The development of pig lungs undergoes embryonic, pseudo glandular, canalicular, and alveolar stages around embryonic day 19 (E19), E25, E60, and E90, respectively (McGeady et al., 2017; Shimamura et al., 2022). The developmental stage at which pig parietal mesothelial progenitor cells (MPCs) could be efficiently harvested was unknown. We harvested the parietal MPCs from the E80 canalicular stage thorax to have enough cell numbers.

To harvest a WT1⁺ developing MPC efficiently, we compared several methods previously reported (Kienzle et al., 2018; Kawai et al., 2019; Pruett et al., 2020; Mierzejewski et al., 2021), including collecting pleural fluid, pinching porcine thoracic walls with tweezers, scaring it with scrapers, or trypsinizing the porcine thoracic wall. Among those methods, trypsinization with a 0.05% trypsin inside the E80 thoracic walls showed the highest yield of MPC collection (Figure 1A). Interestingly, 0.25% trypsin treatment to the thorax did not expand the MPC (Supplementary Figure S1A). Previous papers showed the requirement of EGF for culturing MPCs (Pruett et al., 2020; Mierzejewski et al., 2021). Contrary to expectations, MPC culture with EGF did not offer an apparent effect on MPC colony expansion (Supplementary Figure S1C). To expand MPC efficiently, we coated the cell culture dish with extracellular matrix (ECM) molecules (type I collagen (Col I) and hyaluronic acid (HA)), given their expression in adult mature mesothelial cells (Breborowicz et al., 1996; Saed et al., 1999). We found that the isolated MPC showed the sustained expression of Col I expression and its receptor, integrin beta 1 (ITGB1), but a relatively low expression of HA receptor (CD44) (Figure 1B). Indeed, Col I coating significantly enhanced MPC expansion compared to HA coating (HA) and an uncoated control (Figures 1C, D). Since the gelatin and Col I share the integrin-binding motif RGD sequence (Davidenko et al., 2016), we cultured the MPC on the gelatin-coated dish and confirmed its efficacy in expanding MPCs (Breborowicz et al., 1996), showing WT1⁺ cells (76.7%), α-SMA⁺ cells (7.5%), WT1⁻α-SMA^- cells (3.7%), and WT1⁻α-SMA^- cells (19.5%) (Figure 1E). These results indicate that most cells are WT1⁺ cells. Based on this, we performed all downstream analyses on the gelatin-coated dish. Additionally, we confirmed that mouse MPC can be collected and expanded well after the trypsinization directly on the E17.5 mouse canalicular ∼ sacculation stage thorax, noting that 0.25% trypsin was more effective for mouse MPCs than 0.05% (Supplementary Figure S1D). To compare the efficacy of MPC isolation with another method, we isolated MPCs from WT1-lineage tracing mice prepared by crossing WT1^CreERT2/+ x Rosa26^{tdTomato/tdTomato} mice (Supplementary Figure S2). Briefly, we injected Tamoxifen at E15.5 and E16.5 and examined the proportion of tdTomato cells in lung mesenchyme and thorax fraction at E17.5 by FACS analysis. After the exclusion of hematopoietic cells (CD45), epithelial cells (EpCAM), and endothelial cells (PECAM) as previously reported (Miura et al., 2023), tdTomato⁺ cells were sorted from the lung mesenchyme (Supplementary Figure S2B). Compared with lung mesenchyme, the proportion of tdTomato⁺ cells sorted from the thorax surface was relatively higher (2.70% ± 2.14) but insignificant, while the cell number was very low due to the tissue size. The sorted tdTomato⁺ MPCs expressed WT1 after 3 days of culture (Supplementary Figure S2C). However, it was challenging to expand them efficiently due to the low number of sorted cells (Supplementary Figure S2C). Given that we obviously observed more number of the cells in our method (Supplementary Figure S1E), these results underscore the robustness and effectiveness of our trypsinization-based protocol over the sorting-based method for isolating parietal MPCs in development.

