The MWCNT examined in the present study were produced by CVD as spherical sets of agglomerates of primary MWCNT particles in the range of several tens to several hundreds of micrometers of diameter. In the present study, the aim was to be as close as possible to an in vivo
respiratory exposure and therefore the CNT were dispersed in DPL, a component of pulmonary surfactant. The issue of the biological relevance of dispersion media in the evaluation of the toxicological effects of manufactured nanomaterials is critical (Fu and Sun 2003
; Monteiro-Riviere, Inman et al. 2005
; Lanone and Boczkowski 2006
) and DPL has been extensively used in studies investigating the respiratory effects of air pollution particles (Bachoual et al. 2007
; Baulig et al. 2003
). Using this method of dispersion it was observed that, in addition to the presence of primary MWCNT particles, MWCNT agglomerates of micrometer size were present in solution based on TEM image analysis, light scattering of the suspension and optical microscopy of exposed cells. The relative contribution of the primary particles to the agglomerates is not known when the particles come out of production. Indeed, it is well documented that individual, raw CNT form agglomerates in solution. Dispersing the MWCNT in the 2 other examined media (PBS ad EtOH), resulted also in similar formation of agglomerates accompanied by the presence of primary MWCNT particles. Although agglomerates in DPL were smaller than in PBS, they were still in the μm range. In DPL two types of aggregates corresponding to a bimodal distribution of 22.5 and 480 μm peak diameter in terms of volume fraction, were observed, whereas in PBS a single broad particle distribution from 10 to 400 μm was recorded that may have coalesced from the bimodal PSD distribution function. It needs to be stressed however, that, irrespective of the dispersing media, not all the MWCNT present in the solution formed agglomerates, since individual CNT were observed by TEM analysis of the suspensions. Light scattering reveals a significant relative number concentration of primary MWCNT particles on the order of 108
particles to one, thereby reaching volume fractions up to 13.6% at a modal diameter of 60 ± 5 nm in pure H2
O after sonication. Sonication experiments in 1% EtOH dispersant reveal the partly reversible nature of aggregation and break-up of the MWCNT aggregates into primary particles, but also the re-agglomeration with time once sonication has been halted.
Irrespective of the dispersing media, MWCNT induced a concentration- and time-dependent decrease in mitochondrial metabolism, as revealed by a significant decrease in MTT reduction, but no effect on membrane permeability, as revealed by Neutral Red assay. Gao et al. (Gao et al. 2001
) showed that silica particles, when mixed with Dipalmitoyl phosphatidylcholine (DPPC), lose their cytotoxic potential on NR8383 rat alveolar macrophages. However, in the same study, the authors also fail to demonstrate such protective effect of surfactant coating for kaolin particles. The absence of effect of dispersion in DPL, as compared to other media, on MWCNT-induced cytotoxicity is in agreement with data published by Monteiro-Riviere and collaborators (Monteiro-Riviere, Inman et al. 2005
) and Wick and collaborators (Wick et al. 2007
) showing that cytotoxicity of MWCNT in human epidermal keratinocytes and mesothelial cells was independent of their dispersion in different concentrations of surfactants. However, Monteiro-Riviere et al. showed that HEK cells exposed to CNT dispersed in surfactant produced less IL-8 than cells exposed to CNT alone (Monteiro-Riviere, Inman et al. 2005
). Therefore, if dispersion in various media, and in surfactant in particular, is an important point to study, biological effects can’t be considered as the results of one single end-point. An effect of MWCNT on mitochondrial metabolism without any alteration in cell membrane integrity, agrees with results published by different groups examining separately these parameters in A549 (Davoren et al. 2007
) and other cell types (Muller et al. 2005
; Wick et al. 2007
). This could be explained by the different intracellular targets/mechanisms of the two assays. The metabolic effect of MWCNT on A549 cells further resulted in a decreased cell number after 48 to 72 hr incubation, as shown by decreased cellular DNA content, without any apoptosis or necrosis. The decrease in cell metabolism and number occurred essentially after 24 hr cell incubation with MWCNT and was stable thereafter, suggesting an initial, non-progressive insult. Despite a similar degree of metabolic effect of MWCNT on MeT5A cells, no decrease of total cell number was observed as well as any sign of apoptosis. Since in both cells types, no compensatory modification of proliferation has been observed, the exact mechanisms implied in these events still need to be discovered. A possibility could be a modification of cell architecture, as described by Kaiser et al. (Kaiser et al. 2008
), that could induce subsequent modification in cell physiology, specific to each cell type, or, as proposed by Fung et al. (Fung et al. 1997
), differing repair capacities. However, these possibilities need further study.
