Accepted Manuscript Title: Behavioural toxicity assessment of silver ions and nanoparticles on zebrafish using a locomotion profiling approach Author: Giedr˙e Aˇsmonait˙e Scott Boyer Karine Bresolin de Souza Britt Wassmur Joachim Sturve PII: DOI: Reference:

S0166-445X(16)30019-4 http://dx.doi.org/doi:10.1016/j.aquatox.2016.01.013 AQTOX 4300

To appear in:

Aquatic Toxicology

Received date: Revised date: Accepted date:

15-7-2015 27-11-2015 25-1-2016

Please cite this article as: Aˇsmonait˙e, Giedr˙e, Boyer, Scott, Souza, Karine Bresolin de, Wassmur, Britt, Sturve, Joachim, Behavioural toxicity assessment of silver ions and nanoparticles on zebrafish using a locomotion profiling approach.Aquatic Toxicology http://dx.doi.org/10.1016/j.aquatox.2016.01.013 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

BEHAVIOURAL TOXICITY ASSESSMENT OF SILVER IONS AND NANOPARTICLES ON ZEBRAFISH USING A LOCOMOTION PROFILING APPROACH

Giedrė Ašmonaitėa*, Scott Boyerb, Karine Bresolin de Souzaa, Britt Wassmura, Joachim Sturvea

a

Department of Biological and Environmental Sciences, University of Gothenburg, Medicinaregatan

18 A, Box 463, SE-41390, Göteborg, Sweden b

Swedish Toxicology Sciences Research Centre, Swetox, Forskargatan 20, 15136, Södertälje,

Sweden

*

Corresponding author: [email protected]

Address: Zoologen, Medicinaregatan 18 A, Box 463 40530 Göteborg. Tel.: +47 29 135 659

1

Highlights 

Developmental and behavioral toxicity of Ag+ and AgNPs is comparatively studied



This study presents a novel locomotion profiling approach in FET framework



Locomotion-based behavioral endpoints are sensitive biomarkers for sub-lethal toxicity



Use of dose-response ranking for multiple locomotion variables is commenced



Multivariate statistics eases the analysis of quantitative behavioral data

Abstract Zebrafish (Danio rerio) is not only a widely used species in the Fish Embryo Toxicity (FET) test but also an emerging model in behavioural ecotoxicology. By using automatic behaviour tracking technology, locomotion of developing zebrafish (ZF) larvae can be accurately recorded and potentially used in an ecotoxicological context to detect toxicant-induced behavioural alterations. In this study, we explored if and how quantitative locomotion data can be used for sub-lethal toxicity testing within the FET framework. We exposed ZF embryos to silver ions and nanoparticles, which previously have been reported to cause neurodevelopmental toxicity and behavioural retardation in early-life stages of ZF. Exposure to a broad range of silver (Ag+ and AgNPs) concentrations was conducted, and developmental toxicity was assessed using FET criteria. For behavioural toxicity assessment, locomotion of exposed ZF eleutheroembryos (120 hpf) was quantified according to a customized behavioural assay in an automatic video tracking system. A set of repeated episodes of dark/light stimulation were used to artificially stress ZF and evoke photo-motor responses, which were consequently utilized for locomotion profiling. Our locomotion-based behaviour profiling approach consisted of 1) dose-response ranking for multiple and single locomotion variables; 2) quantitative assessment of locomotion structure; and 3) analysis of ZF responsiveness to darkness stimulation. We documented that both silver forms caused adverse effects on development and inhibited hatchability and, most importantly, altered locomotion. High Ag+ and AgNPs exposures significantly suppressed locomotion and a clear shift in locomotion towards inactivity was reported. Additionally, we noted that low, environmentally relevant Ag+ concentrations may cause subordinate locomotive changes (hyperactivity) in developing fish. Overall, it was concluded that our locomotion-based behaviour-testing scheme can be used jointly with FET and can provide endpoints for sub-lethal toxicity. When combined with multivariate data analysis, this approach facilitated new insights for handling and analysis of data generated by automatized behavioural tracking systems.

2

Keywords: Zebrafish; nanoparticles; locomotion profiling; behaviour; FET

1. Introduction Silver nanoparticles (AgNPs) are the most extensively used inorganic metallic nanoparticles primarily due to their effective antibacterial and antifungal properties (Dziendzikowska et al., 2011; Cleveland et al., 2012; Farkas et al., 2011). Despite the extensive applicability and socio-economic benefits of AgNPs, safety concerns arise when these metallic NPs are released into the environment (Farkas et al., 2011). The predicted environmental concentrations of AgNPs are largely variable, ranging from ng/L to µg/L, depending on the environmental compartment (Fabrega et al., 2011; Gottschalk et al., 2009; Hoque et al., 2012). Silver has the potential to alter the neurodevelopment of zebrafish embryos and cause behavioural retardation (Powers et al., 2010; 2011a; 2011b). AgNPs can passively cross the chorion membrane through pore channels and accumulate in the developing embryos (Lee et al., 2007). Under such circumstances, embryo organogenesis is targeted by both direct and indirect nanoparticle exposure and may cause adverse effects on the developing organism. It was previously reported that AgNPs can interrupt the normal functioning of the nervous system and alter signal transduction processes (Griffitt et al., 2008). The accumulation of AgNPs in the head region of zebrafish embryos was documented, as well as head and eye hypoplasia and cardiac dysfunction (Asharani et al., 2011; Lee et al., 2007; Xin et al., 2015). A recent study by Xin et al. (2015) showed that AgNPs may act as neurotoxicants, affecting the gene expression of neural development-related genes in developing zebrafish. It has been reported that the negative consequences of AgNPs accumulation in the central nervous system might result in the disruption of cardiac rhythm, respiration, and body motion (Asharani et al., 2008; 2011; Lapresta-Fernández, 2012; Powers et al., 2011b). The assessment of fish behaviour is a promising tool for ecotoxicological studies, as it reveals toxicant induced complex physiological consequences at the individual level and has the potential to complement conventional toxicological methods with higher ecological relevance (Scott and Sloman, 2004; Zala and Penn, 2004). Behavioural studies are gaining more recognition within the field of ecotoxicology due to being 10-1000 times more sensitive in comparison to conventional LC50 values (Gerhardt, 2007; Little et al., 1990; Scott and Sloman, 2004) Additionally, behaviour

