General
The present studies were conducted according to the OECD Principles of Good Laboratory Practice [43], which principally meet the United States Environmental Protection Agency Good Laboratory Practice Standards [40] CFR Part 160 (FIFRA) and Part 792 (TSCA)]. The study was conducted referring to OECD Guideline 412 [44].
Test materials and characterization
MWCNT (Graphistrength™ C100, purity >95%), was manufactured by Arkema (Lacq, France). Graphene was produced by thermal shocking of graphite oxide resulting in simultaneous exfoliation and chemical reduction. Graphite nanoplatelets were obtained by flash heating of intercalated graphite. Low surface Carbon Black (37 m2/g) was provided by Evonik-Degussa (Germany). The test materials were examined for the applicable physical-chemical endpoints, including representative images (SEM), particle size distribution (SEM) of both primary structure and agglomerates, crystallite size (X-ray diffraction (XRD)), crystalline phase (XRD), spectroscopy (Raman), surface chemistry (XPS) and secondary ion mass spectrometry (SIMS), dispersability in water and in serum-containing media (by analytical ultracentrifugation (AUC)), pore sizes (Hg intrusion) and the derived values of specific surface area, apparent density, and agglomerate density. Additionally, the materials were examined for endotoxin content using the chromogenic Limulus Amebocyte Lysate Kinetic-QCL Test [45]. No endotoxins were detected in MWCNTs, Graphene, and Graphite Nanoplatelets, and only a negligible endotoxin content was measured in Carbon Black (0.05 EU/mL).
Animals
Permission for animal studies was obtained from the local regulatory agencies, and all protocols were in compliance with the federal guidelines. The studies were performed in an AAALAC-approved laboratory in accordance with the German Animal Welfare Act and the effective European Council Directive. Male Wistar (strain Crl:WI (Han)) rats (7 weeks of age) were obtained from Charles River Laboratories (Sandhofer Weg, Sulzfeld, Germany) and were allowed free access to mouse/rat laboratory diet (Provimi Kliba SA, Kaiseraugst, Switzerland) and water. The animals were housed singly in mesh floored cages in accommodation maintained at 20 to 24°C, with a relative humidity of 30 to 70%, a light/dark cycle of 06.00 to 18.00 h light and 18.00 to 06.00 h dark and were allowed to acclimatize to these conditions for approximately two weeks before commencement of the study.
Exposure regimen/test groups
Study design was similar in all experiments, but different in details as outlined below.
MWCNT and Carbon Black
Groups of 11 male Wistar rats were head-nose exposed to respirable dusts on 6 hours per day, on 5 consecutive days (days 0 to 4). The target concentrations were 0.1, 0.5, or 2.5 mg/m3 (MWCNT), or 0.5, 2.5, or 10 mg/m3 (Carbon Black). A concurrent control group was exposed to conditioned air. On study day 4 (after the last exposure) and 25 (21 days after the last exposure), 6 animals per group were sacrificed and designated for histopathological examinations. On study day 7 (3 days after last exposure) and 28 (24 days after last exposure), the remaining 5 animals per group were sacrificed. The lungs of these animals were lavaged, and BALF was analyzed for markers indicative for injury of the bronchoalveolar region.
Graphene
Groups of 8 male Wistar rats were exposed to target concentrations of 0.5, 2.5, or 10 mg/m3 or to conditioned air for 6 hours per day for 5 consecutive days (days 0 to 4). Animals were sacrificed on study day 7 and 28. On each sacrificing day, 3 animals per group were designated for histopathological examination and the remaining 5 animals per group for bronchoalveolar lavage. Sampling for histopathological examination was performed three days later compared to the other materials due to logistical reasons.
Graphite nanoplatelets
Groups of 8 male Wistar rats were exposed to 0.5, 2.5, or 10 mg/m3 or to conditioned air for 6 hours per day for 5 consecutive days (days 0 to 4). Animals designated for histopathological examinations were sacrificed on study day 4, 25 (only controls and 10 mg/m3), and 95. BALF was collected on study day 7, 28, and 98. On each sacrificing day, 3 animals per group were designated for histopathological examination and 5 animals per group for bronchoalveolar lavage.
Generation of the test atmospheres
Brush dust generators (developed by the Technical University of Karlsruhe in cooperation with BASF, Germany) served for generation of test atmospheres with MWCNT, graphite nanoplatelets, and Carbon Black. In case of graphene, test atmospheres were produced with swinging bed dust generators (in house development by BASF), since it was not possible to generate adequate aerosols of graphene with the brush dust generator.
