Regulatory frameworks are moving towards risk assessment approaches that better protect human health and are not reliant on testing in animals. Therefore, 21st century science is incorporating the use of human-relevant methods that are ethical, scientifically sound, and can accurately predict the toxicity of chemicals. In silico models that consider human-relevant parameters as well as in vitro methods that vary in complexity—spanning from mono-to co-culture systems—are already being used to predict human outcomes. For example, the Organisation for Economic Co-operation and Development (OECD) (OECD, 2021) uses combined information from several sources (e.g. in silico predictions, in chemico, in vitro data) to predict pathological outcomes in humans in response to chemical exposure. Anchored to known mechanisms of human toxicity such mechanism-based approaches enable us to understand whether a chemical will be toxic and through which pathway(s) it may act to cause the adverse outcome (AO) (Clippinger et al., 2018).

Adverse outcome pathways (AOP) are a means to organize known information related to a pathological outcome and understand the mechanism leading to the adverse effect. Starting with a molecular initiating event (MIE) and ending in an AO, AOPs are a sequence of causally linked key events (KE) that span different levels of biological organization—from the molecular to the whole organism level (Ankley et al., 2010). An AOP may not necessarily include every single event that contributes to the development of the AO, but it does include all KEs that are critical for its development (OECD, 2017). In vitro and in silico assays that measure each of the KEs of an AOP can be used to design testing approaches that closely predict human responses and replace the need for in vivo data in order to derive benchmark values for determining the potential adverse health impacts of chemicals. When combined with existing data and physicochemical information related to the test substance, AOP-based testing can help develop integrated approaches that can predict human responses (OECD, 2021). Given the usefulness of AOPs in the risk assessment of chemicals, the OECD launched a program for the development of AOPs in 2012; called the AOP-Wiki, this program is overseen by the Extended Advisory Group on Molecular Screening and Toxicogenomics (EAGMST). Several online resources, including the OECD AOP users’ handbook, are available to aid developers in compiling AOPs on the AOPwiki (OECD, 2018). There are currently more than 300 AOPs online, at various stages of completion. One of these is AOP148 [EGFR Activation Leading to Decreased Lung Function, https://aopwiki.org/aops/148], which is extended and complemented by the AOP network described here.

With a surface area of ∼100 m² and ventilated by 10,000 to 20,000 L of air per day (National Research Council, 1988; Frohlich et al., 2016), the lungs are a major barrier that protect the body from a host of external factors that enter the respiratory system and may cause lung pathologies. Mucociliary clearance (MCC) is a key aspect of the innate immune defense against airborne pathogens and inhaled particles. MCC is governed by the concerted action of its functional components, the cilia and the airway surface liquid (ASL), where the latter comprises mucus and the periciliary layer (PCL) (Bustamante-Marin and Ostrowski, 2017). Healthy subjects produce >10 ml airway secretions daily (King, 2006), which are continuously transported by the mucociliary escalator. Disturbances in any of the processes that regulate ASL volume, mucus production, mucus viscoelastic properties, or ciliary function can cause MCC dysfunction and are linked to airway diseases such as chronic obstructive pulmonary disease (COPD) and asthma, both of which bear a significant risk of increased morbidity and mortality. The mechanism by which exposure to inhaled toxicants might lead to mucus hypersecretion and thereby impact pulmonary function has already been mapped in AOP148 on decreased lung function. However, whether an exposure-related decline in lung function is solely related to excessive production of mucus is debatable, particularly in light of the close relationship between mucus, cilia function, and efficient MCC. To date, no single event has been attributed to MCC impairment. This AOP work evaluates the major processes known to be involved in ensuring efficient MCC and consists of sequences of KEs that culminate in the modulation of ASL, ciliary beat frequency (CBF), and mucus viscosity. Together, these processes impair MCC, which—when persistent—leads to decreased lung function. Evidence was gathered from the peer-reviewed literature from multiple sources (e.g., PubMed, Web of Science, Scopus) by keyword searches. No publication date limit was applied. Both empirical and quantitative evidence was captured, consolidated and transferred to the corresponding KE and KER pages on the AOPwiki following the recommendations in the AOP User’s Handbook.

