Ag Nanoparticles (Ag NM300K) in the Terrestrial Environment: Effects at Population and Cellular Level in Folsomia candida (Collembola)

The effects of nanomaterials have been primarily assessed based on standard ecotoxicity guidelines. However, by adapting alternative measures the information gained could be enhanced considerably, e.g., studies should focus on more mechanistic approaches. Here, the environmental risk posed by the presence of silver nanoparticles (Ag NM300K) in soil was investigated, anchoring population and cellular level effects, i.e., survival, reproduction (28 days) and oxidative stress markers (0, 2, 4, 6, 10 days). The standard species Folsomia candida was used. Measured markers included catalase (CAT), glutathione reductase (GR), glutathione S-transferase (GST), total glutathione (TG), metallothionein (MT) and lipid peroxidation (LPO). Results showed that AgNO3 was more toxic than AgNPs at the population level: reproduction EC20 and EC50 was ca. 2 and 4 times lower, respectively. At the cellular level Correspondence Analysis showed a clear separation between AgNO3 and AgNP throughout time. Results showed differences in the mechanisms, indicating a combined effect of released Ag+ (MT and GST) and of AgNPs (CAT, GR, TG, LPO). Hence, clear advantages from mechanistic approaches are shown, but also that time is of importance when measuring such responses.


Test Materials
Test materials included Ag salt and Ag nanomaterial. The AgNO3 (high-grade, 98.5%-99.9% purity) was purchased from Sigma-Aldrich (St. Louis, MO, USA). The silver nanoparticles (AgNPs) used were the standard reference materials Ag NM300K from the European Commission Joint Research Centre (JRC), fully characterized [20]. The Ag NM300K is dispersed in 4% polyoxyethylene glycerol triolaete and polyoxyethylene (20) sorbitan monolaurate (Tween 20), thus the dispersant was also tested alone.
Ag was spiked as aqueous solution and serially diluted. The soil was pre-moistened before spiking, to obtain a final water holding capacity of 50%, and aged for 72 h before test start. For Ag NM300K, spiking was done individually for each replicate. For AgNO3, the various replicates per treatment were spiked together and then divided into each test vessel as within standard. Concentration range for AgNO3 was: 0, 64, 100, 130, 320, 640 mg Ag/kg soil dry weight (DW) and for AgNP was: 0, 64, 130, 220, 320, 640 mg Ag/kg soil DW. A control dispersant was used adding the same volume as used with the highest concentration of Ag NM300K to assess the effect of the dispersant alone. Test concentration used for the biomarker exposure corresponded to the reproduction EC50 (value selected within the confidence interval). The choice of this EC50 was based on its relevance in Risk Assessment and linkage to reproduction chronic effects. Moreover, the tested concentration should be sub-lethal to ensure organisms' survival for sampling and for mechanistic studies before narcosis (not relevant for biomarkers).

Population Level-Standard Reproduction Test
Tests followed the standard reproduction ISO (International Standardization Organization) test guideline for collembolans [19]. In short, 10 juveniles (10-12 days) were transferred to the test vessels containing the soil. Four replicates were used per treatment. Test ran at 20 °C and 16:8 h (light:dark) photoperiod; food supply and water was replenished every week. Reproduction and adult survival were assessed after 28 days by flotation method to count the number of adults and juveniles.

Cellular Level-Oxidative Stress Biomarkers
Procedures followed the same as in the standard guideline [19] with adaptations [11]. A pool of 50 juveniles of 13-14 days was used as a replicate. Ten (10) replicates (five for MT measurements plus five for the other markers) per treatment were performed. Exposure period included samplings at 0, 2, 4, 6, 10 days. At each sampling time organisms were extracted by flotation, transferred to plaster to absorb the excess water and pooled into microtubes, weighted and snap-frozen in liquid nitrogen, being stored at −80°C until further analysis. Five replicates per condition were used for metallothionein (MT) quantification and the other five for the rest of all biochemical analysis, i.e., catalase (CAT), glutathione reductase (GR), glutathione S-transferase (GST), total glutathione (TG) and lipid peroxidation (LPO). Biomarkers measurements were performed following the procedures as described in Maria et al. [11].

Data analysis
One-way ANOVA and Post Hoc Dunnett's test was used to identify significant differences between control and treatments [21]. The effect concentrations (ECx) were calculated using the Toxicity Relationship Analysis Program (TRAP 1.21) applying the 2-parameters Logistic model. To assess differences between control and control dispersant a t-test (p < 0.05) was used.
Multivariate analysis was done using Correspondence Analysis (CA) including all treatments. The analysis was performed using the software SAS Enterprise Guide 5.1 [22]. To compensate for the different scales of the biomarkers, the response was normalised before use, several different normalisation methods were tested overall giving the same pattern; the present normalisation was based on averaging in relation to the mean.

