Investigations that assess the pulmonary toxicity of TiO₂ particles in vitro have used a variety of models including macrophages (due to their contribution to clearance), epithelial cells (due to their abundance, and expected interaction with particles) and explanted tissue.

Park et al. 26 investigated the cytotoxicity of TiO₂ NPs (21 nm) to BEAS-2B lung epithelial cells, at concentrations ranging from 5 to 40 μg/ml, for up to 96 hours. A dose and time dependent decrease in cell viability was observed. Caspase-3 was activated by TiO₂, with chromosome condensation also apparent, which was suggestive that an apoptotic mechanism of cell death was involved. Reactive oxygen species (ROS) production, glutathione (GSH) depletion, and increased hemeoxygenase (HO-1) expression was evident, thereby implying that an oxidative mechanism drove the cytotoxic response. IL-8, IL-1, IL-6, TNFα, and CXCL2 (neutrophil chemoattractant) cytokine gene expression was increased, insinuating that an inflammatory response was also stimulated by TiO₂. The findings therefore indicated that an oxidant driven response was integral to the toxicity of TiO₂, which had inflammatory and cytotoxic consequences.

Churg et al. 27 determined the impact of TiO₂ exposure (up to 7 days), on the development of a fibrotic response, within the airway wall on rat tracheal explants. Both microparticulate (0.12 μm) and nanoparticulate (0.021 μm) forms of TiO₂ were investigated. No changes in gene expression were observed until day 5. NPs stimulated growth factor expression (such as platelet-derived growth factor (PDGF)) and enhanced pro-collagen expression. The authors suggested that the NPs were more fibrogenic, which is anticipated to contribute to airway obstruction in vivo. However, TGFα and TGFβ expression was increased for microparticles only, but no fibrotic-like morphological changes occurred, and so it was suggested that these mediators were not responsible for driving the fibrotic response mediated by NPs.

Gurr et al. 28 investigated the oxidative damage exhibited by TiO₂ (10, 20 or >200 nm) within bronchial BEAS-2B cells. The oxidative potential of TiO₂ was confirmed by the finding that lipid peroxidation (indicated by increased malondialdehyde (MDA)) and ROS (hydrogen peroxide) and reactive nitrogen species (NO^.) production were enhanced by nanoparticulate forms of TiO₂. DNA adducts were only formed following exposure of cells to TiO₂ NPs. The oxidative and genotoxic potential of nanoparticulate forms of TiO₂ was therefore demonstrated to be superior to that of their larger counterparts.

Simon-Deckers et al. 29 investigated the toxicity of a panel of particles, including TiO₂ (in anatase and rutile forms, and a variety of sizes) and Al₂O₃, to A549 cells (a human carcinoma derived alveolar epithelial type 2 cell line). In general, the cytotoxicity of the particles was low, despite the fact that they were internalised by cells, and contained within cytoplasmic vacuoles or lysosomes. However, it is noteworthy that it is possible for particles to be internalised by cells without having a detrimental impact on normal cell function. The crystal phase of TiO₂ was observed to influence the cytotoxicity exhibited by particles. Specifically, the greater the anatase content of the sample, the greater the ability to induce cell death. In addition, the size of NPs was suggested to contribute to their toxicity, as TiO₂ NPs were more toxic than their larger counterparts. In addition, the chemical composition of NPs was also thought to impact on their toxicity, as 12 nm TiO₂ NPs were more toxic than 11 nm Al₂O₃, despite their similarity in size. Consequently, the size, composition and crystal phase of particles were all suggested to influence NP toxicity, despite the fact that the toxicity of NPs used was generally low. Therefore, it is acknowledged that investigating size dependent effects of particles is confounded by differences in other particles attributes such as the sample phase. Consequently determining the properties of TiO₂ that drive toxicity is difficult, and therefore isolating a particle characteristic that is solely responsible for mediating adverse effects is not likely to be attainable.

Kim et al. 30 assessed TiO₂ (1 μm) mediated cytotoxicity to alveolar macrophages (obtained from rats, via bronchoalveolar lavage), when exposed to concentrations ranging from 0.5 to 5 mg/ml, for up to 5 hours. TiO₂ elicited a dose dependent decrease in cell viability, which was suggested to be mediated by suppressed adenosine triphosphate (ATP) generation, and also reliant on the interaction of TiO₂ with scavenger receptors on the cell surface. It is relevant that the response of silica particles (1.6 μm) was also considered. Silica had a greater cytotoxic potential than TiO₂ and the mechanisms underlying the response were different.

As responses of organs and tissues are likely to involve interactions between different cell types, it is necessary to incorporate this within in vitro tests. Barlow et al. 31 investigated the ability of L2 epithelial cells to modulate macrophage migration in vitro, on exposure to nano and microparticulate forms of TiO₂ (29 nm and 250 nm, respectively) and CB (260 nm and 14 nm diameter respectively). TiO₂ and CB induced a dose dependent decrease in epithelial cell viability, which was greatest for NPs (when exposed at concentrations ranging from 62.5 to 2000 μg/ml for 24 hours). Conditioned medium (obtained from epithelial cells treated with particles) was able to increase macrophage migration, but only for ultrafine carbon black NPs. Therefore, TiO₂ NPs were unable to stimulate the release of chemotaxins from epithelial cells, which the authors suggest could be due to their smaller surface area, when compared to that of ufCB.

From the investigations discussed, it is evident that TiO₂ particles are able to detrimentally affect both lung derived macrophage and epithelial cells. Therefore, in general, particles mediated oxidative, inflammatory and genotoxic effects that eventually culminated in cytotoxicity. However, in some situations there was no evidence of toxicity, mediated by TiO₂. These differences may be derived from the model used, or the particle under investigation. It is of interest that the mechanisms driving the toxicity of TiO₂ in vitro (particularly inflammation and oxidative stress) are also witnessed in vivo (see earlier).

The penetration of TiO₂ particles within the skin has been a focus of a number of in vitro investigations due to the exploitation of particles within sunscreens and cosmetic products. Gamer et al. 32 evaluated the penetration of sunscreen formulations containing micron-sized TiO₂ (up to 400 μg/cm², for 3 to 24 hours) within porcine skin explants. It was observed that the location of both particle types was restricted to the stratum corneum following their topical application. Specifically, particles were deposited on the skin surface, within the outmost layer of the stratum corneum, as evidenced by their removal via washing or tape stripping analysis. There was therefore no evidence that particles penetrated into the deeper stratum corneum layers, epidermis or dermis, and thus the barrier function of the skin was successful in limiting the passage of particles.

