Chemicals and reagents

All used chemicals and solvents were of analytical grade. Most chemicals and salts as well as ethyl methanesulfonate (EMS), Triton™ X-100, low-(LMA; peqGold No. 35-2010) and normal-melting point (NMA; peqGold No. 35-1010) agarose, and thioglycolate broth were obtained from Merck/ Sigma-Aldrich Chemie GmbH (Taufkirchen, Germany) and Dulbecco’s Modified Eagle’s cell culture medium (DMEM) with high glucose (4.5 g/L), GlutaMax™, sodium pyruvate and gentamicin from GIBCO/Invitrogen (Karlsruhe, Germany). Fetal calf serum (FCS) and normal DMEM were sourced from PAN-Biotech (Aidenbach, Germany) and the ready-to-use solution of 10,000 µg/ml streptomycin sulfate and 10,000 U penicillin G, sodium salt from Euroclone (Pero, Italy). ATCC Kaighn's modification of Ham’s F-12 medium (F-12 K), 24-well plates with hydrophobic culture surface (1.9 cm2 cell culture surface per well), one-well Nunc™ Lab-Tek™ II glass chamber slides, and one-well ClipMax chamber slides were purchased from (Thermo Fisher Scientific, Germany), whereas ethidium bromide solution was obtained from Merck-Millipore (Darmstadt, Germany), Vectashield® H-1000 mounting medium from (BIOZOL, Eching, Germany), the Rat CXCL1/CINC-1 DuoSet® ELISA kit from R&D Systems (Bio-Techne GmbH, Wiesbaden, Germany), the “Cytotoxicity Detection Kit” from Roche Diagnostics (Mannheim, Germany), and purpose-made slides with one roughened surface from Menzel Gläser (Brunswick, Germany).

Preparation and physicochemical analysis of the exemplary MMVF sample

For the present study, the bio-soluble stone wool fiber RIF56008, formerly shown to be Note Q-compliant in rat lungs after intratracheal instillation, was chosen as exemplary MMVF. RIF56008 was produced by ROCKWOOL A/S (Hedehusene, Denmark). The specific fiber sample was selected from already-sized fiber fractions obtained during in vivo biopersistence testing in compliance with Note Q of Regulation (EC) 1272/2008. According to Note Q, carcinogen classification does not need to apply if the fiber fraction (length > 20 μm) exhibits a weighted half-life of < 40 days. As in vivo biopersistence testing of MMVF is performed with binder-free fibers, binder-free material was also used for the present in vitro screening approach. RIF56008 was originally produced using cascade spinning technology (Richet 2021) and supplied as bulk material (fiber wool). The material density amounted to 2.7 g/cm3. Chemical characterization of the bulk material was conducted by the Fraunhofer Institute for Silicate Research ISC (Würzburg, Germany). Chemical analysis of SiO2 content was done according to DIN 52340-2:1974. Chemical composition of element oxides was analyzed based on DIN 51086-2:2004, using optical emission spectroscopy inductively coupled to plasma (ICP-OES) and was determined as [wt%]: 38.0 SiO2, 18.5 Al2O3, 0.6 TiO2, 8.7 Fe2O3, 29.2 Σ CaO + MgO, 3.3 Σ Na2O + K2O plus traces of some other metal oxides. For sizing of RIF56008 bulk material, two-step aerosol separation technique was used. In the first step, the bulk material was aerosolized by a suitable dispersion technique, followed by splitting of the airborne fibers into two fractions using an inertial classifier. The coarse particles and fibers were collected by a virtual impactor, whereas the fine particles and fibers were sampled downstream of the separator using a filter. In the present study, a respirable fiber fraction with a geometric mean diameter (GMD) of 0.81 µm for fibers longer than 20 µm was finally used. The GMD of the WHO fiber fraction amounted to 0.63 µm. Re-characterization of the fiber fraction was conducted within this study to confirm appropriate length and fiber distributions (see below), according to the EU protocol ECB/TM/27 rev.7 (European Commission 1999) and the respective German regulation, i.e., “Technische Regel für Gefahrstoffe” (TRGS) 905 (BAuA 2016), to mimic as far as possible the in vivo testing situation.

