Abstract

   Within the human lung, interactions between alveolar epithelial cells
   and resident macrophages shape lung development and function in both
   health and disease. To study these processes, we develop a co-culture
   system combining human pluripotent stem cell-derived alveolar
   epithelial organoids and induced macrophages to create a functional
   environment, termed induced alveolar assembloids. Using single-cell RNA
   sequencing and functional analyses, we identify alveolar type 2-like
   cells producing GM-CSF, which supports macrophage tissue adaptation,
   and macrophage-like cells that secrete interleukin-1β and
   interleukin-6, express surfactant metabolism genes, and demonstrate
   core immune functions. In response to alveolar epithelial injury,
   macrophage-like cells efficiently eliminate damaged cells and absorb
   oxidized lipids. Exposure to bacterial components or infection with
   Mycobacterium tuberculosis reveals that these assembloids replicate key
   aspects of human respiratory defense. These findings highlight the
   potential of induced alveolar assembloids as a platform to investigate
   human lung development, immunity, and disease.

   Subject terms: Stem-cell differentiation, Embryonic stem cells,
   Respiratory system models
     __________________________________________________________________

   Interactions between alveolar epithelial cells and macrophages shape
   lung development and function. Here Kang et al. develop alveolar
   assembloids by co-culturing alveolar organoids with macrophages
   providing a platform to model lung diseases.

Introduction

   The human lung is a complex organ vital for breathing and maintaining
   physiological homeostasis. These processes are executed by the alveolar
   epithelium, which forms a physical barrier that protects the lung from
   potentially harmful inhaled agents, and alveolar macrophages (AMs), an
   abundant immune cell population which takes up residence on the luminal
   surface of alveoli after birth^[44]1,[45]2. Alveolar epithelial cells
   (AECs), comprising alveolar type I (AT1) and type II (AT2) cells,
   facilitate gas exchange and produce surfactant to prevent alveolar
   collapse^[46]3,[47]4 whereas AMs phagocytose inhaled particles,
   pathogens, and apoptotic cells, thereby contributing to the clearance
   of potentially harmful substances from the respiratory system. AMs also
   maintain optimal lung function by clearing excess or aged
   surfactant^[48]5. The coordinated interactions between AECs and
   macrophages are instrumental for both lung development and adult lung
   function, and when these go awry, contribute to the progression of
   respiratory diseases^[49]2,[50]6,[51]7. Such disease-initiating
   mechanisms remain largely unknown due to the lung’s complexity and the
   lack of suitable research models. Although experiments conducted on
   primary human cells have provided important insights, such approaches
   face challenges in terms of scalability, reproducibility, and mimicking
   the complex cellular microenvironment^[52]8,[53]9.

   Organoid technology has been pivotal for enabling studies of
   disease-associated cellular phenotypes and mechanisms in human-relevant
   contexts, and has also been instrumental for studying AEC
   biology^[54]10,[55]11. However, in the lung, as in other organs, the
   presence of immune cells such as macrophages is an important component
   of tissue function in both health and disease, and thus far, studies of
   alveolar epithelial organoids (iAEOs) have not incorporated immune
   cells, significantly limiting insights gained about their physiological
   functions^[56]9. Separately, recent advances in cellular reprogramming
   and differentiation techniques have enabled the generation of induced
   macrophages (iMφ) from human pluripotent stem cells (hPSCs). These iMφ
   possess the basic properties of primary human macrophages, making them
   a highly valuable resource for studying macrophage biology and
   macrophage-related diseases^[57]12,[58]13. While several research
   groups have recently made efforts to integrate macrophages into lung
   organoids or lung slices^[59]14,[60]15, there remains a significant gap
   in understanding how to optimize culture conditions to preserve the
   function and properties of iAEO and iMφ. Overcoming these limitations
   would represent a pivotal step towards investigating the functionality
   and adaptability of iMφ within a tissue-like context.

   Herein, we set out to combine iMφ and iAEOs in an in vitro platform we
   named ‘iAlvAssemb’ to assess their properties and interactions. By
   developing a specialized medium that supported co-culture of iAEOs and
   iMφ, while preserving their distinct functional characteristics, we
   were able to comprehensively characterize their cellular properties,
   evaluate interactive responses, and explore cell-type specific roles in
   tissue homeostasis and disease, including lung injury and infection.

Results

Generation of iAlvAssemb by co-culturing iAEO and iMφ

   To develop iAlvAssemb containing functional macrophages, we first
   generated iAEOs and iMφ from identical hPSC lines as previously
   reported^[61]10,[62]12,[63]13. For the differentiation of iAEO,
   step-wise differentiation was performed and differentiation efficiency
   was validated flow cytometer analysis (Supplementary Fig. [64]1a, b).
   The iAEOs were characterized by staining with Lysotracker
   (Fig. [65]1a), immunostaining of iAEC-specific markers, including
   NKX2.1, SFTPC, Muc1, and HOPX (Fig. [66]1b, [67]c and Supplementary
   Fig. [68]1c), and Transmission Electron Microscopy (TEM) imaging to
   confirm lamellar bodies formation (Fig. [69]1d). Similarly, we
   validated that the iMφ expressed macrophage-specific cell surface
   markers, including CD45, CD14, and CD11c (Fig. [70]1e, f and
   Supplementary Fig. [71]1d), and activated transcription of
   macrophage-related genes such as PTPRC, CD14, and IL1B (Fig. [72]1g).

Fig. 1. Characterization of iAEOs, iMφ, and iAlvAssemb.

   [73]Fig. 1
   [74]Open in a new tab

   a Representative phase-contrast and lysotracker fluorescence images of
   iAEOs. Three independent experiment was repeated independently with
   similar results. b Representative immunostaining images showing the
   expression of the alveolar epithelial markers (NKX2.1, SFTPB, SFTPC,
   HOPX) in iAEOs. c Relative mRNA expression level of alveolar epithelial
   markers (SFTPB, SFTPC, MUC1, HOPX) in iAEOs and undifferentiated hESCs.
   Data were presented as mean ± SD, technical replicates (n = 3). d
   Representative TEM image showing lamellar bodies in iAEO. Two
   independent experiment was repeated independently with similar results.
   e Representative phase-contrast image of iMφ differentiated from hESCs.
   More than three independent experiment was repeated independently with
   similar results. f Representative flow cytometry plots of macrophage
   markers (CD45, CD14, CD11c) on iMφ. g Relative mRNA expression level of
   macrophage markers (PTPRC, CD14, IL1β) in iMφ and undifferentiated
   hESCs. Data were presented as mean ± SD with technical replicates
   (n = 3). h–o iAlvAssemb was created by co-culturing of iAEO and iMφ in
   KSFM media for 7 days. More than three independent experiment was
   repeated independently with similar results. h-i. Representative
   phase-contrast and H&E staining images of iAlvAssemb cultured in KSFM
   media. j Representative cell membrane-linked fluorescence images
   (Green) of iMφ and lysotracker fluorescence images (Red) of Alv-iAEOs
   in AlvAssemb. k Representative immunostaining images showing the
   expression of the alveolar epithelial markers (SFTPB, SFTPC) and iMφ
   marker (MAC387) in AlvAssemb. l, m Cytokine blot array images and
   quantification of secreted cytokines in conditioned media of
   ctrl-iAEOs, ctrl-iMφ and iAlvAssemb cultured in KSFM media. n, o
   Quantification of IL1β and IL6 secretion in conditioned media of
   ctrl-iAEOs, ctrl-iMφ, iAlvAssemb and Human AM. Data were presented as
   mean ± SD with technical replicates for Ctrl-iMφ (n = 6), iAlvAssemb
   (n = 6), Ctrl-iAEO (n = 3), and Human AM (n = 3). Statistical analysis
   was performed using a one-way ANOVA with Dunnett’s multiple comparisons
   test, and p-value (<0.05) was indicated on the graph.

   We then set out to define a cell culture medium that would support
   identity of each cell type. When we evaluated whether iAEOs maintained
   viability and proliferative capacity when cultured in iMφ media, we
   found that iAEOs did not survive (Supplementary Fig. [75]2a), and
   conversely, when we cultured the iMφ in the iAEO medium, growth was
   reduced and the iMφ lost their characteristic surface phenotype
   (Supplementary Fig. [76]2b, c). However, by carefully assessing the
   culture supplements in each growth medium, we successfully identified
   essential and detrimental factors for each cell type. We found that FBS
   was detrimental for iAEO survival (Supplementary Fig. [77]2d, e).
   However, iAEOs could survive and showed increased expression of SFTPA
   and CDC20 in the presence of FBS if the TGFβ inhibitor SB431542 was
   included in the growth medium (Supplementary Fig. [78]2f). Moreover, we
   observed that keratinocyte growth factor (KGF) was essential for
   preserving iAEO identity (data not shown), while expression of the
   surface marker CD14 was decreased in the presence of the GSK inhibitor
   CHIR99021 and dexamethasone (Supplementary Fig. [79]2g, h).
   Particularly, CHIR99021 significantly impacted iMφ identity. Based on
   these findings, we developed an RPMI-based co-culture medium, referred
   to as KSFM, which was supplemented with 2.5 ng/ml FGF7, 2.5uM SB431542,
   10% FBS, and 100 ng/ml M-CSF, and that supported maintenance of both
   iAEO and iMφ identities. In particular, KSFM is a medium that contains
   only low concentrations of SB431542 and KGF added to iMφ media. It can
   maintain similar functions, such as the expression of iMφ markers and
   the LPS response, comparable to those seen with iMφ-specific media
   (Supplemental Fig. [80]2i). We also determined that the ratio of iAEOs
   and iMφ in co-cultures should be kept below 3:1, as a higher ratio of
   iMφ leads to increased levels of IL1β, potentially creating an
   environment similar to an inflammatory response (Supplementary
   Fig. [81]2j-k).

