Differentiation of Human Induced Pluripotent Stem Cells from Patients with Severe COPD into Functional Airway Epithelium

Background: Chronic Obstructive Pulmonary Disease (COPD), a major cause of mortality and disability, is a complex disease with heterogeneous and ill-understood biological mechanisms. Human induced pluripotent stem cells (hiPSCs) are a promising tool to model human disease, including the impact of genetic susceptibility. Methods: We developed a simple and reliable method for reprogramming peripheral blood mononuclear cells into hiPSCs and to differentiate them into air–liquid interface bronchial epithelium within 45 days. Importantly, this method does not involve any cell sorting step. We reprogrammed blood cells from one healthy control and three patients with very severe COPD. Results: The mean cell purity at the definitive endoderm and ventral anterior foregut endoderm (vAFE) stages was >80%, assessed by quantifying C-X-C Motif Chemokine Receptor 4/SRY-Box Transcription Factor 17 (CXCR4/SOX17) and NK2 Homeobox 1 (NKX2.1) expression, respectively. vAFE cells from all four hiPSC lines differentiated into bronchial epithelium in air–liquid interface conditions, with large zones covered by beating ciliated, basal, goblets, club cells and neuroendocrine cells, as found in vivo. The hiPSC-derived airway epithelium (iALI) from patients with very severe COPD and from the healthy control were undistinguishable. Conclusions: iALI bronchial epithelium is ready for better understanding lung disease pathogenesis and accelerating drug discovery.


Introduction
Chronic obstructive pulmonary disease (COPD) is a chronic lung disease characterized by respiratory symptoms associated with chronic airflow limitation. COPD is the third leading cause of death worldwide, and affects approximately 300 million people in the world [1]. Although cigarette smoking has been considered the most frequent cause of COPD, about half of cases are linked to non-tobacco-related risk factors, such as outdoor air pollution, biomass smoke, and occupational exposure to fumes and dust [2]. In COPD, the small conducting airways (<2 mm in diameter) are the major site of airflow obstruction, inflammation, and destruction [3][4][5]. Therefore, in vitro bronchial epithelium models are crucial to better understand and treat COPD.
Induced pluripotent stem cells (iPSCs) represent an attractive solution to model chronic airway diseases because they can yield a virtually unlimited amount of any differentiated cell type [6]. Recently described protocols to differentiate human pluripotent stem cells (PSCs) into bronchial epithelium [7][8][9][10][11][12][13][14][15] rely on the knowledge gathered on normal lung development in mammals [16]. Briefly, lung embryogenesis starts with the definitive endoderm (DE) formation. During the 4th week of human embryonic development, the primitive gut appears and can be divided into foregut, midgut, and hindgut. Early pulmonary development starts from the ventral area of the anterior foregut endoderm (vAFE). From this zone, which is characterized by the expression of the transcription factor NK2 homeobox 1 (NKX2.1), the respiratory diverticulum emerges and forms the trachea, and then bronchi, bronchioles, and alveoli. These steps can be recapitulated in vitro by differentiating PSCs first into DE and then into vAFE [17]. Finally, vAFE cells are differentiated into lung progenitors and bronchial cells. However, the protocols for PSC differentiation into bronchial epithelium present several drawbacks, and many of them have been rarely described in detail. In addition, many of these protocols work only with some pluripotent stem cell lines, often cell lines derived from healthy controls, and require an enrichment step based on the specific selection of NKX2.1+ cells at the vAFE stage using flow cytometry and cell surface markers (e.g., carboxypeptidase M+ cells [13] or CD47 high CD26 low cells [18]), or a final differentiation step in 3D culture conditions. Others require important technical skills and are difficult to replicate [19].
Here, we developed an easy approach to differentiate human iPSCs (hiPSCs) into proximal airway epithelium, without any cell purification steps. Careful in-home reprogramming and then culture adaptation to single-cell passaging, together with precise timing and reagent benchmarking for each differentiation step, led to the successful generation of fully differentiated and functional bronchial epithelium in air-liquid interface (ALI) culture conditions from four hiPSC lines (iALI bronchial epithelium), among which, three were derived from patients with severe COPD. This study highlights the crucial importance of evaluating the cell expansion and differentiation conditions for achieving optimal phenotypic and functional endpoints, such as ciliary beat frequency (CBF), mucus flow velocity, differentiated cells, and transepithelial electrical resistance (TEER). This simple protocol to produce hiPSC-derived iALI bronchial epithelium will facilitate airway disease modeling for developing novel gene/cell therapies, and for drug discovery.

