Fibrous nanoparticles vary in length, shape, diameter, surface area, density, purity, content of transition metals, porosity and chirality. In aqueous milieu, carbonaceous NM tend to agglomerate and are rarely present as single entities . In particular, aggregation/agglomeration of airborne CNF and CNT was previously reported in a series of field workplace studies [25–27]. In the current study, SWCNT appeared mainly as agglomerated structures composed of SWCNT bundled into ropes ranging 65-150 nm in diameter (Figure 1C). CNF were seen as agglomerates incorporating a few individual fibers with lengths varying from 5 to 30 μm, and widths within 80-160 nm range (Figure 1A). In contrast, asbestos fibers (2-30 μm in length and 0.16-0.8 μm in width) were mostly well dispersed with few detectable agglomerated structures (Figure 1B).
Pulmonary clearance of NM depends critically on their size and shape. Biopersistent, high aspect ratio fibers are recognized as a special hazard to the lungs. However, particle-like agglomerated structures of thinner CNT need to be distinguished from the fiber-like thick-walled and rigid assembly of nanofibers . For fiber-like materials (asbestos, CNF, CNT), translocation to other tissues is determined by their dimensions, with the critical diameter < 0.4 μm and length < 10 μm . The particle/agglomerate size as well as surface chemistry may be a significant factor affecting and/or limiting the recognition/engulfment of NM by alveolar macrophages [29, 30]. Exposure of pulmonary cells (e.g. epithelial cells, macrophages, dendritic cells) to NM/fibers may ultimately lead to a broad variety of responses, ranging from cell damage/death (cytotoxicity) to engagement of intracellular signaling pathways facilitating the release of inflammatory mediators. Inflammation in the lung promotes myofibroblast recruitment and transformation, deposition of fibrin degradation products accelerating collagen production and pulmonary fibrosis. However, the prevailing mechanisms driving the fibrosis may be quite diverse for each particle/fiber. Previously, two distinct SWCNT particle morphologies were seen in SWCNT preparations employed for assessment of effects of respirable CNT: agglomerates and dispersed states . Accordingly, two morphologically distinct responses were detected in the lungs as early as 7 days post exposure. Foci of granulomatous inflammation, including discrete granulomas often surrounded by hypertrophied epithelioid macrophages, were associated with deposition of SWCNT agglomerates. The SWCNT materials were clearly visualized within granulomatous lesions interfacing bundles of fibrous connective tissue. In lung regions distant from observed SWCNT agglomerates, morphological alterations were predominantly comprised of diffuse interstitial fibrosis with alveolar wall thickening. This interstitial fibrosis occurred at sites of deposition of more dispersed SWCNT structures . Importantly, deposition of collagen and elastin was also observed in both granulomatous regions as well as in the alveolar walls . In the current study, asbestos fibers did not form agglomerates in either of the aqueous preparation, (Figure 1B) or as deposited within the lungs of exposed animals (data not shown), while CNF fibers formed loosely packed agglomerates in both suspensions and in the lungs. The stiffness/rigidity of CNT or CNF could certainly have an impact on the agglomeration propensity and interactions of these materials with biological systems, and this has been the focus of several recent studies [32, 33]. In the present study, we demonstrated that in an aqueous environment SWCNT and CNF tend to agglomerate, and such agglomerates would no longer obey the rules of non-agglomerating asbestos fibers. Therefore, interactions of agglomerated CNP with biological systems would be defined by the relative proportions of individual fibers vs. agglomerates present. In particular, no granuloma formation was found following exposure to fiber-like CNF particles/agglomerates or asbestos (Figure 9B, C). Therefore, granulomatous lesions formed after SWCNT exposure may be attributed to specific scaffolding features of SWCNT agglomerates. Here we demonstrate that SWCNT agglomerates induce granuloma formation, leading to morphological/structural isolation of SWCNT agglomerates within the lung, presumably making them less damaging to the surrounding pulmonary tissues. Agglomerated SWCNT in the lungs, once walled off by cuboidal cells, are less likely to cause acute inflammatory reactions. Thus, relatively rapid isolation of SWCNT aggregates/agglomerates within granulomas, not observed upon exposure to CNF or asbestos, may contribute to faster resolution of acute SWCNT-induced neutrophilic inflammation and pneumonia (Figure 3A). In addition, the potency for induction of alveolar interstitial fibrosis was as follows: SWCNT > CNF = asbestos (Figure 8A). Asbestos fibers are known to induce "frustrated phagocytosis" causing prolonged oxidative stress. However, no frustrated phagocytosis was seen following exposure of murine or human pulmonary phagocytes to SWCNT [30, 34]. The relatively large calculated effective surface area of SWCNT (138 m2/g vs 21 and 8.3 m2/g for CNF and asbestos, respectively) aids to the adhesion of cell/tissue proteins to the surface of SWCNT [30, 35]. In particular, the ability of SWCNT to serve as a scaffold is beneficial for the adhesion and proliferation of fibroblasts in the lungs and may be essential for their sturdy fibrogenic potential . The unique surface structure of the SWCNT agglomerates and potential affinity to lipid and protein covalent binding and coating provides excellent environment facilitating growth and proliferation of fibroblasts 
