To evaluate airway responsiveness to direct stimulation, we utilized the methacholine challenge test and a whole body plethysmography. Administration of GO at the time of OVA sensitization (GO/SENT) (Figure 1) significantly increased airway responsiveness to methacholine as compared to control (PBS) and OVA-only treated mice (OVA/ALL) up to 2.5-fold (p < 0.05, Figure 2). Importantly, exposure to GO at OVA challenge (GO/CHAL) did not cause a statistically significant AHR elevation in asthmatic mice, suggesting that AHR increase was not due to acute responses to GO administration.

Microscopic evaluation of the lungs of mice in PBS control group reveals normal morphology of conducting and respiratory airways (Figure 3A). Lungs of mice exposed to GO (GO/S/cont group) revealed deposition of brown-pigmented particles in small airways and numerous interstitial aggregates of pigment-laden macrophages (Figure 3B, arrows). GO exposure also caused mild interstitial lymphocytic infiltration without evidence of airway remodeling. In OVA-treated asthmatic animals (Figure 3C), epithelial hypertrophy/hyperplasia, goblet cell hyperplasia, smooth muscle hypertrophy, and intense lymphohystiocytic infiltration were apparent, indicating airway remodeling (Figure 4). In mice treated with GO during the sensitization phase (GO/SENT), interstitial aggregates of GO-laden macrophages were seen (Figure 3D, arrows). Epithelial hypertrophy/hyperplasia in mice of the GO/SENT group was as noticeable as in the OVA/ALL group, and other morphologic features of airway remodeling were more prominent (Figure 4 and Table 1). Compared to the OVA/ALL group, a significant increase in the number of goblet cells (15 ± 1 vs 10 ± 1 per 100 μm), subepithelial fibrosis (32 ± 0.6 μm vs 22 ± 0.8 μm), and smooth muscle layer (24 ± 0.6 μm vs 15 ± 0.9 μm) was seen in OVA/SENT mice.

A robust accumulation of eosinophils in BALF, as expected, accompanied OVA sensitization and challenge in OVA/ALL mice (up to 1.1 ± 0.2 × 10⁶ cells/mL vs none in PBS control). GO exposure at the time of OVA sensitization (GO/SENT group) reduced eosinophil counts compared to OVA/ALL-treated mice ((0.5 ± 0.1) × 10⁶ cells/mL, p < 0.05). GO administration in nonsensitized animals (GO/S/cont group) did not induce eosinophil influx (Figure 5). However, GO treatment on day 28/29 (GO/C/cont group) and during OVA challenge (GO/CHAL group) facilitated transient neutrophil accumulation in BALF (Figure 5); this effect was not observed in mice exposed to GO during OVA sensitization (GO/SENT group) or on day 0 (GO/S/cont). Neutrophil counts in BAL of mice exposed to GO during OVA challenge (GO/CHAL group) were elevated up to (0.6 ± 0.3) × 10⁶vs (0.2 ± 0.1) × 10⁶ cells/mL in OVA-only (OVA/ALL group) mice and none in PBS control animals, respectively (p < 0.05). Interestingly, exposure to GO during OVA sensitization (GO/SENT group) resulted in the elevation of macrophage counts in the lungs on day 31 (up to (1.4 ± 0.1) × 10⁶ cells/mL in GO/SENT mice as compared to (0.9 ± 0.1) × 10⁶ cells/mL in OVA/ALL animals, p < 0.05) (Figure 5). In GO-treated mice, BALF macrophages contained black particles, indicating the uptake/internalization of GO by macrophages (Figure 5E,F). Lymphocyte accumulation was apparent in OVA/ALL mice and GO/SENT, but not in GO/CHAL mice on day 31 after GO exposure (Figure 5). These results indicate that GO aspiration modified the inflammatory cell pattern in asthmatic animals, reducing eosinophilic inflammation in favor of the emergence of monocytes–macrophages.

