Silver Nanoparticles Compromise Neurodevelopment in PC12 Cells: Critical Contributions of Silver Ion, Particle Size, Coating, and Composition

Background Silver exposures are rising because of the increased use of silver nanoparticles (AgNPs) in consumer products. The monovalent silver ion (Ag+) impairs neurodevelopment in PC12 cells and zebrafish. Objectives and methods We compared the effects of AgNPs with Ag+ in PC12 cells for neurodevelopmental end points including cell replication, oxidative stress, cell viability, and differentiation. First, we compared citrate-coated AgNPs (AgNP-Cs) with Ag+, and then we assessed the roles of particle size, coating, and composition by comparing AgNP-C with two different sizes of polyvinylpyrrolidone-coated AgNPs (AgNP-PVPs) or silica nanoparticles. Results In undifferentiated cells, AgNP-C impaired DNA synthesis, but to a lesser extent than an equivalent nominal concentration of Ag+, whereas AgNP-C and Ag+ were equally effective against protein synthesis; there was little or no oxidative stress or loss of viability due to AgNP-C. In contrast, in differentiating cells, AgNP-C evoked robust oxidative stress and impaired differentiation into the acetylcholine phenotype. Although the effects of AgNP-PVP showed similarities to those of AgNP-C, we also found significant differences in potencies and differentiation outcomes that depended both on particle size and coating. None of the effects reflected simple physical attributes of nanoparticles, separate from composition or coating, as equivalent concentrations of silica nanoparticles had no detectable effects. Conclusions AgNP exposure impairs neurodevelopment in PC12 cells. Further, AgNP effects are distinct from those of Ag+ alone and depend on size and coating, indicating that AgNP effects are not due simply to the release of Ag+ into the surrounding environment.


Research
The rapid growth in the commercial use of silver nano particles (AgNPs) is increasing silver exposure in the general population (Wijnhoven et al. 2009). AgNPs are incorporated into prod ucts primarily as an anti microbial, reflecting their release of mono valent silver ion (Ag + ) (Wijnhoven et al. 2009). However, the same mechanisms that make Ag + an anti microbial also render it a potential develop mental neuro toxicant. Silver crosses the placenta and con centrates in the human fetus, achieving higher concentrations than in the mother (Lyon et al. 2002). Animal studies show accumulation in the developing brain, develop mental dys morphology, and behavioral changes in exposed adults (Rungby 1990). Importantly, AgNP exposure via either inhalation or oral routes also leads to Ag accumulation in the adult rodent brain (Wijnhoven et al. 2009), altering the expression of genes involved in neuronal func tion ). We recently showed that in PC12 cells, a wellestablished model of neuronal develop ment, Ag + disrupts key mecha nisms involved in cell replication and neuro differentiation (Powers et al. 2010a); we then demon strated that nervous system develop ment is disrupted in developing zebrafish exposed to Ag + (Powers et al. 2010b). Unlike primary neuronal cultures, PC12 cells provide a homo geneous population that continues to divide until differentiation is triggered by addition of nerve growth factor. Accordingly, this model allows direct study of effects on DNA synthesis associated with cell replication, an important target of neuro toxicants; the cells then differ entiate into distinct acetylcholine (ACh) and dopamine (DA) phenotypes.
