Effects of Humic Acids on the Ecotoxicity of Fe3O4 Nanoparticles and Fe-Ions: Impact of Oxidation and Aging

The magnetite nanoparticles (MNPs) are increasingly produced and studied for various environmental applications, yet the information on their ecotoxicity is scarce. We evaluated the ecotoxicity of MNPs (~7 nm) before and after the addition of humic acids (HAs). White mustard Sinapis alba and unicellular ciliates Paramecium caudatum were used as test species. The MNPs were modified by HAs and oxidized/aged under mild and harsh conditions. Bare MNPs proved not toxic to plants (96 h EC50 > 3300 mg/L) but the addition of HAs and mild oxidation increased their inhibitory effect, especially after harsh oxidation (96 h EC50 = 330 mg/L). Nevertheless, all these formulations could be ranked as ‘not harmful’ to S. alba (i.e., 96 h EC50 > 100 mg/L). The same tendency was observed for ciliates, but the respective EC50 values ranged from ‘harmful’ (24 h EC50 = 10–100 mg/L) to ‘very toxic’ (24 h EC50 < 1 mg/L). The ecotoxicity of Fe-ions with and without the addition of HAs was evaluated in parallel: Fe (II) and Fe (III) ions were toxic to S. alba (96 h EC50 = 35 and 60 mg/L, respectively) and even more toxic to ciliates (24 h EC50 = 1 and 3 mg/L, respectively). Addition of the HAs to Fe-ions yielded the respective complexes not harmful to plants (96h EC50 > 100 mg/L) but toxic to ciliates (24 h EC50 = 10–100 mg/L). These findings will be helpful for the understanding of the environmental fate and toxicity of iron-based NPs.


Introduction
Magnetite (Fe 3 O 4 ) is the most magnetic, naturally existing mineral. Along with the development of nanotechnologies, nanosized magnetite has increasingly been produced and studied for applications in various fields from medicine [1] to environmental remediation [2]. For example, Fe 3 O 4 nanoparticles (NPs), due to their small size, high surface-area-to-volume ratio, the possibility for surface modification and excellent magnetic proper, ties have potential in wastewater treatment as magnetically removable incubated with native humic substances became more oxidized as compared with the control, i.e., bare magnetite.
Dissolution (shedding of metal ions) as a toxicity mechanism of metal-containing NPs has been extensively studied, especially for some types of NPs, such as nanosilver, ZnO NPs and CuO NPs [28]. For example, Navarro et al. [29] evaluated the toxicity of Ag + and AgNPs towards algae Chlamydomonas reinhardtii and showed that toxicity of AgNPs to algae was due to shed Ag-ions and particles contributed as a 'carrier' and a source of the Ag-ions. Heinlaan et al. [30] showed that the toxic effects of ZnO and CuO NPs to crustaceans Daphnia magna and Thamnocephalus platyurus and bacteria Vibrio fischeri were due to solubilized Zn 2+ and Cu 2+ . The same conclusion was reached for algae Pseudokirchneriella subcapitata exposed to ZnO and CuO NPs by Aruoja et al. [31]. Also, Franklin et al. [32] observed that Zn 2+ ions released from the ZnO NPs were highly toxic to algae P. subcapitata.
A meta-analysis of the scientific literature made by Bondarenko et al. [33] covered the (eco)toxicological data for Ag, CuO, and ZnO NPs and respective soluble salts on for algae, crustaceans, fish, bacteria, yeast, nematodes, protozoa and mammalian cell lines. The analysis showed that as a rule, crustaceans, algae and fish proved most sensitive to the studied NPs and at least for AgNPs and ZnO NPs the toxicity was fully explained by solubilized Ag-and Zn-ions, respectively. Analogously, Notter et al. [34] made a meta-analysis of ecotoxicity data for the same types of NPs and the respective metal ions showing that, as a rule, NPs proved less toxic than respective dissolved metal ions. However, to our best knowledge, the information on ecotoxicity of Fe 3 O 4 NPs [35] and especially on the contribution of shed Fe-ions to the overall toxicity and effect of humic substances as toxicity modulators [36] is still limited. The current paper aims to fill that gap.

Synthesis of Fe 3 O 4 and Fe 3 O 4 /HA Nanoparticles
The bare and humic acid (HA)-coated magnetite nanoparticles (Fe 3 O 4 and Fe 3 O 4 /HA NPs) were synthesized with methods described in [37]. Briefly, a commercial sodium salt of humic acids (HA) (Powhumus, the total acidity of the HA was 5.3 mmol/g of acidic COOH and OH-groups, weight-average molecular weight Mw was 9.9 kD; Humintech, Grevenbroich, Germany) was used for the preparation of Fe 3 O 4 /HA and other chemicals were from Sigma-Aldrich (Sigma-Aldrich Chemie GmbH, Steinheim, Germany). For the synthesis of NPs, 6.1 g of FeCl 3 ·6H 2 O and 4.2 g of FeCl 2 4H 2 O was dissolved in 100 mL water and heated to 40 • C, then two solutions, 10 mL of ammonium hydroxide (25%), and 0.7 g of HAs were added rapidly and sequentially. The mixture was stirred at 1000 rpm at 40 • C for 10 min under argon atmosphere and then cooled to room temperature. The black precipitate of Fe 3 O 4 /HA NPs was collected by Nd-magnet (0.3 T) and washed to neutral with distilled water (90 • C) and dried under vacuum at 40 • C. The bare Fe 3 O 4 magnetic nanoparticles were synthesized in a similar way, except that no HA was added.

