FULL PAPER www.advancedscience.com Size-Dependent Pulmonary Impact of Thin Graphene Oxide Sheets in Mice: Toward Safe-by-Design Artur Filipe Rodrigues, Leon Newman, Dhifaf Jasim, Sourav P. Mukherjee, Jun Wang, Isabella A. Vacchi, Cécilia Ménard-Moyon, Alberto Bianco, Bengt Fadeel,* Kostas Kostarelos,* and Cyrill Bussy* 1. Introduction Safety assessment of graphene-based materials (GBMs) including graphene oxide (GO) is essential for their safe use across many sectors of society. In Graphene and other two dimensional (2D) materials have generated substantial particular, the link between specific material properties and biological effects commercial interest due to their unique needs to be further elucidated. Here, the effects of lateral dimensions of GO combination of physicochemical proper- sheets in acute and chronic pulmonary responses after single intranasal ties that are attractive for a wide range of instillation in mice are compared. Micrometer-sized GO induces stronger applications.[1,2] Several methods of large- pulmonary inflammation than nanometer-sized GO, despite reduced scale production have been developed since the first isolation of graphene, yielding a translocation to the lungs. Genome-wide RNA sequencing also reveals range of different forms of graphene-based distinct size-dependent effects of GO, in agreement with the histopathological materials (GBMs) whose properties such results. Although large GO, but not the smallest GO, triggers the formation of as lateral dimensions, thickness, and sur- granulomas that persists for up to 90 days, no pulmonary fibrosis is observed. face chemistry can vary significantly.[3,4] These latter results can be partly explained by Raman imaging, which In particular, the low production cost of evidences the progressive biotransformation of GO into less graphitic graphene oxide (GO) has contributed to the emergence of various commercial structures. The findings demonstrate that lateral dimensions play a products based on GO-based composites, fundamental role in the pulmonary response to GO, and suggest that airborne dispersions, or spray coatings.[5] As a result exposure to micrometer-sized GO should be avoided in the production plant of this sustained commercial expansion, or applications, where aerosolized dispersions are likely to occur. These it is expected that exposure to GBMs via results are important toward the implementation of a safer-by-design inhalation of aerosolized materials will increase, raising concerns about the po- approach for GBM products and applications, for the benefit of workers and tential impact of inhaled GBMs on human end-users. health.[6,7] Dr. A. F. Rodrigues, Dr. L. Newman, Dr. D. Jasim, Prof. K. Kostarelos, Dr. A. F. Rodrigues, Dr. C. Bussy Dr. C. Bussy Lydia Becker Institute of Immunology and Inflammation Nanomedicine Lab School of Health Sciences Faculty of Biology, Medicine and Health University of Manchester University of Manchester Manchester Academic Health Science Centre Manchester Academic Health Science Centre Manchester M13 9PT, UK Manchester M13 9PT, UK Dr. S. P. Mukherjee, Prof. B. Fadeel E-mail: kostas.kostarelos@manchester.ac.uk; Nanosafety & Nanomedicine Laboratory cyrill.bussy@manchester.ac.uk Institute of Environmental Medicine Dr. A. F. Rodrigues, Dr. L. Newman, Dr. D. Jasim, Prof. K. Kostarelos, Karolinska Institutet Dr. C. Bussy Stockholm 171 77, Sweden National Graphene Institute E-mail: bengt.fadeel@ki.se University of Manchester Dr. J. Wang Manchester M13 9PT UK Science for Life Laboratory Department of Biochemistry and Biophysics Stockholm University The ORCID identification number(s) for the author(s) of this article Stockholm 171 65, Sweden can be found under https://doi.org/10.1002/advs.201903200 Dr. I. A. Vacchi, Dr. C. Ménard-Moyon, Prof. A. Bianco University of Strasbourg © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. CNRS KGaA, Weinheim. This is an open access article under the terms of the Immunology, Immunopathology and Therapeutic Chemistry Creative Commons Attribution License, which permits use, distribution UPR 3572, Strasbourg 67 084, France and reproduction in any medium, provided the original work is properly Prof. K. Kostarelos cited. Catalan Institute of Nanoscience and Nanotechnology (ICN2) DOI: 10.1002/advs.201903200 Campus UAB Bellaterra, Barcelona 08193, Spain Adv. Sci. 2020, 7, 1903200 1903200 (1 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com In respect to the pulmonary route of exposure, graphene ditions, we found that the largest GO sheets were causing the nanoplatelets (GNPs) have been the most widely studied GBMs most adverse, widespread, and long-term effects in mice, despite to date, with several reports of transient inflammation after pul- a lower lung deposition, but in agreement with persistent gran- monary exposure.[8–14] In addition, Roberts et al. demonstrated ulomas, while the pulmonary response to the smallest GO was that GNPs with large lateral dimensions (5–20 µm) have a more predominantly acute and transient, in line with a faster clearance detrimental impact on the lungs than their smaller counterparts from the lungs. (<2 µm) following oropharyngeal aspiration,[15] supporting the view that dimensions of these materials are a major determinant of their pulmonary toxicity. However, despite evidence that GNPs 2. Results may persist in the lungs without being inflammogenic even after 2.1. Characteristics of GO Sheets and Benchmark Materials 6 weeks,[16] longer term consequences of material persistence in the alveolar region, such as late onset disease or carcinogenesis, The three GO materials used in the present study were pro- remain largely unknown. duced under endotoxin-free conditions,[22] and yielded single to In contrast, pulmonary responses to GO have not been stud- few-layer sheets with similar surface chemistry and thickness, ied to the same extent. Despite numerous in vitro studies on but with varying lateral dimensions (full characterization was the toxicity of GO, there is not yet a consensus regarding the previously reported in ref. [22] indicative values in Table S1, key properties underlying these effects due to the considerable Supporting Information). Large GO (l-GO) was comprised of variability in the physicochemical properties of the tested mate- micrometer-sized sheets ranging between 1 and 30 µm in lat- rials, as well as the use of different cell models.[7] Indeed, while eral dimensions, whereas small GO (s-GO) was made of sheets studies of the impact of GO on macrophage cell lines have sug- between 50 nm and up to 2 µm. The third materials, so-called gested toxicity,[17] we have recently reported that single to few- ultra-small GO (us-GO) sheets were all smaller than 300 nm, and layer GO is non-cytotoxic for primary human macrophages (at no micrometer-sized sheets were observed in these samples. Fi- the doses tested).[18] On the other hand, endotoxin-free GO sheets nally, multi-walled carbon nanotubes (MWCNT; Mitsui-7), classi- displayed cytotoxicity toward primary human neutrophils, result- fied by the International Agency for Research on Cancer (IARC) ing in the formation of so-called extracellular traps.[19] Further- as a potential human carcinogen on the basis of numerous ani- more, using a lung epithelial cell line, we demonstrated that large mal studies,[24,25] was included as a benchmark material, known GO sheets (5–15 µm) elicited stronger cytotoxic responses com- to induce chronic inflammation and fibrosis. Notably, these nan- pared to small GO sheets (50–200 nm).[20] Similarly, Ma et al. otubes have lengths comparable to the lateral dimensions of the reported enhanced inflammatory responses in the lungs of an- l-GO tested here, ranging from 1 to 20 µm (median = 3.86 µm), imals exposed to large GO sheets (750–1300 nm) when com- and a high aspect ratio, with an average width of 49 ± 13.4 nm.[24] pared to small GO (50–350 nm).[21] However, only the acute in- flammatory response was assessed in the latter study, and ques- tions regarding the potential long-term effects of GO remain to be 2.2. Biodistribution of GO Sheets After Intranasal addressed. Instillation in Mice With this in mind, the aim of the present study was to inves- tigate whether lateral dimensions play a significant role in the In order to estimate the lung burden induced by GO after i.n. pulmonary response to GO at short- and long-term after single instillation (50 µg per mouse), we functionalized GO with NH2 - exposure in mice. Considering the literature, we hypothesized PEG4 -DOTA (GO-DOTA) as previously reported[26] and chelated that micrometer-sized GO sheets would induce a more deleteri- the GO-DOTA complexes with metal isotopes (111 In or 115 In). Iso- ous pulmonary response than nanometer-sized GO after single tope labelling purity and stability in physiological media were intranasal (i.n.) instillation (50 µg per mouse). To this end, we in- tested using 111 In (Figure S1, Supporting Information), which vestigated both acute and chronic responses to three different GO demonstrated the suitability of these probes for biodistribu- materials produced with distinct lateral dimensions,[22] for up to tion studies. Qualitative estimation of biodistribution using au- 90 days after single exposure. Pulmonary responses were char- toradiography of lungs exposed to 50 µg GO-DOTA[111 In] com- acterized in terms of conventional histopathological markers, re- plexes revealed different intensities (Figure 1A), suggesting a cruitment of immune cells, and secretion of inflammatory cy- size-dependent translocation to the lower respiratory tract 1 day tokines. Furthermore, we applied genome-wide RNA sequencing after i.n. instillation. While l-GO translocated less than the other (RNA-seq)[23] in order to evaluate potential differences between two GO materials, s-GO seemed to translocate more effectively the three GO materials at the level of gene expression. However, to the lungs compared to us-GO. For quantitative biodistribu- the probability of inducing such effects could be reduced by a tion (up to 7 days after instillation), we use ICP-MS after chelat- lower deposited dose of larger GO sheets in the alveolar region ing the natural isotope 115 In to the GO-DOTA complexes (Fig- compared to smaller ones, as a result of greater clearance from ure 1B). We again observed a size-dependent deposition in the the upper respiratory tract following intranasal administration. lower respiratory tract (reported in the panel as lungs and tra- To test this possibility, we investigated the biodistribution and chea; Figure 1C), with l-GO corresponding to the lowest de- biodegradation of GO using a combination of inductively coupled posited dose (7.01% ± 4.42%) and s-GO sheets to the highest plasma mass spectrometry (ICP-MS) and Raman spectroscopy, (17.48% ± 7.44%). At an intermediate level, 12.08% (± 5.77%) and then contrasted the inflammatory effects of the three GO of instilled us-GO sheets reached the lower respiratory tract (Ta- materials with their biokinetics (i.e., distribution, biotransforma- ble S2, Supporting Information), and persisted in both lungs tion, and/or clearance from the lungs). Under the tested con- and trachea up to the latest time point tested here (7 days post Adv. Sci. 2020, 7, 1903200 1903200 (2 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com Figure 1. GO sheets translocate to the lower respiratory tract after intranasal instillation. Mice were instilled with DOTA-functionalized GO sheets labeled with 115 In (natural) or 111 In (radioactive), and the organs were collected 1 and 7 days after exposure. A) Autoradiography