Fate and effects of graphene oxide alone and with sorbed benzo(a)pyrene in mussels Mytilus galloprovincialis

Journal of Hazardous Materials 452 (2023) 131280 Available online 28 March 2023 0304-3894/© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/bync-nd/4.0/). Research Article Fate and effects of graphene oxide alone and with sorbed benzo(a)pyrene in mussels Mytilus galloprovincialis Nagore Gonz´ alez-Soto a , Nagore Blasco a , Mireia Irazola b , Eider Bilbao a , Lúcia Guilhermino c , Miren P. Cajaraville a , * a CBET Research Group, Dept. Zoology and Animal Cell Biology, Science and Technology Faculty and Plentzia Marine Station, University of the Basque Country (UPV/ EHU), Basque Country, Spain b Dept. Analytical Chemistry and Plentzia Marine Station, University of the Basque Country (UPV/EHU), Basque Country, Spain c Ecotoxicology Research Group, ICBAS, Institute of Biomedical Sciences of Abel Salazar and Research Group of Ecotoxicology, Stress Ecology and Environmental Health (ECOTOX), CIIMAR, Interdisciplinary Centre of Marine and Environmental Research, University of Porto, Portugal HIGHLIGHTS GRAPHICAL ABSTRACT •GO was found in digestive tract lumen and feces of mussels exposed to GO or GO+BaP. •BaP bioaccumulated in mussels exposed to GO+BaP and especially to BaP. •An environmentally relevant concentration of GO was genotoxic to mussel hemocytes. •Inflammation in digestive gland/gonad and oocyte atresia found in exposed mussels. •Effects of carried BaP and enhanced toxicity of GO+BaP vs GO or BaP were identified. ARTICLE INFO Editor: Karina S. B. Miglioranza Keywords: Graphene family nanomaterials Polycyclic aromatic hydrocarbons Trojan Horse effect Bivalve mollusks Biological responses ABSTRACT Graphene oxide (GO) has gained a great scientific and economic interest due to its unique properties. As incorporation of GO in consumer products is rising, it is expected that GO will end up in oceans. Due to its high surface to volume ratio, GO can adsorb persistent organic pollutants (POPs), such as benzo(a)pyrene (BaP), and act as carrier of POPs, increasing their bioavailability to marine organisms. Thus, uptake and effects of GO in marine biota represent a major concern. This work aimed to assess the potential hazards of GO, alone or with sorbed BaP (GO+BaP), and BaP alone in marine mussels after 7 days of exposure. GO was detected through Raman spectroscopy in the lumen of the digestive tract and in feces of mussels exposed to GO and GO+BaP while BaP was bioaccumulated in mussels exposed to GO+BaP, but especially in those exposed to BaP. Overall, GO acted as a carrier of BaP to mussels but GO appeared to protect mussels towards BaP accumulation. Some effects Abbreviations: CDNB, 2,4-Dinitrochlorobenzene; AChE, Acetylcholinesterase; BaP, Benzo(a)pyrene; Cat, Catalase activity; DMSO, Dimethyl sulfoxide; GC/MSQqQ, Gas chromatography followed by triple quadrupole mass spectrometry; GC–MS, Gas chromatography–mass spectrometry; GPx, Glutathione peroxidase; GST, Glutathione S-transferase; GO-PEI, GO functionalized with polyethyleneimine; GI, Gonad index; GFNs, Graphene family nanomaterials; GO, Graphene oxide; GO+BaP, Graphene oxide with sorbed BaP; IDH, Isocitrate dehydrogenase; NMs, Nanomaterials; NR, Neutral red; NADPH, Nicotinamide adenine dinucleotide phosphate; PFOS, Perfluorooctane sulfonate; POPs, Persistent organic pollutants; PAHs, Polycyclic aromatic hydrocarbons; KP, Potassium phosphate; ROS, Reactive oxygen species; SOD, Superoxide dismutase; Tris, Tris(hydroxymethyl)-aminomethan. * Corresponding author. E-mail address: [email protected] (M.P. Cajaraville). Contents lists available at ScienceDirect Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat https://doi.org/10.1016/j.jhazmat.2023.131280 Received 9 November 2022; Received in revised form 13 March 2023; Accepted 22 March 2023 Journal of Hazardous Materials 452 (2023) 131280 2 observed in mussels exposed to GO+BaP were due to BaP carried onto GO nanoplatelets. Enhanced toxicity of GO+BaP with respect to GO and/or BaP or to controls were identified for other biological responses, demonstrating the complexity of interactions between GO and BaP. 1. Introduction Nanomaterials (NMs) are commonly defined as a diverse class of materials with at least one dimension at the nanoscale (<100 nm) [1]. In 2011, the European Commission defined a nanomaterial as: “a natural, incidental or manufactured material containing particles, in an unbound state or as an aggregate or as an agglomerate and where, for 50% or more of the particles in the number size distribution, one or more external dimensions is in the size range 1–100 nm. In specific cases and where warranted by concerns for the environment, health, safety or competitiveness the number size distribution threshold of 50% may be replaced by a threshold between 1% and 50% [2]. Among NMs, carbon-based NMs have attracted great scientific and technological attention due to the physico-chemical properties of their nanometric structures. Carbon-based NMs include among others, fullerenes, carbon nanotubes, carbon black or graphene; but among all of them, graphene stands out for the unique properties that make it the thinnest, strongest and lightest known material [3]. Graphene family nanomaterials (GFNs) are used in a wide range of applications including electronic devices, a new generation of batteries, sensors [4], biomedical applications [5], anticorrosion coatings [6,7], agricultural procedures [8] or environmental applications such as waste water treatments [4], water desalination [9] and pollutants removal [4, 10]. There are more than 26,000 graphene related patents [11] and more than 100 graphene based products [12]. In fact, the production of GFNs is higher in comparison to the rest of NMs [13] and it is expected to continue growing as the expensive and