Similarly, Pflucker et al. 33 illustrated that TiO₂ NPs (20-50 nm) did not penetrate porcine skin in vitro. A TiO₂ emulsion (4 mg/cm²) was topically applied to excised pig skin for 24 hours. Skin biopsies were taken following exposure, and the penetration of particles assessed using tape stripping. Again TiO₂ was exclusively located within the outermost layer of the stratum corneum, with no distribution to the underlying living cell layers evident. Consequently, this study suggests that the intact stratum corneum is an effective barrier to TiO₂, and that the penetration of particles within the skin is negligible.

In addition, Dussert et al. 34 investigated the distribution of TiO₂ and ZnO NP containing sunscreen formulations within human skin explants. Again, sunscreens were topically applied to skin but no penetration of particles, or intracellular delivery was observed past the stratum corneum skin surface.

The functional implications of the exposure of the various skin cell populations have also been determined within in vitro investigations. From the data available for TiO₂ particles, such studies are potentially more relevant to compromised/damaged/diseased skin than normal healthy skin, due to the negligible penetration of particles within normal skin. There is therefore a need to consider the distribution of particles within damaged or sunburnt skin, which would be likely to promote the penetration of particles, due to damage to the stratum corneum. There is little or no data currently available to assess the impact of such damage on penetration or hazard.

Kiss et al. 22 exposed HaCaT keratinocytes, human dermal fibroblast cells, SZ95 sebaceous gland cells and primary human melanocytes to TiO₂ (9 nm diameter), at concentrations of 0.15 to 15 μg/cm², for up to 4 days. Particles were internalised, and evident in the cytoplasm and perinuclear region of fibroblasts and meloncytes. Particle uptake was also associated within an increase in intracellular calcium in these cell types. However, no particle uptake, or alterations in calcium signalling were observed within keratinocytes or sebocytes. A dose and time dependent decrease in cell proliferation was evident within all cell types, and an increase in cell death (via apoptosis) within fibroblasts was apparent. The direct contact of cells with TiO₂ particles is therefore of concern, as it can disturb normal cell functions within the different skin cell populations, but it is noteworthy that all cell types were not affected similarly, and thus cell specific effects are encountered on exposure to TiO₂ particles. Kiss et al. 22 also demonstrated that the stratum corneum was an effective barrier against micronized TiO₂ in vivo (see earlier). However it would have been of greater interest to determine if NPs are able to penetrate skin, as the in vitro component of the study illustrated that when in direct contact with skin cells TiO₂ NPs are able to elicit toxicity.

Jin et al. 35 determined the cytotoxicity of TiO₂ (20-100 nm) to mouse L929 fibroblasts at concentrations of 3-600 μg/ml, for up to 48 hours. There was a time and dose dependent decrease in cell viability induced by TiO₂. An increase in ROS production and decreases in GSH and SOD activity were observed, thereby implying that oxidative stress was a feature of the response. TiO₂ was also demonstrated to be internalised by phagocytosis, and was contained within lysosomes. Analysis of cell morphology illustrated that cell morphology was detrimentally affected by TiO₂, and that cell adhesion and proliferation was prevented, confirming the decrease in cell survival.

The studies conducted demonstrated that the toxicity of TiO₂ to skin cells is of concern, with oxidative, genotoxic and cytotoxic consequences evident, and likely to act in concert. However, in order to execute these effects, particles would be required to penetrate past the stratum corneum, to reach the underlying cell populations, which is unlikely to occur in healthy skin (see earlier). As noted previously, the penetration of particles within damaged skin should therefore be considered a priority in future experiments.

Only one study could be identified that evaluated the impact of TiO₂ exposure on the GIT, in vitro. Zhang et al. 36 investigated the contribution of photoexcitation, to the cytotoxicity of TiO₂ (21 nm) within human colon carcinoma Ls-174-t cells. Cells were exposed to TiO₂ for 24 hours, at concentrations up to 1000 μg/ml. Limited cytotoxicity of cells was observed, following TiO₂ exposure, in the absence of ultraviolet (UV)A irradiation. In contrast, a high level of TiO₂ induced cell death was observed in the presence of UVA light, which was a dose and time dependent phenomenon. Therefore, the toxicity of TiO₂ had a photocatalytic component. Consequently, TiO₂ particles and light irradiation therapy was suggested by the authors to be a suitable candidate for the treatment of cancer, and warrants further investigation. However, this may be limited by the ability of UV light to penetrate the human body, and the specificity of the response, and so the exploitation of this for tumour treatment will inevitably be restricted.

Hussain et al. 37 compared the impact of a range of NPs, including iron oxide (Fe₃O₄, 30 nm), and TiO₂ (40 nm) on BRL 3A liver cells, at concentrations up to 250 μg/ml, following a 24 hour exposure. Both particle types were able to decrease cell viability, but only at the highest concentration tested.

Linnainmaa et al. 38 assessed the toxicity of TiO₂, in NP (rutile and anatase (20 nm)), and microparticulate (170 nm) forms) to rat liver epithelial cells, in the presence and absence of UVA light. No cytotoxicity or chromosomal damage was observed within cells exposed to all TiO₂ particle types in the presence or absence of UVA.

The two studies that investigated the toxicity of TiO₂ particles to the liver, suggest that TiO₂ exhibits a low level of toxicity to liver cells, but this would require further investigation to be confirmed, due to the limited number of liver cell targets considered within the available studies. The consequences of liver exposure, to TiO₂ are of particular interest due to evidence of preferential NP accumulation within this organ (see earlier).

The interaction of TiO₂ with endothelial cells that line blood vessels has been a focus of investigations, to determine the consequences of particle transport with the blood. In addition, the impact of TiO₂ on cardiomyocyte function has been investigated.

Peters et al. 39 evaluated the impact of a NP panel (including TiO₂, 70 nm) on HDMEC endothelial microvascular cell viability and function. Cells were exposed to TiO₂ at concentrations ranging from 0.5 to 50 μg/ml, for up to 72 hours. NPs were internalised by cells into cytoplasmic vacuoles, but minimal toxicity was associated with this, therefore highlighting that particle uptake does not necessarily impact on cell function. Consequently, TiO₂ was relatively non-toxic, with no cytotoxicity, and minimal IL-8 release stimulated by exposure.

Courtois et al. 40 determined the impact of TiO₂ (15 nm or 0.4 μm diameter) exposure on nitric oxide (NO) mediated relaxation of pulmonary arteries in vitro. Pulmonary arteries were exposed to particles for 24 hours (at concentrations up to 200 μg/ml), in the presence or absence of vascular relaxants. It was found that urban particulate matter was able to impair NO-dependent relaxation within intrapulmonary arteries in vitro and in vivo (following intratracheal exposure). In contrast, manufactured NPs, including TiO₂ did not exhibit this effect. Therefore this study suggested that the pulmonary circulation was not affected by TiO₂ exposure, and that cells exhibited differed in their sensitivity to particles.

