We investigated CBNP-mediated DNA damage and inflammatory responses in Printex 90 instilled mice using multiple doses (i.e., 0.018, 0.054 and 0.162 mg) and post-exposure time-points (i.e., 1, 3 and 28 days), alongside sham controls. These doses equal the pulmonary deposition after 1, 3 and 9 working days for a mouse at the occupational exposure limit of 3.5 mg/m3 CB per 8 hour work shift (as established by the Occupational Safety and Health Administration (OSHA) and the National Institute for Occupational Safety and Health (NIOSH)). These calculations assume that 33.8% of the inhaled mass ends up in the pulmonary region  with a volume of inhaled air per hour of 1.8 L/h  and 8 h working days. CB exposure levels of up to 3.7 mg/m3, 2.2 mg/m3 and 4.2 mg/m3 have been reported for workers implicated in packaging and in handling CB [37–39]. BAL cell influxes revealed CBNP-induced inflammation that peaked 1 and 3 days post-exposure and persisted 28 days thereafter (Table 1). SB were detected in lung and liver tissues and isolated BAL cells relative to controls and persisted to day 28 (Figures 3, 4A and 5). FPG sensitive sites in lung were increased throughout with significant increases occurring on post-exposure days 1 and 3 in comparison to controls (Figure 4B). The expression of Saa3 mRNA in lung tissue was increased at all time-points. This reflects a persistent acute phase response during the entire experiment (Table 2). The correlation between total BAL cells and FPG sensitive sites across all doses and time-points suggests an important role for BAL cell influxes in generation of oxidative stress in the lungs. This study is the first to demonstrate in vivo low dose CBNP-induced SB in lung and liver, which persist 28 days after a single exposure.
Although the primary particle size of Printex 90 (as dry powder) was stated by the supplier to be 14 nm, CBNP suspensions were shown to be highly agglomerated by DLS. When analyzing the unfiltered suspension we were only able to detect particles with a peak size of 2.6 μm. However, this is an artefact of the DLS methodology in which larger agglomerated particles will overshadow smaller particles. Consequently, DLS on unfiltered samples can only detect the largest particles in the suspension. Therefore, we filtered the suspension to remove the largest particles in order to be able to detect the smaller particles. DLS on the filtered suspension confirmed the presence of smaller particles as well (peak size of 190 nm and range of 90-350 nm). This shows that some of the particles in the suspension used for instillation are nano-sized, but the DLS analysis is unable to quantify the proportion of nano-sized particles in the suspension.
As in other studies of NP toxicity, endotoxin present on the NPs can result in inflammatory responses. However, the very low level of endotoxin detected on the CBNPs (0.142 EU/mg CBNP) is not anticipated to result in any significant inflammatory response. In another study, the response of LPS instillation was investigated in mice . Instillation of a high dose of LPS (100 μg = 1200 000 EU) resulted in a high inflammatory response, while a low dose of LPS (0.1 μg = 1200 EU) resulted in a low inflammatory response. We have instilled doses ranging from 0.003 EU (0.142 EU/mg × 0.018 mg) to 0.02 EU (0.142 EU/mg × 0.162 mg). The ensuing doses of endotoxin administered in our study are therefore several orders of magnitude below what is considered a low dose of LPS, and thus are expected to result in very little (if any) endotoxin induced inflammation.
SB in lung and BAL cells demonstrate CBNP-induced DNA damage at the site of exposure (Figures 3 and 4A). Although CBNP-induced DNA SB have been investigated in vitro
[30, 31, 33, 34], only one study has investigated SB in mouse lung in vivo, and this work used a single high dose (i.e., 0.2 mg instillation) . In contrast, we demonstrate that SB in lung occur at much lower doses of exposure (e.g., as low as 0.018 mg). Our BAL findings corroborate two earlier studies demonstrating CBNP-induced SB in C57BL/6 and tumour necrosis factor (TNF) deficient mice exposed by inhalation (i.e., 20 mg/m3, 90 min/day, 4 days) , and apolipoprotein E (ApoE) deficient and C57BL/6 mice exposed by instillation (i.e., 0.054 mg) . Thus, our results indicate that SB result from exposure levels that are much lower than previously reported.