Simulation of Oxidation of Fe 3 O 4 /HA NPs in Mild and Harsh Conditions
Mild oxidation conditions (aging) were simulated by incubation of the stock aqueous suspensions of Fe 3 O 4 /HA in the dark at 5 • C for 90 days. Harsh oxidation conditions were simulated by mechanical dispersion of the Fe 3 O 4 /HA samples in planetary ball mill where the dispersion process takes place between the grinding balls sliding on each other and between the vessel sides and the grinding beads. The Fe 3 O 4 /HA powder was placed in a wolfram carbide cell with wolfram carbide balls (ball-to-sample mass ratios was 7:1) and dispersed in a high-energy ball mill (SPEX SamplePrep 8000 Mixer/Mill, Metuchen, The Netherlands) at 1425 rpm for 10 min. For the preparation of Fe-HA complexes, a stock solution of the HA (50 mg/L) was prepared by dissolving powdered HA in deionized water with drop-wise addition of 0.1 M NaOH till a final pH = 8. Then, Fe(II) and/or Fe(III) chloride stock solutions (FeCl 2 ·4H 2 O, FeCl 3 ·6H 2 O, 1000 mg/L) were added to HA stock (pH 8.0), to obtain the final ratio HA:Fe(II)/Fe(III) in each series of 1:0.15 according to [38].
Altogether nine different preparations were used for biotesting: aqueous suspensions of Fe 3 O 4 , Fe 3 O 4 /HA, HA, soluble salts of Fe 3+ , Fe 2+ , Fe 3+ /Fe 2+ , complexes Fe(II)HA, Fe(III)HA, Fe(II, III)HA. The pH of the suspensions to be tested was adjusted to 7 by diluted HCl and NaOH.

Characterisation of the Microstructure of Magnetic NPs (MNPs)
The phase composition and primary particle size of the Fe 3 O 4, Fe 3 O 4 /HA and Fe 3 O 4 /HAox (after oxidation) were determined by X-ray diffraction analysis (XRD) in the Bragg-Brentano geometry using a Philips X-pert diffractometer (Philips Analytical, Eindhoven, The Netherlands, Cr-Kα radiation, λ = 2.29106 Å). The full width at a half maximum (FWHM) of all reflections was used for particle size determination with the Scherrer equation. In order to quantify the oxidation progress, the (440) reflection was fitted with five different functions in Origin 2019 Pro (OriginLab Corporation, Northampton, MA, USA).
The lattice parameters determined for all samples formulated in this study are smaller than those reported for magnetite 8.396-8.400 Å (ICDD-PDF 19-629), but larger than those for maghemite 8.33-8.34 Å (ICDD-PDF 39-1346). A plausible explanation of this phenomenon could be the process of partial oxidation of Fe 2+ during drying and/or storage and modification resulting in the non-stoichiometric Fe 3-δ O 4 formation where δ can range from zero (stoichiometric magnetite) to 1/3 (completely oxidized) [39]. For magnetite with an ideal Fe 2+ content (assuming the Fe 3 O 4 formula), the mineral phase is known as stoichiometric magnetite (x = 0.50). As magnetite becomes oxidized, the Fe 2+ /Fe 3+ ratio (formula 4) decreases (x < 0.50), with this form denoted as nonstoichiometric or partially oxidized magnetite [39]. The stoichiometry can easily be converted to the following relationship (1):

Characterisation of the Magnetic Properties of Magnetite NPs (MNPs)
Magnetic properties of MNPs dry powders were characterised with Vibrating Sample Magnetometer Lake Shore (Lake Shore Cryotronics, Westerville, OH, USA) at 300 K.

Analysis of the Surface Charge and Hydrodynamic Diameter of Magnetic NPs (MNPs)
Dynamic light scattering (DLS) measurements were conducted with a "Zetasizer 2c" and "Autosizer 2c" equipment (Malvern Panalytical Ltd., Malvern, UK) at 633 nm with a solid-state He-Ne laser at a scattering angle of 173 • at 25 • C. For the DLS analysis, each sample was diluted to approximately 0.1 g/L. Prior to the analysis the magnetic NP suspensions were ultrasonicated for 10 s followed by 100 s of standstill. The average values of the hydrodynamic diameter of NPs were calculated from third-order cumulant fits of the correlation functions using Correlator K7032-09 (Malvern Panalytical Ltd., Malvern, UK). The range of pH was~7. The experiments were performed at constant ionic strengths 0.01 M set by NaCl.