Fig. 2. Transcriptional regulation of iAlvAssemb system.

   [82]Fig. 2
   [83]Open in a new tab

   scRNA-seq was performed with ctrl-iAEOs and iAlvAssemb grown in KSFM
   media with biological replicates (n = 2), and Data was integrated with
   iMφ data ([84]GSE133935), previously reported. a, b The t-SNE plot
   represents the clustering of cell populations based on cell
   type-specific markers (a) and Library ID (b). c, d Violin plot and
   heatmap represents the log -transformed expression values of cell-type
   specific markers, including a box plot that highlights the
   interquartile range (IQR), median, and mean (horizontal line). e Venn
   diagram showing the numbers of DEGs in in both Mφ and AECs. f Heatmap
   of DEGs in both Mφ and AECs. g. Representative enriched pathway in DEGs
   of Mφ and AECs. h Violin plots showing log-transformed expression
   values of Alv-iMφ-enriched genes, includes a box plot highlighting the
   interquartile range (IQR), median and mean (horizontal line). i t-SNE
   plot showing the expression of Alv-iMφ-enriched genes in iAlvAssemb
   system and human lung tissues. j–l ctrl-iMφ, ctrl-iAEO and iAlvAssemb
   was grown in 50% GFR-matrigel with KSFM medium for 7days. j, k
   Representative flow cytometry dot-plots and quantitative analysis of
   macrophage-specific surface markers in ctrl-iMφ, and Alv-iMφ isolated
   from iAlvAssemb. Data were presented as mean ± SD with technical
   replicates (n = 3). l Representative immunostaining images showing the
   expression of the alveolar marker (MUC1) and AM-specific markers
   (MARCO, MSR1). Two independent experiment was repeated independently
   with similar results.

   In the iAlvAssemb system, we observed homing of Alv-iMφ around
   Alv-iAEOs (Fig. [85]1h–k, Supplementary Fig. [86]1e and Supplementary
   Movie [87]1). Immunostaining confirmed cell type-specific markers,
   verifying that the cellular identity of Alv-iMφ and Alv-iAEOs is
   consistent within the iAlvAssemb system. (Fig. [88]1k). In addition, we
   detected elevated secretion of chemokines such as CCL3, CXCL10, CCL5,
   and CCL1 in the iAlvAssemb, as well as cytokines such as IL-6,
   IL1ra/IL1f3, and IL1b (Fig. [89]1l–o), indicating functional maturation
   of macrophages in the Alv-iAEOs niche, providing a dynamic and
   interactive cellular environment within the iAlvAssemb. Taken together,
   these findings demonstrated the successful establishment of the
   iAlvAssemb model.

Interactive communication with Alv-iAEOs and Alv-iMφ within iAlvAssemb

   Next, we sought to investigate interactions between Alv-iAEOs and
   Alv-iMφ in the iAlvAssemb system using single-cell RNA sequencing
   (scRNA-seq) analysis. Cells were first grouped using unsupervised
   dimensionality reduction (t-SNE; t-Distributed Stochastic Neighbor
   Embedding) and then annotated using representative markers for each
   cell type (Fig. [90]2a–d and Supplementary Fig. [91]3). We found that
   the iMφ formed a single cell cluster, whereas the iAEO population was
   composed of three different clusters: AECs, proliferating AECs, and
   pulmonary neuroendocrine cells (Fig. [92]2a–d and Supplementary
   Fig. [93]3). The AECs includes both AT2-like cells and AT1-like cells
   (Supplementary Fig. [94]3). Interestingly, in the iAlvAssemb, Alv-iAECs
   and Alv-iMφ formed distinct clusters relative to the original cells,
   suggesting differential gene expression in these cells upon co-culture
   (Fig. [95]2a, b).

   Differentially expressed genes (DEGs) analysis highlighted changes in
   both Mφ and AECs, with a larger number of DEGs observed in Mφ
   (Fig. [96]2e, f), indicating dynamic cell-cell interactions. Through
   Reactome pathway analysis, we identified an increase in genes related
   to macrophage functions, including interleukin signaling, trans-Golgi
   networks, chemokine signaling, and lipid metabolism. Additionally, we
   also observed an upregulation of genes associated with surfactant
   metabolism (Fig. [97]2f, g). Of particular interest, we found human
   AM-enriched genes such as CD68, IL1b, CD163, ALOX5, MSR1, and MARCO
   were upregulated in the iAlvAssemb cultures (Fig. [98]2h–I and
   Supplementary Fig. [99]1f), indicating the acquisition of an AM-like
   phenotype within the iAlvAssemb. Indeed, through flow cytometry
   analysis, we confirmed the increased expression of surface markers
   CD14, CD16 and CD163 (Fig. [100]2j, k and Supplementary Fig. [101]4),
   and using immunocytochemistry, we validated that expression of
   AM-specific markers (MSR1 and MARCO) were upregulated (Fig. [102]2l).
   Collectively, these data demonstrate that Alv-iMφ acquire an AM-like
   state in iAlvAssemb cultures.

An AM-like Tissue Adaptation of Alv-iMφ in the Alv-iAEO Niche

   To investigate the molecular basis of the AM-like adaptation, we
   performed total RNA sequencing of ctrl-iMφ and isolated Alv-iMφ from
   iAlvAssemb (Fig. [103]3a, b). Overall, Gene Set Enrichment analysis
   (GSEA) revealed an enriched gene-signature of lung macrophage cells in
   Alv-iMφ compared to ctrl-iMφ (Fig. [104]3c).

Fig. 3. Tissue adaptation and maturation of iMφ in iAlvAssemb system.

   [105]Fig. 3
   [106]Open in a new tab

   Ctrl-iMφ and iAlvAssemb was grown in 50% GFR-matrigel with KSFM medium
   for 7 days and Alv- iMφ was isolated from iAlvAssemb. a Schematic
   images showing the method to isolate Alv-iMφ from iAlvAssemb. b
   Representative flow cytometry plots of macrophage markers (CD14, CD45)
   on Ctrl-iMφ and Alv-iMφ. c GSEA-enrichment plot showing gene signature
   of lung macrophage cells on Ctrl-iMφ and Alv-iMφ. d A scatter plot
   showing the correlation of gene expression (n = 7080) between induced
   macrophages (Mono-Mφ, Ctrl-iMφ, Alv-iMφ.) and tissue-resident alveolar
   macrophages (HBAM, original dataset from [107]GSE174659), normalized
   relative to monocytes. correlation coefficient (Pearson’s r) and p
   value was presented in figures. e Sample Distance matrix of indicated
   samples. f Heatmap of mouse AM-specific gene clusters in the lung from
   transcriptome data of macrophages in various tissues (left panel,
   original dataset: [108]GSE63340), Heatmap showing the relative
   expression of human orthologs of mouse AM-specific genes in ctrl-iMφ
   and alv-iMφ (right panel). g Relative mRNA expression level of
   significantly increased genes in undifferented hESC, Ctrl-iMφ, Alv-iMφ,
   and human lung. Alv-i iMφ was isolated from iAlvAssemb. Data were
   presented as mean ± SD. Statistical analysis was performed using an
   unpaired t-test between Ctrl-iMφ (n = 5) and Alv-iMφ (n = 3),
   biological replicates. P-value (<0.05) was indicated on the graph. h
   Representative phase-contrast, fluorescence images and quantifications
   showing phagocytosis ability. Data presented as mean ± SD with
   technical replicates; Ctrl-iMφ (n = 8), Alv-iMφ (n = 6), and Human AM
   (n = 5). Human AM (ACCEGEN) were cultured in Human alveolar macrophage
   media for experimental control. i Representative phase-contrast and
   lipid analogues (D3823, Green) fluorescence images, and quantifications
   showing lipid uptake ability. Data were presented as mean ± SD with
   technical replicates; Ctrl-iMφ (n = 219), Alv-iMφ (n = 185) and Human
   AM (n = 1058). Statistical analysis was performed using a one-way ANOVA
   with Dunnett’s multiple comparisons test, and p-value (<0.05) was
   indicated on the graph.