Reference materials

As vehicle controls and exposure media, standard growth media for primary rat alveolar macrophages (AM) and NRM2 cells were used (see “Cell models”). Ground RIF56008 from the same fiber batch (identical chemical composition), also provided by ROCKWOOL A/S, served as non-fibrous, particle-like reference material for differentiation of chemical and morphological effects. The material control was prepared by effectively destroying fiber morphology by grinding, using a Retsch Vibratory Disc Mill RS 200 (Retsch GmbH, Germany) with the disk material made of wolfram carbide. The material was crushed four times for 30 s at a speed of 1200 revolutions per min (rpm) plus six times for 45 s at 1200 rpm. The ground material was characterized afterwards by scanning electron microscopy (SEM) to show effective depletion of fibers and to characterize the obtained particle fraction (for results see Tables 2 and 3). As bio-insoluble fiber reference, long amosite asbestos (Johns Manville Corp., Littleton, CO, USA) was used, which often served as a positive control in fiber carcinogenicity studies. Raw long amosite asbestos was milled for 30 s at full speed, using a Moulinex grinder (Type AR100G31) to obtain respirable material, which was subsequently characterized by SEM to obtain the length and diameter distribution (for results see Tables 2 and 3).

Analysis of length and diameter distributions and calculation of specific surface area

Length and diameter distributions of RIF56008, RIF56008 ground and amosite asbestos were determined using a scanning electron microscope (SUPRA 55, Carl Zeiss NTS GmbH, Oberkochen, Germany). The general characterization principles used in the context of in vivo biopersistence tests were followed. All materials were initially subjected to low-temperature ashing before being suspended in dispersion medium (Porter et al. 2009; amosite asbestos) or filtered water (RIF56008). Before SEM analysis, the RIF56008 and RIF56008 ground suspensions were sonicated for about 1 min using a Sonorex RK 510H device at 35 kHz and 160 W for 1 min. For amosite asbestos, ultrasonic treatment was done for 10 min using a VS 70 T sonotrode on a Sonoplus HD 2070 ultrasonic homogenizer (Bandelin, Berlin, Germany) at 90% duty cycle and 100% amplitude. Small fractions of the different materials (about 0.01–0.04 mg per 25 mm filter) were then diluted in about 10 ml of filtered water and filtered onto Nuclepore filters (25 mm diameter, pore size 0.2 µm). The fiber containing filter was finally mounted on an aluminum stub and sputtered (Quorum Q 150R ES) with a layer of about 20 nm of gold.

The general guidelines, as described by the World Health Organization's Regional Office for Europe (WHO/EURO) Technical Committee for Monitoring and Evaluating Airborne MMVF (WHO 1985), were followed for counting and size characterization and were adapted to synthetic mineral fibers. In brief, for measurement of length and diameters an SEM magnification of at least 2000× was used. All visible objects were counted. An object was considered as fiber, if the length-to-diameter ratio was at least 3:1. All other objects were considered as particles. Fibers crossing the boundary of the visual field were counted according to the following rules, i.e., fibers with only one end in the field were weighted as half of a fiber, and fibers with neither of their ends in the field were excluded. Fibers diameters were measured at full screen magnification, i.e., up to 18,000×. Length and diameter were recorded individually for each object measured.

A total of about 0.15 mm2 of the filter surface (for 25 mm filters) was examined. For fibers, a size-selected analysis using a minimum of 100 fibers per category for the two length categories < 5 µm and > 20 µm, and a minimum of 200 fibers for the length category > 5 µm as well as < 20 µm was used. The distance between two visual fields analyzed was at least 10 fields. Sizing was stopped when 1 mm2 of the filter surface was examined, even if the minimum number of fibers was not reached for a category. The total number of fibers per filter was determined by normalizing the surface area counted to the total surface area of the filter. For particles, recording was stopped, when a total of 100 particles was reached (for representative SEM pictures see Fig. 1). Additionally, fiber and particle concentrations per mg material were determined for all samples to add in definition of appropriate and relevant concentration levels for cell exposure.

