The breathtaking world of human respiratory in vitro models: Investigating lung diseases and infections in 3D models, organoids, and lung‐on‐chip

The COVID‐19 pandemic illustrated an urgent need for sophisticated, human tissue models to rapidly test and develop effective treatment options against this newly emerging severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2). Thus, in particular, the last 3 years faced an extensive boost in respiratory and pulmonary model development. Nowadays, 3D models, organoids and lung‐on‐chip, respiratory models in perfusion, or precision‐cut lung slices are used to study complex research questions in human primary cells. These models provide physiologically relevant systems for studying SARS‐CoV‐2 and, of course, other respiratory pathogens, but they are, too, suited for studying lung pathologies, such as CF, chronic obstructive pulmonary disease, or asthma, in more detail in terms of viral infection. With these models, the cornerstone has been laid for further advancing the organs by, for example, inclusion of several immune cell types or humoral immune components, combination with other organs in microfluidic organ‐on‐chip devices, standardization and harmonization of the devices for reliable and reproducible drug and vaccine testing in high throughput.


Introduction
The respiratory tract and lungs are engaged in many essential tasks within our body, such as respiration, pulmonary circulation, and immunity.To improve our understanding of (patho-) mechanisms in the respiratory tract and lung, reliable physiological and pathological models are needed.Over the decades, animal models have been instrumental in elucidating biochemical and physiologic processes in cancer and infectious diseases.The unacceptably high failure rates in clinical drug development and the poor translatability from preclinical animal data to humans as well as the gap between animal and human biology and human-specific pathogens ask for a paradigm shift to advanced and qualified human cell-based cultures [1][2][3].
Correspondence: Prof. Doris Wilflingseder and Prof. Wilfried Posch e-mail: doris.wilflingseder@i-med.ac.at; wilfried.posch@i-med.ac.at 3D cultures of the target cells overcome the limitation of 2D cultures/monolayers as shown by various organoids generated from primary cells isolated from patients with cystic fibrosis (CF) and various tumors despite a drift toward clonality seen after a few weeks -but less so as with monolayer cultures [4].The advanced understanding of disease pathwaysmolecular biology, molecular pathology, and biochemistry in particular -has already stimulated the development of different 3D human-based disease models.These illustrate important signatures of various human diseases [5] and, in particular, in vitro respiratory and lung models experienced a boom due to the COVID-19 pandemic.COVID-19 served as a revelation to overcome the obstacles of lacking adequate human laboratory models.Appropriate in vitro models of respiratory tract and lungs allow characterizing initial events upon pathogen transmission at infection sites and a deeper understanding of underlying mechanisms as well as developing strategies for fast and innovative drug testing to minimize the spread of the virus, as illustrated for SARS-CoV-2.Not only SARS-CoV-2 but also respiratory tract infections in general account for approximately three to five million deaths per year [6].Besides respiratory tract infections, other lung diseases, for instance, chronic obstructive pulmonary disease (COPD), CF, or malignant tumors, such as lung cancer, are the leading cause of death in many countries (https://www.who.int/publications/i/item/9789240047761).
Despite the rapid development of highly differentiated respiratory and lung models during the last years that provide valuable information on humoral and cellular responses at barrier sites, these models often still lack essential factors, such as immune cells, humoral immune components, or lung-associated microbiota.Another issue is the broad variation of not only human diseases but also their disease models that need to be harmonized and validated.In this review, we described the current respiratory and lung models and their cultural conditions and highlighted their advances and future challenges.

Cell culture of human airway epithelial cells
Nasal, tracheal, bronchial, small airway, and alveolar epithelial cells from cell lines or of primary origin can be cultured under submerged conditions, in air-liquid interface (ALI), as lung on a chip or in organoid shape (Fig. 1).These cell culture models contain epithelial cells from their particular region of the human respiratory tract/lung in vivo.Club, ciliated, basal, and goblet cells are the main epithelial cells found in the scRNA-seq analysis of the human lung (for a detailed description of the different cell types in the distinct tissue compartments see [7]).

