A 3D Human Airway Model Enables Prediction of Respiratory Toxicity of Inhaled Drugs In Vitro

A 3D Human Airway Model Enables Prediction of Respiratory Toxicity of Inhaled Drugs In Vitro Abstract Respiratory tract toxicity represents a significant cause of attrition of inhaled drug candidates targeting respiratory diseases. One of the key issues to allow early detection of respiratory toxicities is the lack of reliable and predictive in vitro systems. Here, the relevance and value of a physiologically relevant 3D human airway in vitro model (MucilAir) were explored by repeated administration of a set of compounds with (n = 8) or without (n = 7) respiratory toxicity following inhalation in vivo. Predictability for respiratory toxicity was evaluated by readout of cytotoxicity, barrier integrity, viability, morphology, ciliary beating frequency, mucociliary clearance and cytokine release. Interestingly, the data show that in vivo toxicity can be predicted in vitro by studying cell barrier integrity by transepithelial electrical resistance (TEER), and cell viability determined by the Resazurin method. Both read-outs had 88% sensitivity and 100% specificity, respectively, while the former was more accurate with receiver operating characteristic (ROC) AUC of 0.98 (p = .0018) compared with ROC AUC of 0.90 (p = .0092). The loss of cell barrier integrity could mainly, but not fully, be attributed to a loss of cell coverage in 6 out of 7 compounds with reduced TEER. Notably, these effects occurred only at 400 µM, at concentration levels significantly above primary target cell potency, suggesting that greater attention to high local lung concentrations should be taken into account in safety assessment of inhaled drugs. Thus, prediction of respiratory toxicity in 3D human airway in vitro models may result in improved animal welfare and reduced attrition in inhaled drug discovery projects. respiratory toxicology, inhalation toxicology, predictive toxicology, airway epithelium, 3D in vitro model Inhalation is the key route for delivery of therapeutic agents to treat respiratory diseases such as asthma and chronic obstructive pulmonary disease. Inhaled corticosteroids, β2-agonists and antimuscarinics represent standard care in respiratory diseases (Chauhan and Ducharme, 2014) and a number of additional therapeutic opportunities are being explored (Wain et al., 2017). The utility of the inhaled route for drug delivery in respiratory diseases presents a great advantage with topical access to the diseased tissue and a low systemic exposure with associated reduction in the potential for systemic side effects. However, unacceptable toxicity in the respiratory tract represents a major obstacle in the development of inhaled therapies, and has been reported to account for approximately 30% of inhaled project closures (Cook et al., 2014). Respiratory toxicities have typically been identified at a relatively late stage in pre-clinical testing as part of the comprehensive assessment undertaken during in vivo toxicity studies with the aim to identify therapeutic margins, maximal tolerable concentrations and reversibility of any noted adverse effects (Ahuja et al., 2017; Hayes and Bakand, 2014). To this point, improved early assessment and prediction of the toxicity potential of new molecules would add significant value in terms of the potential to reduce late stage compound attrition, increase quality of drug candidates, and improve animal welfare by replacement, reduction or refinement of animal usage in preclinical toxicity testing of novel inhaled drugs. One of the key limitations for early detection of respiratory toxicities so far has been around development of reliable and predictive in vitro models. The airway epithelium represents a barrier with a role in protection of the airways and is one of several key target sites for toxicity in the respiratory tract. Therefore, a major focus has been on developing in vitro models representing this region of the lung. Primary airway epithelial cells cultured at air-liquid interphase represent physiologically relevant models of the human airway consisting of ciliated cells, mucus secreting goblet cells and basal cells (Balharry et al., 2008; Huang et al., 2013), and are now commercialized under trade names such as MucilAir and EpiAirway by Epithelix Sàrl (Geneva, Switzerland) and MatTek Corporation (Ashland, MA, US), respectively. So far, toxicity evaluations in such 3D human airway in vitro models have focused on lung irritancy of environmental toxic agents such as smoke, ozone, formaldehyde, CdCl2, and nanoparticles (Balharry et al., 2008; Cao et al., 2017; Huang et al., 2013; Kuper et al., 2015; Rothen-Rutishauser et al., 2008). Here, we, to the best of our knowledge, for the first time explore the relevance, value and potential application of the MucilAirmodel to predict in vivo respiratory toxicity in the context of inhaled drug candidates for treatment of respiratory diseases. We focused our in vitro investigation in the MucilAir model around a set of 14 inhaled drugs or drug candidates and CdCl2, a known toxic agent, as a positive control. These were divided into compounds with (n = 8) or without (n = 7) respiratory toxicity based on historical data from limited dose-range and dose duration in vivo studies (Table 1). The compounds with toxicities had previously shown to induce a range of different adverse responses in the respiratory tract, mostly observed as histopathological changes in the lung in animal toxicology studies (Table 1). In addition, pentamidine, an approved inhaled drug associated with respiratory adverse events, (TAPS group, 1990, Katzman et al., 1992) and salmeterol in the free base form (Owen et al., 2010) were included as positive controls. A range of different parameters such as cytotoxicity, cell barrier integrity, viability, morphology, ciliary beating frequency (CBF), mucociliary clearance (MCC), and cytokine release were analyzed for predictive accuracy for respiratory toxicity. With this approach, we demonstrate that a 3D human airway in vitro model is capable of predicting the potential for drug-induced adverse effects in the respiratory tract, by monitoring the loss of barrier integrity and cell viability as measured by transepithelial electrical resistance (TEER) and Resazurin, respectively. In conclusion, the 3D human airway in vitro model has a clear potential to reduce compound attrition due to respiratory toxicity, and insight from this work could help to guide the development of future inhaled therapies for respiratory diseases. Table 1. Summary of Toxicity Findings of Studied Drugs Compound  Target  Tox  Toxicity Finding  AZ1  IKK2  Yes  Increased lung weight, inflammatory/epithelial pathology in lung in rat and dog studies.a  AZ2  IKK2  Yes  Increased lung weight, inflammatory/epithelial pathology in lung in rat and dog studies.  AZ3  PDE4  Yes  Acute lung inflammation and nasal inflammation/ulceration in dog (but not in rat)  AZ4  PDE4  Yes  Acute lung inflammation and nasal inflammation/ulceration in dog (not tested in rat).a  D20207  PKC  Yes  Inflammatory/epithelial pathology in rat lung/airways, cough in dogs.a  Pentamidine  Anti-infective  Yes  Marketed product—associated with high incidence of cough, difficulty breathing and burning sensation in throat (TAPS group, 1990; Katzman et al., 1992).  Salmeterol Base  LABA  Yes  Development of this form of salmeterol was discontinued due to respiratory tract irritancy in early rat studies (Owen et al., 2010).  CdCl2    Yes  Acute interstitial pneumonitis and pulmonary edema, leading to interstitial fibrosis, and emphysema.  AZD9819  NEI  No  No adverse effects in respiratory tract in 3-month rat and dog studiesa  Formoterol  LABA  No  Marketed product  Budesonide  GCS  No  Marketed product  Tiotropium  LAMA  No  Marketed product  Cromoglycate  Mast cell stabilizer  No  Marketed product  Glycopyrronium  LAMA  No  Marketed product  AZD9164  LAMA  No  Slight increases in lung weights, no significant histopathology, transient functional effects in rat and dog, and Chronic Obstructive Pulmonary Disease patients (decreased Forced Expiratory Volume 1s).a  Compound  Target  Tox  Toxicity Finding  AZ1  IKK2  Yes  Increased lung weight, inflammatory/epithelial pathology in lung in rat and dog studies.a  AZ2  IKK2  Yes  Increased lung weight, inflammatory/epithelial pathology in lung in rat and dog studies.  AZ3  PDE4  Yes  Acute lung inflammation and nasal inflammation/ulceration in dog (but not in rat)  AZ4  PDE4  Yes  Acute lung inflammation and nasal inflammation/ulceration in dog (not tested in rat).a  D20207  PKC  Yes  Inflammatory/epithelial pathology in rat lung/airways, cough in dogs.a  Pentamidine  Anti-infective  Yes  Marketed product—associated with high incidence of cough, difficulty breathing and burning sensation in throat (TAPS group, 1990; Katzman et al., 1992).  Salmeterol Base  LABA  Yes  Development of this form of salmeterol was discontinued due to respiratory tract irritancy in early rat studies (Owen et al., 2010).  CdCl2    Yes  Acute interstitial pneumonitis and pulmonary edema, leading to interstitial fibrosis, and emphysema.  AZD9819  NEI  No  No adverse effects in respiratory tract in 3-month rat and dog studiesa  Formoterol  LABA  No  Marketed product  Budesonide  GCS  No  Marketed product  Tiotropium  LAMA  No  Marketed product  Cromoglycate  Mast cell stabilizer  No  Marketed product  Glycopyrronium  LAMA  No  Marketed product  AZD9164  LAMA  No  Slight increases in lung weights, no significant histopathology, transient functional effects in rat and dog, and Chronic Obstructive Pulmonary Disease patients (decreased Forced Expiratory Volume 1s).a  a Åberg et al., unpublished data. MATERIALS AND METHODS 3D human airway in vitro model Airway cells were obtained from patients undergoing surgical polypectomy. All experimental procedures were explained in full, and all subjects provided informed consent. The study was conducted according to the declaration of Helsinki on biomedical research (Hong Kong amendment, 1989), and received approval from the local ethics commission. MucilAir (Epithelix Sàrl, Geneva, Switzerland) airway epithelia, reconstituted with a mixture of human nasal cells isolated from a panel of 14 different donors, were cultured at the air liquid interface in MucilAir culture medium (Epithelix Sàrl, Geneva, Switzerland), in 24-well plates with 6.5-mm Transwell inserts (Corning). The cell culture medium was changed at each sampling point or every 2–3 days for those wells not being sampled. The nasal version of MucilAir were used as surrogate for bronchial epithelia due to availability, quality and close resemblance in gene expression levels, morphology and function (Iskandar et al., 2013; McDougall et al., 2008; Thavagnanam et al., 2014). Testing strategy All compounds were synthesized at AstraZeneca, except CdCl2, which was purchased from Sigma. The compounds were dissolved in DMSO and diluted in a buffered saline solution (0.9% NaCl, 1.25 mM CaCl2, 10 mM HEPES) to reduce the final concentration of DMSO to 0.8% (v/v). Test compounds (10 