It is thus critical to assess the extent to which AgNPs can elicit the same or different types of neuro develop mental outcomes as Ag + . In the same PC12 model, high concen trations of AgNPs disrupt the cell membrane and impair mitochondrial function (Hussain et al. 2006) while altering gene expression related to oxidative stress ); however, these studies were not carried out in the context of neuro differentiation (Powers et al. 2010a). In the present study, we carried out extensive experiments on replicating and differentiating PC12 cells, comparing and contrasting the effects of AgNPs with those of Ag + . We then evaluated the roles of par ticle size, coating, and composition, issues of potential importance in assessing the toxicity of different AgNP formulations (Teeguarden et al. 2007). We chose AgNPs covering a range of particle sizes and coatings, charac terizing both their physi cal properties in sus pension as well as their biological effects. We focused first on compari sons between Ag + and citratecoated AgNPs (AgNPC). This was followed by evaluations of the effects of different coatings and sizes, using polyvinyl pyrrolidinecoated AgNPs (AgNPPVPs), and by assessments using uncoated silica nano particles (SiNPs) to determine whether effects could be elicited simply by particles of the same size, regardless of the main composi tion constituent. Nanoparticle coatings are intended to promote stability and dispens ability through surface polarity that prevents agglomeration, and the two types of coatings chosen here are common to many types of nano particles. Commercially available nano particle formu la tions have different ranges of sizes, which can have profound influence over their biologic activities (Teeguarden et al. 2007). Our evalua tions were modeled after our earlier work on the develop mental neuro toxicity of Ag + (Powers et al. 2010a), focusing on anti mitotic effects, inhibition of protein synthesis, oxidative stress, impaired viability, and neuro differentiation into ACh and DA pheno types. Here, we show that the effects of AgNPs do not reflect solely their ability to release soluble Ag + , but instead are influenced by specific nano particle characteristics that dictate different biologic outcomes.

Materials and Methods
Nanoparticle preparation and charac terization. AgNPC was synthesized at Duke University using established methods (Lee volume 119 | number 1 | January 2011 • Environmental Health Perspectives and Meisel 1982). We purchased AgNPPVP and uncoated SiNPs in powder form from Nanostructured & Amorphous Materials Inc. (Houston, TX). Stock suspensions of nano particle powders were prepared in ultra pure water, sonicated continuously at 89-95 W power, with amplitude set at 100%, for 20 min in an ice bath using a Misonix Sonicator 4000 (QSonica LLC, Newton, CT) equipped with a 1/2 in. diameter flat titanium tip.
Both types of AgNPs were prepared in stock solutions equivalent to a nominal con centration of 1 mM Ag. For AgNPC this concentration was achieved by using 1 mM AgNO 3 to synthesize the particles. For AgNPPVP, we weighed out the appropriate volume of powder corresponding to 1 mM Ag, based on the percent composition from the manufacturer's specifications. SiNP stock suspensions were made up to have a particle concentration equivalent to that used for the 10nm AgNPs, based on AgNP and SiNP par ticle volumes and densities. Table 1 describes the concentration of each stock suspension in terms of the nominal Ag concentration (con centration corresponding to the Ag concentra tion that would be achieved if all the Ag were freely dissolved), the concentration of particles, and the mass of particles per milliliter.
We evaluated dry particle size and mor phology using transmission electron micros copy at 160 kV (Tecnai G2 Twin; FEI Company, Hillsboro, OR) by adding 10 µL of the sample to a lacey carbon/copper grid (300 mesh; Electron Microscopy Sciences, Hatfield, PA) and allowing samples to air dry. Images were analyzed using ImagePro, ver sion 4.5 (Media Cybernetics, Inc., Bethesda, MD). Particle size in suspension was assessed using dynamic light scattering using a CGS 3 goniometer (ALVGmbH, Langen, Germany) equipped with a heliumneon laser (633.4 nm). Suspensions were analyzed at 25°C in 5mM diameter cells with the photo multiplier set to a scattering angle of 90°. The nominal Ag concentration in stock suspen sions was measured using inductively coupled plasmaoptical emission spectroscopy (Prism ICP High Dispersion; Teledyne Leeman Labs, Hudson, NH) and graphite flame atomic absorption (PerkinElmer, Waltham, MA). For PVPcoated particles, we meas ured the poly mer concentration by baking the particles at 540°C for 18 hr in a muffle furnace; the pure silver weight was calculated from the weight of residual silver oxide formed at the end of bak ing. The concentration of any free PVP was assessed using a total organic carbon analyzer (TOC5050A; Shimadzu, Columbia, MD).
Cell cultures and assays. All materials and cell culture and assay techniques used in this study have been reported previously and were specifically used in our earlier study of Ag + effects in the PC12 model (Powers et al. 2010a); therefore, we will provide only a brief procedural outline here. For studies in the undifferentiated state, the medium was changed 24 hr after seeding to include test reagents. For studies in differentiating cells, 24 hr after seeding, the medium was changed to include nerve growth factor, and each cul ture was examined under a microscope to ver ify the subsequent outgrowth of neurites. Test agents were added concurrently with the start of nerve growth factor treatment, and cultures were maintained for up to 6 days, with the indicated agents included with every medium change (48hr intervals).
