Iron Oxide Nanoparticle-Induced Neoplastic-Like Cell Transformation in Vitro Is Reduced with a Protective Amorphous Silica Coating

Abstract

Iron oxide nanoparticles (IONP) have recently surged in production and use in a wide variety of biomedical and environmental applications. However, their potential long-term health effects, including carcinogenesis, are unknown. Limited research suggests IONP can induce genotoxicity and neoplastic transformation associated with particle dissolution and release of free iron ions. “Safe by design” strategies involve the modification of particle physicochemical properties to affect subsequent adverse outcomes, such as an amorphous silica coating to reduce IONP dissolution and direct interaction with cells. We hypothesized that long-term exposure to a specific IONP (nFe₂O₃) would induce neoplastic-like cell transformation, which could be prevented with an amorphous silica coating (SiO₂-nFe₂O₃). To test this hypothesis, human bronchial epithelial cells (Beas-2B) were continuously exposed to a 0.6 μg/cm² administered a dose of nFe₂O³ (~0.58 μg/cm delivered dose), SiO₂-nFe₂O₃ (~0.55 μg/cm² delivered dose), or gas metal arc mild steel welding fumes (GMA-MS, ~0.58 μg/cm² delivered dose) for 6.5 months. GMA-MS are composed of roughly 80% iron/iron oxide and were recently classified as a total human carcinogen. Our results showed that low-dose/long-term in vitro exposure to nFe₂O₃ induced a time-dependent neoplastic-like cell transformation, as indicated by increased cell proliferation and attachment-independent colony formation, which closely matched that induced by GMA-MS. This transformation was associated with decreases in intracellular iron, minimal changes in reactive oxygen species (ROS) production, and the induction of double-stranded DNA damage. An amorphous silica-coated but otherwise identical particle (SiO₂-nFe₂O₃) did not induce this neoplastic-like phenotype or changes in the parameters mentioned above. Overall, the presented data suggest the carcinogenic potential of long-term nFe₂O₃ exposure and the utility of an amorphous silica coating in a “safe by design” hazard reduction strategy, within the context of a physiologically relevant exposure scenario (low-dose/long-term), with model validation using GMA-MS.

Graphical Abstract

INTRODUCTION

Nanosized materials, which are specifically engineered to be 100 nm or smaller in at least one dimension, have emerged as novel solutions to critical issues in areas such as medicine, biotechnology, and transportation. Of these, iron oxide nanoparticles (IONP) have been investigated as unique components of MRI imaging techniques,¹ targeted drug delivery systems,² environmental catalysts,³ and incorporation into thermoplastics,^4,5 food products,⁶ and toners for printing equipment.⁷ IONP, as well as many other types of nanometal oxides, have also grown in interest, production, and utilization for consumer products, including agriculture and food applications.^8–10 This surge may translate to an increased risk of exposure for those involved in particle manufacture and use across its life cycle. However, potentially adverse health outcomes following extended periods of exposure to IONP remain unclear.

One of the most common routes for human exposure is the inhalation of particulates in the circulating air, which usually occurs at low doses over long periods of time. This type of exposure can have distinct adverse outcomes for vulnerable populations. Chronic pulmonary exposure to metal-oxide particulates has been associated with cancer development in miners, welders, and other types of industrial workers.¹¹ However, potential cancer-related outcomes induced by IONP, as well as other types of nanometal oxides, remain poorly understood. For IONP, the current standard is to regulate exposure based on fine-sized (100–2500 nm¹²) iron oxide exposure limits, even though evidence suggests that nanosized particulates may pose a greater risk with exposure on a per mass basis than their fine-sized counterparts.^13,14

As a further complication, fine-sized iron oxide has an unclear and controversial toxicity profile. Some researchers suggest this particulate is essentially benign,^15–20 whereas others associate exposure with adverse pulmonary outcomes, including increased risk of lung cancer in iron ore miners.^21–23 Similarly, IONP-induced adverse health outcomes are also largely unclear. It is known there is some risk of occupational exposure to IONP,^24,25 with unknown or unreported adverse outcomes. In laboratory settings, some researchers report no effect following IONP exposure,^26–28 whereas others indicate that IONP or IONP-containing materials may induce genotoxicity,^29–32 neoplastic-like cell transformation,^33,34 in vivo tumor formation,^35,36 and neurotoxicity.³⁷

Gas metal arc welding fumes (GMA) have recently been reclassified as a Group 1 total human carcinogen.³⁸ GMA generated from mild steel welding (GMA-MS) contains roughly 80% of iron/iron oxide particulates that fall within the nanosize range. GMA-MS exposure is associated with fibrotic lesions and elevated lung iron levels in humans,³⁹ as well as lung tumor development^35,36 and lung cancer progression.⁴⁰ Although GMA-MS contains largely insoluble iron, it still provides a useful reference for potential IONP-induced adverse outcomes, particularly those related to the induction or promotion of lung cancer.

The aims of this study were to assess the potential for IONP to induce neoplastic-like cell transformation within the context of an occupationally relevant subchronic in vitro exposure model (low-dose/long-term) and to examine the potential protective qualities of an amorphous silica coating under these conditions. GMA-MS particles were used as a positive control, as they have been previously shown to promote carcinogenic outcomes in vivo.^35,36 Particle-induced changes in intracellular iron, ROS production, and DNA damage associated with the cell transformation were also evaluated throughout the long-term exposure.