Submerged and ALI culture
Primary human airway epithelial (HAE) cells demonstrate a promising source to investigate human lung tissue in different culture conditions (Table 1).HAE cells under submerged conditions are cultivated on plastic typically coated with extracellular matrix proteins such as collagens or cellulose [8,9].2D submerged cultivation of HAE cells results in mainly basal cells, which can be found in the entire airway tree, and is accompanied by the loss of differentiated ciliated and secretory cells.This cultivation does not allow gas exchange and therefore, dependent on the research question, the protocol rarely reflects the in vivo situation.In order to represent a more realistic model, the ALI culture has been developed.Therefore, HAE cells are seeded on a matrix proteincoated microporous membrane, where the apical side of the cell layer is exposed to air while the basolateral side is submerged in a medium [10].The ALI culture allows HAE cells to differentiate into a mucociliary pseudostratified epithelial culture.The 3D ALI culture contains many different functional cell types such as basal, ciliated, and mucus-producing goblet cells [10], while cells isolated from the distal part of the lung like small airway epithelial (SAE) cells contain mainly alveolar epithelial type I (AT1) and AT2 cells [11].HAE cells cultivated in ALI display a good representation of the in vivo situation [12].Another advantage of the ALI culture is the possibility of exposing the cells to pathogens or airborne substances [13,14].

Respiratory models in perfusion
Microphysiological systems (Table 1) composed of interacting organs-on-chip or connected 3D tissue constructs provide more physiologically relevant alternatives to monolayer cultures or animal studies, in particular if studying the interactions of human pathogens at barrier sites.In vitro microfluidic models of the upper and lower respiratory tract [19] down to the alveolus, including dynamic biomechanics to mimic respiration and/or vascularization [15].These microfluidic organ-on-chips show continuous development and enable inclusion of physical, mechanical, and organizational features of the respiratory tract/lung microenvironment [16][17][18].Perfused conditions were illustrated to significantly accelerate ciliogenesis and mucus production of upper and lower respiratory tract epithelial cells (HAE and SAE cells) compared with static cultures, which was inter alia due to the constant supply of nutrients and growth factors, enhanced cell-cell communication due to flow [19].

Organoids
Organoids are 3D structures, which self-organize into airway-like tissue structures (Fig. 2, Table 1) [8].They recapitulation of the essential attributes of their counterpart organs [19].The respiratory cells originate from stem/ progenitor cells (from adult or embryonic tissue, or from iPSC) [19].ASC-derived organoids are typically polarized, cystic structures that consist only of the epithelial organ compartment and express lower complexity compared with most PSC-derived organoids, which also express nonepithelial cells [20].The source of the stem cells has a dramatic influence on the present cell types in lung organoids [21].Organoids derived from hPSC express a more alveolar signature with mainly AT1 and AT2 cells.In contrast, lung organoids extracted from adult tissue via biopsies or bronchoalveolar lavage fluid present basal, club, goblet, and ciliated cells [19].Lung organoids express less basal cells compared with HAE cells but more club and ciliated cells [21].Nevertheless, the coexistence of proximal and distal cell types in lung organoids is characteristic of all extraction methods independent of passage numbers [21].Organoids are embedded in a matrix (e.g.Matrigel) or cultivated in a 3D cell culture system like the Celvivo system.The generated organoids maintain an internally orientated ciliated apical surface.This orientation displays experimental challenges.Therefore, methods to reverse the epithelial polarity in order to face the apical surface outward were developed [22].To improve proximal differentiation, ciliary differentiation can be induced in the alveolar organoids [19].The proximal differentiation increases the  presence of ciliated, basal, and goblet cells while the club cells were decreased.
The human organoid culture can also be differentiated in transwell inserts at ALI or as monolayer in submerged cultures [23].Cultivation as monolayer results in a distal signature with mainly AT1 cells, suggesting that cultivation in monolayers favors the differentiation of AT2 into AT1 cells [21].In contrast, the ALI culture of lung organoids consists of ciliated, goblet, and basal cells resulting in a proximal pseudostratified mucociliary epithelium [21].
Apical-out airway and lung organoids for investigating airborne respiratory challenges under more physiological conditions can be generated in an extracellular matrix-free environ-ment in suspension culture (Fig. 2).Airway organoids can be further differentiated in suspension to develop ciliated and mucusproducing robust organoids [14,24].Such apical-out organoids are highly suited for testing apically applied drugs, for example, antivirals, in high-throughput [25,26].