Cells were harvested and washed, and the DNA, total protein, and membrane protein fractions were isolated and analyzed as described previously (Song et al. 1998). Because neuronal cells contain only a single nucleus (Winick and Noble 1965), measur ing the DNA content in each dish provides a measure of cell number. The protein/DNA ratio was calculated as an index of cell size, and the membrane/total protein ratio was used to assess the rise in membrane complexity that accompanies neurite outgrowth during neuro differentiation. We measured DNA synthe sis by assessing [ 3 H]thymidine incorporation into the DNA fraction (Song et al. 1998); similarly, protein synthesis was assessed by incorporation of [ 3 H]leucine into the pro tein fraction. Oxidative stress was evaluated through measuring the formation of lipid per oxides by reaction of the resultant malondial dehyde (MDA) with thio barbituric acid (Qiao et al. 2005). Cell viability was meas ured by blinded cell counts after trypan blue staining. Differentiation into ACh and DA phenotypes was determined enzymatically by measuring choline acetyltransferase (ChAT) and tyrosine hydroxylase (TH) activities, respectively (Lau et al. 1988;Waymire et al. 1971). We also included samples containing Ag + to serve as a positive test compound for compari son with the effects of the AgNPs; these represent new values, not a restatement of our published results with Ag + (Powers et al. 2010a). The time points for differentiating cells were cho sen based on our prior work with Ag + , show ing progressive loss of DNA content over a span of 4-6 days in culture and changes in neuro transmitter pheno type at the 6day point (Powers et al. 2010a). Loss of viability and oxidative stress produce eventual cell loss, so those measurements were made at 4 days.
We incorporated a number of different controls in our cell culture assays to account for differences specific to each type of nano particle or experi mental condition, and these are described in the figure legends. In our pre vious study (Powers et al. 2010a) we found no effect of nitrate ion on any of these param eters and thus did not include this additional control in the present study. Similarly, we did not include citrate controls because the culture medium already contains citrate in substantial concentrations from the added fetal bovine and horse serum.
Data analysis. All studies were performed on 8-16 separate cultures for each measure and treatment, using 2-4 separate batches of cells. Results are presented as mean ± SE. Treatment effects were established by analysis of variance (ANOVA), followed by Fisher's protected least significant difference test for post hoc compari sons of individual treat ments; data were logtransformed whenever the variance was hetero geneous. In the ini tial test, we evaluated two ANOVA factors (treatment and cell batch) and found that the results did not vary among the different batches of cells, so results across the differ ent batches were normalized and combined for presentation. Significance was assumed at p < 0.05 (twotailed).

Results
Nanoparticle characteristics. The vast majority of nano particles were spherical. AgNPC was poly disperse ( Figure 1A), with an average dry particle size of 6 nm; 85% were < 10 nm, and the remaining 15% were < 63 nm ( Figure 1B). In suspension, particles swelled or aggregated, producing a higher hydro dynamic radius com pared with dry particles. The size remained stable over time ( Figure 1C), indicating either that aggregated particles fell out of suspen sion or that they represented only a small proportion. We examined the particle con centration under culture conditions over the 48hr time period between medium changes, using spectro photometry at 540 and 570 nm to assess absorption by the particles suspended in the cell culture medium, focusing on the highest concentration (100 µM nominal Ag), which would be most likely to aggregate. The suspended nano particle concentration remained unchanged over 48 hr: 0.030 optical density units above culture medium alone at 24 hr, and 0.034 at 48 hr (triplicate samples). Thus, particles in suspension tended to aggre gate somewhat over time but maintained their average size and concentration, indicating that aggregation was not a significant problem. We carried out similar evaluations of AgNPPVP. Dry particles designated to have a 10nm diameter actually averaged 21 nm, with 88% at < 25 nm and the remainder at < 200 nm. The designated 50nm AgNPPVP actually averaged 75 nm, with 57% at < 81 nm and the remainder at < 200 nm. Our analy sis of dry SiNP showed good agreement with the manufacturer's stated description of par ticle size and shape. Similar to AgNPC, both PVPcoated particles and SiNP showed a small degree of aggregation once in suspension, but the effects were of insufficient magnitude to cause major changes in the suspended nano particle concentration ( Figure 1C). We also used a total organic carbon analyzer to measure the PVP concentration in the AgNPPVPs. We found PVP concentrations of 15% and 13% of total AgNP weight for 10 and 50nm particles, respectively, markedly higher than the stated concentrations of 0.2-0.3%. In pre paring our stock concentrations, we used 10% PVP (i.e., between our values and the manufac turer's). The measured values for Ag in the dry AgPVPs were within 15% of those expected.