Materials Characterization
The silver nanoparticles (AgNPs) used were the standard reference materials Ag NM300K from the European Commission Joint Research Centre (JRC), fully characterized [20]. In short, Ag NM300K are spherical and consist of a colloidal dispersion with a nominal silver content of 10.2 w/w %, dispersed in 4% w/w of polyoxyethylene glycerol trioleate and polyoxyethylene (20) sorbitan monolaurate (Tween 20), having > 99% number of particles with a nominal size of about 15 nm, with no coating. Transmisson Electron Microscopy (TEM) indicated a size of 17 ± 8 nm. Smaller nanoparticles of about 5 nm are also present.

Univariate Analysis
For Ag NM300K the control dispersant was used as a reference because for LPO, TG and GR measurements there were differences (p < 0.05) between control and control dispersant ( Figure 2).
For AgNO3, CAT activity decreased after 4 days exposure (p < 0.05) (0.4-fold to control), maintaining a tendency of low values in the remaining exposure time. GR activity shows an increase after 2 and 4 days (p < 0.05) followed by a decrease to levels lower than control at 6 and 10 days exposure. MT shows a similar pattern. GST activity shows an increase-decrease-increase behaviour at 4 (p < 0.05), 6 and 10 (p < 0.05) days respectively and TG increased only at day 4 (p < 0.05). Significant increase in LPO levels was observed at day 4 (1.2-fold, p < 0.05). For Ag NM300K CAT activity was higher after 2, 4 and 10 days (p < 0.05), having a decrease tendency at day 6 (to levels similar to control). GST activity showed an increase (p < 0.05) up to day 4 and then continued on same levels until day 10 (p < 0.05). GR increased only after 4 days (p < 0.05). TG levels were lower than control after 4 days and superior at 10 days. MT levels increased after 6 days (p < 0.05), maintaining the higher level at day 10. LPO increased at day 2 (p < 0.05), after which it decreased to be increased again at day 10 (p < 0.05).

Multivariate Analysis
The multivariate analysis of the data (Correspondence Analysis) enabled an identification of the overall differences between the AgNO3 and AgNP exposures (Figure 3), with a mainly clear separation between the AgNO3 and AgNP throughout time. [It should be noted that whereas Figure 2 shows mean values and standard errors, the multivariate plot displays the individual replicates]. It is seen that LPO and GST were primarily associated with AgNP and MT and TG associated with AgNO3, hence these markers would be the primary identifiers of different exposures. In the later exposure stages (10 days) the GR was most pronounced for the AgNP exposure, when compared to AgNO3 exposure. The larger confidence ellipse (compared to others) related to the AgNO3 at day 6, seem to be related to one replicate having a relative high (again compared to the others) TG.

Population Level
Results showed that AgNO3 displayed higher toxicity than Ag NM300K for Folsomia candida, with increasing difference with higher concentration (EC20 to EC80). For AgNO3, the Effect Concentration (EC) values were within the obtained confidence interval as found by Waalewijn-Kool et al. [8] for F. candida tested under the same conditions. The same authors tested other AgNP (paraffin coated, 3-8 nm, water dispersed) and found no effect on survival or reproduction up to 673 mg Ag/kg soil DW. As concluded by the authors, the internal Ag concentrations for F. candida could not explain the higher toxicity of AgNO3 compared to AgNPs; it has been suggested that the higher internal Ag in F. candida exposed to AgNPs could be because these are taken up on the particulate form. Unlike ZnO NPs [18,23], porewater concentrations could not explain the toxicity of AgNPs. It seems that AgNPs aggregation and sorption to soil parts reduces dissolution. The fate of AgNPs in soil has been reported complex, with e.g., soil type, dissolution (rate), oxidation, nanoparticle size and the type of coating influencing the availability of Ag [8]. For other invertebrates, oligochaete studies has shown that AgNO3 was more toxic than AgNPs [5,6,24,25]. Van der Ploeg et al. [26] observed that low doses of the same Ag NM300K (15 mg Ag/kg soil DW) caused higher effects (for the same mass concentration) than AgNO3 in Lumbricus rubellus longer term reproduction study. Moreover, also focussing on longer term exposures, (Bicho et al., 2015 in preparation) showed that in an Enchytraeus crypticus full life-cycle test 20 mg Ag/kg soil DW of Ag NM300K caused an effect equivalent to the reproduction EC50, although the dose response model estimated an EC50 = 80 mg Ag/kg soil DW.