μl) were applied daily at 4, 40, and 400 µM on the apical surface of MucilAir cultures and left to incubate for 6 h prior to replacing with fresh medium. This was repeated for a period of 12 days. Each treatment was run as 4 replicates on each occasion. Culture supernatants were collected and stored at −80 °C for future analysis. The Transwell membranes containing the 3D airway cells were fixed for 30 min in 4% Formaldehyde-PBS (with Ca/Mg) at RT, washed in PBS and stored at 4 °C until processing for immunohistochemistry. Measurements TEER and CBF were assessed as previously described (Huang et al., 2017). Cell viability was measured by Resazurin test at termination and cytokines assessed at indicated time-points as previously described (Huang et al., 2013). Cytotoxicity was measured by quantification of lactate dehydrogenase (LDH) in supernatants by Cytotoxicity Detection KitPLUS (Roche) according to manufacturer’s instructions. MCC was assessed as previously described (Huang et al., 2017). In short, 30 µm polystyrene microbeads (Sigma,) were added on the apical side of the MucilAir cells. The movement of the microbeads was recorded with a Sony XCD-U100CR camera at 2 frames per second, on an Axiovert 200 M microscope (Zeiss), at room temperature. A total of 50 images were recorded and more than 1000 beads were tracked and the velocity of MCC (µm/s) was calculated using Image-Pro Plus software (version 6.2, Media Cybernetics). Three movies were taken and analyzed per insert. Immunohistochemistry and determination of cell coverage grade Membranes were permeabilised with 0.5% Triton-X100 in PBS buffer for 20 min, blocked with 2.5% BSA, 5% goat serum, 0.05% Tween-20 in PBS for 30 min. Cells was incubated with antiTubulin (NBP2-00812, NovusBio), and antiMuc5AC (MS145P1, Thermo Scientific) in the block buffer for 60 min, followed by Alexa 488 coupled mouse IgG2A (A21131, Invitrogen) and Alexa 568-Mouse IgG1 (A21124, Invitrogen) for 45 min in the dark. All steps were performed at room temperature. The membranes were mounted on glass slides in Vectashield antifade mounting medium containing DAPI stain (Vector Lab) and evaluated under fluorescent microscope Ax70 (Olympus). Images were captured by confocal microscope LSM880 (Zeiss) connected to ZEN 2.3 software (Zeiss) and illustrated by using Photoshop, version 11 (Adobe). The cell coverage grade was determined on blinded samples by visual estimate of the confluency area: 70%–100% (grade 4); 30%–70% (grade 3); 10%–30% (Grade 2); <10% (grade 1); loss of all cells (grade 0). Statistics The method detection limit (MDL) is defined as the minimum value of a substance that can be distinguishable from method blank with 99% confidence and were determined according the U.S. Environmental Protection Agency method (US EPA 2016). Quality of the assays was determined by the Zeta-factor (Z’) method (Zhang et al. 1999). All other statistical calculations were performed by GraphPad prism 7.0 software (GraphPad Software, San Diego, California). Cut-off values were determined by Youden's index (J), i.e. {J = sensitivity + specificity – 1} (Youden, 1950). RESULTS Measurement of Cell Barrier Integrity in 3D Human Airway Epithelial Cell Cultures The fully differentiated MucilAir 3D human airway cultures showed a nonsignificant difference in cell barrier integrity as measured by TEER between assay occasions as determined with 1-way ANOVA with a mean TEER value across studies of 440 (SD 70) Ω cm2 (n = 3). However, we noted an intra-assay time-point dependent shift in TEER values with an average intra-assay coefficient of variance (intra-assay CV) of 16.9%. Therefore, in order to compare compound effects between different assay occasions and assay time-points, the change in TEER value from vehicle was normalized against vehicle control on each day. By this approach, the MDL of the TEER read-out was 25.5% (n = 40) with an intra-assay CV of 9.4% (95% CI 6.7–12.1) and an average Z- prime of 0.75 per plate. In order to predict respiratory toxicity, a test-set of 15 compounds were evaluated in the MucilAir 3D human airway epithelial model, by repeat application of compounds to the apical side, for 6 h per day for 12 days, to represent a prolonged residence time and to mimic the inhaled route. The compounds were divided into 2 different categories based on historical data, compounds with (n = 8) and without (n = 7) demonstrated respiratory toxicity following inhalation (Table 1). Cell barrier integrity was measured for all compounds at 4, 40, and 400 µM and a reduction of TEER over MDL was evident in 7 out of 8 compounds with known toxicity at the highest tested dose of 400 µM after 12 repetitive doses (Figure 1A). Importantly, a reproducible and time-dependent loss of barrier integrity was found for all compounds with known respiratory toxicity in animals (n = 7), while all compounds without respiratory toxicity and pentamidine, a drug with lung irritancy side-effects in human, were below MDL (Figs. 1B and 1C and Supplementary Figure 1A). Figure 1. View largeDownload slide Impact of compounds with and without respiratory toxicity in a 3D human airway in vitro model. Percentage change of cell barrier integrity, as measured by TEER, compared to vehicle control after repeat exposure of compounds to the apical side of the epithelial cell layer. (A) Dose response at day 12 of all 8 compounds with respiratory toxicity (red) and representative compounds without toxicity (green, 3 out of 7 compounds). Dotted line represents the method detection limit (MDL) of significant change compared to vehicle treated cells. All other non-toxic compounds were above MDL. (B) Time course at 400 µM of the same compounds as shown in A. (C) Reduction in cell barrier integrity (represented as % reduction in TEER) at 400 µM on day 12 for compounds with (closed bars) or without (open bars) respiratory toxicity (mean SEM). Figure 1. View largeDownload slide Impact of compounds with and without respiratory toxicity in a 3D human airway in vitro model. Percentage change of cell barrier integrity, as measured by TEER, compared to vehicle control after repeat exposure of compounds to the apical side of the epithelial cell layer. (A) Dose response at day 12 of all 8 compounds with respiratory toxicity (red) and representative compounds without toxicity (green, 3 out of 7 compounds). Dotted line represents the method detection limit (MDL) of significant change compared to vehicle treated cells. All other non-toxic compounds were above MDL. (B) Time course at 400 µM of the same compounds as shown in A. (C) Reduction in cell barrier integrity (represented as % reduction in TEER) at 400 µM on day 12 for compounds with (closed bars) or without (open bars) respiratory toxicity (mean SEM). Next, the most accurate time-point to predict respiratory toxicity was determined by area under the receiver operating characteristic (ROC) curve analysis. A reduction in TEER from baseline significantly predicted toxicity after 1–3 and 8–12 consecutive days of drug treatment, respectively and most accurately at day 12 with ROC AUC of 0.98 (95% CI, 0.93–1.04, p = .0018) (Supplementary Figs. 2B and 4A). To be able to distinguish between toxic and nontoxic compounds, a cut-off value was calculated to provide optimum sensitivity at the highest possible specificity by Youden analysis. The sensitivity and specificity at a cut-off value of 32% decrease in the TEER signal were 88% and 100%, respectively at day 12. Measurement of Cell Viability Cell viability was detected by reduction of resazurin by measuring fluorescence after addition of the detection probe to the cell cultures at the termination of the experiment. The vehicle treated samples had a mean RFU of 12800 with inter-assay CV of 15% (n = 3) and a mean intra-assay CV of 14% (95% CI, 10.2–17.8) resulting in a MDL of 39% (n = 26). In drug treated samples, 5 out of 8 toxic drugs showed a reduction of cell viability beyond MDL, including salmeterol in the free base form with a 48% reduction compared with vehicle (Figure 2A). Cell viability measured at termination of the experiment showed a significant alignment towards TEER (Figure 2B) and predictive accuracy for respiratory toxicity with ROC AUC of 0.90 (95% CI, 0.71–1.09, p = .0092) (Figure 4A). Cell viability had similar sensitivity for predicting toxicity as TEER, with 88% sensitivity and 100% specificity, respectively at cut-off values of 14% reduction in viability compared with vehicle. These data imply that respiratory in vivo toxicity of inhaled drugs could be captured in vitro by assessing both TEER and cell viability. Figure 2. View largeDownload slide Impact of compounds with and without respiratory toxicity on cell viability in a 3D human airway in vitro model. (A) Percentage change of viability, compared to vehicle control measured by resazurin conversion after 12 days of repeat exposure of compounds with (closed bars) or without (open bars) respiratory toxicity (mean SEM). Dotted line represents the method detection limit (MDL) of significant change compared to vehicle treated cells. (B) Regression analysis of the correlation between TEER and viability. Figure 2. View largeDownload slide Impact of compounds with and without respiratory toxicity on cell viability in a 3D human airway in vitro model. (A) Percentage change of viability, compared to vehicle control measured by resazurin conversion after 12 days of repeat exposure of compounds with (closed bars) or without (open bars) respiratory toxicity (mean SEM). Dotted line represents the method detection limit (MDL) of significant change compared to vehicle treated cells. (B) Regression analysis of the correlation between TEER and viability. Correlation of Cell Coverage Against Barrier Integrity and Cell Viability First, the morphology of vehicle treated MucilAir cultures was assessed on whole mount preparations by immunohistochemistry against Tubulin B to detect ciliated cells and cytoskeleton, MUC5AC to detect mucin in goblet cells and DAPI as nuclear stain. The airway epithelium cultures were fully confluent and well differentiated as the majority of cells on the apical side were ciliated epithelial cells, and a minority of goblet cells (Supplementary Figure 2A). In the cross section, the cells were arranged in a pseudostratified pattern with some variations of the thickness of the epithelium. Thus, the 3D human airway epithelial cultures showed a physiological resemblance to a human ciliated epithelial layer in the airway. Next, a cell coverage grading system based on visual estimates of cell confluence area was developed to potentially link loss of TEER and cell viability to cell confluency. First, the epithelial damage was examined for AZ4 at 400 µM on treatment days 0, 1, 8, 12, and after recovery where cells were left untreated for 7 days. Cell coverage decreased with time of drug exposure, correlating with a decrease in TEER signal (Figure 3B). This was visible by a progressive loss of cells, starting at the peripheral area at early time points and a total cell loss at later time points (Figure 3A). Second, the epithelial morphology, TEER signal and