Molecular Signature of Asthma-Enhanced Sensitivity to CuO Nanoparticle Aerosols from 3D Cell Model

Molecular Signature of Asthma-Enhanced Sensitivity to CuO Nanoparticle Aerosols from 3D Cell Model Ingeborg Kooter, † Marit Ilves, ‡,∇ Mariska Grollers-Mulderij, †,∇ Evert Duistermaat, § Peter C. Tromp, † Frieke Kuper, † Pia Kinaret, ∥,⊥ Kai Savolainen, # Dario Greco, ∥,⊥ Piia Karisola, ‡ Joseph Ndika, ‡ and Harri Alenius* ,‡,⊗ † The Netherlands Organization for Applied Scientific Research, TNO, P.O. Box 80015, Utrecht 3584 CB, The Netherlands ‡ Human Microbiome Research, Faculty of Medicine, University of Helsinki, P.O. Box 21, Helsinki 00290, Finland § Triskelion B.V., P.O. Box 844, Zeist 3704 HE, The Netherlands ∥ Faculty of Medicine and Life Sciences, University of Tampere, Tampere FI-33014, Finland ⊥ Institute of Biotechnology, University of Helsinki, P.O. Box 56, Helsinki 00014, Finland # Finnish Institute of Occupational Health, P.O. Box 40, Helsinki 00014, Finland ⊗ Institute of Environmental Medicine, Karolinska Institutet, P.O. Box 210, Stockholm SE-17176, Sweden * SSupporting Information ABSTRACT: More than 5% of any population suffers from asthma, and there are indications that these individuals are more sensitive to nanoparticle aerosols than the healthy population. We used an air−liquid interface model of inhalation exposure to investigate global transcriptomic responses in reconstituted three-dimensional airway epithelia of healthy and asthmatic subjects exposed to pristine (nCuO) and carboxylated (nCuOCOOH)copperoxide nanoparticle aerosols. A dose-dependent increase in cytotoxicity (highest in asthmatic donor cells) and proinflammatory signaling within 24 h confirmed the reliability and sensitivity of the system to detect acute inhalation toxicity. Gene expression changes between nanoparticleexposed versus air-exposed cells were investigated. Hierarchical clustering based on the expression profiles of all differentially expressed genes (DEGs), cell-death-associated DEGs (567 genes), or a subset of 48 highly overlapping DEGs categorized all samples according to “exposure severity”, wherein nanoparticle surface chemistry and asthma are incorporated into the dose−response axis. For example, asthmatics exposed to low and medium dose nCuO clustered with healthy donor cells exposed to medium and high dose nCuO, respectively. Of note, a set of genes with high relevance to mucociliary clearance were observed to distinctly differentiate asthmatic and healthy donor cells. These genes also responded differently to nCuO and nCuOCOOH nanoparticles. Additionally, because response to transition-metal nanoparticles was a highly enriched Gene Ontology term (FDR 8 ×10−13) from the subset of 48 highly overlapping DEGs, these genes may represent biomarkers to a potentially large variety of metal/metal oxide nanoparticles. KEYWORDS: copper(II) oxide, nanoparticles, asthma, 3d human bronchial epithelial cells, air−liquid interface and transcriptomics Asthma is a chronic respiratory disease with extremely high susceptibility to environmental exposures such as allergens, 1 chemicals, 2 and particulate air pollutants. 3−5 Asthmatic airways are typically chronically inflamed and extremely hyper-reactive, with symptoms such as recurrent wheezing, coughing, and shortness of breath. Asthma prevalence has steadily increased over the last three decades and is a major public health burden. The most recent survey estimates that as much as 4.4% of the global population (334 million people) is affected by asthma (Global Asthma Received: March 7, 2019 Accepted: June 12, 2019 Published: June 12, 2019 Article www.acsnano.org Cite This: ACS Nano 2019, 13, 6932−6946 © 2019 American Chemical Society 6932 DOI: 10.1021/acsnano.9b01823 ACS Nano 2019, 13, 6932−6946 This is an open access article published under a Creative Commons Attribution (CC-BY) License, which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited. Downloaded via TAMPERE UNIV on September 25, 2019 at 09:46:40 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. Network, 2014). There is thus a clear need to prioritize development of comprehensive risk assessment frameworks and tools that take this highly susceptible population into account during implementation of regulatory and monitoring programs to curb or follow up on potentially hazardous exposures. The varied and configurable physical and chemical properties of ENM amplify their industrial and medical applicability. Unfortunately, the toxicity of ENM can be mediated by physical and chemical properties; including, but not limited to size, shape, surface charge, aspect ratio, functionalization, etc. 6,7 As a result, the number of toxicologically relevant nanoforms scales dramatically. With regards to health hazard assessment, this poses a tremendous burden (costs and ethical considerations) on the extent of animal testing that will be required. Efficient, accurate and scalable ex vivo test methods which are still applicable to human exposures and can be used to evaluate the potential health hazards associated with ENM in a timely manner, are needed. Three-dimensional cocultures for nanoparticle exposure at an air−liquid interface that mimics the human lung have recently been developed. 