The particles used in this study are described in Table 1. The surface-mean particle size calculated from nitrogen adsorption area measurement was close to the nominal particle size furnished by the vendor except for the supermicron particles of NiO. The adsorbtion surface area was much greater than the geometrical surface for the nominal diameter indicating that the NiO particles either are aggregates or have internal porosity. The natural dusts, prepared by resuspension and aerodynamic separation with an upper size cutpoint of <3 μm, also have high surface area indicating that the nominal PM_2.5 dust had a broad size distribution and contained significant amounts of smaller particles.

The elemental compositions of the natural soil-derived dusts as measured by x-ray fluorescence are dominated by the crustal elements, Al, Si, Ca, Mg, and Ti. The endotoxin content, organic carbon content (functionally defined as the carbon that is volatile in He atmosphere at temperature steps < 550°C) and the elemental carbon content (carbon that is removed in 2% O₂/98% He at temperatures from 550 to 800°C) are listed in Table 2.

Cell viability was evaluated to insure that low levels of cytokine secretion were not caused by cell death during the treatment period. Figure 1 shows the assay results for the highest treatment concentration (53 μg/cm²) of particles and for the concentrations of positive controls listed in Table 1. Only the VOSO₄ positive control caused more than 20% loss of cell viability at the 24-h time point, as measured by the mitochondrial reduction of WST-8.

The release of IL-6 by the immortalized cell line BEAS-2B in response to treatment with pairs of nano- and micron-sized particles is shown in Figure 2. Experiments were conducted in both KGM (2a and 2c) and LHC-9 media (2b and 2d). Graphs in 2a and 2b show typical results obtained from single experiments. Presented in this way the data suggest that the metal oxide particles induce small increases in IL-6 secretion, but that the nano-sized particles are not consistently more potent than the larger particles in each pair. Graphs 2c and 2d present the data in 2a and 2b merged with a followup experiment and also present the results from positive controls run in both experiments. The agreement between the two experiments is good, but the larger data set suggests that the IL-6 response to all the pure oxide manufactured particles is small compared to the positive controls. The pure oxides were also compared to kaolin (pottery clay), and Min-U-Sil (commercial mechanically ground crystalline silica), and neither of these particle was highly potent for inducing IL-6 secretion. The experiments in Figure 2 were conducted 5.3 and 53 μg/cm² particle concentration in media with 0.1% BSA added. Data shown are for the high particle treatment concentration and for the positive control concentrations in Table 1. Similar small responses to pure oxide particles compared to positive controls were seen at the low dose and in experiments with as-formulated media.

To further characterize the response of BEAS-2B cells to the manufactured oxide particles we conducted a separate set of experiments where each cell culture plate contained two of the nano-sized particle types, soil-derived particles from three different sites, and positive controls. Figure 3 shows the merged results of five independent experiments conducted in KGM and LHC-9 media with particles at 53 μg/cm2 and with the positive controls at concentrations listed in Table 1. The manufactured metal and ceramic oxide particles were less potent for the induction of IL-6 release than an equal mass concentration of three different soil-derived dusts. The LPS, TNF-a, and soluble vanadium are positive controls that show that the BEAS-2B cells are capable of producing IL-6 under the experimental conditions. The DD, JE, and MC samples are PM_2.5-enriched dusts prepared from surface soil samples, and are the same materials used in our previous studies of cytokine response in epithelial cells exposed to atmospheric fugitive dust particles 2930. These data strengthen the conclusion that the IL-6 secretion response of BEAS-2B cells to the pure oxides is small compared to other previously studied and environmentally relevant particle types.

Nano-sized SiO₂ particles caused a statistically significant increase in IL-6 compared to both the untreated control and the cells treated with micron-sized SiO₂ particles in six consecutive experiments with BEAS-2B cells indicating a reproducible pattern. For the other nano- and micron-sized particle pairs there was no conclusive evidence for the nanoparticles being consistently more potent than an equal mass concentration of the paired micron-sized particles.