Generated dusts were mixed with compressed air (filtered air pressurized to about 6 bar, flow rate 1.5 ± 0.3 m3/h) in a glass tube, diluted with conditioned air (activated charcoal-filtered air, 22 ±2°C, 50 ± 20% relative humidity, flow rate 4.5 ± 0.3 m3/h) and passed via a cyclone into the inhalation system. The 90-l cylindrical stainless steel inhalation chamber was fed via a cone-shaped inlet at the top and exhausted at the opposite end. The desired inhalation chamber concentrations were achieved by withdrawing/exhausting and replacing a portion of the dust aerosol air with conditioned supply air immediately before entering the chamber (6 m3/h). Mean flow rate through the inhalation chamber, measured at exhaust air, was 5.4 ± 0.3 m3/h for all concentrations, that is, air was changed in the inhalation chambers about 67 times per hour.
A schematic diagram of the inhalation system is shown in Figure 7.
Monitoring and characterization of the test atmosphere
Compressed and conditioned supply air and exhaust air flow rates, chamber temperature and humidity were measured automatically with appropriate sensors/orifice plates; data were saved every 10 s and retained for analysis. To ensure the stability of the dust aerosols, the inhalation chambers were monitored continuously during exposure using scattered light photometers (VisGuard; Sigrist-Photometer AG, Switzerland; for details see [27]). To quantify the atmospheric dust concentration, gravimetric measurements of air samples taken adjacent to the animals’ breathing zone were performed (probe internal diameter 7 mm). A defined volume of sample air was drawn by vacuum pump across a binder-free glass-fiber filter paper (Macherey-Nagel MN 85/90 BF, diameter 4.7 cm).
Aerosol dust concentration was calculated as the increase in weight of the filter after sampling, divided by sample volume at test conditions (22°C, atmospheric pressure, 50% relative humidity). As a rule, two samples were taken per exposure and concentration group. The duration of sampling was adjusted to the test substance concentration in the chamber to obtain a total sample weight of 1 to 5 mg. Thus, the volume of the air samples varied with the atmospheric concentration. To determine the MMAD (the calculated aerodynamic diameter which divides the size distribution in half when measured by mass), cascade impactor measurements were performed with a Sierra Marple 298 cascade impactor. The effective aerodynamic cut-off diameters were 21, 15, 10, 6.5, 3.5, 1, 0.7, and 0.4 μm. To capture the particles < 0.4 μm, the impactor was equipped with a backup filter. The deposition on each impactor stage as well as on the backup filter was determined gravimetrically. Particle size distributions were calculated according to DIN 66141 and DIN 66161, i.e. linear regression of cumulative percent (probit values) versus logarithms of effective cut-off diameters. Particle size distributions measured by cascade impactor were expressed as MMAD and geometrical standard deviation (GSD). Additionally, a light-scattering spectrometer (WELAS 2100; Palas, Karlsruhe, Germany) was used for particle sizes from 0.24 to 10 μm (at least 10 repeats). In the submicrometer range (11 to 1083 nm), particle size distribution was measured with a Scanning Mobility Particle Sizer equipped with a condensation particle counter (SMPS + C) (Grimm Aerosol Technik GmbH, Ainring, Germany). Particles were classified by electrostatic fractionation of the different sized particles. Particle counts in each of the fractions were counted by condensation particle counter. The SMPS spectrometer uses electrical mobility to measure the particle size. This technique utilizes a bipolar charger to impact a known charge distribution on the aerosol sample, and classify particles according to their ability to traverse an electric field. The data are interpreted based on a spherical particle model. While this method is appropriate for sizing spherical particles, it leads to errors in the mobility size data interpretation for agglomerates and aggregates. As the aerosols examined in this paper are mainly irregular agglomerates consisting of either tubes or sheets, the relevance of using SMPS is somewhat limited.
Additionally, samples of the test atmospheres of MWCNTs, graphene and graphite nanoplatelets were collected using gold-coated capillary filters and evaluated by scanning electron microscopy. In case of MWCNTs, the test atmosphere was generated under the same conditions as in the toxicity study at a later time point.
Animal exposure
During exposure, rats were restrained in glass tubes fixed to the inhalation chamber walls with their snouts projecting into the inhalation chamber (head/nose exposure). Overpressure was maintained inside the inhalation chamber to ensure that the aerosol in the animals’ breathing zone was not diluted by laboratory air. The exposure systems were kept under exhaust hoods in an air-conditioned room.
Clinical observations
Health status and cage-side clinical signs were checked at least once daily (on exposure days before, during and after exposure). Body weights were measured weekly.
Hematology and acute phase proteins
Before sacrifice, blood samples for hematology and clinical chemistry were taken from all animals designated for collection of BALF by retrobulbar venous plexus puncture under isoflurane (Isoba®, Essex GmbH Munich, Germany) anesthesia. Hematology (ADVIA120 Instrument, Siemens, Germany) comprised red blood cell counts, hemoglobin, hematocrit, mean corpuscular volume, mean corpuscular hemoglobin content, mean corpuscular hemoglobin concentration, platelets, total white blood cell and differential blood cell counts.