The aforementioned in vivo works investigated pulmonary and BAL SB shortly after CBNP exposure, using recovery time-points ranging from 1 to 24 hours [21, 30, 33]. Taking into consideration that SB are readily repaired within a few hours of induction , we aimed to establish whether de novo DNA SB could be induced at much later post-exposure time-points (e.g., 3 and 28 days). Interestingly, we observed SB at all post-exposure time-points for the mice exposed to the high dose relative to controls in both lung and BAL cells, and these effects often extended to lower dose groups (Figures 3 and 4A). As such, we demonstrate that CBNPs can generate pulmonary and BAL SB even 28 days after a single exposure. In addition, we demonstrate that assessing SB in cells may serve as a potential biomarker of pulmonary genotoxicity in human studies.
Several mechanisms are hypothesized to contribute to CBNP-induced genotoxicity. It has been suggested that CB may be genotoxic because of adsorbed surface compounds, primarily polycyclic aromatic hydrocarbons (PAHs). However, because of the low levels of PAHs found on the CBNP surface (e.g., 72.4 ng/g in the Printex 90 used in this work) and the high affinity of these compounds for the CB particle core, CBNP-induced toxicity is not likely mediated via PAH exposure [26, 27, 42]. Alternatively, CBNPs have been identified as potent ROS generators [14–16] and are also associated with large inflammatory responses, as established by our BAL cell profiles and by the work of others [17–24]. Pulmonary influxes in BAL cells induce ROS production at sites of acute inflammation as a result of degranulation and oxidative burst mechanisms . More importantly, the mutation spectrum arising in vitro following CBNP exposure points to ROS as the primary mutagenic factor , providing substantial evidence that ROS-induced DNA damage occurs upon CBNP exposure. As such, we speculate that oxidative stress could be an important parameter of toxicity requiring further investigation in vivo.
It has been established that oxidative damage to DNA leads to increased mutation frequencies, and there is increasing evidence for a direct association between levels of oxidatively damaged guanine lesions and increased risks of lung cancer in humans [44, 45]. In contrast to BAL cells that are generated from hematopoietic stem cells, mutations in lung can result in permanent genetic changes within tissue and therefore increased cancer risk. As such, we examined lung tissue directly for oxidative damage to DNA. FPG sensitive sites were quantified using the alkaline comet assay, by which oxidized purines, primarily 8-oxo-7, 8-dihydro-2'-deoxyguanosine (8-oxodG) and 2, 6-diamino-4-hydroxy-5-formamidopyrimidine are recognized. 8-oxodG is the most commonly oxidized lesion and is pro-mutagenic due to its ability to base-pair with adenine, resulting in G to T transversions . Previous in vitro work has revealed non-significant increases in the levels of FPG sensitive sites in human Caco-2 cells exposed to CB (20 μg/cm2, 4 h exposure) , whereas significant increases in FPG sensitive sites were found in an immortalized Muta™Mouse lung epithelial cell line (11.3 μg/cm2 or 75 μg/ml, 3 h exposure) . Here, we demonstrate that CBNPs induce FPG sensitive sites in vivo in the lungs. This is consistent with a previous inhalation study in rats where an elevated dose of Printex 90 (i.e. more than 5 mg retained in the lungs) caused increased 8-oxodG, which persisted for 44 weeks . No effects were observed for lower doses. Considering differences between species (e.g., approximate weights of 20 and 200 g and pulmonary surface area of 82.2 and 300 cm2 in mouse and rat respectively [48, 49]) this 5 mg retained dose in our model would translate to instilled doses of 0.5 mg or 1.4 mg according to body weight and lung surface area, which is well above our highest exposure dose. However, levels of 8-oxodG can often be elevated by spurious oxidation  and thus it is possible that effects may have been detected at lower dose levels with lower background levels. A high dose (i.e., 0.050 mg × 6) of Printex 90 repeated six times in six weeks also caused increased immunostaining for 8-oxodG in the lungs of mice . Thus, in the present study we observed increased FPG sensitive sites at much lower CBNP doses (i.e., 0.018 and 0.054 mg) and at later post-exposure recovery time-points relative to previous studies, with significant increases occurring on post-exposure days 1 and 3 (Figure 4B).