The relative responses of BEAS-2B cells in the two types of cell culture media are characteristic of what we observed in a separate study comparing different in vitro lung cell models for particle toxicology [manuscript in preparation]. BEAS-2B cells in KGM media are highly responsive to vanadium and other soluble metals, as has been reported by others 31323334, but show relatively low response to LPS. In LHC-9 media the relative response to LPS and soluble metals is reversed.

The focus of this study was on IL-6 secretion by BEAS-2B cells, but we also measured IL-8 secretion and tested other cell types. Figure 4 shows typical IL-8 secretion data from a typical experiment with treatment at the high particle concentation. Again, any responses to the pure metal oxides are small and near control levels, and similar small responses were also seen at the low treatment concentration. Figure 5 shows typical IL-6 secretion in normal human bronchial epithelial cells (NHBE, Clonetics). The NHBE cells produce much higher control levels of IL-6 than the BEAS-2B cells, but the increased secretion in response to the pure oxide particles is small for both the 53 and 5.3 μg/cm² particle treatmetn concentrations.

During initial experiments we observed apparent artifacts in the measured cytokine concentrations and hypothesized that surface adsorption on some particles reduced the ELISA measurements. Experiments were conducted in which a known amount of recombinant IL-6 was added to suspensions of particles in cell culture media and in media with supplemental protein. The data were consistent with non-specific binding of IL-6 to surfaces. Figure 6a and 6b show the effect of increasing exogenous protein supplementation using bovine serum albumin (BSA) or newborn calf serum on the measured IL-6 concentration in cell-free suspensions of kaolin, nano-sized SiO₂, and TiO₂. Without particles the measured IL-6 in cell culture media was close to the standard in serum-based assay diluent. The presence of 200 μg/mL of particles reduced the measured IL-6 concentration compared to the IL-6 standard in assay diluent, denoted by * in the graph. The measured IL-6 concentration increased with increasing amounts of supplemental protein. Addition of greater than 0.1% BSA and 1% FCS to the media resulted in a statistically significant increase in measured IL-6 compared to the same particle suspension without supplemental protein, denoted by # in the graph. Both the SiO₂ and TiO₂nanoparticles resulted in a statistically significant reduction in measured IL-6 compared to the standard for all mixtures including the highest tested protein supplementation: 3% BSA and 10% FCS. The highest level of protein supplementation eliminated the measurement artifact for 200 ug/mL of kaolin. The results presented were obtained with KGM media, but similar results were seen with LHC-9 media (data not shown).

Figure 6c shows the effect of adding logarithmically increasing concentrations of kaolin and nano-sized Fe₂O₃, SiO₂, and TiO₂ particles to cell culture media without supplemental protein. Particle concentrations of 3.1, 10, and 31 μg/mL did not cause a statistically significant decrease in measured IL-6. The p-values for 100 μg/mL of kaolin and nano-SiO₂ were only slightly greater than 0.05, and the 316 μg/mL concentration of all particles resulted in a statistically significant decrease. The 100 μg/mL particle concentration corresponds to the 53 μg/cm² particle concentration which was used as the high treatment level in the cell response experiments (400 μL suspension in a 0.75 cm² well). Protein concentrations, measured by the Bradford assay and compared to a BSA standard, are 24 and 39 μg/mL for as-formulated KGM and LHC-9 media respectively. Media used for cell culture contains slightly higher protein due to cell secretions (2 – 4 μg/mL increase at 24 h). The use of 0.1% BSA in the cell culture media therefore corresponds to about a 50-fold increase in soluble protein available to block non-specific binding.

The wetted plastic area in the cell culture well is about 2.4 cm². In contrast, at 100 μg/mL of particles and 400 μL of suspension per cell culture well the nitrogen adsorption surface area of the particles ranges from 19 cm² for kaolin to 210 cm² for nano-Al₂O₃, which is 1–2 orders of magnitude more than the wetted plastic surface area. Even if nitrogen adsorption area is a poor surrogate for the surface area available for aqueous protein adsorption, the difference in calculated area of the plastic and the particles supports the hypothesis that IL-6 adsorption on particle surfaces was the mechanism responsible for the measurement artifact.

