CuSO₄·5H₂O (analytical grade) was purchased from Riedel-de-Haën. 63.5 g Cu/L stock solution was prepared in deionised (DI) water and stored in the dark at room temperature. CuSO₄·5H₂O was added to the test media as 100-fold concentrated stock solution in water.

Nano-CuO (advertised particle size 30 nm) was purchased from Sigma-Aldrich. The primary size of nano-CuO was confirmed to be 31 ±12.8 nm in an earlier study by Blinova et al. [20]. TEM and SEM images of nano-CuO preparation are shown elsewhere [20,21]. 63.5 g Cu/L stock suspension on nano-CuO was made in DI water; the suspension was sonicated for 30 min as described earlier [18]. Hydrodynamic diameter of nano-CuO in DI water (measured from 20 mg/L suspension using Zetasizer Nano-ZS; Malvern Instruments, UK) was 195 ± 2 nm. Nano-CuO was added to the test media as 100-fold concentrated stock suspension in water. Hydrodynamic size of nano-CuO in final media was measured as in case of DI water; average hydrodynamic diameter and Pdi (polydispersity index) were calculated from three parallel measurements.

The list of standard ecotoxicological and microbiological media used in this study is presented in Table 1. All mineral salts used for preparation of the media were of analytical grade. Cas-aminoacids (AA) (casein hydrolysate), Tryptone, Yeast extract and Malt extract were from LabM (Lancashire, UK), Peptone was from Difco Laboratories (Beckton Dickinson, MD, USA). The test media were prepared by dissolving the desired amount of ingredients in DI water, autoclaved (121 °C for 15 min) or filter-sterilized (0.1 μm filter pore size, Minisart) and stored at room temperature.

Concentration of Cu in the test media was also determined by AAS-graphite furnace method according to the standard procedures (EVS-EN ISO/IEC 17025:2005) in a certified laboratory of the Institute of Chemistry, Tallinn University of Technology (Estonia). Additionally, pH (Orion PerpHect ROSS) and conductivity (EcoScan CON 5 conductometer; Eutech Instruments, Singapore) of the test media were measured. Conductivity of DI water was measured to be 0.0003 mS/cm.

A copper ion-selective electrode 96–29 ionplus (Orion Research, Thermo Scientific, MA, USA) was used. Before measurement, the electrode was thoroughly washed with DI water, then with 0.025 M H₂SO₄, followed by polishing of the sensor surface with Al₂O₃ polishing strip. Electrode was calibrated daily using 10⁻⁸ M (6.4 × 10⁻⁴ mg Cu/L)–10⁻¹ M (6,350 mg Cu/L) CuSO₄·5H₂O dilutions in DI water. Prior measurement, the ionic strength of all solutions was adjusted by supplementing the sample with 0.1 M NaNO₃. Ionic strength of the test solution was adjusted because of its importance in the response of ISE as shown by Sauvé et al. [22]. Appropriate dilutions of CuSO₄ and nano-CuO were prepared in 5 mL of test media and the measurements were conducted in 30 mL polypropylene tubes. Dilutions of nano-CuO were allowed equilibrate for 2 h at 30 °C before the measurement. The limit of detection (LOD) of Cu-ISE in each test media was calculated as recommended for ion-selective electrodes by IUPAC [23]. Briefly, log(10) of the added Cu was plotted against the electrode potential and the crossing point between the linear increase of the electrode potential and the line representing the electrode background potential was designated as the limit of detection (Cu-ISE_LOD) (Figure S1). The LODs and respective standard deviations were calculated from three independent measurements. Free Cu was calculated using the following equation: (1) Free   Cu ( % ) = Cu   ISE LOD   in   test   media Cu   ISE LOD   in   DI   water × 100