While the role of FGF2 and PDGF in adult mesothelial cell proliferation is known, their impact during development is little known (Mutsaers et al., 1997). To confirm each molecule’s effect on pig developing MPCs, we cultured pig MPC with FGF2 and PDGF-BB for 3 days (Figure 2). PDGF-BB was chosen as the signaling molecule for the PDGF signaling pathway due to its binding potential to all PDGF receptors (Östman, 2017). We found that FGF2 and PDGF-BB treatment increased total cell number as well as the WT1⁺ cell numbers compared to the basal condition control (Figures 2A–D). Ki67 immunostaining confirmed that FGF2 and PDGF-BB significantly increased proliferating cell numbers (Figures 2A, B, E). Notably, FGF2 and PDGF-BB induced a more than four times increase in proliferating Ki67⁺WT1⁺ MPC proportion compared with the control in the short-term culture (Figure 2F). In contrast, the treatment with SU5402, a FGFR inhibitor, and CP 673451, a PDGFR inhibitor, significantly decreased both total and WT1 cell numbers (Figures 2C, D) by inducing 30–40% of cell death, labeled by cleaved caspase 3 (CASP3) 1-day post-treatment (Supplementary Figure S3). These results suggested that the effect of endogenous FGF2 and PDGF activation cultured in the basal medium impacts ∼40% of pig MPC survival and that FGF2 and PDGF signaling may be essential for WT1⁺ MPC maintenance. To investigate the effect of FGF2 and PDGF on MPC pool expansion in the long term, we cultured the MPCs with FGF2 or PDGF-BB for 14 days and analyzed WT1 mRNA expression by qPCR (Figures 2G–I). We found that FGF2 maintained WT1 mRNA expression more than 5 times fold change compared to the control during long-term culture (Figure 2H), while the effect of PDGF-BB pool expansion did not significantly influence the WT1 mRNA expression compared to the control over time (Figure 2I). These results suggest that FGF2 efficiently expands the pig MPC pools, but the PDGF-BB effect on the expansion is temporal and limited.

Under the pig MPC control culture condition, WT1⁻α-SMA⁺ cells were observed (5.8% ± 3.3%) (Figure 2B). We speculated that pig WT1⁺ MPCs could differentiate into smooth muscle cells (SMCs), given that mouse visceral lung mesothelial cells differentiate into smooth muscle cells during mouse lung development (De Langhe et al., 2008; Que et al., 2008). To find which signaling molecules induce pig MPC differentiation into SMC, we cultured pig MPC with various small molecules and inhibitors with different concentrations and screened α-SMA mRNA expression by qPCR analysis (Supplementary Figure S4A). We discovered that the BMP4 and ascorbic acid (AA) condition enhanced α-SMA mRNA expression compared to control among the tested conditions. Since BMP4 more dramatically induced SMC differentiation than AA, we focused on further analyses of BMP signaling. qPCR analyses found that BMP4 treatment showed significantly higher α-SMA mRNA induction both in short-term and long-term cultures, while BMP4 treatment had a transient effect on WT1 mRNA increase only in the short term but did not sustain its impact in the long term (Figures 3B, C). In contrast, dorsomorphin, a BMP4 inhibitor, significantly reduced α-SMA mRNA expression with no significant change in WT1 mRNA expression (Figure 3B). Since the kinetics of WT1 and α-SMA mRNA by BMP4 treatment indicated the MPC differentiation into SMC, we investigated the detailed cell fate change from MPC to SMC by immunostainings in short-term culture (Figure 3A). Consistent with the qPCR observations, immunostaining analysis showed a significantly increased α-SMA⁺ cell proportion (Control: 7.8% ± 1.7% vs. BMP4: 31.4% ± 1.4%) and the number by BMP4 treatment (Figures 3A, D, E), while dorsomorphin significantly reduced the α-SMA⁺ SMC proportion (6.7% ± 3.0%). Unlike FGF2 and PDGF-BB (Figure 2), BMP4 treatment did not alter the total cell number, WT1⁺ MPC numbers, or WT1⁺ proportion but significantly increased Ki67⁺ cells (Figures 3F–H) while inducing about 20% of CASP3⁺ cell death, which might be the cell selection step (Supplementary Figure S3). Indeed, BMP4 selectively eliminates the WT1^-Ki67⁻α-SMA^- unknown cell type while dorsomorphin significantly increased it (Figure 3D). Intriguingly, we observed a significantly increased proportion of WT1⁺α-SMA⁺ cells in WT1⁺ MPCs (control: 9.9% ± 1.9% vs. BMP4 group: 46.4% ± 6.4%) by BMP4 treatment (Figure 3I), but proportion of WT1⁻α-SMA⁺ in SMCs (control: 30.7% ± 15.4% vs. BMP4 group: 43.7% ± 8.6%) (Figure 3J) was not significantly changed. On the other hand, we did not observe any change in the proportion of WT1⁺α-SMA⁺ in α-SMA⁺ cells (Figure 3K). These results indicate that BMP4 treatment primes the pig mesothelial progenitor pools to co-express WT1 and α-SMA, facilitating MPC differentiation into SMCs. Based on these results, including long-term culture, we concluded that the pivotal role of BMP4 is to induce pig parietal MPC differentiation into α-SMA⁺ SMC with losing WT1 expression.