Helfenstein et al. 41 observed that TiO₂ NPs (up to 125 μg/ml, 20-30 nm diameter) were able to affect cardiomyocyte electrophysiology, enhance ROS production, and reduce myofibril organisation. A panel of particles was tested, of which diesel exhaust particles (DEPs) were demonstrated to exhibit the greatest toxic potency, and SWCNTs were found to have limited toxicity in comparison. The findings from this study suggest that cardiac cells may be damaged by TiO₂ or DEP exposure, and so were unable to function normally in vitro.

A limited number of studies have been conducted to determine the impact of TiO₂ particles within in vitro models of the cardiovascular system. However, the findings suggest that TiO₂ may promote an inflammatory response within blood vessels which has the propensity to stimulate or aggravate cardiovascular disease. Consequently, the ability of TiO₂ to contribute to inflammatory mediated disease requires further consideration in vivo. In addition, the ability of TiO₂ to negatively impact of cardiomyocyte function has the ability to disturb normal cardiac electrophysiology. Again, further studies, in vivo, would be required to verify this observation.

Long et al. 42 established the contribution of oxidative stress to the neurotoxicity of TiO₂ NPs (30 nm diameter, but particles agglomerated in culture medium). BV2 microglia cells were exposed to TiO₂ for periods of up to 120 mins, at concentrations ranging from 5 to 120 ppm, and ROS production determined. The oxidative response mediated by TiO₂ within microglia had two components. Firstly, a rapid increase in ROS production was observed at 1-5 minutes, and termed an oxidative burst. This was followed by a greater, sustained ROS release from 60-120 minutes. It is also relevant, that despite having a primary particle size of 30 nm, the particles were observed to aggregate within the cell culture medium, which increased with increasing concentration, and this appeared to promote ROS production, perhaps due to the greater uptake of larger structures by phagocytosis. Internalised particles were found within the cytoplasm. Mitochondria located in the vicinity of the internalised aggregates were found to be swollen and disrupted, which was postulated by the authors to promote an apoptotic or necrotic response.

Similarly, Long et al. 43 investigated the neurotoxicity of nanoparticulate TiO₂ (up to 120 ppm). It was illustrated that particle aggregates were phagocytosed by BV2 microglial cells, and contained within membrane bound vesicles. ROS production by cells was associated with particle exposure, and stimulated the upregulation of genes involved with inflammation, apoptosis, and cell cycling, and a down-regulation in energy metabolism. It was therefore evident that increased ROS production as a consequence of TiO₂ exposure also triggered cytotoxicity via apoptosis. However, TiO₂ exposure was non-toxic to N27 neuronal cells, following a 72 hour exposure, despite being internalised. These results reinforce the finding that particle internalisation does not necessarily equate to the initiation of a toxic response within cells, and that cell types differ in their susceptibility to particle toxicity. In contrast there was evidence of neurone loss within primary cultures of rat striatum within 6 hours of exposure, which was suggested to occur via microglia mediated ROS production.

Information regarding the neurotoxicity of TiO₂ is severely lacking, and restricted to the work conducted by one research group. However, it is evident that the response is likely to be dictated by the cell type under investigation. Again, an oxidant driven response appears to be integral to particle toxicity. At present, there is insufficient evidence to make generalisations regarding TiO₂ mediated neurotoxicity.

Only one study was found that assessed the toxicity of TiO₂ NPs to the kidneys, in vitro. L'azou et al. 44 investigated the effect of TiO₂ (15 or 50 nm) on IP15 mesangial, and LLC-PK₁ proximal tubular epithelial renal cells (at concentrations up to 512 μg/ml, for 24 hours). The different cell types exhibited different sensitivities to the toxicity of TiO₂. No cytotoxicity was observed within IP15 cells on exposure to TiO₂. However, TiO₂ elicited a cytotoxic response within LLC-PK₁ cells, suggesting that this cell type was more sensitive to TiO₂ toxicity. This was postulated to derive from their high endocytic activity, and therefore greater internalisation of particles. The cytotoxic response was also observed to be size dependent, with a greater response exhibited by smaller particles. TiO₂ (15 nm) was also able to induce morphological changes within both cell types (namely cell shrinkage, and detachment), and was internalised into cytoplasmic vacuoles. TiO₂ was unable to induce an increase in ROS production. However, 13 nm carbon black NPs were consistently recognised as being more toxic, in terms of cytotoxicity, and ROS production, which was evident in both cell types. Toxicity to renal cells was therefore observed to be particle size, particle composition, and cell type specific.

The findings from the only study identified as studying the toxicity of TiO₂ particles to the kidneys, suggest that the response of the individual cell populations vary. However, further investigations would be required to identify if the finding was applicable to other particle types, and if the same response is evident in vivo. Therefore it is difficult to draw definitive conclusions about particle toxicity to the kidney, due to insufficient data being available.

It is acknowledged that macrophages are primarily responsible for the clearance of particles at sites of exposure, and accountable for the accumulation of particles within different target sites, which is anticipated to derive from their role within the reticuloendothelial system. The consequences of particle uptake by macrophages is therefore of interest, particularly the initiation of an oxidative driven or inflammatory response, and impact on their phagocytic function. In addition, the ability of particles to affect other immune cell populations, such as lymphocytes, is of interest as particles would be expected to encounter these cells within the blood.

Renwick et al. 45 determined the impact of particle exposure on the phagocytic activity of macrophages. It was observed that TiO₂ (and CB) particles were able to impair J774.2 macrophage cell phagocytic activity (indicated by the uptake of 2 μm latex beads), in a concentration dependent manner. TiO₂ NPs (29 nm) were more effective at inhibiting macrophage phagocytosis than their microparticulate (250 nm) counterparts, and the same response was exhibited by CB particles. It is likely, that the impairment of phagocytosis occurs as a consequence of the fact that cells are overwhelmed by the TiO₂ particle burden, and are therefore unable to further internalise the latex particles. This was suggested to be the case, as an increase in particle-laden macrophages correlated with a decrease in phagocytic cells. The impairment in phagocytic function of macrophages would be anticipated to impact on the clearance of particles, thus increasing their retention, and thereby prolonging their interaction with cells, and enhancing their propensity for eliciting damage.

Afaq et al. 46 demonstrated that alveolar macrophages were increased in rats on intratracheal exposure (2 mg/rat, with observations made for a period of up to 16 days) to TiO₂ NPs (<30 nm), which was maximal at 8 days post exposure. The exposure of alveolar macrophages to TiO₂ was associated with lipid peroxidation, glutathione depletion and enhanced hydrogen peroxide production, illustrating the oxidative response that developed. Exposure was also associated with cytotoxicity.

