The distal lung performs essential respiratory functions that can be compromised by inflammatory, neoplastic or infectious disorders such as COVID-19 pneumonia. The study of these conditions would be facilitated by robust in vitro models based on human cells. The identities of the stem cells that regenerate distal lung epithelium in vivo over the human lifespan have been inferred from mouse studies, despite differences between these species1. In humans, basal stem cells span the entire airway tree, whereas in mouse, club cells2 and/or secretoglobin 1A1 (SCGB1A1)-expressing non-club cells3 renew the distal bronchioles during ageing. In the gas exchange region, mouse alveolar epithelial type II (AT2) cells generate AT1 and AT2 cells during homeostasis4,5, and additional progenitors are recruited in response to severe injury3,6,7,8,9. The presence and/or roles of facultative progenitors in human lung are unknown. Human AT2 cells can be differentiated into AT1 cells, but these cultures are short-lived and do not demonstrate long-term self-renewal or enable expansion for disease modelling4,10,11; furthermore, their requirements for feeder cells impede the definition of specific niche components12,13. We have established long-term organoid culture of distal human lung, including AT2 and basal stem cells, and used this system to validate functional progenitor cells and model SARS-CoV-2 infection.

Clonogenic alveolar and basal organoids

We empirically defined medium conditions to support the clonal expansion of distal human lung progenitors from 136 individuals in collagen/laminin extracellular matrix (ECM) (Fig. 1a, Supplementary Table 1). Together, EGF and the BMP antagonist NOGGIN supported growth of disaggregated distal lung cells or purified epithelial fractions thereof (Extended Data Fig. 1a–c). Single cells (Extended Data Fig. 1d–g) expanded into either SFTPC+HTII-280+ AT2 cystic organoids (Fig. 1b–e) or KRT5+ solid organoids expressing the basal cell marker KRT5 (Fig. 1b, f–h).

Fig. 1: Clonogenic expansion of human distal lung organoids.
figure 1

a. Human distal lung D14 (day 14) organoid cultures contain cystic and solid organoids. Bottom left, brightfield; right, haematoxylin and eosin (H&E). Scale bar, 100 μm. b, Whole-mount immunofluorescence with anti-KRT5 (basal cell), anti-SFTPC (AT2 cell) and anti-SCGB1A1 (club cell) antibodies. Scale bar, 100 μm. ce, Alveolar organoids on D32. c, Cystic AT2 organoid. H&E; scale bar, 25 μm. d, Whole-mount immunofluorescence for anti-SFTPC, anti-HTII-280, phalloidin (Ph) and DAPI; scale bar, 50 μm. e, Anti-KI67 and DAPI fluorescence of an adjacent section to d. fh, Basal organoids on D32. f, H&E; scale bar, 50 μm. g, Whole-mount immunofluorescence for anti-KRT5 and DAPI; scale bar, 100 μm. h, Anti-KI67 and DAPI immunostaining of adjacent section to g. ik, Single-cell RNA-seq of total distal lung organoids at D28. i, t-Distributed stochastic neighbour embedding (t-SNE) plot of 7,285 individual cells demonstrating AT2, basal, and club populations. j, Expression of SFTPC (AT2), KRT5 (basal) and SCGB1A1 (club). k, Feature plots for expression of SFTPC (AT2), KRT5 (basal) and SCGB1A1 (club). lp, Clonogenic AT2 organoid culture. l, Brightfield microscopy of AT2 organoids at D180; scale bar, 200 μm. m, H&E staining from culture as in l; scale bar, 50 μm. n, Transmission electron microscopy of AT2 organoid at D32. LB, lamellar body; scale bar, 5 μm. o, p, Immunofluorescence of AT2 organoid at D32 (o); culture on glass for 10 additional days induces differentiation into AT1 cells (p). Immunofluorescence for anti-HTI-56 (AT1) and anti-HTII-280 (AT2); scale bar, 50 μm. q, scRNA-seq feature plots of 2,780 AT2 organoid cells on D89 of cumulative culture.

Single cell RNA sequencing (scRNA-seq) of distal lung organoids confirmed the presence of populations of SFTPC+ AT2 cells, KRT5+ basal cells and SCGB1A1+ club cells (Fig. 1i, j, Extended Data Fig. 2). Cells that co-expressed KRT5 and SCGB1A1 were found in both basal and club cell populations, suggesting a transitional state (Fig. 1j, k, Extended Data Fig. 2) that was supported by the results of SPADE14 and Monocle15 trajectory analysis (Extended Data Fig. 3, Supplementary Table 2).

Characterization of human AT2 cell organoids

We generated pure AT2 organoids from mixed cultures via uptake of lysosomal dyes into lamellar bodies (Extended Data Fig. 4a, b, Supplementary Fig. 1). EPCAM+LysoTracker+ AT2 cells expanded clonally for up to 180 days (Fig. 1l, m, Extended Data Fig. 4c, d). Chemically defined EGF/NOGGIN medium was sufficient for baseline clonal proliferation of AT2 organoids, which was attenuated by blocking global endogenous WNT biosynthesis (Extended Data Fig. 4e), consistent with the requirement for autocrine WNT signalling in mouse AT2 cells16. Growth was enhanced by adding fibroblast conditioned medium containing serum and WNT agonists (Extended Data Fig. 4f). Transmission electron microscopy revealed microvilli and lamellar bodies characteristic of mature AT2 cells (Fig. 1n, Extended Data Fig. 4g). AT2 organoids showed upregulation of AT1 markers when cultured on glass with serum (Fig. 1o, p).

Single-cell RNA-seq of mixed distal lung organoids or purified alveolar organoids revealed uniformly high expression of canonical AT2 cell markers in alveolar populations (Fig. 1i–k, q, Extended Data Fig. 2, Supplementary Data 1). AT2 cell subsets were not readily observed, and cell cycle-related mRNAs did not localize to a specific AT2 subpopulation (Fig. 1q, Supplementary Data 1, Supplementary Table 3).