We observed pig MPC progenitor pool regulation by FGF2 and PDGF-BB (Figure 2) and differentiation into α-SMA⁺ SMC by BMP4 (Figure 3), but it was unclear whether FGF2 and PDGF-BB influence the SMC pools. To address this, we performed qPCR analyses. We found that the decreased α-SMA mRNA expression by the FGF2 or PDGF-BB over time (Figures 4A, B, Supplementary Figures S4), and the further analysis of IF data showed that the proportion of α-SMA⁺ cells was significantly reduced by the FGF2 or PDGF-BB treatment (Control vs. FGF2 vs. PDGF-BB groups: 7.8% ± 1.7% vs. 2.5% ± 0.5% vs. 3.2% ± 0.4%), while BMP4 significantly induced α-SMA⁺ cells (31.4% ± 1.4%) (Figure 4C). In particular, PDGF-BB showed a dramatic decrease of α-SMA mRNA than FGF2 (Figure 4B). While there were no significant changes in the proportion of proliferating α-SMA⁺ cells, the proportion of WT1⁺α-SMA⁺ cells was significantly decreased by the FGF2 or PDGF treatment (Control vs. FGF2 vs. PDGF-BB groups: 9.9% ± 1.9% vs. 3.8% ± 0.9% vs. 3.4% ± 1.3%) (Figures 4D, E). These results indicate that FGF2 and PDGF play a central role in pig MPC progenitor pool expansion by inhibiting the induction of WT1⁺α-SMA⁺ primed cells, leading to α-SMA⁺ smooth muscle cells (Figure 4F).

Since we found FGF2 and PDGF suppressed BMP4-mediated MPC differentiation into SMC (Figures 2–4), we cultured pig MPCs with the combination of FGF2 and BMP4 (FGF2 + BMP4) or PDGF-BB and BMP4 (PDGF-BB + BMP4) to investigate the potential counter effect. We found that the MPC culture with FGF2 + BMP4 and PDGF-BB + BMP4 significantly suppressed the BMP4-mediated MPC differentiation into SMC with lower α-SMA mRNA expression than the BMP4 group (Figure 5A). This mRNA expression trend was the same in the long-term culture (Figure 5B). Although the short-term treatment with FGF2 + BMP4 and PDGF-BB + BMP4 showed a decrease in WT1 mRNA expression (Figure 5A), the long-term effect with FGF2 + BMP4 exhibited an increase in the WT1 mRNA expression compared to controls (Figure 5B), consistent with the FGF2 effect (Figure 2). The long-term effect of PDGF-BB + BMP4 did not impact the WT1 mRNA expression. Interestingly, the FGF2 + BMP4 or PDGF-BB + BMP4 condition induced more cell proliferation with a higher total cell number than the BMP4 group in the short term (Figures 5C–G). In contrast, FGF2 + PDGF-BB and PDGF-BB + BMP4 conditions significantly increased WT1⁺ MPCs and proliferating cell numbers than the control condition in the short-term but could not sustain WT1 mRNA expression in the long-term (Figures 5A, E, F). FGF2 + PDGF-BB and PDGF-BB + BMP4 conditions significantly decreased α-SMA⁺ cells and showed no increase of primed WT1⁺α-SMA⁺ cells in WT1⁺ cells (Figures 5G, H). As we expected, there was no significant change in WT1⁺α-SMA⁺ cells in α-SMA⁺ cells (Figure 5I). These results suggest the critical role of FGF2 in maintaining the MPC pool and its self-renewal that counteracts the BMP signaling effects on pig MPC differentiation into SMC.

