Investigating pulmonary neuroendocrine cells in human respiratory diseases with airway models

Pulmonary neuroendocrine cells (PNECs) are specialised cells that line the airways of the mammalian lung (Fig. 1). These cells are extremely rare compared to other lung epithelial cells and contribute to only ∼0.5% of the airway epithelium (Travaglini et al., 2020). They were first reported in the scientific literature in the late 1940s, when they were described as ‘Helle Zellen’ (‘bright cells’ in German), and were subsequently found to hold secretory, dense-core vesicles (see Glossary, Box 1) bearing bioactive compounds (Feyrter, 1954; Frohlich, 1949). PNECs are neuroendocrine cells resident in the lung, recognised to be critical airway sensors. Tissue-resident neuroendocrine cells of the pancreas and the intestine sense inputs from the environment and communicate with their surroundings and with other organ systems, such as the immune and nervous systems, to coordinate biological responses (Gunawardene et al., 2011; Röder et al., 2016). By definition, neuroendocrine cells are bifunctional epithelial cells that, in addition to having an endocrine function, have neuronal characteristics and express markers associated with the central nervous system. Canonical PNEC markers include achaete-scute family 1 [ASCL1, encoding a basic helix-loop-helix (bHLH) transcription factor] and chromogranin A (CHGA). Much like neurons, PNECs can process bioactive amines, such as serotonin (5-HT), and produce polypeptide hormones, such as bombesin (encoded by GRP) and calcitonin gene-related peptide 1 (CGRP-1, encoded by CALCA; hereafter CGRP) (Modlin et al., 2006; Garg et al., 2019). Their characteristic distribution, often found at branching points of the airway tree, is considered strategic for sensing environmental cues. They are thought to process information from the surrounding environment, orchestrating local (organ) and systemic (organismic) responses in homeostasis and disease. This cell population has recently emerged as being a multifaceted player involved in chemosensing, regeneration and immune regulation in the lung (Noguchi et al., 2020).

Considering how rare they are in the lung epithelium, PNECs play an outsized role in respiratory disease. Increases in either PNEC numbers or in their secreted neuropeptides have been reported in a variety of respiratory diseases, including chronic obstructive pulmonary disease (COPD), asthma and cystic fibrosis (CF) (Johnson et al., 1988; Xu et al., 1999). PNECs also play a role in cancer. They are the primary cell of origin for small-cell lung cancer (SCLC), which accounts for 25% of aggressive lung cancers (Sutherland et al., 2011; Park et al., 2011). Likewise, PNECs are the presumed cells of origin for other subtypes of pulmonary neuroendocrine cancers, such as the less common large-cell neuroendocrine carcinoma (LCNEC) and lung neuroendocrine tumours or carcinoids (Box 1) (Rekhtman, 2022).

Studies in mice and experiments using in vitro human lung developmental models and RNA sequencing of human fetal lung tissue have helped to shed light on PNEC development. During mouse and human embryonic development, the lung evaginates from the primitive endodermal tube (Box 1), growing from a primary bud stage into a tree-like airway system via a process called branching morphogenesis. This resulting airway system contains a remarkably diverse population of epithelial cells (Fig. 2). The progeny of the cells in the bud tip give rise to all of the differentiated, canonical cell types of the proximal airways (basal cells, club cells, goblet cells, multiciliated cells and PNECs) and of the distal airways (alveolar type 1 and type 2 cells) (Rawlins et al., 2009). Several studies using mouse models have identified Ascl1 and hairy and enhancer of split 1 (Hes1), members of the bHLH transcription factor family, as primary players in the regulation of PNEC differentiation from lung epithelial progenitors (Box 2) (Ito et al., 2000; Borges et al., 1997). PNECs are the first of these mature cell types to be specified in both the human and murine respiratory epithelium (Kuo and Krasnow, 2015; Sunday, 1996). Expression of Ascl1 during embryonic mouse development is detected at embryonic day (E) 13.5 (Borges et al., 1997), whereas several studies observed PNECs in human fetal airways at 8-9 weeks of gestation (Sunday, 1996), before any other mature epithelial cell types appear in the lung.

The multiple progenitor cell states that arise from the bud tip have not been entirely delineated. There is evidence from lineage tracing studies in mice that PNECs in the trachea and intrapulmonary airways (bronchi and bronchioles) can arise from cells that express the basal cell marker TP63 (also known as p63) at E9.5 (Yang et al., 2018a). Nevertheless, loss of these TP63⁺ cells in Tp63 knockout mice does not influence the number of PNECs detected in the intrapulmonary airways at E18.5 (Yang et al., 2018a). Thus, TP63⁺ cells at E9.5 can give rise to PNECs, but TP63 expression is not required for PNEC formation in the developing mouse lung.

Song et al. (2012) provided evidence that intrapulmonary PNECs are derived from a CGRP-expressing precursor that can also give rise to alveolar cells early in mouse development. Lineage tracing of CGRP-expressing cells between E12.5 and E14.5 led to the labelling of intrapulmonary PNECs and of a small fraction of alveolar cells. By contrast, no alveolar cells were labelled when the same lineage tracing was performed at or after E15.5 (Song et al., 2012). Thus, during early mouse development, PNECs share a common progenitor with alveolar cells. In a recently published study, Conchola et al. (2023) identified a group of cells in fetal human airways that were enriched between 8 and 17 weeks post conception. These cells express a unique combination of genes, including SCGB3A2, SFTPB and CFTR, markers canonically associated with secretory, alveolar and ionocyte cell types, respectively. This population of cells was referred to as lower airway progenitors (LAP) and was shown to give rise to PNECs in vitro in lung organoid models (Conchola et al., 2023). It would be of interest to determine whether LAP cells can also give rise to alveolar cells and whether they represent similar progenitors to those observed by Song et al. (2012) in the mouse airways.

Collectively, these studies in mouse models and in cultured human fetal cells indicate that during lung development, PNECs can emerge from both basal cells and LAP cells. In the developing mouse lung, PNECs originating from basal cells appear to be limited to the tracheal region of the lung (Yang et al., 2018a). In the developing human lung, PNECs derived from LAP cells are likely to populate both intrapulmonary and tracheal airways, although this has not been explicitly shown. There is considerably less known about how PNECs are maintained throughout adulthood.