Fig. 1

Representative SEM images of the material samples used in the present study. A RIF56008 (magnification: 2000×), B RIF56008 ground (magnification: 5000×), C amosite asbestos (magnification: 2000×)

Total specific surface area was finally calculated from the objects measured during SEM analysis, assuming cylindrical geometry for fibers and an ellipsoid shape for particles. Length and diameter values were used to calculate the volume and surface of objects. The mass was determined by multiplying the volume with a standard material density of 2.7 g/cm3 for stone wools. To calculate the total surface area the total surface of objects was divided by the total mass of objects.

Approach for definition of dose metrics and test concentrations

For establishment of a meaningful and predictive MMVF-adapted in vitro screening tool, definition of test concentrations is of utmost importance to avoid artificial results based on overload scenarios. Equivalent human exposure conditions should ideally be mimicked or at least included into dose considerations such as appropriate route of exposure as well as dosimetry aspects. Respirability and deposition fractions of fibers in the respiratory tract regions depend amongst others on shape and size. The used approach for definition of meaningful in vitro concentrations thus focused on respective information gathering, calculations and decision making to obtain a scientifically sound basis for the choice of three relevant in vitro concentrations. Extrapolation from the human external dose to deposition in the lung, and the relationship of lung volume to cell and/or cell culture surface of in the vitro model system was considered. Therefore, different topics are discussed below, i.e., (i) definition of relevant metrics; (ii) concentration definition, referring to human exposure situation at workplaces or in private settings; this implies knowledge of occupational exposure limits or derived threshold values as well as measured exposure data, and (iii) calculation/modelling of deposited dose in the lung.

Definition of relevant metrics

While fiber diameter and length influence both deposition, clearance and bioavailability/biopersistence (e.g., Roggli 2015), other dose metrics such as surface area and number of particles are being explored currently as potentially more mechanistically relevant. When comparing different types of particles, the inhaled dose can be expressed in terms of particle volume, particle surface area or number of particles (Oberdörster et al. 1994; Jarabek et al. 2005; Kuempel et al. 2012, 2015). However, in our present approach, we had to compare a mineral wool (RIF56008) and amosite asbestos to a non-fibrous dust control (ground RIF56008). As a common exposure metrics for both fibers and particles was needed, our methodology/approach narrowed down to fiber/particle mass rather than fiber/particle number to be the most adequate metric. Already the used MMVF fraction itself represents a mixture of different size fractions, as does the particle-like material reference. From this point of view, mass is probably the most appropriate and the only possible metric to compare particulate materials of different morphology. Nevertheless, fiber number and specific surface area were reported as additional information and were used for deriving correlations, when considering the two fiber types.

Dose definition: thresholds and exposure data

Measured burden of airborne respirable fibers at workplaces is usually low (< 1 fiber/ml air). Recent studies have shown that occupational exposure concentrations have not increased in the last decades (Marchant et al. 2021) and are still below one fiber per cm3, which represents the most frequent occupational exposure limit for biopersistent mineral wools. However, exceptions might occur during blowing or spraying operations, i.e., during the insulation of aircraft. Here mean levels of up to 1.8 fibers/ml and 4.2 fibers/ml had been detected for fibrous glass and mineral wools, respectively. Mean concentrations during installation of loose fill in confined spaces have revealed up to 8.2 fibers/ml (Occupational Safety and Health Series No. 64, 1989), whereas values up to 10 mg/m3 air were obtained for non-occupational exposure with highest burden assumed in old houses. In Germany, the general dust limit value (“Allgemeiner Staubgrenzwert”) was deduced and laid down in the TRGS 900 (BAuA 2006), which also applies for non-carcinogenic fibers, e.g., mineral wools that had passed the criteria according to Note Q and TRGS 900. This dust limit value was re-evaluated in 2014 by the German Committee on Hazardous Substances (AGS) and was set to 1.25 mg/m3 respirable dust (“Alveolengängige Fraktion”, “A-Staub”; referred to a density of 2.5 g/cm3). The Permanent Senate Commission for the Investigation of Health Hazards of Chemical Compounds in the Work Area of the “Deutsche Forschungsgemeinschaft” (MAK Commission) has defined a limit value for granular biopersistent dust in a comparable range (0.3 mg/m3 respirable dust, referring to a density of 1 g/cm3) (DFG 2014).

