The systemic responses of primary human olfactory neurospheres to four types of commercially-available ZnO nanoparticles are reported here. Two types of ZnO samples had particles of different sizes and were uncoated (Z-COTE and Nanosun), while the other two had particles the same size as Z-COTE but bore different surface coatings (HP1 and MAX), allowing the relative impacts of particle size and applied surface coatings in commercially-produced ZnO samples to be assessed. The responses elicited in human olfactory neurosphere-derived (hONS) cells by all four ZnO samples were generally robust and internally consistent across the comprehensive suite of biological experiments employed, and also across four genetically distinct cell lines in nine cell-stress and viability assays. When first exposed to ZnO nanoparticles, the cells initiated pro-survival responses, generally independent of particle size but slightly dependent on coating, while apoptotic processes dominated after 24 h exposure to three of the ZnO samples - the uncoated Nanosun and Z-COTE, and HP1 coated with triethoxycaprylylsilane. In contrast, cells exposed to MAX, coated with a dimethoxydiphenylsilane/triethoxycaprylylsilane crosspolymer, were still viable at 24 h. Overall, these results suggest that surface coating can be an important factor in mitigating the toxicity of ZnO nanoparticles, while particle size has less impact, at least in the cellular system investigated here.
The minimal toxicity to hONS cells elicited by MAX, compared with HP1, suggests that the magnitude of toxicity reductions conferred by surface coatings may be very sensitive to the chemical composition of the coating molecules; indeed, this has been observed in other cell lines for ZnO nanoparticles bearing a variety of applied coatings . However, other, more subtle characteristics of the coating may also be important. Our comparative analysis of the surface coatings on HP1 and MAX indicates fewer coating molecules per unit surface area on HP1, which could be manifested by HP1 having a thinner coating, a less densely-packed coating, or a patchy coating, and these characteristics could also affect the extent of toxicity mitigation. Furthermore, even if a coating is uniform at the point of manufacture, the strength of interaction between the coating material and the ZnO surface may vary for different coatings (especially if the coating is physically adsorbed) and the coating may partially detach with time and/or in cell culture medium, resulting in a non-uniform coverage of the surface. We have no direct data on the structural characteristics of the coating on HP1 in the cellular system used here. However, consistent with an incomplete surface coverage for HP1 is our observation that the responses in HP1-treated cells clustered with those of cells treated with the uncoated nanoparticles, although delayed by a few hours; in contrast, the cellular responses to MAX were either substantially delayed or minimal. Furthermore, we note that the specific batch of HP1 used here dispersed more readily in cell-culture medium than two other batches of HP1 used by our laboratory, consistent with different extents of surface coverage. We are currently conducting a full physico-chemical analysis of HP1 from the three different batches . Our initial observation of different dispersabilities raises the broader question of the impact of batch-related differences in assessments of nanoparticle toxicity. Given the list of challenges and considerations currently associated with the in vitro testing of nanomaterials, taking into account potential batch-to-batch variations appears to be a daunting prospect, but highlights the importance for full nanoparticle characterisation.
