The metal nanoparticles, including metal oxides, represent one of the major classes of commercial nanomaterials, which are manufactured on a large scale for both industrial and household applications (Chang et al., 2012). Copper (II) oxide nanoparticles (CuO NPs) are increasingly used in several products (Huang et al., 2010; Chang et al., 2012; Rossetto et al., 2014). The wide variety of applications entails the risk of environmental contamination, as a consequence of the environmental release of CuO NP during their production, use and disposal (Weinberg et al., 2011; Sanchís et al., 2013; Fan et al., 2014). This kind of contamination could bias both organisms and ecosystems (Gambardella et al., 2013), as CuO NPs can exert toxic effects but also antimicrobial activity on (environmental) microbes (Bondarenko et al., 2013; Rossetto et al., 2014); hence, they could seriously affect estuarine and coastal environments, considered the ultimate sink for different kinds of NPs (Canesi et al., 2012). Accordingly, the investigation on CuO NP effects in the brackish/marine ecosystems has become a very important issue, and must include information on NP fate, transport and toxicity (Lowry et al., 2012).

Nevertheless, ecotoxicity of CuO NPs, particularly to marine organisms, is still little explored and the available data are often inconsistent, due to different and no standardized experimental approaches that highlighted some procedural limitations, mainly related to the stability of nanomaterials during the test exposure (reviewed by Minetto et al., 2016). The great majority of the studies evaluated endpoints as oxidative stress, genotoxicity, bioaccumulation and behavioral impairments (Bondarenko et al., 2013; Ivask et al., 2013; Minetto et al., 2016; Gonçalves et al., 2017) or soil toxicity (Unrine et al., 2010; Amorim and Scott-Fordsmand, 2012; Amorim et al., 2012; Gomes et al., 2012; Gomes S. I. et al., 2015; Gomes S. I. L. et al., 2015; Gonçalves et al., 2017), with few EC₅₀ identifications, although it is mandatory to define the hazard of CuO NPs in environment and human health and enhance their safe use.

Bacteria are an important component of brackish/marine ecosystems and alterations of the microbial communities could have significant effects on biogeochemical cycling and other critical ecosystem services. Toxicity tests based on microorganisms are gaining popularity even because they are relatively quick, reproducible, cheap and do not imply ethical issues (Parvez et al., 2006). Among bacterial bioassays, V. fischeri luminescence inhibition test, based on the enzymatic activity of the bacterial luciferase, is the most common and well-standardized one (Azur Environmental, 1995). However, this bioassay has some constrains even for conventional contaminants: for example, it cannot be utilized with samples at salinities exceeding a quite narrow range (APAT IRSA-CNR, 2003), or with colored and turbid samples, due to possible interferences with luminescence measurements. As regard other bioassays with bacterial species (B. subtilis, E. coli, L. brevis and S. aureus), most of them are based on the same endpoint, i.e., the inhibition of replication rate of the bacterial culture, but they use different and no standardized protocols (Baek and An, 2011; Kaweeteerawat et al., 2015; Bondarenko et al., 2016). These methods present some criticisms that can influence the NP toxicity assessment and affect the repeatability of the bioassay. For instance, the possible interactions between NPs and organic matter in the exposure medium, the excessively reduced test volumes, without any mixing during the exposure, or the need of correction factors for colored/turbid samples.

In order to provide a useful tool for NP toxicity assessments in marine environments, we recently developed a new bioassay with the marine bacterium Vibrio anguillarum and demonstrated its effectiveness in evaluating the toxicity of a reference toxicant (Rotini et al., 2017). The model organism V. anguillarum is a Gram-negative, short curve-shaped rod bacterium with a polar flagellum. It was chosen because of its intrinsic characteristics: it is halotolerant, ubiquitous and plays important ecological roles in marine/brackish ecosystems (Thompson et al., 2004). The bioassay allows to assess the decrease of bacterial culturability and to determine the EC₅₀ (i.e., the concentration causing the 50% reduction of bacteria actively replicating and forming colonies, after 6-h exposure).

In this study, the suitability of this bioassay in evaluating NP toxicity has been checked. To this end, the study evaluated and compared the ecotoxicity of CuO NPs, CuO-bulk and Cu²⁺ ion (CuSO₄ 5H₂O) in a wide range of salinity (5–35‰), by using the recently proposed test on the marine bacterium V. anguillarum. To deepen the CuO NP behavior during the test and relate it to toxicity, the physicochemical characterization of NPs in the exposure medium was carried out and the size distribution, sedimentation rates and Cu²⁺ dissolution from NPs were analyzed. These data allow a better understanding of NP aggregation dynamics and stability at the different salinities. Furthermore, to provide the most accurate picture, even CuO NP internalization or morphological alterations were evaluated in bacteria at the end of the test exposure.

The bioassay with V. anguillarum allowed to assess the CuO NP toxicity and highlighted a significant variation of the toxic effect with salinity, depending on a different aggregation state of NPs.