Kang et al. 47 determined the mechanisms underlying TiO₂ (25 nm) NP toxicity within lymphocytes, at concentrations ranging from 20 to 100 μg/ml, for up to 48 hours. A particular focus of the study was on the oxidative potential of TiO₂. A dose and time dependent decrease in cell viability was observed. TiO₂ NPs also exhibited a genotoxic effect, whereby the frequency of micronuclei formation increased, which insinuated chromosomal damage occurred on exposure of cells to TiO₂. DNA damage was also observed, using the Comet assay. Importantly, the pre-treatment of cells with the antioxidant N-acetyl cysteine (NAC), decreased TiO₂ mediated DNA damage, thereby inferring that the response was ROS mediated. ROS production was subsequently confirmed to be increased by TiO₂ exposure. The DNA damage inflicted by TiO₂ resulted in the activation of a protective response in the form of increased levels of p53 protein, highlighting an attempt of the cells the repair the damage mediated by TiO₂. TiO₂ mediated increases in ROS production, and therefore DNA damage, was thought to be central to its toxicity within lymphocytes.

Investigation of the impact of TiO₂ particle exposure on immune cell function is lacking. Alveolar macrophages are recognised as being integral to the removal of particles from sites of deposition, and are, in the main, responsible for the accumulation of particles within different target sites. It is therefore of concern that particles appear to impair macrophage phagocytosis, and this should be considered further in future experiments. It is also suggested that particles initiate an oxidative driven response within immune cells that may have inflammatory and/or cytotoxic consequences.

Inflammatory responses have been illustrated as being a prominent feature of a number of studies that investigate the toxicity of nanoparticles such as TiO₂. In vivo, the infiltration of neutrophils has been repeatedly demonstrated to characterise the inflammatory response to TiO₂ NPs 36, in addition to elevated, particle laden macrophage numbers 1012. This is likely to be mediated by increases in inflammatory mediators, such as IL-8, and TNFα that have been observed in vivo 12 and in vitro 18. The inflammogenic nature of TiO₂ has been observed within a number of target sites including the lung, brain, and cardiovascular system, and so is not an organ specific observation. Inflammation has been demonstrated to be recurrently associated with TiO₂exposure, with its manifestation evident within different models, in response to different TiO₂ samples. Therefore, in general the inflammatory response that manifest is not organ, or particle sample specific.

A focus of a number of investigations has been to identify the contribution of oxidative stress to the inflammatory and cytotoxic responses elicited by TiO₂. Studies previously discussed found that oxidative stress was a prominent feature of the response to TiO₂ NPs. These include evidence of increased ROS production, depletion of cellular antioxidants, increase in oxidative products (such as lipid peroxidation) or evidence that toxicity is diminished on pre-treatment with antioxidants. These observations have been made in a variety of cell types, including for example; lung epithelial cells 26, fibroblasts 35, and microglia 42. A summary of the oxidative responses mediated by TiO₂ can be found in table 2.

It is acknowledged that the development of a moderate level of oxidative stress is connected to the initiation of inflammatory responses through the activation of ROS sensitive signalling cascades 48. Consequently Kang et al. 49 compared the ability of nano (21 nm) and microparticulate (1 μm) forms of TiO₂ (0.5-200 μg/ml) to initiate an oxidative response within RAW 264.7 macrophages, following exposure for up to 24 hours. No alterations in cell viability were observed on exposure of cells to both forms of TiO₂. This finding is of importance, as when assessing the processes that underlie particle toxicity, sub-lethal concentrations are required. ROS production was increased by TiO₂, and was greatest in magnitude for the NPs. The ROS sensitive mitogen-activated protein kinase (MAPK) signalling pathway was activated by TiO₂ exposure (indicated by ERK1/2 phosphorylation), which was suggested to initiate the increase in pro-inflammatory mediator production (TNFα, MIP-2). Within the endpoints measured, the response exhibited by NPs was consistently greater than that of their larger counterparts, when administered at an equivalent mass basis. The oxidative stress paradigm therefore held true within this study; specifically that particle mediated ROS production (at a moderate level) stimulates signalling cascades that promote the activation of transcription factors that, in turn, initiate an inflammatory response. Accordingly, the ability of particles to promote ROS production is thought to be central to their ability to regulate inflammatory responses.

In line with these findings, Xia et al. 50 investigated whether TiO₂ (11 nm), ZnO (13 nm), and CeO₂ (8 nm) oxidative stress development or particle dissolution contributed to their toxicity. RAW 264.7 macrophages and BEAS-2B epithelial cells were exposed to the NP panel for up to 15 hours, at concentrations up to 60 μg/ml. However, only ZnO particles were capable of inducing cytotoxicity, within both cell types. In addition, ZnO stimulated an increase in ROS production, which in turn, increased HO-1 expression, and activated the JNK signalling pathway, which paralleled the release of IL-8 and TNFα. Increased calcium release was also mediated by ZnO, and was associated with mitochondrial damage. These findings therefore imply that ZnO elicits an oxidant driven response that was responsible for the initiation of an inflammatory and cytotoxic response, which was absent for TiO₂ and CeO₂. This is an important finding, as contrary to evidence that TiO₂ can exhibit an oxidative mediated response, this does not always transpire, and therefore the experimental model and TiO₂ sample may be influential to the observed response. On the contrary, CeO₂ was able to stimulate a cytoprotective response, whereby pre-treatment of cells with CeO₂ protected against diesel exhaust particle mediated cell damage (which is known to be oxidant driven). All particle types were internalised by an endocytic mechanism, however, only ZnO accumulated within lysosomes, which was suggested to drive their ability to inflict oxidative injury, and promote particle dissolution. ZnO also induced ultrastructural alterations, including nuclear fragmentation, apoptotic body formation and mitochondrial disappearance. ZnO was therefore recognised as being the most toxic particle within the panel, however this was thought to be accounted for, in part, by their dissolution and release of Zn²⁺ ions. The toxicity of the metal oxide particles was therefore suggested to be driven by their oxidant properties, which was dependent on their composition, dissolution and intracellular fate. The composition of NPs has been demonstrated to be integral to their toxicity within this study, and thus not all metal oxide particulates are expected to behave similarly.

