The small airway epithelium as a target for the adverse pulmonary effects of silver nanoparticle inhalation

Abstract

Experimental modeling to identify specific inhalation hazards for nanomaterials has in the main focused on in vivo approaches. However, these models suffer from uncertainties surrounding species-specific differences and cellular targets for biologic response. In terms of pulmonary exposure, approaches which combine ‘inhalation-like’ nanoparticulate aerosol deposition with relevant human cell and tissue air–liquid interface cultures are considered an important complement to in vivo work. In this study, we utilized such a model system to build on previous results from in vivo exposures, which highlighted the small airway epithelium as a target for silver nanoparticle (AgNP) deposition. RNA-SEQ was used to characterize alterations in mRNA and miRNA within the lung. Organotypic-reconstituted 3D human primary small airway epithelial cell cultures (SmallAir) were exposed to the same spark-generated AgNP and at the same dose used in vivo, in an aerosol-exposure air–liquid interface (AE-ALI) system. Adverse effects were characterized using lactate, LDH release and alterations in mRNA and miRNA. Modest toxicological effects were paralleled by significant regulation in gene expression, reflective mainly of specific inflammatory events. Importantly, there was a level of concordance between gene expression changes observed in vitro and in vivo. We also observed a significant correlation between AgNP and mass equivalent silver ion (Ag⁺) induced transcriptional changes in SmallAir cultures. In addition to key mechanistic information relevant for our understanding of the potential health risks associated with AgNP inhalation exposure, this work further highlights the small airway epithelium as an important target for adverse effects.

Keywords:

• Small airway
• inflammation
• genomics
• mechanistic toxicology

Introduction

Respiratory hazards from nanomaterials are an important consideration in circumstances of inadvertent inhalation exposure. As a consequence of their anti-microbial activity, among other properties, silver nanoparticles (AgNPs) are widely used and increasingly incorporated into consumer products (Piccinno et al. 2012; Wang et al. 2014). An understanding of how AgNPs may exert adverse effects on the lung after inhalation has derived mainly from studies of in vivo and in vitro exposure models. Rodent models have been used extensively for toxicity characterization and distribution assessment. One inhalation study in rats found that sub-chronic exposures to AgNPs across 90 days produced histologic changes within the lung, associated with mixed inflammatory cell infiltrate and chronic alveolar inflammation (Sung et al. 2009). Additional sub-chronic studies have observed similar effects, including those on lung function (Ji et al. 2007; Sung et al. 2008; Song et al. 2013). Shorter term inhalation exposures have also revealed that smaller sized AgNPs induce significantly more pulmonary inflammation than larger sized particles (Braakhuis et al. 2014). This difference was attributed in part to increased alveolar dose deposition and dissolution rate of the smaller particles producing increased levels of ionic silver. More recently, we set out to investigate the acute effects of AgNPs inhalation in more detail (Seiffert et al. 2016). Similar to the previous study, we found increased neutrophilia as well as increased inflammatory and tissue injury markers 1 day after exposure, which were diminished at 7 days. We also found dose-dependent AgNP accumulation within macrophages. This localization has been suggested to initiate inflammatory events as a consequence of particle accumulation (Mukherjee et al. 2014).

Growing evidence suggests that the predominant mode of toxicity for AgNP is a consequence of the magnitude of soluble Ag ion release and their interaction with biologic systems (Hadrup and Lam 2014). It is suggested that lysosomal degradation of internalized AgNPs is an important control point for Ag⁺ release (Gliga et al. 2014; Jiang et al. 2015; Setyawati et al. 2014; Arai, Miyayama, and Hirano 2015). Additional mechanisms have been suggested, including the production of reactive oxygen species (ROS) as a consequence of mitochondrial damage (AshaRani et al. 2009; Foldbjerg, Dang, and Autrup 2011). This mode of injury has also been suggested to cause DNA damage resulting in cell cycle arrest at the G2M phase (AshaRani et al. 2009; Lee et al. 2011; Asharani et al. 2012). Other studies however have proposed ROS-independent effects on cellular injury and cell cycle arrest (Chairuangkitti et al. 2013; Sweeney et al. 2016). Investigations are still needed to determine the full profile of how AgNPs exert their effects.

