iPSC-derived mesenchymal cells that support alveolar organoid development

Summary Mesenchymal cells are necessary for organ development. In the lung, distal tip fibroblasts contribute to alveolar and airway epithelial cell differentiation and homeostasis. Here, we report a method for generating human induced pluripotent stem cell (iPSC)-derived mesenchymal cells (iMESs) that can induce human iPSC-derived alveolar and airway epithelial lineages in organoids via epithelial-mesenchymal interaction, without the use of allogenic fetal lung fibroblasts. Through a transcriptome comparison of dermal and lung fibroblasts with their corresponding reprogrammed iPSC-derived iMESs, we found that iMESs had features of lung mesenchyme with the potential to induce alveolar type 2 (AT2) cells. Particularly, RSPO2 and RSPO3 expressed in iMESs directly contributed to AT2 cell induction during organoid formation. We demonstrated that the total iPSC-derived alveolar organoids were useful for characterizing responses to the influenza A (H1N1) virus and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, demonstrating their utility for disease modeling.

Correspondence gotoh.shimpei. 5m@kyoto-u.ac.jp In brief Tamai et al. develop a method for generating iPSC-derived mesenchymal cells (iMESs) capable of developing alveolar organoids (iMES-AOs). iMESs can induce not only alveolar epithelial type II but also type I cells. iMES-AOs provide a platform for modeling lung development and diseases including respiratory viral infection.

INTRODUCTION
Mesenchymal cells provide extracellular matrix proteins and various secreted proteins suited to cell-type-specific tissue microenvironments, and their interaction with epithelial cells is essential for normal organ development, homeostasis, and regeneration. Previous studies have described human pluripotent stem cell (PSC)-derived alveolar epithelial cells in both a fibroblast-dependent Yamamoto et al., 2017) and a fibroblast-free procedure (Jacob et al., 2017;Yamamoto et al., 2017). Challenges to simultaneous differentiation of lung epithelial and mesenchymal cells have been reported (Chen et al., 2017;Dye et al., 2015;Miller et al., 2019), but the process of deriving mesenchymal cells from PSCs remains unknown. Given the potential application of the PSC-derived alveolar organoid (AO) for research in human developmental MOTIVATION Organoid technology is a powerful tool for bioscience research. Previously, human fetal lung fibroblasts (HFLFs) have been used to promote the development of alveolar organoids, but HFLFs have the limitation of being allogenic in organoid systems, and therefore they cannot reproduce an individual's biological environment. To address this limitation, we generated induced pluripotent stem cell-derived mesenchymal cells (iMESs) that have the ability to develop alveolar organoids with the aim of modeling niche environments and diseases.
processes and disease modeling, AOs should include mesenchymal cells. Primary human fetal lung fibroblasts (HFLFs) have been used, but it is often difficult to recapitulate the exact biological environment found in the lung. Although others have reported human PSC-derived lung mesenchymal cells that recapitulated the developmental course of mouse early fetal foregut organogenesis (Han et al., 2020;Kishimoto et al., 2020), AO formation using PSC-derived lung mesenchymal cells has not been accomplished. Hence, it is desirable to generate mesenchymal cells that can support organ development and facilitate further research in embryogenic organogenesis.
In this study, we report a method for generating human induced PSC (iPSC)-derived mesenchymal cells (iMESs) that are able to form AOs (iMES-AOs). We also explored niche factors to induce iPSC-derived alveolar type 2 (iAT2) cells from progenitor cells using transcriptomic analysis of paired isogenic iMESs and the mesenchymal cells of primary fibroblasts. Moreover, we used iMES-AOs in two pandemic respiratory infection models, the influenza A (H1N1) virus and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2).