The question of whether PNECs arise throughout adulthood from neuroendocrine cell proliferation or postmitotic differentiation of non-neuroendocrine cells has not been resolved. Studies labelling PNECs in the murine airway have demonstrated a very low rate (1-2%) of PNEC proliferation (Song et al., 2012). Similarly, in humans, the proliferative fraction of PNECs observed from healthy patient autopsies is 1 to 2% (Boers et al., 1996). Such investigations, coupled with the understanding that PNECs have the capability to give rise to other PNECs in the context of airway injury (Ouadah et al., 2019), suggest that PNEC proliferation could contribute to PNEC maintenance during both injury and homeostasis.

Nevertheless, there is evidence from both mouse and human model systems that non-neuroendocrine cells, such as basal cells, can give rise to PNECs in adult airways. By combining in vivo lineage tracing with single-cell RNA sequencing (scRNA-seq), Montoro et al. (2018) demonstrated that in the mouse tracheal epithelium, PNECs are consistently and directly replenished by basal progenitor cells. Shivaraju et al. (2021) showed that after exposure to hypoxia (Box 1), lineage-traced basal cell progenitors in the murine trachea gave rise to PNECs. Based on pseudotime trajectory analysis of single-cell multiomics data of airway organoids derived from human distal lung tissue, it has been proposed that basal cells might undergo differentiation into PNECs also in adult human lungs (Lee et al., 2023). Lineage-tracing experiments in this human organoid model are needed to confirm the basal cell to PNEC trajectory proposed by pseudotime analysis.

Considering the diverse origins of PNECs during development, it is conceivable that distinct progenitor populations capable of giving rise to PNECs exist in the adult lung. Indeed, a progenitor population similar to fetal LAP cells has been identified in terminal respiratory bronchioles by two research groups (Kadur Lakshminarasimha Murthy et al., 2022; Basil et al., 2022). These progenitors have been demonstrated to serve as progenitors of alveolar cells. Given their resemblance to LAP progenitors, it would be intriguing to investigate, using in vitro models, whether this population could also generate PNECs.

After their specification during lung development, PNECs are found either as scattered solitary cells or enriched in small clusters, referred to as neuroepithelial bodies (NEBs). NEBs are highly innervated and are often found at the branching points of airway tubules (Kuo and Krasnow, 2015; Lauweryns and Peuskens, 1972; Stahlman et al., 1987) (Fig. 1B). Live imaging and PNEC lineage-tracing studies revealed that during embryonic mouse lung development, NEBs are formed through the migration of individual PNECs to airway branch points, where they cluster together (Kuo and Krasnow, 2015; Noguchi et al., 2015) (Fig. 3). After cluster formation, afferent and efferent nerve fibres extend and ramify on the NEBs (Lauweryns and Van Lommel, 1987). In mice, NEBs, rather than solitary PNECs, are selectively supplied by nerves, whereas in humans, there is evidence that solitary PNECs are also innervated (Guha et al., 2012). The innervation of NEBs in this way and their enrichment at bifurcation points where air is trapped suggest that PNECs fulfil a key function as sensors of various stimuli in the lung (Xu et al., 2020).

The function of PNECs as sensors was first described five decades ago. In response to airway hypoxia, increased exocytosis of dense-core vesicles was observed in rabbit lungs (Lauweryns et al., 1977, 1978). Later, the detection of an oxygen-binding protein on the PNEC cell membrane provided further support for the idea that PNECs might function as oxygen sensors in the airway (Youngson et al., 1993). Chemical stimuli have also been shown to trigger responses in PNECs. For example, normal fetal hamster PNECs respond to nicotine by increasing their secretion of 5-HT (Plummer et al., 2000). Human PNECs found in cultures of tracheobronchial epithelium have been shown to express olfactory receptors and to release 5-HT and CGRP in response to volatile chemicals (Gu et al., 2014). In addition to these chemical signals, PNECs can sense mechanical stress. Mechanical stretching of PNECs isolated from rabbit fetal lungs led to the increased release of 5-HT (Pan et al., 2006b). Thus, PNECs can sense different environmental stimuli and respond to these through the release of bioactive compounds.

The bioactive compounds secreted by PNECs influence lung physiology. PNEC signals can, for example, regulate immune system responses. In a mouse model of congenital diaphragmatic hernia with genetic ablation of Robo1 and Robo2, PNEC product secretion was increased, which led to increased infiltration of macrophages and other immune cells into the lung (Branchfield et al., 2016). In a mouse model of allergic asthma, Sui et al. (2018) demonstrated that PNEC-secreted bioactive molecules amplify allergic asthma responses by triggering immune reactions and increasing mucous secretion in the airway. Specifically, PNEC-derived CGRP recruits group 2 innate lymphoid cells to the lung and stimulates their secretion of cytokines. In addition, the neurotransmitter γ-aminobutyric acid (GABA), also secreted by PNECs, promotes goblet cell hyperplasia, leading to increased mucous secretion (Sui et al., 2018).

In addition to immune cells and goblet cells, other cells have been implicated as targets of PNEC signals. Receptors for signalling molecules secreted by PNECs are expressed by a wide array of cell types, including airway epithelial cells, immune cells, neurons, endothelial cells and airway smooth muscle cells, indicating that multiple cell types respond to PNEC-derived signals (Kuo et al., 2022). Myoepithelial cells (Box 1) wrapped around the airway submucosal glands (Box 1) have been recently reported to respond to ATP released from PNECs, triggering myoepithelial contraction and, in turn, promoting mucous ejection (Yu et al., 2022b). Basal cells also respond to CGRP. In the context of hypoxic injury in the mouse airway epithelium, CGRP promoted basal cell proliferation and differentiation into PNECs. Intranasal administration of CGRP was sufficient to mitigate hypoxia-induced airway injury (Shivaraju et al., 2021). This suggests that, in response to lung injury, PNECs regulate protective tissue responses.