Throughout the results, we present the nano particle concentration in two differ ent metrics: the nominal Ag concentration (defined as the equivalent of all the Ag being in free solution) and the number of particles per unit volume. The equivalent amount for each nano particle (mass per unit volume) appears in Table 1.

AgNP-C in undifferentiated cells.
We first compared the anti mitotic effects of AgNPC with those of Ag + . With a 24hr exposure, we found a concentrationdependent decrease in DNA synthesis starting at AgNPC cor responding to a nominal Ag concentration of 1 µM, but in all cases the effects were smaller than those seen with 10 µM Ag + (Figure 2A). To determine if binding of AgNPs to serum proteins was responsible for the smaller effect compared with Ag + , we meas ured DNA syn thesis in cells exposed to AgNPC or Ag + with and without serum for 1 hr, a span in which cells maintain their viability in the absence of serum ( Figure 2B). Removing serum from the medium greatly enhanced the effect of Ag + , reflecting a high degree of binding to serum proteins. However, there was no correspond ing enhancement for AgNPC.
To determine if the reduction in DNA synthesis evoked by AgNPC reflected a spe cific action on mitotic activity, we examined corresponding effects of a 24hr exposure on protein synthesis ( Figure 2C). For Ag + , the reduction in protein synthesis was much smaller than that seen for DNA synthesis. In contrast, for AgNPC, protein synthesis was inhibited to about the same extent as had been observed for its effects on DNA synthesis up to a nominal Ag concentration of 10 µM. However, unlike the situation for DNA syn thesis, the effect on protein synthesis was lost at higher concentrations.
In undifferentiated cells, Ag + produced robust oxidative stress after a 24hr exposure, whereas AgNPC was ineffective ( Figure 2D). Likewise, Ag + was much more cyto toxic, evoking a large reduction in cell viability compared with the much smaller effect of AgNPC ( Figure 2E); similar to the effect on protein synthesis, AgNPC above a nominal Ag concentration of 10 µM became less effec tive. Finally, meas ures of cell number after a 24hr exposure (DNA content) confirmed that Ag + evoked a much greater cell loss than did AgNPC ( Figure 2F) and, again, we saw a non monotonic effect of the nano particles.
AgNP-C in differentiating cells. Unlike the situation in undifferentiated cells, AgNPC exceeding a nominal Ag concentration of 3 µM produced significant oxidative stress after 4 days of exposure, in the same range as Ag + ( Figure 3A). Increased oxidative stress was not secondary to general cytotoxicity, as the AgNPC effect on viability remained much smaller than that of Ag + ( Figure 3B); further more, although the dose-effect relationship was mono tonic for oxidative stress, it was non monotonic for loss of viability. At the same 4day exposure, cell number decreased much more at 10 µM Ag + than at comparable or higher concentrations of AgNPC ( Figure 3C); for AgNPC, we observed small but signifi cant decrements at nominal Ag concentra tions of 10 and 100 µM, albeit not at 30 µM. Furthermore, by 6 days of exposure, cell num ber recovered so that there was no detectable loss at any concentration; in fact, there was an increase in cell number at the lowest AgNPC concentration ( Figure 3C). In contrast, expo sure to Ag + simply produced a progressive cell loss beyond that seen at the 4day point.