The epithelium of the respiratory tract has a powerful defense mechanism against airborne pollutants, owing to the combined performance of mucus-producing goblet cells and ciliated cells that are covered with microtubular projections called cilia. In response to various irritants and pathogens, goblet cells produce and secrete mucus, and the cilia sweep the mucus upward through coordinated beating motions, thus clearing the airways of these substances—a process which is termed MCC. Optimal MCC is dependent on multiple factors, including cilia number and structure, ASL height ¹ , and the physical and chemical properties of mucus. Any disturbances in these factors can lead to impaired MCC. A summary schematic of the AOP network delineating processes that lead from oxidative stress to decreased lung function is presented in Figure 1 and detailed on the AOPwiki (https://aopwiki.org/aops/411, https://aopwiki.org/aops/424, https://aopwiki.org/aops/425).The MIE for this network of AOP is oxidative stress. Oxidative stress is generally regarded as a redox imbalance characterized by the increased production of oxidative species and concurrent depletion of antioxidant defenses. Thus, the overall redox balance of the cell/tissue is tipped in favor of oxidation. Various highly reactive species, collectively referred to as “reactive oxygen species” (ROS) or “reactive nitrogen species” (RNS), are formed continuously at relatively low concentrations during the normal biochemical functioning of cells and tissues. They are highly unstable because they contain unpaired electrons capable of initiating oxidation reactions and include free radicals such as hydroxyl radicals, superoxide anions, oxygen radicals, nitric oxide, and non-free radicals, such as hydrogen peroxide, peroxynitrite and hypochlorous acid (Rahman et al., 2006). However, upon exposure to certain xenobiotics or in the presence of pathogens, cells may form excessive ROS/RNS, which may react with cellular components such as proteins, lipids and nuclear material, leading to the dysfunction of these components and, ultimately, cell death and disease manifestation (Halliwell and Aruoma, 1991; Berlett and Stadtman, 1997). Protective enzymes such as catalase, glutathione peroxidase, superoxide dismutase, and thioredoxin—in combination with radical scavengers such as glutathione, ascorbic acid, uric acid and vitamin E—work in concert to maintain ROS/RNS levels that are not overly damaging to cells and cellular systems (Rahman et al., 2006).

In the lungs, free radical species may be endogenously produced or introduced following exposure to exogenous sources, such as air pollutants, inhaled chemicals/therapeutics, and cigarette smoke (Church and Pryor, 1985). The main cellular sources of reactive species in the lungs include neutrophils, eosinophils, alveolar macrophages, alveolar epithelial cells, bronchial epithelial cells, and endothelial cells (Holland et al., 1990; Kinnula et al., 1992; Kinnula et al., 1995); these cells may increase their ROS/RNS production in response to infection or tissue damage. ROS/RNS generally inflict their effects by remodeling extracellular matrix and stimulating mucus secretion and repair responses (Poli and Parola, 1997). Oxidative stress can lead to a variety of respiratory diseases, such as asthma, acute respiratory distress syndrome and COPD (Rahman and MacNee, 1996; Chabot et al., 1998).With respect to this specific AOP, localized oxidative stress in the airways as a result of cigarette smoke exposure, for example, can cause damage to various proteins linked to the regulation of cilia function. Reduced expression of the CFTR (cystic fibrosis transmembrane conductance regulator) transcript, diminished CFTR protein levels, and altered chloride (Cl⁻) channel gating lead to acquired CFTR dysfunction (Clunes et al., 2012; Braun, 2014), which perturbs the height of the ASL and facilitates cilia collapse. Furthermore, oxidative damage has been reported to decrease the FOXJ1 (forkhead box protein J1) gene and protein expression, a critical protein involved in the assembly of motile cilia (Milara et al., 2012; Brekman et al., 2014; Garcia-Arcos et al., 2016; Valencia-Gattas et al., 2016; Ishikawa and Ito, 2017). Collectively, these perturbations result in decreased MCC from the upper airways.