Differentiation of human basal organoids

Basal cell organoids in mixed distal lung culture grew more rapidly than alveolar organoids and initially formed solid KRT5+ spheroids (Fig. 1a, b, f–h). However, by one month of culture, about 50% of organoids had developed single or occasionally multiple lumens (Fig. 2a, b, Extended Data Fig. 5a–c) with club (SCGB1A1+KRT5) and ciliated (AcTUB+KRT5) cells lining the interior surface (Fig. 2a, b). Basal cultures purified by density sedimentation exhibited serial clonal outgrowth, dependence on EGF and NOGGIN, and cavitation lined by luminal SCGB1A1+ club and AcTUB+ ciliated cells (Fig. 2c, d, Extended Data Fig. 5d–h, Supplementary Videos 1, 2). Similar differentiation occurred when organoids were transferred into 2D air–liquid interface (ALI) cultures (Fig. 2e).

Fig. 2: Basal organoid differentiation and TNFRSF12Ahi progenitor discovery.
figure 2

a, b, Immunofluorescence showing basal organoid differentiation in mixed culture. Lumens are lined with acetylated tubulin+ (AcTUB) ciliated cells, which express the SARS-CoV-2 receptor ACE2 but lack KRT5 (a), and SCGB1A1+ club cells (b). Scale bar, 20 μm. ce, Sedimented basal organoid culture. c, Club cell differentiation in purified basal organoids after 38 days in 3D clonogenic culture, shown by whole-mount fluorescence for anti-SCGB1A1 (red), phalloidin (white), and DAPI (blue); scale bar, 50 μm. d, Ciliated differentiation of organoids as in c. Confocal transmission image; scale bar, 20 μm. White box in inset expanded to main image; arrowheads denote cilia. e, Immunofluorescence showing AcTUB+ ciliated and SCGB1A1+ club cells in an organoid that has been replated in 2D ALI culture. Scale bar, 50 μm. f, g, Single-cell RNA-seq of KRT5+ basal cells from Fig. 1i. f, t-SNE plots showing Basal 1 and Basal 2 subclusters. Basal 1 is subdivided into Basal 1.1 and 1.2, with the latter expressing the proliferative mRNAs CDK1 and PCNA. g, Differential gene expression in Basal 1 and Basal 2 subclusters from a total of 2,303 basal cells. h, Basal marker transcript t-SNE overlays of f (log10 number of unique molecular identifiers (UMI)). i, Relative gene expression kinetic plots of f and g across the pseudotime trajectory (n = 3,721 cells). j, FACS isolation of TNFRSF12Ahi versus TNFRSF12Aneg cells from mixed distal lung organoids, pre-gated on live singlets. k, Brightfield image of organoid culture of cells from j at D14. l, Quantification of k (number of organoids per 1,000 cells); ***P < 0.001 (P = 1.0 × 10−4), two-tailed Student’s t-test. Data in j, k represent n = 5 independent experiments for each TNFRSF12A population.

Source data

Distinct subtypes of airway basal cells

Single-cell RNA-seq clustering of organoid KRT5+ basal cells from multiple individuals reproducibly identified two populations: Basal 1 and Basal 2 (Fig. 2f, g, Extended Data Fig. 6a, Supplementary Data 1). Basal 1 included an actively cycling subpopulation (Basal 1.2) that was enriched in proliferation markers (PCNA, CDK1) and gene cell enrichment analysis (GSEA) cell-cycle-related processes (Fig. 2f, g, Extended Data Fig. 6a–c, Supplementary Table 3). Basal 1, but not Basal 2, expressed canonical lung basal cell mRNAs such as TP6317, integrin-α6 (ITGA6) and integrin-β4 (ITGB4, which encodes a binding partner of ITGA618 and marks mouse lineage-negative epithelial progenitors (LNEPs))7 (Fig. 2h). Basal 2 was enriched in expression of genes related to vesicular transport, endoplasmic reticulum processes and squamous markers (Fig. 2g, Supplementary Table 3).

Characterization of TNFRSF12A+ progenitors

We examined the differentially expressed Basal 1 gene TNFRSF12A (also known as Fn14, TWEAKR), which encodes a membrane receptor (Fig. 2g, h, Supplementary Table 4) because of its potential utility for fluorescence-activated cell sorting (FACS) isolation and its homology to the intestinal stem cell marker TNFRSF1919. Unbiased pseudotime analysis revealed a continuous single-cell trajectory connecting KRT5+ Basal 1 cells to SCGB1A1+ club cells, with TNFRSF12A mRNA being strongly associated with the proliferation marker gene MKI67 (Fig. 2i, Extended Data Fig. 3e). When Basal 1 cells co-expressing EPCAM, ITGA6 and ITGB4 were divided into three fractions (expressing low, medium and high levels of TNFRSF12A mRNA), a proliferative gene module20 was significantly enriched in the highest (TNFRSF12Ahi) versus lowest fraction (TNFRSF12Alo) (Extended Data Fig. 6d–f). To determine whether this enrichment reflected intrinsic proliferative potential, we fractionated total distal lung organoids by FACS into EPCAM+ITGA6+ITGB4+ Basal 1 cells and then into TNFRSF12Ahi and TNFRSF12Aneg subsets (Fig. 2j). When cultured, the TNFRSF12Ahi subset showed 4–12 times greater clonogenic organoid-forming capacity than the TNFRSF12Aneg subset (Fig. 2k, l).