3

studies utilizing fish (eleuthero) embryos are compatible with the 3Rs principle and ethical considerations (Belanger et al., 2010; Embry et al., 2010, Strähle et al., 2012). Zebrafish Danio rerio (ZF) is not only widely used vertebrate model is biomedical research, but also a promising model in behavioural ecotoxicology and can be used for high throughput developmental and behavioural screening (Ahmad et al., 2012). ZF possesses a wide array of sophisticated and complex behaviours (Ali et al., 2012) which can be utilized in ecotoxicological assessment to evaluate potential threats of environmental pollutants. After rapid embryogenesis, larval ZF at 5 days post fertilization have motor and visual functions readily developed and embody a battery of robust locomotion patterns, such as optomotor response, phototaxis and escape responses (Burgess, Granato, 2007). From an integral standpoint, locomotion serves as a key element in foraging, social and defensive activities (Colwill, Creton, 2010). Locomotion-based endpoints can provide essential information not only about the integrity of motor function, which is substantial for fitness and survival of an organism, but also about the functionality of sensory systems. Moreover, with locomotion, it is possible to detect patterns of simple adaptive cognitive behaviours, such as habituation (Best et al., 2008; Roberts et al., 2011; Wong et al., 2010). In the present study, we primarily explored if and how quantitative behaviour tracking (locomotion) data can be used in sub-lethal toxicity screening within FET test framework. We seek to develop a conceptual and methodological approach for ZFEEs locomotion profiling and design a pipeline for data handling. First, we exposed ZF embryos to a wide range of Ag+ and AgNPs concentrations and examined embryonic toxicity. Thereafter, we quantified locomotion of exposed ZF eleutheroembryos (ZFEEs) in a photo-motor test. Retrieved data were analysed to unveil Ag+ and AgNP toxicity on behaviour using locomotion profiling that consisted of 1) dose-response ranking for multiple and single locomotion variables; 2) quantitative assessment of locomotion structure; and 3) analysis of ZFEEs responsiveness to darkness stimulation.

2. Materials and Methods 2.1 Chemicals and Reagents Polyethylene glycol 5000 (PEG-5000) stabilized AgNPs were purchased from Cline Scientific AB (SE), and silver nitrate AgNO3 powder (CAS Nr. 7761-88-8, 204390-1G) was purchased from Sigma Aldrich (DE). For toxicant exposures, AgNO3 and AgNPs were diluted in embryo medium (1 L of buffer contained 245 mg of MgSO4*7H2O, 20.5 mg of KH2PO4, 6 mg of Na2HPO4, 145 mg of 4

CaCl2*2H2O, 37.5 mg of KCl, 875 mg of NaCl dissolved in MiliQ water). The control treatment contained solely embryo medium.

2.2 Physicochemical characterization of AgNPs For AgNPs characterization in ZF embryo medium, dynamic light scattering (DLS, Malvern Zetasizer Nano ZS, ZEN 3600, UK) was used. The hydrodynamic size and zeta potential were measured in triplicates for 6 experimental time-points (0, 24, 48, 72, 96, 120 h) at 1 mg/L concentration. Zeta potential ζ (-mV) was calculated using electrophoretic mobility data with a consideration for ionic strength at 25°C, assuming that the ionic strength of the test medium did not change significantly during the experiment (Ohshima, Hiroyuki, 2012). Transmission electron microscopy (TEM) has been used for AgNPs in stock solution and was performed by the manufacturer (Cline Scientific AB, SE). The poly-L-lysine coated copper grid with attached AgNPs particles was examined using a LEO 912AB Omega electron microscope (Carl Zeiss SMT, Oberkochen, DE). Digital images were captured with a MegaView III camera (SIS, Münster, DE). The diameter of at least 100 particles was measured using the software Image J (National Institute of Health, USA). Additionally, an inductively coupled plasma mass spectroscopy (ICP-MS) was used to determine the Ag+ concentration in the AgNPs stock suspension (ALS Scandinavia, SE).

2.3 Zebrafish husbandry and egg collection Wild-type AB zebrafish (Danio rerio) were raised and bred in the Institute of Neuroscience and Physiology, University of Gothenburg. ZF were kept under the following standard laboratory conditions: 14:10 h light:dark cycle at 27°C in a re-circulating water system. After routine precision breeding, fertilized ZF eggs from at least 3 different breeding pairs were collected and randomly mixed to account for inter-population variability during the experiments (OECD guidelines 236, 2013). The embryos were inspected for viability, and unfertilized eggs and eggs with discrepant developmental stage were excluded.

5

2.4 Exposure design The ZF exposure experiments were static and were arranged independently in triplicates for each silver treatment group (Ag+ and AgNPs). In each experimental trial, ZF embryos were exposed for 5 days to 10 nominal silver concentrations within a range of 2.15 ng/L-2.15 mg/L. For the first 24 h, randomly selected fertilized embryos (~6 hours post fertilization (hpf)), N=30 per concentration) were exposed in glass Petri dishes (30 mL) with known toxicant concentrations. After 24 h, embryos were transferred to U-shaped 96-well microplates containing 200 μL of the exposure solution and kept individually until 96 hpf, when normally developed and hatched individuals were selected for behavioural toxicity assessment.

2.5 Fish Embryo Toxicity Test Embryonic toxicity assessment was performed using the Fish Embryo Toxicity test (FET) guidelines. By using a light stereoscope (EZ4HD LEICA, CH), the developmental endpoints were assessed for certain time points according to OECD recommendations. Mortality (LC50) after 48 h and inhibition of hatchability (EC50) at 6 dpf were determined using Probit model.

2.6 Behavioural toxicity assessment 2.6.1 Behavioural assay The locomotion (movement) assessment of ZFEEs at 120 hpf was performed using the View Point ® automatic behaviour tracking system (Zebralab version 3.22, ViewPoint Life Science Inc., Montreal, CN). Only visually healthy individuals with no apparent malformations or defects were used, as suggested by Padilla et al. (2011) and Powers et al. (2011b). A customized protocol composed of 18 interchanging dark:light cycles (5:5 min) was used to artificially stress ZFEEs and induce the photomotor response of developing larvae (Fig. 1A), as it has been previously described by MacPhail et al., (2012). We composed an assay with interchanging cycles to study the dynamics of photo-motor response in different light and dark phases during the experiment. Every trial was semi-randomized and included ZFEEs from all exposure concentrations, which were tracked simultaneously. Setup was arranged in 48 well-plates, and each well contained 500 μL of treatment/control solutions. Plates were prepared one day before tracking experiments and were considered to be technical replicates. Before every behavioural experiment, ZFEEs were acclimatized in light for 15 min to minimize 6

effects of handling and stabilize the baseline locomotor activity. The locomotion measurements were integrated over 10 second intervals for a 1.5 hour-long experiment. Two-dimensional (horizontal) locomotion variables (distance, duration and activity) were automatically measured in predefined arenas within well-plates. Endpoints on locomotion structure included continuous movements (large movements of a particular direction longer than 6.1 mm), non-ambulatory movements (small movements ranging 2.1-6.1 mm) and inactivity (minor movements ranging 0-2.1 mm), which were set according to Viewpoint recommendations. The movement tracking was registered at a rate of 25 images per second (according to CCIR standards).