EXPERIMENTAL METHODS

Particle Generation, Preparation, Characterization, and Dosimetry.

nFe₂O₃, SiO₂-nFe₂O₃, and SiO₂ were generated at Harvard University Center for Nanotechnology and Nanotoxicology (Boston, MA) using the VEGNES-FSP system.⁴¹ Briefly, this flame spray pyrolysis synthesis platform involves the dispersion of organometallic precursors (iron) and their combustion to form nFe₂O₃. Immediately following combustion, particles were sprayed with SiO₂ vapor to form a nanothin amorphous silica coating,^41,42 before collection on Teflon filters. Particles were then extracted as powders for subsequent experiments. Gas metal arc mild steel welding fumes (GMA-MS) were generated at the Lincoln Electric Co. (St. Louis, MO) and provided by Dr. James M. Antonini from CDC/NIOSH (Morgantown, WV). These fumes were generated in a fume chamber by a skilled welder, and particulates were collected on filters for use in this study.⁴³ Aeroxide TiO₂ (80:20 anatase/rutile structure, Evonik) was supplied from the Degussa Corporation (Parsippany, NJ) and provided by Dr. Tina Sager from CDC/NIOSH (Morgantown, WV).⁴⁴ TiO₂ particulates were sieved at a vibration amplitude of 50 and then filtered three times using Teflon filters (1.18 mm, 250 nm, and 45 nm). Only particles that were able to pass through all three filters were used for treatments. TiO₂ was used as a reference particle control, as the same particle is the source material for NIST Standard Reference Material for TiO₂.⁴⁵

Partial particle characterization in the pristine powder phase was previously determined as described for nFe₂O₃, SiO₂-nFe₂O₃, and SiO₂ (d_XRD, SSA, and iron/iron oxide content),⁴¹ TiO₂ (d_XRD),⁴⁴ and GMA-MS (iron/iron oxide content).⁴³ The d_XRD of GMA-MS was calculated based on representative SEM images. Particles were suspended in sterilized Milli-Q water at a concentration of 1 mg/mL in sterilized glass test tubes. Particle suspensions were sonicated with a cup horn sonicator (Sonics Vibra Cell, Newton, CT) based on sonicator efficiency and calculated critical energy needed to generate stable particle agglomerates (nFe₂O₃/SiO₂-nFe₂O₃, 242 J/mL; SiO₂, 161 J/mL; TiO₂, 161 J/mL; GMA-MS, 242 J/mL)^46–48 and were then serially diluted in BEGM for subsequent treatment. Fresh particle stock suspensions in water were prepared every 4 weeks and sonicated a maximum of eight consecutive times in water, with a hydrodynamic diameter (d_H) in medium determined separately for each sonication.

Hydrodynamic diameter (d_H) and ζ potential for all particles following suspension in Milli-Q water, sonication, and serial dilution in a complete cell culture medium were determined by dynamic light scattering (DLS, ZetaSizer Nano ZS, Model Zen3600, Malvern, Surrey, United Kingdom) using previously described methods.^47,48 The ζ potential was determined using the ZetaSizer Nano Series Dip Cell (Model Zen1002, Malvern, Surrkey, United Kingdom). For scanning electron microscopy (SEM) with energy-dispersive X-ray (EDX), particles in a cell culture medium were further diluted in filtered distilled water (1:10) and agitated manually to maintain dispersion. Following this, particles were vacuum filtered onto a polycarbonate filter, which was then mounted on an aluminum stub and sputter-coated with gold–palladium. SEM images were taken on a Hitachi S4800 field emission scanning electron microscope (Hitachi, Tokyo, Japan) at magnifications of 15 000, 25 000, or 50 000×. For transmission electron microscopy (TEM), one drop of the diluted particle sample was placed on a Formvar-coated 200 mesh copper grid and evaporated overnight before imaging on a JEOL 1400 transmission electron microscope (JEOL, Tokyo, Japan).

Particle release of free iron in H₂O, the culture medium (BEGM), and the phagolysosome-like buffer was assessed using inductively coupled plasma-optical emission spectroscopy (ICP-OES) at CDC/NIOSH (Cincinnati, OH; calculated limit of quantification: 0.0566 μg/mL). Particles were serially diluted in 1 mL of water, BEGM, or phagolysosome-like buffer (prepared as previously described⁴⁹) at the concentration of 20 μg/mL and then stored in a 37 °C incubator to replicate cell culture conditions. Particle-free vehicle controls were also prepared and stored. Then, 72 h later, particle suspensions were centrifuged (13 000 rpm for 10 min) and filtered (0.2 μm of PTFE, ThermoFisher), and the resulting supernatant was assessed for soluble iron content via ICP-OES. The percent of dissolved particle was calculated based on the beginning mass of particle and mass of detectable soluble iron.

Volume-weighted size distribution and effective density of particle suspensions in the medium (calculated as previously described⁵⁰) were incorporated into a distorted grid (DG) fate and transport modeling algorithm in MatLab v.2006b developed at Harvard University^48,51,52 to assess cell deposited dose metrics as a function of exposure time.

Cell Culture and Exposure.

Naïve Beas-2B cells (ATCC) were cultured in serum-free BEGM (basal medium with 1% penicillin/streptomycin, 0.4% bovine pituitary extract, 0.1% hydrocortisone, human epithelial growth factor, epinephrine, transferrin, insulin, retinoic acid, and triiodothyronine) from Lonza (Walkersville, MD) for a minimum of two passages before beginning exposures. Cells were passaged according to the ATCC guidelines. For acute exposures, Beas-2B were seeded at 10 000 cells per well in a 96-well plate (Corning). Then, 24 h postplating, the medium was aspirated out and replaced with specific particle treatments, or a particle-free vehicle control, which were prepared as described above, and serially diluted in BEGM to appropriate doses. Administered doses ranged from 0.2 μg/cm² (0.66 μg/mL) to 2.0 μg/cm² (6.6 μg/mL) for acute studies, and a subtoxic dose was selected for subchronic treatment based on these results and previous literature.⁵³

For subchronic exposures, Beas-2B (passage 3) were seeded at 50 000 cells/well in 6-well plates (Corning), with four biological replicates per treatment group, from which at least three biological replicates were used for any subsequent assay. At 24 h postseeding, the medium was aspirated out and replaced with specific particle treatments, prepared as described above. All particle treatments were administered at 0.6 μg/cm² (2.88 μg/mL). Particle-free vehicle-exposed cells served as passage controls. The medium with particle treatments was replaced every 3 to 4 days. When cells were approximately 80–90% confluent (roughly once a week), cells were washed, trypsin/EDTA was released from the plate, and a random selection of cells was replated at a ψ seeding density of 50 000 cells/well in 75% of the total volume. Particle treatments were added directly to these wells approximately 3 h later in the remaining 25% volume. The remainder of cells from each passage/replating were used for cell transformation and ROS-related assays or frozen back for future studies.

Particle Uptake.