8,9 Using this system in combination with adverse outcome assays, in vitro simulation of particle exposure and potential health hazard has been successfully performed for airborne particles and fibers. 10−14 Metal oxides are one of the most abundantly produced types of engineered nanomaterials (ENM) with production volumes of up to thousands of tons every year. The electrical, optical, and magnetic features of copper oxide (CuO) makes them appealing for a variety of industrial and commercial applications such as electronic chips, solar cells, lithium batteries, paints, processed wood, and plastics. CuO nanomaterials have already been used or could be utilized in food packaging, 15 wound dressings, 16 skin products, and hospital textiles. 17 Production volumes of CuO nanoparticles are expected to reach 1600 tons by the year 2025. 18 Therefore, because CuO has a very high potential for both occupational and consumer exposure, we have used it as a model to investigate potentially enhanced nanoparticle sensitivity within the context of pre-existing asthma. Unraveling the mechanistic interplay between nanoscale materials and asthma has been thus far limited to a handful of studies. 19 As such, employing an in vitro 3D human bronchial epithelial model in tandem with extensive downstream transcriptomic assessment in healthy and vulnerable individuals with a disease-compromised respiratory system is the subject of this study. 3D human bronchial epithelial cells cultured at an air−liquid interface that mimics relevant inhalatory exposure 20 were exposed to aerosols of pristine (nCuO) and carboxylated (nCuOCOOH) copper oxide nanoparticles. We hypothesized that coupling this exposure setup with global transcriptomic assessment will enable identification of altered defense mechanisms and/or enhanced particle sensitivity as a result of pre-existing asthma. In addition, because these primary cells are derived from nasal/ bronchial biopsies of donors, mode-of-action based approaches can inform on biomarker candidates that can be developed and Figure 1. Experimental setup with exposure, nanoparticle, and donor cell characterization. (A) Schematic of aerosolization, dilution, exposures, and implemented downstream bioassays. (B) Scanning electron microscope view of pristine (nCuO, upper panel) and COOHfunctionalized (nCuOCOOH, lower panel) copper oxide nanoparticles on a filter membrane. The filter pore sizes were 0.4 μm (nCuO) and 0.8 μm (nCuOCOOH). Twenty-four hours after the 1 h exposures at four different doses, particle deposition could be visually observed in all nonzero doses. Representative images of an air (zero) versus high dose in healthy and asthmatic donor cells are shown in (C). Cell layers of donor cells are more homogeneous compared to asthmatic cells. Changes in cell cytotoxicity were assessed by measuring release of lactate dehydrogenase (LDH) into culture medium (D). Twenty-four hours after exposures, a dose-dependent increase in relative LDH release, consistent with increasing cytotoxicity, was observed for both nanomaterials and cell types (D). The y-axis represents percent cellular cytotoxicity, with respect to their corresponding air-exposed controls. Maximum cell death observed across all exposures was around 25%. Cytotoxicity was highest in asthmatic donor cells, and pristine nCuO appears to be more cytotoxic than COOH-functionalized nCuO (nCuOCOOH). Bars are the mean with SD of three (healthy) or five (asthmatic) biological replicates. Each value was derived from the mean of three technical replicates. Statistical significance was inferred via two-way ANOVA and *denotes p-value < 0.05. ACS Nano Article DOI: 10.1021/acsnano.9b01823 ACS Nano 2019, 13, 6932−6946 6933 investigated via noninvasive sampling in “high-exposure-risk” and “high-susceptibility”subjects. RESULTS AND DISCUSSION Experimental Setup and Particle Dose Characterization. The experimental setup is depicted in Figure 1A. Cells were exposed to nanoparticle aerosols for 1 h, and all assay samples were collected after a 24 h incubation period. In a single-exposure experiment, air control, low-, mid-, and highdose groups are exposed simultaneously using a Vitrocell exposure system. Each Vitrocell consists of three slots (inserts); thus, every time a test block is exposed, the cell material within its three inserts originates from a single donor only. Previous work has shown that when using this approach, for any parameter, differences in the average for the donors are not affected by differences among sessions, test blocks, or concentrations. 