of lungs dissected from mice instilled with GO-DOTA[111 In] 1 day after exposure. B) Heatmap illustrates tissue distribution and persistence of GO-DOTA[115 In] in the respiratory tract compared to DOTA[115 In] control, at days 1 and 7 after i.n. instillation. Each block represents the mean amount of GO in the respective organ, quantified by ICP-MS, which was normalized as % instilled dose (ID) per gram of dry tissue. C) Quantification of 115 In by ICP-MS in the lungs and trachea from mice exposed to GO-DOTA[115 In] or the DOTA[115 In] control reveals size-dependent distribution of GO in the respiratory tract. Individual data points corresponding to each animal are plotted alongside mean ± SD (n = 4). Data were analyzed using a two-way ANOVA test with post hoc Sidak’s multiple comparisons test. Significant differences between treatments are plotted with (*), whereas differences over time are plotted with (#). In both cases, statistical significance is reported as: (*), p < 0.05; (**), p < 0.01; (***), p < 0.001; (****), p < 0.0001. Remaining organs are plotted in Figure S2, Supporting Information. exposure). After normalizing by the lung dry weight, the uptake The lower amount of l-GO found in the lower respiratory tract of GO sheets by the lungs was calculated to be 11.88 ± 6.38 mg correlated with its greater detection in extra-pulmonary organs g−1 for l-GO, 30.72 ± 13.55 mg g−1 for s-GO, and 20.92 ± 6.37 mg such as the nasal cavity (p < 0.0037 compared to all other treat- g−1 for us-GO (Table S3, Supporting Information). Lung uptake ments) or the gastrointestinal (GI) tract (p = 0.0008 versus DOTA) of s-GO and us-GO was significantly greater than that of the at day 1 post exposure (Figure S2, Supporting Information). Sig- DOTA control (p < 0.0001 for s-GO and p = 0.0102 for us-GO), nificant translocation to the gastrointestinal tract was also ob- which totaled 4.63 ± 2.52 mg g−1 . Although the delivered dose served for s-GO (p = 0.0427) compared to DOTA. Moreover, the to the lungs and trachea was lower than s-GO and us-GO, l-GO amount of s-GO in the kidney was 4.9 and 3.6 times greater than was found to be persistent in the trachea, with a detected dose DOTA (p = 0.0159) and l-GO (p = 0.0315), respectively. These that was about 6.6 times greater than DOTA at day 7 post in- results suggested the translocation of s-GO to the bloodstream, stillation (p = 0.0450). Similar retention was also observed for with subsequent urinary excretion following glomerular filtra- us-GO, with detected doses in the lungs and trachea that were tion, in line with previous observations following other admin- 21 (p = 0.0045) and 5.2 times (p = 0.0003) greater than DOTA, istration routes.[26,27] However, this translocation was not accom- respectively. On the other hand, the amount of s-GO in the panied by a significant retention in organs of the reticuloendothe- lungs and trachea decreased by 73% (p = 0.0002) and 66% (p = lial system such as liver or spleen, compared to the DOTA control 0.0191), respectively, between day 1 and 7, suggesting a greater (p = 0.5014). The continuing decrease in signal from the nasal clearance. cavity after l-GO exposure suggested that these materials were Adv. Sci. 2020, 7, 1903200 1903200 (3 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com efficiently eliminated via mucociliary clearance. On the other lar region, already by day 7 (Figure S3, Supporting Information), hand, us-GO translocated significantly from the nasal cavity to which coincided with areas of granulomatous inflammation. In the brain over time (p = 0.0005), with a retained dose at day 7 agreement with this data, exposure to MWCNTs also increased post instillation that was at least 8 times higher than the other the expression of the known fibrosis marker TGF-𝛽 by 96% (p = treatments (p < 0.0002), most likely because of the smaller di- 0.0453) compared to the vehicle control (Figure 2C). On the other mensions of us-GO. No significant accumulation in the remain- hand, none of the three GO materials induced any obvious colla- ing extra-pulmonary organs was observed for any material com- gen deposition or upregulation of TGF-𝛽 compared to control, in pared to the DOTA control. These results suggested that lateral spite of the formation of persistent granulomas after exposure to dimensions affect the pulmonary deposition of GO sheets and l-GO or s-GO. their biokinetics, which may ultimately determine their biologi- cal impact. Because the amount of GO reaching the lungs was randomly distributed in each lobe (Figure S2, Supporting Infor- mation), we decided for each animal to sample each lobe and then 2.4. Evidence of Biotransformation of GO Sheets in the Lungs pool these samples in order to further characterize the overall pul- monary response (see details below). Lack of fibrosis despite the persistence of GO materials in the lungs could be attributed to the ability of the host tissue to transform the foreign material into a less reactive material.[28] To evaluate this possibility, we performed Raman mapping of 2.3. Tissue Response to GO Sheets after Intranasal Instillation lung sections from exposed animals (Figure 3). The presence of GO in the lungs produced a specific Raman signature (Fig- Pulmonary responses to GO sheets were assessed at days 1, ure S4, Supporting Information), whose evolution over time can 7, 28, and 90 after exposure (Figure 2; Figure S3, Supporting be used as an indicator of biotransformation resulting from the Information). Lung histopathology indicated an acute inflam- loss of the crystalline structure of the material. We therefore an- matory response at days 1 and 7 post instillation, which was alyzed the Raman “fingerprints” of GO in the lungs by Raman characterized by alveolar wall thickening with infiltration of pre- spectroscopy-based imaging (Figure 3). At day 1, whilst l-GO was dominantly mononuclear cells in the interstitium for all three predominantly identified as large agglomerates associated with GO materials. Chronic inflammation was evidenced by the devel- the lung parenchyma, s-GO and us-GO were found in the air- opment of non-necrotizing peri-bronchiolar granulomas in the ways, mainly within circular features resembling individual cells lungs of mice exposed to s-GO and l-GO at days 28 and 90 post (Figure 3A). In agreement with the results obtained from ICP- exposure (Figure 2A). These granulomas were localized only in MS (Figure 1C) and autoradiography (Figure 1A) experiments, a areas with significant agglomeration of GO. Moreover, the size of more widespread and persistent (up to 7 days) signal could be these granulomas correlated with the lateral dimensions of the observed over time for us-GO compared to s-GO, suggesting a administered materials (the larger the GO sheets, the bigger the slower clearance of the former compared to the latter material. granulomas). Quantification of areas showing interstitial infiltra- Interestingly, all three GO materials could still be detected up tion revealed a size-dependent effect of GO materials, with l-GO to 28 days after administration, though the Raman signal cor- inducing the most significant pulmonary infiltration at days 28 (p responding to the starting materials was dampened from day 1 = 0.0218) and 90 (p = 0.0002) compared to the negative control to 28, suggesting that the structure of the materials was differ- (Figure S3, Supporting Information). Similarly, s-GO also trig- ent from the starting materials. Besides the indication of clear- gered a significant granulomatous response 28 days after instilla- ance, the decreased Raman signal intensities over time suggested tion (p = 0.0203). However, this was followed by a reduction of the that all three GO materials underwent biotransformation[29] (Fig- infiltrated areas after 90 days (p = 0.0187). In contrast, us-GO did ure 3B; Figures S5–S7, Supporting Information). Fine spectral not induce a significant granulomatous response in the lungs, analysis revealed that all GO materials exhibited time depen- with the parenchyma recovering completely from acute intersti- dent spectroscopic changes. Despite the apparent increase from tial inflammation at day 28 post instillation, and showing similar 1.32 ± 0.09 to 1.38 ± 0.18, us-GO did not display significant histological features to mice treated with the negative (vehicle) changes in level of crystalline disorder, revealed by the ratio be- control. As expected on the basis of previous reports, MWCNTs tween the D and G bands (ID /IG ).[30] Nevertheless, this material (MWCNT-7), used here as positive control, induced both acute suffered the most dramatic reduction in Raman signal intensity, and chronic interstitial infiltration (Figure S3, Supporting Infor- which suggested that either us-GO sheets were cleared from the mation), showing inflammatory patterns similar to or worse than lungs or biotransformation was leading to their disappearance l-GO at 28 days after exposure (p < 0.0293). Noticeably, chronic via biodegradation. Additional spectroscopic features indicative persistent inflammation in response to MWCNTs only occurred of biotransformation were noted, such as the dislocation of the in areas of material agglomeration in a similar fashion as for l- G band from 1594 to 1603 cm−1 at day 28 post exposure (p = GO and s-GO. Because chronic histopathological inflammation coincided 0.0211) suggesting the emergence of a D’ band that is merged with the apparent persistence of GO, as evidenced by the pres- to the G band. On the other hand, s-GO was the least affected ence of dark brown matter at 90 days post exposure, we also as- material since no spectroscopic features changed significantly sessed whether lung fibrosis was associated to granulomas at the over time, including a constant ID /IG ratio at 1.35 ± 0.12 after two latest time points. To this end, Masson’s trichrome stain- 28 days. Interestingly, l-GO revealed the most significant spectro- ing was used to reveal the deposition of collagen (Figure 2B). scopic changes, as evidenced by Raman analysis, including a de- As expected, MWCNTs promoted interstitial fibrosis in the alveo- crease in ID /IG band from 1.48 ± 0.07 to 1.38 ± 0.08 at day 28 post Adv. Sci. 2020, 7, 1903200 1903200 (4 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com Figure 2. Pulmonary exposure to GO induces size-dependent granulomatous inflammation in the lungs. Lung sections from mice instilled with GO were extracted at 1, 7, 28, and 90 days after i.n. instillation. A) Representative images of sections stained with H&E and Masson’s trichrome at days 28 and 90 were acquired using a 20x magnification. Remaining time points are shown in Figure S4, Supporting Information. Arrows indicate areas of significant immune cell infiltration in response to the presence of GO, with alveolar wall thickening and granuloma formation. Scale bars = 100 µm. B) Despite granuloma formation, l-GO failed to induce significant collagen deposition, a typically hallmark for pulmonary fibrosis. Images correspond to areas highlighted with a yellow box in the l-GO group (MWCNT group is presented in Figure S4, Supporting Information). Scale bars = 50 µm. C) Small portions of each individual lobe were dissected for RNA extraction and RT-qPCR. Gene expression of TGF-𝛽 was quantified as relative fold increase compared to the negative control (Dextrose) at each time point. Data are presented as individual points corresponding to each animal, alongside mean values ± SD are presented (n = 4–5). Statistical analysis was performed using a Kruskal–Wallis test with post hoc Dunn’s multiple comparisons test against the negative control: (*), p < 0.05. The temporal variation of TGF-𝛽 expression in the lungs was also analyzed using a Kruskal–Wallis test with post hoc Dunn’s multiple comparisons test: (#), p < 0.05. Adv. Sci. 2020, 7, 1903200 1903200 (5 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com Figure 3. GO sheets undergo biodegradation and/or clearance from the lungs after pulmonary exposure. Lung cryosections were obtained from mice exposed to l-GO, s-GO, and us-GO. A) Raman maps are plotted overlaying bright-field images, with the intensity scale referring to the correlation between the acquired spectrum for each pixel of the map and the reference spectrum of GO presented in Figure S4, Supporting Information. Presence of s-GO and us-GO is emphasized in inset figures, corresponding to areas highlighted in yellow. Scale bars = 100 µm. B) Representative Raman spectra corresponding to the average spectrum at each time point evidence biotransformation of GO in the lungs, with decreased signal intensity and spectroscopic changes including the shifted position of the G band and the variation in the relative intensity of the D band. instillation (p = 0.0332) and the dislocation of the G band from us-GO suffered mild weight loss, in line with higher lung bur- 1594 to 1602 cm−1 (p = 0.0181). This material presented a higher den, followed by full recovery within 4 days post exposure (Fig- ID /IG band at day 1 post exposure compared to s-GO and us-GO, ure S8A, Supporting Information). Acute pulmonary responses despite their similar structure before administration.