In addition to macrophage NP phagocytosis and activation, airway epithelial cells have also been proposed as a primary site for mediating AgNP adverse effects within the lung. Much of this investigation has focused on the bronchial and alveolar compartments (Foldbjerg et al. 2012; Cronholm et al. 2013; Braakhuis et al. 2014). Interestingly, however, little attention has been devoted to study of the bronchiolar compartment composing the small airways. This is despite evidence that inhalation of AgNP results in deposition in this region. Anderson et al. (2015b) demonstrated Ag deposition in the terminal bronchioles localized to club cells and the sub-epithelial layer after AgNP instillation. Further observations revealed that inhalation of AgNP also resulted in Ag localization mainly to the terminal bronchiole/alveolar duct region (Anderson et al. 2015a). Ag staining on the luminal surface and the sub-epithelial connective tissue of the terminal bronchiole was also observed after AgNP inhalation (Seiffert et al. 2016). The suggestion is that clearance of AgNP through epithelial uptake may point towards a mechanism through which these cells receive doses of Ag capable of eliciting adverse effects and directing adaptive tissue responses, such as the initiation of inflammation.

It was the aim of this current study to investigate the contribution of the small airway epithelium to pulmonary responses to inhaled AgNPs. Initial steps involved characterization of global transcriptomic responses within the rat lung after inhalation of AgNPs. Cataloging both mRNA and microRNA (miRNA) allows for a comprehensive analysis of both adverse pathology and mechanisms of action of AgNP pulmonary effects. This analysis was carried out on lung tissue after bronchoalveolar lavage to focus as much as possible on the resident epithelial compartments. Following on from this, we compared these tissue responses to small airway epithelial cells exposed to the same spark-generated AgNPs across the same dose range using an aerosol-exposure air–liquid interface (AE-ALI) deposition device. The cells comprised a newly developed model of differentiated primary human small airway epithelial cells displaying organotypic 3D structure and club cells typical of small airways found in humans. Correlation between in vivo and in vitro model transcriptomic responses was observed.

Methods

AgNP generation and characterization

AgNP aerosols were produced using a spark generator (DNP 4000; Palas, Karlsruhe, Germany) by the homogeneous nucleation of vapor produced between two electrodes (5 × 1 mm silver wire; Premion™ 99.999% purity, Alfa Aesar™; Heysham, UK) in an inert argon atmosphere. The rate of primary particle production and final size was dependent on the sparking frequency (90–300 Hz). The particles were passed through a krypton-85 charge neutralizer (Model 3077A, TSI Inc., Shoreview, MN) prior to dilution and experimental exposure. A condensation particle counter (CPC model 3775, TSI Inc.) continuously monitored the aerosol particle number concentration during exposures and aerosol mass concentrations were determined gravimetrically using Pallflex^® emfab™ filters (Pall Life Sciences, Ann Arbor, MI) with the aerosol drawn at 2 L min⁻¹, the results are presented in Table 1. The average number-weighted aerosol particle size distribution was determined using a scanning mobility particle sizer (SMPS; model 3936N76, TSI Inc.) and for in vitro exposures results are displayed in Supplementary Figure 1(A). The shape of the aerosol particles was determined with high-resolution transmission electron microscopy (TEM) (JEOL 3000F; JEOL Inc., Tokyo, Japan). A representative TEM image of particles is shown in Supplementary Figure S1(B). Detailed AgNP characterization results for in vivo exposures are previously reported (Seiffert et al. 2016).

Table 1. Parameters for in vivo and in vitro aerosol exposures.

	Group	Exposure length	Mass conc.(µg/m^3)	Number conc.(#/cm^3)	CMD (nm)	GSD	Estimateddose (ng/cm^2)	Estimateddose (#/µm^2)
In vivo		3 h/day for 4 days	670 ± 49	4.6 ± 0.7 × 10^7	14.1 ± 2.3	1.58	4.9	3.3
In vitro	Low	7 min	37.4*	1.2 ± 0.3 × 10^6	16.5 ± 1.1	1.33	1.5	0.4
	Medium	20 min	37.4*	1.0 ± 0.3 × 10^6	16.0 ± 0.9	1.36	4.4	1.3
	High	60 min	37.4*	1.1 ± 0.1 × 10^6	15.8 ± 0.2	1.36	13.2	3.8

*The mass concentration is an average for all the in vitro exposures. CMD: count median diameter; GSD: geometric standard deviation.

In vivo AgNP exposure

The lung tissues processed for transcriptomic analysis were collected from a previous exposure study (Seiffert et al. 2016). Briefly, male Sprague-Dawley rats (10–12 weeks, 250–320 g) (Harlan, UK) were exposed to nose only aerosols of AgNPs under UK home office approval (Project Licence number: PPL 70/7581) with exposure set up and schematic as previously described (Seiffert et al. 2016). Animals were exposed for 3 h on four consecutive days to AgNPs with an estimated lung deposition of 4.9 ng/cm² (Table 1). Animals were sacrificed at 1 day or 7 days after exposure. Control animals were exposed to air using the same system.