We examined lineage relationships between Basal 1 and Basal 2 by fractionating density-sedimented KRT5+ basal organoids by FACS into EPCAM+ITGA6+ITGB4+TNFRSF12Ahi (Basal 1) and EPCAM+ITGA6ITGB4TNFRSF12Aneg (Basal 2) populations (Extended Data Fig. 7a). Clonogenic organoid formation was strongly enriched in Basal 1 versus Basal 2 from three separate individuals (Extended Data Fig. 7b–d). The basal 2-enriched genes SPRR1B and TMSB4X (Fig. 2g, Supplementary Table 3) were transiently induced in Basal 1 cell organoids (Extended Data Fig. 7e, f), suggesting that Basal 2 cells might differentiate from Basal 1.

The NOTCH target gene HES1 was one of the most differentially expressed loci in Basal 1 organoids, in which gene networks included NOTCH1, NOTCH2 and JAG1 (Fig. 2g, Supplementary Table 3). Inhibition of NOTCH significantly increased basal organoid proliferation from TNFRSF12AhiEPCAM+ITGA6+ITGB4+ cells (Extended Data Fig. 7g, h), suggesting that NOTCH signalling restrains growth in these cells. Conversely, NOTCH agonism did not affect proliferation, but did induce expression of SCGB1A1, similar to upper airway cells21,22 (Extended Data Fig. 7i).

Characterization of TNFRSF12A+ cells in distal lung

Immunostaining of human distal lung tissue showed TNFRSF12A+ basal cells that were enriched intermittently at the tips or bases of bronchiolar furrows (Fig. 3a, Extended Data Fig. 8a, b); the latter are recognized as a goblet cell niche23. We found TNFRSF12A in diverse lung stromal and epithelial cells, but it clearly marked a minor population of KRT5+ and p63+ basal cells (Fig. 3a, Extended Data Fig. 8a–c). The TNFRSF12A+ subset of KRT5+ basal cells had a higher mitotic index than total KRT5+ cells in vivo (Fig. 3b, c), consistent with the results of organoid scRNA-seq (Fig. 2i). FACS analysis of human distal lung cells confirmed that TNFRSF12A was expressed in 10.9% of basal cells (Extended Data Fig. 8d, top).

Fig. 3: Characterization of TNFRSF12Ahi Basal 1 cells from intact human lung.
figure 3

a, Left, anti-KRT5 and anti-TNFRSF12A immunofluorescence of human distal lung. Yellow boxed area is expanded with (middle) and without (right) DAPI. Scale bars, 100 μm. b, Immunofluorescence of distal airway TNFRSF12A+ basal cell proliferation showing KRT5, TNFRSF12A, KI67 and DAPI. Scale bar, 100 μm. c, Mitotic index of TNFRSF12A+KRT5+ cells from b; three independent experiments. K5+, total KRT5+; K5+T+, TNFRSF12A+KRT5+. Box plots represent first quartile, median and third quartile; whiskers show minimum and maximum. **P < 0.01 (P = 4.4 × 10−3), two-tailed Student’s t-test. d, Brightfield and anti-KRT5 immunofluorescence of FACS-isolated TNFRSF12Aneg and TNFRSF12Ahi organoid cultures at D14; scale bars, 500 μm. e, Quantification of d (number of organoids per 1,000 cells); data represent n = 5 independent experiments for each TNFRSF12A population. Box plots represent first quartile, median and third quartile; whiskers show minimum and maximum. P = 1.1 × 10−3, two-tailed Student’s t-test. f, H&E and anti-KRT5, anti-TNFRSF12A, anti-SCGB1A1, and anti-AcTUB immunostaining of organoids from the TNFRSF12Ahi fraction of FACS-sorted EPCAM+ITGA6+ITGB4+ distal lung cells. Scale bars, 50 μm. g, H&E and anti-SCGB1A1 and anti-AcTUB immunofluorescence of 2D ALI cultures from basal cell organoids from TNFRSF12Ahi basal cells as in f. Scale bars, 50 μm.

Source data

Upon prospective culture directly from human lung without an organoid intermediate, FACS-isolated EPCAM+ITGA6+ITGB4+TNFRSF12Ahi cells (TNFRSF12Ahi Basal 1 cells; Extended Data Fig. 8d, bottom) showed a 15-fold increase in KRT5+ organoid formation compared with EPCAM+ITGA6+ITGB4+TNFRSF12Aneg cells (TNFRSF12neg Basal 1 cells; Fig. 3d, e). TNFRSF12Ahi basal organoids also differentiated into SCGB1A1+ club and AcTUB+ ciliated cells in prolonged culture (Fig. 3f) or when grown as 2D ALI monolayers (Fig. 3g).

SARS-CoV-2 and H1N1 infection of organoids

Influenza virus H1N1 avidly infected distal lung organoids, which also expressed influenza receptors (Extended Data Fig. 9a–d), similar to proximal airway organoids24,25. Infection of organoids with influenza H1N1 PR8 was inhibited by nucleoside analogues, consistent with previous studies26 (Extended Data Fig. 9e), and screening of diverse antiviral compound classes in 48-well format revealed differential activity (Extended Data Fig. 9f), suggesting that this system could be used for scalable therapeutics discovery.

In COVID-19 pneumonia, severe SARS-CoV-2 infection of the distal lung induces alveolar damage and respiratory failure27. Organoid scRNA-seq at time points before ciliated differentiation identified expression of the SARS-CoV-2 receptor ACE2 and processing protease TMPRSS2 mRNAs predominantly in club and AT2 cells (Extended Data Fig. 10a), consistent with expression of ACE2 in KRT5 differentiated interior luminal cells (Fig. 2a). To facilitate access of SARS-CoV-2 to ACE2-expressing luminal cells, we adapted an apical-out suspension culture polarization method28 to distal lung organoids (Extended Data Fig. 10b). Within 48 h in suspension, organoids reorganized into apical-out epithelial spheroids with microvilli, apical junctions, and motile cilia facing the organoid exterior. Differentiation of outwardly oriented ciliated cells accelerated over five days and progressed over weeks (Fig. 4a, Extended Data Fig. 10c–f, Supplementary Video 3). Apical-out basal organoids also showed an increase in outwardly facing club cells with apical secretory granules (Fig. 4a, Extended Data Fig. 10g), and apical-out AT2 organoids showed differentiation into AT1 cells (Extended Data Fig. 10h–j). Crucially, in apical-out organoids, ACE2 localized to apical cell membranes on the external organoid surface (Fig. 4b, Extended Data Fig. 10k).