Ecotoxicity Testing of Magnetic NPs (MNPs) and Fe-Ions
The toxicity of aqueous suspensions of magnetic NPs towards ciliates was analysed in the concentration range from 0. 33  The ecotoxicity of magnetic NPs (MNPs) and Fe-ions to ciliates were determined using Paramecium caudatum Ehrenberg acute toxicity test performed following the protocol described in [40]. Briefly, the assay is based on the measurement of the mortality of Paramecium caudatum when exposed to toxic substances compared with the control. The assay was performed in the 96-well polystyrene plates; well's size 1 mL; Eppendorf). Stock cultures of P. caudatum were maintained in the mineral Lozin-Lozinskiy nutrient medium of the following composition, mg/L: NaCl-100.0, KCl-10.0, CaCl 2 ·2H 2 O-10.0, MgCl 2 ·6H 2 O-10.0, NaHCO 3 -20.0 (Sigma-Aldrich Chemie GmbH, Steinheim, Germany). The stock cultures were maintained at room temperature (22 ± 2 • C), pH 7.5-8.0, without any organic compounds added. To start the test culture, about 1/3 of the stock culture was transferred into a Petri dish containing fresh nutrient medium and incubated at 22 ± 2 • C in the dark for 24 h.
Using a stereoscopic microscope (Model MC-1, Micromed, Shanghai, China), 10-15 ciliates were transferred using a capillary pipette into each of 3-4 test wells containing fresh incubation medium. The volume of liquid when transferring ciliates into the wells did not exceed 0.02 mL. In general, each series of wells (control and test wells) contained at least 30 ciliates. 0.6 mL of incubation medium was added to the control wells; 0.6 mL of the test sample was added to the test wells. The plates with samples and ciliates were incubated at 22 ± 2 • C in the dark. During the exposure period, no food or any other supplements were added. After 24 h of incubation, the individuals were checked for the viability in each well using a stereoscopic microscope. Freely moving ciliates were considered viable, and immobilized individuals were considered dead. The mean values were calculated and compared with the control values.

Sinapis alba L. Acute Toxicity Test
The toxicity of MNPs and Fe-ions to plants was measured using the white mustard Sinapis alba L. root growth inhibition assay (ISO 18763:2016 [41]) in Phytotoxkit format [42]. Certified, high-quality and commercially available seeds were used for all experiments. Following the Phytotoxkit test format (plate test), the 10 mL of pre-shaken NPs suspension of Fe-ions solutions were poured onto transparent test plates (21 × 15.5 × 0.8 cm covered by a white filter paper, and ten Sinapis alba seeds were placed on the paper. The test plates were closed with a transparent lid and incubated first in a horizontal position at 20 ± 2 • C in darkness for 24 h, and then for 72 h in a vertical position at 24 ± 2 • C and an illumination period of 16 h per day with a light intensity of 4000-7000 lx (light wavelength 400-700 nm, universal white). At the end of the incubation, the length of the main root of the mustard seedlings was measured. The mean values were calculated and compared with the control values. The test was made in three replicates.

Statistical Analysis
The inhibitory effect of tested compounds/dilutions compared to the control was calculated as a percentage (%). From dose-response curves, the EC 50 values (mg/L or mg Fe/L, depending on the compound) were calculated using the probit method (IBM SPSS Statistica 17.0, IBM, Armonk, NY, USA) and expressed as an average value ± standard deviation (SD). ANOVA was used for the analysis of statistically significant variances within and between the test groups. The degree of statistical significance of the results was calculated in the R-studio application. The programs were created in the programming language R (inter-group statistical significance was fixed at p ≤ 0.05).