CFTR is a multi-domain membrane protein belonging to the large family of adenine nucleotide-binding cassette transporters (Riordan, 2008). It is an integral membrane glycoprotein which functions as cyclic adenosine monophosphate (cAMP)-activated and phosphorylation-regulated Cl⁻ channel at the apical membrane of epithelial cells (Farinha et al., 2013). In respiratory epithelia, CFTR mediates fluid and electrolyte transport, and its function is critical to ASL homeostasis. Exposure to inhaled oxidants leads to decreased CFTR gene and protein expression as well as CFTR internalization, which reduces protein presentation at the membrane and reduces or abolishes short-circuit currents (Cantin et al., 2006a; Cantin et al., 2006b; Clunes et al., 2012; Sloane et al., 2012; Rasmussen et al., 2014). Decreased CFTR expression (KE1) in airway epithelium has been observed in cystic fibrosis and after hypoxia and cigarette smoke exposure, resulting in reduced Cl⁻ transport and, ultimately, reduced ASL depth (Alexander et al., 2012; Clunes et al., 2012; Rasmussen et al., 2014; Woodworth, 2015; Raju et al., 2016).

The ASL is a liquid layer on the apical side of the respiratory epithelium, reportedly between 5 and 100 μm in depth (Widdicombe and Widdicombe, 1995). It consists of an inner aqueous PCL, which spans the length of the cilia, and an outer gel-like mucus layer. The PCL has a low viscosity and enables ciliary beating, thereby facilitating the movement of the outer mucus layer toward the glottis and, ultimately, its removal by cough or ingestion (Antunes and Cohen, 2007). Both ASL composition and height are considered critical for its function (Fischer and Widdicombe, 2006). Under physiological conditions, ASL composition and height are regulated through vectorial transport of electrolytes, driven by transepithelial transport and apical secretion of Cl⁻ by (predominantly) CFTR, which results in passive water secretion and, consequently, increased ASL height. Absorption of sodium ions (Na⁺) on the apical side by the epithelial sodium channel (ENaC) and its interaction with the basolateral Na⁺/K⁺-ATPase leads to net absorption of Na⁺, which in turn drives fluid absorption and therefore decreases ASL height (KE2) (Hollenhorst et al., 2011; Althaus, 2013). Impairment of CFTR or ENaC function can lead to the dysfunction of the other ion channel (increased CFTR activity leads to decreased ENaC activity and vice versa) (Hobbs et al., 2013; Munkholm and Mortensen, 2014), resulting in perturbation of ASL height.

The number, structure, and cohesive beating of the motile cilia lining the upper and lower respiratory tract are critical for efficient MCC. Motile cilia are microtubular organelles, 6–7 µm long and 0.2–0.3 µm in diameter (Brooks and Wallingford, 2014; Yaghi and Dolovich, 2016). They protrude from the cell surface and generate directional flow of fluid though coordinated beating. Approximately 50–80% of the human respiratory epithelium is comprised of ciliated cells; each ciliated cell is covered by more than a hundred motile cilia, which move mucus upwards (together with mucus-trapped substances) upward (Yaghi and Dolovich, 2016; Bustamante-Marin and Ostrowski, 2017). Cilia formation is initiated and coordinated by a distinct gene expression program, led by the transcription factor FOXJ1 (Brody et al., 2000; Zhou and Roy, 2015). The multiple motile cilia assembly factors MCIDAS (multiciliate differentiation and DNA synthesis associated cell cycle protein) and GMNC (geminin coiled-coil domain containing) converge in positively regulating FOXJ1 (Stubbs et al., 2012; Arbi et al., 2016; Berta et al., 2016), whereas NOTCH (Notch homolog (Drosophila))-, IL-13 (interleukin-13)- or EGF (epidermal growth factor)-triggered signaling antagonizes FOXJ1-driven multiciliogenesis (Gomperts et al., 2007; Shaykhiev et al., 2013; Gerovac et al., 2014; Gerovac and Fregien, 2016). Although various other factors are involved in multiple motile cilia assembly—including MYB (MYB proto-oncogene), RFX3 (regulatory factor X3), ULK4 (Unc-51 like kinase 4), Wnt signaling, and others—they mostly act upstream or in parallel to FOXJ1 (Tan et al., 2013; Choksi et al., 2014; Liu et al., 2016; Schmid et al., 2017). FOXJ1 appears to be the major factor in multiciliogenesis, whereby its activity is necessary and also sufficient for programming cells to assemble functional motile cilia (Vij et al., 2012; Zhou and Roy, 2015). It is not surprising, therefore, that a decrease in FOXJ1 levels (KE3) inhibits ciliogenesis in multiciliated cells in zebrafish and Xenopus (Stubbs et al., 2008), and knockdown of FOXJ1 results in almost complete absence of cilia in mouse epithelial cells (Chen et al., 1998; Brody et al., 2000). FOXJ1 expression also decreases in cigarette smoke extract-treated human airway epithelial cells, leading to suppression of cilia growth, which can be restored by overexpression of the protein (Brekman et al., 2014).