MWCNT internalization in exposed cells was not seen. It is important to note, however, that while large agglomerates may not get internalized, small individual tubes could be (such as those identified by light scattering experiments reported earlier in this study) although they may not be visible at light microscopy level or even with TEM. Few studies have investigated the relation between agglomerate formation and CNT internalization. Monteiro-Riviere and coworkers (Monteiro-Riviere, Nemanich et al. 2005
, Bussy et al. 2008
) showed that the cytotoxic effect of MWCNT agglomerates to human epidermal keratinocytes resulted from CNT internalization. However, Davoren and coworkers (2007)
found no SWCNT internalization in A549 cells, using a different solubilization media than utilized here. Wörle-Knirsch and colleagues (2006)
found SWCNT present in A549 cells, but prior to cell exposure the CNT underwent an acidic treatment that may have modified surface reactivity (Fu and Sun 2003
) and subsequent CNT internalization. Indeed, CNT uptake by cells is currently being discussed and might depend on the state of functionalization and morphology of the material (Bianco, Kostarelos, and Prato 2005
). Collectively, our results and those of Davoren and coworkers (2007)
suggest that CNT, either SW or MW, are not always internalized in A549 cells, despite their differing degrees of agglomeration. Some studies showed that, when comparing different aggregated nanomaterials, one can’t conclude on the relative cytotoxicity induced by those materials from the agglomerate size range (Soto, Garza, and Murr 2007
). Finally, another important issue is that internalization is not an end-point for particles to have a biological effect. Indeed, Ovrevik et al. (Ovrevik et al. 2006
) showed that silica particles induce a cascade of events prior to any internalization of the particles in A549 cells. Therefore, the relationship between agglomeration and internalization, if very interesting, is maybe not the only relevant parameter to understand biological effects of MWCNT.
The effect of our MWCNT on cell metabolism may rely on mechanisms other than cell internalization. In this context, occurrence of oxidative stress, postulated as a central mechanism in the cellular toxic effects of nanoparticles was explored (Nel et al. 2006
), but expression of neither HO-1, a redox-sensitive antioxidant system, nor GPx or SOD in MWCNT-exposed A549 or MeT5A cells was modified. Furthermore, the expression of the pro-oxidant system NOX4, induced by diesel exhaust particles in A549 cells (Amara et al. 2007
), was unchanged in these cells. Such results could be because of the cancerous nature of A549 cells, as cancer cell lines are known to be refractory to oxidant stress. However, along with exposure to particles, we exposed cells to hydrogen peroxide as a positive control for oxidative stress generation, and found an increase in oxidative stress markers (data not shown). Therefore, other non-oxidant mechanisms may also be important to the effects of MWCNT and asbestos. Activation of cell-surface receptors, involving a redox-independent signaling cascade may be an alternative explanation. Interestingly, Ovrevik and coworkers (2006)
recently showed that silica upregulates IL-8 release from A549 cells through interactions with membrane components prior to particle internalization. However, such a mechanism might not occur in our study since, as opposed to silica, which is acidic, our industrially produced MWCNT present few acidic groups at their surface. The oxidable groups present at their surface could potentially act as free radicals scavengers, as shown by Fenoglio and collaborators (2006). However, investigation of such mechanisms requires further study.
As stated previously, data are scarce in the current literature comparing the effects of CNT with those of asbestos fibers. Wick and coworkers (2007)
showed that agglomerates of CNT induced a similar cytotoxic effect as crocidolite fibers in the human mesothelioma cell line MSTO-211H, but the ultrastructural and molecular basis underlying these effects were not established. A recent pilot study, by Poland and coworkers, also showed similar effects of carbon nanotubes and asbestos fibers in mice which mesothelial lining of the body cavity has been exposed, as a surrogate for the mesothelial lining of the chest cavity (Poland et al., 2008). However, the exact molecular pathways were not deply addressed. In the present study, although both MWCNT and asbestos fibers induced similar alterations in viability of A549 and MeT5A cells: 1)
in contrast to MWCNT, asbestos fibers did not diminish cell number but augmented apoptosis, and 2)
MWCNT were not internalized in cells whereas asbestos fibers were clearly internalized. The nature of the effects induced by MWCNT and asbestos fibers were different, although the present study can’t give definitive answers on that issue.
In conclusion, this study shows that MWCNT produced by CVD for industrial purposes as spherical sets of several hundreds of microns exert adverse biological effects without being internalized by human epithelial and mesothelial pulmonary cells.