The increase in possible oxidatively damaged DNA observed in our study in comparison to controls was much higher than previous in vitro observations. For example,  showed a less than two fold increase over controls in lung epithelial cells, following 3 h of exposure to 75 μg/ml CBNPs (a dose of approximately 11.3 μg/cm2 cells compared to 0.2-2.0 μg/cm2 in the current work ). This might be because of the presence of ROS-producing granulocytes in the mouse lung, as indicated by the close correlation of PMN cells and FPG sensitive sites found. In keeping with this observation, rats exposed by intratracheal administration of 0.64 mg/kg body weight (i.e., our low dose is approximately 0.9 mg/kg body weight in mouse), developed neither inflammation in terms of PMN infiltration nor increased levels of 8-oxodG in lung parenchyma . We speculate that the pulmonary genotoxocity observed in our work is most likely related to oxidative stress mediated by inflammatory cells, which is in accordance with previous work that has shown CBNP-induced mutation frequency increases only after inflammation is established . On the other hand, we have previously found that SB and pro-inflammatory effects occurred independently of each other in vitro
 and in BAL cells in vivo
[21, 34, 54, 55]. In the present study, inflammation and BAL cell SB were observed at all doses and time points, and thus, we are not able to determine if the observed SB are caused by inflammation or whether inflammation and genotoxicity occur independently of each other.
Large inflammatory responses, such as the ones observed in our work, are associated with systemic effects of exposure due to increases in circulating inflammatory cells (e.g., PMN) and molecular mediators of inflammation (e.g., pro-inflammatory cytokines and chemokines). Additionally, NPs themselves have been demonstrated to translocate into systemic circulation  and spark generated ultrafine carbon NPs have been shown to accumulate in liver of rats upon inhalation . Likewise, individual exposure to traffic related air pollution at the home address and by occupation has been associated with an increased risk of hepatic cancer [57, 58]. In order to investigate the possibility of adverse effects in extra-pulmonary tissues, we used the alkaline comet assay to quantify SB in liver tissue of the same mice. We found that SB were elevated on post-exposure days 1 and 28 (Figure 5). The reason for the lack of damage on day 3 remains unclear. It is possible that two separate mechanisms may come into play at different times (e.g., direct instillation effects on day 1 vs. particle relocation to liver or persistent inflammation on day 28), thus not affecting this post-exposure time-point. The results on particle-induced hepatic DNA damage are consistent with our recently published study on DNA damage in mice exposed to Printex 90 by inhalation . Two different mechanisms are hypothesized to be responsible for the observed hepatic DNA damage: 1) direct particle-mediated effect caused by particles translocated from the lungs to the systemic circulation, and 2) indirect effects related to systemic inflammation. To investigate if the hepatic DNA damage resulted from a systemic acute phase response, we measured the mRNA expression of Saa3 in pulmonary and hepatic tissue. As we reported before with mice exposed to CB or diesel exhaust particles by inhalation , we did not detect a change in Saa3 mRNA in the liver in the current study. In contrast, we found a large increase in pulmonary Saa3 mRNA expression. SAA protein can be detected in circulation during pulmonary inflammation and may induce a systemic acute phase response . We have recently reported that the acute phase response is also induced in the lungs of mice exposed to nanotitanium dioxide where increased levels of SAA protein was also detected in lung tissue . We do not know whether the hepatic effects are caused by inflammation or direct effects of translocated particles, but since translocation of the CBNPs is expected to be low, and particles accumulate primarily in Kupffer cells in the liver , the observed hepatic effects are most likely caused by inflammation. To our knowledge, our work is the first to demonstrate that hepatic SB occur as a result of exposure to CBNPs via intratracheal instillation.
There is increasing evidence that both DNA SB and oxidative damage to DNA can lead to increased risk of tumorigenesis and carcinogenesis [63–66]. Likewise, the relationship linking chronic inflammation to increase risk of carcinogenic outcome is increasingly well supported . As such, it is likely that regular exposure to CBNPs is involved in adverse health outcomes that include cancer. Although previous epidemiological data have linked CBNP exposures to pulmonary carcinogenesis [68–70], evidence to date is insufficient to classify these particles as human carcinogens (currently classified as possibly carcinogenic to humans under Group 2B) . As such, further investigations should elucidate additional underlying mechanisms of toxicity induced by CBNPs and their relationship with disease development. This should include simultaneous investigations of oxidative stress, genotoxicicty and inflammation in lung and liver of exposed mice using multiple endpoints (e.g., micronucleus assay, mutation analysis and GSH levels) and in repeated chronic daily inhalation studies to more closely mimic human exposures. Furthermore, as effects were observed at lower doses of exposure, dose-response relationships should be examined to establish the lowest observable effects and no observable effects levels. In addition to mechanistic studies, future works should clearly establish current human exposure levels in order to evaluate the risks of adverse health effects in individuals routinely exposed to CBNPs.