Beas-2B (passage 3) were seeded at 50 000 cells/well in 6-well plates (Corning) and exposed to specific particle treatment 24 h later, as described above. The medium with particle treatment was replaced 3 days later. When cells were 80–90% confluent (1 week later), cells were washed, passaged, and prepared for transmission electron microscopy (TEM) as previously described.⁵⁴ Briefly, cells were suspended in Karnovsky’s fixative and pelleted; then, the fixative was removed and replaced with 4% agarose. After congealment, excess agar was trimmed away, and fresh Karnovsky’s fixative was added overnight. The pellet was rinsed (8% sucrose and 0.9% sodium chloride buffer), and then osmium teroxide was added. En bloc staining was done using 1% tannic acid and 0.5% uranyl acetate. The pellet was dehydrated using a graded series of ethanol, and then cells were embedded in epoxy resin and polymerized. Blocks were sectioned at 70 nm onto 200 mesh copper grids and stained (4% uranyl acetate and Reynold’s Lead Citrate). Imaging was done at 1500–3000× or 10 000–50 000× magnification on a JEOL 1400 transmission electron microscope (JEOL, Tokyo, Japan).

Reactive Oxygen Species Production.

For acute (18 and 24 h) and subchronic exposures (48, 62, 69, 76, 97, 104, 118, 125, 132, 145, and 195 days of continuous exposure), treated Beas-2B cells were seeded at 10 000 cells/well in a clear-bottom black 96-well plate (Corning), stained with 10 μM DCF-H and 1 μM Hoescht-33342 (Sigma-Aldrich, St. Louis, MO), and incubated at 37 °C for 30 min. Cells were washed once with BEGM, and imaged in warm PBS at 20× magnification, with 9 image sites per well. Imaging using DAPI and FITC filters were analyzed using high content screening on an ImageXpress Micro XLS apparatus with MetaXpress v5.3 software (Molecular Devices, San Jose, California). Using the MultiWavelength Application Module in MetaXpress, minimum/maximum cell diameters and minimum intensity thresholds were used to identify each nucleus (DAPI) and intracellular ROS (FITC). Masks were applied to identify cells for each imaged site positively. Only FITC masks containing a nucleus were included in the analysis. The integrated intensity of the fluorescent signal from each cell (evaluated using a cell mask) was calculated and averaged for quantification. Ten μM H₂O₂ treatment was used as a positive control for ROS detection, and particle-only controls were used to verify there was minimal particle interference (data not shown).

Cell Proliferation.

Acute and subchronic exposures were done at previously indicated time points. Treated Beas-2B cells were plated at 10 000 cells/well in a 96-well plate with at least four technical replicates per biological replicate and treatment group. At appropriate time points, cells were incubated with WST-1 reagent (Sigma-Aldrich, St. Louis, MO) at a 1:10 dilution for 2 h; then, absorbance at 450 nm was measured using a spectrophotometer (SpectraMax, Molecular Devices, San Jose, California), according to manufacturer’s instructions.

Attachment-Independent Colony Formation.

A soft agar colony assay was conducted as previously described³⁴ at 55, 83, 111, 138, 174, and 202 days of continuous exposure. Briefly, the 0.5% w/v agar medium was prepared and plated in 6-well plates (1.25% Difco agar, 15% v/v fetal bovine serum, and growth supplements at concentrations consistent with the complete cell culture medium in 2× MEM). After the base layer was solidified, cells were suspended in 33% agar at 1 × 10⁴ cells/well and slowly layered onto precast agar in triplicate 6-well plates. Plates were stored in a 37 °C incubator for 14 days, and colonies were digitally photographed under light microscopy (1.25×). Five representative pictures from each well were taken, totaling 15 pictures per biological replicate. Only colonies >50 μm in diameter were included in counts, which were done manually using ImageJ Grid and CellCounter plugins.

Intracellular Iron and Lysosome Counts.

At select time points during subchronic exposure (91, 132, and 202 days), treated cells were seeded at 10 000 cells per well in a 96-well clear-bottom black plate (Corning). When cells were at least 70% confluent, Beas-2B were fluorescently stained with 1 μM PhenGreen SK, 1 μM Deep Red LysoTracker, and 1 μM Hoechst 33342. Plates were incubated at 37 °C for 30 min. Cells were washed once and imaged in warm Fluorobrite clear DMEM (ThermoFisher) at 20× magnification for intracellular iron and then 40× magnification for lysosome counts, with 9 imaging sites per well for each. Imaging (DAPI, FITC, and Cy5 filters) and analysis were done using the ImageXpress apparatus and MetaXpress software, as described above. Using this program, individual intracellular lysosomes were identified using minimum/maximum diameter masks, counted using the TransFlour HT module, and averaged per cell for each image site and treatment group. The integrated intensity of the PhenGreen fluorescent signal from each cell was calculated using a cell mask and averaged for the quantification of intracellular iron. The PhenGreen fluorescent signal from acute intracellular iron studies was measured using the Varioskan Lux fluorescent plate reader (excitation, 492 nm; emission, 517 nm). The PhenGreen fluorescent signal is quenched upon binding to free iron ions. Therefore, the inverse of the integrated intensity of the fluorescent signal was calculated using the nontreated control cell fluorescence as the threshold (F₀/F), as has been previously described.⁵⁵ Iron specificity of PhenGreen was assessed using FeSO₄ as a positive control and pyridoxal isonicotinoyl hydrazine (PIH, Sigma-Aldrich) as an intracellular iron chelation control).

γH2AX Assay for DNA Damage.

Treated Beas-2B cells from subchronic exposure (91, 132, and 202 days) were seeded at 10 000 cells per well in a 96-well clear-bottom black plate (Corning). When cells were at least 70% confluent, the Beas-2B were fixed in 4% paraformaldehyde, permeabilized with 0.25% Triton-X, and then washed with PBS, all at ambient room temperature. Then, 48 h later, cells were incubated with 0.5% phosphorylated γH2AX primary antibody (1:250, Cell Signaling, Danvers, MA) for 1 h, washed in PBS, and fluorescently tagged using AlexaFlour-555 secondary antibody (1:500) and 1 μM Hoechst 33342. Imaging (at 20× magnification) and analysis were performed using the ImageXpress apparatus and MetaXpress software, as described above. Minimum/maximum cell diameters and minimum intensity thresholds were used to identify each nucleus (DAPI) as described above. The average nuclear intensity of the top 3% of nontreated cells was used as a threshold for positive or negative scoring of all cell treatments to determine the percentage of cells with positive γH2AX damage. H₂O₂ treatment (10 μM) was used as a positive control (data not shown).