Lu et al. 51 had a slightly different focus and determined the impact of TiO₂ (0.2-3 mg/ml) on protein tyrosine nitration (using bovine serum albumin (BSA) as a model protein). Protein nitration was associated with TiO₂ exposure, in the presence of UV light, with anatase TiO₂ samples being more photoactive than rutile forms. Antioxidants, such as GSH and vitamin E were able to prevent against the protein nitration observed. To determine if proteins contained within mouse skin were targets for TiO₂ mediated nitrosation, similar tests were conducted using mouse skin homogenate. Again, anatase forms of TiO₂ were observed to exhibit a greater propensity to nitrosate proteins. This study therefore highlighted the photocatalytic activity of TiO₂, and the superior toxicity of anatase forms, and demonstrated the ability of TiO₂ to induce nitrative stress.

Oxidative responses exhibited by TiO₂ particles have been a focus of a number of studies, and its recurrence has prompted the suggestion that oxidative stress drives the inflammatory and cytotoxic responses evident. It is relevant that the level of oxidative stress, which is inevitably related to the duration or concentration of particles administered, drives the nature of the response that follows particle exposure. Specifically, at moderate levels of oxidative stress, inflammatory responses may be stimulated due to the activation of ROS sensitive signalling pathways 48. At higher levels of oxidative stress, cytotoxicity is evident, as cells are damaged by an overwhelming burden of ROS 48.

The clearance of TiO₂ particles by phagocytic cells has been a particular focus of studies, due to their ability to remove particles from the exposure sites, circulation or secondary target sites. However, a number of non-phagocytic cell populations also have the propensity to internalise particles. Considering the internalisation of particles by cells is of relevance as particle uptake may enhance their toxicity due to their interference with normal cellular physiology and function. However, it is also possible that particles act from outside the cell to elicit toxicity that is mediated by interactions of particles with the cell surface. Furthermore, particles can be internalised by cells and not impact on cell function, which is likely to be driven by the intracellular location in which they accumulate or the extent of uptake.

Stearns et al. 52 demonstrated that 50 nm TiO₂ (40 μg/ml) was internalised by alveolar A549 epithelial cells into membrane bound vesicles after 3, 6 and 24 hour exposures. However, it was observed that the uptake of particles by these cells was limited to their aggregated form. TiO₂ internalisation was suggested to be mediated via phagocytosis, due to the fact that plasma membrane projections surrounded and engulfed the particles prior to uptake.

A focus of a number studies has been to reveal the contribution of macrophage populations to the clearance of particles, particularly within the lung. However, Geiser et al. 53 argued that alveolar macrophages were not primarily responsible for the uptake and clearance of NPs within the lung in vivo. Rats were exposed to TiO₂, via inhalation (0.1 mg/m³), and alveolar macrophages were isolated, and the internalisation of particles assessed, 1 or 24 hours post exposure. Large TiO₂ particles (3-6 μm) were more effectively cleared by alveolar macrophages, than their smaller (20 nm) counterparts. Therefore TiO₂ NPs were able to bypass the important alveolar macrophage mediated clearance mechanism within the lung, due to their small size. These findings could be expected due to the known size limitations of uptake processes such as phagocytosis, which is thought to be restricted to substances that are 1 to 5 μm in size 54. The ability of TiO₂ NPs to evade phagocytosis was confirmed by the findings that they were not enclosed by a vesicular membrane equivalent to that surrounding the larger TiO₂ particles. Therefore it was suggested that phagocytosis was not responsible for NP uptake, and instead it was suggested that a sporadic, non-specific mechanism of uptake enabled NP uptake by macrophages. Alternatively it was suggested that NPs were internalised unintentionally when macrophages phagocytosed other material. The findings of Geiser et al. 53 are also able to provide an explanation for the finding that only TiO₂ agglomerates, and not individual NPs, were internalised by A549 cells in vitro 52. This was also recognised by Stearns et al. 52 who highlighted that there was limited evidence of the uptake of individual particles by cells. As Stearns et al. 52 utilised epithelial cells, and Geiser et al. 53 used macrophages which are professional phagocytes, it is relevant that the processes that drive particle uptake may be cell specific, and reliant on their function.

Rothen-Rutishauser et al. 55 utilised an in vitro airway wall model (triple cell co-culture), containing macrophages, epithelial, and dendritic cells to determine the translocation of particles between the different cell types, and to assess the intracellular fate of particles. Membrane-bound aggregates (>0.2 μm) of TiO₂ were evident within all cell types. However, smaller aggregates (<0.2 μm) were apparent within the cell cytoplasm, but were not membrane bound, suggesting that the entry mechanism of particles was size dependent. However, it is relevant that the behaviour of gold, and polystyrene particles was also assessed, and it was found that their intracellular location differed, which implied that different NPs are internalised by different mechanisms or follow different intracellular trafficking processes.

The uptake of TiO₂ by a variety of cell types has been demonstrated on numerous occasions. The consequences of particle internalisation are anticipated to be oxidative and cytotoxic in nature. It is anticipated that particle physicochemical properties may influence their internalisation by cells. Accordingly, the size and surface charge of particles has the ability to influence their uptake by cells. In addition, particle size and the cell type under investigation also have the potential to determine the mechanism of uptake and intracellular fate of particles.

Evidence of genotoxicity has been previously encountered within a number of studies; micronuclei development is associated with TiO₂ exposure, which is indicative of chromosomal damage, DNA damage has also been observed in response to TiO₂ particulate exposure.

Rahman et al. 56 investigated the potential for TiO₂ to elicit DNA damage within SHE fibroblasts. Micronuclei were evident within cells exposed to nano (<20 nm), but not micro (>200 nm) TiO₂ particles (exposed at concentrations up to 10 μg/cm², for up to 72 hours), which insinuates that chromosomal damage has occurred. The NPs also triggered the induction of apoptosis within cells, which is recognised as a common response to DNA damage.

Karlsson et al. 57 evaluated the ability of a variety of metal oxide NPs (copper oxide (CuO), ZnO, TiO₂, Fe₃O₄) to induce oxidative stress and DNA damage within A549 lung epithelial cells, at concentrations of up to 80 μg/ml for a period of up to 18 hours. CuO and ZnO were able to decrease cell viability, but TiO₂ and Fe₃O₄ were not capable of eliciting such an effect. In addition, all metal oxide particles, except Fe₃O₄ were able to elicit DNA damage, as determined by the Comet assay. The CuO NPs were consistently demonstrated to be the most toxic particle within the panel, which was thought to be drive by their ability to induce oxidative stress, and in turn DNA damage, which prompted cell death. It was suggested that the release of Cu²⁺ ions may contribute to the response, but are not solely accountable for the toxicity of CuO NPs. Importantly, this study illustrated that not all metal oxides behave similarly, and that their dissolution was able to contribute to their toxicity.