2.6.2 Data preparation for analysis Prior to locomotion profiling and statistical data analysis, raw behavioural tracking data were sorted for every tested individual in the experimental cohort, and total values for every locomotion parameter per minute were estimated. Consequently, these values were used to calculate the average and total responses for every dark/light cycle. At least 20 ZFEEs (from three independent experiments) per concentration were used for behavioural toxicity assessment. ZFEEs locomotion data for each exposure concentration from three independent experiments were merged into a single dataset. Measurement values that were outside the range of the mean ± 2 standard deviations were considered outliers and were removed for further data analysis (Selderslaghs et al., 2013).

2.7 Locomotion profiling approach 2.7.1 Dose-response ranking Hierarchical Clustering (HC) and Principal Component Analysis (PCA) were utilized for analysis of data matrices composed of a set of dark and light phase locomotion variables (216 variables) (Supplementary material Table 1). HC (Ward Method with data standardization as implemented in JMP11, Salford Systems) was used to explore the relative ranking of the doses for the total set of dark and light phase locomotion variables. It was also used to test the ability of dark and light phase variables to produce a meaningful dose-related ranking of the groups by themselves and finally to assess the overall clustering separation. After finding that dark and light phase variables did indeed cluster separately in the HC analysis, the dark and light phase variable matrices were analysed separately using PCA to explore the relative placement of the dose groups in multivariate space and 7

to assess the relative contribution of the variables to the placement of groups in that space. Additionally, dose-response relationships between a (single) selected cumulative dark and light phase variable across concentration gradient was established.

2.7.2 Locomotion structure assessment As locomotion activity was differentiated for large, small movements and inactivity, according to a pre-set threshold, a relative proportion of time spent (%) by each type of movements was estimated. These ratios were obtained for ZFEEs in every tested concentration and were compared with control ZFEEs to detect of changes in locomotion structure.

2.7.3 Responsiveness to darkness stimulation The sudden transition from light to dark phase evoked locomotion changes in ZFEEs from low (baseline) to high (elevated) activity levels. The dynamics of this increased activity across repeated darkness cycles was referred as responsiveness to dark stimulation. The responsiveness profiles were created to visually portray changes in locomotion activity (photo-motor response) and contained information about ZFEEs locomotion (measured as distance travelled) in each darkness cycle (Fig. 1 B). Adverse effects on locomotion (for dose-response ranking, locomotion structure and responsiveness) were determined by comparing behavioural responses of individuals from different exposure concentrations to the control ZFEEs for each treatment using one-way ANOVA with Dunnett’s test (SPSS Statistics, IBM, 22.0).

3. Results 3.1 Physicochemical characterization of AgNPs The initial TEM measurements described AgNPs in stock solution as spherical particles with a diameter range of 21.5± 2.9 nm (Fig. 2). DLS measurements showed a bimodal size distribution of AgNPs in the embryo medium (Table 1). The polydispersity index (PDI) and zeta-potential values cumulatively indicated that AgNPs were readily unstable in experimental solutions over time and were prone to form aggregations/agglomerations. Third particle population (third size peak)

8

significantly larger in size could be subsequently identified (Table 1). ICP-MS measurements showed negligible levels of Ag+ in AgNP stock suspension.

3.2 Fish Embryo Toxicity Test The exposure to Ag+ and AgNPs had a negative effect on ZF embryo survival. Ionic silver was slightly more toxic than silver nanoparticles; LC50 values for Ag+ and AgNPs were 0.235 mg/L and 0.306 mg/L, respectively. Both Ag+ and AgNPs inhibited hatchability with EC50 values of 0.034 mg/L and 0.072 mg/L, respectively. Ag+ and AgNPs caused morphological aberrations such as twisted notochord, bent tip of the tail, slightly decaying caudal fins, irregularly shaped or decaying pectoral fins, malformed eyes, yolk sac oedema, or irregular body shape, which was not observed in the control treatment. These morphological abnormalities were not treatment-specific.

3.3 Behavioural toxicity assessment 3.3.1 Endpoint selectivity Overall, higher baseline activity of ZFEEs was recorded in AgNP behavioural experiments compared to Ag+ (Fig. 5, 8, 10). There were principal similarities in the data from the two treatments; ZFEEs swimming activity was notably higher in dark compared to light phases for control individuals. Locomotion variables in light showed low or no discrimination power due to low activity levels and high data variance. Dark variables contained high discriminating ability, and we therefore focused on them primarily for behavioural toxicity assessment. As Ag+ and AgNP experiments were conducted independently, results for each treatment are presented separately.

3.3.2 Locomotion-based dose ranking: Ag+ exposure A behavioural heat-map, based on 108 dark and 108 light phase variables, was obtained from HC and showed a clear dose-response ranking (Fig. 3). For Ag+ treatment, locomotion variables vertically clustered in two distinct clusters for low and high exposures. For dark cycles, within the first cluster (red colour) relatively high locomotion activity, similar to the control level, was reported, whereas in the second cluster (blue colour) – low activity. PCA analysis revealed similar locomotion-based biphasic grouping for dark phase variables (Fig. 4.) 9

To establish dose-response relationships, a single dark and light phase variable was used. For this purpose, we selected a total swimming distance, performed by continuous movements, as a representative locomotion endpoint (Fig. 5). In dark cycles, with increasing concentrations of Ag+, a decrease in ZFEEs locomotion was reported with statistically significant locomotion suppression in concentrations of ≥0.1 mg/L (for all comparisons p ≤0.05). A slight increase in swimming activity in the low concentration range of 2.15 ng/L-0.2 µg/L was documented. In light phases, there was a tendency for hyperactivity across the concentration gradient.

3.3.3 Locomotion-based dose ranking: AgNPs exposure For AgNP exposure, a clustergram for multiple dark and light locomotion variables along the dose gradient was composed (Fig. 6) and three major vertical clusters based on locomotion were identified. Horizontally, the third small cluster on the far right of the dendrogram corresponded to predominantly “inactivity” variables and this cluster had no dose discriminating power (data not shown). PCA analysis on dark phase variables showed binary grouping for low/intermediate and high exposure concentrations (Fig. 7). A dose-response relationship for total swimming distance was established for AgNPs treatment (Fig. 8 A). In concentrations ≥0.47 mg/L, statistically significant monotonic locomotion suppression in dark cycles was reported (for all comparisons p ≤0.05). (Fig. 8 A). Additionally, an increase in swimming activity of AgNPs-exposed ZFEEs during light cycles was detected (≥0.01 mg/L) (Fig. 8 B).