Microstructure of Magnetite Nanoparticles (MNPs)
The crystalline structure of the synthesized MNPs was identified using XRD analysis ( Figure 1). Nanomaterials 2020, 10, x FOR PEER REVIEW 6 of 19 The crystalline structure of the synthesized MNPs was identified using XRD analysis ( Figure 1). The XRD patterns were similar for all studied NP samples and can be interpreted as a facecentered cubic (fcc) lattice with the parameters of 8.383(2), 8.382(6), 8.365 (5) and 8.250(8) Å for the Fe3O4, Fe3O4/HA, Fe3O4/HA harsh oxidation and Fe3O4/HA mild oxidation samples, respectively (Table 1). Finally, the composition of the crystalline component of the pieces can be assigned as follows: Fe2,94O4, Fe2,93O4 and Fe2,84O4 for the Fe3O4, Fe3O4/HA and Fe3O4/HA harsh oxidation samples, respectively. There was no Fe3O4 in the Fe3O4/HA sample after mild oxidation in distilled water during 90 days of aging. Magnetite was likely completely oxidized to maghemite and/or other iron species (Fe(OH)3, FeOOH, etc.). According to [43], the XRD patterns confirm preservation of the spinel structure during oxidation processes: the content of magnetite decreased from ~82.7% to ~79.2%, ~48.3% and 0% for Fe3O4, Fe3O4/HA and Fe3O4/HA harsh oxidation and Fe3O4/HA mild oxidation, respectively ( Figure 2).  The XRD patterns were similar for all studied NP samples and can be interpreted as a face-centered cubic (fcc) lattice with the parameters of 8.383 (2) 3 , FeOOH, etc.). According to [43], the XRD patterns confirm preservation of the spinel structure during oxidation processes: the content of magnetite decreased from~82.7% to~79.2%,~48.3% and 0% for  (Table 1) can be associated with phenol and quinoid units in the HA structure [44]. Notably, the Fe 3 O 4 content in the Fe 3 O 4 /HA harsh oxidation sample was almost halved, assumingly due to the oxidation of magnetite NPs during mechanical treatment (harsh oxidation). The latter shows that HAs were not forming a strong protective shell to the core interacting mainly via Coulombic attraction and hydrogen bonds. Thus, the magnetite gradually oxidized, producing the increasing maghemite shell after each stage of treatment, i.e., first during the treatment with HA and then during the oxidation in harsh conditions in the high-energy ball mill. Magnetite nanoparticles are seemingly very sensitive to oxygen, and in the presence of air, may undergo oxidation to Fe(OH) 3 [45], or to maghemite (γ-Fe 2 O 3 ) phase. Small amounts of O 2 in water could easily oxidize the Fe 2+ species to Fe 3+, becoming a favorable environment for the production of Fe(OH) 3 or γ-Fe 2 O 3 . Depending on pH of the aqueous solution containing Fe 3+ ions, goethite (α-FeOOH) may be formed due to hydrolysis [46]. Full oxidation of magnetite to maghemite was observed in the case of 90 days of aging in mild condition in the presence of dissolved oxygen. This is coherent with the Tombacz et al. [47], where solid phase transformation of magnetite to maghemite and formation of akageneite shell on the magnetic core after storage at 4 • C for 6 in aqueous medium was observed by XRD analysis.
The step-wise oxidation of magnetite is schematically depicted in Figure 2. A slight decrease in the magnetite content in the samples containing humic acids (Fe3O4/HA) from ~82.7 % to ~79.2 % (Table 1) can be associated with phenol and quinoid units in the HA structure [44]. Notably, the Fe3O4 content in the Fe3O4/HA harsh oxidation sample was almost halved, assumingly due to the oxidation of magnetite NPs during mechanical treatment (harsh oxidation). The latter shows that HAs were not forming a strong protective shell to the core interacting mainly via Coulombic attraction and hydrogen bonds. Thus, the magnetite gradually oxidized, producing the increasing maghemite shell after each stage of treatment, i.e., first during the treatment with HA and then during the oxidation in harsh conditions in the high-energy ball mill. Magnetite nanoparticles are seemingly very sensitive to oxygen, and in the presence of air, may undergo oxidation to Fe(OH)3 [45], or to maghemite (γ-Fe2O3) phase. Small amounts of O2 in water could easily oxidize the Fe 2+ species to Fe 3+, becoming a favorable environment for the production of Fe(OH)3 or γ-Fe2O3. Depending on pH of the aqueous solution containing Fe 3+ ions, goethite (α-FeOOH) may be formed due to hydrolysis [46]. Full oxidation of magnetite to maghemite was observed in the case of 90 days of aging in mild condition in the presence of dissolved oxygen. This is coherent with the Tombacz et al. [47], where solid phase transformation of magnetite to maghemite and formation of akageneite shell on the magnetic core after storage at 4 °C for 6 in aqueous medium was observed by XRD analysis.
The step-wise oxidation of magnetite is schematically depicted in Figure 2. The coherent-scattering region size was derived from powder XRD data by Scherrerʹs method. The full width at half maximum (FWHM) of the reflections was used for particle size determination. In order to quantify oxidation, progress the (440) reflection was fitted with Pseudo-Voight function for Fe3O4 and Fe3O4/HA and Voight function for Fe3O4/HAOX in Origin 2019 Pro. While the spherical particle shape remained constant for all modification routes, a slight particle growth during modification with HA can be observed ( Table 1). The diameter of the bare magnetite particles according to XRD analysis was 6.9 nm, i.e., in the size range of superparamagnetic iron oxide NPs (SPIONs) with a high saturation magnetization and a high specific surface area [48,49]. The diameter for Fe3O4/HA particles was 10.3 nm and that of the Fe3O4/HA harsh and mild oxidation particles 7.8 The coherent-scattering region size was derived from powder XRD data by Scherrer's method. The full width at half maximum (FWHM) of the reflections was used for particle size determination. In order to quantify oxidation, progress the (440) reflection was fitted with Pseudo-Voight function for particle shape remained constant for all modification routes, a slight particle growth during modification with HA can be observed ( Table 1). The diameter of the bare magnetite particles according to XRD analysis was 6.9 nm, i.e., in the size range of superparamagnetic iron oxide NPs (SPIONs) with a high saturation magnetization and a high specific surface area [48,49]. The diameter for Fe 3 O 4 /HA particles was 10.3 nm and that of the Fe 3 O 4 /HA harsh and mild oxidation particles 7.8 and 11.03 nm, respectively. All MNPs were polydisperse [50]. According to the coefficient of variation CV and standard deviation value σ of samples, Fe 3 O 4 /HA had a smaller size distribution (12.6%, 1.3) than Fe 3 O 4 (34%, 2.4) , Fe 3 O 4 /HA harsh (24.6%, 1.9) and mild (50.2%, 5.1) oxidations (Table 1). Therefore, considering the values of CV and σ it could be concluded that in the oxidation process, the primary size for the NPs practically remained unchanged.