Indeed, the NEB microenvironment retains a source of facultative epithelial stem cells that mediate airway repair in response to injury (Hong et al., 2001; Reynolds et al., 2000). Two epithelial stem cell populations have been detected in murine NEBs. One population, called variant club cells (vCCs; Box 1), are resistant to naphthalene injury (Box 1) and selectively express uroplakin-3a (Upk3a) (Guha et al., 2012, 2017). The second population constitutes a subset of murine PNECs with stem cell features (NE^stem cells; Box 1), which selectively proliferate and transdifferentiate to other pulmonary fates following naphthalene injury in vivo (Ouadah et al., 2019). Therefore, PNECs may reflect a heterogenous rather than a homogenous population of cells that harbour different characteristics and perform different functions. Consistent with this, two research groups, through integrated scRNA-seq and spatially resolved transcriptomics of the human embryonic fetal lung, independently characterised distinct groups of PNECs present during early lung development. These groups of PNECs were distributed differently along the airway and were defined as progenitor PNECs, gastrin-releasing peptide (GRP)-positive PNECs and ghrelin (GHRL)-positive PNECs (Sountoulidis et al., 2023; He et al., 2022). In the adult lung, fully differentiated PNECs also display a substantial level of heterogeneity. scRNA-seq analysis of adult mouse and human airways identified an extensive repertoire of neuropeptide and peptide hormone (‘peptidergic’) genes expressed in PNECs, revealing an extraordinary diversity and myriad combinations of such peptidergic genes expressed across PNECs (Kuo et al., 2022). Given the known link between PNECs and cancer, it is tempting to speculate that different subtypes of pulmonary neuroendocrine cancers originate from different PNEC subsets. To address this question, studies focusing on how PNEC neoplasms arise, evolve into specific molecular subgroups and then progress are needed.

Progress in PNEC research has drawn attention to this previously underappreciated, rare cell population of the lung, which is emerging as a relevant modulator of lung physiology. The association of PNECs with lung pathologies, such as chronic respiratory diseases and pulmonary neuroendocrine cancers, indicates that research investigating the pathophysiological role of PNECs could be of therapeutic value for treating lung diseases. In the rest of this Review, we summarise what is known about the association of PNECs with lung pathologies, focusing on how new experimental disease models, such as organoids derived from either human pluripotent stem cells (hPSCs) or tissue stem cells, can help us to better understand the contribution of PNECs to respiratory diseases.

The airway epithelium is constantly exposed to external air and is thus liable to injury from pollutants and pathogens carried by aerosols. The lung has therefore evolved several defence responses to deal with diverse types of injury (Planer and Morrisey, 2023). Chemical insults and viral infection can both lead to an altered cellular milieu and, ultimately, to the remodelling of the airway. The hyperactivation of PNECs resulting in increased secretion of bioactive compounds and/or PNEC hyperplasia has been detected in response to an array of insults, including hypoxia, hyperoxia (Box 1) and smoking (Aguayo et al., 1989; Ratcliffe et al., 2016; Shenberger et al., 1997). In some cases, PNECs appear to play a protective role in the airway epithelium. For instance, as we discussed previously, in response to airway hypoxia, PNECs promote basal cell proliferation, thereby mitigating hypoxia-induced injury (Shivaraju et al., 2021). Likewise, exposure to hyperoxia, which increases airway smooth muscle wall thickness and heightens airway reactivity, is also associated with PNEC hyperplasia (Shenberger et al., 1997). This increase in PNEC numbers, coupled with the accompanying increased neuropeptide production in response to hyperoxia, is thought to be protective and counteract airway hyperresponsiveness. These effects are attributed specifically to PNEC-secreted CGRP, which relaxes tracheal smooth muscle (Bhogal et al., 1994).

In contrast, the presence of too many PNECs can also be damaging. Increased PNEC numbers and/or products have been reported in lung diseases, such as CF, asthma and COPD (Johnson et al., 1988; Xu et al., 1999). Hyperoxia-induced PNEC hyperplasia is thought to mediate lung injury in patients with bronchopulmonary dysplasia, a chronic lung disease associated with oxygen supplementation of premature infants. In a baboon model of bronchopulmonary dysplasia, some of the lung defects associated with the disease could be prevented by treatment of the animals with a GRP-blocking antibody, directly linking PNEC-secreted GRP to the disease phenotype (Sunday et al., 1998). PNECs are also strongly associated with lung neuroendocrine cancers and this association could be linked to the dysregulation of the protective and regenerative functions of this cell population. A better understanding of the role that PNECs play in respiratory diseases and lung cancers could contribute to the development of novel therapeutic strategies that target the neuroendocrine system in the lung. In this section, we review current evidence concerning the contribution of PNECs to respiratory diseases.

CF is caused by mutations in the CF transmembrane conductance regulator (CFTR) gene (Riordan et al., 1989). CFTR is expressed by multiple epithelial cell types and encodes a chloride channel (Plasschaert et al., 2018; Shah et al., 2022). In the human airway, impaired CFTR function results in impaired mucous clearance and host defence. As a result, patients with CF suffer persistent bacterial infections and respiratory insufficiency (Elborn, 2016). It is unclear whether CFTR dysfunction influences PNECs in a cell-intrinsic manner. Other airway epithelial cells, including the recently discovered ionocytes, express CFTR at levels much higher than those observed in PNECs (Okuda et al., 2021). Nevertheless, multiple lines of evidence suggest that PNECs contribute to CF disease pathology.