low efficient methods used nowadays for graphene production are improved. Therefore, graphene production is expected to reach 3800 tones, with a worth of 300 millions by 2027 [14]. Graphene oxide (GO) is a precursor in graphene synthesis and one of the most studied graphene derivates [15] whose reactivity and capacity for chemical functionalization [12] are important characteristics related to the presence of functional oxygen groups both on the surface (hydroxyl and epoxy groups) and in the edges (carboxyl groups) of the sheet [16]. Due to the general interest on GO and its increased production in the last years, GO is being released into the environment during its life cycle [17] both through direct release (e.g., sewage effluents, river influx) or indirectly (e.g., aerial deposition, dumping and run off) [18]. Therefore, GO will definitely reach coastal and marine ecosystems [12] and probably will interact with different components of the natural system, which may alter behaviour, transport, fate and toxicity of GO [19]. Transport and fate of GFNs are governed mainly by the stability of suspensions, which may be altered by environmental factors such as salinity, organic matter concentration, oxidation status and bioturbation. In aquatic environments, GO can disperse and form relatively stable suspensions that endure in the water column [18,20]. Such behavior may facilitate the uptake of GO by a large number of organisms through different routes such as ingestion or respiration, as it has been described for other NMs [21]. In addition, the high persistence of GO can result in their bioaccumulation and biomagnification in food webs [4, 22], increasing their potential impact in marine ecosystems, even if the GO concentrations released into the environment are relatively low [23]. Therefore, levels of GO in surface waters should be a primary concern [24]. However, the environmental concentrations of GO are still largely unknown [12]. The presence of GO was already detected in the biomass from wastewater treatment plants [25]. Recent studies consider that the predicted environmental concentration of GO could be similar to that described for other NMs such as multi-walled carbon nanotubes, which is in the range 0.001–1000 μ g/L for aquatic environments [26]. Toxicity of NMs is strongly related to their size, shape, surface properties or chemical composition, which are key characteristics for risk assessment [1,27]. Overall, at cellular level, GFNs have been reported to decrease integrity of the cell´s plasma membrane, possibly as a consequence of entry of nanosheets into cells by direct penetration or endocytosis [18]. Sheets can also disrupt the plasma membrane due to induced invaginations or perforations [28,29], or even by the destructive extraction of lipids [30]. Disruption of the plasma membrane and internalization of GFNs can provoke the formation of reactive oxygen species (ROS) and lead to oxidative stress [29], which is considered one of the main underlying mechanisms of toxicity of NMs [27]. Oxidative damage caused by the increased intracellular production of ROS can lead to mitochondrial and lysosomal dysfunction and finally to a decrease in the viability of hemocyte cells of marine mussels [29]. In addition, oxidative stress and/or physical cell damage can also cause DNA damage resulting in the fragmentation and destruction of nucleic acids [31]. These alterations at the cellular level may lead to effects at higher biological levels, such as reduction of metabolic activity [4,32], histopathological lesions [33,34], alterations in behavior and locomotor functions [35] and adverse impact on the reproduction capacity, growth and survival [36]. In aquatic environments, generally NMs do not appear alone, but are found within complex mixtures of chemical contaminants originated both from natural and anthropogenic sources. Several studies have demonstrated that GFNs show a great adsorption capacity for persistent organic pollutants (POPs) such as polycyclic aromatic hydrocarbons (PAHs), mainly due to their large surface area and hydrophobicity [37–40]. This adsorption produces accumulation of organic pollutants on the surface of the NM, which may increase their uptake by aquatic organisms [41] and their potential adverse effects [39], a phenomenon known as Trojan horse effect. Originally, Limbach et al. [42] introduced the Trojan horse concept to refer to the extended toxicity caused by metal nanoparticles in comparison to their soluble forms, due to the continuous release of metal ions in and out of cells. Later, carrier or Trojan horse effect was introduced to designate the increased uptake and accumulation of environmental pollutants facilitated by NMs [43,44]. Thus, the Trojan horse effect is generally defined as the possible threat of NMs due to their ability to adsorb and carry adsorbed compounds to organisms, a phenomenon known to occur for many NMs and several metal or organic pollutants [45–48]. Most works are based on co-exposure of NMs and environmental pollutants and the toxicity of the mixture is usually compared to that of the dissolved pollutant and not to that of the NM alone [46,48] which could lead to an underestimation of the contribution of the NM to the toxicity of the mixture [46]. Different reviews consider that a Trojan horse effect occurs when the exposure to the mixture causes a significantly higher toxicological effect than the exposure to the compounds separately [45–50]. Among PAHs, benzo(a)pyrene (BaP) is a priority pollutant [51,52], commonly used in ecotoxicology studies and known to cause effects at different levels of biological organization [53]. BaP is a genotoxic and carcinogenic agent capable of producing tissue and DNA damage [54, 55], oxidative stress [56], peroxisome proliferation [57], lysosomal dysfunction [58], endocrine disruption [59], among other effects in marine organisms, including bivalves. In addition, BaP can interact with emerging pollutants of high concern, such as microplastics and NMs, in marine mussels resulting in toxicological interactions [55,60]. Among target organisms, mussels (Mytilus sp.), are considered model organisms for the evaluation of pollutants including micro and nanoscale particulate materials, due to their highly developed mechanisms for cellular internalization of particles through endocytosis and phagocytosis, for N. Gonz´ alez-Soto et al. Journal of Hazardous Materials 452 (2023) 131280 3 physiological functions such as intracellular digestion and cellular immunity [27,61,62]. To improve the basis for the risk assessment of GFNs to the marine environment, it is very important to study the toxicity of GFNs, alone and combined with POPs in marine organisms, such as mussels. Therefore, the aim of this work was to investigate the fate and effects of GO, alone or with sorbed BaP, in adult marine mussels (M. galloprovincialis) using Raman spectroscopy and a battery of biological responses, respectively. Overall, this work contributes to understand the Trojan horse effect of GFNs towards BaP in mussels. 2. Materials and methods 2.1. Obtention of GO and preparation of GO with sorbed BaP Commercial nanoplatelets of graphene oxide (GO) were purchased from Graphenea (San Sebastian, Spain) as stable suspensions. According to the supplier, the concentration of GO in the dispersion was 10 mg/mL. This was experimentally confirmed by measuring GO in aqueous dispersions with a UV–vis spectrophotometer (UV-2550, Shimadzu) at 660 nm, where graphene follows the Lambert-Beer law. For that aim, a calibration curve (Fig. S1) was prepared from dispersions with known GO concentrations. The concentration of GO experimentally determined was 9.91 mg/mL. According to the manufacturer’s information, nanoplatelets showed lateral dimensions ranging from 500 nm to few microns and thickness was <2 nm. Oxygen content was about 40% wt. Characterization of the same batch of GO by transmission electron microscopy and atomic force microscopy was reported previously by Martínez-´ Alvarez et al. [63]. The protocol to prepare GO with sorbed benzo(a)pyrene (BaP) was based on previous work (Martínez-´ Alvarez et al. [63]). Briefly, after preparing the BaP solution of 100 μ g/L containing 0.01% dimethyl sulfoxide (DMSO, purity 99% Sigma, St. Louis, Missouri) in a glass bottle, GO was added in a 0.5 mg:10 mL GO/BaP proportion (weight/- volume). Sorption process was allowed by shaking samples in an orbital shaker (300 rpm) for 24 h in the dark at 21 ±1 ◦C. After 24 h, samples were centrifuged in an Allegra X30R centrifuge (9509 g, 30 min). Supernatant was discarded and the pellet was re-suspended in 50 mL of MilliQ water. Samples were vortexed before dosing into tanks. Samples containing GO alone were processed in the same way, but using only MilliQ water. For sorption experiments [63], samples were prepared in the same way described above. After centrifugation, absence of GO in supernatants was assessed by spectrophotometry at 230 nm. Then, BaP was quantified in supernatants by gas chromatography/mass spectrometry after solid phase micro extraction. Based on the BaP concentration measured in the aqueous phase, the amount of BaP sorbed to GO was indirectly calculated. This process was done in triplicate and vials containing BaP solutions without GO were processed in parallel to monitor potential BaP loss due to evaporation, degradation, sorption onto the vial walls or other factors during the experimental procedure. According to Martínez-´ Alvarez et al. [63] the proportion of BaP sorbed onto GO was 96.7% ±0.5%. 2.2. Sampling and acclimation of mussels Roughly 460 mussels Mytilus galloprovincialis (3.5–4.5 cm shell length) were collected in February 2019 in Mundaka, Basque Country (43◦24 ´04.9”N, 2◦41 ´41.6”W). Mussels were maintained in aquaria facilities at the Plentzia Marine Station (PiE) of the University of the Basque Country (UPV/EHU), for acclimation during 21 days. Acclimation was carried out in a 300 L polypropylene tank with a recirculating seawater system. Marine water was collected with a pump at 10 m depth in the mouth of the Butroi estuary (43◦24′21′′N, 2◦56′47′′W) and filtered (particles ≤3 µm) before reaching the marine station. Mussels were not fed for two days and then they were fed once a day with the Isochrysis galbana microalgae (2 ×10 7 cells/mussel-day) for 19 days. I. galbana (TIso clone) cultures were obtained from the Animal Physiology Laboratory at UPV/EHU. During acclimation, light regime was 12 L/12D and room temperature was kept at 18 ◦C. Water parameters were checked daily with a multichannel probe. The variation of water parameters during the acclimation period was (mean ±standard deviation): salinity of 32.40 ±0.27, pH of 7.25 ±0.35 and temperature of 16.02 ±0.08 ◦C. The dissolved O 2 was always above 80%. Fig. 1. Summary of the experimental design. A: Exposure groups, B: preparation of graphene oxide (GO) with adsorbed benzo(a)pyrene (BaP) and C: exposure set up for the GO+BaP group as an example. N. Gonz´ alez-Soto et al. Journal of Hazardous Materials 452 (2023) 131280 4 2.3. Mussel exposure In order to keep a homogeneous suspension of GO, without aggregation or precipitation, a water recirculation system consisting of two water pumps was installed in the aquaria. Before mussel exposure, tanks were exposed for 24 h to GO at the same concentration tested in order to saturate the system. After 21 days of acclimation, mussels were exposed for 7 days to graphene oxide (GO), graphene oxide with sorbed BaP (GO+BaP) or BaP alone in two 20 L replicate tanks per treatment (GO R1 and GO R2, GO+BaP R1 and GO+BaP R2, BaP R1 and BaP R2) with 57 mussels each (Fig. 1). Two control