Nanotechnologies represent a major stake of the 21^rst century. They are aimed at conceiving, characterizing and producing materials at the scale of the billionth meter; the nanometer. These nanomaterials (among which are particles with a diameter lesser than 100 nm - nanoparticles, NP) present new properties that make them particularly interesting to industrialists. The list of actual uses and applications for nanomaterials is already substantial, and will certainly become exponential in the next future. Such developments, in spite of potential benefits in numerous domains, i.e. in the field of nanomedicine, are accompanied by concerns regarding the potential consequences of human exposure (in the general population or in an occupational context), with particular threat at the respiratory level 1 2 3 . The need for toxicity evaluation is motivated by numerous studies already available in the literature demonstrating that animal exposure to engineered NP is followed, among other pathological manifestations, by the induction of a pulmonary inflammation, associated with lung remodelling 4 5 6 7 8 . However, so far, there is almost no study assessing the consequences of human exposure to manufactured NP. This point is critical since human exposure studies constitute the cornerstone of risk evaluation for health after (chronic) exposure to environmental agents, especially in the occupational context.

During their everyday occupation, welders are exposed to a complex aerosol of gases (e.g. carbon monoxide, ozone), and hazardous metal fumes composed of chain-like agglomerates of particles, with a primary size in the nanometer size range 9 . Indeed, up to 11% of the total mass, and 80% of the total number of particles emitted in welding fumes are NP 10 11 . Various adverse respiratory outcomes have been described in welders, among which inflammation and lung remodeling are largely described 12 13 14 15 16 . However, the underlying responsible mechanisms remain to be elucidated.

We therefore made the hypothesis that these abnormalities could result, at least partly, from an inflammatory effect directly induced by welding fume-issued NP present in the pulmonary tissue. Such hypothesis has never been studied in the literature so far, probably due to the technical challenges related to the analysis and characterization of NP tissular content in biological samples in situ. We therefore setup an original interdisciplinary study, combining the expertise of material science physicists, physico-chemists, biologists and medical doctors. A broad range of imaging and material science techniques (scanning/transmission electron microscopy (S/TEM), energy dispersive X-ray (EDX), and X-ray microfluorescence (μXRF)), was used to characterize the association between lung histological modifications and the presence of NP in 21 well-documented arc welders and matched unexposed individuals. Since this analysis revealed the presence of NP in macrophages in the alveolar lumen and in areas of fibrous tissue in welders, our physico-chemist partner chemically synthesized NP representative of those identified in pulmonary tissue (Fe₂O₃, Fe₃O₄, MnFe₂O₄ and CrOOH), to investigate their effects (in combination or not with exposure to cigarette smoke, as the majority of patients were smokers) on human macrophages. These exposures were performed in combination or not with exposure to cigarette smoke, as the majority of patients were smokers. Our results show that exposure to welding-related NP induce the production of a pro-inflammatory secretrome (increased production of CXCL-8, IL-1ß, TNF-α, CCL-2, −3, −4, and to a lesser extent IL-6, CCL-7 and −22). Moreover, all but Fe₃O₄ NP induce an increased migration of macrophages.Finally, there was no effect of macrophage pro-inflammatory secretome on human primary lung fibroblasts differentiation. These results provide therefore the first evidence of a link between human exposure to NP and long-term pulmonary effects.

In this study a combination of imaging and material science techniques were utilized to comprehensively scrutinize lung tissue samples, and for the first time thoroughly characterize welding-related NP in situ in lung tissue from welders. Moreover, given the tissue localization and the effects of such NP on macrophages in vitro, the data here strongly suggests that these NP could be responsible at least in part for the pulmonary inflammation observed in welders.

To the best of our knowledge, only a very limited number of studies has been performed in welders, identifying the presence of nanoparticles 19 20 21 22 . However, although using the valuable combined (S)TEM/EDX analysis as developed by Abraham and colleagues 23 24 25 26 , these studies present some major limitations, such as a small number of patient(s) 19 21 , or a very specific exposure (aluminum factory, not representative of the general occupational exposure of welders) 20 . Our study in the other hand is the first to provide persuasive evidence not only of the presence of NP in situ in human lung tissue samples, but also the identification of their chemical composition. As expected given the occupational exposure of the patients, iron was the major element present in excess in the arc welders evaluated in our study. Indeed, if the exact chemical composition of welding fumes is dependent on the material being welded and the electrode, in manual arc welding fumes and in accordance with our findings, the elemental composition is predominantly iron present as metal oxide NP of the form (M, Fe)₃O₄, where M may be substituted for Mn and Cr essentially 9 27 28 29 30 . These results led us to chemically synthesize 4 different NP, representative of those found in the welders lungs; 2 speciation of pure iron oxide (Fe₂O₃ and Fe₃O₄), a mixed Mn/Fe oxide (MnFe₂O₄), and an oxyhydroxide CrOOH NP. We chose to generate both Fe₂O₃ and Fe₃O₄ because, although our data didn’t provide any evidence of the exact speciation of these iron oxides, such species have been largely described in the literature regarding welding fume chemical composition 9 27 30 . Moreover, given the observed co-localization of Mn and Fe signals in our lung samples as well as data from the literature 31 , we chose to generate MnFe₂O₄ NP as a mix metal oxide relevant to welding fume exposure. Finally, we chose to generate CrOOH oxyhydroid NP that present the same + III oxidation state and particle size in the same range as the other NP. Indeed, FeCr2O4 particles can be synthesized but not at the nanoscale, making them of no interest for our study.