Evaluation of the Magnetic Properties of the Studied MNPs
The most crucial property of magnetite NPs allowing a variety of applications is their ferrimagnetism. Some magnetic characteristics for the Fe 3 O 4 MNPs are presented in Table 2. The hysteresis loops are closed and symmetrical versus origin of the coordinate system ( Figure S1 in the Supplementary Materials). The shape of the loops evidenced the ferromagnetic character of the material desirable for their application in separation. The respective saturation magnetizations of bare Fe 3 O 4 and Fe 3 O 4 /HA were 68.2 and 30.9 emu/g, respectively, suggesting the content of HA in Fe 3 O 4 /HA about 40% (w/w). The saturation magnetization for samples of MNPs indicates that magnetite nanoparticles stabilized with humic acids exhibited superparamagnetic properties at room temperature ( Figure S1 in Supplementary Materials). The reduced saturation magnetization for Fe 3 O 4 /HA to 30.9 emu g −1 compared to bare magnetite can be explained by a disordered spin canted structure near the surface of NPs. Our harsh oxidation experiment revealed decreasing to 15.7 emu/g in saturation magnetization as well as changes in the shape of the magnetisation curve ( Figure S1). This behavior can be attributed to a change of the composition and to the structure defects arising during oxidation, as shown above by XRD studies (Figure 1). In the absence of a magnetic field, all samples showed a similar low residual magnetism around ±4-7 emu g −1 due to magnetic viscosity for superparamagnetic materials [51]. The further increase of the coercivity can be related to an increasing anisotropy by phase transformation to maghemite [52]. The black aqueous suspensions of Fe 3 O 4 /HA nanoparticles were oxidized to brown suspensions after storage in distilled water for 90 days, indicating the HA coating was not able to protect the magnetite from oxidation and to maintain its saturation magnetization. Due to that, the magnetic properties of aged Fe 3 O 4 /HA were not measured.