Lung-on-chip
The model "lung-on-chip" (Table 1) is used for complex 3D model systems consisting of primary cells, cell lines, pluripotent stem cells (PSCs), or adult stem cells (ASCs) on engineered microdevices [17].This model enables to mimic the in vivo environment through the presence of multi-cellular tissues, which can be combined with electronic sensors to monitor organ and tissue function in the real time [17].Further advantages of the "lung-onchip" system are the combination of engineered microtissues coupled with microfluidic devices (microphysiological systems, see below) and therefore allowing mechanical forces such as stretch and shear, fluid flow, biochemical cues, and electrical or optical signals [27][28][29].

Precision-cut lung slices
Precision-cut lung slices (Table 1) can be cultured from explanted human lungs and therefore precision-cut lung slices contain all cell types as well as the extracellular matrix [30].Due to the maintenance of the complex microarchitecture and functional response, this model enables ex vivo mechanistic studies [31,32].The short duration of viability and functionally is often reported to be only 7-10 days, and the changes in tissue architecture over time display limitations of this model [32].

Co-culture with immune components
HAE or lung organoids cultured on a transwell -submerged or in ALI -can be equipped with other cells of interest, like immune cells.It is increasingly acknowledged that the respiratory immune system plays a fundamental role in epithelial bar-rier integrity and lung homeostasis.Even disruption of the epithelium alone results, for example, upon infection with SARS-CoV-2, in massive production of innate immune components leading to inflammation or attraction of immune cells to infection sites [14].Nonimmune epithelial cells are able to release innate immune responses like intracellular complement and anaphylatoxin or cytokine secretion.An analysis of SARS-CoV-2-infected HAE cells showed a high intracellular complement mobilization and an increased C5a as well as proinflammatory cytokine production [14].Thus, incorporation of immune cells or characterizing humoral immune components (complement products, cytokines) at infection sites is a prerequisite to narrowing the gap between in vitro models and the situation in the host.During initial pathogen invasion, mainly dendritic cells (DCs), macrophages, NK cells, and neutrophils act synergistically at the epithelial entry sites to promote airway inflammation, cytokine, and anaphylatoxin release and lysis of infected cells [14,[33][34][35].Moreover, DCs as sentinels of our immune system take up and process Ag and undergo a change from Ag-catching immature DCs at barrier sites to Agpresenting mature DCs migrating to the proximate lymph node to initiate adaptive immunity [36].Thus, the incorporation of these known modulators of respiratory barrier functions into in vitro models is vital to mimic the situation in the host even more closely.Immune cells are typically added on the basolateral side and therefore interact with the epithelial cells via direct cell-cell contact or cytokine release.The addition of immune cells, such as DCs, macrophages, neutrophils, T, or B cells, mimic the complex cellular network within the lung in a more realistic way and examples of co-cultures are neutrophils with HAE cells [37], macrophages with HAE [38] or SAE cells [33], or dendritic cells with HAE cells [34].These co-culture models are especially of interest to study infection at entry sites as illustrated in Fig. 3 mimicking Aspergillus fumigatus interactions within the barrierimmune cell model.Co-culture of airway epithelium with peripheral blood mononuclear cells allows analysis of the modulation of innate and adaptive immune subsets (NK, monocytes, γδ T cells, CD4, and CD8 T cells) during an infection over a time period of 6 days [39].Of note, the incorporation of immune cells like NK or CD8 T cells requires the matching of the HLA phenotype between immune cells and epithelial cells.Table 1 summarizes the properties, advantages, disadvantages, and examples for applications of the described human in vitro lung models.

