Alveolar wars: The rise of in vitro models to understand human lung alveolar maintenance, regeneration, and disease

Abstract Diseases such as idiopathic pulmonary fibrosis, chronic obstructive pulmonary disease, and bronchopulmonary dysplasia injure the gas‐exchanging alveoli of the human lung. Animal studies have indicated that dysregulation of alveolar cells, including alveolar type II stem/progenitor cells, is implicated in disease pathogenesis. Due to mouse‐human differences, there has been a desperate need to develop human‐relevant lung models that can more closely recapitulate the human lung during homeostasis, injury repair, and disease. Here we discuss how current single‐cell RNA sequencing studies have increased knowledge of the cellular and molecular composition of human lung alveoli, including the identification of molecular heterogeneity, cellular diversity, and previously unknown cell types, some of which arise specifically during disease. For functional analysis of alveolar cells, in vitro human alveolar organoids established from human pluripotent stem cells, embryonic progenitors, and adult tissue from both healthy and diseased lungs have modeled aspects of the cellular and molecular features of alveolar epithelium. Drawbacks of such systems are highlighted, along with possible solutions. Organoid‐on‐a‐chip and ex vivo systems including precision‐cut lung slices can complement organoid studies by providing further cellular and structural complexity of lung tissues, and have been shown to be invaluable models of human lung disease, while the production of acellular and synthetic scaffolds hold promise in lung transplant efforts. Further improvements to such systems will increase understanding of the underlying biology of human alveolar stem/progenitor cells, and could lead to future therapeutic or pharmacological intervention in patients suffering from end‐stage lung diseases.


| INTRODUCTION
The primary function of the lungs is gas exchange and the site for this is the alveoli that are arranged by acini found in the lung parenchyma regions. There is a significant need to understand the mechanisms of alveolar maintenance and damage repair because damage to the alveolar region is a component of chronic adult lung diseases such as chronic obstructive pulmonary disease (COPD) and idiopathic pulmonary fibrosis (IPF) and a cause of acute respiratory failure in pneumonia and acute respiratory distress syndrome (ARDS). In addition, insufficient generation of alveoli results in various neonatal and infant diseases including bronchopulmonary dysplasia (BPD). Despite the pivotal role of alveoli in lung function and disease, and their clinical burden, the pathogenesis of these diverse diseases is incompletely understood and treatment options for patients remain limited. This is partly due to the lack of model systems that allow us to understand human lung biology and disease.
In this review, we summarize our current knowledge of human lung alveoli from decades of animal studies and recent single-cell RNA sequencing analysis (scRNA-seq) (Figure 1). We also highlight recent advances in the available in vitro and ex vivo human lung alveolar model systems and discuss their potential applications and limitations in therapeutic aspects. F I G U R E 1 Cellular composition of the mouse and human lung. A, Schematic of the adult human lung alveoli. The adult human lung is split into five lobes; three on the right, and two on the left. The distal alveolar region has two main epithelial cell types; surfactant-producing AT2 cells and gas-exchanging AT1 cells. A subtype of AT2 cells, named AT2-signaling or AT2-s, has been suggested to show an enrichment of Wnt pathway genes from scRNA-seq analysis, although their presence needs to be verified. 1 Alveolar macrophages exist within the alveolar space, while two populations of mast cells have recently been identified. Fibroblast heterogeneity also exists, with lipofibroblasts, myofibroblasts, and recently identified alveolar fibroblasts located in the alveoli. Cells with dotted outlines have not yet been fully verified. B, Schematic of the adult mouse lung. The mouse lung is also split into five lobes; four on the right, and one on the left. The mouse distal alveolar region possesses at least two subsets of AT2 cells, with AT2 cells expressing Axin2 (Axin2 + AT2) having increased stem cell activity. 2,3 The bronchoalveolar duct junction is an area of transitional epithelium between the alveoli and distal bronchioles, and contain bronchoalveolar stem cells (BASCs); a cell type that expresses both Sftpc and Scgb1a1, and have been shown to differentiate to alveolar and bronchiolar lineages following bleomycin-and naphthalene-induced lung damage, respectively. 4,5 Such a region does not exist in the human lung. Furthermore, basal cells, although present in the human distal lung, are restricted to the trachea and mainstem bronchi of the mouse lung