Because ciliated cell density and the multiple motile cilia length and number per cell correlate with CBF—which is routinely used as a predictor of MCC efficiency (King, 2006)—it follows that, if cilia numbers decrease (KE4), CBF decreases (KE6). Cohesive beating of multiple motile cilia with a specific frequency and pattern propels mucus (and trapped particles or pathogens) upwards, creating a continuous movement (Chilvers and O'Callaghan, 2000). CBF is influenced by several factors, including structural modulation in the cilia and the concentrations of the cyclic nucleotides cAMP and cGMP and intracellular calcium (Ca²⁺) (Rubin, 2002). CBF also depends on the physical and chemical properties of the ASL. If ASL height decreases following, for example, exposure to cigarette smoke, the cilia cannot extend to their full height, and MCC efficiency will drop. In addition, reduced ASL height results in airway dehydration, which increases mucus viscosity (KE5) (Gheber et al., 1998; Lai et al., 2009; Fahy and Dickey, 2010). Increased mucus viscosity, in turn, decreases CBF and slows the transport of mucus on the mucociliary escalator (i.e., decreases MCC; KE7). In chronic inflammatory states, as seen (for example) in the lungs of cystic fibrosis, asthma, or COPD patients, decreased MCC can lead to mucus impaction, resulting in the formation of mucus plugs, which then in turn obstruct the airways and, consequently, lead to decreased lung function (AO) over time (Wanner et al., 1996; Szczesniak et al., 2017; Dunican et al., 2021).

Table 1 presents a summary of supporting evidence for each of the KERs in this AOP. KERs are rated as “strong”, “moderate”, or “weak” on the basis of empirical evidence supporting a change in an upstream KE (KEup) leading to an appropriate change in the immediate downstream KE (KEdown). Other considerations are whether KEups occur at lower doses, earlier time points, and at a higher incidence than KEdowns and if there are any inconsistences in the published data. The experimental evidence for a causal relationship between the KEup and KEdown in this AOP has been provisionally rated as “moderate” or “strong” in most cases.

Exposure to inhaled oxidants, such as cigarette smoke and ozone, leads to decreased CFTR gene and protein expression as well as CFTR internalization (KE1), thereby reducing or abolishing open probabilities, short-circuit currents and subsequently ASL height/volume (KE2) (Kulka et al., 2005; Cantin et al., 2006a; Cantin et al., 2006b; Qu et al., 2009; Clunes et al., 2012; Sloane et al., 2012; Rasmussen et al., 2014). Both reduced mRNA stability (Cantin et al., 2006a) and decreased transcription rates (Bargon et al., 1992a; Bargon et al., 1992b; Rasmussen et al., 2014) reportedly contribute to diminished CFTR mRNA expression. CFTR expression was also modulated by STAT1 (Kulka et al., 2005; Qu et al., 2009) and Nrf2 signaling (Zhang et al., 2015). Additionally, on the post-transcriptional level, CFTR function has been shown to be affected by oxidative stress (Clunes et al., 2012) and ischemia (Brézillon et al., 1997; Bodas et al., 2017).