SARS-CoV-2-related diseases and infections in respiratory/lung models
Primary human nasal, bronchial, and small airway epithelial cells for submerged or ALI cultures as well as organoids are not only commercially available for culturing under healthy conditions but also for various prevalent lung disorders such as CF, COPD, allergic rhinitis, as well as for smokers' tissue.Moreover, stem cells from patient biological samples (bronchoalveolar lavage, biopsies, cell brushings) can be isolated and cultured as self-organizing organoids [5,40].Thus, various features of the disease itself can be compared with healthy tissue equiva-lents, or differences/similarities between microbial infection or any other respiratory challenge can be studied.Healthy versus diseased models can provide information on tissue features, diseaseassociated gene expression, mucociliary clearance, ciliary beat frequency, and inflammatory mechanisms, among others.Therefore, such models are valuable tools for understanding pathologies, reactions to challenges, or treatment options in more detail.The culture of diseased cells is similar to that of healthy ones.Here, we focus on CF and COPD models as well as SARS-CoV-2 and Aspergillus spp. that comprise diseases and infections with a worldwide impact and interconnected pathology mechanisms, for example, during COVID-19-associated pulmonary Aspergillosis (CAPA).

Cystic fibrosis
CF is a chronic lung disorder often accompanied by a plethora of comorbidities due to viral infections, for example, influenza A (H1N1) or co-infections with other bacteria, that is, Pseudomonas aeruginosa [41,42].In the beginning, when SARS-CoV-2 came up, CF patients were advised to protect themselves and self-isolate to avoid coronavirus-mediated worse outcomes in this cohort (rev. in [43]).However, emerging data indicated that they neither had worse clinical outcomes compared with healthy cohorts nor did they illustrate a higher infection rate.Up to date, multiple hypotheses exist to explain these unexpected outcomes related to CF patients in terms of COVID-19 severity and human respiratory models might elucidate some of the mechanisms behind these phenomena.Characterization of -for example -CF versus healthy, differentiated, pseudostratified HAE epithelia revealed a significantly reduced expression of the SARS-CoV-2 receptor angiotensin-converting enzyme 2 (ACE2) on diseased cells (own unpublished results and [44]).Of course, other factors, such as cellular processes, CF medications, and patient-specific features might contribute, can be assessed within in vitro respiratory models.Nevertheless, reduced expression of the ACE2 receptor on diseased cells would explain, at least in part, the lower incidence of individuals suffering from CF for severe COVID-19 compared with healthy individuals, which was reviewed in [43].

Chronic obstructive pulmonary disease
COPD is a progressive and chronic inflammatory disorder affecting all parts of the lung including airways, vasculature, and parenchyma [45].Airway remodeling is fibrotic and irreversible and disease progression is influenced by environmental factors, with smoking being the worst [46].Using in vitro primary models of COPD can provide information on inflammatory mechanisms upon microbial infection, high throughput drug testing, and development in healthy versus diseased organoids.COPD is one of the co-morbidities associated with severe COVID-19, which is partially explained by an elevated expression of ACE2 on small airway epithelium [47,48].Microbial stimuli of COPD small airway chips showed increased secretion of IL-8 and M-CSF, both cytokines stimulating the attraction of neutrophils and macrophages in COPD patients, compared with healthy chips [49].These data also highlight that the diseased phenotype of the tissue is maintained in the models.

Infections
SARS-CoV-2's emergence of COVID-19 boosted the further development of diverse respiratory and lung models, including immune cells, vasculature, and fluidic systems.There are extensive variations in protocols, which will need harmonization in the future to guarantee the highest reproducibility and prediction of data.Nevertheless, current models are sound and provide easy-to-infect tools for evaluating mechanisms at barrier sites for not only SARS-CoV-2 but also other respiratory challenges.SARS-CoV and SARS-CoV-2 variants infect epithelial cells of the upper and lower respiratory tract via binding of spike protein to the host cell receptor, ACE2, and cleavage by host cell proteases [50,51].Inflammatory and cell modulating mechanisms as well as novel therapeutic options mediated at epithelial barrier sites by respiratory viral challenge can be assessed by infection of 3D models of nasal epithelium, upper and lower respiratory tract, or lung.This gives rapid information on the one hand on regional similar-ities/differences and cell types infected; on the other hand, they provide information on the effectiveness of treatment regimen as extensively described by us and others [14,21,26,40,[52][53][54][55][56][57].