Significance statement
Over the last decade, stem cell-derived culture model systems of human lungs have garnered renewed interest, as they recapitulate human lung tissues in a dish. This study summarizes the current concepts and advances in the field of human distal lung alveoli, which is the most critical region for the respiratory function and disease, and thereby has been moving forward so rapidly. Specifically, this study compares the differences in cellular compositions of distal lungs between mouse and human and discusses the current model systems to study maintenance, regeneration, and disease of human lung alveoli, which is difficult to model in animal studies.
preventing the delicate structure of the alveolar sacs from collapsing upon breathing (Figure 2A,B). [6][7][8] AT2 cells also have functions in immune response by having the ability to respond to innate immune stimuli. 9 During development both AT1 and AT2 cells are derived from common multipotent alveolar progenitor cells in the canalicularsaccular phases of human lung development (16-36 postconception weeks), although there is no evidence whether such cells exist in the mature lung. 10,11 The maintenance and regeneration capacity of an adult alveolar epithelium is defined by the presence of AT2 cells which behave as facultative stem cells, with both traditional twodimensional (2D)-cultures of human AT2 cells and later 3D lung organoid studies indicating that AT2 cells can self-renew and differentiate into AT1 cells. 8,9,12,13 Recent work has suggested that there may be an underappreciated heterogeneity in the lung, including within F I G U R E 2 Composition and cellular markers of the healthy human distal lung. A, Representative hematoxylin and eosin (H&E) staining of the healthy human adult distal lung tissue shows open alveolar spaces and thin alveolar walls, with the presence of AT1 and AT2 cells. Scale bar = 100 μm. B, Representative immunofluorescence (IF) staining of the healthy human adult distal lung tissue sections for canonical AT2 marker genes including pro-SFTPC (green, top left and top right), HTII-280 (red, top left and middle), SFTPB (green, top middle), ABCA3 (red, top right; white, bottom middle; pink, bottom right), and LPCAT1 (white, bottom left) and AT1 marker genes including PDPN (red, bottom middle), HTI-56 (green, bottom left), AGER (red, bottom left and right), and CAV1 (green, bottom right). Of note, some AT2 cells do not express HTII-280 (Arrowhead: SFTPB + HTII-280 − cell cluster). Scale bar = 50 μm unless stated otherwise. C, Flow-cytometry analysis cell sorting (FACS) plot of primary human lung cells isolated from a normal background parenchyma lung donor following mechanical and enzymatic tissue dissociation. Cells were analyzed for CD31-APC, CD45-APC, EPCAM-FITC, and HTII-280-PE. AT2 cells represent CD31 − CD45 − EPCAM + HTII-280 + populations, where they consistently represent more than 70% of EPCAM + cells in the distal parenchyma lung tissues. Normal human background tissue was obtained from deidentified lungs of adult donors that were deemed unsuitable for transplantation the AT2 cell population (Figure 1). TM4SF1 + AT2 cells have been suggested to possess better capacity to proliferate and produce AT1 cells when necessary, with increased responsiveness to Wnt signaling demonstrated in human AT2 cell-derived organoid culture. 2 A recent scRNA-seq analysis of selectively enriched epithelial populations from whole human donor lungs also supported the potential heterogeneity of AT2 cells by showing a distinct cluster of AT2 cells, named AT2-signaling, expressing Wnt pathway genes. 1 Additional studies have not reported such AT2 cell subpopulations in their scRNA-seq analysis of whole human lung cells, which may be due to differences in sequencing platforms and cell preparation. 14 However, further validation and phenotypic analysis of these populations is required to understand their functional distinction, if any, in lung maintenance and regeneration. It still remains to be answered: (a) Are certain subpopulations more potent, perhaps having increased capacity for regeneration? (b) Or, do broad AT2 cells have plasticity to be activated upon damage? (c) What are the signals inducing this heterogeneity?
(d) Are specific subsets more prone to become damaged during disease progression? Furthermore, work in the mouse has revealed that airway cells including club cells, bronchioalveolar stem cells (BASCs), and clusters of cells expressing Krt5 contribute to alveolar cells following severe damages, highlighting injury-induced cellular plasticity, but it is currently unknown whether this can also occur in the human lung. 4,5,[15][16][17][18][19] Due to the inability to perform in vivo studies of cells that reside in the human lung alveoli, the establishment of in vitro models such as human lung organoids, and ex vivo cultures such as precision-cut lung slices (PCLS) have been successful in modeling aspects of the human lung alveoli, and were described in further detail later in this review. For example, one such study reported that human AT1 cells can de-differentiate to AT2-like cells in in vitro culture. 20 The lung mesenchyme is an important source of morphogenetic and specification signals, and gives rise to cells including smooth muscle, fibroblasts, and the endothelium ( Figure 1). However, little is known about the cellular diversity and mechanisms of their maintenance, in part due to a lack of defined markers. Traditionally, fibroblasts in the alveoli have been characterized as alveolar fibroblasts and lipofibroblasts, although their exact roles remain to be defined, and there has even been some controversy regarding the presence of lipofibroblasts in the human lung. 21,22 Regional fibroblast heterogeneity has been suggested in the human lung. 23,24 Recent advances in scRNA-seq analysis have begun to prise apart subsets of mesenchymal cells, with multiple distinct stromal cell populations being identified in adult human lungs, including muscle cells, pericytes, and multiple fibroblast populations. COL1A1 + fibroblasts can be split into two subpopulations according to their gene expression profiles gained from scRNA-seq analysis; alveolar fibroblasts, and adventitial fibroblasts localized to vascular adventitia, while two ACTA2-enriched populations comprising myofibroblasts and previously unseen fibromyocytes that exhibit high expression of contractile genes have been observed within proximal lung tissue ( Figure 1A). 1 These cells await further characterization, validation and identification of exact spatial location within the lung. scRNA-seq reported enrichment of Hh target genes in the proximal mesenchyme compared to distal mesenchyme. 25 Differences in contractile forces have also been reported. 26 Such regional differences in stromal populations may partially explain the varied effects of injury and repair in distinct areas of the lung for different respiratory diseases. Unexpected molecular diversity has also been discovered in the endothelium, the cells of which play an important role in vascular homeostasis and allow for efficient gas exchange in the lungs, including two molecularly distinct capillary cell populations located in human alveoli. 1