Overall, it is tempting to attribute the relative cellular responses to the ZnO samples largely, if not completely, to different concentrations of zinc ions sourced from the dissolution of ZnO particles with varying exposed surface areas. It is feasible that a larger area of exposed particle surface might facilitate a more rapid increase in Zn2+ ion concentration compared to a coated or smaller area of exposed surface. Consistent with ZnO nanoparticle literature pointing to zinc ion-mediated toxicity [12, 13], a number of the phenotypic outcomes reported here (loss of cellular viability, increase in caspase 3–7 and decrease in cellular glutathione (GSH)) also have been observed as cellular outcomes following in vitro treatment of neuronal cells with several types of zinc salt . Furthermore, one of the key factors in cytokine stimulation is the rate of intracellular ion release after nanoparticle uptake by phagocytic cells, which appears to be independent of cytotoxicity ; and the increased level of IL-6 at 2 h observed here for the uncoated Nanosun, compared with the uncoated Z-COTE and coated HP1, is consistent with its larger specific surface area and hence a faster release of Zn2+ ions than might be expected for Z-COTE and HP1, with the coating on the latter also diminishing its dissolution rate. (In contrast, the cellular responses to the coated MAX are not consistent with zinc ion-mediated cytotoxicity, and the only significant response by the cells to MAX, namely high levels of IL-6 at 2 h and IL-8 at 6 h, may have been induced by the coating itself; this hypothesis has yet to be tested.) However, at odds with a zinc ion-mediated toxicity profile, hONS cells exposed to the uncoated Nanosun and Z-COTE exhibited similar responses, despite a 2.5-fold difference in powdered surface area. Furthermore, a recent report has questioned the extent to which Zn2+ ions from zinc salts are actually bio-available in cell culture medium . That report instead showed that ionic zinc swiftly forms a range of insoluble carbonate and phosphate-based nanoparticulate complexes when added to cell culture medium. It is thus possible that zinc salts used as a control for dissolved zinc may form and mimic the effects of nanoparticles themselves, compromising their intended purpose. The range of cytotoxicities associated with different counter ions of zinc salts , alongside conflicting reports on zinc salt toxicity [12, 13] compared to ZnO nanoparticles [39–41], and the possibility that zinc ions may themselves form a range of nano-complexes in cell culture, together suggest that further consideration of the role of Zn2+ in ZnO nanoparticle toxicity experiments is warranted.
A distinct spectrum of key cell-stress signalling pathways was most rapidly activated by the uncoated ZnO nanoparticles, followed by the coated HP1 particles. In contrast, cells treated with MAX had much less substantial cell-signalling responses, did not demonstrate a cellular stress response, and did not lose viability. This pattern was also reflected at the transcriptional level, where similar pathways associated with oxidative stress and cell survival were activated by all treatments, but consistently more so in cells treated with the uncoated nanoparticles or HP1 compared to MAX. Treatments that induced the strongest cell-signalling and transcriptional perturbations also showed the greatest cell function responses, resulting in the most significant losses in viability.
Specifically, in cells exposed to Z-COTE, Nanosun or HP1, we observed an early activation of the MAPK and Akt cell-signalling pathways, as well as the NF-kB pathway. This cell-signalling fingerprint is associated with inflammation, proliferation and anti-apoptotic responses and suggests that, in the first instance, these cells mounted a protective response. The activation of these pro-survival responses had substantially diminished by 6-10 h, concomitant with a decrease in cellular GSH levels, and increased proteasome and caspase 3-7 activities, indicative of oxidative stress, protein degradation and initiation of apoptosis, respectively. Late responses in these cells (24 h) included significantly compromised cell membrane and a decrease in most other homeostatic metabolic activities, consistent with a treatment-related reduction in cell viability. The increase in proteasomal activity in cells treated with either of the uncoated nanoparticles or the coated HP1, together with the induction of a range of molecular chaperone genes, may suggest the induction of an unfolded protein response . Disregulation of protein folding has similarly been suggested elsewhere as a mode of action following treatment of cells with micro or nano-sized ZnO particles ; whether this response would be a cause or effect of treatment-related loss of cellular viability remains to be elucidated. Additionally, at 6 h, we found that a number of histone-related genes tended to be up-regulated with increasing treatment toxicity, but down-regulated by treatment with the more benign MAX. Thus, the cytotoxic nanoparticle treatments may have been associated with a suppression of transcriptional activity in addition to an unfolded protein response.
All ZnO nanoparticles induced an early perturbation of Canonical Gene Pathways associated with oxidative stress and cellular stress responses, with stronger perturbations associated with increasing treatment cytotoxicity. At 2 h and 6 h, a number of metallothioneins, molecular chaperonins, zinc finger proteins and solute carrier (SLC) genes were differentially regulated (generally positively) by all treatments, suggesting the induction of mechanisms to modulate intracellular levels of zinc, as well as a generalised cellular stress response. Using immortalised cell cultures treated with ZnO particles, Moos et al. (2011) reported a similar transcriptional profile.