tanks, also with 57 mussels each, were run in parallel (Control R1, Control R2). Mussel samples were taken after 7 days of exposure. An exposure concentration of 500 μ g/L GO was selected, based on environmentally relevant concentrations for multiwalled carbon nanotubes, which range from 1 μ g/L to 1 mg/L [12]. For the GO+BaP exposure groups, the nominal concentration of BaP incubated with GO was 100 μ g/L. Finally, for the BaP exposure groups, 96.7 μ g/L BaP was used as the equivalent BaP concentration sorbed in GO+BaP preparations [63]. During the experiment, water was changed daily. Throughout exposure, mussels were fed once a day with I. galbana (2 ×10 7 cells /mussel-day) two hours before changing water. While water of tanks was changed, 5 mussels per tank were selected randomly and placed in individual glass containers with clean water to collect feces (Fig. 1). The light regime and room temperature were kept at 12 L/12D and 18 ◦C, respectively. Water parameters were checked daily with a multichannel probe: salinity (32.93 ±0.06 PSU), dissolved O 2 (>77%), pH (7.48 ±0.23) and temperature (16.62 ±0.25 ◦C). 2.4. Bioaccumulation of BaP The presence of BaP in seawater was checked by gas chromatography–mass spectrometry (GC–MS) analysis in water collected after 20 h of exposure in all the tanks according to the protocol described in Katsumiti et al. [64]. Briefly, twister stir bars (20 mm length and 0.5 mm film thick, Gerstel GmbH & Co. KG, Mülheim an der Ruhr, Germany) were employed for extraction of BaP from the marine water. Twister bars were introduced in samples (~20 mL) during 195 min. Then, twister bars were cleaned with Milli-Q water and dried with paper tissue. BaP was desorbed from the twister bars using a commercial thermal desorption TDS-2 unit connected to a CIS-4 injector (Gerstel GmbH & Co. KG,Mülheiman der Ruhr, Germany). The desorption unit was then connected to an Agilent 6890 gas chromatograph coupled with an Agilent 5975 mass spectrometer system (Agilent Technologies, Palo Alto, USA) for BaP determination. Mussels for chemical analysis (14–18 mussels per tank) were stored at −40 ◦C and analyzed at IPROMA (Castellon, Spain) to determine bioaccumulation of BaP in whole mussel tissues. Mussels were lyophilized and homogenized. Extraction was performed with acetone and dichloromethane using the QuEChERS method [65]. Concentration of BaP was determined using gas chromatography followed by triple quadrupole mass spectrometry (GC/MS-QqQ) [66], using BCR-682 mussel tissue as reference material. The limit of quantification in the analyses was 5 ng/g dry weight. 2.5. Determination of graphene oxide in mussel tissues and feces Three mussels per tank were dissected, frozen in liquid nitrogen and maintained at −40 ◦C until further analysis. Then 20 µm sections were obtained in a cryostat (Leica CM 3050 S) and observed under an inVia Renishaw microscope in order to get Raman spectra using a 532 nm laser. Conditions were set using the 100x objective as follows: 1–5µm steps in the tissue, 0.2–0.4 s, 10% laser intensity, 1 accumulation and focused in 1200 nm. For each sample, serial cryotome sections were fixed in Baker’s solution (formaldehyde 4% (v/v), NaCl 2% (w/v), calcium acetate 1% (w/v)) for 15 min, rinsed in distilled water and stained 20 s in 0.1% toluidine blue to get the topographic reference of the tissue sample. Mussels´feces were also collected and analyzed by Raman spectroscopy. Mussel feces were completely dried and placed in an aluminum foil before getting the Raman spectrum. Conditions were set using a 100x objective as follows: 100% laser intensity, 1 s, 80 accumulation and focused in 1200 nm. As reference, spectra of microalgae Isochrysis galbana, BaP stock solution and DMSO were obtained in the same conditions. 2.6. Cellular biomarkers in hemocytes Hemolymph of 8 mussels per tank was withdrawn from the posterior adductor muscle and cell viability, catalase activity and DNA damage in terms of micronuclei formation were measured in hemocytes of individual mussels. Neutral red (NR) uptake was assessed according to Borenfreund & Puerner [67] with modifications explained in Gonz´ alez-Soto et al. [60]. Catalase activity (Cat) was assessed according to Aebi [68] as modified in Gonz´ alez-Soto et al. [60]. Protein concentration was measured following the Bradford method 69 to normalize absorbance data. Cat activity was expressed as the consumption of mM H 2 O 2 /min/mg protein. The micronucleus assay was performed according to Duroudier et al. [70]. Micronucleated cells were classified following the accepted criteria for mussels: well-preserved cell cytoplasm, micronuclei not touching the main nucleus, similar or weaker staining than the main nucleus and size of micronuclei ≤1/3 in comparison to the main nucleus. Other nuclear abnormalities such as binucleated cells, occurrence of nucleoplasmic bridges and nuclear buds were scored according to Pinto-Silva et al. [71] and Bolognesi & Fenech [72]. Results are reported in ‰ frequencies. 