In vitro AgNP exposure

Organotypic-reconstituted primary human small airway epithelial cell cultures (SmallAir™), differentiated and maintained at an ALI on 6.5 mm diameter tissue culture inserts were obtained from Epithelix (Geneva, Switzerland). Cultures were maintained at 37 °C in a 5% CO₂ humidified atmosphere with basolateral media (SmallAir™ culture media # EP65SA) exchange and apical washing every 2–3 days. Five different donors without any reported pathology were used for experimental exposures. Cultures were exposed to AgNPs using a Cultex^® RFS exposure module (Cultex^® Laboratories GmbH, Hannover, Germany) with an EDD (electrical deposition device) to enhance the particle deposition efficiency. With temperature at 37 °C, oxygen, CO₂, nitrogen, and relative humidity were maintained at 20, 5, 75 and 80%, respectively. A schematic of the AE-ALI set up is shown in Supplementary Figure S2. Exposures were designed to provide deposition doses in a range including that estimated for our in vivo rat exposures (Seiffert et al. 2016) to the same spark-generated material. This resulted in exposure times of 7, 20 and 60 min. Details of how these exposure doses and therefore durations were estimated, including the calculation method can be found in the Supplementary methods with the results summarized in Table 1. ALI cultures of SmallAir were also exposed to solutions of Ag⁺ in the form of AgNO₃ (Sigma Aldrich, Dorset, UK). Ag mass dose equivalents to AgNP exposures were applied to the apical surface of the cells in 30 µl volume of cell culture media. This minimal volume, recommended by SmallAir manufacturer to have little impact on ALI differentiation status, did not have any significant effects on toxicological endpoints when compared to air alone exposure. After exposure, ALI cultures were returned to fresh culture media under normal culture conditions for a further 6 or 24 h, after which, samples were processed for gene expression and toxicity assessment as detailed in Figure 2.

In vivo rat inhalation RNA-SEQ analysis

After removal of bronchoalveolar lavage (BAL) cells through consecutive washes in PBS, total RNA was extracted from whole lung tissue using the miRNeasy Mini Kit (QIAGEN, Manchester, UK). RNA quality was determined using an Agilent 2100 Bioanalyser and those samples with RIN above 8.0 were used for library preparation. For mRNA sequencing analysis, samples were processed using the TruSeq™ RNA sample preparation method (Illumina, San Diego, CA) as previously described (Meldrum et al. 2016). Processing and sequencing was carried out in conjunction with BGI Tech Solutions (Hong Kong) using 101PE sequencing on the Illumina HiSeq^TM 2000 platform. Raw sequence data was initially processed to remove adaptor sequences, contamination and low-quality reads. Twenty million clean reads for each sample was obtained and brought forward for further processing using CLC Genomics Workbench software (CLCBIO, Aarhus, Denmark). Paired end reads were trimmed to remove remaining adapter and other variable sequences followed by annotation using the Rnor_5.0 (Ensembl release 79) rat reference genome build. For normalization across gene and sample reads, RPKM (Reads Per Kilobase per Million mapped reads) values >0.1 in at least one treatment condition were selected for differential expression analysis. Further comparative analysis and visualization of differentially regulated transcripts was carried out using Qlucore Omics Explorer software (Qlucore, Lund, Sweden). Pathway analysis of differently regulated transcripts was carried out using ingenuity pathway analysis (IPA) software (http://www.ingenuity.comQIAGEN, Manchester, UK). For miRNA sequencing analysis, small RNA between 18 and 30 nucleotides was fractionated and processed using TruSeq™ small RNA library preparation followed by 101 PE sequencing. Raw data was processed, including adaptor removal and annotation using CLC Genomics Workbench software with 10 million clean reads brought forward. Annotation was carried out against the rat miRBase (Release 20 reference dataset).

ICP-MS measurements

Lung tissue was fixed in formalin and 5 µm sections were analyzed for ¹³C and ¹⁰⁷Ag spatial distribution by laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS) using a New Wave Research NWR213 laser ablation system (Electro Scientific Industries, Portland, OR) linked to an iCAP Q ICP-MS (Thermo Fisher Scientific, Hemel Hempstead, UK). AgNP deposition quantification on transwell inserts after AE-ALI exposure in the absence of cells was carried out by acid digestion of the culture membrane prior to iCAP Q ICP-MS analysis. In addition to ¹⁰⁷Ag, ¹⁹³Ir content was used as an internal standard. Further details can be found in the Supplementary methods.

In vitro cytotoxicity assessment

The metabolite lactate was measured in the cell culture media on the basolateral side as well as from a media wash (100 µl/transwell) sample of the apical side of the cultures after treatment. This was used as an indicator of cell stress and altered metabolic activity (Limonciel et al. 2011). Further details of how this assay was performed can be found in the Supplementary methods. Levels of lactate dehydrogenase in these media samples were also assessed as an indicator of compromised membrane activity and direct cytotoxicity. This was carried out using a kit from Roche Diagnostics (Mannheim, Germany) as per manufacturer’s instructions.