Patients' Clinical Characteristics
Patients were younger than 55 years and had severe, early onset COPD (i.e., ratio of forced expiratory volume in one second (FEV1) to forced vital capacity (FVC) <0.70 and FEV1% predicted <50% on postbronchodilator spirometry). More clinical data are available in Supplementary Figure S1 and Appendix A.

HIPSC Differentiation
Differentiation was carried out as described in Figures 1 and 2A, using reagents at the concentrations listed in Supplementary Tables S1 and S2. Cells were plated at high-density (one 35 mm dish for two Transwell inserts) on Geltrex-coated Transwell inserts. During differentiation in hypoxic conditions (5% O 2 , 37 • C), medium was changed every day.

Cell Density and Induction Timing Are Critical for Successful Differentiation and Allows High Rate of Definitive Endoderm Induction
To develop a robust differentiation protocol (Figure 2A), we benchmarked the timing, cell density, and passaging method-three crucial steps for achieving reliable rates of DE purity and quality. We passaged hiPSC lines as single cells because hiPSC clumps were partly resistant to DE induction, as evidenced by OCT4 expression persistence. We obtained optimal cell adaptation by gentle colony dissociation into small clumps for five passages, and then into single cells for at least 5-10 passages, using Versene (EDTA) and

Statistical Analysis
Data are presented as mean and standard deviation (SD) or standard error of the mean (SEM), and graphs were generated with GraphPad (GraphPad Software Prism, v 6.01, San Diego, CA, USA). All shown data are from experiments repeated at least three time.

Cell Density and Induction Timing Are Critical for Successful Differentiation and Allows High Rate of Definitive Endoderm Induction
To develop a robust differentiation protocol (Figure 2A), we benchmarked the timing, cell density, and passaging method-three crucial steps for achieving reliable rates of DE Cells 2022, 11, 2422 6 of 17 purity and quality. We passaged hiPSC lines as single cells because hiPSC clumps were partly resistant to DE induction, as evidenced by OCT4 expression persistence. We obtained optimal cell adaptation by gentle colony dissociation into small clumps for five passages, and then into single cells for at least 5-10 passages, using Versene (EDTA) and Y-27632. Then, we started differentiation by adding activin A and CHIR99021 (a GSK3 inhibitor that acts as a WNT pathway agonist) in the presence of Y-27632 for 1 day (day 1; anterior primitive streak, (APS); Figure 2A,B and Supplementary Table S1), followed by activin A, LDN-193189 (a selective bone morphogenetic protein (BMP) signaling inhibitor that blocks the transcriptional activity of the type I BMP receptors activin receptor-like kinase 1, 2, 3, and 6), and Y-27632 for 1-2 days, leading to DE induction (day 2-3, Figure 2A,B). To optimize the protocol, we tested various intervals between hiPSC plating and APS induction, and different cell densities (from 70 to 130 K cells/cm 2 ) ( Figure 2C,D). Plating cells at too low and too high density led to important cell loss and to persistent OCT4 expression ( Figure 2G). This optimized protocol robustly yielded >80% of C-X-C Motif Chemokine