[22] Follow- were assessed by measuring total protein and lactate dehydroge- ing this initial increase, a decreased ID /IG ratio at day 28 sug- nase (LDH) release to the airways. The analysis of bronchoalve- gested that l-GO had become more amorphous than the other two olar lavage (BAL) fluid revealed size-dependent disruption of the materials, with defects becoming more widespread with time, air-blood barrier, as suggested by the release of proteins to the air- thus reducing the sp2 conjugated areas (i.e., loss of crystalline ways (Figure S8B, Supporting Information). Nevertheless, no sta- structure).[30] tistical significance was observed compared to the negative con- trol, and only MWCNTs inducing 2.7 times more protein exuda- tion 7 days after instillation (p = 0.0325). Exposure to any of the three GO materials was followed by rapid recovery after 7 days, 2.5. Inflammatory Response to GO Sheets in the Airways and which was maintained for up to 28 days. Cytotoxicity levels of Lungs s-GO at 7 days after administration were 2.4-fold greater than ve- hicle control (p = 0.0034), suggesting that the biological response Intranasal instillation with GO dispersions (50 µg per mouse) did to GO may develop over time (Figure S8C, Supporting Informa- not result in acute toxicity, although mice instilled with s-GO and tion), in a similar fashion to the observed granuloma formation Adv. Sci. 2020, 7, 1903200 1903200 (6 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com Figure 4. GO sheets trigger acute recruitment of interstitial macrophages after pulmonary exposure. Lungs of mice exposed to carbon nanomaterials were dissected 1 and 7 days after i.n. instillation, and digested in order to extract cells for flow cytometry. A) Heatmap showing the variation in abundance of immune cell populations relatively to the negative control (Dextrose). Each block represents the mean value, after normalization to the average total number of cells in a specific immune cell population (n = 4). Absolute cell counts for each immune cell population corresponded to the product of the relative abundance in percentage of live CD45+ cells and the total number of live cells determined by Trypan Blue assay. B) Representative images showing the relative abundance of myeloid cells (CD11b+ CD11c− ) 1 day after exposure to carbon nanomaterials (see Figure S10, Supporting Information for gating strategy). Exposure to l-GO and MWCNT increased the relative amount of interstitial macrophages (CD64+ ), in contrast to a decreased population of patrolling monocytes (CD64− Ly6C− ), thus evidencing an acute inflammatory response. (Figure 2). MWCNTs did not induce significant acute effects, de- prolonged. Nevertheless, PMN infiltration to the airways was no spite a mild increase in cytotoxicity 1 day after administration longer detected at 28 days after exposure, which suggested recov- (p = 0.0511). ery from the inflammatory stimulus. These results are in con- Acute responses were further evaluated by quantifying the re- trast with the histopathological evidence of granuloma formation cruitment of leukocytes in the airways by differential staining of (discussed above). A possible explanation could reside in a reduc- the cells recovered from bronchoalveolar lavage at days 1, 7, and tion in the number of inflammatory cells that are recoverable by 28 post exposure (Figure S9, Supporting Information). All three lavage versus those cells that are recruited to the lungs. GO materials induced significant recruitment of polymorphonu- To fully characterize the inflammatory response to GO sheets clear cells (PMN), with s-GO eliciting the most acute effects com- in the lungs, we quantified the recruitment of immune cells to pared to the vehicle control at day 1 (p = 0.0025), in line with a the lung parenchyma using flow cytometry (Figure S10, Sup- higher amount of translocated materials to the lungs (Figure 1C). porting Information). Exposure to any of the three GO mate- At day 7, l-GO promoted significant infiltration of PMNs (p = rials resulted in a general increase in CD64+ CD11b+ intersti- 0.0313), similarly to MWCNTs (p = 0.0063), thus suggesting that tial macrophages and neutrophils (Figure 4A), although only l- the typical inflammatory response to foreign materials could be GO triggered significant recruitment of interstitial macrophages Adv. Sci. 2020, 7, 1903200 1903200 (7 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com Figure 5. Inflammatory responses to l-GO in the lungs are characterized by the secretion of Th1 cytokines. Small portions of each individual lobe were dissected for protein and RNA extraction. A) Heatmap showing the cytokine and chemokine expression levels of each treatment, normalized by protein content. Each block corresponds to the median value, after normalization to the average protein expression level in the negative control (Dextrose). Cytokine and chemokine expression levels were determined using a LEGENDplex Mouse Inflammation Panel kit, and normalized to protein concentration determined by BCA assay. B) Individual data points corresponding to each animal are plotted alongside mean values ± SD (n = 4–5). Statistical analysis of data acquired using the LEGENDplex assay was performed using a 2-way ANOVA test with post hoc Sidak’s multiple comparisons test, whereas gene expression of IL-4 was analyzed using a Kruskal–Wallis test with post hoc Dunn’s multiple comparisons test. In both cases, each treatment was only compared against the negative control: (*), p < 0.05; (**), p < 0.01. The temporal variation in the expression of IL-4 was also analyzed using a Kruskal–Wallis test with post hoc Dunn’s multiple comparisons test: (#), p < 0.05. (p = 0.0025), monocyte-derived dendritic cells (moDCs, p = cruitment, despite a general increase at 7 days post exposure that 0.0213), and activated plasmacytoid DCs (p = 0.0163) 1 day af- suggested a chronic inflammation characterized by an increased ter exposure (Figure S11, Supporting Information). Nevertheless, population of eosinophils (p = 0.0901). immune cell recruitment was mostly resolved 7 days after instil- Cytokine expression, measured using a multiplex array, re- lation of s-GO and us-GO, whereas l-GO seemed to promote a vealed different pro-inflammatory responses for the different ma- prolonged infiltration of monocyte-derived CD11b+ cells (Figure terials at 1 and 7 days after i.n. instillation (Figure 5). All three GO S11, Supporting Information), suggesting the establishment of a materials induced only a mild acute inflammatory response, with more chronic inflammatory response to this type of GO. In con- more pronounced effects observed at day 7 after exposure. Inter- trast, MWCNTs were not as efficient in stimulating granuloma- estingly, and in agreement with histopathological observations, tous interstitial infiltration as l-GO (Figure 4B), despite the appar- most of the measured cytokines were upregulated by exposure to ent increase in interstitial macrophages (p = 0.5884). Similarly l-GO, with the most dramatic effects occurring after 7 days. This to the cytokine expression profile (see below), mice exposed to inflammatory response was characterized by a significantly in- MWCNTs did not induce significant changes in immune cell re- creased expression of IL-1𝛼 (p = 0.0032), IL-6 (p = 0.0406), IL-23 Adv. Sci. 2020, 7, 1903200 1903200 (8 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com (p = 0.0377), IFN-𝛽 (p = 0.0398), and IFN-𝛾 (p = 0.0117), com- for this time point (i.e., not found after either us-GO or s-GO pared to mice treated with 5% dextrose, 7 days after instillation exposure). Further analysis of the latter pathways revealed genes (Figure S12, Supporting Information). Exposure to s-GO did not encoding matrix metalloproteinases (such as MMP3) and cathep- result in significant changes in cytokine expression, despite the sins (including CTSB and CTSZ), as well as the gene encoding apparent increase of IL-1𝛼 (p = 0.6654) and the chemokine MCP- the protein serpin B2 (also known as plasminogen activator in- 1 (p = 0.3053), 1 day after administration (Figure 5). On the other hibitor type 2 or PAI-2). Similarly, when the top most-affected hand, us-GO elicited a mild inflammatory response, stimulating “canonical pathways” were clustered, distinct size-dependent dif- the expression of IL-1𝛼 to similar levels to those observed in mice ferences were observed at day 28 post exposure (Figure S13B, exposed to l-GO, which represented a 3.1-fold increase relative to Supporting Information). The top “diseases and biofunction” vehicle control at day 7 post instillation (p = 0.0057). The latter pathways identified by IPA at day 1, 7, and 28 for each individual cytokine seemed to be upregulated in a size-dependent manner GO material are reported in Figures S14–S16, Supporting Infor- at early time-points, with us-GO inducing the greatest increase mation. Numerous pathways related to lung inflammation were at 1 day after instillation albeit without statistical significance (p affected, which agreed well with other experimental data reported = 0.2816). An opposite trend was found regarding the expres- above. sion of IL-4, with all GO materials inducing a decreased expres- To illustrate these findings, the genes in the “inflammation sion at day 1 despite the non-significance compared to vehicle of the lung” pathway that were found to be significantly per- control (p = 0.1244). This inhibition was followed by a signifi- turbed by l-GO at day 28 are depicted in Figure 6B. These include cant increase in IL-4 expression in mice exposed to s-GO (p = genes such as PYCARD, encoding the protein ASC (apoptosis- 0.0265), to levels similar to the vehicle control at day 28 post ex- associated speck-like protein containing a caspase recruitment posure. Finally, MWCNTs did not induce statistically significant domain) a key adaptor of the inflammasome[32] ; the gene en- changes in cytokine production, despite an apparent increase of coding the IL-1 receptor antagonist; genes encoding mem- pro-inflammatory cytokines at day 7 post exposure. bers of the C-type lectin/C-type lectin-like domain (CTL/CTLD) superfamily[33] ; or the gene encoding MMP12 (Matrix Metal- lopeptidase 12, also known as macrophage metalloelastase or macrophage elastase).