cell viability measured by the Resazurin method were compared between all 15 compounds at all tested concentrations at the end of the experiment. Loss of cell coverage was evident in 6 out of 7 compounds with reduced TEER and correlated to both TEER and cell viability (Figure 3C and Supplementary Figure 2B). The only exception was salmeterol in the free base form, which showed a reduction in TEER and cell viability, while cell coverage remained in the range observed for vehicle treatment. Figure 3. View largeDownload slide Morphological assessment of whole mount preparations of the MucilAir epithelium and cell coverage correlation with TEER and cell viability. A, Representative cell images over time for AZ4 (400 µM) at indicated time points as determined by confocal microscopy against Tubulin B (green), MUC5AC (red) and DAPI (blue). Apical surface view is shown on top and cross-sectional view below on each image. Scale bar represents 30 µm. B, Cell coverage grade (blue squares, mean ± SEM) and cell barrier integrity as measured by TEER (black circles) in the same experimental set-up as in (A). C, Regression analysis of the correlation between cell coverage and viability (red) and TEER (black). Figure 3. View largeDownload slide Morphological assessment of whole mount preparations of the MucilAir epithelium and cell coverage correlation with TEER and cell viability. A, Representative cell images over time for AZ4 (400 µM) at indicated time points as determined by confocal microscopy against Tubulin B (green), MUC5AC (red) and DAPI (blue). Apical surface view is shown on top and cross-sectional view below on each image. Scale bar represents 30 µm. B, Cell coverage grade (blue squares, mean ± SEM) and cell barrier integrity as measured by TEER (black circles) in the same experimental set-up as in (A). C, Regression analysis of the correlation between cell coverage and viability (red) and TEER (black). Measurement of Cytotoxicity, Ciliary Beating, MCC, and Cytokine Release Cytotoxicity as measured by LDH release, cytokine release (Interleukin 8 [IL-18], IL-6, Granulocyte-Macrophage Colony-Stimulating Factor, Growth-Regulated Oncogene (GROα) alpha, Regulated on Activation, Normal T Cell Expressed and Secreted, IL-1α, IL-1β, IP10, Matrix Metallopeptidase 9, Transforming Growth Factor beta), MCC and CBF were assessed on days 2, 4, 8, and 12 in the 3 D human airway in vitro system. A 2- to 3-fold increased release of LDH above the MDL (63 nM) was observed for 4 out of the 8 toxic compounds (Supplementary Figure 3A). The cytotoxicity for these few compounds tended to correlate with loss in tissue integrity, while no increase in LDH was found for compounds without known in vivo toxicity. Regarding cytokine release, only CdCl2 released robust amounts of cytokines as exemplified by IL-8 (30–70 ng/ml), and AZ4 induced IL-8 at low but significant levels at day 8 (3 ng/ml) (Supplementary Figure 3B). However, CBF was reduced in the samples with complete loss of cell coverage (Supplementary Figure 4). Similarly, MCC was assessed in the initial phase of the evaluation with variable results. MCC needs further indepth studies to determine its application in this model. Overall, the separation was insignificant between toxic and nontoxic compounds for LDH, IL-8, and CBF. Thus, these read-outs proved nonpredictive for respiratory toxicity (Figure 4B). Figure 4. View largeDownload slide Determination of the predictivity of the MucilAir™ 3D human airway in vitro model for in vivo respiratory toxicity. (A) Cell barrier integrity (circles) and cell viability (squares) were subjected to receiver operating characteristic (ROC) curve analysis. (B) ROC curve analysis of IL-8 (triangles), LDH (circles) and CBF (squares) at the time point of largest change compared to vehicle treated cells. In both A and B, the dotted line shows the results of random assignment of respiratory toxicity. Figure 4. View largeDownload slide Determination of the predictivity of the MucilAir™ 3D human airway in vitro model for in vivo respiratory toxicity. (A) Cell barrier integrity (circles) and cell viability (squares) were subjected to receiver operating characteristic (ROC) curve analysis. (B) ROC curve analysis of IL-8 (triangles), LDH (circles) and CBF (squares) at the time point of largest change compared to vehicle treated cells. In both A and B, the dotted line shows the results of random assignment of respiratory toxicity. DISCUSSION Toxicity in the respiratory tract represents a major obstacle in the development of inhaled therapeutics, and is often not identified until late stage in vivo toxicity studies resulting in project closure. One of the key issues has been the development of reliable and predictive in vitro systems to permit earlier detection of respiratory toxicities and potential liabilities. This investigation explored the capability of a 3D human airway in vitro system, MucilAir (Huang et al., 2013), to predict respiratory toxicity in vivo, by repeated administration of compounds intended for chronic treatment of respiratory disease by inhalation. To the best of our knowledge, this is the first study with a direct application of inhaled drug candidates in a physiologically relevant 3D human airway in vitro model for evaluation of associated respiratory toxicity in vivo. The data demonstrate that both cell barrier integrity and cell viability as measured by TEER and Resazurin, respectively, can be utilized as predicting factors for respiratory toxicity of inhaled drugs in the MucilAir model (Figure 4A). The TEER signal was significantly reduced for all but one (pentamidine) of the studied compounds with confirmed pulmonary toxicity, while not affected by any of the compounds without respiratory liabilities. The accuracy as determined by area under the ROC curve was above 0.9 after 12 repeated doses with 86% sensitivity at 100% specificity for both TEER and cell viability, which can be considered as excellent predictability of respiratory toxicities (Hajian-Tilaki, 2013, Pencina et al., 2008). The TEER read-out has some clear advantages compared with cell viability. First, due to a higher degree of significance shown here, but more importantly by allowing measurements of toxicity over time. The loss in barrier integrity and cell viability also correlated with cell confluency for 6 out of 7 compounds with a reduction in TEER, suggesting that the major reason for toxicity for these compounds is related to loss in viability. All other measured read-outs were nonsignificant for predicting lung toxicity. For example, the released amounts of LDH and cytokines were barely over MDLs for the majority of treatments. This is in line with observations that 3D human airway cells are relatively resistant to LDH and cytokine release following insults (Zavala et al., 2016). These correlations to respiratory toxicity in vivo only occurred at relatively high compound concentrations, namely 400 µM (Figure 1). Following inhalation, the lung, and the epithelium in particular, is exposed to highly heterogeneous drug concentrations, resulting in high regional concentrations. Additionally, the human lung lining fluid volume is estimated to 30 ml (Rennard et al., 1986), and even modest inhaled doses are likely to result in high instant concentrations when distributed in this small volume. Our recent inhouse experiments suggest that whole lung concentrations reach approximately 70–2000 µM (4 different compounds) in rat inhalation studies at toxicologically relevant dose ranges (Åberg et al., unpublished data). Furthermore, applications of robust margins to any confirmed toxicity in animal studies is expected in the context of respiratory toxicity risk assessment (Alton et al., 2012; Tepper et al., 2014). In addition, as with many other toxicities, nonadaptive adverse tissue responses in lung tend to accumulate over time. Taken together, it is plausible that measurement of cell barrier integrity and cell viability in vitro by repeated administration at high concentrations are relevant for in vivo respiratory toxicity, as a consequence of high local levels in lung and by mimicking accumulation of adverse responses in the respiratory tract over time. Neither barrier integrity nor cell viability were significantly affected compared with vehicle controls by pentamidine, which was included in our evaluations as an agent associated with respiratory toxicity. Pentamidine, is approved for oral inhalation for the prevention of Pneumocystis jiroveci pneumonia in high-risk, HIV-infected patients (NDA 019887, 2010), but has been associated with wheezing and cough in humans (Katzman et al., 1992; TAPS group, 1990). Thus, the tolerability profile appears to be incompatible with chronic therapy. However, to our knowledge, there is limited information on toxicity following inhaled exposure in animals. The data set presented here was based on a limited set of compounds with either confirmed or estimated therapeutic doses at levels generally not >2 mg. In this context pentamidine is an outlier with a therapeutic dose range >100 mg. As our experimental set-up was based on using the same concentration range, despite differences in potency and therapeutic levels, the relevant concentrations may not have been achieved in the case of pentamidine. Another plausible explanation for intact barrier, could be that cough and wheezing by pentamidine in humans are related to stimulation of sensory nerve receptors (Nasra and Belvisi, 2009) or induction of structural and inflammatory airway mucosal changes (Chung and Pavord, 2008) that may not be detected in this in vitro airway model. Salmeterol is approved for inhalation treatment in respiratory indications as a xinfoate salt, and this form is not associated with lung toxicity (Owen et al., 2010). In this study, we included the free base form of salmeterol as predictive parameter for toxicity on the basis of published data suggesting that this form was associated with unacceptable respiratory tract irritancy/toxicity (Owen et al., 2010). Interestingly, we found that salmeterol in the free base form reduced TEER and cell viability, but not cell coverage (Figs. 1 and 2 and Supplementary Figure 3B). TEER measurement is a widely accepted quantitative technique to measure the integrity of tight junction dynamics in cell models, and TEER values are strong indicators of the integrity of the cellular barriers (Srinivasan et al., 2015). Thus, it is possible that salmeterol free base affects the tight junctions at high concentrations resulting in loss of TEER. This may imply that barrier integrity measured by TEER may be indicative of toxicity also in the absence of effects on cell coverage. However, it is also important to consider that the sole reduction in TEER represents a more limited response where translation to toxicity is not absolute, and that the salt form of salmeterol is widely used without association to respiratory toxicity. In addition, in contrast to pentamidine, the therapeutic dose of salmeterol is relatively low. This reflects that the extent of the response, the dosage form as well as the therapeutic dose are