3.3.4 Assessment of locomotion structure In the behavioural assays, ZFEEs locomotion was differentiated to 1) large continuous movements with a direction, 2) small short movements, and 3) inactivity, according to a selected threshold. This movement classification provided valuable information about the locomotion structure of ZFEEs at 120 hpf. On average, in darkness cycles, unexposed individuals were spending 18±6.25 % of time actively swimming with continuous large-scale movements, 45.31±5.7 % swimming with short movements, and 35.8±6.2 % of time resting in an inactive state (Fig. 9). Ag+ treatment had an effect on the locomotion structure of ZFEEs in concentrations of ≥0.001 mg/L. The proportion of time that ZFEEs allocated to swimming with large and small movements significantly decreased, whereas 10

inactivity increased (Fig. 9 A). Similar changes were observed for AgNPs-exposed individuals in a concentration range of ≥0.01 mg/L (Fig. 9 B).

3.3.5 Responsiveness to darkness stimulation The iterative reduction of locomotion in repeated darkness cycles was observed for ZFEEs in exposed and control individuals for both treatments. Robust responsiveness (locomotion) profiles representing a decrease in activity in photo-motor test in dark over time were obtained for high exposure groups to show significantly reduced locomotion activity across a repeated number of darkness cycles (Fig. 10 A, B). The positioning of the curve on the y-axis indicated the level of ZFEEs responsiveness during the photo-motor test. For the first 1-5 dark cycles, locomotion curves had relatively steep slopes, which reached an inception point and flattened out for the final cycles (79). This pattern was recorded for both Ag+ and AgNPs treatments, not only for high exposures, but also for low and intermediate exposure concentrations (Fig. 10 C). Positioning of the locomotion curve below the control baseline displayed hypoactivity across repeated darkness periods. For low Ag+ exposure concentrations, the positioning above the control curve indicated a constant increase in swimming during repeated darkness cycles.

4. Discussion Profiling of ZFEEs locomotion has a great potential to complement zebrafish embryo toxicity (ZFET) test with functional and integrative sub-lethal toxicity endpoints. Numerous descriptive locomotion parameters, containing valuable biological information, are retrieved from automatic tracking systems in controlled dark/light challenges, and once analysed with multivariate data analysis techniques may provide relevant output for behavioural toxicity screening. The novelty of this work was presented by: utilization of multiple locomotion parameters for dose response ranking, description of locomotion structure in dose response setting and assessment of ZFEEs responsiveness upon darkness stimulation (integrity of sensory and motor functions). In the present study, we demonstrated that behavioural heat-maps obtained from HC clustering and accompanied with PCA could be successfully used for both visual comparison and classification of behavioural responses across the exposure gradient. This is particularly useful for detection of subtle

11

behavioural changes and/or identification of meaningful behavioural endpoints that consequently can be used to establish dose-response relationships. For Ag+ exposure, biphasic locomotion response for dose-response ranking was observed, indicating hyperactivity and overreaction to stimulus in a low concentration range, followed by significant reduction in locomotion in a high exposure range. In ecotoxicology, this phenomenon is referred to hormesis, which is characterized by low-dose stimulation and high-dose inhibition. It is an adaptive response that relates to compensatory biological processes following an initial disruption in homeostasis (Calabrese and Baldwin, 2002). Recently, low-dose locomotion excitation has been documented for ethanol (Blaser and Penalosa, 2011), bisphenol A (Saili et al., 2012), neuroactive drugs (Irons et al., 2010), lead (Chen et al., 2012) and other chemical compounds (Ali et al., 2012). The potential for Ag+ to cause hyper-responsiveness to light changes has been previously described by Powers et al. (2011b). It was also reported that in an in vitro neurodevelopmental cell model, Ag+ disrupts cell replication and alters differentiation of neurotransmitter phenotypes (Power et al., 2011a). Importantly, the observed hormetic effect occurred at approximately 500 to 700-fold lower exposure concentrations than observed LC50 values, suggesting that locomotion-based behavioural tracking could provide higher sensitivity measures and could be used for sub-lethal toxicity screening. However, validation of locomotion stimulation trends and assessment of physiological relevance were not feasible within our experimental setup. Therefore, further mechanistic studies are conducted to address this phenomenon. Moreover, not only Ag+ but also AgNPs significantly suppressed ZFEEs locomotion in darkness periods and in high exposure concentrations. Additionally, we reported similar changes in locomotion structure of ZFEEs for both treatments. With increasing exposure concentrations, ZFEEs became less active: a clear shift towards inactivity occurred, as large continuous and small nonambulatory movements significantly decreased. Adverse effects on locomotion were observed in concentrations readily similar to LC50 values, meaning that ZFEEs that survived were negatively affected by toxicant exposure. We believe that silver-induced locomotion alterations can be comprehensively evaluated using the Stepwise Stress Model (SMM). By plotting toxicant concentration on the x-axis, ranges of toxicant doses leading to 1) no behavioural response (resistance), 2) compensatory response (regulation) or 3) serious irreversible effect (toxicity) are identified (Gerhardt, 2007). For Ag+ treatment, the observed significant reduction in swimming performance can be explained by the adverse “toxic” effect. Ag+ exposure had a significant impact on expression of several housekeeping genes (β-actin, EF1α, RPL13α) as well as the target genes (metallothionein (MT) and antioxidant transcription factor (NRF2)) in ≥0.1 mg/L concentrations 12

(Supplementary material Fig. 1), resulting in a drastic deterioration of main physiological functions of developing zebrafish. In contrast, no significant changes in gene expression levels were reported for AgNPs treatment, presumably indicating either reversible damage or NP-specific effects. These findings suggest that, even though Ag+ and AgNPs showed similar locomotion suppression, Ag+ was considerably more potent. In general, locomotion abnormalities can originate from impediments in neuronal innervation, muscle structure and neuromuscular function due to altered vision perception (Wong et al., 2010). In the present study, using FET guidelines, we documented harsh morphological malformations (twisted notochord, decaying and irregularly shaped fins) in body parts related to movement in high exposure concentrations, directly imposing on the reduced ability to perform movements or maintain coordination. Additionally, both Ag+ and AgNPs treatments delayed ZFEEs hatchings, and these findings were consistent with the existing literature (Massarsky et al., 2013; Powers et al., 2011b). Presumably, postponed hatching had an effect on the onset of swimming initiation and delayed development of ZFEEs motoric function. In this context, FET results provided a starting basis for behavioural data interpretation and validation, but more detailed studies should be performed to shed light on the toxicity mechanisms of Ag+ and AgNPs. AgNPs toxicity is highly associated with the release of silver ions from the particle surface, and it is not clear if the toxicity in AgNPs treatment occurred due to the nanoparticles, ions or a combination of the two; this subject remains highly debatable (Kwok et al., 2012). According to NP characterization data, AgNPs were not stable in experimental medium over time and formed aggregations/agglomerations that potentially reduced the bioavailability of nanoparticles during the experiment. Particles were detected by DLS even after 120 h suggesting that AgNPs did not completely dissociate in the exposure solutions. It can be suspected that AgNPs largely remained in particulate form. Therefore, we can only speculate whether the observed behavioural toxicity was arising from AgNP specific-toxicity solely or in combination with Ag+. Habituation is an ancestral form of learning that represents a decrease in response after repeated stimuli that is not due to sensory adaptation or motor fatigue (Rankin et al., 2010) and is gaining popularity in neurobehavioural research (Wong et al., 2010). ZF can habituate to numerous stimuli, including changes in lighting conditions (MacPhail et al., 2009). In this study, we confirmed that ZFEEs have an efficient ability to attenuate locomotion response after repeated dark/light cycles. Therefore, we speculate that with our behavioural assay we observed an intra-session habituation of photo-motor response. Intra-session habituation (within a trial) has been previously indicated in 13