20 Similarly, the differences in averages of the four CuO concentrations are not affected by interdonor variation. Aerosolization of nCuO resulted in agglomeration or aggregation of the particles with a median mass aerodynamic diameter (MMAD) of 1.8 μm (geometric SD = 1.57) and a MMAD of 1.4 μm (geometric SD = 1.48) for nCuOCOOH, measured using an aerodynamic particle sizer (APS) in the high dose (buffer chamber). Agglomerates or aggregates are likely to be the predominant form of the particle that interacted with all types of cells in our experiments. Such agglomeration or aggregation has been reported before. 21,22 Particle concentration was calculated using an APS and a scanning mobility particle sizer (SMPS) simultaneously. APS/ SMPS analyses showed particle concentrations of 6.15 ×105 and 1.65 ×106particles/cm3for nCuO and nCuOCOOH, respectively. Scanning electron microscopy (SEM) images are shown in Figure 1B. The actual exposure concentrations for the low, mid, and high groups were deduced from parallel exposures to be 23, 120, and 470 mg/m3, respectively, for nCuO and 32, 128, and 495 mg/m3for nCuOCOOH. The deposited dose was determined for the low and mid groups to be 14% (nCuO) and 15% (nCuOCOOH) of the actual exposure concentrations. Because particle deposition at constant flow rates is governed by particle diameter, 23 we can directly infer that similar deposition rates (14−15%) will be observed in the high group exposures. Aerosol deposition mechanisms in the lungs, especially with relevance to drug delivery, have been well studied. The deposition of an inhaled particle is dependent on its size. Keeping in mind that the aerosolized nanoparticles consisted of agglomerates with a mean diameter of 1.8 μm (nCuOCOOH nanoparticles) and 1.4 μm (nCuO nanoparticles), the observed 14−15% deposition rate is in line with modeled and experimentally determined 10 to 20% human bronchial airway deposition of unit density particles with a diameter of 1−2μm. 24,25 Real-Life Human Exposure Extrapolation. Extrapolating in vitro doses to human lungs is often problematic because lung morphology, air flow patterns and mucociliary particle clearance are all subject to inter individual and physiological state variability. 26 Nonetheless, we have combined previous calculations from similar exposure scenarios to derive estimates of the equivalent human exposures for the low, mid and high doses used herein. Regulatory occupational exposure limits do not exist for CuO and its derived nanoparticles. Going by the permissible exposure limit (PEL) (5 mg/m3) for respirable dust of particles not otherwise regulated as defined by the US Occupational Safety & Health Administration (2012), the low, mid and high dose exposures corresponded to approximately 1, 6, and 22 8 h work days of constant human conducting airways exposure. These real-life exposures were extrapolated as follows: first we averaged the two concentrations corresponding to low (23 and 32 mg/m3), mid (120 and 128 mg/m3), or high dose (470 and 495 mg/m3) for nCuO and nCuOCOOH particles. Having used the same Vitrocell air−liquid interface exposure system, we next derived mass/surface area exposure concentrations from 27.5 mg/m3(low), 124 mg/m3(mid), and 482.5 mg/m3 (high) concentrations, as described for exposure of CeO2 nanoparticles. 10 Following these calculations (summarized in materials and methods section), at a flow rate of 1.5 mL/min and 15% deposition, donor cells in each Vitrocell insert were exposed to approximately 1.2 μg/cm2(low), 5.6 μg/cm2 (mid), and 21.7 μg/cm2(high) nanoparticles. The average deposited dose normalized by the regional surface area is the default dose metric for respiratory effects of inhaled poorly soluble particles. 27 As such, we extrapolated the mass/area doses of the air−liquid interface to human bronchial epithelium. Incidentally, for particles with average aerodynamic diameters of 1−2μm 24 (note that the MMAD of aggregates/ agglomerates of nCuO and nCuOCOOH nanoparticles in the current exposures are 1.8 and 1.4 μm, respectively), five different deposition models have assigned a bronchial deposition rate of around 10%. Human bronchial airway surface area of 2709−4767 cm2for a healthy adult have been reported. 27,28 To avoid being too conservative in our human equivalent dose estimates, we will use the 4767 cm2bronchial surface area to derive region-specific doses. Finally, in order to relate the human equivalent dose to a real-life scenario, we based our extrapolations on the 5 mg/m3PEL of occupational exposure (NIOSH, 2012). Shvedova and colleagues 29 estimated the workplace nanoparticle human lung burden per day as × × =×× × ×× = (PEL) (ventilation during 8 h working day) (deposition rate) (5 mg/m ) (20 L/min 0.001 m /L 60 min/h 8 h/day)m /day (0.10) 4.8 mg/working day 33 3 = ==μ human equivalent dose lung burden target region surface area 4.8 mg 4767 cm 1.006 g/cm 2 2 =− human equivalent occupationally relevant exposure (workdays) air liquid interface dose/area human equivalent dose This implies, the low (1.2 μg/cm2), mid (5.6 μg/cm2) and high (21.7 μg/cm2) doses used in our study can be extrapolated to 1.2 ÷1.006 = 1.2 days, 5.9 ÷1.006 = 5.8 days and 21.7 ÷1.006 = 21.6 days of constant exposure of the bronchial airway to CuO and CuOCOOH