According to the OECD guidelines (OECD, 2014), the CuO NP behavior in the control and exposure media has been accurately investigated, to link characterization data with ecotoxicological results; this allowed a correct interpretation of the response. In our study, through the analyses by DLS and LUMISizer, the aggregation and stability of the CuO NPs were verified at three different salinities of the exposure media. The DLS analysis (see Table 1 and Figure S1) highlighted a bimodal particle size distribution, with nanoscale (about 80 nm) and microscale (about 200 nm) aggregates, in both mQW and 0.5% saline solution; while, the dispersions in 2.0 and 3.5% saline solutions showed a single size population of microscale aggregates (250–300 nm). The low ζ-potential absolute values also confirmed the aggregation state and instability of the CuO NPs. Although Zeta (ζ-) potential is commonly considered a key parameter to describe the NP behavior in complex environmental media, it is worth to note that the high ionic strength of saline exposure medium, may shield electric charge of NPs, lowering the measured ζ-potential.

The LUMISizer analysis (see Figure 2) confirmed the characterization by DLS and provided an even more accurate representation of the particle size distribution in the exposure media. In fact, LUMISizer better characterized the big size nanoparticle populations: the presence of nano- and micro-scale aggregates in mQW and 0.5% saline solution was confirmed, and also the microscale aggregates in 2.0 and 3.5% saline solutions up to a size of about 1,000 nm were detected. Furthermore, the LUMISizer analysis allowed to quantify the sedimentation rates of CuO NP aggregates in the exposure media (see Figure S2). Only a negligible sedimentation of the CuO NPs as aggregates occurred during the exposure time (6 h), even at the highest salinity. This result was obtained with some dedicated features of the bioassay (i.e., test tube size, exposure volume and continuous agitation), designed to limit the NP sedimentation. Being inversely correlated with NP toxicity (Buffet et al., 2011; Villareal et al., 2014), the NP sedimentation rate represents a relevant parameter to describe the NP behavior in environmental media, particularly for salt water matrix. However, only few studies address the NP sedimentation rate and, at our best knowledge, this study is the only one that measured the sedimentation rates in seawater.

The particle diameter and sedimentation rate increased at increasing salt concentrations, due to the effect of ionic strength. Low salinity (mQW and 0.5% NaCl) allowed optimal and stable dispersion of CuO NPs, without differences between the two media. While, at high salinity (2.0–3.5% NaCl) agglomeration was promoted by the presence of salt ions, which shield the NP charge reducing the repulsive effect among NPs.

The increased particle size at high salinity determines a decrease of the total surface area; this implies a decrease of the superficial reactivity of NPs (because of agglomeration) which, in turn, produces a reduction of the toxic effects. The results obtained with this bioassay confirm that agglomeration and stability of CuO NPs are inversely related to their toxicity. At the lowest salinity (0.5% NaCl), when CuO NP agglomeration was the least and stability of dispersions was optimal, high toxicity (EC₅₀ = 12.6 mg/l) was recorded, suggesting a particularly relevant hazard for these NPs when released in brackish habitats. At the highest salinity, typical of marine environments, the occurring aggregation and sedimentation suggest that Cu NPs may accumulate in sediments (Buffet et al., 2013) and therefore, benthic organisms are supposed to be the most exposed to NPs. However, the fate and bioavailability of NPs in sediments can be highly variable and depend on several abiotic and biotic factors, including ionic strength of water, amount of suspended natural organic matter (Keller et al., 2010), light intensity or temperature (Zhou et al., 2012) and biogenic transformation processes (decomposition, bioturbation, or digestion; Farré et al., 2009).

Table 3 summarizes the studies on CuO NP toxic effects on bacterial species; they show highly variable values of the Effect Concentrations, even in test on the same bacterial species/strain. This depends on the type of nanoparticles and on the experimental conditions (Bondarenko et al., 2013). This variability confirms the difficulty to compare the ecotoxicological assessments of nanomaterials, even among standardized bioassays, and highlights the need of highly reliable and reproducible tests.

The bioassay with V. anguillarum clearly showed a significant decrease of bacterial culturability, at increasing concentration of CuSO₄ 5H₂O and CuO bulk, used in this study as solubility and size controls, respectively. Unlike CuO NPs, size and solubility controls did not elicit different toxicities at different salinities. As expected, a clearly different toxicity was observed among the three copper forms; according with the EC₅₀ values, toxicity ranking is as follows: CuSO₄ 5H₂O > CuO NP > CuO bulk. CuO NPs toxicity is an order of magnitude lower than CuSO₄ 5H₂O but higher than CuO bulk. Importantly, the EC₅₀ values for CuSO₄ 5H₂O are comparable with those obtained by acute tests on other marine species (Adams and Stauber, 2004; Manfra et al., 2015; Rotini et al., 2018) including bacteria (see Table 3); this demonstrates the good sensitivity of our recently proposed bioassay for both conventional contaminants and NPs.