Serous and glandular secretions of the airway epithelium contribute to the ASL, and epithelial ion channel function is critical to ASL homeostasis. Absorption of liquid to and from the mucus layer serves to maintain ASL depth. The regulation of these reabsorption processes is complex and not fully elucidated (Boucher, 2004). Experimental evidence suggests that the balance between Na⁺ absorption and Cl⁻ secretion—mediated by ENaC and CFTR, respectively—plays a major role in these processed, with the ion channels affecting each other’s activity (Boucher, 2003; Boucher, 2004; Schmid et al., 2011). Impaired functioning of the CFTR and ENaC ion channels results in enhanced Na⁺ absorption, reduced Cl⁻ secretion, and consequently, reduced ASL height (KE2). This phenomenon is well known not only from studies in models of cystic fibrosis and acquired CFTR deficiency—even though the exact mechanism of the interaction between these two channels remains to be elucidated (Tarran et al., 2001; Boucher, 2003; Zhang et al., 2013; Hassan et al., 2014; Rasmussen et al., 2014; Woodworth, 2015; Raju et al., 2016)—but also from studies with pharmacological agents that enhance CFTR expression and/or function or perturb the interaction between CFTR and ENaC (Van Goor et al., 2009; Van Goor et al., 2011; Lambert et al., 2014).

Under physiological conditions, ASL height is adjusted to the appropriate height, which helps maintain the PCL depth at approximately the length of the cilia (Antunes and Cohen, 2007). If the airways become “dehydrated” (i.e., the ASL height decreases; KE2), the cilia collapse and ciliary movement is slowed or inhibited (KE6) (Matsui et al., 1998; Tarran et al., 2001; Knowles and Boucher, 2002; Munkholm and Mortensen, 2014). Decreased ASL height also contributes to increased mucus viscosity (KE5), a phenomenon that is well described in cystic fibrosis, where CFTR defect results in decreased ASL height, leading to decreased MCC (KE7) and subsequent mucus plugging (Birket et al., 2014; Birket et al., 2016; Birket et al., 2018).

Free radicals such as super oxides, hydroxyl radicals, and hydrogen peroxides are a common factor in various respiratory diseases, such as acute respiratory distress syndrome, asthma and pneumonia. Oxidative stress (such as that caused by cigarette smoke exposure or irradiation) leads to decreased FOXJ1 gene and protein expression (KE3) as well as to decreased FOXJ1 target gene expression (Milara et al., 2012; Brekman et al., 2014; Garcia-Arcos et al., 2016; Valencia-Gattas et al., 2016; Ishikawa and Ito, 2017). Because FOXJ1 is a key factor of multiple motile cilia assembly in the respiratory airways (Zhou and Roy, 2015), oxidative stress blocks the multiciliogenesis program, which is necessary and also sufficient to program cells to grow functional motile cilia (Hua et al., 2010; Vij et al., 2012). Studies in different model organisms have shown that the loss of FOXJ1 (KE3) results in a loss of multiple motile cilia (KE4) (Chen et al., 1998; Brody et al., 2000; Stubbs et al., 2008; Vij et al., 2012).

Cilia in the respiratory epithelium beat in a coordinated fashion at a frequency of approximately 7–16 Hz, propelling mucus upwards (Joki et al., 1998; Smith et al., 2012; Jing et al., 2017). Many factors have been shown to affect ciliary function, including cilia length, number, structure, orientation, and distribution as well as mucus viscosity, temperature, pH, chemicals, ASL height, and exposure to bacterial and viral pathogens (Kanthakumar et al., 1996; Clary-Meinesz et al., 1998; Joki et al., 1998; Ho et al., 2001; Mall, 2008; Smith et al., 2012; Jing et al., 2017; Snyder et al., 2017). Alterations in normal physiological conditions and healthy cilia number/length/structure (KE4) as well as oxidative stress through exposure to hydrogen peroxide or free radicals typically reduce CBF (KE6) (Burman and Martin, 1986; Clary-Meinesz et al., 1998; Min et al., 1999; Jayathilake et al., 2012).Synchronized ciliary action helps transport mucus from the lungs to the mouth, where it is swallowed or expectorated (Munkholm and Mortensen, 2014). In addition to ASL and mucus properties, the speed of mucus movement—and hence the effectiveness of MCC—is dependent on ciliary amplitude and beat frequency (Rubin, 2002). Aside from genetic defects leading to ciliopathies, there is ample evidence that prolonged exposure to noxious agents, such as cigarette smoke, nitrogen oxide and sulfur dioxide, causes a decrease in CBF (KE6) and, subsequently, MCC (KE7) (Knorst et al., 1994; Yeates et al., 1997; Kakinoki et al., 1998; Cohen et al., 2009; Schmid et al., 2015). CBF also seems to be dependent on mucus viscosity, with the CBF decreasing with increasing viscosity in animal models (Andrade et al., 2005; Kikuchi et al., 2017). This linear correlation between CBF (KE7) and mucus viscosity (KE5) has also been confirmed in mathematical models simulating the two-layer mucociliary transport process (Lee et al., 2011; Sedaghat et al., 2016).