To further study the pulmonary responses to GO exposure, the levels of IL-4, IL-5, IL-13, TNF-α, IL-6, IFN-γ, IL-12p70, IL-10, and IL-17 in BAL fluid of exposed animals were determined. OVA sensitization/challenge resulted in substantial elevation of Th2 cytokines in BALF (OVA/ALL group, Figure 6, p < 0.05). Indeed, in OVA-only treated mice, IL-4 levels peaked at 260 ± 70 pg/mL. However, in GO/SENT-treated mice the levels of IL-4 were significantly decreased and averaged at 80 ± 20 pg/mL (p < 0.05, Figure 6). The levels of IL-5 in BALF were also reduced in GO/SENT mice as compared to OVA-only exposed animals (Figure 6, 400 ± 60 pg/mL vs 270 ± 55 pg/mL, respectively, p < 0.05). IL-13 reached 63 ± 17 pg/mL in OVA-only exposed mice (OVA/ALL), as compared to 21 ± 6 pg/mL in GO/SENT mice or 25 ± 10 pg/mL in GO/CHAL mice (p < 0.05, Figure 5). Increases in IL-6 and TNF-α levels were seen only 48 h after GO administration in both OVA-sensitized and vehicle-treated animals (data not shown). The levels of IFN-γ, IL-12p70, IL-10, and IL-17 in BALF were below the detection limit of the assay. Overall, GO aspiration reduced the release of Th2 cytokines in the lungs of asthmatic mice.

To evaluate the effects of pulmonary GO exposure on adaptive immune responses to OVA sensitization, the levels of total and OVA-specific IgE as well as OVA-specific IgG1 and IgG2a in the serum were assessed. OVA sensitization resulted in a dramatic increase in the serum levels of IgE and IgG1 (Figure 7), as compared to nonsensitized animals. Notably, GO aspiration substantially decreased both total and OVA-specific IgE levels (Figure 7, 2000 ± 230 ng/mL in GO/SENT mice vs 5800 ± 500 ng/mL in OVA/ALL animals, p < 0.05). OVA-specific IgG1, a Th2-associated antibody, was also significantly lower in GO/SENT mice as compared to OVA/ALL counterparts: 40 ± 3 U/mL vs 60 ± 5 U/mL, respectively (p < 0.05). Importantly, a Th1-associated immunoglobulin, IgG2a, was significantly elevated in GO/SENT mice compared to OVA/ALL controls: 2400 ± 730 U/mL vs 250 ± 12 U/mL, respectively (p < 0.05). Administration of GO alone or vehicle did not result in elevation of any of the measured immunoglobulins. Taken together, these results suggest that pulmonary exposure to GO altered immune responses to OVA, reducing Th2-dependent and activating Th1-dependent pathways.

To test whether GO exposure altered the levels of chitinases implicated in asthma development, the content of CHI3L1 and AMCase in BALF of GO-treated mice on day 7 postexposure was detected (Figure 8). In exposed animals, the levels of BALF CHI3L1 was markedly increased and peaked at 147.3 ± 26.9 pg/mL, as compared to 7.2 ± 4.4 pg/mL in control mice (p < 0.05). BALF AMCase levels were 676.9 ± 310.2 ng/mL in exposed mice vs 21.8 ± 10.1 ng/mL in controls (p < 0.05). These data provide the very first evidence that GO could stimulate chitinase accumulation in the lungs of treated mice.

To determine if AM could be the source of CHI3L1 and AMCase in the lungs of GO-treated mice, the production of these chitinases by isolated alveolar macrophages (AM) ex vivo was next assessed (Figure 8). The levels of CHI3L1 in culture supernatants of AM isolated from GO-treated animals peaked at 99.8 ± 35.7 pg/mL, as compared to 7.1 ± 3.6 pg/mL in control cells (p < 0.05). AMCase levels were 632.6 ± 171.4 pg/mL in AM cultures from exposed mice vs undetectable levels in controls. These data showed that GO directly stimulated CHI3L1 and AMCase release by AM.