| Cellular dysfunction in lung diseases
Dysregulation of cellular homeostasis or lack of alveolar structures has been implicated in lung diseases. IPF is an interstitial lung disease characterized by honeycomb lesions, hyperplastic AT2 cells, and fibroblastic foci, the "active lesions" of the disease ( Figure 3A). Prognosis is poor and current treatments are limited, with only two drugs, Pirfenidone and Nintedanib, that possess antifibrotic and antiinflammatory properties, available in the clinic. [32][33][34][35] Although shown to provide a survival advantage to some patients, prevention or reversal of fibrosis has not been achieved, and disease progression is inevitable. 32,33,35 Repeated injury to AT2 cells, possibly in a genetically sensitive background, has been suggested to lead to disease. 36 Additionally, evidence indicating that some familial and sporadic forms of IPF comprise mutations within AT2 cells, such as within the SFTPC gene, provides further support. [37][38][39] Furthermore, telomere shortening within AT2 cells has also been implicated in the disease. 40 One of the key features in IPF is "bronchiolization" of issues such as reliance on exogenous factors for injury initiation and resolution of fibrosis. 38 Therefore, better understanding of human lung alveolar cells is also required to complement animal studies. in vitro models such as lung organoids may prove useful in answering such questions, particularly whether regenerative capacity of AT2 cells in IPF is disrupted, and if so, how to revert their repopulating potential. IL-13 has been implicated in IPF pathogenesis by upregulation in bronchoalveolar lavage fluid of IPF patients. 43 Treatment of human AT2-derived alveolar organoids with IL-13 leads to a reduction of SFTPC + cells. As these organoids were cocultured with supporting stromal cells, it is currently unknown whether the effect of IL-13 is direct, or whether it is acting on the stromal cells. Myofibroblasts are considered to be vital contributors to fibrotic diseases, and activation of fibroblasts to myofibroblasts is thought to be driven by TGFβ signaling, but the exact cellular source remains unknown. A HAS1 hi ECM-producing population has recently been described that localizes to peripheral and subpleural regions of the IPF lung. 42 In addition to fibroblasts, a COL15A1 + endothelial cell population, that is generally observed underlying airways, has been reported in the distal IPF lung close to fibrotic foci and the area of bronchiolization. 31 Phenotypic analysis of these cells needs to be performed, and their role in IPF is currently unknown. Further coculturing these cells with AT2 cells could elucidate their potential roles in modulating AT2 cells in IPF pathogenesis.
COPD encompasses a number of individual progressive lung diseases, including emphysema and chronic bronchitis, and is thought to affect 251 million people globally, with prevalence growing. 44 COPD destroys the alveoli and leads to permanent enlargement of the respiratory airspaces, but disease etiology is unknown, although it is observed that smoking is the leading risk factor in disease initiation. 45 fibroblasts is also observed. 54 Both of these diseases would benefit from the establishment of in vitro models that replicate the histopathological aspects of diseases and are suitable for studying disease mechanisms and therapeutic interventions.