2.7. Enzyme activities in mussel tissues Digestive gland, gills and adductor muscle of 10 mussels per tank were dissected out, frozen in liquid nitrogen and maintained at −80 ◦C until further analysis. Adductor muscle was homogenized in 0.1 M potassium phosphate (KP) buffer (pH 7.2) for acetylcholinesterase (AChE) determination. Gills were cut in two halfs; one half was homogenized in 0.1 M KP buffer (pH 6.5) for glutathione S-transferase (GST) determination and the second half in 0.1 M KP buffer (pH 7.4) for Cat, glutathione peroxidase (GPx) and superoxide dismutase (SOD) determination. Digestive glands were divided in three parts. The first part was used for GST determination, the second for Cat, GPx and SOD, and the last piece was homogenized in 50 mM tris buffer (Tris(hydroxymethyl)-aminomethan, pH 7.8) for isocitrate dehydrogenase (IDH) determination. Tissues were homogenized in each buffer following a 1:10 proportion, tissue weight: volume of buffer. The activity of the enzyme AChE and IDH, GST, GPx and SOD were determined as described in previous studies where some modifications of the original tecnhiques were made [73,74]. Briefly, AChE activity was determined according to Ellman et al. [75] at 412 nm and expressed as the production of 5,5’-dithiobis-(2-nitrobenzoic acid) in nmol/min/mg protein. IDH activity was determined according to Ellis & Goldberg [76] at 340 nm and expressed as the production of nicotinamide adenine dinucleotide phosphate (NADPH) in nmol/min/mg protein. GST activity was determined according to Habig et al. [77] at 340 nm and expressed as the production of 2,4-Dinitrochlorobenzene (CDNB) conjugates with the thiol group of glutathione in nmol/min/mg protein. Cat activity was determined as previously explained. GPx activity was determined according to Floh´ e& Günzler [78] at 340 nm and expressed as the consumption of NADPH in nmol/min/mg protein. SOD activity was determined according to McCord & Fridovich [79] at 550 nm and given in SOD units (1 SOD unit =50% inhibition of the reduction of cytochrome C per mg protein). Each enzyme activity was normalized to N. Gonz´ alez-Soto et al. Journal of Hazardous Materials 452 (2023) 131280 5 protein concentration using Bradford method [69]. 2.8. Histopathology of the digestive gland Digestive glands of 10 mussels per tank were dissected out and processed following a standard protocol for histology [80]. Briefly, tissues were fixed in 4% formalin in individual cassettes and dehydrated through a graded series of ethanol that finished in xylene using an automatic tissue processor (Leica ASP300; Leica Instruments, Wetzlar, Germany). Then, samples were embedded in paraffin and 5 µm sections were cut in a Leitz 1512 microtome (Leica Instruments, Wetzlar, Germany). Slides were dryed in an oven at 37 ◦C (24 h) and stained with hematoxylin/eosin [81] using an autostainer XL V2.02 (Leica). Slides were mounted in DPX and analyzed under a BX51 light microscope (Olympus, Tokyo, Japan). Vacuolization, atrophy and necrosis of the digestive tubule epithelium, fibrosis, hemocytic infiltration, aggregation of brown cells in the connective tissue and in digestive tubules and presence of parasites were assessed in the digestive gland following Villalba et al. [82], Garmendia et al. [83] and Bignell et al. [84]. The prevalence of each alteration (number of individuals showing each pathology divided by the number of individuals of each group) was calculated as percentage. 2.9. Gamete development, gonad index and histopathology of gonad Mantle of the same animals used for the histopathological analysis of the digestive gland were dissected out and processed following the standard protocol for histology described before. Sex ratio, gamete developmental stages and gonad index (GI) were determined. Six gamete stages were distinguished [85] and a gonad index (GI) value, ranging from 0 (resting gonad) to 5 (mature gonad), was assigned to each developmental stage as in Gonz´ alez-Soto et al. [60], adapted from Kim et al. [86]. Oocyte atresia and necrosis, fibrosis, hemocytic infiltration, aggregation of brown cells, and occurrence of parasites were also assessed in gonads following Ortiz-Zarragoitia & Cajaraville [87]. Prevalences were Table 1 Bioaccumulation of BaP in mussel soft tissues (ng/g dry weight) in control mussels and in mussels exposed to GO, GO+BaP, and BaP for 7 days. N =number of mussels is indicated. LoQ =Limit of Quantification: 5 ng/g dry weight. Control GO GO+BaP BaP R1 N=14 R2 N=18 R1 N=16 R2 N=14 R1 N=14 R2 N=17 R1 N=15 R2 N=14 <LoQ <LoQ <LoQ <LoQ 23 28 18,700 18,400 Fig. 2. A-C: Heat maps showing the detection of graphene oxide (GO) in cryostat sections of the digestive gland of different mussels; A: Control mussel, B: mussel exposed to GO for 7 days, C: mussel exposed to GO+BaP for 7 days. D: Raman spectra from the zones highlighted in red in map B (green) and the spectrum obtained for the GO stock (red). E: Raman spectrum from the map in A. Scale bars: A and B: 100 µm and C: 500 µm. N. Gonz´ alez-Soto et al. Journal of Hazardous Materials 452 (2023) 131280 6 calculated as for the digestive gland. In addition, intensity of oocyte atresia was assessed using a semiquantitative scale: 0normal gonad, 1less than a half of follicles are affected, 2about half of follicles are affected, 3more than half of follicles are affected and 4all follicles are affected [86]. Intensity was calculated as Sp/NH, where Sp is the score corresponding to the intensity of atresia and NH is the number of specimens with atresia [83]. 2.10. Whole organism responses Condition index was assessed according to Navarro et al. [88]. Soft tissues of 7 animals per tank were excised from the shells, dried at 80 ◦C for 24 h and weighted. Afterwards, mussel shell lengths were recorded with a Vernier caliper and condition index was calculated as tissue dry weight (g) / [shell length (cm)] 3 . 