In addition, the Cu-ISE results were used to calculate free Cu at EC₅₀ of different organisms for CuSO₄ and nano-CuO. The EC₅₀ values were obtained from earlier published studies (except that of S. cerevisiae in YPD media that was personal communication from Dr. K. Kasemets, NICPB, Estonia). Experimental EC₅₀ values were obtained in exactly the same media (composition of the media was verified) that were used for Cu speciation analysis in this study (Table 1). To calculate the amount of free Cu at EC₅₀ values for CuSO₄ or nano-CuO in different media, the potential of Cu-ISE at EC₅₀ concentrations in these media was measured and compared to CuSO₄ or nano-CuO concentrations that induced similar electrode potential in DI water (Figure S2). The respective concentration in DI water was considered as the concentration of free Cu at this EC₅₀ value.

Chemical equilibrium model Visual MINTEQ 2.51 [28] was used to calculate Cu ion speciation in CuSO₄ solutions prepared in different mineral media. Due to the lack of respective equilibrium models, no calculations were performed for complex organics-containing media. For mineral media, respective pH, concentrations of all main cations and anions were used as input; temperature was set to 23 °C. The sum of free and hydrated Cu ions was considered as the gross free Cu in the test media.

The Gram-negative Cu-sensing Pseudomonas fluorescens OS8::KnCueRPcopAlux, in which bioluminescence is specifically induced by bioavailable Cu ions [15], was used to measure bioavailable Cu. Sensor bacteria were pre-grown overnight on a shaker (200 rpm, 30 °C) in 3 mL of LB medium (Table 1) supplemented with 100 μg/L of kanamycin. 20 mL of fresh LB was inoculated with 1/50 diluted overnight culture, and bacteria were grown until mid-exponential phase (OD₆₀₀ of 0.6), and cells were separated by centrifugation at 5,000 ×g for 10 min. Cell pellet was washed twice with 20 mL of appropriate test medium and further diluted with the same medium until OD₆₀₀ ∼ 0.1 (approximately 10⁶ bacterial cells/mL). 100 μL of CuSO₄ or nano-CuO dilution or medium only (blank medium control) was pipetted onto white polypropylene 96-well microplate (Greiner Bio-one, Germany); 100 μL of bacterial suspension was added to each well and plates were incubated at 30 °C for 2 h. Bioluminescence was measured using Orion II luminometer (Berthold Detection Systems, Germany) and response of sensor bacteria to copper compounds was calculated as follows: (2) Induction   ( fold ) = Biolumines   cence   in   Cu   containing   sample Biolumines   cence   in   blank   medium

Two parallel experiments were included per individual assay and three independent individual assays were performed. Due to the different potency of the cells to induce bioluminescence in ‘nutritionally’ different media absolute induction values in these media differed. Thus, the induction of the bacterial biosensor was expressed as %, where maximum induction value was considered as 100 and induction in blank medium as 1. Limit of the detection (LOD) of the Cu-biosensors was set at 20% induction (Figure S3).

In this study, we analysed the speciation (free ion content) and bioavailability of copper in seventeen different media. The media selection included various standard ecotoxicological test media: 2% NaCl (used for Microtox toxicity test with bioluminescent bacteria Vibrio fischeri), algal medium (used for the toxicity testing with Pseudokirchneriella subcapitata according to OECD 201), two artificial freshwaters (used for tests with crustaceans Daphnia magna according to OECD 202 and with Thamnocephalus platyurus). In addition, common media used for cultivation of microorganisms were included: LB medium (an undefined rich medium that supports the growth of variety of bacteria), malt extract (ME) and YPD (Yeast Peptone Dextrose medium) both used for the cultivation of yeasts (Table 1). Also, M9 and HMM media that were supplemented with 0.5% Cas-amino acids (AA) were studied. The M9 medium has been used in our previous studies for toxicity evaluation of metals and organic chemicals to Escherichia coli [29] and for the analysis of bioavailable metals using metal-inducible bioluminescent bacterial sensors [15,30]. HMM medium has been specifically suggested for the analysis of heavy metals due to its minimal metal-complexing capability [27]. Also 0.9% NaCl and its Cas-amino acid (AA) supplemented versions were tested as 0.1% AA amended 0.9% NaCl has been applied by us earlier to study the bioavailability and toxicity of CuO nanomaterials [18,20,21]. Cu speciation and complexing potential of these different laboratory test media were studied using two methods: (i) a Cu-ISE that responds to free Cu ions and (ii) a Cu-specific bacterial biosensor that responds to bioavailable Cu.