Additional experiments were conducted with nano- and micron-sized SiO₂ at higher concentration of BSA than used for the Figure 2 data. The data, Figure 7, show that the response of BEAS-2B cells to the SiO₂ particles remains small, and this supports the conclusion that, with this in vitro model, pure oxide nanoparticles are not highly potent compared to either the soil-derived dusts or the positive controls. Figure 7 suggests that the measured rank order of response to the nano- and micron-sized particles may switch as the exogeneous protein is increased, but the difference between particle sizes did not reach statistical significance for the 3% BSA concentration. The control levels of IL-6 also showed a statistically significant increase with increased BSA: 3, 9, 15 pg/mL for zero, 0.3 and 3% BSA respectively. Also, 3% BSA appears to reduce the growth rate of the BEAS-2B cells relative to the as-formulated media.

We compared the relative potency of different commercial LPS preparations (Sigma) to verify that the low response to LPS was not due to the specific LPS type selected. Figure 8 shows the response of BEAS-2B cells in KGM (8a) and LHC-9 (8b) media to treatment with 100 and 1000 endotoxin units/mL (EU/mL) of commercial LPS from Escherichia coli 055:B5, Klebsiella pneumoniae, Pseudomonas aeruginosa serotype10. The E. coli and P. aeruginosa LPS had similar potency. The data also show that the response of BEAS-2B cells to all three tested types of LPS was higher when the cells were cultured in LHC-9 media compared to KGM.

The potency to induce IL-6 in BEAS-2B cells does not correlate with the extracellular generation of reactive oxygen species (ROS). Typical data for the generation of ROS by particles in cell-free culture media are shown in Figure 9. These data are measurements of the relative fluorescence 5 minutes after adding 300 μg/mL of particles to KGM media, and are typical of readings taken from 1 to 10 minutes after adding the probe. The nano-sized NiO produced the highest level of fluorescence from DCFH reagent oxidation. For CeO, NiO, SiO₂ and TiO₂ the nanoparticles produced statistically higher fluorescence than the micron-sized particles. Figure 9b compares the nanoparticles of NiO and SiO₂ to the three soil derived dusts.

The nano-sized and micron-sized particles and the positive controls were purchased from commercial suppliers indicated in Table 1. The three soil-derived dusts DD, JE, and MNC were derived from field samples collected on an unpaved desert road in Utah, on an urban street in Ciadud Juàrez, Mexico, and from native soil at a remote desert site in Utah respectively. The DD, JE, and MNC particles are identical to samples 18, 16, 28 respectively in 30 which deals with the correlation between chemical composition of soil-derived dusts and IL-6 induction in BEAS-2B cells. The samples DD and MNC also correspond to UTDG and UTMC in 66 which contains detailed composition data and describes the procedures used for chemical analysis. The field samples were resuspended in the laboratory, and the particles were aerodynamically separated to provide a PM_2.5-enriched material for the cell treatments as previously described 3067.

Particle surface area per mass was measured by nitrogen adsorption (BET single point method) using a Quantachrome Monosorb analyzer. Surface determinations were run in duplicate for both the adsorption and desorption periods. Surface mean diameter was calculated assuming spherical particles and the mineral density from handbooks.

Four different cell culture models were used: the immortalized cell line BEAS-2B in LHC-9 media, BEAS-2B cells in KGM media, A549 cells in DMEM/F12, and normal human bronchial epithelial cells (NHBE) in BEGM media. Table 3 summarizes the details of the protocols used for the different cell culture models. The generic cell culture protocol consisted of growing the cells in an incubator at 37°C/6% CO₂ in 75 or 150 cm² flasks, replacing media every 2–3 days, and passaging before confluence by dislodging with trypsin, washing, and seeding new flasks or treatment wells. The use of LHC-9 media for BEAS-2B cells was based on the original work of Lechner et al. 68, particle studies by other laboratories, for example 69, and our prior studies. 2930 The protocol for BEAS-2B cells in KGM was based on the methods originally developed by the US EPA and since used in multiple studies 31367071. The protocol for the NHBE cells was based on the supplier's recommendation. Particle experiments consisted of seeding 48-well plates (Costar, Fisher Scientific) with cells, allowing a recovery period for the cells to attach and proliferate, replacing the culture media with fresh media containing the treatments, and harvesting the media for cytokine analysis 24 hr after treatment. Culture media were used both as formulated and with 0.1% added bovine serum albumin.