In vitro PCR analysis

Total RNA was isolated using the miRNeasy Mini Kit (QIAGEN) according to the manufacturer’s instructions. For mRNA, cDNA was synthesized from 1 μg RNA using the QuantiTect Reverse Transcription Kit (QIAGEN). RT-PCR reaction mixtures contained 1× Fast SYBR^® Green Master Mix (Thermo Fisher Scientific), primer pairs at a concentration of 200 nM and 2 ng of cDNA in 20 μl reaction volume. Primers were designed using Primer3 software (http://primer3.ut.ee/) and obtained from Integrated DNA Technologies (Leuven, Belgium). Primer sequences can be found in Supplementary Table S1. Human HPRT1 levels were used to normalize expression data using the ΔΔC_T method. No signals were detected in no-template controls. For miRNA, reverse transcription was carried out using a TaqMan MicroRNA reverse transcription kit (Thermo Fisher Scientific) according to the manufacturer’s protocol. hsa-miR-146b, hsa-miR-224, hsa-miR-1298-3p, and hsa-miR-21 were detected using TaqMan MicroRNA assays (Thermo Fisher Scientific). Expression levels were normalized to RNU6B. Standard 40 cycle PCR reactions were used and fluorescent detection was performed using the QuantStudio^® 6 Flex Real-Time PCR System (Thermo Fisher Scientific). Genes were selected for analysis in vitro, based on those which were differentially regulated in vivo by AgNP. In total 50 genes and 4 miRNAs were chosen to cover the most highly regulated transcripts and miRNAs as well as the most prominent pathways identified from RNA-SEQ analysis. miRNAs were selected also on the basis of comparison with IPA regulator analysis and overlap with miRNA-SEQ results (Supplementary Figure S4).

Statistical analysis

For in vitro exposure of HPSAE cultures, all data for cytotoxicity assays and PCR analysis is presented as mean ± standard error of the mean (SEM) for five different donors. Statistical significance was evaluated based on the ratio paired t-test unless otherwise stated.

Results

Lung transcriptomic profiles after AgNP inhalation

To establish the degree to which lung tissue was impacted by AgNP exposure over the short term, we examined global mRNA and miRNA levels using RNA-SEQ. This type of analysis allows for the capture of global transcriptional changes including miRNAs, whose function is to modulate transcriptional events, in order to provide insight into the most relevant cellular and tissue responses underlying the response to toxicological insult. Initial bioinformatic analysis was carried to identify those statistically relevant transcripts, which can be used to provide confident assessment of tissue injury and resolution responses. Importantly, this type of global analysis is used to provide detailed mechanistic information into how toxicological insult occurs at the cellular and molecular level. Hierarchical clustering of 17,804 mRNA transcripts across groups (N = 6 animals per group) revealed 1 outlier at Day 7 control (Air) exposure, which was removed. Differential expression criteria of 1.5-fold change and a statistical p = 0.005 (ANOVA) was applied and resulted in 493 transcripts at Day 1 and 97 transcripts at Day 7. PCA analysis of these transcripts (Figure 1(A)) demonstrates separation of AgNP treatment groups with control air exposure groups clustering together. Hierarchical clustering of this data visualized as a heatmap (Figure 1(B)) demonstrates substantially more upregulated than downregulated transcripts with AgNP treatment. The top 20 upregulated transcripts at Day 1 are listed (Figure 1(C)) and can be broadly characterized as part of inflammatory processes. In addition to the immune cell recruitment signals within these top 20, there was a large representation of genes involved in regulating type II inflammation. The most significant pathways associated with all 493 differentially regulated transcripts were involved in regulating inflammatory processes (Figure 1(D)). Also prominent were pathways involved in DNA damage and cell cycle regulation. A full list of differentially expressed transcripts can be found in Supplementary Figure S3. Examination of transcripts for consistent regulation across Day 1 and Day 7 post-exposure, revealed 16 transcripts including Mt1a, Chia, Duox1 and CD68. This would indicate a persistent response to AgNPs within the lung involving regulation of metal response pathways, oxidative stress and macrophage activity. For miRNA expression analysis small RNA sequences were annotated to 390 known miRNAs. Hierarchical clustering of miRNA revealed an outlier within the Day 7 control (Air) group, which was removed from further processing. PCA analysis of these miRNA (Figure 3(A)) was used to select for those miRNA that separate treatment groups based on ANOVA, using a false discovery rate of 0.05 (p = 0.001)) (Figure 3(A)). This resulted in 12 miRNA sequences and similar to mRNA transcript analysis, hierarchical clustering visualized as a heatmap revealed more upregulated miRNA with AgNP treatment than downregulated (Figure 3(B)). The magnitude of change induced by AgNP was below 3-fold (Figure 3(C)).