Deposited dose in the lung as a basis for in vitro testing

To define a meaningful, data-derived in vitro concentration, conversion of an external exposure level or limit value to a resulting internal deposited dose on the lung surface is necessary. In a first step, the concentration for the RIF56008 fiber sample was defined, and the concentrations of the material reference i.e., ground RIF56008 was then adapted to the fiber sample concentration.

Initially, the GMD of the WHO fiber fraction, as obtained from SEM measurements was subjected to multiple path particle dosimetry (MPPD) calculations. The MPPD model provides a mechanistic modeling to predict deposition and retained doses in lung and has been used in various applications to predict doses of inhaled particles, including elongated mineral particles (e.g., Jarabek et al. 2005; NIOSH 2013; Asgharian et al. 2018). For respective calculations, default values and MPPD settings (human, MPPD 3.04), as defined in TRGS 910 (BAuA 2014) were used (see Supplementary Tab. S1). In addition, models for both rat and human were calculated to bridge the results from in vivo biopersistence studies to the hazard for human beings at workplaces or in private settings. However, MPPD was originally designed for particles and mass median aerodynamic diameter (MMAD) values are a prerequisite for calculation. The MAK Commission stated that for particles including fibrous structures with a diameter of > 0.5 μm the aerodynamic diameter is always the most relevant dimension. Here, the aerodynamic diameter is essentially determined by the diameter with length being of lower influence. For long fibers (length >  > diameter) the MAK Commission concluded on an aerodynamic diameter of 3 times the fiber diameter, supported by data from Sturm et al. (2009, 2021), who theoretically modelled the deposition and clearance of fibers with variable sizes. The authors had chosen for their mathematical modelling of deposition and clearance, fibers with an aspect ratio varying between 3 and 100 and a diameter ranging from 0.001 to 10 µm to cover a broad spectrum of inhalable particles.

As an approach, the general dust limit value (1.35 mg/m3 for a density of 2.7 g/cm3) served as an appropriate external exposure concentration, and the GMD values obtained by SEM (see Table 2), multiplied by a factor of three were used as MMAD values. A density of 2.7 g/cm3 was assumed to be applicable for mineral wools and 3.4 g/cm3 was used for amosite asbestos. Respective results for the deposited fractions are depicted in Table 1 and served as a basis for calculation of the deposited mass per cm2 alveolar surface.

Table 1 Calculation of theoretical deposition fractions in human lung using MPPD 3.04

The present aim was to make a step forward in developing a predictive in vitro screening tool, adapted to MMVF, ideally using human-relevant doses, and taking into account the difference between short-term tests and long-term exposures in humans. But sensitivity of the in vitro test systems used must also be considered adequately. For this reason, we had a closer look into results from in vitro studies carried out with amosite asbestos. In primary human mesothelial LP9 cells, amosite asbestos showed a small but significant effect on cytotoxicity after 24 h of incubation at a concentration of 5 µg/cm2 cell culture surface. Cell proliferation was inhibited, but lactate dehydrogenase (LDH) release, indicative for membrane damage, was not increased (Reamon-Büttner et al. 2021). In another study, focusing on formation of reactive oxygen species (ROS), asbestos showed effects in the range between 2.5 and 10 µg/cm2 (Hansen and Mossman 1987). Ljungman et al. (1994) detected an asbestos-induced tumor necrosis factor-alpha (TNF-α) release at about 20 µg/cm2. From our considerations about dose (metrics) and the calculations of a human equivalent dose (deposited mass on alveolar surface is in the range of 0.5 µg/cm2 per working year) in combination with results from in vitro testing of asbestos fibers (effects at ≥ 5 µg/cm2), we finally suggested 0.5 (in vivo-relevant concentration for occupational exposure), 5 (in vitro concentration at which initial effects are expected to occur) and 50 µg/cm2 (supposed overload concentration) to represent meaningful in vitro concentrations.