literature (Raymond et al., 2012) and is not subjected to mid-term or long-term memory, but rather to spatial working memory (Wong et al., 2010). In this context, we would like to discuss intra-session habituation not as non-associative learning process (Rankin et al., 2010), but more as time-restricted ability to selectively filter out irrelevant stimuli. On the other hand, the observed reduction in locomotion response may also occur due to motor or sensory fatigue and is frequently questioned (Rankin et al., 2010). With our behavioural testing setup it was not feasible to address this concern, as dishabituation of the response was not tested. If elucidated, intra-session habituation may be a useful tool in ecotoxicology to detect alterations in locomotion patterns and potentially can be applied in screening procedures. Habituation responses provide important neurological information about adaptability of sensory systems and development of a cognitive map (Wong et al., 2010). Alterations in habituation caused by toxicant(s) may be a source of valuable neurobiological information and can potentially help to detect anxiety-like behaviours or learning deficits in fish, as habituation is often referred a short-term memory (Müller et al., 1994). Recently, it has been shown that some pharmacological agents can impair habituation in adult zebrafish (Wong et al., 2010); therefore, we suggest that our behavioural protocol could be a valuable tool for the detection of behavioural toxicity of certain chemical compounds that are known to cause neurotoxicity or/and possess endocrine disruption properties. Automated video tracking systems are revolutionizing behavioural research by providing large-scale locomotion data with high accuracy and precision (Stewart et al., 2013). Therefore, it is very important not only to understand prevalent implications on analysis, interpretation and validation of such data, but also to integrate these data in ecotoxicology and behavioural ecology (Gerhardt, 2007; Zala and Penn, 2004). Multivariate “similarity” analysis of locomotion data generated by photographic or video tracking systems is exploited in various contexts of animal behavioural research and can help to address some of these concerns. To this end, the inclusion of HC and PCA in our data analysis can be seen as a natural progression of analysis of digital behavioural data. We anticipate that this type of similarity analysis can not only help better understand the nature of the biological variability, but also assist in distinguishing it from the technical (experimental) variability. Many extrinsic and intrinsic factors, ranging from diurnal and seasonal activity fluctuations to genetic variability, can affect behaviour (Burgess and Granato, 2007; MacPhail et al., 2009; Padilla et al., 2011); therefore, the interpretation of behavioural data remains a great challenge. In addition, such multivariate analysis is particularly suited for addressing the issues of the uniqueness or/and the redundancy of certain behavioural variables, and may facilitate mechanistic interpretation of various changes in behaviour. Importantly, the alternative measurements (variables) gathered during tracking 14

experiments can be introduced to enrich the analysis and can aid in the validation process of behavioural assays. Standardized validation and analysis procedures of behavioural data are not readily available and, most importantly, are hindered by intrinsic data properties such as bimodality and heterogeneity of variance (Gerhardt, 2007). In the present study, overall high variability within and between behavioural datasets was observed. We reported that the baseline activity level of control individuals for the Ag+ and AgNPs treatments (based on absolute measurement values) were naturally different. Differences between two independent exposures were expected, as ZF eggs from different breeding pairs and in unknown proportions were mixed to mimic more ecologically relevant exposure scenarios, as is recommended by FET guidelines (OECD guidelines 236, 2013). The trade-off between data variability and sample size should be considered in this experimental setting. In our study, a relatively large sample size (N>20) from three independent experiments for each exposure concentration was utilized for behavioural testing, and we, therefore, argue that the observed behavioural patterns were readily robust. Additionally, the selection of relevant variables is crucial for data interpretation. HC locomotion profiling and PCA analysis highlighted higher relevance of ZFEEs responses in dark phase compared to light; therefore, higher attention was drawn to the ranking of locomotion-based variables under dark phase(s). It has been demonstrated that AgNPs can act as neurobehavioural disruptors and cause hyperactivity under various lighting conditions (Powers et al., 2011b). According to our findings, AgNPs exposure evoked locomotion/responsiveness in the light phases, but it was difficult to confirm this observation statistically due to the low discrimination power of light cycle variables. This suggests even statistically insignificant behavioural changes could be readily informative and should not be blindly eliminated from data analysis. The demand of ecotoxicological data under the European chemical policy framework, REACH, is expected to increase; therefore, non-animal testing and the development of alternative strategies will be promoted to minimize the use of animals in biomedical research. Notably, embryos and eleutheroembryos are not dependent on exogenous nutrition and, therefore, are not legally protected by European legislation (EU Animal Protection Directive 1996) and are considered as in vitro systems. Recently, the use of ZF embryos for toxicity tests was advocated to be a substitute for some in vivo animal testing (Strähle et al., 2012). Lammer et al. (2009) concluded that FET has already reached a high level of acceptance for international standardization and method validation and will be increasingly used for regulatory purposes in the near future. Behavioural profiling in combination with FET has a great potential to assess sub-lethal effects, especially those on the nervous system (Ali et al., 2012). The behavioural screening techniques that use one alternating stimuli (e.g. light) 15

allow the assessment of non-associative (reflex) behaviours, integrity of motor function, neuronal and physiological integrity and are highly compatible with 3R principle, as it takes advantage of experimental animals and uses them for retrieving high resolution biological data. Once developed, optimized and harmonized behavioural assays could significantly improve the current environmental and hazard risk assessment procedures, as behavioural endpoints possess higher ecological relevance and sensitivity.

5. Conclusions By using this locomotion profiling approach coupled with multivariate data analysis, we have attempted to provide new insights for behavioural toxicity screening. We concluded that locomotion of developing ZFEEs could be successfully used as endpoints of sub-lethal toxicity screening within the FET framework. By using a dose-response ranking approach for multiple locomotion variables, we showed that locomotion-based behavioural heat-maps provide a cumulative overview of largescale datasets and can pinpoint behavioural toxicity trends. We found out that locomotion-based variables can be used to establish concentration-response relationships and provide behavioural toxicity thresholds that can be used as complementary sub-lethal toxicity endpoints within FET. Additionally, the assessment of locomotion structure revealed how different types of body movements can be affected by toxicant exposure, whereas the analysis of responsiveness to dark stimulation suggested that via locomotion it may be feasible to study intra-session habituation. Locomotion shifts towards inactivity in high exposure concentrations were presented. Additionally, we noted that low, environmentally relevant Ag+ concentrations may cause subordinate locomotive changes (hyperactivity), and the biological and ecological significance of such behavioural alterations are being further investigated.

