Involvement of ahr-dependent Cyp1a detoxification activity, oxidative stress and inflammatory regulation in response to graphene oxide exposure in rainbow trout (Oncorhynchus mykiss)

Involvement of ahr-dependent Cyp1a detoxification activity, oxidative stress and inflammatory regulation in response to graphene oxide exposure in rainbow trout (Oncorhynchus mykiss) Gregorio Mol´ es 1 , Ana Valdehita , Mona Connolly , Jos´ e María Navas * Department of Environment and Agronomy, Instituto Nacional de Investigaci´ on y Tecnología Agraria y Alimentaria (INIA), Consejo Superior de Investigaciones Científicas (CSIC), Ctra. de La Coru˜ na, km 7.5, 28040, Madrid, Spain HIGHLIGHTS GRAPHICAL ABSTRACT •96 h exposure in RTL-W1 cells showed cytotoxicity at concentrations ≥4.68 mg/L. •96 h in vivo exposure to GO did not provoke mortality in rainbow trout juveniles. •Histological changes in gills and liver were noted in fish exposed to ≥9.89 mg/L GO. •GO stimulated EROD and BFCOD activities (Cyp1a/Cyp3a) in a concentrationdependent way. •GO induced downregulation of ahr2 and cyp1a, and upregulation of il8 and il1b. ARTICLE INFO Keywords: GO Fish RTL-W1 cytotoxicity OECD TG 203 EROD/BFCOD activity ABSTRACT Graphene oxide (GO) is a very attractive material for use in a vast number of applications. However, before its widespread use, it is important to consider potential issues related to environmental safety to support its safe application. The aim of this study was to investigate effects on fish (rainbow trout) following GO exposure. Using both an in vitro approach with the RTL W1 rainbow trout liver cell line, and in vivo exposures, following OECD TG 203, disturbances at the cellular level as well as in the gills and liver tissue of juvenile trout were assessed. In RTL W1 cells, a time and concentration-dependent loss in cell viability, specifically plasma membrane integrity and lysosomal function, was observed after 96 h of exposure to GO at concentrations ≥18.75 mg/L. Additionally, * Corresponding author. Instituto Nacional de Investigaci´ on y Tecnología Agraria y Alimentaria (INIA). Consejo Superior de Investigaciones Científicas (CSIC), Ctra. de La Coru˜ na, km 7.5, 28040, Madrid, Spain. E-mail addresses: [email protected] (G. Mol´ es), [email protected] (A. Valdehita), [email protected] (M. Connolly), [email protected] (J.M. Navas). 1 Present address: Centro Interdisciplinar de Investigaç˜ ao Marinha e Ambiental (CIIMAR), Terminal de Cruzeiros do Porto de Leix˜ oes, 4450-208 Matosinhos, Portugal. Contents lists available at ScienceDirect Chemosphere journal homepage: www.elsevier.com/locate/chemosphere https://doi.org/10.1016/j.chemosphere.2024.143005 Received 14 May 2024; Received in revised form 20 July 2024; Accepted 1 August 2024 Chemosphere 364 (2024) 143005 Available online 8 August 2024 0045-6535/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC license ( http://creativecommons.org/licenses/bync/4.0/ ). increased reactive oxygen species (ROS) levels were evidenced at concentrations ≥18.75 mg/L, and an enhancement of metabolic activity was noted with concentrations ≥4.68 mg/L. In vivo exposures to GO did not provoke mortality in rainbow trout juveniles following 96 h exposure but led to histological alterations in gills and liver tissues, induction of enzymatic detoxification activities in the liver, as well as aryl hydrocarbon receptor (ahr)–cytochrome P450 1a (cyp1a) gene expression downregulation, and upregulation of pro-inflammatory cytokines il1b and il8 at GO concentrations ≥9.89 mg/L. 1. Introduction Graphene oxide (GO), is a two-dimensional (2D) graphene-related material (GRM) that has been defined as a chemically modified graphene (Bianco et al., 2013; ISO, 2017). It is produced by the oxidation and exfoliation of graphite and has distinct properties such as a hydrophilic nature, semi-conductivity and high adsorption capacity. According to the various synthesis methods, GO can have varying oxygen contents and oxygen-containing functional groups (=O, –OH, -O-, –COOH) decorating the basal planes and the edges (Hummers and Offeman, 1958). The presence of these oxygen functional groups confers active properties that are very attractive for use in a number of applications, for example in biomedical sensors and drug delivery (Chung et al., 2013), re-enforcement of composites, as well as in water purification and desalination systems (Dasmahapatra et al., 2018; Homaeigohar and Elbahri, 2017; Malhotra et al., 2020). However, before GO reaches general and widespread use, it is important to consider potential issues related to environmental safety of this substance to support its safe and sustainable application. Knowledge of adverse effects of GO on aquatic organisms, including fish, is still limited (Dasmahapatra et al., 2018). According to a recent and detailed review of the literature, only 8 studies were identified which have performed toxicity testing of GO in fish (Connolly et al., 2023). The data available often suffers from deficiencies/inconsistencies due to material instability under test conditions, and studies presented effect/no effect values based on nominal rather than measured exposure concentrations. Thus, while no acute toxicity was reported in the studies, results must be taken with caution, as nominal exposure concentrations (or even between 80 and 120% of the nominal concentration as described as stable in the guidance document for testing difficult to test substances (OECD, 2019a) of the Organization for Economic Cooperation and Development, OECD) were likely not reached or maintained. Tests applying standardised guidelines, as these of the OECD (for instance, OECD Test Guideline (TG) 203, Fish acute toxicity test (OECD, 2019b)), including full assessments of materials stability with measurements of concentration maintenance during testing, are needed for hazard assessment and will provide more reliable data. From the existing data in zebrafish, while no mortality was reported, conclusions cannot be made on the acute toxicity due to the instability of exposure concentrations while testing. However, in some cases, histological disorders in gills and liver have been described, suggesting sublethal effects associated with exposure (Chen et al., 2016; Martínez-´ Alvarez et al., 2021; Souza et al., 2017). Moreover, the modulation of enzymes related to oxidative stress and inflammation were observed suggesting that the reactive oxygen species (ROS) produced as a result of GO exposure induced pro-inflammatory cytokines (interleukin-1b (il1b) and interleukin-6 (il6)) gene expression (Chen et al., 2016). There also exists tests using in vitro systems (e.g., fish cell lines) for fish acute toxicity according to losses in cell viability following exposure. Such approaches can generate information that can be used in a tiered approach to testing and assessment, whereby