To investigate if GO, similar to chitins, can directly bind/interact with chitinases AMCase and CHI3L1, we first performed molecular modeling studies to predict the interaction site of GO on chitinases. The preferred binding conformations in each case are shown in Figure 9. The GO was predicted to interact with AMCase and CHI3L1 at two different binding sites, site1 and site2. Residues that are predicted to be within 5 Å of GO at each binding site are listed in Table 2. The preferred binding site, site1, was located in close proximity to the chitin/active binding site in AMCase and the oligosaccharide binding site in CHI3L1, either completely (Figure 9A) or partially (Figure 9B) blocking the binding cavity. The predicted binding energies of GO to AMCase and CHI3L1 at site1 were −20.0 and −19.6 kcal/mol, respectively. Most importantly, the interaction of GO at site1, completely occupying the entrance of the active site cavity, overlaps with the inhibitor binding site in AMCase (Figure 9A). Especially, the predicted binding site residues W31, W99, N100, A295, and E297 (Table 2) were shown to interact with and stabilize AMCase inhibitors bisdionin-C and -F and others.^26,27 However, this was not the case for CHI3L1 protein. GO binding to CHI3L1 at site1 results in the partial occlusion of the entrance to the cavity. Notably, the chitinase-3-like proteins such as CHI3L1 lack Chitinase activity due to mutations within the active site compared to AMCase.²¹ Thus, the occlusion of the entrance to the chitin binding site by GO in AMCase—interfering with the binding/catalysis/hydrolysis of chitin polymers—could lead to the inhibition of its activity and/or signaling. Further, these results also suggest that GO interaction with chitinases, in close proximity to chitin/oligosaccharide binding site leading to immobilization of chitinases on its surface, could act as a signal of chitinase insufficiency, triggering their compensatory synthesis by macrophages.

To verify whether GO inhibits AMCase enzymatic activity, as predicted by molecular modeling, we measured chitinolytic activity in the BAL fluid of mice on day 31 post OVA sensitization and/or GO exposure (Figure 10A). The chitinolytic activity increased significantly in the BAL samples from mice that were OVA-sensitized and challenged (OVA/ALL). Treating these mice with GO during OVA sensitization (GO/SENT group) or OVA challenge (GO/CHAL group) significantly reduced OVA-induced AMCase enzymatic activity in the BALF by 56.2% and 8.5%, respectively (Figure 10A). However, GO administration in nonsensitized (GO/S/cont group) and nonchallenged (GO/CHAL group) animals did not induce any chitinolytic activity (Figure 10A). Further, to provide direct evidence for the specific inhibition of AMCase activity by GO, we evaluated the chitinolytic acivity using recombinant human AMCase (hAMCase), stably expressed in HEK-293 cells (Figure 10B). After 30 min of incubation, GO treatment resulted in a concentration-dependent inhibition of hAMCase activity with an estimated binding constant (K_D) of 326.1 ± 29.9 μg/mL. However, the inhibitory potential of GO was lower compared to a known inhibitor of AMCase, bisdionin C (Figure 10C, inset). The K_D of bisdionin C was estimated to 4.3 ± 1.2 μg/mL. Assuming that the oxygen-containing functionalities—involved in interactions with AMCase—represent only a fraction of the GO surface, our results strongly suggest that GO binds and inhibits the activity of AMCase.

Graphene oxide was synthesized and characterized as described elsewhere.⁶² The zeta potential of GO was −32.4. The average particle thickness was 0.61 nm, while the length and width varied from 20 nm to ∼5 μm. Stock suspensions (1 mg/mL) were prepared before each experiment in PBS, and the pH was adjusted to 7.0; suspensions were sonicated for 5 min with a probe sonicator (VibraCell, Sonics and Materials Inc., Newtown, CT, USA) and sterilized by autoclaving. Stock suspensions were diluted to achieve required concentrations and sonicated (three 1 min cycles) before use. Endotoxin content in GO suspensions was evaluated using Limulus amebocyte lysate chromogenic end point assay kit (Hycult Biotech, Inc., Plymouth Meeting, PA, USA) according to the manufacturer’s instructions and found to be below the detection limit (0.01 EU/mL).

