Carbon nanotubes (CNT) exhibit unique properties, including mechanical, thermal and electrical conductivity, as well as field emission properties. These properties are associated with many applications (car industry, sport accessories, ...), and lead to a steady increase in the industrial production of CNT. However, it is increasingly obvious that exposure to nanoparticles in general, and CNT in particular, can have adverse effects on human health, especially at the level of the pulmonary system, which is a primary route of exposure 1 . This is raising considerable concern 2 3 4 5 6 7 8 9 . Therefore, minimizing the risk associated with CNT exposure is of crucial importance.

Among adverse health effects secondary to exposure to CNT, inflammation and oxidative stress are particularly worrisome because they can be associated with tissue remodeling and impaired function and/or carcinogenesis 10 . Recent data show that the length and rigidity of the nanotubes influences the pro-inflammatory effect of CNT 11 . However, the possible influence of other physicochemical properties remains incompletely understood 12 . Since surface characteristics influence the pro-inflammatory effect of spherical nanoparticles 13 , we hypothesized that embedding CNT in polymers, thus modifying the surface environment of those CNT, could modify their toxicity, and thus represent a useful strategy to decrease adverse health effects of industrially-produced CNT, without affecting their specific properties and further applications. The protective effect of the compatible solute Ectoine against carbon nanoparticle-induced lung inflammation has been nicely demonstrated recently 14 . However, a strategy based on the production of intrinsically safer nanomaterials seems to be much more promising. Such an approach is essential to the further development and safe use of CNT, at an occupational level as well as at the level of the general user.

To assess our hypothesis, we utilized an original approach by coating industrially produced multi-walled CNT (MWCNT) with two different polymers (acid-based and polystyrene-based, respectively) that resulted in different surface environment but similar specific surface areas. To the best of our knowledge, although surface modifications is important to CNT toxicity 15 16 17 18 19 , polymer coating has never been used before as an approach to modulate CNT toxicity. Coating could represent a useful tool, as it allows modifying CNT's external surface environment, without affecting their intrinsic structure. We investigated whether coating modified the toxicity of MWCNT in vitro in murine macrophages and in vivo in mice monitored for 6 months after intratracheal MWCNT instillation. Here, we report that MWCNT-induced cytotoxicity, oxidative stress, and inflammation in both models were increased by acid-based polymer coating and decreased by polystyrene polymer coating. These results demonstrate that surface characteristics play a major role in the biological response to CNT and that modifications of the surface environment by coating with specific compounds may constitute a useful strategy for decreasing CNT toxicity, without affecting their intrinsic structure, already at the time of their production. This may therefore hold promise for improving occupational safety as well as that of general users.

We investigated the effects of industrially-produced MWCNT that were either uncoated (NT1) or coated with carboxylic polyacid polymer (NT2) or a hydrophobic polystyrene polybutadiene polymethylmethacrylate polymer (NT3). These MWCNT were provided by ARKEMA-France (Colombes, France). The two coated MWCNT were composed of 50% MWCNT and 50% coating polymer by weight (determined by Thermo Gravimetric Analysis). Coatings were performed as follows; both coated MWCNT were produced by impregnation of NT1 powder (50% weight) with a solution of polymer (carboxylic polyacid polymer for NT2, and polystyren polybutadiene polymethylmethacrylate for NT3), the solvent being thereafter eliminated by drying under vacuum conditions. Solvent was water for NT2, and methyl ethyl cetone for NT3. NT2 remained coated after dispersion in the culture medium, as revealed by the Toluidine Blue (TB) assay, which measures the stability of polymer grafting 20 (data not shown). Briefly, a 5.10^-4 M TB aqueous solution was prepared and 0.2 ml of a buffer solution of 2-amino 2-methyl propanol was added to reach and maintain the solution pH value at 10. Each MWCNT sample was placed in TB solution at 30°C up to 24 hours. At different time points, uncomplexed TB molecules were removed by rapid washing (30 minutes) of the sample by a basic aqueous solution (NaOH 5.10^-4 M, pH 9). Then, each sample was in 10 ml of an aqueous acetic acid solution (50% v/v) for 24 hours to obtain complete decomplexation of TB from the sample. The decomplexation solution was analyzed by UV/visible spectroscopy at 633 nm. There was no chemical link between the polymers and the nanotubes.

Both coated NT (NT2 and NT3) appear as powders, as the pristine one (NT1), and were dispersed in culture medium for both in vitro and in vivo experiments. The two polymers were also dispersed in culture medium in order to assess their intrinsic effects.

The effects of the three MWCNT were compared to those of crocidolite asbestos fibres (80 nm in diameter) kindly donated by Dr. Ghislaine Lacroix (INERIS, Verneuil-en-Hallatte-France) and of nanosized carbon black (CB, FR103, 95 nm in diameter; Degussa, Dusseldorf, Germany).

We chose to study industrially-produced MWCNT to obtain results relevant to occupational health, in the context of increasing industrial development and use of CNT.