Fig. 4: SARS-CoV-2 infection of apical-out distal lung organoids.
figure 4

a, Immunofluorescence comparing polarization and surface localization of SCGB1A1+ club cells (green) and AcTUB+ ciliated cells (white) upon ECM-embedded non-polarized growth (left) or apical-out suspension culture (right). b, Transverse sections of apical-out basal organoid as in a stained with anti-ACE2, anti-SCGB1A1 and DAPI. c, Quantitative PCR for SARS-CoV-2 unspliced genomic RNA (gRNA, left) and spliced sgRNA (right) in apical-out distal lung organoids, 72 h post-infection (hpi), normalized to U3 snoRNA; n = 2 independent experiments. d, Anti-dsRNA immunofluorescence of apical-out human distal lung organoids, either mock-infected or infected with SARS-CoV-2, 48 hpi. e, Immunofluorescence for SARS-CoV-2 nucleocapsid protein (NP) in infected or mock-infected apical-out organoids at 96 hpi. f, Immunofluorescence of SARS-CoV-2-infected apical-out AT2 organoids with the indicated antibodies at 96 hpi. g, Immunofluorescence colocalization of SARS-CoV-2 NP and SCGB1A1 in apical-out distal lung basal organoids at 96 hpi. h, Immunofluorescence showing cell-type specificity in SARS-CoV-2-infected apical-out organoids. Inf, SARS-CoV-2 infected cell. In ch, organoids were in suspension for 6–10 days (basal) or 3 days (AT2) before infection. Scale bars, 20 μm except for f, h (10 μm).

SARS-CoV-2 infected apical-out mixed distal lung organoids, with induction of unspliced SARS-CoV-2 genomic RNA reaching levels similar to those of the abundantly expressed U3 small nucleolar RNA (Fig. 4c, left). In addition, infected organoids showed replication-specific SARS-CoV-2 spliced subgenomic RNA (sgRNA; Fig. 4c, right) and production of infectious virions with VeroE6 cell plaque formation (35 PFU ml−1 from organoid lysates and 65 PFU ml−1 from organoid supernatants). In basal organoids infected with SARS-CoV-2, double-stranded RNA (dsRNA) appeared by 48 h after infection (Fig. 4d) and SARS-CoV-2 nucleocapsid protein (NP) by 96 h (Fig. 4e). Approximately 10% of AT2 organoids displayed prominent SARS-CoV-2 NP expression in SFTPC+ cells; the remaining organoids were devoid of infection (Fig. 4f). Similarly, SARS-CoV-2 infected about 10% of basal organoids. In 2,621 total distal airway basal cells representing cultures from four individuals (Supplementary Table 5), SARS-CoV-2 infection was not detected in KRT5+ basal or AcTUB+ ciliated cells (odds ratio 0, P < 0.05), in contrast to infection of upper airway ciliated cells by SARS-CoV-2 in 2D ALI culture29,30. However, SARS-CoV-2 NP and dsRNA immunofluorescence were primarily present in SCGB1A1+ club cells (Fig. 4g, h) which were strongly associated with and accounted for 79% of NP- or dsRNA-positive cells (odds ratio 19.33, P < 0.0001); 21% of infected cells lacked SCGB1A1 (Fig. 4g, h, Supplementary Table 5). Overall, these studies indicate that AT2 cells were directly infected by SARS-CoV-2, and suggest that club cells are a distal lung target population.


We have applied long-term human distal lung organoid cultures to progenitor discovery and modelling of infectious disease. Our findings extend upon earlier short-term and feeder-dependent adult lung culture methods4,10,11 and present an alternative to techniques involving the differentiation of inducible pluripotent stem cells31,32,33,34. Organoids contained two related subtypes of KRT5+ human distal lung basal cells—Basal 1 and Basal 2. Notably, the TNFRSF12A-expressing Basal 1 fraction possessed enriched clonogenic progenitor activity, establishing functional precedent for a proliferation-enriched basal cell subtype. Although TNFRSF12A is by no means exclusively present in basal cells, within the basal layer it often localizes to a postulated niche in airway furrow bases and tips23, extending recent notions that the lung epithelium shows spatial specialization35. It is possible that TNFRSF12A or analogous markers could distinguish basal cell progenitor subsets in other tissues. Our organoids also enable facile exploration of SARS-CoV-2 distal lung infection, which is relevant to COVID-19-associated pneumonia27, and implicate SCGB1A1+ club cells as a target whose infection could compromise protective lung glycosaminoglycans and precipitate a vicious infection cycle. We did not observe infection of ciliated cells, in contrast to 2D ALI lung studies29,30; this could require alternative culture conditions. SCGB1A1-negative populations were also infected and are under further investigation; for example, bronchial transient secretory cells express ACE2 and TMPRSS236.

Overall, single-cell analysis of organoid cultures, as described here, may represent a general strategy for identifying and functionally validating candidate stem cells in slowly proliferating tissues. The culture of progenitors for all adult human distal lung epithelial lineages, including alveoli, should substantially enable the modelling of diseases such as neoplastic and interstitial lung conditions12 and allow tissue engineering and precision medicine applications. Finally, this organoid system should facilitate diverse investigations of pulmonary pathogens, including the SARS-CoV-2 distal lung infection that is associated with fulminant respiratory failure.