As is the case for all the respiratory diseases described in this Review, PNEC numbers and their secreted products are increased in the lungs of patients with CF (Johnson et al., 1988). PNEC hyperplasia in CF is presumed to be a response to the chronic hypoxia experienced by these patients. The consequences of increased PNEC numbers in the context of the altered cellular milieu and chronic inflammation present in CF lungs are likely to contribute to the pathology of this disease. A potential problem in the lungs of patients with CF is that persistent mucus obstruction prevents signalling molecules from diffusing properly across the epithelium, thereby preventing PNECs from receiving or responding to these signals (Yu et al., 2022a). PNEC sensing of the inflammatory molecule succinate is compromised in the lungs of pigs with CF (Yu et al., 2022a). PNECs in the submucosal glands respond to succinate by releasing ATP, which stimulates contraction in neighbouring myoepithelial cells, eliciting glandular mucous ejection (Yu et al., 2022a). This mechanism is required to promote host respiratory defences (Yu et al., 2022a). Given that PNEC chemosensory functions are thought to be protective, impaired chemosensing by PNECs in CF could impair host respiratory defence and repair. Indeed, attenuated PNEC communication might be a general theme in the lungs of patients with CF. It has been observed that NEBs of Cftr knockout mice show reduced levels of innervation compared to NEBs from wild-type mice (Pan et al., 2006a). Fewer nerve connections in NEBs could result in blunted PNEC communication with the nervous system. These observations and experimental results in animal models support the hypothesis that PNECs contribute to CF pathogenesis (Fig. 4A). Nevertheless, several questions remain. Is PNEC hyperplasia in CF a response to hypoxic conditions or a mechanism to counterbalance the reduced PNEC chemosensation caused by mucous obstruction? Could CFTR be involved in PNEC development or neuropeptide secretion, and, if so, could this explain the reduced NEB innervation reported in mice lacking CFTR? Further studies are required to address these questions.

Asthma is a chronic disease of the conducting airways that affects over 300 million people worldwide (Vos et al., 2012). It is characterised by inflammation and tissue remodelling as a result of airway hyperreactivity (Papi et al., 2018). Traditionally, two forms of asthma have been defined in the clinic: allergic asthma, resulting from sensitisation to common allergens, involves eosinophilic airway inflammation and hyperreactivity, driven by adaptive T helper 2 cells, and non-allergic (intrinsic) asthma operates independently of the adaptive immunity, with group 2 innate lymphoid cells governing airway inflammation and hyperreactivity (Lambrecht and Hammad, 2015). Cells of the immune system are considered to be the main drivers of disease symptoms (Holgate, 2012); however, this does not exclude a role for the lung epithelium. An altered airway epithelium characterised by PNEC hyperplasia is observed in the lungs of patients with non-allergic asthma (Stanislawski et al., 1981). This increase in PNECs in the lungs of patients with asthma is thought to alter PNEC–immune cell interactions, promoting aberrant immune responses and directly contributing to the disease pathology. Sui et al. (2018) demonstrated that PNECs are essential for allergen-induced asthma-like responses in mice. Ascl1 knockout mice, which lack PNECs, showed a blunted mucosal type 2 response (Box 1) after ovalbumin allergen challenge, which could be rescued by the intratracheal administration of CGRP and GABA (Branchfield et al., 2016). In addition, Branchfield et al. (2016) showed that upregulation of PNEC products in Robo1 and Robo2 double-mutant mice, a genetic model of congenital diaphragmatic hernia, increased immune responses and airway inflammation. Therefore, increased PNEC number in asthmatic lungs can dysregulate immune responses, contributing to asthma pathogenesis.

Given that PNECs are known to contact neurons and are assumed to interact with them, PNEC hyperplasia could also affect PNEC interaction with the nervous system, a known mediator of airway hyperreactivity in asthma. In a preprint, Su et al. (2023 preprint) mapped the neural circuitry involved in establishing allergen-induced airway hyperreactivity in a mouse model of asthma. Interestingly, they showed that Trpv1⁺ neurons, the ablation of which led to reduced airway hyperreactivity in allergen-challenged mice (Tränkner et al., 2014), expressed multiple receptors for PNEC-secreted ligands (Su et al., 2023 preprint). Moreover, Trpv1⁺ neurons have been shown to contact the basal side of individual PNECs (Su et al., 2022). Thus, PNECs may interact with Trpv1⁺ neurons, and PNEC hyperplasia might exacerbate airway hyperreactivity in asthma. Together, these studies suggest that PNECs contribute to asthma pathology (Fig. 4B). Further studies are required to better address whether PNECs are directly involved in establishing airway hyperreactivity and to reveal the detailed mechanisms by which PNECs are activated after an allergen challenge.

COPD is a common respiratory disease and the third leading cause of death worldwide (Vos et al., 2020). It is characterised by the chronic obstruction of the small airways (bronchiolitis) and by alveolar damage (emphysema), which leads to shortness of breath in response to physical exertion (dyspnea) and to a persistent cough (Barnes et al., 2015). The inhalation of cigarette smoke or of other harmful particles is considered a primary driver of COPD, through persistent pulmonary injury that leads to chronic inflammation and tissue remodelling (Mathioudakis et al., 2020). Cigarette smoke has been shown to shift the composition of the airway epithelium – a common feature of most chronic lung diseases – by increasing PNEC numbers (Aguayo et al., 1989; Goldfarbmuren et al., 2020; Gu et al., 2014).

Even though a clear understanding of whether PNEC hyperplasia in smokers' lungs can promote COPD is still lacking, several mechanisms have been proposed. Bombesin-like peptides (BLPs) secreted by PNECs are mitogenic for human bronchial epithelial cells and lung fibroblasts (Aguayo et al., 1990; Willey et al., 1984). As such, they might play a role in the process of airway narrowing, driving airway thickening and peribronchiolar fibrosis in COPD. Supporting this hypothesis, PNEC hyperplasia and increased BLPs in patients who have never smoked appear to cause COPD-associated symptoms of peribronchiolar fibrosis and chronic airflow obstruction (Aguayo et al., 1992). Thus, PNEC hyperplasia might contribute to the development of COPD by promoting aberrant airway remodelling. Further supporting PNEC hyperplasia as a driver of COPD pathogenesis, an increase in PNECs is thought to amplify their chemosensory function, causing heightened airway responsiveness to chemical stimuli. In primary human cell cultures, volatile chemicals have been shown to activate PNECs that express olfactory receptors (Gu et al., 2014). It is tempting to speculate that PNEC hyperplasia in patients with COPD might cause exaggerated reactions to inhaled chemical stimuli that promote airway inflammation, resulting in episodes (called exacerbations) in which symptoms worsen. COPD exacerbations are associated with poor prognosis and have been linked to exposure to air pollutants (Barnes et al., 2015). These observations suggest that PNEC hyperplasia contribute to the development of COPD (Fig. 4C). Studies that address how PNECs respond to air pollutants associated with COPD development and exacerbations are key to understanding how PNECs might contribute to COPD pathogenesis. Future work will also need to investigate whether PNECs are the major contributors that drive airway thickening and airflow obstruction in COPD.