Acute phase proteins: rat α2-macroglobulin was measured with an MTP-ELISA (Kamiya Biomedical Company, Seattle, USA, cat no. KT-353), measured with a Sunrise MTP Reader, Tecan AG, Switzerland, by using the Magellan Software provided by the instrument producer. Haptoglobin was measured photometrically based on the preservation of the hemoglobin peroxidase activity (Tridelta Ltd, Maynooth, Ireland, cat no. TP-801), running on a Hitachi 917 instrument, Roche, Basel Switzerland.
Broncho-alveolar lavage and lung tissue homogenate
To obtain BALF, animals were killed by exsanguination under Narcoren® anesthesia and the lungs lavaged twice with 6 ml (22 ml/kg body weight) physiological saline. The two washes were combined (an average of 11 ml of lavage fluid was recovered per animal) and aliquots of the combined washes were used for the determination of cytology, total protein concentration and enzyme activities, as well as mediators.
Total BALF cell counts were determined with an Advia 120 (Siemens Diagnostics, Fernwald, Germany) haematology analyzer. Counts of macrophages, polymorphonuclear neutrophils, lymphocytes, eosinophils, monocytes and atypical cells were performed on Wright-stained cytocentrifuge slide preparations as described by Warheit and Hartsky [46]. The differential cell count was evaluated manually by counting at least 400 BALF cells per sample. The following parameters were measured with a Hitachi 917 (Roche Diagnostics, Mannheim, Germany) reaction rate analyzer: total protein (turbidimetric method with Benzethonium chloride), LDH (EC 1.1.1.27; kinetic UV test, 340 nm, 37°C acc. to IFCC), ALP (EC 3.1.3.1; kinetic colour test, 450 nm, 37°C acc. to IFCC), NAG (EC 3.2.1.30; colour test, 580 nm, 37°C) [47] and GGT (EC 2.3.2.2, Szasz method) (kinetic colour test, 415 nm, 37°C acc. to IFCC) activities.
In case of MWCNT mediators in BALF were measured at Rules-based Medicine Inc., Austin, TX, USA, with xMAP technology (Luminex Corp., Austin, TX, USA) as described previously [48–51]. These parameters comprised various cytokines, chemokines, adhesion molecules, matrix metalloproteinases, acute phase proteins, signal proteins of apoptosis or cell proliferation: apolipoprotein A1, β-2 microglobulin, calbindin, CD40, CD40L, clusterin, C-reactive protein, cystatin C, epidermal growth factor (EGF), endothelin-1, eotaxin, factor VII, fibroblast growth factor (FGF)-basic, FGF-9, fibrinogen, GCP-2, granulocyte-macrophage colony-stimulating factor (GM-CSF), growth hormone, glutathione-S-transferase (GST)-α, GST-1 Yb, haptoglobin, interferon (IFN)-γ, IgA, IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-10, IL-11, IL-12p70, IL-17, insulin, interferon gamma-induced protein 10 (IP-10), KC/GROα, leptin, leukemia inhibitory factor (LIF), lymphotactin, MCP-1, MCP-3, MCP-5, M-CSF, macrophage-derived chemokine (MDC), MIP-1α, MIP-1β, MIP-1γ, MIP-2, MIP-3β, matrix metalloproteinase-9 (MMP-9), myoglobin, MPO, NGAL, oncostatin M (OSM), osteopontin, regulated on activation, normal T cell expressed and secreted (RANTES), stem cell factor (SCF), serum amyloid P, serum glutamic oxaloacetic transaminase (SGOT), tissue inhibitor of metalloproteinase-1 (TIMP-1), tissue factor, tumor necrosis factor α (TNF-α), thrombopoietin (TPO), vascular cell adhesion molecule-1 (VCAM-1), VEGF, and van Willebrand factor. Additionally, the cytokine-induced neutrophil chemoattractant-1/IL-8 (CINC-1/IL-8) levels in BALF and lung tissue homogenates were measured with an ELISA produced by R&D Systems Inc., Minneapolis, US, (Quantikine rat CINC-1, cat. no. RCN100) by using a Sunrise MTP Reader, Tecan AG, Switzerland, with the Magellan Software provided by the instrument producer. The mediator level changes were regarded relevant when their levels were above the detection limit in controls and/or dose group samples and when they were at least increased above 2-fold in the (high) dose group. Based on these results and taken into account previous data of the whole panel of mediators the following parameters were considered to be most suitable for characterizing of lung inflammation and were measured in case of graphene, graphite nanoplatelets and Carbon Black: rat monocyte chemoattractant protein-1 level (rat MCP-1); instant ELISA, Bender MedSystems, Vienna, Austria (cat. no BMS631INST); rat CINC-1/IL-8 level (ELISA, R&D Systems Inc., Minneapolis, US), (Quantikine rat CINC-1, cat. no. RCN100); macrophage colony stimulating factor (M-CSF; Quantikine Mouse M-CSF ELISA, R&D Systems Inc., Minneapolis, USA) (cat no. MMC00); rodent osteopontin; ELISA, R&D Systems, Inc., Minneapolis, US (Quantikine mouse osteopontin, cat. no. MOST00). The mediators were measured at a Sunrise MTP Reader, Tecan AG, Switzerland, by using the Magellan Software provided by the instrument producer. The monitoring panel was arranged so that different functional groups of mediators were covered, i.e., CC-chemokines (MCP-1); CXC-chemokines (IL-8/CINC-1); hematopoiesis (M-CSF); proliferation of sessile cells (osteopontin).