AgNO3 Mechanisms
Overall, an induction of all measured antioxidant enzymes was observed, with the inhibition of CAT being the exception. Similarly, it has been shown that in C. riparius, exposure to AgNO3 decreases the CAT activity [27]. Also CuCl2 and CuNP have been shown to reduce CAT activity [12], possibly due to direct interaction of Cu with the protein's thiol groups, altering the tertiary structure of the catalase and inhibiting it [28], possibly with a similar mechanism for Ag. On the other hand, CAT has also been reported activated (in other invertebrates) in the presence of AgNO3, e.g., in Eisenia fetida [1,4], and in F. candida when exposed to copper and cadmium [11].
The glutathione-related enzymes, GR and GST present different patterns for activation, GR early and GST later induction. It is known that Ag has a great affinity for thiol groups, besides inducing the production of ROS [29][30][31]. Therefore, the presence of Ag can mobilize the GSH levels in the cell (i.e., binding to this substrate) [32,33], so here it seems that an early activation of GR occurred to compensate the unavailable GSH, i.e., oxidized glutathione. The Ag-GSH detoxification is associated with the GST activation, similar to e.g., the detoxification mechanism of Cd [15], explaining its increase only after 4 days and again after 10 days. Additionally, the initial GR increase followed by a decrease is similar to the response to Cu by F. candida [11]. The GST activity and TG content increase after 4 days may be due to ROS generation, this also related with the LPO levels.
The increase in MT levels must be associated with the Ag chelation. This is in agreement with observations at the gene expression level in E. fetida exposed to AgNO3 [4] and Cu [34], and F. candida exposed to Cd [15]. It is known that Ag can be taken up by Cu transporters and interact with Cu homeostasis, which may contribute to Ag toxic effect [29].
Regarding LPO at day 4, this was similar to the response to Cd in F. candida [11] and Ag in aquatic invertebrates [35]. This could be the result of the imbalance in the redox in the organisms due to CAT reduced activity, as similarly observed to Cu in E. albidus [12]. Such reduced CAT activity leads to accumulation of hydroperoxides, which can be removed via the glutathione cycle enzymes. This is reflected in the initial activation of GR, followed by the increase in GST and TG. When the enzymes activity reach a point of saturation LPO occurs.

AgNP Mechanisms
In contrast to the AgNO3 exposure, CAT activity was significantly increased in the AgNP exposure, except after 6 days, a pattern similarly observed for Cu and Cd in F. candida [11]. The MT induction occurred after 6 and 10 days, i.e., later when compared to AgNO3. It is unknown if for longer exposure periods this would also be followed by a decrease like in AgNO3.
The increase in the glutathiones (higher GST throughout the exposure length, increased GR after 4 days and the increase in TG after 6-10 days), indicate interactions of AgNP with cytosolic and transmembrane proteins, changing the conformation and impairing the antioxidant defenses [36][37][38][39]. Hence, GST levels were continuously high to chelate the radical ligands in thiol groups in glutathione content [4,30,32,33]. The increase in GR was needed to balance the redox potential (GSH recycling), as a result of ROS production from NP interactions [31]. Because NPs can also cause DNA damage, leading to synthesis of nuclear GSH, this may explain the increase in the TG content [33,40,41].

Comparison of Ag Nano and Ag Salt Mechanisms
As discussed so far it is clear that Ag nano and Ag salt cause dissimilar oxidative stress mechanisms of response (see Figure 3). Differences in response patterns for AgNO3 and AgNP have also been described for e.g., the soil invertebrates Eisenia fetida [1,4] and Enchytraeus albidus [5].
The patterns observed in F. candida for GR, TG and MT seem to indicate a delayed effect of AgNP compared to AgNO3 (as shown by some authors [42,43]), suggesting an effect caused by the slower release of Ag or a slower uptake On the other hand, CAT and GST show clearly different patterns, indicating a specific NP effect. As already suggested, AgNPs uptake may be done by different pathways than AgNO3 [29,31,44,45]. There seems to be a combined effect of Ag + and AgNPs which results in a different time of occurrence of events and consequently a different cascade. This is corroborated by the differences caused in terms of LPO, reflecting previous variations in REDOX enzymes. For instance, following the hypothesis of the Ag + release from AgNPs the response of MT, GR and TG could be seen as a delayed response for the AgNP, however this is not the case for CAT, LPO and GST.

Conclusions
Oxidative stress was studied for the first time in F. candida to AgNPs. Reproduction effect concentrations (EC50) caused dissimilar oxidative stress mechanisms, indicating a combined effect of released Ag + (MT and GST) and of AgNPs specifically (CAT, GR, TG, LPO). Ag NM300K were less toxic than AgNO3 in terms of population effects, i.e., survival and reproduction.