Diffuse idiopathic PNEC hyperplasia (DIPNECH) is a rare pulmonary disorder, first described and defined by Aguayo et al. (1992). Patients with DIPNECH manifest symptoms, including cough, dyspnea, wheezing, and mixed obstructive and restrictive defects in pulmonary function tests. DIPNECH is recognised by the World Health Organization as a preinvasive, possibly pre-neoplastic condition (Travis et al., 2015). In most patients, DIPNECH manifests as an indolent and nonprogressive disorder, although a few patients can progress to severe airflow obstruction or develop metastatic carcinoid tumours (Davies et al., 2007; Flint et al., 2019). The symptoms of DIPNECH are accompanied by primary PNEC proliferation that is detectable on histological analyses. Such neuroendocrine cell proliferations include an increase in scattered single neuroendocrine cells, the presence of small nodules (NEBs) or a linear proliferation of PNECs (Aguayo et al., 1992). When neuroendocrine cell proliferations in DIPNECH extend beyond the basement membrane, they are defined as either tumourlets (<5 mm; Box 1) or carcinoid tumours (>5 mm).

PNEC hyperplasia can also be found in other settings, including in smoking-related diseases, reactive proliferation in primary or metastatic cancers, or in response to infection (Rossi et al., 2016). However, PNEC hyperplasia associated with the clinical symptoms of DIPNECH is considered to be a distinct pathological entity. The most effective treatment for patients with DIPNECH is somatostatin analogues (SSAs), which can be effective in palliating chronic respiratory symptoms (Al-Toubah et al., 2020). It is well known that SSAs prevent gastrointestinal neuroendocrine tumours from secreting bioactive substances (Grozinsky-Glasberg et al., 2008). The fact that SSAs effectively treat cough and dyspnea in patients with DIPNECH suggests that the production of bioactive molecules by PNECs in DIPNECH is what causes these symptoms (Al-Toubah et al., 2020). Although evidence clearly linking DIPNECH with carcinogenesis is still lacking, carcinoid tumours in a patient with proven DIPNECH is a predictor of poorer prognosis. In a cohort of patients with resected carcinoid tumours, the occurrence of a more aggressive type of carcinoid (atypical carcinoid), as well as the rate of mediastinal lymph node invasion, were both significantly higher in patients with DIPNECH than in those without DIPNECH (Prieto et al., 2021). Thus, carcinoids in patients with DIPNECH might represent a separate entity from carcinoid tumours present in those without DIPNECH. It follows then that carcinoids that arise from DIPNECH lesions might represent tumours initiating from a different PNEC than carcinoids arising de novo in the absence of prior DIPNECH. Studies are required to test this hypothesis and to elucidate the underlying mechanisms to explain the higher prevalence of atypical carcinoids in patients with DIPNECH. Future work is also needed to understand what triggers PNEC hyperplasia in patients with DIPNECH and how PNEC proliferations contribute to the development of the pathology of this disease.

One-third of all human lung cancers exhibit signs of neuroendocrine differentiation (Travis, 2010). Lung neuroendocrine neoplasms (NENs) have a wide spectrum of clinical behaviours. They can be divided into four histological variants: SCLCs, LCNECs, atypical carcinoids and typical carcinoids (Fig. 5A) (Rekhtman, 2022). Clinically, SCLCs and LCNECs are high-grade carcinomas (Box 1) with a 5-year survival rate of 5% and 15-25%, respectively (Govindan et al., 2006; Travis et al., 2015). By definition, neuroendocrine carcinomas have a mitotic index (Box 1) of more than 20 (per 2 mm²). The overall 10-year survival for atypical carcinoids, classified as intermediate-grade tumors, ranges from 91% for stage I tumours to 18% for stage IV tumours, whereas for typical carcinoids, classified as low-grade tumors, the 10-year survival ranges from 98% for stage I tumours to 49% for stage IV tumours (Yoon et al., 2019). The mitotic index for carcinoids ranges from 2 to 20 (per 2 mm²).

SCLC is the most common and aggressive form of lung NEN, characterised by near-ubiquitous loss of function of the RB1 and TP53 tumour suppressor genes (George et al., 2015). PNECs are recognised as a cell of origin of SCLC. Mouse studies have demonstrated that inactivation of Rb1 and Tp53 in Cgrp-expressing PNECs leads to SCLC (Sutherland et al., 2011; Song et al., 2012). Moreover, NE^stem cells were shown to proliferate slowly and continuously right after Rb1 and Tp53 deletion in Ascl1-expressing mouse PNECs, suggesting that this subpopulation of PNECs represents tumour-initiating cells for SCLC (Ouadah et al., 2019). Nevertheless, NE^stem cells are unlikely to be the exclusive cell of origin for SCLC; other cells are also considered to give rise to SCLC (Huang et al., 2018; Yang et al., 2018b; Ferone et al., 2020). Highlighting the heterogeneity of SCLC, genomic and transcriptomic profiling of mouse and human SCLC has revealed that SCLC comprises four distinct molecular subtypes (SCLC-A, SCLC-N, SCLC-P and SCLC-Y) characterised by the predominant expression of the ASCL1, NEUROD1, POU2F3 or YAP1 transcription factors (Ireland et al., 2020). Each SCLC subtype is suggested to be determined by the identity of the tumour-initiating cell, as well as by genetic drivers (Yang et al., 2018b). For example, amplification or overexpression of the gene MYC in SCLC is postulated to drive the evolution of SCLC-A into SCLC-N and then to SCLC-Y. Time-course scRNA-seq analysis of ASCL1⁺ SCLC cultured in vitro demonstrated the ability of MYC to promote SCLC subtype evolution, indicating that the SCLC-A, SCLC-N and SCLC-Y subtypes might represent different stages of SCLC progression (Ireland et al., 2020). The SCLC-P subtype has been proposed to arise from another rare chemosensory lung cell type, the tuft cell, which shares the expression of several markers with SCLC-P (Huang et al., 2018). Accordingly, SCLCs in genetically engineered mouse models could not give rise to the POU2F3⁺ SCLC-P variant when the tumour was initiated in either PNECs, club cells or alveolar cells using cell type-specific promoters. SCLC-P only arose in this model when tumours were initiated with a general promoter (Ireland et al., 2020). Further highlighting how the cell of origin can play a role in the evolution of SCLC tumours, Yang et al. (2018b) demonstrated that the same genomic alteration in different cell types drives different gene expression programs that enable metastatic progression. Together, these studies have led to the identification of different therapeutic sensitivities across specific SCLC subtypes (Cardnell et al., 2017; Huang et al., 2018; Mollaoglu et al., 2017; Saunders et al., 2015). However, adopting subtype-specific therapeutic strategies might not represent the most effective approach to treating SCLC, considering the dynamic evolution of this cancer and that individual SCLC tumours comprise cells belonging to different subtypes (Huang et al., 2018; Stewart et al., 2020). Future studies that investigate how SCLC progresses and gives rise to intratumoural heterogeneity should inform the development of new therapeutic strategies.