Dunford et al. 58 investigated the DNA damage inflicted by a panel of TiO₂ containing, commercially available sunscreens on fibroblasts, when exposed for a period of up to 60 minutes. It was demonstrated that, on sunlight illumination, the ability of TiO₂ to cause DNA damage was enhanced in plasmid DNA and cells. Anatase forms of TiO₂ were more effective at inducing damage than rutile forms. Mannitol (an antioxidant) was able to protect against DNA damage, therefore implying that TiO₂ mediated ROS production is central to its genotoxic potential. Consequently, absorption of UV light by TiO₂ is thought to stimulate ROS generation, which, in turn, leads to DNA damage.

Nakagawa et al. 59 investigated the genotoxicity of a panel of TiO₂ NPs (in rutile and anatase forms, ranging from 0.021 to 0.255 μm in diameter). A number of different in vitro genotoxicity tests were utilised; a microbial mutation assay (Salmonella typhimurium), the Comet assay to detect DNA damage (mouse lymphoma L5178Y cells) and a chromosome aberration assay (Chinese hamster CHL/IU cells). TiO₂ was used at concentrations up to 3200 μg/ml, with an exposure duration of up to 24 hours, and experiments were conducted in the absence or presence of UV light. Without UV light, TiO₂ NPs induced no, or very limited genotoxicity. However in the presence of UV light TiO₂ elicited DNA damage and chromosome aberrations (but no gene mutations) that was greatest for anatase forms. The phototoxicity of TiO₂ was therefore realised in this study.

Theogaraj et al. 60 investigated the UV dependence of the genotoxic potential of a panel of TiO₂ NPs (in rutile and anatase forms, all of which had a diameter of <60 nm). CHO cells, were exposed to TiO₂ at concentrations up to 5000 μg/ml for a period of 3 hours. In contrast to previously discussed investigations, no chromosomal alterations were observed within exposed cells, in the presence or absence of UV light.

In a different approach, Driscoll et al. 61 determined if there was a relationship between the inflammatory and genotoxic potential of several particles; namely ufCB (14 nm), TiO₂ (0.18 μm) and α quartz (0.9 μm). Rats were exposed to particles via intratracheal instillation, at a dose of 10 or 100 mg/kg, and analysis conducted 15 months post exposure. All particle types induced the infiltration of neutrophils into the lungs, indicating that particles could induce an inflammatory response. Hypoxanthine-guanine phosphoribosyl transferase (hrpt) gene mutation frequency was increased within alveolar type 2 cells, on exposure to particles. This response was replicated in vitro, following the exposure of RLE-6TN alveolar epithelial cells to BAL cells isolated from particle treated animals. It was thus suggested that BAL cell derived ROS were responsible for the mutagenic effects that transpired, and they were suggested to derive from neutrophil activity. Therefore the particle mediated neutrophillic driven inflammatory response within the rat lung, was postulated to increase the frequency of mutations within alveolar epithelial cells, which was a dose and material dependent finding. Accordingly, quartz was consistently found to elicit the greatest genotoxic response, followed by carbon black and then titanium dioxide, and this was related to their inflammogenic potential. Therefore, the particles themselves are not thought to be inherently genotoxic, but the inflammatory response, and in particular neutrophil presence (and therefore increased oxidant burden) instigated is anticipated to mediate this effect.

Concern regarding the genotoxic potential of TiO₂, has also emanated from findings that have illustrated that tumours develop in vivo following a chronic exposure regime (13 see earlier). However, as stated previously this is very much dependent on the exposure conditions, and in particular the exceptionally high doses that were administered within this model.

The ability of TiO₂ NPs to inflict DNA damage has been observed on numerous occasions, and is thought to be driven by particle mediated ROS production, with cell death often stimulated as a protective response. In addition, the manifestation of genotoxic events, as a secondary consequence of an inflammatory response is also evident, and requires further investigation. The appearance of TiO₂ mediated genotoxicity appears to be influenced by the crystal form of the sample, particle dissolution and light conditions. The finding that chronic exposure to TiO₂ can induce tumours in vivo (under extreme exposure conditions), also implies that the genotoxicity exhibited by TiO₂ is not limited to in vitro investigations.

Evaluation of TiO₂ NP effects on the reproductive system is limited to one in vitro study.

Komatsu et al. 62 determined the potential for TiO₂ NPs, DEP and ufCB to impair the function of male mouse reproductive system. This study evaluated the direct effect of NPs on testis-constituent cells, and examined the effect of TiO₂ on mouse Leydig TM3 cells, the testosterone-producing cells of the testis. TiO₂ NPs (25-70 nm) at concentrations 1 to 1000 μg/ml were examined and uptake into Leydig cells was detected using transmission electron microscopy (TEM) or field emission type scanning electron microscopy/energy-dispersive X-ray spectroscopy (FE-SEM/EDS). TiO₂ was more cytotoxic to Leydig cells than the other carbon based particles used in the study. The proliferation of Leydig cells was suppressed transiently by treatment with TiO₂. TiO₂ NPs were taken up by Leydig cells, and in turn affected cell viability, proliferation and gene expression.

Only one available study examined the effects of TiO₂ on male reproductive physiology. The literature highlights toxicity of TiO₂ NPs to male Leydig; however investigations are limited in number and sample size, and therefore further studies would be required to confirm such a finding. No literature examining TiO₂ NP effects on organs or cell types in the female reproductive system were found.

Particle dimensions are recognised as being fundamental to their toxicity. This derives from the fact that NPs have been consistently demonstrated to be capable of eliciting more pronounced toxicity than their larger (microparticulate) counterparts. The size dependency of TiO₂ toxicity has been frequently demonstrated 31224284547, and appears to be applicable to a variety of TiO₂ forms, and occurs regardless of the model used (table 3). However it is relevant that a high degree of particle aggregation and agglomeration is associated with TiO₂ administration, and so exposure to particles is unlikely to occur in a 'nano' form. However, what seems to be critical to the toxic potential of particle samples is the size of particles that make up the agglomerates. It is relevant that the terms agglomeration and aggregation are often used interchangeably within the field of nanotoxicology. However, some authors have suggested that NP aggregation and agglomeration are distinct phenomena with agglomerates formed by clusters of particles that are held together by electrostatic interactions, whereas aggregates are formed from covalently fused or sintered particles that are not easily separated 63.

As discussed, the superior toxicity of nanoparticulate forms of TiO₂ has been repeatedly documented. However, a study conducted by Dick et al. 64 illustrated that compared to other NPs such as nickel, cobalt or carbon black, TiO₂ nanoparticles were less potent at inducing an inflammatory response within the lung, despite their similarity in surface area. This was assumed to be related to their lower capacity to generate free radicals, and therefore perhaps particle composition and surface chemistry are also important contributors to toxic responses 64.