Generation of mesenchymal cells that induce iAT2 cells
We optimized a method to differentiate mesenchymal cells that were able to induce SFTPC + AT2 lineage cells from their progenitor cells (Figures 1A and 1B). AOs were generated in a threedimensional (3D) co-culture of SFTPC-GFP reporter iPSC (B2-3)-derived NKX2-1 + lung progenitors sorted by carboxypeptidase M (CPM) and iMESs derived from their parental 201B7 iPSC line (201B7-iMESs). First, we focused on mesoderm induction followed by differentiation of mesenchymal cells. On day 1, cell clusters started to lose their border sharpness in a medium containing activin A, BMP4, and CHIR99021, and they appeared as primitive streak-like cells expressing T-box T in the EPCAM + cell population ( Figures 1C and 1D). Cells became oblong in shape on day 3, and the EPCAM À cell population became positive for the mesodermal markers NCAM, PDGFRa, and KDR (Evseenko et al., 2010;Sakurai et al., 2006). A new medium containing activin A, KGF, BMP4, FGF2, and FGF10 induced the expression of VIM, THY1, PDGFRa, and KDR on day 7. Because a minor EPCAM + cell population showed insufficient mesenchymal marker expression ( Figure 1D), we purified EPCAM À cells and named them ''iMESs.'' Time course changes of each marker were validated using qRT-PCR, and gene expression levels were compared among iMESs, HFLFs, and human dermal fibroblasts (HDFs) ( Figure S1A). iMESs also expressed FOXF1 and TBX4 as lung mesenchymal markers (Han et al., 2020;Horie et al., 2018) (Figures 1E and S1A).

Generation of iMESs from HFLF-and HDF-derived iPSCs to form AOs
To elucidate the role of iMESs in inducing iAT2 cells, we compared primary fibroblasts with iMESs. Because HDFs cannot generate AOs, we can eliminate unnecessary factors for AO development. We generated iPSCs from HFLFs and HDFs (HFLF-iPSCs and HDF-iPSCs, respectively) that presented expression of undifferentiated markers, normal karyotypes, and trilineage-differentiation potentials (Figures S2A and S2B). Both pre-and post-3D culture samples of HFLFs, HDFs, and their iMESs were recovered ( Figure 2A). FOXF1 was highly expressed in HFLFs and each iMES but was weak in HDFs (Figure 2B). The ability to induce SFTPC-GFP + /EPCAM + iAT2 cells was verified in both HFLF-and HDF-iMESs ( Figures 2C and 2D).
iMESs and HFLFs, but not HDFs, express genes associated with lung development Principal-component analysis of RNA sequencing (RNA-seq) transcriptomes revealed well-separated clusters of each condition. Post-3D culture iMES and HFLF transcriptomes were both (E) Immunofluorescence staining on day 7. Scale bars, 100 mm. Arrowheads: E-cadherin + /FOXF1 À cells. (F and G) Flow cytometry data of the induction efficiency of SFTPC-GFP + /EPCAM + cells and its quantification. iMESs were differentiated from 201B7 and 604A1 iPSCs. HFLFs and HDFs were used as positive and negative controls, respectively. The lung progenitors were differentiated from an SFTPC-GFP reporter iPSC line (B2-3). Negative control data were recorded using AOs consisting of lung progenitors and iMESs, both derived from the non-reporter 201B7. Data are presented as mean ± SEM. **p < 0.01 (Dunn's post hoc test). similar to one another and dissimilar to post-3D culture HDFs, inferred from the relative plot distance between populations (Figure 2E). Selected lineage gene markers for fibroblast, muscle, adipocyte, endothelial, and immune cells were depicted on a heatmap in columns scaled to transcripts per million (TPM). The expression levels of fibroblast markers were considerably high, indicating that iMESs share transcriptomic programs similar to fibroblasts ( Figure 2F). Furthermore, we evaluated respiratory mesenchymal markers, including WNT2, TCF21, FOXF1, NKX6-1, and PRRX1 (Goss et al., 2009;Han et al., 2020;Kishimoto et al., 2020;Park et al., 2019;Yeo et al., 2018) ( Figure 2G). Post-3D culture HFLFs showed the highest expression of WNT2, followed by pre-3D culture HFLFs and HDFs, but HFLF-iMESs, HDF-iMESs, and post-3D HDFs expressed WNT2 at extremely low levels. Although expression levels of TCF21 and FOXF1 were also highest in post-3D culture HFLFs, they were also prominent in post-3D culture HFLF-and HDF-iMESs. NKX6-1 was barely detected in any of the samples, and PRRX1 was highest in HDFs, followed by HFLFs. Gene Ontology (GO) enrichment analysis of biological processes indicated that ''embryonic organ development'' and ''lung development'' were significantly enriched in both up-and down-regulated differentially expressed genes (DEGs) between HDF-iMESs and HDFs ( Figure 2H). DEGs annotated to ''lung development'' were illustrated in a heatmap using four groups of mesenchymal transcriptomes ( Figure 2I). Although WNT5A, FGF7, and PDGFRA are important factors in AT2 cells (Barkauskas et al., 2013;Nabhan et al., 2018;Zepp et al., 2017), they were up-regulated in HDFs. Secreted factors, including RSPO2, WNT11, CCN2, SPARC, BMP4, HHIP, LAMA5, and LOX were up-regulated in iMESs. Expression levels of transcription factors, including FOXF1 and TCF21, were higher in HFLF-and HDF-iMESs and HFLFs than in HDFs, suggesting that iMESs share features of fetal lung fibroblasts. HFLF-iMESs and HFLFs shared expression of EPAS1, yet HDFs expressed EPAS1 at higher levels than HFLFs, which indicates that EPAS1 is not specific to the lung mesenchyme ( Figure 2I). Next, we compared the top 5,000 genes of post-3D culture HFLF-iMESs and HFLFs in a Venn diagram ( Figure 2J). There were 4,220 common genes, of which ''lung development'' was enriched (false discovery rate [FDR] q = 0.001). Genes annotated to ''lung development'' included HHIP, CCN2, SPARC, BMP4, LAMA5, and LOX, sug-gesting that they were important factors for AO generation. Further, the transcription factors FOXF1 and TCF21 were included, and they may be potential markers for lung fibroblasts.