cells are used in low tier tests for screening and any evidenced effects are corroborated using higher tier in vivo testing. Also, when information on in vivo effects is missing the in vitro data can serve in a weight of evidence approach to risk assessment (OECD, 2019c). As well as generating toxicity data, in vitro approaches will contribute to deciphering information on possible mechanisms of toxicity and key events at a cellular level in future adverse outcome pathway (AOP)-directed testing strategies to predict effects at whole organism and indeed at population levels (Halappanavar et al., 2020). To date a number of these in vitro studies using fish cell lines have been performed to assess the toxicity of GRMs including GO (Kalman et al., 2019; Lammel and Navas, 2014; Lammel et al., 2015; Siqueira et al., 2022; Srikanth et al., 2018). In the topminnow fish hepatoma cell line (PLHC-1) Lammel and Navas (2014) observed that GO penetrated the plasma membrane and was accumulated in the cytosol, where it disrupted the mitochondrial membranes and increased ROS levels, contributing to the induction of oxidative stress. In the same cells, as well as in the carp leukocyte cell line (CLC), Kalman et al. (2019) observed GO present within vesicles as well as free in the cytosol and signs of cytotoxicity. In bluegill sun fish fibroblast cells (BF-2), GO induced cytotoxicity and oxidative stress at concentrations ≥10 μ g/mL and ≥40 μ g/mL respectively (Srikanth et al., 2018) and in zebrafish epithelial liver cells (ZFL) GO was internalized interacting with the cell structure/function and increasing ROS production (Siqueira et al., 2022). Therefore, there is evidence of cytotoxicity and disturbances for GO using fish cell lines, however information on whether this translates to effects in vivo is missing. There is some evidence to suggest that such in vitro approaches can be used as screening tools for acute toxicity prediction (Hern´ andez-Moreno et al., 2022) but also other studies that show their poor predictability (Rodrigues et al., 2019). In addition, there is an important lack of information about the mechanisms underlying the toxic action of GO and the possible induction of detoxification activities that could, on the one hand, protect the organism against the toxic insult and, on the other hand, contribute to the catabolism of GO. For these types of investigations in vitro cell lines can serve as invaluable tools. Considering that the molecular structural features of graphene are similar to those of polycyclic aromatic hydrocarbons (PAHs) a recent work explored the possibility that detoxification activities induced by graphene in a fish cell line where similar to those induced by PAHs, and subsequently if induced the enzyme activities would contribute to the metabolism of graphene (Valdehita et al., 2023). Planar aromatic molecules, as those of PAHs, can interact with the aryl hydrocarbon receptor (Ahr) when entering cells inducing the production of Cyp1a that is directly involved in detoxification of these substances through its monooxygenase activities (Denison et al., 2002; Whyte et al., 2000). It has been observed that GO enhances the expression of cyp1a and ahr genes and Cyp1a associated detoxification activities when cells are challenged with a prototypical PAH (Valdehita et al., 2023). The authors proposed that after initial degradation of GO by other mechanisms, intermediate metabolites would appear to be able to interact with the Ahr inducing the corresponding detoxification responses. Taking into account all the above, the objective of the present study was to investigate potential adverse effects on fish following GO exposure and to shed light on processing pathways at the molecular level, applying both in vitro and in vivo approaches. For in vitro testing, the liver RTL-W1 cell line was selected to analyse cell viability, as it has been successfully used in multiple nanomaterial toxicity studies (Bermejo-Nogales et al., 2017; Connolly et al., 2015; Fern´ andez et al., 2013; Galbis-Martínez et al., 2018; Hern´ andez-Moreno et al., 2022; Malh˜ ao et al., 2013; Simon et al., 2014; Valdehita et al., 2023). The liver is the main site of biotransformation and thus this particular cell type was chosen to investigate any potential effects associated with GO G. Mol´ es et al. Chemosphere 364 (2024) 143005 2 metabolism in this organ and the particular involvement of cyp activity. Increased levels of ROS leading to oxidative stress responses has been one of the most cited underlying mechanisms for nanomaterial toxicity (Shvedova et al., 2012; Xia et al., 2006). Therefore, to test the involvement of increased ROS levels in this study, a DCFH-DA probe was used. Cells were exposed to serial dilutions of GO (0.29–75 mg/L) for 24 h and for an extended time of 96 h to match the in vivo exposure duration. Interestingly, the same biotransformation activities can be monitored at the enzymatic activity and molecular level after in vivo exposures. Thus, the OECD TG 203 (Fish Acute Toxicity Test) was applied and following exposure to GO (4.26, 20.6 and 100 mg/L), not only apical endpoints as mortality or behaviour alterations were assessed, but also the induction of detoxification activities dependent on Cyp1a as well as Cyp3a, such as Ethoxyresorufin-O-deethylase (EROD) and 7-Benzyloxy-4-trifluoromethylcoumarin-O-debenzyloxylase (BFCO D) in both liver and gills. Additionally, expression levels of cyp1a, cyp3a and ahr2 genes as well as that of important pro-inflammatory cytokines (il8 and il1b) were also assessed. Specifically, the Ahr is involved in the induction/regulation of Cyp1a. By including the analysis of ahr gene expression, we could monitor if upstream upregulation is occurring. Cyp3a activation is not dependent on the Ahr and thus this cytochrome induction was also included as a reference and to monitor involvement of other pathways in GO biotransformation. By monitoring expression levels of two of the main pro-inflammatory cytokines (il8 and il1b) we could investigate the effect/involvement of potential inflammatory responses on the biotransformation process. The information generated served not only to fill an information gap on the acute toxicity of the GO in fish, but also provided new data, that corroborate findings from in vitro studies, on the likely involvement of Cyp1a in the metabolism of GO. 2. Materials and methods 2.1. Stock dispersions and characterization of GO Graphene oxide was synthesized by a modified Hummers’ method (Hummers and Offeman, 1958) and provided as a powder (GRAnPH®, Grupo Antolin Ingenieria, S.A, Spain). According to the manufacturer and as described in previous work (Kalman et al., 2019), GRAnPH® consists of single or few layers of graphene sheets and exhibits a lateral size <1 μ m. The GO concentration was 100%, so no impurities were expected. The