As with SCLCs, LCNECs and carcinoids are also thought to originate from PNECs (Fisseler-Eckhoff and Demes, 2012). However, less is known about these other histological tumour subtypes. An analysis of mutational and transcriptional patterns in LCNECs found that LCNEC comprises two transcriptionally dissimilar genomic subgroups: type I LCNECs with bi-allelic TP53 along with STK11 and/or KEAP1 alterations, which exhibit a neuroendocrine expression profile, and type II LCNECs, which are enriched for bi-allelic inactivation of TP53 and RB1 and are characterised by the reduced expression of neuroendocrine markers and the upregulation of immune-related pathways (George et al., 2018). Although deleterious mutations in the two tumour suppressors genes TP53 and RB1 are frequently associated with carcinomas, these are rare events within carcinoids, which have a lower mutational burden compared to that of carcinomas, and a higher prevalence of mutations in genes involved in chromatin remodelling (Fernandez-Cuesta et al., 2014). A subgroup of carcinoids, termed ‘supra-carcinoids’, has been identified, which has a carcinoid morphology but molecular characteristics similar to those of LCNEC (Alcala et al., 2019). In this first study, supra-carcinoids accounted for 5.5% of the pulmonary carcinoids analysed (Alcala et al., 2019). Further studies are needed to determine the exact frequency of supra-carcinoids. These observations relating to supra-carcinoids support the hypothesis that, in some rare cases, pulmonary carcinoids can progress into tumours that resemble high-grade malignant carcinomas. The apparently low frequency of this progression event suggests that supra-carcinoids arise from a different cell of origin than the other more common subtypes of carcinoids. More research will be required to explore the clinical implications and specific molecular processes underpinning how PNEC neoplasms arise, evolve into specific molecular subgroups and progress from low- or intermediate- grade tumours into high-grade malignant carcinomas (Fig. 5B). We anticipate that future advancements in in vitro models of the human airway epithelium will prove valuable for investigating the pathological significance of PNECs in respiratory diseases.

The progress of research into PNECs has been significantly hindered by their rarity, the absence of established isolation protocols and the challenges associated with the long-term cultivation of human cells. These challenges are beginning to be addressed through the creation of in vitro models of the human airway, consisting of pseudo-stratified epithelia that contain different lung cell types and recapitulate many features of the in vivo airway epithelium. Cells cultured in these systems can be manipulated to differentiate into PNECs, facilitating the study of these rare cells. Here, we describe several in vitro models of the human airway epithelium (Table 1) that could be used to explore the pathological role of PNECs in respiratory diseases and that might guide the development of new therapies to target the pulmonary neuroendocrine system.

Air–liquid interface (ALI) cultures are the most commonly used models to study the airway epithelium and aim to recapitulate the microenvironment of airway epithelial cells (Whitcutt et al., 1988; Baldassi et al., 2021). In the lung, epithelial cells make up the luminal surface of the respiratory tract and create a tight barrier at the interface between the air in the airways and an intricate capillary network found on the basal side of the airways. ALI systems recreate an air interface and a blood flow-like interface. In ALI culture, cells grown on a porous membrane are exposed to air on their apical surface and are nourished through their basolateral surface, which is in contact with a liquid medium. This configuration allows ciliated and secretory cells to differentiate (Ross et al., 2007; Yamaya et al., 1992), indicating that the exposure of these airway epithelial cells to air promotes their maturation. PNECs have been generated from human induced pluripotent stem cells (hiPSCs) in ALI conditions (Hor et al., 2020). These so-called ‘induced PNECs' could provide researchers with an inexhaustible resource with which to investigate PNEC pathophysiology in lung diseases. These cells are transcriptomically similar to human fetal lung PNECs (Hor et al., 2020). However, their similarity to PNECs in the adult human lung remains unclear, leaving open the question as to their potential application to model pathological conditions of adult respiratory diseases. The generation of PNECs from primary human tracheal cells grown in ALI systems has also been reported (Mou et al., 2021). Although the authors of this study did not address whether these derived PNECs resemble adult human PNECs, it is likely that PNECs derived from primary airway tissue will more accurately resemble their in vivo mature counterparts. Using these tracheal-derived primary cell ALI cultures in combination with CRISPR knockout studies, Goldfarbmuren et al. (2020) investigated lineage relationships among rare airway epithelial cell types: PNECs, tuft cells and ionocytes. Their findings point to a branched lineage progression, in which tuft-like cells give rise to both ionocytes and PNECs. However, lineage-tracing studies are needed to confirm this lineage relationship between tuft cells and PNECs. In a preprint, Waghray et al. (2023 preprint) described an equivalent progenitor population found in the large airways of human adults through deep scRNA-seq (Waghray et al., 2023 preprint). Leveraging human ALI cultures, this replicative rare cell progenitor population was shown to have the potential to differentiate into either mature tuft cells or ionocytes and was therefore termed tuft-ionocyte progenitor (TIP) cell. The authors did not find any evidence that TIP cells could give rise to PNECs. Intriguingly, they showed that cytokines associated with asthma and CF altered the differentiation of TIP cells (Waghray et al., 2023 preprint). Perhaps in the context of other human respiratory diseases, TIP cells could be pathologically redirected to produce PNECs.