Efficient Induction of High Purity NKX2.1+ Lung Progenitors without Cell Sorting
Comparison of various growth factor combinations for vAFE induction showed that DE cells needed minimal cell signaling, and therefore, were grown in RPMI1640 basal medium with B27 minus vitamin A (Figure 2A and Supplementary Tables S1-S3). For efficient vAFE induction, a DE cell population with at least 80% of CXCR4+ cells was required.
Time course experiments showed that at 24-36 h after LDN-193189 addition, there was a narrow window when cells exhibited optimal conditions (i.e., high CXCR4 expression and high viability) for vAFE induction. The 3D bud-like structures emerging between days 4-8 appeared to be a good morphological indicator of vAFE differentiation visible under an optical microscope ( Figure 2B, red arrows). In these conditions, >80% of cells consistently expressed NKX2.1 (assessed by flow cytometry and immunolabelling in five different PSC lines; n = 46 independent experiments) ( Figures 3A,B and S3). The optimum percentage of NKX2.1+ cells (>80%) was reached at~day 3 after vAFE induction ( Figure 3C), as confirmed by immunostaining for NKX2.1 from day 1 to day 4 (Supplementary Figure S4A). This NKX2.1 expression level was required to induce differentiation towards iALI. Pluripotency markers (e.g., OCT4 and NANOG) were strongly downregulated at the vAFE stage, compared with the DE stage ( Figure 3E-G). Positive Controls: Brain mRNA, Thyroid mRNA, HepG2 (Human Liver Cancer Cell Line) mRNA.
NKX2.1 bronchial progenitor cells exhibited a high proliferation rate, assessed by quantifying the expression of the proliferation marker protein Ki-67 (Supplementary Figure S4C). We also detected SOX2, SOX9 expression by immunostaining, as previously reported in vivo during human lung development ( Figure 3D and S4B), but not terminal airway epithelial markers. This confirmed the immature feature of these progenitor cells, and was in agreement with another hiPSC differentiation protocol [18] and human lung development [25]. As NKX2.1 is also expressed in other developing tissues ( Figure 3F), we assessed by RT-qPCR, the purity of NKX2.1+ lung progenitor cells by confirming the absence of thyroid gland-(thyroglobulin, paired box 8), brain-(paired box 6), and liver-specific (alphafetoprotein, confirmed also by immunostaining in Supplementary Figure S4D) cell markers. We did not observe any difference between healthy and COPD cell lines at the bronchial progenitor stage.   We obtained iALI bronchial epithelium from four different hiPSC lines (n > 3 independent experiments per cell line). After mechanical dissociation into small clumps, we plated vAFE cells at high density on Transwell inserts in PneumaCult-Ex Plus medium (day 9, Figure 2A). After 2 days in PneumaCult-Ex Plus medium, we progressively switched to PneumaCult-ALI maintenance medium. Four days after seeding on Transwell inserts, we removed the medium from the apical side to switch to ALI culture ("polarization"). We added DAPT, a γ-secretase inhibitor that blocks NOTCH signal transduction, to the culture medium in the basolateral part of the Transwell from day 14 to day 28, post-plating on Transwell inserts (Figure 2A and Supplementary Table S1).

Epithelium with Barrier Function
HiPSC-derived epithelial cells reached confluence after 4 days of submerged growth conditions (Figure 2A). We observed morphological features consistent with epithelium at late iALI stage (>day 42), zonula occludens 1 expression, and the presence of adherent junctions (junctional complexes) by transmission electron microscopy at day 34 ( Figure 4A). We assessed the barrier integrity during ALI 2D culture by TEER measurement. TEER increased significantly during the differentiation process ( Figure 4H). At day 7 of air liquid interface polarization, it reached~300 Ω·cm 2 and could be maintained for >200 days of culture. TEER values were not significantly different between control and COPD hiPSCderived epithelia at all time points.   The concentration of secreted CCSP ranged from 103.9 ng/mL to 110.9 ng/mL, depending on the experiment ( Figure 4L), and was comparable among cell lines at day 41 of differentiation. We could not detect CCSP in the iCOPD2 cell line.

IALI Bronchial Epithelium Includes
We also detected neuroendocrine cells (CHGA mRNA and protein expression) that could organize into clusters, resembling airway neuroepithelial bodies ( Figure 4F-M). We did not detect SFTPC expression at mRNA level at iALI stage (Supplementary Figure S4E).