Conflict of interest The authors declare that there is no conflict of interest.

16

Acknowledgements This project was funded by the Swedish Research Council FORMAS. We are very grateful Dr. Petronella Kettunen, Dr. Alexandra Abramsson and Dr. Malin Edling for the professional assistance and the opportunity to use the zebrafish facilities in the Institute of Neuroscience and Physiology, University of Gothenburg. For cooperation and technical assistance for nanoparticle characterization, the authors sincerely acknowledge Dr. Geert Cornelis, Dr. Jenny Perez Holmberg, Dr. Lars Skjolding and Dr. Tobias Lammel.

References Ahmad, F., Noldus, L. P. J. J., Tegelenbosch, R. a J., & Richardson, M. K. (2012). Zebrafish embryos and larvae in behavioural assays. Behaviour, 149, 1241–1281. doi:10.1163/1568539X00003020 Ali, S., Champagne, D. L., & Richardson, M. K. (2012). Behavioral profiling of zebrafish embryos exposed to a panel of 60 water-soluble compounds. Behavioural Brain Research, 228(2), 272–283. doi:10.1016/j.bbr.2011.11.020 Asharani, P. V, Lian Wu, Y., Gong, Z., & Valiyaveettil, S. (2008). Toxicity of silver nanoparticles in zebrafish models. Nanotechnology, 19(25), 255102. doi:10.1088/0957-4484/19/25/255102 Asharani, P. V, Lianwu, Y., Gong, Z., & Valiyaveettil, S. (2011). Comparison of the toxicity of silver, gold and platinum nanoparticles in developing zebrafish embryos. Nanotoxicology, 5(1), 43– 54. doi:10.3109/17435390.2010.489207 Belanger, S. E., Balon, E. K., & Rawlings, J. M. (2010). Saltatory ontogeny of fishes and sensitive early

life

stages

for

ecotoxicology

tests.

Aquatic

Toxicology,

97(2),

88–95.

doi:10.1016/j.aquatox.2009.11.020 Blaser, R. E., & Peñalosa, Y. M. (2011). Stimuli affecting zebrafish (Danio rerio) behavior in the light/dark

preference

test.

Physiology

and

Behavior,

104(5),

831–837.

doi:10.1016/j.physbeh.2011.07.029 Burgess, H. A, & Granato, M. (2007). Modulation of locomotor activity in larval zebrafish during light

adaptation.

The

Journal

of

Experimental

Biology,

210(Pt

14),

2526–2539.

doi:10.1242/jeb.003939 17

Calabrese, E. J., & Baldwin, L. A. (2002). Defining hormesis. Human and Expreimental Toxicology, 21(2), 91-97. doi:10.1191/0960327102ht217oa Chen, J., Chen, Y., Liu, W., Bai, C., Liu, X., Liu, K., Huang, C. (2012). Developmental lead acetate exposure induces embryonic toxicity and memory deficit in adult zebrafish. Neurotoxicology and Teratology, 34(6), 581–586. doi:10.1016/j.ntt.2012.09.001 Cleveland, D., Long, S. E., Pennington, P. L., Cooper, E., Fulton, M. H., Scott, G. I., Wood, L. (2012). Pilot estuarine mesocosm study on the environmental fate of Silver nanomaterials leached from

consumer

products.

The

Science

of

the

Total

Environment,

421-422,

267–72.

doi:10.1016/j.scitotenv.2012.01.025 Colwill, R. M., & Creton, R. (2011). Locomotor behaviors in zebrafish (Danio rerio) larvae. Behavioural Processes, 86(2), 222–229. doi:10.1016/j.beproc.2010.12.003 Craig, P. M., Hogstrand, C., Wood, C. M., & McClelland, G. B. (2009). Gene expression endpoints following chronic waterborne copper exposure in a genomic model organism, the zebrafish, Danio rerio. Physiological Genomics, 40(1), 23–33. doi:10.1152/physiolgenomics.00089.2009 Dziendzikowska, K., Gromadzka-Ostrowska, J., Lankoff, a, Oczkowski, M., Krawczyńska, a, Chwastowska, J., Kruszewski, M. (2012). Time-dependent biodistribution and excretion of silver nanoparticles in male Wistar rats. Journal of Applied Toxicology : JAT, 32(11), 920–8. doi:10.1002/jat.2758 Embry, M. R., Belanger, S. E., Braunbeck, T. a, Galay-Burgos, M., Halder, M., Hinton, D. E., Whale, G. (2010). The fish embryo toxicity test as an animal alternative method in hazard and risk assessment and scientific research. Aquatic Toxicology (Amsterdam, Netherlands), 97(2), 79–87. doi:10.1016/j.aquatox.2009.12.008 Fabrega, J., Luoma, S. N., Tyler, C. R., Galloway, T. S., & Lead, J. R. (2011). Silver nanoparticles : Behaviour and effects in the aquatic environment. Environment International, 37(2), 517–531. doi:10.1016/j.envint.2010.10.012 Farkas, J., Christian, P., Gallego-Urrea, J. A., Roos, N., Hassellöv, M., Tollefsen, K. E., & Thomas, K. V. (2011). Uptake and effects of manufactured silver nanoparticles in rainbow trout (Oncorhynchus mykiss) gill cells. Aquatic Toxicology (Amsterdam, Netherlands), 101(1), 117–25. doi:10.1016/j.aquatox.2010.09.010

18

Gerhardt, A. (2007). Aquatic Behavioral Ecotoxicology - Prospects and Limitations. Human and Ecological

Risk

Assessment:

An

International

Journal,

13(3),

481–491.