   To investigate Alv-iMφ‘ maturation, we performed a correlation analysis
   in which we compared gene expression changes in Alv-iMφ, Ctrl-iMφ,
   mono-Mφ, and HBAM (Human Alveolar macrophages from Bronchoalveolar
   lavage) (Fig. [109]3d). The analysis showed the highest correlation
   (Coefficient of Determination (R²) =0.5759) between Alv-iMφ and with
   HBAM ([110]GSE174659). The sample distance matrix analysis also showed
   that Alv-iMφ were most closely located to HBAM, lending further
   supporting to Alv-iMφ undergoing an AM-like maturation (Fig. [111]3e).
   To assess the lung-specific adaptation of Alv-iMφ, we compared it with
   previously reported RNA sequencing data of mouse lung ([112]GSE63340)
   and identified 460 AM-enriched genes among the diverse tissue-resident
   macrophages. Through gene expression analysis of human orthologues, we
   then identified 180 genes that were upregulated in Alv-iMφ
   (Fig. [113]3f), and of those genes validated by RT-PCR that IL1B, IL8,
   MSR1, ALOX5, and GPNMB gene expression was significantly increased
   (Fig. [114]3g and Supplementary Fig. [115]1f).

   We were intrigued to see that mRNA levels of surfactant genes (typical
   markers of AT2 cells) increased in Alv-iMφ, but not in mono- Mφ or
   Ctrl-iMφ (Fig. [116]3f and Supplementary Fig. [117]5a–c), consistent
   with scRNA-seq data of human lung tissue (Supplementary Fig. [118]5d).

   To assess functionality of Alv-iMφ, we examined phagocytosis and lipid
   uptake abilities of ctrl-iMφ and Alv-iMφ. We confirmed that
   phagocytosis was occurring in over 80% of cells in both groups, and
   that the ability to take up lipids was enhanced in Alv-iMφ
   (Fig. [119]3h-i), indicating functional improvement.

Functional Response of Alv-iMφ to Alveolar Damage

   AMs feature highly characteristic thin and elongated pseudopodia that
   enable interaction with inhaled particles and microorganisms by
   extending into the alveolar space^[120]16,[121]17. We observed that
   Alv-iMφ were capable of developing pseudopodia-like structures and
   contained lamellar bodies in the phagosome, indicating functional
   phagocytosis and recycling of surfactant lipids (Fig. [122]4a).

Fig. 4. Phagocytic clearance of damaged AECs by Alv-iMφ.

   [123]Fig. 4
   [124]Open in a new tab

   a Representative TEM images of Ctrl-iAEOs, Ctrl-iMφ, and iAlvAssemb
   cultured in KSFM media. The left panel shows Ctrl-iAEO with lamellar
   bodies indicated by solid arrows, while the middle panel presents Ctrl-
   iMφ. In iAlvAssemb (right panel), iAEO is positioned on the left side
   of the dashed line, with the recruited iMφ on the right. An inset on
   the left highlights the extended pseudopodia in iAlvAssemb, while
   another on the right shows lamellar bodies within the phagosome of iMφ.
   b Relative mRNA expression level of damage-response genes (p21, BAX,
   ISG15) by bleomycin treatment for 48 h. Data were presented as
   mean ± SD with technical replicates; Ctrl-iMφ, Ctrl-AEO, iAlvAssemb
   (n = 6). Statistical analysis was performed using a two-way ANOVA with
   Sidak’s multiple comparisons test, and p-value (<0.05) was indicated on
   the graph. c Representative phase-contrast and live (Green)/dead (Red)
   fluorescence images by bleomycin treatment for 48 hr in Ctrl-iMφ,
   Ctrl-iAEOs, and iAlvAssemb. d Representative phase-contrast and
   peroxidation lipid (Green)/normal lipid (Red) fluorescence images by
   bleomycin treatment for 48 h in Ctrl-iMφ, Ctrl-iAEOs, and iAlvAssemb.

   To evaluate whether the AlvAssemb platform could replicate in vivo
   cellular function, we induced alveolar epithelial injury using
   bleomycin. After 48 h of treatment, expression of damage-response
   genes, including p21, BAX, and ISG15, was upregulated in Ctrl-iAEOs,
   Ctrl-iMφ, and iAlvAssemb (Fig. [125]4b). However, using live/dead
   staining and immunostaining for the cleaved Caspase 3, we detected
   significantly fewer dead cells in the iAlvAssemb than in the Ctr-iAEOs
   only (Fig. [126]4c and Supplementary Fig. [127]6), suggesting that
   Alv-iMφ effectively removes damaged alveolar cells in the iAlvAssemb, a
   crucial in vivo function of AMs in tissue homeostasis. To validate this
   capability, we stained for oxidized lipids given that AMs are known to
   take up oxidized lipids produced from dead cells^[128]18. Our data
   demonstrate that oxidized lipids were indeed present in damaged AECs in
   Ctrl-iAEOs, but not in Alv-iMφ. In contrast, the Alv-iMφ in the
   iAlvAssemb exhibited an oxidized lipid signal, indicating uptake of
   oxidized lipids in Alv-iMφ during phagocytic clearance of damaged
   debris (Fig. [129]4d). Overall, these findings suggest that Alv-iMφ are
   structurally and functionally mature and that the iAlvAssemb represent
   a promising cellular platform for investigating functional human
   cell-cell interactions known to occur in vivo.

GM-CSF, a paracrine for tissue adaptation of Alv-iMφ

   To investigate the factors contributing to tissue adaptation of iMφ in
   the iAlvAssemb, we established a parabiotic organoid culture system in
   which iAEOs and iMφ are sharing the same media but cultured in separate
   matrigel domes within the same dish (Fig. [130]5a). In this setup, we
   detect upregulation of Alv-iMφ-enriched genes (ALOX5, MSR1, CCL4, IL1b,
   and IL8) in the para-iMφ, indicating that the presence of iAEO-derived
   paracrine factors regulate the tissue adaptation of iMφ (Fig. [131]5b).

Fig. 5. Paracrine effects of GM-CSF in iAlvAssemb system.

   [132]Fig. 5
   [133]Open in a new tab

   a A schematic explanation of parabiotic organoid culture systems. b
   Relative mRNA expression level of Alv-iMφ-enriched genes (ALOX5, MSR1,
   CCL4, IL1β, IL8) in the Crtl-iMφ and Alv-iMφ of parabiotic organoid
   system. Data were presented as mean ± SD with biological replicates
   (n = 3). Statistical analysis was performed using an unpaired t-test,
   and p-value (<0.05) was indicated on the graph. c Gene expression of
   SFTPC, GM-CSF and G-CSF in the AECs in Crtl-iAEO and iAlvAssemb on
   t-SNE plot of scRNA-seq analysis. d Representative cytokine blot array
   images of GM-CSF in conditioned media of Crtl-iAEOs, Crtl-iMφ and
   iAlvAssemb. e Relative mRNA expression level of CSF2R in Crtl-iAEOs,
   Crtl-iMφ, iAlvAssemb and human lung. Data were presented as mean ± SD
   with biological replicates (n = 6) for Crtl-iAEOs, Crtl-iMφ, iAlvAssemb
   and statistical analysis was performed using a one-way ANOVA with
   Dunnett’s multiple comparisons test. p-value (<0.05) was indicated on
   the graph. f A schematic explanation of G-CSF or GM-CSF treatment on
   Ctrl- iMφ. g Relative mRNA expression level of Alv-iMφ-enriched genes
   in Ctrl-iMφ with or without treatment of G-CSF and GM-CSF at 10 or
   100 ng/ml for 7 days. Data were presented as mean ± SD with technical
   replicates (n = 6) and statistical analysis was performed using a
   one-way ANOVA with Dunnett’s multiple comparisons test. p- alue (<0.05)
   was indicated on the graph. h, i Representative immunostaining images
   showing the expression of AM specific marker (MSR1, Red) and lipid
   uptake (D3823, Green) in iMφ with or without 100 ng/ml of GM-CSF, and
   quantification of the F.I. Data were presented as mean ± SD with
   technical replicates; n (h) = 5, n (i) = 159. Statistical analysis was
   performed using an unpaired t-test, and p-value (<0.05) was indicated
   on the graph.