Additional experimental details are in the Supplementary Methods. No statistical methods were used to predetermine sample size. The experiments were not randomized and the investigators were not blinded to allocation during experiments and outcome assessment.

Human tissue procurement and processing

All experiments performed in this work were approved by the Stanford University School of Medicine Institutional Review Board and performed under protocol no. 28908. Standard informed consent for research was obtained in writing from all patients who contributed to this study before tissue procurement and all experiments followed relevant guidelines and regulations. Peripheral lung tissue within 1 cm of the visceral pleura was obtained from discarded surgical tissue from lobectomies. For patients with suspected lung cancer, cases with clinical T4 (American Joint Cancer Committee 6th edition) disease (for example, features such as bronchial invasion or parenchymal satellite nodule/metastases) were deferred. Normal tissue was removed from the lung margin most anatomically distal to palpably well-defined lesions, or from uninvolved lobes in the case of pneumonectomies. Samples with tumours containing ill-defined margins were deferred. Tissue was either processed fresh or stored at 4 °C overnight and processed the following morning.

Mixed distal lung organoid culture

To isolate distal airway cells, lung parenchyma from 1 cm from the visceral pleura was mechanically dissociated with Castro scissors, washed and incubated with 5 units per ml porcine elastase (Worthington), 100 Kunitz units per ml DNase I (Worthington), and Normocin (InvivoGen), and resuspended in two tissue volumes of lung organoid medium, comprising advanced DMEM/F12 (Invitrogen) supplemented with 10 mM nicotinamide, N-acetyl cysteine, 1× B27 supplement minus vitamin A, recombinant human NOGGIN (100 ng ml−1, R&D Systems), recombinant human EGF (50 ng ml−1, R&D Systems), and TGF-β inhibitor A83-01 (100 nM, Tocris). This lung organoid medium was used for all experiments except for those shown in Extended Data Fig. 4f. The tissue was then agitated for 1 h at 37 °C and the resultant cell suspension was filtered through 100–40-μm cell strainers and subjected to ammonium chloride red blood cell lysis. The cell pellet was then washed and resuspended in 10 volumes of reduced growth factor Basement Membrane Extract II (Trevigen). Cells in matrix were then plated in 24-well plates in 50-μl droplets, and warm medium was added after the droplets had solidified for 10 min at room temperature Medium was changed every 3–4 days and organoids were passaged every 3–4 weeks by dissociation with TrypLE. Passaging was based on ECM durability and integrity and estimated organoid confluency, judged by estimated organoid volume to volume of the ECM droplet. Distal lung organoids could be passaged for about 6 months with basal organoids initially exhibiting 6–7 doublings every 2 weeks. Alveolar organoids expanded more slowly with an initial rate of 3–4 doublings every 2 weeks but predominated over basal organoids after several months. Calculated from initial cell division rates, the upper limits of basal and alveolar expansion were 219 (524,288-fold) and 216 (65,536-fold), respectively. To rule out contamination by malignant cells, long-term cultures were systematically evaluated for the presence of dysplasia or carcinoma by a board-certified pathologist. In addition, five long-term organoid cultures (2–6 months) underwent targeted next generation sequencing to exclude pathogenic nucleotide variants (see below). Full details are provided in the Supplementary Methods.

Tandem MACS stromal depletion and EPCAM purification of distal lung cells

Distal lung was dissociated as above, and all incubation steps were carried out on ice. We incubated 107 cells with Fc Block (Biolegend 422301) and diluted them 1:100 in FACS buffer (2 mM EDTA and 0.2% fetal calf serum in 1× PBS, pH 7.4), for 10 min. The cells were then mixed with APC-conjugated anti-CD45 antibodies at 1 μg ml−1 in FACS buffer for 30 min, washed, and subjected to two rounds of depletion with magnetic beads according to the manufacturer’s protocol (Miltenyi: anti-human fibroblast 130-050-601, anti-CD31 130-091-935, anti-APC 130-090-855, LS column 130-042-401). Unlabelled cells were then centrifuged at 300g and labelled with a cocktail of 1 μg ml−1 of PerCP-Cy5.5 anti-EPCAM antibody and Zombie Aqua viability stain (Biolegend 423101) diluted 1:400 from stock concentration in FACS buffer.

Organoid cryopreservation and recovery

For cryopreservation and recovery, ECM droplets were dissociated by pipetting in three volumes of PBS with 5 mM EDTA and then incubated on ice for 1 h. Cells were pelleted at 300g for 5 min and resuspended in freezing medium (fetal calf serum (Gibco), 10% v/v DMSO), placed into cryovials and then into Mr. Frosty (Thermo Fisher) containers and stored at –80 °C overnight, followed by transfer to liquid nitrogen vapour phase for long-term storage. Organoids were recovered by quick thaw in a 37 °C water bath followed by washing in organoid medium and plating in ECM with organoid medium plus 10 μM ROCK inhibitor Y-27632 (Tocris).

Screening exogenous growth factors in organoid culture

Distal airway cells were isolated and plated as above with the following exceptions: advanced DMEM/F12 was used instead of organoid medium during elastase digestion of lung tissue, and cells were serially diluted and filtered through a 40-μm cell strainer and counted with a haemocytometer. One thousand viable epithelial cells (by Trypan blue exclusion, size, and morphology) per μl ECM were plated per 5 μl Matrigel droplet per well. Base medium consisted of organoid medium lacking A83-01, EGF, NOGGIN, WNT3A or RSPO1. EGF (final 50 ng ml−1, R&D), NOGGIN (final 100 ng ml−1, R&D), WNT3A (final 100 ng ml−1, R&D), RSPO1 (final 500 ng ml−1, Peprotech) or the PORCUPINE inhibitor C59 (final 1 μM, Biogems) were added singly or in combination to base medium. Images were obtained ten days after primary plating with an inverted light microscope at 5× magnification. Each condition was plated in quadruplicate and organoid formation was quantified using the analyse particle (threshold, 4902 pixels) plugin in ImageJ (see Supplementary Methods).