Statistical Analyses.

Data represent the mean ± SEM from three or more biological replicates and four or more technical replicates per biological replicate. Samples were normalized to the nontreated control for each biological replicate and then combined for all figures. Following normalization, samples were analyzed using a two-way ANOVA with Tukey’s posthoc test for multiple comparisons to determine significance. Statistical significance is indicated in each figure as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. All statistical analyses were performed using GraphPad Prism Software (La Jolla, CA).

RESULTS

Particle Characterization.

First, particles were evaluated for the agglomerate hydrodynamic diameter and ζ potential following preparation and sonication in sterilized water and then serial dilution in complete cell culture media (BEGM), using DLS (Table 1). All test particles had roughly similar primary particle diameters (d_XRD ranging from 15 to 45 nm) and ζ potentials (ranging from –15.63 to –8.84 mV) in BEGM, but with diverse agglomerate hydrodynamic diameters, ranging from 1114 nm (GMA-MS) to 2570 nm (nFe₂O₃).

Table 1.

Colloidal/Physicochemical Characterization of Dispersed Nanoparticles after One Sonication^a

ENM	dXRD (nm)	SSA (m2/g)	ζ (mV) [measured pH]	iron/iron oxide content (%)	dH (Z avg, nm)	PDI	ρENM (g/cm3)	ρEV (g/cm3)
nFe2Os	19.6	42	−8.63 [7.55]	100	2570	0.43	5.242	1.534
SiO2-nFe2O3	21.3	49	−13 [7.54]	66	1895	0.28	5.242	1.273
S1O2	19	147	−15.63 [7.58]	0	1774	0.16	2.648	1.08
gas metal arc mild steel welding fumes	15–45	nd	−10.49 [7.53]	80	1114	0.28	5.7	2.362
TiO2	23	nd	−8.56 [7.59]	0	2085	0.17	4.26	1.391

^aPrimary particle size as determined by X-ray diffraction (d_XRD), surface area (SSA), hydrodynamic diameter (d_H), ζ potentials, iron/iron oxide content, polydispersity index (PDI), raw material density (ρ_ENM), and agglomerate effective density (ρ_EV) of all particle treatments in complete cell culture media (BEGM). nd: no data.

Dispersed particle agglomerates were visualized using TEM (Figure 1A) and SEM (Figure 1B). SEM-EDX (Figure 1C) indicated an amorphous silica coating on SiO₂-nFe₂O₃ and also showed that GMA-MS particulates were primarily composed of insoluble iron and small amounts of other metals.

Figure 1.

Representative SEM and TEM images of particle treatments. Particle agglomeration in a complete cell culture medium was visualized using (A) transmission electron microscopy (TEM), (B) scanning electron microscopy (SEM), and (C) SEM–energy-dispersive X-ray spectroscopy (SEM-EDX). For TEM, magnification ranged from 15 000 to 50 000×.

Acute Cytotoxicity Assessment.

Next, the acute bioeffect of particle treatments (nFe₂O₃, SiO₂-nFe₂O₃, GMA-MS, SiO₂, and TiO₂) was evaluated. Reactive oxygen species (ROS) production and proliferative capacity were used as basic indicators of particle-induced changes in cell activity (Figure 2). Beas-2B cells were exposed to a range of administered doses (0.2–2.0 μg/cm²) of particle treatments for up to 24 h, and ROS production and cell proliferation were assessed. For ROS production, treated Beas-2Bs showed a dose response for all particle treatments, with the greatest ROS response overall at 18 h postexposure (Figure 2A). The 2.0 μg/cm² administered dose of nFe₂O₃, GMA-MS, and TiO₂ all induced significantly elevated ROS production compared to nontreated controls at 18 h postexposure, with GMA-MS- and TiO_2- exposed cells retaining this elevated ROS at 24 h. There was no significant cell death induced by any particle treatments, and higher doses of SiO₂ and the lowest dose of TiO₂ induced significantly elevated proliferation at 24 h postexposure (Figure 2D). Intracellular iron was evaluated using the fluorophore, PhenGreen SK, and iron chelator, PIH. In Beas2B cells, the majority of iron is sequestered and not catalytically active. PIH, as a chelator, and PhenGreen, as a fluorescent marker, actively target chelatable iron within the cell’s catalytically active labile iron pool,⁵⁶ as is indicated with FeSO₄ treatment, which was used as a control. There were no significant changes in intracellular iron across particle treatments at 18 or 24 h postexposure (Figure 3). Cells exposed to higher doses of GMA-MS (0.6 or 2.0 μg/cm² administered dose) showed a trend of elevated intracellular iron at 24 h postexposure. Overall, the acute treatment data supported 0.6 μg/cm² as a subtoxic dose for subchronic treatment.

Figure 2.

Acute particle-induced changes in ROS production (DCF assay) at (A) 18 and (B) 24 h and WST metabolism at (C) 18 and (D) 24 h postexposure. Doses refer to the administered dose. Error bars indicate the mean ± SEM (n = 3). *p < 0.05, ****p < 0.0001 from the nontreated control (NT).

Figure 3.

Acute particle-induced changes in intracellular iron (PhenGreen fluorescence) at (A) 18 and (B) 24 h. Doses refer to the administered dose. Error bars indicate the mean ± SEM (n = 4). Cross hatches indicate the addition of 100 μM pyridoxal isonicotinoyl hydrazine (PIH, iron chelator). In Beas2B cells, the majority of iron is sequestered and not catalytically active. PIH, as a chelator, and PhenGreen, as a fluorescent marker, actively target chelatable iron within the cell’s catalytically active labile iron pool,⁵⁶ which is why it had minimal effect on acutely treated Beas2B, with the exception of FeSO4 treatment. F₀/F: the inverse integrated intensity of the fluorescent signal. *p < 0.05 from the same particle treatment without PIH chelation.

Sub Chronic Cell Transformation.