Cell modelsPrimary rat alveolar macrophages

For this orienting study, primary rat alveolar macrophages (AM) were used as one of the two lung-relevant in vitro cell models. AM are the first side of contact for fibers in the lung and represent a very sensitive test system for in vitro screening experiments with particulate matter (Ziemann et al. 2014, 2017). Cells were isolated from healthy Wistar rats [strain Crl:WI(Han); Charles River, Sulzfeld, Germany] by bronchoalveolar lavage in compliance with the Federal Act on the Protection of Animals (“Tierschutzgesetz”, Bonn, Germany, last revised December 20, 2022). After centrifugation of the cell containing lavage fluid (300 × g, 10 min, 4 °C), the supernatant was discarded, the cell pellet resuspended in cell culture medium, and cells counted and plated at a density of 1.2 × 105 cells in 500 µl of cell culture medium in 24-well plates with hydrophobic culture surface (1.9 cm2). To estimate material uptake by fluorescence-coupled darkfield microscopy, AM were plated at a density of 6 × 105 cells in 2 ml cell culture medium in one-well Nunc™ Lab-Tek™ II glass chamber slides. Before being exposed to the test and reference materials, AM were pre-cultured for 24 h in DMEM with high glucose (4.5 g/l), GlutaMax™, and sodium pyruvate (110 mg/l), supplemented with 10% FCS and 5 ml of a ready-to-use solution of 10,000 µg/ml streptomycin sulfate and 10,000 U penicillin G, sodium salt per 500 ml cell culture medium and at 37 °C and 5% CO2 in a humidified atmosphere using an incubator.

Normal rat mesothelial cells

Normal rat mesothelial (NRM2) cells, as target cells for asbestos-mediated mesothelioma development, served as second cell model. NRM2 cells were a gift of Jeffrey Everitt, MDV, Animal Pathology Core, Duke University School of Medicine (Durham, NC, USA) through James C. Bonner, Department of Biological Sciences, North Carolina State University (Raleigh, NC, USA). For characterization of this normal rat mesothelial cell line see Rutten et al. (1995). Cells were cultured in a 1:1 mixture of ATCC F-12 K and normal DMEM medium, supplemented with 10% FCS-standard and 0.01% gentamicin. Cells were passaged twice a week. For experiments, 5 × 104 cells were plated in 24-well plates and pre-cultured for 24 h, before treatment with the particulate materials. To estimate material uptake by fluorescence-coupled darkfield microscopy, NRM2 cells were plated at a density of 5 × 105 cells in 2 ml cell culture medium in one-well ClipMax chamber slides, and for counting of binucleated cells or mitotic phases, 2.5 × 105, NRM2 cells were plated in one-well ClipMax chamber slides in 3 ml of cell culture medium and were again pre-cultured for 24 h before treatment start.