The Ecotoxicity of Bare Fe 3 O 4 NPs (MNPs) and Humic Acids-Modified MNPs Before and After Oxidation in Mild and Harsh Conditions
The bare MNPs evaluated for the current study's toxic effects were not toxic to plants S. alba in the root length inhibition test: the EC 50 value was not reached even at 3300 mg/L (Table 3). Literature data support the not harmful nature of Fe-oxide NPs. For example, Fe 2 O 3 NPs were not inhibitory in the seed germination test as at 1000 mg/L 63-104% of the Lactuca seeds and 95-100% of the Raphanus seeds germinated [53]. Analogously, the magnetite NPs were not toxic to duckweed Lemna minor growth inhibition assay (EC 50 > 100 mg/L) as well as in Daphnia magna 48 h immobilization assay (EC 50 > 1000 mg/L) [54].  50 is the concentration of a sample reducing the root length or survival of ciliates by 50%; ** highest concentration that was tested; Color code: ≤1 mg/L (red )= very toxic; >1-10 mg/L (orange ) = toxic; >10-100 mg/L (yellow ) = harmful; >100 mg/L (green ) = "not classified/not harmful". EC 50 data not allowing ranking are on white background.
The EC 50 value of bare Fe 2 O 3 NPs to ciliates was not reached at the highest tested concentration (33 mg/L) ( Table 3). According to Aruoja et al. [55], ciliates Tetrahymena thermophila 24 h viability assay yielded an EC 50 value of 26 mg/L and the 72 h EC 50 for algae Pseudokirchneriella subcapita, the toxicity of Fe 2 O 3 NPs in 72 h growth inhibition assay was 1.9 mg/L. In both cases, the toxicity was not due to the solubilisation of NPs (shedding of Fe-ions).
However, the HA-treated MNPs were more inhibitory to S. alba (EC 50~9 00 mg/L) that bare MNPs (EC 50 > 3300 mg/L; Table 3), probably due to the effect of humic acids (EC 50~9 00 mg/L mg/L). The harsh oxidation (but not the mild oxidation) somewhat increased the toxicity of HA-treated MNPs to plants. Despite that, the EC 50 values of all these tested compounds were > 100 mg/L and could be considered not harmful (Table 3, Figure 3B). Nanomaterials 2020, 10, x FOR PEER REVIEW 9 of 19 yielded an EC50 value of 26 mg/L and the 72 h EC50 for algae Pseudokirchneriella subcapita, the toxicity of Fe2O3 NPs in 72 h growth inhibition assay was 1.9 mg/L. In both cases, the toxicity was not due to the solubilisation of NPs (shedding of Fe-ions). However, the HA-treated MNPs were more inhibitory to S. alba (EC50 ~ 900 mg/L) that bare MNPs (EC50 > 3300 mg/L; Table 3), probably due to the effect of humic acids (EC50 ~ 900 mg/L mg/L). The harsh oxidation (but not the mild oxidation) somewhat increased the toxicity of HA-treated MNPs to plants. Despite that, the EC50 values of all these tested compounds were > 100 mg/L and could be considered not harmful (Table 3, Figure 3B).
Concerning the effects of MNPs to ciliates, although there was not a very clear dose-effect relationship, the treatment with HAs and following oxidation increased the toxicity of MNPs ( Figure  3A). Thus, in general, the trends were similar for plants, but the MNPs, especially after the addition of HAs and following oxidation were remarkably more toxic to ciliates (EC50 down to 0.33 mg/L). Thus, the toxicity of iron oxide NPs to ciliates increased in parallel with oxidation of the Fe3O4 following the order: bare Fe3O4 < Fe3O4/HA < Fe3O4/HA mild oxidation < Fe3O4/HA harsh oxidation, to S. alba: bare Fe3O4 < Fe3O4/HA mild oxidation < Fe3O4/HA < Fe3O4/HA harsh oxidation. Summing up, the most toxic compound in both assays were HA-treated MNPs after harsh oxidation (Table 3, Figure 3). The latter may be due to the partial destruction of HAs supramolecule units and released ferrous and ferric ions during harsh mechanical treatment of the Fe3O4/HA as ferrous ions are far more toxic than the inherent toxicity of the iron-containing NPs [34]. The reason for the release of the iron ions from Fe3O4/HA matrix is probably the supramolecular nature of HA that means associates via weak hydrophobic (van der Waals, π-π, CH-π) and hydrogen bonds [56]. As a result of different treatments (includian oxidation), HA can be hydrolyzed, thus leading to the destruction of the protective layer of Fe3O4/HA. Derivatives of HA contribute to the iron ions release from magnetite. Both iron ions, Fe 2 + and Fe 3+ , form strong mixed ligand complexes with HA [57]. Therefore, thermodynamically driven dissolution and subsequent complexation reactions between HA and iron ions can be important for Fe3O4/HA dissolution. The dissolution of Fe3O4/HA was also confirmed by hydrodynamic particle size analysis via DLS showing a decrease in the particle size [58]. The meta-analysis of the ecotoxicity of NPs made by Juganson et al. [35] described Fe-oxide NPs of relatively low toxicity: the toxicity order for seven types of NPs involved in this study was Ag > ZnO > CuO > CeO2 > CNTs > TiO2 > FeOx (Fe2O3, Fe3O4). Notably, FeOx NPs had also the lowest amount of available information on their ecotoxicity compared to other above listed NPs and the Concerning the effects of MNPs to ciliates, although there was not a very clear dose-effect relationship, the treatment with HAs and following oxidation increased the toxicity of MNPs ( Figure 3A). Thus, in general, the trends were similar for plants, but the MNPs, especially after the addition of HAs and following oxidation were remarkably more toxic to ciliates (EC 50 down to 0.33 mg/L). Thus, the toxicity of iron oxide NPs to ciliates increased in parallel with oxidation of the Fe 3 Figure 3). The latter may be due to the partial destruction of HAs supramolecule units and released ferrous and ferric ions during harsh mechanical treatment of the Fe 3 O 4 /HA as ferrous ions are far more toxic than the inherent toxicity of the iron-containing NPs [34]. The reason for the release of the iron ions from Fe 3 O 4 /HA matrix is probably the supramolecular nature of HA that means associates via weak hydrophobic (van der Waals, π-π, CH-π) and hydrogen bonds [56]. As a result of different treatments (includian oxidation), HA can be hydrolyzed, thus leading to the destruction of the protective layer of Fe 3 O 4 /HA. Derivatives of HA contribute to the iron ions release from magnetite. Both iron ions, Fe 2 + and Fe 3+ , form strong mixed ligand complexes with HA [57]. Therefore, thermodynamically driven dissolution and subsequent complexation reactions between HA and iron ions can be important for Fe 3 O 4 /HA dissolution. The dissolution of Fe 3 O 4 /HA was also confirmed by hydrodynamic particle size analysis via DLS showing a decrease in the particle size [58].
The meta-analysis of the ecotoxicity of NPs made by Juganson et al. [35] described Fe-oxide NPs of relatively low toxicity: the toxicity order for seven types of NPs involved in this study was Ag > ZnO > CuO > CeO 2 > CNTs > TiO 2 > FeO x (Fe 2 O 3 , Fe 3 O 4 ). Notably, FeO x NPs had also the lowest amount of available information on their ecotoxicity compared to other above listed NPs and the information on ecotoxicity of FeOx particles started to emerge in 2009, i.e., later than for the other selected NPs and by the publishing of the paper (2015) there was no data about possible mechanism of action of FeOx NPs.

Structure-Bioactivity Relationship for MNPs
Correlating physicochemical properties of studied NPs (primary size, hydrodynamic size, z-potential, percentage of magnetite) and ecotoxicity of MNPs (Table 3, Figure 4) it can be concluded that humic acids can change the toxicity of MNPs in different ways. Importantly, after each step of treatment, whether it be humic acids, harsh oxidation, or 90 days of aging, the percentage of magnetite in MNPs after treatment with HA decreased from 83% to 79% and further to 48% after the harsh oxidation of Fe 3 O 4 /HA NPs. Interestingly, after mild oxidation of Fe 3 O 4 /HA the percentage of magnetite fell practically to zero ( Table 3). The zeta potentials and hydrodynamic diameters of the above described MNPs also changed reflecting the degree of Fe 3 O 4 NPs surface modification by humic acids and hydrolysis of the humic shell during boxidationsn in harsh and mild conditions (Table 3).