During development, mesenchymal β-catenin signaling controls parabronchial smooth muscle cell (PSMC) progenitors in the sub-mesothelial mesenchyme (De Langhe et al., 2008). Wnt signaling is involved in the outer mesothelial pool size of the zebrafish swimbladder during development (Davidenko et al., 2016). However, the molecular characterization of pig MPCs and their maturation during pig lung development have been little studied. To address this issue, we performed immunostaining of WT1 and CALB2 in pig and mouse lung development (Supplementary Figure S5). Developing porcine pleural mesothelial cells expressed high levels of WT1 in the E26 early pseudoglandular stage of porcine lungs, but the relative expression level in the peripheral layer of the lungs was decreased to less than 1/4 of signal intensity in the later stage (Supplementary Figure S5A). In contrast, CALB2 expression was not detected in the peripheral layer of the pig lungs in the E26 and E40 early pseudoglandular stage but appeared in the canalicular stage and afterward (Supplementary Figure S5D). These results indicate that CALB2 is the marker for mesothelial cell maturation during porcine lung development. Intriguingly, CALB2-labeled non-peripheral cells were also observed around the sub-pleural region of the porcine lungs on days 90 and 104 (Supplementary Figure S5D), reminiscent of mouse lung mesothelial cell migration inside of the developing lungs using WT1 lineage tracing analysis (Dixit et al., 2013). We also confirmed that the WT1 expression pattern was also similar during mouse lung development, supported by previous studies (Que et al., 2008; Dixit et al., 2013) (Supplementary Figure S5C), while CALB2 started to be expressed in the sub-peripheral layer from the E14.5 pseudoglandular stage in mouse lung development (Supplementary Figure S5F).

To investigate the common MPC maturation markers across the species, we revisited the deposit single-cell RNA-seq (scRNA-seq) database of developing human (He et al., 2022) and mouse (Negretti et al., 2021) lung mesenchyme (Supplementary Figure S6). We found that WT1 was highly expressed in the early pseudoglandular stage but decreased its expression in the late pseudoglandular and canalicular stages of human and mouse-developing lungs. CALB2, a mature mesothelial cell marker, was slightly observed but not abundant in human lung development. During mouse lung development, CALB2 was observed in non-mesothelial cells. In contrast, mesothelin (MSLN) expression was observed in the late pseudoglandular stage of developing human lungs to the canalicular stage while around the E18 sacculation stage and afterward in the mouse lungs. These results suggest that decreased expression of WT1 and increased MSLN are the evolutionarily conserved markers for MPC maturation, but CALB2 is a pig-specific unique marker for MPC maturation. Based on these results, we examined pig MPC maturation in an in vitro study using WT1, CALB2, and MSLN.

We performed qPCR to screen the most potent signaling molecules regulating pig MPC maturation to CALB2⁺ and MSLN⁺ mature mesothelial cells (Supplementary Figure S4B). Among them, we found that most signaling molecules induced the upregulation of CALB2 and MSLN mRNA. In particular, the GSK3β inhibitor that acts as a Wnt activator (CHIR) showed the most dramatic increase in CALB2 mRNA expression. Thus, we focused on analyzing Wnt signaling using CHIR in the pig MPC maturation. Three days of short-term CHIR treatment increased WT1, CALB2, and MSLN mRNA expressions, while the long-term CHIR treatment lost WT1⁺ MPC pools but relatively sustained CALB2 expression (Figure 6A). Since high WT1 mRNA expression is the landmark for immature MPC pool expansion, these results indicate that the pig MPC maturation by CHIR occurred as a long-term effect (Figure 6A). Interestingly, we also found that long-term treatment with FGF2 or BMP4 significantly increased MSLN mRNA expression compared to the control (Figure 6B). However, FGF2 did not increase the mRNA expression of MSLN and CALB2 in a dose-dependent manner in short-term culture, while BMP4 induced CALB2 mRNA expression in a dose-dependent manner (Supplementary Figure S4C). Furthermore, the CALB2 mRNA upregulation by FGF2 or BMP4 was transient and relatively limited in the long-term treatment compared to the CHIR treatment (Figure 6B). Consistent with the qPCR results, the CALB2 immunostaining exhibited a consistent trend with qPCR results, indicating the increased CALB2⁺ cells by CHIR treatment (Figures 6C, D). As shown in the PDGF-BB effect, CHIR induced Ki67⁺ proliferative WT1⁺ cells and significantly increased total cell number compared to control (Supplementary Figure S7A), while no WT1⁺ cell number or proportional change and reduced α-SMA⁺ cell number (Supplementary Figure S7D). These results indicate that Wnt signaling activation induces pig MPC maturation into MSLN⁺ CALB2⁺ cells, corresponding to the expression pattern of CALB2 in porcine lung development.