Based on previous in vitro and in vivo research, and results from the above acute experiments, an administered dose of 0.6 μg/cm² (2.88 μg/mL) was selected for the subchronic exposure. A recent study by Teeguarden et al.⁵³ exposed Balb/c mice to a similarly sized iron oxide (Fe₃O₄) particle via inhalation at 19.9 mg/m³ over 4 h (80 mg h/m³), which is slightly lower than the OSHA permissible exposure limit (PEL) for fine-sized Fe₂O₃ particles (14 mg/m³ over 8 h or 112 mg h/m³). Following the exposure, the target tissue dose in the bronchial region was calculated to be about 1 μg/cm². Previous research by our group³⁴ also indicated that, at a 0.6 μg/cm² administered dose, nFe₂O₃ induced a neoplastic-like cell transformation in primary human small airway epithelial cells. Therefore, in the present study, Beas-2B cells were exposed to this low dose (0.6 μg/cm²) of nFe₂O₃, SiO₂-nFe₂O₃, GMA-MS, or particle controls (TiO₂, SiO₂) continuously for 6.5 months and assessed throughout for indications of neoplastic-like cell transformation, particularly at time points surrounding two, four, and six months.

Uncontrolled or dysregulated cell proliferation is a classic hallmark of cancer.⁵⁷ Therefore, this was used as an initial indicator of potential cell transformation throughout the exposure period (Figure 4). Beginning at 97 days (~3.25 months) of continuous exposure, nFe₂O₃-exposed cells had a trend of decreased proliferation, which began to increase at 118 days, as compared to nontreated controls. By 125 days (~4.5 months), the nFe₂O₃-exposed cells had almost a 2-fold significant increase in proliferation as compared to nontreated control cells. Elevated proliferation was maintained through the rest of the exposure period (6.5 months). GMA-MS-exposed cells followed a similar trend for this parameter, with significantly elevated proliferation (~1.25-fold) by 145 days (~5 months) of exposure, which was maintained through the rest of the exposure period, as compared to nontreated controls. However, SiO₂-nFe₂O₃-exposed cells had no significant changes in cell proliferation throughout the entire exposure period. TiO₂-exposed cells began to exhibit significantly decreased proliferation at 97 days of continuous exposure, which gradually increased until it became significantly elevated at 145 days. This elevation was maintained through the rest of the exposure period. SiO₂-exposed cells had significantly elevated cell proliferation at 125 days of continuous exposure, which then returned to be roughly similar to that of nontreated control cells by the end of the exposure period.

Figure 4.

Cell proliferation throughout subchronic exposure to the 0.6 μg/cm² administered dose of specific particle treatment (~0.58 μg/cm² nFe₂O₃, ~0.54 μg/cm² SiO₂-nFe₂O₃, ~0.57 μg/cm² GMA-MS, ~0.24 μg/cm² SiO₂, or ~0.57 μg/cm² TiO₂ delivered dose). There were no significant changes in activity of nontreated control cells (Figure S1), so the fold change as compared to nontreated cells was calculated for each time point separately. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 as compared to nontreated control cells at the same time point. Error bars indicate the mean ± SEM (n = 4).

Attachment-independent colony formation is an indicator of neoplastic-like cell transformation because abnormal or neoplastic cells can continue to proliferate without attachment to a solid surface.⁵⁸ This was assessed using soft agar colony formation assay throughout the subchronic exposure period (Figure 5). At 55 days (~2 months) of continuous exposure, there were no significant changes in colony formation across treatment groups. nFe₂O₃- and GMA-MS-exposed cells had a significant increase (1.2- and 1.4-fold) in colony number at 83 days (~2.75 months) postexposure, which was ablated by 111 days (~3.7 months). Colony number then became significantly elevated again (1.6-fold) at 138 days (~4.5 months) as compared to nontreated control cells, which was maintained through the remainder of the exposure period (202 days, ~6.5 months). TiO₂-exposed cells had a significant 1.3-fold increase in colony formation at 111 and 138 days, which then returned to be roughly similar to that of nontreated control cells by 174 days exposure and remained at baseline levels through the end of the exposure period. SiO₂-nFe₂O₃- and SiO₂-exposed cells had no significant changes in attachment-independent colony formation throughout the exposure period.

Figure 5.

Attachment-independent colony formation throughout subchronic exposure to the 0.6 μg/cm² administered dose of specific particle treatment (~0.58 μg/cm² nFe₂O₃, ~0.54 μg/cm² SiO₂-nFe₂O₃, ~0.57 μg/cm² GMA-MS, ~0.24 μg/cm² SiO₂, or ~0.57 μg/cm² TiO₂ delivered dose). (A) Quantification of attachment-independent colonies followed by representative images of colonies from (B) 83 days, (C) 138 days, and (D) 202 days of continuous exposure. There were no significant changes in the activity of nontreated control cells (Figure S2), so the fold change as compared to nontreated cells was calculated for each time point separately. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 as compared to nontreated control cells at the same time point. Error bars indicate the mean ± SEM (n = 4).

Subchronic Exposure-Induced Changes in Cell Response.

Subchronically exposed cells were assessed for changes in intracellular iron, lysosome counts, ROS production, and DNA damage at key points during the subchronic exposure period. Intracellular iron and lysosome counts were evaluated using PhenGreen SK and LysoTracker fluorescent staining and high content image analysis (Figure 6). At all three measured time points (91, 132, and 202 days of continuous exposure), nFe₂O₃-exposed cells had significantly decreased intracellular iron as compared to nontreated control cells (~70%, 55%, and 90% of nontreated control cells at 91, 132, and 202 days of continuous exposure, respectively). GMA-MS-exposed cells had significantly decreased intracellular iron at 91 days of continuous exposure only (~70% that of nontreated control cells) and then returned to be similar to nontreated control cells at 132 days, which was maintained through the end of the exposure period. GMA-MS-exposed cells also had significantly fewer average lysosomes per cell than nontreated controls at 202 days of continuous exposure, but all other treatment groups showed no significant changes in this parameter at this or any other time points measured.

Figure 6.