Figure 1. AgNP aerosol exposure alters rat lung mRNA expression. Sprague-Dawley rats (n = 5–6 per group) were exposed to control gas (Air) or AgNP Aerosol (4.9 ng/cm²) (NP). Bronchoalveolar lavage was performed and lung tissue was removed for mRNA expression analysis by RNA-SEQ, 1 and 7 days post-exposure. PCA analysis (A) and hierarchical clustering (B) were used to visualize patterns of differentially expressed transcripts. The most highly upregulated transcripts by AgNP are displayed as fold over control (C) and ordered according to Day 1 regulation. Pathway analysis of differentially regulated transcripts by AgNP (Day 1 and 7 combined) is displayed as a function of significant association (D).

Figure 2. Schematic representation of small airway in vitro cell exposure and toxicological assessment. Endpoint analysis was carried out 6 and 24 h post-exposure.

Figure 3. AgNP aerosol exposure alters rat lung miRNA expression. Sprague-Dawley rats (n = 5–6 per group) were exposed to control gas (Air) or AgNP Aerosol (4.9 ng/cm²) (NP). Bronchoalveolar lavage was performed and lung tissue was removed for miRNA expression analysis by RNA-SEQ, 1 and 7 days post-exposure. PCA analysis (A) and hierarchical clustering (B) were used to visualize patterns of differentially expressed miRNAs. Details of AgNP induced fold expression changes for miRNAs are also displayed (C).

Analysis of Ag levels within lung

To evaluate levels of Ag distribution within the distal rat lung 1 day after inhalation of AgNPs, laser ablation ICP-MS was used. Initial observations revealed that the small airway bronchiolar-sized airways appeared to display higher levels of ¹⁰⁷Ag than surrounding alveolar structures (Figure 4(A)). Quantification of ¹⁰⁷Ag content normalized to ¹³C within the airway walls confirmed higher levels within the small airway bronchiolar structures when compared to alveolar walls (Figure 4(B)).

Figure 4. Analysis of ¹⁰⁷Ag levels with rat lung after AgNP exposure. After AgNP exposure, lung tissue was fixed and processed for laser ablation ICP-MS assessment of ¹⁰⁷Ag levels. Representative light microscopy image (A; left panel) and associated map of spatial distribution of ¹⁰⁷Ag (A; right panel) for the distal lung. Quantification of airway wall ¹⁰⁷Ag as a proportion of ¹³C content was carried out for both bronchiolar and alveolar structures for four separate airway structure types in each animal (n = 3 animals) (B). Results are expressed as mean ± standard error of mean (SEM) normalized units and statistical analysis performed by t-test (*p < 0.05).

Small airway epithelium exposure to AgNP aerosol in vitro

SmallAir epithelial cultures are organotypic differentiated cultures of human primary small airway epithelial cells (Huang et al. 2017). These cultures are maintained at the ALI and comprise ciliated, secretory as well as basal cells within a pseudostratified structure and express CC10, a marker of small airway club cells. Considering this in vitro culture as a close representative of the human small airway epithelium in vivo, we investigated the effects of AgNP aerosol exposure using an ALI particle deposition device. In order to directly compare effects with our in vivo inhalation observations we used the same spark-generated material for our in vitro ALI exposures. The size of the AgNPs used for in vitro exposure was slightly larger than used for in vivo exposure (∼16 versus 14 nm) (Table 1). We used three different exposure times (7–60 min) to the AgNP aerosol to provide a range of dose estimates for both mass and particle number that would cover the dose estimate for in vivo lung exposure (Table 1). In addition to estimates of in vitro exposure dose for these times (Table 1), we also quantified the amount of Ag deposited onto cell culture inserts in the absence of biologic material using ICP-MS. The quantities for deposited ¹⁰⁷Ag were 1.56 ± 0.24 ng/cm² for 7 min exposure, 2.49 ± 0.15 ng/cm² for 20 min exposure, and 4.59 ± 0.24 ng/cm² for 60 min exposure (n = 3 for all measurements).