Specific-pathogen-free adult female BALB/c mice (7–8 weeks old) were supplied by Jackson Laboratories (Bar Harbor, ME, USA). Animals were individually housed in the National Institute for Occupational Safety and Health (NIOSH) facilities approved by Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC). Mice were acclimated for at least 1 week. Sterile Sani-Chip bedding (Harlan Teklad, Madison, WI, USA) was changed weekly. Animals were supplied with water and food (Harlan Teklad, 7913, NIH-31 modified mouse/rat diet, irradiated) ad libitum and housed under controlled light, temperature, and humidity conditions. All experiments were conducted under a protocol approved by the Animal Care and Use Committee of NIOSH (protocol #10-AS-M-009). Six animals per study group were utilized for all in vivo assays, and all experiments were repeated at least two times.

As female mice of any strain are generally less aggressive toward each other and are less apt to fight than the male mice, they were preferred for studying GO treatment during OVA sensitization and/or challenge. While it is known that hormonal influences in females might interfere with toxicity studies, it is generally accepted that mice from the same lot may be used when large toxicology studies are employed for comparing different treatments/exposures. Thus, any hormonal effects would also be accounted for in the PBS control and OVA/ALL group; so any effects observed would be due to GO exposure.

On day 0 of an experiment, mice were sensitized to chicken egg ovalbumin (Hyglos GmbH, Germany) by intraperitoneal injection of a sterile suspension containing OVA (20 μg/mouse) and aluminum hydroxide (Al) adjuvant (1.5 μg/mouse, Sigma, St. Louis, MO, USA) in 100 μL of PBS (Figure 1). Additionally, a group of mice were exposed to GO or vehicle by pharyngeal aspiration on day 0: a suspension of GO (80 μg/mouse in divalent ion-free PBS) or PBS was placed posterior in the throat, and the tongue was held until the suspension was aspirated into the lungs. Booster injection of OVA (ip, 20 μg/mouse) or vehicle was administered on day 14. On days 28 and 29, groups were challenged with pharyngeal aspiration of OVA (10 μg/mouse/day) in PBS or treated with GO along with OVA (40 μg/mouse/day, 80 μg cumulative dose). Experimental groups are listed as follows (Figure 1): treatment group name (treatment at day 0; day 14; days 28–29): OVA/ALL (OVA sensitization; OVA boosting; OVA challenge), GO/SENT (OVA sensitization + GO; OVA boosting; OVA challenge); GO/CHAL (OVA sensitization; OVA boosting; OVA challenge + GO). Control groups included animals treated with GO only (on day 0 or days 28–29, GO/s/cont and GO/C/cont, respectively) or vehicle (PBS). Two days (48 h) after the last OVA challenge (day 31), mice were sacrificed or used for airway hyperresponsiveness evaluation. Additional groups of animals were used for chitinase measurements and were exposed to GO only as described above.

The responsiveness of mouse airways was measured on day 31 using a noninvasive whole-body plethysmograph (WBP) system (Buxco Systems Inc., Troy, NY, USA). Briefly, mice were placed in WBP of approximately 300 mL in which the animals were unrestrained and exposed to increasing concentrations (10, 25, 50 mg/mL) of aerosolized methacholine (MCh; Sigma-Aldrich). A bias flow of HEPA filtered room air of 1 LPM was drawn through each WBP. After acclimatization, the baseline was measured for 5 min prior to MCh exposure. After the baseline measurements, MCh is nebulized for 1.5 min, and pressure changes within the chamber due to respiration were monitored and recorded (2 min/concentration). The pressure signals were postprocessed using Matlab (Mathworks, Inc.) to calculate airway reactivity and expressed as enhanced pause (Penh) values. Penh is a reasonable analogue of airway responsiveness to a nonspecific inhaled stimulus, such as MCh, which provides an accepted measure for comparison between the experimental groups.