The Ecotoxicity of Fe-Ions before and after Addition of Humic Acids
It is still unclear whether the toxicity of iron oxide NPs is specifically related to nanoparticle properties (such as nano-size that also translates into big relative surface area etc.) or is due to the 2+ 3+

The Ecotoxicity of Fe-Ions before and after Addition of Humic Acids
It is still unclear whether the toxicity of iron oxide NPs is specifically related to nanoparticle properties (such as nano-size that also translates into big relative surface area etc.) or is due to the toxicity of shed Fe 2+ or Fe 3+ ions, or, both Juganson et al. [35]. Wang et al. [59] studied the toxicity of Fe 2 O 3 NPs to bioluminescent bacteria Photobacterium phosphoreum and observed EC 50 value of 200 mg/L (a moderate toxic effect) and showed that the toxic effect was caused solely by NPs and not by shed ions as the NPs were practically nonsoluble. Fe 3+ ions, however, were very toxic to P. phosphoreum, EC 50 about 0.03 mg Fe/L.
As in the environment the NPs become in contact with humic substances that may affect their behavior and properties, we evaluated the toxicity of Fe 3 O 4 -NPs and iron ions (Fe 2+ and Fe 3+ ) in the presence of humic acids (HA), a natural organic polymer that easily forms complexes with Fe 2+ and Fe 3+ and also binds to NPs surface. As test species, plants S. alba and ciliates P. caudatum were used. The Fe(II, III)-humic complexes and Fe 3 O 4 -HA NPs were synthesized as described in Materials and Methods. Test organisms were exposed to: control-culture medium without Fe and HA; samples-Fe(II) or/and Fe(III), HA, Fe(II, III)HA, Fe(II)HA, Fe(III)HA, Fe(II, III)HA.
The current study shows that the EC 50 values of the Fe-ions to ciliates ranged from 1-3 mg/L whereas Fe 3+ ions were slightly less toxic than Fe 2+ ions ( Figure 5A) and to S. alba from 35-60 mg/L and again, Fe 3+ ions were slightly less toxic than Fe 2+ ions ( Figure 5B, Table 4). Lower Fe 3+ toxicity to plants may be due to their lower bioavailability [61] as Fe 2+ is more soluble and can be more easily absorbed by plants [62]. In general, Fe 2+ and Fe 3+ ions were about tenfold more toxic to ciliates than to plants S. alba. The mixture of Fe 2+ /Fe 3+ showed different type of dose-response compared to Fe 2+ and Fe 3+ ions separately when exposed to ciliates ( Figure 5A), but not in case of S. alba. The difference in dose-response may be a result of contribution of several species of ferrous and ferric ions (aqua-ions Fe 2+ , Fe 3+ and hydroxocomplexes FeOH + , Fe(OH) 2+ ) due to their hydrolysis in biotesting environment used for ciliates. Nanomaterials 2020, 10, x FOR PEER REVIEW 12 of 19

Effects of Humic Acids on the Ecotoxicity of Fe 2+ and Fe 3+
The addition of humic acids to iron ions reduced their toxicity to S. alba about 10-fold and EC50 values of Fe-HA complexes proved higher than 100 mg/L, i.e., the complexes could be considered not harmful. The same trend was observed in case of toxicity to ciliates: the addition of HA decreased the toxicity of Fe-ions about 10-fold. However, Fe 3+ /HA still remained toxic (EC50 about 25 mg/L) ( Figure  5, Table 4). The HA showed also some inhibitory effect to both, ciliates and plants but in general their EC50 values to plants were not exceeding 100 mg/L ( Figure 5B) and most probably the same would have been true for ciliates although so high concentration was not tested ( Figure 5A).
The addition of humic acids to binary iron solutions lead to a significant reduction in toxicity (EC50 changed from 0.5 to > 100 mg Fe/L) probably due to the formation of chelate-type ternary iron ions humic acids complexes. Mitigating effect of HA in Fe-ions and HA complexes could be explained by the fact that Fe(II) and Fe(III) are strongly bound to acidic functional groups of HAs [63][64][65][66][67]. It has been reported that complexes of Fe 2+ and Fe 3+ with humic acids increased Fe-uptake in plants Garcia [38]. These results are coherent with the fact that after the addition of humic acids Fe 2+ and Fe 3+ became less toxic to plants (Table 4). Thermodynamic constants for 1:1 complexes of Fe(III) with weak binding sites in humic substances like carboxylic groups and strong binding sites like phenolic groups are four to 40