Taken together, these results show that coating MWCNT with an acidic polymer reduced their specific surface five-fold but enhanced their dispersion in culture medium and their internalization by macrophages both in vitro and in vivo, compared to uncoated MWCNT. These results agree with earlier data 33 showing that the presence of acidic groups such as phenolic or carboxyl groups on the surface of carbon materials facilitates the internalization of these materials by mouse macrophages. The marked internalization of NT2 by mouse macrophages in our study was associated with the induction of a cytotoxic, oxidative, and inflammatory response despite the marked decrease in specific surface, compared to NT1. A similar direct relationship between the degree of internalization and the severity of the inflammatory response was described with metallic nanoparticles 34 . In contrast to the in vitro experiments, in which the oxidative and inflammatory effects of NT1 and NT2 were similar, the in vivo experiments showed stronger oxidative and inflammatory responses to NT2 than to NT1, suggesting that cells other than macrophages were involved in the response to NT2 in vivo. This hypothesis is supported by the smaller size and larger number of NT2 agglomerates in the alveolar wall, suggesting enhanced contact of alveolar cells with NT2 compared to NT1. Accordingly, Shvedova and coworkers 35 demonstrated that exposure of mice to relatively well-dispersed single-walled CNT by inhalation produced a stronger inflammatory response than did pharyngeal aspiration of an agglomerated particle suspension (micrometer-size agglomerates). A stronger inflammatory response to NT2 could result from their acidic surface, in addition to their better internalization. Indeed, we recently obtained data showing that the inflammatory response of murine macrophages exposed to MWCNT generated by catalyst chemical vapor deposition was more marked when acidic groups were present on the MWCNT surface, compared to identically produced MWCNT of similar length that did not have acidic groups on their surface (Cyrill Bussy, unpublished data). Moreover, Hamilton and colleagues 36 recently showed that titanium dioxide nanoparticles induced inflammasome activation and inflammatory cytokine release through a cathepsin B-mediated mechanism involving lysosomal disruption and, therefore, acidification of the intracellular compartment, thus linking acidification to inflammatory response. It should be noted, however, that the inflammatory response to NT2, evaluated based on BAL fluid cellularity and lung gene expression, resolved between 1 week and 1 month post-instillation, which probably explained the absence of persistent histological abnormalities. The absence of fibrosis may also reflect the strain of mice used in this study as compared to others 37 .

In contrast to the acidic polymer, the polystyrene polymer suppressed the cytotoxicity, oxidative stress, and inflammation observed both in vitro and in vivo when the nanotubes were not coated. Several mechanisms may explain this result. First, the coated MWCNT contained 50% polymer by weight and therefore contained half the amount of nanotube material supplied by NT1. To investigate the potential role for this difference, we performed in vivo experiments, comparing the effects of 100 μg of NT1 to those of a "double dose" of NT2 and NT3 (200 μg, thus containing the same amount of MWCNT than 100 μg of NT1). These experiments showed that BAL fluid cellularity 24 h after exposure to 200 μg NT2 or NT3 was similar to that seen with the original dose of 100 μg (data not shown), excluding a role for reduced MWCNT mass in the absence of NT3 toxicity. Second, the specific surface area was smaller with the coated MWCNT than with NT1. However, NT2 and NT3 had similar specific surface areas, yet different toxicities. Third, cell internalization was less marked with NT3 than with the other MWCNT. Similarly, Pan and coworkers 38 showed that coating TiO₂ nanoparticles with a dense grafted polymer brush that exhibited hydrophobic properties, also exhibited by our NT3, prevented adherence to the cell membrane and hence penetration into the cell and induction of oxidative stress. Indeed, cells adhere less well to hydrophobic than to hydrophilic biomaterials 39 . However, the reduced internalization of NT3 was probably not the only mechanism explaining the absence of toxicity of this material, since the in vivo experiments showed similar numbers of alveolar macrophages with MWCNT-containing vacuoles in the NT1 and NT3 groups. Fourth, NT3 may have a decreased ability to induce an oxidative and inflammatory response once internalized, compared to the other two MWCNT. This mechanism is consistent with the presence on the NT3 surface of large numbers of ozone oxidizable groups, compared to the other two MWCNT. These oxidizable groups may act as free radical scavengers 40 thereby decreasing the oxidative stress and consecutive inflammation in NT3-exposed cells and animals. Further studies are needed to determine whether such a mechanism is at play.

We administered MWCNT intratracheally to investigate their effects on the lung. We are aware that inhalation should be preferred over intratracheal instillation for toxicological investigations 41 but we believe our study is relevant to the analysis of the importance of surface characteristics as a determinant of MWCNT toxicity, as we compared the effects of the three materials administered to mice in the same way and using the same experimental set-up. Furthermore, the consistency of the results from our simultaneous in vitro and in vivo investigations supports the validity of our findings.

In conclusion, to the best of our knowledge, this study supplies the first evidence that using polymer-coating is a useful strategy to diminish adverse effects of CNT. Such tool to modify the surface environment of industrially produced CNT may hold promise for improving occupational safety, as well as general user's safety.

The authors declare that they have no competing interests.

LT, CB, JB, and SL designed the study. LT, CB and ASD performed the biological assays. AS and MR performed the physicochemical measurements. JB and SL drafted the manuscript, and CB and MR helped with the final version. All authors read and approved the final manuscript.