   We detected high expression of GM-CSF in SFTPC^+ alveolar type 2 cells
   (AT2) by scRNA-seq (Fig. [134]5c). Using cytokine array analysis, we
   documented that GM-CSF is secreted in both iAEOs and iAlvAssemb, and
   confirmed by RT-PCR that the GM-CSF receptor (CSF2RA) was expressed in
   both Ctrl-iMφ and iAlvAssemb (Fig. [135]5d, e). When we treated
   Ctrl-iMφ with GM-CSF (but not G-CSF), we found that expression of
   Alv-iMφ-enriched genes was increased (ALOX5, MSR1, CCL4, IL1b, and IL8)
   (Fig. [136]5f, g). Immunostaining confirmed increased expression of
   MSR1, a marker specific to AMs, as well as enhanced lipid uptake by
   GM-CSF treatment (Fig. [137]5h, i and Supplementary Fig. [138]7a, b).
   Moreover, by applying a GM-CSF blocking antibody, we could partially
   inhibit the upregulation of Alv-iMφ-enriched genes in the iAlvAssemb
   (Supplementary Fig. [139]7c). To clarify the role of GM-CSF in Alv-iMφ
   adaptation, we created CSF2-/- iAEOs through CRISPR/Cas9-based gene
   editing (Supplementary Fig. [140]8a–c and Fig. [141]6a–d) and verified
   the absence of GM-CSF secretion using enzyme-linked immunosorbent assay
   (ELISA) (Fig. [142]5l). When iMφ were co-cultured with CSF2-/- iAEOs,
   gene expression for IL1b, IL8, CCL4, and ALOX5, typically elevated in
   Alv-iMφ due to interaction with iAEOs, was not increased (Fig. [143]6f,
   g). Furthermore, lipid uptake was reduced in Alv-iMφ co-cultured with
   CSF2-/- iAEOs (Fig. [144]6h, i), indicating a significant role of
   GM-CSF in the tissue adaptation of Alv-iMφ in iAlvAssemb.

Fig. 6. The role of GM-CSF on the tissue adaptation of Alv-iMφ.

   [145]Fig. 6
   [146]Open in a new tab

   a Illustration of genomic DNA showing target of the dual guide RNA and
   genotyping primers used to produce CSF^-/- iAEO. b Schematic image
   showing the strategy for CSF2^-/- iAEO generation. c Agarose gel
   electrophoresis of PCR amplified products for CSF2 genotyping. d Phase
   contrast images and fluorescence images for Lysotracker staining of
   CSF2-/- iAEO lines cultured in iAEO media. e Quantification of GM-CSF
   secretion in conditioned media of WT-, CSF-/- iAEO #1, and #9 lines
   cultured in KSFM media. Data are presented as mean ± SD with technical
   replicates (n = 3). ND, not detected. (m–p). f Schematic images showing
   the parabiotic organoid cultures of iMφ with Ctrl-iMφ, wt-iAEO or
   CSF-/--iAEO in KSFM media for 7 days. g Relative mRNA expression level
   of Alv-iMφ-enriched genes (IL1β, IL8, CCL4, ALOX5 and MSR1). Data
   presented as mean ± SD with technical replicates (n = 3). Statistical
   analysis was performed using a one-way ANOVA with Dunnett’s multiple
   comparisons test, and p value (<0.05) was indicated on the graph. h.
   Representative florescence images showing the lipid uptake (D3823,
   Green) in para-iMφ with WT-, CSF-/- iAEO #1, and #9. i. Quantification
   of the F.I. Data were presented as mean ± SD with technical replicates;
   WT (n = 209), #1 (n = 180) and #9 (n = 224). Statistical analysis was
   performed using a one-way ANOVA with Dunnett’s multiple comparisons
   test, and p-value (<0.05) was indicated on the graph.

   However, the surfactant-related genes that were confirmed to be
   increased in Alv- iMφ in Supplementary Fig. [147]4c were not elevated
   by GM-CSF treatment (Supplementary Fig. [148]9a). Instead, a
   significant amount of SFTPC mRNA was observed in exosomes secreted from
   Ctrl-iAEOs (Supplementary Fig. [149]9b, c). This suggests that the
   SFTPC mRNA detected in Alv- iMφ may result from exosome delivery from
   AT2 cells to Alv- iMφ. These findings are consistent with the
   observation of SFTPC mRNA in various non-parenchymal cells, such as
   macrophages and endothelial cells, in human lung tissue (Supplementary
   Fig. [150]4d), which might imply exosome-mediated cell interaction.

Interleukins-mediated surfactant regulation in AECs

   In the iAlvAssemb t-SNE plot, we noted a distinct cluster of AECs,
   reflecting significant gene expression changes resulting from
   interactions with Alv-iMφ (Fig. [151]7a). Upregulated genes were
   associated with interleukin signaling, immune system, and chemokine
   signaling pathways (Supplementary Fig. [152]10a–d). Additionally, we
   detected an increase in genes related to surfactant metabolism,
   including SFTPA1 and SFTPA2 (Fig. [153]7b–d), without any change in the
   expression of markers for AT1 cells (Fig. [154]7e). We validated these
   gene alterations by RT-PCR (Fig. [155]7f), and noted similar gene
   expression increases in indirect parabiotic organoids culture
   (Fig. [156]7g, h), suggesting that paracrine signaling instructs these
   gene expression changes. Notably, we found that IL-1β and IL-6
   treatment led to upregulation of both SFTPA1 and SFTPA2 in a
   concentration-dependent manner (Fig. [157]7i). Conversely, at higher
   concentrations, both IL-1β and IL-6 inhibited SFTPC expression. The iMφ
   differentiated from IL1B-/- or IL6-/- hESCs exhibited increased IL-6
   secretion in IL1B-/- iMφ, and vice versa, likely due to functional
   compensation between these cytokines (Supplementary Fig. [158]11a–f).
   In the co-culture with IL6-/- iMφ, which lacked IL-6 but showed
   increased IL-1β levels, there was a significant increase in the
   expression of surfactant genes SFTPA1, SFTPC, along with a reduction in
   the expression of IL1R1 and SLC34A2 (Supplementary Fig. [159]11g, h).
   Conversely, co-culture with IL1B-/- iMφ led to increased expression of
   IL1R1 and SLC34A2 (Supplementary Fig. [160]11h), indicating that IL-6
   specifically impacts the expression of IL1R1 and SLC34A2 while both
   IL-1β and IL-6 influence surfactant-related gene expression.

Fig. 7. Paracrine effects of interleukin to Alv-iAEOs.

   [161]Fig. 7
   [162]Open in a new tab

   a t-SNE plots showing single cell distribution of individual replicates
   in Crtl-iAEOs and Alv-iAEOs. b, c Representative enrichment pathways in
   DEG analysis and Violin plots showing log expression level of
   surfactant metabolism related genes of Alv-iAEOs. Each violin includes
   a box plot showing the interquartile range (IQR), with horizontal lines
   indicating the median and mean. d, e Expression of genes related with
   AT2 cells (SFTPA1, SFTPA2, NAPSA, ABCA3 and IL1R1) (d) and AT1 cells
   (AGER, CLIC5) (e) on t-SNE. f Relative mRNA expression level of
   surfactant metabolism (SFTPA1, SFTPA2, SFTPC, SLC34A2 and ABCA3) and
   immune related genes (IL1R1, LCN2, IL8 and CXCL2) in Ctrl-iAEOs and
   Alv-iAEOs isolated from iAlvAssemb cultured in KSFM media. Data were
   presented as mean ± SD with biological replicates; Ctrl-iAEO (n = 5),
   Alv-iAEO (n = 7). Statistical analysis was performed using an unpaired
   t-test, and p-value (<0.05) was indicated on the graph. g, h A
   schematic explanation (h) of parabiotic culture of iAEOs with iMφ in
   KSFM media for 7 days. h Relative mRNA expression level of surfactant
   metabolism (SFTPA1, SFTPA2, SFTPC, SLC34A2 and ABCA3) and immune
   related genes (IL1R1and LCN2) in ctrl-iAEO and para-iAEO. Data were
   presented as mean ± SD with biological replicates; Ctrl-iAEO (n = 4),
   Para-iAEO (n = 3). Statistical analysis was performed using an unpaired
   t-test, and p-value (<0.05) was indicated on the graph. i Relative mRNA
   expression level of surfactant metabolism (SFTPA1, SFTPA2, SFTPC,
   SLC34A2 and ABCA3) and immune related genes (IL1R1 and LCN2) in iAEOs
   with or without IL1β or IL6 in KSFM media for 7days. Data were
   presented as mean ± SD with technical replicates (n = 3). Statistical
   analysis was performed using a one-way ANOVA with Dunnett’s multiple
   comparisons test, and p-value (<0.05) was indicated on the graph.

Lipopolysaccharide (LPS) response and Mycobacterium tuberculosis (M.tb)
infection of iAlvAssemb

   To assess iAlvAssemb’s capability to model immune responses resulting
   from lung infection, we performed scRNA-seq analysis after LPS
   response. In t-SNE analysis, we noted distinct clustering of
   macrophages and IL1R1^+ AECs in LPS-treated vs. non-treated iAlvAssemb
   (Fig. [163]8a–f). Amongst DEGs, we found 37 common alterations in AECs
   and Mφ, suggesting shared changes in antigen presentation, ER-phagosome
   pathway, and interferon signaling (Fig. [164]8g–i, and Supplementary
   Fig. [165]12a, b). Some cytokines, chemokines, and surface molecules
   like as IL6, IL1B, CXCL11, CXCL9, CD38, CD274, SLAMF7 were upregulated
   only in Alv-iMφ (Fig. [166]8j–l), while in AECs, IL32 expression was
   altered and expression of AT1 cell markers (HOPX, AQP4, CLIC5 and AGER)
   was reduced (Fig. [167]8m). Notably, there was a significant decrease
   in the expression of ribosomal proteins and translation-related genes
   in Alv-iMφs, but not in AECs (Fig. [168]8i and Supplementary
   Fig. [169]12c), suggesting a possible differential regulation of cell
   type-specific translation by infection,

Fig. 8. iAlvAssemb as a disease modeling system with LPS.