Effects of Humic Acids on the Ecotoxicity of Fe 2+ and Fe 3+
The addition of humic acids to iron ions reduced their toxicity to S. alba about 10-fold and EC 50 values of Fe-HA complexes proved higher than 100 mg/L, i.e., the complexes could be considered not harmful. The same trend was observed in case of toxicity to ciliates: the addition of HA decreased the toxicity of Fe-ions about 10-fold. However, Fe 3+ /HA still remained toxic (EC 50 about 25 mg/L) ( Figure 5, Table 4). The HA showed also some inhibitory effect to both, ciliates and plants but in general their EC 50 values to plants were not exceeding 100 mg/L ( Figure 5B) and most probably the same would have been true for ciliates although so high concentration was not tested ( Figure 5A).
The addition of humic acids to binary iron solutions lead to a significant reduction in toxicity (EC 50 changed from 0.5 to > 100 mg Fe/L) probably due to the formation of chelate-type ternary iron ions humic acids complexes.
Mitigating effect of HA in Fe-ions and HA complexes could be explained by the fact that Fe(II) and Fe(III) are strongly bound to acidic functional groups of HAs [63][64][65][66][67]. It has been reported that complexes of Fe 2+ and Fe 3+ with humic acids increased Fe-uptake in plants Garcia [38]. These results are coherent with the fact that after the addition of humic acids Fe 2+ and Fe 3+ became less toxic to plants (Table 4). Thermodynamic constants for 1:1 complexes of Fe(III) with weak binding sites in humic substances like carboxylic groups and strong binding sites like phenolic groups are four to 40 times greater than the respective stability constants for Fe(II) complexes [68]. That could be a reason for more pronounced mitigating effect of HA in case of Fe 3+ (Table 4).
3.6. Acute Toxicity of Fe 2+ /Fe 3+ with and without Addition of Humic Acids: Effect of Aging The influence of the aging on the toxicity of HA, Fe 2+ /Fe 3+ and Fe(II,III)HA was studied using S. alba and P. caudatum as test species (Table 5, Figure 6). To simulate the aging in mild conditions, the preparations were incubated in water in the dark at 5 • C during 90 days.
In case of ciliates P. caudatum ( Figure 6A), there were no statistically significant differences in toxicity of as-prepared and aged Fe 2+ /Fe 3+ : both preparations were 'very toxic' (EC 50 < 1 mg/L). However, the aging had pronounced effect on toxicity to ciliates in case of humic acids as well as Fe-ions modified with HAs. In both cases the toxicity remarkably increased after aging (Table 5). Interestingly, the toxicity of aged Fe(II,III)HA (EC 50 = 1.4 mg Fe/L) was higher than the toxicity of HA (EC 50 = 1.93 mg/L). The increase of toxicity can be explained by the formulation of hydroxocomplexes of iron ions with humic acids, as well as hydrolysis of humic acids with decomposition into low molecular fractions and fulvic acids, that may affect P. caudatum.

Conclusions
In the current study, the magnetite NPs (MNPs) were studied for ecotoxicity (ciliates and plants were used as test species) before and after addition of humic acids, to obtain new scientific information on toxicity of magnetite NPs in environmentally relevant conditions. As Fe-ions maybe shed from MNPs and these ions are far more toxic than MNPs, their toxic effects were studied in parallel. In general, MNPs and Fe-ions were remarkably more inhibitory to ciliates Paramecium caudatum than to plants Sinapis alba. The MNPs (with or without HA) had no remarkable inhibitory effects to plants S. alba even after harsh oxidation (96 h EC 50~3 00 mg/L). However, Fe-ions were toxic to S. alba (EC 50 25-60 mg/L) but the toxic effect was mitigated after addition of HAs.
To ciliates MNPs were toxic whereas the complexing with HAs and aging increased the toxic effects down to 24 h EC 50 = 0.33 mg/L after harsh oxidation. The toxicity of Fe-ions to ciliates (EC 50 1-3 mg/L) was reduced after addition of HAs about 10-fold but the complexes still remained toxic to ciliates. Thus, the addition of HAs and aging increased the toxicity of MNPs and mitigated the toxic effect of Fe-ions.
The changes in toxicity were accompanied with the changes in physicochemical properties of MNPs: after each step of treatment as by humic acids, harsh oxidation or 90-days aging the percentage of stoichiometric magnetite decreased from 83 to 79%, to 48%, and to practically to zero, respectively. The zeta potential and hydrodynamic diameter of the above described MNPs also changed reflecting the degree of Fe 3 O 4 NPs surface modification by humic acids and hydrolysis of humic shell during both oxidation in harsh and mild conditions. Supplementary Materials: The following are available online at http://www.mdpi.com/2079-4991/10/10/2011/s1, Figure S1. Hysteresis loops at 300 K for Fe 3 O 4 , Fe 3 O 4 /HA and Fe 3 O 4 /HAox; Table S1. Exposure concentration (mg Fe/L) and concentration for soluble Fe-salt and studied MNPs (mg/L); Table S2. p-Values calculated with R using ANOVA for the toxicity of MNPs and Fe-ions to P. caudatum and S. alba; Table S3. p-Values for the toxicity (EC 50 ) of humic acids, as prepared and aged Fe 2+ /Fe 3+ ions to P. caudatum and S. alba.
Author Contributions: Investigation, software, formal analysis, L.B.; data curation, methodology, writingoriginal draft preparation, writing-ecotoxicity data interpretation, review and editing, A.K.; resources, investigation of samples microstructure using XRD, magnetic analysis, writing-review and editing, G.D.; experimental studies using biotesting, P.U.; conceptualization, general leadership of the work, data analysis, writing-original draft preparation, writing-review and editing K.K. and V.T. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the Russian Foundation for Basic Research №19-33-90149. Anne Kahru acknowledges the financial support from the European Regional Development Fund project TK134. Acknowledgments: Gulzhian Dzhardimalieva performed this study in accordance with the state tasks, state registration AAAA-A19-119032690060-9. Vera Terekhova acknowledges the Moscow State University for supporting this work with a grant for Leading Scientific Schools «Depository of the Living Systems» within the framework of the MSU Development Program.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

HA
Humic acids HS Humic substances NPs Nanoparticles MNPs Magnetite nanoparticles SPIONs Superparamagnet iron oxide nanoparticles XRD X-ray diffraction analysis DOM Dissolved organic matter nZVI Nano zerovalent iron NOM Natural organic matter DLS Dynamic light scattering