Functional Multi-Ciliated Cell Airway Epithelium
We also observed cilia beating by optical microscopy and by TUBIV immunofluorescent labeling ( Figure 5A). We identified abundant multi-ciliated cells in all four iPSC lines after 30 days of differentiation. We observed dynein axonemal heavy chain 5 staining along the ciliary axoneme ( Figure 5B). Morphological analysis of multi-ciliated cells by optical and transmission electron microscopy ( Figure 5C,D) indicated that the cilium structure was characterized by nine peripheral doublet pairs and a central pair of singlet microtubules ( Figure 5D), typical of motile cilia [26].
We then measured cilium length in iALI cultures, in fresh bronchial epithelial cells obtained by endoscopic brushing, and in classical ALI-cultured airway epithelium by optical and scanning electron microscopy. The mean cilium length was similar in ALI and iALI cultures ( Figure 5E), without any obvious difference among the different samples. We could observe cilia beating using a high-speed camera after isolation of iALI epithelium patches (Supplementary Movie S1), on Transwell membranes (Supplementary Movie S2), and after live immunostaining using SiR-conjugated fluorogenic probes for tubulin (Supplementary Movie S3). cent labeling ( Figure 5A). We identified abundant multi-ciliated cells in all four iPSC lines after 30 days of differentiation. We observed dynein axonemal heavy chain 5 staining along the ciliary axoneme ( Figure 5B). Morphological analysis of multi-ciliated cells by optical and transmission electron microscopy ( Figure 5C, D) indicated that the cilium structure was characterized by nine peripheral doublet pairs and a central pair of singlet microtubules ( Figure 5D), typical of motile cilia [26]. To assess the muco-ciliary clearance capacity of 2D cultures, we recorded CBF and muco-ciliary flow. The CBF of iALI cultures (14.3 ± 1.8 Hz) was similar to that of primary airway epithelium in ALI culture ( Figure 5F) [27]. Cultures presented structures with high density of ciliated cells that were actively beating, giving rise occasionally to localized vortexes ( Figure 5F, right bottom panel, Supplementary Movie S2). The estimated flow velocity of the vortex was 5.6 ± 6.5 µm/s. We could observe beating cilia in iALI bronchial epithelia for~400 days without cell passaging and without aneuploidy appearance (Supplementary Figure S2B). Moreover, we could passage cultures at least three times after iALI generation.