| Lung in a dish: In vitro systems to model lung regeneration and disease
Desperate demands to find a suitable model to complement animal studies and provide more relevance to human biology led to development of 3D organoid cultures which are self-organizing, multicellular structures formed from stem cells. 55 The ability to establish and expand organoids from various human tissues including patients raises the possibility of using organoids in translational applications, such as in vitro disease modeling, personalized therapy and regenerative medicine as well as the establishment of organoid biobanks. 55

| Alveolar organoids
The first human alveolar organoids from adult tissue were formed from AT2 cells isolated by a surface marker HTII-280 and cocultured with MRC5 fibroblasts. 12  activation of Wnt signaling has been found to promote AT2 maturation. 62 However, culture of hPSCs in Collagen I gels along with GSK3 inhibition using the small molecule CHIR99021 (CHIR) promoted proliferation and inhibited differentiation, whereas withdrawal of the inhibition induced multilineage maturation of proximal and distal fates. 66 This finding could not be recapitulated using the Wnt ligand WNT3a, perhaps suggesting a method of action other than canonical Wnt signaling. This work contradicts previous studies, in which CHIR removal drove a proximal fate and continued presence led to AT2 cell fate, but may be due to differences in experimental set-up. 59,62 In addition to studying cell regulation, organoids can also be used to model disease. Using hPSCs, organoids were produced using CRISPR/Cas9 that successfully modeled Hermansky-Pudlak syndrome for improved coculture models. Furthermore, lung-on-a-chip technology will enable analysis of multiple cell type interactions in microfluidic devices that will better recapitulate the complexity of the lung. found that respiratory sensitizers such as HClpt increase TNF-α and IL1-α levels. 83 Furthermore, gene transfer to human lung epithelium has been assessed using PCLS from macroscopically normal human lung tissues, where a LacZ-expressing adenovirus reporter was instilled into bronchioles, and traced following 4 days in culture. 84 The use of PCLS allowed visualization of β-galactosidase in the lungs, providing the potential tissue-relevant preclinical model of gene therapy.

| Precision-cut lung slices
Despite the benefits, PCLS remain controversial, with some questioning how closely they recapitulate the in vivo tissue, in addition to their reliability during long-term culture, an important consideration, particularly in applications such as drug testing of slowly metabolized chemicals. 85 Furthermore, they are typically a static system, making analysis of breathing-related diseases such as BPD difficult, although strategies have been developed that apply stretch to cultures, through methods such as suturing slices to a flexible membrane. 86 Cell trafficking from the blood into the lungs, and vice versa, cannot be assessed, although future engineering efforts, particularly using technologies such as "organ-on-a-chip" may be able to solve this limitation. Recruitable immune cells can also not be analyzed, and heterogeneity can exist in different regions even within a single lobe, although this could be rectified in part by sampling multiple regions.
Donor variation is also an issue due to differences in genetic background, although the same problem arises in all models that utilize human tissue, and highlights the potential of personalized therapies.
As tissue viability dramatically declines following isolation, improvements to cryopreservation efforts have allowed for immediate freezing and long-term storage of tissue that can be used later in PCLS with no decline in cell viability and only slight decline in metabolic activity. 87