ALI culture systems provide a potentially useful in vitro tool with which to study respiratory airways and PNEC biology in healthy and diseased states (Baldassi et al., 2021). However, it is important to note that these systems often require extended differentiation periods before PNECs emerge. Furthermore, the long-term expansion potential of primary or hiPSC-derived PNECs in these systems has yet to be fully determined. To overcome these limitations, further research is needed to optimise the culture techniques used to derive PNECs in ALI culture systems.

Precision-cut lung slices (PCLSs) offer promise as an ex vivo platform for translational research. PCLSs are obtained by infusing freshly retrieved whole lungs or individual lobes with warm low-melting agarose solution, which fills the airway to maintain the expanded state of the lung and prevents tissue damage during the slicing process (Placke, 1987). The solidified infused tissue is then sliced with a vibrating microtome, resulting in slices of uniform thickness (100-500 μm), which can be preserved and studied ex vivo (Liu et al., 2019).

A key advantage of PCLSs is that they retain all cell types present in the original lung tissue, including crucial non-epithelial components, such as stromal cells and immune cells. This provides researchers with the opportunity to explore interactions between the immune system and other resident lung cells (Temann et al., 2017). For instance, PCLSs could be used to investigate how the interactions between PNECs and immune cells influence immune responses and contribute to respiratory diseases, such as asthma and COPD. PCLSs also preserve the spatial cellular architecture of lung tissue, which enables ‘intact’ NEBs from rabbits and mice to be studied in a relatively native environment. Fu et al. (1999, 2002) used PCLSs from rabbit and hamster lungs to study the effect of hypoxia on NEBs. These two studies revealed that NEBs release 5-HT under hypoxic conditions and in response to ATP, and that this response was mediated through the inhibition of voltage-gated K⁺ channels present on NEB cells (Fu et al., 1999, 2002). In another study, De Proost et al. (2008) performed live-cell imaging of NEBs in murine PCLSs to investigate the response of NEBs and surrounding cells to applied stimuli. In response to a high K⁺ concentration, NEB cells and vCCs that encircle them increased their cytoplasmic Ca²⁺, whereas other epithelial cells did not. Remarkably, vCCs exhibited a delayed response in comparison to NEB cells, hinting at an indirect activation mechanism mediated by NEBs (De Proost et al., 2008). Together, these studies highlight the benefit of using PCLS approaches to conduct comparative studies of NEBs, at different developmental stages and in different species, including humans, for which NEB research is limited.

PCLSs have also been used to investigate human respiratory diseases (Liu et al., 2019). However, PCLSs undergo a gradual decline in cellular integrity and functionality over time. PCLSs that have been cultured for a prolonged period reduce their production of cytokines and chemokines when exposed to external stimuli (Liu et al., 2019; Temann et al., 2017). Moreover, after the first week of culture, the airway epithelium in PCLSs undergoes a significant decline in cilia functionality, whereas the alveolar epithelium and vascular system undergo severe cell loss within the first 2 weeks (Preuß et al., 2022). The short window of cellular viability in PCLSs thus prevents the modelling of the long-term effects of respiratory diseases, such as chronic inflammation and cancer progression, and also implies that gene-editing experiments would be nearly impossible or extremely challenging to conduct in this system. The viability of PNECs and their functionality over time in human PCLSs is unclear and future studies will be needed to address this. This will lay the foundation for utilizing human PCLSs as a valuable tool in which to investigate the biology of PNECs and their impact on respiratory diseases.

hPSCs – either embryonic stem cells or hiPSCs – provide an unlimited source of cells that can be differentiated into lung progenitor cells. The ability to generate airway epithelia from hPSCs has provided an effective approach to study lung development, model respiratory diseases and conduct drug screening (Goldsteen et al., 2021). Several differentiation protocols to derive airway epithelial cells from hPSCs in two-dimensional (2D) cultures (Firth et al., 2014; Huang et al., 2014; Mou et al., 2012; Wong et al., 2012) and in three-dimensional (3D) organoids (Chen et al., 2017; Dye et al., 2015; Gotoh et al., 2014; Konishi et al., 2016; McCauley et al., 2017) have been established that aim to recapitulate lung development in vitro through the precise activation and/or inhibition of key signalling pathways in a carefully timed manner.

Airway epithelial lineages arise from progenitors present on growing bud tips during branching morphogenesis (Rawlins et al., 2009), which in turn emerge from the anterior ventral side of the definitive endoderm (Morrisey and Hogan, 2010) (Fig. 2). Therefore, the process of differentiating hPSCs into pulmonary tissue commences with differentiation into the definitive endoderm, then the anterior foregut endoderm (AFE), the ventral AFE and, finally, the specification of lung and airway lineages. The activation of activin A signalling drives the differentiation of hPSCs into the definitive endoderm, which can be further anteriorised and differentiated towards the AFE through the dual inhibition of tansforming growth factor β (TGF-β) and bone morphogenetic protein (BMP) signalling (Green et al., 2011). The subsequent addition of different combinations of Wingless-related integration site (WNT) signals, fibroblast growth factors (FGFs), BMPs, sonic hedgehog protein (SHH), keratinocyte growth factors (KGFs) and retinoic acid can result in the formation of lung progenitors and cell cultures that express a wide range of lung epithelial markers (Firth et al., 2014; Huang et al., 2014; Mou et al., 2012; Wong et al., 2012). Chen et al. (2017) reported the creation of 3D hPSC-derived lung organoids that contained mesoderm and developed into branching structures, resembling the proximodistal architecture and cell lineages of the airway tree. Interestingly, when transplanted under the kidney capsule of immunodeficient mice, these organoids were also able to grow NEBs. Whether these organoids contained significant numbers of PNECs when they were grown in vitro was not reported. It would be of interest to conduct future experiments to investigate whether manipulating Notch signalling, which is known to influence proximal and distal cell fate differentiation (Yamamoto et al., 2017), can promote neuroendocrine fate in these organoids.