All surgical procedures were conducted under the approval of the Columbia University Institutional Animal Care and Use Committee and USAMRMC Animal Care and Use Review Office (ACURO). For pig experiment, Timed-pregnant Yucatan miniature sows were obtained from Sinclair BioResources. For mouse experiment, CD-1 mice (male, 8 weeks), (female, 8 weeks) were purchased from Charles River Laboratories. For WT1 lineage-tracing mouse analysis, WT1^{tm2(cre/ERT2)Wtp} mouse (male, 8 weeks), Gt (ROSA)26Sor^{tm14(CAG-tdTomato)Hze} (female, 8 weeks) were purchased from Jackson Laboratory (Farmington, CT).

E80 Yucatan pig embryos were surgically harvested from Yucatan pig sows. After euthanasia, the thorax was harvested. Two methods for MPC isolation were compared as shown in Figure 1: 1) Mesothelial tissue was isolated from the medial side of the chest wall of E80 pigs with a cell scraper (Fisher Scientific, Waltham, MA) and incubated in a 50 mL tube containing 2 mL of 0.05% or 0.25% trypsin-EDTA (Thermofisher Scientific, Waltham, MA) solution at 37°C for 20 min 2) 0.05% or 0.25% trypsin-EDTA were directly treated on the thoracic wall, followed by incubation for at 37°C for 20 min. After trypsin-EDTA treatment in either method, the dissociated cell pellet was washed with PBS × 3 times by centrifugation (350 × g for 5 min) at 4°C. The cell pellet was then incubated with Red Blood Cell (RBC) lysis buffer (Biolegend, San Diego, CA) at 4°C for 5 min. After centrifugation (350 × g for 5 min, at 4°C), the supernatant was aspirated and the cell pellet was three times with PBS. After washing with PBS, the cell pellet was filtered through a cell strainer (pore size 40 um, MTC Bio) and seeded onto type I collagen (10 ug/cm², rat tail derived, Sigma-Aldrich, Burlington, MA)-coated six well tissue culture plates. MPCs (P0) were seeded in MPC culture medium (DMEM (high glucose, Gibco) + 10% FBS (Clytia, Marlborough, MA) + 1% pen/strep (Gibco)) for 7 days. During culture, the MPC culture medium was replaced every other day. For passage, MPCs were washed with PBS and dissociated in 0.05% trypsin-EDTA for 5 min at 37°C. For MPC culture and analysis, passage 6-8 MPCs grown on gelatin-coated tissue culture plates were used.

Mouse parietal MPCs were isolated from the thorax of E17.5 embryos with the pre-treatment of 0.5 mL of 0.05% or 0.25% trypsin-EDTA onto the median side of the thorax. Then, the parietal mouse MPCs were treated as pig MPC isolation. The mouse MPC was cultured on a type I collagen (10 ug/cm²)-coated 24well tissue culture plate in MPC culture medium with the replacement of the cell culture media every other day. Mouse MPCs isolated from E17.5 WT1 lineage-tracing mouse embryo were conducted with the same way. For quantification of tdTomato⁺ mouse parietal MPCs isolated from E17.5 WT1 lineage-tracing mouse (3 mg/kg tamoxifen administration into pregnant mother at E15.5, E16.5 by oral gavage feeding) by flow cytometry. The isolated mouse MPCs were stained with DAPI (Cayman Chemical, Ann Arbor, MI), rat to mouse EpCAM antibody (BV711-conjugated, Biolegend), rat to mouse PECAM antibody (APC-conjugated, Biolegend), CD45 (BV605-conjugated, BD Biosciences, San Jose, CA), and subsequently analyzed by flow cytometry (Sony MA900, Sony Biotechnology, San Jose, CA). The sorted CD45⁻ EpCAM⁻ PECAM⁻ tdTomato⁺ fractions were cultured on a type I collagen-coated 24 well tissue culture plate. After 3 days of culture, the cell nucleus was stained with WT1 and NucBlue™ Live ReadyProbes™ Reagent (Fisher Scientific). More detailed information of antibodies were described in Supplementary Table S2. The stained cells were visualized with a Leica DMI microscope (Leica, New York, NY).