nanoparticle aerosols in an occupational setting. Therefore, all three doses employed in the described air−liquid interface exposures can be ACS Nano Article DOI: 10.1021/acsnano.9b01823 ACS Nano 2019, 13, 6932−6946 6934 described as human-relevant, from a realistic exposure perspective. Visual Microscopic Inspection and Cellular Cytotoxicity Assessment. Microscopic examinations prior to nanoparticle exposure of the epithelia reveal vacuoles formed by enlarged cells. Such structures could also be observed in the epithelia constituted with cells from healthy donors, even though to a lesser extent (Figure 1C, air exposure). These vacuoles do not compromise the barrier function of the epithelia, since the regularly measured transepithelial electrical resistance of the epithelia derived from asthmatic donor cells was within the normal range (>100 Ω*cm2). In addition, when these asthmatic 3D cultures are established, culture media is absent from the apical surface of the epithelia, further confirming the structural integrity of the epithelial surface (information obtained via personal correspondence with Epithelix). Nanoparticle exposure of cells resulted in visual observable particle deposition at all doses, with the cells of asthmatic origin showing a less homogeneous cell layer compared to the cells from healthy donors (Figure 1C, high dose exposure). In both the healthy and asthmatic cells, we also observed cilia beating before and immediately after the exposures. Twenty-four hours after the 1 h exposures, most of the nanoparticle agglomerates/aggregates were located on the outer side of the insert, possibly moved there by beating cilia. LDH measurements showed that the cellular cytotoxicity was less than 25% in all exposures (Figure 1D). Compared to unexposed cells, a significant (p-value < 0.05) increase in cytotoxicity was observed in both healthy (high dose only) and asthmatic cells (low, mid, and high dose), with the highest levels of cytotoxicity detected in asthmatics. With impaired tracheobronchial mucociliary clearance observed even in the airways of nonsymptomatic asthmatic subjects, 30 a higher nanoparticle-induced cellular cytotoxicity in asthmatic donor cells confirms the reliability of the model to distinguish asthmatic and healthy bronchial airways. Inefficient particle clearance prolongs epithelium−nanoparticle interaction, which may lead to exacerbated particle-induced cytotoxicity from dissolved Cu2+ ions or enhanced particle uptake. NCuO induced greater cytotoxicity when compared to nCuOCOOH in both healthy and asthmatic donor cells. Studies have shown that cell-binding events and internalization of nanoparticles are largely mediated by their surface chemistries (reviewed by Mu et al. 31 ). Phospholipids, containing negatively charged phosphate groups, are the main components of pulmonary surfactant and cell membranes. The relatively lower apoptotic potential of the nCuOCOOH particles may be due (in part) to the fact that in an aqueous environment such as the surface of the respiratory epithelium deprotonation of the COOH functional group yields anionic nCuOCOO‑nanoparticles with poor membrane binding efficiency (hence uptake). This line of reasoning is consistent with studies showing enhanced Figure 2. Cytokine profiling after nanoparticle exposures. Release of selected pro-inflammatory cytokines, previously associated with exposure to CuO nanoparticles, including their agglomerates/aggregates, was carried out after exposures (A). Twenty-four hours after exposures, a dose-dependent release of IL-6 and IL-8 was measured from cell culture medium. No clear dose-dependent pattern was observed for MCP-1 release. Overall, the expression trend across doses for each nanoparticle was similar between protein (released cytokines) and mRNA measured via microarray-based gene expression profiling (B). Statistical significance was inferred using two-way ANOVA. Transcript and protein levels of IL-6 and IL-8 progressively increase with dose, with the most significant differences observed between air-exposed and high dose-exposed donor cells in both asthmatic and healthy donor cells. A significant difference between asthmatic and healthy donor cells was only observed for IL-6 and IL-8 mRNA levels in cells exposed to the highest concentrations of nCuO. Degree of significance is represented by an *, where *,**,***, and **** indicate pvalues <0.05, <0.01, <0.001, and <0.0001, respectively. ACS Nano Article DOI: 10.1021/acsnano.9b01823 ACS Nano 2019, 13, 6932−6946 6935 cytotoxicity of positively charged metal and metalloid nanoparticles relative to their neutral counterparts. 32 Selected Pro-inflammatory Cytokines Are Elevated in a Dose-Dependent Manner. Being more than just a barrier, the airway epithelium synthesizes and releases potent immunomodulators like chemokines, cytokines, growth factors, and antimicrobial peptides in response to an external stimulant. These responses typify an intact defense response characteristic of human airway epithelia. 