Intracellular iron and lysosome response during the subchronic exposure. (A) Quantification of intracellular iron fluorescent signal, (B) representative images of 20× fluorescent staining, (C) average lysosome count per cell, and (D) representative images of 40× fluorescent staining. Images were taken at key time points during subchronic exposure to the 0.6 μg/cm² administered dose of specific particle treatments (~0.58 μg/cm² nFe₂O₃, ~0.54 μg/cm² SiO₂-nFe₂O₃, ~0.57 μg/cm²GMA-MS, ~0.24 μg/cm² SiO₂, or ~0.57 μg/cm² TiO₂ delivered dose). For images, blue represents nuclear staining, green represents intracellular iron staining, and red represents lysosome staining. **p < 0.01, ***p < 0.001, ****p <0.0001 as compared to nontreated control cells. Error bars: mean ± SEM (n = 3). F₀/F: the inverse integrated intensity of the fluorescent signal.

ROS production was also assessed throughout the subchronic exposure period (Figure 7). At the first measured time point for this parameter (48 days/1.5 months continuous exposure), all treatment groups had significantly decreased ROS as compared to nontreated controls. nFe₂O₃-exposed cells showed a trend of increased ROS production at 104, 125, and 145 days, which then decreased to become less than that of production by nontreated control cells by the end of the exposure period. Both GMA-MS- and SiO₂-nFe₂O₃-exposed cells had slight and sometimes significantly decreased ROS production throughout the entire subchronic exposure period compared to nontreated controls. SiO₂- and TiO₂-exposed cells had no significant changes in ROS production throughout the exposure period, with the exception of the first time point measured (48 days).

Figure 7.

ROS production throughout subchronic exposure to the 0.6 μg/cm² administered dose of specific particle treatments (~0.58 μg/cm² nFe₂O₃, ~0.54 μg/cm² SiO₂-nFe₂O₃, ~0.57 μg/cm² GMA-MS, ~0.24 μg/cm² SiO₂, or ~0.57 μg/cm² TiO₂ delivered dose). There were no significant changes in ROS production of nontreated control cells (Figure S3), so the fold change as compared to nontreated cells was calculated for each time point separately. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 as compared to nontreated control cells at the same time point. Error bars indicate the mean ± SEM (n = 4). Genomic instability and DNA damage are known to be key events associated with carcinogenesis.⁵⁹ There are several types of DNA damage that have been implicated in carcinogenesis adverse outcome pathways, such as double-stranded DNA damage, which is known to have a critical impact on cell phenotype.^60,61 To assess double-stranded DNA damage of the transformed cells, γH2AX fluorescence was evaluated at key time points during the subchronic exposure period (Figure 8) as an indicator of double-stranded DNA breaks.⁶² At 132 days (~4.5 months) of continuous exposure, there were significantly more nFe₂O₃-exposed cells that were γH2AX positive as compared to SiO₂-nFe₂O₃-treated cells. By 202 days of continuous exposure, nFe₂O₃-exposed cells had significantly elevated γH2AX as compared to both nontreated and SiO₂-nFe₂O₃-treated cells. GMA-MS cells also showed significant DNA damage at 202 days of continuous exposure as compared to nontreated control cells.

Particle Dosimetry.

A critical issue with in vitro nanoparticle exposures is the difficulty in ascertaining accurate dosimetry. Particle agglomeration and the settling rate will impact how the known administered dose may compare to actual delivered dose, while particle uptake may play a key role in the induction of adverse outcomes. First, previously calculated particle agglomeration (Table 1) and measured particle density were used to calculate the particle settling rate. Raw particle density is known to relate to effective density in a cell culture medium, and effective density is a key factor in the calculation of the in vitro particle deposition rate.⁵⁰

These values, along with the hydrodynamic diameter, were used to model particle deposition (Table S1). All particles except for SiO₂ had greater than 90% particle deposition (average fraction deposited) within the first 24 h following treatment, whereas SiO₂ had roughly 70% deposition (average fraction deposited) by 24 h following exposure (0.6 μg/cm² administered dose or ~0.42 μg/cm² delivered dose). During the subchronic exposure period (6.5 months), particle suspensions in water were freshly prepared every 4 weeks and were sonicated before each exposure (every 3–4 days), followed by dilution in BEGM to 0.6 μg/cm² administrated dose (Figure 8). With repeated sonications (≤8 times), all particles decreased in the average agglomerate hydrodynamic diameter, but deposition remained >86% (>0.52 μg/cm² delivered dose for subchronic exposures), with the exception of SiO₂. For SiO₂, agglomerate size and deposition decreased roughly 7-fold over the course of eight repeated sonications.

Figure 8.

Double-stranded DNA damage (p-γH2AX) at key points during subchronic exposure to 0.6 μg/cm² of selected particle treatments (~0.58 μg/cm² nFe₂O₃, ~0.54 μg/cm² SiO₂-nFe₂O₃, ~0.57 μg/cm² GMA-MS, ~0.24 μg/cm² SiO₂, or ~0.57 μg/cm² TiO₂ delivered dose). (A) Quantification of the integrated intensity of fluorescent signals, representative images of fluorescent staining at (B) 91 days, (C) 132 days, and (D) 202 days of continuous exposure. *p < 0.05 as compared to nontreated control cells at the same time point. #p < 0.05 as compared to SiO₂-nFe₂O_3-exposed cells at the same time point. Error bars indicate the mean ± SEM (n = 3).

Qualitative particle uptake by Beas-2B cells was evaluated using TEM (Figure 9) following 1 week of continuous exposure, with two consecutive particle treatments. All particle treatments except SiO₂ were clearly shown to be taken up by cells and were shown encased in membrane-bound organelles, which were likely endosomes or phagolysosomes.

Figure 9.

Representative transmission electron microscopy (TEM) images of Beas-2B uptake of particles following 1 week of continuous exposure (two consecutive particle treatments). Arrows refer to the location of higher magnification images.

Particle dissolution of nFe₂O₃, GMA-MS, and SiO₂-nFe₂O₃ was assessed in sterilized Milli-Q water (pH 6.40), BEGM (pH 7.46), or a phagolysosome-like buffer (pH 4.55) over 72 h by ICP-OES (Figure 10). In both water and BEGM, SiO₂-nFe₂O₃ had the greatest iron release as compared to other particle treatments. In phagolysosome-like buffer, GMA-MS had the greatest iron release, while nFe₂O₃ had the least. However, dissolution at only 1.2% over 72 h.