   [170]Fig. 8
   [171]Open in a new tab

   a t-SNE Plot of Merged iAlvAssembs with and without 1 ng/ml LPS
   Treatment for 24 h in KSFM media (K-means clustering). b, c t-SNE Plot
   showing the expression of genes related with macrophage (PTPRC, CD14)
   and alveolar epithelial markers (EPCAM, NKX2.1, SFTPC, MUC1). d Cell
   clustering and cell type annotation of iAlvassemb system. e Violin
   plots showing the log gene expression of cell type specific markers in
   each cluster. Each violin includes a box plot showing the interquartile
   range (IQR), with horizontal lines indicating the median and mean. f
   t-SNE plots showing the difference in cell clusters based on library
   id, and expression of CD68 and IL1R1. g, h Venn diagram and heatmap of
   DEG analysis with LPS-response in AEC and iMφ. i List of enriched
   Reactome Pathways in iAlvAssemb after LPS treatment. j–m Violin plots
   showing gene expression related with interleukin signaling (j),
   chemotaxis (k), Mφ (l) and AEC-specific markers (m) after LPS
   treatment. Each violin includes a box plot showing the interquartile
   range (IQR), with horizontal lines indicating the median and mean.

   We also found that expression of pathogen entry receptors was altered
   in co-culture of iAEOs and iMφ (Supplementary Fig. [172]13a and
   [173]S13b). Specifically, in Alv-iMφ, we detected and validated by
   RT-PCR increased expression of CD44 (Supplementary Fig. [174]13c, d), a
   cell surface glycoprotein that mediates recruitment of inflammatory
   cells and that is a macrophage binding site for Mycobacterium
   tuberculosis (M.tb). We also show that CD44 is upregulated in response
   to GM- CSF stimulation (Supplementary Fig. [175]13e). To model
   infection in the iAlvAssemb platform, we employed fluorescence-tagged
   M.tb (H37Rv) and examined whether iMφ were permissive for M.tb (H37Rv)
   from a multiplicity of infection (MOI), ranging from 1 to 20 bacteria
   per cell, in a dose-dependent manner. We found that the highest
   infection led to effective intracellular replication (Supplementary
   Fig. [176]14), demonstrated that the virus preferentially infected iMφ
   (Supplementary Fig. [177]13f–g), consistent with its known in vivo
   tropism for AM as primary targets. Infection efficiency was even higher
   in isolated Alv-iMφ (Supplementary Fig. [178]13h, i), emphasizing the
   relevance of AM in M.tb infection. To leverage the iAlvAssemb model
   system for M.tb infection, we measured cellular fate decision, a
   hallmark of M.tb intracellular replication. iMφ and Alv-iMφ underwent
   apoptosis and necrosis upon M.tb infection and, Alv-iMφ were more
   sensitive to the M.tb infection resulting in a high level of necrotic
   cell death compared to iMφ (Supplementary Fig. [179]15a, b). To extend
   our understanding of the necrotic cell death, we analyzed metabolic
   activity in Alv-iMφ, and found that the major cellular energy-producing
   metabolisms such as glycolysis and oxidative phosphorylation were
   decreased in the Alv-iMφ by M.tb infection (Supplementary
   Fig. [180]15c–g) which recapitulated the metabolic responses of
   mono-Mφ. Taken together, these findings support that iAlvAssemb can be
   leveraged as human disease-relevant model system to investigate
   inflammation and respiratory infectious disease^[181]13, offering
   insights into lung microenvironment tropism and infection dynamics.

Discussion

   Our study makes a significant stride towards modeling the intricate
   cellular dynamics within the human lung environment, and specifically
   focusing on the interactions between AECs and AMs. By leveraging
   organoid technology, we derived both iAEOs and iMφ from hPSCs, and
   successfully established the iAlvAssemb system, which incorporates
   functional macrophages. Our advance opens up new avenues for modeling
   human lung infection and disease, gaining novel insights that could
   inform the development of therapeutic strategies for lung diseases, and
   testing compounds for efficacy in preclinical disease paradigms.

   We found that both iAEOs and iMφ, when co-cultured in the iAlvAssemb
   model, underwent distinct changes, highlighting how their interactive
   milieu intricately influences and shapes both cellular identity and
   function. Our work also underlines the importance of carefully
   designing and optimizing a co-culture medium capable of sustaining the
   unique identities of both cell types, an instrumental component of
   setting up the iAlvAssemb. Importantly, we demonstrate that the
   iAlvAssemb exhibits cellular characteristics and interactions akin to
   those of lung tissue, as evidenced by our data showing that the iMφ
   acquired a global gene expression profile similar to human tissue
   macrophages and upregulated specific genes known to be upregulated in
   human AMs. Our finding aligns well with the tissue adaptation of
   macrophages, in which the tissue environment shapes the identity and
   function of macrophages^[182]19,[183]20. Interestingly, we also
   observed an increase in the expression of surfactant genes in Alv-iMφ,
   suggesting that AMs not only contribute to surfactant clearance but
   also play a role in its production. Collectively, the iAlvAssemb
   provides a versatile platform for studying macrophage biology within a
   tissue-like context, facilitating investigations into their functions
   and contributions to pulmonary physiology and disease.

   A significant finding was the Alv-iMφ were capable of executing
   functional phagocytosis and lipid uptake, and that they were
   morphologically similar to AM. Our data demonstrate that the Alv-iMφ
   remove damaged alveolar cells and absorb oxidized lipids, validating
   the iAlvAssemb’s potential in modeling diseases associated with
   macrophage dysfunction or alveolar damage. These functional data are
   particularly significant given the role of macrophages in maintaining
   homeostasis in lung tissue and their involvement in various lung
   diseases.

   Our study also noted the impact of paracrine factors, and particularly
   GM-CSF expressed by AT2-like cells in iAEOs, in influencing the
   characteristics of Alv-iMφ. This observation unveils a potential
   mechanism through which epithelial-macrophage interactions could
   dictate the characteristics and functions of AMs within the human lung.
   Interestingly, the importance of GM-CSF in mouse AM development was
   previously reported in a study on fetal lung development, further
   supporting the significance of this interaction^[184]21,[185]22.
   Additionally, we observed that co-culture with iMφ led to changes in
   expression of genes related to epithelial cell defense in iAEOs, and
   particularly increased levels of surfactant protein A (SFTPA), an
   essential component of the lung’s innate immune
   response^[186]23,[187]24. Intriguingly, we demonstrate mechanistically
   that the changes in iAEOs were mediated by interleukin-1b and
   interleukin-6 secreted by Alv-iMφ, shedding light on how macrophages
   shape the functional characteristics of AECs.

   Furthermore, our study demonstrated that iAlvAssemb generate an immune
   response in response to LPS stimulation, suggesting utility as a 3D
   cellular in vitro model for studying the lung immune response to a
   bacterial infection. The LPS-induced changes in chemokines and
   cytokines we annotated, along with distinct clustering patterns and
   altered gene expression profiles in macrophages and AECs, serve as
   benchmarks against which other infectious agents or co-administered
   therapeutic agents can be investigated in the iAlvAssemb model. To this
   end, we demonstrate the ability of iAlvAssemb to recapitulate the
   cellular tropism observed in M.tb infection, a significant step towards
   establishing a robust in vitro model for studying lung infections.
   Taken together, we propose that iAlvAssemb is capable of replicating
   aspects of the elusive immune responses of the human lung, thereby
   making it a valuable research platform for studying cell-specific
   defense responses and their interactions.

   In conclusion, we here describe the iAlvAssemb platform, a model that
   enables understanding cellular interactions representative of the human
   lung environment and mechanistic interrogations of macrophage-mediated
   immune responses. By combining iAEOs and iMφ in a dynamic tissue
   environment, iAlvAssemb enables disease modeling studies of the complex
   interplay between AECs and AMs, interrogations of downstream mechanisms
   impacting on cellular identity and function, and development of
   therapeutic strategies for lung diseases. Despite these significant
   advances, we recognize that our model does not encompass other cell
   types that also contribute to lung tissue function, such as
   fibroblasts, endothelial cells, and diverse immune cells. Furthermore,
   the detailed mechanisms of which types of iMφ participate in homing and
   what factors induce homing remain unknown. Future efforts are needed to
   further refine iAlvAssemb to better replicate the highly complex lung
   tissue, along with more sophisticated analyses of cell-cell
   interactions.

