Taken together, our results show the progressive and dose-dependent development of lung fibrosis in mice exposed to CeO2 NP. They also demonstrate that the blockade of macrophagic autophagy protects from alveolar but not bronchiolar fibrosis, probably via the modulation of macrophage polarization in favor of a M2 phenotype.
The progressive and dose-dependent induction of pulmonary fibrosis that we describe here in response to CeO2 NP single administration is in accordance with data from literature using other CeO2 NP (1, 5–9). The same is true for the sustained pulmonary inflammation that we detected and which has been shown to be slow to resolve post exposure (16). We didn't however observe the formation of granuloma as described for example by Park and colleagues in CeO2-exposed animals (8), but this is probably linked to the more relevant dose of CeO2 NP that we used in our study; 5–50 µg per mouse versus 3 mg in Park study. Our results also confirm and extend the data obtained by Ma and colleagues, showing that rats exposure to CeO2 NP lead to the induction of M1 phenotype in BAL macrophages of these animals (17). It must be noted however that 28 days after the initial administration, mRNA expression of Arginase 1 was increased in Park study, thus suggesting a shift toward M2 phenotype that was not observed in our WT animals, although care must be taken as only one single M2 marker was targeted in Park's work.
The development of bronchiolar fibrosis in response to CeO2 NP was not prevented by the blockade of macrophagic autophagy. Interestingly, this could be paralleled by the very low Ce elemental signal observed by µXRF in bronchiolar regions, as well as the almost lack of macrophages present in these areas, suggesting that CeO2-induced bronchiolar fibrosis is independent of macrophages. Although we did not investigate further the biological mechanism(s) underlying this bronchiolar fibrosis, the occurrence of epithelial-mesenchymal transition (EMT) could be an explanation. Indeed, Ma and colleagues have recently demonstrated that exposure to CeO2 NP induces EMT in alveolar type II cells that ultimately plays a role in lung fibrosis (5). Similarly to what we have demonstrated with in vitro exposure of fibroblast to the supernatant of Ce-exposed macrophages, it could be interesting to explore the effect of the secretome from Ce-exposed epithelial cell on fibroblast to myofibroblast differentiation in our system.
The protection against alveolar fibrogenesis observed in Ce-exposed Atg5+/− mice could have been the result of decreased amounts of Ce present in alveolar regions of Atg5+/− mice. Indeed, autophagy is known to interplay with macrophage phagocytosis (18–20), and M2 polarization has been shown to enhance Si-NP uptake by macrophages (21). Although we did not strictly quantify the Ce content in WT and Atg5+/− mice exposed to CeO2 NP, the similar µXRF signals obtained in both animal genotypes in terms of both intensity and tissue distribution, together with the similar total number of macrophages present in the lungs of WT and Atg5+/− animals strongly suggest that the protection against alveolar fibrosis observed in Atg5+/− animals is probably not the result of a decreased amount of CeO2 NP in these individuals. CeO2 is highly reactive, and it has been suggested that a change in the Ce3+/Ce4+ ratio may play a significant role in toxicity determination (22), and could thus contribute to lung fibrosis development. In our experiment, while Ce in CeO2 NP was Ce4+, Ce in lung tissue was only 33–36% pure Ce4+, indicating a change in speciation after NP administration. However, both WT and Atg5+/− mice showed a similar modification of their Ce speciation, with 45–48% as Ce3+, and around 18% presenting a mix Ce3+/Ce4+ form. Therefore, the specific protection against alveolar fibrosis observed in Atg5+/− mice could not be attributed to modifications of Ce speciation.
The modification of macrophage polarization related to their autophagy status could also represent an interesting candidate to explore the underlying mechanism that occurs in Atg5+/− mice (23). Indeed, macrophages are divided into two distinct sub-populations defined as classically activated pro-inflammatory M1 subtype and alternatively activated M2 subtype responsible for anti-inflammatory, tissue repair and remodeling (Novak and Koh 2013). Both M1 and M2 macrophages have been noted to be involved in the pathogenesis of pulmonary fibrosis, this translating the plasticity of macrophage polarization depending the micro-environment stimuli and signals (25). In our experimental model of lung fibrosis, we demonstrated that macrophages from autophagy-deficient mice tended to polarize into M2a and M2c phenotype whereas their wild-type counterparts exhibiting proficient autophagy skewed towards the M1 subtype. Interestingly, the induction of autophagy by advanced glycation end products or rapamycin triggered macrophage polarization toward M1 phenotype, as well as a sustained inflammation in mice and patients, resulting in delayed wound healing. Moreover, the inhibition of autophagy reduced M1 population but no changes of M2 subtype, which is in accordance with our results (26). However, it must be noted that in mice under high fat diet, there was an increased M1 and decreased M2 polarization in macrophages with defective autophagy, leading to hepatic inflammation and the progression to liver injury (27). This underlines the importance of the environmental context in the overall autophagy effect, which could be a clue to the effects observed in our study, as to understand the underlying mechanism of macrophage specific polarization in absence of autophagy.
Finally, our results suggest that macrophagic autophagy can facilitate fibrosis. Although not in accordance with the initial acceptance that autophagy might be protective in lung fibrosis (11, 13, 28), our data are in line with more recent results from the literature showing an increased number of LC3-II puncta, an index of active autophagy, in fibroblast foci of patients with idiopathic pulmonary fibrosis (IPF), the most common form of pulmonary fibrosis (13, 29). In vitro, these authors also showed that autophagy is induced by TGFß, and that it is necessary for TGFß-induced fibrosis in both non-IPF and IPF fibroblasts (7). Interestingly, such a cell-specific role for autophagy has been recently described in chronic obstructive pulmonary disease (COPD) where autophagy is believed to be protective except in macrophages (10–12, 28). In lung fibrosis, the majority of the studies has focused on lung epithelial cells and/or fibroblasts (28), as the two major cell type involved in fibrogenesis (15). However, the recruitment of inflammatory cells, leading to the activation of effector cells, is often considered as a first trigger of fibrosis (30), and as such, the activation of macrophages occupies a pivotal role in the translation of injury to aberrant repair in lung fibrosis (31, 32). Interestingly, the LC3 puncta observed by Ghavami and colleagues in fibroblast foci of IPF patients are coherent with the presence of macrophages (29).