important factors to take into consideration in the context of decision making about compound progression. This study could potentially be biased due to a low study power (n = 15) and inclusion of a limited set of chemical drug classes, with a focus on kinase-inhibitors and phosphodiesterase-inhibitors in the group of toxic compounds. In addition, most of the toxicities reported here were observed in rats and dogs, while the in vitro model is based on human cells. As mechanisms for toxicity were not determined, there is a potential for species differences. However, based on the outcome of these experiments, it appears that these differences were not significant, as reflected in the robust predictability. Furthermore, the in vivo responses comprised of a range of phenotypes such as bronchopneumonia, alveolitis, edema and/or changes in cell morphology, while the in vitro model evidently only addresses the direct impact on the epithelium. There are evident opportunities to further enhance the power of the predictability assessment by incorporation of novel inhaled drug candidates as they become available, as well as drugs with lung irritancy and further assess viability in this model during the treatment phase. Also development of more complex coculture systems and dry powder aerosol delivery systems could potentially enhance the resolution by replicating the multicellular environment of the respiratory tract and the in vivo exposure conditions. In addition, further evaluation of viability measures in cell line systems, such as THP-1 cells, are warranted due to the close alignment between cell viability and respiratory toxicity in the 3D human airway in vitro systems in this study. In a drug discovery context, these results suggest that measurements of cell barrier integrity and cell viability in a 3D human airway in vitro model represent a valuable addition to the tools which can be utilized to reduce the risk of attrition of inhaled drug candidates in comprehensive regulatory in vivo toxicity studies. Furthermore, our data emphasize that a larger attention to deleterious effects in vitro at concentrations well above an acceptable therapeutic index should be taken into account in predicting risk for respiratory toxicity. In summary, this approach may potentially reduce the risk for late-stage drug attrition, reduce the number of animals used in early preclinical toxicity assessments and thus improve the safety profile of novel drugs for human respiratory diseases. SUPPLEMENTARY DATA Supplementary data are available at Toxicological Sciences online. FUNDING This work was funded by Drug Safety and Metabolism IMED Biotech Unit at AstraZeneca, and most of the practical part of this work was performed at Epithelix Sàrl (Geneva, Switzerland). ACKNOWLEDGMENT We thank Marta Perez Alcazar for her technical help for the confocal imaging. REFERENCES Ahuja V., Bokan S., Sharma S. ( 2017). Predicting toxicities in humans by nonclinical safety testing: An update with particular reference to anticancer compounds. Drug Discov. Today  22, 127– 132. http://dx.doi.org/10.1016/j.drudis.2016.09.007 Google Scholar CrossRef Search ADS PubMed  Alton E. W., Boushey H. A., Garn H., Green F. H., Hodges M., Martin R. J., Murdoch R. D., Renz H., Shrewsbury S. B., Seguin R. ( 2012). Clinical expert panel on monitoring potential lung toxicity of inhaled oligonucleotides: Consensus points and recommendations. Nucleic. Acid Ther . 22, 246– 254. Google Scholar PubMed  Balharry D., Sexton K., BéruBé K. A. ( 2008). An in vitro approach to assess the toxicity of inhaled tobacco smoke components: Nicotine, cadmium, formaldehyde and urethane. Toxicology  244, 66– 76. Google Scholar CrossRef Search ADS PubMed  Cao X., Muskhelishvili L., Latendresse J., Richter P., Heflich R. H. ( 2017). Evaluating the toxicity of cigarette whole smoke solutions in an air-liquid-interface human in vitro airway tissue model. Toxicol. Sci . 156, 14– 24. Google Scholar PubMed  Chauhan B. F., Ducharme F. M. ( 2014). Addition to inhaled corticosteroids of long-acting beta2-agonists versus anti-leukotrienes for chronic asthma. Cochrane Database Syst. Rev . 1, CD003137. Chung K. F., Pavord I. D. ( 2008). Prevalence, pathogenesis, and causes of chronic cough. Lancet  371, 1364– 1374. http://dx.doi.org/10.1016/S0140-6736(08)60595-4 Google Scholar CrossRef Search ADS PubMed  Cook D., Brown D., Alexander R., March R., Morgan P., Satterthwaite G., Pangalos M. N. ( 2014). Lessons learned from the fate of AstraZeneca’s drug pipeline: A five-dimensional framework. Nat. Rev. Drug Discov . 13, 419– 431. Google Scholar CrossRef Search ADS PubMed  US EPA ( 2016). Definition and Procedure for the Determination of the Method Detection Limit, Revision 2, 40 CFR 136, Appendix B, 2016. Available on www.epa.org, last accessed in November 28, 2017. Hayes A. J., Bakand S. ( 2014). Toxicological perspectives of inhaled therapeutics and nanoparticles. Expert Opin. Drug Metab. Toxicol . 10, 933– 947. http://dx.doi.org/10.1517/17425255.2014.916276 Google Scholar CrossRef Search ADS PubMed  Hajian-Tilaki K. ( 2013). Receiver operating characteristic (ROC) curve analysis for medical diagnostic test evaluation. Caspian J. Intern. Med . 4, 627– 635. Google Scholar PubMed  Huang S., Wiszniewski L., Constant S., Roggen E. ( 2013). Potential of in vitro reconstituted 3D human airway epithelia (MucilAir™) to assess respiratory sensitizers. Toxicol. In Vitro  27, 1151– 1156. Google Scholar CrossRef Search ADS PubMed  Huang S., Boda B., Vernaz J., Ferreira E., Wiszniewski L., Constant S. ( 2017). Establishment and characterization of an in vitro human small airway model (SmallAirTM). Eur. J. Pharm. Biopharm . 118, 68– 72. Google Scholar CrossRef Search ADS PubMed  Iskandar A. R., Martin F., Talikka M., Schlage W. K., Kostadinova R., Mathis C., Hoeng J., Peitsch M. C. ( 2013). Systems approaches evaluating the perturbation of xenobiotic metabolism in response to cigarette smoke exposure in nasal and bronchial tissues. Biomed Res. Int . 2013, 512086. Google Scholar CrossRef Search ADS PubMed  Katzman M., Meade W., Iglar K., Rachlis A., Berger P., Chan C. K. ( 1992). High incidence of bronchospasm with regular administration of aerosolized pentamidine. Chest  101, 79– 81. Google Scholar CrossRef Search ADS PubMed  Kuper C. F., Gröllers-Mulderij M., Maarschalkerweerd T., Meulendijks N. M. M., Reus A., van Acker F., Zondervan-van den Beuken E. K., Wouters M. E. L., Bijlsma S., Kooter I. M. ( 2015). Toxicity assessment of aggregated/agglomerated cerium oxide nanoparticles in an in vitro 3D airway model: The influence of mucociliary clearance. Toxicol. In Vitro  29, 389– 397. Google Scholar CrossRef Search ADS PubMed  McDougall C. M., Blaylock M. G., Douglas J. G., Brooker R. J., Helms P. J., Walsh G. M. ( 2008). Nasal epithelial cells as surrogates for bronchial epithelial cells in airway inflammation studies. Am. J. Resp. Cell Mol. Biol . 39, 560– 568. Google Scholar CrossRef Search ADS   Nasra J., Belvisi M. G. ( 2009). Modulation of sensory nerve function and the cough reflex: Understanding disease pathogenesis. Pharmacol. Ther . 124, 354– 375. http://dx.doi.org/10.1016/j.pharmthera.2009.09.006 Google Scholar CrossRef Search ADS PubMed  NDA 019887. ( 2010). United States Food and Drug Administration. Data file retrieved from https://www.accessdata.fda.gov/drugsatfda_docs/label/2011/019887s014lbl.pdf, last accessed November 28, 2017. Owen K., Beck S. L., Damment S. J. P. ( 2010). The preclinical toxicology of salmeterol hydroxynaphthoate. Hum. Exp. Toxicol . 29, 393– 407. http://dx.doi.org/10.1177/0960327110363335 Google Scholar CrossRef Search ADS PubMed  Pencina M. J., D' Agostino R. B., D' Agostino R. B., Vasan R. S. ( 2008). Evaluating the added predictive ability of a new marker: From area under the ROC curve to reclassification and beyond. Stat. Med . 27, 157– 172. Google Scholar CrossRef Search ADS PubMed  Rennard S. I. G., Basset D., Lecossier K. M., O’Donell P., Pinkston P. G. M., Crystal R. G. ( 1986). Estimation of volume of epithelial lining fluid recovered by lavage using urea as marker of dilution. J. Appl. Physiol . 60, 532– 538. Google Scholar CrossRef Search ADS PubMed  Rothen-Rutishauser B., Blank F., Mühlfeld C., Gehr P. ( 2008). In vitro models of the human epithelial airway barrier to study the toxic potential of particulate matter. Expert Opin. Drug Metab. Toxicol . 4, 1075– 1089. Google Scholar CrossRef Search ADS PubMed  Srinivasan B., Kolli A. R., Esch M. B., Abaci H. E., Shuler M. L., Hickman J. J. ( 2015). TEER measurement techniques for in vitro barrier model systems. J. Lab. Autom . 20, 107– 126. Google Scholar CrossRef Search ADS PubMed  Thavagnanam S., Parker J. C., McBrien M. E., Skibinski G., Shields M. D., Heaney L. G. ( 2014). Nasal epithelial cells can act as a physiological surrogate for paediatric asthma studies. PLoS One  9, e85802. Google Scholar CrossRef Search ADS PubMed  Toronto Aerosolized Pentamidine Study (TAPS) Group. ( 1990). Acute pulmonary effects of aerosolized pentamidine: A randomized controlled study. Chest  98, 907– 910. CrossRef Search ADS PubMed  Tepper J., Ochoa R., Rix P., Elliott G., Hoglen N., Poulin D., Parsley E., Masamune H. ( 2014). A 26-week toxicity assessment of AIR001 (sodium nitrite) by inhalation exposure in rats and by intravenous administration in dogs. Int. J. Toxicol . 33, 162– 174. Google Scholar CrossRef Search ADS PubMed  Wain L. V., Shrine N., Artigas M. S., Erzurumluoglu A. M., Noyvert B., Bossini-Castillo L., Obeidat M., Henry A. P., Portelli M. A., Hall R. J., et al.   ( 2017). Genome-wide association analyses for lung function and chronic obstructive pulmonary disease identify new loci and potential druggable targets. Nat. Genet . 49, 416– 425. Google Scholar CrossRef Search ADS PubMed  Youden W. J. ( 1950). Index for rating diagnostic tests. Cancer  3, 32– 35. http://dx.doi.org/10.1002/1097-0142(1950)3:1<32::AID-CNCR2820030106>3.0.CO;2-3 Google Scholar CrossRef Search ADS PubMed  Zavala J., O'Brien B., Lichtveld K., Sexton K. G., Rusyn I., Jaspers I., Vizuete W. ( 2016). Assessment of biological responses of EpiAirway 3-D cell constructs versus A549 cells for determining toxicity of ambient air pollution. Inhal. Toxicol . 28, 251– 259. Google Scholar CrossRef Search ADS PubMed  Zhang J. H., Chung T. D. Y., Oldenburg K. R. ( 1999). A simple statistical parameter for use in evaluation and validation of high throughput screening assays. J. Biomol. Screen.  4, 67– 73. http://dx.doi.org/10.1177/108705719900400206 Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press on behalf of the Society of Toxicology. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Toxicological Sciences Oxford University Press

A 3D Human Airway Model Enables Prediction of Respiratory Toxicity of Inhaled Drugs In Vitro

Loading next page...
 
/lp/ou_press/a-3d-human-airway-model-enables-prediction-of-respiratory-toxicity-of-7jQ5qqTQ4F
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press on behalf of the Society of Toxicology. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com
ISSN
1096-6080
eISSN
1096-0929
D.O.I.