Ascorbate prevents oxidative stress and cell loss caused by exposure to Ag + (Powers et al. 2010a). Accordingly, we performed comple mentary experiments with AgNPC at a nomi nal Ag concentration of 10 µM. We found the same increase in MDA in the presence of ascor bate (mean ± SE:10 µM, 17 ± 2% increase;   Figure 3A,C). Effects of nano particle coating, size, and composition. In undifferentiated cells, a 24hr exposure to AgNPPVP with manufacturer designated diameters of either 10 or 50 nm at a nominal Ag concentration of 30 µM evoked signifi cant decreases in DNA synthe sis ( Figure 4A). Notably, the decrement for the 50nm AgNPPVP exceeded that caused by the smaller AgNPPVP or by AgNPC. SiNP had a smaller (non significant) effect than any of the AgNPs. Similar measures of protein synthesis showed no discernible effect of either 10 or 50 nm AgNPPVP at nominal Ag concentrations of 10 or 30 µM, whereas 10 µM AgNPC clearly inhibited synthesis ( Figure 4B). SiNPs had no effect. With the same 24hr exposure to a nominal 10 µM Ag concentration, AgNPs reduced cell num ber, with the greatest effect from AgNPPVP 50 nm, followed by AgNPC, and no effect for AgNPPVP 10 nm ( Figure 4C); however, at a 30 µM nominal Ag, the effect of 50 nm AgNPPVP or AgNPC was reduced or lost. Again, SiNPs had no effect.
We next compared the effects of particle size, coating, and composition in differentiat ing cells. With a 4day exposure, we found oxidative stress for all three types of AgNPs, whereas SiNP was ineffective ( Figure 5A). Either size of AgNPPVP decreased cell num ber at 4 days ( Figure 5B); by 6 days, the effect regressed to normal for the smaller AgNPPVP particle but not for the larger particle. AgNPC at the same nominal concentration had no measurable effect on cell number at either time point (Figure 5B, replicating the results seen in Figure 3C). Both sizes of AgNPPVP increased the index of cell size at 6 days: 9 ± 3% (mean ± SE) increase in the total protein/DNA ratio for 10 nm AgNPPVP (p < 0.01; n = 20), 7 ± 2% increase for 50 nm AgNPPVP (p < 0.04; n = 20; control ratio To achieve 100 µM AgNP-C, the culture medium was diluted 10% with AgNP-C stock solution; isotonic NaCl and NaHCO 3 were then added to achieve isotonicity and to match the NaHCO 3 concentration normally in the medium; accordingly, these samples have separate controls with the same additions. HC control, high concentration control. *Significantly different from the corresponding control (p < 0.05 or better) by Fisher's protected least significant difference test.  27.8 ± 0.4 µg/µg) but had no significant effect on the membrane/total protein ratio (data not shown). Samples with corresponding concen trations of AgNPC run concurrently with the AgNPPVP showed no significant differ ence in total protein/DNA. Finally, 10 nm AgNPPVP enhanced differentiation into the DA pheno type, as indicated by a significant increase in TH activity relative to control values, but neither 50 nm AgNPPVP nor AgNPC had a comparable effect ( Figure 5C). In contrast, all three types of AgNPs sup pressed the ACh pheno type, as shown by deficits in ChAT activity, but AgNPC and 50 nm AgNPPVP were more effective than the smallerdiameter AgNPPVP. Although the under lying components differed, all of the AgNPs increased the ratio of TH/ChAT ( Figure 5D), reflecting diversion of cells toward the DA phenotype and away from the ACh phenotype; the net effect was greatest for the largerdiameter AgNPPVP.