doi:10.1080/10807030701340839 Gonzalez, P., Baudrimont, M., Boudou, A., & Bourdineaud, J. P. (2006). Comparative effects of direct cadmium contamination on gene expression in gills, liver, skeletal muscles and brain of the zebrafish (Danio rerio). BioMetals, 19(3), 225–235. doi:10.1007/s10534-005-5670-x Gottschalk, F., Sonderer, T., Scholz, R.W., Nowack, B., 2009. Modeled Environmental Concentrations of Engineered Nanomaterials (TiO2, ZnO, Ag, CNT, Fullerenes) for Different Regions. Environmental Science & Technology 43, 9216-9222. Griffitt, R. J., Luo, J., Gao, J., Bonzongo, J.-C., & Barber, D. S. (2008). Effects of particle composition and species on toxicity of metallic nanomaterials in aquatic organisms. Environmental Toxicology and Chemistry / SETAC, 27(9), 1972–8. doi:10.1897/08-002.1 Hoque, M. E., Khosravi, K., Newman, K., & Metcalfe, C. D. (2012). Detection and characterization of silver nanoparticles in aqueous matrices using asymmetric-flow field flow fractionation with inductively coupled plasma mass spectrometry. Journal of Chromatography A, 1233, 109–115. doi:10.1016/j.chroma.2012.02.011 Irons, T. D., MacPhail, R. C., Hunter, D. L., & Padilla, S. (2010). Acute neuroactive drug exposures alter locomotor activity in larval zebrafish. Neurotoxicology and Teratology, 32(1), 84–90. doi:10.1016/j.ntt.2009.04.066 Kwok, K. W. H., Auffan, M., Badireddy, A. R., Nelson, C. M., Wiesner, M. R., Chilkoti, A., Hinton, D. E. (2012). Uptake of silver nanoparticles and toxicity to early life stages of Japanese medaka (Oryzias latipes): effect of coating materials. Aquatic Toxicology (Amsterdam, Netherlands), 120121, 59–66. doi:10.1016/j.aquatox.2012.04.012 Lammer, E., Carr, G. J., Wendler, K., Rawlings, J. M., Belanger, S. E., & Braunbeck, T. (2009). Is the fish embryo toxicity test (FET) with the zebrafish (Danio rerio) a potential alternative for the fish acute toxicity test? Comparative Biochemistry and Physiology - C Toxicology and Pharmacology, 149(2), 196–209. doi:10.1016/j.cbpc.2008.11.006 Lapresta-Fernández, A., Fernández, A., & Blasco, J. (2012). Nanoecotoxicity effects of engineered silver and gold nanoparticles in aquatic organisms. TrAC Trends in Analytical Chemistry, 32(797), 40–59. doi:10.1016/j.trac.2011.09.007 19

Lee, K. J., Nallathamby, P. D., Browning, L. M., Osgood, C. J., & Xu, X.-H. N. (2007). In vivo imaging of transport and biocompatibility of single silver nanoparticles in early development of zebrafish embryos. ACS Nano, 1(2), 133–43. doi:10.1021/nn700048y Little, E. E., Archeski, R. D., Flerov, B. a., & Kozlovskaya, V. I. (1990). Behavioral indicators of sublethal toxicity in rainbow trout. Archives of Environmental Contamination and Toxicology, 19(3), 380–385. doi:10.1007/BF01054982 MacPhail, R. C., Brooks, J., Hunter, D. L., Padnos, B., Irons, T. D., & Padilla, S. (2009). Locomotion in larval zebrafish: Influence of time of day, lighting and ethanol. NeuroToxicology, 30(1), 52–58. doi:10.1016/j.neuro.2008.09.011 Massarsky, A., Dupuis, L., Taylor, J., Eisa-Beygi, S., Strek, L., Trudeau, V. L., & Moon, T. W. (2013). Assessment of nanosilver toxicity during zebrafish (Danio rerio) development. Chemosphere, 92(1), 59–66. doi:10.1016/j.chemosphere.2013.02.060 Müller, U., Cristina, N., Li, Z. W., Wolfer, D. P., Lipp, H. P., Rülicke, T., Weissmann, C. (1994). Behavioral and anatomical deficits in mice homozygous for a modified beta-amyloid precursor protein gene. Cell, 79(5), 755–765. doi:10.1016/0092-8674(94)90066-3 OECD guidelines. (2013). Test No. 236: Fish Embryo Acute Toxicity (FET) Test. OECD Guidelines for

the

Testing

of

Chemicals,

Section

2,

OECD

Publishing,

(July),

1–22.

doi:10.1787/9789264203709-en Ohshima, Hiroyuki (2012) Electrical Phenomena at Interfaces and Biointerfaces: Fundamentals and Applications in Nano-, Bio-, and Environmental Sciences. John Wiley & Sons, Inc. (Hoboken, NJ) Padilla, S., Hunter, D. L., Padnos, B., Frady, S., & MacPhail, R. C. (2011). Assessing locomotor activity in larval zebrafish: Influence of extrinsic and intrinsic variables. Neurotoxicology and Teratology, 33(6), 624–630. doi:10.1016/j.ntt.2011.08.005 Powers, C. M., Levin, E. D., Seidler, F. J., & Slotkin, T. (2011a). Silver exposure in developing zebrafish produces persistent synaptic and behavioral changes. Neurotoxicology and Teratology, 33(2), 329–332. doi:10.1016/j.ntt.2010.10.006 Powers, C. M., Slotkin, T. a, Seidler, F. J., Badireddy, A. R., & Padilla, S. (2011b). Silver nanoparticles alter zebrafish development and larval behavior: distinct roles for particle size, coating and composition. Neurotoxicology and Teratology, 33(6), 708–14. doi:10.1016/j.ntt.2011.02.002 20

Powers, C. M., Yen, J., Linney, E. a, Seidler, F. J., & Slotkin, T. (2010). Silver exposure in developing zebrafish (Danio rerio): persistent effects on larval behavior and survival. Neurotoxicology and Teratology, 32(3), 391–7. doi:10.1016/j.ntt.2010.01.009 Raymond, J., Chanin, S., Stewart, A. M., Kyzar, E., Gaikwad, S., Roth, A., Bruce, I., Fryar, C., Varga, D., Enriquez, J., Bagawandoss, S., Pham, M., Zapolsky, I., Green, J., Desmond, D., Hester, J., Kalueff, A. (2012). Assessing Habituation Phenotypes in Adult Zebrafish: Intra- and Inter-Trial Habituation in the Novel Tank Test. 66. Rankin, C. H., Abrams, T., Barry, R. J., Bhatnagar, S., Clayton, D., Colombo, J., Thompson, R. F. (2010). NIH Public Access, 92(2), 135–138. doi:10.1016/j.nlm.2008.09.012 Roberts, A. C., Reichl, J., Song, M. Y., Dearinger, A. D., Moridzadeh, N., Lu, E. D. Pearce, K., Esdin, J., Glanzman, D. L. (2011). Habituation of the C-Start response in larval zebrafish exhibits several distinct phases and sensitivity to NMDA receptor blockade. PLOS One, 6(12). doi: 10.1371/journal.pone.0029132 Saili, K. S., Corvi, M. M., Weber, D. N., Patel, A. U., Das, S. R., Przybyla, J., Tanguay, R. L. (2012). Neurodevelopmental low-dose bisphenol A exposure leads to early life-stage hyperactivity and learning deficits in adult zebrafish. Toxicology, 291(1-3), 83–92. doi:10.1016/j.tox.2011.11.001 Scott, G. R., & Sloman, K. a. (2004). The effects of environmental pollutants on complex fish behaviour: Integrating behavioural and physiological indicators of toxicity. Aquatic Toxicology, 68(4), 369–392. doi:10.1016/j.aquatox.2004.03.016 Selderslaghs, I. W. T., Hooyberghs, J., Blust, R., & Witters, H. E. (2013). Assessment of the developmental neurotoxicity of compounds by measuring locomotor activity in zebrafish embryos and larvae. Neurotoxicology and Teratology, 37, 44–56. doi:10.1016/j.ntt.2013.01.003 Stewart, A. M., Cachat, J., Green, J., Gaikwad, S., Kyzar, E., Roth, A., Kalueff, A. V. (2013). Constructing the habituome for phenotype-driven zebrafish research. Behavioural Brain Research, 236(1), 110–117. doi:10.1016/j.bbr.2012.08.026 Strähle, U., Scholz, S., Geisler, R., Greiner, P., Hollert, H., Rastegar, S., Braunbeck, T. (2012). Zebrafish embryos as an alternative to animal experiments-A commentary on the definition of the onset of protected life stages in animal welfare regulations. Reproductive Toxicology, 33(2), 128– 132. doi:10.1016/j.reprotox.2011.06.121