To investigate the MPC cell fate by signaling molecules, MPCs were cultured in the MPC culture medium with various signaling molecules (FGF2 (2/10/20 ng/mL) (Peprotech, Cranbury, NJ), PDGF-BB (1/5/15 ng/mL) (R&D systems, Minneapolis, MN), BMP4 (5/25/50 ng/mL) (R&D systems), retinoic acid (0.1/0.5/1 uM) (RA, Sigma-Aldrich), CHIR99021 (0.3/1.5/3 uM) (MedChem Express, Hoboken, NJ), ascorbic acid (5/25/50 ug/mL) (AA, Thermofisher Scientific), purmorphamine (0.2/1/2uM) (sonic hedgehog (Shh) activator, Tocris, Westwoods Bus Park Ellisville, MO)) and the inhibitors (SU5402 as FGFR inhibitor (30 nM) (MedChem Express), CP673451 as PDGFR inhibitor (4 nM) (MedChem Express), and dorsomorphin (2 uM) (Tocris) for 3, 10, or 14 days. During MPC culture with signaling molecules for immunofluorescence (IF) and qPCR analysis, 20 ng/mL of FGF2, 15 ng/mL of PDGF-BB, 50 ng/mL of BMP4, and three uM of CHIR 99021 were used. The MPC culture medium, including signaling molecules, was replaced every other day and the MPCs were passaged at day 3, 6, and 10 to avoid full confluency.

mRNA was isolated from MPCs with Direct-zol RNA Microprep isolation kit (Zymo Research, Irvine, CA) after lysis of MPCs with IBI isolate total reagent (IBI Scientific, Dubuque, IA). For cDNA synthesis, the isolated mRNA was mixed with PrimeScript RT Master Mix (Takara bio, San Jose, CA), followed by cDNA synthesis protocol. For RT-qPCR analysis, the synthesized cDNA was mixed with qPCR primers (see each primer sequence in Supplementary Table S1) and Luna universal qPCR Master Mix (New England Biolabs (NEB), Ipswich, MA). RT-qPCR was conducted with Quantstudio (Thermofisher Scientific). mRNA expression of each gene was normalized with the housekeeping gene (GAPDH). The relative mRNA expression of the genes was normalized with the control group (MPC culture in DMEM +10% FBS +1% pen/strep).

For cell sample preparation, MPCs were fixed with 3.7% paraformaldehyde (PFA) for 10 min at room temperature. For tissue sample preparation, 10um-frozen sectioned tissue samples were washed with PBS 3 times, followed by antigen retrieval with citrate-based buffer (Vector Laboratories) in the microwave for 8 min. After washing the cells and the tissue samples with PBS 3 times, the primary antibodies in dilution solution (0.25% triton X-100 (Sigma-Aldrich) + 0.75% BSA in PBS) were treated to the samples and incubated at 4°C for overnight. After 3 times PBS wash on the following day, the secondary antibodies and DAPI were treated (0.75% BSA in PBS) for 1 h at room temperature. Then, the sample was mounted with a coverglass, anti-fade reagent (Invitrogen). For pig cell/tissue CALB2 staining, primary antibody-treated samples were treated with HRP conjugated anti-chicken antibody (in PBS) and incubated for 30 min at room temperature. After PBS wash, Cy3 tyramide (AAT Bioquest, Pleasanton, CA) (1:1000 diluted in 100 mM borate +0.1% Tween-20 + 0.003% H₂O₂ solution (pH 8.5)) was treated in the samples and incubated for 15 min at room temperature in the dark. After PBS wash, the samples were mounted with a coverglass and an anti-fade reagent (Invitrogen). The cell samples were visualized with a Leica DMI microscope. The tissue samples were visualized with a Zeiss confocal microscope (Zeiss, White Plains, NY). More detailed information of the used antibodies in this study was described in Supplementary Table S2.

For human and mouse RNA-seq data analysis, we utilized the database from the previous studies. The hyperlink of the database is described in Data Availability Statement.

Quantification of cell number in the phase contrast images was conducted by ImageJ. For immunostained cell (single-immunostained and co-immunostained cell population) and DAPI-stained cell counting from IF images, Cellpose software was used. The mean fluorescence intensity (MFI) of each IF sample was measured in the non-overlapping random fields using ImageJ software. Data analysis was performed using Prism 10. Data acquired by performing biological replicates ((n = 3) for RT-qPCR and phase contrast images, (n = 4) for IF images) of three or four independent experiments are presented as the mean ± standard deviation (SD). Statistical significance was determined using a one-way ANOVA or a two-tailed t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: non-significant.