Adverse effects of AgNP exposure

Exposure of SmallAIR cultures to AgNP resulted in no significant alterations in gross cytotoxicity indicators with cell stress marker lactate unaltered across the range of exposures (Figure 5(A)). LDH assessment, however, did reveal significant differences compared to control after 6 h at the highest 13.2 ng/cm² dose (Figure 5(B)). Basolateral levels at both 6 and 24 h indicated an increase with exposure, which did not reach statistical significance. Gene expression analysis was also carried out across these exposure ranges. Air only exposures at 7, 20 and 60 min were performed to control for system exposure effects and analyzed for a range of stress responses genes at 24 h post-exposure. For 20 and 60 min, exposure in the AE-ALI system resulted in significant alterations in stress response genes (Supplementary Figure S5). As the 7 min exposure did not result in such stress responses and since we cannot be confident of how system exposure may alter or confound gene expression results, further interrogation was only carried at the lower dose of AgNP (7 min; 1.5 ng/cm²). The majority of the 50 transcripts chosen from the in vivo RNA-SEQ analysis were upregulated by AgNP exposure (Figure 5(C); right hand panel). Of these 50 genes, 24 displayed significant alterations in expression in response to in vitro AgNP at 24 h (Figure 5(C); left hand panel). Of these 24, 20 were regulated in the same direction as in vivo. While these in vitro changes may capture the airway epithelial cell response to AgNP treatment, the level of correlation between in vitro and in vivo was not statistically different (Pearson’s correlation). Two of the four miRNAs selected for analysis were significantly altered by AgNP in vitro (Figure 5(C)). These were, however, not regulated in the same direction as in vivo.

Figure 5. Exposure of small airway epithelium to AgNP aerosol. Human primary small airway epithelial cell cultures (SmallAIR) were exposed to AgNP aerosols using an AE-ALI deposition device. After 7, 20 or 60 min of exposure (respective doses 1.5, 4.4 and 13.2 ng/cm²), cultures were incubated for 6 or 24 h and apical wash and basolateral media was assessed for levels of lactate (A) or LDH (B). RNA was also assessed for relative expression of 50 mRNA and 4 miRNA transcripts displayed as fold over control values for 7 min exposure (C; left panel). For comparison, in vivo lung AgNP transcript fold change alterations from RNA-SEQ data were also included (C; right panel). All AgNP measurements were normalized to air-exposed control cultures. Statistical significance was calculated between unexposed and exposed conditions using paired t-test (⁻p < 0.1, *p < 0.05, **p < 0.01).

The mechanisms through which AgNPs have been suggested to mediate their effects include NP dissolution and cellular alterations in response to silver ion (Ag⁺) levels. To investigate whether Ag⁺ impact cellular responses to the same extent as AgNP, we exposed SmallAIR cultures to AgNO₃ doses covering both in vivo and in vitro AgNP exposure ranges. Levels of lactate from these exposures were observed as significantly decreased at both 6 and 24 h exposure, indicating an impact on normal cellular metabolism (Figure 6(A)). Apical LDH levels were increased at 24 hrs at the highest Ag⁺ dose indicating some level of toxicity (Figure 6(B)). Genes were selected for Ag⁺ analysis from those which were significantly regulated by AgNP in vitro and in vivo. For the most part Ag⁺ treatment displayed minimal effect on gene expression at 6 h, while increased significantly for 25 out of 28 genes analyzed at 24 h exposure (Figure 6(C)). Direct comparison of Ag⁺ gene expression to AgNP in vitro revealed a significant correlation (Pearson correlation coefficient r = 0.4124; p = 0.019).

Figure 6. Exposure of small airway epithelium to Ag⁺ solution. Human primary small airway epithelial cell cultures (SmallAIR) were exposed to AgNO₃ at a dose range of 0.5–13.2 ng/cm² for either 6 or 24 h. Apical media wash and basolateral media was assessed for levels of lactate (A) or LDH (B). RNA was also assessed for relative expression of 28 mRNA and 4 miRNA transcripts displayed as fold over control values (C). All Ag⁺ measurements were normalized to air-exposed control cultures. Statistical significance was calculated between unexposed and exposed conditions using paired t-test (*p < 0.05, **p < 0.01, ***p < 0.001).

Discussion

Acute pulmonary exposure to AgNP in vivo causes a substantial increase in neutrophil accumulation within the lung at Day 1, diminished at Day 7, with little effect on other inflammatory cell recruitment (Seiffert et al. 2016). This response was paralleled by changes in neutrophil chemoattractant transcripts within lung tissue, including cxcl5 as the fourth highest regulated after 1 day, again diminished after Day 7. Other transcripts for genes associated with neutrophils were also regulated in a similar way, including cxcl1, cxcl2 and cd177. Macrophage levels were modestly increased on AgNP exposure (Seiffert et al. 2016). Ccl22 and ccl1, which regulate monocyte and macrophage recruitment, were among the top 10 most highly regulated genes at Day 1. This together with changes in expression of genes associated with macrophage function, mmp12, retnla and trem1 indicate the involvement of these cells in addition to neutrophils as primary responders to AgNP exposure. Complement genes including C3 were also increased to a significant level. In addition, there was a significant proportion of transcripts involved in regulating type II inflammation within the top 20 genes. This may be important for pulmonary hazard assessment given the prominence this immune profile has in asthma and allergic airway disease (Meldrum et al. 2017). In addition to pro-inflammatory events, there was also evidence of inflammation resolution processes with the induction of IL1R2 expression.