It is generally accepted that CuO NP toxicity is higher than CuO-bulk, depending on size, surface characteristics, dissolution, and exposure routes (reviewed by Chang et al., 2012), on the contrary, the contribution of dissolved Cu²⁺ ions to the observed toxicity of CuO NPs is still under discussion (Ivask et al., 2013; Gonçalves et al., 2017). Several studies ascribe primarily to the Cu²⁺ ion dissolution the CuO NP toxicity in biological systems (Heinlaan et al., 2008; Aruoja et al., 2009; Kahru and Dubourguier, 2010; Mortimer et al., 2010; Bondarenko et al., 2012; Kasemets et al., 2013). A solubility-dependent toxicity of CuO NPs has been observed in Daphnia magna (Heinlaan et al., 2008; Blinova et al., 2010; Fan et al., 2012; Jo et al., 2012), Cyprinus carpio (Zhao et al., 2011) and zebrafish embryos (Lin et al., 2015). On the contrary, just as many other studies found that the toxic effects of CuO NPs do not depend from the ion dissolution (Heinlaan et al., 2008; Jiang et al., 2009; Baek and An, 2011; Isani et al., 2013). Agreeing with this last group, our results indicate a modest Cu²⁺ dissolution with an inverse salinity-dependence; as a consequence, the toxic effects observed for CuO NPs might be due to intrinsic toxicity mechanisms related to the nano-form as, for instance, aggregation plays a key role. This was already suggested by other authors (Buffet et al., 2013; Gonçalves et al., 2017), although the issue deserves further investigations. Both SEM and TEM analyses did not evidenced morphological differences between control and CuO NPs exposed bacteria. Interestingly, the SEM images (see Figure 1) showed two surface structures present on both control and exposed bacteria: outer membrane vesicles (a.k.a. blebs), known to be unique for Vibrio strains, and fibrils. Blebs and fibrils are signature of starvation and are produced by the bacterial cells in response to the growth arrest (Östling et al., 1993). This response was expected because during the test bacteria are exposed to the toxicant in the absence of nutrients. The TEM images again do not suggest CuO NP internalization into or intimate adhesion to bacteria. Some opaque particles seem to be included into the bacterial cells, but their small size and shape does not support their identification as the NPs used in this study. Hence, our results do not chime with the evidence by Kaweeteerawat et al. (2015), which found nano Cu species strongly bound to or internalized within E. coli cells and stated that both nano Cu and nano CuO can be internalized into the bacterial cell. Probably the different findings can be ascribed to the size of NPs and their aggregates that, in our study, are not compatible with internalization into a Vibrio bacterial cell.

The bioassay with V. anguillarum benefits from several procedural points in assessing the NP toxicity. As a first point, it assesses culturability, i.e. the bacterial capability to actively replicate and form colonies after the exposure to NPs in a saline medium. The endpoint culturability allows an easier comparison of results with those from the most common ecotoxicological bioassays for marine environments, which often have survival/mortality as endpoint; this will facilitate its introduction in test batteries. Furthermore, it is interesting to highlight that the reduction of bacterial culturability can be also due to the reversible VBNC (Viable But NonCulturable) state which is known to be induced/triggered by a variety of stressors such as out range of growth temperature, oxygen concentrations, heavy metals, etc. (Oliver, 2005, 2009), hence increasing the sensitivity of the test. Moreover, the exposure is carried out in simple saline solution, avoiding any possible interferences of nutrients with the NPs, known to modify the reproducibility of results by increasing or reducing NP bioavailability (Ivask et al., 2013; Kasemets et al., 2013). Again, the short time of exposure (6 h) is long enough to observe and evaluate acute effects on the bacterial population and short enough to limit sedimentation and aggregation of metal NPs in salt water, ensuring repeatability of the exposure conditions. Last but not least, this bioassay has no limitations for colored/turbid samples, and is applicable on a wider range of salinities, common limitations of the existing methods with microorganisms. This feature of the bioassay gives the possibility to carry out contemporary assays in a wide range of salinity (0.5–3.5%), which is known to affect NP behavior and toxicity (Corsi et al., 2014).

In this study, a recently designed bioassay with the marine bacterium V. anguillarum allowed to assess the toxic effects of CuO NPs, showing a clear dose-response relationships and a crucial role of salinity and particle aggregation in the observed toxicity. Results highlighted the high toxicity of CuO NPs, particularly at low salinity, and pointed out a relevant hazard for these NPs in brackish environments. While, at salinities typical of marine environments the high aggregation and sedimentation rate of these NPs suggest a possible accumulation in sediments, increasing the risk of exposure for benthic organisms. The influence of salinity on the NP toxicity is a hitherto little explored issue and further ecotoxicity assessments, including different NPs and bioassays are particularly needed. Moreover, our results demonstrated the effectiveness of V. anguillarum bioassay as a promising tool for the risk assessment of nanomaterials and confirmed its useful application on conventional contaminants too.

LM, AR, MT, and LoM conceived and designed the study. AR, AT, and LoM performed ecotoxicological analyses. MB, CM, and RC performed chemical analyses. FI and GB performed microscopy. MT choose the suitable Vibrio strain and the cultural conditions. AR, AT, and LoM wrote the manuscript; all authors contributed to the discussion and approved the final manuscript.