Particles were weighed, mixed with cell culture media, and resuspended by sonication and vortexing immediately before adding to the cell wells. The treatment experimental design consisted of multiple treatments applied to cells from a single passage to minimize confounding of comparisons by passage-to-passage variation of the cultured cells. Each multiwell cell culture plate included positive and negative controls. Results were replicated using additional independent passages of cells. Particle treatment concentrations were 0.53, 5.3 and 53 μg/cm² which correspond to 1, 10 and 100 μg/mL respectively for the cell culture plates and media volumes used.

Viable cell count was measured using the Cell Counting Kit -8 assay from Dojindo Laboratories. This assay assumes that the relative number of cells is linear with the metabolic activity indicated by mitocondrial reduction of WST-8 (2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, monosodium salt) to produce a water-soluble product. Data are quantified by absorption at 450 nm, corrected by absorption at 650 nm, using a Molecular Devices plate reader.

The cytokines IL-6 and IL-8 were measured using an enzyme-linked immunosorbent assay (ELISA). Antibodies, cytokine standard, and avidin horseradish peroxidase (AVHRP) for IL-6 were obtained from eBioscience; the IL-8 assay used a Human CXCL8/IL-8 DuoSet kit (DY208) from R&D Systems. Nunc MaxiSorp immuno plates (Fisher Scientific) were coated overnight with antibody; room-temperature incubations on a plate rocker were carried out with sample and standards, biotin-conjugated antibody, and AVHRP with thorough washing with 0.05% Tween-20 in PBS pH 7.0 between steps. After a 1-hour incubation with ABTS substrate, the plates were read at 405 nm using a Molecular Devices plate reader with SoftMax Pro software, and cytokine concentrations were calculated against a standard curve prepared in duplicate.

The experiments to quantify particle artifacts in the cytokine analysis used the addition of known amounts of recombinant human IL-6 from R&D Systems to various aqueous phases including water, phosphate buffered saline, LHC-9 or KGM cell culture media, cell culture media supplemented with serum or BSA, and media conditioned by growing cells for 24-h. Mixtures of IL-6 in the aqueous phase were incubated alone or with particles, centrifuged to separate the particles, decanted, and stored frozen until analyzed by ELISA. Cytokine quantification was by both direct comparison of measured concentrations in particle-free versus particle-treated samples and by a standard curve using recombinant IL-6 in serum-based assay diluent and prepared in duplicate at the time of analysis. Most cell-free experiments used concentrations similar to typical cell culture experiments, for example, 100 μg/mL particles, 200 pg/mL IL-6. However, the hypothesis-testing experimental conditions were varied over a wide range using logarithmically spaced nominal values: particles from zero to 400 μg/mL, IL-6 from zero to 10,000 pg/mL, incubation time from 5 minutes to overnight.

Endotoxin was measured using the chromogenic Limulus Amebocyte Lysate (LAL) assay kit QCL-1000 (Cambrex Biosciences).

The generation of reactive oxygen species in cell-free media was determined using the dichlorodihydrofluorescein diacetate (DCFH-DA) reagent (Molecular Probes #D-399). The method was based on studies of extracellular ROS generation by silica 57. For cell-free experiments the acetate was cleaved to form the non-fluorescent precursor by treatment with 0.1 M NaOH followed by neutralization and dilution to working concentration. Dichlorodihydrofluorescein (DCFH) is converted to the fluorescent product dichlorofluorescein by oxidation. Particles were added to 10 μM DCFH at 10–1000 μg/mL final concentration, incubated, and read at fixed intervals starting immediately after DCFH addition. Data in Figure 9 were from the third reading, nominally starting at 5 minutes. Fluorescence was read on a Perkin Elmer Victor 3 V Multilabel counter using 485 nm excitation and 535 nm emission. Freshly diluted hydrogen peroxide was used as a positive control.

Data were analyzed with JMP software (SAS Institute). Paired comparisons were made using Student's t-test, comparison of multiple treatments to a common control used one-way ANOVA with Dunnett's test, and p < 0.05 was considered significant.