Bermudez et al. 7 exposed rats to TiO₂ via inhalation, to investigate their pulmonary toxicity. Despite the fact that the NPs had a primary particle size of 21 nm, the aerosol generated contained particle aggregates (1.37 μm). This is a common experience in the exposure of animals or cell culture models to nanomaterials, and is likely to be encountered within the exposure of humans. However, despite the formation of aggregates, the NPs still exerted toxicity. Similarly, aggregates of TiO₂ NPs (1.44 μm) have been demonstrated to be more toxic than similarly sized aggregates of their larger counterparts 3. However, Grassian, et al. 10 demonstrated that the properties of agglomerates of particles that formed during aerosol generation were dependent on the primary particle size. Specifically, 21 nm particle based agglomerates were less dense than their 5 nm counterparts whose agglomerates contained particles that were more tightly packed. It was found that 21 nm TiO₂ was more toxic than 5 nm particles following inhalation and instillation exposure. The agglomeration state of particles was therefore suggested to dictate their toxic potential. Consequently, although 21 nm particles were larger, it is anticipated that they would form agglomerates that would more easily deagglomerate due to the weaker interactions that held the particles together. As a result, 21 nm particles would be available as smaller structures to stimulate an enhanced toxic response. Furthermore the aggregates formed were larger within inhalation preparations than intratracheal suspensions, and so this may explain why intratracheal exposures produced a greater toxic response.

The greater toxicity of NPs, compared to their larger counterparts, is anticipated to be driven by the greater surface area of smaller particles, when administered at an equivalent mass dose. In fact, the surface area of particles has been previously suggested to dictate the toxic potential of particles 656667. In line with this hypothesis, Sager et al. 68 addressed which dose metric was most influential in driving the toxicity of TiO₂; the surface area or mass of particles administered. NPs (suspended in BALF) were observed to agglomerate to a diameter of 200-300 nm (despite having a primary particle size of 21 nm). This finding emphasises that particle exposure often does not occur in an individual particle form. Microparticle (1 μm) and NP particles were able to induce an inflammatory response (neutrophil infiltration, LDH activity, TNFα, MIP and IL-1β release), subsequent to intratracheal administration of rats. It was of interest that a higher mass of microparticulate TiO₂ was required to obtain the same inflammogenic response as nanoparticle TiO₂. Therefore, the inflammatory and cytotoxic potential of NPs was greater than that of microparticles, when administered at an equivalent mass dose. However, when the dose of particles administered was equivalent, in terms of surface area delivered (which necessitates the administration of a much higher mass of microparticles), both particle types behaved similarly. These findings suggests that the size, and surface area of particles is integral to their toxicity. Similarly, Sager et al. 69 determined whether the surface area of ufCB and TiO₂ NPs drives their inflammatory potential. Particles were administered to rats via intratracheal instillation, and toxicological observations made up to 42 days post exposure. At equivalent surface area doses, both particle types induced an inflammatory response that was characterised by the infiltration of neutrophils at day 1 which was of a similar magnitude. However the response decreased with time for ufCB, but for TiO₂ the response was sustained in nature. This pattern of response was similar for BALF protein levels. Consequently, the composition of particles may also contribute to their toxicity. This observation remains debatable as others have disputed this finding 11.

The size (and therefore surface area) of TiO₂ nanoparticles is known to be fundamental to their toxicity. The aggregation and agglomeration state of nanoparticles is also likely to be influential to their toxicity. Consequently, it is often the case that although the primary particle size is stipulated by investigators; cells and animals are not exposed to the particles in this form, 27323335424352707172. However it is often observed that agglomerates/aggregates of nanoparticles are more toxic than similarly sized agglomerates/aggregates of their larger counterparts. Therefore, the propensity of particles to aggregate has prompted investigators to prevent against its occurrence through the use of sonication, or inclusion of dispersants such as within particle suspensions, but this also needs to take into account what is relevant to human exposure.

As for other nanoparticles, the surface of TiO₂ particulates can be altered through the attachment of surface moieties, a process which is termed particle functionalisation. The surface of TiO₂ particles can also be modified through coating with aluminium oxide, or silica, which has generally been encountered within sunscreens to enhance protection from UV radiation 9.

Warheit et al. 9 investigated the impact of surface treatments (alumina or silica) on TiO₂ toxicity, within the lungs (see earlier). It was demonstrated that those exhibiting the greatest toxicity, were those that contained the highest aluminium oxide and/or silica content on the particle surface. However, overall it was observed that TiO₂ particles induced low pulmonary toxicity, despite the ability of surface coatings to influence this.

Oberdorster 73 demonstrated that the surface properties of TiO₂ were able to influence their toxicity. This was demonstrated via the intratracheal exposure of rats (500 μg/animal, for 24 hours) to TiO₂ (20 nm) that remained uncoated (hydrophilic) or received a silane coating (hydrophobic). The samples varied, with regards to their inflammatory potency within the lungs, so that the uncoated particles induced a lower response, than their coated counterparts.

Hohr et al. 74 investigated the impact of surface properties on TiO₂ toxicity. Microparticulate (180 nm diameter), uncoated and coated (via methylation) NPs (20-30 nm diameter) of TiO₂ were exposed to rats, via intratracheal instillation (1 or 6 mg/rat), and analysis conducted 16 hours post exposure. Administration of all particle types stimulated the infiltration of neutrophils, with the effect most pronounced with administration of NPs. It is also of interest that the coated (hydrophobic) NPs tended to stimulate neutrophil recruitment to a lesser extent than the uncoated (hydrophillic) particles. Protein, LDH, TNFα and MIP-2 levels in BAL also exhibited a similar pattern, with regards to the potency of the different samples, but overall the impact of surface methylation on TiO₂ toxicity was negligible. However, it was concluded that the toxicity of TiO₂ was driven by the surface area of particles, and not their surface coating. Therefore, a stronger inflammatory response was associated with NPs compared with their larger counterparts and so the effect of methylation was negligible.

Singh et al. 75 determined the impact of TiO₂ size (microparticles: 40-300 nm, NPs: 20-80 nm), surface modification (using methylation, to make particles more hydrophobic) and radical generating potential on their uptake and toxicity within A549 lung epithelial cells. TiO₂ particles, regardless of their size or methylation were phagocytosed as clusters of particles. Internalisation of NPs via clathrin mediated endocytosis, was only apparent for small particle clusters that were less than 30 nm, illustrating that this may be a size dependent phenomenon. ROS generation, and IL-8 production exhibited by cells exposed to NPs was greater than that of microparticles, and the findings were unaffected by the methylation of the particle surface. An important observation was that both particle types were present in an aggregated form, and that the NPs were in a smaller size range (<500 nm) than their larger counterparts (2000-5000 nm). Therefore the superior toxicity of NPs held true, despite their tendency to aggregate. Therefore, even when in an aggregated form the toxicity of TiO₂ particles was assumed to be primarily driven by their surface area.