10.1093/toxsci/kfx255
Publisher site
See Article on Publisher Site

Abstract

Abstract Respiratory tract toxicity represents a significant cause of attrition of inhaled drug candidates targeting respiratory diseases. One of the key issues to allow early detection of respiratory toxicities is the lack of reliable and predictive in vitro systems. Here, the relevance and value of a physiologically relevant 3D human airway in vitro model (MucilAir) were explored by repeated administration of a set of compounds with (n = 8) or without (n = 7) respiratory toxicity following inhalation in vivo. Predictability for respiratory toxicity was evaluated by readout of cytotoxicity, barrier integrity, viability, morphology, ciliary beating frequency, mucociliary clearance and cytokine release. Interestingly, the data show that in vivo toxicity can be predicted in vitro by studying cell barrier integrity by transepithelial electrical resistance (TEER), and cell viability determined by the Resazurin method. Both read-outs had 88% sensitivity and 100% specificity, respectively, while the former was more accurate with receiver operating characteristic (ROC) AUC of 0.98 (p = .0018) compared with ROC AUC of 0.90 (p = .0092). The loss of cell barrier integrity could mainly, but not fully, be attributed to a loss of cell coverage in 6 out of 7 compounds with reduced TEER. Notably, these effects occurred only at 400 µM, at concentration levels significantly above primary target cell potency, suggesting that greater attention to high local lung concentrations should be taken into account in safety assessment of inhaled drugs. Thus, prediction of respiratory toxicity in 3D human airway in vitro models may result in improved animal welfare and reduced attrition in inhaled drug discovery projects. respiratory toxicology, inhalation toxicology, predictive toxicology, airway epithelium, 3D in vitro model Inhalation is the key route for delivery of therapeutic agents to treat respiratory diseases such as asthma and chronic obstructive pulmonary disease. Inhaled corticosteroids, β2-agonists and antimuscarinics represent standard care in respiratory diseases (Chauhan and Ducharme, 2014) and a number of additional therapeutic opportunities are being explored (Wain et al., 2017). The utility of the inhaled route for drug delivery in respiratory diseases presents a great advantage with topical access to the diseased tissue and a low systemic exposure with associated reduction in the potential for systemic side effects. However, unacceptable toxicity in the respiratory tract represents a major obstacle in the development of inhaled therapies, and has been reported to account for approximately 30% of inhaled project closures (Cook et al., 2014). Respiratory toxicities have typically been identified at a relatively late stage in pre-clinical testing as part of the comprehensive assessment undertaken during in vivo toxicity studies with the aim to identify therapeutic margins, maximal tolerable concentrations and reversibility of any noted adverse effects (Ahuja et al., 2017; Hayes and Bakand, 2014). To this point, improved early assessment and prediction of the toxicity potential of new molecules would add significant value in terms of the potential to reduce late stage compound attrition, increase quality of drug candidates, and improve animal welfare by replacement, reduction or refinement of animal usage in preclinical toxicity testing of novel inhaled drugs. One of the key limitations for early detection of respiratory toxicities so far has been around development of reliable and predictive in vitro models. The airway epithelium represents a barrier with a role in protection of the airways and is one of several key target sites for toxicity in the respiratory tract. Therefore, a major focus has been on developing in vitro models representing this region of the lung. Primary airway epithelial cells cultured at air-liquid interphase represent physiologically relevant models of the human airway consisting of ciliated cells, mucus secreting goblet cells and basal cells (Balharry et al., 2008; Huang et al., 2013), and are now commercialized under trade names such as MucilAir and EpiAirway by Epithelix Sàrl (Geneva, Switzerland) and MatTek Corporation (Ashland, MA, US), respectively. So far, toxicity evaluations in such 3D human airway in vitro models have focused on lung irritancy of environmental toxic agents such as smoke, ozone, formaldehyde, CdCl2, and nanoparticles (Balharry et al., 2008; Cao et al., 2017; Huang et al., 2013; Kuper et al., 2015; Rothen-Rutishauser et al., 2008). Here, we, to the best of our knowledge, for the first time explore the relevance, value and potential application of the MucilAirmodel to predict in vivo respiratory toxicity in the context of inhaled drug candidates for treatment of respiratory diseases. We focused our in vitro investigation in the MucilAir model around a set of 14 inhaled drugs or drug candidates and CdCl2, a known toxic agent, as a positive control. These were divided into compounds with (n = 8) or without (n = 7) respiratory toxicity based on historical data from limited dose-range and dose duration in vivo studies (Table 1). The compounds with toxicities had previously shown to induce a range of different adverse responses in the respiratory tract, mostly observed as histopathological changes in the lung in animal toxicology studies (Table 1). In addition, pentamidine, an approved inhaled drug associated with respiratory adverse events, (TAPS group, 1990, Katzman et al., 1992) and salmeterol in the free base form (Owen et al., 2010) were included as positive controls. A range of different parameters such as cytotoxicity, cell barrier integrity, viability, morphology, ciliary beating frequency (CBF), mucociliary clearance (MCC), and cytokine release were analyzed for predictive accuracy for respiratory toxicity. With this approach, we demonstrate that a 3D human airway in vitro model is capable of predicting the potential for drug-induced adverse effects in the respiratory tract, by monitoring the loss of barrier integrity and cell viability as measured by transepithelial electrical resistance (TEER) and Resazurin, respectively. In conclusion, the 3D human airway in vitro model has a clear potential to reduce compound attrition due to respiratory toxicity, and insight from this work could help to guide the development of future inhaled therapies for respiratory diseases. Table 1. Summary of Toxicity Findings of Studied Drugs Compound  Target  Tox  Toxicity Finding  AZ1  IKK2  Yes  Increased lung weight, inflammatory/epithelial pathology in lung in rat and dog studies.a  AZ2  IKK2  Yes  Increased lung weight, inflammatory/epithelial pathology in lung in rat and dog studies.  AZ3  PDE4  Yes  Acute lung inflammation and nasal inflammation/ulceration in dog (but not in rat)  AZ4  PDE4  Yes  Acute lung inflammation and nasal inflammation/ulceration in dog (not tested in rat).a  D20207  PKC  Yes  Inflammatory/epithelial pathology in rat lung/airways, cough in dogs.a  Pentamidine  Anti-infective  Yes  Marketed product—associated with high incidence of cough, difficulty breathing and burning sensation in throat (TAPS group, 1990; Katzman et al., 1992).  Salmeterol Base  LABA  Yes  Development of this form of salmeterol was discontinued due to respiratory tract irritancy in early rat studies (Owen et al., 2010).  CdCl2    Yes  Acute interstitial pneumonitis and pulmonary edema, leading to interstitial fibrosis, and emphysema.  AZD9819  NEI  No  No adverse effects in respiratory tract in 3-month rat and dog studiesa  Formoterol  LABA  No  Marketed product  Budesonide  GCS  No  Marketed product  Tiotropium  LAMA  No  Marketed product  Cromoglycate  Mast cell stabilizer  No  Marketed product  Glycopyrronium  LAMA  No  Marketed product  AZD9164  LAMA  No  Slight increases in lung weights, no significant histopathology, transient functional effects in rat and dog, and Chronic Obstructive Pulmonary Disease patients (decreased Forced Expiratory Volume 1s).a  Compound  Target  Tox  Toxicity Finding  AZ1  IKK2  Yes  Increased lung weight, inflammatory/epithelial pathology in lung in rat and dog studies.a  AZ2  IKK2  Yes  Increased lung weight, inflammatory/epithelial pathology in lung in rat and dog studies.  AZ3  PDE4  Yes  Acute lung inflammation and nasal inflammation/ulceration in dog (but not in rat)  AZ4  PDE4  Yes  Acute lung inflammation and nasal inflammation/ulceration in dog (not tested in rat).a  D20207  PKC  Yes  Inflammatory/epithelial pathology in rat lung/airways, cough in dogs.a  Pentamidine  Anti-infective  Yes  Marketed product—associated with high incidence of cough, difficulty breathing and burning sensation in throat (TAPS group, 1990; Katzman et al., 1992).  Salmeterol Base  LABA  Yes  Development of this form of salmeterol was discontinued due to respiratory tract irritancy in early rat studies (Owen et al., 2010).  CdCl2    Yes  Acute interstitial pneumonitis and pulmonary edema, leading to interstitial fibrosis, and emphysema.  AZD9819  NEI  No  No adverse effects in respiratory tract in 3-month rat and dog studiesa  Formoterol  LABA  No  Marketed product  Budesonide  GCS  No  Marketed product  Tiotropium  LAMA  No  Marketed product  Cromoglycate  Mast cell stabilizer  No  Marketed product  Glycopyrronium  LAMA  No  Marketed product  AZD9164  LAMA  No  Slight increases in lung weights, no significant histopathology, transient functional effects in rat and dog, and Chronic Obstructive Pulmonary Disease patients (decreased Forced Expiratory Volume 1s).a  a Åberg et al., unpublished data. MATERIALS AND METHODS 3D human airway in vitro model Airway cells were obtained from patients undergoing surgical polypectomy. All experimental procedures were explained in full, and all subjects provided informed consent. The study was conducted according to the declaration of Helsinki on biomedical research (Hong Kong amendment, 1989), and received approval from the local ethics commission. MucilAir (Epithelix Sàrl, Geneva, Switzerland) airway epithelia, reconstituted with a mixture of human nasal cells isolated from a panel of 14 different donors, were cultured at the air liquid interface in MucilAir culture medium (Epithelix Sàrl, Geneva, Switzerland), in 24-well plates with 6.5-mm Transwell inserts (Corning). The cell culture medium was changed at each sampling point or every 2–3 days for those wells not being sampled. The nasal version of MucilAir were used as surrogate for bronchial epithelia due to availability, quality and close resemblance in gene expression levels, morphology and function (Iskandar et al., 2013; McDougall et al., 2008; Thavagnanam et al., 2014). Testing strategy All compounds were synthesized at AstraZeneca, except CdCl2, which was purchased from Sigma. The compounds were dissolved in DMSO and diluted in a buffered saline solution (0.9% NaCl, 1.25 mM CaCl2, 10 mM HEPES) to reduce the final concentration of DMSO to 0.8% (v/v). Test compounds (10 μl) were applied daily at 4, 40, and 400 µM on the apical surface of