[34] Notably and in agreement with the his- 2.6. Transcriptomics Analysis of Lungs Exposed to GO Sheets tological data, none of the GO materials were found to affect fi- brosis pathways. Finally, network analysis of DEGs performed at To gain further molecular insight into the pulmonary responses day 28 post l-GO exposure revealed that both TNF-𝛼 and IFN- to the three different GO materials, we performed RNA sequenc- ɣ acted as central nodes or putative regulators of the affected ing on total RNA extracted from lung tissues at days 1, 7, and genes, as shown in Figure 6C; Figures S14–S16, Supporting 28 post exposure. Sequencing was performed using the Illumina Information. HiSeq2500 platform, and sequencing reads were aligned against the Mus musculus (GRCm38) genome. RNA sequencing gener- ated 500 GB of raw sequence data with very high sequencing 3. Discussion depth (>15 million reads per sample). Differentially expressed genes (DEGs) displaying a fold change (logFC) ≥ 0.5, and p ≤ Understanding the potential hazard associated to a pulmonary 0.05 were considered for functional (bioinformatics) analysis. En- exposure to GBMs or other 2D materials is a major goal of scien- riched networks, molecular functions, and pathways were elabo- tists, regulatory agencies, and industrial stakeholders. However, rated using the Ingenuity Pathway Analysis (IPA) tool.[31] Venn while there are several studies on the pulmonary impact of GNPs, diagrams were used to group unique and common DEGs iden- very few studies have been performed to date on other GBMs tified in mice exposed to GO materials at 1, 7, and 28 days including GO,[7] and they focused primarily on acute response post exposure (Figure S13A, Supporting Information). RNA se- with a maximum follow-up of 14–21 days post exposure.[35,36] quencing results displayed in this manner revealed distinct time- Herein we studied the biodistribution, biotransformation, and dependent patterns of gene expression for all three GO mate- pulmonary response to GO in mice for up to 90 days after sin- rials. Markedly, gene expression changes were attenuated at 28 gle i.n. instillation of GO sheets with varying lateral dimensions days post exposure for us-GO when compared to days 1 and 7, in (50 µg per animal). We demonstrated a size-dependent transloca- line with the observation that histopathological signs of inflam- tion to the lower airways. In the airways, the formation of material mation were largely resolved at this time-point. In contrast, for agglomerates was also size-dependent: the larger the materials, l-GO, the number of DEGs (both up- and downregulated genes) the greater the agglomeration. However, regardless of their pri- continued to increase up to 28 days post exposure, indicative of a mary lateral dimensions, all materials were submitted to clear- long-lasting pulmonary response to this material. ance and had undergone a time-dependent biotransformation, In order to compare the impact of the three different GO ma- which left behind materials with an altered crystalline structure. terials, we then performed hierarchical clustering analysis of the In terms of pulmonary response, adverse outcomes were also top “diseases and biofunction” pathways identified by IPA at day size-dependent; with micrometer-sized GO inducing more pro- 28 (Figure 6A). The data showed remarkable lateral dimension- nounced and long-lasting effects when compared to nanometer- dependent differences in affected pathways. Notably, while all sized GO. Importantly, only the material suspensions containing three GO materials affected pathways related to inflammation or micrometer-sized GO sheets triggered the formation of granulo- immune cell function, several pathways annotated as “lung can- mas that persisted for up to 90 days. Yet no significant pulmonary cer related” in IPA were only affected in mice exposed to l-GO fibrosis was observed after exposure to these large materials. Adv. Sci. 2020, 7, 1903200 1903200 (9 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com Figure 6. RNA-seq analysis reveals size-dependent effects of GO in inflammation-related pathways. A) Hierarchical clustering analysis of the top diseases and biofunctions pathways identified by IPA in lung tissue samples of mice exposed for 28 days to us-GO, s-GO, and l-GO. DEGs having ≥0.5 log fold change and ≥ 0.05 FDR were included in the analysis. The color coding in the heatmap depicts the p-values for the pathways shown. B) IPA analysis was conducted on DEGs identified at day 28 post exposure. The pathway inflammation of the lung was significantly affected in lungs of animals exposed to l-GO. Red denotes upregulated genes, and green denotes genes that are downregulated. C) Network analysis of DEGs identified in the lungs of mice at 28 days post-exposure to l-GO. IPA suggested that IFN-𝛾 and TNF-𝛼 (highlighted in yellow) are main regulators of the affected pathways by l-GO. Adv. Sci. 2020, 7, 1903200 1903200 (10 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com 3.1. Lateral Dimensions Govern the Biodistribution and biological response to these materials (i.e., mild and transient in- Pulmonary Responses to GO flammation, with full histological recovery by day 28 after expo- sure), and highlighted us-GO as the less reactive materials of all In comparison to s-GO or us-GO, greater amounts of l-GO three, as previously suggested.[38] were detected in the nasal cavity and GI tract, while lower Altogether, our findings strongly suggest that the primary lat- amounts were measured in the lungs. These findings suggest eral dimensions of GO sheets, together with their ability to form that micrometer-sized GO sheets were less able to penetrate material agglomerates in the airways, are major determinant fac- the lung airways than the other two materials, and more likely tors of their biodistribution and ensuing adverse effects in lungs. to undergo mucociliary clearance from the upper airways, as previously postulated when considering solely the nanomaterial size.[37] However, despite the lowest dose delivered to the lower airways, l-GO induced the strongest granulomatous inflamma- 3.2. Persistent Granulomatous Response Attenuated by tion (histopathology) and the highest impact at the molecular Transformation of GO level (whole lung RNA sequencing), hence suggesting a higher material reactivity, in agreement with previous studies report- Granulomas form when the immune system attempts to wall ing larger materials to be more deleterious.[15,20,21,38] In addition, off a foreign agent that cannot be eliminated.[42] This physio- granuloma formation after l-GO exposure appeared to be strongly logical process is for instance used to isolate infectious agents correlated with the formation of material agglomerates, which such as Mycobacterium tuberculosis in the lungs of infected in- were the largest amongst the three materials and persisted for dividuals who will not develop the disease.[43] Here, both s-GO up to 90 days, as revealed by Raman imaging. and l-GO induced the formation of granulomas that persisted The greatest lung burden was found in s-GO exposed mice, for up to 90 days, and this formation appeared to be correlated which in turn elicited the strongest response in the airways. In- with the presence of larger material agglomerates for these two deed, a greater neutrophil infiltration to the airways correlated materials. Noticeably, the formation of granulomas after l-GO well with a greater delivered dose; however, s-GO did not exert a exposure followed the recruitment of interstitial macrophages strong inflammatory cytokine secretion or molecular impact in and moDCs, which have been previously described in granulo- lungs, suggesting that the biological effects of s-GO might reside matous response to both sterile and infectious agents,[44,45] and primarily in the airways rather than the lung parenchyma. As in are known to stimulate Th1 polarized immune responses during l-GO exposed mice, material agglomerates persisting for up to granulomatous inflammation.[46] At day 7, the enhanced secre- 90 days were found in s-GO exposed animals. Comparable ma- tion of both IL-6 and IFN-𝛾 was also an event favoring a Th1 re- terial agglomerates persisting over 90 days were also reported by sponse to l-GO. In agreement with those findings, Ma et al. found Bengtson et al. after single i.t. instillation.[39] Likewise, Li et al. that large GO sheets (750–1300 nm) induced greater Th1 inflam- have demonstrated a similar persistence of small GO sheets (10– mation in mouse lungs than small GO sheets (50–350 nm).[21] 800 nm) for an equivalent period of time (3 months), also after Here, l-GO sheets also inhibited an anti-inflammatory Th2 im- i.t. instillation.[40] In that study, GO sheets were causing acute mune response by decreasing the expression of IL-4,[47] in agree- lung injury with dose-dependent inflammation and cytotoxicity, ment with previous works in which GO was shown to attenuate followed by pulmonary fibrosis starting 1 month after exposure. Th2 response in mice.[48] Overall, the modulation of the different Despite having similar lateral dimensions, the s-GO sheets used inflammation mediators observed here after l-GO exposure was here exerted a much milder response (i.e., acute inflammation consistent with the sustainment of a pro-inflammatory granulo- without fibrosis), which could be explained by a 3.5 times lower matous response.[45] delivered dose to the lungs than in Li et al.[40] However, an important complication of persistent granuloma- Despite the presence of materials in the lower airways evi- tous inflammation is fibrosis, which may cause permanent tissue denced by Raman imaging at 1 and 7 days, us-GO was the ma- remodeling and damages, even after the causative agent has been terial that least interacted with the lung epithelium, supporting eliminated.[42] In addition, granuloma-driven fibrosis after pul- the idea that material agglomeration and persistence were size- monary exposure to carbon nanomaterials has been previously dependent. Differences in lung burden between us-GO and s-GO ascribed for inducing loss of lung function and further patho- (or l-GO) can be first explained by the greater ability of us-GO to logical sequelae, including cancer.[49,50] Here, granulomatous in- translocate to extra-pulmonary organs (such as the brain, possi- flammation combined with marked fibrosis was only observed bly via the olfactory tract[37] ), as measured by ICPMS. Secondly, in MWCNT (Mitsui-7) exposed animals, in agreement with pre- us-GO sheets were distributed more evenly within the lung air- vious reports.[44,51] Indeed, despite the persistence of lung gran- ways, rarely forming material agglomerates, as revealed by Ra- ulomas after s-GO or l-GO exposure for up to 90 days, there was man imaging. This pattern would favor us-GO likelihood of being no evidence of fibrosis at any time point. Bengtson et al. reported taken up by lung macrophages and drained to lymph nodes.[37] similar absence of fibrosis for up to 90 days after GO intratracheal Early material presence (i.e., day 1 and 7) could hence be due to instillation in mice, in spite of observing inflammation, acute a slow uptake by airway macrophages, with degradation and/or phase response, or BAL cell genotoxicity.[39] In contrast, Li et al. clearance from the alveolar region occurring over several months, showed that GO materials could induce lung fibrosis as early as 7 as suggested for other nanomaterials.