2.11. Data analysis Statistical analyses were carried out with the aid of the statistical package SPSS 24 (IBM Analytics, Armonk, NY), and the significance level was 0.05. All data sets were tested for normality and homogeneity of variance using Kolmogorov-Smirnov´s and Levene’s tests, respectively. Normally distributed data, which met the assumptions of homogeneity of variances, were assessed via one-way ANOVA. When significant differences were found, the Tukey’s post-hoc test was used to Fig. 3. A: Neutral red uptake (given as absorbance/10 6 cells); B: catalase activity (given as mM H 2 O 2 /min/mg prot) in hemocytes of control mussels and in mussels exposed for 7 days to GO, GO+BaP and BaP. Box-plots show median value (horizontal line), 25%−75% quartiles (box) and standard deviation (whiskers). Dots denote outliers. Letters indicate significant differences among treatments (one-way ANOVA with Tukey’s post-hoc, p <0.05); C-J: Light micrographs of mussel hemocytes during the micronucleus assay showing C) normal hemocyte of a control mussel; D) hemocyte of a mussel exposed to GO showing a micronucleus (arrow); E) hemocyte of a mussel exposed to GO showing a nuclear bud; F) binucleated cell with nucleoplasmic bridge of a mussel exposed to GO; G) binuclear hemocyte of a mussel exposed to BaP; H) hemocyte of a mussel exposed to GO+BaP showing a micronucleus (arrow); I) hemocyte of a mussel exposed to GO+BaP showing a nuclear bud; J) binucleated cell with nucleoplasmic bridge of a mussel exposed to GO+BaP. Scale bar: 10 µm. Table 2 Frequency (‰) of micronuclei, binucleated cells, binucleated cells with nucleoplasmic bridges and nuclear buds in control mussels (N =8) and mussels exposed for 7 days to GO (N =7), GO+BaP (N =6) and BaP (N =8). Letters denote statistical differences among groups (p <0.05 after Kruskal-Wallis test followed by Dunn’s post-hoc). Micronuclei Binucleated cells Binucleated cells with nucleoplasmic bridges Nuclear buds Control 0 A 0.125 ±0.35 2.38 ±2.67 2.63 ±2.13 A GO 1.86 ±1.35 B 0.14 ±0.38 5.43 ±2.51 8.71 ±2.93 B GO+BaP 3.67 ±3.08 B 0.50 ±0.55 6.33 ±2.58 7.17 ±1.72 AB BaP 0.88 ±0.83 AB 0.63 ±0.52 3.88 ±2.23 5.88 ±1.73 AB N. Gonz´ alez-Soto et al. Journal of Hazardous Materials 452 (2023) 131280 7 identify significant different treatments. Data which did not met the above assumptions were analyzed by the one-way Kruskal-Wallis test, followed by the Dunn’s post-hoc test when significant differences were found. For histopathological data expressed as percentages, the X 2 test was used [89]. When no differences were found between the two replicate tanks, data sets were mixed and displayed as: Control, GO, GO+BaP, BaP. However, when differences were found between replicates, data sets were maintained separate and displayed as: Control R1, Control R2, GO R1, GO R2, GO+BaP R1, GO+BaP R2, BaP R1, BaP R2. 3. Results 3.1. Bioaccumulation of BaP BaP concentration in control mussels and in mussels exposed to GO was below the detection limit (Table 1). BaP was bioaccumulated in mussels exposed to GO+BaP (23–28 ng/g dry weight), but especially in mussels exposed to BaP (18,400–18,700 ng/g dry weight) (Table 1). In water, BaP was detected only in the tanks exposed to BaP alone. 3.2. Determination of graphene oxide in mussel tissues and feces GO was detected only in the lumen of the digestive tract of mussels exposed to both GO and GO+BaP (Fig. 2), indicating that even if GO was internalized, it was not accumulated in the rest of mussel tissues after 7 days of exposure. Accordingly, GO was detected in feces of mussels exposed to GO and GO+BaP from day one of exposure (Fig. S2). Spectra of microalgae Isochrysis galbana, BaP stock solution and DMSO were not observed in mussel tissues or feces. 3.3. Cellular biomarkers in hemocytes Viability of hemocytes decreased in mussels exposed to GO+BaP and BaP in comparison to controls (Fig. 3A). No differences were observed among groups in the activity of catalase (Cat) (Fig. 3B). Regarding genotoxicity, higher micronuclei frequency was observed in hemocytes of mussels exposed to GO and GO+BaP in comparison to controls (Table 2, Fig. 3D, G). In addition, higher frequency of nuclear buds was observed in hemocytes of mussels exposed to GO than in control mussels (Table 2, Fig. 3E, H). No differences were observed among groups in binucleated cells (Table 2, Fig. 3J) and binucleated cells with a nucleoplasmic bridge (Table 2, Fig. 3F, I). 3.4. Enzyme activities in mussel tissues There were statistically significant differences in the activity of acetylcholinesterase (AChE) between replicates and thus, data were treated separately (Fig. 4A). Activity of AChE was lower in one of the replicates exposed to GO+BaP (R1) and in both replicates exposed to BaP in comparison to one of the control replicates (R1) (Fig. 4A). Inhibition of isocitrate dehydrogenase (IDH) was observed in mussels exposed to BaP in comparison to mussels exposed to GO and GO+BaP (Fig. 4B). Similarly, inhibition of Glutathione-S-Transferase (GST) was observed in the digestive gland of mussels exposed to GO+BaP and BaP in comparison to control mussels (Fig. 4C), while no response was observed in gills (Fig. S3A). (caption on next column) Fig. 4. Acetylcholinesterase (AChE) activity in the aductor muscle (given as nmol/min/mg prot); B) Isocitrate dehydrogenase (IDH) activity in the digestive gland (given as nmol/min/mg prot); C) Glutathione-S-Tramsferase (GST) activity in the digestive gland (given as nmol/min/mg prot) control mussels and in mussels exposed for 7 days to GO, GO+BaP and BaP. Box-plots show median value (horizontal line), 25%−75% quartiles (box) and standard deviation (whiskers). Dots denote outliers. Letters denote statistical differences among means (Kruskal-Wallis test followed by Dunn’s post-hoc in AChE and GST and one-way ANOVA followed by Tukey’s post-hoc in IDH, p <0.05). N. Gonz´ alez-Soto et al. Journal of Hazardous Materials 452 (2023) 131280 8 Regarding antioxidant enzymes, induction of Cat activity was observed in the digestive gland of mussels exposed to GO+BaP in comparison to controls (Fig. 5A), but no response was observed in gills (Fig. S3B). No differences were found in the activity of glutathione peroxidase (GPx) in the digestive gland (Fig. S3C), but inhibition was observed in gills of mussels exposed to GO in comparison to mussels exposed to BaP (Fig. 5B). Superoxide dismutase (SOD) was inhibited in the digestive gland of mussels exposed to BaP in comparison to control mussels and mussels exposed to GO+BaP, while induction of SOD was observed in mussels exposed to GO+BaP in comparison to mussels exposed to GO and BaP (Fig. 5C). However, no clear trend was observed in gills (Fig. S3C). 