MucilAir cultures and left to incubate for 6 h prior to replacing with fresh medium. This was repeated for a period of 12 days. Each treatment was run as 4 replicates on each occasion. Culture supernatants were collected and stored at −80 °C for future analysis. The Transwell membranes containing the 3D airway cells were fixed for 30 min in 4% Formaldehyde-PBS (with Ca/Mg) at RT, washed in PBS and stored at 4 °C until processing for immunohistochemistry. Measurements TEER and CBF were assessed as previously described (Huang et al., 2017). Cell viability was measured by Resazurin test at termination and cytokines assessed at indicated time-points as previously described (Huang et al., 2013). Cytotoxicity was measured by quantification of lactate dehydrogenase (LDH) in supernatants by Cytotoxicity Detection KitPLUS (Roche) according to manufacturer’s instructions. MCC was assessed as previously described (Huang et al., 2017). In short, 30 µm polystyrene microbeads (Sigma,) were added on the apical side of the MucilAir cells. The movement of the microbeads was recorded with a Sony XCD-U100CR camera at 2 frames per second, on an Axiovert 200 M microscope (Zeiss), at room temperature. A total of 50 images were recorded and more than 1000 beads were tracked and the velocity of MCC (µm/s) was calculated using Image-Pro Plus software (version 6.2, Media Cybernetics). Three movies were taken and analyzed per insert. Immunohistochemistry and determination of cell coverage grade Membranes were permeabilised with 0.5% Triton-X100 in PBS buffer for 20 min, blocked with 2.5% BSA, 5% goat serum, 0.05% Tween-20 in PBS for 30 min. Cells was incubated with antiTubulin (NBP2-00812, NovusBio), and antiMuc5AC (MS145P1, Thermo Scientific) in the block buffer for 60 min, followed by Alexa 488 coupled mouse IgG2A (A21131, Invitrogen) and Alexa 568-Mouse IgG1 (A21124, Invitrogen) for 45 min in the dark. All steps were performed at room temperature. The membranes were mounted on glass slides in Vectashield antifade mounting medium containing DAPI stain (Vector Lab) and evaluated under fluorescent microscope Ax70 (Olympus). Images were captured by confocal microscope LSM880 (Zeiss) connected to ZEN 2.3 software (Zeiss) and illustrated by using Photoshop, version 11 (Adobe). The cell coverage grade was determined on blinded samples by visual estimate of the confluency area: 70%–100% (grade 4); 30%–70% (grade 3); 10%–30% (Grade 2); <10% (grade 1); loss of all cells (grade 0). Statistics The method detection limit (MDL) is defined as the minimum value of a substance that can be distinguishable from method blank with 99% confidence and were determined according the U.S. Environmental Protection Agency method (US EPA 2016). Quality of the assays was determined by the Zeta-factor (Z’) method (Zhang et al. 1999). All other statistical calculations were performed by GraphPad prism 7.0 software (GraphPad Software, San Diego, California). Cut-off values were determined by Youden's index (J), i.e. {J = sensitivity + specificity – 1} (Youden, 1950). RESULTS Measurement of Cell Barrier Integrity in 3D Human Airway Epithelial Cell Cultures The fully differentiated MucilAir 3D human airway cultures showed a nonsignificant difference in cell barrier integrity as measured by TEER between assay occasions as determined with 1-way ANOVA with a mean TEER value across studies of 440 (SD 70) Ω cm2 (n = 3). However, we noted an intra-assay time-point dependent shift in TEER values with an average intra-assay coefficient of variance (intra-assay CV) of 16.9%. Therefore, in order to compare compound effects between different assay occasions and assay time-points, the change in TEER value from vehicle was normalized against vehicle control on each day. By this approach, the MDL of the TEER read-out was 25.5% (n = 40) with an intra-assay CV of 9.4% (95% CI 6.7–12.1) and an average Z- prime of 0.75 per plate. In order to predict respiratory toxicity, a test-set of 15 compounds were evaluated in the MucilAir 3D human airway epithelial model, by repeat application of compounds to the apical side, for 6 h per day for 12 days, to represent a prolonged residence time and to mimic the inhaled route. The compounds were divided into 2 different categories based on historical data, compounds with (n = 8) and without (n = 7) demonstrated respiratory toxicity following inhalation (Table 1). Cell barrier integrity was measured for all compounds at 4, 40, and 400 µM and a reduction of TEER over MDL was evident in 7 out of 8 compounds with known toxicity at the highest tested dose of 400 µM after 12 repetitive doses (Figure 1A). Importantly, a reproducible and time-dependent loss of barrier integrity was found for all compounds with known respiratory toxicity in animals (n = 7), while all compounds without respiratory toxicity and pentamidine, a drug with lung irritancy side-effects in human, were below MDL (Figs. 1B and 1C and Supplementary Figure 1A). Figure 1. View largeDownload slide Impact of compounds with and without respiratory toxicity in a 3D human airway in vitro model. Percentage change of cell barrier integrity, as measured by TEER, compared to vehicle control after repeat exposure of compounds to the apical side of the epithelial cell layer. (A) Dose response at day 12 of all 8 compounds with respiratory toxicity (red) and representative compounds without toxicity (green, 3 out of 7 compounds). Dotted line represents the method detection limit (MDL) of significant change compared to vehicle treated cells. All other non-toxic compounds were above MDL. (B) Time course at 400 µM of the same compounds as shown in A. (C) Reduction in cell barrier integrity (represented as % reduction in TEER) at 400 µM on day 12 for compounds with (closed bars) or without (open bars) respiratory toxicity (mean SEM). Figure 1. View largeDownload slide Impact of compounds with and without respiratory toxicity in a 3D human airway in vitro model. Percentage change of cell barrier integrity, as measured by TEER, compared to vehicle control after repeat exposure of compounds to the apical side of the epithelial cell layer. (A) Dose response at day 12 of all 8 compounds with respiratory toxicity (red) and representative compounds without toxicity (green, 3 out of 7 compounds). Dotted line represents the method detection limit (MDL) of significant change compared to vehicle treated cells. All other non-toxic compounds were above MDL. (B) Time course at 400 µM of the same compounds as shown in A. (C) Reduction in cell barrier integrity (represented as % reduction in TEER) at 400 µM on day 12 for compounds with (closed bars) or without (open bars) respiratory toxicity (mean SEM). Next, the most accurate time-point to predict respiratory toxicity was determined by area under the receiver operating characteristic (ROC) curve analysis. A reduction in TEER from baseline significantly predicted toxicity after 1–3 and 8–12 consecutive days of drug treatment, respectively and most accurately at day 12 with ROC AUC of 0.98 (95% CI, 0.93–1.04, p = .0018) (Supplementary Figs. 2B and 4A). To be able to distinguish between toxic and nontoxic compounds, a cut-off value was calculated to provide optimum sensitivity at the highest possible specificity by Youden analysis. The sensitivity and specificity at a cut-off value of 32% decrease in the TEER signal were 88% and 100%, respectively at day 12. Measurement of Cell Viability Cell viability was detected by reduction of resazurin by measuring fluorescence after addition of the detection probe to the cell cultures at the termination of the experiment. The vehicle treated samples had a mean RFU of 12800 with inter-assay CV of 15% (n = 3) and a mean intra-assay CV of 14% (95% CI, 10.2–17.8) resulting in a MDL of 39% (n = 26). In drug treated samples, 5 out of 8 toxic drugs showed a reduction of cell viability beyond MDL, including salmeterol in the free base form with a 48% reduction compared with vehicle (Figure 2A). Cell viability measured at termination of the experiment showed a significant alignment towards TEER (Figure 2B) and predictive accuracy for respiratory toxicity with ROC AUC of 0.90 (95% CI, 0.71–1.09, p = .0092) (Figure 4A). Cell viability had similar sensitivity for predicting toxicity as TEER, with 88% sensitivity and 100% specificity, respectively at cut-off values of 14% reduction in viability compared with vehicle. These data imply that respiratory in vivo toxicity of inhaled drugs could be captured in vitro by assessing both TEER and cell viability. Figure 2. View largeDownload slide Impact of compounds with and without respiratory toxicity on cell viability in a 3D human airway in vitro model. (A) Percentage change of viability, compared to vehicle control measured by resazurin conversion after 12 days of repeat exposure of compounds with (closed bars) or without (open bars) respiratory toxicity (mean SEM). Dotted line represents the method detection limit (MDL) of significant change compared to vehicle treated cells. (B) Regression analysis of the correlation between TEER and viability. Figure 2. View largeDownload slide Impact of compounds with and without respiratory toxicity on cell viability in a 3D human airway in vitro model. (A) Percentage change of viability, compared to vehicle control measured by resazurin conversion after 12 days of repeat exposure of compounds with (closed bars) or without (open bars) respiratory toxicity (mean SEM). Dotted line represents the method detection limit (MDL) of significant change compared to vehicle treated cells. (B) Regression analysis of the correlation between TEER and viability. Correlation of Cell Coverage Against Barrier Integrity and Cell Viability First, the morphology of vehicle treated MucilAir cultures was assessed on whole mount preparations by immunohistochemistry against Tubulin B to detect ciliated cells and cytoskeleton, MUC5AC to detect mucin in goblet cells and DAPI as nuclear stain. The airway epithelium cultures were fully confluent and well differentiated as the majority of cells on the apical side were ciliated epithelial cells, and a minority of goblet cells (Supplementary Figure 2A). In the cross section, the cells were arranged in a pseudostratified pattern with some variations of the thickness of the epithelium. Thus, the 3D human airway epithelial cultures showed a physiological resemblance to a human ciliated epithelial layer in the airway. Next, a cell coverage grading system based on visual estimates of cell confluence area was developed to potentially link loss of TEER and cell viability to cell confluency. First, the epithelial damage was examined for AZ4 at 400 µM on treatment days 0, 1, 8, 12, and after recovery where cells were left untreated for 7 days. Cell coverage decreased with time of drug exposure, correlating with a decrease in TEER signal (Figure 3B). This was visible by a progressive loss of cells, starting at the peripheral area at early time points and a total cell loss at later time points (Figure 3A). Second, the epithelial morphology, TEER signal and cell viability measured by