[41] In fact, Raman signal in- days.[40] Taken together, these different studies suggest that GO- tensity for us-GO in lung sections was reduced with time. Overall, induced fibrosis might be GO material specific, even though dif- the fate (i.e., biodistribution, agglomeration and clearance pro- ferences in the delivered dose to the lower airways in these differ- file) of us-GO sheets in the lungs agreed well with the observed ent studies should not be neglected. In the present study, the lack Adv. Sci. 2020, 7, 1903200 1903200 (11 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com of pulmonary fibrosis despite the formation of granulomas could be of interest in future work to study this putative link to lung be explained by a long-lasting Th1 polarized immune response diseases in the context of chronic GO exposure. Similarly, further (combined with an inhibition of pro-fibrotic Th2 immune re- analysis of RNA sequencing data revealed that both TNF-𝛼 and sponse, as mentioned above), and an elevated level of interferons IFN-ɣ were central regulators for the affected genes, highlighting (i.e., IFN-𝛽 (ELISA measured, significant at 7 days after l-GO ex- the key role of inflammation mediators in the regulation of gene posure) or IFN-𝛾 (revealed by analysis of RNA sequencing)), espe- expression in response to l-GO exposure. cially after l-GO exposure. Type I interferons (such as IFN-𝛽) have More importantly, several pathways annotated as “cancer- indeed been associated with the development of chronic inflam- related” were affected by l-GO (especially at day 28), but not by mation in response to particulate matter by triggering the pro- s-GO or us-GO, including the gene encoding serpin B2, a pro- duction of chemokines that further attract inflammatory cells.[52] tein involved in both inflammation and cancer.[59] Although these Therefore, which material features and biological mechanisms pathways were significantly perturbed, activation scores for each are specifically associated to the observed differences in differ- individual gene were not. When putting together granulomas for- ent GO materials ability to induce fibrosis should be investigated mation and cancer-related pathways activation, both observed af- further. ter l-GO exposure, it is important to consider that the multinu- Besides precluding the development of fibrosis, the sustained cleated macrophages present within granulomas may in fact ac- recruitment of inflammatory cells especially neutrophils and quire epithelial cell like features. Indeed, a recent study suggested macrophages, via for instance interferons, could also fulfil a that granuloma macrophages and pre-malignant epithelial cells highly beneficial role in terms of GO metabolism and elimination may share common mechanisms of adaptation to chronic geno- from the lungs. Indeed, previous studies have shown that neu- toxic stress.[60] In tuberculosis, granuloma macrophages are also trophils and macrophages could mediate the enzymatic degra- known to adopt an epithelioid appearance, and to undergo re- dation of carbon nanomaterials in mouse lungs, and the fibrotic programming events involving E-cadherin-dependent formation response was more pronounced in enzyme-deficient mice when of epithelial-like cell-cell junctions.[43] Thus, one should be cau- compared to wild-type mice.[53,54] Taken together, these results tious concluding that the present transcriptomics results point demonstrated that biodegradation is endogenously alleviating toward a definite pro-carcinogenic potential of l-GO. In fact, it the adverse effects of carbon nanomaterials, including lung fi- remains possible that some of the gene expression changes ev- brosis. In line with these data, Girish et al. provided the first evi- idenced here are reflective of the granulomatous inflammation dence of in vivo degradation of oxidized graphene in several mice or other underlining biological processes sharing commonalities organs.[55] More recently, we have also demonstrated that GO can with cancer pathways such as EMT, as previously reported for re- be enzymatically degraded in a matter of hours when incubated duced GO.[61] Therefore, more detailed analysis of the potential ex vivo with activated human neutrophils.[29] In the present study, carcinogenicity of micrometer-sized GO sheets is warranted. In changes in the Raman signature of GO present in lung sections particular, attention should be paid to comparing the gene expres- were also suggestive of biodegradative processes. In particular, sion profiles in granuloma macrophages and in the whole lung l-GO suffered pronounced structural changes in the lungs and tissue. displayed a gradual but advanced carbon amorphization. Overall, the lack of fibrosis despite the persistence of granulo- mas and material agglomerates might hence be explained by the biotransformation of GO materials into less reactive materials, 3.4. Considerations Regarding the Exposure Model which in turn did not appear to induce pro-fibrotic Th2 immune and Its Implications response. Therefore, further investigations should focus on the association between biotransformation of GO and Th1/Th2 po- The present results highlight some values of i.n. instillation in larized immune responses, and the role of macrophages in GO comparison to or alongside i.t. instillation. Although both meth- degradation. ods provide good control of the administered doses, i.t. instilla- tion enables a superior delivery of the nominal dose of materi- als to the lower respiratory tract. On the other hand, i.n. instilla- tion does not bypass the upper respiratory tract or the mucocil- 3.3. RNA Sequencing Provides Insights into Pathway of liary clearance, thus providing a more realistic assessment of the Pulmonary Responses to GO ability of materials with varying dimensions to reach the lower airways, and a better evaluation of their impact considering the Transcriptomics approaches afford a considerable degree of sen- actual delivered dose. However, material agglomeration in the sitivity and are vastly superior to traditional microarray-based nasal cavity upon i.n. instillation may affect the delivered dose approaches.[21] Using RNA sequencing, we found distinct, size- to the lungs due to potential retention in the cavity, as observed dependent effects at gene expression level matching the size- here for l-GO. Conversely, the nasal cavity of mice has significant dependent histopathological findings; thus confirming the abil- anatomical and physiological differences compared to humans, ity of transcriptomics approaches to detect biologically relevant including a much greater surface area of the olfactory epithelium changes.[56] In particular, there was a statistically significant up- relative to the total mucosal surface.[62] regulation of MMP12 in mice exposed to l-GO; this is noteworthy In our view, the main limitation of the present study was, how- as genetic variants of the gene encoding this protein have been ever, the delivery of a single dose administered as a bolus; even associated with different human lung diseases,[57] including em- though this dose was similar to that of several other in vivo stud- physema, commonly associated to smoking.[58] It may therefore ies of GBMs.[7] Using the allometric relationships between mice Adv. Sci. 2020, 7, 1903200 1903200 (12 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com and humans, we estimated that the GO dose used here in mice The morphology (AFM and TEM imaging) and surface properties (XPS (50 µg) would correspond to a lung burden equivalent to a hu- analysis) of the three GO materials tested in this study have been pre- man occupational exposure lasting from 2.3 months up to ap- viously fully characterized and reported.[22] Biodistribution studies were performed after chemical functionalization of GO sheets with NH2 -PEG4 - proximately 23.5 years at maximum recommended levels, that DOTA via epoxide ring opening reaction, as previously reported.[26] Physic- is, 50 µg m−3 (see calculations in Supporting Information). In ochemical characterization of GO-DOTA sheets, including labeling stabil- another study using gold standard sub-chronic inhalation expo- ity, demonstrated the maintenance of the oxidation degree and structural sure of rats, the highest dose deposited in the lungs at aerosol properties, making these probes suitable for biodistribution studies. For concentration of 9.78 mg m−3 , was estimated to be 0.515 mg GO those studies, GO-DOTA probes were labeled after chelation of metal iso- (dimensions 0.5–5 µm).[36] Following the allometric relationships topes, either the natural 115 In (Merck-Sigma, UK) or the radioactive 111 In (Mallinckrodt, UK), as previously described.[26] The MWCNTs (MWCNT- between rats and mice,[70,71] we estimated that this is equivalent 7, Mitsui-7) used in this study were kindly provided by Prof Ulla Vogel to a deposited dose of 64 µg per day in the mouse lungs over (National Research Centre for the Working Environment, Denmark, and a period of 5 days (see calculations in Supporting Information). Technical University of Denmark). Therefore, the single bolus dose of 50 µg used here is not exces- Animal Exposures: Six to 8-week old female C57BL/6 mice (Envigo, sively high compared to standard inhalation studies. In order to UK) were allowed to acclimatize for 7 days prior to any experiment. An- more adequately replicate the type of exposures observed in oc- imals were kept in groups of five with free access to water and food, under cupational settings and establish exposure limits for occupational a regular 12 h light/dark cycle between 7 am and 7 pm, at a temperature of 19–22 °C and relative humidity of 45–65%. All procedures were conducted health as well as inform safety risk assessments, it would, how- after ethical approval from the UK Home Office, under Project License no. ever, be important to apply repeated administration of GO, using 70/7763, in accordance with the ARRIVE guidelines for animal research.