3.5. Histopathology of digestive gland Statistical differences were found between replicates for some histopathological alterations. In those cases data sets for the two replicates were treated separately (Table 3). Fibrosis (Fig. 6A) was widespread in all mussels including controls. All groups showed high prevalences, significantly higher in BaP R1 than in GO R1 (Table 3). Prevalence of hemocytic infiltration (Fig. 6B) was similar in control R2 and all exposure groups but it was significantly higher in the GO R2 group than in control R1 (Table 3). Overall, aggregation of brown cells (Fig. 6C, D) were more frequent in exposed groups than in controls, at least in one of the replicates (Table 3). Aggregation of brown cells in the connective tissue was significantly higher in mussels exposed to GO, GO+BaP and BaP than in controls (Table 3) while in the digestive tract epithelium, it was significantly higher in GO R2 and GO+BaP R1 than in control R1 (Table 3). In addition, aggregation of brown cells in the digestive tract epithelium was significantly higher in GO+BaP R1 than in GO R1 (Table 3). Areas of necrosis of digestive tubule epithelium were found in both control and exposed mussels. Prevalence of necrosis was significantly higher in GO+BaP R2 and BaP R2 compared to control R2 (Table 3). Necrosis of digestive tubule epithelium appeared to be associated to the occurrence of an intracellular ciliated protozoan. It was found in the digestive epithelium of mussels of all experimental groups, with its highest prevalence in controls, especially in comparison to mussels exposed to BaP (Table 3). In addition, the protozoan Nematopsis sp. was the most common parasite in the digestive gland. It appeared in the connective tissue of almost all mussels, with high prevalences in all the replicates (Table 3). Other parasites such as Mytilicola intestinalis were also observed in the digestive gland of some mussels, but with low prevalences (Table 3). 3.6. Gamete development, gonad index and histopathology of mussel gonad Mussels from different groups were in a similar gamete development stage. Spawning was the predominant stage (Fig. S4) as expected for the experimental period. There were no differences in the sex ratio and gonad index among groups (Fig. S4). Fibrosis (Fig. 6F), hemocytic infiltration (Fig. 6G) and aggregation of brown cells (Fig. 6H) were widely observed in the gonad of both control and exposed mussels. Prevalence of fibrosis was significantly higher in BaP R2 mussels than in Control R2 mussels (Table 4). Hemocytic (caption on next column) Fig. 5. Catalase (Cat) activity in the digestive gland (given as mM H 2 O 2 /min/ mg prot); B: Glutathione Peroxidase (GPx) activity in gills (given as nmol/min/ mg prot); C: Superoxide dismutase (SOD) activity in the digestive gland (U/mg prot) of control mussels and in mussels exposed for 7 days to GO, GO+BaP and BaP. Box-plots show median value (horizontal line), 25%−75% quartiles (box) and standard deviation (whiskers). Dots denote outliers. Letters denote statistical differences among means (one-way ANOVA followed by Tukey’s post-hoc in Cat and SOD, Kruskal-Wallis test followed by Dunn’s post-hoc in GPx, p <0.05). N. Gonz´ alez-Soto et al. Journal of Hazardous Materials 452 (2023) 131280 9 infiltration occurred both in the connective tissue and within the gonad follicles and prevalences were higher in GO+BaP R1 and BaP R2 than in both control replicates and in GO R1 (Table 4). Aggregation of brown cells, both in the connective tissues and gonad follicles, was higher in mussels exposed to GO and to BaP than in controls (Table 4). However, the main histopathological alteration observed was oocyte atresia (Fig. 6F) with prevalences over 75% in all control and exposed groups (Table 4). Control mussels showed low intensity of atresia which can be linked to the developmental stage of the gonad (Table 4, Fig. S4). On the other hand, stages 2 and 3 of oocyte atresia were only recorded in exposed mussels and the highest intensity was found in mussels exposed to GO+BaP (Table 4). Nematopsis sp. protozoan was found in the connective tissue of the gonad of almost all mussels (Table 4). 3.7. Whole organism responses No differences were recorded among groups in mussel condition index (Fig. S5). 4. Discussion In this work, the toxic effects of short-term (7 days) exposure to GO alone (500 µg/L) and to GO with sorbed BaP (500 µg/L GO incubated with 100 µg/L BaP) has been assessed using a wide selection of biological responses, from molecular to organism levels. For comparison, the effects of exposure to BaP alone at a concentration of 96.7 µg/L, equivalent to that sorbed in GO+BaP preparations [63], were also assessed. In parallel, the fate of GO nanoplatelets in mussels was studied by Raman spectroscopy, as well as BaP bioaccumulation. After 20 h of exposure, BaP or GO could not be detected in the water of tanks exposed to GO+BaP, suggesting that mussels had uptaken the GO nanoplatelets with sorbed BaP. Little is known about the desorption dynamics of PAHs from GO and other GFNs in the marine environment. In aqueous solutions, adsorption of phenanthrene into magnetic graphene is mostly irreversible [90]. The stability of GO with sorbed BaP could explain that BaP was undetectable in water in GO+BaP tanks as mussels would internalize BaP adsorbed into GO nanoplatelets. However, the concentration of BaP bioaccumulated in mussels was much lower in mussels exposed to GO+BaP compared to those exposed to dissolved BaP. Sanchis et al. [41] reported that the Trojan horse effect is more likely to occur with organic pollutants that present an intermediate affinity to carbon-based NMs, than with non polar compounds showing aromatic rings such as BaP, as the most non polar compounds