H1N1 and SARS-CoV-2 infection of iMES-AOs induce intrinsic interferon responses
We applied iMES-AOs to disease modeling of acute respiratory viral infections. iMES-AOs (P2) on day 12 were infected with H1N1 or SARS-CoV-2 for 3 days and then collected for analysis ( Figure 4A). The viral titer was higher in H1N1-infected iMES-AOs than in the no-cell control (6.1 ± 0.1 versus 2.6 ± 0.1, log 10 PFU/ mL) ( Figure 4B). The nucleoprotein of H1N1 was detected in EPCAM + epithelial cells, and some infected cells, including SFTPC-GFP + iAT2 cells, fell into the lumen ( Figure 4C). We observed MX1 + cells, showing a type I interferon response induced by H1N1 infection ( Figure 4D). In contrast, the nucleocapsid protein of SARS-CoV-2 was not stained in SARS-CoV-2-infected iMES-AOs. We speculated that inaccessibility to the apical inside of AOs interfered with the efficient infection of SARS-CoV-2, while H1N1 could invade AOs from the basolateral side, where sialic acid is present. Thus, we dissociated whole gels of iMES-AOs and then incubated the gels in a viral solution of SARS-CoV-2 for 2 h, as described previously (Mulay et al., 2021). After washing, they were re-cultured in a 3D culture with Matrigel ( Figure 4E). Three days later, the viral titer increased compared with both the no-cell control and the previous nondissociated samples (3.2 ± 0.2 versus 4.5 ± 0.2 versus 5.7 ± 0.1, log 10 PFU/mL) ( Figure 4F), suggesting that the virus could approach the apical side of organoids where abundant receptors are present. The nucleocapsid protein of SARS-CoV-2 was also  Figure 4G). NaPi2b and the nucleocapsid protein were co-stained, indicating infected AT2 cells ( Figure 4H), and MX1 + cells indicated that the type I interferon response was induced by SARS-CoV-2 infection ( Figure 4I).