same batch was used throughout the study to rule out variations due to the use of different ones. The GO stock suspensions were prepared according to Lammel et al. (2013). Briefly, GO was dispersed in sterile Milli-Q water (1–10 g/L) and sonicated at 37 kHz for 30 min (total energy delivered: 36,826 J) in an ultrasonic ice-water bath (Fisherbrand S-series; Thermo Fisher Scientific, Waltham, MA, US) using sterile 30 mL Pyrex® glass tubes (POBEL, Madrid, ES). Following sonication, the suspensions were centrifuged at 1300 g for 30 min at 16 ◦C to remove large aggregates/agglomerates. The concentration in supernatants was estimated by means of a concentration-absorbance standard curve generated from aliquots of the original, i.e., not centrifuged stock suspensions. Absorbance measurements of GO serial dilutions were performed in a quartz spectrophotometer cuvette using a multimode microplate reader Spark 20 M (TECAN, M¨ annedorf, CH) at its peak wavelength (230 nm). All GO stock dispersions were freshly prepared for each toxicity assay. Afterwards, working dispersions where prepared by dilution of the GO stock suspension supernatant with culture medium (in vitro test) or aquarium water (in vivo test). Hydrodynamic size and size frequency distribution of GO dispersions during in vitro and in vivo tests were monitored by dynamic light scattering (DLS) analysis using a Zetasizer Nano-ZS apparatus (Malvern Instruments Ltd., Malvern, UK). Milli-Q water, Leibovitz’s L-15 medium and aquarium water were used as background controls. At least six measurements (10 runs, 10s/run) were taken of each sample. To monitor the stability over time several measurements were taken throughout the exposure period. For this the attenuator and the optimal measurement position were automatically determined in the first measurement of each concentration in each sample and thereafter they were fixed in order to monitor any changes (Pulido-Reyes et al., 2024). Data was analysed using Zetasizer Software version 6.34 (Malvern Instruments Ltd.). Z-potential measurements were also performed using disposable capillary cuvettes (Malvern Instruments Ltd.). Three measurements were taken of each sample. The number of runs was set automatically. GO morphology was also assessed by means of transmission electron microscopy (TEM) at 18.75 and 4.6 μ g/mL, in milliQ water and in L-15 medium. Samples were prepared by placing a drop of GO suspensions on carbon-coated copper TEM grids and allowed to evaporate at room temperature before analysis. TEM analysis was carried out using a JEOL 1400 Plus TEM (JEOL Ltd., Tokyo, JP). 2.2. In vitro cytotoxicity assessment 2.2.1. RTL-W1 cells culture RTL-W1 cells are derived from rainbow trout liver and have the appearance of bile preductural epithelial cells (Malh˜ ao et al., 2013). These cells have been used in long term experiments up to 7 days with no obvious loss in functioning under control conditions. This points to their utility as a valuable tool for long term in vitro ecotoxicity studies (Galbis-Martínez et al., 2018; Valdehita et al., 2023). The cells were cultured in 75 cm 2 Cell Star cell culture flasks (Greiner Bio-One GmbH, Frickenhausen, DE) at 20 ◦C in Leibovitz’s L-15 culture medium (Gibco; Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (FBS) (Sigma Aldrich; Merck group, Darmstadt, DE), and 1% penicillin/streptomycin (P/S) solution (Lonza; Thermo Fisher Scientific). 2.2.2. Cytotoxicity assays Cell viability was measured according to a modified version (Lammel et al., 2013; Lammel and Navas, 2014) of a protocol described by Dayeh et al. (2005). This protocol uses three assays based on different toxicological endpoints (AlamarBlue™ for metabolic activity; 5-Carboxyfluorescein diacetate acetoxymethyl ester (CFDA-AM) for cell membrane integrity, and neutral red uptake (NRU) for lysosomal function). RTL-W1 cells were seeded (2.5 ×10 4 cells/well) into transparent flat-bottom 96-well plates (Greiner Bio-one GmbH). Working concentrations of GO were prepared by 1:2 serial dilutions (0.29–75 mg/L) in complete L-15 medium (Gibco; Thermo Fisher Scientific) and vortexed just before applying them to the cells. After 24 and 96 h of GO exposure (final volume of 200 μ L/well), the medium was removed and cells were washed twice with 200 μ L PBS. Wells received 100 μ L of 1.25% (v/v) AlamarBlue™ (Invitrogen; Thermo Fisher Scientific) and 4 μ M CFDAAM (Invitrogen; Thermo Fisher Scientific) prepared in serum-free/ phenol red-free L-15 medium (Gibco; Thermo Fisher Scientific). Resazurin (7-Hydroxy-3H-phenoxazin-3-one 10-oxide), the active compound in the commercial solution AlamarBlue™ yields a fluorescent signal in response to metabolic activity. It is reduced to the fluorescent compound resorufin and a decrease in fluorescence is indicative for diminished metabolic activity and cytotoxicity. The CFDA-AM assay is based on the conversion to the fluorescent product 5-carboxyfluorescein (5-CF) by cytosolic esterases, which are only retained in cells with intact plasma membranes. Fluorescence was measured on a microplate reader Spark 20 M (Tecan) at a wavelength of 535/590 nm (excitation/emission) for AlamarBlue™, or at 485/535 nm for CFDA-AM after 30 min of incubation at 20 ◦C in the dark. Cells were washed with PBS and incubated with 100 μ L of neutral red (Sigma Aldrich; Merck group) solution (33 μ g/mL in serum-free/phenol red-free L-15) for 1 h at 20 ◦C in the dark. Neutral red is a fluorescence probe that accumulates in the lysosomes and in cells with damaged lysosomal function less NR is taken up and/or retained. After incubation, cells were rinsed with PBS and the retained dye was extracted with 100 μ L of an acidified solution (1% glacial acetic acid, 50% ethanol and 49% Milli-Q water). Thereafter, fluorescence was G. Mol´ es et al. Chemosphere 364 (2024) 143005 3 measured at 532/680 nm (excitation/emission). The fluorescence values were corrected for the cell-free control results and normalized against the medium control values. 2.2.3. Intracellular reactive oxygen species (ROS) Intracellular ROS levels were measured by the dichlorofluorescein (DCF) assay using the 6-carboxy-2 ′ 7 ′ dichlorodihydrofluorescein diacetate (DCFH-DA) probe (Merck group) reported by Wang and Joseph (1999) and adapted for the RTL-W1 cell line. Cell seeding and treatment with GO was performed in 96-well plates as described above. Cells treated with chloramine-T trihydrate (0.04 mM−10 mM) were used as positive controls. After 24 h or 96 h of exposure with increasing concentrations of GO, cells were washed with PBS and incubated with 100 μ L of 100 μ M DCFH-DA probe (in serumfree/phenol red-free L-15 medium) at 20 ◦C for 30 min in the dark. After removal of the DCFH-DA probe, the cells were washed with PBS and incubated with 100 μ L of serum-free/phenol red-free L-15 medium. Fluorescence was measured on a microplate reader Spark 20 M (Tecan) at a wavelength of 485/530 nm (excitation/emission), immediately (t0) and every 15 min over 60 min. ROS levels were quantified as the percentage increase in fluorescence per well over a 30 min period using the formula [(Ft30 −Ft0)/Ft0 × 100], where Ft30 and Ft0 are the fluorescence measured at time 30 and 0 min, respectively. The results are presented as percentage of the fluorescence of untreated control cells. 