Figure 10.

Particle dissolution following a 72 h suspension in sterilized water (pH 6.40), complete cell culture medium (BEGM, pH 7.46), or phagolysosome-like buffer (pH 4.55) by ICP-OES. Error bars indicate the mean ± SEM (n = 3). Limit of quantification: 0.0283 μg/mL. Limit of detection: 0.0089 μg/mL, with a maximum of 1.5% dissolved particle measured over 72 h.

DISCUSSION

Within the context of a subchronic in vitro exposure model, the present study showed a particle type and time-dependent neoplastic-like cell transformation. Long-term exposure to a 0.6 μg/cm² administered dose (~0.58 μg/cm² delivered dose) of nFe₂O₃ induced neoplastic-like transformation in Beas-2B cells beginning at 4–4.5 months, which was associated with decreases in intracellular iron, minimal changes in ROS production, and significantly elevated cell proliferation, colony formation, and double-stranded DNA damage. The observed response was similar to that of GMA-MS-exposed cells (0.6 μg/cm² administered dose, ~0.57 μg/cm² delivered dose). Conversely, Beas-2B cells exposed to amorphous silica-coated nFe₂O₃ particles (SiO₂-nFe₂O₃, 0.6 μg/cm² administered dose, ~0.54 μg/cm² delivered dose) did not undergo a neoplastic-like transformation, nor exhibit significant changes in any of the above-measured parameters.

Acute assessment of cytotoxicity and ROS production (Figure 2) showed that nFe₂O₃, GMA-MS, and TiO₂ exposure induced significantly elevated ROS production at 18 h, which was maintained through 24 h for GMA-MS and TiO₂ treatments, at the highest dose used (2.0 μg/cm²). However, this influx did not have a significant impact on acute cytotoxicity or cell proliferation for nFe₂O₃ or GMA-MS. There were also no significant changes in intracellular iron (Figure 2) across particle treatment groups, although GMA-MS induced a trend of increased intracellular iron at 24 h postexposure at both the 0.6 and 2.0 μg/cm² administered dose.

Based on the presented acute response, target lung tissue dose determined from in vivo results,⁵³ and previous cell transformation data,⁶³ a subtoxic administered dose (0.6 μg/cm²) was selected for subchronic exposure. Such low-dose/long-term exposure is more physiologically representative of occupational and environmental exposure conditions than a high-dose/acute exposure scenario and may be more likely to contribute to the human development of chronic conditions.⁶⁴ Throughout the subchronic exposure period, cells were assessed for neoplastic-like transformation, using cell proliferation (Figure 4) and attachment-independent colony formation (Figure 5) as primary indications. Uncontrolled or dysregulated cell proliferation can indicate some type of abnormal cell phenotype,⁶⁵ while a soft agar colony formation assay is the recommended technique by OECD to assess continued proliferation when cells are plated in nonadherent conditions,⁵⁸ which is a hallmark of abnormal or cancer-like phenotype in cells.⁶⁵ Both nFe₂O₃ and GMA-MS-exposed-cells followed a similar trend of decreased proliferation at roughly 3–3.5 months, which became significantly elevated by 4.5–5 months, and was maintained through the rest of the 6.5-month exposure period. The initial lower proliferation observed in the first few months of exposure may have indicated some type of chronic cell injury or stress, leading to cell adaptation, and ultimately resulting in significantly elevated proliferation and overall cell transformation by the end of the exposure period. The 4.5–5-month time point, when elevated cell proliferation was first observed, coincided with significant elevation of attachment-independent colony formation, again indicating particle-induced cell transformation by nFe₂O₃ and GMA-MS. For nFe₂O₃ exposure, this trend has been previously reported³⁴ and, at the time, was suggested to be attributable to chronic particle injury, which may have resulted in chronically stressed cells and resulting adaptations.

TiO₂ exposure induced a similar trend in cell proliferation as GMA-MS and nFe₂O₃, but without a sustained neoplastic-like transformation. TiO₂-exposed cells also showed increased attachment-independent colony formation at 4.5–5 months, but this effect was gone by the end of the exposure period, even though cell proliferation remained elevated. This suggested that TiO₂ may have the potential to induce cell transformation, but a more transient kind. This is mirrored in the current literature, which has shown TiO₂ exposure to be associated with in vitro cytotoxicity^66,67 and in vivo tumorigenesis only at high doses.⁶⁸

nFe₂O₃- and GMA-MS-induced cell transformation was hypothesized to be reliant on particle-induced changes in intracellular iron, ROS production, and double-stranded DNA damage. It is generally believed that if a particle is engulfed by the cell via phagocytosis or endocytosis, the resulting phagosome will fuse with a lysosome for degradation. Once in this acidic environment, iron-containing particles may dissolve and release free iron ions into the cell’s labile iron pool. This increase in catalytically available iron can participate in the Fenton reaction and generate ROS,^69,70 which is known to induce DNA lesions,⁷¹ and may ultimately result in the observed neoplastic-like phenotype predicted with subchronic nFe₂O₃ or GMA-MS exposure. Therefore, this proposed mechanism was evaluated using changes in intracellular iron, ROS production, and double-stranded DNA damage at key points during the subchronic exposure for nFe₂O₃-, SiO_2-nFe₂O₃-, or GMA-MS-exposed cells, which, to our knowledge, has not been previously reported for any of these particles. However, our data showed that nFe₂O₃ and GMA-MS induced decreased intracellular iron and minimal changes in ROS production, but still induced significant double-stranded DNA damage. Therefore, the observed subchronic effects were likely not due to an influx of intracellular iron, but rather some kind of alternative mechanism. The minimal changes in ROS production observed by nFe₂O₃- and GMA-MS-exposed cells align with some previous literature, which has shown human lung adenocarcinoma cells to have minimal oxidative stress response with IONP exposure, even though the exposure still resulted in increased DNA damage.⁷² Cadmium, a heavy metal which is a known carcinogen, induced decreased oxidative stress⁷³ and minimal changes in related gene expression⁷⁴ with low-dose/long-term exposure, as well. For cadmium, adaptation and tolerance to oxidative stress were believed to be critical mechanisms involved in its overall toxicity and carcinogenic capacity.⁷⁵ Arsenic, another heavy metal, was reported to have a similar effect. Arsenic-transformed Beas-2B cells exhibited a decrease in ROS generation, which was posited to be responsible for observed increases in cell growth and colony formation.⁷⁶

However, the underlying mechanism of carcinogenic agent-induced cell transformation is not totally understood. The results presented here, as well as similarly conducted low-dose/long-term exposure studies, suggest other mechanisms involved in the nFe₂O₃ induced-cell transformation and DNA damage beyond the hypothesized intracellular iron-ROS-DNA damage pathway, such as a direct nanoparticle surface interaction with exposed cells, cellular uptake efficiency, and intracellular distribution of the engulfed particles, which we are currently investigating in our lab.