There is a general belief that a fraction of toxicity of metal-containing nanomaterials may result from dissolved metal ions [45]. The release of Cu ions from CuO nanoparticles was reported as the main case of CuO toxicity for the crustaceans Thamnocephalus platyurus, the bacteria Vibrio fischeri [18] and Escherichia coli [21], the algae Pseudokirchneriella subcapitata [46] and the nematodes Caenorhabditis elegans [47]. Recently, Puzyn et al. [48] have developed a quantitative structure-activity relationship (QSAR) model for metal containing NPs using the formation of metal ions as a single predictor for their toxicity. Therefore, we decided to use a Cu-ISE and bacterial Cu-biosensor in parallel, to study the speciation of Cu in suspensions of CuO nanoparticles. Due to the poor bioluminescent response of Cu-biosensor bacteria in mineral media (Osterhout’s medium, AFW1, AFW2, algal medium, 2% NaCl and 0.9% NaCl; the induction of bioluminescence in these media was only about 10% of that in organics-containing media; data not shown) discussed above, we did not include these media in speciation analysis of nano-CuO.

Interestingly, the limits of detection of Cu-ISE to CuSO₄ (0.021 ± 0.005 mg Cu/L (Table 2)) and nano-CuO (0.015 ± 0.003 mg Cu/L (Table 3)) were almost identical. This indicates that at these low concentrations all Cu from nano-CuO was likely dissolved and present in the form of free ion. Analogously to CuSO₄, the fraction of fee Cu ions decreased when nano-CuO was introduced to various laboratory test media (Figure 3). Although there was generally a good correlation between the free Cu ions in CuSO₄ and in nano-CuO-spiked test media (r = 0.837; Figure 4) there were several exceptions. Specifically, in LB medium, HMM and 0.9% NaCl supplemented with 0.5%AA the concentration of free Cu in CuSO₄ was not the best predictor for the free Cu concentration in suspensions of nano-CuO. Notably, more free Cu was detected in nano-CuO suspensions than could be predicted from the results of CuSO₄ (Figure 4, the ‘outlier’-media are marked with brown color). Therefore, we suggest that differently from CuSO₄, where the only process affecting the Cu speciation was interaction of Cu ions with media components, additional processes take place in case of nano-CuO. These processes likely include agglomeration of nanoparticles, release of Cu ions from the CuO, and finally, interaction of the released Cu ions with the media components. Unfortunately, dissolution and speciation of dissolved metals from metal-containing nanomaterials has not been studied in a systematic manner. Only a recent report by Gunawan et al. [49] demonstrated differential dissolution of CuO nanoparticles in different media. Their results showed that in organics-containing ‘rich’ media the dissolution of CuO was indeed high if compared to that in water or in saline solution. Similar observation has been done for nano ZnO in a study by Li et al. [50] where the authors suggested that the affinity of Zn for ligands present in these media was responsible for dissolution of ZnO. Organic ligand-enhanced dissolution due to the presence of proteins and organic substances in the test media has been observed also in other studies concerning CdSe, iron oxides, aluminium oxides and aluminium oxyhydroxides [51]. One reason for the enhanced dissolution may be the decrease in effective hydrodynamic size of the nanomaterials as a result from coating of the particles with organic molecules. This may lead to stabilization of the nanomaterials aggregates and increased dissolution. Thus, in our experiments differential dissolution of nano-CuO due to media components and differential speciation of dissolved Cu in the media is a very likely scenario. Behaviour of nano-CuO in 0.9% NaCl amended with different amounts of amino acids is a relevant example. Differently from CuSO₄ in case of which the fraction of free Cu decreased with increasing amino acid content, very similar Cu-ISE_LOD as well as Cu-biosensor_LOD values for nano-CuO were obtained (Table 3). Thus, it could be supposed that with increasing amino acid content in the meedium, the dissolution of Cu from nano-Cu increased. In accordance with the hypothesis about organic ligand-induced decrease in effective hydrodynamic diameter of the nanomaterials, we observed that the Dh of nano-CuO decreased when Cas-aminoacids were added to saline (Table 3).