Single-cell RNA-seq of unfractionated organoid cultures

Lung organoid cultures from separate individuals were dissociated 4 weeks after primary plating and subjected to droplet-based scRNA-seq with the 10x Genomics Single Cell 3′ platform with a 5-nucleotide UMI, according to the manufacturer’s protocol. Cell capture, library preparation, and sequencing were performed as previously described37. For scRNA-seq analysis in Fig. 1k, a modified Kruskal–Wallis rank sum test was performed to determine the significance of differential marker gene expression for AT2 (SFTPC), basal (KRT5), and club (SCGB1A1) cells, with all P < 0.001. Principal component analysis, t-SNE, unsupervised graph-based clustering, statistical testing, and the pseudotime trajectory for all scRNA-seq analyses are described in Supplementary Methods and Supplementary Data 1.

Single-cell RNA-seq of purified AT2 organoid cultures

LysoTracker+ AT2 cells38 from unfractionated organoids were purified by FACS and cultured for two months with one passage. These were dissociated and subjected to droplet-based scRNA-seq with the 10x Genomics Chromium Single Cell 3′ platform v2 according to the manufacturer’s protocol. The library was sequenced using paired-end sequencing (26 bp read 1 and 98 bp read 2) with a single sample index (8 bp) on an Illumina NextSeq 500. Data preprocessing and principal component analysis were carried out with CellRanger v1.2. Subsequent analysis is described in Supplementary Methods and Supplementary Data 1.

Electron microscopy

Organoid cultures were fixed in ECM with 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4), dehydrated, embedded in epoxy resin and visualized with a JEOL (model JEM1400) transmission-electron microscope with a LaB6 emitter at 120 kV.

Histology and immunocytochemistry

Organoids were fixed with 2% paraformaldehyde at 4 °C overnight, paraffin embedded and sectioned (10–20 μm) as previously described37. Sections were deparaffinized and stained with H&E for histological analysis. Antibodies used for immunocytochemistry staining following standard staining protocols39 are listed in Supplementary Methods and images were acquired on a Leica-SP8 confocal microscope.

RNA fluorescent in situ hybridization

RNA in situ hybridization was performed as described40 and probe sequences are provided in Supplementary Methods.

Whole-mount organoid confocal immunofluorescence microscopy

Intact, uninfected organoids were fixed in 2% paraformaldehyde in 100 mM phosphate buffer (pH 7.4) (4% paraformaldehyde for infected organoids) for 1 h at room temperature, washed with PBS with 100 mM glycine, permeabilized in 0.5% Triton X-100 in PBS for 1 h, then incubated in staining buffer (4% BSA, 0.05% Tween-20 in PBS pH 7.4, 10% goat/donkey serum) for an additional hour, followed by incubation with primary antibody for 24 h at room temperature in staining buffer. Whole mounts were then washed with PBS-T and incubated with fluorescent secondary antibodies, phalloidin and DAPI, for 4 h at room temperature in staining buffer. Following additional washes, whole mounts were submerged in mounting medium (Vectashield, Vector Laboratories) and mounted on chambered coverslips for imaging in four channels using Zeiss LSM 700 or 900 confocal microscopes. 3D rendering of confocal image stacks was performed using Volocity Image Analysis software (Quorum Technologies Inc., Guelph, Ontario). For Fig. 4h, which required five colours, cilia were distinguished by staining with two fluorescent secondary antibodies and merging the colocalized voxels into a pseudocoloured channel using Volocity software. Lectin staining (FITC-Sambuca Nigrin, Vector Labs FL-1301; Biotin-Maackia Amurensis, Vector Labs FL-1301) was carried out according to the manufacturer’s protocol after fixation of organoids with 0.1% paraformaldehyde in PBS for 1 h at room temperature followed by blocking with avidin/biotin (Vector Labs SP-2001). Biotin-Maackia Amurensis lectin was labelled with streptavidin–PE conjugate (Thermo Fisher SA10041) and after washing lectin staining was imaged in a Keyence BZ-X700.

Next generation sequencing of organoid cultures

Ten organoid cultures were sequenced using a commercial targeted resequencing assay with end-to-end coverage of 131 cancer genes and companion software (TOMA COMPASS Tumour Mutational Profiling System, Foster City, CA) to determine the presence of oncogenic mutations in long-term organoid cultures. Libraries were sequenced on an Illumina NextSeq 500. Nonsynonymous variants are listed in Supplementary Table 6. Variant call files are provided in Supplementary Data 2.

Density sedimentation of basal cells

Organoid cultures within 2–3 weeks of primary plating were dissociated with 1 U ml−1 neutral protease (Worthington, Cat LS02100) and 100 KU DNase I in lung organoid medium. Basal organoids were then collected by gravity sedimentation and the supernatant was either aspirated or collected for downstream use. Basal organoids were then further fractionated on a custom Ficoll-Paque gradient (4 vol Ficoll-Paque to 1 vol PBS) and centrifuged at 300g for 10 min at room temperature. Supernatant was aspirated and the organoid pellet was resuspended in 10 ml PBS in a 15-ml conical tube, collected by gravity sedimentation, and plated into ECM as described above.