Sterility testing and treatment of cells

RIF56008, ground RIF56008, and amosite asbestos were initially tested for sterility by adding a defined amount of the fiber/particle suspensions to thioglycolate broth and incubating two independent samples per material at 34/35 °C for 14 days. Saline (0.9%) served as negative and Bacillus subtilis (DSM 10, DSMZ-German Collection for Microorganisms and Cell Cultures, Brunswig, Germany) as positive control. After 14 days, turbidity was checked by the naked eye. Additionally, endotoxin was measured by a commercial service laboratory (Lonza Verviers SPRL, Verviers, Belgium), as endotoxin might lead to unspecific pro-inflammatory effects, and might disturb Enzyme-linked Immunosorbent Assay (ELISA) measurement of the cytokine-induced neutrophil chemoattractant 1 (CINC-1). Sterility and endotoxin testing both did not point to relevant contaminations with bacteria or fungi. Endotoxin was not detected, even at the lowest dilution (1:10). All values were below the detection limit of 0.05 EU/ml.

For cell treatment, the different materials were accurately weighed, sterilized for 4 h at 160 °C in a drying oven and then dispersed in the respective cell culture medium to generate concentrated stock dispersions. Stock dispersions were subsequently homogenized by ultrasonication for 1 min (RIF56008 and ground RIF56008) using a Sonorex Super RK 514 BH ultrasonic water bath or for two times 5 min using a VS 70 T sonotrode on a Sonoplus HD 2070 ultrasonic homogenizer (both Bandelin, Berlin, Germany) at 90% duty cycle and 100% amplitude (amosite asbestos). After sonication the resulting stock dispersions were finally diluted with cell culture medium to get two-fold (AM) or finally concentrated (NRM2) incubation dispersions. The experiments were performed with the carefully chosen concentrations of 0.5, 5, and 50 µg/cm2, corresponding to 1, 10, and 100 µg/ml incubation volume.

For AM, 500 µl of the twofold-concentrated incubation dispersions for both the alkaline comet and LDH release assays (performed in parallel), cell counting, and CINC-1 release were then carefully added to the respective cell-containing wells of 24-well plates, resulting in a total incubation volume of 1 ml/well. For fluorescence-coupled darkfield microscopy, 2 ml of the twofold-concentrated incubation dispersions were added to cell-containing one-well glass chamber slides. As negative/vehicle control, 500 µl (24-well plate) or 2 ml (glass chamber slides) of cell culture medium were added to the respective culture vessels. For NRM2 cells, a complete medium exchange was performed and 1 ml (24-well plates) or 4 ml (one-well chamber slides) of the finally concentrated material dispersions or vehicles were added. Both cell types were then exposed to the different fibers/particles or the vehicle controls for 4, 24 or 48 h (depending on the endpoint) at 37 °C and 5% CO2 in a humidified culture atmosphere using an incubator. The methodological positive control cultures received 500 µl (AM) or 1 ml (NRM2) of cell culture medium during the incubation period, and the methodological positive controls ethyl methanesulfonate (EMS) and Triton™ X-100 were added to the respective wells for 1 h or 5–15 min, respectively, before the end of cell incubation.

At the end of cell treatment, samples of the culture medium were carefully taken for analysis of LDH activity and CINC-1 release in the respective experiments. For determination of DNA strand breaks using the comet assay, and for automatic cell counting, AM were placed on ice for 10 min to enable cell detachment without usage of enzymes like trypsin, to avoid unspecific membrane damage or cell activation. For cell detachment, NRM2 cells were trypsinized and then subjected to the comet assay procedure or automatic cell counting. Determination of cell number, as one endpoint for fiber screening, was done for both cell types using an automatic CASY cell counting device (OLS, OMNI Life Sciences, Bremen, Germany), CASYton isotonic measuring buffer, and CASYcups as measuring vessels, and setting cell-type specific size borders.