2.2.4. Interference and fluorescence quenching Before starting any experiment, potential interferences due to autofluorescence or to fluorescence quenching phenomena of GO suspensions, were assessed. Autofluorescence in cell culture medium or adhered to the cells after the treatments was measured simulating the same conditions of the cytotoxicity assays, but without adding the corresponding fluorophores. Non-treated cells served as a reference. Fluorescence quenching was assessed in the presence of cells to simulate a more realistic assay scenario. For that, cells were seeded and exposed to GO (0.29–75 μ g/mL) in the same way as for each assay. After exposure and washing twice with PBS, the cells were incubated in the dark with the conversion products that are formed in the course of each assay at the maximal concentration and to 10% of the maximal concentration that can be expected to be formed in the respective assays, namely, AlamarBlue™ (1 and 0.1 μ M of resorufin), CFDA-AM (4 and 0.4 μ M of 5carboxyfluorescein, 5-CF), NRU (33 and 3.3 μ g/ml of protonated neutral red), and ROS (100 and 10 μ M of DCF). Finally, fluorescence readings of exposed cells were taken at the corresponding wavelengths used for each assay. 2.2.5. Internalization of GO Transmission electron microscopy (TEM) was used to investigate the potential internalization of GO in RTL-W1 cells. For that, cells were seeded on poly-L-lysine coated cover slips in a 24-well plate (1.0 ×10 5 cells/well) and were exposed to 4.6 and 18.75 μ g/mL of GO for 24 and 96 h. After exposure, samples were prepared as described by Lammel et al. (2013) and included washing steps (Millonig phosphate buffer, pH 7.3), primary fixation (4% paraformaldehyde/2.5% glutaraldehyde), post fixation (1% osmium tetroxide), gradual dehydration steps (30–100% acetone), embedding (gradual infiltration with Spurr’s resin) and a polymerisation step (65 ◦C, 48 h). Ultrathin sections were stained in uranyl acetate and lead citrate and viewed in a JOEL 1010 JEM TEM (JEOL Ltd, Tokyo, JP). 2.3. Fish maintenance and in vivo exposure to GO Juveniles of rainbow trout (Oncorhynchus mykiss) with an average weight of 2.7 ±0.08 g and size 6.3 ±0.07 cm were obtained from a local trout farm (Felechosa, Asturias, Spain) and acclimated at the INIA (CSIC) fish facilities (Carretera de la Coru˜ na, km 7,5, Madrid, Spain). Prior to distribution into the experimental aquariums, the fish were kept during twelve days in a 450 L tank with a water recirculation system at 15.4 ◦C and photoperiod of 12 h light/day. Once a day, fish were fed with commercial diet (BioMar Inicio Plus 801, BIOMAR, Aarhus, DK) for trout at a rate of 2% of their body weight. After this period, the fish were transferred to rectangular 33 L glass tanks (10 fish/tank) with filtered and reconstituted water to acclimatize for 7 days under semi-static conditions prior to the start of experiments. During this acclimatization period, fish were maintained under controlled conditions and fed with a commercial diet once a day (2% of their weight). Feeding was stopped 24 h prior to starting the experiment. The survival rate at the end of the acclimatization period was 98.7%. The experiment started with 7 fish/tank, and the remaining fish up to the 10 per tank used in the acclimation period were transferred to the 450 L maintenance tank. The acute toxicity test was developed according to OECD TG 203 (OECD, 2019b). During the whole experiment, the test conditions and the water parameters were within the recommended OECD range, namely, the photoperiod was 12/12 h, temperature 13.8 ±0.05 ◦C, conductivity 272.2 ±1.5 μ S/cm, water pH 7.9 ±0.08 and the dissolved oxygen remained above 95%. A limit test under static conditions was carried out exposing fish to 100 mg/L of GO and the corresponding control for 96 h. Additionally, two concentrations of 4.26 and 20.66 mg/L of GO were also tested. These treatments were applied in duplicate, so that after exposure, one of the duplicates served to observe the evolution of fish in a depuration phase of 96 h. Due to the partial precipitation problems of GO in aquarium water seen in preliminary tests, turbines (Voyager nano, SICCE, Pozzoleone, IT) were installed in the aquariums to create a stream of water and to improve maintenance of the test material in the water column. All the GO concentrations used were monitored during the experiment by UV/ Vis spectrophotometry using a Spark 20 M (Tecan) spectrophotometer at 230 nm and a standard curve of GO (2.2–142 mg/L) in aquarium water. Absorbance was measured at the beginning of the experiment and every 24 h until the end of the exposure. Samples for analysis were collected in the same way at each time point. A pool of 3 samples (3 ml each) per aquarium was analysed, collected along the centerline at mid-depth. The stability of all GO dispersions also was checked by DLS. Mortalities, appearance, and swimming behaviour of fish were checked at 0, 2, 5, 24 h and then, twice per day until 96 h. At the end of the exposure period, one batch (one aquarium with 7 fish for each GO concentration) was sacrificed while the other fish were put in aquariums with clean water for a depuration period of another 96 h. While this depuration phase is not part of the OECD TG203 we considered it useful to monitor if the fish can recover from any adverse effects observed during the acute exposure. During this depuration period the fish were fed once a day at 2% body weight. For the sacrifices, the fish were anesthetized with 250 mg/ L of ethyl 3-aminobenzoate methanesulfonate (MS-222) (Sigma-Aldrich; Merck group), weighed, sized, and finally euthanized by decapitation. Livers and gills were collected for histological examination, enzyme activities, and gene expression analyses. Samples for biochemistry and molecular biology were stored at −80 ◦C until analyses. All the procedures were performed in accordance with the Spanish Royal Decree 53/2013 (RD, 2013) and European legislation (Directive, 2010/63/EU) (EC, 2010) for the use of laboratory animals for scientific purposes. The experiments were performed after receiving a favourable report of the committees and responsible competent authorities for animal experimentation (PROEX 089.2/21). 