SLC26A4 is an epithelial cell expressed anion transporter that has been suggested as a regulator of mucin production and inflammatory events in airway conditions such as asthma and COPD (Nakagami et al. 2008; Nakao et al. 2008; Gorrieri et al. 2016; Suzuki et al. 2016). Expression of this gene was observed as the second most highly regulated transcript (Figure 1(C)). Duox1 expression was also increased in our study and has been previously observed to mediate small airway epithelial repair and bronchiolar reepithelialization in a model of naphthalene-induced small airway injury (Gorissen et al. 2013). The Club cell marker CLDN10 (Zemke et al. 2009; Fukumoto et al. 2016) was also significantly altered in the lungs of these animals by AgNP. Regulation of other epithelial specific genes including ect2, sftpa1, sftpd, muc1 and cftr was also observed. This data suggests the airway epithelium including the small airway epithelium as responding to NP inhalation exposure. We addressed whether Ag was deposited in the distal airways, in particular the bronchiolar and alveolar regions, using laser ablation ICP-MS detection. We found higher levels of Ag in the bronchiolar wall when compared to alveolar (Figure 4). All of this evidence strongly suggests the small airway epithelium as a target for AgNP inhalation exposure.

To date, the investigation of the small airway epithelium as a primary responder to NP exposure has been exclusively limited to in vitro studies of model cell lines and primary cells in submerged culture, including (Ng et al. 2013; Snyder-Talkington et al. 2013; Sisler et al. 2016; Stueckle et al. 2017). Submerged culture is not optimal for differentiation of airway cells towards in vivo functionality and does not allow for representative NP deposition to occur using aerosolized material. These limitations can be overcome with the use of ALI culture. In our study, we use such an approach to investigate deposited NP aerosol effects on ALI cultures of small airway epithelial cells for the first time utilizing SmallAIR (Huang et al. 2017). Using an electrostatic charge coupled particle deposition device we exposed these cultures to the same AgNP aerosol used for an in vivo rat inhalation study to investigate how small airway epithelial cells may account for primary responses within the lung.

Estimation of deposition by ICP-MS analysis compared well with the manufacturer’s estimated deposition efficiency of 40% (Aufderheide et al. 2011). Distribution of Ag across the 6.5 mm insert membrane was relatively homogeneous (data not shown) but some caution should be taken using larger inserts due to potential uneven concentric deposition, which may require optimization. Caution should also be taken with exposure duration within the system, where greater than 7 min exposure to air alone resulted in adverse cell gene expression. Despite this, when we compared the results from an estimated dose (1.5 ng/cm²) in vitro, which was lower than that estimated in vivo, we still achieved a level of concordance in transcript regulation indicative of similar biologic responses. The application of ionic silver to ALI cultures at doses covering the estimates of in vivo exposure to Ag also resulted in concordant transcript expression changes. While comparison of in vitro to in vivo effects must take into consideration multiple factors including species effects, other cell types and particle handling, that may account for incomplete concordance, we believe we have established an in vitro system and exposure set-up that can attribute, in part, the pulmonary stress response to AgNP inhalation to the small airway compartment.