| Biology meets engineering
Recapitulation of structural and cellular complexity of the lung encourages the use of bioengineering approaches to gain better microenvironmental control, and have led to the development of organ-or lung-on-a-chip technologies. In addition to designing cellcell interactions, 'on-a-chip' technology has also provided the opportunity to better model tissue-tissue and multiorgan interactions. One of their benefits is that mechanical forces, such as stretch and airflow can be implemented, and their effects on cell behavior monitored.
This is an important consideration in creating more physiologically relevant cell models, particularly in the lung, as the cellular response of AT2 cells to the mechanical tension during alveolar development and regeneration of mouse lungs has been described. 88 Figure 2B). Furthermore, HTII-280 is a less useful marker for subculture of AT2 cells due to loss of expression during culture. 67 Alternative isolation strategies are therefore currently under investigation.
Lysotracker, a fluorescent dye that labels acidic components within lysosomes, has been shown to successfully label the lamellar bodies of AT2 cells, allowing for both live cell imaging and selection of high-expressing cells using fluorescence-activated cell sorting (FACS). 61,67,80 More recently, NaPi2b, a sodium phosphate cotransporter that is highly expressed on the surface of AT2 cells, has been shown to isolate a HTII-280 + equivalent population of SFTPC + cells. 67 This same marker was also used for passaging hPSC-derived organoids, where it was found to be more useful for subculturing AT2 cells than HTII-280. 67 Furthermore, MHCII has been previously used to isolate AT2 cells from mouse lungs, and the MHCII antigens HLA-DR and HLA-DP are expressed on the surface of human AT2 cells. [103][104][105] Advances in scRNA-seq techniques may begin to identify more reliable surface markers for isolation of human AT2 cells, and establishment of reporter cell lines will be helpful in subculturing of AT2 cells. 61,62 Furthermore, establishment of collaborative online tools such as LungMAP (www.lungmap.net) and the Human Lung Cell Atlas will aid researchers in establishing improved isolation strategies. 106 It is important to consider that reliance on single lineage marker strategies can be unreliable, particularly when multiple cell types may express the same marker (Table 2) is also a concern, and cells will have to be of an assured quality. 110 Such issues need to be considered and addressed before being used in the clinic, and patients monitored post-transplantation. It has also been noted that culture differences between separate laboratories can lead to considerable heterogeneity within a single induced-hPSC cell line, while interpatient heterogeneity also exists. [111][112][113] Another consideration that needs to be made for both adult-and hPSCderived organoids is the method in which the cells are cultured, with most relying on matrigel, a mixture of undefined ECM components secreted by Engelbreth-Holm-Swarm mouse sarcoma cells. 114 As a result, matrigel is not approved for use in humans. The undefined nature of matrigel may also result in issues such as inhibition of cell maturity or differentiation. Therefore, this has led to increased efforts in producing defined, synthetic hydrogels, such as those utilized for intestinal organoids to aid in colonic wound healing. 115
The possibility of using a patient's own epithelial cells without gene correction has been explored. A rare population of human basal cells expressing SOX9 were enriched in 2D culture following bronchoscopic brushings from Bronchiectasis patients, and instilled back into the individual patients' lung lobes, leading to thinner bronchial walls and improved pulmonary function. 125 The direct contribution of delivered cells and the long-term effects of this procedure are currently unknown, and reproducibility needs to be proven. Such findings highlight the increasing "stem cell hype" that is present within both the media and wider scientific community. This can lead to over-emphasis of findings, which can have a host of consequences including misleading the public, reduction in methodical scientific approaches, and even potentially harmful premature clinical use. 126 The importance to carefully validate scientific and clinical findings, as well as increase study sizes, is therefore evident.

| CONCLUSION AND FUTURE PERSPECTIVES
In vitro human alveolar models provide a new powerful platform to investigate cellular behavior and activity of human lung alveolar stem/ progenitor cells, cell-cell crosstalk, host-pathogen interactions, as well as to conduct drug screenings and toxicity assays. Despite these advances, improved model systems to recapitulate complexity of in vivo human lungs are still required. Additionally, recent scRNAseq technologies have elucidated cellular and molecular heterogeneity and diversity within the human lung, with the observance of "aberrant" cell types arising during disease states that are not present within healthy lungs. Additional single-cell "omics" approaches with spatial information are becoming available. Such cells await further characterization and phenotypic analysis to assess their roles in disease pathogenesis, but their identification may hold the key to better understanding lung diseases and repair mechanisms.
Use of human lung-derived cells in autologous transplantation may promise a future treatment of lung disease, although a number of hurdles, including the efficacy of transplantation and the method of administration, need to be overcome before these translational efforts will be realized in the clinic. Engrafting cells onto cellular scaffolds prior to transplantation may assist in increasing cell survivability and maturation, while production of injectable synthetic hydrogels could improve cell delivery. Most importantly, common standard principles for tissue acquisition and processing are required. Human alveolar lung models are invaluable tools to address these questions, and may one day lead to therapeutic regeneration of the human lung.

ACKNOWLEDGMENTS
We would like to thank Vishal Menon for valuable scientific comments; Irina Pshenichnaya (Histology), Peter Humphreys (Imaging),

CONFLICT OF INTEREST
The authors declared no potential conflicts of interest.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.