Interestingly, given its role as a key tumor-suppressor protein regulating cellular apoptosis in response to cyto or genotoxic insults, we did not observe a statistically significant increase in p53 phosphorylated at Serine 15 (a site closely associated with DNA-damage response ), nor did we see an increase in the activation of either of the cell-cycle checkpoint proteins, Chk1 or Chk2. This contrasts with previous reports of p53 protein up-regulation following in vitro treatment with ZnO nanoparticles [41, 45, 46] using different cell lines. However, consistent with the absence of activation of the cell-signalling pathways involving p-p53, pChk1 and pChk2 here, we also observed that a number of transcripts encoding DNA-damage processing proteins were generally down-regulated at 6 h, with the exception of the POLH gene, which was up-regulated by all treatments. The POLH gene encodes a polymerase that accurately replicates past thymine-thymine dimers (typically associated with UV-induced DNA damage) during translesion synthesis (TLS), but is otherwise a low-fidelity polymerase when copying undamaged DNA . The up-regulation of a key TLS enzyme alongside the down-regulation of enzymes associated with DNA repair, as well as activation of proliferative pathways and lack of cell-cycle checkpoint activation, may suggest that the hONS cells responded to damage in DNA induced by ZnO nanoparticles by preferentially by-passing lesions rather than repairing them.
There is little reported elsewhere in the literature concerning the impact of ZnO nanoparticles on the activation and function of DNA-damage processing mechanisms, such as base excision repair, nucleotide excision repair, or translesion synthesis. The general down-regulation of transcripts associated with DNA-damage processing reported here may suggest that, even though ZnO nanoparticles have been shown elsewhere to induce DNA-damage [20, 41, 45, 48], primary hONS cells did not respond to this genotoxic insult by activating DNA-damage repair mechanisms, or alternatively that the DNA was not damaged in these experiments. Consistent with the former, Hackenberg  showed, following repeated exposure to sub-cytotoxic doses of ZnO nanoparticles, that damage to DNA in human nasal mucosa mini-organ cultures was not repaired, and in fact damage increased during a 24 h recovery period, compared to repair observed in methyl methanesulfonate-treated cells. Different surface coatings have been shown previously to protect cell viability by reducing the generation of ROS from ZnO nanoparticles , but not necessarily by reducing nanoparticle-mediated genotoxicity . ZnO nanoparticle-mediated genotoxicity, therefore, may not be a major trigger for concomitant cytotoxicity. Whether ZnO nanoparticle-mediated DNA-damage might be associated with direct inhibition of damage-processing proteins by the particles themselves, or whether the generalised stress responses of increasingly unviable cells overwhelms or bypasses such mechanisms, or a combination of the two, would be an interesting avenue to pursue.
Here, we utilised classical suspension to expose cells rather than the novel technique of delivering nanoparticle aerosols to the air-liquid-interface (ALI) of cell cultures e.g. . The two techniques can produce differences in assay parameters, including deposition kinetics, nanoparticle agglomeration and dissolution, and the influence of cell-culture medium on the surface characteristics of the nanoparticles , which can influence biological end-points. Literature comparing the effects of nanoparticles delivered via the two techniques indicate general agreement, but also some differences, in cellular responses over time, including cell membrane integrity, gene expression of pro-inflammatory and oxidative stress markers, and cell viability [50, 51]. It was further reported that while toxicity assessments at the ALI are likely to produce the more relevant biological responses, the influence of gas-derived effects can mask those of the nanoparticles  and that, for the moment, classical suspension exposures remain a valuable complementary technique when assessing the toxicity of nanoparticles to cells relevant for inhalation exposure.
The actual nanoparticle dose to cells in the olfactory bulb will depend in part on the numbers of nanoparticles inhaled through the nose, and their state of agglomeration. Work by Guilherme and Kimbell  showed good agreement between experimentally derived calculations and computer modelling predictions for particle sizes up to 100 nm, suggesting that approximately 20% of nanoparticles deposited in the olfactory region will translocate to the olfactory bulb . While we determined sizes of the individual ZnO nanoparticles and time-dependent changes in sizes of agglomerates forming in cell-culture medium, we did not characterise the particle sizes in aerosol form. Thus, it is difficult to estimate what the nasally-inhaled dose of ZnO particles would be, with subsequent translocation to the olfactory mucosa, that would produce the dose used here for the suspension assays.