Dosimetry comparisons to real-world exposure scenarios are an important component for how our study may be considered for risk assessment. There is limited information on AgNP exposures in the workplace or public exposure from the use of products containing AgNPs. Measurements within a manufacturing facility have indicated typical workplace exposure levels ∼1 µg/m³ with maximum concentrations up to 300 µg/m³ (Lee et al. 2012). If one is to consider cumulative dose these concentrations are comparable to those used in our study. Our findings also point towards the mechanisms through which AgNP have their effects at the level of material characteristic, biologic stress response and specific cellular injury. Both the nanoparticulate and dissolved component of AgNPs have been suggested to contribute to adverse pulmonary effects (Kawata, Osawa, and Okabe 2009), with the percentage of Ag⁺ ion contributing to cellular responses estimated to be between 10 and 60% (Beer et al. 2012; Maurer-Jones et al. 2013; Comfort, Maurer, and Hussain 2014). From our SmallAir ALI cultures, we found a significant correlation between AgNP and Ag⁺ regulated RNA transcripts. This is similar to other studies (Bouwmeester et al. 2011; van der Zande et al. 2016) and suggests a high proportion of the effects of AgNP may be attributable to the dissolved fraction. Silver ions are biologically active and are known to interact with cellular components including proteins such as cell membrane receptors and those containing metal ions (Lansdown 2006). Cellular defense systems are in place to mitigate the effects of metal ion exposure such as the metallothioneins and anti-oxidant response pathways such as Nrf2. Examination of mRNA-SEQ data from rat lung exposed to AgNP aerosol revealed multiple pathways and genes outlining specific stress responses. Metallothioneins aim to sequester metal ions preventing intracellular biochemical disruption and both MT1A and MT2A were upregulated by AgNP. Also prominent among these responses was an oxidative stress response, which included upregulation of SLC7A11, NOXO1, DUOXA1, SOD2 and HMOX1, all Nrf2 dependent genes involved in defense against oxidant and electrophilic stress (Jennings et al. 2013). A mechanism previously proposed to outline AgNP toxicity involves disruption of the mitochondrial respiratory chain by Ag⁺, leading to production of ROS and interruption of ATP synthesis, in turn causing oxidative DNA damage, which may also involve direct Ag⁺ interaction with DNA leading to cell cycle arrest in the G(2)/M phase (AshaRani et al. 2009; Lee et al. 2011). Our data would generally support aspects of this mechanism, where we observe markers of oxidative stress and regulation of genes involved in G2/M phase cell cycle arrest. With the alteration in levels of the mitochondrial genes, NOXO1 and SOD2, there is also evidence that mitochondrial disruption occurs with AgNP exposure. We did not observe alterations in lactate levels in our in vitro exposures, indicating mitochondrial injury was insufficient to push metabolism towards glycolysis, as has been previously observed with nano-silver exposure (Chen et al. 2014). Other mechanisms suggested to mediate Ag nanomaterial effects include inhibition of thioredoxin reductase activity and selenium metabolism (Srivastava et al. 2011). We did not observe any alterations in transcripts involved in this pathway.

miRNAs are small endogenous noncoding RNAs of 19–24 nucleotides in length with the potential to control the expression of hundreds of mRNA at the post-transcriptional level. In our study, we identified differentially regulated miRNA transcripts using RNA-SEQ and found a small number of miRNAs altered by AgNP exposure. These included mir146, mir155, mir21 and mir224 all of which have been observed regulated as part of inflammatory processes in response to nanoparticulate exposure within the lung (Moschos et al. 2007; Izzotti et al. 2009; Bourdon et al. 2012; Dorhoi et al. 2013; Fossati et al. 2014; Malmhall et al. 2014). Analysis of select miRNAs in our ALI cultures after AgNP exposure revealed differences when compared to in vivo exposure. Only mir224 demonstrated significant regulation in response to AgNP but in the opposite direction to in vivo changes. Regulation of miRNAs by Ag ion treatment displayed alterations in levels with increases at 6 h and downregulation at 24 h and may indicate similar effects could occur with AgNP treatment. Given these temporal differences, the limited numbers and the low magnitude of change it is difficult to see how changes in miRNAs may be used as sensitive biomarkers for specific injury or provide substantial insight into mechanisms of toxicological insult from this study.

Conclusion

In conclusion, our findings support the suggestion that a component of the in vivo pulmonary response to inhaled AgNP arises from the small airway epithelium. They also support the principle that using differentiated in vitro models in an environment representative of in vivo exposure as a relevant and appropriate model to capture detailed cellular responses when pursuing in vitro culture alternatives. Transcriptomic analysis of lung tissue provided important information on specific cell type responses to AgNP, including small airway effects and immune cell recruitment and regulation. It also allowed us to more clearly define inflammatory phenotypes such as complement activation and type II inflammation responses, as well as inflammation resolution indicators. From a cellular signaling perspective, we observed responses indicative of DNA damage, cell cycle effects, oxidative stress as well as metal overload. This information together with our ability, in part, to capture many of these tissue responses using cultures of small airway epithelial cells exposed to AgNP and Ag⁺ ions adds significant detail and strengthens our mechanistic understanding of how acute AgNP inhalation exposure impacts the lung.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The original in vivo study was part funded by grants from US National Institute for Environmental Health Sciences grant number [U19ES019536] (http://www.niehs.nih.gov/research/supported/index.cfm), and from the UK National Environmental Research Council grant [NE/H012893] (http://www.nerc.ac.uk/research/). Transcriptomic analysis of lung tissue samples was funded by Public Health England. The in vitro aspect of this research was part funded by the National Institute for Health Research Health Protection Research Unit (NIHR HPRU) in Health Impact of Environmental Hazards at King’s College London in partnership with Public Health England. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, the Department of Health or Public Health England.