Estimation of cellular uptake

As an aid in interpretation of cyto- and genotoxicity data, uptake of the different particulate materials was estimated using fluorescence-coupled darkfield microscopy. Therefore, AM and NRM2 cells grown and incubated for 24 h in glass chamber slides/one-well ClipMax chamber slides, were washed, subsequently fixed with cold methanol/acetic acid solution (3:1), air dried, and finally stained with 4′,6-diamidino-2-phenylindole (DAPI). Slides were then mounted using Vectashield® H-1000 and particle/fiber internalization was visualized and documented using an enhanced dark field illumination system in fluorescence mode (CytoViva®, Auburn, AL, USA), attached to a standard light microscope. Additionally, light microscopy served as a screening tool for evaluation of both cell density, cellular uptake, and cell morphology as well as for estimation of density and homogeneity of the fiber/particle dispersions. Light-microscopic pictures were taken using a camera-equipped Nikon ECLIPSE TS 100 infinity-corrected inverse microscope.

Lactate dehydrogenase (LDH) release assay

LDH release, indicative for membrane damage, was chosen as cytotoxicity endpoint, as it had previously been used to compare different MMVF in vitro in rat alveolar macrophages (e.g., Luoto et al. 1994), and had also been shown to respond to fiber treatment in vivo, as measured in cell-free bronchoalveolar lavage fluid (e.g., Adamis et al. 2001). To measure LDH release, culture supernatants were sampled at the end of cell treatment, centrifuged at 425 × g for 10 min to clean the supernatants from residual fibers/particles and stored in 1.5 ml reaction tubes. LDH activity was subsequently measured by transferring 100 µl of supernatant per well into a 96-well plate and adding 100 µl of reaction mixture of the “Cytotoxicity Detection Kit”. After incubation for about 15 min at room temperature in the dark, photometric measurement at 490 and 630 nm was performed, using a microplate reader. Percent cytotoxicity was finally calculated using the delta optical density (OD) of the two wavelengths, subtracting the blank value, and setting the negative/vehicle control values to 1 (main tests) or the result of the Triton™ X-100-treated cells to 100% (mechanical influence on membrane damage) to finally calculate relative cytotoxicity.

In vitro alkaline comet assay

To look for DNA strand break induction, cells were subjected to the in vitro alkaline comet assay according to Singh et al. (1988) after 24 h of pre-culture followed by 24 h of incubation. The comet assay represents an indicator test for detection of genotoxicity, which was previously used with AM to estimate the genotoxic potential of alkaline earth silica wools (Ziemann et al. 2014).

In the in vitro alkaline comet assay, all steps after the end of fiber/particle treatment were done under red light to avoid unspecific DNA damage due to UV‐irradiation. Detached cells were transferred to 1.5 ml reaction cups and pelleted by centrifugation for 5 min (900 rpm; Heraeus Biofuge 15, Thermo Scientific, Germany). Cells were subsequently re‐suspended in 80 μL of 0.75% (w/v) pre‐conditioned LMA and applied to purpose-made slides with one roughened surface, which had been pre‐coated with 0.5% (w/v) of NMA. The gels were then overlaid with a cover slip, allowed to set at 4 °C, before adding an additional layer of 100 μL of 0.75% LMA, which was again covered. Slides were then transferred to 4 °C. After removal of the cover slips, the slides were incubated in lysis solution (2.5 M NaCl, 100 mM Na2EDTA, 10 mM Tris‐HCl, 8 g/L NaOH, 1% Triton™ X-100, 10% dimethyl sulfoxide) overnight at 4 °C. After cell lysis, slides were placed into a pre‐cooled horizontal electrophoresis tank (Agagel Maxi, Biometra, Germany) and covered with pre‐cooled electrophoresis buffer (300 mM NaOH, 1 mM Na2EDTA, pH > 13). DNA was allowed to unwind for 20 min to generate DNA-single strand breaks, before electrophoresis was performed at fixed 0.7 V/cm and 300 mA for 20 min. Finally, slides were removed, neutralized by three changes of neutralizing buffer (0.4 M Tris‐HCl pH 7.4), and stained with 80 μL of a 20 μg/mL ethidium bromide solution.