DISCUSSION
In this study, we featured iMESs that efficiently induced SFTPC-GFP + /EPCAM + iAT2 cells via epithelial-mesenchymal interactions. Although the iMES transcriptome did not completely match that of HFLFs, our approach to generate human PSCderived mesenchymal cells that are suited for inducing tissue epithelial cells, such as iAT1 and iAT2 cells, might inform us of the most crucial factors in organogenesis.
Previous studies have reported a method of induction of respiratory mesenchymal cells (Han et al., 2020;Kishimoto et al., 2020). In these studies, NKX6-1 was induced as a respiratory mesenchymal marker with retinoic acid, BMP4, and Hedgehog agonists, followed by a low-dose WNT agonist. Intriguingly, although iMESs do not express NKX6-1, we succeeded in expressing SFTPC-GFP in alveolar epithelial cells in iMES-AOs. Indeed, we observed a low NKX6-1 expression level, even in HFLFs; thus, we speculated that NKX6-1 might not be requisite for inducing AOs. Moreover, a recent study reported the differentiation of mouse PSC-derived lung-specific mesenchymal cells via TBX4 + state (Alber et al., 2022). Although AT1/AT2 markers, including Sftpc, Ager, and Hopx, were not robustly induced in the study, it was demonstrated that lung progenitor cells expressed early distal lung epithelial markers, such as Sox9 and Etv5, indicating insufficient inducing factors of AT1/AT2 cells. iMES transcriptome analysis and subsequent validation revealed that RSPO2 and RSPO3 expressed in iMESs could promote iAT2 cell induction. This is consistent with a recent report that revealed that RSPO2 + mesenchymal cells were adjacent to human fetal lung bud tip progenitors and that RSPO2 might potentially possess a pivotal role in proximal-distal patterning (Hein et al., 2022). The low expression level of WNT2 was unexpected because Wnt2/2b signaling has been reported to be essential for lung endoderm specification (Goss et al., 2009). However, RSPO2/RSPO3 expressed in iMESs could promote iAT2 cell induction, substituting for the WNT2 expressed in HFLFs for activation of canonical WNT signaling. The lack of canonical Wnt ligand gene expression in HDFs might be one reason for their inability to induce iAT2 cells. On the other hand, it was an unexpected result that WNT5A, FGF7, and PDGFRA, known to be important factors in AT2 cells (Barkauskas et al., 2013;Nabhan et al., 2018;Zepp et al., 2017), appeared to be up-regulated in HDFs, compared with iMESs and HFLFs, although FGF7 was supplemented in the alveolarization medium. In a previous study, we reported that AOs could not develop without mesenchymal cells if CHIR99021, SB431542, and Y27632 were not added . Therefore, there should be cell-cell interactions between iMESs and alveolar epithelial cells that at least complement the role of SB431542. All in all, further studies are needed to clarify AT2 cell induction down-stream pathways. We also elucidated that iMES-AOs and HFLF-AOs included multiple cell types using scRNA-seq transcriptomics. Expression levels of each lineage marker tended to be higher in iMES-AOs, suggesting that iMESs could induce more mature respiratory epithelial cells compared with HFLFs. FOXJ1 + /RSPH1 + iCilia observed in iMES-AOs at P2 co-expressed SFTPC; hence, they may be cells that would differentiate to mature multiciliated cells from SFTPC + distal tip cells. TM4SF1 + cells were numerous in iMES-AOs at P2. TM4SF1 has been noted as a marker of Wntresponsive alveolar epithelial progenitor lineage (Zacharias et al., 2018), and further verification is needed on whether TM4SF1 + cells in iMES-AOs could serve as progenitors. In the analysis of mesenchymal cells, the IRX3 + FOXO1 + TCF21 + FOXF1 À cluster and the CXCL12 + FOXF1 + TCF21 + cluster were distinctive in iMESs. Further studies are needed to determine if these clusters have the power to induce respiratory epithelial cells, particularly iCilia that could be induced by iMESs but not by HFLFs.
In conclusion, iMES-AOs may be used to investigate the central mechanism of alveolar differentiation through epithelialmesenchymal interactions, which would be helpful for disease models, drug screening, and niche reconstruction for in vivo lung regeneration in the future.

Limitations of the study
It is advantageous that the ratio of SFTPC-GFP + /EPCAM + cells increased until P3 in iMES-AOs, but in this study, we did not validate the phenotypes of iMES-AOs in long-term passages. We applied the iMES-AO platform to the H1N1 and SARS-CoV-2 infection models, and iMES-AOs were analyzed at 72 h postinfection because the highest viral titers were observed in the

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Shimpei .
Materials availability 201B7 and 604A1 were obtained from CiRA at Kyoto University. TIG120 was obtained from JCRB (JCRB0542). B2-3, HFA and GC23 are available from the lead contact upon request. HFLF has been discontinued (DV Biologics) but is available from the lead contact upon request for use in academia.
Data and code availability d Bulk RNA (GSE188822) and single-cell RNA-seq data (GSE188823) have been deposited in the NCBI Gene Expression Omnibus. d This paper does not report original code. d Additional Supplemental Items are available from Mendeley Data: http://dx.doi.org/10.17632/6pgb6rx8t8.1. d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.