33 Depending on the dose and type of material, ENM exposures typically trigger inflammation, marked by elevated levels of pro-inflammatory cytokines. Twenty-four hours after the nanoparticle exposures, we investigated changes in IL-6, IL-8, and MCP-1 since induction of these cytokines has been observed in bronchial epithelial cell cultures exposed to CuO and other metal oxide nanoparticles such as CeO2, TiO2, and ZnO. 34−38 IL-8 was the most elevated of all three cytokines. Increased levels of IL-6 and IL-8 were measured in three healthy and five asthmatic donor cells (Suppl. Figure 1A). Release of IL-6 and IL-8 was dose dependent, being most significant (p-value < 0.0001) in donor cells exposed to the highest dose of nCuO nanoparticles. MCP-1 was found to be elevated (only in response to pristine nCuO nanoparticles) in two out of three and two out of five healthy and asthmatic donor cells, respectively (Suppl. Figure 1B). IL-6 is a pleiotropic cytokine with a wide range of biological activities in immune regulation, hematopoiesis, inflammation, and oncogenesis, 39 while IL-8 and MCP-1 are neutrophiland monocyte/basophil-attracting chemokines, respectively. Given that the maximum level of MCP-1 measured across all donor cells was 4to 75-fold less than that of IL-6 and IL-8 respectively, assay detection limits coupled to technical variability might explain the observed inconsistency in MCP-1 induction. No significant difference in IL-6 and IL-8 secretion was observed between healthy and asthmatic donor cells (Figure 2A), even after normalizing cytokine release to percentage cell viability (data not shown). The expression trend of all three cytokines, as measured by microarray analysis was similar to secreted protein abundance (Figure 2B). Although macrophages are very often emphasized to be the first line of defense in pulmonary ENM-induced responses, 40 release of IL-8 suggests that in the bronchial areas of the lungs, epithelial cells are the responsible cell type that play a role in triggering innate immunity responses and neutrophil influx into the lung tissue to uptake/internalize foreign particles. This is in line with the well-studied role of the respiratory epithelium in cytokine-mediated innate defense (reviewed in Whitsett and Alenghat 41 ). In terms of the nanomaterial type, nCuO appears to be more bioreactive than nCuOCOOH, as the highest levels of all three pro-inflammatory cytokines were observed after nCuO exposure (Figure 2A,B). Figure 3. Transcriptomic profiling of asthmatics versus healthy when exposed to air, nCuO, or nCuOCOOH. The gene expression of asthmatics was compared to that of healthy donor cells after exposure to control air and three doses of nanoparticle-derived aerosols. (A) The most upregulated genes in asthma air/healthy air consists of genes that represent a highly significant (FDR 1 ×10−22) functional enrichment of extracellular matrix organization. (B) Venn comparisons of differentially expressed genes (DEGs) identified in asthma/healthy exposed to air to the combined DEGs from asthma/healthy exposed to low-, mid-, and high-dose nCuO and low-, mid-, and high-dose nCuOCOOH reveal there is very little overlap across identified DEGs. This suggests there is a strong interaction between nanoparticle exposure and asthmatic phenotype. (C) K-means clustering of only the 164 DEGs between asthma and healthy exposed to air, across all exposures, separates airexposed asthmatics and air-exposed healthy cells to the most distant clusters. Clusters of healthy donor cells exposed to control air are colored brown, and those for asthmatic donor cells exposed to control air are colored pink. ACS Nano Article DOI: 10.1021/acsnano.9b01823 ACS Nano 2019, 13, 6932−6946 6936 Transcriptomic Profiling of Control and Asthmatic Donor Bronchial Epithelia Identifies Known AsthmaRelated Genes. Previous studies have reported similarities between the mRNA expression profiles of epithelial cells cultured at the air−liquid interface and that of tracheal and bronchial brushings from human airways. 42,43 To validate the disease model in this study, we examined whether the relative expression of known asthma-associated genes, for the corresponding tissue type (bronchial epithelium), was consistent with the published literature. We identified genes related to asthma by microarray-based comparative transcriptomics on total RNA isolated from asthmatic and healthy donor cells that had only been exposed to control air. In addition, potential disease-modulating effects resulting from particle exposure were also investigated by analysis of differentially expressed genes (DEGs) in healthy versus asthmatic bronchial epithelial donor cells exposed at the air− liquid interface to CuO and CuOCOOH nanoparticle-derived aerosols. The ENM doses were selected such that cytotoxicity was less than 30% even with the highest dose. By this approach, gene expression profiling identifies early transcriptomic responses (biomarkers) that reflect the bioreactivity of CuO nanoparticles and not just genes that are related to general cell death. Pathway analysis of genes identified as upregulated genes in asthmatic donor cells revealed a highly significant (FDR 1 × 10−22) enrichment of genes that are functionally involved in extracellular matrix organization (Figure 3A). The most upregulated of which are collagen (COL1AI, COL4A1, COL4A2, COL5A2 and COL7A1), TNC (tenascin C, a fibronectin binding protein), cadherin (CDH2, CDH4 and CDH11), MRC2 (mannose receptor C type 2), and MMP13 (Matrix Metallopeptidase13). The relative expression of these genes is in line with the increased expression of extracellular matrix proteins (fibronectin, MMP9, and MMP12) observed within the airway smooth muscle of asthma patients when compared to nonasthmatic controls. 