Lung tissues were harvested on day 31 of the study and inflation fixed in situ with 4% paraformaldehyde at 10 cm H₂O for 10 min with the chest cavity open. Paraffin-embedded tissue was cut at a thickness of 5 μm, stained with hematoxylin and eosin (H&E), and examined microscopically. Sample identification was coded to ensure unbiased evaluation. Additionally, a set of lung tissue sections with a constant thickness of 5 μm was stained using PAS/diastase to highlight mucus-secreting goblet cells, by trichrome blue to highlight peribronchial fibrous tissue, and by desmin immunostain (Ventana Medical Systems, Inc.) to highlight the peribronchial smooth muscle. Unbiased morphometric analysis was performed on these stained lung tissue sections according to the principles and guidelines of basic methodological standards in lung morphometry.⁶³ The thickness of respiratory epithelium, subepithelial smooth muscle layer, and subepithelial fibrous layer was measured using a calibrated micrometric analyzer (Spot Insight software 5.1). At least 10 randomly chosen regions on each slide were analyzed for estimating different morphological features including number of goblet cells, thickness of smooth muscle, and fibrosis in each case. A standard area measuring 16 400 μm² was used for the enumeration of goblet cells. For standardization, the number of mucus-secreting goblet cells in respiratory epithelium was calculated as number of cells per 100 μm using the following formula: total number of goblet cells/total airway circumference × 100.⁶⁴ These evaluations were carried out by a single board-certified pathologist (DWG) familiar with the guidelines and standards of toxicologic pathology criteria and nomenclature for mouse lungs.^63,65−67 All morphometric measurements were performed in an unbiased (i.e., blinded) fashion, where the identities of the various groups were masked and were not known by the reading pathologist.

Twenty-four hours after the last OVA challenge (day 31), mice were sacrificed by intraperitoneal injection of sodium pentobarbital and exsanguinated. The trachea was cannulated with a blunted 22-gauge needle, and bronchoalveolar lavage was performed with cold sterile Ca²⁺/Mg²⁺-free PBS at a volume of 0.7 mL for the first lavage (kept separate) and 0.8 mL for subsequent lavages. A total of 5 mL of bronchoalveolar lavage fluid per mouse was collected and pooled in sterile centrifuge tubes. Pooled BAL cells were washed in Ca²⁺/Mg²⁺-free PBS by alternate centrifugation (200g, 10 min, 4 °C). Cell-free first-fraction BALF aliquots were frozen and stored at −80 °C until processed.

The degree of pulmonary inflammatory response was estimated by the total cell counts, as well as macrophages, neutrophils, eosinophils, and lymphocytes recruited into the mouse lungs and recovered from the BALF. Alveolar macrophages, neutrophils, eosinophils, and lymphocytes were identified in cytospin preparations stained with a Hema-3 kit (Fisher Scientific, Pittsburgh, PA, USA) by characteristic cell morphology, and differential counts of BAL cells were performed. Three hundred cells per slide were counted.

Levels of cytokines were assayed in the acellular BAL fluid. The concentrations of TNF-α, IFN-g, IL-12p70, IL-10, IL-4, IL-5, IL-13, and IL-17 (sensitivity of assays is 5–7.3 pg/mL) were determined using the BD cytometric bead array, mouse Inflammation kit (BD Biosciences, San Diego, CA, USA).

Levels of total IgE, as well as OVA-specific IgE, IgG2a, and IgG1a, were determined in mouse serum (day 31) using commercially available ELISA kits (Alpha Diagnostic Intl. Inc., San Antonio, TX, USA) according to the manufacturer’s instructions. Detection of OVA-specific immunoglobulins was used for additional identification of Th1- or Th2-inducing effects of GO.

BAL cell pellets were washed three times with PBS and suspended with RPMI-1640 containing 10% heat-inactivated fetal bovine serum, 2 mM l-glutamine, 200 U/mL penicillin, and 200 mg/mL streptomycin (Invitrogen). The cell suspension was cultured overnight in 25 cm² flasks at 37 °C in a 5% CO₂ humidified milieu to permit the adherence of AMs. Then the nonadherent cells were removed by three washes with 2% RPMI-1640. The purity of adherent AMs was identified to be greater than 95% by morphology. The adherent cells were collected and seeded in 96-well plates (1.8 × 10⁶ cells/plate). Forty-eight hours later, culture supernatants were collected and frozen at −80 °C for chitinase assay.