Discussion
Here, we described the generation of iALI bronchial epithelium that represents an attractive alternative to animal models and ex vivo cultures of differentiated bronchial epithelium from endobronchial biopsies. Our differentiation protocol offers a virtually unlimited source of homogeneous reliable human bronchial epithelium. Importantly, this protocol was carried out successfully by ten different members of our research group, and at least three times for each cell lines.
In vitro models of human epithelia in ALI culture represent useful platforms to promote the differentiation and maturation of epithelial cells and allow the modeling of infections and environmental exposures. The generation of mature bronchial epithelium from hiPSCs is a powerful way to explore and recapitulate in vitro human airway development through a series of steps that mimic the normal in vivo embryonic development. Furthermore, iALI is an unlimited source of airway epithelium. HiPSC differentiation provides also a mean to characterize the different signaling pathways involved in airway lineage specification and differentiation [28]. Besides being an excellent tool for modeling human airway development, iALI represents an optimal platform for therapeutic innovation, extensive drug screening, and for cell-based therapy.
The limitations of our iALI system are mainly linked to the potential lack of purity of iPSC-derived airway progenitors and the difficulties to achieve fully matured airway epithelium. However, recent single cell transcriptomic analyses indicated that human airway primary cells from bronchial biopsies and adult human alveolar epithelium share a common signature with iPSC-derived lung epithelium [29,30]. Although iALI bronchial epithelium generation is slower than that of ALI epithelium obtained from airway tissue samples, it provides a potentially unlimited quantity of epithelium from a given donor, thus, avoiding batch heterogeneity due to multiple donors.
We identified several critical factors that ensure the efficiency and reproducibility of airway epithelium differentiation from human PSCs. First, PSCs must be adapted to single-cell culture for homogenous cell seeding. When we tried to plate non-adapted cells (i.e., large clumps or high cell density), cell loss was reduced, but differentiation was hampered ( Figure 2D,G). This could be explained by the sustained expression of pluripotency transcription factors within clumps and/or by altered YAP/TAZ signaling activity. Second, the homogeneity of DE and vAFE cell populations (CXCR4 and NKX2.1 expression in ≥80% of cells at the relevant step) was a good predictor of the final success. Based on the work by Matsuno et al. [17], we found that APS induction by activation of the activin A/nodal and WNT pathways for 24 h, followed by two additional days of activin A activity and TGFβ pathway inhibition for DE induction, without addition of other cytokines or small molecules during the vAFE stage, was an effective strategy. Both SOX2 and SOX9 were expressed at the vAFE stage with many double positive cells, in accordance with previous studies reporting the presence of these bipotent cells specifically in human PSC, further strengthening the iALI model [25]. To efficiently isolate NKX2.1+ bronchial progenitors during hiPSC differentiation, several cell surface molecules specifically expressed in these cells have been tested. Carboxypeptidase M (CPM), a specific marker of NKX2.1+ airway progenitors that generate type II alveolar epithelial progenitor cells, was proposed as a cell surface marker for sorting NKX2-1+ cells derived from human iPSCs [13,31]. However, CPM is strongly expressed in hepatoblasts and fetal liver progenitor cells and is present during the hepatic specification of iPSC-derived endoderm cells. This may limit its use for sorting lung progenitor during endoderm cell differentiation [18,32]. Hawkins et al. reported that sorting CD47 high CD26 low cells allowed enriching the NKX2-1+ lung progenitor population from 62% to 70%. Therefore, improving NKX2-1+ lung progenitor sorting based on cell surface markers could help to refine our differentiation strategy. Nevertheless, we found that this step was not necessary for robust bronchial epithelium induction, thus, overcoming a major bottleneck of directed differentiation protocols.
Another key point was the use of the PneumaCult differentiation medium. This proprietary medium, the composition of which is not disclosed, efficiently promotes the differentiation of primary cells obtained from bronchial biopsies. Although this medium might contain a NOTCH pathway inhibitor, we added DAPT to our differentiation protocol. Indeed, NOTCH signaling inhibition promotes the differentiation into multi-ciliated cell at the expenses of club cells [33]. This protocol generated epithelia containing CCSP+/MUC5AC+, CCSP+/MUC5AC-, and CCSP-/MUC5AC+ cells, although basal and ciliated cells were predominant. Interestingly, we detected also rare cells, such as chromogranin A-expressing neuro-endocrine cells. Altogether, these features suggest that the generated epithelia reproduced many features of a fully differentiated bronchiolar epithelium [34]. The physiological relevance of the model was reinforced by the detection of plugs of mucus (alcian blue and periodic acid-Schiff staining), the formation of vortexes of muco-ciliary clearance, cilium length, and CBF, as observed in vivo.
Besides its reproducibility and simplicity, our protocol provides a 2D bronchial epithelium, unlike other methods that lead to 3D ciliated organoids [12,14,15]. To the best of our knowledge, these three COPD hiPSC lines are the first described in the literature, although difficulties could have been expected given the previously reported relative CD34 deficiency [35]. Moreover, one COPD hiPSC-derived epithelium culture could be kept consistently differentiated for~400 days at the time of writing. As expected for a disease with multifactorial genetic susceptibility to environmental triggers (e.g., cigarette smoke), and considering that cell reprogramming erases most epigenetic marks, the DE cells and iALI bronchial epithelia derived from the COPD hiPSC lines were mostly identical to those derived from the healthy donor. One notable difference was basal MUC5B secretion that was increased in all iALI bronchial epithelia derived from COPD hiPSC lines. It will be interesting to challenge these iALI epithelia with smoke extract or pollution particles and investigate whether mucins are induced, as observed in smokers and patients with COPD [36,37].