In order to evaluate if mediators were not washed out into the BALF (i.e. remained in the lung), the parameters were measured also in lung tissue homogenates: After bronchoalveolar lavage was performed the right lung portion of each animal was resected and stored at −80°C until lung tissue homogenate preparation: 0.2 g of the main lobe (lobus caudalis dexter) was mixed with 0.8 ml ice-cold Tissue Protein Extraction Reagent (T-PER, cat. no. 78510, Pierce Biotechnology, Rockford, IL, USA) added with a Complete Protease Inhibitor Cocktail (cat no., 11 873 580 001, Roche, Basel, Switzerland), homogenized for up to 40 seconds with an Ultra-Turrax (IKA, Staufen, Germany), and the homogenate centrifuged at 14,000 g and 4°C for 5 min.
In case of MWCNT, mediators in lung tissue homogenate were measured at Rules-based Medicine Inc., Austin, TX, USA, with xMAP technology (Luminex Corp., Austin, TX, USA) as mentioned above. The mediator level changes were regarded relevant when their levels were above the detection limit in controls and/or dose group samples and when they were at least increased above 2-fold in the (high) dose group. Comparing the mediator levels in BALF and lung tissue homogenates, together with results of previous studies, revealed that two mediators (IL-1α and TNFα) were consistently more pronounced increased in the lung tissue. Therefore, both parameters (cytokines indicating local inflammation) were used for monitoring in the studies with Graphene, Graphite Nanoplatelets and Carbon Black: rat IL-1α; FlowCytomix Rat IL-1α Simplex Kit; Bender MedSystems, Vienna, Austria (cat. no. BMS8627FF). The measurement was performed at the FACS Calibur flow cytometer, Becton Dickinson, Heidelberg, Germany and the evaluation was made with the FlowCytomix Pro Software, vs. 2.3, Bender MedSystems, Vienna, Austria; rat tumor necrosis factor-alpha (rat TNFα); Quantikine rat TNFα/TNFSF1A ELISA; R&D Systems Inc., Minneapolis, US, (cat. no. RTA100). The measurements were performed with a Sunrise MTP Reader, Tecan AG, Switzerland, by using the Magellan Software provided by the instrument producer.
Pathology
Animals were euthanized by exsanguination under Narcoren® anesthesia. Gross necropsy was carried out. Lungs were instilled at a pressure of 20 to 30 cm of water. Weights of brain with olfactory bulb, lungs, and mediastinal lymph nodes were determined.
Brain, head (with oropharynx), larynx, lungs, mediastinal lymph nodes, and trachea were fixed in 4% buffered formaldehyde (corresponding to 10% formalin), paraffin embedded, sectioned and stained with hematoxylin-eosin for histopathology.
Light microscopic examination was performed on the respiratory tract comprising nasal cavity (four levels), larynx (three levels), trachea (longitudinal with carina), lung (five lobes), and mediastinal lymph nodes.
Statistical analysis
Dunnett’s test [52, 53] was used for simultaneous comparison of all concentration groups with the control group for body weights and body weight changes. Clinical pathology parameters were analyzed by non-parametric one-way analysis using the Kruskal-Wallis test (two-sided). If the resulting p-value was equal or less than 0.05, a pair-wise comparison of each dose group with the control group was performed using Wilcoxon-test or Mann–Whitney-U-test (two-sided) for the equal medians. Statistical significance was defined as p ≤ 0.05 compared with the control group [54]. Organ weights were compared among groups by nonparametric one-way analysis using the two-sided Kruskal–Wallis test, followed by a two-sided Wilcoxon test for the hypothesis of equal medians in case of p ≤ 0.05 [55–57].