Subchronic GMA-MS exposure induced a similar response in cells compared to nFe₂O₃ treatment. Previous GMA-MS in vivo pulmonary exposure has shown minimal inflammatory or oxidative stress response, and no changes in iron homeostasis markers,⁷⁷ even though long-term exposure did induce genotoxicity⁷⁸ and tumor formation.³⁶ This aligns with the presented in vitro GMA-MS-induced cell transformation, which occurred without direct increases in intracellular iron or ROS production. GMA-MS-exposed cells also had a decreasing average number of lysosomes per cell over time (Figure 6C), suggesting cellular changes in autophagy, which have been previously shown to act as a protective mechanism in some types of particle-induced cell transformation.⁷⁹ Autophagy dysfunction as a potential underlying mechanism is also under further investigation in our lab.

As was hypothesized, compared to nFe₂O₃ treatment, SiO_2-nFe₂O₃-exposed cells did not show any significant changes in cell proliferation or attachment-independent colony formation throughout the subchronic exposure period and also did not exhibit any significant changes in intracellular iron, ROS production, or DNA damage. This suggested the protective capacity of the amorphous silica coating by reducing the overall cell interaction with nFe₂O₃. Many researchers have shown the utility of using a surface coating to alter particle function.^80–82 Previous studies have also shown the potential for an amorphous silica coating to reduce or postpone inflammatory outcomes in vivo with other types of nanometal oxides,⁸³ but this has not been previously shown for SiO_2-nFe₂O₃ within the context of a low-dose/long-term in vitro exposure model, as is presented here.

A complicating factor for in vitro nanotoxicology is the lack of information regarding dosimetry. In the present study, when able to do so, we evaluated how the known administered dose compared to the actual delivered dose of test particles. If the known administered dose and calculated delivered dose to the cells vary significantly, this may be able to account for unexpected observed cell response. Particle characterization was done in the present study via DLS (Table 1), SEM-EDX, and TEM (Figures 1 and 9) in conditions that replicated those of exposure. The delivered dose of each particle in a complete cell culture medium (BEGM) was calculated using the distorted grid (DG) dosimetry model. Due to the extended exposure period (6.5 months), particle suspensions in Milli-Q water were freshly prepared every 4 weeks, and sonicated before each particle treatment (every 3–4 days). This repeated sonication (≤8 times) affected the average agglomerate hydrodynamic diameter for particle treatments. However, regardless of hydrodynamic diameter, all test particles except amorphous SiO₂ had >86% deposition within the first 24 h after exposure (Table S1). Previous research has shown that particle settling rates are dependent on a complex interplay of each particle’s effective density, protein corona formation, and medium composition and viscosity.⁸⁴ The cell culture medium used in this study (BEGM) was serum-free and had a low density/viscosity. This, combined with the relatively high effective density of the metal-oxide particles, resulted in rapid particle deposition regardless of agglomerate size or the number of sonications. Amorphous SiO₂, however, had a much lower effective density, as well as steadily decreasing agglomerate size with repeated sonications, which resulted in the observed low deposition (≤70% after 24 h). Cells treated with amorphous SiO₂ did not receive a similar deposited dose as compared to other particle treatments; therefore, no definitive conclusions about the potential toxicity of amorphous SiO₂ can be made compared to other treatments used in the present study. Information about amorphous SiO₂ toxicity can be inferred from previous research, which has shown the overall benign^85,86 or protective^{83,85,87–89} qualities of amorphous SiO₂, especially when used as a coating for other types of metallic nanoparticles. Furthermore, in the present study, SiO₂-coated nFe₂O₃ had a similar deposition to its uncoated counterpart (nFe₂O₃), resulting in the direct contact of cells with amorphous SiO₂, but did not induce neoplastic-like cell transformation, which may further support the potentially protective/benign qualities of amorphous SiO₂. However, this example clearly emphasizes the importance of thorough particle characterization before exposure begins to ensure consistent dosing each time.

CONCLUSIONS

Subchronic in vitro exposure to nFe₂O₃ (~0.58 μg/cm² delivered the dose, for 6.5 months) induced a neoplastic-like cell transformation in human lung bronchial epithelial cells (Beas-2B), as characterized by elevated cell proliferation, increased attachment-independent colony formation, and induction of DNA damage. An amorphous silica-coated but otherwise identical nFe₂O₃ particle (SiO₂-nFe₂O₃, ~0.54 μg/cm² delivered dose) had markedly diminished neoplastic-like cell transformation, as indicated by no significant changes in any of the parameters described above.

GMA-MS subchronic exposure (~0.57 μg/cm² delivered dose) induced a neoplastic-like cell transformation in Beas-2B cells similar to that induced by nFe₂O₃ exposure, which was also associated with the induction of double-stranded DNA damage, but overall decreased ROS production and no significant changes in intracellular iron. Overall, this suggests that ROS may play a limited and not critical role in particle-induced neoplastic-like transformation in vitro or carcinogenesis in vivo.

This study provides support for physicochemical property-dependent modes of toxicity for nanometal oxide-induced carcinogenic potential and helps establish a “safe by design” hazard reduction strategy for emerging nanotechnology. Furthermore, this study gives evidence for the utility of a low-dose/long-term in vitro exposure model to assess the risk of cancer-related outcomes for occupationally relevant nanometal oxide exposures.

ABBREVIATIONS

IONP

iron oxide nanoparticles

GMA-MS

gas metal arc mild steel welding fumes

ROS

reactive oxygen species

nd

no data