2.4. Histological analysis Directly after extraction liver and gill fish tissues were immediately fixed overnight by immersion in 4% formaldehyde - 1% glutaraldehyde (McDowell and Trump, 1976), dehydrated, embedded in 2-hydroxyethyl methacrylate polymer resin (Technovit 7100, Heraeus Kultzer, Wehrheim, DE), sectioned (3 μ m) using a microtome and stained with toluidine blue. Such a stain provides good contrast between the blue G. Mol´ es et al. Chemosphere 364 (2024) 143005 4 stained cellular nuclei, purple polysaccharides and red-purple glycosaminoglycans in the tissues. The sections were viewed under a light microscope. 2.5. Enzyme activities in gill and liver EROD and BFCOD activities were monitored in gills and livers of rainbow trout after 96 h exposure to GO and after 96 h of depuration (see above). Tissue fragments (10 mg approx.) were homogenized in 250 μ L of ice-cold homogenization buffer (0.1 M Tris HCL pH 7.5, 1 mM EDTA, 0.25 M sucrose, 150 mM KCL, 20% v/v glycerol, 1 mM DTT and 5 μ g/mL of pepstatin A, aprotinin and leupeptin) using a TissueLyser II (QIAGEN, Venlo, NL) for 30s at 30Hz. Homogenates were then centrifuged at 6000 g for 10 min at 4 ◦C and the supernatants centrifuged at 16000 g for 60 min at 4 ◦C. The resulting pellets were dissolved in 100 μ L homogenization buffer and used for EROD and BFCOD analysis. EROD activity was measured at room temperature following the methodology established by Burke and Mayer (1974) as detailed in Valdehita et al. (2012). 96-well plates (Greiner Bio-one GmbH) were used for carrying out the measurements. Each sample (10 μ L undiluted) was analysed in duplicate. The reaction was followed by reading the fluorescence every 10 min over 40 min on a microplate reader Spark 20 M (Tecan) at 532/590 nm (excitation/emission). EROD activity was expressed as pmol of resorufin produced in 1 min per mg of protein. A resorufin standard curve was used to quantify the pmol formed during the assay. Sample protein concentrations were quantified using a fluorescamine-based assay (Udenfriend et al., 1972) and a bovine serum albumin (BSA) standard curve. BFCOD activity was measured at 30 ◦C as described by Thibaut et al. (2006) in 96-well plates (Greiner Bio-One GmbH). The reaction was monitored in duplicate (10 μ L undiluted samples) reading the fluorescence every 10 min over 40 min on a microplate reader Spark 20 M (Tecan) at 409/530 nm (excitation/emission). BFCOD activity was expressed as pmol of 7-hydroxy-4-trifluoromethyl coumarin (HFC) produced in 1 min per mg of protein. HFC pmol production was calculated using a standard curve. Protein concentration was quantified using a fluorescamine-based assay. 2.6. Gene expression analyses Total RNA of gills and livers was isolated using TRIzol™ reagent (Invitrogen; ThermoFisher Scientific) following manufacturer’s instructions. Tissue fragments were homogenized in 1 mL of TRIzol™ reagent using a TissueLyser II (QIAGEN) for 50s at 30Hz. Purity and concentration of RNA was determined by NanoDrop One spectrophotometer (ThermoFisher Scientific). DNAse treatment of RNAs (10 μ g) was done using TURBO DNA-free Kit (Invitrogen; ThermoFisher Scientific). For cDNA synthesis, 1 μ g of treated RNA was reverse-transcribed using iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Inc, Hercules, CA, US) in a reaction volume of 20 μ L. As an exogenous internal reference, 0.5 ng of mRNA from the luciferase (luc1) gene (L4561, Promega, Madison, WI, US) was spiked into each reverse transcription reaction. Quantitative real-time PCR (qPCR) assays were run in duplicate for each sample to analyse the expression of cyp1a, ahr2, cyp3a, il1b, il8 and luc1 on 96-well PCR plates using a 7500 Fast Real-Time PCR System (Applied Biosystems, Waltham, MA, US) with default settings for the fluorescence detection system. Each reaction contained 10 μ L (2x) SYBR Green Quantimix easy master mix (Biotools B&M labs, Madrid, ES), 0.3 μ L (10 μ M) of each primer (Supplementary Table S1), nuclease-free water and 1 μ L of non-diluted cDNA sample in a final reaction volume of 20 μ L. After an amplification stage, a melt curve was included to verify the product specificity and to ensure the absence of primer dimers. Standard curve dilution series were generated from a pool of cDNA samples to calculate the RT-qPCR assay efficiencies (96–105%) and correlation coefficients (R 2 ≥0.99). Data were analysed with 7500 Software v2.06 (Life Technologies Corp.) and the relative expression levels of each gene were calculated using the 2 -(ΔΔCt) method. For data normalization the expression of luciferase (Supplementary Table S1) as a reference gene was used and the control group during the exposure period was used as the reference group. 2.7. Statistical analysis Data are expressed as mean ±standard error of the mean (SEM) of at least three independent experiments in the case of the in vitro assays. All statistical analyses were performed using Prism 9.4.1 (GraphPad Software, Inc., CA, US). The normality distribution and the homogeneity of variance was checked by Kolmogorov-Smirnov test and Brown-Forsythe test, respectively. Significant differences between control and treated groups were tested by one-way analysis of variance (ANOVA) (p <0.05) followed by a post hoc Dunnett’s multiple comparison test. 3. Results 3.1. Characterization of GO dispersions Hydrodynamic size frequency distribution of the GO stock and dispersions during in vitro and in vivo toxicity testing were determined by DLS (Table 1). GO stock suspensions in Milli-Q water were stable and characterised by two peaks; a main peak (~98% of the total distribution according to intensity) with a hydrodynamic diameter (HDD) of ~263 nm and a minor peak (~2% of the total distribution) with ~4360 nm (Table 1 and Valdehita et al., 2023). When the GO stocks were diluted in L-15 culture medium used during the in vitro test, HDD size distributions were maintained, with a slight increase in the main size distribution peak to 519 ±29 nm evidenced for the highest exposure concentration (75 mg/L) (Table 1). However, during the in vivo test the HDD increased drastically in all concentrations following preparation (t 0h) in aquarium water, reaching values of 2383 ±179 nm in the main peak (95.2%) and 5239 ±87 nm in the minor peak (4.8%). No major changes in size distribution were observed after 96 h (Table 1). According to Z-Potential measurements, all values were outside the stable range (≤± 30 mV) (Clogston and Patri, 2011). The highest values were obtained with 100 mg/L of GO (−16.3 ±0.47 mV) and this value was maintained throughout 96 h. Concentrations of 4.26 and 20.66 mg/L of GO had Z-Potential values which progressively decreased after 96 h, and most notably for the lowest concentration (−14.2 to −7.3 mV), indicating a drop in stability over time (Table 1). Morphology and structure of GO dispersions were also characterized by TEM and the images revealed no remarkable differences in size or shape between Milli-Q water or L-15 culture medium dispersions at both GO concentrations tested (18.75 and 4.6 μ g/mL) showing similar results to those previously reported (Kalman et al., 2019; Valdehita et al., 2023). 