Overall, there was a good correlation between LODs of Cu-ISE and Cu-biosensor bacteria for nano-CuO (r = 0.869; Figure 2). However, in most of the tested media, the LOD of the Cu-biosensor was remarkably up to 5-fold lower than that of Cu-ISE (Figure 2; Table 3). This interesting observation suggests that in nano-CuO suspensions, bacterial cells were also able to access fractions of Cu other than just the free ion form detected by Cu-ISE. This finding is similar to our previous reports showing partial bioavailability of particulate matter bound Cu in soil and sediment samples [52]. Thus, we propose that in case of particle-containing samples (including Cu-containing nanomaterials), more Cu may be bioavailable to living (e.g., microbial) cells than could be predicted based on only dissolved fraction of Cu.

Usually, the difference in toxicity of copper to several freshwater and saltwater organisms varies remarkably, ranging from 0.005 to 10 mg/L [53]. When a certain organism group is considered, the variation is smaller, but still considerable; in a recent review [5] acute toxicity data (48–96 h LC₅₀) of Cu²⁺ for fish from the literature were compared. Thirteen EC₅₀ values for zebrafish, rainbow trout, trout, common carp, gibel carp and mullet were included; these values ranged for almost two orders of magnitudes from 0.03 till 1.4 mg/L, the median value being 0.21 mg/L. According to [53], this variability may be due to the species differences but also due to differential speciation of copper in different test environments (media) used to perform these toxicity assays.

Here, we collected ‘in house’ toxicity data (EC₅₀ values) for CuSO₄ and nano-CuO which were determined using standard ecotoxicological test organisms (protozoa, crustaceans, algae, bacteria) and yeasts (Table 4). All these toxicity tests were performed in the same media that were used for Cu speciation and bioavailability analysis in the current study. The collected EC₅₀ values of these organisms towards CuSO₄ differed by 4 orders of magnitude (from 0.02 to 368 mg/L, Table 4). Using data from Cu-ISE, we calculated the amount of free Cu ions at each of the EC₅₀ values. When we normalized the CuSO₄ EC₅₀ values for free Cu in the test media, the difference in EC₅₀ values comprised just 1.8 orders of magnitude. Thus, in case of CuSO₄, the free ion concentration may be considered as relatively suitable parameter for toxicity prediction, as also earlier demonstrated by Apte et al. [44]. To our surprise, a good agreement between the EC₅₀ values and free Cu content was observed in our experiments, even though they were conducted with different organisms of substantially different biological complexity (ranging from bacteria to algae and crustaceans). This brings us to the conclusion that the apparently big differences observed in toxicity of CuSO₄ towards these organisms could be attributed to the properties of the testing media rather than to inherent differences in sensitivity between the test organisms. Unfortunately, we were unable to calculate the bioavailable Cu corresponding to respective EC₅₀ values using the Cu-biosensor bacteria as all the EC₅₀ values for CuSO₄ exhibited already strong toxic effects towards the Cu-biosensor (Figure 1). However, as the working principle of biosensor bacteria is to respond to subtoxic amounts of heavy metals, high toxicity of EC₅₀ concentrations of Cu compounds to sensor bacteria was anticipated.