Aspergillus spp
Aspergillus spp. is a saprophytic, ubiquitous fungus, which is inhaled daily at huge levels by humans.Healthy hosts are able to successfully clear the fungi by mucociliary clearance and tissueresident phagocytes [58].In particular, immune-suppressed individuals are prone to "Aspergillus-related" diseases, such as allergic bronchopulmonary aspergillosis, aspergillomas, or invasive aspergillosis as reviewed in Lass-Flörl et al. [59].Recently, CAPA raised concerns about contributing to an increased mortality rate during COVID-19, and a broad spectrum of Aspergillus species was detected in a laboratory-based study on multiple biomarker testing of CAPA [60].3D models of the respiratory tract and lung are highly suited to assess the first interactions of Aspergillus species with the human host and co-cultures with immune cells like DCs and macrophages revealed full functionality upon infection with the fungus [33,34,61].In co-culture studies with macrophages, Luvanda et al. [34] illustrated that corticosteroids like dexamethasone create an immunosuppressive microenvironment, thus promoting enhanced A. fumigatus growth and invasion.Moreover, the authors illustrated that this went along with corticosteroid-induced macrophage M2 polarization within the immune-competent tissue environment.Thus, such models are highly suited for studying first encounters with fungal pathogens and characterize the effects of antifungals in terms of epithelial immune modulation.

COVID-19-associated pulmonary Aspergillosis
The outbreak of the current SARS-CoV-2 pandemic led to an increase in intensive care patients with severe pulmonary disorders [62].Thus, several reports of CAPA have raised concerns that this superinfection contributes to increased mortality [62].Since reports from CAPA varied among hospitals and countries with incidences between 3% and 33% [63,64], a laboratory-based study on multiple biomarker testing in the diagnosis of CAPA was performed by Lass-Flörl et al. [60].In general, 3D respiratory models are also appropriate for studying SARS-CoV-2/Aspergillus coinfections at the interface of environment/lung/immunity in more detail.

Outlook
In particular, due to the COVID-19 pandemics the last 3 years faced an extensive boost in respiratory and pulmonary model development.These models provide physiologically relevant systems for studying respiratory challenges, such as SARS-CoV-2 and other pathogens, or lung pathologies, such as CF, COPD, or asthma, in more detail.While ALI approaches could serve as platforms for personalized approaches to test saliva or serum for their neutralizing capacities of respiratory challenges following vaccination or convalescence, organoids emerged as promising tools for drug efficacy by high-throughput screens.With these models the cornerstone has been laid for further advancing the organs by, for example, the inclusion of several immune cell types or humoral immune components, combination with other organs in microfluidic organ-on-chip devices, standardization and harmonization of the devices for reliable and reproducible drug and vaccine testing in high throughput.Due to the accelerated development of complex, physiologically relevant, and personalized respiratory models from healthy and diseased states, these models are promising approaches to obtain clinically relevant data on human (patho-)physiology during infection and disease.Moreover, these models might also in the near future contribute to drastic reductions or -even better -replacement of animals for drug or vaccine testing purposes, for detailed molecular characterization of (patho-)physiologic mechanisms within host cells to find new targets against infectious challenges.

Figure 1 .
Figure 1.Standardized primary epithelial barrier models exist for the nasal, upper, and lower respiratory tract and lung.

Figure 3 .
Figure 3. Interactions of DCs (green) with Aspergillus fumigatus (red) within a differentiated respiratory tissue model (white).While in uninfected controls (left), CFSE-labelled DCs remained basolaterally located over time, the cells sensed fungal conidia (red) already after 1 h (upper panel), colocalized with these (yellow) after 3 h (middle panel), and were then found alongside fungal hyphae after 24 h.Respiratory tissues were destroyed by fungal toxins after 24 h.Bars represent 20 μm.

Table 1 .
Properties, advantages, disadvantages, and examples for applications of the various respiratory models.