3.2. In vitro cytotoxicity in RTL-W1 cells Potential interferences of GO dispersions with cytotoxicity assay reagents (AlamarBlue™, CFDA-AM, NRU or ROS assays) were tested in the presence of cells before starting the tests. No autofluorescence was detected for GO suspensions in culture medium or in exposed cells under assay conditions. For all fluorophores a GO concentration dependent attenuation of fluorescence intensity was observed. However, the degree of quenching was only significant (higher than 10%) with the highest concentration of the GO suspension tested (75 mg/L) with the AlamarBlue™ assay (Valdehita et al., 2023, supplementary data), therefore only this particular result should be interpreted with caution. The degree of quenching was independent of the fluorophore concentration. 3.2.1. AlamarBlue™ assay Metabolic activity was assessed by the AlamarBlue™ assay after GO exposure for 24 and 96 h, (Fig. 1A). The results in our study show that G. Mol´ es et al. Chemosphere 364 (2024) 143005 5 after 24 h of exposure the metabolic activity remained unchanged. After 96 h there was a statistically significant increase in metabolic activity reaching a maximum following exposure to 9.38 mg/L of GO and then gradually dropped to basal levels with 75 mg/L. A potential fluorescence quenching effect could also be contributing to the observed drop at 75 mg/L as interference of GO was seen at this concentration (Valdehita et al., 2023). 3.2.2. CFDA-AM assay The membrane integrity was assessed by the CFDA-AM assay after GO exposure for 24 h and 96 h. The effect was time and concentration dependent, after 24 h the membrane integrity remained unchanged while after 96 h there was a statistically significant decrease in cell viability at concentrations ≥18.75 mg/L (Fig. 1B). 3.2.3. Neutral red uptake assay In our study a statistically significant decrease in fluorescence intensity indicating a damage in lysosomal functioning was only observed after 96 h of exposure with concentrations of ≥18.75 mg/L (Fig. 1C). 3.2.4. Generation of intracellular ROS GO induced intracellular ROS formation in a concentration and time dependent manner (Fig. 1D). After 24 h of exposure, a statistically significant increase in ROS with respect to the control was observed at the highest GO concentration (75 mg/L). After 96 h of exposure there was a statistically significant increase in ROS levels at concentrations ≥18.75 mg/L, reaching a maximum of 421.2% increase at 75 mg/L. 3.3. Internalization and intracellular fate of GO TEM micrographs of ultrathin sections were used to investigate the internalization and fate of GO in treated RTL-W1 cells. Numerous GO nanoplatelets were observed adjacent to the cell membrane. Although GO could be observed inside the cells after 24 h of treatment, the GO internalization was time dependent and more evident after 96 h. GO crossed the plasma membrane and was accumulated inside the cells as aggregate-like structures of different size and compactness either in intracellular vesicles (Fig. 2B) or freely localized in the cytosol (Fig. 2C). On some occasions, the cells responded with the formation of thick intermediate filament bundles (Fig. 2D). 3.4. In vivo toxicity tests in rainbow trout 3.4.1. Fish acute toxicity test An acute toxicity test exposing rainbow trout juveniles to 4.2, 20.6 and 100 mg/L of GO was designed following the requirements described in OECD TG 203. Due to the instability of GO dispersions detected in preliminary tests, we added turbines in each aquarium to create a stream of water to improve the stability. The GO test concentration in each aquarium was monitored by UV/Vis spectrophotometry at the beginning of the experiment and then every 24 h up to 96 h (Fig. 3). Despite the use of turbines a decrease in the concentration of GO in the water column was observed. Only the concentration of 100 mg/L showed enough stability so that the measured concentration was above 80 % of the nominal concentration (n.c.) as recommended by the TG 203 (Fig. 3). However, following OECD GD 317 recommendations on Aquatic and Sediment Toxicological Testing of Nanomaterials (OECD, 2022), geometric mean concentrations (g.m.c.) were calculated to determine the real exposure concentration in all the aquaria during the exposure period, resulting in values of 1.1, 9.89 and 88.5 mg/L for the used nominal concentrations of 4.2, 20.6 and 100 mg/L, respectively (Fig. 3). After 96 h of exposure, only one fish died in one of the batches with the treatment of 4.2 mg/L. There was no mortality during the depuration phase. No visible abnormalities concerning equilibrium and swimming Table 1 Hydrodynamic size distribution of GO dispersions during in vitro and in vivo tests test. Time (h) Dispersion medium Concentration (mg/L) Z-ave a (nm) PDI b Average HDD c nm ±SEM (%) d Z-Potential mV Peak 1 Peak 2 Peak 3 0 Milli-Q H 2 O 1000 e g g g g – – 750 f g g g g – – 96 Milli-Q H 2 O 750 f 188 ±5 0.24 ±0.01 253 ±13 (98) 4417 ±302 (2) – – 0 L-15 4.6 g g g g g – 18.75 g g g g g – 75 g g g g g – 96 L-15 4.6 156 ±12 0.60 ±0.04 331 ±19 (86) 30 ±4 (10) 7.6 ±0.1 (4) – 18.75 203 ±17 0.42 ±0.06 304 ±20 (90) 45 ±5 (7) 4531 ±142 (3) – 75 345 ±24 0.49 ±0.02 519 ±29 (85) 119 ±9 (9) 4497 ±148 (6) – 0 Aquarium 4.26 1723 ±143 0.45 ±0.05 1614 ±182 (94) 181 ±22 (2.4) 5160 ±355 (3.6) −14.2 ±0.2 H 2 O 20.66 2544 ±110 0.38 ±0.02 1957 ±114 (85.1) 4384 ±615 (14.5) 465 (0.4) −14.7 ±0.3 100 2409 ±140 0.35 ±0.07 2383 ±179 (95.2) 5239 ±87 (4.8) –−16.3 ±0.5 24 Aquarium 4.26 3540 ±375 0.62 ±0.12 1722 ±414 (100) – – – H 2 O 20.66 3516 ±51 0.37 ±0.05 2236 ±244 (100) – – – 100 2750 ±186 0.26 ±0.02 2605 ±240 (99.1) 5444 ±116 (0.9) – – 48 Aquarium 4.26 1629 ±491 0.83 ±0.05 1192 ±183 (46.6) 242 ±26 (43.6) 4922 ±162 (9.8) −10 ±0.5 H 2 O 20.66 5065 ±357 0.56 ±0.07 1266 ±211 (100) – – −14.6 ±0.5 100 2621 ±193 0.22 ±0.02 2675 ±219 (98) 5304 ±115 (2) –−15.5 ±0.2 72 Aquarium 4.26 2090 ±331 0.55 ±0.08 1533 ±183 (94.4) 5111 ±147 (5.6) – – H 2 O 20.66 5451 ±509 0.72 ±0.07 1190 ±194 (55.6) 248 ±56 (44.4) – – 100 2565 ±200 0.24 ±0.07 2585 ±210 (100) – – – 96 Aquarium 4.26 2120 ±221 0.48 ±0.07 1722 ±185 (93.7) 5006 ±205 (6.4) –−7.3 ±0.5 H 2 O 20.66 4742 ±2434 0.99 ±0.01 179 ±74 (100) – – −10.3 ±1.3 100 2741 ±168 0.22 ±0.07 2797 ±204 (100) – – −15.2 ±0.2 Dynamic Light Scattering analysis of GO dispersions in aquarium water. a Z-average size (Z-ave). b Polydispersity index (PDI). c Average Hydrodynamic Diameter (HDD). d relative intensities of the size peak (%). e GO stock before centrifugation. f GO stock after centrifugation. g Values from Valdehita et al., 2023. Data are represented as mean ±SEM (n ≥6). G. Mol´ es et al. Chemosphere 364 (2024) 143005 6 behaviour were detected during the exposure period in control or treatment groups at concentrations ≤20.66 mg/L (n.c.). At concentration of 100 mg/L (n.c.) it was difficult to properly observe the fish due to the opacity of the suspensions. No changes in the pH, dissolved oxygen and temperature were observed during the test. 3.4.2. Histopathological alterations Control fish showed a normal morphology of gills and liver (Fig. 4A and C) but some disturbances were observed in fish exposed to GO (Fig. 4B and D). For example, in fish exposed to 100 mg/L (n.c)/88.5 mg/L (g.m.c.) GO, gills exhibited signs of secondary lamellar fusion, obliteration of inter-lamellar spaces due to hyperplasia of epithelial cells, and inflammation (Fig. 4 B). Livers in the control group exhibited typical architecture, showing hepatocytes with normal nucleus (Fig. 4C). Conversely, fish exposed to different concentrations of GO presented many hepatocytes with nuclei exhibiting hyperchromatic pyknotic appearance and a higher degree of vacuolization than the control group (Fig. 4D). 