FACS isolation and culture of AT2 cells

Organoids were dissociated with TrypLE followed by neutralization with 10% volume fetal calf serum, subjected to DNase at 100 KU ml−1, washed with lung organoid medium and then incubated with 100 cell pellet volumes of lung organoid medium with 10 nM LysoTracker Red DND-99 (Thermo Fisher L7528) at 37 °C for 30 min. Cells were then washed and resuspended in FACS buffer as described above, incubated with Fc block, and then incubated on ice with labelling cocktail consisting of 1 μg ml−1 of PerCP-Cy5.5 anti-EPCAM antibody and Zombie Aqua viability stain (Biolegend 423101) diluted 1:400 from stock concentration in FACS buffer. EPCAMhi and LysoTrackerhi cells were sorted into lung organoid medium with 10 μM Y-27632 (Tocris 1254) and cultured in ECM and lung organoid medium with Y-27632 for 24 h, followed by lung organoid medium without Y-27632. Pure AT2 organoid growth was enhanced by the addition of 1:1 vol:vol serum-containing L-cell conditioned medium (L-WRN CM) containing WNT3A, R-SPONDIN3 and NOGGIN and supplemented with recombinant EGF41. The full gating strategy is provided in Supplementary Fig. 1. All FACS antibodies were purchased from Biolegend. Qualitatively identical results could also be obtained with anti-HTII-280 (AT2 marker, Terrace Biotech) FACS purification in lieu of LysoTracker. AT2 organoid cells were transdifferentiated into AT1 cells by TrypLE dissociation from ECM and seeding onto chambered glass coverslips followed by culturing with advanced DMEM/F12 and 5% fetal calf serum42.

Colour mixing studies with lentivirally transduced GFP and mCherry

FACS EPCAM+ stroma-depleted organoids at D14 were infected with lentivirus at an estimated MOI of 0.9 as described previously43 with third-generation lentiviral vectors (PGK-GFP T2A Puro, SBI cat. no. CD550A-1; mCherry modified from pLentiCRISPRv1 (Addgene no. 49545) to incorporate an EF-1a-mCherry P2A Puro cassette, a gift from Paul Rack). Ninety-six hours after infection, organoids were treated with puromycin at a concentration of 600 ng ml−1 for 48 h to select for transduced cells. Two weeks after selection, GFP-expressing organoids or mCherry-expressing organoids were dissociated into single cells and mixed in a 1:1 ratio and scored as monochromatic or mixed after 28 days of each passage. The same approach was employed for purified AT2 and basal cultures after respective isolation strategies from an initial FACS-purified, EPCAM+, stroma-depleted organoid starter culture.

Flow cytometry analysis of resident basal cells from adult human lung

Adult human lung tissue was procured and dissociated as above but cells were labelled with Zombie Aqua live:dead stain as above, washed with FACS buffer, and then fixed in 2% PFA in PBS overnight at 4 °C. Cells were then stained using the whole-mount procedure as described above with the omission of PBS-glycine washing. Fixed and permeabilized cells were then incubated with 1:400 dilution of Alexa Fluor 647 conjugated mouse anti-human cytokeratin 5 antibody (Abcam) for 24 h at 4 °C in permeabilization buffer. Cells were then washed with FACS buffer and labelled with PE conjugated mouse anti-human TNFRSF12A antibody (clone ITEM-4, Biolegend) for 30 min on ice, followed by washing and analysis on a BD Aria Fusion instrument. The full gating strategy and qPCR validation of the ITEM-4 antibody are detailed in Supplementary Fig. 1.

FACS isolation of TNFRSF12Ahi and TNFRSF12Aneg basal cells

Single-cell suspensions from either fresh human distal lung or primary organoid culture at approximately 4 weeks of culture were dissociated as above, treated with Fc Block (BioLegend), and incubated in FACS buffer with Zombie Aqua 1:400, 1 μg ml−1 PerCP-Cy5.5 anti-human EpCAM (CD326), 1 μg ml−1 APC anti-human ITGA6 (CD49f), 2 μg ml−1 FITC anti-human ITGB4 (CD104), and 1 μg ml−1 PE anti-human TNFRSF12A (CD266). Thirty minutes after labelling, the cells were washed twice with FACS buffer and sorted for EPCAMhi, ITGA6/ITGB4hi, TNFRSF12Ahi and TNFRSF12Aneg. The full gating strategy is provided in Supplementary Fig. 1. More than 5,000 cells were sorted into Eppendorf tubes with lung organoid medium and 10 μM ROCK inhibitor Y-27632. All FACS antibodies were purchased from Biolegend.

Culture of TNFRSF12Ahi and TNFRSF12Aneg basal cells

Cells were seeded in ECM and submerged in lung organoid medium with 10 μM ROCK inhibitor Y-27632. The seeding density for cells FACS-isolated from organoid culture was 1,000 cells per well at a density of 100 cells per μl ECM. The seeding density for cells FACS-isolated from fresh human distal lung was 3,000 cells per well at a density of 300 cells per μl ECM. After 24 h, the medium was changed to remove ROCK inhibitor, thereafter it was changed every 72 h. Organoid formation was manually quantified 14 days after plating by two independent observers.

H1N1 organoid influenza assay

Unfractionated cultures containing AT2, basal, and club cell types at 2–3 weeks were infected in triplicate with H1N1 strain PR8 modified to express GFP upon viral replication44 after 24 h of pretreatment with antiviral compounds. ECM was dispersed by addition of 5 mM EDTA in PBS, followed by washing and inoculation with PR8-H1N1–GFP reporter virus at an estimated MOI of 1 in medium containing either vehicle or antivirals. After 12 h (one influenza infection cycle), intact organoid GFP expression was either visualized by fluorescence microscopy with a Keyence BZ-X700 automated microscope, or dissociated to single cells and fixed with 0.1% PFA in PBS followed by FACS quantification of GFP+ cells (gating strategy is provided in Supplementary Fig. 1). Antiviral dose response curves were generated using four-parameter nonlinear regression curve fitting with GraphPad Prism 7 (GraphPad Software, San Diego, CA). H1N1 tropism was assessed in a manner similar to above with the exception that Ficoll-sedimented basal cell fraction versus non-basal fractions were dissociated into single cells, counted, and infected with an estimated MOI of 1 in organoid medium for 1 h at 37 °C, followed by washing and reseeding into ECM, culturing for 16 h, and then being subjected to dissociation and FACS as above.