[63] inhalation or oro-pharyngeal instillation, in future work. Experiments were carried out using four to five animals per group. All GO materials were dispersed in an aqueous solution of 5% (m/v) dextrose in ultrapure water (5% dextrose), at a concentration of 1 mg mL−1 , about 30 min before administration. MWCNTs were dispersed at the same con- 4. Conclusions centration in an aqueous solution of 0.9% (m/v) sodium chloride supple- mented with 0.5% (m/v) bovine serum albumin (BSA), followed by bath The present study demonstrates that lateral dimensions play a sonication for 10 min at 80 W (VWR, UK). Both dispersants were sterile fundamental role in the pulmonary response to GO after in- filtered (Merck Millipore, PES membrane, 0.2 µm, 33 mm) prior to use. tranasal instillation in mice. Micrometer-sized GO induced the Animals were anesthesized by inhalation of 2.5% isoflurane in 100% oxy- strongest adverse response, including granulomas persisting for gen flowing at 2 L min−1 . Mice in experimental groups were instilled with up to 90 days, despite a reduced translocation to the lungs; 50 µL of the respective nanomaterial dispersions (i.e., 50 µg per mouse), whereas animals exposed to nanometer-sized GO displayed full which were equally distributed in each nostril. The same volume of 5% dextrose solution was given to the control group. During administration, histological recovery by day 28. Whole tissue molecular analysis the animals were held in a supine position, tilted to about 60°, in order to of the lung response to the different tested materials confirmed effectively introduce the full dose. Mice were observed until full recovery, the very distinct and size-dependent effect of each GO. Impor- which occurred within 5 min after instillation. tantly, it highlighted some common pathways between cancer de- Biodistribution Studies: Mice exposed to GO-DOTA labeled with the velopment and response to micrometer-sized GO exposure that natural isotope 115 In (for ICP-MS) were dissected 1 and 7 days after i.n. in- require further investigations. Finally, our results revealed that stillation, and organs of interest were extracted and washed in PBS 1x to re- move excess blood. Each sample was placed in a small glass vial and dried regardless of their dimensions GO sheets while in the lungs under a fume hood for several days. Both wet and dry weights were mea- are enduring in situ biotransformation leading to a degradation sured using a microbalance (Sartorius MC210S, Goettingen, Germany) of their crystalline structure, which may partly be ascribed to prior to tissue solubilization. In the case of GO-DOTA labelled with ra- the absence of lung fibrosis despite the long-term persistence dioactive 111 In (for autoradiography), lungs were extracted 1 day after ad- of both granulomas and material agglomerates in the case of ministration, and only tissue-dried to remove excess liquid. micrometer-sized GO. Based on our results, it is unlikely that oc- ICP-MS. Dried organs were solubilized following a protocol involving multiple rounds of exposure to 1 mL of 30% hydrogen peroxide (Merck- cupational exposure to nanometer-sized GO (s-GO and us-GO) at Sigma Aldrich, UK) at 140 °C until drying, followed by another step using realistic air concentrations would induce significant pulmonary 70% nitric acid (Merck-Sigma Aldrich, UK). At the end of the solubiliza- toxicity. Considering the growing usage of GO or other GBMs in tion process, each sample was reconstituted with 5 mL of 2% nitric acid. spray coating for anti-corrosion solutions, we believe the present Removal of organic matter was completed by filtering through a 0.2 µm findings will be essential toward the implementation of a safer- PES membrane (Merck Millipore, UK). Inductively-coupled plasma mass by-design approach in the industrial development of GBM en- spectrometry (ICP-MS) measurements were performed at the Manchester hanced products or applications, for the benefit of workers and Analytical Geochemistry Unit, University of Manchester, using an Agilent 7500 Series ICP-MS system (Agilent, UK). end-users. Autoradiography. Deposition of GO-DOTA in the lungs was investi- gated by chelating with 111 In. As a vehicle control, mice were exposed to the DOTA[111 In] probe. All probes were administered with similar radioactivity (5.5–6.5 MBq), corresponding to the same volume of in- 5. Experimental Section stilled dose. Extracted lungs were fixed to two autoradiography cas- Material Preparation: GO dispersions were prepared following a mod- settes (Kodak Biomax Cassette) using transparent sticky tape, and left ified Hummers’ method under endotoxin-free conditions, as previously in contact with super sensitive plates (Cyclone storage phosphor screen, described.[22] Briefly, the difference in the production of these materials Packard) for 3 days. Autoradiographs were scanned using a cyclone phos- consisted in the application of ultrasounds to break down the GO sheets, phor detector (Packard Biosciences, UK) and processed using OptiQuant after oxidation of graphite flakes. l-GO was not sonicated, whereas s-GO software (version 3.1, Packard Bioscience). Signal intensity scales were and us-GO underwent sonication steps of 5 min and 4 h in a bath son- matched to take into account minor differences in background of the two icator operating at 80W (VWR, UK) before purification by centrifugation. cassettes. Adv. Sci. 2020, 7, 1903200 1903200 (13 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com Sample Collection: Mice were sacrificed at days 1, 7, 28, or 90 days After adding 50 µL of stop solution, optical absorbance at 490 nm after exposure by overdose via intraperitoneal injection of 0.2 mL pento- was measured using a FLUOstar Omega plate reader (BMG Labtech, barbitone. Except at day 90, BAL was performed by injecting the airways UK). with 1.5 mL of Hank’s balanced salt solution (HBSS, Gibco, Thermo Fisher Differential Cell Counting. Cells in BAL fluid were quantified using a Scientific, UK), which were slowly flushed up and down prior to extraction haemocytometer after Trypan Blue exclusion staining. Live cells that did to a 1.5 mL microcentrifuge tube, which was immediately placed on ice not internalize the dye were visualized under a PrimoVert inverted micro- until processing. At all considered time points, the lungs were carefully scope (Carl Zeiss, UK) using a 20x objective. Cell suspensions underwent dissected from the large airways and small portions of each lobe were a cyto-centrifugation step at 600 rpm (34 g) for 5 min at 4 °C, followed extracted for RNA extraction. Tissue samples for protein analysis were by fixation in ice-cold 100% methanol for 10 min. Differential cell staining also collected at days 1 and 7 post exposure. Tissue samples were placed was performed using the Kwik-Diff kit (Thermo Fisher Scientific, Shandon, in freshly prepared RIPA buffer or RNAlater (Invitrogen, Thermo Fisher UK) according to the manufacturer’s protocol. Stained cells were imaged Scientific, UK), respectively, and placed immediately on ice. Samples in under an Olympus BX41 microscope (Olympus, UK). RNAlater solution were incubated overnight at 4 °C prior to storage at Flow Cytometry: In a separate experiment, mice were intranasally in- −20 °C, until further processing. The remaining lung tissue was fixed in 4% stilled with carbon nanomaterials and sacrificed 1 and 7 days post ex- paraformaldehyde (PFA) overnight at 4 °C. The left lung was transferred to posure. Lungs were carefully dissected from the large airways and split a solution of 70% ethanol before paraffin embedding for histology, while into small portions using scissors. These lung portions were transferred the right lung was cryopreserved in 30% (m/v) sucrose dissolved in ul- to tubes containing 2 mL of digestion buffer, comprised by Liberase-TM trapure water before snap-freezing in optical cutting temperature (OCT) (Roche, UK) at a concentration of 0.4 Wünsch units mL−1 and DNase I medium for cryo-sectioning. (Roche, UK) at a concentration of 0.01 mg mL−1 , in RPMI-1640 cell cul- Histopathology: Paraffin-embedded lung samples were sectioned us- ture medium (Life Technologies, Thermo Fisher Scientific, UK), supple- ing a Leica RM2255 microtome set at a thickness of 5 µm. Hematoxylin mented with 10% fetal bovine serum (FBS, Gibco, Thermo Fisher Sci- and eosin or Masson’s trichrome staining was employed to measure entific, UK). After shaking for 30 min at 37 °C, the digested lungs were pathological changes and collagen deposition, respectively. Images were filtered through a 100 µm mesh filter and supplemented with cell cul- collected using a Pannoramic 250 Flash slide scanner (3D Histech, Hun- ture medium with 5 mm EDTA. The resulting cell suspension underwent gary), in bright-field mode, and analyzed using Pannoramic Viewer (ver- lysis of red blood cells using Red Blood Cell Lysing Buffer Hybri-Max sion 1.15.4, 3D Histech, Hungary). Areas of evident interstitial infiltration, (Merck-Sigma Aldrich, UK) before centrifugation at 500 g for 7 min to characterized by alveolar wall thickening and granuloma formation, were remove cell debris. Obtained cell suspensions were quantified by Trypan manually segmented using ImageJ software (version 1.51, National Insti- blue exclusion before staining for flow cytometry. The staining protocol tutes of Health, Bethesda, MD). Cell infiltration was defined as the sum involved the incubation with the Live/Dead UV viability dye (Life Tech- of these manually segmented areas divided by the total area of the lung nologies, Thermo Fisher Scientific, UK) for 20 min at room temperature, section. Infiltrate size corresponded to the area of each segmented area. before washing and staining with a mixture of Fc blocking agent (anti- Raman Mapping: Lung samples snap-frozen in OCT were sectioned mouse CD16/CD32) and fluorochrome-conjugated antibodies (see Table using a Leica CM3050S cryotome set at a thickness of 20 µm. Cryo-sections S4, Supporting Information) for 30 min at 4 °C. Cells were finally fixed of lungs exposed to GO materials were gently rinsed with PBS 1x, ultrapure with 2% PFA in cell culture medium for 20 min at room temperature, and water, and 100% methanol, in order to remove any excess of OCT com- washed prior to flow cytometry measurements. Data were acquired us- pound and salts, and to post-fix the section. Once dried, lung sections ing a BD Fortessa flow cytometer (BD Biosciences, UK), and then pro- were imaged under an Olympus BX41 microscope, part of the DXRxi Ra- cessed using FlowJo (version 10, FlowJo, LLC, USA). Appropriate single- man system (Thermo Scientific, UK), using a 50x objective. Raman spectra color and fluorescence minus one (FMO) controls were performed for all were acquired using a laser of 𝜆 = 633 nm operating at 0.4 mW, through fluorochromes, in order to compensate any spectral overlap and poten- a 50 µm pinhole aperture with an exposure time of 0.125 s. Raman corre- tial non-specific binding and/or fluorescence emission in undesired chan- lation maps were plotted using the OMNICxi software (Thermo Scientific, nels. Cell populations were defined as shown in Figure S10, Supporting UK), in order to reflect the spatial distribution and transformation of GO in Information. Cell events were gated after excluding cell debris and dou- the lungs. An arbitrary color scale was defined at each pixel to describe the blets, using both forward (FSC) and side scatter (SSC). Leukocytes were similarity between the acquired Raman spectrum at that coordinate and defined by selecting CD45+ cells that had low fluorescence intensity for a reference spectrum of GO, focusing on a region comprising both the D the Live/Dead signal. Neutrophils were discriminated based on their high and G bands (1145–1810 cm−1 ). Black color represents background sig- expression of Ly6G and CD11b, and confirmed by their intermediate SSC nal, whereas white color represents the highest similarity to the reference values, indicating their granularity. The remaining cells were gated based spectrum, indicating the presence of GO. Red color represents a weaker on their expression of Siglec-F. Siglec-F+ cells were distinguished based on signal of the D and G bands compared to the reference spectrum. their expression of CD11b, with eosinophils expressing CD11b and hav- BAL Analysis: BAL fluid samples were centrifuged at 1500 rpm (214 g) ing high SSC. Alveolar macrophages, on the other hand, were CD11b− for 5 min at 4 °C using a Hettich Universal 320R centrifuge (Hettich Zen- and confirmed to be CD11c+ and CD64+ . Myeloid cells were further gated trifuger, Germany). Supernatant was collected for the quantification of to- based on their expression of CD11b and CD11c. Dendritic cells (DCs) were tal protein and amount of released lactate dehydrogenase (LDH), whereas considered CD11c+ cells, and based on their expression of CD11b, Ly6C the pellet was re-suspended in 0.5 mL of PBS for differential cell counting. and MHC-II they could be further classified: i) conventional DCs (cDCs) BCA Assay. Protein release in BAL fluid was quantified using the bicin- were considered CD11b− CD11c+ Ly6C− MHC-II+ ; ii) plasmacytoid cells choninic acid (BCA) protein assay (Pierce, Thermo Fisher Scientific, UK), (pDCs) could be found in the CD11b− CD11c+ Ly6C+ MHC-II−/int pop- according to the manufacturer’s instructions. Briefly, 200 µL of BCA ulation; iii) CD11b+ DCs differed from cDCs due to their expression of reagent mixture were added to 25 µL of each sample, and incubated at CD11b (CD11b+ CD11c+ Ly6C− MHC-II+ ); and iv) monocyte-derived DCs 37 °C for 30 min in a 96-well plate (Corning, UK). Optical absorbance (moDCs) were CD11b+ CD11c+ Ly6C+ MHC-II−/int . Monocytic cells were at 562 nm was measured using a FLUOstar Omega plate reader (BMG gated on the CD11b+ CD11c− population, with interstitial macrophages Labtech, UK). The protein concentration of each sample was determined expressing CD64. Monocytes were divided into classical (Ly6C+ ) or pa- via extrapolation from the BSA standard curve, defined with albumin stan- trolling (Ly6C− ), depending on their activation state. Lymphocytes were dards (0–2000 µg mL−1 ). Standard solutions and BCA reagent mixture identified as CD3+ , from the population of CD11b− CD11c− cells, and fur- were freshly prepared before each assay. ther classified either as CD4+ or CD8+ T cells. The obtained percentages LDH Assay. LDH content was measured using the CytoTox 96 Non- for each cell population were multiplied by the number of live cells after Radioactive Cytotoxicity Assay (Promega, UK). In this case, 50 µL Trypan blue exclusion, rendering absolute cell counts. of each sample were mixed with 50 µL of LDH substrate mix, fol- Cytokine Analysis: Tissue incubated in RIPA buffer was homogenized lowed by incubation in the dark for 15 min at room temperature. with 5 mm stainless steel beads using a TissueLyser LT system (QIAGEN, Adv. Sci. 2020, 7, 1903200 1903200 (14 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com UK) operating at 50 Hz for 5 min, followed by centrifugation at 2600 g Bioinformatics Analysis: The quality check involving the analysis of for 5 min to remove cell debris. The obtained supernatant was utilized sequence quality, GC content, the presence of adaptors, overrepre- in subsequent protein analysis. Protein concentration was determined us- sented sequences, duplication level in order to detect sequencing er- ing the BCA assay, as described above. Quantification of a set of inflam- rors, PCR artifacts or adapter contaminations was conducted using soft- matory cytokines and chemokines (IL-1𝛼, IL-1𝛽, IL-6, IL-10, IL-12p70, IL- ware tools FastQC. Fastq-Screen was performed to check the composi- 17A, IL-23, IL-27, MCP-1, IFN-𝛽, IFN-𝛾, TNF-𝛼, and GM-CSF) was carried tion of the libraries. The mRNA sequencing data were analyzed using the out using a LEGENDplex Mouse Inflammation kit (BioLegend, UK) ac- rnaseq pipeline (https://github.com/nf -core/rnaseq). The preprocessed cording to the manufacturer’s recommendations. Fluorescence intensi- sequencing reads were aligned against the Mus musculus (GRCm38) refer- ties of the multiple analytes were recorded using a BD FACSCanto flow ence genome using STAR.[64] Subsequently, the read-counts to genes were cytometer (BD Biosciences, UK). Data analysis was performed with spe- generated using the software featureCounts.[65] Full-length transcripts cialized software provided by the manufacturer of the assay Cytokine con- representing multiple splice variants for each gene locus were assem- centration was normalized by total protein concentration as determined bled and quantitated using StringTie.[66] Extensive quality-control on the above. results was performed using RSeQC,[67] dupRadar,[68] and Preseq (http: RNA Extraction and RT-qPCR: Tissue samples in RNAlater solution //smithlabresearch.org/software/preseq/) which generated RNA quality (Invitrogen, Thermo Fisher Scientific, UK) were homogenized using a Tis- control metrics, technical/biological read duplication level, and library sueLyser LT (QIAGEN, UK) as described above. Homogenized samples complexity estimation. Differentially expressed genes between different were then loaded onto spin cartridges containing silica membranes, for treatments of samples were determined using edgeR.[69] Differential gene extraction of total RNA performed using a PureLink RNA Mini kit (Invit- expression levels between the healthy control lung tissue sample and the rogen, Thermo Fisher Scientific, UK) according to the manufacturer’s in- three different GOs (us-GO, s-GO, l-GO) were estimated with a t-test. p structions. Total RNA concentration and purity were calculated by mea- values were corrected with the Benjamini–Hochberg algorithm (false dis- suring the optical density at 230, 260, and 280 nm, using a Biophotome- covery rate; FDR). DEGs having a fold change (logFC) ≥ 0.5, p-value ≤ 0.05 ter Plus spectrophotometer (Eppendorf AG, Germany). For quality con- were considered for functional analysis. Ingenuity Pathway Analysis (appli- trol purposes, RNA samples were required to have absorbance ratios at cation version 220 217, content version 16 542 223) (licensed by Ingenuity 260 nm/280 nm and 260 nm/230 nm above 1.8. First-strand cDNA was Systems, Redwood City, CA) was employed to study biological processes, created from 1 µg of RNA in a total volume of 20 µL using the High Capac- canonical pathways, upstream transcriptional regulators, and gene net- ity cDNA Reverse Transcription kit (Invitrogen, Thermo Fisher Scientific, works, based on the DEGs. UK). The cDNA synthesis reaction followed a protocol of 10 min at 25 °C, Statistical Analysis: All experiments were analyzed using GraphPad followed by 2 h at 37 °C, and 5 min at 85 °C, executed by a CFX96 real- Prism software (version 6.01, GraphPad Inc., USA). Data were ana- time PCR system (BioRad, UK). Quantitative PCR was performed using lyzed using analysis of variance (two-way ANOVA) with post hoc Sidak’s the CFX96 real-time PCR detection system (BioRad, UK). Each sample con- multiple comparisons test or Kruskal–Wallis test with post hoc Dunn’s sisted of 2 µL of cDNA obtained from reverse transcription, each primer at multiple comparisons test. In both cases, p < 0.05 was considered 500 nm, and 10 µL of PowerUp SYBR Green Master Mix (Applied Biosys- significant. tems, Thermo Fisher Scientific, UK) in a 20 µL reaction. Primers are listed in Table S5, Supporting Information. PCR protocol consisted of an initial activation step at 50 °C for 2 min, followed by a denaturation step at 95 °C for 2 min, prior to 40 cycles of amplification (denaturation at 95 °C for 15 s, Supporting Information then annealing and elongation at 60 °C for 1 min). Melting curves between 65 and 95 °C were also obtained, in order to confirm amplification speci- Supporting Information is available from the Wiley Online Library or from ficity and the absence of artefacts such as primer dimers. No-template the author. controls did not return any Ct values. Each sample was analyzed in du- plicate, and the mean Ct value was used for further calculations. Relative changes of gene expression of each treatment were calculated according to the Livak method as 2−ΔΔCt , where the obtained ΔCt values are sub- Acknowledgements tracted by those obtained with the negative control (Dextrose). The ΔCt This work was supported by the European Commission through the Future values were calculated by subtracting the Ct value of each individual gene Emerging Technologies (FET) Graphene Flagship project (grant agree- by the Ct value of the housekeeping gene in each sample, in order to nor- ment no. 696656). This work was also partly supported by the Centre malize the amount of each transcript. National de la Recherche Scientifique (CNRS), by the International Cen- RNA Sequencing: Total RNA from tissue sample was extracted as de- ter for Frontier Research in Chemistry (icFRC) and by the Agence Na- scribed above. RNA quantity and RNA integrity number (RIN) were as- tionale de la Recherche (ANR) through the LabEx project Chemistry of sessed by Agilent 2200 bioanalyzer. Sample RNA with RIN values below 8 Complex Systems (ANR-10-LABX-0026_CSC). A.F.R. wishes to acknowl- was discarded. The isolated RNA samples were stored at −80 °C until RNA edge a studentship from the UKRI- Engineering and Physical Sciences Re- sequencing was carried out. Illumina TruSeq Stranded mRNA Library Prep search Council (EPSRC) Centre for Doctoral Training in the Science and Kit (Illumina, Inc., San Diego, CA) was used to prepare mRNA sequenc- Applications of Graphene and Related Nanomaterials (GrapheneNOW- ing libraries. Briefly, 1000 ng of total RNA was used for mRNA isolation NANO CDT; EP/L01548X/1). LN is indebted to EPSRC for his studentship using poly dT-coated beads. After purification, the mRNA was chemically in the framework of the UKRI-EPSRC-funded North West Nanoscience fragmented into small pieces, which was subsequently used as a template Doctoral Training Centre (NOWNANO DTC; EP/G03737X/1). The authors for cDNA synthesis using reverse transcriptase followed by short frag- thank Paul Lythgoe, Manchester Analytical Geochemistry Unit, Williamson ment removal from another purification step. After end-repair, adapter lig- Research Centre, University of Manchester, for assistance with ICP-MS. ation, and index codes adding for each sample, PCR amplification was con- The authors are also grateful to Peter Walker and Grace Bako, Histol- ducted. Unaligned adapters were removed twice after adapter ligation and ogy facility at the Faculty of Biology, Medicine and Health, University of PCR amplification. The quality of the libraries were examined by Caliper Manchester, for their advice and assistance in tissue histology prepara- LabChip GX/HT DNA high sensitivity and the quantitation of libraries tion and analysis, and to Gareth Howell, Flow Cytometry Facility, Fac- were measured by Qubit dsDNA HS. All 36 libraries were paired-end se- ulty of Biology, Medicine and Health, University of Manchester, for ad- quenced on a HiSeq2500 (HiSeq Control Software 2.2.58/RTA 1.18.64) vice regarding flow cytometry analysis. The authors acknowledge Sci- with a 2 × 126 setup using “HiSeq SBS Kit v4” chemistry. The conver- ence for Life Laboratory, Stockholm, Sweden, National Genomics Infras- sion from Bcl to FastQ was performed using “bcl2fastq2 v2.19” from the tructure (NGI), Stockholm, Sweden, and UPPMAX, Uppsala, Sweden, CASAVA software suite. The quality scale used is Sanger/phred33/Illumina for access to the sequencing platform and the computational 1.8+. infrastructure. Adv. Sci. 2020, 7, 1903200 1903200 (15 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com Conflict of Interest [14] L. Ma-Hock, V. Strauss, S. Treumann, K. Küttler, W. Wohlleben, T. Hof- mann, S. Gröters, K. Wiench, B. van Ravenzwaay, R. Landsiedel, Part. The authors declare no conflict of interest. Fibre Toxicol. 2013, 10, 23. [15] J. R. Roberts, R. R. Mercer, A. B. Stefaniak, M. S. Seehra, U. K. Ged- dam, I. S. Chaudhuri, A. Kyrlidis, V. K. Kodali, T. Sager, A. Kenyon, S. A. Bilgesu, T. Eye, J. F. Scabilloni, S. S. Leonard, N. R. Fix, D. Schwegler- Author Contributions Berry, B. Y. Farris, M. G. Wolfarth, D. W. Porter, V. Castranova, A. Erdely, Part. Fibre Toxicol. 2015, 13, 34. A.F.R. implemented the experiments and analyzed the data, under the su- [16] A. Schinwald, F. Murphy, A. Askounis, V. Koutsos, K. Sefiane, K. Don- pervision of C.B. and K.K. L.N. and D.A.J. performed biodistribution experi- aldson, C. J. Campbell, Nanotoxicology 2014, 8, 824. ments. I.A.V., C.M.M., and A.B. prepared the functionalized nanomaterials [17] G. Qu, S. Liu, S. Zhang, L. Wang, X. Wang, B. Sun, N. Yin, X. Gao, T. for biodistribution experiments. S.P.M. performed R.N.A. sequencing and Xia, J.