3.4.3. EROD/BFCOD activities in liver and gills Significant increments of EROD and BFCOD activities with respect to the control group were measured in livers after 96 h of exposure and 96 h of depuration (Fig. 5A and C). Both activities are commonly used to monitor Cyp1a and Cyp3a function, respectively. In our study, the analyses of EROD activity showed that GO was able to stimulate the Cyp1a enzyme activity in a concentration dependent manner (Fig. 5A). During the exposure period, the basal activity detected in the control group was of 6 pmol resorufin/min/mg protein. The concentration of 100 mg/L (n. Fig. 1. Cytotoxicity of GO in RTL-W1 cells. Effect of increasing concentrations of GO (0.29–75 mg/L) after 24 and 96 h. Cytotoxicity was assessed by means of the AlamarBlue™ (A), CFDA-AM (B) and NRU (C) assays. Additionally, the intracellular ROS generation was assessed (D). Lines and symbols represent the mean ±SEM of at least three independent experiments. Statistically significant differences with respect to the control group (One-way ANOVA, Dunnett’s Post-hoc test) are indicated as follows: *p <0.05, **p <0.01, ***p <0.001. Fig. 2. Intracellular fate of GO nanoplatelets. TEM images of non-exposed (A) and 96 h GO exposed RTL-W1 cells (B, C and D). B) Internalized GO concentrated in cytoplasmic vesicles (white arrows). C) GO freely-localized in the cytosol (white arrows). D) Formation of thick intermediate filament bundles (white arrows). G. Mol´ es et al. Chemosphere 364 (2024) 143005 7 c)/88.5 mg/L (g.m.c.) GO provoked a significant increase of 5.7 fold the control group reaching 34 pmol resorufin/min/mg protein. The levels of EROD activity kept rising during the depuration period until levels reached 514 pmol resorufin/min/mg protein in the fish that had been exposed to this highest concentration (Fig. 5A). During the depuration period the levels of EROD activity in the control group also experienced an increase compared to that seen during the exposure period (Fig. 5A). Regarding BFCOD activity, the analyses showed that there was no effect during the exposure period but there was a significant increase in the activity during the depuration period with GO concentrations of 20.6 and 100 mg/L. Maximum levels of 593 pmol HFC/min/mg were reached in a similar way to that observed with the EROD activity (Fig. 5C). In gills, no significant increments with respect to the control group could be detected with any concentration of GO for either EROD or BFCOD activities. The basal values detected in the control group were of 1 pmol resorufin/min/mg and 6.7 pmol HFC/min/mg, respectively (Fig. 5B and D). 3.4.4. Gene expression analyses Transcription levels of genes associated with the metabolism of xenobiotics/detoxification activities (cyp1a, cyp3a and ahr2) and response to inflammatory challenge (il8 and il1b) were analysed by RT-qPCR in liver and gills after 96 h of exposure and a 96 h depuration period (Fig. 6). Exogenous luc1 reference gene was used to normalize the results under the experimental conditions. In liver, the expression levels of cyp1a, cyp3a and ahr2 during the exposure period showed a tendency to decrease according to an increased concentration of GO (Fig. 6A, C and 6E), although it was only significant for ahr2 at 100 mg/L (n.c) (88.5 mg/L (g.m.c)) of GO (Fig. 6C). In the same period, a significant increment of il8 expression was detected with 20.6 mg/L (n.c) (9.89 mg/L (g.m.c)) of GO (Fig. 6G). During the depuration period, a strong and significant down regulation of cyp1a and ahr2 expression was detected in a GO concentration dependent manner (Fig. 6A and C). On the contrary, no changes in the cyp3a and il8 expression was detected compared to control fish levels (Fig. 6E and G). il1b expression in livers have not been presented due to low amplification levels achieved. In gills, a significant down regulation in cyp1a and ahr2 expression was observed during the exposure period (Fig. 6B and D) and an up regulation in il1b in a GO concentration dependent manner (Fig. 6F). No significant changes in il8 expression were detected (Fig. 6H). During the depuration period a significant decrease of cyp1a expression was only detected with the highest concentration of GO (100 mg/L) (Fig. 6B). There were no changes detected in ahr2, il1b and il8 expression levels from control fish during this period (Fig. 6D, F and 6H). 4. Discussion In this study, we have performed and shown how an assessment of the aquatic toxicity of a commercial GO (GRAnPH®) to fish can be performed by integrating in vitro and in vivo approaches using a hepatic cell line and juveniles of rainbow trout, respectively. First, in an alternative to animal use approach and in keeping with the 3Rs principles of Replacement, Reduction and Refinement of animal testing, we assessed the effects of GO on the cellular level in vitro using the cell line RTL-W1. We chose a cell line of hepatic origin since the liver plays an essential role in detoxification processes. Testing the toxicity of some nanomaterials is particularly challenging given their particulate nature, often instability of the suspensions in aqueous media, and the potential interferences associated with some assay reagents (Kalman et al., 2019). In this study, along the experimental period and for each GO concentration, no major differences in hydrodynamic size of GO in dispersions were observed by DLS in L-15 culture media except for the highest concentration of GO tested (75 mg/L), where a tendency of GO to agglomerate was detected. The results from the cytotoxicity assays showed no reduction in viability at exposure concentrations ≤75 mg/L following 24 h exposure. There were, however, concentration dependent losses in viability (plasma membrane damage and lysosomal disruption) measured after 96 h, showing a time dependent influence on cytotoxicity potential with