Thevenot et al. 76 determined the impact of TiO₂ NPs, with various surface modifications (-COOH, -OH or -NH₂ functional groups), on the survival of a variety of cancer cell lines. Exposure concentrations were high (up to 10 mg/ml) and occurred for a duration of 3 to 24 hours. The cancer cell lines used were B16F10 and BF16F1 melanoma, Lewis lung carcinoma, JHU prostate cancer, and 3T3 fibroblast cell lines. The different cell lines showed different sensitivities to the cytotoxicity of TiO₂ NPs. 3T3, B16F10 and BF16F1 cells were unresponsive to all types of TiO₂. However, TiO₂ decreased the viability of JHU and LLC cells in a dose dependent manner. In addition, by altering the surface chemistry of TiO₂ particles, the toxicity of TiO₂ could be modified. In general, NH₂, and OH surface modified TiO₂ exhibited greater toxicity than COOH modified TiO₂. This study therefore illustrated that TiO₂ toxicity was very much dependent on the cell type in question, the surface chemistry of TiO₂ NPs, and concentration of particles used.

Rehn et al. 70 illustrated that unmodified, or silane coated forms of TiO₂ were non toxic to the lungs. Rats were exposed to the different forms of TiO₂ (20 nm) via intratracheal instillation (up to 1.2 mg/rat) and the inflammatory and genotoxic effects within the lung were determined, at 3, 21 and 90 days post exposure. All forms of TiO₂ were able to stimulate a modest increase in neutrophils and macrophages within the lungs (at day 3), but this was not persistent in nature, and was not associated with TNFα production. In addition, no genotoxicity was evident with TiO₂ exposure. In contrast, quartz induced a persistent and progressive inflammatory response, with a genotoxic element to the response also observed. However, the only genotoxic endpoint that was assessed was the detection of 8 oxy-guanine, so perhaps a more comprehensive genotoxic study should be considered for TiO₂.

The modification of the surface of TiO₂ particles is able to influence its toxicity, however, this is likely to be dependent on the modification, and cell type in question.

TiO₂ exists in two main crystal phases, termed rutile and anatase. These TiO₂ forms vary with regards to their crystalline structure and surface properties, which is responsible for differences within their toxicity 77. Anatase has been demonstrated to be the most toxic form of TiO₂, with a number of previously mentioned studies supporting this conclusion (see for example 17). The photoactivity of TiO₂ is also dictated by the crystal phase, and therefore surface characteristics of particles (such as oxygen vacancies on the particle surface), with anatase forms having a greater capacity to generate ROS on exposure to light (see later). A summary of the contribution of particle of crystallinity to TiO₂ toxicity is demonstrated in table 4.

Sayes et al. 71 assessed the in vitro toxicity of anatase (10 nm), rutile (5 nm) or anatase/rutile (3 nm) TiO₂ samples to A549 epithelial cells and HDF fibroblasts, at concentrations ranging from 3 μg/ml to 30 mg/ml, for up to 48 hours. The level of cytotoxicity exhibited by the different particle types varied. Accordingly, anatase particles were the most cytotoxic, while the rutile TiO₂ particles were the least toxic. This response was paralleled within the release of IL-8 from cells. TiO₂ particle stimulated ROS generation and photoactivity was also greatest for pure anatase samples. Overall, the results implied that the anatase samples had the greatest toxic potency. Consequently, the authors suggested that it was the phase of TiO₂ that was integral to its toxic potential. Therefore, despite the relatively large surface area of rutile TiO₂ NPs (112 m²/g), their surface chemistry is thought to be less reactive than that of anatase samples (that had a larger surface area of 153 m²/g), and so rutile particles are less toxic. The findings also suggested that the ability of TiO₂ particles to generate ROS, governs their cytotoxic and inflammatory potential, which is dictated by the crystal structure of TiO₂, and hence toxicity is not solely driven by surface area. However, rutile NPs were observed to agglomerate to the greatest extent, and so this may also explain why this particle type was observed to be less toxic. Although the findings demonstrate that particle crystallinity is important to TiO₂ toxicity, the relevance of the findings is questionable, as in general, toxicity was only observed at higher concentrations (greater than 300 μg/ml), which are not deemed to be physiologically relevant (see later).

Pan et al. 72 investigated the toxicity of a panel of TiO₂ particles to primary human dermal fibroblasts. Uncoated rutile TiO₂ (15 nm), polymer coated rutile TiO₂ or uncoated anatase TiO₂ (200 nm) particles were exposed to cells for up to 11 days, at concentrations up to 0.8 mg/ml. Uncoated rutile TiO₂ caused alterations in cell morphology, so that cells had a smaller cell area, became elongated, and detached from the culture surface. Anatase TiO₂ caused a greater magnitude of damage to cells, with severe morphological changes observed including breakage of actin filaments, and plasma membrane rupture within exposed cells. Rutile TiO₂ did not impact on cell viability, but did slow cell proliferation rate but were internalised by cells, and were contained within cytoplasmic vesicles. Anatase TiO₂ was internalised by cells, and could access the nucleus. H₂O₂ production by cells was greatest with anatase particles, implying that their ability to generate ROS was greatest within the TiO₂ particle panel. Anatase particles were therefore more potent that rutile particles at inducing cell damage. Rutile particles were then coated in order to prevent against particle adhesion to the plasma membrane, and thus limit their potential for internalisation. Coated particles were not evident on the cell surface, or within the cell interior. Consequently, a polymer coating abolished particle toxicity, which was suggested to derive from the lack of particle adherence to the cell surface, and thus limited internalisation, so that they were unable to impact on normal cell function. This study therefore demonstrates that particle coatings, and therefore surface chemistry, are able to modify particle toxicity. However, this requires further investigation, and may be dependent on the surface coating and cell type in question.

The phase composition of TiO₂ has been proposed to be influential in dictating the toxic potency of particles. This is likely to derive from their substantially different surface chemistry, with anatase forms exhibiting the greatest photocatalytic and biological activity. The importance of crystallinity, to the toxicity of particles has also been demonstrated for SiO₂ (see for example Kaewamatawong et al. 78).

Furthermore the ability of UVA or visible light to increase the toxic potency of TiO₂ particles, through increased ROS production, has been a focus of a number of previously discussed studies [see for example 365158], but does not always transpire 3860. This phenomenon is anticipated to be exploited, particularly within the treatment of cancer 36. The crystal form of TiO₂ has been suggested to be responsible for driving the photoactivity of particles, due to the importance of surface properties 77.