We demonstrated that exposure of macrophages to the synthesized representative NP results in the production of a pro-inflammatory secretome, by itself able to further activate the migration of macrophages in vitro. Although we did not conduct experiments to identify the specific protein(s) responsible for the activation of macrophage migration, several candidates present in macrophage pro-inflammatory secretome can be proposed; CCL-2, CCL-3, CXCL-8, as they directly act as efficient monocyte, lymphocyte and neutrophil chemoattractant 32 33 34 . Moreover, it is important to underline that mediators such as TNF-α and IL-1ß, highly secreted by macrophages in response to NP (Figure 4), could also participate in the perpetuation of inflammation initiated by NP exposure since they have been shown to further induce the expression of chemotactic cytokines such as CCL-2, −3 and CXCL-2, thus sustaining the pulmonary increases of neutrophils, lymphocytes, and macrophages 32 35 36 . Similar events have been described in BAL fluid from animals exposed to welding fumes or other dusts 5 13 37 38 39 40 41 42 . These events could happen in welders, given that we were able to detect some of these cytokines in lung tissue sections of the patients analyzed in our study. Not all NP produced the same effects in vitro, although utilized at the same mass concentration; Fe₃O₄ was overall the less reactive NP, particularly as compared to Fe₂O₃, which underlines the importance of NP speciation in their individual effects. Such differences in NP effects could be related to different degrees of solubility. However, observation by TEM of 3 months old NP suspensions revealed that all nanoparticles appear similar to those observed in the initial suspension, except for MnFe₂O₄ nanoparticles that appear slightly amorphous at their surface (data not shown). This most probably indicates a very small or no solubility at all of the nanoparticle of interest, as we should have observed a diminution of the nanoparticle diameter in case of solubility. This is also in accordance with data from literature demonstrating that iron and chromium oxides are very poorly soluble in water 43 44 45 46 47 . Interestingly, we did not observe any synergistic effect between NP when used as a 1:1:1:1 mix, suggesting a common biological mechanism of action. An LPS contamination of our NP could have been responsible for the M1-like phenotype of THP-1 macrophages in vitro 48 . However, the endotoxin content of the NP solutions was evaluated and we didn’t measure any detectable levels of endotoxin, thus ruling out endotoxin contamination (data not shown). Altogether, our data strongly suggests a role for the NP identified in lung tissue samples from welders in the development of their pulmonary inflammation. This could have great implications in other populations of workers occupationally exposed to NP.

As stated previously, the majority of the patients studied here were either current or former smokers which could have been troublesome during the interpretation of the data. Indeed, the presence of NP has been described in cigarette smoke 49 . Therefore, the welding origin of the NP found in welders lungs could be questioned. However, only Al and Si NP were observed in control patients, irrespective of their smoking status, thus ruling out a potential misinterpretation of the findings related to cigarette smoke NP. Moreover, we did not observe any synergistic effect between cigarette smoke and metallic NP exposure in vitro. Although we observed only a limited effect of cigarette smoke exposure alone, this underlines the unique effect of welding-related NP.

NP-loaded macrophages were present not only in the alveolar lumen but also in the fibrous tissue of lung tissue sections from welders. In chronic repetitive injury situations leading to fibrosis development (such as welding fume exposure in an occupational context), it is known that inflammation invariably precedes fibrosis 50 . Since we detected a pro-inflammatory secretome in macrophages exposed to NP, we hypothesized that this secretome could induce fibroblast differentiation into myofibroblasts. Indeed, myofibroblasts represent the major cellular effector of fibrogenic reactions, and fibroblasts have long been considered as the only cell type of origin for myofiboblasts 50 . However, the data here did not indicate any myofibroblastic differentiation. These findings could be explained by the absence of TGF-ß production in NP-exposed macrophage supernatant, as TGF-ß is the main cytokine involved in fibroblasts differentiation 50 . Moreover, one can’t rule out the influence of kinetic aspects of our experimental set-up, where macrophages were exposed for 48 hours, leading to a M1-like pro-inflammatory phenotype, without the possibility of developing a later M2 phenotype, where macrophages could act as the primary effectors of later stages of repair and/or later proliferative and remodeling phases. Beside fibroblasts, fibrocytes and epithelial cells have also been described as sources of myofibroblasts 50 51 . Their response to supernatant from NP-exposed macrophages could have been interesting to evaluate given the presence of specific cytokines such as CCL-2 and CCL-3. Indeed, CCL-2 is a chemoattractant for fibrocytes 52 , and CCL-3 has been involved in bleomycin-induced recruitment of bone marrow-derived macrophages and fibrocytes, and the subsequent development of pulmonary fibrosis 53 . This should deserve further studies.

In this study, thanks to the extensive physico-chemical characterization of lung tissue sample from welders, we evidenced the presence of welding-related NP in situ. Moreover, our data strongly suggest that these NP could participate in the pulmonary alterations observed in welders. These findings provide the first evidence of a link between human exposure to NP and long-term pulmonary effects.

Please refer to online data supplement (Additional file 6) for additional details.

NP: Nanoparticle; (S)TEM: (Scanning) transmission electron microscopy; EDX: Energy dispersive X-ray; μ-XRF: X-ray microfluorescence; XRD: X-ray diffraction.

The authors declare that they have no competing interests.

PA, ASD, JB and SL designed the study. BC performed the face-to-face interview, PA and JCP selected the patients, and PA and FGS performed the histological analysis. MABG performed the mineralogic analyses, and ASD all biological experiments. ASD, BF, JD and SL performed the μXRF experiments, and ASD performed all other imaging experiments. GB and OD synthesized and characterized the NP. PA, ASD and SL drafted the manuscript, and SL and JB were the main additional participants to its further elaboration. All authors read and approved the final manuscript.