44 Araujo and colleagues 44 also observed that deposition of type-I and type-III collagens correlated with the clinical severity of asthma. Asthma heterogeneity is widely acknowledged, with allergic asthma, mediated by allergen-specific T helper type 2 (TH2) cells, the most common and most studied form of asthma. 45 Being that its hallmark mechanistic feature is the synthesis and release of the TH2 cytokines, IL-4, IL-5, and IL-13, 46 it is no surprise that their expression levels were close to background in the studied airway epithelial cell models. However, we did detect significantly elevated expression of IL-33 and TSLP (known inducers of TH2-type proinflammatory mediators) in asthmatic donor cells (Suppl. Figure 2A). Similarly, elevated levels of IL33 or TSLP, coupled with a concomitant correlation to disease severity, have been observed in bronchial biopsies and cultured bronchial epithelial cells derived from asthmatic patients. 47−49 In terms of morphological changes, mucus accumulation (implicated in airway obstruction) is a prominent feature of asthma. Mucus hypersecretion, marked by an increase in the number of goblet cells and upregulation of mucin genes (notably MUC5AC and MUC5AB), as well as altered mucus clearance due to a decrease in the number of ciliated cells and/ or cilia viability, are thought to both contribute to mucus accumulation in asthmatic airways. Incidentally, we identified modest but significant decrease in expression of FOXJ1 (marker of ciliated epithelial cells) and an increase in both MUC5AC (marker of goblet cells) and MUC5B in asthmatic donor cells (Suppl. Figure 2B). Several other mucin and mucin-like genes, MUC1, MUCL1, MUC2, MUC4, and MUC7, were also identified as significantly upregulated in MucilAir cells from asthmatic donors (Suppl. Figure 2C). A hierarchical cluster consisting of these upregulated mucin genes clearly separates asthmatic from control donor cells (Suppl. Figure 2D). To answer whether particle exposure modulates the asthmatic genotype, we compared gene expression changes between asthmatic and healthy donor cells exposed to air or nanoparticle-derived aerosols. A Venn distribution of the topmost DEGs (fold change cutoff≥1.5-fold and a Benjamini−Hochberg FDR of at most 5%) between healthy and asthmatic donor cells exposed to Air, nCuO, or nCuOCOOH is shown in Figure 3B. K-means clustering of all exposure groups, based only on the genes that were differentially expressed between asthmatic and healthy donor cells exposed to the control air stream, is depicted in Figure 3C. Within this dendrogram, the healthy and asthmatic groups occupy clusters that are farthest from each other. Conversely, if in addition to the genes that were different between asthmatics and healthy exposed to air those genes differing between asthmatic and healthy donor cells exposed to either nCuO or nCuOCOOH nanoparticle aerosols are included prior to hierarchical cluster generation, partitioning of the resulting cluster dendrogram then becomes based on exposure dose, with the asthmatic air and healthy air groups occupying closely related clusters (Suppl. Figure 3A,B). In summary, when comparing air-exposed healthy and asthmatics, the main difference between all samples is driven by disease-related genes, and when the exposure to nanomaterials is considered, the main difference between samples is driven by the exposure dose. The top pathways enriched by the genes that are differentially expressed between asthmatics and healthy, when unexposed (air) or exposed to either nCuO or nCuOCOOH, are shown in Suppl. Figure 3C. An alteration from asthma-relevant pathways (extracellular matrix organization, FDR 1 ×10−22; collagen metabolic process, FDR 1 ×10−17; extracellular matrix part, FDR 3 ×10−16) to pathways associated with adverse effects of metal oxide nanomaterials (DNA damage response, FDR 1 ×10−6to 1 ×10−2; response to metal ion, FDR 6 ×10−14 to 1 ×10−8; response to inorganic substance, FDR 2 ×10−7) is observed. Taken together, this indicates that in response to nanoparticle exposure, the effects of the disease (i.e., asthma) are overshadowed by the adverse exposure outcomes at the doses tested. From a mechanistic viewpoint, a notable limitation of the current study is that Mucilair asthmatic donor cells are obtained based only