Coded slides were subsequently analyzed microscopically for induction of DNA damage using a camera-equipped Axioskop fluorescence microscope (Carl Zeiss, Göttingen, Germany) with a 40x/0.9 mm Korr Plan-Neofluar Ph3 objective and the Comet Assay III software (Perceptive Instruments, Bury St Edmunds, UK). As a measure for DNA damage, DNA migration out of the cell nucleus was analyzed with the amount of DNA in the comet tail, i.e., the tail intensity (TI) as main and most accepted measure for DNA migration, as given in OECD 489 (OECD 2016). Slides were analyzed under the following criteria i.e., acceptable staining, evaluation of at least 100 nuclei per slide, evaluation of nuclei in the middle of the slide only, avoiding regions with bubbles, and no analysis of overlapping nuclei/comets. Comets without head, also called “hedgehogs” were excluded from analysis. Finally, the median of the single cell data per slide, the mean TI of three replicates per experiment and the means ± SD of the mean TI of three independent experiments were calculated.

CINC-1 release

To look for release of CINC-1, as a pro-inflammatory chemokine marker, the incubation supernatants of AM and NRM2 cells were frozen and stored at − 80 °C until measurement. CINC-1 was quantified (undiluted for AM, diluted 1:10 for NRM2) using a CINC-specific ELISA kit in 96-well plates (i.e., Rat CXCL1/CINC-1 DuoSet ELISA). Measurements were performed according to the manufacturer’s protocol. OD of each well was finally measured at 450 nm. Wavelength correction was performed automatically using a wavelength of 570 nm, which was subtracted from the 450 nm readings to correct for optical imperfections in the plate.

Counting of binucleated NRM2 cells

As certain asbestos and glass and stone wool fibers were previously shown to induce bi- and multinucleated cells in human mesothelial cells (Pelin et al. 1995), analysis of binucleated NRM2 cells was included as fiber morphology-dependent endpoint, most likely based on disturbance of cell division by physical effects on both chromosomes and cytoskeleton. For counting of bi-/multinucleated cells, NRM2 cells, pre-cultured in one-well ClipMax chamber slides, were exposed to RIF56008, ground RIF56008 or amosite asbestos for 48 h to enable sufficient cell division. Cultures were then washed, subsequently fixed with cold methanol/acetic acid solution (3:1), air dried, and stained using a standard Giemsa staining protocol. Two thousand NRM2 cells were then analyzed light microscopically for occurrence of bi-/multinucleated cells using a Leica DM4000 B automated upright light microscopy system equipped with a Leica N PLAN L 100x/0.75 objective and a DFC295 camera. As a measure for cytotoxicity/cell proliferation, automatic cell counting was performed using parallel cultures.

Statistical analyses

For LDH release arithmetic means of at least three independent experiments with three biological replicates each and for CINC-1 release, arithmetic means of up to five biological replicates, measured in duplicate, were subjected to statistical analysis using the Student’s t‐test for unpaired values, two-tailed, combined with normality (Shapiro–Wilk) and equal variance testing (Brown–Forsythe). For the in vitro alkaline comet assay mean TI values of the three independent experiments (three biological replicates each) were statistically analyzed. Samples were calculated from the median TI values of the three biological replicates derived from at least 100 nuclei per replicate. Due to the hierarchical nature of in vitro comet assay data and concentration-dependent change of single cell data variance (Møller and Loft 2014), equal variance was not assumed. Therefore, comet assay data were statistically analyzed using the Welch’s t-test, one-tailed, combined with normality (Shapiro–Wilk) and equal variance testing (Brown-Forsythe) for pairwise comparison to the negative control. Based on the right skewed data distribution, log transformation of comet assay data might be an option for statistical evaluation but can render testing hypersensitive in combination with the Welch’s t-test. The appropriate statistical method for statistical evaluation of in vitro comet assay data is still under debate. This also includes pairwise versus multiple testing approaches. Differences from the negative control were considered statistically significant at p ≤ 0.05. “Pearson Product Moment Correlation” was used for detection of correlations between material characteristics and biological effects. All statistical tests were done using SigmaPlot 14.0 (Systat Software GmbH, Germany).