-J. Chen, G.-B. Jiang, ACS Nano 2013, 7, 5732. the bioinformatics analysis, while J.W. performed statistical analysis of se- [18] S. P. Mukherjee, K. Kostarelos, B. Fadeel, Adv. Healthcare Mater. 2018, quencing data, both under the supervision of B.F. A.F.R., K.K., and C.B. 7, 1700815. conceived the overall design of the study. A.F.R., C.B., and B.F. wrote the [19] S. P. Mukherjee, B. Lazzaretto, K. Hultenby, L. Newman, A. F. Ro- manuscript, which was revised by all authors. drigues, N. Lozano, K. Kostarelos, P. Malmberg, B. Fadeel, Chem 2018, 4, 334. [20] S. Vranic, A. F. Rodrigues, M. Buggio, L. Newman, M. R. H. White, D. Keywords G. Spiller, C. Bussy, K. Kostarelos, ACS Nano 2018, 12, 1373. [21] J. Ma, R. Liu, X. Wang, Q. Liu, Y. Chen, R. P. Valle, Y. Y. Zuo, T. Xia, S. graphene oxide, inflammation, lung, macrophages, mice, RNA sequencing Liu, ACS Nano 2015, 9, 10498. [22] A. F. Rodrigues, L. Newman, N. Lozano, S. P. Mukherjee, B. Fadeel, Received: December 23, 2019 C. Bussy, K. Kostarelos, 2D Mater. 2018, 5, 035020. Revised: April 7, 2020 [23] N. Feliu, P. Kohonen, J. Ji, Y. Zhang, H. L. Karlsson, L. Palmberg, A. Published online: May 7, 2020 Nyström, B. Fadeel, ACS Nano 2015, 9, 146. [24] D. W. Porter, A. F. Hubbs, R. R. Mercer, N. Wu, M. G. Wolfarth, K. Sri- ram, S. Leonard, L. Battelli, D. Schwegler-Berry, S. Friend, M. Andrew, B. T. Chen, S. Tsuruoka, M. Endo, V. Castranova, Toxicology 2010, 269, [1] A. Zurutuza, C. Marinelli, Nat. Nanotechnol. 2014, 9, 730. 136. [2] K. S. Novoselov, V. I. Fal’ko, L. Colombo, P. R. Gellert, M. G. Schwab, [25] Y. Grosse, D. Loomis, G. KZ, B. Lauby-Secretan, F. El Ghissassi,V. K. Kim, Nature 2012, 490, 192. Bouvard, L. Benbrahim-Tallaa, N. Guha, C. Scoccianti, H. Mattock, [3] P. Wick, A. E. Louw-Gaume, M. Kucki, H. F. Krug, K. Kostarelos, B. K. Straif, Lancet Oncol. 2014, 15, 1427. Fadeel, K. A. Dawson, A. Salvati, E. Vázquez, L. Ballerini, M. Treti- [26] A. F. Rodrigues, L. Newman, D. A. Jasim, I. A. Vacchi, C. Ménard- ach, F. Benfenati, E. Flahaut, L. Gauthier, M. Prato, A. Bianco, Angew. Moyon, L. E. Crica, A. Bianco, K. Kostarelos, C. Bussy, Arch. Toxicol. Chem., Int. Ed. 2014, 53, 7714. 2018, 92, 3359. [4] A. Bianco, H.-M. M. Cheng, T. Enoki, Y. Gogotsi, R. H. Hurt, N. Ko- [27] D. A. Jasim, S. Murphy, L. Newman, A. Mironov, E. Prestat, J. Mc- ratkar, T. Kyotani, M. Monthioux, C. R. Park, J. M. D. D. Tascon, J. Caffrey, C. Ménard-Moyon, A. F. Rodrigues, A. Bianco, S. Haigh, R. Zhang, Carbon 2013, 65, 1. Lennon, K. Kostarelos, ACS Nano 2016, 10, 10753. [5] A. P. Kauling, A. T. Seefeldt, D. P. Pisoni, R. C. Pradeep, R. Bentini, R. [28] R. Kurapati, J. Russier, M. A. Squillaci, E. Treossi, C. Ménard-Moyon, V. B. Oliveira, K. S. Novoselov, A. H. C. Neto, Adv. Mater. 2018, 30, A. E. Del Rio-Castillo, E. Vazquez, P. Samorì, V. Palermo, A. Bianco, 1803784. Small 2015, 11, 3985. [6] M. V. D. Z. Park, E. A. J. Bleeker, W. Brand, F. R. Cassee, M. van Elk, I. [29] S. P. Mukherjee, A. R. Gliga, B. Lazzaretto, B. Brandner, M. Fielden, Gosens, W. H. de Jong, J. A. J. Meesters, W. J. G. M. Peijnenburg, J. T. C. Vogt, L. Newman, A. F. Rodrigues, W. Shao, P. M. Fournier, M. S. K. Quik, R. J. Vandebriel, A. J. A. M. Sips, ACS Nano 2017, 11, 9574. Toprak, A. Star, K. Kostarelos, K. Bhattacharya, B. Fadeel, Nanoscale [7] B. Fadeel, C. Bussy, S. Merino, E. Vázquez, E. Flahaut, F. Mouchet, 2018, 10, 1180. L. Evariste, L. Gauthier, A. J. Koivisto, U. Vogel, C. Martín, L. G. Del- [30] A. C. Ferrari, J. Robertson, Phys. Rev. B 2000, 61, 14095. ogu, T. Buerki-Thurnherr, P. Wick, D. Beloin-Saint-Pierre, R. Hischier, [31] A. Krämer, J. Green, J. Pollard, S. Tugendreich, Bioinformatics 2014, M. Pelin, F. C. Carniel, M. Tretiach, F. Cesca, F. Benfenati, D. Scaini, 30, 523. L. Ballerini, K. Kostarelos, M. Prato, A. Bianco, ACS Nano 2018, 12, [32] M. S. J. Mangan, E. J. Olhava, W. R. Roush, H. M. Seidel, G. D. Glick, 10582. E. Latz, Nat. Rev. Drug Discovery 2018, 17, 588. [8] A. Schinwald, F. A. Murphy, A. Jones, W. MacNee, K. Donaldson, ACS [33] I. M. Dambuza, G. D. Brown, Curr. Opin. Immunol. 2015, 32, 21. Nano 2012, 6, 736. [34] A. Y. Hendrix, F. Kheradmand, Prog. Mol. Biol. Transl. Sci. 2017, 148, [9] A. Schinwald, F. A. Murphy, A. Askounis, V. Koutsos, K. Sefiane, K. 1. Donaldson, C. J. Campbell, Nanotoxicology 2014, 8, 824. [35] S. G. Han, J. K. Kim, J. H. Shin, J. H. Hwang, J. S. Lee, T. G. Kim, J. [10] E.-J. Park, G.-L. Lee, B. S. Han, B.-S. Lee, S. Lee, M.-H. Cho, J.-H. Kim, H. Lee, G. H. Lee, K. S. Kim, H. S. Lee, N. W. Song, K. Ahn, I. J. Yu, D.-W. Kim, Arch. Toxicol. 2015, 89, 1557. Biomed Res. Int. 2015, 2015, 1. [11] E.-J. Park, S. J. Lee, K. Lee, Y. C. Choi, B.-S. Lee, G.-H. Lee, D.-W. Kim, [36] Y. H. Kim, M. S. Jo, J. K. Kim, J. H. Shin, J. E. Baek, H. S. Park, H. J. J. Appl. Toxicol. 2017, 37, 296. An, J. S. Lee, B. W. Kim, H. P. Kim, K. H. Ahn, K. S. Jeon, S. M. Oh, [12] J. H. Shin, S. G. Han, J. K. Kim, B. W. Kim, J. H. Hwang, J. S. Lee, J. H. J. H. Lee, T. Workman, E. M. Faustman, I. J. Yu, Nanotoxicology 2018, Lee, J. E. Baek, T. G. Kim, K. S. Kim, H. S. Lee, N. W. Song, K. Ahn, I. 12, 224. J. Yu, Nanotoxicology 2015, 9, 1023. [37] G. Oberdorster, E. Oberdorster, J. Oberdorster, G. Oberdörster, E. [13] J. K. Kim, J. H. Shin, J. S. Lee, J. H. Hwang, J. H. Lee, J. E. Baek, T. G. Oberdörster, J. Oberdörster, Environ. Health Perspect. 2005, 113, 823. Kim, B. W. Kim, J. S. Kim, G. H. Lee, K. Ahn, S. G. Han, D. Bello, I. J. [38] C. Bussy, H. Ali-Boucetta, K. Kostarelos, Acc. Chem. Res. 2013, 46, Yu, Nanotoxicology 2016, 10, 891. 692. Adv. Sci. 2020, 7, 1903200 1903200 (16 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.advancedsciencenews.com www.advancedscience.com [39] S. Bengtson, K. B. Knudsen, Z. O. Kyjovska, T. Berthing, V. Skaug, [55] C. M. Girish, A. Sasidharan, G. S. Gowd, S. Nair, M. Koyakutty, Adv. M. Levin, I. K. Koponen, A. Shivayogimath, T. J. Booth, B. Alonso, A. Healthcare Mater. 2013, 2, 1489. Pesquera, A. Zurutuza, B. L. Thomsen, J. T. Troelsen, N. R. Jacobsen, [56] P. M. Costa, I. Gosens, A. Williams, L. Farcal, D. Pantano, D. M. U. Vogel, PLoS One 2017, 12, e0178355. Brown, V. Stone, F. R. Cassee, S. Halappanavar, B. Fadeel, J. Appl. [40] B. Li, J. Z. Yang, Q. Huang, Y. Zhang, C. Peng, Y. J. Zhang, Y. He, J. Y. Toxicol. 2018, 38, 385. Shi, W. X. Li, J. Hu, C. Fan, NPG Asia Mater. 2013, 5, e44. [57] G. M. Hunninghake, M. H. Cho, Y. Tesfaigzi, M. E. Soto-Quiros, L. [41] M. Semmler, J. Seitz, F. Erbe, P. Mayer, J. Heyder, G. Oberdörster, W. Avila, J. Lasky-Su, C. Stidley, E. Melén, C. Söderhäll, J. Hallberg, I. G. Kreyling, Inhalation Toxicol. 2004, 16, 453. Kull, J. Kere, M. Svartengren, G. Pershagen, M. Wickman, C. Lange, [42] G. T. Williams, W. J. Williams, Journal of Clinical Pathology 1983, 36, D. L. Demeo, C. P. Hersh, B. J. Klanderman, B. A. Raby, D. Sparrow, S. 723. D. Shapiro, E. K. Silverman, A. A. Litonjua, S. T. Weiss, J. C. Celedón, [43] M. R. Cronan, R. W. Beerman, A. F. Rosenberg, J. W. Saelens, M. G. N. Engl. J. Med. 2009, 361, 2599. Johnson, S. H. Oehlers, D. M. Sisk, K. L. Jurcic Smith, N. A. Medvitz, [58] R. D. Hautamaki, D. K. Kobayashi, R. M. Senior, S. D. Shapiro, Sci- S. E. Miller, L. A. Trinh, S. E. Fraser, J. F. Madden, J. Turner, J. E. Stout, ence. 1997, 277, 2002. S. Lee, D. M. Tobin, Immunity 2016, 45, 861. [59] D. R. Croucher, D. N. Saunders, S. Lobov, M. Ranson, Nat. Rev. Can- [44] I. Huizar, A. Malur, Y. A. Midgette, C. Kukoly, P. Chen, P. C. Ke, R. cer 2008, 8, 535. Podila, A. M. Rao, C. J. Wingard, L. Dobbs, B. P. Barna, M. S. Kavuru, [60] L. Herrtwich, I. Nanda, K. Evangelou, T. Nikolova, V. Horn, Sagar, D. M. J. Thomassen, Am. J. Respir. Cell Mol. Biol. 2011, 45, 858. Erny, J. Stefanowski, L. Rogell, C. Klein, K. Gharun, M. Follo, M. Seidl, [45] Y. Pacheco, M. Ponchon, S. Lebecque, A. Calender, J. F. Bernaudin, B. Kremer, N. Münke, J. Senges, M. Fliegauf, T. Aschman, D. Pfeifer, D. Valeyre, M. Iglarz, D. S. Strasser, R. Studer, D. Freti, T. Renno, A. S. Sarrazin, M. H. Sieweke, D. Wagner, C. Dierks, T. Haaf, T. Ness, Bentaher, Experimental Lung Research 2018, 44, 25. M. M. Zaiss, R. E. Voll, S. D. Deshmukh, M. Prinz, T. Goldmann, [46] B. Ten Berge, A. Kleinjan, F. Muskens, H. Hammad, H. C. Hoogste- C. Hölscher, A. E. Hauser, A. J. Lopez-Contreras, D. Grün, V. Gor- den, R. W. Hendriks, B. N. Lambrecht, B. Van den Blink, Respiratory goulis, A. Diefenbach, P. Henneke, A. Triantafyllopoulou, Cell 2016, Research 2012, 13, 33. 167, 1264. [47] T. A. Wynn, Nat. Rev. Immunol. 2015, 15, 271. [61] Y. Liao, W. Wang, X. Huang, Y. Sun, S. Tian, P. Cai, Sci. Rep. 2018, 8, [48] M. R. Shurin, N. Yanamala, E. R. Kisin, A. V. Tkach, G. V. Shurin, A. 15188. R. Murray, H. D. Leonard, J. S. Reynolds, D. W. Gutkin, A. Star, B. [62] G. Oberdörster, A. Maynard, K. Donaldson, V. Castranova, J. Fitz- Fadeel, K. Savolainen, V. E. Kagan, A. A. Shvedova, ACS Nano 2014, patrick, K. Ausman, J. Carter, B. Karn, W. Kreyling, D. Lai, S. Olin, 8, 5585. N. Monteiro-Riviere, D. Warheit, H. Yang, Part. Fibre Toxicol. 2005, 2, [49] A. A. Shvedova, E. Kisin, A. R. Murray, V. J. Johnson, O. Gorelik, S. 8. Arepalli, A. F. Hubbs, R. R. Mercer, P. Keohavong, N. Sussman, J. Jin, [63] C. Kilkenny, W. J. Browne, I. C. Cuthill, M. Emerson, D. G. Altman, J. Yin, S. Stone, B. T. Chen, G. Deye, A. Maynard, V. Castranova, P. Animals. 2010, 4, 35. A. Baron, V. E. Kagan, Am J Physiol Lung Cell Mol Physiol. 2008, 295, [64] A. Dobin, C. A. Davis, F. Schlesinger, J. Drenkow, C. Zaleski, S. Jha, P. L552. Batut, M. Chaisson, T. R. Gingeras, Bioinformatics 2013, 29, 15. [50] S. Toyokuni, Adv. Drug Delivery Rev. 2013, 65, 2098. [65] Y. Liao, G. K. Smyth, W. Shi, Bioinformatics 2014, 30, 923. [51] A. A. Shvedova, E. R. Kisin, R. Mercer, A. R. Murray, V. J. Johnson, [66] M. Pertea, G. M. Pertea, C. M. Antonescu, T. C. Chang, J. T. Mendell, A. I. Potapovich, Y. Y. Tyurina, O. Gorelik, S. Arepalli, D. Schwegler- S. L. Salzberg, Nat. Biotechnol. 2015, 33, 290. Berry, A. F. Hubbs, J. Antonini, D. E. Evans, B.-K. Ku, D. Ramsey, A. [67] L. Wang, S. Wang, W. Li, Bioinformatics 2012, 28, 2184. Maynard, V. E. Kagan, V. Castranova, P. Baro, Am. J. Physiol. – Lung [68] S. Sayols, D. Scherzinger, H. Klein, BMC Bioinformatics 2016, 17, 428. Cell. Mol. Physiol. 2005, 289, L698. [69] M. D. Robinson, D. J. McCarthy, G. K. Smyth, Bioinformatics 2010, 26, [52] G. Giordano, S. van den Brûle, S. Lo Re, P. Triqueneaux, F. 139. Uwambayinema, Y. Yakoub, I. Couillin, B. Ryffel, T. Michiels, J. C. Re- [70] U.S. EPA. Methods For Derivation Of Inhalation Reference Concen- nauld, D. Lison, F. Huaux, Toxicol. Sci. 2010, 116, 682. trations (RfCs) And Application Of Inhalation Dosimetry. U.S. Envi- [53] A. A. Shvedova, A. A. Kapralov, W. H. Feng, E. R. Kisin, A. R. Murray, ronmental Protection Agency, Office of Research and Development, R. R. Mercer, C. M. St. Croix, M. A. Lang, S. C. Watkins, N. V. Konduru, Office of Health and Environmental Assessment, Washington, DC, B. L. Allen, J. Conroy, G. P. Kotchey, B. M. Mohamed, A. D. Meade, Y. EPA/600/8-90/066F. Volkov, A. Star, B. Fadeel, V. E. Kagan, PLoS One 2012, 7, e30923. [71] K. C. Stone, R. R. Mercer, P. Gehr, B. Stockstill, J. D. Crapo, Am. J. [54] V. E. Kagan, A. A. Kapralov, C. M. St. Croix, S. C. Watkins, E. R. Kisin, Respir. Cell Mol. Biol. 1992, 6, 235. G. P. Kotchey, K. Balasubramanian, I. I. Vlasova, J. Yu, K. Kim, W. Seo, [72] V. Castranova, P. A. Schulte, R. D. Zumwalde, Acc. Chem. Res. 2013, R. K. Mallampalli, A. Star, A. A. Shvedova, ACS Nano 2014, 8, 5610. 46, 642. Adv. Sci. 2020, 7, 1903200 1903200 (17 of 17) © 2020 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim