]anthracene, dibenzothiophene, pyrene) (Goodale et al. 2013; Incardona et al. 2006; NIST 2016) oxygenated (i.e. 1,4-Benzoquinone, 1,4-Naphthoquinone,1,2-Naphthoquinone) (Knecht et al. 2013; Santos et al. 2016), and nitro-PAHs (i.e. 3-nitrobenzanthrone, 5-nitroacenaphthalene, 9-nitrophenanthrene) (Chlebowski et al. 2017b; Santos et al. 2016). These observed developmental effects from some PAHs are known to be mediated by AHR (Goodale et al. 2015). The mixture of PAHs and additional compounds present in PM, may also be AHR-mediated, as suggested by the observed CYP1A expression following SRM1649b exposure. Similar developmental toxicity has been observed following exposure to the well-studied AHR agonist 2, 3, 7, 8-tetrachlorodibenzo-p-dioxin (TCDD) (Carney et al. 2004; Peterson et al. 1993), further indicating the AHR-mediated response to SRM1649b. Beyond PAHs, SRM1649b contains a number of compounds including dioxins and polychlorinated biphenyls (PCBs) that also should be further explored as possible compounds causing the proposed AHR-mediated response to SRM1649b. The potential for AHR- and/or CYP1A-independent toxicity following PM exposure in zebrafish has yet to be thoroughly investigated but is likely based on previous research with individual compounds that are frequently present in ambient PM (Gerlach et al. 2014; Incardona et al. 2005; Scott et al. 2011).

Zebrafish are a surrogate model to study the systemic effects of PM-related human health effects. While the primary human exposure route to PM is via inhalation, study of the systemic effects of an immense variety of PM samples is currently unobtainable using other models. When considering inhalation exposures, future research into the physicochemical changes in PM from an aqueous exposure scenario are necessary. However, in utero exposures occur via maternal exposure while fetal lung fluid is present, an issue not frequently addressed in traditional inhalation toxicology research. Due to the known developmental human health effects associated with PM exposures (DeFranco et al. 2015; Volk et al. 2013), use of the developmental zebrafish model in these exposure conditions is relevant to human exposures. Recent research has shown that compounds in PM associated with human health effects also result in elevated developmental toxicity in zebrafish (Mesquita et al. 2014) and zebrafish skin has been proposed as a predictor of responses in mammalian lung epithelium (Stevens et al. 2017). This platform allows for rapid collection of biological activity data and potential mechanistic information in a model that has conserved developmental processes and high genetic homology to humans (Howe et al. 2013). Harnessing the advantages of zebrafish to study developmental and systemic effects related to PM has the potential to address many of the research gaps that remain regarding PM-associated health effects in humans (i.e. underlying mechanisms, causal components, and variability in PM composition and the resultant health implications).

Conclusions

The developmental zebrafish is a powerful HTS platform that readily supports multidimensional readouts of mixture bioactivity. Our report suggests that future studies could leverage the model with a standardized protocol like the one we presented to conduct high-throughput PM screens with high data concordance between PM samples and research groups. The limited existing data from ambient PM exposures is generally concordant with our findings that PM preparations, especially the soluble fraction from DMSO extraction, is highly bioactive. This concordance was found despite varying methods used between research groups (Duan et al. 2017; Mesquita et al. 2016). Thus, we are optimistic that the method standardization outlined here will enable robust comparisons of toxicity of different PM samples across the field of study. To perform efficient HTS of ambient PM in the future, a single method was selected that best queried multiple dimensions of chemical phenotype, from gross toxicity to subtle behavioral changes and an initial foray into underlying signaling events such as the CYP1A adaptive response to PM. Use of this consistent, scalable, and sensitive platform creates the potential of a surrogate in vivo system for PM hazard identification and a path to better understand the mechanisms underlying PM-associated human health effects.