Finally, the link between decreased MCC and decreased lung function (AO) is well established through observations in patients with ciliary defects (e.g., primary ciliary dyskinesia) and cystic fibrosis. Failure to clear mucus from the lungs causes mucus build up, which can lead to mucus plugging in the airways and, consequently, leads to decreased lung function over time (Mossberg et al., 1978; Regnis et al., 1994; Wanner et al., 1996; Robinson and Bye, 2002; Kerem et al., 2014; Szczesniak et al., 2017). Mucus plugging due to decreased MCC is also considered a major cause of airway obstruction and airflow limitation in COPD patients (Okajima et al., 2020; Dunican et al., 2021) and asthmatics (Maxwell, 1985; Kuyper et al., 2003).

Here, we have presented an AOP that links oxidative stress resulting from inhalation exposure to toxicants to impaired lung function via a decrease in MCC. Given the individual and public health burden of the consequences of lung function impairment, gaining a greater understanding of the underlying mechanisms of this pathology is extremely important in the risk assessment of inhaled toxic chemicals. There is strong empirical evidence to support several of the KERs in this AOP, particularly at the cellular level (i.e., oxidative stress leading to decreased CFTR function [KERA], decreased CFTR function leading to decreased ASL height [KERB], and oxidative stress leading to decreased CBF [KERJ]). However, additional evidence on causality is required to attribute stronger plausibility to KERs such as that between decreased ASL height and decreased CBF (KERC), which we evaluated as being weak. Future studies, using in vitro assays such as those outlined in Supplementary Table S1, that directly assess this linkage as well as the KERs we determined to have moderate plausibility (i.e., KERD, KERE, KERF, KERH, KERI, KERK, and KERL) will help greatly strengthen this AOP overall.

An integrated assessment of substances with the potential to be inhaled, either intentionally or unintentionally, could incorporate inhalation exposure and dosimetry modelling to inform an in vitro assessment approach with appropriate exposure techniques and cell systems for assessing the KEs in this AOP (EPA’s Office of Chemical Safety and Pollution Prevention, 2019). Standardization and robustness testing of assays against explicit performance criteria using suitable reference materials can greatly increase the level of confidence in their use for KE assessment (Petersen EJ. et al., 2021; Petersen E. J. et al., 2021). Much of the empirical evidence that supports the KERs in the qualitative AOP described here was obtained from in vitro studies using well-established methodologies for biological endpoint assessment (Supplementary Table S1). Being chemical-agnostic, this AOP can be applied to a variety of substances that share the AO. For example, impaired MCC and decreased lung function have a long-known relationship with smoking, but little is known about the consequences of the long-term use of alternative inhaled nicotine delivery products such as electronic cigarettes and heated tobacco products. This AOP can form the basis of an assessment strategy for evaluating the effects of exposure to aerosol from these products on the basis of the KEs identified here.

AOPs such as this one can play a central role in risk assessment strategies for a wide variety of regulatory purposes by providing mechanistic support to an integrated approach to testing and assessment (IATA; (Clippinger et al., 2018)) or defined approach (DA). IATAs are flexible frameworks that can be adapted to best address the regulatory question or purpose at hand. Unlike the assessment process within IATA that involves some level of expert judgement, DA uses rule-based fixed data interpretation procedure. Both DA and IATAs are a means to integrate existing data on a chemical (e.g., physicochemical properties and ADME [absorption, distribution, metabolism, and excretion] information) with an AOP-based in vitro testing strategy to generate data that does not currently exist (Willett, 2019). An important feature of these approaches is that they should also include a measure of uncertainty to facilitate regulatory decision-making. AOPs can be used in an iterative fashion to identify and reduce or resolve, where possible, areas of uncertainty by generating data to fill those knowledge gaps. Furthermore, a quantitative AOP could provide data that would be translated to prediction models for human risk assessment through the application of in vitro to in vivo extrapolation (IVIVE) approaches.