significant reductions at concentrations ≥18.75 mg/L. Interestingly instead of a decrease in metabolic activity, a concentration-dependent increase was detected at lower concentrations, which may be related with cellular detoxification activities involving increased activity of mitochondrial oxidoreductases. However, this increased activity gradually decreased from the concentration of 18.75 mg/L until reaching basal levels with the highest concentration of GO. The observed interference due to the quenching caused by GO in the fluorescence measurements could have contributed to this described drop. Similar results were also observed after 7 days exposure in RTL-W1 cells using the same GO nanoplatelets (Valdehita et al., 2023). An increase in metabolic activity was also described by Lammel et al. (2013) in human liver cancer cell line Hep G2 cells after 72 h exposure with GO, however this was not the case in GO exposed PLHC-1 topminnow fish hepatoma cells (Lammel et al., 2015), and a decrease of metabolic activity was observed in CLC cells after 72 h exposure with GO (Kalman et al., 2019). The reason for these differences could be associated with differences among material properties, cell lines (physiology/culture conditions) or the degree of GO interaction and or internalization in the distinct cells. What is clear however is that effects can only be evidenced after an extended exposure duration (e.g. >24–96 h). This is important as traditionally experiments with cell culture are performed over a 24 h period. In the CFDA-AM assay, GO caused a time and concentrationdependent decrease in fluorescence intensity indicating that there was plasma membrane damage that could be due to the physical interaction of GO with the phospholipid bilayer. TEM micrographs of ultrathin sections of RTL-W1 corroborated that GO nanoplatelets were able to penetrate through the plasma membrane. Plasma membrane damage by GRM has been also reported previously in human erythrocytes and Fig. 3. GO concentration throughout the acute toxicity test. GO concentrations were determined by UV/Vis spectrophotometry (270 nm) in aquarium water every 24 h. Results are expressed as percentage of the nominal concentration. To determine the real exposure concentrations to which the fish were exposed during the test, a geometric mean was calculated for the respective nominal concentrations. G. Mol´ es et al. Chemosphere 364 (2024) 143005 8 different cells lines such as PLHC-1 and the carp leukocyte cell line CLC (Kalman et al., 2019; Lammel et al., 2013; Lammel and Navas, 2014; Liao et al., 2011). RTL-W1 cells appear to respond to GO damage with the formation of thick intermediate filament bundles (Fig. 2D) that may support the plasma membrane and prevent further loss of structural integrity. A similar response was observed previously in Hep G2 cells after GO exposure (Lammel et al., 2013). In our study, GO was observed inside the cell either freely localized in the cytosol or enclosed in intracellular vesicles. According to Kalman et al. (2019), the different localization inside cells could be dependent on the properties of the nanomaterial and the cell line used, leading to different intracellular fate and toxicity. Lammel and Navas (2014) observed freely-localized GO in the cytosol of PLHC-1 cells that physically interacted with the mitochondria and nucleus, suggesting that free unprocessed material could cause greater cellular damage than encapsulated/vesicle contained forms. In our study, no evidence of interactions with intracellular organelles was observed by TEM. However, a moderate impairment of lysosomal function was detected with the NRU assay after 96 h of exposure at concentrations of ≥18.75 mg/L. Induction of oxidative stress is considered one of the principal mechanisms underlying nanomaterial toxicity (Shvedova et al., 2012; Xia et al., 2006). In our study, GO induced the generation of intracellular ROS in a concentrationand time-dependent manner, with increases seen both after 24 and 96 h. Increase in levels seen already at 24 h when no other effects could be evidenced suggest that this is a very sensitive early effect marker, while the much higher levels of intracellular ROS at 96 h exposure agree with the significant levels of cytotoxicity observed at this exposure timepoint. The ROS levels measured here after 96 h exposure in RTL-W1 cells were considerably higher than those obtained in Hep G2 and PLHC-1 cells treated with GO nanoplatelets (Lammel et al., 2013, 2014). Results from the CFDA-AM and NRU assays suggested loss of plasma membrane integrity and impairment of lysosomal function, respectively, which could be related with the physical injury caused by the interaction with GO. The increase in metabolic activity (AlamarBlue™ assay) could be associated with the initiation of energy-dependent processes involved in membrane repair or isolation of the internalized GO. In order to investigate if these effects seen at the cellular level translate to mortality or disturbances on the whole organism, an in vivo acute toxicity test with juveniles of rainbow trout was performed following OECD TG 203 (OECD, 2019b). According to TG 203, the concentration of the chemical being tested should be preferably maintained at ≥80% of the nominal concentration throughout the test. To aid GO concentration maintenance, we introduced turbines in all the aquariums. Although the turbines improved dispersion of GO in aquariums, only the concentration of 100 mg/L was maintained at levels above 80% of the nominal values after 96 h. The results of UV/Vis spectrophotometry showed that there is a greater precipitation and loss of GO from the water column at lower concentrations of GO in the dispersions. Nevertheless, an estimation of the real exposure concentrations was obtained by taking measurements at various time intervals and applying a geometric mean approach, according to the indications of OECD GD 317 (OECD, 2022). During the experiment, only one fish died after 96 h of exposure in Fig. 4. Light microscopy of histological sections of rainbow trout gills and liver stained with toluidine blue after 96 h of GO exposure. A) Gills of control group exhibiting normal morphology. The primary lamellae (PL), secondary lamellae (SL), chondrocytes (CH) and erythrocytes within the capillary lumen (E) can be observed. B) Gills of fish exposed to 100 mg/L GO showing lamellar fusion (LF) and obliteration of inter-lamellar spaces due to hyperplasia and inflammation (IS). C) Liver of control group exhibiting normal morphology. The hepatic lobule (HL), the sinusoids (S), the Kupffer cells (KC) and the typical nucleus of the hepatocytes (N) can be observed. D) Liver of fish exposed to 20.6 mg/L GO showing pyknotic nuclei (PN) and vacuole formation (V). Scale bars: A - B) 50 μ m and C - D) 10 μ m. G. Mol´ es et al. Chemosphere 364 (2024) 143005 9