the Resazurin method were compared between all 15 compounds at all tested concentrations at the end of the experiment. Loss of cell coverage was evident in 6 out of 7 compounds with reduced TEER and correlated to both TEER and cell viability (Figure 3C and Supplementary Figure 2B). The only exception was salmeterol in the free base form, which showed a reduction in TEER and cell viability, while cell coverage remained in the range observed for vehicle treatment. Figure 3. View largeDownload slide Morphological assessment of whole mount preparations of the MucilAir epithelium and cell coverage correlation with TEER and cell viability. A, Representative cell images over time for AZ4 (400 µM) at indicated time points as determined by confocal microscopy against Tubulin B (green), MUC5AC (red) and DAPI (blue). Apical surface view is shown on top and cross-sectional view below on each image. Scale bar represents 30 µm. B, Cell coverage grade (blue squares, mean ± SEM) and cell barrier integrity as measured by TEER (black circles) in the same experimental set-up as in (A). C, Regression analysis of the correlation between cell coverage and viability (red) and TEER (black). Figure 3. View largeDownload slide Morphological assessment of whole mount preparations of the MucilAir epithelium and cell coverage correlation with TEER and cell viability. A, Representative cell images over time for AZ4 (400 µM) at indicated time points as determined by confocal microscopy against Tubulin B (green), MUC5AC (red) and DAPI (blue). Apical surface view is shown on top and cross-sectional view below on each image. Scale bar represents 30 µm. B, Cell coverage grade (blue squares, mean ± SEM) and cell barrier integrity as measured by TEER (black circles) in the same experimental set-up as in (A). C, Regression analysis of the correlation between cell coverage and viability (red) and TEER (black). Measurement of Cytotoxicity, Ciliary Beating, MCC, and Cytokine Release Cytotoxicity as measured by LDH release, cytokine release (Interleukin 8 [IL-18], IL-6, Granulocyte-Macrophage Colony-Stimulating Factor, Growth-Regulated Oncogene (GROα) alpha, Regulated on Activation, Normal T Cell Expressed and Secreted, IL-1α, IL-1β, IP10, Matrix Metallopeptidase 9, Transforming Growth Factor beta), MCC and CBF were assessed on days 2, 4, 8, and 12 in the 3 D human airway in vitro system. A 2- to 3-fold increased release of LDH above the MDL (63 nM) was observed for 4 out of the 8 toxic compounds (Supplementary Figure 3A). The cytotoxicity for these few compounds tended to correlate with loss in tissue integrity, while no increase in LDH was found for compounds without known in vivo toxicity. Regarding cytokine release, only CdCl2 released robust amounts of cytokines as exemplified by IL-8 (30–70 ng/ml), and AZ4 induced IL-8 at low but significant levels at day 8 (3 ng/ml) (Supplementary Figure 3B). However, CBF was reduced in the samples with complete loss of cell coverage (Supplementary Figure 4). Similarly, MCC was assessed in the initial phase of the evaluation with variable results. MCC needs further indepth studies to determine its application in this model. Overall, the separation was insignificant between toxic and nontoxic compounds for LDH, IL-8, and CBF. Thus, these read-outs proved nonpredictive for respiratory toxicity (Figure 4B). Figure 4. View largeDownload slide Determination of the predictivity of the MucilAir™ 3D human airway in vitro model for in vivo respiratory toxicity. (A) Cell barrier integrity (circles) and cell viability (squares) were subjected to receiver operating characteristic (ROC) curve analysis. (B) ROC curve analysis of IL-8 (triangles), LDH (circles) and CBF (squares) at the time point of largest change compared to vehicle treated cells. In both A and B, the dotted line shows the results of random assignment of respiratory toxicity. Figure 4. View largeDownload slide Determination of the predictivity of the MucilAir™ 3D human airway in vitro model for in vivo respiratory toxicity. (A) Cell barrier integrity (circles) and cell viability (squares) were subjected to receiver operating characteristic (ROC) curve analysis. (B) ROC curve analysis of IL-8 (triangles), LDH (circles) and CBF (squares) at the time point of largest change compared to vehicle treated cells. In both A and B, the dotted line shows the results of random assignment of respiratory toxicity. DISCUSSION Toxicity in the respiratory tract represents a major obstacle in the development of inhaled therapeutics, and is often not identified until late stage in vivo toxicity studies resulting in project closure. One of the key issues has been the development of reliable and predictive in vitro systems to permit earlier detection of respiratory toxicities and potential liabilities. This investigation explored the capability of a 3D human airway in vitro system, MucilAir (Huang et al., 2013), to predict respiratory toxicity in vivo, by repeated administration of compounds intended for chronic treatment of respiratory disease by inhalation. To the best of our knowledge, this is the first study with a direct application of inhaled drug candidates in a physiologically relevant 3D human airway in vitro model for evaluation of associated respiratory toxicity in vivo. The data demonstrate that both cell barrier integrity and cell viability as measured by TEER and Resazurin, respectively, can be utilized as predicting factors for respiratory toxicity of inhaled drugs in the MucilAir model (Figure 4A). The TEER signal was significantly reduced for all but one (pentamidine) of the studied compounds with confirmed pulmonary toxicity, while not affected by any of the compounds without respiratory liabilities. The accuracy as determined by area under the ROC curve was above 0.9 after 12 repeated doses with 86% sensitivity at 100% specificity for both TEER and cell viability, which can be considered as excellent predictability of respiratory toxicities (Hajian-Tilaki, 2013, Pencina et al., 2008). The TEER read-out has some clear advantages compared with cell viability. First, due to a higher degree of significance shown here, but more importantly by allowing measurements of toxicity over time. The loss in barrier integrity and cell viability also correlated with cell confluency for 6 out of 7 compounds with a reduction in TEER, suggesting that the major reason for toxicity for these compounds is related to loss in viability. All other measured read-outs were nonsignificant for predicting lung toxicity. For example, the released amounts of LDH and cytokines were barely over MDLs for the majority of treatments. This is in line with observations that 3D human airway cells are relatively resistant to LDH and cytokine release following insults (Zavala et al., 2016). These correlations to respiratory toxicity in vivo only occurred at relatively high compound concentrations, namely 400 µM (Figure 1). Following inhalation, the lung, and the epithelium in particular, is exposed to highly heterogeneous drug concentrations, resulting in high regional concentrations. Additionally, the human lung lining fluid volume is estimated to 30 ml (Rennard et al., 1986), and even modest inhaled doses are likely to result in high instant concentrations when distributed in this small volume. Our recent inhouse experiments suggest that whole lung concentrations reach approximately 70–2000 µM (4 different compounds) in rat inhalation studies at toxicologically relevant dose ranges (Åberg et al., unpublished data). Furthermore, applications of robust margins to any confirmed toxicity in animal studies is expected in the context of respiratory toxicity risk assessment (Alton et al., 2012; Tepper et al., 2014). In addition, as with many other toxicities, nonadaptive adverse tissue responses in lung tend to accumulate over time. Taken together, it is plausible that measurement of cell barrier integrity and cell viability in vitro by repeated administration at high concentrations are relevant for in vivo respiratory toxicity, as a consequence of high local levels in lung and by mimicking accumulation of adverse responses in the respiratory tract over time. Neither barrier integrity nor cell viability were significantly affected compared with vehicle controls by pentamidine, which was included in our evaluations as an agent associated with respiratory toxicity. Pentamidine, is approved for oral inhalation for the prevention of Pneumocystis jiroveci pneumonia in high-risk, HIV-infected patients (NDA 019887, 2010), but has been associated with wheezing and cough in humans (Katzman et al., 1992; TAPS group, 1990). Thus, the tolerability profile appears to be incompatible with chronic therapy. However, to our knowledge, there is limited information on toxicity following inhaled exposure in animals. The data set presented here was based on a limited set of compounds with either confirmed or estimated therapeutic doses at levels generally not >2 mg. In this context pentamidine is an outlier with a therapeutic dose range >100 mg. As our experimental set-up was based on using the same concentration range, despite differences in potency and therapeutic levels, the relevant concentrations may not have been achieved in the case of pentamidine. Another plausible explanation for intact barrier, could be that cough and wheezing by pentamidine in humans are related to stimulation of sensory nerve receptors (Nasra and Belvisi, 2009) or induction of structural and inflammatory airway mucosal changes (Chung and Pavord, 2008) that may not be detected in this in vitro airway model. Salmeterol is approved for inhalation treatment in respiratory indications as a xinfoate salt, and this form is not associated with lung toxicity (Owen et al., 2010). In this study, we included the free base form of salmeterol as predictive parameter for toxicity on the basis of published data suggesting that this form was associated with unacceptable respiratory tract irritancy/toxicity (Owen et al., 2010). Interestingly, we found that salmeterol in the free base form reduced TEER and cell viability, but not cell coverage (Figs. 1 and 2 and Supplementary Figure 3B). TEER measurement is a widely accepted quantitative technique to measure the integrity of tight junction dynamics in cell models, and TEER values are strong indicators of the integrity of the cellular barriers (Srinivasan et al., 2015). Thus, it is possible that salmeterol free base affects the tight junctions at high concentrations resulting in loss of TEER. This may imply that barrier integrity measured by TEER may be indicative of toxicity also in the absence of effects on cell coverage. However, it is also important to consider that the sole reduction in TEER represents a more limited response where translation to toxicity is not absolute, and that the salt form of salmeterol is widely used without association to respiratory toxicity. In addition, in contrast to pentamidine, the therapeutic dose of salmeterol is relatively low. This reflects that the extent of the response, the dosage form as well as the