would be irreversibly bound to the carbon-based NMs decreasing their bioavailability. Alternatively, it could be that BaP bioaccumulation was low in mussels exposed to GO+BaP because of the excretion of GO nanoplatelets with sorbed BaP. This idea is supported by the detection of GO in feces of mussels exposed to GO and to GO+BaP. At studying the stability of GO in tanks with and without oysters, Khan et al. [34,91] concluded that the animals were actively removing the GO from the water, in line with our results. Filtration is the common pathway for the internalization of different NMs during the feeding process of filter-feeding organisms such as mussels. After a first contact with the gills, NM aggregates are ingested and, depending on their size, they can be accumulated in the digestive gland and/or translocated to hemolymph and then distributed to other organs [92]. In the present work GO was not detected in gills, probably due to the short contact between GO and gills. This could explain the lack of effects on enzyme activities studied in gills of mussels exposed to GO and GO+BaP in comparison to controls. Presence of GO was confirmed through Raman spectroscopy in the lumen of the digestive tract and feces of mussels exposed to both GO and GO+BaP, suggesting that GO tends to accumulate in organs related with food intake, as previously reported by Josende and colleagues [93]. The presence of GO in the digestive tract has been previously reported in exposures to other aquatic invertebrates, such as nematodes [36] and crustaceans [93–98], but not in bivalves. To the best of our knowledge this is the first time that the presence of GO in the digestive tract of bivalves was confirmed through Raman spectroscopy. As the digestive gland is one of the most important organs in mussels, responsible for intracellular digestion and also for antioxidant defense and pollutant sequestration and detoxification, significant damages in this organ could lead to impact at the organism level [92,99,100]. GO could then be translocated from the digestive system to the rest of organs [101,102]. However, Bortolozzo et al. [36] observed that carboxyl groups in GO edges can avoid permeability of GO nanoplatelets through the intestine in nematodes. This could explain the presence of GO in feces after each digestive cycle and the absence of GO out of the digestive tract in this work, although there is yet no conclusive data on the distribution and excretion strategies for GFNs inside animal bodies [5]. Even though GO was not detected in hemocytes by Raman spectroscopy, cytotoxicity occurred in hemocytes of mussels exposed to GO+BaP and BaP with respect to controls. Further, genotoxic effects were observed in hemocytes of mussels exposed to GO and GO+BaP in comparison to controls. DNA damage has been previously reported in other organisms exposed to GO [103–106], but there is a single work reporting genotoxicity of GFNs to mussels [107]. Genotoxicity may be provoked by the physical damage caused by nanoplatelets and/or through oxidative stress [93,101]. In most publications the comet assay has been used to measure genotoxicity after exposure to GFNs, but DNA strand breaks determined by the comet assay can be reversible [108]. Flasz et al. [105] reported that DNA damage measured in crickets Acheta domesticus by the comet assay after 5 and 25 days of GO exposure was completely reversed after 10 days of depuration. On the other hand, Table 3 Prevalence of histopathological alterations in the digestive gland of both replicates (R1 and R2) of control mussels and mussels exposed to GO, GO+BaP and BaP for 7 days. Data are shown in percentages of 10 mussels per experimental replicate tank. Letters denote statistical differences for each alteration (X2 test, p <0,05). Inflammatory responses Parasites Fibrosis Hemocytic infiltration Brown cells in the connective tissue Brown cells in the digestive tract epithelium Necrosis of digestive tubule epithelium Intracellular ciliated protozoan Nematopsis sp. Mytilicola intestinalis Control R1 90 AB 10 A 0 A 10 A 50 AB 80 A 80 10 Control R2 90 AB 40 AB 30 A 30 ABC 80 A 70 A 100 20 GO R1 80 A 50 AB 60 B 20 AB 50 AB 60 AB 90 10 GO R2 60 AB 60 B 60 B 60 CB 60 AB 30 AB 100 0 GO+BaP R1 90 AB 40 AB 60 B 70 C 60 AB 60 AB 80 20 GO+BaP R2 90 AB 50 AB 40 B 20 AB 30 B 40 AB 80 10 BaP R1 100 B 40 AB 70 B 10 A 80 A 40 B 90 10 BaP R2 90 A B 50 AB 60 B 30 ABC 20 B 40 B 70 20 N. Gonz´ alez-Soto et al. Journal of Hazardous Materials 452 (2023) 131280 16 [a]pyrene exposure in mussel Mytilus galloprovincialis. Arch Environ Contam Toxicol 63, 534–543. [127] Guo, B., Feng, D., Xu, Z., Qi, P., Yan, X., 2021. Axute benzo[a]pyrene exposure induced oxidative stress, neurotxocity and epigenetic change in blood clam Tegillarca granosa. Sci Rep 11, 18744. [128] Hu, X., Wei, Z., Mu, L., 2017. Graphene oxide nanosheets at trace concentrations elicit neurotoxicity in the offsprung of zebrafish. Carbon 117, 182–191. [129] Urban-Malinga, B., Jakubowska, M., Hallmann, A., Dabrowska, A., 2021. Do the graphene nanoflakes pose a potential threat to the polychaete Hediste diversicolor? Chemosphere 269, 128685. [130] Ding, J., Chen, S., Qu, M., Wang, Y., Di, Y., 2020. Trophic transfer affects cytogenetic and antioxidant responses of the mussel Mytilus galloprovincialis to copper and benzo( α )pyrene. Mar Environ Res 154, 104848. [131] Cheung, C.C.C., Siu, W.H.L., Richardson, B.J., Luca-Abbott, S.B.D., Lam, P.K.S., 2004. Antioxidant responses to benzo[a]pyrene and Aroclor 1254 exposure in the greenlipped mussel, Perna viridis. Environ Pollut 128, 393–403. [132] Wang, C., Zhao, Y., Zheng, R., Ding, X., Wei, W., Zuo, Z., Chen, Y., 2006. Effects of tributyltin, benzo[a]pyrene, and their mixture on antioxidant defense systems in Sebastiscus marmoratus. Ecotoxicol Environ Saf 65, 381–387. N. Gonz´ alez-Soto et al.