on symptomatic, as such given the limited number of asthmatic donors (N= 5), it is impossible to investigate particle exposure susceptibilities of different asthma subtypes. Expression Profile of DEGs Incorporate Nanomaterial Dose, Functionalization and Asthma into the Dose− Response Axis. To characterize the gene expression profiles that represent adverse exposure to both nCuO and nCuOCOOH as a function of dose, functional group, and disease state, a total of 12 contrast sets for identification of DEGs were specified. The number of DEGs identified in each of the specified contrast sets (A−L) is shown in Suppl. Figure 4A.In total, 6523 DEGs were observed across all comparisons. The number of DEGs correlated with the dose (progressive increase in the number of DEGs from low dose to high dose). No obvious trend could be identified from the number ACS Nano Article DOI: 10.1021/acsnano.9b01823 ACS Nano 2019, 13, 6932−6946 6937 of DEGs with respect to material type or disease state. However, the total number of DEGs from contrasts reflecting exposure to nanoparticles irrespective of disease state, or disease state exposures irrespective of the particle type, was similar for healthy and asthmatics, while an excess of 1627 DEGs was observed in nCuO relative to the nCuOCOOH exposures (Suppl. Figure 4B). This supports our previously mentioned observation that following exposure to CuO nanomaterials the difference between healthy and asthmatic cells is mainly driven by genes that are modulated in response Figure 4. Global differential expressed genes and genes related to regulation of cell death differentiate exposures according to dose, material type, and disease state. In total, 6523 differentially expressed genes (DEGs) were identified when exposed healthy [H] and asthmatic [A] cells were compared to their corresponding air-exposure controls. These genes were then obtained from the normalized expression matrix and used for principal component analysis (PCA). The top two components, explaining about 65% of the variance between exposures, are shown in (A). Colored ovals with dashed lines depict the different doses from air (zero) to high dose (A, left to right). Circular symbols represent healthy donors and triangles represent asthmatic donors. It can be seen from PCA that asthmatic donor cells exposed to the relatively lower dose are overrepresented in the regions where the low, mid and high doses overlap. This indicates interaction between disease and nanoparticle exposure and is further highlighted by the fact that all of the genes that are differentially expressed between baseline asthmatics and healthy donor cells were also differentially expressed in response to nanoparticle exposure (B). Approximately 9% (567 genes) of the DEGs between exposed and unexposed donor cells were identified by pathway enrichment analysis to represent GO biological processes corresponding to positive and negative regulation of cell death. Venn distribution of these nanoparticle/air DEGs related to cell death is shown in (C). K-means clustering, exclusively based on these cell death related genes, differentiates and groups the samples according to dose, surface chemistry, and disease state (D). The primary distribution of samples across the various branches of the dendrogram can be attributed to the nanoparticle deposited dose, wherein all high-dose exposures were grouped together in branch 1, airexposed controls in branch 2A, and mid dose in branch 2B (D). All low-dose exposures are closest to the air-exposed controls in branch 2A, except nCuO-low dose (asthmatics), which clustered with the mid doses (branch 2B), and all mid doses clustered in branch 2A except nCuO-mid (asthmatics), which clustered in branch 1 with the high doses. ACS Nano Article DOI: 10.1021/acsnano.9b01823 ACS Nano 2019, 13, 6932−6946 6938 to the nanomaterial exposure (Suppl. Figure 3C). On this basis, the main difference between asthmatic and healthy cells in response to nCuO/nCuOCOOH is more likely to be as a result of differences in sensitivity than being due to a different effect of these materials on asthmatic airways. However, a limitation of the current study protocol is that, in the absence of transcriptomic data over multiple time points, we cannot rule out the possibility that the enhanced sensitivity of the asthmatic epithelium derivesfromdelayed(andnot insufficient) self-regulation upon nanoparticle-induced irritation of the airways. In order to visualize possible trends in the data arising from the expression profile of genes that significantly change in response to CuO and CuOCOOH exposure, we performed a principal component analysis (PCA) that was based exclusively on the 6523 genes that were significantly differentially expressed between exposed cells and their corresponding air controls. The top two principal components, explaining about 65% of the variation, are shown in Figure 4A. The different dose groups (air, low, mid and high) in the PCA plot are highlighted with distinct colors and oval shapes (legend). Interestingly there are regions of overlap between adjacent doses (low/mid, mid/high) wherein asthmatic cells exposed to nCuO from the lower dose are overrepresented. For example, in the overlap between low and mid doses, low nCuOasthmatic cells cluster closer to the mid dose exposures than the other three [(1) nCuO-healthy, (2) nCuOCOOH-healthy, and (3) nCuOCOOH-asthmatic] low dose groups. The same holds true for the mid/high intersection. This is an indication that the presence of asthma enhances the sensitivity to the adverse effects of nCuO exposure, and this sensitivity is diminished by functionalization to a COOH group (nCuOCOOH). This observation is in line with cytotoxicity measurements in healthy and asthmatic cells after nCuO and nCuOCOOH exposures (Figure 1D). We also observed that the genes which are differentially expressed between asthmatic and healthy donor cells exposed to only to control air were differentially expressed between nanoparticle-exposed and airexposed healthy and asthmatic donor cells (Figure 4B), meaning that the genes which reflect the asthmatic genotype in this cohort are also involved in the response to nanoparticle exposure. This could explain why the asthma versus healthy gene expression profiles were so different for donor cells exposed to either air or aggregates/agglomerates of nCuO or nCuOCOOH nanoparticles (Figure 3A). It is possible that asthmatics are more susceptible to nanoparticle exposure because the expression of some of the genes required to mount a (possibly protective) response to nanoparticle exposure is skewed in asthmatics, resulting in a delayed onset of the mechanisms required to protect against the adverse effects caused by these particles. Expression of Genes Involved in Regulation of Cell Death Correlates with Cellular Cytotoxicity Assay. Because the dose−response differentiation was similar for both the cytotoxicity assay and expression profile of all DEGs, we next investigated whether this distinction is consistent for a subset of DEGs that represent changes in cell viability. To this end, we performed gene ontology (GO) based biological process enrichment analysis using all 6523 DEGs as input list. We used this global approach as opposed to biological process Figure 5. Scatter plot of genes related to organization of the cilia 186 differentially expressed genes were identified by pathway analysis to represent biological functions corresponding to cilia organization/cilia assembly (A). Average relative expression for each of these genes were obtained by Z-score normalization of their mRNA intensity values across all samples for the nCuO and nCuOCOOH exposures. Z-scores were then averaged for air, low, mid, and high nanoparticle doses. A scatter plot of air-exposed healthy versus asthmatic donor cells is shown in (B). As an example of a pre-existing distinction between healthy and asthmatic airways, MCIDAS and RAB3IP genes with potential relevance in the functionality of asthmatic airways are highlighted. Following exposure to nCuO (C, upper panel) or nCuOCOOH (C, lower panel) nanoparticles/nanoparticle agglomerates, a bigger change in expression (green lines) of cilia-related genes can be observed in healthy when compared to asthmatic donor cells. Furthermore, the low-dose response was distinct in all exposures, except in asthmatics exposed to nCuO (dashed red circles), wherein the low-dose response overlaps with the midand high-dose responses. ACS Nano Article DOI: 10.1021/acsnano.9b01823 ACS Nano 2019, 13, 6932−6946 6939 enrichment analysis for each of the contrast sets depicted in Suppl. Figure 4A because the expression profile of the combined DEGs correlated better with dose, functional group, and disease state than the number of DEGs for every contrast. The top five enriched nonredundant GO terms were regulation of cell death (GO:0010941), cilium organization Figure 6. Shared DEGs highlight core molecular signature of nanosized CuO exposure and integrates all four tested parameters into a dose− response gradient. Comparing nanoparticle-exposed healthy or asthmatic donor cells to their corresponding air-exposed controls resulted in a total of 12 contrast sets. Forty-eight differentially expressed genes were identified to be shared between 9 or more of these contrasts sets; i.e., 33 genes were shared across 9 contrasts, 12 genes in 10 contrasts, and 3 genes in 11 contrasts. A hierarchical cluster (upper panel) based on these reoccurring 48 differentially expressed genes separates all groups along a dose−response gradient, which takes into account the material surface chemistry and tissue health status. Response to transition metal nanoparticle, attributed to a subset of five metallothionein 1 genes, was identified as a highly enriched (FDR 9 ×10−13) biological process within these 48 shared DEGs. Connections between gene nodes represent physical, predicted, and genetic interactions as well as shared protein domains and pathways. ACS Nano Article DOI: 10.1021/acsnano.9b01823 ACS Nano 2019, 13, 6932−6946 6940