therapeutic dose are important factors to take into consideration in the context of decision making about compound progression. This study could potentially be biased due to a low study power (n = 15) and inclusion of a limited set of chemical drug classes, with a focus on kinase-inhibitors and phosphodiesterase-inhibitors in the group of toxic compounds. In addition, most of the toxicities reported here were observed in rats and dogs, while the in vitro model is based on human cells. As mechanisms for toxicity were not determined, there is a potential for species differences. However, based on the outcome of these experiments, it appears that these differences were not significant, as reflected in the robust predictability. Furthermore, the in vivo responses comprised of a range of phenotypes such as bronchopneumonia, alveolitis, edema and/or changes in cell morphology, while the in vitro model evidently only addresses the direct impact on the epithelium. There are evident opportunities to further enhance the power of the predictability assessment by incorporation of novel inhaled drug candidates as they become available, as well as drugs with lung irritancy and further assess viability in this model during the treatment phase. Also development of more complex coculture systems and dry powder aerosol delivery systems could potentially enhance the resolution by replicating the multicellular environment of the respiratory tract and the in vivo exposure conditions. In addition, further evaluation of viability measures in cell line systems, such as THP-1 cells, are warranted due to the close alignment between cell viability and respiratory toxicity in the 3D human airway in vitro systems in this study. In a drug discovery context, these results suggest that measurements of cell barrier integrity and cell viability in a 3D human airway in vitro model represent a valuable addition to the tools which can be utilized to reduce the risk of attrition of inhaled drug candidates in comprehensive regulatory in vivo toxicity studies. Furthermore, our data emphasize that a larger attention to deleterious effects in vitro at concentrations well above an acceptable therapeutic index should be taken into account in predicting risk for respiratory toxicity. In summary, this approach may potentially reduce the risk for late-stage drug attrition, reduce the number of animals used in early preclinical toxicity assessments and thus improve the safety profile of novel drugs for human respiratory diseases. SUPPLEMENTARY DATA Supplementary data are available at Toxicological Sciences online. FUNDING This work was funded by Drug Safety and Metabolism IMED Biotech Unit at AstraZeneca, and most of the practical part of this work was performed at Epithelix Sàrl (Geneva, Switzerland). ACKNOWLEDGMENT We thank Marta Perez Alcazar for her technical help for the confocal imaging. REFERENCES Ahuja V., Bokan S., Sharma S. ( 2017). Predicting toxicities in humans by nonclinical safety testing: An update with particular reference to anticancer compounds. Drug Discov. Today  22, 127– 132. http://dx.doi.org/10.1016/j.drudis.2016.09.007 Google Scholar CrossRef Search ADS PubMed  Alton E. W., Boushey H. A., Garn H., Green F. H., Hodges M., Martin R. J., Murdoch R. D., Renz H., Shrewsbury S. B., Seguin R. ( 2012). Clinical expert panel on monitoring potential lung toxicity of inhaled oligonucleotides: Consensus points and recommendations. Nucleic. Acid Ther . 22, 246– 254. Google Scholar PubMed  Balharry D., Sexton K., BéruBé K. A. ( 2008). An in vitro approach to assess the toxicity of inhaled tobacco smoke components: Nicotine, cadmium, formaldehyde and urethane. Toxicology  244, 66– 76. Google Scholar CrossRef Search ADS PubMed  Cao X., Muskhelishvili L., Latendresse J., Richter P., Heflich R. H. ( 2017). Evaluating the toxicity of cigarette whole smoke solutions in an air-liquid-interface human in vitro airway tissue model. Toxicol. Sci . 156, 14– 24. Google Scholar PubMed  Chauhan B. F., Ducharme F. M. ( 2014). Addition to inhaled corticosteroids of long-acting beta2-agonists versus anti-leukotrienes for chronic asthma. Cochrane Database Syst. Rev . 1, CD003137. Chung K. F., Pavord I. D. ( 2008). Prevalence, pathogenesis, and causes of chronic cough. Lancet  371, 1364– 1374. http://dx.doi.org/10.1016/S0140-6736(08)60595-4 Google Scholar CrossRef Search ADS PubMed  Cook D., Brown D., Alexander R., March R., Morgan P., Satterthwaite G., Pangalos M. N. ( 2014). Lessons learned from the fate of AstraZeneca’s drug pipeline: A five-dimensional framework. Nat. Rev. Drug Discov . 13, 419– 431. Google Scholar CrossRef Search ADS PubMed  US EPA ( 2016). Definition and Procedure for the Determination of the Method Detection Limit, Revision 2, 40 CFR 136, Appendix B, 2016. Available on www.epa.org, last accessed in November 28, 2017. Hayes A. J., Bakand S. ( 2014). Toxicological perspectives of inhaled therapeutics and nanoparticles. Expert Opin. Drug Metab. Toxicol . 10, 933– 947. http://dx.doi.org/10.1517/17425255.2014.916276 Google Scholar CrossRef Search ADS PubMed  Hajian-Tilaki K. ( 2013). Receiver operating characteristic (ROC) curve analysis for medical diagnostic test evaluation. Caspian J. Intern. Med . 4, 627– 635. Google Scholar PubMed  Huang S., Wiszniewski L., Constant S., Roggen E. ( 2013). Potential of in vitro reconstituted 3D human airway epithelia (MucilAir™) to assess respiratory sensitizers. Toxicol. In Vitro  27, 1151– 1156. Google Scholar CrossRef Search ADS PubMed  Huang S., Boda B., Vernaz J., Ferreira E., Wiszniewski L., Constant S. ( 2017). Establishment and characterization of an in vitro human small airway model (SmallAirTM). Eur. J. Pharm. Biopharm . 118, 68– 72. Google Scholar CrossRef Search ADS PubMed  Iskandar A. R., Martin F., Talikka M., Schlage W. K., Kostadinova R., Mathis C., Hoeng J., Peitsch M. C. ( 2013). Systems approaches evaluating the perturbation of xenobiotic metabolism in response to cigarette smoke exposure in nasal and bronchial tissues. Biomed Res. Int . 2013, 512086. Google Scholar CrossRef Search ADS PubMed  Katzman M., Meade W., Iglar K., Rachlis A., Berger P., Chan C. K. ( 1992). High incidence of bronchospasm with regular administration of aerosolized pentamidine. Chest  101, 79– 81. Google Scholar CrossRef Search ADS PubMed  Kuper C. F., Gröllers-Mulderij M., Maarschalkerweerd T., Meulendijks N. M. M., Reus A., van Acker F., Zondervan-van den Beuken E. K., Wouters M. E. L., Bijlsma S., Kooter I. M. ( 2015). Toxicity assessment of aggregated/agglomerated cerium oxide nanoparticles in an in vitro 3D airway model: The influence of mucociliary clearance. Toxicol. In Vitro  29, 389– 397. Google Scholar CrossRef Search ADS PubMed  McDougall C. M., Blaylock M. G., Douglas J. G., Brooker R. J., Helms P. J., Walsh G. M. ( 2008). Nasal epithelial cells as surrogates for bronchial epithelial cells in airway inflammation studies. Am. J. Resp. Cell Mol. Biol . 39, 560– 568. Google Scholar CrossRef Search ADS   Nasra J., Belvisi M. G. ( 2009). Modulation of sensory nerve function and the cough reflex: Understanding disease pathogenesis. Pharmacol. Ther . 124, 354– 375. http://dx.doi.org/10.1016/j.pharmthera.2009.09.006 Google Scholar CrossRef Search ADS PubMed  NDA 019887. ( 2010). United States Food and Drug Administration. Data file retrieved from https://www.accessdata.fda.gov/drugsatfda_docs/label/2011/019887s014lbl.pdf, last accessed November 28, 2017. Owen K., Beck S. L., Damment S. J. P. ( 2010). The preclinical toxicology of salmeterol hydroxynaphthoate. Hum. Exp. Toxicol . 29, 393– 407. http://dx.doi.org/10.1177/0960327110363335 Google Scholar CrossRef Search ADS PubMed  Pencina M. J., D' Agostino R. B., D' Agostino R. B., Vasan R. S. ( 2008). Evaluating the added predictive ability of a new marker: From area under the ROC curve to reclassification and beyond. Stat. Med . 27, 157– 172. Google Scholar CrossRef Search ADS PubMed  Rennard S. I. G., Basset D., Lecossier K. M., O’Donell P., Pinkston P. G. M., Crystal R. G. ( 1986). Estimation of volume of epithelial lining fluid recovered by lavage using urea as marker of dilution. J. Appl. Physiol . 60, 532– 538. Google Scholar CrossRef Search ADS PubMed  Rothen-Rutishauser B., Blank F., Mühlfeld C., Gehr P. ( 2008). In vitro models of the human epithelial airway barrier to study the toxic potential of particulate matter. Expert Opin. Drug Metab. Toxicol . 4, 1075– 1089. Google Scholar CrossRef Search ADS PubMed  Srinivasan B., Kolli A. R., Esch M. B., Abaci H. E., Shuler M. L., Hickman J. J. ( 2015). TEER measurement techniques for in vitro barrier model systems. J. Lab. Autom . 20, 107– 126. Google Scholar CrossRef Search ADS PubMed  Thavagnanam S., Parker J. C., McBrien M. E., Skibinski G., Shields M. D., Heaney L. G. ( 2014). Nasal epithelial cells can act as a physiological surrogate for paediatric asthma studies. PLoS One  9, e85802. Google Scholar CrossRef Search ADS PubMed  Toronto Aerosolized Pentamidine Study (TAPS) Group. ( 1990). Acute pulmonary effects of aerosolized pentamidine: A randomized controlled study. Chest  98, 907– 910. CrossRef Search ADS PubMed  Tepper J., Ochoa R., Rix P., Elliott G., Hoglen N., Poulin D., Parsley E., Masamune H. ( 2014). A 26-week toxicity assessment of AIR001 (sodium nitrite) by inhalation exposure in rats and by intravenous administration in dogs. Int. J. Toxicol . 33, 162– 174. Google Scholar CrossRef Search ADS PubMed  Wain L. V., Shrine N., Artigas M. S., Erzurumluoglu A. M., Noyvert B., Bossini-Castillo L., Obeidat M., Henry A. P., Portelli M. A., Hall R. J., et al.   ( 2017). Genome-wide association analyses for lung function and chronic obstructive pulmonary disease identify new loci and potential druggable targets. Nat. Genet . 49, 416– 425. Google Scholar CrossRef Search ADS PubMed  Youden W. J. ( 1950). Index for rating diagnostic tests. Cancer  3, 32– 35. http://dx.doi.org/10.1002/1097-0142(1950)3:1<32::AID-CNCR2820030106>3.0.CO;2-3 Google Scholar CrossRef Search ADS PubMed  Zavala J., O'Brien B., Lichtveld K., Sexton K. G., Rusyn I., Jaspers I., Vizuete W. ( 2016). Assessment of biological responses of EpiAirway 3-D cell constructs versus A549 cells for determining toxicity of ambient air pollution. Inhal. Toxicol . 28, 251– 259. Google Scholar CrossRef Search ADS PubMed  Zhang J. H., Chung T. D. Y., Oldenburg K. R. ( 1999). A simple statistical parameter for use in evaluation and validation of high throughput screening assays. J. Biomol. Screen.  4, 67– 73. http://dx.doi.org/10.1177/108705719900400206 Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press on behalf of the Society of Toxicology. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com

Journal

Toxicological SciencesOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve Freelancer

DeepDyve Pro

Price
FREE
$49/month

$360/year
Save searches from
Google Scholar,
PubMed
Create lists to
organize your research
Export lists, citations
Read DeepDyve articles
Abstract access only
Unlimited access to over
18 million full-text articles
Print
20 pages/month
PDF Discount
20% off