Conclusions
In conclusion, we described an easy and reliable method to drive PSC differentiation into 2D multicellular bronchial epithelium. This method is highly reproducible, efficient, does not require cell sorting, and is achievable using blood cells from patients with polygenic lung diseases. Our protocol recapitulates the generation of bronchial airway during lung development, particularly the distal bronchial pattern. The protocol will also allow the studying of chronic airway diseases, especially those that concern mainly the small airways, such as cystic fibrosis, COPD, severe asthma, and idiopathic pulmonary fibrosis [38].

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/cells11152422/s1, Figure S1: Clinical characteristics of patients with COPD; Figure S2: Genetic integrity of the hiPSC lines used for differentiation into iALI bronchial epithelia; Figure S3: NKX2.1 and CXCR4 FACS gating strategy; Figure S4: Characterization of vAFE progenitors: SOX9 expression and NKX2.1 expression changes during v-AFE induction; Table S1: Medium composition by culture period; Table S2: Molecules and used concentration; Table S3: List of reagents and consumables; Table S4: List and sequences of the primers used for RT-qPCR; Movie S1: iALI bronchial epithelium obtained from the iCOP9 hiPSC cell line (X20), Movie S2: iALI bronchial epithelium obtained from the iCOPD8 hiPSC cell line (X40); Movie S3: iALI bronchial epithelium obtained from the iCOPD9 hiPSC cell line after live immunofluorescence for TUBIV [39,40]. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patient(s) to publish this paper. Data Availability Statement: Not applicable.

COPD Patients' Characteristics and History
The healthy control (41 years at inclusion) had normal lung function, without respiratory symptoms from childhood, and had no family history of chronic airway diseases.
Patients with severe COPD were recruited in the framework of the INVECCO project. Patients had normal blood level of α-1 anti-trypsin and none of them carried any TERT mutations, to avoid any monogenic COPD form.
Early disease onset was suggested by the early appearance of symptoms, at the mean age of 35 years.
Disease was revealed by dyspnea, pneumothorax, or acute exacerbation. The mean time between first symptoms and COPD diagnosis was 12 years. Family history of COPD (first-degree relative) was found for two of the three enrolled patients. All of them were at least GOLD stage 3. Environmental exposures were identified for all three patients: early and heavy active tobacco exposure, in utero exposure to tobacco, and to second-hand smoke in early childhood. Their mean cigarette consumption was 50 pack-years (range, . Consumption of cannabis and intravenous heroin was recorded, in agreement with the already reported association between substance consumption and severe emphysema. Dyspnea was severe, requiring long-term home oxygen support for two patients. No cardiovascular comorbidity, diabetes, cancer, or pulmonary hypertension was described, but they had severe osteoporosis despite their young age. Two of the three patients with COPD had recurrent spontaneous pneumothorax. Two patients had three or more exacerbations per year that required hospitalization. All three patients received at baseline long-acting beta2-agonist (LABA) and anti-cholinergic (LAMA) drugs, but not inhaled corticosteroid (ICS). Chest CT imaging showed severe apical centrilobular emphysema, basal bronchiectasis, and increased wall thickness ( Figure 1A, left panel). Lung function declined fast in these patients with COPD. The mean change in FEV1 (mL/year) was −25.3 (SD, 43.3) mL/year ( Figure 1A, right panel).
At the time of inclusion, all three patients with COPD were on the lung transplantation waiting list. At the time of manuscript submission, patient iCOPD2 was admitted in the intensive care unit for a severe COPD exacerbation that required mechanical ventilation. Patient iCOPD8 refused lung transplantation. Patient iCOPD9 was programmed for single lung transplantation (pre-transplant pleurodesis due to iterative pneumothorax). The lung transplant was delayed for more than 24 months due to overweight. Recently, a localized lung adenocarcinoma was discovered. Bronchoscopic lung volume reduction has been proposed to this patient due the severe emphysema phenotype.