Some efforts have been made to specifically differentiate PNECs from 2D and 3D hPSC-derived cultures. As discussed in Box 2, Notch signalling inhibition drives the neuroendocrine fate. Chen et al. (2019) used this knowledge to induce PNECs from hPSC-derived lung progenitors grown in two dimensions. They further demonstrated that the proportion of PNECs in 2D hPSC-derived lung progenitor cultures treated with a Notch inhibitor could be further enriched by shRNA-mediated knockdown of RB1. Reduced levels of RB1 expression resulted in the activation of a transcriptional program in PNECs that resembles that observed in human SCLC, indicating that this culture system might enable modelling of early stages of cancer development (Chen et al., 2019). Notch inhibition has also been used to derive PNECs from hPSCs in monolayer ALI cultures (Hor et al., 2020) and in airway organoids (Konishi et al., 2016). These models could inform our studies and understanding of the role of PNECs in respiratory diseases; however, we first need to better understand to what extent these derived PNECs resemble endogenous PNECs isolated from adult human lung tissue.

To date, several approaches have been used to generate airway organoids from tissue-resident, endogenous stem cells derived from adult or fetal lungs (van der Vaart and Clevers, 2021; Barkauskas et al., 2017). These systems offer a twofold benefit as they possess the self-renewal capacity of hPSCs and are less likely to generate off-target cell types, compared to cultures derived from hPSCs.

Until 2020, PNECs had not been reported in any airway organoids created from lung-derived pluripotent stem cells (Rock et al., 2009; Tan et al., 2017; Nikolić et al., 2017; Hild and Jaffe, 2016; Sachs et al., 2019). This was overcome by Miller et al. (2020) using SMAD signalling, followed by SMAD inhibition, to induce human bud tip progenitor differentiation, which created airway organoids that consisted of all the mature cell types of the proximal airways, including PNECs. The same group also investigated lineage relationships in the developing embryonic lung, applying a single-cell barcode-based lineage-tracing method to track the fate of progenitor cells during airway organoid differentiation in vitro. Their results support a model in which bud tip progenitors give rise to both LAP (see Introduction) and basal cells, which then give rise to distinct cell types. Notably, they observed that LAPs make a greater contribution to the generation of PNECs compared to that by basal cells in this airway organoid model (Conchola et al., 2023). Although these organoid systems can produce PNECs, the precise in vitro conditions responsible for driving the specification of lung progenitors into PNEC precursors and their subsequent maturation into differentiated PNECs remains unclear. Understanding these conditions could provide valuable insights into the lineage specification mechanisms guiding PNEC development during lung development. Strategies to efficiently obtain large numbers of PNECs in fetal and adult tissue-derived organoids could advance the study of this rare cell population in the future.

The intricate ways in which PNECs impact lung pathologies have only recently been understood and most remain largely unexplored. As discussed in this Review, a range of different in vitro airway models now provide us with platforms to investigate PNEC pathophysiology in human respiratory diseases. Below, we outline open questions relevant to the respiratory diseases discussed in this Review and propose how different model systems could be used to address these questions.

Understanding the mechanism by which PNECs are activated by known allergens in asthma and how they subsequently amplify immune responses could enable the design of PNEC-targeted therapies to dampen the aberrant airway inflammatory response observed in asthma. ALI culture systems containing PNECs could be leveraged to explore this question.

ALI culture systems could also be used to investigate the impact of volatile compounds and air pollutants on PNECs in the context of COPD. Such experiments could provide insights into the connection between the increase in exacerbation episodes in patients with COPD and their increased exposure to air pollutants. It would also be interesting to address the extent to which PNEC-secreted products, such as bombesins, contribute to the thickening of airways, leading to airflow obstruction in COPD. Exposing PCLSs to PNEC products may serve to decipher whether these products are sufficient to induce a COPD phenotype in the airways.

Given our limited knowledge regarding the involvement of PNECs in CF, deriving hiPSCs from patients with CF and using them to differentiate airway epithelia in vitro could enable us to investigate this further. For instance, it would be beneficial to investigate the effect of CFTR dysfunction on PNEC function and how this might contribute to the promotion of the pathological lung environment in CF.

There remains an unmet need for therapies specifically targeting lung NENs. Studying the dynamics of PNEC differentiation could contribute to resolving how lung NENs arise, and how they evolve into specific molecular subgroups and progress. A notable future breakthrough in this endeavour would be the successful generation of organoids containing PNECs from adult human tissue, a feat that has yet to be accomplished.

We predict that organoids derived from hPSCs or from tissue stem cells will significantly enhance our understanding of the pathophysiology of PNECs and advance the modelling of human respiratory diseases. Additionally, the implementation of state-of-the-art gene-editing techniques, such as CRISPR/Cas9, will enable researchers to genetically manipulate and label PNECs in vitro to facilitate their study. CRISPR/Cas9 could also be used, for example, to engineer organoid models of pulmonary cancers by introducing cancer-associated genetic alterations found in patients with lung cancer. The same approach could be used to introduce genetic variants that are linked (by genome-wide association studies) to an increased predisposition to respiratory diseases, such as asthma and COPD. Such an approach could assist in uncovering the molecular mechanisms that underlie these predispositions. It might also be possible to use this approach to obtain a better understanding of the factors that induce PNEC hyperplasia and pathology in patients with DIPNECH, by incorporating genetic variant candidates from familial DIPNECH into organoid models. Although airway organoids are valuable models, they are limited in their capacity to represent the complexity of the actual tissue as they lack crucial non-epithelial components. An intriguing, yet unexplored, strategy to address this limitation could be the co-culture of airway organoid-derived cells with PCLSs.

Our understanding of PNEC biology and the contribution of PNECS to respiratory diseases will be enhanced by the development of more precise in vitro models of the airway. As PNECs rely on interactions with non-epithelial cells, including with immune cells, neurons, endothelial cells and smooth muscle cells, there will be a need in the future to integrate these cellular components into existing models. As human airway epithelial models become more sophisticated, they will serve as valuable resources for advancing our understanding of PNEC biology. We predict that this progress will promote the development of new treatment strategies for respiratory diseases targeting PNECs or their secreted products.

N.C. and T.D. were supported by the European Molecular Biology Laboratory (EMBL) Barcelona. We thank Rossana Passannante for their comments on the manuscript.