Discussion
Our results provide some of the first evidence that AgNPs can act as develop mental neuro toxicants in a model of neuronal cell replica tion and differentiation. They further point to compound effects that depend not only on the release of Ag + but also on particle size, coating, and composition. Indeed, if AgNPs acted solely by releasing soluble Ag + , then all of their effects would resemble those of lower concen trations of the soluble ion, because a large pro portion of the Ag + is not dissolved. In that case, we would expect to see a decline in AgNP effect with increasing particle size because the smaller surfacetovolume ratio of larger par ticles would render less of the Ag + available to dissolve. Some of our findings followed this predicted pattern, but others clearly did not. The effects of AgNPC on undifferentiated cells were in the same direction but decidedly smaller than those from the same concentra tion of freely dissolved Ag + for comparisons of DNA synthesis, oxidative stress, viability, and cell loss. However, this was not true for the role of plasma protein binding on DNA syn thesis, nor for the effects on protein synthesis; for the latter, AgNPC was as effective as Ag + and showed a non monotonic effect that was not seen in our earlier work with Ag + (Powers et al. 2010a). The same pattern of some simi larities, coupled with important dichotomies, was apparent in differentiating cells. AgNPC produced oxidative stress, decreased viability, cell loss, and impaired neurite formation, in each case requiring a higher concentration to produce effects equivalent to those seen from the freely dissolved Ag + . Likewise, in our earlier work with Ag + , we found a biphasic effect on DNA content between 4 and 6 days of expo sure just as seen here for AgNPC, but involv ing a lower Ag + concentration. Nevertheless, AgNPC failed to evoke the cell enlargement (increased total protein/DNA) seen with Ag + ; furthermore, ascorbate did not protect cells from the oxidative stress and cell loss caused by AgNPC, whereas the same treatment protects against Ag + (Powers et al. 2010a). Finally, the experi ments using the two sizes of AgNPPVP showed a relationship opposite what would be expected just from release of Ag + from the particle surface: At the same nominal Ag con centration, the larger nano particle had greater effects on DNA synthesis and content and caused a higher degree of disruption in oxida tive stress and neuro transmitter phenotype. Clearly, then, the neuro toxic actions of AgNPs involve significant contributions from nano particle formulation, albeit not from only the physical dimensions, because SiNPs were generally ineffective in producing the effects seen with the AgNPs. Nanoparticles can produce their unique effects either through altering access to the interior of the cell (pharmaco kinetic effects) or through eliciting responses that differ from those of the freely dissolved materials (pharmaco dynamic effect). Our results for effects on DNA synthesis with  AgNP-C 100 µM 2 × 10 12 HC control HC control 6 × 10 10 2 × 10 11 6 × 10 11 AgNP-C 3 µM 10 µM 30 µM 6 × 10 10 2 × 10 11 6 × 10 11 3 µM 1 µM 10 µM 30 µM 6 × 10 10 2 × 10 11 2 × 10 10 6 × 10 11 AgNP-C 3 µM 1 µM 10 µM 30 µM 6 × 10 10 2 × 10 11 2 × 10 10 6 × 10 11 AgNP-C 3 µM 1 µM 10 µM 30 µM 6 × 10 10 2 × 10 11 2 × 10 10 6 × 10 11 volume 119 | number 1 | January 2011 • Environmental Health Perspectives and without serum effectively eliminate the possibility that a reduced effect of AgNPs results from binding to serum proteins. In fact, we found the opposite, namely, that removal of serum greatly enhanced the effect of Ag + but not that of AgNPC. Similarly, we can rule out the possibility that nano particle aggre gation limits the concentration of Ag + avail able for biologic effects for two reasons. First, we saw little evidence for significant changes in the net nano particle size or concentration over time. Second, aggregation would produce a parallel change in all the measured effects, whereas we saw a mono tonic concentrationresponse curve for some variables but non monotonic effects for others. Instead, our results provide conclusive evi dence for unique biologic effects of AgNPs, distinct from the actions of freely dissolved Ag + and unrelated to simple pharmaco kinetic attributes. Four key findings support this interpretation: • The lack of AgNPC selectivity toward DNA versus protein synthesis in undiffer entiated cells (whereas Ag + is highly selective toward the former macro molecule) • The inability of ascorbate to protect cells from oxidative stress and cell loss caused by AgNPC (whereas the same anti oxidant is protective against Ag + ) (Powers et al. 2010a), which implies that cell loss from AgNPC reflects a different under lying mecha nism and that, for the nano material, oxidative stress is a result of cyto toxicity, not a cause of it • The greater inhibition of protein synthesis at lower AgNPC concentrations and a loss of effect at higher concentrations, totally distinct from the mono tonic dose-effect relation ship for Ag + (Powers et al. 2010a), which indi cates that low AgNPC concentrations dis rupt protein synthesis through a mechanism unrelated to freely dissolved Ag + • The restricted effect of AgNPC to suppress the ACh phenotype (Ag + affects both ACh and DA phenotypes) (Powers et al. 2010a).