Methods

Human pluripotent stem cells (hPSCs)

   In this study, one hESC lines (H9) and one hiPSC (CRLe2-iPSC) were
   used. hESCs (H9) were obtained from the WiCell Research Institute
   (Wicell, Madison, WI) and hiPSC were reprogrammed from skin fibroblast
   (ATCC, Manassas, VA) by episomal vector. hPSCs were maintained in
   mTeSR-1 medium (STEMCELL^TM Technologies, Vancouver, Canada) and
   subcultured with ReLeSR (STEMCELL^TM technologies) as following
   manufacturer’s instructions. This study was approved by Korean Public
   IRB (No. P01-2014-ES-01-09, and P01-201609-31-002).

Differentiation and maintenance of Alveolar Epithelial Organoids (iAEO) from
hPSCs

   iAEOs were differentiated from hPSCs by previously
   reported^[188]10,[189]25, with modifications. Briefly, hPSCs were grown
   in 1% matrigel (Corning, N.Y.)-coated dish with the mTeSR-1 medium. To
   differentiate into definitive endoderm, hESC were dissociated into
   single cells using Accutase and transferred to matrigel-coated dishes
   at a density of 2*10^5 cells/cm^2. Cells were cultured for 24 h in
   mTeSR-1 medium with 10 μM Y27632 (R&D Systems, Minneapolis, MN, USA),
   and differentiated into definitive endoderm by using STEMdiff™
   Definitive Endoderm Kit (STEMCELL^TM technologies) for 72 h. For
   anterior foregut specifications, definitive endoderm cells were
   transferred to matrigel-coating dish with 1:4 split and exposure to
   2 μM dorsomorphin (Sigma-Aldrich, Burlington, MO) and 10 μM SB4531542
   (Tocris Bioscience, Bristol, UK) in SFM medium (75% IMDM, 25 % Ham’F12,
   0.5X N2, 0.5X B27, 0.056% BSA, 1% P/S, 1% Glutamax, 50 μg/ml Ascorbic
   Acid, 0.4 μM Monothioglycerol) for 72 h. Then, Lung endoderm cells were
   induced by SFM medium supplemented with 3.5 μM CHIR 99021 (Tocris),
   10 ng/ml hrBMP4 (R&D Systems) and 100 nM Retinoic Acid (Sigma-Aldrich)
   in cells on day9. On differentiation day 15, cells were detached into
   small clumps by treatment of Gentle Cell Dissociation Reagent
   (STEMCELL^TM technologies), and embedded matrigel dome in the SFM
   medium supplemented with 3.5 μM CHIR 99021, 10 ng/ml FGF7 (R&D
   Systems), 10 ng/ml FGF10 (R&D Systems), 500 nM Retinoic Acid. Seven
   days later, NKX2.1 positive cells were sorted by FACSAria™ III Cell
   Sorter (BD Biosciences, San Jose, CA) for the CD26-PE^-/CD47-PerCP
   Cy5^+ cells and re-embedded in a matrigel dome. Further differentiation
   into AECs, cells were differentiated by SFM medium supplemented with
   3.5 μM CHIR9921, 10 ng/ml FGF7, 50 nM Dexamethasone (Sigma-Aldrich),
   100 μM 8-br-cAMP (Biolog, Hayward, CA), 100 μM IBMX (Sigma-Aldrich). On
   days 30-35, CHIR9921 was withdrawn and later it was added back. To
   enrich iAEOs were differentiated from hESCs by previously reported
   iAEOs were differentiated from hESCs by previously reported iAEO, the
   organoids were selectively cultured to identify clones that stained
   positively with Lysotracker after low-density single-cell seeding. For
   long-term maintenance, organoids were cultured in a matrigel dome with
   SFM medium supplemented with 3.5 μM CHIR9921, 10 ng/ml FGF7, 50 nM
   Dexamethasone, 100 μM 8-br-cAMP, 100 μM IBMX, and 10uM SB431542, and
   passaged by small clump every 2weeks.

Differentiation and maintenance of iMφ from hPSCs

   Differentiation of iMφ from hPSC was performed by the method we
   reported previously^[190]13. Briefly, hPSCs colonies were cultured in
   mTESR medium until they reached a size of approximately 2 mm, and
   mesoderm specification was performed by HDM medium (APEL2 (STEMCELL^TM
   technologies), 1x Insulin-Transferrin-Selenium-X (Thermo Fisher
   Scientific Inc., Waltham, MA)) supplemented with 100 ng/ml BMP4 (R&D
   Systems) for 48 h (D0-2). Subsequently, the concentration of BMP4 was
   reduced to 10 ng/ml, and the cells were incubated for an additional two
   days (D2-4). To induction of hematopoietic stem and progenitor cells,
   the medium was replaced with HDM medium containing 40 ng/ml VEGF (R&D
   Systems) and 50 ng/ml SCF (R&D Systems) for 2 days (D4-6), and then
   further differentiated in the HDM medium supplemented with 10 ng/ml TPO
   (R&D Systems), 50 ng/ml IL-3 (PeproTech, Cranbury, NJ), 50 ng/ml IL-6
   (PeproTech), 50 ng/ml Flt-3L (R&D Systems) and 50 ng/ml SCF for 10days.
   On day 15, floating cells were harvested and transferred into a 6-well
   plate and culture with RPMI 1640 (ThermoFisher Scientific, Gibco) media
   supplemented with FBS and 100 ng/ml M-CSF (PeproTech) for macrophage
   differentiation. The iMφ were scalable for up to 2 months, with
   sub-culturing every 7–10 days.

Generation iAlvAssemb by co-culturing of iAEO and iMφ

   iAEO was dissociated into a large clump by a few pipetting with 200 μl
   tips, and parts of iAEO were dissociated with single cells to count
   cells. To co-culture, 3 ~ 5*10^5 of iAEO and 0.5 ~ 1*10^5 of iMφ were
   mixed with 500 μl of 50% GFR-matrigels and made a 20ul matrigel dome.
   The co-culture medium was used RPMI1640 medium supplemented with 10%
   FBS, 100 ng/ml MCSF, 2.5 μM SB431542, and 2.5 ng/ml hrFGF7. Y27632 was
   exclusively added on the first day of culturing, and the co-cultures
   were typically incubated for one week before further analysis. For
   enhanced AEC integrity, particularly in AT2 cells with higher
   expression of SFTPC, low-dose dexamethasone (<5 nM) could be optionally
   included.

   To induce damages in iAlvAssemb, Bleomycin (TCI, Tokyo, Japan) was
   applied at a concentration of 40 μg/ml for 48 h. Additionally, LPS
   (Sigma-Aldrich) was treated at a concentration of 1 ng/ml for 24 h. To
   evaluate the effectiveness of the paracrine factors, GM-CSF (R&D
   Systems), G-CSF (R&D Systems), and an anti-human GM-CSF monoclonal
   antibody (Mabtech, Nacka Strand, Sweden) were administered to iMφ. In
   the case of IL-1β (R&D Systems), and IL-6 (R&D Systems), they were
   treated in iAEOs using co-culture media.

Live staining with fluorescence dye

   To perform live staining of the iAEO, the organoids were treated with
   LysoTracker Red DND-99 (ThermoFisher Scientific, Invitrogen), which was
   diluted at a ratio of 1:20000, for a duration of 2 h. For tracking iMφ
   in the iAlvAssemb, the iMφ cells were labeled using the PKH67 Green
   Fluorescent Cell Linker Mini Kit for General Cell Membrane Labeling
   (Sigma-Aldrich), following the instructions provided by the
   manufacturer. Live cell imaging was carried out using either a
   fluorescence microscope from Olympus (Tokyo, Japan) or a confocal
   microscope from Carl Zeiss AG (Oberkochen, Germany).

Real Time-Polymerase Chain Reaction (RT-PCR)

   Total RNA was extracted by Easyblue^TM Total RNA Extraction Kit (iNtRON
   Biotechnology, Seongnam-si, Korea) following the manufacturer’s
   instructions. cDNA was synthesized from 1.2 μg of total RNA using
   SuperScript^TM IV First-Strand Synthesis System (Thermo Fisher
   Scientific). Quantitative RT-PCR was performed using a 1/10
   concentration of synthetic cDNA with Fast SYBR Green Master Mix
   (Applied Biosystems, Waltham, MA) on an 7500 Fast Real-Time PCR
   instrument system (Applied Biosystems). Gene expression levels were
   normalized to hGAPDH or hTBP. The primer sequences employed in the
   study are presented in Supplementary Data [191]1.

Immunocytochemistry (ICC)

   Organoids in the matrigel dome were fixed in 4% PFA (Biosesang,
   Yongin-si, Korea) for 2-3 min, followed by three washes with dPBS.
   Subsequently, the samples were treated with 30% sucrose overnight.
   Then, samples were embedded in Tissue-Tek® O.C.T. Compound (Sakura
   Finetek USA, Torrance, CA) and rapidly frozen using liquid nitrogen.
   Then, 10 μm sections were cut and affixed onto slide glass (MATSUNAMI,
   Japan).