21

Wong, K., Elegante, M., Bartels, B., Elkhayat, S., Tien, D., Roy, S., Kalueff, A. V. (2010). Analysing habituation responses to novelty in zebrafish (Danio rerio). Behavioural Brain Research, 208(2), 450–457. doi:10.1016/j.bbr.2009.12.023 Wu, S. M., Tsai, P. R., & Yan, C. J. (2012). Maternal cadmium exposure induces mt2 and smtB mRNA expression in zebrafish (Danio rerio) females and their offspring. Comparative Biochemistry and Physiology - C Toxicology and Pharmacology, 156(1), 1–6. doi:10.1016/j.cbpc.2012.02.001 Xin, Q., Rotchell, J. M., Cheng, J.,Yi, J., Zhang, Q. (2015). Silver nanoparticles affect the neural development of zebrafish embryos. J. Appl. Toxicol., 9999: doi: 10.1002/jat.3164 Zala, S. M., & Penn, D. J. (2004). Abnormal behaviours induced by chemical pollution: A review of the

evidence

and

new

challenges.

Animal

Behaviour,

68(4),

649–664.

doi:10.1016/j.anbehav.2004.01.005

22

Figure captions Figure 1 Graphic representation of behavioural assay used in the present study (A). Grey areas reflect dark phases (0% light saturation), white areas – light phases (100 % light saturation). Profile of ZFEEs responsiveness (B) consists of average responses in each darkness cycle. Figure 2 TEM image of colloidal AgNPs in stock suspension. Figure 3 Hierarchical clustering for dark (left) and light (right) phase variables. Locomotion profiles are hierarchically clustered to link the Ag+ exposure concentration and the locomotion variable. Each square represents the average relative locomotion value for groups of exposed ZFEEs (N>20). Red colour indicates relatively high locomotion activity, and blue indicates low activity. Figure 4 PCA for dark phase locomotion variables at different Ag+ exposure concentrations. High exposure groups are shown in the blue circle and lower doses/control are indicated in the red circle on the scores plot (left). Loadings plot (right) indicated high correlation between most dark variables. Figure 5 Concentration-dependent locomotion in darkness cycles (A) and light (B) cycles for Ag+. The grey line indicates a control response; data presented as the means ±SE of total distance swam by continuous movements. Statistically significant differences (from control treatment), p ≤0.05 are indicated with asterisks (*). Figure 6 Hierarchical clustering for dark (left) and light (right) phase variables. Locomotion profiles are hierarchically clustered to link AgNPs exposure concentration and locomotion variable. Each square represents average relative locomotion value for groups of exposed ZFEEs (N>20). Red colour indicates relatively high locomotion activity, blue – low activity. Figure 7 PCA for dark phase locomotion variables at different AgNPs exposure concentrations. High exposure groups are shown in the blue circle and lower doses/control are indicated in the red circle on the scores plot (left). Loadings plot (right) indicated high correlation between most dark variables. Figure 8 Concentration-dependent locomotion responses in darkness cycles (A) and light (B) cycles for AgNPs. The grey line indicates the control response; data presented as the means ± SE of total distance that ZFEEs swam by large movements. Statistically significant differences (from control treatment), p ≤0.05 are indicated with asterisks (*).

23

Figure 9 Proportion of time (duration) that ZFEEs allocated to large, small movements or inactivity in darkness periods for Ag+ (A) and AgNPs (B). Data presented as the means ± SE; red line represents locomotion structure of unexposed individuals. Statistically significant differences (from control treatment), p ≤0.05 are indicated with asterisks (*). Figure 10 Locomotion profiles, representing a gradual decrease in responsiveness (total distance swam by large movements) over repeated dark cycles for Ag+ (A, C) and AgNPs (B) exposed ZFEEs. Data presented as the means ± SE; statistically significant differences (from control treatment), p ≤0.05 are indicated with asterisks (*).

Fig. 1

24

Fig. 2

25

Fig. 3

26

Fig. 4

27

Fig. 5

Fig. 6 28

Fig. 7

Fig. 8

29

Fig. 9

Fig. 10

30

Table 1 Characterization of AgNPs in the ZF embryo medium 1 mg/L at different experimental time-points (mean ±SD).

Timepoint (h) Mili-Q (t=0) t=0 t=24

Size peak 1 Polydispersity Size peak 1 (% index (PDI) (nm) Intensity)

Size peak 2 Size peak 2 (% (nm) Intensity)

Zeta potential, ζ (-mV)

0.35 ± 0.18

64.2 ± 28.4

8.2 ± 2.0

12.9 ± 1.9

NR

0.12 ± 0.10 0.32 ± 0.13

40.7 ± 2.2 90.8 ± 0.9 33.9 ± 4.4 88.3 ± 2.4 108.2 ± 72.9 ± 25.5 121.2 1196.0 ± 89.3 ± 19.8 1736.0 5.5 ± 7.9 44.4 ± 52.7 199.7 ± 52.0 ± 9.1 297.4

6.01 ± 0.5 5.78 ± 1.3

9.2 ± 0.9 11.6 ± 2.2

10.66 ± 7.55 0.01 ± 0.25

12.4 ± 13.9

22.9 ± 23.2

14.93 ± 5.33

63.0 ± 116.7

10.7 ± 19.8

3.54 ± 1.44

0 304.2 384.8

0

11.53 ± 3.39

37.7 ± 4.5

22.94 ± 1.37

46.0 ± 6.2

6.7 ± 0.6

19.1 ±1.6

NR

t=48

0.92 ± 0.17

t=72

0.89 ± 0.14

t=96

0.92 ± 0.12

t=120

0.45 ± 0.07

Mili-Q 0.29 ± 0.21 (t=120) NR-not reported

86.7 ± 2.6

80.9 ± 1.6

±

31