A comparison of AgNPC with the two sizes of AgNPPVP readily illustrates the roles of nano particle coating and size. Particle coatings clearly affected biological outcomes: One or both PVPcoated particles had greater effects than AgNPC toward cell loss, cell size, and promotion of TH activity, yet the AgNPPVPs had no effect on protein synthe sis. If the coating simply altered the dissolu tion of Ag + , then all the comparative effects would have been similar. At the same time, the largerdiameter AgNPPVP had greater effects than the smaller nano particles on most of the outcomes. Studies with gold nano particles show that particles ≥ 50 nm are actively taken up into cells, whereas smaller particles are not (Johnston et al. 2010), thus providing a possible explanation for the generally greater effects seen here for the larger AgNPPVP. However, this was not the case for the effects on differentiation into neuro transmitter pheno types, where the 10nm AgNPPVP had promotional effects on DA greater than those obtained with the 50nm nano particle. Thus, AgNPs not only elicit effects distinct from those of Ag + , but also display important dif ferences that are dictated by particle size and coating and specific to each biological process. This finding strongly indicates that AgNPs act biologically as nano particles and not just as a source of Ag + .
Although we present strong evidence that AgNPs work through a combination of Ag + release and mechanisms that reflect actions of the AgNPs themselves, more work is clearly needed to understand the mechanisms under lying nano particle effects and the inter actions of nano particles with extra cellular and intra cellular components. With regard to the for mer, our data show that immediate, anti mitotic effects of AgNPs are not sensitive to the pres ence of serum proteins, but it is certainly likely that inter actions could occur with the more prolonged exposures that would occur in vivo or as particles interact with proteins on the cell    surface. Indeed, addition of serum mitigates the loss of viability during a 24hr exposure to AgNPs in mouse keratino cytes (Murdock et al. 2008). Secondly, the diminished effects of AgNPC versus Ag + toward oxidative stress in undifferentiated cells and toward viability in either differentiation state, as well as non monotonic effects for these and other variables, may reflect protective actions of the citrate coating. Soluble citrate could supplement cellu lar metabolic and biosynthesis demands (Bauer et al. 2005), thereby ameliorating effects of AgNP exposure. This could also explain why the 50nm AgNPPVP, despite its larger size, had generally greater effects than AgNPC at the same nominal Ag concentration.
Given the rapid growth in AgNP use (Wijnhoven et al. 2009), detailed studies of the biological effects on neuro develop ment are critically important. In vitro models, such as that used here, can guide future in vivo studies to focus on critical stages of neuronal vulnerability, such as neuro transmitter tar gets and under lying cellu lar mechanisms. Our volume 119 | number 1 | January 2011 • Environmental Health Perspectives findings point to the likelihood that AgNPs are develop mental neuro toxicants that will display a wide window of vulnerable stages, ranging from events in early develop ment (mitosis, cell survival) through later stages of neuro differentiation. The non monotonic dose-response relationships seen with the nano particles, along with the dependence on coating and size, point to multiple mecha nisms of action rather than a single mecha nism. Accordingly, the effects of in vivo exposures may differ substantially at different develop mental stages and at different ends of the dose-response continuum. One uniform finding, however, was that the AgNPs, like Ag + , divert the end phenotype away from ACh and toward DA, albeit by different contribu tory mechanisms for each individual agent. If this occurs in vivo, we would expect to find substantial miswiring of ACh and DA circuits. Accordingly, we would then predict that the specific neuro behavioral outcomes of AgNP exposure are likely to include adverse effects on cognitive, reward, and motor performance. We are currently examining outcomes of AgNP exposure in developing zebrafish to determine if, as predicted by these in vitro studies, AgNPs are develop mental neuro toxicants in vivo.