. It is noteworthy that CNF utilized in the current study share several physical properties with MWCNT, such as a relatively large diameter (as compared to SWCNT) which may contribute to higher stiffness, less "tangling" and lower agglomeration propensity for CNF as compared to SWCNT. Recently, MWCNT have been reported to elicit "asbestos-like" pathogenicity in rodent models, including lung injury and mesothelioma formation [32, 33]. In the current study, we showed that the sub-acute inflammatory, immunologic and fibrogenic outcomes of pulmonary exposure to CNF are similar to asbestos. The carcinogenic potential of CNF as well as other relatively long-term outcomes is a matter for future investigations. Our data showed that agglomerated SWCNT and CNF do not behave as single, fibrous entities but rather as agglomerated particles, and subsequently do not follow the HARN paradigm. As demonstrated by Wang et al. , SWCNT directly stimulate fibroblast proliferation and collagen production in a cell culture system - in line with the known fact that lung fibroblasts like to grow upon SWCNT. This effect does not involve frustrated phagocytosis, as macrophages were not present in the system, but appears to involve the activation of matrix metalloproteinases (MMPs). It was recently shown that up-regulation of MMP-12 and cathespin K by SWCNT in co-culture of epithelial/mesenchymal lung cells and BAL macrophages was due to cell-type specific interactions . The mechanisms of MMPs activation in response to SWCNT - known to cause the formation of irreversible interstitial fibrosis with airway alteration and changes in pulmonary functions found in mice [10, 40, 41] - resembled those that play a pivotal role in the pathogenesis of idiopathic fibrosis and obstructive airway disease in humans ."
Along with shape, size and structure, chemical composition of NM may also contribute to the inflammatory and toxic outcomes. SWCNT and CNF are produced predominantly by HiPco, chemical vapor deposition, laser ablation and arc discharge techniques involving utilization of various transition metal catalysts . Catalytically competent metal-containing NM may synergistically enhance oxidative stress damaging the cells and tissues . Kagan et al.,  was one of the first to document - using EPR spectroscopy of ascorbate radicals as well as adducts with a spin-trap, 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) - the production of hydroxyl radicals generated by iron admixtures in unpurified SWCNT as well as the suppression of these signals by an iron chelator, desferroxamine. SWCNT, CNF and asbestos employed in the current study were found to have 0.23, 1.4, and 18% iron, respectively. Iron content of unpurified SWCNT was previously implicated in enhanced oxidative stress, depletion of antioxidant reserves and accumulation of lipid peroxidation products after SWCNT exposures. Neither Fe-rich (unpurified) nor purified SWCNT were able to induce intracellular production of superoxide or nitric oxide by RAW 264.7 macrophages . Instead, extracellularly generated highly reactive hydroxyl radicals, particularly in the presence of Fe-rich SWCNT, were reported to enhance the oxidative burst and cause oxidative stress via extracellular oxidation . One should, however, keep in mind that SWCNT and CNF synthesized by high-pressure CO conversion (HipCo) or CVD methodology results in accumulation of mostly elemental iron embedded in the core crystalline structure of CNP and thereby not readily mobilized in water in ionic redox-active form. It is likely that uptake by professional phagocytes whereby SWCNT may be localized in acidic environments of phago-lysosomes will lead to the release of ionic iron, hence cause the toxic effects. This secondarily ionized form of iron may be involved in redox-cycling mechanisms and facilitate the development of oxidative stress. However, the significance of this oxidative stress for triggering non-specific peroxidation reactions is likely limited. In fact, mass-spectrometry based global lipidomics analysis of pulmonary lipid peroxidation after the inhalation exposure of mice to iron-rich SWCNT revealed a highly selective non-random pattern of phospholipid peroxidation whereby only anionic phospholipids cardiolipin, phosphatidylserine and phosphatidylinositol underwend oxidative modification . These data are in sharp contrast with the expected random profile of peroxidation of the most abundant polyunsaturated species of phosphatidylcholine and phosphatidylethanolamine observed during non-selective transition metal-catalyzed peroxidation of phospholipids in tissues. In the current study, exposure to SWCNT, CNF and asbestos resulted in increased accumulation of biomarkers of oxidative stress, e.g. 4-hydroxynonenal (4-HNE) and protein carbonyls found in mouse lungs (Figure 5). Elevated levels of 4-HNE were found on days 1, 7 and 28 in SWCNT or CNF -treated animals, in contrast to the peak seen on day 28 in asbestos-treated mice (28 d). The most prominent induction of oxidative stress (4-HNE and protein carbonyls) occurred after SWCNT-exposure. We speculate that the relatively high surface area of carbonaceous NM facilitates the efficient interactions of catalytically active Fe with cellular components and pulmonary tissues; thereby explaining why SWCNT elicited the most pronounced oxidative stress, cell damage, granulomatous inflammation and fibrosis.
