In recent years, a wide number of potential graphene applications have emerged across different research and innovation fields [3, 7, 8, 48, 49]. The growing interest in this material has led to an increase in its production—and, consequently, in human exposure to it. Many of these applications—e.g., face masks, sensors, and smart clothes—involve daily use and thus continuous exposure [50, 51, 52]. In order to create safe-by-design protocols, it is essential to study how graphene and GRMs interact with different human biological barriers, especially those that will come into direct contact with them [2, 3]. Therefore, assessing how graphene interacts with the respiratory system is essential—especially the interaction with the first chain of defense, the respiratory epithelium. These studies are crucial, for example, for setting occupational exposure limits. On the other hand, it is necessary to establish standardized criteria for this kind of studies [1, 3]. The scientific community must conduct multiple studies, evaluating the potential impact of different GRMs at different doses and exposure times. In addition, it is necessary to define the most appropriate biological model to conduct these studies . Finally, for an adequate assessment of their toxicity, different GRMs should be well-characterized through standardized protocols .
The major potential routes of graphene into the body are inhalation, ingestion, and dermal adsorption . Exposure to graphene is variable during its production process, involving direct interaction with the respiratory tract if adequate personal protective equipment is not used . Concerns about the toxic effect of graphene on the lungs also extend to its integration into everyday products such as face masks  and biomedical applications such as intranasal immunization . Moreover, different studies on the biodistribution of graphene have demonstrated the presence of graphene in the lung after intravenous [56, 57], oral , and intraperitoneal administration [59, 60]. This suggests that the lung could also be damaged when other administration routes are used.
Different studies have evaluated the pulmonary toxicity of graphene in murine models in recent years, with contradictory results [3, 61, 62, 63, 64, 65, 66]. This is because the impact of graphene depends on its different physicochemical characteristics, concentration, and exposure time . Bussy et al. observed recently that GO inhalation could induce lung granulomas that persist up to 90 days after exposition . This suggests that in vivo studies must evaluate its long-term effects. However, this is not very common. On the other hand, the in vivo studies published to date, evaluating different conditions and scenarios, required very large numbers of mice. To ensure the 3Rs principle is followed, as well as reducing costs and time, it is essential to refine the in vivo exposure conditions prior to conducting the experiments, by using standardized in vitro toxicity assessment protocols. However, the choice of cellular models for in vitro study is a crucial issue that should not be taken lightly [68, 69].
In this work, we propose a model using primary normal human bronchial epithelial (NHBE) cells, which have been used previously to study particle-generated lung toxicity [22, 24, 25, 70]. At present, the gold standard to study graphene-induced lung toxicity is the lung tumor cell line A549 [71, 72]. Tumor cell models are cost-efficient, easy to use, and provide an unlimited supply of material. However, they do not have the same characteristics as normal cells, particularly regarding the composition and net charge of the plasma membrane or the oxidative stress response—all of which are critical for interacting with GRMs . Indeed, some studies using A549 cells showed no toxicity after exposure to high doses (≥ 50 µg/mL) of graphene, indicating that this cell line is highly resistant to graphene-induced toxicity [73, 74, 75].
The use of the NHBE model offers, therefore, a more realistic scenario for toxicity assessment. In this work, we have proposed a series of simple and reproducible toxicity determination procedures for identifying variations in cell viability, from slight to acute effects. The results indicate that low doses of different GRMs induced a significant increase in NHBE cell death, an effect not observed in A549 cells (Figs. 2–4). The results obtained in A549 cells were similar to those reported in previous works [44, 63]. Differences were only due to the intrinsic characteristics of tumoral cells A549—i.e., membrane dynamics and resistance to oxidative stress .
However, to avoid underestimating the real impact of GRM-based toxicity on lung cells, in addition to the cell model used it is also crucial to combine different approaches. It is possible that studies published to date quantifying cytotoxicity by classical methods underestimate the real in vitro cytotoxic impact of GRMs. In our study, observed necrosis and apoptosis in cells exposed for seven days (Fig. 2C; Supplementary Fig. 1C) to 5, 50, and 100 µg/mL doses was much higher, since it was related to a very small proportion of surviving cells (Fig. 4). The substantial increase in cell death at seven-day exposures led us to focus our attention on a 24-hour exposure time—which is also the standard exposure time in toxicity studies. Moreover, our study further evaluated other indirect parameters of cell damage, such as alteration in Ca2+ homeostasis and ROS levels. We observed that low doses of GRMs altered these parameters only in NHBE cells (Fig. 5).
Our study assessed the toxicity of three well-characterized GRMs with different lateral sizes and oxidation degrees. Regarding necrosis, 5 and 50 µg/mL GO (more oxidized) generated an immediate and acute increase in this parameter compared to FLG and sFLG, which was maintained over time (Supplementary Fig. 4). On the other hand, the size of the graphene was determinant in cytotoxicity at long times and low doses, as suggested by the high toxicity effect of seven-day sFLG exposure. This difference was not observed at higher doses, since the level of cytotoxicity generated was extremely high. This trend was not observed regarding apoptosis, highlighting again the importance of combining different approaches to assess toxicity in the same study.
It has been fully demonstrated that small particles have a harmful effect on the lung , and graphene is no exception. The toxicity of many of these particles has been studied previously using the NHBE cell line. Therefore, finally, to put our results into context, we compared graphene-induced toxicity levels in NHBE cells with those of other toxic particles analyzed using the same cell model. The toxicity levels induced by 5 µg/mL doses of GO, FLG, and sFLG were comparable to those generated by low doses of toxic compounds such as DEPs  and cigarette smoke extracts . For 50 µg/mL doses (particularly FLG), toxicity levels were similar to those induced by high doses of DEP compounds or electronic cigarette smoke extracts [22, 24, 25]. For example, DEPs are generated by diesel engines, one of the most important sources of anthropogenic particulate matter emissions. These particles generate cytotoxicity in a wide variety of cells, including NHBE [78, 79, 80]. Remarkably, different studies show that exposure to even low doses of these toxic compounds have a detrimental effect on human health [22, 24, 25, 70]. The results obtained in our study allow us to conclude that, for NHBE cells, a 5 µg/mL dose of GRMs (considered as low) generated toxicity after 24 hours of exposure, and a dose of 50 µg/mL was as toxic as higher doses of other, well-studied toxic nanoparticles.