Suspension culture to generate apical-out polarity in lung organoids

Lung organoids grown embedded in 50 μl ECM droplets were transferred to suspension culture as described28 with modifications. In brief, ECM-embedded organoids were dislodged gently by pipetting using sterile LoBind tips (Eppendorf 22493008) and placed in 15 ml LoBind conical tubes (Eppendorf 30122216) containing ice-cold 5 mM EDTA in PBS. Five millilitres of EDTA solution was used per ECM droplet (3 ECM droplets per 15-ml conical tube) rotating for 1 h at 4 °C on a rotating platform. Organoids were centrifuged at 200g for 3 min at 4 °C and the supernatant was removed. The pellet was resuspended in growth medium in ultra-low attachment six-well tissue culture plates (Corning Costar 3471). Suspended organoids were incubated at 37 °C with 5% CO2 for different times (range 0–30 days) to characterize apical-out polarity, ciliogenesis and differentiation, and to prepare apical-out organoids for infection experiments with SARS-CoV-2.

SARS-CoV2 infection of human distal lung organoids

VeroE6 cells were obtained from ATCC as mycoplasma-free stocks and maintained in supplemented DMEM with 10% FBS. SARS-CoV-2 (USA-WA1/2020) was passaged in VeroE6 cells in DMEM with 2% FBS. Titres were determined by plaque assay on VeroE6 cells using Avicel (FMC Biopolymer) and crystal violet (Sigma), the viral genome sequence was verified, and all infections were performed with passage 3 virus. Organoids were counted and passaged into suspension medium for 6–8 days and then resuspended in virus medium or an equal volume of mock medium at an MOI of 1 relative to total organoid cells in the sample, and then incubated at 37 °C under 5% CO2 for 2 h. Organoids were then plated in suspension in lung organoid medium (apical-out organoids). At the indicated time points, apical-out organoids were washed with lung organoid medium and PBS and resuspended in TRIzol LS (Thermo Fisher), freshly-made 4% PFA in PBS, or 250 μl lung organoid medium. Cells resuspended in lung organoid medium were lysed by freezing at −80 °C. Culture supernatants were preserved in TRIzol LS or added directly to plaque assay monolayers. All SARS-CoV-2 work was performed in a class II biosafety cabinet under BSL3 conditions at Stanford University.

qPCR analysis of SARS-CoV-2 RNA

RNA from SARS-CoV-2-infected organoids was extracted by adding 750 μl TRIzol (Thermo Fisher Scientific), incubating at 55 °C for 5 min and then adding 150 μl chloroform. After mixing each sample by vortexing for 7 s, the samples were incubated at 25 °C for 5 min and then centrifuged at 12,000 rpm for 15 min at 4 °C. The aqueous layer was carefully removed from each sample, mixed with two volumes of 100% ethanol and purified using an RNA Clean & Concentrator-25 kit (Zymo Research) as per the manufacturer’s instructions. All RNA samples were treated with DNase (Turbo DNA-free kit, Thermo Fisher Scientific). The Brilliant II SYBR Green QRT–PCR 1-Step Master Mix (VWR) was used to convert RNA into cDNA and to amplify specific RNA regions on the CFX96 Touch real-time PCR detection system (Bio-Rad). The reverse transcription reaction was performed for 30 min at 50 °C, 10 min at 95 °C, followed by two-step qPCR with 95 °C for 10 s and 55 °C for 30 s, for a total of 40 cycles. Two primer sets were used to amplify either non-spliced SARS-CoV-2 genomic RNA (gRNA) spanning nucleotide positions 14221–14306, or spliced SARS-CoV-2 sgRNA30. Primer sequences are in Supplementary Table 7.

TNFRSF12A immunostaining of intact distal lung

Optimal staining of human distal lung tissue was achieved from specimens fixed within 30 min of primary surgical resections in 4% paraformaldehyde in PBS. Specimens were incubated in fixative overnight at 4 °C, transferred to 30% sucrose, and embedded into OCT. Frozen sections were cut at 10 μm, subjected to citrate-based antigen retrieval (Vector Labs) at 70 °C for 30 min, and then blocked for 1 h with 10% goat serum in IF wash buffer as described above. Mouse anti-TNFRSF12A (clone ITEM-4, Biolegend) was used for Fig. 3a and polyclonal rabbit anti-TNFRSF12A (ThermoFisher PA5-20275) was used for Fig. 3b, f and Extended Data Fig. 845.

Live-imaging and confocal microscopy of immobilized apical-out lung organoids

Live organoids were held between two coverslips in a viewing chamber (Lab-Tek II two-chambered coverglass) and filmed using a Nikon TE2000E microscope using differential interference contrast (DIC) microscopy with a 63× objective. Samples were kept at 37 °C with 5% CO2 during imaging. Digital videos were collected by a Hamamatsu high-resolution ORCA-285 digital camera and rendered using OpenLab 5.5.2 software (Improvision). After recording, samples were fixed and stained without removal from the chambers and transferred to the confocal microscope for immunofluorescence microscopy.

Statistics and reproducibility

Unless stated otherwise, all data are representative of at least two independent experiments with each independent experiment carried out using an organoid culture derived from one individual. Box plot bounds span first through third quartiles, horizontal lines represent median values, and whiskers represent data range minima or maxima or, in the case of outliers, 1.5 times the interquartile range with outliers represented by data points. t-tests were two-tailed and P values are denoted as *P < 0.05, **P ≤ 0.01, and ***P ≤ 0.001. Full details are provided in Supplementary Methods.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.