The levels of chitinase 3-like protein 1 and acidic mammalian chitinase (AMCase or CHIA) were evaluated in BAL fluid of mice (7 days after exposure) and macrophage culture supernatants using commercially available ELISA kits. Mouse CHI3L1 ELISA assay (R&D Systems) was performed according to the manufacturer’s instructions. In brief, samples, control, and standards (4000 pg/mL) were added to a microplate precoated with a polyclonal antibody specific for mouse CHI3L1. Unbound substances were washed away, and an enzyme-linked polyclonal antibody specific for mouse CHI3L1 was added, followed by a wash to remove unbound antibody–enzyme reagents. The AMCase ELISA kit was purchased from Cusabio Biotech Co. Ltd. (China, http://www.cusabio.com/). The assay was performed according to the manufacturer’s instructions. In brief, samples and standards (20 ng/mL) were added to a microplate precoated with an antibody specific for mouse CHIA. Unbound substances were removed by washing, followed by the addition of biotin-conjugated antibody specific for CHIA and subsequent washing. Horseradish peroxidase was added, and any unbound avidin–enzyme reagents were washed away. For color development in both assays, substrate solution was added, and the optical density readings at 450–540 nm were proportional to the amount of initially bound CHI3L1/CHIA.

The three-dimensional structure of GO was docked to the crystal structure of AMCase (pdbid: 3FXY:A) and CHI3L1 (pdbid: 1HJV:A) using AutoDock Vina⁶⁸ available at http://vina.scripps.edu. While the presence of rotatable bonds on the GO ligand structure imparted flexibility, structures of chitinases were considered to be rigid for docking. The grid box was centered at coordinates 23.994, −28.478, −35.646 with 80 Å units in the x, y, and z directions for AMCase and at 24.202, 39.202, 34.858 with 80 Å units in the x, y, and z directions for CHI3L1. This grid box covered the entire structure of each chitinase, making the docking unbiased for different binding sites. The resulting orientations in each case were clustered based on the position of binding on the receptor structure. The best ligand-bound chitinase receptor structure in each case was chosen based on lowest energy as well as the total number of conformations in that binding site.

The activity of AMCase was measured fluorimetrically on the basis of enzymatic hydrolysis of the 4-methylumbelliferyl-β-d-N,N′,N″-triacetylchitotriose (4-MU-chitotrioside, a substrate for endochitinase) with production of 4-methylumbelliferone (4-MU) using CycLex acidic mammalian chitinase fluorometric assay kit (MBL International, Woburn, MA, USA). A 10 μL volume of BAL samples (after 200-fold dilution) was used for the enzymatic assay. Fluorescence was measured at an excitation of 340 nm and emission of 460 nm for 120 min with 5–10 min intervals at 30 °C. The activity of AMCase in different samples was estimated by dividing the relative fluorescence vs reaction time (min). The inhibitory effects of a known chitinase inhibitor, bisdionin C (IC₅₀: 20 μM), and GO were investigated using recombinant human AMCase. Briefly, 10 μL of purified hAMCase (0.3 ng/μL) and 10 μL of 0.2 mM 4-MU-chitotrioside were added to 80 μL of chitinase assay buffer containing various concentrations of GO (0–1.2 mg/mL) or bisdionin C (0–8 μg/mL or 0–20 μM). The results of the inhibition assays are reported as percent decrease in activity (ratio of hAMCase activity in the presence and absence of inhibitor). All samples were assayed in triplicates. The binding constants of GO and bisdionon C to AMCase were estimated by fitting the percent decrease in relative fluorescence using nonlinear regression.

Values are presented as mean values ± SEM. The significance of treatment-related differences was evaluated using either two-tailed Student’s t test or a nonparametric Kruskal–Wallis ANOVA on Ranks followed by the Holm–Sidak test. p values of less than 0.05 were considered to be statistically significant.