   For immunostaining, cryosections were re-fixed in 4% PFA for 10 min,
   followed by washing with dPBS. Next, the samples were permeabilized
   using 0.1% Triton X-100 for 30 min and washed three times with dPBS. To
   block non-specific binding, the washed samples were treated with 3%
   BSA. Primary antibodies were diluted in 1% BSA and incubated with the
   samples overnight at 4 °C. Following fluorescence-conjugated antibodies
   were utilized:Anti-TTF1 antibody (NKX2-1) (Abcam, Cambridge, UK),
   Anti-Mature SP-B (SevenHills bioreagents, Cincinnati, OH), Anti-Mature
   SP-C (SevenHills bioreagents), Anti-HOP (E1) (Santa Cruz Biotechnology,
   Dallas, TX), Anti-MUC1 (Santa Cruz Biotechnology), Anti-MAC387 (Abcam),
   Anti-MARCO (Abcam), and Anti-CD204 (MSR1, Abcam). Subsequently, the
   samples were washed with 0.02% TBST and incubated with secondary
   antibodies (Jackson Laboratory, Bar Harbor, ME) and Hoechst (Thermo
   Fisher Scientific) for 1 h at room temperature. After three additional
   washes in 0.02% TBST, the samples were mounted on slides using
   Fluorescence Mounting Medium (Dako, Carpinteria, CA) and covered with a
   cover glass. Finally, fluorescence samples were visualized using
   confocal microscopy.

ELISA

   IL-6, IL1β, and GM-CSF secreted from cells were analyzed using the
   Quantikine ELISA Kit (R&D system, MN, United States) according to the
   manufacturer’s instructions. Briefly, we collected the culture medium
   and added it to the microplate strip well with control medium. We then
   covered the strip and incubate it at room temperature for 1-2 h. After
   four washes, we added conjugates of IL-6, IL1β, and GM-CSF each and
   incubated at room temperature for 2 h with cover. After another wash,
   we added substrate solution and incubated the microplate at room
   temperature for 20 min in dim light, after which we added stop solution
   and analyzed the results using a microplate reader at 450 nm.

Cytokine array

   Cytokines were analyzed using the Proteome Profiler Human Cytokine
   Array kit (R&D system, MN, United States). While blocking the membrane,
   the collected cell culture medium is mixed with array buffer and human
   cytokine array detection antibody cocktail and incubated for one hour
   at room temperature. Add sample mixture to membrane and incubate at 4
   °C overnight with shaking. After washing, add streptavidin-HRP solution
   and incubate at room temperature for 30 min with shaking. After another
   washing, add chemi reagent mix to the membrane and incubate at room
   temperature for one minute. Remove remaining reagent mix and analyze
   with autoradiography cassette.

Gene editing

   To produce gene-edited cell lines, we designed six guide RNA target
   sequences per gene and verified their editing efficiency by
   transfection into 293 T cells. Two guide RNAs with high efficiency were
   selected and used to produce knockout (KO) lines for CSF2^-/- iAEO and
   IL6^-/-,IL1b^-/- hESC. The target sequences of sgRNA were presented in
   Supplementary Data [192]1.

   For generating knock-out iAEOs and hESC, the iAEOs and hESC were
   dissociated into single cells using Accutase. Total 10ug of plasmids
   encoding dual guide RNA, CRISPR/Cas9-RFP, and p53DD were then
   transiently introduced into the 1*10^6 cells by Neon electroporation
   system (Thermo Fisher scientific). The cell type-specific parameters
   for electroporation were as follows: hESCs: 1350vols, 20 ms width, 1
   pulse, iAEOs: 1400 volts, 30 ms width, 1 pulse. 24 h later,
   RFP-positive cells were sorted and seeded at low density for clonal
   selection, followed by genotyping to select knock-out lines.

Exosome isolation

   Exosomes were isolated by ExoQuick-TC^TM Exosome Precipitation Solution
   (System Biosciences, CA, United States) following the manufacturer’s
   instructions. Briefly, we collected the culture medium and centrifuged
   it for 15 min at 3000 g. We then transferred the supernatant to a clean
   tube and added 1/5 of the volume of ExoQuick-TC. After incubating the
   well-mixed solution for 12 h at 4°C, we centrifuged for 30 min at
   1500 g, aspirated the supernatant and extracted total RNA from the
   pellet.

Flow cytometry analysis

   Macrophages were collected and incubated with fluorescence-conjugated
   antibodies for 30 min at 37°C or 2 h at room temperature. Following
   fluorescence-conjugated antibodies were utilized: BB700 anti-human CD16
   Antibody (BD biosciences), CD45 monoclonal Antibody (2D1) conjugated to
   APC (eBioscience^TM, San Diego, CA.), BB515 Mouse Anti-Human CD45 (BD
   biosciences), Pharmingen^TM PE Mouse Anti-Human CD14 (BD biosciences),
   APC anti-human CD11c antibody (BioLegend,Cat #: 301614) and
   Pharmingen^TM FITC Mouse Anti-Human CD163 (BD biosciences). The stained
   cells were detected using a BD FACSVerse™ flow cytometer (BD
   Biosciences), and the data were analyzed using FlowJo_v10 software
   (FlowJo, LLC, Ashland, OR). To analyze apoptosis and necrosis, we mixed
   1000,000 cells each with 1 ul of CellEvent Caspase-3/7 Green Detection
   Reagent (Thermo, #[193]C10427). Following gentle vortexing, samples
   were incubated for 60 min at RT, shielded from light. As a final step,
   1 ul of a 1 mM solution of SYTOX AADvanced dead cell stain was added.
   Subsequently, the cells were subjected to analysis using a flow
   cytometer (CytoFLEX LX, Beckman Coulter), and the data were analyzed
   using FlowJo_v10 software (FlowJo, LLC, Ashland, OR).

Hematoxylin and Eosin (H&E) staining

   The samples on the slide were dehydrated sequentially using 70% and 50%
   ethanol for 2 min each. Following dehydration, the samples were stained
   with hematoxylin solution (Sigma-Aldrich) for 7 min. They were then
   washed in tap water until the water became colorless, which took
   approximately 15 min. In the subsequent step, the samples were stained
   with Eosin (Sigma-Aldrich) for 1 min. They were then sequentially
   dipped in 50%, 70%, 80%, 90%, and 95% ethanol solutions. Subsequently,
   the samples were incubated with 100% ethanol three times, with each
   incubation lasting for 2 min. Finally, the samples were mounted using a
   mounting solution (Thermo Fisher Scientific). The images were
   visualized using a BX53 microscope from Olympus.

Phagocytosis assay

   To compare the phagocytosis ability of iMφ or Alv-iMφ cells isolated
   from iAlvAssemb, the EZCellTM Phagocytosis Assay Kit (Red E.coli) from
   BioVision Inc. (Waltham, MA) was utilized. Human AM as an experimental
   control were purchased from ACCEGEN (Fairfield, New Jersey). On the 7th
   day of co-culture, the Matrigel dome was scraped off with a tip from
   the culture dish and then gently pipetted to separate the Matrigel from
   the iAlvAssemb. After centrifugation, the Matrigel layer was removed
   from the cell pellet and fresh medium was added. Following gentle
   pipetting to separate the iAEOs and adherent macrophages, the mixture
   was passed through a 70 μm cell strainer. Through this process, we
   successfully separated the Alv-iMφ from the iAlvAssemb, as the larger
   iAEOs remained on the cell strainer, while most of the Alv-iMφ and some
   dissociated iAEOs passed through with the medium. The isolated ctrl-iMφ
   and Alv-iMφ were then incubated with a diluted Red E.coli slurry in
   complete media for 2 h. After incubation, the samples were washed three
   times with Phagocytosis Assay Buffer. The fluorescent signals were
   observed using confocal microscopy. The fluorescence intensity per cell
   was quantified by the Image J software.

Lipid uptake

   iMφ or Alv-iMφ cells isolated from iAlvAssemb were treated with 5 μM
   BODIPYTM 500/510 (Thermo Fisher Scientific, Invitrogen) for 30 min.
   Subsequently, the cells were washed multiple times with media to remove
   any unbound dye. The absorbed fluorescent fatty acid analog was then
   visualized using confocal microscopy. The fluorescence of the cells was
   quantified using the Image J software.

Live-dead assay and live staining of lipid peroxidation

   iAEO, iMφ, and iAlvAssemb cells were cultured in confocal dishes, and
   after 5 days, they were treated with 40 ng/ml of bleomycin in the media
   for 48 h to induce cellular damages. For the live-dead assay, the
   Cyto3D Live-Dead Assay kit from TheWell Bioscience (North Brunswick,
   NJ) was used. The Cyto3D reagent in the kit was added to the dishes at
   a ratio of 1:50 of the total volume, and the organoids were incubated
   at 37 °C for 10 min. Subsequently, the cells were washed multiple times
   with medium.