Inflammatory milieu in the lung launches a wide variety of signaling events engaging innate immunity and governing systemic/adaptive immune response. The pulmonary innate immune system provides rapid recognition of inhaled agents while orchestrating defensive responses. Exposure to airborne NM could engage pulmonary innate immunity at many levels. A number of recent publications have reported the effects of carbonaceous NM on the immune system [45–47]. It has been shown that splenic T cell dysfunction and impaired systemic immunity was associated with release of TGF-β and subsequent expression of IL-10 and PGE2 in the spleen . Here, we observed that SWCNT elicited the most prominent release of TGF-β as compared to CNF and asbestos. Increased TGF-β found in BAL on day 7 post exposure to SWCNT was accompanied by slightly suppressed spleen T cell proliferation. At this time-point, release of TGF-β in response to CNF and asbestos was significantly lower as compared to that of SWCNT. Exposure to CNF or asbestos, in contrast to SWCNT, did not suppress T cell proliferation on day 7. Surprisingly, a slight stimulation of the spleen T cell responsiveness was observed on day 7 in animals exposed to asbestos (Figure 10). This stimulation may be partially attributable to a marked increase in IL-12 in the lung (Figure 6). However, 28 days post exposure, spleen T cell proliferation was suppressed in both CNF and asbestos treated animals, while pulmonary levels of TGF- β were not markedly changed. These data suggest that splenic T cell suppression at later time points (28 days) is not likely due to TGF- β release. CNF appears to have effects similar to asbestos causing a "delayed" immune suppression, which occurred when the acute inflammation was resolved. It has been reported that asbestos-related immune suppression followed 3 and 6 months after asbestos instillation . One could expect that SWCNT with the highest surface area (138 m2/g) would elicit the strongest acute inflammation and release of TGF-β, as compared to CNF (21 m2/g) and asbestos (8.3 m2/g). At equivalent mass doses, CNF and asbestos are generally less capable of inducing TGF-β release in the lung; therefore, the peripheral tolerance/suppression observed is most likely driven by different mechanisms possibly involving suppressive antigen presenting cells (APC)  and regulatory T cell induction.
In order to address whether specific surface area and/or particle number derived from toxicological studies is useful as a dose metric for hazard identification and risk assessment, we attempted to correlate the inflammatory (PMN) responses observed in the lungs of mice to either specific surface area (measured by BET) or number of particles/agglomerates in the given amount (mass) of NM (Figure 11A). The dose of SWCNT given to animals (40 μg/mouse) was equivalent to the specific surface area of SWCNT of 4.16 × 10-2 m2/mouse, while CNF and asbestos doses (120 μg/mouse) were equal to 5.4 × 10-3 m2/mouse and 9.6 × 10-4 m2/mouse, respectively. The data presented in Figure 11 indicate that although the specific surface area of SWCNT (measured by BET) given to mice was 43 times higher as compared to asbestos, PMN counts in BAL fluid of mice exposed to SWCNT were only 5.7 fold higher (1 day post exposure). Moreover, on day 7 post exposure PMN counts in animals exposed to asbestos were 12.2 fold higher as compared to SWCNT. Accordingly, particle number alone does not seem to be a reliable factor in dose metrics for assessment of NM exposure outcomes. At the same concentration, SWCNT and CNF suspensions had a much lower number of particles/structures due to agglomeration as compared to non-agglomerated asbestos (Figure 1D); however, the neutrophilic infiltration in the lung of animals exposed to SWCNT and CNF was greater as compared to asbestos-exposed mice (24 h post exposure). These data suggest that specific surface area (measured by BET) or particle/agglomerate numbers do not provide a reliable basis for predicting biological outcomes of exposure to carbonaceous NM and is therefore not an efficient dose metric for the assessment of pulmonary outcomes in response to agglomerating fibrous NM.
Indeed, PMN counts in BAL fluid of mice exposed to SWCNT or CNF were increased by ~5.7 and ~2.8 fold, respectively, as compared to asbestos. Of note, the calculated effective surface areas of SWCNT and CNF agglomerates delivered to the lung were ~5.8- and ~2.6- fold higher (vs. asbestos, respectively, day 1 post exposure). Correlations between various pulmonary outcomes and effective surface area of NM administered to the animals are presented on Figure 11. Protein levels in BAL fluid of mice exposed to NM (day 1 post exposure) were well correlated with the calculated effective surface area of particle agglomerates given to animals: Pearson's correlation coefficient was 0.997, p < 0.05. Additionally, our data suggests that the high aspect ratio nanoparticle (HARN) paradigm  is not fully applicable for the assessment of the hazardous effects of carbonaceous fibrous NM. Therefore, in addition to mass dose, the effective surface area of NM structures should be experimentally determined by detailed analysis of NM agglomerates. Effective surface area of NP agglomerates may be useful as a predictive dose metric of pulmonary toxicity - acute inflammation, pulmonary damage and fibrosis - induced by SWCNT or CNF and could thus be utilized for health hazard and risk assessment of fibrous carbonaceous NM.