   To visualize lipid peroxidation in live cells, BODIPYTM 581/591 C11
   (Lipid peroxidation sensor) from Thermo Fisher Scientific, Invitrogen
   was added to the media at a final concentration of 2 μM and incubated
   for 30 min. Following the incubation, the cells were washed multiple
   times with the used media. The fluorescent signals were then observed
   using confocal microscopy.

Transmission electron microscopy

   Organoid Matrigel domes were fixed in a solution of 2% glutaraldehyde
   and 2% paraformaldehyde in DPBS (pH 7.4) for 24 h at 4 °C. After
   washing, the samples were postfixed in 1% osmium tetroxide (Electron
   Microscopy Sciences, Hatfield, PA) for 1 h at 4 °C. Dehydration was
   performed using graded ethanol and propylene oxide, followed by
   infiltration with propylene oxide and EMBed 812 resin mixtures. The
   samples were embedded in pure EMBed 812 resin (EMS) and polymerized at
   60 °C for 24 h. Ultrathin sections (70 nm) were prepared using an
   ultramicrotome (Leica, Wetzlar, Germany), stained with uranyl acetate
   and lead citrate, and examined using a transmission electron microscope
   (FEI, Tecnai G2 spirit TWIN, Thermos Fisher Scientific). The work was
   conducted atthe EM & Histology Core Facility at the BioMedical Research
   Center, KAIST.

Single-cell RNA sequencing

   For the iAEOs, iAlvAssemb, and LPS-exposed iAlvAssemb assay groups, two
   biological replicates were collected, and single-cell RNA sequencing
   (scRNA-seq) was conducted by NICEM research facilities in Seoul
   National University (Seoul, Korea). Additionally, scRNA-seq was
   performed on control macrophages, and their sequencing data were
   analyzed by merging with our previously published [194]GSE133935
   data^[195]26. To explore gene expression at the single-cell level in
   human lung tissue, the scRNA-seq data generated by K.J. Travaglini et
   al.^[196]27 were analyzed and visualized using the web-based cellxgene
   analyzer. The URL provided allows access to this resource:
   [197]https://cellxgene.cziscience.com/d/krasnow_lab_human_l
   ung_cell_atlas_ 10x-1.cxg/. The scRNA-seq procedure took place in the
   same room as we reported in a previous publication^[198]26. Briefly,
   the organoids were disaggregated into individual cells using the Neural
   Tissue Dissociation Kit (Miltenyi Biotec, NRW, Germany). The cells were
   washed with a solution containing 0.04% BSA three times. The viability
   of the cells was assessed using the Countess™ II system (Thermo Fisher
   Scientific). For library construction, the 10X Chromium Single Cell 3’
   Reagent Kit v3.1 (10X Genomics, Pleasanton, CA) was employed. The
   resulting libraries were sequenced on the Illumina NovaSeq 6000
   platform (Illumina, San Diego, CA), and the initial sequencing data
   were converted into FASTQ files using 10x Cell Ranger software (10X
   Genomics, CA). To process the data, we followed the standard sequencing
   protocol provided by 10x Genomics, which involved trimming the barcode
   and unique molecular identifier (UMI) sequence to 26 bp and the mRNA
   sequence to 98 bp. Subsequently, the FASTQ files were aligned to the
   human reference genome (GRCh38). We utilized 10x Cell Ranger software
   for preliminary data analysis, which generated a file containing a
   barcode table, a gene table, and a gene expression matrix. Seurat
   objects were created by using Seurat v5 and were normalized by using
   normalizing SCTransform, which followed by integration using
   Harmony{Korsunsky, 2019 #33}. Integrated object was converted into
   cloupe file by using loupeR package
   ([199]https://github.com/10XGenomics/loupeR). We utilized Loupe Cell
   Browser 8.0 (10x Genomics) to perform quality control, analysis, and
   exploration of the single-cell RNA-seq data.

Bulk RNA sequencing

   To perform a comprehensive gene expression analysis of Ctrl-iMφ and
   Alv-iMφ, Bulk RNA sequencing was carried out by Macrogen (Seoul,
   Korea). Four groups of samples were prepared, including Ctrl-iMφ,
   Alv-iMφ isolated from iAlvAssemb (7 days), PBMC-derived monocytes (day
   0), and monocyte-derived macrophages (day 7). Libraries were prepared
   using the TruSeq Stranded mRNA LT Sample Prep Kit (Illumina, San Diego,
   CA) and sequenced on an Illumina platform.The sequencing data underwent
   trimming and mapping to the human reference genome (GRCh38) using
   HISAT2. Transcript assembly was performed using StringTie. The
   expression levels of each transcript were normalized using fragments
   per kilobase of transcript per million reads mapped (FPKM). The entire
   dataset was log2-transformed (FPKM + 1) for further normalization.

   For GSEA analysis, Ctrl-iMφ (60-day) and Alv-iMφ from iAlvAssemb
   (7-day) samples were used. GSEA Desktop v4.1.0 with C8: cell type
   signature gene sets from Human MSigDB Collections was employed. To
   compare with AM in human tissue, only peripheral blood-derived monocyte
   (HPCM) and Human BAL-derived AM (HBAM) data were extracted from a
   published dataset ([200]GSE174659) by Czarnewski P et al. ^[201]28. A
   total of 7080 genes with a p-value < 0.05 in HPCM and HBAM were
   selected, and the fold-change (FC) value of expression in monocytes
   versus macrophages was calculated. The correlation of iMφ, Alv-iMφ, and
   monocyte-derived Mφ with tissue-resident AM was independently
   determined by calculating the Pearson correlation coefficient (r) using
   Prism software (GraphPad Software, San Diego, CA). Sample distance
   analysis was performed using the Multi Experiment Viewer software (MeV
   version 4.9.0; [202]http://mev.tm4.org/). Differentially expressed
   genes were sorted based on a fold-change of gene expression (>1.5), and
   a heatmap was generated using MeV to visualize genes that were
   upregulated in both Alv-iMφ/mono and HBAM/HPCM. Pathway enrichment
   analysis using the Reactome Database ([203]https://reactome.org/) was
   conducted in Cytoscape software ([204]www.cytoscape.org).
   Representative enriched pathway were listed in supplementay
   Fig. [205]2B.

   To examine the expression changes of AM-enriched genes in Alv-iMφ, we
   utilized bulk RNA-seq data from a study by Y. Labin et al. ^[206]19.
   This dataset provided a comparison of gene expression across various
   tissue macrophages in mice. Using relative expression analysis, we
   identified 480 genes that were enriched in mouse AMs. Subsequently, we
   investigated the expression changes of the human orthologues of these
   genes in Alv-iMφ compare to ctrl-iMφ.

Immuno-metabolism analysis

   To perform immuno-metabolism analysis using RNA sequencing data, we
   cultured iMφ with and without GM-CSF (10 ng/ml) for 7 days. The cells
   were then infected with M.tb (H37Rv) at MOIs of 5 and 10 for 48 h.
   Following infection, RNA sequencing libraries were prepared, and raw
   data were processed using the STAR 2.7.11b aligner^[207]29, followed by
   gene expression quantification with RSEM^3. We employed the METAFlux
   pipeline for metabolic flux analysis. Briefly, the metabolic reaction
   activity score (MRAS) for input data was calculated and metabolic
   models were constructed. Then flux balance analysis (FBA) was applied
   to optimize metabolic fluxes. Finally, the results were visualized
   using integrated plotting functions. For detailed steps, we referred to
   the METAFlux tutorial^[208]30.

Statistics & reproducibility

   Replication details are provided in the figure legends, with each
   experiment independently repeated at least twice. In experiments on iMφ
   tissue adaptation, we noted that when iMφ maturity was high before
   co-culture, the increase in tissue adaptation-related gene expression
   was either minimal or not observable. consequently, this data was
   excluded, and only iMφ cells within two month post-differentiation were
   used in our study

   Sample size were indicated in Figure legends and the results are
   presented as the mean ± standard deviation (SD). The statistical
   significance of the data was determined using a two-tailed Student’s
   t-test or one-way ANOVA with Dunnett’s multiple comparisons test,
   performed by GraphPad Prism 9. P-values lower than 0.05 are plotted in
   the graph.

Graphical Schematics

   The graphical schematics illustrating the research methods were created
   by the author using PowerPoint software. All items were designed as
   vector-based graphics and were drawn manually without using external
   sources.

Reporting summary

   Further information on research design is available in the [209]Nature
   Portfolio Reporting Summary linked to this article.

Supplementary information

   [210]Supplementary Infomation^ (2.2MB, pdf)
   [211]41467_2025_58450_MOESM2_ESM.pdf^ (42.1KB, pdf)

   Description of Additional Supplementary Files
   [212]Supplementary Movie 1^ (13.5MB, mp4)
   [213]Supplementary Data File 1^ (21.2KB, xlsx)
   [214]Reporting Summary^ (136.1KB, pdf)
   [215]Transparent Peer Review file^ (2.9MB, pdf)

Source data

   [216]Source data^ (87.6KB, xlsx)

Acknowledgements